drm/amdgpu: move get_user_pages out of amdgpu_ttm_tt_pin_userptr v6

That avoids lock inversion between the BO reservation lock
and the anon_vma lock.

v2:
* Changed amdgpu_bo_list_entry.user_pages to an array of pointers
* Lock mmap_sem only for get_user_pages
* Added invalidation of unbound userpointer BOs
* Fixed memory leak and page reference leak

v3 (chk):
* Revert locking mmap_sem only for_get user_pages
* Revert adding invalidation of unbound userpointer BOs
* Sanitize and fix error handling

v4 (chk):
* Init userpages pointer everywhere.
* Fix error handling when get_user_pages() fails.
* Add invalidation of unbound userpointer BOs again.

v5 (chk):
* Add maximum number of tries.

v6 (chk):
* Fix error handling when we run out of tries.

Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> (v4)
Acked-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -471,6 +471,8 @@ struct amdgpu_bo_list_entry {
 	struct ttm_validate_buffer	tv;
 	struct amdgpu_bo_va		*bo_va;
 	uint32_t			priority;
+	struct page			**user_pages;
+	int				user_invalidated;
 };
 
 struct amdgpu_bo_va_mapping {
@@ -2365,11 +2367,14 @@ int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type,
 		       struct amdgpu_ring **out_ring);
 void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *rbo, u32 domain);
 bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
+int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages);
 int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
 				     uint32_t flags);
 struct mm_struct *amdgpu_ttm_tt_get_usermm(struct ttm_tt *ttm);
 bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
 				  unsigned long end);
+bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
+				       int *last_invalidated);
 bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
 uint32_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
 				 struct ttm_mem_reg *mem);

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

@@ -200,6 +200,7 @@ void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
 
 		list_add_tail(&list->array[i].tv.head,
 			      &bucket[priority]);
+		list->array[i].user_pages = NULL;
 	}
 
 	/* Connect the sorted buckets in the output list. */

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

@@ -111,6 +111,7 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 	p->uf_entry.priority = 0;
 	p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
 	p->uf_entry.tv.shared = true;
+	p->uf_entry.user_pages = NULL;
 
 	drm_gem_object_unreference_unlocked(gobj);
 	return 0;
@@ -297,6 +298,7 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 
 	list_for_each_entry(lobj, validated, tv.head) {
 		struct amdgpu_bo *bo = lobj->robj;
+		bool binding_userptr = false;
 		struct mm_struct *usermm;
 		uint32_t domain;
 
@@ -304,6 +306,15 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 		if (usermm && usermm != current->mm)
 			return -EPERM;
 
+		/* Check if we have user pages and nobody bound the BO already */
+		if (lobj->user_pages && bo->tbo.ttm->state != tt_bound) {
+			size_t size = sizeof(struct page *);
+
+			size *= bo->tbo.ttm->num_pages;
+			memcpy(bo->tbo.ttm->pages, lobj->user_pages, size);
+			binding_userptr = true;
+		}
+
 		if (bo->pin_count)
 			continue;
 
@@ -334,6 +345,11 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 			}
 			return r;
 		}
+
+		if (binding_userptr) {
+			drm_free_large(lobj->user_pages);
+			lobj->user_pages = NULL;
+		}
 	}
 	return 0;
 }
@@ -342,8 +358,10 @@ 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_bo_list_entry *e;
 	struct list_head duplicates;
 	bool need_mmap_lock = false;
+	unsigned i, tries = 10;
 	int r;
 
 	INIT_LIST_HEAD(&p->validated);
@@ -364,9 +382,81 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	if (need_mmap_lock)
 		down_read(&current->mm->mmap_sem);
 
-	r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true, &duplicates);
-	if (unlikely(r != 0))
-		goto error_reserve;
+	while (1) {
+		struct list_head need_pages;
+		unsigned i;
+
+		r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
+					   &duplicates);
+		if (unlikely(r != 0))
+			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) {
+
+			e = &p->bo_list->array[i];
+
+			if (amdgpu_ttm_tt_userptr_invalidated(e->robj->tbo.ttm,
+				 &e->user_invalidated) && e->user_pages) {
+
+				/* We acquired a page array, but somebody
+				 * invalidated it. Free it an try again
+				 */
+				release_pages(e->user_pages,
+					      e->robj->tbo.ttm->num_pages,
+					      false);
+				drm_free_large(e->user_pages);
+				e->user_pages = NULL;
+			}
+
+			if (e->robj->tbo.ttm->state != tt_bound &&
+			    !e->user_pages) {
+				list_del(&e->tv.head);
+				list_add(&e->tv.head, &need_pages);
+
+				amdgpu_bo_unreserve(e->robj);
+			}
+		}
+
+		if (list_empty(&need_pages))
+			break;
+
+		/* Unreserve everything again. */
+		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+
+		/* We tried to often, just abort */
+		if (!--tries) {
+			r = -EDEADLK;
+			goto error_free_pages;
+		}
+
+		/* Fill the page arrays for all useptrs. */
+		list_for_each_entry(e, &need_pages, tv.head) {
+			struct ttm_tt *ttm = e->robj->tbo.ttm;
+
+			e->user_pages = drm_calloc_large(ttm->num_pages,
+							 sizeof(struct page*));
+			if (!e->user_pages) {
+				r = -ENOMEM;
+				goto error_free_pages;
+			}
+
+			r = amdgpu_ttm_tt_get_user_pages(ttm, e->user_pages);
+			if (r) {
+				drm_free_large(e->user_pages);
+				e->user_pages = NULL;
+				goto error_free_pages;
+			}
+		}
+
+		/* And try again. */
+		list_splice(&need_pages, &p->validated);
+	}
 
 	amdgpu_vm_get_pt_bos(&fpriv->vm, &duplicates);
 
@@ -398,10 +488,26 @@ error_validate:
 		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
 	}
 
-error_reserve:
+error_free_pages:
+
 	if (need_mmap_lock)
 		up_read(&current->mm->mmap_sem);
 
+	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;
+
+			release_pages(e->user_pages,
+				      e->robj->tbo.ttm->num_pages,
+				      false);
+			drm_free_large(e->user_pages);
+		}
+	}
+
 	return r;
 }
 

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

@@ -274,18 +274,23 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 
 	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) {
 		down_read(&current->mm->mmap_sem);
+
+		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
+						 bo->tbo.ttm->pages);
+		if (r)
+			goto unlock_mmap_sem;
+
 		r = amdgpu_bo_reserve(bo, true);
-		if (r) {
-			up_read(&current->mm->mmap_sem);
-			goto release_object;
-		}
+		if (r)
+			goto free_pages;
 
 		amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
 		amdgpu_bo_unreserve(bo);
-		up_read(&current->mm->mmap_sem);
 		if (r)
-			goto release_object;
+			goto free_pages;
+
+		up_read(&current->mm->mmap_sem);
 	}
 
 	r = drm_gem_handle_create(filp, gobj, &handle);
@@ -297,6 +302,12 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 	args->handle = handle;
 	return 0;
 
+free_pages:
+	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages, false);
+
+unlock_mmap_sem:
+	up_read(&current->mm->mmap_sem);
+
 release_object:
 	drm_gem_object_unreference_unlocked(gobj);
 

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

@@ -508,22 +508,18 @@ struct amdgpu_ttm_tt {
 	uint32_t		userflags;
 	spinlock_t              guptasklock;
 	struct list_head        guptasks;
+	atomic_t		mmu_invalidations;
 };
 
-/* prepare the sg table with the user pages */
-static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
+int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
 {
-	struct amdgpu_device *adev = amdgpu_get_adev(ttm->bdev);
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
-	unsigned pinned = 0, nents;
+	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+	unsigned pinned = 0;
 	int r;
 
-	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
-	enum dma_data_direction direction = write ?
-		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
-
 	if (gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) {
-		/* check that we only pin down anonymous memory
+		/* check that we only use anonymous memory
 		   to prevent problems with writeback */
 		unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
 		struct vm_area_struct *vma;
@@ -536,7 +532,7 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 	do {
 		unsigned num_pages = ttm->num_pages - pinned;
 		uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE;
-		struct page **pages = ttm->pages + pinned;
+		struct page **p = pages + pinned;
 		struct amdgpu_ttm_gup_task_list guptask;
 
 		guptask.task = current;
@@ -545,7 +541,7 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 		spin_unlock(&gtt->guptasklock);
 
 		r = get_user_pages(current, current->mm, userptr, num_pages,
-				   write, 0, pages, NULL);
+				   write, 0, p, NULL);
 
 		spin_lock(&gtt->guptasklock);
 		list_del(&guptask.list);
@@ -558,6 +554,25 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 
 	} while (pinned < ttm->num_pages);
 
+	return 0;
+
+release_pages:
+	release_pages(pages, pinned, 0);
+	return r;
+}
+
+/* prepare the sg table with the user pages */
+static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_get_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned nents;
+	int r;
+
+	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+	enum dma_data_direction direction = write ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+
 	r = sg_alloc_table_from_pages(ttm->sg, ttm->pages, ttm->num_pages, 0,
 				      ttm->num_pages << PAGE_SHIFT,
 				      GFP_KERNEL);
@@ -576,9 +591,6 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 
 release_sg:
 	kfree(ttm->sg);
-
-release_pages:
-	release_pages(ttm->pages, pinned, 0);
 	return r;
 }
 
@@ -803,6 +815,7 @@ int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
 	gtt->userflags = flags;
 	spin_lock_init(&gtt->guptasklock);
 	INIT_LIST_HEAD(&gtt->guptasks);
+	atomic_set(&gtt->mmu_invalidations, 0);
 
 	return 0;
 }
@@ -840,9 +853,21 @@ bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
 	}
 	spin_unlock(&gtt->guptasklock);
 
+	atomic_inc(&gtt->mmu_invalidations);
+
 	return true;
 }
 
+bool amdgpu_ttm_tt_userptr_invalidated(struct ttm_tt *ttm,
+				       int *last_invalidated)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	int prev_invalidated = *last_invalidated;
+
+	*last_invalidated = atomic_read(&gtt->mmu_invalidations);
+	return prev_invalidated != *last_invalidated;
+}
+
 bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm)
 {
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;

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

@@ -95,6 +95,7 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
 	entry->priority = 0;
 	entry->tv.bo = &vm->page_directory->tbo;
 	entry->tv.shared = true;
+	entry->user_pages = NULL;
 	list_add(&entry->tv.head, validated);
 }
 
@@ -1186,6 +1187,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		entry->priority = 0;
 		entry->tv.bo = &entry->robj->tbo;
 		entry->tv.shared = true;
+		entry->user_pages = NULL;
 		vm->page_tables[pt_idx].addr = 0;
 	}