1 files changed, 322 insertions, 270 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index e80fc38141b5..663043c8f0f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -31,45 +31,79 @@
 #include <drm/drm_syncobj.h>
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_gmc.h"
+#include "amdgpu_gem.h"
 
 static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 				      struct drm_amdgpu_cs_chunk_fence *data,
 				      uint32_t *offset)
 {
 	struct drm_gem_object *gobj;
+	struct amdgpu_bo *bo;
 	unsigned long size;
+	int r;
 
 	gobj = drm_gem_object_lookup(p->filp, data->handle);
 	if (gobj == NULL)
 		return -EINVAL;
 
-	p->uf_entry.robj = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+	bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
 	p->uf_entry.priority = 0;
-	p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
+	p->uf_entry.tv.bo = &bo->tbo;
 	p->uf_entry.tv.shared = true;
 	p->uf_entry.user_pages = NULL;
 
-	size = amdgpu_bo_size(p->uf_entry.robj);
-	if (size != PAGE_SIZE || (data->offset + 8) > size)
-		return -EINVAL;
+	drm_gem_object_put_unlocked(gobj);
+
+	size = amdgpu_bo_size(bo);
+	if (size != PAGE_SIZE || (data->offset + 8) > size) {
+		r = -EINVAL;
+		goto error_unref;
+	}
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		r = -EINVAL;
+		goto error_unref;
+	}
 
 	*offset = data->offset;
 
-	drm_gem_object_put_unlocked(gobj);
+	return 0;
 
-	if (amdgpu_ttm_tt_get_usermm(p->uf_entry.robj->tbo.ttm)) {
-		amdgpu_bo_unref(&p->uf_entry.robj);
-		return -EINVAL;
-	}
+error_unref:
+	amdgpu_bo_unref(&bo);
+	return r;
+}
+
+static int amdgpu_cs_bo_handles_chunk(struct amdgpu_cs_parser *p,
+				      struct drm_amdgpu_bo_list_in *data)
+{
+	int r;
+	struct drm_amdgpu_bo_list_entry *info = NULL;
+
+	r = amdgpu_bo_create_list_entry_array(data, &info);
+	if (r)
+		return r;
 
+	r = amdgpu_bo_list_create(p->adev, p->filp, info, data->bo_number,
+				  &p->bo_list);
+	if (r)
+		goto error_free;
+
+	kvfree(info);
 	return 0;
+
+error_free:
+	if (info)
+		kvfree(info);
+
+	return r;
 }
 
-static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
+static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
-	union drm_amdgpu_cs *cs = data;
 	uint64_t *chunk_array_user;
 	uint64_t *chunk_array;
 	unsigned size, num_ibs = 0;
@@ -163,6 +197,19 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 
 			break;
 
+		case AMDGPU_CHUNK_ID_BO_HANDLES:
+			size = sizeof(struct drm_amdgpu_bo_list_in);
+			if (p->chunks[i].length_dw * sizeof(uint32_t) < size) {
+				ret = -EINVAL;
+				goto free_partial_kdata;
+			}
+
+			ret = amdgpu_cs_bo_handles_chunk(p, p->chunks[i].kdata);
+			if (ret)
+				goto free_partial_kdata;
+
+			break;
+
 		case AMDGPU_CHUNK_ID_DEPENDENCIES:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
@@ -183,9 +230,13 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 		goto free_all_kdata;
 	}
 
-	if (p->uf_entry.robj)
+	if (p->uf_entry.tv.bo)
 		p->job->uf_addr = uf_offset;
 	kfree(chunk_array);
+
+	/* Use this opportunity to fill in task info for the vm */
+	amdgpu_vm_set_task_info(vm);
+
 	return 0;
 
 free_all_kdata:
@@ -257,7 +308,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 		return;
 	}
 
-	total_vram = adev->mc.real_vram_size - adev->vram_pin_size;
+	total_vram = adev->gmc.real_vram_size - atomic64_read(&adev->vram_pin_size);
 	used_vram = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
 
@@ -302,8 +353,8 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	*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;
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc)) {
+		u64 total_vis_vram = adev->gmc.visible_vram_size;
 		u64 used_vis_vram =
 			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 
@@ -346,8 +397,8 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	struct ttm_operation_ctx ctx = {
 		.interruptible = true,
 		.no_wait_gpu = false,
-		.allow_reserved_eviction = false,
-		.resv = bo->tbo.resv
+		.resv = bo->tbo.resv,
+		.flags = 0
 	};
 	uint32_t domain;
 	int r;
@@ -359,7 +410,7 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	 * to move it. Don't move anything if the threshold is zero.
 	 */
 	if (p->bytes_moved < p->bytes_moved_threshold) {
-		if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
+		if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 		    (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
@@ -377,13 +428,12 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	}
 
 retry:
-	amdgpu_ttm_placement_from_domain(bo, domain);
+	amdgpu_bo_placement_from_domain(bo, domain);
 	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 
 	p->bytes_moved += ctx.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)
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+	    amdgpu_bo_in_cpu_visible_vram(bo))
 		p->bytes_moved_vis += ctx.bytes_moved;
 
 	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
@@ -409,14 +459,13 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 	     p->evictable = list_prev_entry(p->evictable, tv.head)) {
 
 		struct amdgpu_bo_list_entry *candidate = p->evictable;
-		struct amdgpu_bo *bo = candidate->robj;
+		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(candidate->tv.bo);
 		struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-		u64 initial_bytes_moved, bytes_moved;
 		bool update_bytes_moved_vis;
 		uint32_t other;
 
 		/* If we reached our current BO we can forget it */
-		if (candidate->robj == validated)
+		if (bo == validated)
 			break;
 
 		/* We can't move pinned BOs here */
@@ -435,18 +484,14 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 			continue;
 
 		/* 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);
+				!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+				amdgpu_bo_in_cpu_visible_vram(bo);
+		amdgpu_bo_placement_from_domain(bo, other);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
-		bytes_moved = atomic64_read(&adev->num_bytes_moved) -
-			initial_bytes_moved;
-		p->bytes_moved += bytes_moved;
+		p->bytes_moved += ctx.bytes_moved;
 		if (update_bytes_moved_vis)
-			p->bytes_moved_vis += bytes_moved;
+			p->bytes_moved_vis += ctx.bytes_moved;
 
 		if (unlikely(r))
 			break;
@@ -485,7 +530,7 @@ static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 	int r;
 
 	list_for_each_entry(lobj, validated, tv.head) {
-		struct amdgpu_bo *bo = lobj->robj;
+		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(lobj->tv.bo);
 		bool binding_userptr = false;
 		struct mm_struct *usermm;
 
@@ -496,8 +541,8 @@ static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 		/* Check if we have user pages and nobody bound the BO already */
 		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm) &&
 		    lobj->user_pages) {
-			amdgpu_ttm_placement_from_domain(bo,
-							 AMDGPU_GEM_DOMAIN_CPU);
+			amdgpu_bo_placement_from_domain(bo,
+							AMDGPU_GEM_DOMAIN_CPU);
 			r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 			if (r)
 				return r;
@@ -525,29 +570,46 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 				union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
 	struct amdgpu_bo_list_entry *e;
 	struct list_head duplicates;
-	unsigned i, tries = 10;
+	struct amdgpu_bo *gds;
+	struct amdgpu_bo *gws;
+	struct amdgpu_bo *oa;
+	unsigned tries = 10;
 	int r;
 
 	INIT_LIST_HEAD(&p->validated);
 
-	p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
-	if (p->bo_list) {
-		amdgpu_bo_list_get_list(p->bo_list, &p->validated);
-		if (p->bo_list->first_userptr != p->bo_list->num_entries)
-			p->mn = amdgpu_mn_get(p->adev);
+	/* p->bo_list could already be assigned if AMDGPU_CHUNK_ID_BO_HANDLES is present */
+	if (cs->in.bo_list_handle) {
+		if (p->bo_list)
+			return -EINVAL;
+
+		r = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle,
+				       &p->bo_list);
+		if (r)
+			return r;
+	} else if (!p->bo_list) {
+		/* Create a empty bo_list when no handle is provided */
+		r = amdgpu_bo_list_create(p->adev, p->filp, NULL, 0,
+					  &p->bo_list);
+		if (r)
+			return r;
 	}
 
+	amdgpu_bo_list_get_list(p->bo_list, &p->validated);
+	if (p->bo_list->first_userptr != p->bo_list->num_entries)
+		p->mn = amdgpu_mn_get(p->adev, AMDGPU_MN_TYPE_GFX);
+
 	INIT_LIST_HEAD(&duplicates);
 	amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd);
 
-	if (p->uf_entry.robj)
+	if (p->uf_entry.tv.bo && !ttm_to_amdgpu_bo(p->uf_entry.tv.bo)->parent)
 		list_add(&p->uf_entry.tv.head, &p->validated);
 
 	while (1) {
 		struct list_head need_pages;
-		unsigned i;
 
 		r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
 					   &duplicates);
@@ -557,17 +619,9 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 			goto error_free_pages;
 		}
 
-		/* Without a BO list we don't have userptr BOs */
-		if (!p->bo_list)
-			break;
-
 		INIT_LIST_HEAD(&need_pages);
-		for (i = p->bo_list->first_userptr;
-		     i < p->bo_list->num_entries; ++i) {
-			struct amdgpu_bo *bo;
-
-			e = &p->bo_list->array[i];
-			bo = e->robj;
+		amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+			struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
 
 			if (amdgpu_ttm_tt_userptr_invalidated(bo->tbo.ttm,
 				 &e->user_invalidated) && e->user_pages) {
@@ -586,7 +640,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 				list_del(&e->tv.head);
 				list_add(&e->tv.head, &need_pages);
 
-				amdgpu_bo_unreserve(e->robj);
+				amdgpu_bo_unreserve(bo);
 			}
 		}
 
@@ -605,7 +659,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 		/* Fill the page arrays for all userptrs. */
 		list_for_each_entry(e, &need_pages, tv.head) {
-			struct ttm_tt *ttm = e->robj->tbo.ttm;
+			struct ttm_tt *ttm = e->tv.bo->ttm;
 
 			e->user_pages = kvmalloc_array(ttm->num_pages,
 							 sizeof(struct page*),
@@ -658,35 +712,29 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
 				     p->bytes_moved_vis);
-	if (p->bo_list) {
-		struct amdgpu_bo *gds = p->bo_list->gds_obj;
-		struct amdgpu_bo *gws = p->bo_list->gws_obj;
-		struct amdgpu_bo *oa = p->bo_list->oa_obj;
-		struct amdgpu_vm *vm = &fpriv->vm;
-		unsigned i;
 
-		for (i = 0; i < p->bo_list->num_entries; i++) {
-			struct amdgpu_bo *bo = p->bo_list->array[i].robj;
+	gds = p->bo_list->gds_obj;
+	gws = p->bo_list->gws_obj;
+	oa = p->bo_list->oa_obj;
 
-			p->bo_list->array[i].bo_va = amdgpu_vm_bo_find(vm, bo);
-		}
+	amdgpu_bo_list_for_each_entry(e, p->bo_list)
+		e->bo_va = amdgpu_vm_bo_find(vm, ttm_to_amdgpu_bo(e->tv.bo));
 
-		if (gds) {
-			p->job->gds_base = amdgpu_bo_gpu_offset(gds);
-			p->job->gds_size = amdgpu_bo_size(gds);
-		}
-		if (gws) {
-			p->job->gws_base = amdgpu_bo_gpu_offset(gws);
-			p->job->gws_size = amdgpu_bo_size(gws);
-		}
-		if (oa) {
-			p->job->oa_base = amdgpu_bo_gpu_offset(oa);
-			p->job->oa_size = amdgpu_bo_size(oa);
-		}
+	if (gds) {
+		p->job->gds_base = amdgpu_bo_gpu_offset(gds) >> PAGE_SHIFT;
+		p->job->gds_size = amdgpu_bo_size(gds) >> PAGE_SHIFT;
+	}
+	if (gws) {
+		p->job->gws_base = amdgpu_bo_gpu_offset(gws) >> PAGE_SHIFT;
+		p->job->gws_size = amdgpu_bo_size(gws) >> PAGE_SHIFT;
+	}
+	if (oa) {
+		p->job->oa_base = amdgpu_bo_gpu_offset(oa) >> PAGE_SHIFT;
+		p->job->oa_size = amdgpu_bo_size(oa) >> PAGE_SHIFT;
 	}
 
-	if (!r && p->uf_entry.robj) {
-		struct amdgpu_bo *uf = p->uf_entry.robj;
+	if (!r && p->uf_entry.tv.bo) {
+		struct amdgpu_bo *uf = ttm_to_amdgpu_bo(p->uf_entry.tv.bo);
 
 		r = amdgpu_ttm_alloc_gart(&uf->tbo);
 		p->job->uf_addr += amdgpu_bo_gpu_offset(uf);
@@ -698,18 +746,12 @@ error_validate:
 
 error_free_pages:
 
-	if (p->bo_list) {
-		for (i = p->bo_list->first_userptr;
-		     i < p->bo_list->num_entries; ++i) {
-			e = &p->bo_list->array[i];
-
-			if (!e->user_pages)
-				continue;
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		if (!e->user_pages)
+			continue;
 
-			release_pages(e->user_pages,
-				      e->robj->tbo.ttm->num_pages);
-			kvfree(e->user_pages);
-		}
+		release_pages(e->user_pages, e->tv.bo->ttm->num_pages);
+		kvfree(e->user_pages);
 	}
 
 	return r;
@@ -721,9 +763,11 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 	int r;
 
 	list_for_each_entry(e, &p->validated, tv.head) {
-		struct reservation_object *resv = e->robj->tbo.resv;
+		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
+		struct reservation_object *resv = bo->tbo.resv;
+
 		r = amdgpu_sync_resv(p->adev, &p->job->sync, resv, p->filp,
-				     amdgpu_bo_explicit_sync(e->robj));
+				     amdgpu_bo_explicit_sync(bo));
 
 		if (r)
 			return r;
@@ -766,17 +810,88 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error,
 	kfree(parser->chunks);
 	if (parser->job)
 		amdgpu_job_free(parser->job);
-	amdgpu_bo_unref(&parser->uf_entry.robj);
+	if (parser->uf_entry.tv.bo) {
+		struct amdgpu_bo *uf = ttm_to_amdgpu_bo(parser->uf_entry.tv.bo);
+
+		amdgpu_bo_unref(&uf);
+	}
 }
 
-static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
+static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 {
-	struct amdgpu_device *adev = p->adev;
+	struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched);
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_device *adev = p->adev;
 	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_list_entry *e;
 	struct amdgpu_bo_va *bo_va;
 	struct amdgpu_bo *bo;
-	int i, r;
+	int r;
+
+	/* Only for UVD/VCE VM emulation */
+	if (ring->funcs->parse_cs || ring->funcs->patch_cs_in_place) {
+		unsigned i, j;
+
+		for (i = 0, j = 0; i < p->nchunks && j < p->job->num_ibs; i++) {
+			struct drm_amdgpu_cs_chunk_ib *chunk_ib;
+			struct amdgpu_bo_va_mapping *m;
+			struct amdgpu_bo *aobj = NULL;
+			struct amdgpu_cs_chunk *chunk;
+			uint64_t offset, va_start;
+			struct amdgpu_ib *ib;
+			uint8_t *kptr;
+
+			chunk = &p->chunks[i];
+			ib = &p->job->ibs[j];
+			chunk_ib = chunk->kdata;
+
+			if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
+				continue;
+
+			va_start = chunk_ib->va_start & AMDGPU_GMC_HOLE_MASK;
+			r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
+			if (r) {
+				DRM_ERROR("IB va_start is invalid\n");
+				return r;
+			}
+
+			if ((va_start + chunk_ib->ib_bytes) >
+			    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+				DRM_ERROR("IB va_start+ib_bytes is invalid\n");
+				return -EINVAL;
+			}
+
+			/* the IB should be reserved at this point */
+			r = amdgpu_bo_kmap(aobj, (void **)&kptr);
+			if (r) {
+				return r;
+			}
+
+			offset = m->start * AMDGPU_GPU_PAGE_SIZE;
+			kptr += va_start - offset;
+
+			if (ring->funcs->parse_cs) {
+				memcpy(ib->ptr, kptr, chunk_ib->ib_bytes);
+				amdgpu_bo_kunmap(aobj);
+
+				r = amdgpu_ring_parse_cs(ring, p, j);
+				if (r)
+					return r;
+			} else {
+				ib->ptr = (uint32_t *)kptr;
+				r = amdgpu_ring_patch_cs_in_place(ring, p, j);
+				amdgpu_bo_kunmap(aobj);
+				if (r)
+					return r;
+			}
+
+			j++;
+		}
+	}
+
+	if (!p->job->vm)
+		return amdgpu_cs_sync_rings(p);
+
 
 	r = amdgpu_vm_clear_freed(adev, vm, NULL);
 	if (r)
@@ -806,29 +921,26 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 			return r;
 	}
 
-	if (p->bo_list) {
-		for (i = 0; i < p->bo_list->num_entries; i++) {
-			struct dma_fence *f;
-
-			/* ignore duplicates */
-			bo = p->bo_list->array[i].robj;
-			if (!bo)
-				continue;
+	amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+		struct dma_fence *f;
 
-			bo_va = p->bo_list->array[i].bo_va;
-			if (bo_va == NULL)
-				continue;
+		/* ignore duplicates */
+		bo = ttm_to_amdgpu_bo(e->tv.bo);
+		if (!bo)
+			continue;
 
-			r = amdgpu_vm_bo_update(adev, bo_va, false);
-			if (r)
-				return r;
+		bo_va = e->bo_va;
+		if (bo_va == NULL)
+			continue;
 
-			f = bo_va->last_pt_update;
-			r = amdgpu_sync_fence(adev, &p->job->sync, f, false);
-			if (r)
-				return r;
-		}
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		if (r)
+			return r;
 
+		f = bo_va->last_pt_update;
+		r = amdgpu_sync_fence(adev, &p->job->sync, f, false);
+		if (r)
+			return r;
 	}
 
 	r = amdgpu_vm_handle_moved(adev, vm);
@@ -843,11 +955,18 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	if (amdgpu_vm_debug && p->bo_list) {
+	r = reservation_object_reserve_shared(vm->root.base.bo->tbo.resv);
+	if (r)
+		return r;
+
+	p->job->vm_pd_addr = amdgpu_gmc_pd_addr(vm->root.base.bo);
+
+	if (amdgpu_vm_debug) {
 		/* Invalidate all BOs to test for userspace bugs */
-		for (i = 0; i < p->bo_list->num_entries; i++) {
+		amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+			struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
+
 			/* ignore duplicates */
-			bo = p->bo_list->array[i].robj;
 			if (!bo)
 				continue;
 
@@ -855,78 +974,6 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 		}
 	}
 
-	return r;
-}
-
-static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
-				 struct amdgpu_cs_parser *p)
-{
-	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
-	struct amdgpu_vm *vm = &fpriv->vm;
-	struct amdgpu_ring *ring = p->job->ring;
-	int r;
-
-	/* Only for UVD/VCE VM emulation */
-	if (p->job->ring->funcs->parse_cs) {
-		unsigned i, j;
-
-		for (i = 0, j = 0; i < p->nchunks && j < p->job->num_ibs; i++) {
-			struct drm_amdgpu_cs_chunk_ib *chunk_ib;
-			struct amdgpu_bo_va_mapping *m;
-			struct amdgpu_bo *aobj = NULL;
-			struct amdgpu_cs_chunk *chunk;
-			uint64_t offset, va_start;
-			struct amdgpu_ib *ib;
-			uint8_t *kptr;
-
-			chunk = &p->chunks[i];
-			ib = &p->job->ibs[j];
-			chunk_ib = chunk->kdata;
-
-			if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
-				continue;
-
-			va_start = chunk_ib->va_start & AMDGPU_VA_HOLE_MASK;
-			r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
-			if (r) {
-				DRM_ERROR("IB va_start is invalid\n");
-				return r;
-			}
-
-			if ((va_start + chunk_ib->ib_bytes) >
-			    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
-				DRM_ERROR("IB va_start+ib_bytes is invalid\n");
-				return -EINVAL;
-			}
-
-			/* the IB should be reserved at this point */
-			r = amdgpu_bo_kmap(aobj, (void **)&kptr);
-			if (r) {
-				return r;
-			}
-
-			offset = m->start * AMDGPU_GPU_PAGE_SIZE;
-			kptr += va_start - offset;
-
-			memcpy(ib->ptr, kptr, chunk_ib->ib_bytes);
-			amdgpu_bo_kunmap(aobj);
-
-			r = amdgpu_ring_parse_cs(ring, p, j);
-			if (r)
-				return r;
-
-			j++;
-		}
-	}
-
-	if (p->job->vm) {
-		p->job->vm_pd_addr = amdgpu_bo_gpu_offset(vm->root.base.bo);
-
-		r = amdgpu_bo_vm_update_pte(p);
-		if (r)
-			return r;
-	}
-
 	return amdgpu_cs_sync_rings(p);
 }
 
@@ -935,14 +982,15 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 {
 	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
-	int i, j;
 	int r, ce_preempt = 0, de_preempt = 0;
+	struct amdgpu_ring *ring;
+	int i, j;
 
 	for (i = 0, j = 0; i < parser->nchunks && j < parser->job->num_ibs; i++) {
 		struct amdgpu_cs_chunk *chunk;
 		struct amdgpu_ib *ib;
 		struct drm_amdgpu_cs_chunk_ib *chunk_ib;
-		struct amdgpu_ring *ring;
+		struct drm_sched_entity *entity;
 
 		chunk = &parser->chunks[i];
 		ib = &parser->job->ibs[j];
@@ -964,27 +1012,24 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 				return -EINVAL;
 		}
 
-		r = amdgpu_queue_mgr_map(adev, &parser->ctx->queue_mgr, chunk_ib->ip_type,
-					 chunk_ib->ip_instance, chunk_ib->ring, &ring);
+		r = amdgpu_ctx_get_entity(parser->ctx, chunk_ib->ip_type,
+					  chunk_ib->ip_instance, chunk_ib->ring,
+					  &entity);
 		if (r)
 			return r;
 
-		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE) {
-			parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
-			if (!parser->ctx->preamble_presented) {
-				parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
-				parser->ctx->preamble_presented = true;
-			}
-		}
+		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE)
+			parser->job->preamble_status |=
+				AMDGPU_PREAMBLE_IB_PRESENT;
 
-		if (parser->job->ring && parser->job->ring != ring)
+		if (parser->entity && parser->entity != entity)
 			return -EINVAL;
 
-		parser->job->ring = ring;
+		parser->entity = entity;
 
-		r =  amdgpu_ib_get(adev, vm,
-					ring->funcs->parse_cs ? chunk_ib->ib_bytes : 0,
-					ib);
+		ring = to_amdgpu_ring(entity->rq->sched);
+		r =  amdgpu_ib_get(adev, vm, ring->funcs->parse_cs ?
+				   chunk_ib->ib_bytes : 0, ib);
 		if (r) {
 			DRM_ERROR("Failed to get ib !\n");
 			return r;
@@ -998,12 +1043,13 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 	}
 
 	/* UVD & VCE fw doesn't support user fences */
+	ring = to_amdgpu_ring(parser->entity->rq->sched);
 	if (parser->job->uf_addr && (
-	    parser->job->ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
-	    parser->job->ring->funcs->type == AMDGPU_RING_TYPE_VCE))
+	    ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
+	    ring->funcs->type == AMDGPU_RING_TYPE_VCE))
 		return -EINVAL;
 
-	return amdgpu_ctx_wait_prev_fence(parser->ctx, parser->job->ring->idx);
+	return amdgpu_ctx_wait_prev_fence(parser->ctx, parser->entity);
 }
 
 static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
@@ -1019,24 +1065,23 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
 		sizeof(struct drm_amdgpu_cs_chunk_dep);
 
 	for (i = 0; i < num_deps; ++i) {
-		struct amdgpu_ring *ring;
 		struct amdgpu_ctx *ctx;
+		struct drm_sched_entity *entity;
 		struct dma_fence *fence;
 
 		ctx = amdgpu_ctx_get(fpriv, deps[i].ctx_id);
 		if (ctx == NULL)
 			return -EINVAL;
 
-		r = amdgpu_queue_mgr_map(p->adev, &ctx->queue_mgr,
-					 deps[i].ip_type,
-					 deps[i].ip_instance,
-					 deps[i].ring, &ring);
+		r = amdgpu_ctx_get_entity(ctx, deps[i].ip_type,
+					  deps[i].ip_instance,
+					  deps[i].ring, &entity);
 		if (r) {
 			amdgpu_ctx_put(ctx);
 			return r;
 		}
 
-		fence = amdgpu_ctx_get_fence(ctx, ring,
+		fence = amdgpu_ctx_get_fence(ctx, entity,
 					     deps[i].handle);
 		if (IS_ERR(fence)) {
 			r = PTR_ERR(fence);
@@ -1059,7 +1104,7 @@ static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
 {
 	int r;
 	struct dma_fence *fence;
-	r = drm_syncobj_find_fence(p->filp, handle, &fence);
+	r = drm_syncobj_find_fence(p->filp, handle, 0, &fence);
 	if (r)
 		return r;
 
@@ -1148,71 +1193,80 @@ static void amdgpu_cs_post_dependencies(struct amdgpu_cs_parser *p)
 	int i;
 
 	for (i = 0; i < p->num_post_dep_syncobjs; ++i)
-		drm_syncobj_replace_fence(p->post_dep_syncobjs[i], p->fence);
+		drm_syncobj_replace_fence(p->post_dep_syncobjs[i], 0, p->fence);
 }
 
 static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 			    union drm_amdgpu_cs *cs)
 {
-	struct amdgpu_ring *ring = p->job->ring;
-	struct drm_sched_entity *entity = &p->ctx->rings[ring->idx].entity;
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct drm_sched_entity *entity = p->entity;
+	enum drm_sched_priority priority;
+	struct amdgpu_ring *ring;
+	struct amdgpu_bo_list_entry *e;
 	struct amdgpu_job *job;
-	unsigned i;
 	uint64_t seq;
 
 	int r;
 
-	amdgpu_mn_lock(p->mn);
-	if (p->bo_list) {
-		for (i = p->bo_list->first_userptr;
-		     i < p->bo_list->num_entries; ++i) {
-			struct amdgpu_bo *bo = p->bo_list->array[i].robj;
-
-			if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {
-				amdgpu_mn_unlock(p->mn);
-				return -ERESTARTSYS;
-			}
-		}
-	}
-
 	job = p->job;
 	p->job = NULL;
 
-	r = drm_sched_job_init(&job->base, &ring->sched, entity, p->filp);
-	if (r) {
-		amdgpu_job_free(job);
-		amdgpu_mn_unlock(p->mn);
-		return r;
+	r = drm_sched_job_init(&job->base, entity, p->filp);
+	if (r)
+		goto error_unlock;
+
+	/* No memory allocation is allowed while holding the mn lock */
+	amdgpu_mn_lock(p->mn);
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
+
+		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {
+			r = -ERESTARTSYS;
+			goto error_abort;
+		}
 	}
 
 	job->owner = p->filp;
-	job->fence_ctx = entity->fence_context;
 	p->fence = dma_fence_get(&job->base.s_fence->finished);
 
-	r = amdgpu_ctx_add_fence(p->ctx, ring, p->fence, &seq);
-	if (r) {
-		dma_fence_put(p->fence);
-		dma_fence_put(&job->base.s_fence->finished);
-		amdgpu_job_free(job);
-		amdgpu_mn_unlock(p->mn);
-		return r;
-	}
-
+	amdgpu_ctx_add_fence(p->ctx, entity, p->fence, &seq);
 	amdgpu_cs_post_dependencies(p);
 
+	if ((job->preamble_status & AMDGPU_PREAMBLE_IB_PRESENT) &&
+	    !p->ctx->preamble_presented) {
+		job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
+		p->ctx->preamble_presented = true;
+	}
+
 	cs->out.handle = seq;
 	job->uf_sequence = seq;
 
 	amdgpu_job_free_resources(job);
-	amdgpu_ring_priority_get(job->ring, job->base.s_priority);
 
 	trace_amdgpu_cs_ioctl(job);
+	amdgpu_vm_bo_trace_cs(&fpriv->vm, &p->ticket);
+	priority = job->base.s_priority;
 	drm_sched_entity_push_job(&job->base, entity);
 
+	ring = to_amdgpu_ring(entity->rq->sched);
+	amdgpu_ring_priority_get(ring, priority);
+
+	amdgpu_vm_move_to_lru_tail(p->adev, &fpriv->vm);
+
 	ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);
 	amdgpu_mn_unlock(p->mn);
 
 	return 0;
+
+error_abort:
+	dma_fence_put(&job->base.s_fence->finished);
+	job->base.s_fence = NULL;
+	amdgpu_mn_unlock(p->mn);
+
+error_unlock:
+	amdgpu_job_free(job);
+	return r;
 }
 
 int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
@@ -1239,6 +1293,12 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	if (r)
 		goto out;
 
+	r = amdgpu_cs_dependencies(adev, &parser);
+	if (r) {
+		DRM_ERROR("Failed in the dependencies handling %d!\n", r);
+		goto out;
+	}
+
 	r = amdgpu_cs_parser_bos(&parser, data);
 	if (r) {
 		if (r == -ENOMEM)
@@ -1250,16 +1310,10 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 
 	reserved_buffers = true;
 
-	r = amdgpu_cs_dependencies(adev, &parser);
-	if (r) {
-		DRM_ERROR("Failed in the dependencies handling %d!\n", r);
-		goto out;
-	}
-
 	for (i = 0; i < parser.job->num_ibs; i++)
 		trace_amdgpu_cs(&parser, i);
 
-	r = amdgpu_cs_ib_vm_chunk(adev, &parser);
+	r = amdgpu_cs_vm_handling(&parser);
 	if (r)
 		goto out;
 
@@ -1283,9 +1337,8 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
 			 struct drm_file *filp)
 {
 	union drm_amdgpu_wait_cs *wait = data;
-	struct amdgpu_device *adev = dev->dev_private;
 	unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
-	struct amdgpu_ring *ring = NULL;
+	struct drm_sched_entity *entity;
 	struct amdgpu_ctx *ctx;
 	struct dma_fence *fence;
 	long r;
@@ -1294,15 +1347,14 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
 	if (ctx == NULL)
 		return -EINVAL;
 
-	r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr,
-				 wait->in.ip_type, wait->in.ip_instance,
-				 wait->in.ring, &ring);
+	r = amdgpu_ctx_get_entity(ctx, wait->in.ip_type, wait->in.ip_instance,
+				  wait->in.ring, &entity);
 	if (r) {
 		amdgpu_ctx_put(ctx);
 		return r;
 	}
 
-	fence = amdgpu_ctx_get_fence(ctx, ring, wait->in.handle);
+	fence = amdgpu_ctx_get_fence(ctx, entity, wait->in.handle);
 	if (IS_ERR(fence))
 		r = PTR_ERR(fence);
 	else if (fence) {
@@ -1334,7 +1386,7 @@ static struct dma_fence *amdgpu_cs_get_fence(struct amdgpu_device *adev,
 					     struct drm_file *filp,
 					     struct drm_amdgpu_fence *user)
 {
-	struct amdgpu_ring *ring;
+	struct drm_sched_entity *entity;
 	struct amdgpu_ctx *ctx;
 	struct dma_fence *fence;
 	int r;
@@ -1343,14 +1395,14 @@ static struct dma_fence *amdgpu_cs_get_fence(struct amdgpu_device *adev,
 	if (ctx == NULL)
 		return ERR_PTR(-EINVAL);
 
-	r = amdgpu_queue_mgr_map(adev, &ctx->queue_mgr, user->ip_type,
-				 user->ip_instance, user->ring, &ring);
+	r = amdgpu_ctx_get_entity(ctx, user->ip_type, user->ip_instance,
+				  user->ring, &entity);
 	if (r) {
 		amdgpu_ctx_put(ctx);
 		return ERR_PTR(r);
 	}
 
-	fence = amdgpu_ctx_get_fence(ctx, ring, user->seq_no);
+	fence = amdgpu_ctx_get_fence(ctx, entity, user->seq_no);
 	amdgpu_ctx_put(ctx);
 
 	return fence;
@@ -1599,7 +1651,7 @@ int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 
 	if (!((*bo)->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)) {
 		(*bo)->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-		amdgpu_ttm_placement_from_domain(*bo, (*bo)->allowed_domains);
+		amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
 		r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
 		if (r)
 			return r;