From: Stefan Hajnoczi <stefa...@redhat.com>

virtio_queue_aio_attach_host_notifier() does not require the AioContext
lock. Stop taking the lock and add an explicit smp_wmb() because we were
relying on the implicit barrier in the AioContext lock before.

Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
Reviewed-by: Eric Blake <ebl...@redhat.com>
Reviewed-by: Kevin Wolf <kw...@redhat.com>
Message-ID: <20231204164259.1515217-3-stefa...@redhat.com>
Signed-off-by: Kevin Wolf <kw...@redhat.com>
---
 hw/scsi/virtio-scsi-dataplane.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/hw/scsi/virtio-scsi-dataplane.c b/hw/scsi/virtio-scsi-dataplane.c
index 1e684beebe..135e23fe54 100644
--- a/hw/scsi/virtio-scsi-dataplane.c
+++ b/hw/scsi/virtio-scsi-dataplane.c
@@ -149,23 +149,17 @@ int virtio_scsi_dataplane_start(VirtIODevice *vdev)
 
     memory_region_transaction_commit();
 
-    /*
-     * These fields are visible to the IOThread so we rely on implicit barriers
-     * in aio_context_acquire() on the write side and aio_notify_accept() on
-     * the read side.
-     */
     s->dataplane_starting = false;
     s->dataplane_started = true;
+    smp_wmb(); /* paired with aio_notify_accept() */
 
     if (s->bus.drain_count == 0) {
-        aio_context_acquire(s->ctx);
         virtio_queue_aio_attach_host_notifier(vs->ctrl_vq, s->ctx);
         virtio_queue_aio_attach_host_notifier_no_poll(vs->event_vq, s->ctx);
 
         for (i = 0; i < vs->conf.num_queues; i++) {
             virtio_queue_aio_attach_host_notifier(vs->cmd_vqs[i], s->ctx);
         }
-        aio_context_release(s->ctx);
     }
     return 0;
 
-- 
2.43.0


Reply via email to