From 54fb2a5cd0baf8e97d743de411e2f832d1afa68d Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 24 Nov 2015 14:30:56 -0500 Subject: drm/amdgpu: call hpd_irq_event on resume Need to call this on resume if displays changes during suspend in order to properly be notified of changes. Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index d5b421330145..58cb6987b078 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1788,6 +1788,7 @@ int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon) } drm_kms_helper_poll_enable(dev); + drm_helper_hpd_irq_event(dev); if (fbcon) { amdgpu_fbdev_set_suspend(adev, 0); -- cgit From 288912cb95d156ad9fbb626b0b05d714c2ec8c8e Mon Sep 17 00:00:00 2001 From: Jammy Zhou Date: Tue, 24 Nov 2015 16:55:20 +0800 Subject: drm/amdgpu: use $(src) in Makefile (v2) This can solve the path problem when compile amdgpu with DKMS. v2: agd: rebase on current drm-next Signed-off-by: Jammy Zhou Acked-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/Makefile | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 04c270757030..a5c3aa0de773 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -2,10 +2,12 @@ # Makefile for the drm device driver. This driver provides support for the # Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. -ccflags-y := -Iinclude/drm -Idrivers/gpu/drm/amd/include/asic_reg \ - -Idrivers/gpu/drm/amd/include \ - -Idrivers/gpu/drm/amd/amdgpu \ - -Idrivers/gpu/drm/amd/scheduler +FULL_AMD_PATH=$(src)/.. + +ccflags-y := -Iinclude/drm -I$(FULL_AMD_PATH)/include/asic_reg \ + -I$(FULL_AMD_PATH)/include \ + -I$(FULL_AMD_PATH)/amdgpu \ + -I$(FULL_AMD_PATH)/scheduler amdgpu-y := amdgpu_drv.o -- cgit From 7946b878038d5113262d6eae09ff22fb2956351f Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 24 Nov 2015 10:14:28 -0500 Subject: drm/amdgpu: add a callback for reading the bios from the rom directly MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This is necessary when the vbios image is not directly accessible via the rom BAR or legacy vga location. Reviewed-by: Jammy Zhou Reviewed-by: Christian König Reviewed-by: Monk Liu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 251b14736de9..1799973e7744 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1826,6 +1826,8 @@ struct amdgpu_cu_info { */ struct amdgpu_asic_funcs { bool (*read_disabled_bios)(struct amdgpu_device *adev); + bool (*read_bios_from_rom)(struct amdgpu_device *adev, + u8 *bios, u32 length_bytes); int (*read_register)(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 reg_offset, u32 *value); void (*set_vga_state)(struct amdgpu_device *adev, bool state); @@ -2232,6 +2234,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) #define amdgpu_asic_set_vce_clocks(adev, ev, ec) (adev)->asic_funcs->set_vce_clocks((adev), (ev), (ec)) #define amdgpu_asic_get_gpu_clock_counter(adev) (adev)->asic_funcs->get_gpu_clock_counter((adev)) #define amdgpu_asic_read_disabled_bios(adev) (adev)->asic_funcs->read_disabled_bios((adev)) +#define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l)) #define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v))) #define amdgpu_asic_get_cu_info(adev, info) (adev)->asic_funcs->get_cu_info((adev), (info)) #define amdgpu_gart_flush_gpu_tlb(adev, vmid) (adev)->gart.gart_funcs->flush_gpu_tlb((adev), (vmid)) -- cgit From 1eb22bd38ac58a1be6e7e0ae6dd4ab6d604a27d6 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 24 Nov 2015 10:34:45 -0500 Subject: drm/amdgpu: add read_bios_from_rom callback for CI parts MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Read the vbios image directly from the rom. Reviewed-by: Jammy Zhou Reviewed-by: Christian König Reviewed-by: Monk Liu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cik.c | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index 484710cfdf82..61689f094ba8 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -929,6 +929,37 @@ static bool cik_read_disabled_bios(struct amdgpu_device *adev) return r; } +static bool cik_read_bios_from_rom(struct amdgpu_device *adev, + u8 *bios, u32 length_bytes) +{ + u32 *dw_ptr; + unsigned long flags; + u32 i, length_dw; + + if (bios == NULL) + return false; + if (length_bytes == 0) + return false; + /* APU vbios image is part of sbios image */ + if (adev->flags & AMD_IS_APU) + return false; + + dw_ptr = (u32 *)bios; + length_dw = ALIGN(length_bytes, 4) / 4; + /* take the smc lock since we are using the smc index */ + spin_lock_irqsave(&adev->smc_idx_lock, flags); + /* set rom index to 0 */ + WREG32(mmSMC_IND_INDEX_0, ixROM_INDEX); + WREG32(mmSMC_IND_DATA_0, 0); + /* set index to data for continous read */ + WREG32(mmSMC_IND_INDEX_0, ixROM_DATA); + for (i = 0; i < length_dw; i++) + dw_ptr[i] = RREG32(mmSMC_IND_DATA_0); + spin_unlock_irqrestore(&adev->smc_idx_lock, flags); + + return true; +} + static struct amdgpu_allowed_register_entry cik_allowed_read_registers[] = { {mmGRBM_STATUS, false}, {mmGB_ADDR_CONFIG, false}, @@ -2267,6 +2298,7 @@ int cik_set_ip_blocks(struct amdgpu_device *adev) static const struct amdgpu_asic_funcs cik_asic_funcs = { .read_disabled_bios = &cik_read_disabled_bios, + .read_bios_from_rom = &cik_read_bios_from_rom, .read_register = &cik_read_register, .reset = &cik_asic_reset, .set_vga_state = &cik_vga_set_state, -- cgit From 95addb2ae06f46cd13346ed87bbda3990438a69b Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 24 Nov 2015 10:37:54 -0500 Subject: drm/amdgpu: add read_bios_from_rom callback for VI parts MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Read the vbios image directly from the rom. Reviewed-by: Jammy Zhou Reviewed-by: Christian König Reviewed-by: Monk Liu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/vi.c | 33 +++++++++++++++++++++++++++++++++ 1 file changed, 33 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 2adc1c855e85..bf5bd931b8ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -376,6 +376,38 @@ static bool vi_read_disabled_bios(struct amdgpu_device *adev) WREG32_SMC(ixROM_CNTL, rom_cntl); return r; } + +static bool vi_read_bios_from_rom(struct amdgpu_device *adev, + u8 *bios, u32 length_bytes) +{ + u32 *dw_ptr; + unsigned long flags; + u32 i, length_dw; + + if (bios == NULL) + return false; + if (length_bytes == 0) + return false; + /* APU vbios image is part of sbios image */ + if (adev->flags & AMD_IS_APU) + return false; + + dw_ptr = (u32 *)bios; + length_dw = ALIGN(length_bytes, 4) / 4; + /* take the smc lock since we are using the smc index */ + spin_lock_irqsave(&adev->smc_idx_lock, flags); + /* set rom index to 0 */ + WREG32(mmSMC_IND_INDEX_0, ixROM_INDEX); + WREG32(mmSMC_IND_DATA_0, 0); + /* set index to data for continous read */ + WREG32(mmSMC_IND_INDEX_0, ixROM_DATA); + for (i = 0; i < length_dw; i++) + dw_ptr[i] = RREG32(mmSMC_IND_DATA_0); + spin_unlock_irqrestore(&adev->smc_idx_lock, flags); + + return true; +} + static struct amdgpu_allowed_register_entry tonga_allowed_read_registers[] = { {mmGB_MACROTILE_MODE7, true}, }; @@ -1368,6 +1400,7 @@ static uint32_t vi_get_rev_id(struct amdgpu_device *adev) static const struct amdgpu_asic_funcs vi_asic_funcs = { .read_disabled_bios = &vi_read_disabled_bios, + .read_bios_from_rom = &vi_read_bios_from_rom, .read_register = &vi_read_register, .reset = &vi_asic_reset, .set_vga_state = &vi_vga_set_state, -- cgit From f930b2e8628f4e6e818ea69a426a2b732679e15b Mon Sep 17 00:00:00 2001 From: "monk.liu" Date: Thu, 29 Oct 2015 15:33:06 +0800 Subject: drm/amdgpu: Use new read bios from rom callback MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Read the vbios directly from the rom. In some cases, e.g., virtualization, the rom is not available via the BAR or other means. Access it directly. This is an updated version of Monks original patch which uses family specific callbacks and unifies some of the validation checking. Reviewed-by: Christian König Reviewed-by: Jammy Zhou Signed-off-by: Monk Liu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c | 58 +++++++++++++++++++++++++++----- 1 file changed, 50 insertions(+), 8 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c index c44c0c6afd1b..80add22375ee 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c @@ -35,6 +35,13 @@ * BIOS. */ +#define AMD_VBIOS_SIGNATURE " 761295520" +#define AMD_VBIOS_SIGNATURE_OFFSET 0x30 +#define AMD_VBIOS_SIGNATURE_SIZE sizeof(AMD_VBIOS_SIGNATURE) +#define AMD_VBIOS_SIGNATURE_END (AMD_VBIOS_SIGNATURE_OFFSET + AMD_VBIOS_SIGNATURE_SIZE) +#define AMD_IS_VALID_VBIOS(p) ((p)[0] == 0x55 && (p)[1] == 0xAA) +#define AMD_VBIOS_LENGTH(p) ((p)[2] << 9) + /* If you boot an IGP board with a discrete card as the primary, * the IGP rom is not accessible via the rom bar as the IGP rom is * part of the system bios. On boot, the system bios puts a @@ -58,7 +65,7 @@ static bool igp_read_bios_from_vram(struct amdgpu_device *adev) return false; } - if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) { + if (size == 0 || !AMD_IS_VALID_VBIOS(bios)) { iounmap(bios); return false; } @@ -74,7 +81,7 @@ static bool igp_read_bios_from_vram(struct amdgpu_device *adev) bool amdgpu_read_bios(struct amdgpu_device *adev) { - uint8_t __iomem *bios, val1, val2; + uint8_t __iomem *bios, val[2]; size_t size; adev->bios = NULL; @@ -84,10 +91,10 @@ bool amdgpu_read_bios(struct amdgpu_device *adev) return false; } - val1 = readb(&bios[0]); - val2 = readb(&bios[1]); + val[0] = readb(&bios[0]); + val[1] = readb(&bios[1]); - if (size == 0 || val1 != 0x55 || val2 != 0xaa) { + if (size == 0 || !AMD_IS_VALID_VBIOS(val)) { pci_unmap_rom(adev->pdev, bios); return false; } @@ -101,6 +108,38 @@ bool amdgpu_read_bios(struct amdgpu_device *adev) return true; } +static bool amdgpu_read_bios_from_rom(struct amdgpu_device *adev) +{ + u8 header[AMD_VBIOS_SIGNATURE_END+1] = {0}; + int len; + + if (!adev->asic_funcs->read_bios_from_rom) + return false; + + /* validate VBIOS signature */ + if (amdgpu_asic_read_bios_from_rom(adev, &header[0], sizeof(header)) == false) + return false; + header[AMD_VBIOS_SIGNATURE_END] = 0; + + if ((!AMD_IS_VALID_VBIOS(header)) || + 0 != memcmp((char *)&header[AMD_VBIOS_SIGNATURE_OFFSET], + AMD_VBIOS_SIGNATURE, + strlen(AMD_VBIOS_SIGNATURE))) + return false; + + /* valid vbios, go on */ + len = AMD_VBIOS_LENGTH(header); + len = ALIGN(len, 4); + adev->bios = kmalloc(len, GFP_KERNEL); + if (!adev->bios) { + DRM_ERROR("no memory to allocate for BIOS\n"); + return false; + } + + /* read complete BIOS */ + return amdgpu_asic_read_bios_from_rom(adev, adev->bios, len); +} + static bool amdgpu_read_platform_bios(struct amdgpu_device *adev) { uint8_t __iomem *bios; @@ -113,7 +152,7 @@ static bool amdgpu_read_platform_bios(struct amdgpu_device *adev) return false; } - if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) { + if (size == 0 || !AMD_IS_VALID_VBIOS(bios)) { return false; } adev->bios = kmemdup(bios, size, GFP_KERNEL); @@ -230,7 +269,7 @@ static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev) break; } - if (i == 0 || adev->bios[0] != 0x55 || adev->bios[1] != 0xaa) { + if (i == 0 || !AMD_IS_VALID_VBIOS(adev->bios)) { kfree(adev->bios); return false; } @@ -319,6 +358,9 @@ bool amdgpu_get_bios(struct amdgpu_device *adev) r = igp_read_bios_from_vram(adev); if (r == false) r = amdgpu_read_bios(adev); + if (r == false) { + r = amdgpu_read_bios_from_rom(adev); + } if (r == false) { r = amdgpu_read_disabled_bios(adev); } @@ -330,7 +372,7 @@ bool amdgpu_get_bios(struct amdgpu_device *adev) adev->bios = NULL; return false; } - if (adev->bios[0] != 0x55 || adev->bios[1] != 0xaa) { + if (!AMD_IS_VALID_VBIOS(adev->bios)) { printk("BIOS signature incorrect %x %x\n", adev->bios[0], adev->bios[1]); goto free_bios; } -- cgit From a9906fde57b690c5e606daf8c5994850a5abfc84 Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Mon, 23 Nov 2015 10:32:37 +0100 Subject: drm/amdgpu: Use unlocked gem unreferencing MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit For drm_gem_object_unreference callers are required to hold dev->struct_mutex, which these paths don't. Enforcing this requirement has become a bit more strict with commit ef4c6270bf2867e2f8032e9614d1a8cfc6c71663 Author: Daniel Vetter Date: Thu Oct 15 09:36:25 2015 +0200 drm/gem: Check locking in drm_gem_object_unreference Cc: Alex Deucher Reviewed-by: Christian König Signed-off-by: Daniel Vetter Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c index 6fcbbcc2e99e..cfb6caad2a73 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c @@ -263,7 +263,7 @@ out_unref: } if (fb && ret) { - drm_gem_object_unreference(gobj); + drm_gem_object_unreference_unlocked(gobj); drm_framebuffer_unregister_private(fb); drm_framebuffer_cleanup(fb); kfree(fb); -- cgit From ccba7691a580a0967f60a512473ce699b9edac0d Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 24 Nov 2015 17:43:42 -0500 Subject: drm/amdgpu: add EDC support for CZ (v3) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This adds EDC support for CZ. EDC = Error Correction and Detection This code properly initializes the EDC hardware and resets the error counts. This is done in late_init since it requires the IB pool which is not initialized during hw_init. v2: fix the IB size as noted by Felix, fix shader pgm register programming v3: use the IB for the shaders as suggested by Christian Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 331 +++++++++++++++++++++++++++++++++- 1 file changed, 330 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index e1dcab98e249..07c1ec30259c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -964,6 +964,322 @@ static int gfx_v8_0_mec_init(struct amdgpu_device *adev) return 0; } +static const u32 vgpr_init_compute_shader[] = +{ + 0x7e000209, 0x7e020208, + 0x7e040207, 0x7e060206, + 0x7e080205, 0x7e0a0204, + 0x7e0c0203, 0x7e0e0202, + 0x7e100201, 0x7e120200, + 0x7e140209, 0x7e160208, + 0x7e180207, 0x7e1a0206, + 0x7e1c0205, 0x7e1e0204, + 0x7e200203, 0x7e220202, + 0x7e240201, 0x7e260200, + 0x7e280209, 0x7e2a0208, + 0x7e2c0207, 0x7e2e0206, + 0x7e300205, 0x7e320204, + 0x7e340203, 0x7e360202, + 0x7e380201, 0x7e3a0200, + 0x7e3c0209, 0x7e3e0208, + 0x7e400207, 0x7e420206, + 0x7e440205, 0x7e460204, + 0x7e480203, 0x7e4a0202, + 0x7e4c0201, 0x7e4e0200, + 0x7e500209, 0x7e520208, + 0x7e540207, 0x7e560206, + 0x7e580205, 0x7e5a0204, + 0x7e5c0203, 0x7e5e0202, + 0x7e600201, 0x7e620200, + 0x7e640209, 0x7e660208, + 0x7e680207, 0x7e6a0206, + 0x7e6c0205, 0x7e6e0204, + 0x7e700203, 0x7e720202, + 0x7e740201, 0x7e760200, + 0x7e780209, 0x7e7a0208, + 0x7e7c0207, 0x7e7e0206, + 0xbf8a0000, 0xbf810000, +}; + +static const u32 sgpr_init_compute_shader[] = +{ + 0xbe8a0100, 0xbe8c0102, + 0xbe8e0104, 0xbe900106, + 0xbe920108, 0xbe940100, + 0xbe960102, 0xbe980104, + 0xbe9a0106, 0xbe9c0108, + 0xbe9e0100, 0xbea00102, + 0xbea20104, 0xbea40106, + 0xbea60108, 0xbea80100, + 0xbeaa0102, 0xbeac0104, + 0xbeae0106, 0xbeb00108, + 0xbeb20100, 0xbeb40102, + 0xbeb60104, 0xbeb80106, + 0xbeba0108, 0xbebc0100, + 0xbebe0102, 0xbec00104, + 0xbec20106, 0xbec40108, + 0xbec60100, 0xbec80102, + 0xbee60004, 0xbee70005, + 0xbeea0006, 0xbeeb0007, + 0xbee80008, 0xbee90009, + 0xbefc0000, 0xbf8a0000, + 0xbf810000, 0x00000000, +}; + +static const u32 vgpr_init_regs[] = +{ + mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0xffffffff, + mmCOMPUTE_RESOURCE_LIMITS, 0, + mmCOMPUTE_NUM_THREAD_X, 256*4, + mmCOMPUTE_NUM_THREAD_Y, 1, + mmCOMPUTE_NUM_THREAD_Z, 1, + mmCOMPUTE_PGM_RSRC2, 20, + mmCOMPUTE_USER_DATA_0, 0xedcedc00, + mmCOMPUTE_USER_DATA_1, 0xedcedc01, + mmCOMPUTE_USER_DATA_2, 0xedcedc02, + mmCOMPUTE_USER_DATA_3, 0xedcedc03, + mmCOMPUTE_USER_DATA_4, 0xedcedc04, + mmCOMPUTE_USER_DATA_5, 0xedcedc05, + mmCOMPUTE_USER_DATA_6, 0xedcedc06, + mmCOMPUTE_USER_DATA_7, 0xedcedc07, + mmCOMPUTE_USER_DATA_8, 0xedcedc08, + mmCOMPUTE_USER_DATA_9, 0xedcedc09, +}; + +static const u32 sgpr1_init_regs[] = +{ + mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0x0f, + mmCOMPUTE_RESOURCE_LIMITS, 0x1000000, + mmCOMPUTE_NUM_THREAD_X, 256*5, + mmCOMPUTE_NUM_THREAD_Y, 1, + mmCOMPUTE_NUM_THREAD_Z, 1, + mmCOMPUTE_PGM_RSRC2, 20, + mmCOMPUTE_USER_DATA_0, 0xedcedc00, + mmCOMPUTE_USER_DATA_1, 0xedcedc01, + mmCOMPUTE_USER_DATA_2, 0xedcedc02, + mmCOMPUTE_USER_DATA_3, 0xedcedc03, + mmCOMPUTE_USER_DATA_4, 0xedcedc04, + mmCOMPUTE_USER_DATA_5, 0xedcedc05, + mmCOMPUTE_USER_DATA_6, 0xedcedc06, + mmCOMPUTE_USER_DATA_7, 0xedcedc07, + mmCOMPUTE_USER_DATA_8, 0xedcedc08, + mmCOMPUTE_USER_DATA_9, 0xedcedc09, +}; + +static const u32 sgpr2_init_regs[] = +{ + mmCOMPUTE_STATIC_THREAD_MGMT_SE0, 0xf0, + mmCOMPUTE_RESOURCE_LIMITS, 0x1000000, + mmCOMPUTE_NUM_THREAD_X, 256*5, + mmCOMPUTE_NUM_THREAD_Y, 1, + mmCOMPUTE_NUM_THREAD_Z, 1, + mmCOMPUTE_PGM_RSRC2, 20, + mmCOMPUTE_USER_DATA_0, 0xedcedc00, + mmCOMPUTE_USER_DATA_1, 0xedcedc01, + mmCOMPUTE_USER_DATA_2, 0xedcedc02, + mmCOMPUTE_USER_DATA_3, 0xedcedc03, + mmCOMPUTE_USER_DATA_4, 0xedcedc04, + mmCOMPUTE_USER_DATA_5, 0xedcedc05, + mmCOMPUTE_USER_DATA_6, 0xedcedc06, + mmCOMPUTE_USER_DATA_7, 0xedcedc07, + mmCOMPUTE_USER_DATA_8, 0xedcedc08, + mmCOMPUTE_USER_DATA_9, 0xedcedc09, +}; + +static const u32 sec_ded_counter_registers[] = +{ + mmCPC_EDC_ATC_CNT, + mmCPC_EDC_SCRATCH_CNT, + mmCPC_EDC_UCODE_CNT, + mmCPF_EDC_ATC_CNT, + mmCPF_EDC_ROQ_CNT, + mmCPF_EDC_TAG_CNT, + mmCPG_EDC_ATC_CNT, + mmCPG_EDC_DMA_CNT, + mmCPG_EDC_TAG_CNT, + mmDC_EDC_CSINVOC_CNT, + mmDC_EDC_RESTORE_CNT, + mmDC_EDC_STATE_CNT, + mmGDS_EDC_CNT, + mmGDS_EDC_GRBM_CNT, + mmGDS_EDC_OA_DED, + mmSPI_EDC_CNT, + mmSQC_ATC_EDC_GATCL1_CNT, + mmSQC_EDC_CNT, + mmSQ_EDC_DED_CNT, + mmSQ_EDC_INFO, + mmSQ_EDC_SEC_CNT, + mmTCC_EDC_CNT, + mmTCP_ATC_EDC_GATCL1_CNT, + mmTCP_EDC_CNT, + mmTD_EDC_CNT +}; + +static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) +{ + struct amdgpu_ring *ring = &adev->gfx.compute_ring[0]; + struct amdgpu_ib ib; + struct fence *f = NULL; + int r, i; + u32 tmp; + unsigned total_size, vgpr_offset, sgpr_offset; + u64 gpu_addr; + + /* only supported on CZ */ + if (adev->asic_type != CHIP_CARRIZO) + return 0; + + /* bail if the compute ring is not ready */ + if (!ring->ready) + return 0; + + tmp = RREG32(mmGB_EDC_MODE); + WREG32(mmGB_EDC_MODE, 0); + + total_size = + (((ARRAY_SIZE(vgpr_init_regs) / 2) * 3) + 4 + 5 + 2) * 4; + total_size += + (((ARRAY_SIZE(sgpr1_init_regs) / 2) * 3) + 4 + 5 + 2) * 4; + total_size += + (((ARRAY_SIZE(sgpr2_init_regs) / 2) * 3) + 4 + 5 + 2) * 4; + total_size = ALIGN(total_size, 256); + vgpr_offset = total_size; + total_size += ALIGN(sizeof(vgpr_init_compute_shader), 256); + sgpr_offset = total_size; + total_size += sizeof(sgpr_init_compute_shader); + + /* allocate an indirect buffer to put the commands in */ + memset(&ib, 0, sizeof(ib)); + r = amdgpu_ib_get(ring, NULL, total_size, &ib); + if (r) { + DRM_ERROR("amdgpu: failed to get ib (%d).\n", r); + return r; + } + + /* load the compute shaders */ + for (i = 0; i < ARRAY_SIZE(vgpr_init_compute_shader); i++) + ib.ptr[i + (vgpr_offset / 4)] = vgpr_init_compute_shader[i]; + + for (i = 0; i < ARRAY_SIZE(sgpr_init_compute_shader); i++) + ib.ptr[i + (sgpr_offset / 4)] = sgpr_init_compute_shader[i]; + + /* init the ib length to 0 */ + ib.length_dw = 0; + + /* VGPR */ + /* write the register state for the compute dispatch */ + for (i = 0; i < ARRAY_SIZE(vgpr_init_regs); i += 2) { + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1); + ib.ptr[ib.length_dw++] = vgpr_init_regs[i] - PACKET3_SET_SH_REG_START; + ib.ptr[ib.length_dw++] = vgpr_init_regs[i + 1]; + } + /* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */ + gpu_addr = (ib.gpu_addr + (u64)vgpr_offset) >> 8; + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2); + ib.ptr[ib.length_dw++] = mmCOMPUTE_PGM_LO - PACKET3_SET_SH_REG_START; + ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr); + ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr); + + /* write dispatch packet */ + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3); + ib.ptr[ib.length_dw++] = 8; /* x */ + ib.ptr[ib.length_dw++] = 1; /* y */ + ib.ptr[ib.length_dw++] = 1; /* z */ + ib.ptr[ib.length_dw++] = + REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1); + + /* write CS partial flush packet */ + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0); + ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4); + + /* SGPR1 */ + /* write the register state for the compute dispatch */ + for (i = 0; i < ARRAY_SIZE(sgpr1_init_regs); i += 2) { + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1); + ib.ptr[ib.length_dw++] = sgpr1_init_regs[i] - PACKET3_SET_SH_REG_START; + ib.ptr[ib.length_dw++] = sgpr1_init_regs[i + 1]; + } + /* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */ + gpu_addr = (ib.gpu_addr + (u64)sgpr_offset) >> 8; + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2); + ib.ptr[ib.length_dw++] = mmCOMPUTE_PGM_LO - PACKET3_SET_SH_REG_START; + ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr); + ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr); + + /* write dispatch packet */ + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3); + ib.ptr[ib.length_dw++] = 8; /* x */ + ib.ptr[ib.length_dw++] = 1; /* y */ + ib.ptr[ib.length_dw++] = 1; /* z */ + ib.ptr[ib.length_dw++] = + REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1); + + /* write CS partial flush packet */ + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0); + ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4); + + /* SGPR2 */ + /* write the register state for the compute dispatch */ + for (i = 0; i < ARRAY_SIZE(sgpr2_init_regs); i += 2) { + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1); + ib.ptr[ib.length_dw++] = sgpr2_init_regs[i] - PACKET3_SET_SH_REG_START; + ib.ptr[ib.length_dw++] = sgpr2_init_regs[i + 1]; + } + /* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */ + gpu_addr = (ib.gpu_addr + (u64)sgpr_offset) >> 8; + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 2); + ib.ptr[ib.length_dw++] = mmCOMPUTE_PGM_LO - PACKET3_SET_SH_REG_START; + ib.ptr[ib.length_dw++] = lower_32_bits(gpu_addr); + ib.ptr[ib.length_dw++] = upper_32_bits(gpu_addr); + + /* write dispatch packet */ + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3); + ib.ptr[ib.length_dw++] = 8; /* x */ + ib.ptr[ib.length_dw++] = 1; /* y */ + ib.ptr[ib.length_dw++] = 1; /* z */ + ib.ptr[ib.length_dw++] = + REG_SET_FIELD(0, COMPUTE_DISPATCH_INITIATOR, COMPUTE_SHADER_EN, 1); + + /* write CS partial flush packet */ + ib.ptr[ib.length_dw++] = PACKET3(PACKET3_EVENT_WRITE, 0); + ib.ptr[ib.length_dw++] = EVENT_TYPE(7) | EVENT_INDEX(4); + + /* shedule the ib on the ring */ + r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL, + AMDGPU_FENCE_OWNER_UNDEFINED, + &f); + if (r) { + DRM_ERROR("amdgpu: ib submit failed (%d).\n", r); + goto fail; + } + + /* wait for the GPU to finish processing the IB */ + r = fence_wait(f, false); + if (r) { + DRM_ERROR("amdgpu: fence wait failed (%d).\n", r); + goto fail; + } + + tmp = REG_SET_FIELD(tmp, GB_EDC_MODE, DED_MODE, 2); + tmp = REG_SET_FIELD(tmp, GB_EDC_MODE, PROP_FED, 1); + WREG32(mmGB_EDC_MODE, tmp); + + tmp = RREG32(mmCC_GC_EDC_CONFIG); + tmp = REG_SET_FIELD(tmp, CC_GC_EDC_CONFIG, DIS_EDC, 0) | 1; + WREG32(mmCC_GC_EDC_CONFIG, tmp); + + + /* read back registers to clear the counters */ + for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++) + RREG32(sec_ded_counter_registers[i]); + +fail: + fence_put(f); + amdgpu_ib_free(adev, &ib); + + return r; +} + static void gfx_v8_0_gpu_early_init(struct amdgpu_device *adev) { u32 gb_addr_config; @@ -4458,6 +4774,19 @@ static int gfx_v8_0_early_init(void *handle) return 0; } +static int gfx_v8_0_late_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int r; + + /* requires IBs so do in late init after IB pool is initialized */ + r = gfx_v8_0_do_edc_gpr_workarounds(adev); + if (r) + return r; + + return 0; +} + static int gfx_v8_0_set_powergating_state(void *handle, enum amd_powergating_state state) { @@ -4995,7 +5324,7 @@ static int gfx_v8_0_priv_inst_irq(struct amdgpu_device *adev, const struct amd_ip_funcs gfx_v8_0_ip_funcs = { .early_init = gfx_v8_0_early_init, - .late_init = NULL, + .late_init = gfx_v8_0_late_init, .sw_init = gfx_v8_0_sw_init, .sw_fini = gfx_v8_0_sw_fini, .hw_init = gfx_v8_0_hw_init, -- cgit From d033a6de80054139b4358db12cf6bb8d6cf58853 Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Thu, 5 Nov 2015 15:23:09 +0800 Subject: drm/amd: abstract kernel rq and normal rq to priority of run queue MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Allows us to set priorities in the scheduler. Signed-off-by: Chunming Zhou Reviewed-by: Christian König Reviewed-by: Junwei Zhang --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 11 +++++------ drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 +- 3 files changed, 7 insertions(+), 8 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 1799973e7744..5f975030bb72 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1044,7 +1044,7 @@ struct amdgpu_ctx_mgr { struct idr ctx_handles; }; -int amdgpu_ctx_init(struct amdgpu_device *adev, bool kernel, +int amdgpu_ctx_init(struct amdgpu_device *adev, enum amd_sched_priority pri, struct amdgpu_ctx *ctx); void amdgpu_ctx_fini(struct amdgpu_ctx *ctx); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index fec65f01c031..c1f2308b4115 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -25,7 +25,7 @@ #include #include "amdgpu.h" -int amdgpu_ctx_init(struct amdgpu_device *adev, bool kernel, +int amdgpu_ctx_init(struct amdgpu_device *adev, enum amd_sched_priority pri, struct amdgpu_ctx *ctx) { unsigned i, j; @@ -42,10 +42,9 @@ int amdgpu_ctx_init(struct amdgpu_device *adev, bool kernel, /* create context entity for each ring */ for (i = 0; i < adev->num_rings; i++) { struct amd_sched_rq *rq; - if (kernel) - rq = &adev->rings[i]->sched.kernel_rq; - else - rq = &adev->rings[i]->sched.sched_rq; + if (pri >= AMD_SCHED_MAX_PRIORITY) + return -EINVAL; + rq = &adev->rings[i]->sched.sched_rq[pri]; r = amd_sched_entity_init(&adev->rings[i]->sched, &ctx->rings[i].entity, rq, amdgpu_sched_jobs); @@ -103,7 +102,7 @@ static int amdgpu_ctx_alloc(struct amdgpu_device *adev, return r; } *id = (uint32_t)r; - r = amdgpu_ctx_init(adev, false, ctx); + r = amdgpu_ctx_init(adev, AMD_SCHED_PRIORITY_NORMAL, ctx); mutex_unlock(&mgr->lock); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 58cb6987b078..84775964b138 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1528,7 +1528,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, return r; } - r = amdgpu_ctx_init(adev, true, &adev->kernel_ctx); + r = amdgpu_ctx_init(adev, AMD_SCHED_PRIORITY_KERNEL, &adev->kernel_ctx); if (r) { dev_err(adev->dev, "failed to create kernel context (%d).\n", r); return r; -- cgit From 8cdacf4457670cbd341285a2c76c542ce19ad74b Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Mon, 30 Nov 2015 14:13:11 -0500 Subject: amdgpu/gfxv8: Add missing break to switch statement from states init code MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 07c1ec30259c..296471c4059c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -1923,6 +1923,7 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev) adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden; WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden); } + break; case CHIP_FIJI: for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) { switch (reg_offset) { -- cgit From 90bea0abf61d020ff8a33bde5f93fedfbaba06d6 Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 1 Dec 2015 11:47:21 -0500 Subject: amdgpu/gfxv8: Cleanup of gfx_v8_0_tiling_mode_table_init() (v2) Simplification and LOC reduction of function gfx_v8_0_tiling_mode_table_init() v2: remove spurious break bug: https://bugs.freedesktop.org/show_bug.cgi?id=93236 Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 2286 +++++++++++++-------------------- 1 file changed, 898 insertions(+), 1388 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 296471c4059c..5c0400255f6d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -1639,1407 +1639,917 @@ static int gfx_v8_0_sw_fini(void *handle) static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev) { + uint32_t *modearray, *mod2array; const u32 num_tile_mode_states = 32; const u32 num_secondary_tile_mode_states = 16; - u32 reg_offset, gb_tile_moden, split_equal_to_row_size; + u32 reg_offset; - switch (adev->gfx.config.mem_row_size_in_kb) { - case 1: - split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_1KB; - break; - case 2: - default: - split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_2KB; - break; - case 4: - split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_4KB; - break; - } + modearray = adev->gfx.config.tile_mode_array; + mod2array = adev->gfx.config.macrotile_mode_array; + + for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) + modearray[reg_offset] = 0; + + for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) + mod2array[reg_offset] = 0; switch (adev->asic_type) { case CHIP_TOPAZ: - for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 1: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 2: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 3: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 4: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 5: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 6: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 8: - gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | - PIPE_CONFIG(ADDR_SURF_P2)); - break; - case 9: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 10: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 11: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 13: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 14: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 15: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 16: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 18: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 19: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 20: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 21: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 22: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 24: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 25: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 26: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 27: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 28: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 29: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 7: - case 12: - case 17: - case 23: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden); - } - for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 1: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 2: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 3: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 4: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 5: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 6: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 8: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 9: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 10: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 11: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 12: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 13: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 14: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 7: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden); - } + modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | + PIPE_CONFIG(ADDR_SURF_P2)); + modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + + mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + + for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) + if (reg_offset != 7 && reg_offset != 12 && reg_offset != 17 && + reg_offset != 23) + WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]); + + for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) + if (reg_offset != 7) + WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]); + break; case CHIP_FIJI: - for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 1: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 2: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 3: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 4: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 5: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 6: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 7: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 8: - gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16)); - break; - case 9: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 10: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 11: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 12: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 13: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 14: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 15: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 16: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 17: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 18: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 19: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 20: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 21: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 22: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 23: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 24: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 25: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 26: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 27: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 28: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 29: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 30: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - default: - gb_tile_moden = 0; - break; - } - adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden); - } - for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 1: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 2: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 3: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 4: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 5: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 6: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 8: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 9: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 10: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 11: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 12: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 13: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 14: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_4_BANK)); - break; - case 7: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - } - adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden); - } + modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16)); + modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[12] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + + mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_4_BANK)); + + for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) + WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]); + + for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) + if (reg_offset != 7) + WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]); + break; case CHIP_TONGA: - for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 1: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 2: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 3: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 4: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 5: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 6: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 7: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 8: - gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16)); - break; - case 9: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 10: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 11: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 12: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 13: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 14: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 15: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 16: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 17: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 18: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 19: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 20: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 21: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 22: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 23: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 24: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 25: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 26: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 27: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 28: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 29: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 30: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P4_16x16) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden); - } - for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 1: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 2: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 3: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 4: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 5: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 6: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 8: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 9: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 10: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 11: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 12: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 13: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_4_BANK)); - break; - case 14: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | - NUM_BANKS(ADDR_SURF_4_BANK)); - break; - case 7: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden); - } + modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[7] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16)); + modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[12] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[17] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[23] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[30] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + + mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_4_BANK)); + mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) | + NUM_BANKS(ADDR_SURF_4_BANK)); + + for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) + WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]); + + for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) + if (reg_offset != 7) + WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]); + break; case CHIP_STONEY: - for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 1: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 2: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 3: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 4: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 5: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 6: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 8: - gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | - PIPE_CONFIG(ADDR_SURF_P2)); - break; - case 9: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 10: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 11: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 13: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 14: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 15: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 16: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 18: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 19: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 20: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 21: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 22: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 24: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 25: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 26: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 27: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 28: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 29: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 7: - case 12: - case 17: - case 23: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden); - } - for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 1: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 2: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 3: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 4: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 5: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 6: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 8: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 9: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 10: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 11: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 12: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 13: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 14: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 7: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden); - } + modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | + PIPE_CONFIG(ADDR_SURF_P2)); + modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + + mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + + for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) + if (reg_offset != 7 && reg_offset != 12 && reg_offset != 17 && + reg_offset != 23) + WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]); + + for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) + if (reg_offset != 7) + WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]); + break; - case CHIP_CARRIZO: default: - for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 1: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 2: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 3: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 4: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 5: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 6: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); - break; - case 8: - gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | - PIPE_CONFIG(ADDR_SURF_P2)); - break; - case 9: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 10: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 11: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 13: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 14: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 15: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 16: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 18: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 19: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 20: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 21: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 22: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 24: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 25: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 26: - gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); - break; - case 27: - gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 28: - gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); - break; - case 29: - gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | - PIPE_CONFIG(ADDR_SURF_P2) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | - SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); - break; - case 7: - case 12: - case 17: - case 23: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden); - } - for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) { - switch (reg_offset) { - case 0: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 1: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 2: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 3: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 4: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 5: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 6: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 8: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 9: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 10: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 11: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 12: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 13: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | - NUM_BANKS(ADDR_SURF_16_BANK)); - break; - case 14: - gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | - BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | - MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | - NUM_BANKS(ADDR_SURF_8_BANK)); - break; - case 7: - /* unused idx */ - continue; - default: - gb_tile_moden = 0; - break; - }; - adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden; - WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden); - } + dev_warn(adev->dev, + "Unknown chip type (%d) in function gfx_v8_0_tiling_mode_table_init() falling through to CHIP_CARRIZO\n", + adev->asic_type); + + case CHIP_CARRIZO: + modearray[0] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[1] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[2] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[3] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[4] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[5] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[6] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); + modearray[8] = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | + PIPE_CONFIG(ADDR_SURF_P2)); + modearray[9] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[10] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[11] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[13] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[14] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[15] = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[16] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + modearray[18] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[19] = (ARRAY_MODE(ARRAY_1D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[20] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[21] = (ARRAY_MODE(ARRAY_3D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[22] = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[24] = (ARRAY_MODE(ARRAY_2D_TILED_THICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[25] = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[26] = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1)); + modearray[27] = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[28] = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2)); + modearray[29] = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8)); + + mod2array[0] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[1] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[2] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[3] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[4] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[5] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[6] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + mod2array[8] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[9] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[10] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[11] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[12] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[13] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) | + NUM_BANKS(ADDR_SURF_16_BANK)); + mod2array[14] = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) | + BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) | + MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) | + NUM_BANKS(ADDR_SURF_8_BANK)); + + for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) + if (reg_offset != 7 && reg_offset != 12 && reg_offset != 17 && + reg_offset != 23) + WREG32(mmGB_TILE_MODE0 + reg_offset, modearray[reg_offset]); + + for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) + if (reg_offset != 7) + WREG32(mmGB_MACROTILE_MODE0 + reg_offset, mod2array[reg_offset]); + + break; } } @@ -4957,7 +4467,7 @@ static void gfx_v8_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr, EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) | EVENT_INDEX(5))); amdgpu_ring_write(ring, addr & 0xfffffffc); - amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) | + amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) | DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0)); amdgpu_ring_write(ring, lower_32_bits(seq)); amdgpu_ring_write(ring, upper_32_bits(seq)); -- cgit From 544b8a74c72d39aebbbf482ccad8e7762a8b2a1e Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 1 Dec 2015 11:48:32 -0500 Subject: amdgpu/gfxv8: Simplification of gfx_v8_0_create_bitmask() Simplification of the function gfx_v8_0_create_bitmask(). Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 5c0400255f6d..b5396db0e6cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -2555,13 +2555,7 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev) static u32 gfx_v8_0_create_bitmask(u32 bit_width) { - u32 i, mask = 0; - - for (i = 0; i < bit_width; i++) { - mask <<= 1; - mask |= 1; - } - return mask; + return (u32)((1ULL << bit_width) - 1); } void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num) -- cgit From 0d07db7e10f415d7ec0efde54f10a0905a0910b9 Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 1 Dec 2015 10:42:28 -0500 Subject: amdgpu/gfxv8: Simplification in gfx_v8_0_enable_gui_idle_interrupt() Simplified the function by folding the two paths into one. Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index b5396db0e6cb..33e3c0d63c66 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -2818,17 +2818,11 @@ static void gfx_v8_0_enable_gui_idle_interrupt(struct amdgpu_device *adev, { u32 tmp = RREG32(mmCP_INT_CNTL_RING0); - if (enable) { - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, 1); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, 1); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, 1); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, 1); - } else { - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, 0); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, 0); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, 0); - tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, 0); - } + tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, enable ? 1 : 0); + tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, enable ? 1 : 0); + tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, enable ? 1 : 0); + tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, enable ? 1 : 0); + WREG32(mmCP_INT_CNTL_RING0, tmp); } -- cgit From ba98f9e56c77f2b7a28f9d025f8780098221e1b6 Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Thu, 26 Nov 2015 16:33:58 +0800 Subject: drm/amdgpu: add err check for pin userptr MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Missing error check if the operation failed. Signed-off-by: Chunming Zhou Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index d4bac5f49939..8051cb9b8c1e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -587,9 +587,13 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm, uint32_t flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem); int r; - if (gtt->userptr) - amdgpu_ttm_tt_pin_userptr(ttm); - + if (gtt->userptr) { + r = amdgpu_ttm_tt_pin_userptr(ttm); + if (r) { + DRM_ERROR("failed to pin userptr\n"); + return r; + } + } gtt->offset = (unsigned long)(bo_mem->start << PAGE_SHIFT); if (!ttm->num_pages) { WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n", -- cgit From eb64526f5a5a382df7018a3e69061b9e3f20e32d Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Thu, 3 Dec 2015 12:23:28 -0500 Subject: amdgpu/gfxv8: Remove magic numbers from function gfx_v8_0_tiling_mode_table_init() Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 33e3c0d63c66..29e02e0fa3d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -1640,8 +1640,8 @@ static int gfx_v8_0_sw_fini(void *handle) static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev) { uint32_t *modearray, *mod2array; - const u32 num_tile_mode_states = 32; - const u32 num_secondary_tile_mode_states = 16; + const u32 num_tile_mode_states = ARRAY_SIZE(adev->gfx.config.tile_mode_array); + const u32 num_secondary_tile_mode_states = ARRAY_SIZE(adev->gfx.config.macrotile_mode_array); u32 reg_offset; modearray = adev->gfx.config.tile_mode_array; -- cgit From 9c4153b1eef9bc8da6a624252a3a25790b705136 Mon Sep 17 00:00:00 2001 From: jimqu Date: Fri, 4 Dec 2015 17:17:00 +0800 Subject: drm/amdgpu: add spin lock to protect freed list in vm (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit there is a protection fault about freed list when OCL test. add a spin lock to protect it. v2: drop changes in vm_fini Signed-off-by: JimQu Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 16 +++++++++++++--- 2 files changed, 15 insertions(+), 3 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 5f975030bb72..3b5d3706f0cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -955,6 +955,8 @@ struct amdgpu_vm { struct amdgpu_vm_id ids[AMDGPU_MAX_RINGS]; /* for interval tree */ spinlock_t it_lock; + /* protecting freed */ + spinlock_t freed_lock; }; struct amdgpu_vm_manager { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index ae037e5b6ad0..fce4c6d952c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -885,17 +885,21 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *mapping; int r; + spin_lock(&vm->freed_lock); while (!list_empty(&vm->freed)) { mapping = list_first_entry(&vm->freed, struct amdgpu_bo_va_mapping, list); list_del(&mapping->list); - + spin_unlock(&vm->freed_lock); r = amdgpu_vm_bo_update_mapping(adev, vm, mapping, 0, 0, NULL); kfree(mapping); if (r) return r; + spin_lock(&vm->freed_lock); } + spin_unlock(&vm->freed_lock); + return 0; } @@ -1150,10 +1154,13 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev, spin_unlock(&vm->it_lock); trace_amdgpu_vm_bo_unmap(bo_va, mapping); - if (valid) + if (valid) { + spin_lock(&vm->freed_lock); list_add(&mapping->list, &vm->freed); - else + spin_unlock(&vm->freed_lock); + } else { kfree(mapping); + } return 0; } @@ -1186,7 +1193,9 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, interval_tree_remove(&mapping->it, &vm->va); spin_unlock(&vm->it_lock); trace_amdgpu_vm_bo_unmap(bo_va, mapping); + spin_lock(&vm->freed_lock); list_add(&mapping->list, &vm->freed); + spin_unlock(&vm->freed_lock); } list_for_each_entry_safe(mapping, next, &bo_va->invalids, list) { list_del(&mapping->list); @@ -1247,6 +1256,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm) INIT_LIST_HEAD(&vm->cleared); INIT_LIST_HEAD(&vm->freed); spin_lock_init(&vm->it_lock); + spin_lock_init(&vm->freed_lock); pd_size = amdgpu_vm_directory_size(adev); pd_entries = amdgpu_vm_num_pdes(adev); -- cgit From b516a9efb7af38733ba2308fc7c0f49f034f53bd Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Fri, 4 Dec 2015 09:45:43 +0100 Subject: drm: Move LEAVE/ENTER_ATOMIC_MODESET to fbdev helpers This is only used for kgdb (and previously panic) handlers in the fbdev emulation, so belongs there. Note that this means we'll leave behind a forward declaration, but once all the helper vtables are consolidated (in the next patch) that will make more sense. v2: fixup radone/amdgpu. Signed-off-by: Daniel Vetter Link: http://patchwork.freedesktop.org/patch/msgid/1449218769-16577-3-git-send-email-daniel.vetter@ffwll.ch Reviewed-by: Thierry Reding (v2) --- drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index a53d756672fe..fdc1be8550da 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -35,6 +35,7 @@ #include #include #include +#include #include #include #include -- cgit From e410b5cbabe70b1f1da08146481f6bd7e904643c Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Mon, 7 Dec 2015 15:02:52 +0800 Subject: drm/amdgpu: fix the lost duplicates checking MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Chunming Zhou Reviewed-by: Christian König Reviewed-by: Jammy Zhou Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index f6ea4b43a60c..9c253c535d26 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -477,6 +477,14 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, if (domain == AMDGPU_GEM_DOMAIN_CPU) goto error_unreserve; } + list_for_each_entry(entry, &duplicates, head) { + domain = amdgpu_mem_type_to_domain(entry->bo->mem.mem_type); + /* if anything is swapped out don't swap it in here, + just abort and wait for the next CS */ + if (domain == AMDGPU_GEM_DOMAIN_CPU) + goto error_unreserve; + } + r = amdgpu_vm_update_page_directory(adev, bo_va->vm); if (r) goto error_unreserve; -- cgit From 13a3d91f17a5f7ed2acd275d18b6acfdb131fb15 Mon Sep 17 00:00:00 2001 From: Ville Syrjälä Date: Wed, 9 Dec 2015 16:20:18 +0200 Subject: drm: Pass 'name' to drm_encoder_init() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Done with coccinelle for the most part. However, it thinks '...' is part of the semantic patch, so I put an 'int DOTDOTDOT' placeholder in its place and got rid of it with sed afterwards. @@ identifier dev, encoder, funcs; @@ int drm_encoder_init(struct drm_device *dev, struct drm_encoder *encoder, const struct drm_encoder_funcs *funcs, int encoder_type + ,const char *name, int DOTDOTDOT ) { ... } @@ identifier dev, encoder, funcs; @@ int drm_encoder_init(struct drm_device *dev, struct drm_encoder *encoder, const struct drm_encoder_funcs *funcs, int encoder_type + ,const char *name, int DOTDOTDOT ); @@ expression E1, E2, E3, E4; @@ drm_encoder_init(E1, E2, E3, E4 + ,NULL ) v2: Add ', or NULL...' to @name kernel doc (Jani) Annotate the function with __printf() attribute (Jani) Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter Link: http://patchwork.freedesktop.org/patch/msgid/1449670818-2966-1-git-send-email-ville.syrjala@linux.intel.com --- drivers/gpu/drm/amd/amdgpu/dce_v10_0.c | 14 +++++++------- drivers/gpu/drm/amd/amdgpu/dce_v11_0.c | 14 +++++++------- drivers/gpu/drm/amd/amdgpu/dce_v8_0.c | 14 +++++++------- 3 files changed, 21 insertions(+), 21 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index 4dcc8fba5792..093599aba64b 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -3729,7 +3729,7 @@ static void dce_v10_0_encoder_add(struct amdgpu_device *adev, case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); drm_encoder_helper_add(encoder, &dce_v10_0_dac_helper_funcs); break; case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: @@ -3740,15 +3740,15 @@ static void dce_v10_0_encoder_add(struct amdgpu_device *adev, if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { amdgpu_encoder->rmx_type = RMX_FULL; drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_LVDS); + DRM_MODE_ENCODER_LVDS, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder); } else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) { drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder); } else { drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_TMDS); + DRM_MODE_ENCODER_TMDS, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder); } drm_encoder_helper_add(encoder, &dce_v10_0_dig_helper_funcs); @@ -3766,13 +3766,13 @@ static void dce_v10_0_encoder_add(struct amdgpu_device *adev, amdgpu_encoder->is_ext_encoder = true; if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_LVDS); + DRM_MODE_ENCODER_LVDS, NULL); else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); else drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs, - DRM_MODE_ENCODER_TMDS); + DRM_MODE_ENCODER_TMDS, NULL); drm_encoder_helper_add(encoder, &dce_v10_0_ext_helper_funcs); break; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 8f1e51128b33..8701661a8868 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -3722,7 +3722,7 @@ static void dce_v11_0_encoder_add(struct amdgpu_device *adev, case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); drm_encoder_helper_add(encoder, &dce_v11_0_dac_helper_funcs); break; case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: @@ -3733,15 +3733,15 @@ static void dce_v11_0_encoder_add(struct amdgpu_device *adev, if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { amdgpu_encoder->rmx_type = RMX_FULL; drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_LVDS); + DRM_MODE_ENCODER_LVDS, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder); } else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) { drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder); } else { drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_TMDS); + DRM_MODE_ENCODER_TMDS, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder); } drm_encoder_helper_add(encoder, &dce_v11_0_dig_helper_funcs); @@ -3759,13 +3759,13 @@ static void dce_v11_0_encoder_add(struct amdgpu_device *adev, amdgpu_encoder->is_ext_encoder = true; if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_LVDS); + DRM_MODE_ENCODER_LVDS, NULL); else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); else drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs, - DRM_MODE_ENCODER_TMDS); + DRM_MODE_ENCODER_TMDS, NULL); drm_encoder_helper_add(encoder, &dce_v11_0_ext_helper_funcs); break; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index 42d954dc436d..d0e128c24813 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -3659,7 +3659,7 @@ static void dce_v8_0_encoder_add(struct amdgpu_device *adev, case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); drm_encoder_helper_add(encoder, &dce_v8_0_dac_helper_funcs); break; case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: @@ -3670,15 +3670,15 @@ static void dce_v8_0_encoder_add(struct amdgpu_device *adev, if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { amdgpu_encoder->rmx_type = RMX_FULL; drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_LVDS); + DRM_MODE_ENCODER_LVDS, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder); } else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) { drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder); } else { drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_TMDS); + DRM_MODE_ENCODER_TMDS, NULL); amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder); } drm_encoder_helper_add(encoder, &dce_v8_0_dig_helper_funcs); @@ -3696,13 +3696,13 @@ static void dce_v8_0_encoder_add(struct amdgpu_device *adev, amdgpu_encoder->is_ext_encoder = true; if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_LVDS); + DRM_MODE_ENCODER_LVDS, NULL); else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_DAC); + DRM_MODE_ENCODER_DAC, NULL); else drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs, - DRM_MODE_ENCODER_TMDS); + DRM_MODE_ENCODER_TMDS, NULL); drm_encoder_helper_add(encoder, &dce_v8_0_ext_helper_funcs); break; } -- cgit From 3b55ddadef631d8081cb2f73c6c7cb80c634ba2b Mon Sep 17 00:00:00 2001 From: Flora Cui Date: Wed, 2 Dec 2015 09:56:06 +0800 Subject: drm/amdgpu/gfx8: Enable interrupt on ME1_PIPE3 Otherwise FW cannot see the RLC ACK for the memory clean request It's for Stoney. Signed-off-by: Flora Cui Reviewed-by: Alex Deucher Reviewed-by: Jammy Zhou --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 29e02e0fa3d4..6816a1a60a97 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -3756,6 +3756,11 @@ static int gfx_v8_0_cp_compute_resume(struct amdgpu_device *adev) tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x53); WREG32(mmCP_HQD_PERSISTENT_STATE, tmp); mqd->cp_hqd_persistent_state = tmp; + if (adev->asic_type == CHIP_STONEY) { + tmp = RREG32(mmCP_ME1_PIPE3_INT_CNTL); + tmp = REG_SET_FIELD(tmp, CP_ME1_PIPE3_INT_CNTL, GENERIC2_INT_ENABLE, 1); + WREG32(mmCP_ME1_PIPE3_INT_CNTL, tmp); + } /* activate the queue */ mqd->cp_hqd_active = 1; -- cgit From c27816a883575037ec67e7f92fbd214648a1c1cb Mon Sep 17 00:00:00 2001 From: Flora Cui Date: Tue, 8 Dec 2015 11:23:29 +0800 Subject: drm/amdgpu/gfx8: update PA_SC_RASTER_CONFIG:PKR_MAP only Use default value as a base. Signed-off-by: Flora Cui Reviewed-by: Jammy Zhou --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 6816a1a60a97..b20369e4ea74 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -2630,7 +2630,7 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev, mutex_lock(&adev->grbm_idx_mutex); for (i = 0; i < se_num; i++) { gfx_v8_0_select_se_sh(adev, i, 0xffffffff); - data = 0; + data = RREG32(mmPA_SC_RASTER_CONFIG); for (j = 0; j < sh_per_se; j++) { switch (enabled_rbs & 3) { case 0: -- cgit From abdfb850ca99a27557564e4fcf4b137e209ebfef Mon Sep 17 00:00:00 2001 From: Flora Cui Date: Fri, 20 Nov 2015 11:40:53 +0800 Subject: drm/amdgpu: update rev id register for VI Change-Id: I2ae9bb4a929f7c0c8783e0be563ae04be77596e2 Signed-off-by: Flora Cui Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/vi.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index bf5bd931b8ca..30b408f1d5c0 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -1386,15 +1386,12 @@ int vi_set_ip_blocks(struct amdgpu_device *adev) static uint32_t vi_get_rev_id(struct amdgpu_device *adev) { - if (adev->asic_type == CHIP_TOPAZ) - return (RREG32(mmPCIE_EFUSE4) & PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID_MASK) - >> PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID__SHIFT; - else if (adev->flags & AMD_IS_APU) + if (adev->flags & AMD_IS_APU) return (RREG32_SMC(ATI_REV_ID_FUSE_MACRO__ADDRESS) & ATI_REV_ID_FUSE_MACRO__MASK) >> ATI_REV_ID_FUSE_MACRO__SHIFT; else - return (RREG32(mmCC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK) - >> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT; + return (RREG32(mmPCIE_EFUSE4) & PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID_MASK) + >> PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID__SHIFT; } static const struct amdgpu_asic_funcs vi_asic_funcs = -- cgit From 2c1a27840394428d8f44fe5d7509dd20574d0573 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 7 Dec 2015 17:02:53 -0500 Subject: drm/amdgpu: add more debugging output for driver failures Add more fine grained debugging output for init/fini/suspend/ resume failures. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 73 +++++++++++++++++++++++------- 1 file changed, 57 insertions(+), 16 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 84775964b138..54af6ce7901f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1214,12 +1214,14 @@ static int amdgpu_early_init(struct amdgpu_device *adev) } else { if (adev->ip_blocks[i].funcs->early_init) { r = adev->ip_blocks[i].funcs->early_init((void *)adev); - if (r == -ENOENT) + if (r == -ENOENT) { adev->ip_block_status[i].valid = false; - else if (r) + } else if (r) { + DRM_ERROR("early_init %d failed %d\n", i, r); return r; - else + } else { adev->ip_block_status[i].valid = true; + } } else { adev->ip_block_status[i].valid = true; } @@ -1237,20 +1239,28 @@ static int amdgpu_init(struct amdgpu_device *adev) if (!adev->ip_block_status[i].valid) continue; r = adev->ip_blocks[i].funcs->sw_init((void *)adev); - if (r) + if (r) { + DRM_ERROR("sw_init %d failed %d\n", i, r); return r; + } adev->ip_block_status[i].sw = true; /* need to do gmc hw init early so we can allocate gpu mem */ if (adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_GMC) { r = amdgpu_vram_scratch_init(adev); - if (r) + if (r) { + DRM_ERROR("amdgpu_vram_scratch_init failed %d\n", r); return r; + } r = adev->ip_blocks[i].funcs->hw_init((void *)adev); - if (r) + if (r) { + DRM_ERROR("hw_init %d failed %d\n", i, r); return r; + } r = amdgpu_wb_init(adev); - if (r) + if (r) { + DRM_ERROR("amdgpu_wb_init failed %d\n", r); return r; + } adev->ip_block_status[i].hw = true; } } @@ -1262,8 +1272,10 @@ static int amdgpu_init(struct amdgpu_device *adev) if (adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_GMC) continue; r = adev->ip_blocks[i].funcs->hw_init((void *)adev); - if (r) + if (r) { + DRM_ERROR("hw_init %d failed %d\n", i, r); return r; + } adev->ip_block_status[i].hw = true; } @@ -1280,12 +1292,16 @@ static int amdgpu_late_init(struct amdgpu_device *adev) /* enable clockgating to save power */ r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev, AMD_CG_STATE_GATE); - if (r) + if (r) { + DRM_ERROR("set_clockgating_state(gate) %d failed %d\n", i, r); return r; + } if (adev->ip_blocks[i].funcs->late_init) { r = adev->ip_blocks[i].funcs->late_init((void *)adev); - if (r) + if (r) { + DRM_ERROR("late_init %d failed %d\n", i, r); return r; + } } } @@ -1306,10 +1322,15 @@ static int amdgpu_fini(struct amdgpu_device *adev) /* ungate blocks before hw fini so that we can shutdown the blocks safely */ r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev, AMD_CG_STATE_UNGATE); - if (r) + if (r) { + DRM_ERROR("set_clockgating_state(ungate) %d failed %d\n", i, r); return r; + } r = adev->ip_blocks[i].funcs->hw_fini((void *)adev); /* XXX handle errors */ + if (r) { + DRM_DEBUG("hw_fini %d failed %d\n", i, r); + } adev->ip_block_status[i].hw = false; } @@ -1318,6 +1339,9 @@ static int amdgpu_fini(struct amdgpu_device *adev) continue; r = adev->ip_blocks[i].funcs->sw_fini((void *)adev); /* XXX handle errors */ + if (r) { + DRM_DEBUG("sw_fini %d failed %d\n", i, r); + } adev->ip_block_status[i].sw = false; adev->ip_block_status[i].valid = false; } @@ -1335,9 +1359,15 @@ static int amdgpu_suspend(struct amdgpu_device *adev) /* ungate blocks so that suspend can properly shut them down */ r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev, AMD_CG_STATE_UNGATE); + if (r) { + DRM_ERROR("set_clockgating_state(ungate) %d failed %d\n", i, r); + } /* XXX handle errors */ r = adev->ip_blocks[i].funcs->suspend(adev); /* XXX handle errors */ + if (r) { + DRM_ERROR("suspend %d failed %d\n", i, r); + } } return 0; @@ -1351,8 +1381,10 @@ static int amdgpu_resume(struct amdgpu_device *adev) if (!adev->ip_block_status[i].valid) continue; r = adev->ip_blocks[i].funcs->resume(adev); - if (r) + if (r) { + DRM_ERROR("resume %d failed %d\n", i, r); return r; + } } return 0; @@ -1484,8 +1516,10 @@ int amdgpu_device_init(struct amdgpu_device *adev, return -EINVAL; } r = amdgpu_atombios_init(adev); - if (r) + if (r) { + dev_err(adev->dev, "amdgpu_atombios_init failed\n"); return r; + } /* Post card if necessary */ if (!amdgpu_card_posted(adev)) { @@ -1499,21 +1533,26 @@ int amdgpu_device_init(struct amdgpu_device *adev, /* Initialize clocks */ r = amdgpu_atombios_get_clock_info(adev); - if (r) + if (r) { + dev_err(adev->dev, "amdgpu_atombios_get_clock_info failed\n"); return r; + } /* init i2c buses */ amdgpu_atombios_i2c_init(adev); /* Fence driver */ r = amdgpu_fence_driver_init(adev); - if (r) + if (r) { + dev_err(adev->dev, "amdgpu_fence_driver_init failed\n"); return r; + } /* init the mode config */ drm_mode_config_init(adev->ddev); r = amdgpu_init(adev); if (r) { + dev_err(adev->dev, "amdgpu_init failed\n"); amdgpu_fini(adev); return r; } @@ -1570,8 +1609,10 @@ int amdgpu_device_init(struct amdgpu_device *adev, * explicit gating rather than handling it automatically. */ r = amdgpu_late_init(adev); - if (r) + if (r) { + dev_err(adev->dev, "amdgpu_late_init failed\n"); return r; + } return 0; } -- cgit From c648ed7c5c7f0e3bb4ab11bf08bccf99b42a4cbb Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Thu, 10 Dec 2015 15:50:02 +0800 Subject: drm/amdgpu: handle error case for ctx MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Properly handle ctx init failure. Signed-off-by: Chunming Zhou Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index c1f2308b4115..15e341634536 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -56,7 +56,6 @@ int amdgpu_ctx_init(struct amdgpu_device *adev, enum amd_sched_priority pri, for (j = 0; j < i; j++) amd_sched_entity_fini(&adev->rings[j]->sched, &ctx->rings[j].entity); - kfree(ctx); return r; } } @@ -103,8 +102,12 @@ static int amdgpu_ctx_alloc(struct amdgpu_device *adev, } *id = (uint32_t)r; r = amdgpu_ctx_init(adev, AMD_SCHED_PRIORITY_NORMAL, ctx); + if (r) { + idr_remove(&mgr->ctx_handles, *id); + *id = 0; + kfree(ctx); + } mutex_unlock(&mgr->lock); - return r; } -- cgit From 37cd0ca204a55e123fca9ce411e6571ac49fa8f7 Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Thu, 10 Dec 2015 15:45:11 +0800 Subject: drm/amdgpu: unify AMDGPU_CTX_MAX_CS_PENDING and amdgpu_sched_jobs MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Chunming Zhou Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 5 ++--- drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 25 ++++++++++++++++++------- 2 files changed, 20 insertions(+), 10 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 3b5d3706f0cb..c3996e0e2e7e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1023,11 +1023,9 @@ int amdgpu_vm_free_job(struct amdgpu_job *job); * context related structures */ -#define AMDGPU_CTX_MAX_CS_PENDING 16 - struct amdgpu_ctx_ring { uint64_t sequence; - struct fence *fences[AMDGPU_CTX_MAX_CS_PENDING]; + struct fence **fences; struct amd_sched_entity entity; }; @@ -1036,6 +1034,7 @@ struct amdgpu_ctx { struct amdgpu_device *adev; unsigned reset_counter; spinlock_t ring_lock; + struct fence **fences; struct amdgpu_ctx_ring rings[AMDGPU_MAX_RINGS]; }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index 15e341634536..ee121ec2917b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -35,15 +35,24 @@ int amdgpu_ctx_init(struct amdgpu_device *adev, enum amd_sched_priority pri, ctx->adev = adev; kref_init(&ctx->refcount); spin_lock_init(&ctx->ring_lock); - for (i = 0; i < AMDGPU_MAX_RINGS; ++i) - ctx->rings[i].sequence = 1; + ctx->fences = kzalloc(sizeof(struct fence *) * amdgpu_sched_jobs * + AMDGPU_MAX_RINGS, GFP_KERNEL); + if (!ctx->fences) + return -ENOMEM; + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { + ctx->rings[i].sequence = 1; + ctx->rings[i].fences = (void *)ctx->fences + sizeof(struct fence *) * + amdgpu_sched_jobs * i; + } if (amdgpu_enable_scheduler) { /* create context entity for each ring */ for (i = 0; i < adev->num_rings; i++) { struct amd_sched_rq *rq; - if (pri >= AMD_SCHED_MAX_PRIORITY) + if (pri >= AMD_SCHED_MAX_PRIORITY) { + kfree(ctx->fences); return -EINVAL; + } rq = &adev->rings[i]->sched.sched_rq[pri]; r = amd_sched_entity_init(&adev->rings[i]->sched, &ctx->rings[i].entity, @@ -56,6 +65,7 @@ int amdgpu_ctx_init(struct amdgpu_device *adev, enum amd_sched_priority pri, for (j = 0; j < i; j++) amd_sched_entity_fini(&adev->rings[j]->sched, &ctx->rings[j].entity); + kfree(ctx->fences); return r; } } @@ -71,8 +81,9 @@ void amdgpu_ctx_fini(struct amdgpu_ctx *ctx) return; for (i = 0; i < AMDGPU_MAX_RINGS; ++i) - for (j = 0; j < AMDGPU_CTX_MAX_CS_PENDING; ++j) + for (j = 0; j < amdgpu_sched_jobs; ++j) fence_put(ctx->rings[i].fences[j]); + kfree(ctx->fences); if (amdgpu_enable_scheduler) { for (i = 0; i < adev->num_rings; i++) @@ -241,7 +252,7 @@ uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring, unsigned idx = 0; struct fence *other = NULL; - idx = seq % AMDGPU_CTX_MAX_CS_PENDING; + idx = seq % amdgpu_sched_jobs; other = cring->fences[idx]; if (other) { signed long r; @@ -276,12 +287,12 @@ struct fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx, } - if (seq + AMDGPU_CTX_MAX_CS_PENDING < cring->sequence) { + if (seq + amdgpu_sched_jobs < cring->sequence) { spin_unlock(&ctx->ring_lock); return NULL; } - fence = fence_get(cring->fences[seq % AMDGPU_CTX_MAX_CS_PENDING]); + fence = fence_get(cring->fences[seq % amdgpu_sched_jobs]); spin_unlock(&ctx->ring_lock); return fence; -- cgit From b70f014d58b976930b9e8337980954a958c01d7f Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Thu, 10 Dec 2015 15:46:50 +0800 Subject: drm/amdgpu: change default sched jobs to 32 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Change the default scheduler queue size from 16 to 32. Signed-off-by: Chunming Zhou Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 0508c5cd103a..642e305bcbcd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -79,7 +79,7 @@ int amdgpu_vm_fault_stop = 0; int amdgpu_vm_debug = 0; int amdgpu_exp_hw_support = 0; int amdgpu_enable_scheduler = 1; -int amdgpu_sched_jobs = 16; +int amdgpu_sched_jobs = 32; int amdgpu_sched_hw_submission = 2; int amdgpu_enable_semaphores = 0; @@ -155,7 +155,7 @@ module_param_named(exp_hw_support, amdgpu_exp_hw_support, int, 0444); MODULE_PARM_DESC(enable_scheduler, "enable SW GPU scheduler (1 = enable (default), 0 = disable)"); module_param_named(enable_scheduler, amdgpu_enable_scheduler, int, 0444); -MODULE_PARM_DESC(sched_jobs, "the max number of jobs supported in the sw queue (default 16)"); +MODULE_PARM_DESC(sched_jobs, "the max number of jobs supported in the sw queue (default 32)"); module_param_named(sched_jobs, amdgpu_sched_jobs, int, 0444); MODULE_PARM_DESC(sched_hw_submission, "the max number of HW submissions (default 2)"); -- cgit From 91acbeb68ab10c0c0f65f30b5b7fddbde4c97dd2 Mon Sep 17 00:00:00 2001 From: Christian König Date: Mon, 14 Dec 2015 16:42:31 +0100 Subject: drm/amdgpu: fix user fence handling MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This fixes a random corruption under memory pressure. We need to fence the BO for the user fence as well, otherwise it might be swapped out and the GPU could write the fence value to an undesired location. Signed-off-by: Christian König Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 3 +- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 63 ++++++++++++++++++++++------------ 2 files changed, 44 insertions(+), 22 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 5a5f04d0902d..048cfe073dae 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1264,7 +1264,8 @@ struct amdgpu_cs_parser { struct ww_acquire_ctx ticket; /* user fence */ - struct amdgpu_user_fence uf; + struct amdgpu_user_fence uf; + struct amdgpu_bo_list_entry uf_entry; }; struct amdgpu_job { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 4f352ec9dec4..25a3e2485cc2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -127,6 +127,37 @@ int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type, return 0; } +static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p, + struct drm_amdgpu_cs_chunk_fence *fence_data) +{ + struct drm_gem_object *gobj; + uint32_t handle; + + handle = fence_data->handle; + gobj = drm_gem_object_lookup(p->adev->ddev, p->filp, + fence_data->handle); + if (gobj == NULL) + return -EINVAL; + + p->uf.bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj)); + p->uf.offset = fence_data->offset; + + if (amdgpu_ttm_tt_has_userptr(p->uf.bo->tbo.ttm)) { + drm_gem_object_unreference_unlocked(gobj); + return -EINVAL; + } + + p->uf_entry.robj = amdgpu_bo_ref(p->uf.bo); + p->uf_entry.prefered_domains = AMDGPU_GEM_DOMAIN_GTT; + p->uf_entry.allowed_domains = AMDGPU_GEM_DOMAIN_GTT; + p->uf_entry.priority = 0; + p->uf_entry.tv.bo = &p->uf_entry.robj->tbo; + p->uf_entry.tv.shared = true; + + drm_gem_object_unreference_unlocked(gobj); + return 0; +} + int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data) { union drm_amdgpu_cs *cs = data; @@ -207,28 +238,15 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data) case AMDGPU_CHUNK_ID_FENCE: size = sizeof(struct drm_amdgpu_cs_chunk_fence); - if (p->chunks[i].length_dw * sizeof(uint32_t) >= size) { - uint32_t handle; - struct drm_gem_object *gobj; - struct drm_amdgpu_cs_chunk_fence *fence_data; - - fence_data = (void *)p->chunks[i].kdata; - handle = fence_data->handle; - gobj = drm_gem_object_lookup(p->adev->ddev, - p->filp, handle); - if (gobj == NULL) { - ret = -EINVAL; - goto free_partial_kdata; - } - - p->uf.bo = gem_to_amdgpu_bo(gobj); - amdgpu_bo_ref(p->uf.bo); - drm_gem_object_unreference_unlocked(gobj); - p->uf.offset = fence_data->offset; - } else { + if (p->chunks[i].length_dw * sizeof(uint32_t) < size) { ret = -EINVAL; goto free_partial_kdata; } + + ret = amdgpu_cs_user_fence_chunk(p, (void *)p->chunks[i].kdata); + if (ret) + goto free_partial_kdata; + break; case AMDGPU_CHUNK_ID_DEPENDENCIES: @@ -391,6 +409,9 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) p->vm_bos = amdgpu_vm_get_bos(p->adev, &fpriv->vm, &p->validated); + if (p->uf.bo) + list_add(&p->uf_entry.tv.head, &p->validated); + if (need_mmap_lock) down_read(¤t->mm->mmap_sem); @@ -488,8 +509,8 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bo for (i = 0; i < parser->num_ibs; i++) amdgpu_ib_free(parser->adev, &parser->ibs[i]); kfree(parser->ibs); - if (parser->uf.bo) - amdgpu_bo_unref(&parser->uf.bo); + amdgpu_bo_unref(&parser->uf.bo); + amdgpu_bo_unref(&parser->uf_entry.robj); } static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p, -- cgit From a1493cd575678910f90d488069a8a2c41a144fda Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 9 Dec 2015 15:36:40 -0500 Subject: drm/amdgpu: limit visible vram if it's smaller than the BAR MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In some cases the amount of vram may be less than the BAR size, if so, limit visible vram to the amount of actual vram, not the BAR size. Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c | 4 ++++ drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 4 ++++ 2 files changed, 8 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index 7427d8cd4c43..868505753a9a 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -370,6 +370,10 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev) adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL; adev->mc.visible_vram_size = adev->mc.aper_size; + /* In case the PCI BAR is larger than the actual amount of vram */ + if (adev->mc.visible_vram_size > adev->mc.real_vram_size) + adev->mc.visible_vram_size = adev->mc.real_vram_size; + /* unless the user had overridden it, set the gart * size equal to the 1024 or vram, whichever is larger. */ diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index cb0e50ebb528..a2d869d30cf4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -476,6 +476,10 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev) adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL; adev->mc.visible_vram_size = adev->mc.aper_size; + /* In case the PCI BAR is larger than the actual amount of vram */ + if (adev->mc.visible_vram_size > adev->mc.real_vram_size) + adev->mc.visible_vram_size = adev->mc.real_vram_size; + /* unless the user had overridden it, set the gart * size equal to the 1024 or vram, whichever is larger. */ -- cgit From 5b0112356cf9a735632b26ff5f3450e1716c8598 Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Thu, 10 Dec 2015 17:34:33 +0800 Subject: drm/amdgpu: restrict the sched jobs number to power of two MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Chunming Zhou Reviewed-by: Christian König CC: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 9 +++++++++ 2 files changed, 11 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index ee121ec2917b..17d1fb12128a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -252,7 +252,7 @@ uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring, unsigned idx = 0; struct fence *other = NULL; - idx = seq % amdgpu_sched_jobs; + idx = seq & (amdgpu_sched_jobs - 1); other = cring->fences[idx]; if (other) { signed long r; @@ -292,7 +292,7 @@ struct fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx, return NULL; } - fence = fence_get(cring->fences[seq % amdgpu_sched_jobs]); + fence = fence_get(cring->fences[seq & (amdgpu_sched_jobs - 1)]); spin_unlock(&ctx->ring_lock); return fence; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 54af6ce7901f..587ff7145361 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -949,6 +949,15 @@ static bool amdgpu_check_pot_argument(int arg) */ static void amdgpu_check_arguments(struct amdgpu_device *adev) { + if (amdgpu_sched_jobs < 4) { + dev_warn(adev->dev, "sched jobs (%d) must be at least 4\n", + amdgpu_sched_jobs); + amdgpu_sched_jobs = 4; + } else if (!amdgpu_check_pot_argument(amdgpu_sched_jobs)){ + dev_warn(adev->dev, "sched jobs (%d) must be a power of 2\n", + amdgpu_sched_jobs); + amdgpu_sched_jobs = roundup_pow_of_two(amdgpu_sched_jobs); + } /* vramlimit must be a power of two */ if (!amdgpu_check_pot_argument(amdgpu_vram_limit)) { dev_warn(adev->dev, "vram limit (%d) must be a power of 2\n", -- cgit From 3c0eea6c35d932c4d25070868067dc9cd9ceab91 Mon Sep 17 00:00:00 2001 From: Christian König Date: Fri, 11 Dec 2015 14:39:05 +0100 Subject: drm/amdgpu: put VM page tables directly into duplicates list MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit They share the reservation object with the page directory anyway. Signed-off-by: Christian König Reviewed-by: Alex Deucher Reviewed-by: Chunming Zhou --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 3 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 12 +++++++----- 4 files changed, 12 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index c3996e0e2e7e..dc3dab539e4c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -982,7 +982,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm); struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, - struct list_head *head); + struct list_head *validated, + struct list_head *duplicates); int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_sync *sync); void amdgpu_vm_flush(struct amdgpu_ring *ring, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 1d44d508d4d4..9591c13781bd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -386,13 +386,13 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) amdgpu_cs_buckets_get_list(&buckets, &p->validated); } + INIT_LIST_HEAD(&duplicates); p->vm_bos = amdgpu_vm_get_bos(p->adev, &fpriv->vm, - &p->validated); + &p->validated, &duplicates); if (need_mmap_lock) down_read(¤t->mm->mmap_sem); - INIT_LIST_HEAD(&duplicates); r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true, &duplicates); if (unlikely(r != 0)) goto error_reserve; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index fc32fc01a64b..7fe7f8afa5ff 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -460,7 +460,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, tv.shared = true; list_add(&tv.head, &list); - vm_bos = amdgpu_vm_get_bos(adev, bo_va->vm, &list); + vm_bos = amdgpu_vm_get_bos(adev, bo_va->vm, &list, &duplicates); if (!vm_bos) return; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index fce4c6d952c8..f6c1d6f0bf37 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -78,14 +78,16 @@ static unsigned amdgpu_vm_directory_size(struct amdgpu_device *adev) * amdgpu_vm_get_bos - add the vm BOs to a validation list * * @vm: vm providing the BOs - * @head: head of validation list + * @validated: head of validation list + * @duplicates: head of duplicates list * * Add the page directory to the list of BOs to * validate for command submission (cayman+). */ struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev, - struct amdgpu_vm *vm, - struct list_head *head) + struct amdgpu_vm *vm, + struct list_head *validated, + struct list_head *duplicates) { struct amdgpu_bo_list_entry *list; unsigned i, idx; @@ -103,7 +105,7 @@ struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev, list[0].priority = 0; list[0].tv.bo = &vm->page_directory->tbo; list[0].tv.shared = true; - list_add(&list[0].tv.head, head); + list_add(&list[0].tv.head, validated); for (i = 0, idx = 1; i <= vm->max_pde_used; i++) { if (!vm->page_tables[i].bo) @@ -115,7 +117,7 @@ struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev, list[idx].priority = 0; list[idx].tv.bo = &list[idx].robj->tbo; list[idx].tv.shared = true; - list_add(&list[idx++].tv.head, head); + list_add(&list[idx++].tv.head, duplicates); } return list; -- cgit From 56467ebfb254836dc30eb45d4ac8a46a400bfad6 Mon Sep 17 00:00:00 2001 From: Christian König Date: Fri, 11 Dec 2015 15:16:32 +0100 Subject: drm/amdgpu: split VM PD and PT handling during CS MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This way we avoid the extra allocation for the page directory entry. Signed-off-by: Christian König Reviewed-by: Alex Deucher Reviewed-by: Chunming Zhou --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 10 +++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 9 +++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 16 +++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 45 ++++++++++++++++++++------------- 4 files changed, 51 insertions(+), 29 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index dc3dab539e4c..40850afa763f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -980,10 +980,11 @@ struct amdgpu_vm_manager { void amdgpu_vm_manager_fini(struct amdgpu_device *adev); int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm); -struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev, - struct amdgpu_vm *vm, - struct list_head *validated, - struct list_head *duplicates); +void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, + struct list_head *validated, + struct amdgpu_bo_list_entry *entry); +struct amdgpu_bo_list_entry *amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, + struct list_head *duplicates); int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_sync *sync); void amdgpu_vm_flush(struct amdgpu_ring *ring, @@ -1253,6 +1254,7 @@ struct amdgpu_cs_parser { unsigned nchunks; struct amdgpu_cs_chunk *chunks; /* relocations */ + struct amdgpu_bo_list_entry vm_pd; struct amdgpu_bo_list_entry *vm_bos; struct list_head validated; struct fence *fence; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 9591c13781bd..3fb21ecd29e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -387,8 +387,7 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) } INIT_LIST_HEAD(&duplicates); - p->vm_bos = amdgpu_vm_get_bos(p->adev, &fpriv->vm, - &p->validated, &duplicates); + amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd); if (need_mmap_lock) down_read(¤t->mm->mmap_sem); @@ -397,6 +396,12 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) if (unlikely(r != 0)) goto error_reserve; + p->vm_bos = amdgpu_vm_get_pt_bos(&fpriv->vm, &duplicates); + if (!p->vm_bos) { + r = -ENOMEM; + goto error_validate; + } + r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &p->validated); if (r) goto error_validate; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 7fe7f8afa5ff..ea0fe94e4b54 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -448,6 +448,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, { struct ttm_validate_buffer tv, *entry; struct amdgpu_bo_list_entry *vm_bos; + struct amdgpu_bo_list_entry vm_pd; struct ww_acquire_ctx ticket; struct list_head list, duplicates; unsigned domain; @@ -460,14 +461,18 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, tv.shared = true; list_add(&tv.head, &list); - vm_bos = amdgpu_vm_get_bos(adev, bo_va->vm, &list, &duplicates); - if (!vm_bos) - return; + amdgpu_vm_get_pd_bo(bo_va->vm, &list, &vm_pd); /* Provide duplicates to avoid -EALREADY */ r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates); if (r) - goto error_free; + goto error_print; + + vm_bos = amdgpu_vm_get_pt_bos(bo_va->vm, &duplicates); + if (!vm_bos) { + r = -ENOMEM; + goto error_unreserve; + } list_for_each_entry(entry, &list, head) { domain = amdgpu_mem_type_to_domain(entry->bo->mem.mem_type); @@ -489,10 +494,9 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, error_unreserve: ttm_eu_backoff_reservation(&ticket, &list); - -error_free: drm_free_large(vm_bos); +error_print: if (r && r != -ERESTARTSYS) DRM_ERROR("Couldn't update BO_VA (%d)\n", r); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index f6c1d6f0bf37..592be6438a6c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -75,39 +75,50 @@ static unsigned amdgpu_vm_directory_size(struct amdgpu_device *adev) } /** - * amdgpu_vm_get_bos - add the vm BOs to a validation list + * amdgpu_vm_get_pd_bo - add the VM PD to a validation list * * @vm: vm providing the BOs * @validated: head of validation list + * @entry: entry to add + * + * Add the page directory to the list of BOs to + * validate for command submission. + */ +void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, + struct list_head *validated, + struct amdgpu_bo_list_entry *entry) +{ + entry->robj = vm->page_directory; + entry->prefered_domains = AMDGPU_GEM_DOMAIN_VRAM; + entry->allowed_domains = AMDGPU_GEM_DOMAIN_VRAM; + entry->priority = 0; + entry->tv.bo = &vm->page_directory->tbo; + entry->tv.shared = true; + list_add(&entry->tv.head, validated); +} + +/** + * amdgpu_vm_get_bos - add the vm BOs to a validation list + * + * @vm: vm providing the BOs * @duplicates: head of duplicates list * * Add the page directory to the list of BOs to * validate for command submission (cayman+). */ -struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev, - struct amdgpu_vm *vm, - struct list_head *validated, - struct list_head *duplicates) +struct amdgpu_bo_list_entry *amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, + struct list_head *duplicates) { struct amdgpu_bo_list_entry *list; unsigned i, idx; - list = drm_malloc_ab(vm->max_pde_used + 2, + list = drm_malloc_ab(vm->max_pde_used + 1, sizeof(struct amdgpu_bo_list_entry)); - if (!list) { + if (!list) return NULL; - } /* add the vm page table to the list */ - list[0].robj = vm->page_directory; - list[0].prefered_domains = AMDGPU_GEM_DOMAIN_VRAM; - list[0].allowed_domains = AMDGPU_GEM_DOMAIN_VRAM; - list[0].priority = 0; - list[0].tv.bo = &vm->page_directory->tbo; - list[0].tv.shared = true; - list_add(&list[0].tv.head, validated); - - for (i = 0, idx = 1; i <= vm->max_pde_used; i++) { + for (i = 0, idx = 0; i <= vm->max_pde_used; i++) { if (!vm->page_tables[i].bo) continue; -- cgit From ee1782c3f27fec5462363af48f27811b049155ab Mon Sep 17 00:00:00 2001 From: Christian König Date: Fri, 11 Dec 2015 21:01:23 +0100 Subject: drm/amdgpu: keep the PTs validation list in the VM v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This avoids allocating it on the fly. v2: fix grammar in comment Signed-off-by: Christian König Reviewed-by: Alex Deucher Reviewed-by: Chunming Zhou --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 8 ++---- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 7 +---- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 9 +----- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 50 +++++++++++++++------------------ 4 files changed, 27 insertions(+), 47 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 40850afa763f..d4e9272b60e8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -917,8 +917,8 @@ struct amdgpu_ring { #define AMDGPU_VM_FAULT_STOP_ALWAYS 2 struct amdgpu_vm_pt { - struct amdgpu_bo *bo; - uint64_t addr; + struct amdgpu_bo_list_entry entry; + uint64_t addr; }; struct amdgpu_vm_id { @@ -983,8 +983,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, struct list_head *validated, struct amdgpu_bo_list_entry *entry); -struct amdgpu_bo_list_entry *amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, - struct list_head *duplicates); +void amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, struct list_head *duplicates); int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_sync *sync); void amdgpu_vm_flush(struct amdgpu_ring *ring, @@ -1255,7 +1254,6 @@ struct amdgpu_cs_parser { struct amdgpu_cs_chunk *chunks; /* relocations */ struct amdgpu_bo_list_entry vm_pd; - struct amdgpu_bo_list_entry *vm_bos; struct list_head validated; struct fence *fence; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 3fb21ecd29e0..6ce595ff1aff 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -396,11 +396,7 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) if (unlikely(r != 0)) goto error_reserve; - p->vm_bos = amdgpu_vm_get_pt_bos(&fpriv->vm, &duplicates); - if (!p->vm_bos) { - r = -ENOMEM; - goto error_validate; - } + amdgpu_vm_get_pt_bos(&fpriv->vm, &duplicates); r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &p->validated); if (r) @@ -483,7 +479,6 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bo if (parser->bo_list) amdgpu_bo_list_put(parser->bo_list); - drm_free_large(parser->vm_bos); for (i = 0; i < parser->nchunks; i++) drm_free_large(parser->chunks[i].kdata); kfree(parser->chunks); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index ea0fe94e4b54..8c5687e4a6d1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -447,7 +447,6 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va, uint32_t operation) { struct ttm_validate_buffer tv, *entry; - struct amdgpu_bo_list_entry *vm_bos; struct amdgpu_bo_list_entry vm_pd; struct ww_acquire_ctx ticket; struct list_head list, duplicates; @@ -468,12 +467,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, if (r) goto error_print; - vm_bos = amdgpu_vm_get_pt_bos(bo_va->vm, &duplicates); - if (!vm_bos) { - r = -ENOMEM; - goto error_unreserve; - } - + amdgpu_vm_get_pt_bos(bo_va->vm, &duplicates); list_for_each_entry(entry, &list, head) { domain = amdgpu_mem_type_to_domain(entry->bo->mem.mem_type); /* if anything is swapped out don't swap it in here, @@ -494,7 +488,6 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev, error_unreserve: ttm_eu_backoff_reservation(&ticket, &list); - drm_free_large(vm_bos); error_print: if (r && r != -ERESTARTSYS) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 592be6438a6c..e0fa9d9ff5c2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -98,40 +98,27 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, } /** - * amdgpu_vm_get_bos - add the vm BOs to a validation list + * amdgpu_vm_get_bos - add the vm BOs to a duplicates list * * @vm: vm providing the BOs * @duplicates: head of duplicates list * - * Add the page directory to the list of BOs to - * validate for command submission (cayman+). + * Add the page directory to the BO duplicates list + * for command submission. */ -struct amdgpu_bo_list_entry *amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, - struct list_head *duplicates) +void amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, struct list_head *duplicates) { - struct amdgpu_bo_list_entry *list; - unsigned i, idx; - - list = drm_malloc_ab(vm->max_pde_used + 1, - sizeof(struct amdgpu_bo_list_entry)); - if (!list) - return NULL; + unsigned i; /* add the vm page table to the list */ - for (i = 0, idx = 0; i <= vm->max_pde_used; i++) { - if (!vm->page_tables[i].bo) + for (i = 0; i <= vm->max_pde_used; ++i) { + struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry; + + if (!entry->robj) continue; - list[idx].robj = vm->page_tables[i].bo; - list[idx].prefered_domains = AMDGPU_GEM_DOMAIN_VRAM; - list[idx].allowed_domains = AMDGPU_GEM_DOMAIN_VRAM; - list[idx].priority = 0; - list[idx].tv.bo = &list[idx].robj->tbo; - list[idx].tv.shared = true; - list_add(&list[idx++].tv.head, duplicates); + list_add(&entry->tv.head, duplicates); } - - return list; } /** @@ -474,7 +461,7 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, /* walk over the address space and update the page directory */ for (pt_idx = 0; pt_idx <= vm->max_pde_used; ++pt_idx) { - struct amdgpu_bo *bo = vm->page_tables[pt_idx].bo; + struct amdgpu_bo *bo = vm->page_tables[pt_idx].entry.robj; uint64_t pde, pt; if (bo == NULL) @@ -651,7 +638,7 @@ static int amdgpu_vm_update_ptes(struct amdgpu_device *adev, /* walk over the address space and update the page tables */ for (addr = start; addr < end; ) { uint64_t pt_idx = addr >> amdgpu_vm_block_size; - struct amdgpu_bo *pt = vm->page_tables[pt_idx].bo; + struct amdgpu_bo *pt = vm->page_tables[pt_idx].entry.robj; unsigned nptes; uint64_t pte; int r; @@ -1083,9 +1070,11 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, /* walk over the address space and allocate the page tables */ for (pt_idx = saddr; pt_idx <= eaddr; ++pt_idx) { struct reservation_object *resv = vm->page_directory->tbo.resv; + struct amdgpu_bo_list_entry *entry; struct amdgpu_bo *pt; - if (vm->page_tables[pt_idx].bo) + entry = &vm->page_tables[pt_idx].entry; + if (entry->robj) continue; r = amdgpu_bo_create(adev, AMDGPU_VM_PTE_COUNT * 8, @@ -1102,8 +1091,13 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, goto error_free; } + entry->robj = pt; + entry->prefered_domains = AMDGPU_GEM_DOMAIN_VRAM; + entry->allowed_domains = AMDGPU_GEM_DOMAIN_VRAM; + entry->priority = 0; + entry->tv.bo = &entry->robj->tbo; + entry->tv.shared = true; vm->page_tables[pt_idx].addr = 0; - vm->page_tables[pt_idx].bo = pt; } return 0; @@ -1334,7 +1328,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm) } for (i = 0; i < amdgpu_vm_num_pdes(adev); i++) - amdgpu_bo_unref(&vm->page_tables[i].bo); + amdgpu_bo_unref(&vm->page_tables[i].entry.robj); kfree(vm->page_tables); amdgpu_bo_unref(&vm->page_directory); -- cgit From 41869c1c7fe583dec932eb3d87de2e010b30a737 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Thu, 17 Dec 2015 09:57:49 -0500 Subject: drm/amdgpu: fix dp link rate selection (v2) Need to properly handle the max link rate in the dpcd. This prevents some cases where 5.4 Ghz is selected when it shouldn't be. v2: simplify logic, add array bounds check Reviewed-by: Tom St Denis Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/atombios_dp.c | 96 ++++++++++++-------------------- 1 file changed, 36 insertions(+), 60 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c index 92b6acadfc52..21aacc1f45c1 100644 --- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c +++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c @@ -243,7 +243,7 @@ static void amdgpu_atombios_dp_get_adjust_train(const u8 link_status[DP_LINK_STA /* convert bits per color to bits per pixel */ /* get bpc from the EDID */ -static int amdgpu_atombios_dp_convert_bpc_to_bpp(int bpc) +static unsigned amdgpu_atombios_dp_convert_bpc_to_bpp(int bpc) { if (bpc == 0) return 24; @@ -251,64 +251,32 @@ static int amdgpu_atombios_dp_convert_bpc_to_bpp(int bpc) return bpc * 3; } -/* get the max pix clock supported by the link rate and lane num */ -static int amdgpu_atombios_dp_get_max_dp_pix_clock(int link_rate, - int lane_num, - int bpp) -{ - return (link_rate * lane_num * 8) / bpp; -} - /***** amdgpu specific DP functions *****/ -/* First get the min lane# when low rate is used according to pixel clock - * (prefer low rate), second check max lane# supported by DP panel, - * if the max lane# < low rate lane# then use max lane# instead. - */ -static int amdgpu_atombios_dp_get_dp_lane_number(struct drm_connector *connector, +static int amdgpu_atombios_dp_get_dp_link_config(struct drm_connector *connector, const u8 dpcd[DP_DPCD_SIZE], - int pix_clock) -{ - int bpp = amdgpu_atombios_dp_convert_bpc_to_bpp(amdgpu_connector_get_monitor_bpc(connector)); - int max_link_rate = drm_dp_max_link_rate(dpcd); - int max_lane_num = drm_dp_max_lane_count(dpcd); - int lane_num; - int max_dp_pix_clock; - - for (lane_num = 1; lane_num < max_lane_num; lane_num <<= 1) { - max_dp_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(max_link_rate, lane_num, bpp); - if (pix_clock <= max_dp_pix_clock) - break; - } - - return lane_num; -} - -static int amdgpu_atombios_dp_get_dp_link_clock(struct drm_connector *connector, - const u8 dpcd[DP_DPCD_SIZE], - int pix_clock) + unsigned pix_clock, + unsigned *dp_lanes, unsigned *dp_rate) { - int bpp = amdgpu_atombios_dp_convert_bpc_to_bpp(amdgpu_connector_get_monitor_bpc(connector)); - int lane_num, max_pix_clock; - - if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) == - ENCODER_OBJECT_ID_NUTMEG) - return 270000; - - lane_num = amdgpu_atombios_dp_get_dp_lane_number(connector, dpcd, pix_clock); - max_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(162000, lane_num, bpp); - if (pix_clock <= max_pix_clock) - return 162000; - max_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(270000, lane_num, bpp); - if (pix_clock <= max_pix_clock) - return 270000; - if (amdgpu_connector_is_dp12_capable(connector)) { - max_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(540000, lane_num, bpp); - if (pix_clock <= max_pix_clock) - return 540000; + unsigned bpp = + amdgpu_atombios_dp_convert_bpc_to_bpp(amdgpu_connector_get_monitor_bpc(connector)); + static const unsigned link_rates[3] = { 162000, 270000, 540000 }; + unsigned max_link_rate = drm_dp_max_link_rate(dpcd); + unsigned max_lane_num = drm_dp_max_lane_count(dpcd); + unsigned lane_num, i, max_pix_clock; + + for (lane_num = 1; lane_num <= max_lane_num; lane_num <<= 1) { + for (i = 0; i < ARRAY_SIZE(link_rates) && link_rates[i] <= max_link_rate; i++) { + max_pix_clock = (lane_num * link_rates[i] * 8) / bpp; + if (max_pix_clock >= pix_clock) { + *dp_lanes = lane_num; + *dp_rate = link_rates[i]; + return 0; + } + } } - return drm_dp_max_link_rate(dpcd); + return -EINVAL; } static u8 amdgpu_atombios_dp_encoder_service(struct amdgpu_device *adev, @@ -422,6 +390,7 @@ void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector, { struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector); struct amdgpu_connector_atom_dig *dig_connector; + int ret; if (!amdgpu_connector->con_priv) return; @@ -429,10 +398,14 @@ void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector, if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) { - dig_connector->dp_clock = - amdgpu_atombios_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock); - dig_connector->dp_lane_count = - amdgpu_atombios_dp_get_dp_lane_number(connector, dig_connector->dpcd, mode->clock); + ret = amdgpu_atombios_dp_get_dp_link_config(connector, dig_connector->dpcd, + mode->clock, + &dig_connector->dp_lane_count, + &dig_connector->dp_clock); + if (ret) { + dig_connector->dp_clock = 0; + dig_connector->dp_lane_count = 0; + } } } @@ -441,14 +414,17 @@ int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector, { struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector); struct amdgpu_connector_atom_dig *dig_connector; - int dp_clock; + unsigned dp_lanes, dp_clock; + int ret; if (!amdgpu_connector->con_priv) return MODE_CLOCK_HIGH; dig_connector = amdgpu_connector->con_priv; - dp_clock = - amdgpu_atombios_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock); + ret = amdgpu_atombios_dp_get_dp_link_config(connector, dig_connector->dpcd, + mode->clock, &dp_lanes, &dp_clock); + if (ret) + return MODE_CLOCK_HIGH; if ((dp_clock == 540000) && (!amdgpu_connector_is_dp12_capable(connector))) -- cgit From 005ae95e6ec119c64e2d16eb65a94c49e1dcf9f0 Mon Sep 17 00:00:00 2001 From: Felix Kuehling Date: Mon, 23 Nov 2015 17:43:48 -0500 Subject: drm/amdgpu: Fix off-by-one errors in amdgpu_vm_bo_map MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit eaddr is sometimes treated as the last address inside the address range, and sometimes as the first address outside the range. This was resulting in errors when a test filled up the entire address space. Make it consistent to always be the last address within the range. Signed-off-by: Felix.Kuehling Reviewed-by: Christian König Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index e0fa9d9ff5c2..d6ff5dad98f6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -1010,13 +1010,13 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, return -EINVAL; /* make sure object fit at this offset */ - eaddr = saddr + size; + eaddr = saddr + size - 1; if ((saddr >= eaddr) || (offset + size > amdgpu_bo_size(bo_va->bo))) return -EINVAL; last_pfn = eaddr / AMDGPU_GPU_PAGE_SIZE; - if (last_pfn > adev->vm_manager.max_pfn) { - dev_err(adev->dev, "va above limit (0x%08X > 0x%08X)\n", + if (last_pfn >= adev->vm_manager.max_pfn) { + dev_err(adev->dev, "va above limit (0x%08X >= 0x%08X)\n", last_pfn, adev->vm_manager.max_pfn); return -EINVAL; } @@ -1025,7 +1025,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, eaddr /= AMDGPU_GPU_PAGE_SIZE; spin_lock(&vm->it_lock); - it = interval_tree_iter_first(&vm->va, saddr, eaddr - 1); + it = interval_tree_iter_first(&vm->va, saddr, eaddr); spin_unlock(&vm->it_lock); if (it) { struct amdgpu_bo_va_mapping *tmp; @@ -1046,7 +1046,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, INIT_LIST_HEAD(&mapping->list); mapping->it.start = saddr; - mapping->it.last = eaddr - 1; + mapping->it.last = eaddr; mapping->offset = offset; mapping->flags = flags; -- cgit From 3a2c788d95a24dc4cf720ddd19c1b115a03f41bf Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Tue, 25 Aug 2015 15:57:43 +0800 Subject: drm/amdgpu: share struct amdgpu_pm_state_type with powerplay module rename amdgpu_pm_state_type to amd_pm_state_type Signed-off-by: Rex Zhu Acked-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 28 ++-------------------------- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 8 ++++---- 2 files changed, 6 insertions(+), 30 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index d4e9272b60e8..d454ad6ff798 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1299,31 +1299,7 @@ struct amdgpu_wb { int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb); void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb); -/** - * struct amdgpu_pm - power management datas - * It keeps track of various data needed to take powermanagement decision. - */ -enum amdgpu_pm_state_type { - /* not used for dpm */ - POWER_STATE_TYPE_DEFAULT, - POWER_STATE_TYPE_POWERSAVE, - /* user selectable states */ - POWER_STATE_TYPE_BATTERY, - POWER_STATE_TYPE_BALANCED, - POWER_STATE_TYPE_PERFORMANCE, - /* internal states */ - POWER_STATE_TYPE_INTERNAL_UVD, - POWER_STATE_TYPE_INTERNAL_UVD_SD, - POWER_STATE_TYPE_INTERNAL_UVD_HD, - POWER_STATE_TYPE_INTERNAL_UVD_HD2, - POWER_STATE_TYPE_INTERNAL_UVD_MVC, - POWER_STATE_TYPE_INTERNAL_BOOT, - POWER_STATE_TYPE_INTERNAL_THERMAL, - POWER_STATE_TYPE_INTERNAL_ACPI, - POWER_STATE_TYPE_INTERNAL_ULV, - POWER_STATE_TYPE_INTERNAL_3DPERF, -}; enum amdgpu_int_thermal_type { THERMAL_TYPE_NONE, @@ -1605,8 +1581,8 @@ struct amdgpu_dpm { /* vce requirements */ struct amdgpu_vce_state vce_states[AMDGPU_MAX_VCE_LEVELS]; enum amdgpu_vce_level vce_level; - enum amdgpu_pm_state_type state; - enum amdgpu_pm_state_type user_state; + enum amd_pm_state_type state; + enum amd_pm_state_type user_state; u32 platform_caps; u32 voltage_response_time; u32 backbias_response_time; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 22a8c7d3a3ab..eea1933947ae 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -52,7 +52,7 @@ static ssize_t amdgpu_get_dpm_state(struct device *dev, { struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = ddev->dev_private; - enum amdgpu_pm_state_type pm = adev->pm.dpm.user_state; + enum amd_pm_state_type pm = adev->pm.dpm.user_state; return snprintf(buf, PAGE_SIZE, "%s\n", (pm == POWER_STATE_TYPE_BATTERY) ? "battery" : @@ -351,7 +351,7 @@ void amdgpu_dpm_thermal_work_handler(struct work_struct *work) container_of(work, struct amdgpu_device, pm.dpm.thermal.work); /* switch to the thermal state */ - enum amdgpu_pm_state_type dpm_state = POWER_STATE_TYPE_INTERNAL_THERMAL; + enum amd_pm_state_type dpm_state = POWER_STATE_TYPE_INTERNAL_THERMAL; if (!adev->pm.dpm_enabled) return; @@ -379,7 +379,7 @@ void amdgpu_dpm_thermal_work_handler(struct work_struct *work) } static struct amdgpu_ps *amdgpu_dpm_pick_power_state(struct amdgpu_device *adev, - enum amdgpu_pm_state_type dpm_state) + enum amd_pm_state_type dpm_state) { int i; struct amdgpu_ps *ps; @@ -516,7 +516,7 @@ static void amdgpu_dpm_change_power_state_locked(struct amdgpu_device *adev) { int i; struct amdgpu_ps *ps; - enum amdgpu_pm_state_type dpm_state; + enum amd_pm_state_type dpm_state; int ret; /* if dpm init failed */ -- cgit From 7e85be994828b15014e4748e4b5eca015be6adce Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Mon, 21 Sep 2015 14:29:10 +0800 Subject: drm/amdgpu: mv some definition from amdgpu_acpi.c to amdgpu_acpi.h These will be shared with the new powerplay module. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c | 56 ------------------------------- drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h | 57 +++++++++++++++++++++++++++++--- 2 files changed, 53 insertions(+), 60 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index a142d5ae148d..5df5b8390b03 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -32,63 +32,7 @@ #include "amdgpu_acpi.h" #include "atom.h" -#define ACPI_AC_CLASS "ac_adapter" - extern void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev); - -struct atif_verify_interface { - u16 size; /* structure size in bytes (includes size field) */ - u16 version; /* version */ - u32 notification_mask; /* supported notifications mask */ - u32 function_bits; /* supported functions bit vector */ -} __packed; - -struct atif_system_params { - u16 size; /* structure size in bytes (includes size field) */ - u32 valid_mask; /* valid flags mask */ - u32 flags; /* flags */ - u8 command_code; /* notify command code */ -} __packed; - -struct atif_sbios_requests { - u16 size; /* structure size in bytes (includes size field) */ - u32 pending; /* pending sbios requests */ - u8 panel_exp_mode; /* panel expansion mode */ - u8 thermal_gfx; /* thermal state: target gfx controller */ - u8 thermal_state; /* thermal state: state id (0: exit state, non-0: state) */ - u8 forced_power_gfx; /* forced power state: target gfx controller */ - u8 forced_power_state; /* forced power state: state id */ - u8 system_power_src; /* system power source */ - u8 backlight_level; /* panel backlight level (0-255) */ -} __packed; - -#define ATIF_NOTIFY_MASK 0x3 -#define ATIF_NOTIFY_NONE 0 -#define ATIF_NOTIFY_81 1 -#define ATIF_NOTIFY_N 2 - -struct atcs_verify_interface { - u16 size; /* structure size in bytes (includes size field) */ - u16 version; /* version */ - u32 function_bits; /* supported functions bit vector */ -} __packed; - -#define ATCS_VALID_FLAGS_MASK 0x3 - -struct atcs_pref_req_input { - u16 size; /* structure size in bytes (includes size field) */ - u16 client_id; /* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */ - u16 valid_flags_mask; /* valid flags mask */ - u16 flags; /* flags */ - u8 req_type; /* request type */ - u8 perf_req; /* performance request */ -} __packed; - -struct atcs_pref_req_output { - u16 size; /* structure size in bytes (includes size field) */ - u8 ret_val; /* return value */ -} __packed; - /* Call the ATIF method */ /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h index 01a29c3d7011..51ce3e337503 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h @@ -24,11 +24,60 @@ #ifndef AMDGPU_ACPI_H #define AMDGPU_ACPI_H -struct amdgpu_device; -struct acpi_bus_event; +#define ACPI_AC_CLASS "ac_adapter" -int amdgpu_atif_handler(struct amdgpu_device *adev, - struct acpi_bus_event *event); +struct atif_verify_interface { + u16 size; /* structure size in bytes (includes size field) */ + u16 version; /* version */ + u32 notification_mask; /* supported notifications mask */ + u32 function_bits; /* supported functions bit vector */ +} __packed; + +struct atif_system_params { + u16 size; /* structure size in bytes (includes size field) */ + u32 valid_mask; /* valid flags mask */ + u32 flags; /* flags */ + u8 command_code; /* notify command code */ +} __packed; + +struct atif_sbios_requests { + u16 size; /* structure size in bytes (includes size field) */ + u32 pending; /* pending sbios requests */ + u8 panel_exp_mode; /* panel expansion mode */ + u8 thermal_gfx; /* thermal state: target gfx controller */ + u8 thermal_state; /* thermal state: state id (0: exit state, non-0: state) */ + u8 forced_power_gfx; /* forced power state: target gfx controller */ + u8 forced_power_state; /* forced power state: state id */ + u8 system_power_src; /* system power source */ + u8 backlight_level; /* panel backlight level (0-255) */ +} __packed; + +#define ATIF_NOTIFY_MASK 0x3 +#define ATIF_NOTIFY_NONE 0 +#define ATIF_NOTIFY_81 1 +#define ATIF_NOTIFY_N 2 + +struct atcs_verify_interface { + u16 size; /* structure size in bytes (includes size field) */ + u16 version; /* version */ + u32 function_bits; /* supported functions bit vector */ +} __packed; + +#define ATCS_VALID_FLAGS_MASK 0x3 + +struct atcs_pref_req_input { + u16 size; /* structure size in bytes (includes size field) */ + u16 client_id; /* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */ + u16 valid_flags_mask; /* valid flags mask */ + u16 flags; /* flags */ + u8 req_type; /* request type */ + u8 perf_req; /* performance request */ +} __packed; + +struct atcs_pref_req_output { + u16 size; /* structure size in bytes (includes size field) */ + u8 ret_val; /* return value */ +} __packed; /* AMD hw uses four ACPI control methods: * 1. ATIF -- cgit From 66dc0ddd02a1fdc6678e0b3ffcb257630462afaa Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Fri, 18 Sep 2015 16:35:17 +0800 Subject: drm/amdgpu: mv amdgpu_acpi.h to amd/include/amd_acpi.h This will be shared with the new powerplay module. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h | 494 ----------------------- drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c | 2 +- 3 files changed, 2 insertions(+), 496 deletions(-) delete mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 5df5b8390b03..5cd7b736a9de 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -29,7 +29,7 @@ #include #include #include "amdgpu.h" -#include "amdgpu_acpi.h" +#include "amd_acpi.h" #include "atom.h" extern void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h deleted file mode 100644 index 51ce3e337503..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h +++ /dev/null @@ -1,494 +0,0 @@ -/* - * Copyright 2012 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef AMDGPU_ACPI_H -#define AMDGPU_ACPI_H - -#define ACPI_AC_CLASS "ac_adapter" - -struct atif_verify_interface { - u16 size; /* structure size in bytes (includes size field) */ - u16 version; /* version */ - u32 notification_mask; /* supported notifications mask */ - u32 function_bits; /* supported functions bit vector */ -} __packed; - -struct atif_system_params { - u16 size; /* structure size in bytes (includes size field) */ - u32 valid_mask; /* valid flags mask */ - u32 flags; /* flags */ - u8 command_code; /* notify command code */ -} __packed; - -struct atif_sbios_requests { - u16 size; /* structure size in bytes (includes size field) */ - u32 pending; /* pending sbios requests */ - u8 panel_exp_mode; /* panel expansion mode */ - u8 thermal_gfx; /* thermal state: target gfx controller */ - u8 thermal_state; /* thermal state: state id (0: exit state, non-0: state) */ - u8 forced_power_gfx; /* forced power state: target gfx controller */ - u8 forced_power_state; /* forced power state: state id */ - u8 system_power_src; /* system power source */ - u8 backlight_level; /* panel backlight level (0-255) */ -} __packed; - -#define ATIF_NOTIFY_MASK 0x3 -#define ATIF_NOTIFY_NONE 0 -#define ATIF_NOTIFY_81 1 -#define ATIF_NOTIFY_N 2 - -struct atcs_verify_interface { - u16 size; /* structure size in bytes (includes size field) */ - u16 version; /* version */ - u32 function_bits; /* supported functions bit vector */ -} __packed; - -#define ATCS_VALID_FLAGS_MASK 0x3 - -struct atcs_pref_req_input { - u16 size; /* structure size in bytes (includes size field) */ - u16 client_id; /* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */ - u16 valid_flags_mask; /* valid flags mask */ - u16 flags; /* flags */ - u8 req_type; /* request type */ - u8 perf_req; /* performance request */ -} __packed; - -struct atcs_pref_req_output { - u16 size; /* structure size in bytes (includes size field) */ - u8 ret_val; /* return value */ -} __packed; - -/* AMD hw uses four ACPI control methods: - * 1. ATIF - * ARG0: (ACPI_INTEGER) function code - * ARG1: (ACPI_BUFFER) parameter buffer, 256 bytes - * OUTPUT: (ACPI_BUFFER) output buffer, 256 bytes - * ATIF provides an entry point for the gfx driver to interact with the sbios. - * The AMD ACPI notification mechanism uses Notify (VGA, 0x81) or a custom - * notification. Which notification is used as indicated by the ATIF Control - * Method GET_SYSTEM_PARAMETERS. When the driver receives Notify (VGA, 0x81) or - * a custom notification it invokes ATIF Control Method GET_SYSTEM_BIOS_REQUESTS - * to identify pending System BIOS requests and associated parameters. For - * example, if one of the pending requests is DISPLAY_SWITCH_REQUEST, the driver - * will perform display device detection and invoke ATIF Control Method - * SELECT_ACTIVE_DISPLAYS. - * - * 2. ATPX - * ARG0: (ACPI_INTEGER) function code - * ARG1: (ACPI_BUFFER) parameter buffer, 256 bytes - * OUTPUT: (ACPI_BUFFER) output buffer, 256 bytes - * ATPX methods are used on PowerXpress systems to handle mux switching and - * discrete GPU power control. - * - * 3. ATRM - * ARG0: (ACPI_INTEGER) offset of vbios rom data - * ARG1: (ACPI_BUFFER) size of the buffer to fill (up to 4K). - * OUTPUT: (ACPI_BUFFER) output buffer - * ATRM provides an interfacess to access the discrete GPU vbios image on - * PowerXpress systems with multiple GPUs. - * - * 4. ATCS - * ARG0: (ACPI_INTEGER) function code - * ARG1: (ACPI_BUFFER) parameter buffer, 256 bytes - * OUTPUT: (ACPI_BUFFER) output buffer, 256 bytes - * ATCS provides an interface to AMD chipset specific functionality. - * - */ -/* ATIF */ -#define ATIF_FUNCTION_VERIFY_INTERFACE 0x0 -/* ARG0: ATIF_FUNCTION_VERIFY_INTERFACE - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * WORD - version - * DWORD - supported notifications mask - * DWORD - supported functions bit vector - */ -/* Notifications mask */ -# define ATIF_DISPLAY_SWITCH_REQUEST_SUPPORTED (1 << 0) -# define ATIF_EXPANSION_MODE_CHANGE_REQUEST_SUPPORTED (1 << 1) -# define ATIF_THERMAL_STATE_CHANGE_REQUEST_SUPPORTED (1 << 2) -# define ATIF_FORCED_POWER_STATE_CHANGE_REQUEST_SUPPORTED (1 << 3) -# define ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST_SUPPORTED (1 << 4) -# define ATIF_DISPLAY_CONF_CHANGE_REQUEST_SUPPORTED (1 << 5) -# define ATIF_PX_GFX_SWITCH_REQUEST_SUPPORTED (1 << 6) -# define ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST_SUPPORTED (1 << 7) -# define ATIF_DGPU_DISPLAY_EVENT_SUPPORTED (1 << 8) -/* supported functions vector */ -# define ATIF_GET_SYSTEM_PARAMETERS_SUPPORTED (1 << 0) -# define ATIF_GET_SYSTEM_BIOS_REQUESTS_SUPPORTED (1 << 1) -# define ATIF_SELECT_ACTIVE_DISPLAYS_SUPPORTED (1 << 2) -# define ATIF_GET_LID_STATE_SUPPORTED (1 << 3) -# define ATIF_GET_TV_STANDARD_FROM_CMOS_SUPPORTED (1 << 4) -# define ATIF_SET_TV_STANDARD_IN_CMOS_SUPPORTED (1 << 5) -# define ATIF_GET_PANEL_EXPANSION_MODE_FROM_CMOS_SUPPORTED (1 << 6) -# define ATIF_SET_PANEL_EXPANSION_MODE_IN_CMOS_SUPPORTED (1 << 7) -# define ATIF_TEMPERATURE_CHANGE_NOTIFICATION_SUPPORTED (1 << 12) -# define ATIF_GET_GRAPHICS_DEVICE_TYPES_SUPPORTED (1 << 14) -#define ATIF_FUNCTION_GET_SYSTEM_PARAMETERS 0x1 -/* ARG0: ATIF_FUNCTION_GET_SYSTEM_PARAMETERS - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * DWORD - valid flags mask - * DWORD - flags - * - * OR - * - * WORD - structure size in bytes (includes size field) - * DWORD - valid flags mask - * DWORD - flags - * BYTE - notify command code - * - * flags - * bits 1:0: - * 0 - Notify(VGA, 0x81) is not used for notification - * 1 - Notify(VGA, 0x81) is used for notification - * 2 - Notify(VGA, n) is used for notification where - * n (0xd0-0xd9) is specified in notify command code. - * bit 2: - * 1 - lid changes not reported though int10 - */ -#define ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS 0x2 -/* ARG0: ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * DWORD - pending sbios requests - * BYTE - panel expansion mode - * BYTE - thermal state: target gfx controller - * BYTE - thermal state: state id (0: exit state, non-0: state) - * BYTE - forced power state: target gfx controller - * BYTE - forced power state: state id - * BYTE - system power source - * BYTE - panel backlight level (0-255) - */ -/* pending sbios requests */ -# define ATIF_DISPLAY_SWITCH_REQUEST (1 << 0) -# define ATIF_EXPANSION_MODE_CHANGE_REQUEST (1 << 1) -# define ATIF_THERMAL_STATE_CHANGE_REQUEST (1 << 2) -# define ATIF_FORCED_POWER_STATE_CHANGE_REQUEST (1 << 3) -# define ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST (1 << 4) -# define ATIF_DISPLAY_CONF_CHANGE_REQUEST (1 << 5) -# define ATIF_PX_GFX_SWITCH_REQUEST (1 << 6) -# define ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST (1 << 7) -# define ATIF_DGPU_DISPLAY_EVENT (1 << 8) -/* panel expansion mode */ -# define ATIF_PANEL_EXPANSION_DISABLE 0 -# define ATIF_PANEL_EXPANSION_FULL 1 -# define ATIF_PANEL_EXPANSION_ASPECT 2 -/* target gfx controller */ -# define ATIF_TARGET_GFX_SINGLE 0 -# define ATIF_TARGET_GFX_PX_IGPU 1 -# define ATIF_TARGET_GFX_PX_DGPU 2 -/* system power source */ -# define ATIF_POWER_SOURCE_AC 1 -# define ATIF_POWER_SOURCE_DC 2 -# define ATIF_POWER_SOURCE_RESTRICTED_AC_1 3 -# define ATIF_POWER_SOURCE_RESTRICTED_AC_2 4 -#define ATIF_FUNCTION_SELECT_ACTIVE_DISPLAYS 0x3 -/* ARG0: ATIF_FUNCTION_SELECT_ACTIVE_DISPLAYS - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - selected displays - * WORD - connected displays - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * WORD - selected displays - */ -# define ATIF_LCD1 (1 << 0) -# define ATIF_CRT1 (1 << 1) -# define ATIF_TV (1 << 2) -# define ATIF_DFP1 (1 << 3) -# define ATIF_CRT2 (1 << 4) -# define ATIF_LCD2 (1 << 5) -# define ATIF_DFP2 (1 << 7) -# define ATIF_CV (1 << 8) -# define ATIF_DFP3 (1 << 9) -# define ATIF_DFP4 (1 << 10) -# define ATIF_DFP5 (1 << 11) -# define ATIF_DFP6 (1 << 12) -#define ATIF_FUNCTION_GET_LID_STATE 0x4 -/* ARG0: ATIF_FUNCTION_GET_LID_STATE - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * BYTE - lid state (0: open, 1: closed) - * - * GET_LID_STATE only works at boot and resume, for general lid - * status, use the kernel provided status - */ -#define ATIF_FUNCTION_GET_TV_STANDARD_FROM_CMOS 0x5 -/* ARG0: ATIF_FUNCTION_GET_TV_STANDARD_FROM_CMOS - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * BYTE - 0 - * BYTE - TV standard - */ -# define ATIF_TV_STD_NTSC 0 -# define ATIF_TV_STD_PAL 1 -# define ATIF_TV_STD_PALM 2 -# define ATIF_TV_STD_PAL60 3 -# define ATIF_TV_STD_NTSCJ 4 -# define ATIF_TV_STD_PALCN 5 -# define ATIF_TV_STD_PALN 6 -# define ATIF_TV_STD_SCART_RGB 9 -#define ATIF_FUNCTION_SET_TV_STANDARD_IN_CMOS 0x6 -/* ARG0: ATIF_FUNCTION_SET_TV_STANDARD_IN_CMOS - * ARG1: - * WORD - structure size in bytes (includes size field) - * BYTE - 0 - * BYTE - TV standard - * OUTPUT: none - */ -#define ATIF_FUNCTION_GET_PANEL_EXPANSION_MODE_FROM_CMOS 0x7 -/* ARG0: ATIF_FUNCTION_GET_PANEL_EXPANSION_MODE_FROM_CMOS - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * BYTE - panel expansion mode - */ -#define ATIF_FUNCTION_SET_PANEL_EXPANSION_MODE_IN_CMOS 0x8 -/* ARG0: ATIF_FUNCTION_SET_PANEL_EXPANSION_MODE_IN_CMOS - * ARG1: - * WORD - structure size in bytes (includes size field) - * BYTE - panel expansion mode - * OUTPUT: none - */ -#define ATIF_FUNCTION_TEMPERATURE_CHANGE_NOTIFICATION 0xD -/* ARG0: ATIF_FUNCTION_TEMPERATURE_CHANGE_NOTIFICATION - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - gfx controller id - * BYTE - current temperature (degress Celsius) - * OUTPUT: none - */ -#define ATIF_FUNCTION_GET_GRAPHICS_DEVICE_TYPES 0xF -/* ARG0: ATIF_FUNCTION_GET_GRAPHICS_DEVICE_TYPES - * ARG1: none - * OUTPUT: - * WORD - number of gfx devices - * WORD - device structure size in bytes (excludes device size field) - * DWORD - flags \ - * WORD - bus number } repeated structure - * WORD - device number / - */ -/* flags */ -# define ATIF_PX_REMOVABLE_GRAPHICS_DEVICE (1 << 0) -# define ATIF_XGP_PORT (1 << 1) -# define ATIF_VGA_ENABLED_GRAPHICS_DEVICE (1 << 2) -# define ATIF_XGP_PORT_IN_DOCK (1 << 3) - -/* ATPX */ -#define ATPX_FUNCTION_VERIFY_INTERFACE 0x0 -/* ARG0: ATPX_FUNCTION_VERIFY_INTERFACE - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * WORD - version - * DWORD - supported functions bit vector - */ -/* supported functions vector */ -# define ATPX_GET_PX_PARAMETERS_SUPPORTED (1 << 0) -# define ATPX_POWER_CONTROL_SUPPORTED (1 << 1) -# define ATPX_DISPLAY_MUX_CONTROL_SUPPORTED (1 << 2) -# define ATPX_I2C_MUX_CONTROL_SUPPORTED (1 << 3) -# define ATPX_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION_SUPPORTED (1 << 4) -# define ATPX_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION_SUPPORTED (1 << 5) -# define ATPX_GET_DISPLAY_CONNECTORS_MAPPING_SUPPORTED (1 << 7) -# define ATPX_GET_DISPLAY_DETECTION_PORTS_SUPPORTED (1 << 8) -#define ATPX_FUNCTION_GET_PX_PARAMETERS 0x1 -/* ARG0: ATPX_FUNCTION_GET_PX_PARAMETERS - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * DWORD - valid flags mask - * DWORD - flags - */ -/* flags */ -# define ATPX_LVDS_I2C_AVAILABLE_TO_BOTH_GPUS (1 << 0) -# define ATPX_CRT1_I2C_AVAILABLE_TO_BOTH_GPUS (1 << 1) -# define ATPX_DVI1_I2C_AVAILABLE_TO_BOTH_GPUS (1 << 2) -# define ATPX_CRT1_RGB_SIGNAL_MUXED (1 << 3) -# define ATPX_TV_SIGNAL_MUXED (1 << 4) -# define ATPX_DFP_SIGNAL_MUXED (1 << 5) -# define ATPX_SEPARATE_MUX_FOR_I2C (1 << 6) -# define ATPX_DYNAMIC_PX_SUPPORTED (1 << 7) -# define ATPX_ACF_NOT_SUPPORTED (1 << 8) -# define ATPX_FIXED_NOT_SUPPORTED (1 << 9) -# define ATPX_DYNAMIC_DGPU_POWER_OFF_SUPPORTED (1 << 10) -# define ATPX_DGPU_REQ_POWER_FOR_DISPLAYS (1 << 11) -#define ATPX_FUNCTION_POWER_CONTROL 0x2 -/* ARG0: ATPX_FUNCTION_POWER_CONTROL - * ARG1: - * WORD - structure size in bytes (includes size field) - * BYTE - dGPU power state (0: power off, 1: power on) - * OUTPUT: none - */ -#define ATPX_FUNCTION_DISPLAY_MUX_CONTROL 0x3 -/* ARG0: ATPX_FUNCTION_DISPLAY_MUX_CONTROL - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - display mux control (0: iGPU, 1: dGPU) - * OUTPUT: none - */ -# define ATPX_INTEGRATED_GPU 0 -# define ATPX_DISCRETE_GPU 1 -#define ATPX_FUNCTION_I2C_MUX_CONTROL 0x4 -/* ARG0: ATPX_FUNCTION_I2C_MUX_CONTROL - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - i2c/aux/hpd mux control (0: iGPU, 1: dGPU) - * OUTPUT: none - */ -#define ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION 0x5 -/* ARG0: ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - target gpu (0: iGPU, 1: dGPU) - * OUTPUT: none - */ -#define ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION 0x6 -/* ARG0: ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - target gpu (0: iGPU, 1: dGPU) - * OUTPUT: none - */ -#define ATPX_FUNCTION_GET_DISPLAY_CONNECTORS_MAPPING 0x8 -/* ARG0: ATPX_FUNCTION_GET_DISPLAY_CONNECTORS_MAPPING - * ARG1: none - * OUTPUT: - * WORD - number of display connectors - * WORD - connector structure size in bytes (excludes connector size field) - * BYTE - flags \ - * BYTE - ATIF display vector bit position } repeated - * BYTE - adapter id (0: iGPU, 1-n: dGPU ordered by pcie bus number) } structure - * WORD - connector ACPI id / - */ -/* flags */ -# define ATPX_DISPLAY_OUTPUT_SUPPORTED_BY_ADAPTER_ID_DEVICE (1 << 0) -# define ATPX_DISPLAY_HPD_SUPPORTED_BY_ADAPTER_ID_DEVICE (1 << 1) -# define ATPX_DISPLAY_I2C_SUPPORTED_BY_ADAPTER_ID_DEVICE (1 << 2) -#define ATPX_FUNCTION_GET_DISPLAY_DETECTION_PORTS 0x9 -/* ARG0: ATPX_FUNCTION_GET_DISPLAY_DETECTION_PORTS - * ARG1: none - * OUTPUT: - * WORD - number of HPD/DDC ports - * WORD - port structure size in bytes (excludes port size field) - * BYTE - ATIF display vector bit position \ - * BYTE - hpd id } reapeated structure - * BYTE - ddc id / - * - * available on A+A systems only - */ -/* hpd id */ -# define ATPX_HPD_NONE 0 -# define ATPX_HPD1 1 -# define ATPX_HPD2 2 -# define ATPX_HPD3 3 -# define ATPX_HPD4 4 -# define ATPX_HPD5 5 -# define ATPX_HPD6 6 -/* ddc id */ -# define ATPX_DDC_NONE 0 -# define ATPX_DDC1 1 -# define ATPX_DDC2 2 -# define ATPX_DDC3 3 -# define ATPX_DDC4 4 -# define ATPX_DDC5 5 -# define ATPX_DDC6 6 -# define ATPX_DDC7 7 -# define ATPX_DDC8 8 - -/* ATCS */ -#define ATCS_FUNCTION_VERIFY_INTERFACE 0x0 -/* ARG0: ATCS_FUNCTION_VERIFY_INTERFACE - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * WORD - version - * DWORD - supported functions bit vector - */ -/* supported functions vector */ -# define ATCS_GET_EXTERNAL_STATE_SUPPORTED (1 << 0) -# define ATCS_PCIE_PERFORMANCE_REQUEST_SUPPORTED (1 << 1) -# define ATCS_PCIE_DEVICE_READY_NOTIFICATION_SUPPORTED (1 << 2) -# define ATCS_SET_PCIE_BUS_WIDTH_SUPPORTED (1 << 3) -#define ATCS_FUNCTION_GET_EXTERNAL_STATE 0x1 -/* ARG0: ATCS_FUNCTION_GET_EXTERNAL_STATE - * ARG1: none - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * DWORD - valid flags mask - * DWORD - flags (0: undocked, 1: docked) - */ -/* flags */ -# define ATCS_DOCKED (1 << 0) -#define ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST 0x2 -/* ARG0: ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) - * WORD - valid flags mask - * WORD - flags - * BYTE - request type - * BYTE - performance request - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * BYTE - return value - */ -/* flags */ -# define ATCS_ADVERTISE_CAPS (1 << 0) -# define ATCS_WAIT_FOR_COMPLETION (1 << 1) -/* request type */ -# define ATCS_PCIE_LINK_SPEED 1 -/* performance request */ -# define ATCS_REMOVE 0 -# define ATCS_FORCE_LOW_POWER 1 -# define ATCS_PERF_LEVEL_1 2 /* PCIE Gen 1 */ -# define ATCS_PERF_LEVEL_2 3 /* PCIE Gen 2 */ -# define ATCS_PERF_LEVEL_3 4 /* PCIE Gen 3 */ -/* return value */ -# define ATCS_REQUEST_REFUSED 1 -# define ATCS_REQUEST_COMPLETE 2 -# define ATCS_REQUEST_IN_PROGRESS 3 -#define ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION 0x3 -/* ARG0: ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION - * ARG1: none - * OUTPUT: none - */ -#define ATCS_FUNCTION_SET_PCIE_BUS_WIDTH 0x4 -/* ARG0: ATCS_FUNCTION_SET_PCIE_BUS_WIDTH - * ARG1: - * WORD - structure size in bytes (includes size field) - * WORD - client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) - * BYTE - number of active lanes - * OUTPUT: - * WORD - structure size in bytes (includes size field) - * BYTE - number of active lanes - */ - -#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c index 5a8fbadbd27b..3c895863fcf5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c @@ -11,7 +11,7 @@ #include #include -#include "amdgpu_acpi.h" +#include "amd_acpi.h" struct amdgpu_atpx_functions { bool px_params; -- cgit From 3f1d35a03b3cc7e0fb85c92f4ac6eafd1780d5dc Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Tue, 15 Sep 2015 14:44:44 +0800 Subject: drm/amdgpu: implement new cgs interface for acpi function Add a new driver internal interface for accessing ACPI methods. These will be used by various new components including powerplay. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 220 +++++++++++++++++++++++++++++++- 1 file changed, 218 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 8e995148f56e..f901cdce7c5f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -24,6 +24,7 @@ #include #include #include +#include #include #include #include @@ -32,7 +33,6 @@ #include "atom.h" #include "amdgpu_ucode.h" - struct amdgpu_cgs_device { struct cgs_device base; struct amdgpu_device *adev; @@ -736,6 +736,221 @@ static int amdgpu_cgs_get_firmware_info(void *cgs_device, return 0; } +/** \brief evaluate acpi namespace object, handle or pathname must be valid + * \param cgs_device + * \param info input/output arguments for the control method + * \return status + */ + +#if defined(CONFIG_ACPI) +static int amdgpu_cgs_acpi_eval_object(void *cgs_device, + struct cgs_acpi_method_info *info) +{ + CGS_FUNC_ADEV; + acpi_handle handle; + struct acpi_object_list input; + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + union acpi_object *params = NULL; + union acpi_object *obj = NULL; + uint8_t name[5] = {'\0'}; + struct cgs_acpi_method_argument *argument = NULL; + uint32_t i, count; + acpi_status status; + int result; + uint32_t func_no = 0xFFFFFFFF; + + handle = ACPI_HANDLE(&adev->pdev->dev); + if (!handle) + return -ENODEV; + + memset(&input, 0, sizeof(struct acpi_object_list)); + + /* validate input info */ + if (info->size != sizeof(struct cgs_acpi_method_info)) + return -EINVAL; + + input.count = info->input_count; + if (info->input_count > 0) { + if (info->pinput_argument == NULL) + return -EINVAL; + argument = info->pinput_argument; + func_no = argument->value; + for (i = 0; i < info->input_count; i++) { + if (((argument->type == ACPI_TYPE_STRING) || + (argument->type == ACPI_TYPE_BUFFER)) + && (argument->pointer == NULL)) + return -EINVAL; + argument++; + } + } + + if (info->output_count > 0) { + if (info->poutput_argument == NULL) + return -EINVAL; + argument = info->poutput_argument; + for (i = 0; i < info->output_count; i++) { + if (((argument->type == ACPI_TYPE_STRING) || + (argument->type == ACPI_TYPE_BUFFER)) + && (argument->pointer == NULL)) + return -EINVAL; + argument++; + } + } + + /* The path name passed to acpi_evaluate_object should be null terminated */ + if ((info->field & CGS_ACPI_FIELD_METHOD_NAME) != 0) { + strncpy(name, (char *)&(info->name), sizeof(uint32_t)); + name[4] = '\0'; + } + + /* parse input parameters */ + if (input.count > 0) { + input.pointer = params = + kzalloc(sizeof(union acpi_object) * input.count, GFP_KERNEL); + if (params == NULL) + return -EINVAL; + + argument = info->pinput_argument; + + for (i = 0; i < input.count; i++) { + params->type = argument->type; + switch (params->type) { + case ACPI_TYPE_INTEGER: + params->integer.value = argument->value; + break; + case ACPI_TYPE_STRING: + params->string.length = argument->method_length; + params->string.pointer = argument->pointer; + break; + case ACPI_TYPE_BUFFER: + params->buffer.length = argument->method_length; + params->buffer.pointer = argument->pointer; + break; + default: + break; + } + params++; + argument++; + } + } + + /* parse output info */ + count = info->output_count; + argument = info->poutput_argument; + + /* evaluate the acpi method */ + status = acpi_evaluate_object(handle, name, &input, &output); + + if (ACPI_FAILURE(status)) { + result = -EIO; + goto error; + } + + /* return the output info */ + obj = output.pointer; + + if (count > 1) { + if ((obj->type != ACPI_TYPE_PACKAGE) || + (obj->package.count != count)) { + result = -EIO; + goto error; + } + params = obj->package.elements; + } else + params = obj; + + if (params == NULL) { + result = -EIO; + goto error; + } + + for (i = 0; i < count; i++) { + if (argument->type != params->type) { + result = -EIO; + goto error; + } + switch (params->type) { + case ACPI_TYPE_INTEGER: + argument->value = params->integer.value; + break; + case ACPI_TYPE_STRING: + if ((params->string.length != argument->data_length) || + (params->string.pointer == NULL)) { + result = -EIO; + goto error; + } + strncpy(argument->pointer, + params->string.pointer, + params->string.length); + break; + case ACPI_TYPE_BUFFER: + if (params->buffer.pointer == NULL) { + result = -EIO; + goto error; + } + memcpy(argument->pointer, + params->buffer.pointer, + argument->data_length); + break; + default: + break; + } + argument++; + params++; + } + +error: + if (obj != NULL) + kfree(obj); + kfree((void *)input.pointer); + return result; +} +#else +static int amdgpu_cgs_acpi_eval_object(void *cgs_device, + struct cgs_acpi_method_info *info) +{ + return -EIO; +} +#endif + +int amdgpu_cgs_call_acpi_method(void *cgs_device, + uint32_t acpi_method, + uint32_t acpi_function, + void *pinput, void *poutput, + uint32_t output_count, + uint32_t input_size, + uint32_t output_size) +{ + struct cgs_acpi_method_argument acpi_input[2] = { {0}, {0} }; + struct cgs_acpi_method_argument acpi_output = {0}; + struct cgs_acpi_method_info info = {0}; + + acpi_input[0].type = CGS_ACPI_TYPE_INTEGER; + acpi_input[0].method_length = sizeof(uint32_t); + acpi_input[0].data_length = sizeof(uint32_t); + acpi_input[0].value = acpi_function; + + acpi_input[1].type = CGS_ACPI_TYPE_BUFFER; + acpi_input[1].method_length = CGS_ACPI_MAX_BUFFER_SIZE; + acpi_input[1].data_length = input_size; + acpi_input[1].pointer = pinput; + + acpi_output.type = CGS_ACPI_TYPE_BUFFER; + acpi_output.method_length = CGS_ACPI_MAX_BUFFER_SIZE; + acpi_output.data_length = output_size; + acpi_output.pointer = poutput; + + info.size = sizeof(struct cgs_acpi_method_info); + info.field = CGS_ACPI_FIELD_METHOD_NAME | CGS_ACPI_FIELD_INPUT_ARGUMENT_COUNT; + info.input_count = 2; + info.name = acpi_method; + info.pinput_argument = acpi_input; + info.output_count = output_count; + info.poutput_argument = &acpi_output; + + return amdgpu_cgs_acpi_eval_object(cgs_device, &info); +} + static const struct cgs_ops amdgpu_cgs_ops = { amdgpu_cgs_gpu_mem_info, amdgpu_cgs_gmap_kmem, @@ -768,7 +983,8 @@ static const struct cgs_ops amdgpu_cgs_ops = { amdgpu_cgs_set_camera_voltages, amdgpu_cgs_get_firmware_info, amdgpu_cgs_set_powergating_state, - amdgpu_cgs_set_clockgating_state + amdgpu_cgs_set_clockgating_state, + amdgpu_cgs_call_acpi_method, }; static const struct cgs_os_ops amdgpu_cgs_os_ops = { -- cgit From 5e6186991a75ea192d7dd88b9d3f7e166eaae801 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Wed, 23 Sep 2015 20:11:54 +0800 Subject: drm/amdgpu: implement cgs interface to query system info Add a query to get the bus number and function of the device. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index f901cdce7c5f..19f46d0b6ca7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -736,6 +736,28 @@ static int amdgpu_cgs_get_firmware_info(void *cgs_device, return 0; } +static int amdgpu_cgs_query_system_info(void *cgs_device, + struct cgs_system_info *sys_info) +{ + CGS_FUNC_ADEV; + + if (NULL == sys_info) + return -ENODEV; + + if (sizeof(struct cgs_system_info) != sys_info->size) + return -ENODEV; + + switch (sys_info->info_id) { + case CGS_SYSTEM_INFO_ADAPTER_BDF_ID: + sys_info->value = adev->pdev->devfn | (adev->pdev->bus->number << 8); + break; + default: + return -ENODEV; + } + + return 0; +} + /** \brief evaluate acpi namespace object, handle or pathname must be valid * \param cgs_device * \param info input/output arguments for the control method @@ -985,6 +1007,7 @@ static const struct cgs_ops amdgpu_cgs_ops = { amdgpu_cgs_set_powergating_state, amdgpu_cgs_set_clockgating_state, amdgpu_cgs_call_acpi_method, + amdgpu_cgs_query_system_info, }; static const struct cgs_os_ops amdgpu_cgs_os_ops = { -- cgit From 47bf18b5b257d5a385b7d447a29f97301f5b2282 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Thu, 17 Sep 2015 16:34:14 +0800 Subject: drm/amdgpu: add new cgs interface to get display info (v2) Add new CGS interfaces to query display info across modules. This is nedded by the powerplay module for synchronizing with the display module. v2: (agd): fold in refresh rate fix, rebase Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 40 +++++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 19f46d0b6ca7..8f758eaeef9f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -758,6 +758,45 @@ static int amdgpu_cgs_query_system_info(void *cgs_device, return 0; } +static int amdgpu_cgs_get_active_displays_info(void *cgs_device, + struct cgs_display_info *info) +{ + CGS_FUNC_ADEV; + struct amdgpu_crtc *amdgpu_crtc; + struct drm_device *ddev = adev->ddev; + struct drm_crtc *crtc; + uint32_t line_time_us, vblank_lines; + + if (info == NULL) + return -EINVAL; + + if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) { + list_for_each_entry(crtc, + &ddev->mode_config.crtc_list, head) { + amdgpu_crtc = to_amdgpu_crtc(crtc); + if (crtc->enabled) { + info->active_display_mask |= (1 << amdgpu_crtc->crtc_id); + info->display_count++; + } + if (info->mode_info != NULL && + crtc->enabled && amdgpu_crtc->enabled && + amdgpu_crtc->hw_mode.clock) { + line_time_us = (amdgpu_crtc->hw_mode.crtc_htotal * 1000) / + amdgpu_crtc->hw_mode.clock; + vblank_lines = amdgpu_crtc->hw_mode.crtc_vblank_end - + amdgpu_crtc->hw_mode.crtc_vdisplay + + (amdgpu_crtc->v_border * 2); + info->mode_info->vblank_time_us = vblank_lines * line_time_us; + info->mode_info->refresh_rate = drm_mode_vrefresh(&amdgpu_crtc->hw_mode); + info->mode_info->ref_clock = adev->clock.spll.reference_freq; + info->mode_info++; + } + } + } + + return 0; +} + /** \brief evaluate acpi namespace object, handle or pathname must be valid * \param cgs_device * \param info input/output arguments for the control method @@ -1006,6 +1045,7 @@ static const struct cgs_ops amdgpu_cgs_ops = { amdgpu_cgs_get_firmware_info, amdgpu_cgs_set_powergating_state, amdgpu_cgs_set_clockgating_state, + amdgpu_cgs_get_active_displays_info, amdgpu_cgs_call_acpi_method, amdgpu_cgs_query_system_info, }; -- cgit From 1f7371b2a5faf139465f0af386cccbb54b832534 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 2 Dec 2015 17:46:21 -0500 Subject: drm/amd/powerplay: add basic powerplay framework amdgpu_pp_ip_funcs is introduced to handle the two code paths, the legacy one and the new powerplay implementation. CONFIG_DRM_AMD_POWERPLAY kernel configuration option is introduced for the powerplay component. v4: squash in fixes v3: register debugfs file when powerplay module enable v2: add amdgpu_ucode_init_bo in hw init when amdgpu_powerplay enable. Signed-off-by: Rex Zhu Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/Makefile | 12 +- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 5 + drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 1 + drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 280 ++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h | 33 +++ drivers/gpu/drm/amd/amdgpu/cik.c | 11 +- drivers/gpu/drm/amd/amdgpu/vi.c | 7 +- 7 files changed, 340 insertions(+), 9 deletions(-) create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index a5c3aa0de773..16603a0f87e7 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -7,7 +7,8 @@ FULL_AMD_PATH=$(src)/.. ccflags-y := -Iinclude/drm -I$(FULL_AMD_PATH)/include/asic_reg \ -I$(FULL_AMD_PATH)/include \ -I$(FULL_AMD_PATH)/amdgpu \ - -I$(FULL_AMD_PATH)/scheduler + -I$(FULL_AMD_PATH)/scheduler \ + -I$(FULL_AMD_PATH)/powerplay/inc amdgpu-y := amdgpu_drv.o @@ -46,6 +47,7 @@ amdgpu-y += \ # add SMC block amdgpu-y += \ amdgpu_dpm.o \ + amdgpu_powerplay.o \ cz_smc.o cz_dpm.o \ tonga_smc.o tonga_dpm.o \ fiji_smc.o fiji_dpm.o \ @@ -96,6 +98,14 @@ amdgpu-$(CONFIG_VGA_SWITCHEROO) += amdgpu_atpx_handler.o amdgpu-$(CONFIG_ACPI) += amdgpu_acpi.o amdgpu-$(CONFIG_MMU_NOTIFIER) += amdgpu_mn.o +ifneq ($(CONFIG_DRM_AMD_POWERPLAY),) + +include drivers/gpu/drm/amd/powerplay/Makefile + +amdgpu-y += $(AMD_POWERPLAY_FILES) + +endif + obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o CFLAGS_amdgpu_trace_points.o := -I$(src) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index d454ad6ff798..6f08d39a3232 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -52,6 +52,7 @@ #include "amdgpu_irq.h" #include "amdgpu_ucode.h" #include "amdgpu_gds.h" +#include "amd_powerplay.h" #include "gpu_scheduler.h" @@ -85,6 +86,7 @@ extern int amdgpu_enable_scheduler; extern int amdgpu_sched_jobs; extern int amdgpu_sched_hw_submission; extern int amdgpu_enable_semaphores; +extern int amdgpu_powerplay; #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS 3000 #define AMDGPU_MAX_USEC_TIMEOUT 100000 /* 100 ms */ @@ -2036,6 +2038,9 @@ struct amdgpu_device { /* interrupts */ struct amdgpu_irq irq; + /* powerplay */ + struct amd_powerplay powerplay; + /* dpm */ struct amdgpu_pm pm; u32 cg_flags; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 642e305bcbcd..09248a640dbb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -82,6 +82,7 @@ int amdgpu_enable_scheduler = 1; int amdgpu_sched_jobs = 32; int amdgpu_sched_hw_submission = 2; int amdgpu_enable_semaphores = 0; +int amdgpu_powerplay = 0; MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes"); module_param_named(vramlimit, amdgpu_vram_limit, int, 0600); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c new file mode 100644 index 000000000000..5dd2a4c1a70d --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -0,0 +1,280 @@ +/* + * Copyright 2015 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ +#include "atom.h" +#include "amdgpu.h" +#include "amd_shared.h" +#include +#include +#include "amdgpu_pm.h" +#include +#include "amdgpu_powerplay.h" +#include "cik_dpm.h" +#include "vi_dpm.h" + +static int amdgpu_powerplay_init(struct amdgpu_device *adev) +{ + int ret = 0; + struct amd_powerplay *amd_pp; + + amd_pp = &(adev->powerplay); + + if (amdgpu_powerplay) { +#ifdef CONFIG_DRM_AMD_POWERPLAY + struct amd_pp_init *pp_init; + + pp_init = kzalloc(sizeof(struct amd_pp_init), GFP_KERNEL); + + if (pp_init == NULL) + return -ENOMEM; + + pp_init->chip_family = adev->family; + pp_init->chip_id = adev->asic_type; + pp_init->device = amdgpu_cgs_create_device(adev); + + ret = amd_powerplay_init(pp_init, amd_pp); + kfree(pp_init); +#endif + } else { + amd_pp->pp_handle = (void *)adev; + + switch (adev->asic_type) { +#ifdef CONFIG_DRM_AMDGPU_CIK + case CHIP_BONAIRE: + case CHIP_HAWAII: + amd_pp->ip_funcs = &ci_dpm_ip_funcs; + break; + case CHIP_KABINI: + case CHIP_MULLINS: + case CHIP_KAVERI: + amd_pp->ip_funcs = &kv_dpm_ip_funcs; + break; +#endif + case CHIP_TOPAZ: + amd_pp->ip_funcs = &iceland_dpm_ip_funcs; + break; + case CHIP_TONGA: + amd_pp->ip_funcs = &tonga_dpm_ip_funcs; + break; + case CHIP_CARRIZO: + amd_pp->ip_funcs = &cz_dpm_ip_funcs; + break; + default: + ret = -EINVAL; + break; + } + } + return ret; +} + +static int amdgpu_pp_early_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int ret = 0; + + ret = amdgpu_powerplay_init(adev); + if (ret) + return ret; + + if (adev->powerplay.ip_funcs->early_init) + ret = adev->powerplay.ip_funcs->early_init( + adev->powerplay.pp_handle); + return ret; +} + +static int amdgpu_pp_sw_init(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->sw_init) + ret = adev->powerplay.ip_funcs->sw_init( + adev->powerplay.pp_handle); + +#ifdef CONFIG_DRM_AMD_POWERPLAY + if (amdgpu_powerplay) { + adev->pm.dpm_enabled = true; + amdgpu_pm_sysfs_init(adev); + } +#endif + + return ret; +} + +static int amdgpu_pp_sw_fini(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->sw_fini) + ret = adev->powerplay.ip_funcs->sw_fini( + adev->powerplay.pp_handle); + if (ret) + return ret; + +#ifdef CONFIG_DRM_AMD_POWERPLAY + if (amdgpu_powerplay) { + amdgpu_pm_sysfs_fini(adev); + amd_powerplay_fini(adev->powerplay.pp_handle); + } +#endif + + return ret; +} + +static int amdgpu_pp_hw_init(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (amdgpu_powerplay && adev->firmware.smu_load) + amdgpu_ucode_init_bo(adev); + + if (adev->powerplay.ip_funcs->hw_init) + ret = adev->powerplay.ip_funcs->hw_init( + adev->powerplay.pp_handle); + + return ret; +} + +static int amdgpu_pp_hw_fini(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->hw_fini) + ret = adev->powerplay.ip_funcs->hw_fini( + adev->powerplay.pp_handle); + + if (amdgpu_powerplay && adev->firmware.smu_load) + amdgpu_ucode_fini_bo(adev); + + return ret; +} + +static int amdgpu_pp_suspend(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->suspend) + ret = adev->powerplay.ip_funcs->suspend( + adev->powerplay.pp_handle); + return ret; +} + +static int amdgpu_pp_resume(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->resume) + ret = adev->powerplay.ip_funcs->resume( + adev->powerplay.pp_handle); + return ret; +} + +static int amdgpu_pp_set_clockgating_state(void *handle, + enum amd_clockgating_state state) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->set_clockgating_state) + ret = adev->powerplay.ip_funcs->set_clockgating_state( + adev->powerplay.pp_handle, state); + return ret; +} + +static int amdgpu_pp_set_powergating_state(void *handle, + enum amd_powergating_state state) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->set_powergating_state) + ret = adev->powerplay.ip_funcs->set_powergating_state( + adev->powerplay.pp_handle, state); + return ret; +} + + +static bool amdgpu_pp_is_idle(void *handle) +{ + bool ret = true; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->is_idle) + ret = adev->powerplay.ip_funcs->is_idle( + adev->powerplay.pp_handle); + return ret; +} + +static int amdgpu_pp_wait_for_idle(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->wait_for_idle) + ret = adev->powerplay.ip_funcs->wait_for_idle( + adev->powerplay.pp_handle); + return ret; +} + +static int amdgpu_pp_soft_reset(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->soft_reset) + ret = adev->powerplay.ip_funcs->soft_reset( + adev->powerplay.pp_handle); + return ret; +} + +static void amdgpu_pp_print_status(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->print_status) + adev->powerplay.ip_funcs->print_status( + adev->powerplay.pp_handle); +} + +const struct amd_ip_funcs amdgpu_pp_ip_funcs = { + .early_init = amdgpu_pp_early_init, + .late_init = NULL, + .sw_init = amdgpu_pp_sw_init, + .sw_fini = amdgpu_pp_sw_fini, + .hw_init = amdgpu_pp_hw_init, + .hw_fini = amdgpu_pp_hw_fini, + .suspend = amdgpu_pp_suspend, + .resume = amdgpu_pp_resume, + .is_idle = amdgpu_pp_is_idle, + .wait_for_idle = amdgpu_pp_wait_for_idle, + .soft_reset = amdgpu_pp_soft_reset, + .print_status = amdgpu_pp_print_status, + .set_clockgating_state = amdgpu_pp_set_clockgating_state, + .set_powergating_state = amdgpu_pp_set_powergating_state, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h new file mode 100644 index 000000000000..da5cf47cfd99 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h @@ -0,0 +1,33 @@ +/* + * Copyright 2015 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __AMDGPU_POPWERPLAY_H__ +#define __AMDGPU_POPWERPLAY_H__ + +#include "amd_shared.h" + +extern const struct amd_ip_funcs amdgpu_pp_ip_funcs; + +#endif /* __AMDSOC_DM_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index 61689f094ba8..c7c298b88170 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -65,6 +65,7 @@ #include "oss/oss_2_0_sh_mask.h" #include "amdgpu_amdkfd.h" +#include "amdgpu_powerplay.h" /* * Indirect registers accessor @@ -1953,7 +1954,7 @@ static const struct amdgpu_ip_block_version bonaire_ip_blocks[] = .major = 7, .minor = 0, .rev = 0, - .funcs = &ci_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, @@ -2021,7 +2022,7 @@ static const struct amdgpu_ip_block_version hawaii_ip_blocks[] = .major = 7, .minor = 0, .rev = 0, - .funcs = &ci_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, @@ -2089,7 +2090,7 @@ static const struct amdgpu_ip_block_version kabini_ip_blocks[] = .major = 7, .minor = 0, .rev = 0, - .funcs = &kv_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, @@ -2157,7 +2158,7 @@ static const struct amdgpu_ip_block_version mullins_ip_blocks[] = .major = 7, .minor = 0, .rev = 0, - .funcs = &kv_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, @@ -2225,7 +2226,7 @@ static const struct amdgpu_ip_block_version kaveri_ip_blocks[] = .major = 7, .minor = 0, .rev = 0, - .funcs = &kv_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 30b408f1d5c0..8e4c026824d1 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -71,6 +71,7 @@ #include "uvd_v5_0.h" #include "uvd_v6_0.h" #include "vce_v3_0.h" +#include "amdgpu_powerplay.h" /* * Indirect registers accessor @@ -1130,7 +1131,7 @@ static const struct amdgpu_ip_block_version topaz_ip_blocks[] = .major = 7, .minor = 1, .rev = 0, - .funcs = &iceland_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_GFX, @@ -1177,7 +1178,7 @@ static const struct amdgpu_ip_block_version tonga_ip_blocks[] = .major = 7, .minor = 1, .rev = 0, - .funcs = &tonga_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, @@ -1313,7 +1314,7 @@ static const struct amdgpu_ip_block_version cz_ip_blocks[] = .major = 8, .minor = 0, .rev = 0, - .funcs = &cz_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs }, { .type = AMD_IP_BLOCK_TYPE_DCE, -- cgit From ba5c2a87b0c614013296901205ed693007964a59 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Fri, 6 Nov 2015 20:33:24 -0500 Subject: drm/amdgpu: disable legacy path of firmware check if powerplay is enabled Powerplay will use a different interface once it's integrated. These legacy pathes will be removed once powerplay is enabled by default. Signed-off-by: Rex Zhu Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 80 ++++++++++++++++++---------------- drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 24 +++++----- 2 files changed, 55 insertions(+), 49 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index b20369e4ea74..d90aae08445e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -2902,16 +2902,18 @@ static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev) gfx_v8_0_rlc_reset(adev); - if (!adev->firmware.smu_load) { - /* legacy rlc firmware loading */ - r = gfx_v8_0_rlc_load_microcode(adev); - if (r) - return r; - } else { - r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, - AMDGPU_UCODE_ID_RLC_G); - if (r) - return -EINVAL; + if (!amdgpu_powerplay) { + if (!adev->firmware.smu_load) { + /* legacy rlc firmware loading */ + r = gfx_v8_0_rlc_load_microcode(adev); + if (r) + return r; + } else { + r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, + AMDGPU_UCODE_ID_RLC_G); + if (r) + return -EINVAL; + } } gfx_v8_0_rlc_start(adev); @@ -3802,35 +3804,37 @@ static int gfx_v8_0_cp_resume(struct amdgpu_device *adev) if (!(adev->flags & AMD_IS_APU)) gfx_v8_0_enable_gui_idle_interrupt(adev, false); - if (!adev->firmware.smu_load) { - /* legacy firmware loading */ - r = gfx_v8_0_cp_gfx_load_microcode(adev); - if (r) - return r; - - r = gfx_v8_0_cp_compute_load_microcode(adev); - if (r) - return r; - } else { - r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, - AMDGPU_UCODE_ID_CP_CE); - if (r) - return -EINVAL; - - r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, - AMDGPU_UCODE_ID_CP_PFP); - if (r) - return -EINVAL; - - r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, - AMDGPU_UCODE_ID_CP_ME); - if (r) - return -EINVAL; + if (!amdgpu_powerplay) { + if (!adev->firmware.smu_load) { + /* legacy firmware loading */ + r = gfx_v8_0_cp_gfx_load_microcode(adev); + if (r) + return r; - r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, - AMDGPU_UCODE_ID_CP_MEC1); - if (r) - return -EINVAL; + r = gfx_v8_0_cp_compute_load_microcode(adev); + if (r) + return r; + } else { + r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, + AMDGPU_UCODE_ID_CP_CE); + if (r) + return -EINVAL; + + r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, + AMDGPU_UCODE_ID_CP_PFP); + if (r) + return -EINVAL; + + r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, + AMDGPU_UCODE_ID_CP_ME); + if (r) + return -EINVAL; + + r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, + AMDGPU_UCODE_ID_CP_MEC1); + if (r) + return -EINVAL; + } } r = gfx_v8_0_cp_gfx_resume(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 7253132f04b8..8091c1c37c4e 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -727,18 +727,20 @@ static int sdma_v3_0_start(struct amdgpu_device *adev) { int r, i; - if (!adev->firmware.smu_load) { - r = sdma_v3_0_load_microcode(adev); - if (r) - return r; - } else { - for (i = 0; i < adev->sdma.num_instances; i++) { - r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, - (i == 0) ? - AMDGPU_UCODE_ID_SDMA0 : - AMDGPU_UCODE_ID_SDMA1); + if (!amdgpu_powerplay) { + if (!adev->firmware.smu_load) { + r = sdma_v3_0_load_microcode(adev); if (r) - return -EINVAL; + return r; + } else { + for (i = 0; i < adev->sdma.num_instances; i++) { + r = adev->smu.smumgr_funcs->check_fw_load_finish(adev, + (i == 0) ? + AMDGPU_UCODE_ID_SDMA0 : + AMDGPU_UCODE_ID_SDMA1); + if (r) + return -EINVAL; + } } } -- cgit From 1b5708ffb1032a2f24b4224320753532303c1ae4 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Tue, 10 Nov 2015 18:25:24 -0500 Subject: drm/amdgpu: export amd_powerplay_func to amdgpu and other ip block Update amdgpu to deal with the new powerplay module properly. v2: squash in fixes v3: squash in Rex's power state reporting fix Signed-off-by: Rex Zhu Acked-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 46 ++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 215 +++++++++++++++++++++------------ 2 files changed, 180 insertions(+), 81 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 6f08d39a3232..d9ef4d25be79 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -2264,20 +2264,54 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) #define amdgpu_dpm_set_power_state(adev) (adev)->pm.funcs->set_power_state((adev)) #define amdgpu_dpm_post_set_power_state(adev) (adev)->pm.funcs->post_set_power_state((adev)) #define amdgpu_dpm_display_configuration_changed(adev) (adev)->pm.funcs->display_configuration_changed((adev)) -#define amdgpu_dpm_get_sclk(adev, l) (adev)->pm.funcs->get_sclk((adev), (l)) -#define amdgpu_dpm_get_mclk(adev, l) (adev)->pm.funcs->get_mclk((adev), (l)) #define amdgpu_dpm_print_power_state(adev, ps) (adev)->pm.funcs->print_power_state((adev), (ps)) -#define amdgpu_dpm_debugfs_print_current_performance_level(adev, m) (adev)->pm.funcs->debugfs_print_current_performance_level((adev), (m)) -#define amdgpu_dpm_force_performance_level(adev, l) (adev)->pm.funcs->force_performance_level((adev), (l)) #define amdgpu_dpm_vblank_too_short(adev) (adev)->pm.funcs->vblank_too_short((adev)) -#define amdgpu_dpm_powergate_uvd(adev, g) (adev)->pm.funcs->powergate_uvd((adev), (g)) -#define amdgpu_dpm_powergate_vce(adev, g) (adev)->pm.funcs->powergate_vce((adev), (g)) #define amdgpu_dpm_enable_bapm(adev, e) (adev)->pm.funcs->enable_bapm((adev), (e)) #define amdgpu_dpm_set_fan_control_mode(adev, m) (adev)->pm.funcs->set_fan_control_mode((adev), (m)) #define amdgpu_dpm_get_fan_control_mode(adev) (adev)->pm.funcs->get_fan_control_mode((adev)) #define amdgpu_dpm_set_fan_speed_percent(adev, s) (adev)->pm.funcs->set_fan_speed_percent((adev), (s)) #define amdgpu_dpm_get_fan_speed_percent(adev, s) (adev)->pm.funcs->get_fan_speed_percent((adev), (s)) +#define amdgpu_dpm_get_sclk(adev, l) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->get_sclk((adev)->powerplay.pp_handle, (l)) : \ + (adev)->pm.funcs->get_sclk((adev), (l)) + +#define amdgpu_dpm_get_mclk(adev, l) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->get_mclk((adev)->powerplay.pp_handle, (l)) : \ + (adev)->pm.funcs->get_mclk((adev), (l)) + + +#define amdgpu_dpm_force_performance_level(adev, l) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->force_performance_level((adev)->powerplay.pp_handle, (l)) : \ + (adev)->pm.funcs->force_performance_level((adev), (l)) + +#define amdgpu_dpm_powergate_uvd(adev, g) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->powergate_uvd((adev)->powerplay.pp_handle, (g)) : \ + (adev)->pm.funcs->powergate_uvd((adev), (g)) + +#define amdgpu_dpm_powergate_vce(adev, g) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->powergate_vce((adev)->powerplay.pp_handle, (g)) : \ + (adev)->pm.funcs->powergate_vce((adev), (g)) + +#define amdgpu_dpm_debugfs_print_current_performance_level(adev, m) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->print_current_performance_level((adev)->powerplay.pp_handle, (m)) : \ + (adev)->pm.funcs->debugfs_print_current_performance_level((adev), (m)) + +#define amdgpu_dpm_get_current_power_state(adev) \ + (adev)->powerplay.pp_funcs->get_current_power_state((adev)->powerplay.pp_handle) + +#define amdgpu_dpm_get_performance_level(adev) \ + (adev)->powerplay.pp_funcs->get_performance_level((adev)->powerplay.pp_handle) + +#define amdgpu_dpm_dispatch_task(adev, event_id, input, output) \ + (adev)->powerplay.pp_funcs->dispatch_tasks((adev)->powerplay.pp_handle, (event_id), (input), (output)) + #define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a)) /* Common functions */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index eea1933947ae..235fae54724b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -30,10 +30,16 @@ #include #include +#include "amd_powerplay.h" + static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev); void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev) { + if (amdgpu_powerplay) + /* TODO */ + return; + if (adev->pm.dpm_enabled) { mutex_lock(&adev->pm.mutex); if (power_supply_is_system_supplied() > 0) @@ -52,7 +58,12 @@ static ssize_t amdgpu_get_dpm_state(struct device *dev, { struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = ddev->dev_private; - enum amd_pm_state_type pm = adev->pm.dpm.user_state; + enum amd_pm_state_type pm; + + if (amdgpu_powerplay) { + pm = amdgpu_dpm_get_current_power_state(adev); + } else + pm = adev->pm.dpm.user_state; return snprintf(buf, PAGE_SIZE, "%s\n", (pm == POWER_STATE_TYPE_BATTERY) ? "battery" : @@ -66,40 +77,57 @@ static ssize_t amdgpu_set_dpm_state(struct device *dev, { struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = ddev->dev_private; + enum amd_pm_state_type state; - mutex_lock(&adev->pm.mutex); if (strncmp("battery", buf, strlen("battery")) == 0) - adev->pm.dpm.user_state = POWER_STATE_TYPE_BATTERY; + state = POWER_STATE_TYPE_BATTERY; else if (strncmp("balanced", buf, strlen("balanced")) == 0) - adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED; + state = POWER_STATE_TYPE_BALANCED; else if (strncmp("performance", buf, strlen("performance")) == 0) - adev->pm.dpm.user_state = POWER_STATE_TYPE_PERFORMANCE; + state = POWER_STATE_TYPE_PERFORMANCE; else { - mutex_unlock(&adev->pm.mutex); count = -EINVAL; goto fail; } - mutex_unlock(&adev->pm.mutex); - /* Can't set dpm state when the card is off */ - if (!(adev->flags & AMD_IS_PX) || - (ddev->switch_power_state == DRM_SWITCH_POWER_ON)) - amdgpu_pm_compute_clocks(adev); + if (amdgpu_powerplay) { + amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_ENABLE_USER_STATE, &state, NULL); + } else { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.user_state = state; + mutex_unlock(&adev->pm.mutex); + + /* Can't set dpm state when the card is off */ + if (!(adev->flags & AMD_IS_PX) || + (ddev->switch_power_state == DRM_SWITCH_POWER_ON)) + amdgpu_pm_compute_clocks(adev); + } fail: return count; } static ssize_t amdgpu_get_dpm_forced_performance_level(struct device *dev, - struct device_attribute *attr, - char *buf) + struct device_attribute *attr, + char *buf) { struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = ddev->dev_private; - enum amdgpu_dpm_forced_level level = adev->pm.dpm.forced_level; - return snprintf(buf, PAGE_SIZE, "%s\n", - (level == AMDGPU_DPM_FORCED_LEVEL_AUTO) ? "auto" : - (level == AMDGPU_DPM_FORCED_LEVEL_LOW) ? "low" : "high"); + if (amdgpu_powerplay) { + enum amd_dpm_forced_level level; + + level = amdgpu_dpm_get_performance_level(adev); + return snprintf(buf, PAGE_SIZE, "%s\n", + (level == AMD_DPM_FORCED_LEVEL_AUTO) ? "auto" : + (level == AMD_DPM_FORCED_LEVEL_LOW) ? "low" : "high"); + } else { + enum amdgpu_dpm_forced_level level; + + level = adev->pm.dpm.forced_level; + return snprintf(buf, PAGE_SIZE, "%s\n", + (level == AMDGPU_DPM_FORCED_LEVEL_AUTO) ? "auto" : + (level == AMDGPU_DPM_FORCED_LEVEL_LOW) ? "low" : "high"); + } } static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev, @@ -112,7 +140,6 @@ static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev, enum amdgpu_dpm_forced_level level; int ret = 0; - mutex_lock(&adev->pm.mutex); if (strncmp("low", buf, strlen("low")) == 0) { level = AMDGPU_DPM_FORCED_LEVEL_LOW; } else if (strncmp("high", buf, strlen("high")) == 0) { @@ -123,7 +150,11 @@ static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev, count = -EINVAL; goto fail; } - if (adev->pm.funcs->force_performance_level) { + + if (amdgpu_powerplay) + amdgpu_dpm_force_performance_level(adev, level); + else { + mutex_lock(&adev->pm.mutex); if (adev->pm.dpm.thermal_active) { count = -EINVAL; goto fail; @@ -131,6 +162,9 @@ static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev, ret = amdgpu_dpm_force_performance_level(adev, level); if (ret) count = -EINVAL; + else + adev->pm.dpm.forced_level = level; + mutex_unlock(&adev->pm.mutex); } fail: mutex_unlock(&adev->pm.mutex); @@ -197,7 +231,7 @@ static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev, int err; int value; - if(!adev->pm.funcs->set_fan_control_mode) + if (!adev->pm.funcs->set_fan_control_mode) return -EINVAL; err = kstrtoint(buf, 10, &value); @@ -294,7 +328,10 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj, struct amdgpu_device *adev = dev_get_drvdata(dev); umode_t effective_mode = attr->mode; - /* Skip attributes if DPM is not enabled */ + if (amdgpu_powerplay) + return 0; /* to do */ + + /* Skip limit attributes if DPM is not enabled */ if (!adev->pm.dpm_enabled && (attr == &sensor_dev_attr_temp1_crit.dev_attr.attr || attr == &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr || @@ -635,49 +672,54 @@ done: void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable) { - if (adev->pm.funcs->powergate_uvd) { - mutex_lock(&adev->pm.mutex); - /* enable/disable UVD */ + if (amdgpu_powerplay) amdgpu_dpm_powergate_uvd(adev, !enable); - mutex_unlock(&adev->pm.mutex); - } else { - if (enable) { + else { + if (adev->pm.funcs->powergate_uvd) { mutex_lock(&adev->pm.mutex); - adev->pm.dpm.uvd_active = true; - adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD; + /* enable/disable UVD */ + amdgpu_dpm_powergate_uvd(adev, !enable); mutex_unlock(&adev->pm.mutex); } else { - mutex_lock(&adev->pm.mutex); - adev->pm.dpm.uvd_active = false; - mutex_unlock(&adev->pm.mutex); + if (enable) { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.uvd_active = true; + adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD; + mutex_unlock(&adev->pm.mutex); + } else { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.uvd_active = false; + mutex_unlock(&adev->pm.mutex); + } + amdgpu_pm_compute_clocks(adev); } - amdgpu_pm_compute_clocks(adev); } } void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable) { - if (adev->pm.funcs->powergate_vce) { - mutex_lock(&adev->pm.mutex); - /* enable/disable VCE */ + if (amdgpu_powerplay) amdgpu_dpm_powergate_vce(adev, !enable); - - mutex_unlock(&adev->pm.mutex); - } else { - if (enable) { + else { + if (adev->pm.funcs->powergate_vce) { mutex_lock(&adev->pm.mutex); - adev->pm.dpm.vce_active = true; - /* XXX select vce level based on ring/task */ - adev->pm.dpm.vce_level = AMDGPU_VCE_LEVEL_AC_ALL; + amdgpu_dpm_powergate_vce(adev, !enable); mutex_unlock(&adev->pm.mutex); } else { - mutex_lock(&adev->pm.mutex); - adev->pm.dpm.vce_active = false; - mutex_unlock(&adev->pm.mutex); + if (enable) { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.vce_active = true; + /* XXX select vce level based on ring/task */ + adev->pm.dpm.vce_level = AMDGPU_VCE_LEVEL_AC_ALL; + mutex_unlock(&adev->pm.mutex); + } else { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.vce_active = false; + mutex_unlock(&adev->pm.mutex); + } + amdgpu_pm_compute_clocks(adev); } - - amdgpu_pm_compute_clocks(adev); } } @@ -685,10 +727,13 @@ void amdgpu_pm_print_power_states(struct amdgpu_device *adev) { int i; - for (i = 0; i < adev->pm.dpm.num_ps; i++) { - printk("== power state %d ==\n", i); + if (amdgpu_powerplay) + /* TO DO */ + return; + + for (i = 0; i < adev->pm.dpm.num_ps; i++) amdgpu_dpm_print_power_state(adev, &adev->pm.dpm.ps[i]); - } + } int amdgpu_pm_sysfs_init(struct amdgpu_device *adev) @@ -698,8 +743,11 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev) if (adev->pm.sysfs_initialized) return 0; - if (adev->pm.funcs->get_temperature == NULL) - return 0; + if (!amdgpu_powerplay) { + if (adev->pm.funcs->get_temperature == NULL) + return 0; + } + adev->pm.int_hwmon_dev = hwmon_device_register_with_groups(adev->dev, DRIVER_NAME, adev, hwmon_groups); @@ -748,32 +796,43 @@ void amdgpu_pm_compute_clocks(struct amdgpu_device *adev) if (!adev->pm.dpm_enabled) return; - mutex_lock(&adev->pm.mutex); + if (amdgpu_powerplay) { + int i = 0; + + amdgpu_display_bandwidth_update(adev); + mutex_lock(&adev->ring_lock); + for (i = 0; i < AMDGPU_MAX_RINGS; i++) { + struct amdgpu_ring *ring = adev->rings[i]; + if (ring && ring->ready) + amdgpu_fence_wait_empty(ring); + } + mutex_unlock(&adev->ring_lock); - /* update active crtc counts */ - adev->pm.dpm.new_active_crtcs = 0; - adev->pm.dpm.new_active_crtc_count = 0; - if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) { - list_for_each_entry(crtc, - &ddev->mode_config.crtc_list, head) { - amdgpu_crtc = to_amdgpu_crtc(crtc); - if (crtc->enabled) { - adev->pm.dpm.new_active_crtcs |= (1 << amdgpu_crtc->crtc_id); - adev->pm.dpm.new_active_crtc_count++; + amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_DISPLAY_CONFIG_CHANGE, NULL, NULL); + } else { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.new_active_crtcs = 0; + adev->pm.dpm.new_active_crtc_count = 0; + if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) { + list_for_each_entry(crtc, + &ddev->mode_config.crtc_list, head) { + amdgpu_crtc = to_amdgpu_crtc(crtc); + if (crtc->enabled) { + adev->pm.dpm.new_active_crtcs |= (1 << amdgpu_crtc->crtc_id); + adev->pm.dpm.new_active_crtc_count++; + } } } - } - - /* update battery/ac status */ - if (power_supply_is_system_supplied() > 0) - adev->pm.dpm.ac_power = true; - else - adev->pm.dpm.ac_power = false; - - amdgpu_dpm_change_power_state_locked(adev); + /* update battery/ac status */ + if (power_supply_is_system_supplied() > 0) + adev->pm.dpm.ac_power = true; + else + adev->pm.dpm.ac_power = false; - mutex_unlock(&adev->pm.mutex); + amdgpu_dpm_change_power_state_locked(adev); + mutex_unlock(&adev->pm.mutex); + } } /* @@ -787,7 +846,13 @@ static int amdgpu_debugfs_pm_info(struct seq_file *m, void *data) struct drm_device *dev = node->minor->dev; struct amdgpu_device *adev = dev->dev_private; - if (adev->pm.dpm_enabled) { + if (!adev->pm.dpm_enabled) { + seq_printf(m, "dpm not enabled\n"); + return 0; + } + if (amdgpu_powerplay) { + amdgpu_dpm_debugfs_print_current_performance_level(adev, m); + } else { mutex_lock(&adev->pm.mutex); if (adev->pm.funcs->debugfs_print_current_performance_level) amdgpu_dpm_debugfs_print_current_performance_level(adev, m); -- cgit From 4630f0faae80fd2252cc85accdbc8353b0444dd9 Mon Sep 17 00:00:00 2001 From: Jammy Zhou Date: Wed, 22 Jul 2015 09:54:16 +0800 Subject: drm/amd/powerplay: add Carrizo smu support This implements the SMU firmware manager interface for CZ. Some header files are moved from amdgpu folder to powerplay as well. v3: delete peci sub-module. v2: use cgs interface directly add load_mec_firmware function Signed-off-by: Rex Zhu Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h | 185 ------------------------- drivers/gpu/drm/amd/amdgpu/smu8.h | 72 ---------- drivers/gpu/drm/amd/amdgpu/smu8_fusion.h | 127 ----------------- drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h | 147 -------------------- 4 files changed, 531 deletions(-) delete mode 100644 drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/smu8.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/smu8_fusion.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h b/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h deleted file mode 100644 index 273616ab43db..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h +++ /dev/null @@ -1,185 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef CZ_PP_SMC_H -#define CZ_PP_SMC_H - -#pragma pack(push, 1) - -/* Fan control algorithm:*/ -#define FDO_MODE_HARDWARE 0 -#define FDO_MODE_PIECE_WISE_LINEAR 1 - -enum FAN_CONTROL { - FAN_CONTROL_FUZZY, - FAN_CONTROL_TABLE -}; - -enum DPM_ARRAY { - DPM_ARRAY_HARD_MAX, - DPM_ARRAY_HARD_MIN, - DPM_ARRAY_SOFT_MAX, - DPM_ARRAY_SOFT_MIN -}; - -/* - * Return codes for driver to SMC communication. - * Leave these #define-s, enums might not be exactly 8-bits on the microcontroller. - */ -#define PPSMC_Result_OK ((uint16_t)0x01) -#define PPSMC_Result_NoMore ((uint16_t)0x02) -#define PPSMC_Result_NotNow ((uint16_t)0x03) -#define PPSMC_Result_Failed ((uint16_t)0xFF) -#define PPSMC_Result_UnknownCmd ((uint16_t)0xFE) -#define PPSMC_Result_UnknownVT ((uint16_t)0xFD) - -#define PPSMC_isERROR(x) ((uint16_t)0x80 & (x)) - -/* - * Supported driver messages - */ -#define PPSMC_MSG_Test ((uint16_t) 0x1) -#define PPSMC_MSG_GetFeatureStatus ((uint16_t) 0x2) -#define PPSMC_MSG_EnableAllSmuFeatures ((uint16_t) 0x3) -#define PPSMC_MSG_DisableAllSmuFeatures ((uint16_t) 0x4) -#define PPSMC_MSG_OptimizeBattery ((uint16_t) 0x5) -#define PPSMC_MSG_MaximizePerf ((uint16_t) 0x6) -#define PPSMC_MSG_UVDPowerOFF ((uint16_t) 0x7) -#define PPSMC_MSG_UVDPowerON ((uint16_t) 0x8) -#define PPSMC_MSG_VCEPowerOFF ((uint16_t) 0x9) -#define PPSMC_MSG_VCEPowerON ((uint16_t) 0xA) -#define PPSMC_MSG_ACPPowerOFF ((uint16_t) 0xB) -#define PPSMC_MSG_ACPPowerON ((uint16_t) 0xC) -#define PPSMC_MSG_SDMAPowerOFF ((uint16_t) 0xD) -#define PPSMC_MSG_SDMAPowerON ((uint16_t) 0xE) -#define PPSMC_MSG_XDMAPowerOFF ((uint16_t) 0xF) -#define PPSMC_MSG_XDMAPowerON ((uint16_t) 0x10) -#define PPSMC_MSG_SetMinDeepSleepSclk ((uint16_t) 0x11) -#define PPSMC_MSG_SetSclkSoftMin ((uint16_t) 0x12) -#define PPSMC_MSG_SetSclkSoftMax ((uint16_t) 0x13) -#define PPSMC_MSG_SetSclkHardMin ((uint16_t) 0x14) -#define PPSMC_MSG_SetSclkHardMax ((uint16_t) 0x15) -#define PPSMC_MSG_SetLclkSoftMin ((uint16_t) 0x16) -#define PPSMC_MSG_SetLclkSoftMax ((uint16_t) 0x17) -#define PPSMC_MSG_SetLclkHardMin ((uint16_t) 0x18) -#define PPSMC_MSG_SetLclkHardMax ((uint16_t) 0x19) -#define PPSMC_MSG_SetUvdSoftMin ((uint16_t) 0x1A) -#define PPSMC_MSG_SetUvdSoftMax ((uint16_t) 0x1B) -#define PPSMC_MSG_SetUvdHardMin ((uint16_t) 0x1C) -#define PPSMC_MSG_SetUvdHardMax ((uint16_t) 0x1D) -#define PPSMC_MSG_SetEclkSoftMin ((uint16_t) 0x1E) -#define PPSMC_MSG_SetEclkSoftMax ((uint16_t) 0x1F) -#define PPSMC_MSG_SetEclkHardMin ((uint16_t) 0x20) -#define PPSMC_MSG_SetEclkHardMax ((uint16_t) 0x21) -#define PPSMC_MSG_SetAclkSoftMin ((uint16_t) 0x22) -#define PPSMC_MSG_SetAclkSoftMax ((uint16_t) 0x23) -#define PPSMC_MSG_SetAclkHardMin ((uint16_t) 0x24) -#define PPSMC_MSG_SetAclkHardMax ((uint16_t) 0x25) -#define PPSMC_MSG_SetNclkSoftMin ((uint16_t) 0x26) -#define PPSMC_MSG_SetNclkSoftMax ((uint16_t) 0x27) -#define PPSMC_MSG_SetNclkHardMin ((uint16_t) 0x28) -#define PPSMC_MSG_SetNclkHardMax ((uint16_t) 0x29) -#define PPSMC_MSG_SetPstateSoftMin ((uint16_t) 0x2A) -#define PPSMC_MSG_SetPstateSoftMax ((uint16_t) 0x2B) -#define PPSMC_MSG_SetPstateHardMin ((uint16_t) 0x2C) -#define PPSMC_MSG_SetPstateHardMax ((uint16_t) 0x2D) -#define PPSMC_MSG_DisableLowMemoryPstate ((uint16_t) 0x2E) -#define PPSMC_MSG_EnableLowMemoryPstate ((uint16_t) 0x2F) -#define PPSMC_MSG_UcodeAddressLow ((uint16_t) 0x30) -#define PPSMC_MSG_UcodeAddressHigh ((uint16_t) 0x31) -#define PPSMC_MSG_UcodeLoadStatus ((uint16_t) 0x32) -#define PPSMC_MSG_DriverDramAddrHi ((uint16_t) 0x33) -#define PPSMC_MSG_DriverDramAddrLo ((uint16_t) 0x34) -#define PPSMC_MSG_CondExecDramAddrHi ((uint16_t) 0x35) -#define PPSMC_MSG_CondExecDramAddrLo ((uint16_t) 0x36) -#define PPSMC_MSG_LoadUcodes ((uint16_t) 0x37) -#define PPSMC_MSG_DriverResetMode ((uint16_t) 0x38) -#define PPSMC_MSG_PowerStateNotify ((uint16_t) 0x39) -#define PPSMC_MSG_SetDisplayPhyConfig ((uint16_t) 0x3A) -#define PPSMC_MSG_GetMaxSclkLevel ((uint16_t) 0x3B) -#define PPSMC_MSG_GetMaxLclkLevel ((uint16_t) 0x3C) -#define PPSMC_MSG_GetMaxUvdLevel ((uint16_t) 0x3D) -#define PPSMC_MSG_GetMaxEclkLevel ((uint16_t) 0x3E) -#define PPSMC_MSG_GetMaxAclkLevel ((uint16_t) 0x3F) -#define PPSMC_MSG_GetMaxNclkLevel ((uint16_t) 0x40) -#define PPSMC_MSG_GetMaxPstate ((uint16_t) 0x41) -#define PPSMC_MSG_DramAddrHiVirtual ((uint16_t) 0x42) -#define PPSMC_MSG_DramAddrLoVirtual ((uint16_t) 0x43) -#define PPSMC_MSG_DramAddrHiPhysical ((uint16_t) 0x44) -#define PPSMC_MSG_DramAddrLoPhysical ((uint16_t) 0x45) -#define PPSMC_MSG_DramBufferSize ((uint16_t) 0x46) -#define PPSMC_MSG_SetMmPwrLogDramAddrHi ((uint16_t) 0x47) -#define PPSMC_MSG_SetMmPwrLogDramAddrLo ((uint16_t) 0x48) -#define PPSMC_MSG_SetClkTableAddrHi ((uint16_t) 0x49) -#define PPSMC_MSG_SetClkTableAddrLo ((uint16_t) 0x4A) -#define PPSMC_MSG_GetConservativePowerLimit ((uint16_t) 0x4B) - -#define PPSMC_MSG_InitJobs ((uint16_t) 0x252) -#define PPSMC_MSG_ExecuteJob ((uint16_t) 0x254) - -#define PPSMC_MSG_NBDPM_Enable ((uint16_t) 0x140) -#define PPSMC_MSG_NBDPM_Disable ((uint16_t) 0x141) - -#define PPSMC_MSG_DPM_FPS_Mode ((uint16_t) 0x15d) -#define PPSMC_MSG_DPM_Activity_Mode ((uint16_t) 0x15e) - -#define PPSMC_MSG_PmStatusLogStart ((uint16_t) 0x170) -#define PPSMC_MSG_PmStatusLogSample ((uint16_t) 0x171) - -#define PPSMC_MSG_AllowLowSclkInterrupt ((uint16_t) 0x184) -#define PPSMC_MSG_MmPowerMonitorStart ((uint16_t) 0x18F) -#define PPSMC_MSG_MmPowerMonitorStop ((uint16_t) 0x190) -#define PPSMC_MSG_MmPowerMonitorRestart ((uint16_t) 0x191) - -#define PPSMC_MSG_SetClockGateMask ((uint16_t) 0x260) -#define PPSMC_MSG_SetFpsThresholdLo ((uint16_t) 0x264) -#define PPSMC_MSG_SetFpsThresholdHi ((uint16_t) 0x265) -#define PPSMC_MSG_SetLowSclkIntrThreshold ((uint16_t) 0x266) - -#define PPSMC_MSG_ClkTableXferToDram ((uint16_t) 0x267) -#define PPSMC_MSG_ClkTableXferToSmu ((uint16_t) 0x268) -#define PPSMC_MSG_GetAverageGraphicsActivity ((uint16_t) 0x269) -#define PPSMC_MSG_GetAverageGioActivity ((uint16_t) 0x26A) -#define PPSMC_MSG_SetLoggerBufferSize ((uint16_t) 0x26B) -#define PPSMC_MSG_SetLoggerAddressHigh ((uint16_t) 0x26C) -#define PPSMC_MSG_SetLoggerAddressLow ((uint16_t) 0x26D) -#define PPSMC_MSG_SetWatermarkFrequency ((uint16_t) 0x26E) - -/* REMOVE LATER*/ -#define PPSMC_MSG_DPM_ForceState ((uint16_t) 0x104) - -/* Feature Enable Masks*/ -#define NB_DPM_MASK 0x00000800 -#define VDDGFX_MASK 0x00800000 -#define VCE_DPM_MASK 0x00400000 -#define ACP_DPM_MASK 0x00040000 -#define UVD_DPM_MASK 0x00010000 -#define GFX_CU_PG_MASK 0x00004000 -#define SCLK_DPM_MASK 0x00080000 - -#if !defined(SMC_MICROCODE) -#pragma pack(pop) - -#endif - -#endif diff --git a/drivers/gpu/drm/amd/amdgpu/smu8.h b/drivers/gpu/drm/amd/amdgpu/smu8.h deleted file mode 100644 index d758d07b6a31..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/smu8.h +++ /dev/null @@ -1,72 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef SMU8_H -#define SMU8_H - -#pragma pack(push, 1) - -#define ENABLE_DEBUG_FEATURES - -struct SMU8_Firmware_Header { - uint32_t Version; - uint32_t ImageSize; - uint32_t CodeSize; - uint32_t HeaderSize; - uint32_t EntryPoint; - uint32_t Rtos; - uint32_t UcodeLoadStatus; - uint32_t DpmTable; - uint32_t FanTable; - uint32_t PmFuseTable; - uint32_t Globals; - uint32_t Reserved[20]; - uint32_t Signature; -}; - -struct SMU8_MultimediaPowerLogData { - uint32_t avgTotalPower; - uint32_t avgGpuPower; - uint32_t avgUvdPower; - uint32_t avgVcePower; - - uint32_t avgSclk; - uint32_t avgDclk; - uint32_t avgVclk; - uint32_t avgEclk; - - uint32_t startTimeHi; - uint32_t startTimeLo; - - uint32_t endTimeHi; - uint32_t endTimeLo; -}; - -#define SMU8_FIRMWARE_HEADER_LOCATION 0x1FF80 -#define SMU8_UNBCSR_START_ADDR 0xC0100000 - -#define SMN_MP1_SRAM_START_ADDR 0x10000000 - -#pragma pack(pop) - -#endif diff --git a/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h b/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h deleted file mode 100644 index 5c9cc3c0bbfa..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h +++ /dev/null @@ -1,127 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef SMU8_FUSION_H -#define SMU8_FUSION_H - -#include "smu8.h" - -#pragma pack(push, 1) - -#define SMU8_MAX_CUS 2 -#define SMU8_PSMS_PER_CU 4 -#define SMU8_CACS_PER_CU 4 - -struct SMU8_GfxCuPgScoreboard { - uint8_t Enabled; - uint8_t spare[3]; -}; - -struct SMU8_Port80MonitorTable { - uint32_t MmioAddress; - uint32_t MemoryBaseHi; - uint32_t MemoryBaseLo; - uint16_t MemoryBufferSize; - uint16_t MemoryPosition; - uint16_t PollingInterval; - uint8_t EnableCsrShadow; - uint8_t EnableDramShadow; -}; - -/* Clock Table Definitions */ -#define NUM_SCLK_LEVELS 8 -#define NUM_LCLK_LEVELS 8 -#define NUM_UVD_LEVELS 8 -#define NUM_ECLK_LEVELS 8 -#define NUM_ACLK_LEVELS 8 - -struct SMU8_Fusion_ClkLevel { - uint8_t GnbVid; - uint8_t GfxVid; - uint8_t DfsDid; - uint8_t DeepSleepDid; - uint32_t DfsBypass; - uint32_t Frequency; -}; - -struct SMU8_Fusion_SclkBreakdownTable { - struct SMU8_Fusion_ClkLevel ClkLevel[NUM_SCLK_LEVELS]; - struct SMU8_Fusion_ClkLevel DpmOffLevel; - /* SMU8_Fusion_ClkLevel PwrOffLevel; */ - uint32_t SclkValidMask; - uint32_t MaxSclkIndex; -}; - -struct SMU8_Fusion_LclkBreakdownTable { - struct SMU8_Fusion_ClkLevel ClkLevel[NUM_LCLK_LEVELS]; - struct SMU8_Fusion_ClkLevel DpmOffLevel; - /* SMU8_Fusion_ClkLevel PwrOffLevel; */ - uint32_t LclkValidMask; - uint32_t MaxLclkIndex; -}; - -struct SMU8_Fusion_EclkBreakdownTable { - struct SMU8_Fusion_ClkLevel ClkLevel[NUM_ECLK_LEVELS]; - struct SMU8_Fusion_ClkLevel DpmOffLevel; - struct SMU8_Fusion_ClkLevel PwrOffLevel; - uint32_t EclkValidMask; - uint32_t MaxEclkIndex; -}; - -struct SMU8_Fusion_VclkBreakdownTable { - struct SMU8_Fusion_ClkLevel ClkLevel[NUM_UVD_LEVELS]; - struct SMU8_Fusion_ClkLevel DpmOffLevel; - struct SMU8_Fusion_ClkLevel PwrOffLevel; - uint32_t VclkValidMask; - uint32_t MaxVclkIndex; -}; - -struct SMU8_Fusion_DclkBreakdownTable { - struct SMU8_Fusion_ClkLevel ClkLevel[NUM_UVD_LEVELS]; - struct SMU8_Fusion_ClkLevel DpmOffLevel; - struct SMU8_Fusion_ClkLevel PwrOffLevel; - uint32_t DclkValidMask; - uint32_t MaxDclkIndex; -}; - -struct SMU8_Fusion_AclkBreakdownTable { - struct SMU8_Fusion_ClkLevel ClkLevel[NUM_ACLK_LEVELS]; - struct SMU8_Fusion_ClkLevel DpmOffLevel; - struct SMU8_Fusion_ClkLevel PwrOffLevel; - uint32_t AclkValidMask; - uint32_t MaxAclkIndex; -}; - - -struct SMU8_Fusion_ClkTable { - struct SMU8_Fusion_SclkBreakdownTable SclkBreakdownTable; - struct SMU8_Fusion_LclkBreakdownTable LclkBreakdownTable; - struct SMU8_Fusion_EclkBreakdownTable EclkBreakdownTable; - struct SMU8_Fusion_VclkBreakdownTable VclkBreakdownTable; - struct SMU8_Fusion_DclkBreakdownTable DclkBreakdownTable; - struct SMU8_Fusion_AclkBreakdownTable AclkBreakdownTable; -}; - -#pragma pack(pop) - -#endif diff --git a/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h b/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h deleted file mode 100644 index f8ba071f39c8..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h +++ /dev/null @@ -1,147 +0,0 @@ -// CZ Ucode Loading Definitions -#ifndef SMU_UCODE_XFER_CZ_H -#define SMU_UCODE_XFER_CZ_H - -#define NUM_JOBLIST_ENTRIES 32 - -#define TASK_TYPE_NO_ACTION 0 -#define TASK_TYPE_UCODE_LOAD 1 -#define TASK_TYPE_UCODE_SAVE 2 -#define TASK_TYPE_REG_LOAD 3 -#define TASK_TYPE_REG_SAVE 4 -#define TASK_TYPE_INITIALIZE 5 - -#define TASK_ARG_REG_SMCIND 0 -#define TASK_ARG_REG_MMIO 1 -#define TASK_ARG_REG_FCH 2 -#define TASK_ARG_REG_UNB 3 - -#define TASK_ARG_INIT_MM_PWR_LOG 0 -#define TASK_ARG_INIT_CLK_TABLE 1 - -#define JOB_GFX_SAVE 0 -#define JOB_GFX_RESTORE 1 -#define JOB_FCH_SAVE 2 -#define JOB_FCH_RESTORE 3 -#define JOB_UNB_SAVE 4 -#define JOB_UNB_RESTORE 5 -#define JOB_GMC_SAVE 6 -#define JOB_GMC_RESTORE 7 -#define JOB_GNB_SAVE 8 -#define JOB_GNB_RESTORE 9 - -#define IGNORE_JOB 0xff -#define END_OF_TASK_LIST (uint16_t)0xffff - -// Size of DRAM regions (in bytes) requested by SMU: -#define SMU_DRAM_REQ_MM_PWR_LOG 48 - -#define UCODE_ID_SDMA0 0 -#define UCODE_ID_SDMA1 1 -#define UCODE_ID_CP_CE 2 -#define UCODE_ID_CP_PFP 3 -#define UCODE_ID_CP_ME 4 -#define UCODE_ID_CP_MEC_JT1 5 -#define UCODE_ID_CP_MEC_JT2 6 -#define UCODE_ID_GMCON_RENG 7 -#define UCODE_ID_RLC_G 8 -#define UCODE_ID_RLC_SCRATCH 9 -#define UCODE_ID_RLC_SRM_ARAM 10 -#define UCODE_ID_RLC_SRM_DRAM 11 -#define UCODE_ID_DMCU_ERAM 12 -#define UCODE_ID_DMCU_IRAM 13 - -#define UCODE_ID_SDMA0_MASK 0x00000001 -#define UCODE_ID_SDMA1_MASK 0x00000002 -#define UCODE_ID_CP_CE_MASK 0x00000004 -#define UCODE_ID_CP_PFP_MASK 0x00000008 -#define UCODE_ID_CP_ME_MASK 0x00000010 -#define UCODE_ID_CP_MEC_JT1_MASK 0x00000020 -#define UCODE_ID_CP_MEC_JT2_MASK 0x00000040 -#define UCODE_ID_GMCON_RENG_MASK 0x00000080 -#define UCODE_ID_RLC_G_MASK 0x00000100 -#define UCODE_ID_RLC_SCRATCH_MASK 0x00000200 -#define UCODE_ID_RLC_SRM_ARAM_MASK 0x00000400 -#define UCODE_ID_RLC_SRM_DRAM_MASK 0x00000800 -#define UCODE_ID_DMCU_ERAM_MASK 0x00001000 -#define UCODE_ID_DMCU_IRAM_MASK 0x00002000 - -#define UCODE_ID_SDMA0_SIZE_BYTE 10368 -#define UCODE_ID_SDMA1_SIZE_BYTE 10368 -#define UCODE_ID_CP_CE_SIZE_BYTE 8576 -#define UCODE_ID_CP_PFP_SIZE_BYTE 16768 -#define UCODE_ID_CP_ME_SIZE_BYTE 16768 -#define UCODE_ID_CP_MEC_JT1_SIZE_BYTE 384 -#define UCODE_ID_CP_MEC_JT2_SIZE_BYTE 384 -#define UCODE_ID_GMCON_RENG_SIZE_BYTE 4096 -#define UCODE_ID_RLC_G_SIZE_BYTE 2048 -#define UCODE_ID_RLC_SCRATCH_SIZE_BYTE 132 -#define UCODE_ID_RLC_SRM_ARAM_SIZE_BYTE 8192 -#define UCODE_ID_RLC_SRM_DRAM_SIZE_BYTE 4096 -#define UCODE_ID_DMCU_ERAM_SIZE_BYTE 24576 -#define UCODE_ID_DMCU_IRAM_SIZE_BYTE 1024 - -#define NUM_UCODES 14 - -typedef struct { - uint32_t high; - uint32_t low; -} data_64_t; - -struct SMU_Task { - uint8_t type; - uint8_t arg; - uint16_t next; - data_64_t addr; - uint32_t size_bytes; -}; -typedef struct SMU_Task SMU_Task; - -struct TOC { - uint8_t JobList[NUM_JOBLIST_ENTRIES]; - SMU_Task tasks[1]; -}; - -// META DATA COMMAND Definitions -#define METADATA_CMD_MODE0 0x00000103 -#define METADATA_CMD_MODE1 0x00000113 -#define METADATA_CMD_MODE2 0x00000123 -#define METADATA_CMD_MODE3 0x00000133 -#define METADATA_CMD_DELAY 0x00000203 -#define METADATA_CMD_CHNG_REGSPACE 0x00000303 -#define METADATA_PERFORM_ON_SAVE 0x00001000 -#define METADATA_PERFORM_ON_LOAD 0x00002000 -#define METADATA_CMD_ARG_MASK 0xFFFF0000 -#define METADATA_CMD_ARG_SHIFT 16 - -// Simple register addr/data fields -struct SMU_MetaData_Mode0 { - uint32_t register_address; - uint32_t register_data; -}; -typedef struct SMU_MetaData_Mode0 SMU_MetaData_Mode0; - -// Register addr/data with mask -struct SMU_MetaData_Mode1 { - uint32_t register_address; - uint32_t register_mask; - uint32_t register_data; -}; -typedef struct SMU_MetaData_Mode1 SMU_MetaData_Mode1; - -struct SMU_MetaData_Mode2 { - uint32_t register_address; - uint32_t register_mask; - uint32_t target_value; -}; -typedef struct SMU_MetaData_Mode2 SMU_MetaData_Mode2; - -// Always write data (even on a save operation) -struct SMU_MetaData_Mode3 { - uint32_t register_address; - uint32_t register_mask; - uint32_t register_data; -}; -typedef struct SMU_MetaData_Mode3 SMU_MetaData_Mode3; - -#endif -- cgit From 7ff1d70a40f468e20a6cae5311800cd18e680865 Mon Sep 17 00:00:00 2001 From: yanyang1 Date: Wed, 19 Aug 2015 12:22:34 +0800 Subject: drm/amd/powerplay: Move smu7*.h from amdgpu to powerplay. Move smu7.h, smu7_discrete.h and smu7_fusion.h from amdgpu to powerplay. Reviewed-by: Alex Deucher Signed-off-by: yanyang1 --- drivers/gpu/drm/amd/amdgpu/smu7.h | 170 ---------- drivers/gpu/drm/amd/amdgpu/smu7_discrete.h | 514 ----------------------------- drivers/gpu/drm/amd/amdgpu/smu7_fusion.h | 300 ----------------- 3 files changed, 984 deletions(-) delete mode 100644 drivers/gpu/drm/amd/amdgpu/smu7.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/smu7_discrete.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/smu7_fusion.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/smu7.h b/drivers/gpu/drm/amd/amdgpu/smu7.h deleted file mode 100644 index 75a380a15292..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/smu7.h +++ /dev/null @@ -1,170 +0,0 @@ -/* - * Copyright 2013 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef SMU7_H -#define SMU7_H - -#pragma pack(push, 1) - -#define SMU7_CONTEXT_ID_SMC 1 -#define SMU7_CONTEXT_ID_VBIOS 2 - - -#define SMU7_CONTEXT_ID_SMC 1 -#define SMU7_CONTEXT_ID_VBIOS 2 - -#define SMU7_MAX_LEVELS_VDDC 8 -#define SMU7_MAX_LEVELS_VDDCI 4 -#define SMU7_MAX_LEVELS_MVDD 4 -#define SMU7_MAX_LEVELS_VDDNB 8 - -#define SMU7_MAX_LEVELS_GRAPHICS SMU__NUM_SCLK_DPM_STATE // SCLK + SQ DPM + ULV -#define SMU7_MAX_LEVELS_MEMORY SMU__NUM_MCLK_DPM_LEVELS // MCLK Levels DPM -#define SMU7_MAX_LEVELS_GIO SMU__NUM_LCLK_DPM_LEVELS // LCLK Levels -#define SMU7_MAX_LEVELS_LINK SMU__NUM_PCIE_DPM_LEVELS // PCIe speed and number of lanes. -#define SMU7_MAX_LEVELS_UVD 8 // VCLK/DCLK levels for UVD. -#define SMU7_MAX_LEVELS_VCE 8 // ECLK levels for VCE. -#define SMU7_MAX_LEVELS_ACP 8 // ACLK levels for ACP. -#define SMU7_MAX_LEVELS_SAMU 8 // SAMCLK levels for SAMU. -#define SMU7_MAX_ENTRIES_SMIO 32 // Number of entries in SMIO table. - -#define DPM_NO_LIMIT 0 -#define DPM_NO_UP 1 -#define DPM_GO_DOWN 2 -#define DPM_GO_UP 3 - -#define SMU7_FIRST_DPM_GRAPHICS_LEVEL 0 -#define SMU7_FIRST_DPM_MEMORY_LEVEL 0 - -#define GPIO_CLAMP_MODE_VRHOT 1 -#define GPIO_CLAMP_MODE_THERM 2 -#define GPIO_CLAMP_MODE_DC 4 - -#define SCRATCH_B_TARG_PCIE_INDEX_SHIFT 0 -#define SCRATCH_B_TARG_PCIE_INDEX_MASK (0x7< Date: Mon, 17 Aug 2015 14:15:20 +0800 Subject: drm/amd/powerplay: add header file for tonga smu and dpm These headers provide the SMU interface used by the driver. Reviewed-by: Alex Deucher Signed-off-by: yanyang1 --- drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h | 198 ------------------------------- 1 file changed, 198 deletions(-) delete mode 100644 drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h b/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h deleted file mode 100644 index 811781f69482..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h +++ /dev/null @@ -1,198 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef TONGA_PP_SMC_H -#define TONGA_PP_SMC_H - -#pragma pack(push, 1) - -#define PPSMC_SWSTATE_FLAG_DC 0x01 -#define PPSMC_SWSTATE_FLAG_UVD 0x02 -#define PPSMC_SWSTATE_FLAG_VCE 0x04 -#define PPSMC_SWSTATE_FLAG_PCIE_X1 0x08 - -#define PPSMC_THERMAL_PROTECT_TYPE_INTERNAL 0x00 -#define PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL 0x01 -#define PPSMC_THERMAL_PROTECT_TYPE_NONE 0xff - -#define PPSMC_SYSTEMFLAG_GPIO_DC 0x01 -#define PPSMC_SYSTEMFLAG_STEPVDDC 0x02 -#define PPSMC_SYSTEMFLAG_GDDR5 0x04 - -#define PPSMC_SYSTEMFLAG_DISABLE_BABYSTEP 0x08 - -#define PPSMC_SYSTEMFLAG_REGULATOR_HOT 0x10 -#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_ANALOG 0x20 -#define PPSMC_SYSTEMFLAG_12CHANNEL 0x40 - -#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_MASK 0x07 -#define PPSMC_EXTRAFLAGS_AC2DC_DONT_WAIT_FOR_VBLANK 0x08 - -#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTODPMLOWSTATE 0x00 -#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTOINITIALSTATE 0x01 - -#define PPSMC_EXTRAFLAGS_AC2DC_GPIO5_POLARITY_HIGH 0x10 -#define PPSMC_EXTRAFLAGS_DRIVER_TO_GPIO17 0x20 -#define PPSMC_EXTRAFLAGS_PCC_TO_GPIO17 0x40 - -#define PPSMC_DPM2FLAGS_TDPCLMP 0x01 -#define PPSMC_DPM2FLAGS_PWRSHFT 0x02 -#define PPSMC_DPM2FLAGS_OCP 0x04 - -#define PPSMC_DISPLAY_WATERMARK_LOW 0 -#define PPSMC_DISPLAY_WATERMARK_HIGH 1 - -#define PPSMC_STATEFLAG_AUTO_PULSE_SKIP 0x01 -#define PPSMC_STATEFLAG_POWERBOOST 0x02 -#define PPSMC_STATEFLAG_PSKIP_ON_TDP_FAULT 0x04 -#define PPSMC_STATEFLAG_POWERSHIFT 0x08 -#define PPSMC_STATEFLAG_SLOW_READ_MARGIN 0x10 -#define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 0x20 -#define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS 0x40 - -#define FDO_MODE_HARDWARE 0 -#define FDO_MODE_PIECE_WISE_LINEAR 1 - -enum FAN_CONTROL { - FAN_CONTROL_FUZZY, - FAN_CONTROL_TABLE -}; - -#define PPSMC_Result_OK ((uint16_t)0x01) -#define PPSMC_Result_NoMore ((uint16_t)0x02) -#define PPSMC_Result_NotNow ((uint16_t)0x03) -#define PPSMC_Result_Failed ((uint16_t)0xFF) -#define PPSMC_Result_UnknownCmd ((uint16_t)0xFE) -#define PPSMC_Result_UnknownVT ((uint16_t)0xFD) - -typedef uint16_t PPSMC_Result; - -#define PPSMC_isERROR(x) ((uint16_t)0x80 & (x)) - -#define PPSMC_MSG_Halt ((uint16_t)0x10) -#define PPSMC_MSG_Resume ((uint16_t)0x11) -#define PPSMC_MSG_EnableDPMLevel ((uint16_t)0x12) -#define PPSMC_MSG_ZeroLevelsDisabled ((uint16_t)0x13) -#define PPSMC_MSG_OneLevelsDisabled ((uint16_t)0x14) -#define PPSMC_MSG_TwoLevelsDisabled ((uint16_t)0x15) -#define PPSMC_MSG_EnableThermalInterrupt ((uint16_t)0x16) -#define PPSMC_MSG_RunningOnAC ((uint16_t)0x17) -#define PPSMC_MSG_LevelUp ((uint16_t)0x18) -#define PPSMC_MSG_LevelDown ((uint16_t)0x19) -#define PPSMC_MSG_ResetDPMCounters ((uint16_t)0x1a) -#define PPSMC_MSG_SwitchToSwState ((uint16_t)0x20) -#define PPSMC_MSG_SwitchToSwStateLast ((uint16_t)0x3f) -#define PPSMC_MSG_SwitchToInitialState ((uint16_t)0x40) -#define PPSMC_MSG_NoForcedLevel ((uint16_t)0x41) -#define PPSMC_MSG_ForceHigh ((uint16_t)0x42) -#define PPSMC_MSG_ForceMediumOrHigh ((uint16_t)0x43) -#define PPSMC_MSG_SwitchToMinimumPower ((uint16_t)0x51) -#define PPSMC_MSG_ResumeFromMinimumPower ((uint16_t)0x52) -#define PPSMC_MSG_EnableCac ((uint16_t)0x53) -#define PPSMC_MSG_DisableCac ((uint16_t)0x54) -#define PPSMC_DPMStateHistoryStart ((uint16_t)0x55) -#define PPSMC_DPMStateHistoryStop ((uint16_t)0x56) -#define PPSMC_CACHistoryStart ((uint16_t)0x57) -#define PPSMC_CACHistoryStop ((uint16_t)0x58) -#define PPSMC_TDPClampingActive ((uint16_t)0x59) -#define PPSMC_TDPClampingInactive ((uint16_t)0x5A) -#define PPSMC_StartFanControl ((uint16_t)0x5B) -#define PPSMC_StopFanControl ((uint16_t)0x5C) -#define PPSMC_NoDisplay ((uint16_t)0x5D) -#define PPSMC_HasDisplay ((uint16_t)0x5E) -#define PPSMC_MSG_UVDPowerOFF ((uint16_t)0x60) -#define PPSMC_MSG_UVDPowerON ((uint16_t)0x61) -#define PPSMC_MSG_EnableULV ((uint16_t)0x62) -#define PPSMC_MSG_DisableULV ((uint16_t)0x63) -#define PPSMC_MSG_EnterULV ((uint16_t)0x64) -#define PPSMC_MSG_ExitULV ((uint16_t)0x65) -#define PPSMC_PowerShiftActive ((uint16_t)0x6A) -#define PPSMC_PowerShiftInactive ((uint16_t)0x6B) -#define PPSMC_OCPActive ((uint16_t)0x6C) -#define PPSMC_OCPInactive ((uint16_t)0x6D) -#define PPSMC_CACLongTermAvgEnable ((uint16_t)0x6E) -#define PPSMC_CACLongTermAvgDisable ((uint16_t)0x6F) -#define PPSMC_MSG_InferredStateSweep_Start ((uint16_t)0x70) -#define PPSMC_MSG_InferredStateSweep_Stop ((uint16_t)0x71) -#define PPSMC_MSG_SwitchToLowestInfState ((uint16_t)0x72) -#define PPSMC_MSG_SwitchToNonInfState ((uint16_t)0x73) -#define PPSMC_MSG_AllStateSweep_Start ((uint16_t)0x74) -#define PPSMC_MSG_AllStateSweep_Stop ((uint16_t)0x75) -#define PPSMC_MSG_SwitchNextLowerInfState ((uint16_t)0x76) -#define PPSMC_MSG_SwitchNextHigherInfState ((uint16_t)0x77) -#define PPSMC_MSG_MclkRetrainingTest ((uint16_t)0x78) -#define PPSMC_MSG_ForceTDPClamping ((uint16_t)0x79) -#define PPSMC_MSG_CollectCAC_PowerCorreln ((uint16_t)0x7A) -#define PPSMC_MSG_CollectCAC_WeightCalib ((uint16_t)0x7B) -#define PPSMC_MSG_CollectCAC_SQonly ((uint16_t)0x7C) -#define PPSMC_MSG_CollectCAC_TemperaturePwr ((uint16_t)0x7D) -#define PPSMC_MSG_ExtremitiesTest_Start ((uint16_t)0x7E) -#define PPSMC_MSG_ExtremitiesTest_Stop ((uint16_t)0x7F) -#define PPSMC_FlushDataCache ((uint16_t)0x80) -#define PPSMC_FlushInstrCache ((uint16_t)0x81) -#define PPSMC_MSG_SetEnabledLevels ((uint16_t)0x82) -#define PPSMC_MSG_SetForcedLevels ((uint16_t)0x83) -#define PPSMC_MSG_ResetToDefaults ((uint16_t)0x84) -#define PPSMC_MSG_SetForcedLevelsAndJump ((uint16_t)0x85) -#define PPSMC_MSG_SetCACHistoryMode ((uint16_t)0x86) -#define PPSMC_MSG_EnableDTE ((uint16_t)0x87) -#define PPSMC_MSG_DisableDTE ((uint16_t)0x88) -#define PPSMC_MSG_SmcSpaceSetAddress ((uint16_t)0x89) -#define PPSMC_MSG_SmcSpaceWriteDWordInc ((uint16_t)0x8A) -#define PPSMC_MSG_SmcSpaceWriteWordInc ((uint16_t)0x8B) -#define PPSMC_MSG_SmcSpaceWriteByteInc ((uint16_t)0x8C) -#define PPSMC_MSG_ChangeNearTDPLimit ((uint16_t)0x90) -#define PPSMC_MSG_ChangeSafePowerLimit ((uint16_t)0x91) -#define PPSMC_MSG_DPMStateSweepStart ((uint16_t)0x92) -#define PPSMC_MSG_DPMStateSweepStop ((uint16_t)0x93) -#define PPSMC_MSG_OVRDDisableSCLKDS ((uint16_t)0x94) -#define PPSMC_MSG_CancelDisableOVRDSCLKDS ((uint16_t)0x95) -#define PPSMC_MSG_ThrottleOVRDSCLKDS ((uint16_t)0x96) -#define PPSMC_MSG_CancelThrottleOVRDSCLKDS ((uint16_t)0x97) -#define PPSMC_MSG_GPIO17 ((uint16_t)0x98) -#define PPSMC_MSG_API_SetSvi2Volt_Vddc ((uint16_t)0x99) -#define PPSMC_MSG_API_SetSvi2Volt_Vddci ((uint16_t)0x9A) -#define PPSMC_MSG_API_SetSvi2Volt_Mvdd ((uint16_t)0x9B) -#define PPSMC_MSG_API_GetSvi2Volt_Vddc ((uint16_t)0x9C) -#define PPSMC_MSG_API_GetSvi2Volt_Vddci ((uint16_t)0x9D) -#define PPSMC_MSG_API_GetSvi2Volt_Mvdd ((uint16_t)0x9E) - -#define PPSMC_MSG_BREAK ((uint16_t)0xF8) - -#define PPSMC_MSG_Test ((uint16_t)0x100) -#define PPSMC_MSG_DRV_DRAM_ADDR_HI ((uint16_t)0x250) -#define PPSMC_MSG_DRV_DRAM_ADDR_LO ((uint16_t)0x251) -#define PPSMC_MSG_SMU_DRAM_ADDR_HI ((uint16_t)0x252) -#define PPSMC_MSG_SMU_DRAM_ADDR_LO ((uint16_t)0x253) -#define PPSMC_MSG_LoadUcodes ((uint16_t)0x254) - -typedef uint16_t PPSMC_Msg; - -#define PPSMC_EVENT_STATUS_THERMAL 0x00000001 -#define PPSMC_EVENT_STATUS_REGULATORHOT 0x00000002 -#define PPSMC_EVENT_STATUS_DC 0x00000004 -#define PPSMC_EVENT_STATUS_GPIO17 0x00000008 - -#pragma pack(pop) - -#endif -- cgit From 770911a3cfbb43b67b5ea3189b624e4fe2cb27c1 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Mon, 9 Nov 2015 17:34:31 -0500 Subject: drm/amd/powerplay: add/update headers for Fiji SMU and DPM New headers for Fiji. Reviewed-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h | 182 -------------------------------- 1 file changed, 182 deletions(-) delete mode 100644 drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h b/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h deleted file mode 100644 index 3c4824082990..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h +++ /dev/null @@ -1,182 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef FIJI_PP_SMC_H -#define FIJI_PP_SMC_H - -#pragma pack(push, 1) - -#define PPSMC_SWSTATE_FLAG_DC 0x01 -#define PPSMC_SWSTATE_FLAG_UVD 0x02 -#define PPSMC_SWSTATE_FLAG_VCE 0x04 - -#define PPSMC_THERMAL_PROTECT_TYPE_INTERNAL 0x00 -#define PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL 0x01 -#define PPSMC_THERMAL_PROTECT_TYPE_NONE 0xff - -#define PPSMC_SYSTEMFLAG_GPIO_DC 0x01 -#define PPSMC_SYSTEMFLAG_STEPVDDC 0x02 -#define PPSMC_SYSTEMFLAG_GDDR5 0x04 - -#define PPSMC_SYSTEMFLAG_DISABLE_BABYSTEP 0x08 - -#define PPSMC_SYSTEMFLAG_REGULATOR_HOT 0x10 -#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_ANALOG 0x20 - -#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_MASK 0x07 -#define PPSMC_EXTRAFLAGS_AC2DC_DONT_WAIT_FOR_VBLANK 0x08 - -#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTODPMLOWSTATE 0x00 -#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTOINITIALSTATE 0x01 - -#define PPSMC_DPM2FLAGS_TDPCLMP 0x01 -#define PPSMC_DPM2FLAGS_PWRSHFT 0x02 -#define PPSMC_DPM2FLAGS_OCP 0x04 - -#define PPSMC_DISPLAY_WATERMARK_LOW 0 -#define PPSMC_DISPLAY_WATERMARK_HIGH 1 - -#define PPSMC_STATEFLAG_AUTO_PULSE_SKIP 0x01 -#define PPSMC_STATEFLAG_POWERBOOST 0x02 -#define PPSMC_STATEFLAG_PSKIP_ON_TDP_FAULT 0x04 -#define PPSMC_STATEFLAG_POWERSHIFT 0x08 -#define PPSMC_STATEFLAG_SLOW_READ_MARGIN 0x10 -#define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 0x20 -#define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS 0x40 - -#define FDO_MODE_HARDWARE 0 -#define FDO_MODE_PIECE_WISE_LINEAR 1 - -enum FAN_CONTROL { - FAN_CONTROL_FUZZY, - FAN_CONTROL_TABLE -}; - -//Gemini Modes -#define PPSMC_GeminiModeNone 0 //Single GPU board -#define PPSMC_GeminiModeMaster 1 //Master GPU on a Gemini board -#define PPSMC_GeminiModeSlave 2 //Slave GPU on a Gemini board - -#define PPSMC_Result_OK ((uint16_t)0x01) -#define PPSMC_Result_NoMore ((uint16_t)0x02) -#define PPSMC_Result_NotNow ((uint16_t)0x03) -#define PPSMC_Result_Failed ((uint16_t)0xFF) -#define PPSMC_Result_UnknownCmd ((uint16_t)0xFE) -#define PPSMC_Result_UnknownVT ((uint16_t)0xFD) - -typedef uint16_t PPSMC_Result; - -#define PPSMC_isERROR(x) ((uint16_t)0x80 & (x)) - -#define PPSMC_MSG_Halt ((uint16_t)0x10) -#define PPSMC_MSG_Resume ((uint16_t)0x11) -#define PPSMC_MSG_EnableDPMLevel ((uint16_t)0x12) -#define PPSMC_MSG_ZeroLevelsDisabled ((uint16_t)0x13) -#define PPSMC_MSG_OneLevelsDisabled ((uint16_t)0x14) -#define PPSMC_MSG_TwoLevelsDisabled ((uint16_t)0x15) -#define PPSMC_MSG_EnableThermalInterrupt ((uint16_t)0x16) -#define PPSMC_MSG_RunningOnAC ((uint16_t)0x17) -#define PPSMC_MSG_LevelUp ((uint16_t)0x18) -#define PPSMC_MSG_LevelDown ((uint16_t)0x19) -#define PPSMC_MSG_ResetDPMCounters ((uint16_t)0x1a) -#define PPSMC_MSG_SwitchToSwState ((uint16_t)0x20) -#define PPSMC_MSG_SwitchToSwStateLast ((uint16_t)0x3f) -#define PPSMC_MSG_SwitchToInitialState ((uint16_t)0x40) -#define PPSMC_MSG_NoForcedLevel ((uint16_t)0x41) -#define PPSMC_MSG_ForceHigh ((uint16_t)0x42) -#define PPSMC_MSG_ForceMediumOrHigh ((uint16_t)0x43) -#define PPSMC_MSG_SwitchToMinimumPower ((uint16_t)0x51) -#define PPSMC_MSG_ResumeFromMinimumPower ((uint16_t)0x52) -#define PPSMC_MSG_EnableCac ((uint16_t)0x53) -#define PPSMC_MSG_DisableCac ((uint16_t)0x54) -#define PPSMC_DPMStateHistoryStart ((uint16_t)0x55) -#define PPSMC_DPMStateHistoryStop ((uint16_t)0x56) -#define PPSMC_CACHistoryStart ((uint16_t)0x57) -#define PPSMC_CACHistoryStop ((uint16_t)0x58) -#define PPSMC_TDPClampingActive ((uint16_t)0x59) -#define PPSMC_TDPClampingInactive ((uint16_t)0x5A) -#define PPSMC_StartFanControl ((uint16_t)0x5B) -#define PPSMC_StopFanControl ((uint16_t)0x5C) -#define PPSMC_NoDisplay ((uint16_t)0x5D) -#define PPSMC_HasDisplay ((uint16_t)0x5E) -#define PPSMC_MSG_UVDPowerOFF ((uint16_t)0x60) -#define PPSMC_MSG_UVDPowerON ((uint16_t)0x61) -#define PPSMC_MSG_EnableULV ((uint16_t)0x62) -#define PPSMC_MSG_DisableULV ((uint16_t)0x63) -#define PPSMC_MSG_EnterULV ((uint16_t)0x64) -#define PPSMC_MSG_ExitULV ((uint16_t)0x65) -#define PPSMC_PowerShiftActive ((uint16_t)0x6A) -#define PPSMC_PowerShiftInactive ((uint16_t)0x6B) -#define PPSMC_OCPActive ((uint16_t)0x6C) -#define PPSMC_OCPInactive ((uint16_t)0x6D) -#define PPSMC_CACLongTermAvgEnable ((uint16_t)0x6E) -#define PPSMC_CACLongTermAvgDisable ((uint16_t)0x6F) -#define PPSMC_MSG_InferredStateSweep_Start ((uint16_t)0x70) -#define PPSMC_MSG_InferredStateSweep_Stop ((uint16_t)0x71) -#define PPSMC_MSG_SwitchToLowestInfState ((uint16_t)0x72) -#define PPSMC_MSG_SwitchToNonInfState ((uint16_t)0x73) -#define PPSMC_MSG_AllStateSweep_Start ((uint16_t)0x74) -#define PPSMC_MSG_AllStateSweep_Stop ((uint16_t)0x75) -#define PPSMC_MSG_SwitchNextLowerInfState ((uint16_t)0x76) -#define PPSMC_MSG_SwitchNextHigherInfState ((uint16_t)0x77) -#define PPSMC_MSG_MclkRetrainingTest ((uint16_t)0x78) -#define PPSMC_MSG_ForceTDPClamping ((uint16_t)0x79) -#define PPSMC_MSG_CollectCAC_PowerCorreln ((uint16_t)0x7A) -#define PPSMC_MSG_CollectCAC_WeightCalib ((uint16_t)0x7B) -#define PPSMC_MSG_CollectCAC_SQonly ((uint16_t)0x7C) -#define PPSMC_MSG_CollectCAC_TemperaturePwr ((uint16_t)0x7D) -#define PPSMC_MSG_ExtremitiesTest_Start ((uint16_t)0x7E) -#define PPSMC_MSG_ExtremitiesTest_Stop ((uint16_t)0x7F) -#define PPSMC_FlushDataCache ((uint16_t)0x80) -#define PPSMC_FlushInstrCache ((uint16_t)0x81) -#define PPSMC_MSG_SetEnabledLevels ((uint16_t)0x82) -#define PPSMC_MSG_SetForcedLevels ((uint16_t)0x83) -#define PPSMC_MSG_ResetToDefaults ((uint16_t)0x84) -#define PPSMC_MSG_SetForcedLevelsAndJump ((uint16_t)0x85) -#define PPSMC_MSG_SetCACHistoryMode ((uint16_t)0x86) -#define PPSMC_MSG_EnableDTE ((uint16_t)0x87) -#define PPSMC_MSG_DisableDTE ((uint16_t)0x88) -#define PPSMC_MSG_SmcSpaceSetAddress ((uint16_t)0x89) -#define PPSMC_MSG_SmcSpaceWriteDWordInc ((uint16_t)0x8A) -#define PPSMC_MSG_SmcSpaceWriteWordInc ((uint16_t)0x8B) -#define PPSMC_MSG_SmcSpaceWriteByteInc ((uint16_t)0x8C) - -#define PPSMC_MSG_BREAK ((uint16_t)0xF8) - -#define PPSMC_MSG_Test ((uint16_t)0x100) -#define PPSMC_MSG_DRV_DRAM_ADDR_HI ((uint16_t)0x250) -#define PPSMC_MSG_DRV_DRAM_ADDR_LO ((uint16_t)0x251) -#define PPSMC_MSG_SMU_DRAM_ADDR_HI ((uint16_t)0x252) -#define PPSMC_MSG_SMU_DRAM_ADDR_LO ((uint16_t)0x253) -#define PPSMC_MSG_LoadUcodes ((uint16_t)0x254) - -typedef uint16_t PPSMC_Msg; - -#define PPSMC_EVENT_STATUS_THERMAL 0x00000001 -#define PPSMC_EVENT_STATUS_REGULATORHOT 0x00000002 -#define PPSMC_EVENT_STATUS_DC 0x00000004 -#define PPSMC_EVENT_STATUS_GPIO17 0x00000008 - -#pragma pack(pop) - -#endif -- cgit From 3a74f6f27328ff4b9784a6d16f2aafa62081d9c7 Mon Sep 17 00:00:00 2001 From: Jammy Zhou Date: Tue, 21 Jul 2015 14:01:50 +0800 Subject: drm/amdgpu: add amdgpu.powerplay module option This option can be used to enable the new powerplay implementation, and it is disabled by default. Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 09248a640dbb..a15ca4c7cf60 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -165,6 +165,11 @@ module_param_named(sched_hw_submission, amdgpu_sched_hw_submission, int, 0444); MODULE_PARM_DESC(enable_semaphores, "Enable semaphores (1 = enable, 0 = disable (default))"); module_param_named(enable_semaphores, amdgpu_enable_semaphores, int, 0644); +#ifdef CONFIG_DRM_AMD_POWERPLAY +MODULE_PARM_DESC(powerplay, "Powerplay component (1 = enable, 0 = disable (default))"); +module_param_named(powerplay, amdgpu_powerplay, int, 0444); +#endif + static struct pci_device_id pciidlist[] = { #ifdef CONFIG_DRM_AMDGPU_CIK /* Kaveri */ -- cgit From 899fa4c04e3894007174f8fd49f86154440afc91 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Tue, 29 Sep 2015 14:58:53 -0400 Subject: drm/amd/amdgpu: enable powerplay and smc firmware loading for Fiji. Switch over to handling in the powerplay module. Reviewed-by: Alex Deucher Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 3 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 3 +++ drivers/gpu/drm/amd/amdgpu/vi.c | 2 +- 3 files changed, 7 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 8f758eaeef9f..a611401a29d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -703,6 +703,9 @@ static int amdgpu_cgs_get_firmware_info(void *cgs_device, case CHIP_TONGA: strcpy(fw_name, "amdgpu/tonga_smc.bin"); break; + case CHIP_FIJI: + strcpy(fw_name, "amdgpu/fiji_smc.bin"); + break; default: DRM_ERROR("SMC firmware not supported\n"); return -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index 5dd2a4c1a70d..1a824f05a28b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -77,6 +77,9 @@ static int amdgpu_powerplay_init(struct amdgpu_device *adev) case CHIP_TONGA: amd_pp->ip_funcs = &tonga_dpm_ip_funcs; break; + case CHIP_FIJI: + amd_pp->ip_funcs = &fiji_dpm_ip_funcs; + break; case CHIP_CARRIZO: amd_pp->ip_funcs = &cz_dpm_ip_funcs; break; diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 8e4c026824d1..e51070e9697a 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -1246,7 +1246,7 @@ static const struct amdgpu_ip_block_version fiji_ip_blocks[] = .major = 7, .minor = 1, .rev = 0, - .funcs = &fiji_dpm_ip_funcs, + .funcs = &amdgpu_pp_ip_funcs, }, { .type = AMD_IP_BLOCK_TYPE_DCE, -- cgit From 76c8cc6b3ba2186215322cf45d6547d66713bd7b Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Sat, 17 Oct 2015 17:57:58 +0800 Subject: drm/amdgpu: enable powerplay module by default for tonga. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index 1a824f05a28b..cbb00e29b56c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -96,6 +96,14 @@ static int amdgpu_pp_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int ret = 0; + switch (adev->asic_type) { + case CHIP_TONGA: + amdgpu_powerplay = 1; + break; + default: + break; + } + ret = amdgpu_powerplay_init(adev); if (ret) return ret; -- cgit From edb611c1e1e7647510185a3fcde5914f761afd75 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Tue, 20 Oct 2015 11:05:45 +0800 Subject: drm/amdgpu: enable powerplay module by default for fiji. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index cbb00e29b56c..1ff6fd54df61 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -96,13 +96,16 @@ static int amdgpu_pp_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int ret = 0; +#ifdef CONFIG_DRM_AMD_POWERPLAY switch (adev->asic_type) { case CHIP_TONGA: + case CHIP_FIJI: amdgpu_powerplay = 1; break; default: break; } +#endif ret = amdgpu_powerplay_init(adev); if (ret) -- cgit From 3af76f23a45b75441b8eac30aa5a7d957e699e73 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Thu, 15 Oct 2015 17:23:43 +0800 Subject: drm/amdgpu: export fan control functions to amdgpu Hook up the amdgpu thermal control callbacks for powerplay. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 30 +++++++++++++++++++++++++----- 1 file changed, 25 insertions(+), 5 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index d9ef4d25be79..8b1ff13d50cc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -2259,7 +2259,6 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) #define amdgpu_display_resume_mc_access(adev, s) (adev)->mode_info.funcs->resume_mc_access((adev), (s)) #define amdgpu_emit_copy_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_copy_buffer((ib), (s), (d), (b)) #define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b)) -#define amdgpu_dpm_get_temperature(adev) (adev)->pm.funcs->get_temperature((adev)) #define amdgpu_dpm_pre_set_power_state(adev) (adev)->pm.funcs->pre_set_power_state((adev)) #define amdgpu_dpm_set_power_state(adev) (adev)->pm.funcs->set_power_state((adev)) #define amdgpu_dpm_post_set_power_state(adev) (adev)->pm.funcs->post_set_power_state((adev)) @@ -2267,10 +2266,31 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) #define amdgpu_dpm_print_power_state(adev, ps) (adev)->pm.funcs->print_power_state((adev), (ps)) #define amdgpu_dpm_vblank_too_short(adev) (adev)->pm.funcs->vblank_too_short((adev)) #define amdgpu_dpm_enable_bapm(adev, e) (adev)->pm.funcs->enable_bapm((adev), (e)) -#define amdgpu_dpm_set_fan_control_mode(adev, m) (adev)->pm.funcs->set_fan_control_mode((adev), (m)) -#define amdgpu_dpm_get_fan_control_mode(adev) (adev)->pm.funcs->get_fan_control_mode((adev)) -#define amdgpu_dpm_set_fan_speed_percent(adev, s) (adev)->pm.funcs->set_fan_speed_percent((adev), (s)) -#define amdgpu_dpm_get_fan_speed_percent(adev, s) (adev)->pm.funcs->get_fan_speed_percent((adev), (s)) + +#define amdgpu_dpm_get_temperature(adev) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->get_temperature((adev)->powerplay.pp_handle) : \ + (adev)->pm.funcs->get_temperature((adev)) + +#define amdgpu_dpm_set_fan_control_mode(adev, m) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->set_fan_control_mode((adev)->powerplay.pp_handle, (m)) : \ + (adev)->pm.funcs->set_fan_control_mode((adev), (m)) + +#define amdgpu_dpm_get_fan_control_mode(adev) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->get_fan_control_mode((adev)->powerplay.pp_handle) : \ + (adev)->pm.funcs->get_fan_control_mode((adev)) + +#define amdgpu_dpm_set_fan_speed_percent(adev, s) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->set_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \ + (adev)->pm.funcs->set_fan_speed_percent((adev), (s)) + +#define amdgpu_dpm_get_fan_speed_percent(adev, s) \ + amdgpu_powerplay ? \ + (adev)->powerplay.pp_funcs->get_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \ + (adev)->pm.funcs->get_fan_speed_percent((adev), (s)) #define amdgpu_dpm_get_sclk(adev, l) \ amdgpu_powerplay ? \ -- cgit From 8804b8d5b05bbf5aea205e49fa4ed8240eb1728d Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Tue, 10 Nov 2015 18:29:11 -0500 Subject: drm/amdgpu: enable sysfs interface for powerplay Same interface exposed in pre-powerplay dpm code. Signed-off-by: Rex Zhu Reviewed-by: Jammy Zhou Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 235fae54724b..40ae30530a01 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -184,10 +184,10 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev, struct amdgpu_device *adev = dev_get_drvdata(dev); int temp; - if (adev->pm.funcs->get_temperature) - temp = amdgpu_dpm_get_temperature(adev); - else + if (!amdgpu_powerplay && !adev->pm.funcs->get_temperature) temp = 0; + else + temp = amdgpu_dpm_get_temperature(adev); return snprintf(buf, PAGE_SIZE, "%d\n", temp); } @@ -215,8 +215,10 @@ static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev, struct amdgpu_device *adev = dev_get_drvdata(dev); u32 pwm_mode = 0; - if (adev->pm.funcs->get_fan_control_mode) - pwm_mode = amdgpu_dpm_get_fan_control_mode(adev); + if (!amdgpu_powerplay && !adev->pm.funcs->get_fan_control_mode) + return -EINVAL; + + pwm_mode = amdgpu_dpm_get_fan_control_mode(adev); /* never 0 (full-speed), fuse or smc-controlled always */ return sprintf(buf, "%i\n", pwm_mode == FDO_PWM_MODE_STATIC ? 1 : 2); @@ -231,7 +233,7 @@ static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev, int err; int value; - if (!adev->pm.funcs->set_fan_control_mode) + if (!amdgpu_powerplay && !adev->pm.funcs->set_fan_control_mode) return -EINVAL; err = kstrtoint(buf, 10, &value); @@ -328,9 +330,6 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj, struct amdgpu_device *adev = dev_get_drvdata(dev); umode_t effective_mode = attr->mode; - if (amdgpu_powerplay) - return 0; /* to do */ - /* Skip limit attributes if DPM is not enabled */ if (!adev->pm.dpm_enabled && (attr == &sensor_dev_attr_temp1_crit.dev_attr.attr || @@ -341,6 +340,9 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj, attr == &sensor_dev_attr_pwm1_min.dev_attr.attr)) return 0; + if (amdgpu_powerplay) + return effective_mode; + /* Skip fan attributes if fan is not present */ if (adev->pm.no_fan && (attr == &sensor_dev_attr_pwm1.dev_attr.attr || -- cgit From e61710c59dd205b48413762b2aedd46e86df3c45 Mon Sep 17 00:00:00 2001 From: Jammy Zhou Date: Tue, 10 Nov 2015 18:31:08 -0500 Subject: drm/amdgpu: support per device powerplay enablement (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The amdgu_powerplay variable is global for multiple GPU instances. v2: fold in Flora's module option change, protect adev reference in macros Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 71 ++++++++++++++------------- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 30 +++++------ drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 15 +++--- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 4 +- drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 2 +- 6 files changed, 65 insertions(+), 61 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 8b1ff13d50cc..637eff393bc7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -2040,6 +2040,7 @@ struct amdgpu_device { /* powerplay */ struct amd_powerplay powerplay; + bool pp_enabled; /* dpm */ struct amdgpu_pm pm; @@ -2268,68 +2269,68 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) #define amdgpu_dpm_enable_bapm(adev, e) (adev)->pm.funcs->enable_bapm((adev), (e)) #define amdgpu_dpm_get_temperature(adev) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->get_temperature((adev)->powerplay.pp_handle) : \ - (adev)->pm.funcs->get_temperature((adev)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->get_temperature((adev)->powerplay.pp_handle) : \ + (adev)->pm.funcs->get_temperature((adev)) #define amdgpu_dpm_set_fan_control_mode(adev, m) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->set_fan_control_mode((adev)->powerplay.pp_handle, (m)) : \ - (adev)->pm.funcs->set_fan_control_mode((adev), (m)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->set_fan_control_mode((adev)->powerplay.pp_handle, (m)) : \ + (adev)->pm.funcs->set_fan_control_mode((adev), (m)) #define amdgpu_dpm_get_fan_control_mode(adev) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->get_fan_control_mode((adev)->powerplay.pp_handle) : \ - (adev)->pm.funcs->get_fan_control_mode((adev)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->get_fan_control_mode((adev)->powerplay.pp_handle) : \ + (adev)->pm.funcs->get_fan_control_mode((adev)) #define amdgpu_dpm_set_fan_speed_percent(adev, s) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->set_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \ - (adev)->pm.funcs->set_fan_speed_percent((adev), (s)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->set_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \ + (adev)->pm.funcs->set_fan_speed_percent((adev), (s)) #define amdgpu_dpm_get_fan_speed_percent(adev, s) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->get_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \ - (adev)->pm.funcs->get_fan_speed_percent((adev), (s)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->get_fan_speed_percent((adev)->powerplay.pp_handle, (s)) : \ + (adev)->pm.funcs->get_fan_speed_percent((adev), (s)) #define amdgpu_dpm_get_sclk(adev, l) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->get_sclk((adev)->powerplay.pp_handle, (l)) : \ + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->get_sclk((adev)->powerplay.pp_handle, (l)) : \ (adev)->pm.funcs->get_sclk((adev), (l)) #define amdgpu_dpm_get_mclk(adev, l) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->get_mclk((adev)->powerplay.pp_handle, (l)) : \ - (adev)->pm.funcs->get_mclk((adev), (l)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->get_mclk((adev)->powerplay.pp_handle, (l)) : \ + (adev)->pm.funcs->get_mclk((adev), (l)) #define amdgpu_dpm_force_performance_level(adev, l) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->force_performance_level((adev)->powerplay.pp_handle, (l)) : \ - (adev)->pm.funcs->force_performance_level((adev), (l)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->force_performance_level((adev)->powerplay.pp_handle, (l)) : \ + (adev)->pm.funcs->force_performance_level((adev), (l)) #define amdgpu_dpm_powergate_uvd(adev, g) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->powergate_uvd((adev)->powerplay.pp_handle, (g)) : \ - (adev)->pm.funcs->powergate_uvd((adev), (g)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->powergate_uvd((adev)->powerplay.pp_handle, (g)) : \ + (adev)->pm.funcs->powergate_uvd((adev), (g)) #define amdgpu_dpm_powergate_vce(adev, g) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->powergate_vce((adev)->powerplay.pp_handle, (g)) : \ - (adev)->pm.funcs->powergate_vce((adev), (g)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->powergate_vce((adev)->powerplay.pp_handle, (g)) : \ + (adev)->pm.funcs->powergate_vce((adev), (g)) #define amdgpu_dpm_debugfs_print_current_performance_level(adev, m) \ - amdgpu_powerplay ? \ - (adev)->powerplay.pp_funcs->print_current_performance_level((adev)->powerplay.pp_handle, (m)) : \ - (adev)->pm.funcs->debugfs_print_current_performance_level((adev), (m)) + (adev)->pp_enabled ? \ + (adev)->powerplay.pp_funcs->print_current_performance_level((adev)->powerplay.pp_handle, (m)) : \ + (adev)->pm.funcs->debugfs_print_current_performance_level((adev), (m)) #define amdgpu_dpm_get_current_power_state(adev) \ - (adev)->powerplay.pp_funcs->get_current_power_state((adev)->powerplay.pp_handle) + (adev)->powerplay.pp_funcs->get_current_power_state((adev)->powerplay.pp_handle) #define amdgpu_dpm_get_performance_level(adev) \ - (adev)->powerplay.pp_funcs->get_performance_level((adev)->powerplay.pp_handle) + (adev)->powerplay.pp_funcs->get_performance_level((adev)->powerplay.pp_handle) -#define amdgpu_dpm_dispatch_task(adev, event_id, input, output) \ +#define amdgpu_dpm_dispatch_task(adev, event_id, input, output) \ (adev)->powerplay.pp_funcs->dispatch_tasks((adev)->powerplay.pp_handle, (event_id), (input), (output)) #define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index a15ca4c7cf60..b5dbbb573491 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -82,7 +82,7 @@ int amdgpu_enable_scheduler = 1; int amdgpu_sched_jobs = 32; int amdgpu_sched_hw_submission = 2; int amdgpu_enable_semaphores = 0; -int amdgpu_powerplay = 0; +int amdgpu_powerplay = -1; MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes"); module_param_named(vramlimit, amdgpu_vram_limit, int, 0600); @@ -166,7 +166,7 @@ MODULE_PARM_DESC(enable_semaphores, "Enable semaphores (1 = enable, 0 = disable module_param_named(enable_semaphores, amdgpu_enable_semaphores, int, 0644); #ifdef CONFIG_DRM_AMD_POWERPLAY -MODULE_PARM_DESC(powerplay, "Powerplay component (1 = enable, 0 = disable (default))"); +MODULE_PARM_DESC(powerplay, "Powerplay component (1 = enable, 0 = disable, -1 = auto (default))"); module_param_named(powerplay, amdgpu_powerplay, int, 0444); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 40ae30530a01..3b78982abaf1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -36,7 +36,7 @@ static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev); void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev) { - if (amdgpu_powerplay) + if (adev->pp_enabled) /* TODO */ return; @@ -60,7 +60,7 @@ static ssize_t amdgpu_get_dpm_state(struct device *dev, struct amdgpu_device *adev = ddev->dev_private; enum amd_pm_state_type pm; - if (amdgpu_powerplay) { + if (adev->pp_enabled) { pm = amdgpu_dpm_get_current_power_state(adev); } else pm = adev->pm.dpm.user_state; @@ -90,7 +90,7 @@ static ssize_t amdgpu_set_dpm_state(struct device *dev, goto fail; } - if (amdgpu_powerplay) { + if (adev->pp_enabled) { amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_ENABLE_USER_STATE, &state, NULL); } else { mutex_lock(&adev->pm.mutex); @@ -113,7 +113,7 @@ static ssize_t amdgpu_get_dpm_forced_performance_level(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = ddev->dev_private; - if (amdgpu_powerplay) { + if (adev->pp_enabled) { enum amd_dpm_forced_level level; level = amdgpu_dpm_get_performance_level(adev); @@ -151,7 +151,7 @@ static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev, goto fail; } - if (amdgpu_powerplay) + if (adev->pp_enabled) amdgpu_dpm_force_performance_level(adev, level); else { mutex_lock(&adev->pm.mutex); @@ -184,7 +184,7 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev, struct amdgpu_device *adev = dev_get_drvdata(dev); int temp; - if (!amdgpu_powerplay && !adev->pm.funcs->get_temperature) + if (!adev->pp_enabled && !adev->pm.funcs->get_temperature) temp = 0; else temp = amdgpu_dpm_get_temperature(adev); @@ -215,7 +215,7 @@ static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev, struct amdgpu_device *adev = dev_get_drvdata(dev); u32 pwm_mode = 0; - if (!amdgpu_powerplay && !adev->pm.funcs->get_fan_control_mode) + if (!adev->pp_enabled && !adev->pm.funcs->get_fan_control_mode) return -EINVAL; pwm_mode = amdgpu_dpm_get_fan_control_mode(adev); @@ -233,7 +233,7 @@ static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev, int err; int value; - if (!amdgpu_powerplay && !adev->pm.funcs->set_fan_control_mode) + if (!adev->pp_enabled && !adev->pm.funcs->set_fan_control_mode) return -EINVAL; err = kstrtoint(buf, 10, &value); @@ -340,7 +340,7 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj, attr == &sensor_dev_attr_pwm1_min.dev_attr.attr)) return 0; - if (amdgpu_powerplay) + if (adev->pp_enabled) return effective_mode; /* Skip fan attributes if fan is not present */ @@ -674,7 +674,7 @@ done: void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable) { - if (amdgpu_powerplay) + if (adev->pp_enabled) amdgpu_dpm_powergate_uvd(adev, !enable); else { if (adev->pm.funcs->powergate_uvd) { @@ -701,7 +701,7 @@ void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable) void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable) { - if (amdgpu_powerplay) + if (adev->pp_enabled) amdgpu_dpm_powergate_vce(adev, !enable); else { if (adev->pm.funcs->powergate_vce) { @@ -729,7 +729,7 @@ void amdgpu_pm_print_power_states(struct amdgpu_device *adev) { int i; - if (amdgpu_powerplay) + if (adev->pp_enabled) /* TO DO */ return; @@ -745,7 +745,7 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev) if (adev->pm.sysfs_initialized) return 0; - if (!amdgpu_powerplay) { + if (!adev->pp_enabled) { if (adev->pm.funcs->get_temperature == NULL) return 0; } @@ -798,7 +798,7 @@ void amdgpu_pm_compute_clocks(struct amdgpu_device *adev) if (!adev->pm.dpm_enabled) return; - if (amdgpu_powerplay) { + if (adev->pp_enabled) { int i = 0; amdgpu_display_bandwidth_update(adev); @@ -852,7 +852,7 @@ static int amdgpu_debugfs_pm_info(struct seq_file *m, void *data) seq_printf(m, "dpm not enabled\n"); return 0; } - if (amdgpu_powerplay) { + if (adev->pp_enabled) { amdgpu_dpm_debugfs_print_current_performance_level(adev, m); } else { mutex_lock(&adev->pm.mutex); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index 1ff6fd54df61..6b46fbfd6be4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -40,7 +40,7 @@ static int amdgpu_powerplay_init(struct amdgpu_device *adev) amd_pp = &(adev->powerplay); - if (amdgpu_powerplay) { + if (adev->pp_enabled) { #ifdef CONFIG_DRM_AMD_POWERPLAY struct amd_pp_init *pp_init; @@ -100,11 +100,14 @@ static int amdgpu_pp_early_init(void *handle) switch (adev->asic_type) { case CHIP_TONGA: case CHIP_FIJI: - amdgpu_powerplay = 1; + adev->pp_enabled = (amdgpu_powerplay == 0) ? false : true; break; default: + adev->pp_enabled = (amdgpu_powerplay > 0) ? true : false; break; } +#else + adev->pp_enabled = false; #endif ret = amdgpu_powerplay_init(adev); @@ -127,7 +130,7 @@ static int amdgpu_pp_sw_init(void *handle) adev->powerplay.pp_handle); #ifdef CONFIG_DRM_AMD_POWERPLAY - if (amdgpu_powerplay) { + if (adev->pp_enabled) { adev->pm.dpm_enabled = true; amdgpu_pm_sysfs_init(adev); } @@ -148,7 +151,7 @@ static int amdgpu_pp_sw_fini(void *handle) return ret; #ifdef CONFIG_DRM_AMD_POWERPLAY - if (amdgpu_powerplay) { + if (adev->pp_enabled) { amdgpu_pm_sysfs_fini(adev); amd_powerplay_fini(adev->powerplay.pp_handle); } @@ -162,7 +165,7 @@ static int amdgpu_pp_hw_init(void *handle) int ret = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (amdgpu_powerplay && adev->firmware.smu_load) + if (adev->pp_enabled && adev->firmware.smu_load) amdgpu_ucode_init_bo(adev); if (adev->powerplay.ip_funcs->hw_init) @@ -181,7 +184,7 @@ static int amdgpu_pp_hw_fini(void *handle) ret = adev->powerplay.ip_funcs->hw_fini( adev->powerplay.pp_handle); - if (amdgpu_powerplay && adev->firmware.smu_load) + if (adev->pp_enabled && adev->firmware.smu_load) amdgpu_ucode_fini_bo(adev); return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index d90aae08445e..6c063a4a2c38 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -2902,7 +2902,7 @@ static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev) gfx_v8_0_rlc_reset(adev); - if (!amdgpu_powerplay) { + if (!adev->pp_enabled) { if (!adev->firmware.smu_load) { /* legacy rlc firmware loading */ r = gfx_v8_0_rlc_load_microcode(adev); @@ -3804,7 +3804,7 @@ static int gfx_v8_0_cp_resume(struct amdgpu_device *adev) if (!(adev->flags & AMD_IS_APU)) gfx_v8_0_enable_gui_idle_interrupt(adev, false); - if (!amdgpu_powerplay) { + if (!adev->pp_enabled) { if (!adev->firmware.smu_load) { /* legacy firmware loading */ r = gfx_v8_0_cp_gfx_load_microcode(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 8091c1c37c4e..c741c091bc9e 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -727,7 +727,7 @@ static int sdma_v3_0_start(struct amdgpu_device *adev) { int r, i; - if (!amdgpu_powerplay) { + if (!adev->pp_enabled) { if (!adev->firmware.smu_load) { r = sdma_v3_0_load_microcode(adev); if (r) -- cgit From d0dd7f0cc345fc8757148004639e1993ba183bd6 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 11 Nov 2015 19:45:06 -0500 Subject: drm/amdgpu: store pcie gen mask and link width We'll need this later for pcie dpm. Reviewed-by: Jammy Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 4 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 78 ++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/cik.c | 24 ++++----- drivers/gpu/drm/amd/amdgpu/vi.c | 13 ++--- 4 files changed, 99 insertions(+), 20 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 637eff393bc7..26d9134f7c0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1637,8 +1637,12 @@ struct amdgpu_pm { const struct firmware *fw; /* SMC firmware */ uint32_t fw_version; const struct amdgpu_dpm_funcs *funcs; + uint32_t pcie_gen_mask; + uint32_t pcie_mlw_mask; }; +void amdgpu_get_pcie_info(struct amdgpu_device *adev); + /* * UVD */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 587ff7145361..65531463f88e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -38,6 +38,7 @@ #include "amdgpu_i2c.h" #include "atom.h" #include "amdgpu_atombios.h" +#include "amd_pcie.h" #ifdef CONFIG_DRM_AMDGPU_CIK #include "cik.h" #endif @@ -1932,6 +1933,83 @@ retry: return r; } +void amdgpu_get_pcie_info(struct amdgpu_device *adev) +{ + u32 mask; + int ret; + + if (pci_is_root_bus(adev->pdev->bus)) + return; + + if (amdgpu_pcie_gen2 == 0) + return; + + if (adev->flags & AMD_IS_APU) + return; + + ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask); + if (!ret) { + adev->pm.pcie_gen_mask = (CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1 | + CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN2 | + CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN3); + + if (mask & DRM_PCIE_SPEED_25) + adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1; + if (mask & DRM_PCIE_SPEED_50) + adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2; + if (mask & DRM_PCIE_SPEED_80) + adev->pm.pcie_gen_mask |= CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3; + } + ret = drm_pcie_get_max_link_width(adev->ddev, &mask); + if (!ret) { + switch (mask) { + case 32: + adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X1); + break; + case 16: + adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X1); + break; + case 12: + adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X12 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X1); + break; + case 8: + adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X8 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X1); + break; + case 4: + adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X4 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X1); + break; + case 2: + adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X2 | + CAIL_PCIE_LINK_WIDTH_SUPPORT_X1); + break; + case 1: + adev->pm.pcie_mlw_mask = CAIL_PCIE_LINK_WIDTH_SUPPORT_X1; + break; + default: + break; + } + } +} /* * Debugfs diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index c7c298b88170..fd9c9588ef46 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -32,6 +32,7 @@ #include "amdgpu_vce.h" #include "cikd.h" #include "atom.h" +#include "amd_pcie.h" #include "cik.h" #include "gmc_v7_0.h" @@ -1595,8 +1596,8 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev) { struct pci_dev *root = adev->pdev->bus->self; int bridge_pos, gpu_pos; - u32 speed_cntl, mask, current_data_rate; - int ret, i; + u32 speed_cntl, current_data_rate; + int i; u16 tmp16; if (pci_is_root_bus(adev->pdev->bus)) @@ -1608,23 +1609,20 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev) if (adev->flags & AMD_IS_APU) return; - ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask); - if (ret != 0) - return; - - if (!(mask & (DRM_PCIE_SPEED_50 | DRM_PCIE_SPEED_80))) + if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 | + CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3))) return; speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL); current_data_rate = (speed_cntl & PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK) >> PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT; - if (mask & DRM_PCIE_SPEED_80) { + if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) { if (current_data_rate == 2) { DRM_INFO("PCIE gen 3 link speeds already enabled\n"); return; } DRM_INFO("enabling PCIE gen 3 link speeds, disable with amdgpu.pcie_gen2=0\n"); - } else if (mask & DRM_PCIE_SPEED_50) { + } else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) { if (current_data_rate == 1) { DRM_INFO("PCIE gen 2 link speeds already enabled\n"); return; @@ -1640,7 +1638,7 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev) if (!gpu_pos) return; - if (mask & DRM_PCIE_SPEED_80) { + if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) { /* re-try equalization if gen3 is not already enabled */ if (current_data_rate != 2) { u16 bridge_cfg, gpu_cfg; @@ -1735,9 +1733,9 @@ static void cik_pcie_gen3_enable(struct amdgpu_device *adev) pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16); tmp16 &= ~0xf; - if (mask & DRM_PCIE_SPEED_80) + if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3) tmp16 |= 3; /* gen3 */ - else if (mask & DRM_PCIE_SPEED_50) + else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2) tmp16 |= 2; /* gen2 */ else tmp16 |= 1; /* gen1 */ @@ -2450,6 +2448,8 @@ static int cik_common_early_init(void *handle) return -EINVAL; } + amdgpu_get_pcie_info(adev); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index e51070e9697a..25d620790607 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -31,6 +31,7 @@ #include "amdgpu_vce.h" #include "amdgpu_ucode.h" #include "atom.h" +#include "amd_pcie.h" #include "gmc/gmc_8_1_d.h" #include "gmc/gmc_8_1_sh_mask.h" @@ -1052,9 +1053,6 @@ static int vi_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk) static void vi_pcie_gen3_enable(struct amdgpu_device *adev) { - u32 mask; - int ret; - if (pci_is_root_bus(adev->pdev->bus)) return; @@ -1064,11 +1062,8 @@ static void vi_pcie_gen3_enable(struct amdgpu_device *adev) if (adev->flags & AMD_IS_APU) return; - ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask); - if (ret != 0) - return; - - if (!(mask & (DRM_PCIE_SPEED_50 | DRM_PCIE_SPEED_80))) + if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 | + CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3))) return; /* todo */ @@ -1473,6 +1468,8 @@ static int vi_common_early_init(void *handle) if (amdgpu_smc_load_fw && smc_enabled) adev->firmware.smu_load = true; + amdgpu_get_pcie_info(adev); + return 0; } -- cgit From cfd316d59e203985699495147a973ba058ff5478 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 11 Nov 2015 20:35:32 -0500 Subject: drm/amdgpu/cgs: add sys info query for pcie gen and link width Needed by powerplay to properly handle pcie dpm switching. Reviewed-by: Jammy Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index a611401a29d5..6fa0feac27f8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -754,6 +754,12 @@ static int amdgpu_cgs_query_system_info(void *cgs_device, case CGS_SYSTEM_INFO_ADAPTER_BDF_ID: sys_info->value = adev->pdev->devfn | (adev->pdev->bus->number << 8); break; + case CGS_SYSTEM_INFO_PCIE_GEN_INFO: + sys_info->value = adev->pm.pcie_gen_mask; + break; + case CGS_SYSTEM_INFO_PCIE_MLW: + sys_info->value = adev->pm.pcie_mlw_mask; + break; default: return -ENODEV; } -- cgit From 6e378858df1df362d8db814666d0caa0703087a3 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Tue, 10 Nov 2015 10:50:25 -0500 Subject: drm/amd/amdgpu: add gfx clock gating support for Fiji. Reviewed-by: Alex Deucher Reviewed-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 254 ++++++++++++++++++++++++++++++++++ 1 file changed, 254 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 6c063a4a2c38..13235d84e5a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -66,6 +66,27 @@ #define MACRO_TILE_ASPECT(x) ((x) << GB_MACROTILE_MODE0__MACRO_TILE_ASPECT__SHIFT) #define NUM_BANKS(x) ((x) << GB_MACROTILE_MODE0__NUM_BANKS__SHIFT) +#define RLC_CGTT_MGCG_OVERRIDE__CPF_MASK 0x00000001L +#define RLC_CGTT_MGCG_OVERRIDE__RLC_MASK 0x00000002L +#define RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK 0x00000004L +#define RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK 0x00000008L +#define RLC_CGTT_MGCG_OVERRIDE__CGLS_MASK 0x00000010L +#define RLC_CGTT_MGCG_OVERRIDE__GRBM_MASK 0x00000020L + +/* BPM SERDES CMD */ +#define SET_BPM_SERDES_CMD 1 +#define CLE_BPM_SERDES_CMD 0 + +/* BPM Register Address*/ +enum { + BPM_REG_CGLS_EN = 0, /* Enable/Disable CGLS */ + BPM_REG_CGLS_ON, /* ON/OFF CGLS: shall be controlled by RLC FW */ + BPM_REG_CGCG_OVERRIDE, /* Set/Clear CGCG Override */ + BPM_REG_MGCG_OVERRIDE, /* Set/Clear MGCG Override */ + BPM_REG_FGCG_OVERRIDE, /* Set/Clear FGCG Override */ + BPM_REG_FGCG_MAX +}; + MODULE_FIRMWARE("amdgpu/carrizo_ce.bin"); MODULE_FIRMWARE("amdgpu/carrizo_pfp.bin"); MODULE_FIRMWARE("amdgpu/carrizo_me.bin"); @@ -4301,9 +4322,242 @@ static int gfx_v8_0_set_powergating_state(void *handle, return 0; } +static void fiji_send_serdes_cmd(struct amdgpu_device *adev, + uint32_t reg_addr, uint32_t cmd) +{ + uint32_t data; + + gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff); + + WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff); + WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff); + + data = RREG32(mmRLC_SERDES_WR_CTRL); + data &= ~(RLC_SERDES_WR_CTRL__WRITE_COMMAND_MASK | + RLC_SERDES_WR_CTRL__READ_COMMAND_MASK | + RLC_SERDES_WR_CTRL__P1_SELECT_MASK | + RLC_SERDES_WR_CTRL__P2_SELECT_MASK | + RLC_SERDES_WR_CTRL__RDDATA_RESET_MASK | + RLC_SERDES_WR_CTRL__POWER_DOWN_MASK | + RLC_SERDES_WR_CTRL__POWER_UP_MASK | + RLC_SERDES_WR_CTRL__SHORT_FORMAT_MASK | + RLC_SERDES_WR_CTRL__BPM_DATA_MASK | + RLC_SERDES_WR_CTRL__REG_ADDR_MASK | + RLC_SERDES_WR_CTRL__SRBM_OVERRIDE_MASK); + data |= (RLC_SERDES_WR_CTRL__RSVD_BPM_ADDR_MASK | + (cmd << RLC_SERDES_WR_CTRL__BPM_DATA__SHIFT) | + (reg_addr << RLC_SERDES_WR_CTRL__REG_ADDR__SHIFT) | + (0xff << RLC_SERDES_WR_CTRL__BPM_ADDR__SHIFT)); + + WREG32(mmRLC_SERDES_WR_CTRL, data); +} + +static void fiji_update_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + /* It is disabled by HW by default */ + if (enable) { + /* 1 - RLC memory Light sleep */ + temp = data = RREG32(mmRLC_MEM_SLP_CNTL); + data |= RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK; + if (temp != data) + WREG32(mmRLC_MEM_SLP_CNTL, data); + + /* 2 - CP memory Light sleep */ + temp = data = RREG32(mmCP_MEM_SLP_CNTL); + data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK; + if (temp != data) + WREG32(mmCP_MEM_SLP_CNTL, data); + + /* 3 - RLC_CGTT_MGCG_OVERRIDE */ + temp = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE); + data &= ~(RLC_CGTT_MGCG_OVERRIDE__CPF_MASK | + RLC_CGTT_MGCG_OVERRIDE__RLC_MASK | + RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK | + RLC_CGTT_MGCG_OVERRIDE__GRBM_MASK); + + if (temp != data) + WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data); + + /* 4 - wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + + /* 5 - clear mgcg override */ + fiji_send_serdes_cmd(adev, BPM_REG_MGCG_OVERRIDE, CLE_BPM_SERDES_CMD); + + /* 6 - Enable CGTS(Tree Shade) MGCG /MGLS */ + temp = data = RREG32(mmCGTS_SM_CTRL_REG); + data &= ~(CGTS_SM_CTRL_REG__SM_MODE_MASK); + data |= (0x2 << CGTS_SM_CTRL_REG__SM_MODE__SHIFT); + data |= CGTS_SM_CTRL_REG__SM_MODE_ENABLE_MASK; + data &= ~CGTS_SM_CTRL_REG__OVERRIDE_MASK; + data &= ~CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK; + data |= CGTS_SM_CTRL_REG__ON_MONITOR_ADD_EN_MASK; + data |= (0x96 << CGTS_SM_CTRL_REG__ON_MONITOR_ADD__SHIFT); + if (temp != data) + WREG32(mmCGTS_SM_CTRL_REG, data); + udelay(50); + + /* 7 - wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + } else { + /* 1 - MGCG_OVERRIDE[0] for CP and MGCG_OVERRIDE[1] for RLC */ + temp = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE); + data |= (RLC_CGTT_MGCG_OVERRIDE__CPF_MASK | + RLC_CGTT_MGCG_OVERRIDE__RLC_MASK | + RLC_CGTT_MGCG_OVERRIDE__MGCG_MASK | + RLC_CGTT_MGCG_OVERRIDE__GRBM_MASK); + if (temp != data) + WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data); + + /* 2 - disable MGLS in RLC */ + data = RREG32(mmRLC_MEM_SLP_CNTL); + if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) { + data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK; + WREG32(mmRLC_MEM_SLP_CNTL, data); + } + + /* 3 - disable MGLS in CP */ + data = RREG32(mmCP_MEM_SLP_CNTL); + if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) { + data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK; + WREG32(mmCP_MEM_SLP_CNTL, data); + } + + /* 4 - Disable CGTS(Tree Shade) MGCG and MGLS */ + temp = data = RREG32(mmCGTS_SM_CTRL_REG); + data |= (CGTS_SM_CTRL_REG__OVERRIDE_MASK | + CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK); + if (temp != data) + WREG32(mmCGTS_SM_CTRL_REG, data); + + /* 5 - wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + + /* 6 - set mgcg override */ + fiji_send_serdes_cmd(adev, BPM_REG_MGCG_OVERRIDE, SET_BPM_SERDES_CMD); + + udelay(50); + + /* 7- wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + } +} + +static void fiji_update_coarse_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, temp1, data, data1; + + temp = data = RREG32(mmRLC_CGCG_CGLS_CTRL); + + if (enable) { + /* 1 enable cntx_empty_int_enable/cntx_busy_int_enable/ + * Cmp_busy/GFX_Idle interrupts + */ + gfx_v8_0_enable_gui_idle_interrupt(adev, true); + + temp1 = data1 = RREG32(mmRLC_CGTT_MGCG_OVERRIDE); + data1 &= ~RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK; + if (temp1 != data1) + WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data1); + + /* 2 wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + + /* 3 - clear cgcg override */ + fiji_send_serdes_cmd(adev, BPM_REG_CGCG_OVERRIDE, CLE_BPM_SERDES_CMD); + + /* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + + /* 4 - write cmd to set CGLS */ + fiji_send_serdes_cmd(adev, BPM_REG_CGLS_EN, SET_BPM_SERDES_CMD); + + /* 5 - enable cgcg */ + data |= RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK; + + /* enable cgls*/ + data |= RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK; + + temp1 = data1 = RREG32(mmRLC_CGTT_MGCG_OVERRIDE); + data1 &= ~RLC_CGTT_MGCG_OVERRIDE__CGLS_MASK; + + if (temp1 != data1) + WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data1); + + if (temp != data) + WREG32(mmRLC_CGCG_CGLS_CTRL, data); + } else { + /* disable cntx_empty_int_enable & GFX Idle interrupt */ + gfx_v8_0_enable_gui_idle_interrupt(adev, false); + + /* TEST CGCG */ + temp1 = data1 = RREG32(mmRLC_CGTT_MGCG_OVERRIDE); + data1 |= (RLC_CGTT_MGCG_OVERRIDE__CGCG_MASK | + RLC_CGTT_MGCG_OVERRIDE__CGLS_MASK); + if (temp1 != data1) + WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data1); + + /* read gfx register to wake up cgcg */ + RREG32(mmCB_CGTT_SCLK_CTRL); + RREG32(mmCB_CGTT_SCLK_CTRL); + RREG32(mmCB_CGTT_SCLK_CTRL); + RREG32(mmCB_CGTT_SCLK_CTRL); + + /* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + + /* write cmd to Set CGCG Overrride */ + fiji_send_serdes_cmd(adev, BPM_REG_CGCG_OVERRIDE, SET_BPM_SERDES_CMD); + + /* wait for RLC_SERDES_CU_MASTER & RLC_SERDES_NONCU_MASTER idle */ + gfx_v8_0_wait_for_rlc_serdes(adev); + + /* write cmd to Clear CGLS */ + fiji_send_serdes_cmd(adev, BPM_REG_CGLS_EN, CLE_BPM_SERDES_CMD); + + /* disable cgcg, cgls should be disabled too. */ + data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | + RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK); + if (temp != data) + WREG32(mmRLC_CGCG_CGLS_CTRL, data); + } +} +static int fiji_update_gfx_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + if (enable) { + /* CGCG/CGLS should be enabled after MGCG/MGLS/TS(CG/LS) + * === MGCG + MGLS + TS(CG/LS) === + */ + fiji_update_medium_grain_clock_gating(adev, enable); + fiji_update_coarse_grain_clock_gating(adev, enable); + } else { + /* CGCG/CGLS should be disabled before MGCG/MGLS/TS(CG/LS) + * === CGCG + CGLS === + */ + fiji_update_coarse_grain_clock_gating(adev, enable); + fiji_update_medium_grain_clock_gating(adev, enable); + } + return 0; +} + static int gfx_v8_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + switch (adev->asic_type) { + case CHIP_FIJI: + fiji_update_gfx_clock_gating(adev, + state == AMD_CG_STATE_GATE ? true : false); + break; + default: + break; + } return 0; } -- cgit From a0d69786b5e8c39f8e8a24422a05d3f1f4434b4d Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Tue, 10 Nov 2015 11:27:39 -0500 Subject: drm/amd/amdgpu: add gmc clock gating support for Fiji. Reviewed-by: Alex Deucher Reviewed-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 172 ++++++++++++++++++++++++++++++++++ 1 file changed, 172 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index a2d869d30cf4..6e2331f70b39 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -1328,9 +1328,181 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev, return 0; } +static void fiji_update_mc_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t data; + + if (enable) { + data = RREG32(mmMC_HUB_MISC_HUB_CG); + data |= MC_HUB_MISC_HUB_CG__ENABLE_MASK; + WREG32(mmMC_HUB_MISC_HUB_CG, data); + + data = RREG32(mmMC_HUB_MISC_SIP_CG); + data |= MC_HUB_MISC_SIP_CG__ENABLE_MASK; + WREG32(mmMC_HUB_MISC_SIP_CG, data); + + data = RREG32(mmMC_HUB_MISC_VM_CG); + data |= MC_HUB_MISC_VM_CG__ENABLE_MASK; + WREG32(mmMC_HUB_MISC_VM_CG, data); + + data = RREG32(mmMC_XPB_CLK_GAT); + data |= MC_XPB_CLK_GAT__ENABLE_MASK; + WREG32(mmMC_XPB_CLK_GAT, data); + + data = RREG32(mmATC_MISC_CG); + data |= ATC_MISC_CG__ENABLE_MASK; + WREG32(mmATC_MISC_CG, data); + + data = RREG32(mmMC_CITF_MISC_WR_CG); + data |= MC_CITF_MISC_WR_CG__ENABLE_MASK; + WREG32(mmMC_CITF_MISC_WR_CG, data); + + data = RREG32(mmMC_CITF_MISC_RD_CG); + data |= MC_CITF_MISC_RD_CG__ENABLE_MASK; + WREG32(mmMC_CITF_MISC_RD_CG, data); + + data = RREG32(mmMC_CITF_MISC_VM_CG); + data |= MC_CITF_MISC_VM_CG__ENABLE_MASK; + WREG32(mmMC_CITF_MISC_VM_CG, data); + + data = RREG32(mmVM_L2_CG); + data |= VM_L2_CG__ENABLE_MASK; + WREG32(mmVM_L2_CG, data); + } else { + data = RREG32(mmMC_HUB_MISC_HUB_CG); + data &= ~MC_HUB_MISC_HUB_CG__ENABLE_MASK; + WREG32(mmMC_HUB_MISC_HUB_CG, data); + + data = RREG32(mmMC_HUB_MISC_SIP_CG); + data &= ~MC_HUB_MISC_SIP_CG__ENABLE_MASK; + WREG32(mmMC_HUB_MISC_SIP_CG, data); + + data = RREG32(mmMC_HUB_MISC_VM_CG); + data &= ~MC_HUB_MISC_VM_CG__ENABLE_MASK; + WREG32(mmMC_HUB_MISC_VM_CG, data); + + data = RREG32(mmMC_XPB_CLK_GAT); + data &= ~MC_XPB_CLK_GAT__ENABLE_MASK; + WREG32(mmMC_XPB_CLK_GAT, data); + + data = RREG32(mmATC_MISC_CG); + data &= ~ATC_MISC_CG__ENABLE_MASK; + WREG32(mmATC_MISC_CG, data); + + data = RREG32(mmMC_CITF_MISC_WR_CG); + data &= ~MC_CITF_MISC_WR_CG__ENABLE_MASK; + WREG32(mmMC_CITF_MISC_WR_CG, data); + + data = RREG32(mmMC_CITF_MISC_RD_CG); + data &= ~MC_CITF_MISC_RD_CG__ENABLE_MASK; + WREG32(mmMC_CITF_MISC_RD_CG, data); + + data = RREG32(mmMC_CITF_MISC_VM_CG); + data &= ~MC_CITF_MISC_VM_CG__ENABLE_MASK; + WREG32(mmMC_CITF_MISC_VM_CG, data); + + data = RREG32(mmVM_L2_CG); + data &= ~VM_L2_CG__ENABLE_MASK; + WREG32(mmVM_L2_CG, data); + } +} + +static void fiji_update_mc_light_sleep(struct amdgpu_device *adev, + bool enable) +{ + uint32_t data; + + if (enable) { + data = RREG32(mmMC_HUB_MISC_HUB_CG); + data |= MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_HUB_MISC_HUB_CG, data); + + data = RREG32(mmMC_HUB_MISC_SIP_CG); + data |= MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_HUB_MISC_SIP_CG, data); + + data = RREG32(mmMC_HUB_MISC_VM_CG); + data |= MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_HUB_MISC_VM_CG, data); + + data = RREG32(mmMC_XPB_CLK_GAT); + data |= MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK; + WREG32(mmMC_XPB_CLK_GAT, data); + + data = RREG32(mmATC_MISC_CG); + data |= ATC_MISC_CG__MEM_LS_ENABLE_MASK; + WREG32(mmATC_MISC_CG, data); + + data = RREG32(mmMC_CITF_MISC_WR_CG); + data |= MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_CITF_MISC_WR_CG, data); + + data = RREG32(mmMC_CITF_MISC_RD_CG); + data |= MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_CITF_MISC_RD_CG, data); + + data = RREG32(mmMC_CITF_MISC_VM_CG); + data |= MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_CITF_MISC_VM_CG, data); + + data = RREG32(mmVM_L2_CG); + data |= VM_L2_CG__MEM_LS_ENABLE_MASK; + WREG32(mmVM_L2_CG, data); + } else { + data = RREG32(mmMC_HUB_MISC_HUB_CG); + data &= ~MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_HUB_MISC_HUB_CG, data); + + data = RREG32(mmMC_HUB_MISC_SIP_CG); + data &= ~MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_HUB_MISC_SIP_CG, data); + + data = RREG32(mmMC_HUB_MISC_VM_CG); + data &= ~MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_HUB_MISC_VM_CG, data); + + data = RREG32(mmMC_XPB_CLK_GAT); + data &= ~MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK; + WREG32(mmMC_XPB_CLK_GAT, data); + + data = RREG32(mmATC_MISC_CG); + data &= ~ATC_MISC_CG__MEM_LS_ENABLE_MASK; + WREG32(mmATC_MISC_CG, data); + + data = RREG32(mmMC_CITF_MISC_WR_CG); + data &= ~MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_CITF_MISC_WR_CG, data); + + data = RREG32(mmMC_CITF_MISC_RD_CG); + data &= ~MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_CITF_MISC_RD_CG, data); + + data = RREG32(mmMC_CITF_MISC_VM_CG); + data &= ~MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK; + WREG32(mmMC_CITF_MISC_VM_CG, data); + + data = RREG32(mmVM_L2_CG); + data &= ~VM_L2_CG__MEM_LS_ENABLE_MASK; + WREG32(mmVM_L2_CG, data); + } +} + static int gmc_v8_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + switch (adev->asic_type) { + case CHIP_FIJI: + fiji_update_mc_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE ? true : false); + fiji_update_mc_light_sleep(adev, + state == AMD_CG_STATE_GATE ? true : false); + break; + default: + break; + } return 0; } -- cgit From 3c997d2412572a9306b7cebd713271c0fdc1350c Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Wed, 11 Nov 2015 11:49:11 -0500 Subject: drm/amdgpu: add sdma clock gating support for Fiji. Reviewed-by: Alex Deucher Reviewed-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 105 +++++++++++++++++++++++++++++++++ 1 file changed, 105 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index c741c091bc9e..ad54c46751b0 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -1429,9 +1429,114 @@ static int sdma_v3_0_process_illegal_inst_irq(struct amdgpu_device *adev, return 0; } +static void fiji_update_sdma_medium_grain_clock_gating( + struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + if (enable) { + temp = data = RREG32(mmSDMA0_CLK_CTRL); + data &= ~(SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK); + if (data != temp) + WREG32(mmSDMA0_CLK_CTRL, data); + + temp = data = RREG32(mmSDMA1_CLK_CTRL); + data &= ~(SDMA1_CLK_CTRL__SOFT_OVERRIDE7_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE6_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE5_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE4_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE3_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE2_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE1_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE0_MASK); + + if (data != temp) + WREG32(mmSDMA1_CLK_CTRL, data); + } else { + temp = data = RREG32(mmSDMA0_CLK_CTRL); + data |= SDMA0_CLK_CTRL__SOFT_OVERRIDE7_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE6_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE5_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE4_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE3_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE2_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE1_MASK | + SDMA0_CLK_CTRL__SOFT_OVERRIDE0_MASK; + + if (data != temp) + WREG32(mmSDMA0_CLK_CTRL, data); + + temp = data = RREG32(mmSDMA1_CLK_CTRL); + data |= SDMA1_CLK_CTRL__SOFT_OVERRIDE7_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE6_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE5_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE4_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE3_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE2_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE1_MASK | + SDMA1_CLK_CTRL__SOFT_OVERRIDE0_MASK; + + if (data != temp) + WREG32(mmSDMA1_CLK_CTRL, data); + } +} + +static void fiji_update_sdma_medium_grain_light_sleep( + struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + if (enable) { + temp = data = RREG32(mmSDMA0_POWER_CNTL); + data |= SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK; + + if (temp != data) + WREG32(mmSDMA0_POWER_CNTL, data); + + temp = data = RREG32(mmSDMA1_POWER_CNTL); + data |= SDMA1_POWER_CNTL__MEM_POWER_OVERRIDE_MASK; + + if (temp != data) + WREG32(mmSDMA1_POWER_CNTL, data); + } else { + temp = data = RREG32(mmSDMA0_POWER_CNTL); + data &= ~SDMA0_POWER_CNTL__MEM_POWER_OVERRIDE_MASK; + + if (temp != data) + WREG32(mmSDMA0_POWER_CNTL, data); + + temp = data = RREG32(mmSDMA1_POWER_CNTL); + data &= ~SDMA1_POWER_CNTL__MEM_POWER_OVERRIDE_MASK; + + if (temp != data) + WREG32(mmSDMA1_POWER_CNTL, data); + } +} + static int sdma_v3_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + switch (adev->asic_type) { + case CHIP_FIJI: + fiji_update_sdma_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE ? true : false); + fiji_update_sdma_medium_grain_light_sleep(adev, + state == AMD_CG_STATE_GATE ? true : false); + break; + default: + break; + } return 0; } -- cgit From 6cec2655fa988b4df605e46d4b5c7fbe50056dd5 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Thu, 12 Nov 2015 16:59:47 -0500 Subject: drm/amd/powerplay: add parts of system clock gating support for Fiji. (v2) Removed fiji_mgcg_cgcg_init that is affected and redundant for new implementation. v2: re-add mgcg_cgcg init Reviewed-by: Alex Deucher Reviewed-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/vi.c | 86 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 86 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 25d620790607..67845ee644ad 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -1543,9 +1543,95 @@ static int vi_common_soft_reset(void *handle) return 0; } +static void fiji_update_bif_medium_grain_light_sleep(struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + temp = data = RREG32_PCIE(ixPCIE_CNTL2); + + if (enable) + data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK | + PCIE_CNTL2__MST_MEM_LS_EN_MASK | + PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK; + else + data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK | + PCIE_CNTL2__MST_MEM_LS_EN_MASK | + PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK); + + if (temp != data) + WREG32_PCIE(ixPCIE_CNTL2, data); +} + +static void fiji_update_hdp_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + temp = data = RREG32(mmHDP_HOST_PATH_CNTL); + + if (enable) + data &= ~HDP_HOST_PATH_CNTL__CLOCK_GATING_DIS_MASK; + else + data |= HDP_HOST_PATH_CNTL__CLOCK_GATING_DIS_MASK; + + if (temp != data) + WREG32(mmHDP_HOST_PATH_CNTL, data); +} + +static void fiji_update_hdp_light_sleep(struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + temp = data = RREG32(mmHDP_MEM_POWER_LS); + + if (enable) + data |= HDP_MEM_POWER_LS__LS_ENABLE_MASK; + else + data &= ~HDP_MEM_POWER_LS__LS_ENABLE_MASK; + + if (temp != data) + WREG32(mmHDP_MEM_POWER_LS, data); +} + +static void fiji_update_rom_medium_grain_clock_gating(struct amdgpu_device *adev, + bool enable) +{ + uint32_t temp, data; + + temp = data = RREG32_SMC(ixCGTT_ROM_CLK_CTRL0); + + if (enable) + data &= ~(CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK | + CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK); + else + data |= CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE0_MASK | + CGTT_ROM_CLK_CTRL0__SOFT_OVERRIDE1_MASK; + + if (temp != data) + WREG32_SMC(ixCGTT_ROM_CLK_CTRL0, data); +} + static int vi_common_set_clockgating_state(void *handle, enum amd_clockgating_state state) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + switch (adev->asic_type) { + case CHIP_FIJI: + fiji_update_bif_medium_grain_light_sleep(adev, + state == AMD_CG_STATE_GATE ? true : false); + fiji_update_hdp_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE ? true : false); + fiji_update_hdp_light_sleep(adev, + state == AMD_CG_STATE_GATE ? true : false); + fiji_update_rom_medium_grain_clock_gating(adev, + state == AMD_CG_STATE_GATE ? true : false); + break; + default: + break; + } return 0; } -- cgit From 9c97e75f0fe6f98285127fb0424862087916e83f Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Fri, 20 Nov 2015 13:33:44 -0500 Subject: amdgpu/powerplay: Add Stoney to list of early init cases Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index 6b46fbfd6be4..4f6740cc1e69 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -81,6 +81,7 @@ static int amdgpu_powerplay_init(struct amdgpu_device *adev) amd_pp->ip_funcs = &fiji_dpm_ip_funcs; break; case CHIP_CARRIZO: + case CHIP_STONEY: amd_pp->ip_funcs = &cz_dpm_ip_funcs; break; default: -- cgit From 7fb72a1fc01cc1a8de533abc80b9eaf0120e8529 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Thu, 19 Nov 2015 13:35:30 +0800 Subject: drm/amd/powerplay: export interface to DAL to init/change display configuration. Signed-off-by: Rex Zhu Signed-off-by: David Rokhvarg Reviewed-by: Alex Deucher Reviewed-by: Jammy Zhou --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 26d9134f7c0d..a3fc43e52483 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1639,6 +1639,7 @@ struct amdgpu_pm { const struct amdgpu_dpm_funcs *funcs; uint32_t pcie_gen_mask; uint32_t pcie_mlw_mask; + struct amd_pp_display_configuration pm_display_cfg;/* set by DAL */ }; void amdgpu_get_pcie_info(struct amdgpu_device *adev); -- cgit From dbd29f0d8513b0d660807beb13df94b00402ea57 Mon Sep 17 00:00:00 2001 From: Jammy Zhou Date: Wed, 21 Oct 2015 17:15:45 +0800 Subject: drm/amdgpu: rename tonga_smumgr.h to tonga_smum.h This conflicts with the tonga_smumgr.h from powerplay in DKMS environement Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher Signed-off-by: Jordan Lazare --- drivers/gpu/drm/amd/amdgpu/tonga_dpm.c | 2 +- drivers/gpu/drm/amd/amdgpu/tonga_smc.c | 2 +- drivers/gpu/drm/amd/amdgpu/tonga_smum.h | 42 +++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h | 42 ------------------------------- 4 files changed, 44 insertions(+), 44 deletions(-) create mode 100644 drivers/gpu/drm/amd/amdgpu/tonga_smum.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c b/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c index 204903897b4f..f4a1346525fe 100644 --- a/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c @@ -24,7 +24,7 @@ #include #include "drmP.h" #include "amdgpu.h" -#include "tonga_smumgr.h" +#include "tonga_smum.h" MODULE_FIRMWARE("amdgpu/tonga_smc.bin"); diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_smc.c b/drivers/gpu/drm/amd/amdgpu/tonga_smc.c index 5421309c1862..361c49a82323 100644 --- a/drivers/gpu/drm/amd/amdgpu/tonga_smc.c +++ b/drivers/gpu/drm/amd/amdgpu/tonga_smc.c @@ -25,7 +25,7 @@ #include "drmP.h" #include "amdgpu.h" #include "tonga_ppsmc.h" -#include "tonga_smumgr.h" +#include "tonga_smum.h" #include "smu_ucode_xfer_vi.h" #include "amdgpu_ucode.h" diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_smum.h b/drivers/gpu/drm/amd/amdgpu/tonga_smum.h new file mode 100644 index 000000000000..c031ff99fe3e --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/tonga_smum.h @@ -0,0 +1,42 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef TONGA_SMUMGR_H +#define TONGA_SMUMGR_H + +#include "tonga_ppsmc.h" + +int tonga_smu_init(struct amdgpu_device *adev); +int tonga_smu_fini(struct amdgpu_device *adev); +int tonga_smu_start(struct amdgpu_device *adev); + +struct tonga_smu_private_data +{ + uint8_t *header; + uint32_t smu_buffer_addr_high; + uint32_t smu_buffer_addr_low; + uint32_t header_addr_high; + uint32_t header_addr_low; +}; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h b/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h deleted file mode 100644 index c031ff99fe3e..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h +++ /dev/null @@ -1,42 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef TONGA_SMUMGR_H -#define TONGA_SMUMGR_H - -#include "tonga_ppsmc.h" - -int tonga_smu_init(struct amdgpu_device *adev); -int tonga_smu_fini(struct amdgpu_device *adev); -int tonga_smu_start(struct amdgpu_device *adev); - -struct tonga_smu_private_data -{ - uint8_t *header; - uint32_t smu_buffer_addr_high; - uint32_t smu_buffer_addr_low; - uint32_t header_addr_high; - uint32_t header_addr_low; -}; - -#endif -- cgit From b57fd5663e34f024406936283475f4c0eadbfc96 Mon Sep 17 00:00:00 2001 From: Jammy Zhou Date: Wed, 21 Oct 2015 17:18:10 +0800 Subject: drm/amdgpu: rename fiji_smumgr.h to fiji_smum.h This conflicts with fiji_smumgr.h from powerplay in DKMS environment Signed-off-by: Jammy Zhou Reviewed-by: Alex Deucher Signed-off-by: Jordan Lazare --- drivers/gpu/drm/amd/amdgpu/fiji_dpm.c | 2 +- drivers/gpu/drm/amd/amdgpu/fiji_smc.c | 2 +- drivers/gpu/drm/amd/amdgpu/fiji_smum.h | 42 ++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h | 42 -------------------------------- 4 files changed, 44 insertions(+), 44 deletions(-) create mode 100644 drivers/gpu/drm/amd/amdgpu/fiji_smum.h delete mode 100644 drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c b/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c index 8f9845d9a986..4b0e45a27129 100644 --- a/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c @@ -24,7 +24,7 @@ #include #include "drmP.h" #include "amdgpu.h" -#include "fiji_smumgr.h" +#include "fiji_smum.h" MODULE_FIRMWARE("amdgpu/fiji_smc.bin"); diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_smc.c b/drivers/gpu/drm/amd/amdgpu/fiji_smc.c index bda1249eb871..e35340afd3db 100644 --- a/drivers/gpu/drm/amd/amdgpu/fiji_smc.c +++ b/drivers/gpu/drm/amd/amdgpu/fiji_smc.c @@ -25,7 +25,7 @@ #include "drmP.h" #include "amdgpu.h" #include "fiji_ppsmc.h" -#include "fiji_smumgr.h" +#include "fiji_smum.h" #include "smu_ucode_xfer_vi.h" #include "amdgpu_ucode.h" diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_smum.h b/drivers/gpu/drm/amd/amdgpu/fiji_smum.h new file mode 100644 index 000000000000..1cef03deeac3 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/fiji_smum.h @@ -0,0 +1,42 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef FIJI_SMUMGR_H +#define FIJI_SMUMGR_H + +#include "fiji_ppsmc.h" + +int fiji_smu_init(struct amdgpu_device *adev); +int fiji_smu_fini(struct amdgpu_device *adev); +int fiji_smu_start(struct amdgpu_device *adev); + +struct fiji_smu_private_data +{ + uint8_t *header; + uint32_t smu_buffer_addr_high; + uint32_t smu_buffer_addr_low; + uint32_t header_addr_high; + uint32_t header_addr_low; +}; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h b/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h deleted file mode 100644 index 1cef03deeac3..000000000000 --- a/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h +++ /dev/null @@ -1,42 +0,0 @@ -/* - * Copyright 2014 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef FIJI_SMUMGR_H -#define FIJI_SMUMGR_H - -#include "fiji_ppsmc.h" - -int fiji_smu_init(struct amdgpu_device *adev); -int fiji_smu_fini(struct amdgpu_device *adev); -int fiji_smu_start(struct amdgpu_device *adev); - -struct fiji_smu_private_data -{ - uint8_t *header; - uint32_t smu_buffer_addr_high; - uint32_t smu_buffer_addr_low; - uint32_t header_addr_high; - uint32_t header_addr_low; -}; - -#endif -- cgit From 9b08a306476d25c9f5721eccbc43e90bb23c5f58 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Mon, 23 Nov 2015 11:20:36 -0500 Subject: drm/amd/amdgpu: add uvd6.0 clock gating support. (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit v2: fix bug in register mask setting. Reviewed-by: Alex Deucher Reviewed-by: Christian König Acked-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c | 261 +++++++++++++++++++++++++++++++++- 1 file changed, 259 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index 121915bbc3b6..3d5913926436 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -279,6 +279,234 @@ static void uvd_v6_0_mc_resume(struct amdgpu_device *adev) WREG32(mmUVD_VCPU_CACHE_SIZE2, size); } +static void cz_set_uvd_clock_gating_branches(struct amdgpu_device *adev, + bool enable) +{ + u32 data, data1; + + data = RREG32(mmUVD_CGC_GATE); + data1 = RREG32(mmUVD_SUVD_CGC_GATE); + if (enable) { + data |= UVD_CGC_GATE__SYS_MASK | + UVD_CGC_GATE__UDEC_MASK | + UVD_CGC_GATE__MPEG2_MASK | + UVD_CGC_GATE__RBC_MASK | + UVD_CGC_GATE__LMI_MC_MASK | + UVD_CGC_GATE__IDCT_MASK | + UVD_CGC_GATE__MPRD_MASK | + UVD_CGC_GATE__MPC_MASK | + UVD_CGC_GATE__LBSI_MASK | + UVD_CGC_GATE__LRBBM_MASK | + UVD_CGC_GATE__UDEC_RE_MASK | + UVD_CGC_GATE__UDEC_CM_MASK | + UVD_CGC_GATE__UDEC_IT_MASK | + UVD_CGC_GATE__UDEC_DB_MASK | + UVD_CGC_GATE__UDEC_MP_MASK | + UVD_CGC_GATE__WCB_MASK | + UVD_CGC_GATE__VCPU_MASK | + UVD_CGC_GATE__SCPU_MASK; + data1 |= UVD_SUVD_CGC_GATE__SRE_MASK | + UVD_SUVD_CGC_GATE__SIT_MASK | + UVD_SUVD_CGC_GATE__SMP_MASK | + UVD_SUVD_CGC_GATE__SCM_MASK | + UVD_SUVD_CGC_GATE__SDB_MASK | + UVD_SUVD_CGC_GATE__SRE_H264_MASK | + UVD_SUVD_CGC_GATE__SRE_HEVC_MASK | + UVD_SUVD_CGC_GATE__SIT_H264_MASK | + UVD_SUVD_CGC_GATE__SIT_HEVC_MASK | + UVD_SUVD_CGC_GATE__SCM_H264_MASK | + UVD_SUVD_CGC_GATE__SCM_HEVC_MASK | + UVD_SUVD_CGC_GATE__SDB_H264_MASK | + UVD_SUVD_CGC_GATE__SDB_HEVC_MASK; + } else { + data &= ~(UVD_CGC_GATE__SYS_MASK | + UVD_CGC_GATE__UDEC_MASK | + UVD_CGC_GATE__MPEG2_MASK | + UVD_CGC_GATE__RBC_MASK | + UVD_CGC_GATE__LMI_MC_MASK | + UVD_CGC_GATE__LMI_UMC_MASK | + UVD_CGC_GATE__IDCT_MASK | + UVD_CGC_GATE__MPRD_MASK | + UVD_CGC_GATE__MPC_MASK | + UVD_CGC_GATE__LBSI_MASK | + UVD_CGC_GATE__LRBBM_MASK | + UVD_CGC_GATE__UDEC_RE_MASK | + UVD_CGC_GATE__UDEC_CM_MASK | + UVD_CGC_GATE__UDEC_IT_MASK | + UVD_CGC_GATE__UDEC_DB_MASK | + UVD_CGC_GATE__UDEC_MP_MASK | + UVD_CGC_GATE__WCB_MASK | + UVD_CGC_GATE__VCPU_MASK | + UVD_CGC_GATE__SCPU_MASK); + data1 &= ~(UVD_SUVD_CGC_GATE__SRE_MASK | + UVD_SUVD_CGC_GATE__SIT_MASK | + UVD_SUVD_CGC_GATE__SMP_MASK | + UVD_SUVD_CGC_GATE__SCM_MASK | + UVD_SUVD_CGC_GATE__SDB_MASK | + UVD_SUVD_CGC_GATE__SRE_H264_MASK | + UVD_SUVD_CGC_GATE__SRE_HEVC_MASK | + UVD_SUVD_CGC_GATE__SIT_H264_MASK | + UVD_SUVD_CGC_GATE__SIT_HEVC_MASK | + UVD_SUVD_CGC_GATE__SCM_H264_MASK | + UVD_SUVD_CGC_GATE__SCM_HEVC_MASK | + UVD_SUVD_CGC_GATE__SDB_H264_MASK | + UVD_SUVD_CGC_GATE__SDB_HEVC_MASK); + } + WREG32(mmUVD_CGC_GATE, data); + WREG32(mmUVD_SUVD_CGC_GATE, data1); +} + +static void tonga_set_uvd_clock_gating_branches(struct amdgpu_device *adev, + bool enable) +{ + u32 data, data1; + + data = RREG32(mmUVD_CGC_GATE); + data1 = RREG32(mmUVD_SUVD_CGC_GATE); + if (enable) { + data |= UVD_CGC_GATE__SYS_MASK | + UVD_CGC_GATE__UDEC_MASK | + UVD_CGC_GATE__MPEG2_MASK | + UVD_CGC_GATE__RBC_MASK | + UVD_CGC_GATE__LMI_MC_MASK | + UVD_CGC_GATE__IDCT_MASK | + UVD_CGC_GATE__MPRD_MASK | + UVD_CGC_GATE__MPC_MASK | + UVD_CGC_GATE__LBSI_MASK | + UVD_CGC_GATE__LRBBM_MASK | + UVD_CGC_GATE__UDEC_RE_MASK | + UVD_CGC_GATE__UDEC_CM_MASK | + UVD_CGC_GATE__UDEC_IT_MASK | + UVD_CGC_GATE__UDEC_DB_MASK | + UVD_CGC_GATE__UDEC_MP_MASK | + UVD_CGC_GATE__WCB_MASK | + UVD_CGC_GATE__VCPU_MASK | + UVD_CGC_GATE__SCPU_MASK; + data1 |= UVD_SUVD_CGC_GATE__SRE_MASK | + UVD_SUVD_CGC_GATE__SIT_MASK | + UVD_SUVD_CGC_GATE__SMP_MASK | + UVD_SUVD_CGC_GATE__SCM_MASK | + UVD_SUVD_CGC_GATE__SDB_MASK; + } else { + data &= ~(UVD_CGC_GATE__SYS_MASK | + UVD_CGC_GATE__UDEC_MASK | + UVD_CGC_GATE__MPEG2_MASK | + UVD_CGC_GATE__RBC_MASK | + UVD_CGC_GATE__LMI_MC_MASK | + UVD_CGC_GATE__LMI_UMC_MASK | + UVD_CGC_GATE__IDCT_MASK | + UVD_CGC_GATE__MPRD_MASK | + UVD_CGC_GATE__MPC_MASK | + UVD_CGC_GATE__LBSI_MASK | + UVD_CGC_GATE__LRBBM_MASK | + UVD_CGC_GATE__UDEC_RE_MASK | + UVD_CGC_GATE__UDEC_CM_MASK | + UVD_CGC_GATE__UDEC_IT_MASK | + UVD_CGC_GATE__UDEC_DB_MASK | + UVD_CGC_GATE__UDEC_MP_MASK | + UVD_CGC_GATE__WCB_MASK | + UVD_CGC_GATE__VCPU_MASK | + UVD_CGC_GATE__SCPU_MASK); + data1 &= ~(UVD_SUVD_CGC_GATE__SRE_MASK | + UVD_SUVD_CGC_GATE__SIT_MASK | + UVD_SUVD_CGC_GATE__SMP_MASK | + UVD_SUVD_CGC_GATE__SCM_MASK | + UVD_SUVD_CGC_GATE__SDB_MASK); + } + WREG32(mmUVD_CGC_GATE, data); + WREG32(mmUVD_SUVD_CGC_GATE, data1); +} + +static void uvd_v6_0_set_uvd_dynamic_clock_mode(struct amdgpu_device *adev, + bool swmode) +{ + u32 data, data1 = 0, data2; + + /* Always un-gate UVD REGS bit */ + data = RREG32(mmUVD_CGC_GATE); + data &= ~(UVD_CGC_GATE__REGS_MASK); + WREG32(mmUVD_CGC_GATE, data); + + data = RREG32(mmUVD_CGC_CTRL); + data &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK | + UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK); + data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK | + 1 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_GATE_DLY_TIMER) | + 4 << REG_FIELD_SHIFT(UVD_CGC_CTRL, CLK_OFF_DELAY); + + data2 = RREG32(mmUVD_SUVD_CGC_CTRL); + if (swmode) { + data &= ~(UVD_CGC_CTRL__UDEC_RE_MODE_MASK | + UVD_CGC_CTRL__UDEC_CM_MODE_MASK | + UVD_CGC_CTRL__UDEC_IT_MODE_MASK | + UVD_CGC_CTRL__UDEC_DB_MODE_MASK | + UVD_CGC_CTRL__UDEC_MP_MODE_MASK | + UVD_CGC_CTRL__SYS_MODE_MASK | + UVD_CGC_CTRL__UDEC_MODE_MASK | + UVD_CGC_CTRL__MPEG2_MODE_MASK | + UVD_CGC_CTRL__REGS_MODE_MASK | + UVD_CGC_CTRL__RBC_MODE_MASK | + UVD_CGC_CTRL__LMI_MC_MODE_MASK | + UVD_CGC_CTRL__LMI_UMC_MODE_MASK | + UVD_CGC_CTRL__IDCT_MODE_MASK | + UVD_CGC_CTRL__MPRD_MODE_MASK | + UVD_CGC_CTRL__MPC_MODE_MASK | + UVD_CGC_CTRL__LBSI_MODE_MASK | + UVD_CGC_CTRL__LRBBM_MODE_MASK | + UVD_CGC_CTRL__WCB_MODE_MASK | + UVD_CGC_CTRL__VCPU_MODE_MASK | + UVD_CGC_CTRL__JPEG_MODE_MASK | + UVD_CGC_CTRL__SCPU_MODE_MASK); + data1 |= UVD_CGC_CTRL2__DYN_OCLK_RAMP_EN_MASK | + UVD_CGC_CTRL2__DYN_RCLK_RAMP_EN_MASK; + data1 &= ~UVD_CGC_CTRL2__GATER_DIV_ID_MASK; + data1 |= 7 << REG_FIELD_SHIFT(UVD_CGC_CTRL2, GATER_DIV_ID); + data2 &= ~(UVD_SUVD_CGC_CTRL__SRE_MODE_MASK | + UVD_SUVD_CGC_CTRL__SIT_MODE_MASK | + UVD_SUVD_CGC_CTRL__SMP_MODE_MASK | + UVD_SUVD_CGC_CTRL__SCM_MODE_MASK | + UVD_SUVD_CGC_CTRL__SDB_MODE_MASK); + } else { + data |= UVD_CGC_CTRL__UDEC_RE_MODE_MASK | + UVD_CGC_CTRL__UDEC_CM_MODE_MASK | + UVD_CGC_CTRL__UDEC_IT_MODE_MASK | + UVD_CGC_CTRL__UDEC_DB_MODE_MASK | + UVD_CGC_CTRL__UDEC_MP_MODE_MASK | + UVD_CGC_CTRL__SYS_MODE_MASK | + UVD_CGC_CTRL__UDEC_MODE_MASK | + UVD_CGC_CTRL__MPEG2_MODE_MASK | + UVD_CGC_CTRL__REGS_MODE_MASK | + UVD_CGC_CTRL__RBC_MODE_MASK | + UVD_CGC_CTRL__LMI_MC_MODE_MASK | + UVD_CGC_CTRL__LMI_UMC_MODE_MASK | + UVD_CGC_CTRL__IDCT_MODE_MASK | + UVD_CGC_CTRL__MPRD_MODE_MASK | + UVD_CGC_CTRL__MPC_MODE_MASK | + UVD_CGC_CTRL__LBSI_MODE_MASK | + UVD_CGC_CTRL__LRBBM_MODE_MASK | + UVD_CGC_CTRL__WCB_MODE_MASK | + UVD_CGC_CTRL__VCPU_MODE_MASK | + UVD_CGC_CTRL__SCPU_MODE_MASK; + data2 |= UVD_SUVD_CGC_CTRL__SRE_MODE_MASK | + UVD_SUVD_CGC_CTRL__SIT_MODE_MASK | + UVD_SUVD_CGC_CTRL__SMP_MODE_MASK | + UVD_SUVD_CGC_CTRL__SCM_MODE_MASK | + UVD_SUVD_CGC_CTRL__SDB_MODE_MASK; + } + WREG32(mmUVD_CGC_CTRL, data); + WREG32(mmUVD_SUVD_CGC_CTRL, data2); + + data = RREG32_UVD_CTX(ixUVD_CGC_CTRL2); + data &= ~(REG_FIELD_MASK(UVD_CGC_CTRL2, DYN_OCLK_RAMP_EN) | + REG_FIELD_MASK(UVD_CGC_CTRL2, DYN_RCLK_RAMP_EN) | + REG_FIELD_MASK(UVD_CGC_CTRL2, GATER_DIV_ID)); + data1 &= (REG_FIELD_MASK(UVD_CGC_CTRL2, DYN_OCLK_RAMP_EN) | + REG_FIELD_MASK(UVD_CGC_CTRL2, DYN_RCLK_RAMP_EN) | + REG_FIELD_MASK(UVD_CGC_CTRL2, GATER_DIV_ID)); + data |= data1; + WREG32_UVD_CTX(ixUVD_CGC_CTRL2, data); +} + /** * uvd_v6_0_start - start UVD block * @@ -303,8 +531,19 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) uvd_v6_0_mc_resume(adev); - /* disable clock gating */ - WREG32(mmUVD_CGC_GATE, 0); + /* Set dynamic clock gating in S/W control mode */ + if (adev->cg_flags & AMDGPU_CG_SUPPORT_UVD_MGCG) { + if (adev->flags & AMD_IS_APU) + cz_set_uvd_clock_gating_branches(adev, false); + else + tonga_set_uvd_clock_gating_branches(adev, false); + uvd_v6_0_set_uvd_dynamic_clock_mode(adev, true); + } else { + /* disable clock gating */ + uint32_t data = RREG32(mmUVD_CGC_CTRL); + data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK; + WREG32(mmUVD_CGC_CTRL, data); + } /* disable interupt */ WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1)); @@ -758,6 +997,24 @@ static int uvd_v6_0_process_interrupt(struct amdgpu_device *adev, static int uvd_v6_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + bool enable = (state == AMD_CG_STATE_GATE) ? true : false; + + if (!(adev->cg_flags & AMDGPU_CG_SUPPORT_UVD_MGCG)) + return 0; + + if (enable) { + if (adev->flags & AMD_IS_APU) + cz_set_uvd_clock_gating_branches(adev, enable); + else + tonga_set_uvd_clock_gating_branches(adev, enable); + uvd_v6_0_set_uvd_dynamic_clock_mode(adev, true); + } else { + uint32_t data = RREG32(mmUVD_CGC_CTRL); + data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK; + WREG32(mmUVD_CGC_CTRL, data); + } + return 0; } -- cgit From 0689a5701358656fe16e83df7c6c654185593208 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Mon, 23 Nov 2015 16:57:53 -0500 Subject: drm/amd/amdgpu: add vce3.0 clock gating support. (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit v2: fix grbm locking Reviewed-by: Alex Deucher Reviewed-by: Christian König Acked-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 149 +++++++++++++++++++++++++++++++++- 1 file changed, 148 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index 370c6c9d81c2..35f48ad7644d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -103,6 +103,108 @@ static void vce_v3_0_ring_set_wptr(struct amdgpu_ring *ring) WREG32(mmVCE_RB_WPTR2, ring->wptr); } +static void vce_v3_0_override_vce_clock_gating(struct amdgpu_device *adev, bool override) +{ + u32 tmp, data; + + tmp = data = RREG32(mmVCE_RB_ARB_CTRL); + if (override) + data |= VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK; + else + data &= ~VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK; + + if (tmp != data) + WREG32(mmVCE_RB_ARB_CTRL, data); +} + +static void vce_v3_0_set_vce_sw_clock_gating(struct amdgpu_device *adev, + bool gated) +{ + u32 tmp, data; + /* Set Override to disable Clock Gating */ + vce_v3_0_override_vce_clock_gating(adev, true); + + if (!gated) { + /* Force CLOCK ON for VCE_CLOCK_GATING_B, + * {*_FORCE_ON, *_FORCE_OFF} = {1, 0} + * VREG can be FORCE ON or set to Dynamic, but can't be OFF + */ + tmp = data = RREG32(mmVCE_CLOCK_GATING_B); + data |= 0x1ff; + data &= ~0xef0000; + if (tmp != data) + WREG32(mmVCE_CLOCK_GATING_B, data); + + /* Force CLOCK ON for VCE_UENC_CLOCK_GATING, + * {*_FORCE_ON, *_FORCE_OFF} = {1, 0} + */ + tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING); + data |= 0x3ff000; + data &= ~0xffc00000; + if (tmp != data) + WREG32(mmVCE_UENC_CLOCK_GATING, data); + + /* set VCE_UENC_CLOCK_GATING_2 */ + tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING_2); + data |= 0x2; + data &= ~0x2; + if (tmp != data) + WREG32(mmVCE_UENC_CLOCK_GATING_2, data); + + /* Force CLOCK ON for VCE_UENC_REG_CLOCK_GATING */ + tmp = data = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + data |= 0x37f; + if (tmp != data) + WREG32(mmVCE_UENC_REG_CLOCK_GATING, data); + + /* Force VCE_UENC_DMA_DCLK_CTRL Clock ON */ + tmp = data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL); + data |= VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | + 0x8; + if (tmp != data) + WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data); + } else { + /* Force CLOCK OFF for VCE_CLOCK_GATING_B, + * {*, *_FORCE_OFF} = {*, 1} + * set VREG to Dynamic, as it can't be OFF + */ + tmp = data = RREG32(mmVCE_CLOCK_GATING_B); + data &= ~0x80010; + data |= 0xe70008; + if (tmp != data) + WREG32(mmVCE_CLOCK_GATING_B, data); + /* Force CLOCK OFF for VCE_UENC_CLOCK_GATING, + * Force ClOCK OFF takes precedent over Force CLOCK ON setting. + * {*_FORCE_ON, *_FORCE_OFF} = {*, 1} + */ + tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING); + data |= 0xffc00000; + if (tmp != data) + WREG32(mmVCE_UENC_CLOCK_GATING, data); + /* Set VCE_UENC_CLOCK_GATING_2 */ + tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING_2); + data |= 0x10000; + if (tmp != data) + WREG32(mmVCE_UENC_CLOCK_GATING_2, data); + /* Set VCE_UENC_REG_CLOCK_GATING to dynamic */ + tmp = data = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + data &= ~0xffc00000; + if (tmp != data) + WREG32(mmVCE_UENC_REG_CLOCK_GATING, data); + /* Set VCE_UENC_DMA_DCLK_CTRL CG always in dynamic mode */ + tmp = data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL); + data &= ~(VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | + 0x8); + if (tmp != data) + WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data); + } + vce_v3_0_override_vce_clock_gating(adev, false); +} + /** * vce_v3_0_start - start VCE block * @@ -121,7 +223,7 @@ static int vce_v3_0_start(struct amdgpu_device *adev) if (adev->vce.harvest_config & (1 << idx)) continue; - if(idx == 0) + if (idx == 0) WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); else @@ -174,6 +276,10 @@ static int vce_v3_0_start(struct amdgpu_device *adev) /* clear BUSY flag */ WREG32_P(mmVCE_STATUS, 0, ~1); + /* Set Clock-Gating off */ + if (adev->cg_flags & AMDGPU_CG_SUPPORT_VCE_MGCG) + vce_v3_0_set_vce_sw_clock_gating(adev, false); + if (r) { DRM_ERROR("VCE not responding, giving up!!!\n"); mutex_unlock(&adev->grbm_idx_mutex); @@ -609,6 +715,47 @@ static int vce_v3_0_process_interrupt(struct amdgpu_device *adev, static int vce_v3_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + bool enable = (state == AMD_CG_STATE_GATE) ? true : false; + int i; + + if (!(adev->cg_flags & AMDGPU_CG_SUPPORT_VCE_MGCG)) + return 0; + + mutex_lock(&adev->grbm_idx_mutex); + for (i = 0; i < 2; i++) { + /* Program VCE Instance 0 or 1 if not harvested */ + if (adev->vce.harvest_config & (1 << i)) + continue; + + if (i == 0) + WREG32_P(mmGRBM_GFX_INDEX, 0, + ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + else + WREG32_P(mmGRBM_GFX_INDEX, + GRBM_GFX_INDEX__VCE_INSTANCE_MASK, + ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + + if (enable) { + /* initialize VCE_CLOCK_GATING_A: Clock ON/OFF delay */ + uint32_t data = RREG32(mmVCE_CLOCK_GATING_A); + data &= ~(0xf | 0xff0); + data |= ((0x0 << 0) | (0x04 << 4)); + WREG32(mmVCE_CLOCK_GATING_A, data); + + /* initialize VCE_UENC_CLOCK_GATING: Clock ON/OFF delay */ + data = RREG32(mmVCE_UENC_CLOCK_GATING); + data &= ~(0xf | 0xff0); + data |= ((0x0 << 0) | (0x04 << 4)); + WREG32(mmVCE_UENC_CLOCK_GATING, data); + } + + vce_v3_0_set_vce_sw_clock_gating(adev, enable); + } + + WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + mutex_unlock(&adev->grbm_idx_mutex); + return 0; } -- cgit From 0bbb81761864c63b5f323b61b48ee892d472c880 Mon Sep 17 00:00:00 2001 From: Eric Huang Date: Tue, 24 Nov 2015 10:53:27 -0500 Subject: drm/amd/amdgpu: enable uvd&vce clock gating for Fiji. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Alex Deucher Reviewed-by: Christian König Acked-by: Jammy Zhou Signed-off-by: Eric Huang --- drivers/gpu/drm/amd/amdgpu/vi.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 67845ee644ad..652e76644c31 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -1442,7 +1442,8 @@ static int vi_common_early_init(void *handle) break; case CHIP_FIJI: adev->has_uvd = true; - adev->cg_flags = 0; + adev->cg_flags = AMDGPU_CG_SUPPORT_UVD_MGCG | + AMDGPU_CG_SUPPORT_VCE_MGCG; adev->pg_flags = 0; adev->external_rev_id = adev->rev_id + 0x3c; break; -- cgit From bd90dd89e53c383d1fea973721c8e7864946a254 Mon Sep 17 00:00:00 2001 From: Qiang Yu Date: Wed, 2 Dec 2015 10:56:57 +0800 Subject: drm/amdgpu: Prepare DKMS build for powerplay module. Signed-off-by: Qiang Yu Reviewed-by: Jammy Zhou --- drivers/gpu/drm/amd/amdgpu/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 16603a0f87e7..66f729eaf00b 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -100,7 +100,7 @@ amdgpu-$(CONFIG_MMU_NOTIFIER) += amdgpu_mn.o ifneq ($(CONFIG_DRM_AMD_POWERPLAY),) -include drivers/gpu/drm/amd/powerplay/Makefile +include $(FULL_AMD_PATH)/powerplay/Makefile amdgpu-y += $(AMD_POWERPLAY_FILES) -- cgit From 7ad4e7f09372946d1bfd5359c45ccce024d0689e Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Mon, 7 Dec 2015 16:42:35 +0800 Subject: drm/amd/powerplay: fix bug that dpm funcs in debugfs/sysfs missing. in dpm module, sysfs init func move to late_init from sw_init. Reviewed-by: Alex Deucher Change-Id: Ice4a73212d8e3106d05f04a27043820ffd32929e Signed-off-by: Rex Zhu --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index 4f6740cc1e69..6cbbae7367f7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -121,6 +121,19 @@ static int amdgpu_pp_early_init(void *handle) return ret; } + +static int amdgpu_pp_late_init(void *handle) +{ + int ret = 0; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (adev->powerplay.ip_funcs->late_init) + ret = adev->powerplay.ip_funcs->late_init( + adev->powerplay.pp_handle); + + return ret; +} + static int amdgpu_pp_sw_init(void *handle) { int ret = 0; @@ -282,7 +295,7 @@ static void amdgpu_pp_print_status(void *handle) const struct amd_ip_funcs amdgpu_pp_ip_funcs = { .early_init = amdgpu_pp_early_init, - .late_init = NULL, + .late_init = amdgpu_pp_late_init, .sw_init = amdgpu_pp_sw_init, .sw_fini = amdgpu_pp_sw_fini, .hw_init = amdgpu_pp_hw_init, -- cgit From 1ea6c1e8e40c5cccc3572d1221c0770fc0c437f3 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Mon, 23 Nov 2015 14:50:10 +0800 Subject: drm/amd/powerplay: check whether enable dpm in powerplay. Change-Id: I0a2dbf8ef7d4a3e9788fe211fc5964dd2487c519 Signed-off-by: Rex Zhu --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index 6cbbae7367f7..b8b4a4781f4d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -145,8 +145,11 @@ static int amdgpu_pp_sw_init(void *handle) #ifdef CONFIG_DRM_AMD_POWERPLAY if (adev->pp_enabled) { - adev->pm.dpm_enabled = true; amdgpu_pm_sysfs_init(adev); + if (amdgpu_dpm == 0) + adev->pm.dpm_enabled = false; + else + adev->pm.dpm_enabled = true; } #endif -- cgit From 898b1dead9a99aeeb103febacf838c7c71d58292 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 8 Dec 2015 17:28:28 -0500 Subject: drm/amdgpu/powerplay: enable sysfs and debugfs interfaces late MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit To avoid users accessing them before the module has finished initializing them and make sure they are only created if dpm has properly initialized. Reviewed-by: Christian König Reviewed-by: Jammy Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index b8b4a4781f4d..ddb90ebd0e7c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -131,6 +131,10 @@ static int amdgpu_pp_late_init(void *handle) ret = adev->powerplay.ip_funcs->late_init( adev->powerplay.pp_handle); +#ifdef CONFIG_DRM_AMD_POWERPLAY + if (adev->pp_enabled) + amdgpu_pm_sysfs_init(adev); +#endif return ret; } @@ -145,7 +149,6 @@ static int amdgpu_pp_sw_init(void *handle) #ifdef CONFIG_DRM_AMD_POWERPLAY if (adev->pp_enabled) { - amdgpu_pm_sysfs_init(adev); if (amdgpu_dpm == 0) adev->pm.dpm_enabled = false; else -- cgit From 53d8eabe3df36015daf40a7a9bfad9b2ffafc6bd Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 21 Dec 2015 17:07:40 -0500 Subject: amd/powerplay: disable powerplay by default initially Hopefully we can enable this by default once we get more upstream feedback on stability, etc. Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index ddb90ebd0e7c..5ee9a0690278 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -101,7 +101,7 @@ static int amdgpu_pp_early_init(void *handle) switch (adev->asic_type) { case CHIP_TONGA: case CHIP_FIJI: - adev->pp_enabled = (amdgpu_powerplay == 0) ? false : true; + adev->pp_enabled = (amdgpu_powerplay > 0) ? true : false; break; default: adev->pp_enabled = (amdgpu_powerplay > 0) ? true : false; -- cgit From b92c26d1808ccd2acac17b660dce750ba9b0a0db Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Mon, 4 Jan 2016 23:43:47 +0300 Subject: drm/amdgpu/cgs: cleanup some indenting This code is indented too far. Also we normally use spaces to align if statement conditions. Signed-off-by: Dan Carpenter Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 6fa0feac27f8..59485d0b3cfb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -843,15 +843,15 @@ static int amdgpu_cgs_acpi_eval_object(void *cgs_device, if (info->input_count > 0) { if (info->pinput_argument == NULL) return -EINVAL; - argument = info->pinput_argument; - func_no = argument->value; - for (i = 0; i < info->input_count; i++) { - if (((argument->type == ACPI_TYPE_STRING) || - (argument->type == ACPI_TYPE_BUFFER)) - && (argument->pointer == NULL)) - return -EINVAL; - argument++; - } + argument = info->pinput_argument; + func_no = argument->value; + for (i = 0; i < info->input_count; i++) { + if (((argument->type == ACPI_TYPE_STRING) || + (argument->type == ACPI_TYPE_BUFFER)) && + (argument->pointer == NULL)) + return -EINVAL; + argument++; + } } if (info->output_count > 0) { -- cgit From e1de741529412f31cb53899307bd63286918c49c Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Tue, 29 Dec 2015 11:57:38 +0800 Subject: drm/amdgpu: fix NULL in vm_grab_id while S3 back vm_manager_fini shouldn't be in suspend phase. Signed-off-by: Chunming Zhou Reviewed-by: Ken Wang --- drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c | 1 - drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 1 - 2 files changed, 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index dababe40a685..3f956065d069 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -1016,7 +1016,6 @@ static int gmc_v7_0_suspend(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; if (adev->vm_manager.enabled) { - amdgpu_vm_manager_fini(adev); gmc_v7_0_vm_fini(adev); adev->vm_manager.enabled = false; } diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index adc25f87fc18..c0c9a0101eb4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -1037,7 +1037,6 @@ static int gmc_v8_0_suspend(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; if (adev->vm_manager.enabled) { - amdgpu_vm_manager_fini(adev); gmc_v8_0_vm_fini(adev); adev->vm_manager.enabled = false; } -- cgit From 1dab5f067e1371e5ed88aa6c47a3cd800f061cfa Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 10:21:46 -0500 Subject: amdgpu/vce3: Cleanup harvest config function. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Basic LOC reduction. Signed-off-by: Tom St Denis Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 20 ++++++-------------- 1 file changed, 6 insertions(+), 14 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index 35f48ad7644d..0de86dea4bfa 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -314,14 +314,11 @@ static int vce_v3_0_start(struct amdgpu_device *adev) static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev) { u32 tmp; - unsigned ret; /* Fiji, Stoney are single pipe */ if ((adev->asic_type == CHIP_FIJI) || - (adev->asic_type == CHIP_STONEY)){ - ret = AMDGPU_VCE_HARVEST_VCE1; - return ret; - } + (adev->asic_type == CHIP_STONEY)) + return AMDGPU_VCE_HARVEST_VCE1; /* Tonga and CZ are dual or single pipe */ if (adev->flags & AMD_IS_APU) @@ -335,19 +332,14 @@ static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev) switch (tmp) { case 1: - ret = AMDGPU_VCE_HARVEST_VCE0; - break; + return AMDGPU_VCE_HARVEST_VCE0; case 2: - ret = AMDGPU_VCE_HARVEST_VCE1; - break; + return AMDGPU_VCE_HARVEST_VCE1; case 3: - ret = AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1; - break; + return AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1; default: - ret = 0; + return 0; } - - return ret; } static int vce_v3_0_early_init(void *handle) -- cgit From 92988e604b8a7c7d541396905433021eedfc28bb Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 10:35:56 -0500 Subject: amdgpu/vce3: Simplify idle and wait for idle code MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit More LOC reductions in VCE3 code. This patch simplifies the is_idle and wait_for_idle logic. Signed-off-by: Tom St Denis Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 30 +++++------------------------- 1 file changed, 5 insertions(+), 25 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index 0de86dea4bfa..f20529d5cb84 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -535,17 +535,9 @@ static bool vce_v3_0_is_idle(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 mask = 0; - int idx; - - for (idx = 0; idx < 2; ++idx) { - if (adev->vce.harvest_config & (1 << idx)) - continue; - if (idx == 0) - mask |= SRBM_STATUS2__VCE0_BUSY_MASK; - else - mask |= SRBM_STATUS2__VCE1_BUSY_MASK; - } + mask |= (adev->vce.harvest_config & (1<<0)) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK; + mask |= (adev->vce.harvest_config & (1<<1)) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK; return !(RREG32(mmSRBM_STATUS2) & mask); } @@ -554,23 +546,11 @@ static int vce_v3_0_wait_for_idle(void *handle) { unsigned i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - u32 mask = 0; - int idx; - for (idx = 0; idx < 2; ++idx) { - if (adev->vce.harvest_config & (1 << idx)) - continue; - - if (idx == 0) - mask |= SRBM_STATUS2__VCE0_BUSY_MASK; - else - mask |= SRBM_STATUS2__VCE1_BUSY_MASK; - } - - for (i = 0; i < adev->usec_timeout; i++) { - if (!(RREG32(mmSRBM_STATUS2) & mask)) + for (i = 0; i < adev->usec_timeout; i++) + if (vce_v3_0_is_idle(handle)) return 0; - } + return -ETIMEDOUT; } -- cgit From af18b0f7fb635e5b18d541bf05058723412d749f Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 10:40:16 -0500 Subject: amdgpu/vce3: Simplify vce_v3_0_soft_reset() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit LOC reduction and simplification. Signed-off-by: Tom St Denis Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index f20529d5cb84..d50db761e327 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -558,17 +558,10 @@ static int vce_v3_0_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 mask = 0; - int idx; - for (idx = 0; idx < 2; ++idx) { - if (adev->vce.harvest_config & (1 << idx)) - continue; + mask |= (adev->vce.harvest_config & (1<<0)) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK; + mask |= (adev->vce.harvest_config & (1<<1)) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK; - if (idx == 0) - mask |= SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK; - else - mask |= SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK; - } WREG32_P(mmSRBM_SOFT_RESET, mask, ~(SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK | SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK)); -- cgit From 81da2edef935adc6091deb20963c29d8463f1440 Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 10:42:39 -0500 Subject: amdgpu/vce3: Simplify vce_v3_0_process_interrupt() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Fold two cases into one for a LOC reduction. Signed-off-by: Tom St Denis Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index d50db761e327..d3e7ba60d99c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -663,10 +663,8 @@ static int vce_v3_0_process_interrupt(struct amdgpu_device *adev, switch (entry->src_data) { case 0: - amdgpu_fence_process(&adev->vce.ring[0]); - break; case 1: - amdgpu_fence_process(&adev->vce.ring[1]); + amdgpu_fence_process(&adev->vce.ring[entry->src_data]); break; default: DRM_ERROR("Unhandled interrupt: %d %d\n", -- cgit From 74af12762de65cbbc11e11501dacd652fae75de9 Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Mon, 4 Jan 2016 10:46:41 -0500 Subject: amdgpu/vce3: Remove magic constants from harvest register masks. Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index d3e7ba60d99c..ad56b1f49b09 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -536,8 +536,8 @@ static bool vce_v3_0_is_idle(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 mask = 0; - mask |= (adev->vce.harvest_config & (1<<0)) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK; - mask |= (adev->vce.harvest_config & (1<<1)) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK; + mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK; + mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK; return !(RREG32(mmSRBM_STATUS2) & mask); } @@ -559,8 +559,8 @@ static int vce_v3_0_soft_reset(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 mask = 0; - mask |= (adev->vce.harvest_config & (1<<0)) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK; - mask |= (adev->vce.harvest_config & (1<<1)) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK; + mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK; + mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK; WREG32_P(mmSRBM_SOFT_RESET, mask, ~(SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK | -- cgit From 691ca86a3a15a9c5054edb3c517a1aeb40e9fadc Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 10:55:34 -0500 Subject: amdgpu/vce3: Simplify vce_v3_0_hw_init and ensure both rings default to not ready. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Simplified the ring test and added logic to ensure rings are marked not ready by default. Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher Reviewed-by: Christian König --- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 24 +++++++++--------------- 1 file changed, 9 insertions(+), 15 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index ad56b1f49b09..e99af81e4aec 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -414,28 +414,22 @@ static int vce_v3_0_sw_fini(void *handle) static int vce_v3_0_hw_init(void *handle) { - struct amdgpu_ring *ring; - int r; + int r, i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; r = vce_v3_0_start(adev); if (r) return r; - ring = &adev->vce.ring[0]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; - return r; - } + adev->vce.ring[0].ready = false; + adev->vce.ring[1].ready = false; - ring = &adev->vce.ring[1]; - ring->ready = true; - r = amdgpu_ring_test_ring(ring); - if (r) { - ring->ready = false; - return r; + for (i = 0; i < 2; i++) { + r = amdgpu_ring_test_ring(&adev->vce.ring[i]); + if (r) + return r; + else + adev->vce.ring[i].ready = true; } DRM_INFO("VCE initialized successfully.\n"); -- cgit From 9e4e1ae82f2479a32bee61132206367a710e961e Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 13:01:49 -0500 Subject: amdgpu/dce11: Remove division from dce_v11_0_vblank_wait() Mimics odd behaviour where (i++ % 100 == 0) is true in the first iteration of each loop... Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/dce_v11_0.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 8701661a8868..80be62a0b99b 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -211,7 +211,7 @@ static bool dce_v11_0_is_counter_moving(struct amdgpu_device *adev, int crtc) */ static void dce_v11_0_vblank_wait(struct amdgpu_device *adev, int crtc) { - unsigned i = 0; + unsigned i = 100; if (crtc >= adev->mode_info.num_crtc) return; @@ -223,14 +223,16 @@ static void dce_v11_0_vblank_wait(struct amdgpu_device *adev, int crtc) * wait for another frame. */ while (dce_v11_0_is_in_vblank(adev, crtc)) { - if (i++ % 100 == 0) { + if (i++ == 100) { + i = 0; if (!dce_v11_0_is_counter_moving(adev, crtc)) break; } } while (!dce_v11_0_is_in_vblank(adev, crtc)) { - if (i++ % 100 == 0) { + if (i++ == 100) { + i = 0; if (!dce_v11_0_is_counter_moving(adev, crtc)) break; } -- cgit From 15c3277ff60286860b3dc98651fc241875e678fa Mon Sep 17 00:00:00 2001 From: Tom St Denis Date: Tue, 15 Dec 2015 13:03:43 -0500 Subject: amdgpu/dce11: Add test for crtc < 0 to various DCEv11 functions To be consistent with other DCE11 functions test for crtc < 0. Signed-off-by: Tom St Denis Reviewed-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/dce_v11_0.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 80be62a0b99b..8e67249d4367 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -213,7 +213,7 @@ static void dce_v11_0_vblank_wait(struct amdgpu_device *adev, int crtc) { unsigned i = 100; - if (crtc >= adev->mode_info.num_crtc) + if (crtc < 0 || crtc >= adev->mode_info.num_crtc) return; if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK)) @@ -241,7 +241,7 @@ static void dce_v11_0_vblank_wait(struct amdgpu_device *adev, int crtc) static u32 dce_v11_0_vblank_get_counter(struct amdgpu_device *adev, int crtc) { - if (crtc >= adev->mode_info.num_crtc) + if (crtc < 0 || crtc >= adev->mode_info.num_crtc) return 0; else return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]); @@ -3386,7 +3386,7 @@ static void dce_v11_0_crtc_vblank_int_ack(struct amdgpu_device *adev, { u32 tmp; - if (crtc >= adev->mode_info.num_crtc) { + if (crtc < 0 || crtc >= adev->mode_info.num_crtc) { DRM_DEBUG("invalid crtc %d\n", crtc); return; } @@ -3401,7 +3401,7 @@ static void dce_v11_0_crtc_vline_int_ack(struct amdgpu_device *adev, { u32 tmp; - if (crtc >= adev->mode_info.num_crtc) { + if (crtc < 0 || crtc >= adev->mode_info.num_crtc) { DRM_DEBUG("invalid crtc %d\n", crtc); return; } -- cgit From 75ac63dbc3b0f4d3af67a5857790749e954e2ba6 Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Wed, 6 Jan 2016 16:38:48 +0800 Subject: drm/amd/powerplay: fix Smatch static checker warnings with indenting (v2) v2: AGD: rebase on upstream Signed-off-by: Rex Zhu Reviewed-by: Alex Deucher Reviewed-by: Ken Wang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 3b78982abaf1..4386cbac7f97 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -807,7 +807,7 @@ void amdgpu_pm_compute_clocks(struct amdgpu_device *adev) struct amdgpu_ring *ring = adev->rings[i]; if (ring && ring->ready) amdgpu_fence_wait_empty(ring); - } + } mutex_unlock(&adev->ring_lock); amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_DISPLAY_CONFIG_CHANGE, NULL, NULL); -- cgit From 9354573d76f599e05a34e0b468ffce681769115f Mon Sep 17 00:00:00 2001 From: Rex Zhu Date: Wed, 6 Jan 2016 17:08:46 +0800 Subject: drm/amdgpu: Show gpu load when display gpu performance for Ci. Reviewed-by: Alex Deucher Signed-off-by: Rex Zhu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/ci_dpm.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c index 57a2e347f04d..8b4731d4e10e 100644 --- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c @@ -1395,7 +1395,6 @@ static void ci_thermal_stop_thermal_controller(struct amdgpu_device *adev) ci_fan_ctrl_set_default_mode(adev); } -#if 0 static int ci_read_smc_soft_register(struct amdgpu_device *adev, u16 reg_offset, u32 *value) { @@ -1405,7 +1404,6 @@ static int ci_read_smc_soft_register(struct amdgpu_device *adev, pi->soft_regs_start + reg_offset, value, pi->sram_end); } -#endif static int ci_write_smc_soft_register(struct amdgpu_device *adev, u16 reg_offset, u32 value) @@ -6084,11 +6082,23 @@ ci_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev, struct amdgpu_ps *rps = &pi->current_rps; u32 sclk = ci_get_average_sclk_freq(adev); u32 mclk = ci_get_average_mclk_freq(adev); + u32 activity_percent = 50; + int ret; + + ret = ci_read_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, AverageGraphicsA), + &activity_percent); + + if (ret == 0) { + activity_percent += 0x80; + activity_percent >>= 8; + activity_percent = activity_percent > 100 ? 100 : activity_percent; + } seq_printf(m, "uvd %sabled\n", pi->uvd_enabled ? "en" : "dis"); seq_printf(m, "vce %sabled\n", rps->vce_active ? "en" : "dis"); seq_printf(m, "power level avg sclk: %u mclk: %u\n", sclk, mclk); + seq_printf(m, "GPU load: %u %%\n", activity_percent); } static void ci_dpm_print_power_state(struct amdgpu_device *adev, -- cgit From 191caba6cc4c95ba282304a10a2a9cba85450317 Mon Sep 17 00:00:00 2001 From: Christian König Date: Tue, 15 Dec 2015 11:10:30 +0100 Subject: drm/amdgpu: add warning to amdgpu_bo_gpu_offset() v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Check if there really is a valid offset for the BO. v2: user WARN_ON_ONCE Signed-off-by: Christian König Reviewed-by: Michel Dänzer Reviewed-by: Alex Deucher (v1) Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index ea756e77b023..5107fb291bdb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -96,6 +96,7 @@ static inline void amdgpu_bo_unreserve(struct amdgpu_bo *bo) */ static inline u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo) { + WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM); return bo->tbo.offset; } -- cgit From ba228ac8f512c9cd09cb4245c424ab1632da0c24 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 23 Dec 2015 11:25:43 -0500 Subject: drm/amdgpu/cgs: add an interface to access PCI resources MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This provides an interface to get access to the base address of PCI resources (MMIO, DOORBELL, etc.). Only MMIO and DOORBELL are implemented right now. This is necessary to properly utilize shared drivers on platform devices. IP modules can use this interface to get the base address of the resource and add any additional offset and set the size when setting up the platform driver(s). Acked-by: Dave Airlie Acked-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c | 36 +++++++++++++++++++++++++++++++++ 1 file changed, 36 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c index 59485d0b3cfb..a081dda9fa2f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c @@ -398,6 +398,41 @@ static void amdgpu_cgs_write_pci_config_dword(void *cgs_device, unsigned addr, WARN(ret, "pci_write_config_dword error"); } + +static int amdgpu_cgs_get_pci_resource(void *cgs_device, + enum cgs_resource_type resource_type, + uint64_t size, + uint64_t offset, + uint64_t *resource_base) +{ + CGS_FUNC_ADEV; + + if (resource_base == NULL) + return -EINVAL; + + switch (resource_type) { + case CGS_RESOURCE_TYPE_MMIO: + if (adev->rmmio_size == 0) + return -ENOENT; + if ((offset + size) > adev->rmmio_size) + return -EINVAL; + *resource_base = adev->rmmio_base; + return 0; + case CGS_RESOURCE_TYPE_DOORBELL: + if (adev->doorbell.size == 0) + return -ENOENT; + if ((offset + size) > adev->doorbell.size) + return -EINVAL; + *resource_base = adev->doorbell.base; + return 0; + case CGS_RESOURCE_TYPE_FB: + case CGS_RESOURCE_TYPE_IO: + case CGS_RESOURCE_TYPE_ROM: + default: + return -EINVAL; + } +} + static const void *amdgpu_cgs_atom_get_data_table(void *cgs_device, unsigned table, uint16_t *size, uint8_t *frev, uint8_t *crev) @@ -1041,6 +1076,7 @@ static const struct cgs_ops amdgpu_cgs_ops = { amdgpu_cgs_write_pci_config_byte, amdgpu_cgs_write_pci_config_word, amdgpu_cgs_write_pci_config_dword, + amdgpu_cgs_get_pci_resource, amdgpu_cgs_atom_get_data_table, amdgpu_cgs_atom_get_cmd_table_revs, amdgpu_cgs_atom_exec_cmd_table, -- cgit From 5f2323658e4829ffb893553297e64795a90cbcd9 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 6 Nov 2015 01:29:08 -0500 Subject: drm/amdgpu: add irq domain support MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Hardware blocks on the GPU like ACP generate interrupts in the GPU interrupt controller, but are driven by a separate driver. Add an irq domain to the GPU driver so that blocks like ACP can register a Linux interrupt. Acked-by: Dave Airlie Acked-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 108 +++++++++++++++++++++++++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h | 9 +++ drivers/gpu/drm/amd/amdgpu/cik_ih.c | 6 ++ drivers/gpu/drm/amd/amdgpu/cz_ih.c | 7 +++ drivers/gpu/drm/amd/amdgpu/iceland_ih.c | 7 +++ drivers/gpu/drm/amd/amdgpu/tonga_ih.c | 7 +++ 6 files changed, 136 insertions(+), 8 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 7c42ff670080..3006182c5d0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -312,6 +312,7 @@ int amdgpu_irq_add_id(struct amdgpu_device *adev, unsigned src_id, } adev->irq.sources[src_id] = source; + return 0; } @@ -335,15 +336,19 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev, return; } - src = adev->irq.sources[src_id]; - if (!src) { - DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id); - return; - } + if (adev->irq.virq[src_id]) { + generic_handle_irq(irq_find_mapping(adev->irq.domain, src_id)); + } else { + src = adev->irq.sources[src_id]; + if (!src) { + DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id); + return; + } - r = src->funcs->process(adev, src, entry); - if (r) - DRM_ERROR("error processing interrupt (%d)\n", r); + r = src->funcs->process(adev, src, entry); + if (r) + DRM_ERROR("error processing interrupt (%d)\n", r); + } } /** @@ -461,3 +466,90 @@ bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src, return !!atomic_read(&src->enabled_types[type]); } + +/* gen irq */ +static void amdgpu_irq_mask(struct irq_data *irqd) +{ + /* XXX */ +} + +static void amdgpu_irq_unmask(struct irq_data *irqd) +{ + /* XXX */ +} + +static struct irq_chip amdgpu_irq_chip = { + .name = "amdgpu-ih", + .irq_mask = amdgpu_irq_mask, + .irq_unmask = amdgpu_irq_unmask, +}; + +static int amdgpu_irqdomain_map(struct irq_domain *d, + unsigned int irq, irq_hw_number_t hwirq) +{ + if (hwirq >= AMDGPU_MAX_IRQ_SRC_ID) + return -EPERM; + + irq_set_chip_and_handler(irq, + &amdgpu_irq_chip, handle_simple_irq); + return 0; +} + +static struct irq_domain_ops amdgpu_hw_irqdomain_ops = { + .map = amdgpu_irqdomain_map, +}; + +/** + * amdgpu_irq_add_domain - create a linear irq domain + * + * @adev: amdgpu device pointer + * + * Create an irq domain for GPU interrupt sources + * that may be driven by another driver (e.g., ACP). + */ +int amdgpu_irq_add_domain(struct amdgpu_device *adev) +{ + adev->irq.domain = irq_domain_add_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID, + &amdgpu_hw_irqdomain_ops, adev); + if (!adev->irq.domain) { + DRM_ERROR("GPU irq add domain failed\n"); + return -ENODEV; + } + + return 0; +} + +/** + * amdgpu_irq_remove_domain - remove the irq domain + * + * @adev: amdgpu device pointer + * + * Remove the irq domain for GPU interrupt sources + * that may be driven by another driver (e.g., ACP). + */ +void amdgpu_irq_remove_domain(struct amdgpu_device *adev) +{ + if (adev->irq.domain) { + irq_domain_remove(adev->irq.domain); + adev->irq.domain = NULL; + } +} + +/** + * amdgpu_irq_create_mapping - create a mapping between a domain irq and a + * Linux irq + * + * @adev: amdgpu device pointer + * @src_id: IH source id + * + * Create a mapping between a domain irq (GPU IH src id) and a Linux irq + * Use this for components that generate a GPU interrupt, but are driven + * by a different driver (e.g., ACP). + * Returns the Linux irq. + */ +unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id) +{ + adev->irq.virq[src_id] = irq_create_mapping(adev->irq.domain, src_id); + + return adev->irq.virq[src_id]; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h index 17b01aef4278..e124b59f39c1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h @@ -24,6 +24,7 @@ #ifndef __AMDGPU_IRQ_H__ #define __AMDGPU_IRQ_H__ +#include #include "amdgpu_ih.h" #define AMDGPU_MAX_IRQ_SRC_ID 0x100 @@ -65,6 +66,10 @@ struct amdgpu_irq { /* interrupt ring */ struct amdgpu_ih_ring ih; const struct amdgpu_ih_funcs *ih_funcs; + + /* gen irq stuff */ + struct irq_domain *domain; /* GPU irq controller domain */ + unsigned virq[AMDGPU_MAX_IRQ_SRC_ID]; }; void amdgpu_irq_preinstall(struct drm_device *dev); @@ -90,4 +95,8 @@ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src, bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type); +int amdgpu_irq_add_domain(struct amdgpu_device *adev); +void amdgpu_irq_remove_domain(struct amdgpu_device *adev); +unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c index 8993c50cb89f..30c9b3beeef9 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c @@ -274,6 +274,11 @@ static void cik_ih_set_rptr(struct amdgpu_device *adev) static int cik_ih_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int ret; + + ret = amdgpu_irq_add_domain(adev); + if (ret) + return ret; cik_ih_set_interrupt_funcs(adev); @@ -300,6 +305,7 @@ static int cik_ih_sw_fini(void *handle) amdgpu_irq_fini(adev); amdgpu_ih_ring_fini(adev); + amdgpu_irq_remove_domain(adev); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c index bc751bfbcae2..c79638f8e732 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c @@ -253,8 +253,14 @@ static void cz_ih_set_rptr(struct amdgpu_device *adev) static int cz_ih_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int ret; + + ret = amdgpu_irq_add_domain(adev); + if (ret) + return ret; cz_ih_set_interrupt_funcs(adev); + return 0; } @@ -278,6 +284,7 @@ static int cz_ih_sw_fini(void *handle) amdgpu_irq_fini(adev); amdgpu_ih_ring_fini(adev); + amdgpu_irq_remove_domain(adev); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c index 779532d350ff..679e7394a495 100644 --- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c @@ -253,8 +253,14 @@ static void iceland_ih_set_rptr(struct amdgpu_device *adev) static int iceland_ih_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int ret; + + ret = amdgpu_irq_add_domain(adev); + if (ret) + return ret; iceland_ih_set_interrupt_funcs(adev); + return 0; } @@ -278,6 +284,7 @@ static int iceland_ih_sw_fini(void *handle) amdgpu_irq_fini(adev); amdgpu_ih_ring_fini(adev); + amdgpu_irq_remove_domain(adev); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c index 743c372837aa..b6f7d7bff929 100644 --- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c @@ -273,8 +273,14 @@ static void tonga_ih_set_rptr(struct amdgpu_device *adev) static int tonga_ih_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int ret; + + ret = amdgpu_irq_add_domain(adev); + if (ret) + return ret; tonga_ih_set_interrupt_funcs(adev); + return 0; } @@ -301,6 +307,7 @@ static int tonga_ih_sw_fini(void *handle) amdgpu_irq_fini(adev); amdgpu_ih_ring_fini(adev); + amdgpu_irq_add_domain(adev); return 0; } -- cgit From 888c9e33e4c5a503285921046c621f7c73199d2f Mon Sep 17 00:00:00 2001 From: Chunming Zhou Date: Wed, 13 Jan 2016 12:55:18 +0800 Subject: drm/amdgpu: fix lost sync_to if scheduler is enabled. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit when scheduler is enabled, the semaphore isn't used at all. Signed-off-by: Chunming Zhou Reviewed-by: Christian König Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c index dd005c336c97..181ce39ef5e5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c @@ -293,7 +293,8 @@ int amdgpu_sync_rings(struct amdgpu_sync *sync, fence = to_amdgpu_fence(sync->sync_to[i]); /* check if we really need to sync */ - if (!amdgpu_fence_need_sync(fence, ring)) + if (!amdgpu_enable_scheduler && + !amdgpu_fence_need_sync(fence, ring)) continue; /* prevent GPU deadlocks */ @@ -303,7 +304,7 @@ int amdgpu_sync_rings(struct amdgpu_sync *sync, } if (amdgpu_enable_scheduler || !amdgpu_enable_semaphores) { - r = fence_wait(&fence->base, true); + r = fence_wait(sync->sync_to[i], true); if (r) return r; continue; -- cgit From d83b1e8132f91ba9e038b5206dc81d32ecac1f75 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 18 Dec 2015 11:06:42 -0500 Subject: drm/amdgpu/cz: add code to enable forcing UVD clocks MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit UVD DPM works similarly to SCLK DPM. Add a similar interface for UVD for forcing the UVD clocks. Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cz_dpm.c | 129 ++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/cz_dpm.h | 1 + 2 files changed, 130 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c index 8035d4d6a4f5..5ccea9f9c634 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c @@ -1078,6 +1078,37 @@ static uint32_t cz_get_eclk_level(struct amdgpu_device *adev, return i; } +static uint32_t cz_get_uvd_level(struct amdgpu_device *adev, + uint32_t clock, uint16_t msg) +{ + int i = 0; + struct amdgpu_uvd_clock_voltage_dependency_table *table = + &adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table; + + switch (msg) { + case PPSMC_MSG_SetUvdSoftMin: + case PPSMC_MSG_SetUvdHardMin: + for (i = 0; i < table->count; i++) + if (clock <= table->entries[i].vclk) + break; + if (i == table->count) + i = table->count - 1; + break; + case PPSMC_MSG_SetUvdSoftMax: + case PPSMC_MSG_SetUvdHardMax: + for (i = table->count - 1; i >= 0; i--) + if (clock >= table->entries[i].vclk) + break; + if (i < 0) + i = 0; + break; + default: + break; + } + + return i; +} + static int cz_program_bootup_state(struct amdgpu_device *adev) { struct cz_power_info *pi = cz_get_pi(adev); @@ -1739,6 +1770,104 @@ static int cz_dpm_unforce_dpm_levels(struct amdgpu_device *adev) return 0; } + +static int cz_dpm_uvd_force_highest(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + int ret = 0; + + if (pi->uvd_dpm.soft_min_clk != pi->uvd_dpm.soft_max_clk) { + pi->uvd_dpm.soft_min_clk = + pi->uvd_dpm.soft_max_clk; + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetUvdSoftMin, + cz_get_uvd_level(adev, + pi->uvd_dpm.soft_min_clk, + PPSMC_MSG_SetUvdSoftMin)); + if (ret) + return ret; + } + + return ret; +} + +static int cz_dpm_uvd_force_lowest(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + int ret = 0; + + if (pi->uvd_dpm.soft_max_clk != pi->uvd_dpm.soft_min_clk) { + pi->uvd_dpm.soft_max_clk = pi->uvd_dpm.soft_min_clk; + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetUvdSoftMax, + cz_get_uvd_level(adev, + pi->uvd_dpm.soft_max_clk, + PPSMC_MSG_SetUvdSoftMax)); + if (ret) + return ret; + } + + return ret; +} + +static uint32_t cz_dpm_get_max_uvd_level(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + + if (!pi->max_uvd_level) { + cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxUvdLevel); + pi->max_uvd_level = cz_get_argument(adev) + 1; + } + + if (pi->max_uvd_level > CZ_MAX_HARDWARE_POWERLEVELS) { + DRM_ERROR("Invalid max uvd level!\n"); + return -EINVAL; + } + + return pi->max_uvd_level; +} + +static int cz_dpm_unforce_uvd_dpm_levels(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + struct amdgpu_uvd_clock_voltage_dependency_table *dep_table = + &adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table; + uint32_t level = 0; + int ret = 0; + + pi->uvd_dpm.soft_min_clk = dep_table->entries[0].vclk; + level = cz_dpm_get_max_uvd_level(adev) - 1; + if (level < dep_table->count) + pi->uvd_dpm.soft_max_clk = dep_table->entries[level].vclk; + else + pi->uvd_dpm.soft_max_clk = + dep_table->entries[dep_table->count - 1].vclk; + + /* get min/max sclk soft value + * notify SMU to execute */ + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetUvdSoftMin, + cz_get_uvd_level(adev, + pi->uvd_dpm.soft_min_clk, + PPSMC_MSG_SetUvdSoftMin)); + if (ret) + return ret; + + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetUvdSoftMax, + cz_get_uvd_level(adev, + pi->uvd_dpm.soft_max_clk, + PPSMC_MSG_SetUvdSoftMax)); + if (ret) + return ret; + + DRM_DEBUG("DPM uvd unforce state min=%d, max=%d.\n", + pi->uvd_dpm.soft_min_clk, + pi->uvd_dpm.soft_max_clk); + + return 0; +} + static int cz_dpm_force_dpm_level(struct amdgpu_device *adev, enum amdgpu_dpm_forced_level level) { diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.h b/drivers/gpu/drm/amd/amdgpu/cz_dpm.h index 99e1afc89629..6a6a6fe7967f 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.h +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.h @@ -183,6 +183,7 @@ struct cz_power_info { uint32_t voltage_drop_threshold; uint32_t gfx_pg_threshold; uint32_t max_sclk_level; + uint32_t max_uvd_level; /* flags */ bool didt_enabled; bool video_start; -- cgit From 044c0629b55d73235161557a48870d663d0072f5 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 18 Dec 2015 11:25:16 -0500 Subject: drm/amdgpu/cz: add code to enable forcing VCE clocks MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit VCE DPM works similarly to SCLK DPM. Add a similar interface for VCE for forcing the VCE clocks. Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cz_dpm.c | 98 ++++++++++++++++++++++++++++++++++++- drivers/gpu/drm/amd/amdgpu/cz_dpm.h | 1 + 2 files changed, 98 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c index 5ccea9f9c634..02cba49f13a8 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c @@ -1770,7 +1770,6 @@ static int cz_dpm_unforce_dpm_levels(struct amdgpu_device *adev) return 0; } - static int cz_dpm_uvd_force_highest(struct amdgpu_device *adev) { struct cz_power_info *pi = cz_get_pi(adev); @@ -1868,6 +1867,103 @@ static int cz_dpm_unforce_uvd_dpm_levels(struct amdgpu_device *adev) return 0; } +static int cz_dpm_vce_force_highest(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + int ret = 0; + + if (pi->vce_dpm.soft_min_clk != pi->vce_dpm.soft_max_clk) { + pi->vce_dpm.soft_min_clk = + pi->vce_dpm.soft_max_clk; + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetEclkSoftMin, + cz_get_eclk_level(adev, + pi->vce_dpm.soft_min_clk, + PPSMC_MSG_SetEclkSoftMin)); + if (ret) + return ret; + } + + return ret; +} + +static int cz_dpm_vce_force_lowest(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + int ret = 0; + + if (pi->vce_dpm.soft_max_clk != pi->vce_dpm.soft_min_clk) { + pi->vce_dpm.soft_max_clk = pi->vce_dpm.soft_min_clk; + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetEclkSoftMax, + cz_get_uvd_level(adev, + pi->vce_dpm.soft_max_clk, + PPSMC_MSG_SetEclkSoftMax)); + if (ret) + return ret; + } + + return ret; +} + +static uint32_t cz_dpm_get_max_vce_level(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + + if (!pi->max_vce_level) { + cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxEclkLevel); + pi->max_vce_level = cz_get_argument(adev) + 1; + } + + if (pi->max_vce_level > CZ_MAX_HARDWARE_POWERLEVELS) { + DRM_ERROR("Invalid max vce level!\n"); + return -EINVAL; + } + + return pi->max_vce_level; +} + +static int cz_dpm_unforce_vce_dpm_levels(struct amdgpu_device *adev) +{ + struct cz_power_info *pi = cz_get_pi(adev); + struct amdgpu_vce_clock_voltage_dependency_table *dep_table = + &adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table; + uint32_t level = 0; + int ret = 0; + + pi->vce_dpm.soft_min_clk = dep_table->entries[0].ecclk; + level = cz_dpm_get_max_vce_level(adev) - 1; + if (level < dep_table->count) + pi->vce_dpm.soft_max_clk = dep_table->entries[level].ecclk; + else + pi->vce_dpm.soft_max_clk = + dep_table->entries[dep_table->count - 1].ecclk; + + /* get min/max sclk soft value + * notify SMU to execute */ + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetEclkSoftMin, + cz_get_eclk_level(adev, + pi->vce_dpm.soft_min_clk, + PPSMC_MSG_SetEclkSoftMin)); + if (ret) + return ret; + + ret = cz_send_msg_to_smc_with_parameter(adev, + PPSMC_MSG_SetEclkSoftMax, + cz_get_eclk_level(adev, + pi->vce_dpm.soft_max_clk, + PPSMC_MSG_SetEclkSoftMax)); + if (ret) + return ret; + + DRM_DEBUG("DPM vce unforce state min=%d, max=%d.\n", + pi->vce_dpm.soft_min_clk, + pi->vce_dpm.soft_max_clk); + + return 0; +} + static int cz_dpm_force_dpm_level(struct amdgpu_device *adev, enum amdgpu_dpm_forced_level level) { diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.h b/drivers/gpu/drm/amd/amdgpu/cz_dpm.h index 6a6a6fe7967f..5df8c1faab51 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.h +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.h @@ -184,6 +184,7 @@ struct cz_power_info { uint32_t gfx_pg_threshold; uint32_t max_sclk_level; uint32_t max_uvd_level; + uint32_t max_vce_level; /* flags */ bool didt_enabled; bool video_start; -- cgit From 5f57642672dd90e3183efcd9e6afbff1305a133a Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 18 Dec 2015 11:28:49 -0500 Subject: drm/amdgpu/cz: force uvd clocks when sclks are forced MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cz_dpm.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c index 02cba49f13a8..bdf5a22ba450 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c @@ -1971,23 +1971,47 @@ static int cz_dpm_force_dpm_level(struct amdgpu_device *adev, switch (level) { case AMDGPU_DPM_FORCED_LEVEL_HIGH: + /* sclk */ ret = cz_dpm_unforce_dpm_levels(adev); if (ret) return ret; ret = cz_dpm_force_highest(adev); + if (ret) + return ret; + + /* uvd */ + ret = cz_dpm_unforce_uvd_dpm_levels(adev); + if (ret) + return ret; + ret = cz_dpm_uvd_force_highest(adev); if (ret) return ret; break; case AMDGPU_DPM_FORCED_LEVEL_LOW: + /* sclk */ ret = cz_dpm_unforce_dpm_levels(adev); if (ret) return ret; ret = cz_dpm_force_lowest(adev); + if (ret) + return ret; + + /* uvd */ + ret = cz_dpm_unforce_uvd_dpm_levels(adev); + if (ret) + return ret; + ret = cz_dpm_uvd_force_lowest(adev); if (ret) return ret; break; case AMDGPU_DPM_FORCED_LEVEL_AUTO: + /* sclk */ ret = cz_dpm_unforce_dpm_levels(adev); + if (ret) + return ret; + + /* uvd */ + ret = cz_dpm_unforce_uvd_dpm_levels(adev); if (ret) return ret; break; -- cgit From 403664bbf840e80a45c4b97115ac518b40487f91 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 18 Dec 2015 11:33:30 -0500 Subject: drm/amdgpu/cz: force vce clocks when sclks are forced MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cz_dpm.c | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c index bdf5a22ba450..4dd17f2dd905 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c @@ -1984,6 +1984,14 @@ static int cz_dpm_force_dpm_level(struct amdgpu_device *adev, if (ret) return ret; ret = cz_dpm_uvd_force_highest(adev); + if (ret) + return ret; + + /* vce */ + ret = cz_dpm_unforce_vce_dpm_levels(adev); + if (ret) + return ret; + ret = cz_dpm_vce_force_highest(adev); if (ret) return ret; break; @@ -2001,6 +2009,14 @@ static int cz_dpm_force_dpm_level(struct amdgpu_device *adev, if (ret) return ret; ret = cz_dpm_uvd_force_lowest(adev); + if (ret) + return ret; + + /* vce */ + ret = cz_dpm_unforce_vce_dpm_levels(adev); + if (ret) + return ret; + ret = cz_dpm_vce_force_lowest(adev); if (ret) return ret; break; @@ -2012,6 +2028,11 @@ static int cz_dpm_force_dpm_level(struct amdgpu_device *adev, /* uvd */ ret = cz_dpm_unforce_uvd_dpm_levels(adev); + if (ret) + return ret; + + /* vce */ + ret = cz_dpm_unforce_vce_dpm_levels(adev); if (ret) return ret; break; @@ -2154,7 +2175,8 @@ static int cz_update_vce_dpm(struct amdgpu_device *adev) pi->vce_dpm.hard_min_clk = table->entries[table->count-1].ecclk; } else { /* non-stable p-state cases. without vce.Arbiter.EcclkHardMin */ - pi->vce_dpm.hard_min_clk = table->entries[0].ecclk; + /* leave it as set by user */ + /*pi->vce_dpm.hard_min_clk = table->entries[0].ecclk;*/ } cz_send_msg_to_smc_with_parameter(adev, -- cgit From cc29ec874b3727e3f5a606bd1875e34f6e3f1593 Mon Sep 17 00:00:00 2001 From: Geliang Tang Date: Wed, 13 Jan 2016 22:48:42 +0800 Subject: drm/amdgpu: use kobj_to_dev() Use kobj_to_dev() instead of open-coding it. Signed-off-by: Geliang Tang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 4386cbac7f97..7d8d84eaea4a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -326,7 +326,7 @@ static struct attribute *hwmon_attributes[] = { static umode_t hwmon_attributes_visible(struct kobject *kobj, struct attribute *attr, int index) { - struct device *dev = container_of(kobj, struct device, kobj); + struct device *dev = kobj_to_dev(kobj); struct amdgpu_device *adev = dev_get_drvdata(dev); umode_t effective_mode = attr->mode; -- cgit From eceb8a1562db6fc3e1c21cd54a4be0189aa1d0e3 Mon Sep 17 00:00:00 2001 From: Christian König Date: Mon, 11 Jan 2016 15:35:21 +0100 Subject: drm/amdgpu: move VM page tables to the LRU end on CS v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This makes it less likely to run into an ENOMEM because VM page tables are evicted last. v2: move the BOs in the LRU tail after validation Signed-off-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 7 ++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 27 +++++++++++++++++++++++++++ 3 files changed, 35 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 003959f99251..313b0cc8d676 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -987,6 +987,8 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, struct list_head *validated, struct amdgpu_bo_list_entry *entry); void amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, struct list_head *duplicates); +void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev, + struct amdgpu_vm *vm); int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_sync *sync); void amdgpu_vm_flush(struct amdgpu_ring *ring, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index ce0254d4dcd7..1fffc338f69d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -428,8 +428,10 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &duplicates); error_validate: - if (r) + if (r) { + amdgpu_vm_move_pt_bos_in_lru(p->adev, &fpriv->vm); ttm_eu_backoff_reservation(&p->ticket, &p->validated); + } error_reserve: if (need_mmap_lock) @@ -473,8 +475,11 @@ static int cmp_size_smaller_first(void *priv, struct list_head *a, **/ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bool backoff) { + struct amdgpu_fpriv *fpriv = parser->filp->driver_priv; unsigned i; + amdgpu_vm_move_pt_bos_in_lru(parser->adev, &fpriv->vm); + if (!error) { /* Sort the buffer list from the smallest to largest buffer, * which affects the order of buffers in the LRU list. diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 8f7688e598a1..aefc668e6b5d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -119,6 +119,33 @@ void amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, struct list_head *duplicates) list_add(&entry->tv.head, duplicates); } + +} + +/** + * amdgpu_vm_move_pt_bos_in_lru - move the PT BOs to the LRU tail + * + * @adev: amdgpu device instance + * @vm: vm providing the BOs + * + * Move the PT BOs to the tail of the LRU. + */ +void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev, + struct amdgpu_vm *vm) +{ + struct ttm_bo_global *glob = adev->mman.bdev.glob; + unsigned i; + + spin_lock(&glob->lru_lock); + for (i = 0; i <= vm->max_pde_used; ++i) { + struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry; + + if (!entry->robj) + continue; + + ttm_bo_move_to_lru_tail(&entry->robj->tbo); + } + spin_unlock(&glob->lru_lock); } /** -- cgit From d8e0cae645504a787b05abfb91afee8cd7fa1701 Mon Sep 17 00:00:00 2001 From: Christian König Date: Mon, 11 Jan 2016 15:35:22 +0100 Subject: drm/amdgpu: validate duplicates first MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Most VM BOs end up in the duplicates list, validate it first make -ENOMEM less likely. Signed-off-by: Christian König Reviewed-by: Chunming Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 1fffc338f69d..6f89f8e034d0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -421,11 +421,11 @@ static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p) amdgpu_vm_get_pt_bos(&fpriv->vm, &duplicates); - r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &p->validated); + r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &duplicates); if (r) goto error_validate; - r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &duplicates); + r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &p->validated); error_validate: if (r) { -- cgit From e9c5e7402dad6f4f04c2430db6f283512bcd4392 Mon Sep 17 00:00:00 2001 From: Dave Airlie Date: Thu, 14 Jan 2016 08:07:55 +1000 Subject: drm/amdgpu: add missing irq.h include this fixes the build on arm. Signed-off-by: Dave Airlie --- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 3006182c5d0d..f594cfaa97e5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -25,6 +25,7 @@ * Alex Deucher * Jerome Glisse */ +#include #include #include #include -- cgit