Re: [dpdk-dev] [PATCH v2] net/virtio: handle Tx checksums correctly for tunnel packets

[Olivier Matz <olivier.matz@6wind.com>]

On Mon, Oct 18, 2021 at 10:04:46AM +0300, Andrew Rybchenko wrote:
> On 10/15/21 11:32 AM, Olivier Matz wrote:
> > On Thu, Oct 14, 2021 at 07:12:29AM +0000, Xia, Chenbo wrote:
> >>> -----Original Message-----
> >>> From: Ivan Malov <ivan.malov@oktetlabs.ru>
> >>> Sent: Friday, September 17, 2021 2:50 AM
> >>> To: dev@dpdk.org
> >>> Cc: Maxime Coquelin <maxime.coquelin@redhat.com>; stable@dpdk.org; Andrew
> >>> Rybchenko <andrew.rybchenko@oktetlabs.ru>; Xia, Chenbo <chenbo.xia@intel.com>;
> >>> Yuanhan Liu <yuanhan.liu@linux.intel.com>; Olivier Matz
> >>> <olivier.matz@6wind.com>
> >>> Subject: [PATCH v2] net/virtio: handle Tx checksums correctly for tunnel
> >>> packets
> >>>
> >>> Tx prepare method calls rte_net_intel_cksum_prepare(), which
> >>> handles tunnel packets correctly, but Tx burst path does not
> >>> take tunnel presence into account when computing the offsets.
> >>>
> >>> Fixes: 58169a9c8153 ("net/virtio: support Tx checksum offload")
> >>> Cc: stable@dpdk.org
> >>>
> >>> Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
> >>> Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
> >>> ---
> >>>  drivers/net/virtio/virtqueue.h | 9 ++++++---
> >>>  1 file changed, 6 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
> >>> index 03957b2bd0..b83ff32efb 100644
> >>> --- a/drivers/net/virtio/virtqueue.h
> >>> +++ b/drivers/net/virtio/virtqueue.h
> >>> @@ -620,19 +620,21 @@ static inline void
> >>>  virtqueue_xmit_offload(struct virtio_net_hdr *hdr, struct rte_mbuf *cookie)
> >>>  {
> >>>  	uint64_t csum_l4 = cookie->ol_flags & PKT_TX_L4_MASK;
> >>> +	uint16_t o_l23_len = (cookie->ol_flags & PKT_TX_TUNNEL_MASK) ?
> >>> +			     cookie->outer_l2_len + cookie->outer_l3_len : 0;
> >>>
> >>>  	if (cookie->ol_flags & PKT_TX_TCP_SEG)
> >>>  		csum_l4 |= PKT_TX_TCP_CKSUM;
> >>>
> >>>  	switch (csum_l4) {
> >>>  	case PKT_TX_UDP_CKSUM:
> >>> -		hdr->csum_start = cookie->l2_len + cookie->l3_len;
> >>> +		hdr->csum_start = o_l23_len + cookie->l2_len + cookie->l3_len;
> >>>  		hdr->csum_offset = offsetof(struct rte_udp_hdr, dgram_cksum);
> >>>  		hdr->flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
> >>>  		break;
> >>>
> >>>  	case PKT_TX_TCP_CKSUM:
> >>> -		hdr->csum_start = cookie->l2_len + cookie->l3_len;
> >>> +		hdr->csum_start = o_l23_len + cookie->l2_len + cookie->l3_len;
> >>>  		hdr->csum_offset = offsetof(struct rte_tcp_hdr, cksum);
> >>>  		hdr->flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
> >>>  		break;
> >>> @@ -650,7 +652,8 @@ virtqueue_xmit_offload(struct virtio_net_hdr *hdr, struct
> >>> rte_mbuf *cookie)
> >>>  			VIRTIO_NET_HDR_GSO_TCPV6 :
> >>>  			VIRTIO_NET_HDR_GSO_TCPV4;
> >>>  		hdr->gso_size = cookie->tso_segsz;
> >>> -		hdr->hdr_len = cookie->l2_len + cookie->l3_len + cookie->l4_len;
> >>> +		hdr->hdr_len = o_l23_len + cookie->l2_len + cookie->l3_len +
> >>> +			       cookie->l4_len;
> >>>  	} else {
> >>>  		ASSIGN_UNLESS_EQUAL(hdr->gso_type, 0);
> >>>  		ASSIGN_UNLESS_EQUAL(hdr->gso_size, 0);
> >>> --
> >>> 2.20.1
> >>
> >> Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
> >>
> > 
> > I have one comment to mention that from application perspective, it has
> > to take care that the driver does not support outer tunnel offload (this
> > matches the advertised capabilities). For instance, in case of a vxlan
> > tunnel, if the outer checksum needs to be calculated, it has to be done
> > by the application. In short, the application can ask to offload the
> > inner part if no offload is required on the outer part.
> 
> Yes. It is really complicated.
> 
> Let's name below case (A).
> 
> > Also, since grep "PKT_TX_TUNNEL" in driver/net/ixgbe gives nothing, it
> > seems the ixgbe driver does not support the same offload request than
> > described in this patch:
> >   (m->ol_flags & PKT_TX_TUNNEL_MASK) == PKT_TX_TUNNEL_XXXXX
> >   m->outer_l2_len = outer l2 length
> >   m->outer_l3_len = outer l3 length
> >   m->l2_len = outer l4 length + tunnel len + inner l2 len
> >   m->l3_len = inner l3 len
> >   m->l4_len = inner l4 len
> 
> True.
> 
> Let's name below case (B).
> 
> > An alternative for doing the same (that would work with ixgbe and
> > current virtio) is to give:
> >   (m->ol_flags & PKT_TX_TUNNEL_MASK) == 0
> >   m->l2_len = outer lengths + tunnel len + inner l2 len
> >   m->l3_len = inner l3 len
> >   m->l4_len = inner l4 len
> 
> Possible, but I think it is a bit less friendly.
> If application supports both cases with and without
> outer IPv4 checksum offload, it will require more extra
> steps from the application to remove tunnel mask and
> collapse outer lengths into l2_len (in addition outer
> IPv4 checksum calculation in SW).
> 
> > I think a capability may be missing to differentiate which drivers
> > support which mode.
> 
> Yes, something like TX_INNER_CKSUM.
> Some drivers already support it, some do not.
>  > Or, all drivers could be fixed to support both modes
> > (and this would make this patch valid).
> 
> I disagree with the point that something is required to
> make the patch valid. Before the patch the driver supports
> just (B). After the patch the driver supports both (A) and (B).
> It makes the driver to support more cases.

Yes, I misspoke, I wanted to say that today it is difficult for an
application to use this new mode because it does not know which driver
supports it. Fixing all drivers would solve the issue without a
capability.

A drivers/hw supporting one mode can easily support the other one (this
patch demonstrates it), so maybe it's not worth having a capability.

But yes, this is not a good reason to reject the patch.

So:
Reviewed-by: Olivier Matz <olivier.matz@6wind.com>


> Please, note that rte_net_intel_cksum_prepare() supports
> both (A) and (B) as well.