Merge tag 'drm-fixes-for-v4.14-rc1' of git://people.freedesktop.org/~airlied/linux

Pull drm AMD fixes from Dave Airlie:
 "Just had a single AMD fixes pull from Alex for rc1"

* tag 'drm-fixes-for-v4.14-rc1' of git://people.freedesktop.org/~airlied/linux:
  drm/amdgpu: revert "fix deadlock of reservation between cs and gpu reset v2"
  drm/amdgpu: remove duplicate return statement
  drm/amdgpu: check memory allocation failure
  drm/amd/amdgpu: fix BANK_SELECT on Vega10 (v2)
  drm/amdgpu: inline amdgpu_ttm_do_bind again
  drm/amdgpu: fix amdgpu_ttm_bind
  drm/amdgpu: remove the GART copy hack
  drm/ttm:fix wrong decoding of bo_count
  drm/ttm: fix missing inc bo_count
  drm/amdgpu: set sched_hw_submission higher for KIQ (v3)
  drm/amdgpu: move default gart size setting into gmc modules
  drm/amdgpu: refine default gart size
  drm/amd/powerplay: ACG frequency added in PPTable
  drm/amdgpu: discard commands of killed processes
  drm/amdgpu: fix and cleanup shadow handling
  drm/amdgpu: add automatic per asic settings for gart_size
  drm/amdgpu/gfx8: fix spelling typo in mqd allocation
  drm/amd/powerplay: unhalt mec after loading
  drm/amdgpu/virtual_dce: Virtual display doesn't support disable vblank immediately
  drm/amdgpu: Fix huge page updates with CPU

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -76,7 +76,7 @@
 extern int amdgpu_modeset;
 extern int amdgpu_vram_limit;
 extern int amdgpu_vis_vram_limit;
-extern unsigned amdgpu_gart_size;
+extern int amdgpu_gart_size;
 extern int amdgpu_gtt_size;
 extern int amdgpu_moverate;
 extern int amdgpu_benchmarking;

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c

@@ -155,7 +155,6 @@ static const struct kfd2kgd_calls kfd2kgd = {
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
 {
 	return (struct kfd2kgd_calls *)&kfd2kgd;
-	return (struct kfd2kgd_calls *)&kfd2kgd;
 }
 
 static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -1079,6 +1079,9 @@ static int amdgpu_cs_process_syncobj_out_dep(struct amdgpu_cs_parser *p,
 					     GFP_KERNEL);
 	p->num_post_dep_syncobjs = 0;
 
+	if (!p->post_dep_syncobjs)
+		return -ENOMEM;
+
 	for (i = 0; i < num_deps; ++i) {
 		p->post_dep_syncobjs[i] = drm_syncobj_find(p->filp, deps[i].handle);
 		if (!p->post_dep_syncobjs[i])
@@ -1150,7 +1153,6 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	cs->out.handle = amdgpu_ctx_add_fence(p->ctx, ring, p->fence);
 	job->uf_sequence = cs->out.handle;
 	amdgpu_job_free_resources(job);
-	amdgpu_cs_parser_fini(p, 0, true);
 
 	trace_amdgpu_cs_ioctl(job);
 	amd_sched_entity_push_job(&job->base);
@@ -1208,10 +1210,7 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		goto out;
 
 	r = amdgpu_cs_submit(&parser, cs);
-	if (r)
-		goto out;
 
-	return 0;
 out:
 	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
 	return r;

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -1062,11 +1062,11 @@ static void amdgpu_check_arguments(struct amdgpu_device *adev)
 		amdgpu_sched_jobs = roundup_pow_of_two(amdgpu_sched_jobs);
 	}
 
-	if (amdgpu_gart_size < 32) {
+	if (amdgpu_gart_size != -1 && amdgpu_gart_size < 32) {
 		/* gart size must be greater or equal to 32M */
 		dev_warn(adev->dev, "gart size (%d) too small\n",
 			 amdgpu_gart_size);
-		amdgpu_gart_size = 32;
+		amdgpu_gart_size = -1;
 	}
 
 	if (amdgpu_gtt_size != -1 && amdgpu_gtt_size < 32) {
@@ -2622,12 +2622,6 @@ static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
 			goto err;
 		}
 
-		r = amdgpu_ttm_bind(&bo->shadow->tbo, &bo->shadow->tbo.mem);
-		if (r) {
-			DRM_ERROR("%p bind failed\n", bo->shadow);
-			goto err;
-		}
-
 		r = amdgpu_bo_restore_from_shadow(adev, ring, bo,
 						 NULL, fence, true);
 		if (r) {

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -76,7 +76,7 @@
 
 int amdgpu_vram_limit = 0;
 int amdgpu_vis_vram_limit = 0;
-unsigned amdgpu_gart_size = 256;
+int amdgpu_gart_size = -1; /* auto */
 int amdgpu_gtt_size = -1; /* auto */
 int amdgpu_moverate = -1; /* auto */
 int amdgpu_benchmarking = 0;
@@ -128,7 +128,7 @@ module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
 MODULE_PARM_DESC(vis_vramlimit, "Restrict visible VRAM for testing, in megabytes");
 module_param_named(vis_vramlimit, amdgpu_vis_vram_limit, int, 0444);
 
-MODULE_PARM_DESC(gartsize, "Size of PCIE/IGP gart to setup in megabytes (32, 64, etc.)");
+MODULE_PARM_DESC(gartsize, "Size of GART to setup in megabytes (32, 64, etc., -1=auto)");
 module_param_named(gartsize, amdgpu_gart_size, uint, 0600);
 
 MODULE_PARM_DESC(gttsize, "Size of the GTT domain in megabytes (-1 = auto)");

drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c

@@ -56,18 +56,6 @@
  * Common GART table functions.
  */
 
-/**
- * amdgpu_gart_set_defaults - set the default gart_size
- *
- * @adev: amdgpu_device pointer
- *
- * Set the default gart_size based on parameters and available VRAM.
- */
-void amdgpu_gart_set_defaults(struct amdgpu_device *adev)
-{
-	adev->mc.gart_size = (uint64_t)amdgpu_gart_size << 20;
-}
-
 /**
  * amdgpu_gart_table_ram_alloc - allocate system ram for gart page table
  *

drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h

@@ -56,7 +56,6 @@ struct amdgpu_gart {
 	const struct amdgpu_gart_funcs *gart_funcs;
 };
 
-void amdgpu_gart_set_defaults(struct amdgpu_device *adev);
 int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev);
 void amdgpu_gart_table_ram_free(struct amdgpu_device *adev);
 int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev);

drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c

@@ -108,10 +108,10 @@ bool amdgpu_gtt_mgr_is_allocated(struct ttm_mem_reg *mem)
  *
  * Allocate the address space for a node.
  */
-int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
-			 struct ttm_buffer_object *tbo,
-			 const struct ttm_place *place,
-			 struct ttm_mem_reg *mem)
+static int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
+				struct ttm_buffer_object *tbo,
+				const struct ttm_place *place,
+				struct ttm_mem_reg *mem)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
 	struct amdgpu_gtt_mgr *mgr = man->priv;
@@ -143,12 +143,8 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
 					fpfn, lpfn, mode);
 	spin_unlock(&mgr->lock);
 
-	if (!r) {
+	if (!r)
 		mem->start = node->start;
-		if (&tbo->mem == mem)
-			tbo->offset = (tbo->mem.start << PAGE_SHIFT) +
-			    tbo->bdev->man[tbo->mem.mem_type].gpu_offset;
-	}
 
 	return r;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c

@@ -221,8 +221,9 @@ int amdgpu_irq_init(struct amdgpu_device *adev)
 
 	spin_lock_init(&adev->irq.lock);
 
-	/* Disable vblank irqs aggressively for power-saving */
-	adev->ddev->vblank_disable_immediate = true;
+	if (!adev->enable_virtual_display)
+		/* Disable vblank irqs aggressively for power-saving */
+		adev->ddev->vblank_disable_immediate = true;
 
 	r = drm_vblank_init(adev->ddev, adev->mode_info.num_crtc);
 	if (r) {

drivers/gpu/drm/amd/amdgpu/amdgpu_object.c

@@ -91,7 +91,10 @@ static void amdgpu_ttm_placement_init(struct amdgpu_device *adev,
 
 	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
 		places[c].fpfn = 0;
-		places[c].lpfn = 0;
+		if (flags & AMDGPU_GEM_CREATE_SHADOW)
+			places[c].lpfn = adev->mc.gart_size >> PAGE_SHIFT;
+		else
+			places[c].lpfn = 0;
 		places[c].flags = TTM_PL_FLAG_TT;
 		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC)
 			places[c].flags |= TTM_PL_FLAG_WC |
@@ -446,17 +449,16 @@ static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
 	if (bo->shadow)
 		return 0;
 
-	bo->flags |= AMDGPU_GEM_CREATE_SHADOW;
-	memset(&placements, 0,
-	       (AMDGPU_GEM_DOMAIN_MAX + 1) * sizeof(struct ttm_place));
-
-	amdgpu_ttm_placement_init(adev, &placement,
-				  placements, AMDGPU_GEM_DOMAIN_GTT,
-				  AMDGPU_GEM_CREATE_CPU_GTT_USWC);
+	memset(&placements, 0, sizeof(placements));
+	amdgpu_ttm_placement_init(adev, &placement, placements,
+				  AMDGPU_GEM_DOMAIN_GTT,
+				  AMDGPU_GEM_CREATE_CPU_GTT_USWC |
+				  AMDGPU_GEM_CREATE_SHADOW);
 
 	r = amdgpu_bo_create_restricted(adev, size, byte_align, true,
 					AMDGPU_GEM_DOMAIN_GTT,
-					AMDGPU_GEM_CREATE_CPU_GTT_USWC,
+					AMDGPU_GEM_CREATE_CPU_GTT_USWC |
+					AMDGPU_GEM_CREATE_SHADOW,
 					NULL, &placement,
 					bo->tbo.resv,
 					0,
@@ -484,30 +486,28 @@ int amdgpu_bo_create(struct amdgpu_device *adev,
 {
 	struct ttm_placement placement = {0};
 	struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
+	uint64_t parent_flags = flags & ~AMDGPU_GEM_CREATE_SHADOW;
 	int r;
 
-	memset(&placements, 0,
-	       (AMDGPU_GEM_DOMAIN_MAX + 1) * sizeof(struct ttm_place));
+	memset(&placements, 0, sizeof(placements));
+	amdgpu_ttm_placement_init(adev, &placement, placements,
+				  domain, parent_flags);
 
-	amdgpu_ttm_placement_init(adev, &placement,
-				  placements, domain, flags);
-
-	r = amdgpu_bo_create_restricted(adev, size, byte_align, kernel,
-					domain, flags, sg, &placement,
-					resv, init_value, bo_ptr);
+	r = amdgpu_bo_create_restricted(adev, size, byte_align, kernel, domain,
+					parent_flags, sg, &placement, resv,
+					init_value, bo_ptr);
 	if (r)
 		return r;
 
-	if (amdgpu_need_backup(adev) && (flags & AMDGPU_GEM_CREATE_SHADOW)) {
-		if (!resv) {
-			r = ww_mutex_lock(&(*bo_ptr)->tbo.resv->lock, NULL);
-			WARN_ON(r != 0);
-		}
+	if ((flags & AMDGPU_GEM_CREATE_SHADOW) && amdgpu_need_backup(adev)) {
+		if (!resv)
+			WARN_ON(reservation_object_lock((*bo_ptr)->tbo.resv,
+							NULL));
 
 		r = amdgpu_bo_create_shadow(adev, size, byte_align, (*bo_ptr));
 
 		if (!resv)
-			ww_mutex_unlock(&(*bo_ptr)->tbo.resv->lock);
+			reservation_object_unlock((*bo_ptr)->tbo.resv);
 
 		if (r)
 			amdgpu_bo_unref(bo_ptr);

drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c

@@ -170,6 +170,16 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		     unsigned irq_type)
 {
 	int r;
+	int sched_hw_submission = amdgpu_sched_hw_submission;
+
+	/* Set the hw submission limit higher for KIQ because
+	 * it's used for a number of gfx/compute tasks by both
+	 * KFD and KGD which may have outstanding fences and
+	 * it doesn't really use the gpu scheduler anyway;
+	 * KIQ tasks get submitted directly to the ring.
+	 */
+	if (ring->funcs->type == AMDGPU_RING_TYPE_KIQ)
+		sched_hw_submission = max(sched_hw_submission, 256);
 
 	if (ring->adev == NULL) {
 		if (adev->num_rings >= AMDGPU_MAX_RINGS)
@@ -178,8 +188,7 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		ring->adev = adev;
 		ring->idx = adev->num_rings++;
 		adev->rings[ring->idx] = ring;
-		r = amdgpu_fence_driver_init_ring(ring,
-			amdgpu_sched_hw_submission);
+		r = amdgpu_fence_driver_init_ring(ring, sched_hw_submission);
 		if (r)
 			return r;
 	}
@@ -218,8 +227,7 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		return r;
 	}
 
-	ring->ring_size = roundup_pow_of_two(max_dw * 4 *
-					     amdgpu_sched_hw_submission);
+	ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
 
 	ring->buf_mask = (ring->ring_size / 4) - 1;
 	ring->ptr_mask = ring->funcs->support_64bit_ptrs ?

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c

@@ -761,35 +761,11 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
 	sg_free_table(ttm->sg);
 }
 
-static int amdgpu_ttm_do_bind(struct ttm_tt *ttm, struct ttm_mem_reg *mem)
-{
-	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	uint64_t flags;
-	int r;
-
-	spin_lock(&gtt->adev->gtt_list_lock);
-	flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, mem);
-	gtt->offset = (u64)mem->start << PAGE_SHIFT;
-	r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
-		ttm->pages, gtt->ttm.dma_address, flags);
-
-	if (r) {
-		DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
-			  ttm->num_pages, gtt->offset);
-		goto error_gart_bind;
-	}
-
-	list_add_tail(&gtt->list, &gtt->adev->gtt_list);
-error_gart_bind:
-	spin_unlock(&gtt->adev->gtt_list_lock);
-	return r;
-
-}
-
 static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 				   struct ttm_mem_reg *bo_mem)
 {
 	struct amdgpu_ttm_tt *gtt = (void*)ttm;
+	uint64_t flags;
 	int r = 0;
 
 	if (gtt->userptr) {
@@ -809,9 +785,24 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 	    bo_mem->mem_type == AMDGPU_PL_OA)
 		return -EINVAL;
 
-	if (amdgpu_gtt_mgr_is_allocated(bo_mem))
-	    r = amdgpu_ttm_do_bind(ttm, bo_mem);
+	if (!amdgpu_gtt_mgr_is_allocated(bo_mem))
+		return 0;
 
+	spin_lock(&gtt->adev->gtt_list_lock);
+	flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
+	gtt->offset = (u64)bo_mem->start << PAGE_SHIFT;
+	r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
+		ttm->pages, gtt->ttm.dma_address, flags);
+
+	if (r) {
+		DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
+			  ttm->num_pages, gtt->offset);
+		goto error_gart_bind;
+	}
+
+	list_add_tail(&gtt->list, &gtt->adev->gtt_list);
+error_gart_bind:
+	spin_unlock(&gtt->adev->gtt_list_lock);
 	return r;
 }
 
@@ -824,20 +815,39 @@ bool amdgpu_ttm_is_bound(struct ttm_tt *ttm)
 
 int amdgpu_ttm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *bo_mem)
 {
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
 	struct ttm_tt *ttm = bo->ttm;
+	struct ttm_mem_reg tmp;
+
+	struct ttm_placement placement;
+	struct ttm_place placements;
 	int r;
 
 	if (!ttm || amdgpu_ttm_is_bound(ttm))
 		return 0;
 
-	r = amdgpu_gtt_mgr_alloc(&bo->bdev->man[TTM_PL_TT], bo,
-				 NULL, bo_mem);
-	if (r) {
-		DRM_ERROR("Failed to allocate GTT address space (%d)\n", r);
+	tmp = bo->mem;
+	tmp.mm_node = NULL;
+	placement.num_placement = 1;
+	placement.placement = &placements;
+	placement.num_busy_placement = 1;
+	placement.busy_placement = &placements;
+	placements.fpfn = 0;
+	placements.lpfn = adev->mc.gart_size >> PAGE_SHIFT;
+	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
+
+	r = ttm_bo_mem_space(bo, &placement, &tmp, true, false);
+	if (unlikely(r))
 		return r;
-	}
 
-	return amdgpu_ttm_do_bind(ttm, bo_mem);
+	r = ttm_bo_move_ttm(bo, true, false, &tmp);
+	if (unlikely(r))
+		ttm_bo_mem_put(bo, &tmp);
+	else
+		bo->offset = (bo->mem.start << PAGE_SHIFT) +
+			bo->bdev->man[bo->mem.mem_type].gpu_offset;
+
+	return r;
 }
 
 int amdgpu_ttm_recover_gart(struct amdgpu_device *adev)

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h

@@ -62,10 +62,6 @@ extern const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func;
 extern const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func;
 
 bool amdgpu_gtt_mgr_is_allocated(struct ttm_mem_reg *mem);
-int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
-			 struct ttm_buffer_object *tbo,
-			 const struct ttm_place *place,
-			 struct ttm_mem_reg *mem);
 uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man);
 
 uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man);

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c

@@ -165,14 +165,6 @@ static int amdgpu_vm_validate_level(struct amdgpu_vm_pt *parent,
 	unsigned i;
 	int r;
 
-	if (parent->bo->shadow) {
-		struct amdgpu_bo *shadow = parent->bo->shadow;
-
-		r = amdgpu_ttm_bind(&shadow->tbo, &shadow->tbo.mem);
-		if (r)
-			return r;
-	}
-
 	if (use_cpu_for_update) {
 		r = amdgpu_bo_kmap(parent->bo, NULL);
 		if (r)
@@ -1277,7 +1269,7 @@ static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
 	/* In the case of a mixed PT the PDE must point to it*/
 	if (p->adev->asic_type < CHIP_VEGA10 ||
 	    nptes != AMDGPU_VM_PTE_COUNT(p->adev) ||
-	    p->func == amdgpu_vm_do_copy_ptes ||
+	    p->src ||
 	    !(flags & AMDGPU_PTE_VALID)) {
 
 		dst = amdgpu_bo_gpu_offset(entry->bo);
@@ -1294,9 +1286,23 @@ static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
 	entry->addr = (dst | flags);
 
 	if (use_cpu_update) {
+		/* In case a huge page is replaced with a system
+		 * memory mapping, p->pages_addr != NULL and
+		 * amdgpu_vm_cpu_set_ptes would try to translate dst
+		 * through amdgpu_vm_map_gart. But dst is already a
+		 * GPU address (of the page table). Disable
+		 * amdgpu_vm_map_gart temporarily.
+		 */
+		dma_addr_t *tmp;
+
+		tmp = p->pages_addr;
+		p->pages_addr = NULL;
+
 		pd_addr = (unsigned long)amdgpu_bo_kptr(parent->bo);
 		pde = pd_addr + (entry - parent->entries) * 8;
 		amdgpu_vm_cpu_set_ptes(p, pde, dst, 1, 0, flags);
+
+		p->pages_addr = tmp;
 	} else {
 		if (parent->bo->shadow) {
 			pd_addr = amdgpu_bo_gpu_offset(parent->bo->shadow);
@@ -1610,7 +1616,6 @@ error_free:
  *
  * @adev: amdgpu_device pointer
  * @exclusive: fence we need to sync to
- * @gtt_flags: flags as they are used for GTT
  * @pages_addr: DMA addresses to use for mapping
  * @vm: requested vm
  * @mapping: mapped range and flags to use for the update
@@ -1624,7 +1629,6 @@ error_free:
  */
 static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
 				      struct dma_fence *exclusive,
-				      uint64_t gtt_flags,
 				      dma_addr_t *pages_addr,
 				      struct amdgpu_vm *vm,
 				      struct amdgpu_bo_va_mapping *mapping,
@@ -1679,11 +1683,7 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
 		}
 
 		if (pages_addr) {
-			if (flags == gtt_flags)
-				src = adev->gart.table_addr +
-					(addr >> AMDGPU_GPU_PAGE_SHIFT) * 8;
-			else
-				max_entries = min(max_entries, 16ull * 1024ull);
+			max_entries = min(max_entries, 16ull * 1024ull);
 			addr = 0;
 		} else if (flags & AMDGPU_PTE_VALID) {
 			addr += adev->vm_manager.vram_base_offset;
@@ -1728,10 +1728,10 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	struct amdgpu_vm *vm = bo_va->base.vm;
 	struct amdgpu_bo_va_mapping *mapping;
 	dma_addr_t *pages_addr = NULL;
-	uint64_t gtt_flags, flags;
 	struct ttm_mem_reg *mem;
 	struct drm_mm_node *nodes;
 	struct dma_fence *exclusive;
+	uint64_t flags;
 	int r;
 
 	if (clear || !bo_va->base.bo) {
@@ -1751,15 +1751,10 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 		exclusive = reservation_object_get_excl(bo->tbo.resv);
 	}
 
-	if (bo) {
+	if (bo)
 		flags = amdgpu_ttm_tt_pte_flags(adev, bo->tbo.ttm, mem);
-		gtt_flags = (amdgpu_ttm_is_bound(bo->tbo.ttm) &&
-			adev == amdgpu_ttm_adev(bo->tbo.bdev)) ?
-			flags : 0;
-	} else {
+	else
 		flags = 0x0;
-		gtt_flags = ~0x0;
-	}
 
 	spin_lock(&vm->status_lock);
 	if (!list_empty(&bo_va->base.vm_status))
@@ -1767,8 +1762,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	spin_unlock(&vm->status_lock);
 
 	list_for_each_entry(mapping, &bo_va->invalids, list) {
-		r = amdgpu_vm_bo_split_mapping(adev, exclusive,
-					       gtt_flags, pages_addr, vm,
+		r = amdgpu_vm_bo_split_mapping(adev, exclusive, pages_addr, vm,
 					       mapping, flags, nodes,
 					       &bo_va->last_pt_update);
 		if (r)

drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c

@@ -4579,9 +4579,9 @@ static int gfx_v8_0_mqd_init(struct amdgpu_ring *ring)
 	mqd->compute_misc_reserved = 0x00000003;
 	if (!(adev->flags & AMD_IS_APU)) {
 		mqd->dynamic_cu_mask_addr_lo = lower_32_bits(ring->mqd_gpu_addr
-					     + offsetof(struct vi_mqd_allocation, dyamic_cu_mask));
+					     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
 		mqd->dynamic_cu_mask_addr_hi = upper_32_bits(ring->mqd_gpu_addr
-					     + offsetof(struct vi_mqd_allocation, dyamic_cu_mask));
+					     + offsetof(struct vi_mqd_allocation, dynamic_cu_mask));
 	}
 	eop_base_addr = ring->eop_gpu_addr >> 8;
 	mqd->cp_hqd_eop_base_addr_lo = eop_base_addr;
@@ -4768,8 +4768,8 @@ static int gfx_v8_0_kiq_init_queue(struct amdgpu_ring *ring)
 		mutex_unlock(&adev->srbm_mutex);
 	} else {
 		memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation));
-		((struct vi_mqd_allocation *)mqd)->dyamic_cu_mask = 0xFFFFFFFF;
-		((struct vi_mqd_allocation *)mqd)->dyamic_rb_mask = 0xFFFFFFFF;
+		((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
 		mutex_lock(&adev->srbm_mutex);
 		vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
 		gfx_v8_0_mqd_init(ring);
@@ -4792,8 +4792,8 @@ static int gfx_v8_0_kcq_init_queue(struct amdgpu_ring *ring)
 
 	if (!adev->gfx.in_reset && !adev->gfx.in_suspend) {
 		memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation));
-		((struct vi_mqd_allocation *)mqd)->dyamic_cu_mask = 0xFFFFFFFF;
-		((struct vi_mqd_allocation *)mqd)->dyamic_rb_mask = 0xFFFFFFFF;
+		((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
+		((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
 		mutex_lock(&adev->srbm_mutex);
 		vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0);
 		gfx_v8_0_mqd_init(ring);

drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c

@@ -124,7 +124,7 @@ static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
 
 static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 {
-	uint32_t tmp, field;
+	uint32_t tmp;
 
 	/* Setup L2 cache */
 	tmp = RREG32_SOC15(GC, 0, mmVM_L2_CNTL);
@@ -143,9 +143,8 @@ static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
 	WREG32_SOC15(GC, 0, mmVM_L2_CNTL2, tmp);
 
-	field = adev->vm_manager.fragment_size;
 	tmp = mmVM_L2_CNTL3_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
 	WREG32_SOC15(GC, 0, mmVM_L2_CNTL3, tmp);
 

drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c

@@ -332,7 +332,24 @@ static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
 	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
 	adev->mc.visible_vram_size = adev->mc.aper_size;
 
-	amdgpu_gart_set_defaults(adev);
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_HAINAN:    /* no MM engines */
+		default:
+			adev->mc.gart_size = 256ULL << 20;
+			break;
+		case CHIP_VERDE:    /* UVD, VCE do not support GPUVM */
+		case CHIP_TAHITI:   /* UVD, VCE do not support GPUVM */
+		case CHIP_PITCAIRN: /* UVD, VCE do not support GPUVM */
+		case CHIP_OLAND:    /* UVD, VCE do not support GPUVM */
+			adev->mc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
 	gmc_v6_0_vram_gtt_location(adev, &adev->mc);
 
 	return 0;

drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c

@@ -386,7 +386,27 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
 	if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
 		adev->mc.visible_vram_size = adev->mc.real_vram_size;
 
-	amdgpu_gart_set_defaults(adev);
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_TOPAZ:     /* no MM engines */
+		default:
+			adev->mc.gart_size = 256ULL << 20;
+			break;
+#ifdef CONFIG_DRM_AMDGPU_CIK
+		case CHIP_BONAIRE: /* UVD, VCE do not support GPUVM */
+		case CHIP_HAWAII:  /* UVD, VCE do not support GPUVM */
+		case CHIP_KAVERI:  /* UVD, VCE do not support GPUVM */
+		case CHIP_KABINI:  /* UVD, VCE do not support GPUVM */
+		case CHIP_MULLINS: /* UVD, VCE do not support GPUVM */
+			adev->mc.gart_size = 1024ULL << 20;
+			break;
+#endif
+		}
+	} else {
+		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
 	gmc_v7_0_vram_gtt_location(adev, &adev->mc);
 
 	return 0;

drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c

@@ -562,7 +562,26 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
 	if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
 		adev->mc.visible_vram_size = adev->mc.real_vram_size;
 
-	amdgpu_gart_set_defaults(adev);
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_POLARIS11: /* all engines support GPUVM */
+		case CHIP_POLARIS10: /* all engines support GPUVM */
+		case CHIP_POLARIS12: /* all engines support GPUVM */
+		default:
+			adev->mc.gart_size = 256ULL << 20;
+			break;
+		case CHIP_TONGA:   /* UVD, VCE do not support GPUVM */
+		case CHIP_FIJI:    /* UVD, VCE do not support GPUVM */
+		case CHIP_CARRIZO: /* UVD, VCE do not support GPUVM, DCE SG support */
+		case CHIP_STONEY:  /* UVD does not support GPUVM, DCE SG support */
+			adev->mc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
 	gmc_v8_0_vram_gtt_location(adev, &adev->mc);
 
 	return 0;

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c

@@ -499,7 +499,21 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev)
 	if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
 		adev->mc.visible_vram_size = adev->mc.real_vram_size;
 
-	amdgpu_gart_set_defaults(adev);
+	/* set the gart size */
+	if (amdgpu_gart_size == -1) {
+		switch (adev->asic_type) {
+		case CHIP_VEGA10:  /* all engines support GPUVM */
+		default:
+			adev->mc.gart_size = 256ULL << 20;
+			break;
+		case CHIP_RAVEN:   /* DCE SG support */
+			adev->mc.gart_size = 1024ULL << 20;
+			break;
+		}
+	} else {
+		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
+	}
+
 	gmc_v9_0_vram_gtt_location(adev, &adev->mc);
 
 	return 0;

drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c

@@ -138,7 +138,7 @@ static void mmhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
 
 static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 {
-	uint32_t tmp, field;
+	uint32_t tmp;
 
 	/* Setup L2 cache */
 	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
@@ -157,9 +157,8 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL2, tmp);
 
-	field = adev->vm_manager.fragment_size;
 	tmp = mmVM_L2_CNTL3_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 9);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, tmp);
 

drivers/gpu/drm/amd/include/vi_structs.h

@@ -419,8 +419,8 @@ struct vi_mqd_allocation {
 	struct vi_mqd mqd;
 	uint32_t wptr_poll_mem;
 	uint32_t rptr_report_mem;
-	uint32_t dyamic_cu_mask;
-	uint32_t dyamic_rb_mask;
+	uint32_t dynamic_cu_mask;
+	uint32_t dynamic_rb_mask;
 };
 
 struct cz_mqd {

drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c

@@ -1558,7 +1558,8 @@ static int vega10_populate_smc_link_levels(struct pp_hwmgr *hwmgr)
 */
 
 static int vega10_populate_single_gfx_level(struct pp_hwmgr *hwmgr,
-		uint32_t gfx_clock, PllSetting_t *current_gfxclk_level)
+		uint32_t gfx_clock, PllSetting_t *current_gfxclk_level,
+		uint32_t *acg_freq)
 {
 	struct phm_ppt_v2_information *table_info =
 			(struct phm_ppt_v2_information *)(hwmgr->pptable);
@@ -1609,6 +1610,8 @@ static int vega10_populate_single_gfx_level(struct pp_hwmgr *hwmgr,
 			cpu_to_le16(dividers.usPll_ss_slew_frac);
 	current_gfxclk_level->Did = (uint8_t)(dividers.ulDid);
 
+	*acg_freq = gfx_clock / 100; /* 100 Khz to Mhz conversion */
+
 	return 0;
 }
 
@@ -1689,7 +1692,8 @@ static int vega10_populate_all_graphic_levels(struct pp_hwmgr *hwmgr)
 	for (i = 0; i < dpm_table->count; i++) {
 		result = vega10_populate_single_gfx_level(hwmgr,
 				dpm_table->dpm_levels[i].value,
-				&(pp_table->GfxclkLevel[i]));
+				&(pp_table->GfxclkLevel[i]),
+				&(pp_table->AcgFreqTable[i]));
 		if (result)
 			return result;
 	}
@@ -1698,7 +1702,8 @@ static int vega10_populate_all_graphic_levels(struct pp_hwmgr *hwmgr)
 	while (i < NUM_GFXCLK_DPM_LEVELS) {
 		result = vega10_populate_single_gfx_level(hwmgr,
 				dpm_table->dpm_levels[j].value,
-				&(pp_table->GfxclkLevel[i]));
+				&(pp_table->GfxclkLevel[i]),
+				&(pp_table->AcgFreqTable[i]));
 		if (result)
 			return result;
 		i++;

drivers/gpu/drm/amd/powerplay/inc/smu9_driver_if.h

@@ -315,10 +315,12 @@ typedef struct {
   uint8_t      AcgEnable[NUM_GFXCLK_DPM_LEVELS];
   GbVdroopTable_t AcgBtcGbVdroopTable;
   QuadraticInt_t  AcgAvfsGb;
-  uint32_t     Reserved[4];
+
+  /* ACG Frequency Table, in Mhz */
+  uint32_t     AcgFreqTable[NUM_GFXCLK_DPM_LEVELS];
 
   /* Padding - ignore */
-  uint32_t     MmHubPadding[7]; /* SMU internal use */
+  uint32_t     MmHubPadding[3]; /* SMU internal use */
 
 } PPTable_t;
 

drivers/gpu/drm/amd/powerplay/smumgr/smu7_smumgr.c

@@ -380,7 +380,8 @@ static int smu7_populate_single_firmware_entry(struct pp_smumgr *smumgr,
 		entry->num_register_entries = 0;
 	}
 
-	if (fw_type == UCODE_ID_RLC_G)
+	if ((fw_type == UCODE_ID_RLC_G)
+		|| (fw_type == UCODE_ID_CP_MEC))
 		entry->flags = 1;
 	else
 		entry->flags = 0;

drivers/gpu/drm/amd/scheduler/gpu_scheduler.c

@@ -205,17 +205,32 @@ void amd_sched_entity_fini(struct amd_gpu_scheduler *sched,
 			   struct amd_sched_entity *entity)
 {
 	struct amd_sched_rq *rq = entity->rq;
+	int r;
 
 	if (!amd_sched_entity_is_initialized(sched, entity))
 		return;
-
 	/**
 	 * The client will not queue more IBs during this fini, consume existing
-	 * queued IBs
+	 * queued IBs or discard them on SIGKILL
 	*/
-	wait_event(sched->job_scheduled, amd_sched_entity_is_idle(entity));
-
+	if ((current->flags & PF_SIGNALED) && current->exit_code == SIGKILL)
+		r = -ERESTARTSYS;
+	else
+		r = wait_event_killable(sched->job_scheduled,
+					amd_sched_entity_is_idle(entity));
 	amd_sched_rq_remove_entity(rq, entity);
+	if (r) {
+		struct amd_sched_job *job;
+
+		/* Park the kernel for a moment to make sure it isn't processing
+		 * our enity.
+		 */
+		kthread_park(sched->thread);
+		kthread_unpark(sched->thread);
+		while (kfifo_out(&entity->job_queue, &job, sizeof(job)))
+			sched->ops->free_job(job);
+
+	}
 	kfifo_free(&entity->job_queue);
 }
 

drivers/gpu/drm/ttm/ttm_bo.c

@@ -109,8 +109,8 @@ static ssize_t ttm_bo_global_show(struct kobject *kobj,
 	struct ttm_bo_global *glob =
 		container_of(kobj, struct ttm_bo_global, kobj);
 
-	return snprintf(buffer, PAGE_SIZE, "%lu\n",
-			(unsigned long) atomic_read(&glob->bo_count));
+	return snprintf(buffer, PAGE_SIZE, "%d\n",
+				atomic_read(&glob->bo_count));
 }
 
 static struct attribute *ttm_bo_global_attrs[] = {

drivers/gpu/drm/ttm/ttm_bo_util.c

@@ -469,6 +469,7 @@ static int ttm_buffer_object_transfer(struct ttm_buffer_object *bo,
 	 * TODO: Explicit member copy would probably be better here.
 	 */
 
+	atomic_inc(&bo->glob->bo_count);
 	INIT_LIST_HEAD(&fbo->ddestroy);
 	INIT_LIST_HEAD(&fbo->lru);
 	INIT_LIST_HEAD(&fbo->swap);