Linux Virtualization - Mar 2023 - [PATCH net v2 2/2] virtio_net: free xdp shinfo frags when build_skb_from_xdp_buff() fails

On Wed, 15 Mar 2023 10:14:34 +0800, Yunsheng Lin <linyunsheng at
huawei.com> wrote:
> On 2023/3/15 9:52, Xuan Zhuo wrote:
> > build_skb_from_xdp_buff() may return NULL, in this case
> > we need to free the frags of xdp shinfo.
> >
> > Fixes: fab89bafa95b ("virtio-net: support multi-buffer xdp")
> > Signed-off-by: Xuan Zhuo <xuanzhuo at linux.alibaba.com>
> > Acked-by: Michael S. Tsirkin <mst at redhat.com>
> > ---
> >  drivers/net/virtio_net.c | 5 ++++-
> >  1 file changed, 4 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 8ecf7a341d54..2396c28c0122 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -1273,9 +1273,12 @@ static struct sk_buff *receive_mergeable(struct
net_device *dev,
> >
> >  		switch (act) {
> >  		case XDP_PASS:
> > +			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp,
xdp_frags_truesz);
> > +			if (unlikely(!head_skb))
> > +				goto err_xdp_frags;
>
> LGTM.
> Reviewed-by: Yunsheng Lin <linyunsheng at huawei.com>
>
> Note, "stats->drops++; dev_kfree_skb(head_skb);" is also done
for the above case,
> I assume it is ok as other "goto err_xdp_frags" case also do
that.

Yes. It's ok.

Thanks.

>
> > +
> >  			if (unlikely(xdp_page != page))
> >  				put_page(page);
> > -			head_skb = build_skb_from_xdp_buff(dev, vi, &xdp,
xdp_frags_truesz);
> >  			rcu_read_unlock();
> >  			return head_skb;
> >  		case XDP_TX:
> >