Linux Virtualization - Dec 2016 - [PATCH net] virtio-net: correctly enable multiqueue

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 at mit.edu>
Tested-by: Theodore Ts'o <tytso at mit.edu>
Cc: Neil Horman <nhorman at tuxdriver.com>
Cc: Michael S. Tsirkin <mst at redhat.com>
Fixes: 49000102901 ("virtio-net: enable multiqueue by default")
Signed-off-by: Jason Wang <jasowang at redhat.com>
---
 drivers/net/virtio_net.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index b425fa1..fe9f772 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -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. */
-- 
2.7.4

On Tue, Dec 13, 2016 at 02:23:05PM +0800, Jason Wang
wrote:
> 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 at mit.edu>
> Tested-by: Theodore Ts'o <tytso at mit.edu>
> Cc: Neil Horman <nhorman at tuxdriver.com>
> Cc: Michael S. Tsirkin <mst at redhat.com>
> Fixes: 49000102901 ("virtio-net: enable multiqueue by default")
> Signed-off-by: Jason Wang <jasowang at redhat.com>
Acked-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/net/virtio_net.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index b425fa1..fe9f772 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -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. */
I note that virtnet_set_channels also plays with affinity
directly. Can this be changed to rely on cpu notifiers
somehow?

> -- 
> 2.7.4

From: Jason Wang <jasowang at redhat.com>
Date: Tue, 13 Dec 2016 14:23:05 +0800
> 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 at mit.edu>
> Tested-by: Theodore Ts'o <tytso at mit.edu>
> Cc: Neil Horman <nhorman at tuxdriver.com>
> Cc: Michael S. Tsirkin <mst at redhat.com>
> Fixes: 49000102901 ("virtio-net: enable multiqueue by default")
> Signed-off-by: Jason Wang <jasowang at redhat.com>
Applied, thanks Jason.