Re: [PATCH v2 2/3] virtio_net: Add missing virtio header in skb for XDP_PASS

From: Jason Wang
Date: Wed Jan 24 2024 - 22:48:42 EST


On Wed, Jan 24, 2024 at 5:16 PM Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx> wrote:
>
> On Wed, 24 Jan 2024 16:57:20 +0800, Liang Chen <liangchen.linux@xxxxxxxxx> wrote:
> > For the XDP_PASS scenario of the XDP path, the skb constructed with
> > xdp_buff does not include the virtio header. Adding the virtio header
> > information back when creating the skb.
> >
> > Signed-off-by: Liang Chen <liangchen.linux@xxxxxxxxx>
> > ---
> > drivers/net/virtio_net.c | 6 ++++++
> > 1 file changed, 6 insertions(+)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index b56828804e5f..2de46eb4c661 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -1270,6 +1270,9 @@ static struct sk_buff *receive_small_xdp(struct net_device *dev,
> > if (unlikely(!skb))
> > goto err;
> >
> > + /* Store the original virtio header for subsequent use by the driver. */
> > + memcpy(skb_vnet_common_hdr(skb), &virtnet_xdp.hdr, vi->hdr_len);
>
> About this, a spec is waiting for voting.
>

A pointer?

> This may change the logic of the csum offset and so on.

Btw, doesn't it need a new feature bit?

Thanks

>
> Please not do this.
>
> Thanks.
>
>
> > +
> > if (metasize)
> > skb_metadata_set(skb, metasize);
> >
> > @@ -1635,6 +1638,9 @@ static struct sk_buff *receive_mergeable_xdp(struct net_device *dev,
> > head_skb = build_skb_from_xdp_buff(dev, vi, xdp, xdp_frags_truesz);
> > if (unlikely(!head_skb))
> > break;
> > + /* Store the original virtio header for subsequent use by the driver. */
> > + memcpy(skb_vnet_common_hdr(head_skb), &virtnet_xdp.hdr, vi->hdr_len);
> > +
> > return head_skb;
> >
> > case XDP_TX:
> > --
> > 2.40.1
> >
>