Merge branch 'drm-next' of git://people.freedesktop.org/~airlied/linux

Pull drm updates from Dave Airlie:
 "I Was Almost Tempted To Capitalise Every Word, but then I decided I
  couldn't read it myself!

  I've also got one pull request for the sti driver outstanding.  It
  relied on a commit in Greg's tree and I didn't find out in time, that
  commit is in your tree now so I might send that along once this is
  merged.

  I also had the accidental misfortune to have access to a Skylake on my
  desk for a few days, and I've had to encourage Intel to try harder,
  which seems to be happening now.

  Here is the main drm-next pull request for 4.4.

  Highlights:

  New driver:
        vc4 driver for the Rasberry Pi VPU.
        (From Eric Anholt at Broadcom.)

  Core:
        Atomic fbdev support
        Atomic helpers for runtime pm
        dp/aux i2c STATUS_UPDATE handling
        struct_mutex usage cleanups.
        Generic of probing support.

  Documentation:
        Kerneldoc for VGA switcheroo code.
        Rename to gpu instead of drm to reflect scope.

  i915:
        Skylake GuC firmware fixes
        HPD A support
        VBT backlight fallbacks
        Fastboot by default for some systems
        FBC work
        BXT/SKL workarounds
        Skylake deeper sleep state fixes

  amdgpu:
        Enable GPU scheduler by default
        New atombios opcodes
        GPUVM debugging options
        Stoney support.
        Fencing cleanups.

  radeon:
        More efficient CS checking

  nouveau:
        gk20a instance memory handling improvements.
        Improved PGOB detection and GK107 support
        Kepler GDDR5 PLL statbility improvement
        G8x/GT2xx reclock improvements
        new userspace API compatiblity fixes.

  virtio-gpu:
        Add 3D support - qemu 2.5 has it merged for it's gtk backend.

  msm:
        Initial msm88896 (snapdragon 8200)

  exynos:
        HDMI cleanups
        Enable mixer driver byt default
        Add DECON-TV support

  vmwgfx:
        Move to using memremap + fixes.

  rcar-du:
        Add support for R8A7793/4 DU

  armada:
        Remove support for non-component mode
        Improved plane handling
        Power savings while in DPMS off.

  tda998x:
        Remove unused slave encoder support
        Use more HDMI helpers
        Fix EDID read handling

  dwhdmi:
        Interlace video mode support for ipu-v3/dw_hdmi
        Hotplug state fixes
        Audio driver integration

  imx:
        More color formats support.

  tegra:
        Minor fixes/improvements"

[ Merge fixup: remove unused variable 'dev' that had all uses removed in
  commit 4e270f088011: "drm/gem: Drop struct_mutex requirement from
  drm_gem_mmap_obj" ]

* 'drm-next' of git://people.freedesktop.org/~airlied/linux: (764 commits)
  drm/vmwgfx: Relax irq locking somewhat
  drm/vmwgfx: Properly flush cursor updates and page-flips
  drm/i915/skl: disable display side power well support for now
  drm/i915: Extend DSL readout fix to BDW and SKL.
  drm/i915: Do graphics device reset under forcewake
  drm/i915: Skip fence installation for objects with rotated views (v4)
  vga_switcheroo: Drop client power state VGA_SWITCHEROO_INIT
  drm/amdgpu: group together common fence implementation
  drm/amdgpu: remove AMDGPU_FENCE_OWNER_MOVE
  drm/amdgpu: remove now unused fence functions
  drm/amdgpu: fix fence fallback check
  drm/amdgpu: fix stoping the scheduler timeout
  drm/amdgpu: cleanup on error in amdgpu_cs_ioctl()
  drm/i915: Fix locking around GuC firmware load
  drm/amdgpu: update Fiji's Golden setting
  drm/amdgpu: update Fiji's rev id
  drm/amdgpu: extract common code in vi_common_early_init
  drm/amd/scheduler: don't oops on failure to load
  drm/amdgpu: don't oops on failure to load (v2)
  drm/amdgpu: don't VT switch on suspend
  ...

9 files changed, 215 insertions, 242 deletions

diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
index 2c7a25c71af2..a09cf8529b9f 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
@@ -146,73 +146,73 @@
 
 static const struct drm_ioctl_desc vmw_ioctls[] = {
 	VMW_IOCTL_DEF(VMW_GET_PARAM, vmw_getparam_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_ALLOC_DMABUF, vmw_dmabuf_alloc_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_UNREF_DMABUF, vmw_dmabuf_unref_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_CURSOR_BYPASS,
 		      vmw_kms_cursor_bypass_ioctl,
-		      DRM_MASTER | DRM_CONTROL_ALLOW | DRM_UNLOCKED),
+		      DRM_MASTER | DRM_CONTROL_ALLOW),
 
 	VMW_IOCTL_DEF(VMW_CONTROL_STREAM, vmw_overlay_ioctl,
-		      DRM_MASTER | DRM_CONTROL_ALLOW | DRM_UNLOCKED),
+		      DRM_MASTER | DRM_CONTROL_ALLOW),
 	VMW_IOCTL_DEF(VMW_CLAIM_STREAM, vmw_stream_claim_ioctl,
-		      DRM_MASTER | DRM_CONTROL_ALLOW | DRM_UNLOCKED),
+		      DRM_MASTER | DRM_CONTROL_ALLOW),
 	VMW_IOCTL_DEF(VMW_UNREF_STREAM, vmw_stream_unref_ioctl,
-		      DRM_MASTER | DRM_CONTROL_ALLOW | DRM_UNLOCKED),
+		      DRM_MASTER | DRM_CONTROL_ALLOW),
 
 	VMW_IOCTL_DEF(VMW_CREATE_CONTEXT, vmw_context_define_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_UNREF_CONTEXT, vmw_context_destroy_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_CREATE_SURFACE, vmw_surface_define_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_UNREF_SURFACE, vmw_surface_destroy_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_REF_SURFACE, vmw_surface_reference_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
-	VMW_IOCTL_DEF(VMW_EXECBUF, NULL, DRM_AUTH | DRM_UNLOCKED |
+		      DRM_AUTH | DRM_RENDER_ALLOW),
+	VMW_IOCTL_DEF(VMW_EXECBUF, NULL, DRM_AUTH |
 		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_FENCE_WAIT, vmw_fence_obj_wait_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_FENCE_SIGNALED,
 		      vmw_fence_obj_signaled_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_FENCE_UNREF, vmw_fence_obj_unref_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_FENCE_EVENT, vmw_fence_event_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_GET_3D_CAP, vmw_get_cap_3d_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 
 	/* these allow direct access to the framebuffers mark as master only */
 	VMW_IOCTL_DEF(VMW_PRESENT, vmw_present_ioctl,
-		      DRM_MASTER | DRM_AUTH | DRM_UNLOCKED),
+		      DRM_MASTER | DRM_AUTH),
 	VMW_IOCTL_DEF(VMW_PRESENT_READBACK,
 		      vmw_present_readback_ioctl,
-		      DRM_MASTER | DRM_AUTH | DRM_UNLOCKED),
+		      DRM_MASTER | DRM_AUTH),
 	VMW_IOCTL_DEF(VMW_UPDATE_LAYOUT,
 		      vmw_kms_update_layout_ioctl,
-		      DRM_MASTER | DRM_UNLOCKED),
+		      DRM_MASTER),
 	VMW_IOCTL_DEF(VMW_CREATE_SHADER,
 		      vmw_shader_define_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_UNREF_SHADER,
 		      vmw_shader_destroy_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_GB_SURFACE_CREATE,
 		      vmw_gb_surface_define_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_GB_SURFACE_REF,
 		      vmw_gb_surface_reference_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_SYNCCPU,
 		      vmw_user_dmabuf_synccpu_ioctl,
-		      DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_RENDER_ALLOW),
 	VMW_IOCTL_DEF(VMW_CREATE_EXTENDED_CONTEXT,
 		      vmw_extended_context_define_ioctl,
-		      DRM_AUTH | DRM_UNLOCKED | DRM_RENDER_ALLOW),
+		      DRM_AUTH | DRM_RENDER_ALLOW),
 };
 
 static struct pci_device_id vmw_pci_id_list[] = {
@@ -643,7 +643,7 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
 	init_waitqueue_head(&dev_priv->fence_queue);
 	init_waitqueue_head(&dev_priv->fifo_queue);
 	dev_priv->fence_queue_waiters = 0;
-	atomic_set(&dev_priv->fifo_queue_waiters, 0);
+	dev_priv->fifo_queue_waiters = 0;
 
 	dev_priv->used_memory_size = 0;
 
@@ -752,8 +752,8 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
 	ttm_lock_set_kill(&dev_priv->fbdev_master.lock, false, SIGTERM);
 	dev_priv->active_master = &dev_priv->fbdev_master;
 
-	dev_priv->mmio_virt = ioremap_cache(dev_priv->mmio_start,
-					    dev_priv->mmio_size);
+	dev_priv->mmio_virt = memremap(dev_priv->mmio_start,
+				       dev_priv->mmio_size, MEMREMAP_WB);
 
 	if (unlikely(dev_priv->mmio_virt == NULL)) {
 		ret = -ENOMEM;
@@ -907,7 +907,7 @@ out_no_irq:
 out_no_device:
 	ttm_object_device_release(&dev_priv->tdev);
 out_err4:
-	iounmap(dev_priv->mmio_virt);
+	memunmap(dev_priv->mmio_virt);
 out_err3:
 	vmw_ttm_global_release(dev_priv);
 out_err0:
@@ -958,7 +958,7 @@ static int vmw_driver_unload(struct drm_device *dev)
 		pci_release_regions(dev->pdev);
 
 	ttm_object_device_release(&dev_priv->tdev);
-	iounmap(dev_priv->mmio_virt);
+	memunmap(dev_priv->mmio_virt);
 	if (dev_priv->ctx.staged_bindings)
 		vmw_binding_state_free(dev_priv->ctx.staged_bindings);
 	vmw_ttm_global_release(dev_priv);
@@ -1062,14 +1062,6 @@ static struct vmw_master *vmw_master_check(struct drm_device *dev,
 	mutex_unlock(&dev->master_mutex);
 
 	/*
-	 * Taking the drm_global_mutex after the TTM lock might deadlock
-	 */
-	if (!(flags & DRM_UNLOCKED)) {
-		DRM_ERROR("Refusing locked ioctl access.\n");
-		return ERR_PTR(-EDEADLK);
-	}
-
-	/*
 	 * Take the TTM lock. Possibly sleep waiting for the authenticating
 	 * master to become master again, or for a SIGTERM if the
 	 * authenticating master exits.
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
index f19fd39b43e1..a8ae9dfb83b7 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
@@ -375,7 +375,7 @@ struct vmw_private {
 	uint32_t stdu_max_height;
 	uint32_t initial_width;
 	uint32_t initial_height;
-	u32 __iomem *mmio_virt;
+	u32 *mmio_virt;
 	uint32_t capabilities;
 	uint32_t max_gmr_ids;
 	uint32_t max_gmr_pages;
@@ -440,13 +440,12 @@ struct vmw_private {
 	spinlock_t waiter_lock;
 	int fence_queue_waiters; /* Protected by waiter_lock */
 	int goal_queue_waiters; /* Protected by waiter_lock */
-	int cmdbuf_waiters; /* Protected by irq_lock */
-	int error_waiters; /* Protected by irq_lock */
-	atomic_t fifo_queue_waiters;
+	int cmdbuf_waiters; /* Protected by waiter_lock */
+	int error_waiters; /* Protected by waiter_lock */
+	int fifo_queue_waiters; /* Protected by waiter_lock */
 	uint32_t last_read_seqno;
-	spinlock_t irq_lock;
 	struct vmw_fence_manager *fman;
-	uint32_t irq_mask;
+	uint32_t irq_mask; /* Updates protected by waiter_lock */
 
 	/*
 	 * Device state
@@ -914,9 +913,9 @@ void vmw_kms_idle_workqueues(struct vmw_master *vmaster);
 bool vmw_kms_validate_mode_vram(struct vmw_private *dev_priv,
 				uint32_t pitch,
 				uint32_t height);
-u32 vmw_get_vblank_counter(struct drm_device *dev, int crtc);
-int vmw_enable_vblank(struct drm_device *dev, int crtc);
-void vmw_disable_vblank(struct drm_device *dev, int crtc);
+u32 vmw_get_vblank_counter(struct drm_device *dev, unsigned int pipe);
+int vmw_enable_vblank(struct drm_device *dev, unsigned int pipe);
+void vmw_disable_vblank(struct drm_device *dev, unsigned int pipe);
 int vmw_kms_present(struct vmw_private *dev_priv,
 		    struct drm_file *file_priv,
 		    struct vmw_framebuffer *vfb,
@@ -1206,4 +1205,30 @@ static inline void vmw_fifo_resource_dec(struct vmw_private *dev_priv)
 {
 	atomic_dec(&dev_priv->num_fifo_resources);
 }
+
+/**
+ * vmw_mmio_read - Perform a MMIO read from volatile memory
+ *
+ * @addr: The address to read from
+ *
+ * This function is intended to be equivalent to ioread32() on
+ * memremap'd memory, but without byteswapping.
+ */
+static inline u32 vmw_mmio_read(u32 *addr)
+{
+	return READ_ONCE(*addr);
+}
+
+/**
+ * vmw_mmio_write - Perform a MMIO write to volatile memory
+ *
+ * @addr: The address to write to
+ *
+ * This function is intended to be equivalent to iowrite32 on
+ * memremap'd memory, but without byteswapping.
+ */
+static inline void vmw_mmio_write(u32 value, u32 *addr)
+{
+	WRITE_ONCE(*addr, value);
+}
 #endif
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c
index 567ddede51d1..8e689b439890 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c
@@ -142,8 +142,8 @@ static bool vmw_fence_enable_signaling(struct fence *f)
 	struct vmw_fence_manager *fman = fman_from_fence(fence);
 	struct vmw_private *dev_priv = fman->dev_priv;
 
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
-	u32 seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE);
+	u32 *fifo_mem = dev_priv->mmio_virt;
+	u32 seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE);
 	if (seqno - fence->base.seqno < VMW_FENCE_WRAP)
 		return false;
 
@@ -386,14 +386,14 @@ static bool vmw_fence_goal_new_locked(struct vmw_fence_manager *fman,
 				      u32 passed_seqno)
 {
 	u32 goal_seqno;
-	u32 __iomem *fifo_mem;
+	u32 *fifo_mem;
 	struct vmw_fence_obj *fence;
 
 	if (likely(!fman->seqno_valid))
 		return false;
 
 	fifo_mem = fman->dev_priv->mmio_virt;
-	goal_seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE_GOAL);
+	goal_seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE_GOAL);
 	if (likely(passed_seqno - goal_seqno >= VMW_FENCE_WRAP))
 		return false;
 
@@ -401,8 +401,8 @@ static bool vmw_fence_goal_new_locked(struct vmw_fence_manager *fman,
 	list_for_each_entry(fence, &fman->fence_list, head) {
 		if (!list_empty(&fence->seq_passed_actions)) {
 			fman->seqno_valid = true;
-			iowrite32(fence->base.seqno,
-				  fifo_mem + SVGA_FIFO_FENCE_GOAL);
+			vmw_mmio_write(fence->base.seqno,
+				       fifo_mem + SVGA_FIFO_FENCE_GOAL);
 			break;
 		}
 	}
@@ -430,18 +430,18 @@ static bool vmw_fence_goal_check_locked(struct vmw_fence_obj *fence)
 {
 	struct vmw_fence_manager *fman = fman_from_fence(fence);
 	u32 goal_seqno;
-	u32 __iomem *fifo_mem;
+	u32 *fifo_mem;
 
 	if (fence_is_signaled_locked(&fence->base))
 		return false;
 
 	fifo_mem = fman->dev_priv->mmio_virt;
-	goal_seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE_GOAL);
+	goal_seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE_GOAL);
 	if (likely(fman->seqno_valid &&
 		   goal_seqno - fence->base.seqno < VMW_FENCE_WRAP))
 		return false;
 
-	iowrite32(fence->base.seqno, fifo_mem + SVGA_FIFO_FENCE_GOAL);
+	vmw_mmio_write(fence->base.seqno, fifo_mem + SVGA_FIFO_FENCE_GOAL);
 	fman->seqno_valid = true;
 
 	return true;
@@ -453,9 +453,9 @@ static void __vmw_fences_update(struct vmw_fence_manager *fman)
 	struct list_head action_list;
 	bool needs_rerun;
 	uint32_t seqno, new_seqno;
-	u32 __iomem *fifo_mem = fman->dev_priv->mmio_virt;
+	u32 *fifo_mem = fman->dev_priv->mmio_virt;
 
-	seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE);
+	seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE);
 rerun:
 	list_for_each_entry_safe(fence, next_fence, &fman->fence_list, head) {
 		if (seqno - fence->base.seqno < VMW_FENCE_WRAP) {
@@ -477,7 +477,7 @@ rerun:
 
 	needs_rerun = vmw_fence_goal_new_locked(fman, seqno);
 	if (unlikely(needs_rerun)) {
-		new_seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE);
+		new_seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE);
 		if (new_seqno != seqno) {
 			seqno = new_seqno;
 			goto rerun;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
index 80c40c31d4f8..a8baf5f5e765 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
@@ -36,7 +36,7 @@ struct vmw_temp_set_context {
 
 bool vmw_fifo_have_3d(struct vmw_private *dev_priv)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+	u32 *fifo_mem = dev_priv->mmio_virt;
 	uint32_t fifo_min, hwversion;
 	const struct vmw_fifo_state *fifo = &dev_priv->fifo;
 
@@ -60,15 +60,15 @@ bool vmw_fifo_have_3d(struct vmw_private *dev_priv)
 	if (!(dev_priv->capabilities & SVGA_CAP_EXTENDED_FIFO))
 		return false;
 
-	fifo_min = ioread32(fifo_mem  + SVGA_FIFO_MIN);
+	fifo_min = vmw_mmio_read(fifo_mem  + SVGA_FIFO_MIN);
 	if (fifo_min <= SVGA_FIFO_3D_HWVERSION * sizeof(unsigned int))
 		return false;
 
-	hwversion = ioread32(fifo_mem +
-			     ((fifo->capabilities &
-			       SVGA_FIFO_CAP_3D_HWVERSION_REVISED) ?
-			      SVGA_FIFO_3D_HWVERSION_REVISED :
-			      SVGA_FIFO_3D_HWVERSION));
+	hwversion = vmw_mmio_read(fifo_mem +
+				  ((fifo->capabilities &
+				    SVGA_FIFO_CAP_3D_HWVERSION_REVISED) ?
+				   SVGA_FIFO_3D_HWVERSION_REVISED :
+				   SVGA_FIFO_3D_HWVERSION));
 
 	if (hwversion == 0)
 		return false;
@@ -85,13 +85,13 @@ bool vmw_fifo_have_3d(struct vmw_private *dev_priv)
 
 bool vmw_fifo_have_pitchlock(struct vmw_private *dev_priv)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+	u32  *fifo_mem = dev_priv->mmio_virt;
 	uint32_t caps;
 
 	if (!(dev_priv->capabilities & SVGA_CAP_EXTENDED_FIFO))
 		return false;
 
-	caps = ioread32(fifo_mem + SVGA_FIFO_CAPABILITIES);
+	caps = vmw_mmio_read(fifo_mem + SVGA_FIFO_CAPABILITIES);
 	if (caps & SVGA_FIFO_CAP_PITCHLOCK)
 		return true;
 
@@ -100,7 +100,7 @@ bool vmw_fifo_have_pitchlock(struct vmw_private *dev_priv)
 
 int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+	u32  *fifo_mem = dev_priv->mmio_virt;
 	uint32_t max;
 	uint32_t min;
 
@@ -137,19 +137,19 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 	if (min < PAGE_SIZE)
 		min = PAGE_SIZE;
 
-	iowrite32(min, fifo_mem + SVGA_FIFO_MIN);
-	iowrite32(dev_priv->mmio_size, fifo_mem + SVGA_FIFO_MAX);
+	vmw_mmio_write(min, fifo_mem + SVGA_FIFO_MIN);
+	vmw_mmio_write(dev_priv->mmio_size, fifo_mem + SVGA_FIFO_MAX);
 	wmb();
-	iowrite32(min,  fifo_mem + SVGA_FIFO_NEXT_CMD);
-	iowrite32(min,  fifo_mem + SVGA_FIFO_STOP);
-	iowrite32(0, fifo_mem + SVGA_FIFO_BUSY);
+	vmw_mmio_write(min,  fifo_mem + SVGA_FIFO_NEXT_CMD);
+	vmw_mmio_write(min,  fifo_mem + SVGA_FIFO_STOP);
+	vmw_mmio_write(0, fifo_mem + SVGA_FIFO_BUSY);
 	mb();
 
 	vmw_write(dev_priv, SVGA_REG_CONFIG_DONE, 1);
 
-	max = ioread32(fifo_mem + SVGA_FIFO_MAX);
-	min = ioread32(fifo_mem  + SVGA_FIFO_MIN);
-	fifo->capabilities = ioread32(fifo_mem + SVGA_FIFO_CAPABILITIES);
+	max = vmw_mmio_read(fifo_mem + SVGA_FIFO_MAX);
+	min = vmw_mmio_read(fifo_mem  + SVGA_FIFO_MIN);
+	fifo->capabilities = vmw_mmio_read(fifo_mem + SVGA_FIFO_CAPABILITIES);
 
 	DRM_INFO("Fifo max 0x%08x min 0x%08x cap 0x%08x\n",
 		 (unsigned int) max,
@@ -157,7 +157,7 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 		 (unsigned int) fifo->capabilities);
 
 	atomic_set(&dev_priv->marker_seq, dev_priv->last_read_seqno);
-	iowrite32(dev_priv->last_read_seqno, fifo_mem + SVGA_FIFO_FENCE);
+	vmw_mmio_write(dev_priv->last_read_seqno, fifo_mem + SVGA_FIFO_FENCE);
 	vmw_marker_queue_init(&fifo->marker_queue);
 
 	return 0;
@@ -165,31 +165,23 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 
 void vmw_fifo_ping_host(struct vmw_private *dev_priv, uint32_t reason)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
-	static DEFINE_SPINLOCK(ping_lock);
-	unsigned long irq_flags;
+	u32 *fifo_mem = dev_priv->mmio_virt;
 
-	/*
-	 * The ping_lock is needed because we don't have an atomic
-	 * test-and-set of the SVGA_FIFO_BUSY register.
-	 */
-	spin_lock_irqsave(&ping_lock, irq_flags);
-	if (unlikely(ioread32(fifo_mem + SVGA_FIFO_BUSY) == 0)) {
-		iowrite32(1, fifo_mem + SVGA_FIFO_BUSY);
+	preempt_disable();
+	if (cmpxchg(fifo_mem + SVGA_FIFO_BUSY, 0, 1) == 0)
 		vmw_write(dev_priv, SVGA_REG_SYNC, reason);
-	}
-	spin_unlock_irqrestore(&ping_lock, irq_flags);
+	preempt_enable();
 }
 
 void vmw_fifo_release(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+	u32  *fifo_mem = dev_priv->mmio_virt;
 
 	vmw_write(dev_priv, SVGA_REG_SYNC, SVGA_SYNC_GENERIC);
 	while (vmw_read(dev_priv, SVGA_REG_BUSY) != 0)
 		;
 
-	dev_priv->last_read_seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE);
+	dev_priv->last_read_seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE);
 
 	vmw_write(dev_priv, SVGA_REG_CONFIG_DONE,
 		  dev_priv->config_done_state);
@@ -213,11 +205,11 @@ void vmw_fifo_release(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 
 static bool vmw_fifo_is_full(struct vmw_private *dev_priv, uint32_t bytes)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
-	uint32_t max = ioread32(fifo_mem + SVGA_FIFO_MAX);
-	uint32_t next_cmd = ioread32(fifo_mem + SVGA_FIFO_NEXT_CMD);
-	uint32_t min = ioread32(fifo_mem + SVGA_FIFO_MIN);
-	uint32_t stop = ioread32(fifo_mem + SVGA_FIFO_STOP);
+	u32  *fifo_mem = dev_priv->mmio_virt;
+	uint32_t max = vmw_mmio_read(fifo_mem + SVGA_FIFO_MAX);
+	uint32_t next_cmd = vmw_mmio_read(fifo_mem + SVGA_FIFO_NEXT_CMD);
+	uint32_t min = vmw_mmio_read(fifo_mem + SVGA_FIFO_MIN);
+	uint32_t stop = vmw_mmio_read(fifo_mem + SVGA_FIFO_STOP);
 
 	return ((max - next_cmd) + (stop - min) <= bytes);
 }
@@ -260,7 +252,6 @@ static int vmw_fifo_wait(struct vmw_private *dev_priv,
 			 unsigned long timeout)
 {
 	long ret = 1L;
-	unsigned long irq_flags;
 
 	if (likely(!vmw_fifo_is_full(dev_priv, bytes)))
 		return 0;
@@ -270,16 +261,8 @@ static int vmw_fifo_wait(struct vmw_private *dev_priv,
 		return vmw_fifo_wait_noirq(dev_priv, bytes,
 					   interruptible, timeout);
 
-	spin_lock(&dev_priv->waiter_lock);
-	if (atomic_add_return(1, &dev_priv->fifo_queue_waiters) > 0) {
-		spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-		outl(SVGA_IRQFLAG_FIFO_PROGRESS,
-		     dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
-		dev_priv->irq_mask |= SVGA_IRQFLAG_FIFO_PROGRESS;
-		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-		spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
-	}
-	spin_unlock(&dev_priv->waiter_lock);
+	vmw_generic_waiter_add(dev_priv, SVGA_IRQFLAG_FIFO_PROGRESS,
+			       &dev_priv->fifo_queue_waiters);
 
 	if (interruptible)
 		ret = wait_event_interruptible_timeout
@@ -295,14 +278,8 @@ static int vmw_fifo_wait(struct vmw_private *dev_priv,
 	else if (likely(ret > 0))
 		ret = 0;
 
-	spin_lock(&dev_priv->waiter_lock);
-	if (atomic_dec_and_test(&dev_priv->fifo_queue_waiters)) {
-		spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-		dev_priv->irq_mask &= ~SVGA_IRQFLAG_FIFO_PROGRESS;
-		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-		spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
-	}
-	spin_unlock(&dev_priv->waiter_lock);
+	vmw_generic_waiter_remove(dev_priv, SVGA_IRQFLAG_FIFO_PROGRESS,
+				  &dev_priv->fifo_queue_waiters);
 
 	return ret;
 }
@@ -321,7 +298,7 @@ static void *vmw_local_fifo_reserve(struct vmw_private *dev_priv,
 				    uint32_t bytes)
 {
 	struct vmw_fifo_state *fifo_state = &dev_priv->fifo;
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+	u32  *fifo_mem = dev_priv->mmio_virt;
 	uint32_t max;
 	uint32_t min;
 	uint32_t next_cmd;
@@ -329,9 +306,9 @@ static void *vmw_local_fifo_reserve(struct vmw_private *dev_priv,
 	int ret;
 
 	mutex_lock(&fifo_state->fifo_mutex);
-	max = ioread32(fifo_mem + SVGA_FIFO_MAX);
-	min = ioread32(fifo_mem + SVGA_FIFO_MIN);
-	next_cmd = ioread32(fifo_mem + SVGA_FIFO_NEXT_CMD);
+	max = vmw_mmio_read(fifo_mem + SVGA_FIFO_MAX);
+	min = vmw_mmio_read(fifo_mem + SVGA_FIFO_MIN);
+	next_cmd = vmw_mmio_read(fifo_mem + SVGA_FIFO_NEXT_CMD);
 
 	if (unlikely(bytes >= (max - min)))
 		goto out_err;
@@ -342,7 +319,7 @@ static void *vmw_local_fifo_reserve(struct vmw_private *dev_priv,
 	fifo_state->reserved_size = bytes;
 
 	while (1) {
-		uint32_t stop = ioread32(fifo_mem + SVGA_FIFO_STOP);
+		uint32_t stop = vmw_mmio_read(fifo_mem + SVGA_FIFO_STOP);
 		bool need_bounce = false;
 		bool reserve_in_place = false;
 
@@ -376,8 +353,8 @@ static void *vmw_local_fifo_reserve(struct vmw_private *dev_priv,
 				fifo_state->using_bounce_buffer = false;
 
 				if (reserveable)
-					iowrite32(bytes, fifo_mem +
-						  SVGA_FIFO_RESERVED);
+					vmw_mmio_write(bytes, fifo_mem +
+						       SVGA_FIFO_RESERVED);
 				return (void __force *) (fifo_mem +
 							 (next_cmd >> 2));
 			} else {
@@ -427,7 +404,7 @@ void *vmw_fifo_reserve_dx(struct vmw_private *dev_priv, uint32_t bytes,
 }
 
 static void vmw_fifo_res_copy(struct vmw_fifo_state *fifo_state,
-			      u32 __iomem *fifo_mem,
+			      u32  *fifo_mem,
 			      uint32_t next_cmd,
 			      uint32_t max, uint32_t min, uint32_t bytes)
 {
@@ -439,17 +416,16 @@ static void vmw_fifo_res_copy(struct vmw_fifo_state *fifo_state,
 	if (bytes < chunk_size)
 		chunk_size = bytes;
 
-	iowrite32(bytes, fifo_mem + SVGA_FIFO_RESERVED);
+	vmw_mmio_write(bytes, fifo_mem + SVGA_FIFO_RESERVED);
 	mb();
-	memcpy_toio(fifo_mem + (next_cmd >> 2), buffer, chunk_size);
+	memcpy(fifo_mem + (next_cmd >> 2), buffer, chunk_size);
 	rest = bytes - chunk_size;
 	if (rest)
-		memcpy_toio(fifo_mem + (min >> 2), buffer + (chunk_size >> 2),
-			    rest);
+		memcpy(fifo_mem + (min >> 2), buffer + (chunk_size >> 2), rest);
 }
 
 static void vmw_fifo_slow_copy(struct vmw_fifo_state *fifo_state,
-			       u32 __iomem *fifo_mem,
+			       u32  *fifo_mem,
 			       uint32_t next_cmd,
 			       uint32_t max, uint32_t min, uint32_t bytes)
 {
@@ -457,12 +433,12 @@ static void vmw_fifo_slow_copy(struct vmw_fifo_state *fifo_state,
 	    fifo_state->dynamic_buffer : fifo_state->static_buffer;
 
 	while (bytes > 0) {
-		iowrite32(*buffer++, fifo_mem + (next_cmd >> 2));
+		vmw_mmio_write(*buffer++, fifo_mem + (next_cmd >> 2));
 		next_cmd += sizeof(uint32_t);
 		if (unlikely(next_cmd == max))
 			next_cmd = min;
 		mb();
-		iowrite32(next_cmd, fifo_mem + SVGA_FIFO_NEXT_CMD);
+		vmw_mmio_write(next_cmd, fifo_mem + SVGA_FIFO_NEXT_CMD);
 		mb();
 		bytes -= sizeof(uint32_t);
 	}
@@ -471,10 +447,10 @@ static void vmw_fifo_slow_copy(struct vmw_fifo_state *fifo_state,
 static void vmw_local_fifo_commit(struct vmw_private *dev_priv, uint32_t bytes)
 {
 	struct vmw_fifo_state *fifo_state = &dev_priv->fifo;
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
-	uint32_t next_cmd = ioread32(fifo_mem + SVGA_FIFO_NEXT_CMD);
-	uint32_t max = ioread32(fifo_mem + SVGA_FIFO_MAX);
-	uint32_t min = ioread32(fifo_mem + SVGA_FIFO_MIN);
+	u32  *fifo_mem = dev_priv->mmio_virt;
+	uint32_t next_cmd = vmw_mmio_read(fifo_mem + SVGA_FIFO_NEXT_CMD);
+	uint32_t max = vmw_mmio_read(fifo_mem + SVGA_FIFO_MAX);
+	uint32_t min = vmw_mmio_read(fifo_mem + SVGA_FIFO_MIN);
 	bool reserveable = fifo_state->capabilities & SVGA_FIFO_CAP_RESERVE;
 
 	if (fifo_state->dx)
@@ -507,11 +483,11 @@ static void vmw_local_fifo_commit(struct vmw_private *dev_priv, uint32_t bytes)
 		if (next_cmd >= max)
 			next_cmd -= max - min;
 		mb();
-		iowrite32(next_cmd, fifo_mem + SVGA_FIFO_NEXT_CMD);
+		vmw_mmio_write(next_cmd, fifo_mem + SVGA_FIFO_NEXT_CMD);
 	}
 
 	if (reserveable)
-		iowrite32(0, fifo_mem + SVGA_FIFO_RESERVED);
+		vmw_mmio_write(0, fifo_mem + SVGA_FIFO_RESERVED);
 	mb();
 	up_write(&fifo_state->rwsem);
 	vmw_fifo_ping_host(dev_priv, SVGA_SYNC_GENERIC);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
index 0a970afed93b..b8c6a03c8c54 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
@@ -64,7 +64,7 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data,
 		break;
 	case DRM_VMW_PARAM_FIFO_HW_VERSION:
 	{
-		u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+		u32 *fifo_mem = dev_priv->mmio_virt;
 		const struct vmw_fifo_state *fifo = &dev_priv->fifo;
 
 		if ((dev_priv->capabilities & SVGA_CAP_GBOBJECTS)) {
@@ -73,11 +73,11 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data,
 		}
 
 		param->value =
-			ioread32(fifo_mem +
-				 ((fifo->capabilities &
-				   SVGA_FIFO_CAP_3D_HWVERSION_REVISED) ?
-				  SVGA_FIFO_3D_HWVERSION_REVISED :
-				  SVGA_FIFO_3D_HWVERSION));
+			vmw_mmio_read(fifo_mem +
+				      ((fifo->capabilities &
+					SVGA_FIFO_CAP_3D_HWVERSION_REVISED) ?
+				       SVGA_FIFO_3D_HWVERSION_REVISED :
+				       SVGA_FIFO_3D_HWVERSION));
 		break;
 	}
 	case DRM_VMW_PARAM_MAX_SURF_MEMORY:
@@ -122,6 +122,22 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data,
 	return 0;
 }
 
+static u32 vmw_mask_multisample(unsigned int cap, u32 fmt_value)
+{
+	/* If the header is updated, update the format test as well! */
+	BUILD_BUG_ON(SVGA3D_DEVCAP_DXFMT_BC5_UNORM + 1 != SVGA3D_DEVCAP_MAX);
+
+	if (cap >= SVGA3D_DEVCAP_DXFMT_X8R8G8B8 &&
+	    cap <= SVGA3D_DEVCAP_DXFMT_BC5_UNORM)
+		fmt_value &= ~(SVGADX_DXFMT_MULTISAMPLE_2 |
+			       SVGADX_DXFMT_MULTISAMPLE_4 |
+			       SVGADX_DXFMT_MULTISAMPLE_8);
+	else if (cap == SVGA3D_DEVCAP_MULTISAMPLE_MASKABLESAMPLES)
+		return 0;
+
+	return fmt_value;
+}
+
 static int vmw_fill_compat_cap(struct vmw_private *dev_priv, void *bounce,
 			       size_t size)
 {
@@ -147,7 +163,8 @@ static int vmw_fill_compat_cap(struct vmw_private *dev_priv, void *bounce,
 	for (i = 0; i < max_size; ++i) {
 		vmw_write(dev_priv, SVGA_REG_DEV_CAP, i);
 		compat_cap->pairs[i][0] = i;
-		compat_cap->pairs[i][1] = vmw_read(dev_priv, SVGA_REG_DEV_CAP);
+		compat_cap->pairs[i][1] = vmw_mask_multisample
+			(i, vmw_read(dev_priv, SVGA_REG_DEV_CAP));
 	}
 	spin_unlock(&dev_priv->cap_lock);
 
@@ -162,7 +179,7 @@ int vmw_get_cap_3d_ioctl(struct drm_device *dev, void *data,
 		(struct drm_vmw_get_3d_cap_arg *) data;
 	struct vmw_private *dev_priv = vmw_priv(dev);
 	uint32_t size;
-	u32 __iomem *fifo_mem;
+	u32 *fifo_mem;
 	void __user *buffer = (void __user *)((unsigned long)(arg->buffer));
 	void *bounce;
 	int ret;
@@ -202,7 +219,8 @@ int vmw_get_cap_3d_ioctl(struct drm_device *dev, void *data,
 		spin_lock(&dev_priv->cap_lock);
 		for (i = 0; i < num; ++i) {
 			vmw_write(dev_priv, SVGA_REG_DEV_CAP, i);
-			*bounce32++ = vmw_read(dev_priv, SVGA_REG_DEV_CAP);
+			*bounce32++ = vmw_mask_multisample
+				(i, vmw_read(dev_priv, SVGA_REG_DEV_CAP));
 		}
 		spin_unlock(&dev_priv->cap_lock);
 	} else if (gb_objects) {
@@ -211,7 +229,7 @@ int vmw_get_cap_3d_ioctl(struct drm_device *dev, void *data,
 			goto out_err;
 	} else {
 		fifo_mem = dev_priv->mmio_virt;
-		memcpy_fromio(bounce, &fifo_mem[SVGA_FIFO_3D_CAPS], size);
+		memcpy(bounce, &fifo_mem[SVGA_FIFO_3D_CAPS], size);
 	}
 
 	ret = copy_to_user(buffer, bounce, size);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c
index 9498a5e33c12..0c7e1723292c 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c
@@ -36,15 +36,13 @@ irqreturn_t vmw_irq_handler(int irq, void *arg)
 	struct vmw_private *dev_priv = vmw_priv(dev);
 	uint32_t status, masked_status;
 
-	spin_lock(&dev_priv->irq_lock);
 	status = inl(dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
-	masked_status = status & dev_priv->irq_mask;
-	spin_unlock(&dev_priv->irq_lock);
+	masked_status = status & READ_ONCE(dev_priv->irq_mask);
 
 	if (likely(status))
 		outl(status, dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
 
-	if (!masked_status)
+	if (!status)
 		return IRQ_NONE;
 
 	if (masked_status & (SVGA_IRQFLAG_ANY_FENCE |
@@ -72,8 +70,8 @@ static bool vmw_fifo_idle(struct vmw_private *dev_priv, uint32_t seqno)
 void vmw_update_seqno(struct vmw_private *dev_priv,
 			 struct vmw_fifo_state *fifo_state)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
-	uint32_t seqno = ioread32(fifo_mem + SVGA_FIFO_FENCE);
+	u32 *fifo_mem = dev_priv->mmio_virt;
+	uint32_t seqno = vmw_mmio_read(fifo_mem + SVGA_FIFO_FENCE);
 
 	if (dev_priv->last_read_seqno != seqno) {
 		dev_priv->last_read_seqno = seqno;
@@ -178,8 +176,9 @@ int vmw_fallback_wait(struct vmw_private *dev_priv,
 	}
 	finish_wait(&dev_priv->fence_queue, &__wait);
 	if (ret == 0 && fifo_idle) {
-		u32 __iomem *fifo_mem = dev_priv->mmio_virt;
-		iowrite32(signal_seq, fifo_mem + SVGA_FIFO_FENCE);
+		u32 *fifo_mem = dev_priv->mmio_virt;
+
+		vmw_mmio_write(signal_seq, fifo_mem + SVGA_FIFO_FENCE);
 	}
 	wake_up_all(&dev_priv->fence_queue);
 out_err:
@@ -189,65 +188,51 @@ out_err:
 	return ret;
 }
 
-void vmw_seqno_waiter_add(struct vmw_private *dev_priv)
+void vmw_generic_waiter_add(struct vmw_private *dev_priv,
+			    u32 flag, int *waiter_count)
 {
-	spin_lock(&dev_priv->waiter_lock);
-	if (dev_priv->fence_queue_waiters++ == 0) {
-		unsigned long irq_flags;
-
-		spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-		outl(SVGA_IRQFLAG_ANY_FENCE,
-		     dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
-		dev_priv->irq_mask |= SVGA_IRQFLAG_ANY_FENCE;
+	spin_lock_bh(&dev_priv->waiter_lock);
+	if ((*waiter_count)++ == 0) {
+		outl(flag, dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
+		dev_priv->irq_mask |= flag;
 		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-		spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
 	}
-	spin_unlock(&dev_priv->waiter_lock);
+	spin_unlock_bh(&dev_priv->waiter_lock);
 }
 
-void vmw_seqno_waiter_remove(struct vmw_private *dev_priv)
+void vmw_generic_waiter_remove(struct vmw_private *dev_priv,
+			       u32 flag, int *waiter_count)
 {
-	spin_lock(&dev_priv->waiter_lock);
-	if (--dev_priv->fence_queue_waiters == 0) {
-		unsigned long irq_flags;
-
-		spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-		dev_priv->irq_mask &= ~SVGA_IRQFLAG_ANY_FENCE;
+	spin_lock_bh(&dev_priv->waiter_lock);
+	if (--(*waiter_count) == 0) {
+		dev_priv->irq_mask &= ~flag;
 		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-		spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
 	}
-	spin_unlock(&dev_priv->waiter_lock);
+	spin_unlock_bh(&dev_priv->waiter_lock);
 }
 
+void vmw_seqno_waiter_add(struct vmw_private *dev_priv)
+{
+	vmw_generic_waiter_add(dev_priv, SVGA_IRQFLAG_ANY_FENCE,
+			       &dev_priv->fence_queue_waiters);
+}
+
+void vmw_seqno_waiter_remove(struct vmw_private *dev_priv)
+{
+	vmw_generic_waiter_remove(dev_priv, SVGA_IRQFLAG_ANY_FENCE,
+				  &dev_priv->fence_queue_waiters);
+}
 
 void vmw_goal_waiter_add(struct vmw_private *dev_priv)
 {
-	spin_lock(&dev_priv->waiter_lock);
-	if (dev_priv->goal_queue_waiters++ == 0) {
-		unsigned long irq_flags;
-
-		spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-		outl(SVGA_IRQFLAG_FENCE_GOAL,
-		     dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
-		dev_priv->irq_mask |= SVGA_IRQFLAG_FENCE_GOAL;
-		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-		spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
-	}
-	spin_unlock(&dev_priv->waiter_lock);
+	vmw_generic_waiter_add(dev_priv, SVGA_IRQFLAG_FENCE_GOAL,
+			       &dev_priv->goal_queue_waiters);
 }
 
 void vmw_goal_waiter_remove(struct vmw_private *dev_priv)
 {
-	spin_lock(&dev_priv->waiter_lock);
-	if (--dev_priv->goal_queue_waiters == 0) {
-		unsigned long irq_flags;
-
-		spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-		dev_priv->irq_mask &= ~SVGA_IRQFLAG_FENCE_GOAL;
-		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-		spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
-	}
-	spin_unlock(&dev_priv->waiter_lock);
+	vmw_generic_waiter_remove(dev_priv, SVGA_IRQFLAG_FENCE_GOAL,
+				  &dev_priv->goal_queue_waiters);
 }
 
 int vmw_wait_seqno(struct vmw_private *dev_priv,
@@ -304,7 +289,6 @@ void vmw_irq_preinstall(struct drm_device *dev)
 	if (!(dev_priv->capabilities & SVGA_CAP_IRQMASK))
 		return;
 
-	spin_lock_init(&dev_priv->irq_lock);
 	status = inl(dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
 	outl(status, dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
 }
@@ -327,30 +311,3 @@ void vmw_irq_uninstall(struct drm_device *dev)
 	status = inl(dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
 	outl(status, dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
 }
-
-void vmw_generic_waiter_add(struct vmw_private *dev_priv,
-			    u32 flag, int *waiter_count)
-{
-	unsigned long irq_flags;
-
-	spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-	if ((*waiter_count)++ == 0) {
-		outl(flag, dev_priv->io_start + VMWGFX_IRQSTATUS_PORT);
-		dev_priv->irq_mask |= flag;
-		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-	}
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
-}
-
-void vmw_generic_waiter_remove(struct vmw_private *dev_priv,
-			       u32 flag, int *waiter_count)
-{
-	unsigned long irq_flags;
-
-	spin_lock_irqsave(&dev_priv->irq_lock, irq_flags);
-	if (--(*waiter_count) == 0) {
-		dev_priv->irq_mask &= ~flag;
-		vmw_write(dev_priv, SVGA_REG_IRQMASK, dev_priv->irq_mask);
-	}
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irq_flags);
-}
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
index 15a6c01cd016..9fcd7f82995c 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
@@ -78,7 +78,7 @@ int vmw_cursor_update_image(struct vmw_private *dev_priv,
 	cmd->cursor.hotspotX = hotspotX;
 	cmd->cursor.hotspotY = hotspotY;
 
-	vmw_fifo_commit(dev_priv, cmd_size);
+	vmw_fifo_commit_flush(dev_priv, cmd_size);
 
 	return 0;
 }
@@ -123,14 +123,14 @@ err_unreserve:
 void vmw_cursor_update_position(struct vmw_private *dev_priv,
 				bool show, int x, int y)
 {
-	u32 __iomem *fifo_mem = dev_priv->mmio_virt;
+	u32 *fifo_mem = dev_priv->mmio_virt;
 	uint32_t count;
 
-	iowrite32(show ? 1 : 0, fifo_mem + SVGA_FIFO_CURSOR_ON);
-	iowrite32(x, fifo_mem + SVGA_FIFO_CURSOR_X);
-	iowrite32(y, fifo_mem + SVGA_FIFO_CURSOR_Y);
-	count = ioread32(fifo_mem + SVGA_FIFO_CURSOR_COUNT);
-	iowrite32(++count, fifo_mem + SVGA_FIFO_CURSOR_COUNT);
+	vmw_mmio_write(show ? 1 : 0, fifo_mem + SVGA_FIFO_CURSOR_ON);
+	vmw_mmio_write(x, fifo_mem + SVGA_FIFO_CURSOR_X);
+	vmw_mmio_write(y, fifo_mem + SVGA_FIFO_CURSOR_Y);
+	count = vmw_mmio_read(fifo_mem + SVGA_FIFO_CURSOR_COUNT);
+	vmw_mmio_write(++count, fifo_mem + SVGA_FIFO_CURSOR_COUNT);
 }
 
 int vmw_du_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv,
@@ -1155,7 +1155,8 @@ int vmw_kms_write_svga(struct vmw_private *vmw_priv,
 	if (vmw_priv->capabilities & SVGA_CAP_PITCHLOCK)
 		vmw_write(vmw_priv, SVGA_REG_PITCHLOCK, pitch);
 	else if (vmw_fifo_have_pitchlock(vmw_priv))
-		iowrite32(pitch, vmw_priv->mmio_virt + SVGA_FIFO_PITCHLOCK);
+		vmw_mmio_write(pitch, vmw_priv->mmio_virt +
+			       SVGA_FIFO_PITCHLOCK);
 	vmw_write(vmw_priv, SVGA_REG_WIDTH, width);
 	vmw_write(vmw_priv, SVGA_REG_HEIGHT, height);
 	vmw_write(vmw_priv, SVGA_REG_BITS_PER_PIXEL, bpp);
@@ -1181,8 +1182,8 @@ int vmw_kms_save_vga(struct vmw_private *vmw_priv)
 		vmw_priv->vga_pitchlock =
 		  vmw_read(vmw_priv, SVGA_REG_PITCHLOCK);
 	else if (vmw_fifo_have_pitchlock(vmw_priv))
-		vmw_priv->vga_pitchlock = ioread32(vmw_priv->mmio_virt +
-						   SVGA_FIFO_PITCHLOCK);
+		vmw_priv->vga_pitchlock = vmw_mmio_read(vmw_priv->mmio_virt +
+							SVGA_FIFO_PITCHLOCK);
 
 	if (!(vmw_priv->capabilities & SVGA_CAP_DISPLAY_TOPOLOGY))
 		return 0;
@@ -1230,8 +1231,8 @@ int vmw_kms_restore_vga(struct vmw_private *vmw_priv)
 		vmw_write(vmw_priv, SVGA_REG_PITCHLOCK,
 			  vmw_priv->vga_pitchlock);
 	else if (vmw_fifo_have_pitchlock(vmw_priv))
-		iowrite32(vmw_priv->vga_pitchlock,
-			  vmw_priv->mmio_virt + SVGA_FIFO_PITCHLOCK);
+		vmw_mmio_write(vmw_priv->vga_pitchlock,
+			       vmw_priv->mmio_virt + SVGA_FIFO_PITCHLOCK);
 
 	if (!(vmw_priv->capabilities & SVGA_CAP_DISPLAY_TOPOLOGY))
 		return 0;
@@ -1263,7 +1264,7 @@ bool vmw_kms_validate_mode_vram(struct vmw_private *dev_priv,
 /**
  * Function called by DRM code called with vbl_lock held.
  */
-u32 vmw_get_vblank_counter(struct drm_device *dev, int crtc)
+u32 vmw_get_vblank_counter(struct drm_device *dev, unsigned int pipe)
 {
 	return 0;
 }
@@ -1271,7 +1272,7 @@ u32 vmw_get_vblank_counter(struct drm_device *dev, int crtc)
 /**
  * Function called by DRM code called with vbl_lock held.
  */
-int vmw_enable_vblank(struct drm_device *dev, int crtc)
+int vmw_enable_vblank(struct drm_device *dev, unsigned int pipe)
 {
 	return -ENOSYS;
 }
@@ -1279,7 +1280,7 @@ int vmw_enable_vblank(struct drm_device *dev, int crtc)
 /**
  * Function called by DRM code called with vbl_lock held.
  */
-void vmw_disable_vblank(struct drm_device *dev, int crtc)
+void vmw_disable_vblank(struct drm_device *dev, unsigned int pipe)
 {
 }
 
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c b/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c
index c22e2df1b336..b1fc1c02792d 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c
@@ -717,6 +717,8 @@ static int vmw_stdu_crtc_page_flip(struct drm_crtc *crtc,
 						   &event->event.tv_usec,
 						   true);
 		vmw_fence_obj_unreference(&fence);
+	} else {
+		vmw_fifo_flush(dev_priv, false);
 	}
 
 	return ret;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
index 03f63c749c02..7d620e82e000 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
@@ -1291,6 +1291,8 @@ int vmw_gb_surface_define_ioctl(struct drm_device *dev, void *data,
 	uint32_t size;
 	uint32_t backup_handle;
 
+	if (req->multisample_count != 0)
+		return -EINVAL;
 
 	if (unlikely(vmw_user_surface_size == 0))
 		vmw_user_surface_size = ttm_round_pot(sizeof(*user_srf)) +