Merge tag 'drm-for-v4.14' of git://people.freedesktop.org/~airlied/linux

Pull drm updates from Dave Airlie:
 "This is the main drm pull request for 4.14 merge window.

  I'm sending this early, as my continuing journey into fatherhood is
  occurring really soon now, I'm going to be mostly useless for the next
  couple of weeks, though I may be able to read email, I doubt I'll be
  doing much patch applications or git sending. If anything urgent pops
  up I've asked Daniel/Jani/Alex/Sean to try and direct stuff towards
  you.

  Outside drm changes:

  Some rcar-du updates that touch the V4L tree, all acks should be in
  place. It adds one export to the radix tree code for new i915 use
  case. There are some minor AGP cleanups (don't see that too often).
  Changes to the vbox driver in staging to avoid breaking compilation.

  Summary:

  core:
   - Atomic helper fixes
   - Atomic UAPI fixes
   - Add YCBCR 4:2:0 support
   - Drop set_busid hook
   - Refactor fb_helper locking
   - Remove a bunch of internal APIs
   - Add a bunch of better default handlers
   - Format modifier/blob plane property added
   - More internal header refactoring
   - Make more internal API names consistent
   - Enhanced syncobj APIs (wait/signal/reset/create signalled)

  bridge:
   - Add Synopsys Designware MIPI DSI host bridge driver

  tiny:
   - Add Pervasive Displays RePaper displays
   - Add support for LEGO MINDSTORMS EV3 LCD

  i915:
   - Lots of GEN10/CNL  support patches
   - drm syncobj support
   - Skylake+ watermark refactoring
   - GVT vGPU 48-bit ppgtt support
   - GVT performance improvements
   - NOA change ioctl
   - CCS (color compression) scanout support
   - GPU reset improvements

  amdgpu:
   - Initial hugepage support
   - BO migration logic rework
   - Vega10 improvements
   - Powerplay fixes
   - Stop reprogramming the MC
   - Fixes for ACP audio on stoney
   - SR-IOV fixes/improvements
   - Command submission overhead improvements

  amdkfd:
   - Non-dGPU upstreaming patches
   - Scratch VA ioctl
   - Image tiling modes
   - Update PM4 headers for new firmware
   - Drop all BUG_ONs.

  nouveau:
   - GP108 modesetting support.
   - Disable MSI on big endian.

  vmwgfx:
   - Add fence fd support.

  msm:
   - Runtime PM improvements

  exynos:
   - NV12MT support
   - Refactor KMS drivers

  imx-drm:
   - Lock scanout channel to improve memory bw
   - Cleanups

  etnaviv:
   - GEM object population fixes

  tegra:
   - Prep work for Tegra186 support
   - PRIME mmap support

  sunxi:
   - HDMI support improvements
   - HDMI CEC support

  omapdrm:
   - HDMI hotplug IRQ support
   - Big driver cleanup
   - OMAP5 DSI support

  rcar-du:
   - vblank fixes
   - VSP1 updates

  arcgpu:
   - Minor fixes

  stm:
   - Add STM32 DSI controller driver

  dw_hdmi:
   - Add support for Rockchip RK3399
   - HDMI CEC support

  atmel-hlcdc:
   - Add 8-bit color support

  vc4:
   - Atomic fixes
   - New ioctl to attach a label to a buffer object
   - HDMI CEC support
   - Allow userspace to dictate rendering order on submit ioctl"

* tag 'drm-for-v4.14' of git://people.freedesktop.org/~airlied/linux: (1074 commits)
  drm/syncobj: Add a signal ioctl (v3)
  drm/syncobj: Add a reset ioctl (v3)
  drm/syncobj: Add a syncobj_array_find helper
  drm/syncobj: Allow wait for submit and signal behavior (v5)
  drm/syncobj: Add a CREATE_SIGNALED flag
  drm/syncobj: Add a callback mechanism for replace_fence (v3)
  drm/syncobj: add sync obj wait interface. (v8)
  i915: Use drm_syncobj_fence_get
  drm/syncobj: Add a race-free drm_syncobj_fence_get helper (v2)
  drm/syncobj: Rename fence_get to find_fence
  drm: kirin: Add mode_valid logic to avoid mode clocks we can't generate
  drm/vmwgfx: Bump the version for fence FD support
  drm/vmwgfx: Add export fence to file descriptor support
  drm/vmwgfx: Add support for imported Fence File Descriptor
  drm/vmwgfx: Prepare to support fence fd
  drm/vmwgfx: Fix incorrect command header offset at restart
  drm/vmwgfx: Support the NOP_ERROR command
  drm/vmwgfx: Restart command buffers after errors
  drm/vmwgfx: Move irq bottom half processing to threads
  drm/vmwgfx: Don't use drm_irq_[un]install
  ...

1 files changed, 91 insertions, 40 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 5599c01b265d..269b835571eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -54,7 +54,7 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 
 	*offset = data->offset;
 
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 
 	if (amdgpu_ttm_tt_get_usermm(p->uf_entry.robj->tbo.ttm)) {
 		amdgpu_bo_unref(&p->uf_entry.robj);
@@ -90,7 +90,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 	}
 
 	/* get chunks */
-	chunk_array_user = (uint64_t __user *)(uintptr_t)(cs->in.chunks);
+	chunk_array_user = u64_to_user_ptr(cs->in.chunks);
 	if (copy_from_user(chunk_array, chunk_array_user,
 			   sizeof(uint64_t)*cs->in.num_chunks)) {
 		ret = -EFAULT;
@@ -110,7 +110,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 		struct drm_amdgpu_cs_chunk user_chunk;
 		uint32_t __user *cdata;
 
-		chunk_ptr = (void __user *)(uintptr_t)chunk_array[i];
+		chunk_ptr = u64_to_user_ptr(chunk_array[i]);
 		if (copy_from_user(&user_chunk, chunk_ptr,
 				       sizeof(struct drm_amdgpu_cs_chunk))) {
 			ret = -EFAULT;
@@ -121,7 +121,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 		p->chunks[i].length_dw = user_chunk.length_dw;
 
 		size = p->chunks[i].length_dw;
-		cdata = (void __user *)(uintptr_t)user_chunk.chunk_data;
+		cdata = u64_to_user_ptr(user_chunk.chunk_data);
 
 		p->chunks[i].kdata = kvmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
 		if (p->chunks[i].kdata == NULL) {
@@ -223,10 +223,11 @@ static s64 bytes_to_us(struct amdgpu_device *adev, u64 bytes)
  * ticks. The accumulated microseconds (us) are converted to bytes and
  * returned.
  */
-static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
+static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
+					      u64 *max_bytes,
+					      u64 *max_vis_bytes)
 {
 	s64 time_us, increment_us;
-	u64 max_bytes;
 	u64 free_vram, total_vram, used_vram;
 
 	/* Allow a maximum of 200 accumulated ms. This is basically per-IB
@@ -238,11 +239,14 @@ static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
 	 */
 	const s64 us_upper_bound = 200000;
 
-	if (!adev->mm_stats.log2_max_MBps)
-		return 0;
+	if (!adev->mm_stats.log2_max_MBps) {
+		*max_bytes = 0;
+		*max_vis_bytes = 0;
+		return;
+	}
 
 	total_vram = adev->mc.real_vram_size - adev->vram_pin_size;
-	used_vram = atomic64_read(&adev->vram_usage);
+	used_vram = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
 
 	spin_lock(&adev->mm_stats.lock);
@@ -280,23 +284,46 @@ static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
 		adev->mm_stats.accum_us = max(min_us, adev->mm_stats.accum_us);
 	}
 
-	/* This returns 0 if the driver is in debt to disallow (optional)
+	/* This is set to 0 if the driver is in debt to disallow (optional)
 	 * buffer moves.
 	 */
-	max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
+	*max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
+
+	/* Do the same for visible VRAM if half of it is free */
+	if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
+		u64 total_vis_vram = adev->mc.visible_vram_size;
+		u64 used_vis_vram =
+			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
+
+		if (used_vis_vram < total_vis_vram) {
+			u64 free_vis_vram = total_vis_vram - used_vis_vram;
+			adev->mm_stats.accum_us_vis = min(adev->mm_stats.accum_us_vis +
+							  increment_us, us_upper_bound);
+
+			if (free_vis_vram >= total_vis_vram / 2)
+				adev->mm_stats.accum_us_vis =
+					max(bytes_to_us(adev, free_vis_vram / 2),
+					    adev->mm_stats.accum_us_vis);
+		}
+
+		*max_vis_bytes = us_to_bytes(adev, adev->mm_stats.accum_us_vis);
+	} else {
+		*max_vis_bytes = 0;
+	}
 
 	spin_unlock(&adev->mm_stats.lock);
-	return max_bytes;
 }
 
 /* Report how many bytes have really been moved for the last command
  * submission. This can result in a debt that can stop buffer migrations
  * temporarily.
  */
-void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes)
+void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
+				  u64 num_vis_bytes)
 {
 	spin_lock(&adev->mm_stats.lock);
 	adev->mm_stats.accum_us -= bytes_to_us(adev, num_bytes);
+	adev->mm_stats.accum_us_vis -= bytes_to_us(adev, num_vis_bytes);
 	spin_unlock(&adev->mm_stats.lock);
 }
 
@@ -304,7 +331,7 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 				 struct amdgpu_bo *bo)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	u64 initial_bytes_moved;
+	u64 initial_bytes_moved, bytes_moved;
 	uint32_t domain;
 	int r;
 
@@ -314,17 +341,35 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	/* Don't move this buffer if we have depleted our allowance
 	 * to move it. Don't move anything if the threshold is zero.
 	 */
-	if (p->bytes_moved < p->bytes_moved_threshold)
-		domain = bo->prefered_domains;
-	else
+	if (p->bytes_moved < p->bytes_moved_threshold) {
+		if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
+		    (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
+			/* And don't move a CPU_ACCESS_REQUIRED BO to limited
+			 * visible VRAM if we've depleted our allowance to do
+			 * that.
+			 */
+			if (p->bytes_moved_vis < p->bytes_moved_vis_threshold)
+				domain = bo->preferred_domains;
+			else
+				domain = bo->allowed_domains;
+		} else {
+			domain = bo->preferred_domains;
+		}
+	} else {
 		domain = bo->allowed_domains;
+	}
 
 retry:
 	amdgpu_ttm_placement_from_domain(bo, domain);
 	initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
 	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
-	p->bytes_moved += atomic64_read(&adev->num_bytes_moved) -
-		initial_bytes_moved;
+	bytes_moved = atomic64_read(&adev->num_bytes_moved) -
+		      initial_bytes_moved;
+	p->bytes_moved += bytes_moved;
+	if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
+	    bo->tbo.mem.mem_type == TTM_PL_VRAM &&
+	    bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT)
+		p->bytes_moved_vis += bytes_moved;
 
 	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
 		domain = bo->allowed_domains;
@@ -350,7 +395,8 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 		struct amdgpu_bo_list_entry *candidate = p->evictable;
 		struct amdgpu_bo *bo = candidate->robj;
 		struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-		u64 initial_bytes_moved;
+		u64 initial_bytes_moved, bytes_moved;
+		bool update_bytes_moved_vis;
 		uint32_t other;
 
 		/* If we reached our current BO we can forget it */
@@ -370,10 +416,17 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 
 		/* Good we can try to move this BO somewhere else */
 		amdgpu_ttm_placement_from_domain(bo, other);
+		update_bytes_moved_vis =
+			adev->mc.visible_vram_size < adev->mc.real_vram_size &&
+			bo->tbo.mem.mem_type == TTM_PL_VRAM &&
+			bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT;
 		initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
-		p->bytes_moved += atomic64_read(&adev->num_bytes_moved) -
+		bytes_moved = atomic64_read(&adev->num_bytes_moved) -
 			initial_bytes_moved;
+		p->bytes_moved += bytes_moved;
+		if (update_bytes_moved_vis)
+			p->bytes_moved_vis += bytes_moved;
 
 		if (unlikely(r))
 			break;
@@ -554,8 +607,10 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		list_splice(&need_pages, &p->validated);
 	}
 
-	p->bytes_moved_threshold = amdgpu_cs_get_threshold_for_moves(p->adev);
+	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
+					  &p->bytes_moved_vis_threshold);
 	p->bytes_moved = 0;
+	p->bytes_moved_vis = 0;
 	p->evictable = list_last_entry(&p->validated,
 				       struct amdgpu_bo_list_entry,
 				       tv.head);
@@ -579,8 +634,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		goto error_validate;
 	}
 
-	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved);
-
+	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
+				     p->bytes_moved_vis);
 	fpriv->vm.last_eviction_counter =
 		atomic64_read(&p->adev->num_evictions);
 
@@ -619,10 +674,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	}
 
 error_validate:
-	if (r) {
-		amdgpu_vm_move_pt_bos_in_lru(p->adev, &fpriv->vm);
+	if (r)
 		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
-	}
 
 error_free_pages:
 
@@ -670,21 +723,18 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
  * If error is set than unvalidate buffer, otherwise just free memory
  * used by parsing context.
  **/
-static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bool backoff)
+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;
 
-	if (!error) {
-		amdgpu_vm_move_pt_bos_in_lru(parser->adev, &fpriv->vm);
-
+	if (!error)
 		ttm_eu_fence_buffer_objects(&parser->ticket,
 					    &parser->validated,
 					    parser->fence);
-	} else if (backoff) {
+	else if (backoff)
 		ttm_eu_backoff_reservation(&parser->ticket,
 					   &parser->validated);
-	}
 
 	for (i = 0; i < parser->num_post_dep_syncobjs; i++)
 		drm_syncobj_put(parser->post_dep_syncobjs[i]);
@@ -737,7 +787,8 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 
 	if (amdgpu_sriov_vf(adev)) {
 		struct dma_fence *f;
-		bo_va = vm->csa_bo_va;
+
+		bo_va = fpriv->csa_va;
 		BUG_ON(!bo_va);
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
@@ -774,7 +825,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 
 	}
 
-	r = amdgpu_vm_clear_invalids(adev, vm, &p->job->sync);
+	r = amdgpu_vm_clear_moved(adev, vm, &p->job->sync);
 
 	if (amdgpu_vm_debug && p->bo_list) {
 		/* Invalidate all BOs to test for userspace bugs */
@@ -984,7 +1035,7 @@ static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
 {
 	int r;
 	struct dma_fence *fence;
-	r = drm_syncobj_fence_get(p->filp, handle, &fence);
+	r = drm_syncobj_find_fence(p->filp, handle, &fence);
 	if (r)
 		return r;
 
@@ -1383,7 +1434,7 @@ int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 	if (fences == NULL)
 		return -ENOMEM;
 
-	fences_user = (void __user *)(uintptr_t)(wait->in.fences);
+	fences_user = u64_to_user_ptr(wait->in.fences);
 	if (copy_from_user(fences, fences_user,
 		sizeof(struct drm_amdgpu_fence) * fence_count)) {
 		r = -EFAULT;
@@ -1436,7 +1487,7 @@ amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 			    addr > mapping->last)
 				continue;
 
-			*bo = lobj->bo_va->bo;
+			*bo = lobj->bo_va->base.bo;
 			return mapping;
 		}
 
@@ -1445,7 +1496,7 @@ amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 			    addr > mapping->last)
 				continue;
 
-			*bo = lobj->bo_va->bo;
+			*bo = lobj->bo_va->base.bo;
 			return mapping;
 		}
 	}