drm/ttm: Make LRU removal optional v2

We are already doing this for DMA-buf imports and also for
amdgpu VM BOs for quite a while now.

If this doesn't run into any problems we are probably going
to stop removing BOs from the LRU altogether.

v2: drop BUG_ON from ttm_bo_add_to_lru

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Tested-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -585,7 +585,7 @@ static int reserve_bo_and_vm(struct kgd_mem *mem,
 	amdgpu_vm_get_pd_bo(vm, &ctx->list, &ctx->vm_pd[0]);
 
 	ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list,
-				     false, &ctx->duplicates);
+				     false, &ctx->duplicates, true);
 	if (!ret)
 		ctx->reserved = true;
 	else {
@@ -658,7 +658,7 @@ static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
 	}
 
 	ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list,
-				     false, &ctx->duplicates);
+				     false, &ctx->duplicates, true);
 	if (!ret)
 		ctx->reserved = true;
 	else
@@ -1808,7 +1808,8 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
 	}
 
 	/* Reserve all BOs and page tables for validation */
-	ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates);
+	ret = ttm_eu_reserve_buffers(&ticket, &resv_list, false, &duplicates,
+				     true);
 	WARN(!list_empty(&duplicates), "Duplicates should be empty");
 	if (ret)
 		goto out_free;
@@ -2014,7 +2015,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
 	}
 
 	ret = ttm_eu_reserve_buffers(&ctx.ticket, &ctx.list,
-				     false, &duplicate_save);
+				     false, &duplicate_save, true);
 	if (ret) {
 		pr_debug("Memory eviction: TTM Reserve Failed. Try again\n");
 		goto ttm_reserve_fail;

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

@@ -648,7 +648,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	}
 
 	r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
-				   &duplicates);
+				   &duplicates, true);
 	if (unlikely(r != 0)) {
 		if (r != -ERESTARTSYS)
 			DRM_ERROR("ttm_eu_reserve_buffers failed.\n");

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

@@ -79,7 +79,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	list_add(&csa_tv.head, &list);
 	amdgpu_vm_get_pd_bo(vm, &list, &pd);
 
-	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true);
 	if (r) {
 		DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r);
 		return r;

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

@@ -171,7 +171,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
 
 	amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
 
-	r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates);
+	r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, true);
 	if (r) {
 		dev_err(adev->dev, "leaking bo va because "
 			"we fail to reserve bo (%d)\n", r);
@@ -608,7 +608,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 
 	amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
 
-	r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates);
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, true);
 	if (r)
 		goto error_unref;
 

drivers/gpu/drm/qxl/qxl_release.c

@@ -256,7 +256,7 @@ int qxl_release_reserve_list(struct qxl_release *release, bool no_intr)
 		return 0;
 
 	ret = ttm_eu_reserve_buffers(&release->ticket, &release->bos,
-				     !no_intr, NULL);
+				     !no_intr, NULL, true);
 	if (ret)
 		return ret;
 

drivers/gpu/drm/radeon/radeon_gem.c

@@ -559,7 +559,7 @@ static void radeon_gem_va_update_vm(struct radeon_device *rdev,
 	if (!vm_bos)
 		return;
 
-	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true);
 	if (r)
 		goto error_free;
 

drivers/gpu/drm/radeon/radeon_object.c

@@ -539,7 +539,7 @@ int radeon_bo_list_validate(struct radeon_device *rdev,
 	u64 bytes_moved_threshold = radeon_bo_get_threshold_for_moves(rdev);
 
 	INIT_LIST_HEAD(&duplicates);
-	r = ttm_eu_reserve_buffers(ticket, head, true, &duplicates);
+	r = ttm_eu_reserve_buffers(ticket, head, true, &duplicates, true);
 	if (unlikely(r != 0)) {
 		return r;
 	}

drivers/gpu/drm/ttm/ttm_bo.c

@@ -173,19 +173,20 @@ void ttm_bo_add_to_lru(struct ttm_buffer_object *bo)
 
 	reservation_object_assert_held(bo->resv);
 
-	if (!(bo->mem.placement & TTM_PL_FLAG_NO_EVICT)) {
-		BUG_ON(!list_empty(&bo->lru));
+	if (!list_empty(&bo->lru))
+		return;
 
-		man = &bdev->man[bo->mem.mem_type];
-		list_add_tail(&bo->lru, &man->lru[bo->priority]);
+	if (bo->mem.placement & TTM_PL_FLAG_NO_EVICT)
+		return;
+
+	man = &bdev->man[bo->mem.mem_type];
+	list_add_tail(&bo->lru, &man->lru[bo->priority]);
+	kref_get(&bo->list_kref);
+
+	if (bo->ttm && !(bo->ttm->page_flags &
+			 (TTM_PAGE_FLAG_SG | TTM_PAGE_FLAG_SWAPPED))) {
+		list_add_tail(&bo->swap, &bdev->glob->swap_lru[bo->priority]);
 		kref_get(&bo->list_kref);
-
-		if (bo->ttm && !(bo->ttm->page_flags &
-				 (TTM_PAGE_FLAG_SG | TTM_PAGE_FLAG_SWAPPED))) {
-			list_add_tail(&bo->swap,
-				      &bdev->glob->swap_lru[bo->priority]);
-			kref_get(&bo->list_kref);
-		}
 	}
 }
 EXPORT_SYMBOL(ttm_bo_add_to_lru);

drivers/gpu/drm/ttm/ttm_execbuf_util.c

@@ -69,7 +69,8 @@ void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket,
 	list_for_each_entry(entry, list, head) {
 		struct ttm_buffer_object *bo = entry->bo;
 
-		ttm_bo_add_to_lru(bo);
+		if (list_empty(&bo->lru))
+			ttm_bo_add_to_lru(bo);
 		reservation_object_unlock(bo->resv);
 	}
 	spin_unlock(&glob->lru_lock);
@@ -93,7 +94,7 @@ EXPORT_SYMBOL(ttm_eu_backoff_reservation);
 
 int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket,
 			   struct list_head *list, bool intr,
-			   struct list_head *dups)
+			   struct list_head *dups, bool del_lru)
 {
 	struct ttm_bo_global *glob;
 	struct ttm_validate_buffer *entry;
@@ -172,11 +173,11 @@ int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket,
 		list_add(&entry->head, list);
 	}
 
-	if (ticket)
-		ww_acquire_done(ticket);
-	spin_lock(&glob->lru_lock);
-	ttm_eu_del_from_lru_locked(list);
-	spin_unlock(&glob->lru_lock);
+	if (del_lru) {
+		spin_lock(&glob->lru_lock);
+		ttm_eu_del_from_lru_locked(list);
+		spin_unlock(&glob->lru_lock);
+	}
 	return 0;
 }
 EXPORT_SYMBOL(ttm_eu_reserve_buffers);
@@ -203,7 +204,10 @@ void ttm_eu_fence_buffer_objects(struct ww_acquire_ctx *ticket,
 			reservation_object_add_shared_fence(bo->resv, fence);
 		else
 			reservation_object_add_excl_fence(bo->resv, fence);
-		ttm_bo_add_to_lru(bo);
+		if (list_empty(&bo->lru))
+			ttm_bo_add_to_lru(bo);
+		else
+			ttm_bo_move_to_lru_tail(bo, NULL);
 		reservation_object_unlock(bo->resv);
 	}
 	spin_unlock(&glob->lru_lock);

drivers/gpu/drm/virtio/virtgpu_ioctl.c

@@ -63,7 +63,7 @@ int virtio_gpu_object_list_validate(struct ww_acquire_ctx *ticket,
 	struct virtio_gpu_object *qobj;
 	int ret;
 
-	ret = ttm_eu_reserve_buffers(ticket, head, true, NULL);
+	ret = ttm_eu_reserve_buffers(ticket, head, true, NULL, true);
 	if (ret != 0)
 		return ret;
 

drivers/gpu/drm/vmwgfx/vmwgfx_resource.c

@@ -464,7 +464,8 @@ vmw_resource_check_buffer(struct ww_acquire_ctx *ticket,
 	val_buf->bo = &res->backup->base;
 	val_buf->num_shared = 0;
 	list_add_tail(&val_buf->head, &val_list);
-	ret = ttm_eu_reserve_buffers(ticket, &val_list, interruptible, NULL);
+	ret = ttm_eu_reserve_buffers(ticket, &val_list, interruptible, NULL,
+				     true);
 	if (unlikely(ret != 0))
 		goto out_no_reserve;
 

drivers/gpu/drm/vmwgfx/vmwgfx_validation.h

@@ -169,7 +169,7 @@ vmw_validation_bo_reserve(struct vmw_validation_context *ctx,
 			  bool intr)
 {
 	return ttm_eu_reserve_buffers(&ctx->ticket, &ctx->bo_list, intr,
-				      NULL);
+				      NULL, true);
 }
 
 /**

include/drm/ttm/ttm_bo_driver.h

@@ -769,7 +769,10 @@ static inline void ttm_bo_unreserve(struct ttm_buffer_object *bo)
 {
 	if (!(bo->mem.placement & TTM_PL_FLAG_NO_EVICT)) {
 		spin_lock(&bo->bdev->glob->lru_lock);
-		ttm_bo_add_to_lru(bo);
+		if (list_empty(&bo->lru))
+			ttm_bo_add_to_lru(bo);
+		else
+			ttm_bo_move_to_lru_tail(bo, NULL);
 		spin_unlock(&bo->bdev->glob->lru_lock);
 	}
 	reservation_object_unlock(bo->resv);

include/drm/ttm/ttm_execbuf_util.h

@@ -70,6 +70,7 @@ extern void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket,
  * @list:    thread private list of ttm_validate_buffer structs.
  * @intr:    should the wait be interruptible
  * @dups:    [out] optional list of duplicates.
+ * @del_lru: true if BOs should be removed from the LRU.
  *
  * Tries to reserve bos pointed to by the list entries for validation.
  * If the function returns 0, all buffers are marked as "unfenced",
@@ -98,7 +99,7 @@ extern void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket,
 
 extern int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket,
 				  struct list_head *list, bool intr,
-				  struct list_head *dups);
+				  struct list_head *dups, bool del_lru);
 
 /**
  * function ttm_eu_fence_buffer_objects.