[PATCH net] virtio-net: fix the race between channels setting and refill

[PATCH net] virtio-net: fix the race between channels setting and refill

[Jason Wang]

Commit 55257d72bd1c51f25106350f4983ec19f62ed1fa (virtio-net: fill only rx queues
which are being used) tries to refill on demand when changing the number of
channels by call try_refill_recv() directly, this may race:

- the refill work who may do the refill in the same time
- the try_refill_recv() called in bh since napi was not disabled

Which may led guest complain during setting channels:

virtio_net virtio0: input.1:id 0 is not a head!

Solve this issue by scheduling a refill work which can guarantee the
serialization of refill.

Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/net/virtio_net.c |    5 +----
 1 files changed, 1 insertions(+), 4 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index c9e0038..47b4882 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -901,7 +901,6 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs)
 	struct scatterlist sg;
 	struct virtio_net_ctrl_mq s;
 	struct net_device *dev = vi->dev;
-	int i;
 
 	if (!vi->has_cvq || !virtio_has_feature(vi->vdev, VIRTIO_NET_F_MQ))
 		return 0;
@@ -915,10 +914,8 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs)
 			 queue_pairs);
 		return -EINVAL;
 	} else {
-		for (i = vi->curr_queue_pairs; i < queue_pairs; i++)
-			if (!try_fill_recv(&vi->rq[i], GFP_KERNEL))
-				schedule_delayed_work(&vi->refill, 0);
 		vi->curr_queue_pairs = queue_pairs;
+		schedule_delayed_work(&vi->refill, 0);
 	}
 
 	return 0;
-- 
1.7.1

Re: [PATCH net] virtio-net: fix the race between channels setting and refill

[David Miller]

From: Jason Wang <jasowang@redhat.com>
Date: Wed,  3 Jul 2013 20:15:52 +0800

> Commit 55257d72bd1c51f25106350f4983ec19f62ed1fa (virtio-net: fill only rx queues
> which are being used) tries to refill on demand when changing the number of
> channels by call try_refill_recv() directly, this may race:
> 
> - the refill work who may do the refill in the same time
> - the try_refill_recv() called in bh since napi was not disabled
> 
> Which may led guest complain during setting channels:
> 
> virtio_net virtio0: input.1:id 0 is not a head!
> 
> Solve this issue by scheduling a refill work which can guarantee the
> serialization of refill.
> 
> Cc: Sasha Levin <sasha.levin@oracle.com>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>

Michael, please review.

Thanks.

Re: [PATCH net] virtio-net: fix the race between channels setting and refill

[Asias He]

On Wed, Jul 03, 2013 at 08:15:52PM +0800, Jason Wang wrote:
> Commit 55257d72bd1c51f25106350f4983ec19f62ed1fa (virtio-net: fill only rx queues
> which are being used) tries to refill on demand when changing the number of
> channels by call try_refill_recv() directly, this may race:
> 
> - the refill work who may do the refill in the same time
> - the try_refill_recv() called in bh since napi was not disabled
> 
> Which may led guest complain during setting channels:
> 
> virtio_net virtio0: input.1:id 0 is not a head!
> 
> Solve this issue by scheduling a refill work which can guarantee the
> serialization of refill.
> 
> Cc: Sasha Levin <sasha.levin@oracle.com>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>

Reviewed-by: Asias He <asias@redhat.com>

> ---
>  drivers/net/virtio_net.c |    5 +----
>  1 files changed, 1 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index c9e0038..47b4882 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -901,7 +901,6 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs)
>  	struct scatterlist sg;
>  	struct virtio_net_ctrl_mq s;
>  	struct net_device *dev = vi->dev;
> -	int i;
>  
>  	if (!vi->has_cvq || !virtio_has_feature(vi->vdev, VIRTIO_NET_F_MQ))
>  		return 0;
> @@ -915,10 +914,8 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs)
>  			 queue_pairs);
>  		return -EINVAL;
>  	} else {
> -		for (i = vi->curr_queue_pairs; i < queue_pairs; i++)
> -			if (!try_fill_recv(&vi->rq[i], GFP_KERNEL))
> -				schedule_delayed_work(&vi->refill, 0);
>  		vi->curr_queue_pairs = queue_pairs;
> +		schedule_delayed_work(&vi->refill, 0);
>  	}

Also, this is not on the data path, it is simpler this way.

>  
>  	return 0;
> -- 
> 1.7.1
> 
> _______________________________________________
> Virtualization mailing list
> Virtualization@lists.linux-foundation.org
> https://lists.linuxfoundation.org/mailman/listinfo/virtualization

-- 
Asias

Re: [PATCH net] virtio-net: fix the race between channels setting and refill

[Rusty Russell]

Jason Wang <jasowang@redhat.com> writes:
> Commit 55257d72bd1c51f25106350f4983ec19f62ed1fa (virtio-net: fill only rx queues
> which are being used) tries to refill on demand when changing the number of
> channels by call try_refill_recv() directly, this may race:
>
> - the refill work who may do the refill in the same time
> - the try_refill_recv() called in bh since napi was not disabled
>
> Which may led guest complain during setting channels:
>
> virtio_net virtio0: input.1:id 0 is not a head!
>
> Solve this issue by scheduling a refill work which can guarantee the
> serialization of refill.
>
> Cc: Sasha Levin <sasha.levin@oracle.com>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>

Thanks.

Applied,
Rusty.

Re: [PATCH net] virtio-net: fix the race between channels setting and refill

[Michael S. Tsirkin]

On Wed, Jul 03, 2013 at 08:15:52PM +0800, Jason Wang wrote:
> Commit 55257d72bd1c51f25106350f4983ec19f62ed1fa (virtio-net: fill only rx queues
> which are being used) tries to refill on demand when changing the number of
> channels by call try_refill_recv() directly, this may race:
> 
> - the refill work who may do the refill in the same time
> - the try_refill_recv() called in bh since napi was not disabled
> 
> Which may led guest complain during setting channels:
> 
> virtio_net virtio0: input.1:id 0 is not a head!
> 
> Solve this issue by scheduling a refill work which can guarantee the
> serialization of refill.
> 
> Cc: Sasha Levin <sasha.levin@oracle.com>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>

Acked-by: Michael S. Tsirkin <mst@redhat.com>

> ---
>  drivers/net/virtio_net.c |    5 +----
>  1 files changed, 1 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index c9e0038..47b4882 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -901,7 +901,6 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs)
>  	struct scatterlist sg;
>  	struct virtio_net_ctrl_mq s;
>  	struct net_device *dev = vi->dev;
> -	int i;
>  
>  	if (!vi->has_cvq || !virtio_has_feature(vi->vdev, VIRTIO_NET_F_MQ))
>  		return 0;
> @@ -915,10 +914,8 @@ static int virtnet_set_queues(struct virtnet_info *vi, u16 queue_pairs)
>  			 queue_pairs);
>  		return -EINVAL;
>  	} else {
> -		for (i = vi->curr_queue_pairs; i < queue_pairs; i++)
> -			if (!try_fill_recv(&vi->rq[i], GFP_KERNEL))
> -				schedule_delayed_work(&vi->refill, 0);
>  		vi->curr_queue_pairs = queue_pairs;
> +		schedule_delayed_work(&vi->refill, 0);
>  	}
>  
>  	return 0;
> -- 
> 1.7.1