Re: [RFC PATCH 2/9] virtio_net: set up xdp for multi buffer packets

[Heng Qi <hengqi@xxxxxxxxxxxxxxxxx>]

在 2022/12/6 下午1:29, Jason Wang 写道:
> On Tue, Nov 22, 2022 at 3:44 PM Heng Qi <hengqi@xxxxxxxxxxxxxxxxx> wrote:
> > When the xdp program sets xdp.frags, which means it can process
> > multi-buffer packets, so we continue to open xdp support when
> > features such as GRO_HW are negotiated.
> >
> > Signed-off-by: Heng Qi <hengqi@xxxxxxxxxxxxxxxxx>
> > Reviewed-by: Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx>
> > ---
> >   drivers/net/virtio_net.c | 30 +++++++++++++++++-------------
> >   1 file changed, 17 insertions(+), 13 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index c5046d21b281..8f7d207d58d6 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -3080,14 +3080,21 @@ static int virtnet_xdp_set(struct net_device *dev, struct bpf_prog *prog,
> >          u16 xdp_qp = 0, curr_qp;
> >          int i, err;
> >
> > -       if (!virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS)
> > -           && (virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_TSO4) ||
> > -               virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_TSO6) ||
> > -               virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_ECN) ||
> > -               virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_UFO) ||
> > -               virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_CSUM))) {
> > -               NL_SET_ERR_MSG_MOD(extack, "Can't set XDP while host is implementing GRO_HW/CSUM, disable GRO_HW/CSUM first");
> > -               return -EOPNOTSUPP;
> > +       if (!virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS)) {
> > +               if (virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_CSUM)) {
> > +                       NL_SET_ERR_MSG_MOD(extack, "Can't set XDP without frags while guest is implementing GUEST_CSUM");
> > +                       return -EOPNOTSUPP;
> > +               }
> > +
> > +               if (prog && !prog->aux->xdp_has_frags) {
> > +                       if (virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_TSO4) ||
> > +                           virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_TSO6) ||
> > +                           virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_ECN) ||
> > +                           virtio_has_feature(vi->vdev, VIRTIO_NET_F_GUEST_UFO)) {
> > +                               NL_SET_ERR_MSG_MOD(extack, "Can't set XDP without frags while guest is implementing GUEST_GRO_HW");
> > +                               return -EOPNOTSUPP;
> > +                       }
> > +               }
> >          }
> >
> >          if (vi->mergeable_rx_bufs && !vi->any_header_sg) {
> > @@ -3095,8 +3102,8 @@ static int virtnet_xdp_set(struct net_device *dev, struct bpf_prog *prog,
> >                  return -EINVAL;
> >          }
> >
> > -       if (dev->mtu > max_sz) {
> > -               NL_SET_ERR_MSG_MOD(extack, "MTU too large to enable XDP");
> > +       if (prog && !prog->aux->xdp_has_frags && dev->mtu > max_sz) {
> > +               NL_SET_ERR_MSG_MOD(extack, "MTU too large to enable XDP without frags");
> >                  netdev_warn(dev, "XDP requires MTU less than %lu\n", max_sz);
> >                  return -EINVAL;
> >          }
> > @@ -3218,9 +3225,6 @@ static int virtnet_set_features(struct net_device *dev,
> >          int err;
> >
> >          if ((dev->features ^ features) & NETIF_F_GRO_HW) {
> > -               if (vi->xdp_enabled)
> > -                       return -EBUSY;
> This seems suspicious, GRO_HW could be re-enabled accidentally even if
> it was disabled when attaching an XDP program that is not capable of
> doing multi-buffer XDP?

Yes, we shouldn't drop this check, because GRO_HW is unfriendly to xdp 
programs without xdp.frags.

Thanks.

> Thanks
>
> > -
> >                  if (features & NETIF_F_GRO_HW)
> >                          offloads = vi->guest_offloads_capable;
> >                  else
> > --
> > 2.19.1.6.gb485710b