virtio-net: correctly enable multiqueue
authorJason Wang <jasowang@redhat.com>
Tue, 13 Dec 2016 06:23:05 +0000 (14:23 +0800)
committerDavid S. Miller <davem@davemloft.net>
Tue, 13 Dec 2016 15:37:38 +0000 (10:37 -0500)
Commit 4490001029012539937ff02778fe6180613fa949 ("virtio-net: enable
multiqueue by default") blindly set the affinity instead of queues
during probe which can cause a mismatch of #queues between guest and
host. This patch fixes it by setting queues.

Reported-by: Theodore Ts'o <tytso@mit.edu>
Tested-by: Theodore Ts'o <tytso@mit.edu>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Fixes: 49000102901 ("virtio-net: enable multiqueue by default")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/virtio_net.c

index b425fa1013af918529f9d66e54652308dd06361a..fe9f7726e99053c5ec6cfbb46320f16cc39773ad 100644 (file)
@@ -1930,7 +1930,9 @@ static int virtnet_probe(struct virtio_device *vdev)
                goto free_unregister_netdev;
        }
 
-       virtnet_set_affinity(vi);
+       rtnl_lock();
+       virtnet_set_queues(vi, vi->curr_queue_pairs);
+       rtnl_unlock();
 
        /* Assume link up if device can't report link status,
           otherwise get link status from config. */