Re: [PATCH 3/4] virtio_ring: introduce a per virtqueue waitqueue

On Sun, Jan 29, 2023 at 3:37 PM Hillf Danton [off-list ref] wrote:

On Mon, 26 Dec 2022 15:49:07 +0800 Jason Wang [off-list ref]

quoted

@@ -2654,6 +2663,8 @@ static void vring_free(struct virtqueue *_vq)
 {
      struct vring_virtqueue *vq = to_vvq(_vq);

+     wake_up_interruptible(&vq->wq);
+
      if (vq->we_own_ring) {
              if (vq->packed_ring) {
                      vring_free_queue(vq->vq.vdev,

@@ -2863,4 +2874,22 @@ const struct vring *virtqueue_get_vring(struct virtqueue *vq)
 }
 EXPORT_SYMBOL_GPL(virtqueue_get_vring);

+int virtqueue_wait_for_used(struct virtqueue *_vq)
+{
+     struct vring_virtqueue *vq = to_vvq(_vq);
+
+     /* TODO: Tweak the timeout. */
+     return wait_event_interruptible_timeout(vq->wq,
+            virtqueue_is_broken(_vq) || more_used(vq), HZ);
+}
+EXPORT_SYMBOL_GPL(virtqueue_wait_for_used);

        waker           waiter
        ---             ---
        vring_del_virtqueue
          vring_free(_vq);
            wakeup
          kfree(vq);
                        get on CPU a tick later
                        uaf ?

Exactly, this wakeup of vring_free is not needed. It's up to the
driver to do the proper wake up to avoid race when subsystem un
registration.

Thanks