drm/virtio: implement blob resources: fix stride discrepancy

The old transfer ioctls may work on blob resources, and there is no
TRANSFER_BLOB hypercall now for simplicity.

The guest may have a image view on the blob resources such that the
stride is not equal to width * bytes_per_pixel.

For host-only blobs, we can repurpose the transfer ioctls to synchronize
caches as well.  For guest-only blobs, these operations are undefined
for now so leave them out.

Also, with seamless Wayland integration between guest/host looking
increasingly attractive, it also makes sense to keep track of
one value for stride.
Signed-off-by: Gurchetan Singh <gurchetansingh@chromium.org>
Acked-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20200924003214.662-16-gurchetansingh@chromium.orgSigned-off-by: Gerd Hoffmann <kraxel@redhat.com>

drivers/gpu/drm/virtio/virtgpu_drv.h

@@ -351,12 +351,16 @@ void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev,
 void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev,
 					  uint32_t ctx_id,
 					  uint64_t offset, uint32_t level,
+					  uint32_t stride,
+					  uint32_t layer_stride,
 					  struct drm_virtgpu_3d_box *box,
 					  struct virtio_gpu_object_array *objs,
 					  struct virtio_gpu_fence *fence);
 void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 					uint32_t ctx_id,
 					uint64_t offset, uint32_t level,
+					uint32_t stride,
+					uint32_t layer_stride,
 					struct drm_virtgpu_3d_box *box,
 					struct virtio_gpu_object_array *objs,
 					struct virtio_gpu_fence *fence);

drivers/gpu/drm/virtio/virtgpu_ioctl.c

@@ -312,6 +312,7 @@ static int virtio_gpu_transfer_from_host_ioctl(struct drm_device *dev,
 	struct virtio_gpu_device *vgdev = dev->dev_private;
 	struct virtio_gpu_fpriv *vfpriv = file->driver_priv;
 	struct drm_virtgpu_3d_transfer_from_host *args = data;
+	struct virtio_gpu_object *bo;
 	struct virtio_gpu_object_array *objs;
 	struct virtio_gpu_fence *fence;
 	int ret;
@@ -325,6 +326,17 @@ static int virtio_gpu_transfer_from_host_ioctl(struct drm_device *dev,
 	if (objs == NULL)
 		return -ENOENT;
 
+	bo = gem_to_virtio_gpu_obj(objs->objs[0]);
+	if (bo->guest_blob && !bo->host3d_blob) {
+		ret = -EINVAL;
+		goto err_put_free;
+	}
+
+	if (!bo->host3d_blob && (args->stride || args->layer_stride)) {
+		ret = -EINVAL;
+		goto err_put_free;
+	}
+
 	ret = virtio_gpu_array_lock_resv(objs);
 	if (ret != 0)
 		goto err_put_free;
@@ -334,9 +346,10 @@ static int virtio_gpu_transfer_from_host_ioctl(struct drm_device *dev,
 		ret = -ENOMEM;
 		goto err_unlock;
 	}
+
 	virtio_gpu_cmd_transfer_from_host_3d
-		(vgdev, vfpriv->ctx_id, offset, args->level,
-		 &args->box, objs, fence);
+		(vgdev, vfpriv->ctx_id, offset, args->level, args->stride,
+		 args->layer_stride, &args->box, objs, fence);
 	dma_fence_put(&fence->f);
 	virtio_gpu_notify(vgdev);
 	return 0;
@@ -354,6 +367,7 @@ static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data,
 	struct virtio_gpu_device *vgdev = dev->dev_private;
 	struct virtio_gpu_fpriv *vfpriv = file->driver_priv;
 	struct drm_virtgpu_3d_transfer_to_host *args = data;
+	struct virtio_gpu_object *bo;
 	struct virtio_gpu_object_array *objs;
 	struct virtio_gpu_fence *fence;
 	int ret;
@@ -363,6 +377,12 @@ static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data,
 	if (objs == NULL)
 		return -ENOENT;
 
+	bo = gem_to_virtio_gpu_obj(objs->objs[0]);
+	if (bo->guest_blob && !bo->host3d_blob) {
+		ret = -EINVAL;
+		goto err_put_free;
+	}
+
 	if (!vgdev->has_virgl_3d) {
 		virtio_gpu_cmd_transfer_to_host_2d
 			(vgdev, offset,
@@ -370,6 +390,12 @@ static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data,
 			 objs, NULL);
 	} else {
 		virtio_gpu_create_context(dev, file);
+
+		if (!bo->host3d_blob && (args->stride || args->layer_stride)) {
+			ret = -EINVAL;
+			goto err_put_free;
+		}
+
 		ret = virtio_gpu_array_lock_resv(objs);
 		if (ret != 0)
 			goto err_put_free;
@@ -381,8 +407,9 @@ static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data,
 
 		virtio_gpu_cmd_transfer_to_host_3d
 			(vgdev,
-			 vfpriv ? vfpriv->ctx_id : 0, offset,
-			 args->level, &args->box, objs, fence);
+			 vfpriv ? vfpriv->ctx_id : 0, offset, args->level,
+			 args->stride, args->layer_stride, &args->box, objs,
+			 fence);
 		dma_fence_put(&fence->f);
 	}
 	virtio_gpu_notify(vgdev);

drivers/gpu/drm/virtio/virtgpu_vq.c

@@ -1017,6 +1017,8 @@ virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev,
 void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 					uint32_t ctx_id,
 					uint64_t offset, uint32_t level,
+					uint32_t stride,
+					uint32_t layer_stride,
 					struct drm_virtgpu_3d_box *box,
 					struct virtio_gpu_object_array *objs,
 					struct virtio_gpu_fence *fence)
@@ -1025,12 +1027,14 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 	struct virtio_gpu_transfer_host_3d *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
 	bool use_dma_api = !virtio_has_dma_quirk(vgdev->vdev);
+
+	if (virtio_gpu_is_shmem(bo) && use_dma_api) {
 		struct virtio_gpu_object_shmem *shmem = to_virtio_gpu_shmem(bo);
 
-	if (use_dma_api)
 		dma_sync_sg_for_device(vgdev->vdev->dev.parent,
 				       shmem->pages->sgl, shmem->pages->nents,
 				       DMA_TO_DEVICE);
+	}
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
@@ -1043,6 +1047,8 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 	convert_to_hw_box(&cmd_p->box, box);
 	cmd_p->offset = cpu_to_le64(offset);
 	cmd_p->level = cpu_to_le32(level);
+	cmd_p->stride = cpu_to_le32(stride);
+	cmd_p->layer_stride = cpu_to_le32(layer_stride);
 
 	virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence);
 }
@@ -1050,6 +1056,8 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev,
 					  uint32_t ctx_id,
 					  uint64_t offset, uint32_t level,
+					  uint32_t stride,
+					  uint32_t layer_stride,
 					  struct drm_virtgpu_3d_box *box,
 					  struct virtio_gpu_object_array *objs,
 					  struct virtio_gpu_fence *fence)
@@ -1069,6 +1077,8 @@ void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev,
 	convert_to_hw_box(&cmd_p->box, box);
 	cmd_p->offset = cpu_to_le64(offset);
 	cmd_p->level = cpu_to_le32(level);
+	cmd_p->stride = cpu_to_le32(stride);
+	cmd_p->layer_stride = cpu_to_le32(layer_stride);
 
 	virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence);
 }