1 files changed, 385 insertions, 205 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 4e51dce3aab5..2b931e855abd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -61,7 +61,7 @@
 #include "amdgpu_res_cursor.h"
 #include "bif/bif_4_1_d.h"
 
-MODULE_IMPORT_NS(DMA_BUF);
+MODULE_IMPORT_NS("DMA_BUF");
 
 #define AMDGPU_TTM_VRAM_MAX_DW_READ	((size_t)128)
 
@@ -102,23 +102,19 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 	/* Don't handle scatter gather BOs */
 	if (bo->type == ttm_bo_type_sg) {
 		placement->num_placement = 0;
-		placement->num_busy_placement = 0;
 		return;
 	}
 
 	/* Object isn't an AMDGPU object so ignore */
 	if (!amdgpu_bo_is_amdgpu_bo(bo)) {
 		placement->placement = &placements;
-		placement->busy_placement = &placements;
 		placement->num_placement = 1;
-		placement->num_busy_placement = 1;
 		return;
 	}
 
 	abo = ttm_to_amdgpu_bo(bo);
 	if (abo->flags & AMDGPU_GEM_CREATE_DISCARDABLE) {
 		placement->num_placement = 0;
-		placement->num_busy_placement = 0;
 		return;
 	}
 
@@ -127,17 +123,18 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 	case AMDGPU_PL_GWS:
 	case AMDGPU_PL_OA:
 	case AMDGPU_PL_DOORBELL:
+	case AMDGPU_PL_MMIO_REMAP:
 		placement->num_placement = 0;
-		placement->num_busy_placement = 0;
 		return;
 
 	case TTM_PL_VRAM:
 		if (!adev->mman.buffer_funcs_enabled) {
 			/* Move to system memory */
 			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
+
 		} else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
-			   amdgpu_bo_in_cpu_visible_vram(abo)) {
+			   amdgpu_res_cpu_visible(adev, bo->resource)) {
 
 			/* Try evicting to the CPU inaccessible part of VRAM
 			 * first, but only set GTT as busy placement, so this
@@ -149,8 +146,7 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 							AMDGPU_GEM_DOMAIN_CPU);
 			abo->placements[0].fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
 			abo->placements[0].lpfn = 0;
-			abo->placement.busy_placement = &abo->placements[1];
-			abo->placement.num_busy_placement = 1;
+			abo->placements[0].flags |= TTM_PL_FLAG_DESIRED;
 		} else {
 			/* Move to GTT memory */
 			amdgpu_bo_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT |
@@ -192,7 +188,6 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 	struct amdgpu_job *job;
 	void *cpu_addr;
 	uint64_t flags;
-	unsigned int i;
 	int r;
 
 	BUG_ON(adev->mman.buffer_funcs->copy_max_bytes <
@@ -231,7 +226,8 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4 + num_bytes,
-				     AMDGPU_IB_POOL_DELAYED, &job);
+				     AMDGPU_IB_POOL_DELAYED, &job,
+				     AMDGPU_KERNEL_JOB_ID_TTM_MAP_BUFFER);
 	if (r)
 		return r;
 
@@ -241,7 +237,7 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 	dst_addr = amdgpu_bo_gpu_offset(adev->gart.bo);
 	dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8;
 	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
-				dst_addr, num_bytes, false);
+				dst_addr, num_bytes, 0);
 
 	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
 	WARN_ON(job->ibs[0].length_dw > num_dw);
@@ -258,16 +254,9 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 		dma_addr = &bo->ttm->dma_address[mm_cur->start >> PAGE_SHIFT];
 		amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, cpu_addr);
 	} else {
-		dma_addr_t dma_address;
-
-		dma_address = mm_cur->start;
-		dma_address += adev->vm_manager.vram_base_offset;
+		u64 pa = mm_cur->start + adev->vm_manager.vram_base_offset;
 
-		for (i = 0; i < num_pages; ++i) {
-			amdgpu_gart_map(adev, i << PAGE_SHIFT, 1, &dma_address,
-					flags, cpu_addr);
-			dma_address += PAGE_SIZE;
-		}
+		amdgpu_gart_map_vram_range(adev, pa, 0, num_pages, flags, cpu_addr);
 	}
 
 	dma_fence_put(amdgpu_job_submit(job));
@@ -289,20 +278,24 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
  * move and different for a BO to BO copy.
  *
  */
-int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
-			       const struct amdgpu_copy_mem *src,
-			       const struct amdgpu_copy_mem *dst,
-			       uint64_t size, bool tmz,
-			       struct dma_resv *resv,
-			       struct dma_fence **f)
+__attribute__((nonnull))
+static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
+				      const struct amdgpu_copy_mem *src,
+				      const struct amdgpu_copy_mem *dst,
+				      uint64_t size, bool tmz,
+				      struct dma_resv *resv,
+				      struct dma_fence **f)
 {
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
 	struct amdgpu_res_cursor src_mm, dst_mm;
 	struct dma_fence *fence = NULL;
 	int r = 0;
+	uint32_t copy_flags = 0;
+	struct amdgpu_bo *abo_src, *abo_dst;
 
 	if (!adev->mman.buffer_funcs_enabled) {
-		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		dev_err(adev->dev,
+			"Trying to move memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
@@ -311,7 +304,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 
 	mutex_lock(&adev->mman.gtt_window_lock);
 	while (src_mm.remaining) {
-		uint64_t from, to, cur_size;
+		uint64_t from, to, cur_size, tiling_flags;
+		uint32_t num_type, data_format, max_com, write_compress_disable;
 		struct dma_fence *next;
 
 		/* Never copy more than 256MiB at once to avoid a timeout */
@@ -328,8 +322,31 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 		if (r)
 			goto error;
 
-		r = amdgpu_copy_buffer(ring, from, to, cur_size,
-				       resv, &next, false, true, tmz);
+		abo_src = ttm_to_amdgpu_bo(src->bo);
+		abo_dst = ttm_to_amdgpu_bo(dst->bo);
+		if (tmz)
+			copy_flags |= AMDGPU_COPY_FLAGS_TMZ;
+		if ((abo_src->flags & AMDGPU_GEM_CREATE_GFX12_DCC) &&
+		    (abo_src->tbo.resource->mem_type == TTM_PL_VRAM))
+			copy_flags |= AMDGPU_COPY_FLAGS_READ_DECOMPRESSED;
+		if ((abo_dst->flags & AMDGPU_GEM_CREATE_GFX12_DCC) &&
+		    (dst->mem->mem_type == TTM_PL_VRAM)) {
+			copy_flags |= AMDGPU_COPY_FLAGS_WRITE_COMPRESSED;
+			amdgpu_bo_get_tiling_flags(abo_dst, &tiling_flags);
+			max_com = AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_MAX_COMPRESSED_BLOCK);
+			num_type = AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_NUMBER_TYPE);
+			data_format = AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_DATA_FORMAT);
+			write_compress_disable =
+				AMDGPU_TILING_GET(tiling_flags, GFX12_DCC_WRITE_COMPRESS_DISABLE);
+			copy_flags |= (AMDGPU_COPY_FLAGS_SET(MAX_COMPRESSED, max_com) |
+				       AMDGPU_COPY_FLAGS_SET(NUMBER_TYPE, num_type) |
+				       AMDGPU_COPY_FLAGS_SET(DATA_FORMAT, data_format) |
+				       AMDGPU_COPY_FLAGS_SET(WRITE_COMPRESS_DISABLE,
+							     write_compress_disable));
+		}
+
+		r = amdgpu_copy_buffer(ring, from, to, cur_size, resv,
+				       &next, false, true, copy_flags);
 		if (r)
 			goto error;
 
@@ -341,9 +358,7 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev,
 	}
 error:
 	mutex_unlock(&adev->mman.gtt_window_lock);
-	if (f)
-		*f = dma_fence_get(fence);
-	dma_fence_put(fence);
+	*f = fence;
 	return r;
 }
 
@@ -383,11 +398,12 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
 	    (abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE)) {
 		struct dma_fence *wipe_fence = NULL;
 
-		r = amdgpu_fill_buffer(abo, AMDGPU_POISON, NULL, &wipe_fence,
-					false);
+		r = amdgpu_fill_buffer(abo, 0, NULL, &wipe_fence,
+				       false, AMDGPU_KERNEL_JOB_ID_MOVE_BLIT);
 		if (r) {
 			goto error;
 		} else if (wipe_fence) {
+			amdgpu_vram_mgr_set_cleared(bo->resource);
 			dma_fence_put(fence);
 			fence = wipe_fence;
 		}
@@ -408,40 +424,56 @@ error:
 	return r;
 }
 
-/*
- * amdgpu_mem_visible - Check that memory can be accessed by ttm_bo_move_memcpy
+/**
+ * amdgpu_res_cpu_visible - Check that resource can be accessed by CPU
+ * @adev: amdgpu device
+ * @res: the resource to check
  *
- * Called by amdgpu_bo_move()
+ * Returns: true if the full resource is CPU visible, false otherwise.
  */
-static bool amdgpu_mem_visible(struct amdgpu_device *adev,
-			       struct ttm_resource *mem)
+bool amdgpu_res_cpu_visible(struct amdgpu_device *adev,
+			    struct ttm_resource *res)
 {
-	u64 mem_size = (u64)mem->size;
 	struct amdgpu_res_cursor cursor;
-	u64 end;
 
-	if (mem->mem_type == TTM_PL_SYSTEM ||
-	    mem->mem_type == TTM_PL_TT)
+	if (!res)
+		return false;
+
+	if (res->mem_type == TTM_PL_SYSTEM || res->mem_type == TTM_PL_TT ||
+	    res->mem_type == AMDGPU_PL_PREEMPT || res->mem_type == AMDGPU_PL_DOORBELL ||
+	    res->mem_type == AMDGPU_PL_MMIO_REMAP)
 		return true;
-	if (mem->mem_type != TTM_PL_VRAM)
+
+	if (res->mem_type != TTM_PL_VRAM)
 		return false;
 
-	amdgpu_res_first(mem, 0, mem_size, &cursor);
-	end = cursor.start + cursor.size;
+	amdgpu_res_first(res, 0, res->size, &cursor);
 	while (cursor.remaining) {
+		if ((cursor.start + cursor.size) > adev->gmc.visible_vram_size)
+			return false;
 		amdgpu_res_next(&cursor, cursor.size);
+	}
 
-		if (!cursor.remaining)
-			break;
+	return true;
+}
 
-		/* ttm_resource_ioremap only supports contiguous memory */
-		if (end != cursor.start)
-			return false;
+/*
+ * amdgpu_res_copyable - Check that memory can be accessed by ttm_bo_move_memcpy
+ *
+ * Called by amdgpu_bo_move()
+ */
+static bool amdgpu_res_copyable(struct amdgpu_device *adev,
+				struct ttm_resource *mem)
+{
+	if (!amdgpu_res_cpu_visible(adev, mem))
+		return false;
 
-		end = cursor.start + cursor.size;
-	}
+	/* ttm_resource_ioremap only supports contiguous memory */
+	if (mem->mem_type == TTM_PL_VRAM &&
+	    !(mem->placement & TTM_PL_FLAG_CONTIGUOUS))
+		return false;
 
-	return end <= adev->gmc.visible_vram_size;
+	return true;
 }
 
 /*
@@ -471,14 +503,16 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
 
 	if (!old_mem || (old_mem->mem_type == TTM_PL_SYSTEM &&
 			 bo->ttm == NULL)) {
+		amdgpu_bo_move_notify(bo, evict, new_mem);
 		ttm_bo_move_null(bo, new_mem);
-		goto out;
+		return 0;
 	}
 	if (old_mem->mem_type == TTM_PL_SYSTEM &&
 	    (new_mem->mem_type == TTM_PL_TT ||
 	     new_mem->mem_type == AMDGPU_PL_PREEMPT)) {
+		amdgpu_bo_move_notify(bo, evict, new_mem);
 		ttm_bo_move_null(bo, new_mem);
-		goto out;
+		return 0;
 	}
 	if ((old_mem->mem_type == TTM_PL_TT ||
 	     old_mem->mem_type == AMDGPU_PL_PREEMPT) &&
@@ -488,22 +522,26 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
 			return r;
 
 		amdgpu_ttm_backend_unbind(bo->bdev, bo->ttm);
+		amdgpu_bo_move_notify(bo, evict, new_mem);
 		ttm_resource_free(bo, &bo->resource);
 		ttm_bo_assign_mem(bo, new_mem);
-		goto out;
+		return 0;
 	}
 
 	if (old_mem->mem_type == AMDGPU_PL_GDS ||
 	    old_mem->mem_type == AMDGPU_PL_GWS ||
 	    old_mem->mem_type == AMDGPU_PL_OA ||
 	    old_mem->mem_type == AMDGPU_PL_DOORBELL ||
+	    old_mem->mem_type == AMDGPU_PL_MMIO_REMAP ||
 	    new_mem->mem_type == AMDGPU_PL_GDS ||
 	    new_mem->mem_type == AMDGPU_PL_GWS ||
 	    new_mem->mem_type == AMDGPU_PL_OA ||
-	    new_mem->mem_type == AMDGPU_PL_DOORBELL) {
+	    new_mem->mem_type == AMDGPU_PL_DOORBELL ||
+	    new_mem->mem_type == AMDGPU_PL_MMIO_REMAP) {
 		/* Nothing to save here */
+		amdgpu_bo_move_notify(bo, evict, new_mem);
 		ttm_bo_move_null(bo, new_mem);
-		goto out;
+		return 0;
 	}
 
 	if (bo->type == ttm_bo_type_device &&
@@ -515,27 +553,28 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
 		abo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 	}
 
-	if (adev->mman.buffer_funcs_enabled) {
-		if (((old_mem->mem_type == TTM_PL_SYSTEM &&
-		      new_mem->mem_type == TTM_PL_VRAM) ||
-		     (old_mem->mem_type == TTM_PL_VRAM &&
-		      new_mem->mem_type == TTM_PL_SYSTEM))) {
-			hop->fpfn = 0;
-			hop->lpfn = 0;
-			hop->mem_type = TTM_PL_TT;
-			hop->flags = TTM_PL_FLAG_TEMPORARY;
-			return -EMULTIHOP;
-		}
+	if (adev->mman.buffer_funcs_enabled &&
+	    ((old_mem->mem_type == TTM_PL_SYSTEM &&
+	      new_mem->mem_type == TTM_PL_VRAM) ||
+	     (old_mem->mem_type == TTM_PL_VRAM &&
+	      new_mem->mem_type == TTM_PL_SYSTEM))) {
+		hop->fpfn = 0;
+		hop->lpfn = 0;
+		hop->mem_type = TTM_PL_TT;
+		hop->flags = TTM_PL_FLAG_TEMPORARY;
+		return -EMULTIHOP;
+	}
 
+	amdgpu_bo_move_notify(bo, evict, new_mem);
+	if (adev->mman.buffer_funcs_enabled)
 		r = amdgpu_move_blit(bo, evict, new_mem, old_mem);
-	} else {
+	else
 		r = -ENODEV;
-	}
 
 	if (r) {
 		/* Check that all memory is CPU accessible */
-		if (!amdgpu_mem_visible(adev, old_mem) ||
-		    !amdgpu_mem_visible(adev, new_mem)) {
+		if (!amdgpu_res_copyable(adev, old_mem) ||
+		    !amdgpu_res_copyable(adev, new_mem)) {
 			pr_err("Move buffer fallback to memcpy unavailable\n");
 			return r;
 		}
@@ -545,10 +584,10 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
 			return r;
 	}
 
-out:
-	/* update statistics */
+	/* update statistics after the move */
+	if (evict)
+		atomic64_inc(&adev->num_evictions);
 	atomic64_add(bo->base.size, &adev->num_bytes_moved);
-	amdgpu_bo_move_notify(bo, evict, new_mem);
 	return 0;
 }
 
@@ -561,7 +600,6 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev,
 				     struct ttm_resource *mem)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bdev);
-	size_t bus_size = (size_t)mem->size;
 
 	switch (mem->mem_type) {
 	case TTM_PL_SYSTEM:
@@ -572,9 +610,6 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev,
 		break;
 	case TTM_PL_VRAM:
 		mem->bus.offset = mem->start << PAGE_SHIFT;
-		/* check if it's visible */
-		if ((mem->bus.offset + bus_size) > adev->gmc.visible_vram_size)
-			return -EINVAL;
 
 		if (adev->mman.aper_base_kaddr &&
 		    mem->placement & TTM_PL_FLAG_CONTIGUOUS)
@@ -590,6 +625,12 @@ static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev,
 		mem->bus.is_iomem = true;
 		mem->bus.caching = ttm_uncached;
 		break;
+	case AMDGPU_PL_MMIO_REMAP:
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		mem->bus.offset += adev->rmmio_remap.bus_addr;
+		mem->bus.is_iomem = true;
+		mem->bus.caching = ttm_uncached;
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -607,6 +648,8 @@ static unsigned long amdgpu_ttm_io_mem_pfn(struct ttm_buffer_object *bo,
 
 	if (bo->resource->mem_type == AMDGPU_PL_DOORBELL)
 		return ((uint64_t)(adev->doorbell.base + cursor.start)) >> PAGE_SHIFT;
+	else if (bo->resource->mem_type == AMDGPU_PL_MMIO_REMAP)
+		return ((uint64_t)(adev->rmmio_remap.bus_addr + cursor.start)) >> PAGE_SHIFT;
 
 	return (adev->gmc.aper_base + cursor.start) >> PAGE_SHIFT;
 }
@@ -654,10 +697,11 @@ struct amdgpu_ttm_tt {
  * memory and start HMM tracking CPU page table update
  *
  * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only
- * once afterwards to stop HMM tracking
+ * once afterwards to stop HMM tracking. Its the caller responsibility to ensure
+ * that range is a valid memory and it is freed too.
  */
-int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
-				 struct hmm_range **range)
+int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo,
+				 struct amdgpu_hmm_range *range)
 {
 	struct ttm_tt *ttm = bo->tbo.ttm;
 	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
@@ -667,9 +711,6 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
 	bool readonly;
 	int r = 0;
 
-	/* Make sure get_user_pages_done() can cleanup gracefully */
-	*range = NULL;
-
 	mm = bo->notifier.mm;
 	if (unlikely(!mm)) {
 		DRM_DEBUG_DRIVER("BO is not registered?\n");
@@ -693,7 +734,7 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages,
 
 	readonly = amdgpu_ttm_tt_is_readonly(ttm);
 	r = amdgpu_hmm_range_get_pages(&bo->notifier, start, ttm->num_pages,
-				       readonly, NULL, pages, range);
+				       readonly, NULL, range);
 out_unlock:
 	mmap_read_unlock(mm);
 	if (r)
@@ -704,38 +745,6 @@ out_unlock:
 	return r;
 }
 
-/* amdgpu_ttm_tt_discard_user_pages - Discard range and pfn array allocations
- */
-void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm,
-				      struct hmm_range *range)
-{
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-
-	if (gtt && gtt->userptr && range)
-		amdgpu_hmm_range_get_pages_done(range);
-}
-
-/*
- * amdgpu_ttm_tt_get_user_pages_done - stop HMM track the CPU page table change
- * Check if the pages backing this ttm range have been invalidated
- *
- * Returns: true if pages are still valid
- */
-bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
-				       struct hmm_range *range)
-{
-	struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm);
-
-	if (!gtt || !gtt->userptr || !range)
-		return false;
-
-	DRM_DEBUG_DRIVER("user_pages_done 0x%llx pages 0x%x\n",
-		gtt->userptr, ttm->num_pages);
-
-	WARN_ONCE(!range->hmm_pfns, "No user pages to check\n");
-
-	return !amdgpu_hmm_range_get_pages_done(range);
-}
 #endif
 
 /*
@@ -745,12 +754,12 @@ bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm,
  * that backs user memory and will ultimately be mapped into the device
  * address space.
  */
-void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages)
+void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct amdgpu_hmm_range *range)
 {
 	unsigned long i;
 
 	for (i = 0; i < ttm->num_pages; ++i)
-		ttm->pages[i] = pages ? pages[i] : NULL;
+		ttm->pages[i] = range ? hmm_pfn_to_page(range->hmm_range.hmm_pfns[i]) : NULL;
 }
 
 /*
@@ -778,7 +787,7 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_device *bdev,
 	/* Map SG to device */
 	r = dma_map_sgtable(adev->dev, ttm->sg, direction, 0);
 	if (r)
-		goto release_sg;
+		goto release_sg_table;
 
 	/* convert SG to linear array of pages and dma addresses */
 	drm_prime_sg_to_dma_addr_array(ttm->sg, gtt->ttm.dma_address,
@@ -786,6 +795,8 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_device *bdev,
 
 	return 0;
 
+release_sg_table:
+	sg_free_table(ttm->sg);
 release_sg:
 	kfree(ttm->sg);
 	ttm->sg = NULL;
@@ -828,8 +839,7 @@ static void amdgpu_ttm_gart_bind_gfx9_mqd(struct amdgpu_device *adev,
 	int num_xcc = max(1U, adev->gfx.num_xcc_per_xcp);
 	uint64_t page_idx, pages_per_xcc;
 	int i;
-	uint64_t ctrl_flags = (flags & ~AMDGPU_PTE_MTYPE_VG10_MASK) |
-			AMDGPU_PTE_MTYPE_VG10(AMDGPU_MTYPE_NC);
+	uint64_t ctrl_flags = AMDGPU_PTE_MTYPE_VG10(flags, AMDGPU_MTYPE_NC);
 
 	pages_per_xcc = total_pages;
 	do_div(pages_per_xcc, num_xcc);
@@ -868,6 +878,7 @@ static void amdgpu_ttm_gart_bind(struct amdgpu_device *adev,
 		amdgpu_gart_bind(adev, gtt->offset, ttm->num_pages,
 				 gtt->ttm.dma_address, flags);
 	}
+	gtt->bound = true;
 }
 
 /*
@@ -894,7 +905,7 @@ static int amdgpu_ttm_backend_bind(struct ttm_device *bdev,
 	if (gtt->userptr) {
 		r = amdgpu_ttm_tt_pin_userptr(bdev, ttm);
 		if (r) {
-			DRM_ERROR("failed to pin userptr\n");
+			dev_err(adev->dev, "failed to pin userptr\n");
 			return r;
 		}
 	} else if (ttm->page_flags & TTM_TT_FLAG_EXTERNAL) {
@@ -959,16 +970,12 @@ int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo)
 		return 0;
 
 	addr = amdgpu_gmc_agp_addr(bo);
-	if (addr != AMDGPU_BO_INVALID_OFFSET) {
-		bo->resource->start = addr >> PAGE_SHIFT;
+	if (addr != AMDGPU_BO_INVALID_OFFSET)
 		return 0;
-	}
 
 	/* allocate GART space */
 	placement.num_placement = 1;
 	placement.placement = &placements;
-	placement.num_busy_placement = 1;
-	placement.busy_placement = &placements;
 	placements.fpfn = 0;
 	placements.lpfn = adev->gmc.gart_size >> PAGE_SHIFT;
 	placements.mem_type = TTM_PL_TT;
@@ -1024,7 +1031,7 @@ static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev,
 	/* if the pages have userptr pinning then clear that first */
 	if (gtt->userptr) {
 		amdgpu_ttm_tt_unpin_userptr(bdev, ttm);
-	} else if (ttm->sg && gtt->gobj->import_attach) {
+	} else if (ttm->sg && drm_gem_is_imported(gtt->gobj)) {
 		struct dma_buf_attachment *attach;
 
 		attach = gtt->gobj->import_attach;
@@ -1318,10 +1325,11 @@ uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem)
 
 	if (mem && (mem->mem_type == TTM_PL_TT ||
 		    mem->mem_type == AMDGPU_PL_DOORBELL ||
-		    mem->mem_type == AMDGPU_PL_PREEMPT)) {
+		    mem->mem_type == AMDGPU_PL_PREEMPT ||
+		    mem->mem_type == AMDGPU_PL_MMIO_REMAP)) {
 		flags |= AMDGPU_PTE_SYSTEM;
 
-		if (ttm->caching == ttm_cached)
+		if (ttm && ttm->caching == ttm_cached)
 			flags |= AMDGPU_PTE_SNOOPED;
 	}
 
@@ -1387,7 +1395,8 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo,
 	 */
 	dma_resv_for_each_fence(&resv_cursor, bo->base.resv,
 				DMA_RESV_USAGE_BOOKKEEP, f) {
-		if (amdkfd_fence_check_mm(f, current->mm))
+		if (amdkfd_fence_check_mm(f, current->mm) &&
+		    !(place->flags & TTM_PL_FLAG_CONTIGUOUS))
 			return false;
 	}
 
@@ -1472,10 +1481,12 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4, AMDGPU_IB_POOL_DELAYED,
-				     &job);
+				     &job,
+				     AMDGPU_KERNEL_JOB_ID_TTM_ACCESS_MEMORY_SDMA);
 	if (r)
 		goto out;
 
+	mutex_lock(&adev->mman.gtt_window_lock);
 	amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm);
 	src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) +
 		src_mm.start;
@@ -1484,12 +1495,13 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 		swap(src_addr, dst_addr);
 
 	amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr, dst_addr,
-				PAGE_SIZE, false);
+				PAGE_SIZE, 0);
 
 	amdgpu_ring_pad_ib(adev->mman.buffer_funcs_ring, &job->ibs[0]);
 	WARN_ON(job->ibs[0].length_dw > num_dw);
 
 	fence = amdgpu_job_submit(job);
+	mutex_unlock(&adev->mman.gtt_window_lock);
 
 	if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout))
 		r = -ETIMEDOUT;
@@ -1727,7 +1739,10 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev)
 		reserve_size =
 			amdgpu_atomfirmware_get_fw_reserved_fb_size(adev);
 
-	if (!adev->bios && adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3))
+	if (!adev->bios &&
+	    (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)))
 		reserve_size = max(reserve_size, (uint32_t)280 << 20);
 	else if (!reserve_size)
 		reserve_size = DISCOVERY_TMR_OFFSET;
@@ -1741,25 +1756,21 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev)
 						 &ctx->c2p_bo,
 						 NULL);
 		if (ret) {
-			DRM_ERROR("alloc c2p_bo failed(%d)!\n", ret);
+			dev_err(adev->dev, "alloc c2p_bo failed(%d)!\n", ret);
 			amdgpu_ttm_training_reserve_vram_fini(adev);
 			return ret;
 		}
 		ctx->init = PSP_MEM_TRAIN_RESERVE_SUCCESS;
 	}
 
-	if (!adev->gmc.is_app_apu) {
-		ret = amdgpu_bo_create_kernel_at(
-			adev, adev->gmc.real_vram_size - reserve_size,
-			reserve_size, &adev->mman.fw_reserved_memory, NULL);
-		if (ret) {
-			DRM_ERROR("alloc tmr failed(%d)!\n", ret);
-			amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory,
-					      NULL, NULL);
-			return ret;
-		}
-	} else {
-		DRM_DEBUG_DRIVER("backdoor fw loading path for PSP TMR, no reservation needed\n");
+	ret = amdgpu_bo_create_kernel_at(
+		adev, adev->gmc.real_vram_size - reserve_size, reserve_size,
+		&adev->mman.fw_reserved_memory, NULL);
+	if (ret) {
+		dev_err(adev->dev, "alloc tmr failed(%d)!\n", ret);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL,
+				      NULL);
+		return ret;
 	}
 
 	return 0;
@@ -1781,7 +1792,7 @@ static int amdgpu_ttm_pools_init(struct amdgpu_device *adev)
 	for (i = 0; i < adev->gmc.num_mem_partitions; i++) {
 		ttm_pool_init(&adev->mman.ttm_pools[i], adev->dev,
 			      adev->gmc.mem_partitions[i].numa.node,
-			      false, false);
+			      TTM_ALLOCATION_POOL_BENEFICIAL_ORDER(get_order(SZ_2M)));
 	}
 	return 0;
 }
@@ -1800,6 +1811,59 @@ static void amdgpu_ttm_pools_fini(struct amdgpu_device *adev)
 	adev->mman.ttm_pools = NULL;
 }
 
+/**
+ * amdgpu_ttm_mmio_remap_bo_init - Allocate the singleton 4K MMIO_REMAP BO
+ * @adev: amdgpu device
+ *
+ * Allocates a one-page (4K) GEM BO in AMDGPU_GEM_DOMAIN_MMIO_REMAP when the
+ * hardware exposes a remap base (adev->rmmio_remap.bus_addr) and the host
+ * PAGE_SIZE is <= AMDGPU_GPU_PAGE_SIZE (4K). The BO is created as a regular
+ * GEM object (amdgpu_bo_create).
+ *
+ * Return:
+ *  * 0 on success or intentional skip (feature not present/unsupported)
+ *  * negative errno on allocation failure
+ */
+static int amdgpu_ttm_mmio_remap_bo_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo_param bp;
+	int r;
+
+	/* Skip if HW doesn't expose remap, or if PAGE_SIZE > AMDGPU_GPU_PAGE_SIZE (4K). */
+	if (!adev->rmmio_remap.bus_addr || PAGE_SIZE > AMDGPU_GPU_PAGE_SIZE)
+		return 0;
+
+	memset(&bp, 0, sizeof(bp));
+
+	/* Create exactly one GEM BO in the MMIO_REMAP domain. */
+	bp.type        = ttm_bo_type_device;          /* userspace-mappable GEM */
+	bp.size        = AMDGPU_GPU_PAGE_SIZE;        /* 4K */
+	bp.byte_align  = AMDGPU_GPU_PAGE_SIZE;
+	bp.domain      = AMDGPU_GEM_DOMAIN_MMIO_REMAP;
+	bp.flags       = 0;
+	bp.resv        = NULL;
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
+
+	r = amdgpu_bo_create(adev, &bp, &adev->rmmio_remap.bo);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * amdgpu_ttm_mmio_remap_bo_fini - Free the singleton MMIO_REMAP BO
+ * @adev: amdgpu device
+ *
+ * Frees the kernel-owned MMIO_REMAP BO if it was allocated by
+ * amdgpu_ttm_mmio_remap_bo_init().
+ */
+static void amdgpu_ttm_mmio_remap_bo_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->rmmio_remap.bo);
+	adev->rmmio_remap.bo = NULL;
+}
+
 /*
  * amdgpu_ttm_init - Init the memory management (ttm) as well as various
  * gtt/vram related fields.
@@ -1816,29 +1880,36 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 
 	mutex_init(&adev->mman.gtt_window_lock);
 
+	dma_set_max_seg_size(adev->dev, UINT_MAX);
 	/* No others user of address space so set it to 0 */
 	r = ttm_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev,
 			       adev_to_drm(adev)->anon_inode->i_mapping,
 			       adev_to_drm(adev)->vma_offset_manager,
-			       adev->need_swiotlb,
-			       dma_addressing_limited(adev->dev));
+			       (adev->need_swiotlb ?
+				TTM_ALLOCATION_POOL_USE_DMA_ALLOC : 0) |
+			       (dma_addressing_limited(adev->dev) ?
+				TTM_ALLOCATION_POOL_USE_DMA32 : 0) |
+			       TTM_ALLOCATION_POOL_BENEFICIAL_ORDER(get_order(SZ_2M)));
 	if (r) {
-		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
+		dev_err(adev->dev,
+			"failed initializing buffer object driver(%d).\n", r);
 		return r;
 	}
 
 	r = amdgpu_ttm_pools_init(adev);
 	if (r) {
-		DRM_ERROR("failed to init ttm pools(%d).\n", r);
+		dev_err(adev->dev, "failed to init ttm pools(%d).\n", r);
 		return r;
 	}
 	adev->mman.initialized = true;
 
-	/* Initialize VRAM pool with all of VRAM divided into pages */
-	r = amdgpu_vram_mgr_init(adev);
-	if (r) {
-		DRM_ERROR("Failed initializing VRAM heap.\n");
-		return r;
+	if (!adev->gmc.is_app_apu) {
+		/* Initialize VRAM pool with all of VRAM divided into pages */
+		r = amdgpu_vram_mgr_init(adev);
+		if (r) {
+			dev_err(adev->dev, "Failed initializing VRAM heap.\n");
+			return r;
+		}
 	}
 
 	/* Change the size here instead of the init above so only lpfn is affected */
@@ -1867,19 +1938,19 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 		return r;
 
 	/*
-	 *The reserved vram for driver must be pinned to the specified
-	 *place on the VRAM, so reserve it early.
+	 * The reserved VRAM for the driver must be pinned to a specific
+	 * location in VRAM, so reserve it early.
 	 */
 	r = amdgpu_ttm_drv_reserve_vram_init(adev);
 	if (r)
 		return r;
 
 	/*
-	 * only NAVI10 and onwards ASIC support for IP discovery.
-	 * If IP discovery enabled, a block of memory should be
-	 * reserved for IP discovey.
+	 * only NAVI10 and later ASICs support IP discovery.
+	 * If IP discovery is enabled, a block of memory should be
+	 * reserved for it.
 	 */
-	if (adev->mman.discovery_bin) {
+	if (adev->discovery.reserve_tmr) {
 		r = amdgpu_ttm_reserve_tmr(adev);
 		if (r)
 			return r;
@@ -1917,63 +1988,89 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 		DRM_DEBUG_DRIVER("Skipped stolen memory reservation\n");
 	}
 
-	DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
+	dev_info(adev->dev, "amdgpu: %uM of VRAM memory ready\n",
 		 (unsigned int)(adev->gmc.real_vram_size / (1024 * 1024)));
 
 	/* Compute GTT size, either based on TTM limit
 	 * or whatever the user passed on module init.
 	 */
-	if (amdgpu_gtt_size == -1)
-		gtt_size = ttm_tt_pages_limit() << PAGE_SHIFT;
-	else
-		gtt_size = (uint64_t)amdgpu_gtt_size << 20;
+	gtt_size = ttm_tt_pages_limit() << PAGE_SHIFT;
+	if (amdgpu_gtt_size != -1) {
+		uint64_t configured_size = (uint64_t)amdgpu_gtt_size << 20;
+
+		drm_warn(&adev->ddev,
+			"Configuring gttsize via module parameter is deprecated, please use ttm.pages_limit\n");
+		if (gtt_size != configured_size)
+			drm_warn(&adev->ddev,
+				"GTT size has been set as %llu but TTM size has been set as %llu, this is unusual\n",
+				configured_size, gtt_size);
+
+		gtt_size = configured_size;
+	}
 
 	/* Initialize GTT memory pool */
 	r = amdgpu_gtt_mgr_init(adev, gtt_size);
 	if (r) {
-		DRM_ERROR("Failed initializing GTT heap.\n");
+		dev_err(adev->dev, "Failed initializing GTT heap.\n");
 		return r;
 	}
-	DRM_INFO("amdgpu: %uM of GTT memory ready.\n",
+	dev_info(adev->dev, "amdgpu: %uM of GTT memory ready.\n",
 		 (unsigned int)(gtt_size / (1024 * 1024)));
 
-	/* Initiailize doorbell pool on PCI BAR */
+	if (adev->flags & AMD_IS_APU) {
+		if (adev->gmc.real_vram_size < gtt_size)
+			adev->apu_prefer_gtt = true;
+	}
+
+	/* Initialize doorbell pool on PCI BAR */
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_DOORBELL, adev->doorbell.size / PAGE_SIZE);
 	if (r) {
-		DRM_ERROR("Failed initializing doorbell heap.\n");
+		dev_err(adev->dev, "Failed initializing doorbell heap.\n");
 		return r;
 	}
 
 	/* Create a boorbell page for kernel usages */
 	r = amdgpu_doorbell_create_kernel_doorbells(adev);
 	if (r) {
-		DRM_ERROR("Failed to initialize kernel doorbells.\n");
+		dev_err(adev->dev, "Failed to initialize kernel doorbells.\n");
 		return r;
 	}
 
+	/* Initialize MMIO-remap pool (single page 4K) */
+	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_MMIO_REMAP, 1);
+	if (r) {
+		dev_err(adev->dev, "Failed initializing MMIO-remap heap.\n");
+		return r;
+	}
+
+	/* Allocate the singleton MMIO_REMAP BO (4K) if supported */
+	r = amdgpu_ttm_mmio_remap_bo_init(adev);
+	if (r)
+		return r;
+
 	/* Initialize preemptible memory pool */
 	r = amdgpu_preempt_mgr_init(adev);
 	if (r) {
-		DRM_ERROR("Failed initializing PREEMPT heap.\n");
+		dev_err(adev->dev, "Failed initializing PREEMPT heap.\n");
 		return r;
 	}
 
 	/* Initialize various on-chip memory pools */
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GDS, adev->gds.gds_size);
 	if (r) {
-		DRM_ERROR("Failed initializing GDS heap.\n");
+		dev_err(adev->dev, "Failed initializing GDS heap.\n");
 		return r;
 	}
 
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GWS, adev->gds.gws_size);
 	if (r) {
-		DRM_ERROR("Failed initializing gws heap.\n");
+		dev_err(adev->dev, "Failed initializing gws heap.\n");
 		return r;
 	}
 
 	r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_OA, adev->gds.oa_size);
 	if (r) {
-		DRM_ERROR("Failed initializing oa heap.\n");
+		dev_err(adev->dev, "Failed initializing oa heap.\n");
 		return r;
 	}
 	if (amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE,
@@ -2005,12 +2102,16 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev)
 		/* return the FW reserved memory back to VRAM */
 		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL,
 				      NULL);
+		amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL,
+				      NULL);
 		if (adev->mman.stolen_reserved_size)
 			amdgpu_bo_free_kernel(&adev->mman.stolen_reserved_memory,
 					      NULL, NULL);
 	}
 	amdgpu_bo_free_kernel(&adev->mman.sdma_access_bo, NULL,
 					&adev->mman.sdma_access_ptr);
+
+	amdgpu_ttm_mmio_remap_bo_fini(adev);
 	amdgpu_ttm_fw_reserve_vram_fini(adev);
 	amdgpu_ttm_drv_reserve_vram_fini(adev);
 
@@ -2023,15 +2124,20 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev)
 		drm_dev_exit(idx);
 	}
 
-	amdgpu_vram_mgr_fini(adev);
+	if (!adev->gmc.is_app_apu)
+		amdgpu_vram_mgr_fini(adev);
 	amdgpu_gtt_mgr_fini(adev);
 	amdgpu_preempt_mgr_fini(adev);
+	amdgpu_doorbell_fini(adev);
+
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GDS);
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GWS);
 	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_OA);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_DOORBELL);
+	ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_MMIO_REMAP);
 	ttm_device_fini(&adev->mman.bdev);
 	adev->mman.initialized = false;
-	DRM_INFO("amdgpu: ttm finalized\n");
+	dev_info(adev->dev, "amdgpu: ttm finalized\n");
 }
 
 /**
@@ -2063,8 +2169,9 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 					  DRM_SCHED_PRIORITY_KERNEL, &sched,
 					  1, NULL);
 		if (r) {
-			DRM_ERROR("Failed setting up TTM BO move entity (%d)\n",
-				  r);
+			dev_err(adev->dev,
+				"Failed setting up TTM BO move entity (%d)\n",
+				r);
 			return;
 		}
 
@@ -2072,15 +2179,18 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 					  DRM_SCHED_PRIORITY_NORMAL, &sched,
 					  1, NULL);
 		if (r) {
-			DRM_ERROR("Failed setting up TTM BO move entity (%d)\n",
-				  r);
+			dev_err(adev->dev,
+				"Failed setting up TTM BO move entity (%d)\n",
+				r);
 			goto error_free_entity;
 		}
 	} else {
 		drm_sched_entity_destroy(&adev->mman.high_pr);
 		drm_sched_entity_destroy(&adev->mman.low_pr);
-		dma_fence_put(man->move);
-		man->move = NULL;
+		/* Drop all the old fences since re-creating the scheduler entities
+		 * will allocate new contexts.
+		 */
+		ttm_resource_manager_cleanup(man);
 	}
 
 	/* this just adjusts TTM size idea, which sets lpfn to the correct value */
@@ -2103,7 +2213,7 @@ static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
 				  struct dma_resv *resv,
 				  bool vm_needs_flush,
 				  struct amdgpu_job **job,
-				  bool delayed)
+				  bool delayed, u64 k_job_id)
 {
 	enum amdgpu_ib_pool_type pool = direct_submit ?
 		AMDGPU_IB_POOL_DIRECT :
@@ -2113,7 +2223,7 @@ static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
 						    &adev->mman.high_pr;
 	r = amdgpu_job_alloc_with_ib(adev, entity,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
-				     num_dw * 4, pool, job);
+				     num_dw * 4, pool, job, k_job_id);
 	if (r)
 		return r;
 
@@ -2134,7 +2244,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 		       uint64_t dst_offset, uint32_t byte_count,
 		       struct dma_resv *resv,
 		       struct dma_fence **fence, bool direct_submit,
-		       bool vm_needs_flush, bool tmz)
+		       bool vm_needs_flush, uint32_t copy_flags)
 {
 	struct amdgpu_device *adev = ring->adev;
 	unsigned int num_loops, num_dw;
@@ -2144,7 +2254,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 	int r;
 
 	if (!direct_submit && !ring->sched.ready) {
-		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		dev_err(adev->dev,
+			"Trying to move memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
@@ -2152,7 +2263,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 	num_loops = DIV_ROUND_UP(byte_count, max_bytes);
 	num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->copy_num_dw, 8);
 	r = amdgpu_ttm_prepare_job(adev, direct_submit, num_dw,
-				   resv, vm_needs_flush, &job, false);
+				   resv, vm_needs_flush, &job, false,
+				   AMDGPU_KERNEL_JOB_ID_TTM_COPY_BUFFER);
 	if (r)
 		return r;
 
@@ -2160,8 +2272,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 		uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
 
 		amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_offset,
-					dst_offset, cur_size_in_bytes, tmz);
-
+					dst_offset, cur_size_in_bytes, copy_flags);
 		src_offset += cur_size_in_bytes;
 		dst_offset += cur_size_in_bytes;
 		byte_count -= cur_size_in_bytes;
@@ -2180,7 +2291,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 
 error_free:
 	amdgpu_job_free(job);
-	DRM_ERROR("Error scheduling IBs (%d)\n", r);
+	dev_err(adev->dev, "Error scheduling IBs (%d)\n", r);
 	return r;
 }
 
@@ -2188,7 +2299,8 @@ static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data,
 			       uint64_t dst_addr, uint32_t byte_count,
 			       struct dma_resv *resv,
 			       struct dma_fence **fence,
-			       bool vm_needs_flush, bool delayed)
+			       bool vm_needs_flush, bool delayed,
+			       u64 k_job_id)
 {
 	struct amdgpu_device *adev = ring->adev;
 	unsigned int num_loops, num_dw;
@@ -2201,7 +2313,7 @@ static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data,
 	num_loops = DIV_ROUND_UP_ULL(byte_count, max_bytes);
 	num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->fill_num_dw, 8);
 	r = amdgpu_ttm_prepare_job(adev, false, num_dw, resv, vm_needs_flush,
-				   &job, delayed);
+				   &job, delayed, k_job_id);
 	if (r)
 		return r;
 
@@ -2221,11 +2333,78 @@ static int amdgpu_ttm_fill_mem(struct amdgpu_ring *ring, uint32_t src_data,
 	return 0;
 }
 
+/**
+ * amdgpu_ttm_clear_buffer - clear memory buffers
+ * @bo: amdgpu buffer object
+ * @resv: reservation object
+ * @fence: dma_fence associated with the operation
+ *
+ * Clear the memory buffer resource.
+ *
+ * Returns:
+ * 0 for success or a negative error code on failure.
+ */
+int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo,
+			    struct dma_resv *resv,
+			    struct dma_fence **fence)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct amdgpu_res_cursor cursor;
+	u64 addr;
+	int r = 0;
+
+	if (!adev->mman.buffer_funcs_enabled)
+		return -EINVAL;
+
+	if (!fence)
+		return -EINVAL;
+
+	*fence = dma_fence_get_stub();
+
+	amdgpu_res_first(bo->tbo.resource, 0, amdgpu_bo_size(bo), &cursor);
+
+	mutex_lock(&adev->mman.gtt_window_lock);
+	while (cursor.remaining) {
+		struct dma_fence *next = NULL;
+		u64 size;
+
+		if (amdgpu_res_cleared(&cursor)) {
+			amdgpu_res_next(&cursor, cursor.size);
+			continue;
+		}
+
+		/* Never clear more than 256MiB at once to avoid timeouts */
+		size = min(cursor.size, 256ULL << 20);
+
+		r = amdgpu_ttm_map_buffer(&bo->tbo, bo->tbo.resource, &cursor,
+					  1, ring, false, &size, &addr);
+		if (r)
+			goto err;
+
+		r = amdgpu_ttm_fill_mem(ring, 0, addr, size, resv,
+					&next, true, true,
+					AMDGPU_KERNEL_JOB_ID_TTM_CLEAR_BUFFER);
+		if (r)
+			goto err;
+
+		dma_fence_put(*fence);
+		*fence = next;
+
+		amdgpu_res_next(&cursor, size);
+	}
+err:
+	mutex_unlock(&adev->mman.gtt_window_lock);
+
+	return r;
+}
+
 int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 			uint32_t src_data,
 			struct dma_resv *resv,
 			struct dma_fence **f,
-			bool delayed)
+			bool delayed,
+			u64 k_job_id)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
@@ -2234,7 +2413,8 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 	int r;
 
 	if (!adev->mman.buffer_funcs_enabled) {
-		DRM_ERROR("Trying to clear memory with ring turned off.\n");
+		dev_err(adev->dev,
+			"Trying to clear memory with ring turned off.\n");
 		return -EINVAL;
 	}
 
@@ -2254,7 +2434,7 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 			goto error;
 
 		r = amdgpu_ttm_fill_mem(ring, src_data, to, cur_size, resv,
-					&next, true, delayed);
+					&next, true, delayed, k_job_id);
 		if (r)
 			goto error;
 
@@ -2294,7 +2474,7 @@ int amdgpu_ttm_evict_resources(struct amdgpu_device *adev, int mem_type)
 		man = ttm_manager_type(&adev->mman.bdev, mem_type);
 		break;
 	default:
-		DRM_ERROR("Trying to evict invalid memory type\n");
+		dev_err(adev->dev, "Trying to evict invalid memory type\n");
 		return -EINVAL;
 	}