Re: [PATCH net-next 2/4] virtio-net: reprobe csum related fields for skb passed by XDP

[Heng Qi <hengqi@linux.alibaba.com>]

On Mon, Jun 19, 2023 at 07:27:26AM -0400, Michael S. Tsirkin wrote:
> On Mon, Jun 19, 2023 at 06:57:36PM +0800, Heng Qi wrote:
> > Currently, the VIRTIO_NET_F_GUEST_CSUM (corresponds to NETIF_F_RXCSUM
> > for netdev) feature of the virtio-net driver conflicts with the loading
> > of XDP, which is caused by the problem described in [1][2], that is,
> > XDP may cause errors in partial csumed-related fields which can lead
> > to packet dropping.
> > 
> > In addition, when communicating between vm and vm on the same host, the
> > receiving side vm will receive packets marked as
> > VIRTIO_NET_HDR_F_NEEDS_CSUM, but after these packets are processed by
> > XDP, the VIRTIO_NET_HDR_F_NEEDS_CSUM and skb CHECKSUM_PARTIAL flags will
> > be cleared, causing the packet dropping.
> > 
> > This patch introduces a helper function, which will try to solve the
> > above problems in the subsequent patch.
> > 
> > [1] commit 18ba58e1c234 ("virtio-net: fail XDP set if guest csum is negotiated")
> > [2] commit e59ff2c49ae1 ("virtio-net: disable guest csum during XDP set")
> > 
> > Signed-off-by: Heng Qi <hengqi@linux.alibaba.com>
> > Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> 
> 
> squash this and patch 1 into where the helpers are used.
> 
> in particular so we don't get warnings with bisect.

Ok. Will modify.

Thanks.

> 
> > ---
> >  drivers/net/virtio_net.c | 38 ++++++++++++++++++++++++++++++++++++++
> >  1 file changed, 38 insertions(+)
> > 
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 36cae78f6311..07b4801d689c 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -1663,6 +1663,44 @@ static int virtnet_flow_dissect_udp_tcp(struct virtnet_info *vi, struct sk_buff
> >  	return 0;
> >  }
> >  
> > +static int virtnet_set_csum_after_xdp(struct virtnet_info *vi,
> > +				      struct sk_buff *skb,
> > +				      __u8 flags)
> > +{
> > +	int err;
> > +
> > +	/* When XDP program is loaded, for example, the vm-vm scenario
> > +	 * on the same host, packets marked as VIRTIO_NET_HDR_F_NEEDS_CSUM
> > +	 * will travel. Although these packets are safe from the point of
> > +	 * view of the vm, to avoid modification by XDP and successful
> > +	 * forwarding in the upper layer, we re-probe the necessary checksum
> > +	 * related information: skb->csum_{start, offset}, pseudo-header csum.
> > +	 *
> > +	 * This benefits us:
> > +	 * 1. XDP can be loaded when there's _F_GUEST_CSUM.
> > +	 * 2. The device verifies the checksum of packets , especially
> > +	 *    benefiting for large packets.
> > +	 * 3. In the same-host vm-vm scenario, packets marked as
> > +	 *    VIRTIO_NET_HDR_F_NEEDS_CSUM are no longer dropped after being
> > +	 *    processed by XDP.
> > +	 */
> > +	if (flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
> > +		err = virtnet_flow_dissect_udp_tcp(vi, skb);
> > +		if (err < 0)
> > +			return -EINVAL;
> > +
> > +		skb->ip_summed = CHECKSUM_PARTIAL;
> > +	} else if (flags && VIRTIO_NET_HDR_F_DATA_VALID) {
> > +		/* We want to benefit from this: XDP guarantees that packets marked
> > +		 * as VIRTIO_NET_HDR_F_DATA_VALID still have correct csum after they
> > +		 * are processed.
> > +		 */
> > +		skb->ip_summed = CHECKSUM_UNNECESSARY;
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> >  static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> >  			void *buf, unsigned int len, void **ctx,
> >  			unsigned int *xdp_xmit,
> > -- 
> > 2.19.1.6.gb485710b