[PATCH 08/11] drm/virtio: move the check for vqs_ready earlier

Chia-I Wu olvaffe at gmail.com
Wed Feb 5 18:19:52 UTC 2020


When vqs_ready is false, vq should be considered invalid and we
should not check vq->num_free.  After this change, a fenced command
queued before the vqs are ready will have fence id 0 and will be
considered done.

Signed-off-by: Chia-I Wu <olvaffe at gmail.com>
---
 drivers/gpu/drm/virtio/virtgpu_vq.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c
index 299470aac281a..0bf82cff8da37 100644
--- a/drivers/gpu/drm/virtio/virtgpu_vq.c
+++ b/drivers/gpu/drm/virtio/virtgpu_vq.c
@@ -333,6 +333,14 @@ static bool virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
 again:
 	spin_lock(&vgdev->ctrlq.qlock);
 
+	if (!vgdev->vqs_ready) {
+		spin_unlock(&vgdev->ctrlq.qlock);
+
+		if (fence && vbuf->objs)
+			virtio_gpu_array_unlock_resv(vbuf->objs);
+		return notify;
+	}
+
 	if (vq->num_free < elemcnt) {
 		spin_unlock(&vgdev->ctrlq.qlock);
 		wait_event(vgdev->ctrlq.ack_queue, vq->num_free >= elemcnt);
@@ -351,11 +359,6 @@ static bool virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
 		}
 	}
 
-	if (!vgdev->vqs_ready) {
-		spin_unlock(&vgdev->ctrlq.qlock);
-		return notify;
-	}
-
 	ret = virtqueue_add_sgs(vq, sgs, outcnt, incnt, vbuf, GFP_ATOMIC);
 	WARN_ON(ret);
 
-- 
2.25.0.341.g760bfbb309-goog



More information about the dri-devel mailing list