[xdp-hints] Re: [PATCH bpf-next v7 17/18] selftests/bpf: Use AF_INET for TX in xdp_metadata

[Larysa Zaremba <larysa.zaremba@intel.com>]

On Mon, Nov 20, 2023 at 01:15:41PM -0800, Stanislav Fomichev via xdp-hints wrote:
> Date: Mon, 20 Nov 2023 13:15:41 -0800
> From: Stanislav Fomichev <sdf@google.com>
> To: Larysa Zaremba <larysa.zaremba@intel.com>
> CC: bpf@vger.kernel.org, ast@kernel.org, daniel@iogearbox.net,
>  andrii@kernel.org, martin.lau@linux.dev, song@kernel.org, yhs@fb.com,
>  john.fastabend@gmail.com, kpsingh@kernel.org, haoluo@google.com,
>  jolsa@kernel.org, David Ahern <dsahern@gmail.com>, Jakub Kicinski
>  <kuba@kernel.org>, Willem de Bruijn <willemb@google.com>, Jesper Dangaard
>  Brouer <hawk@kernel.org>, Anatoly Burakov <anatoly.burakov@intel.com>,
>  Alexander Lobakin <alexandr.lobakin@intel.com>, Magnus Karlsson
>  <magnus.karlsson@gmail.com>, Maryam Tahhan <mtahhan@redhat.com>,
>  xdp-hints@xdp-project.net, netdev@vger.kernel.org, Willem de Bruijn
>  <willemdebruijn.kernel@gmail.com>, Alexei Starovoitov
>  <alexei.starovoitov@gmail.com>, Tariq Toukan <tariqt@mellanox.com>, Saeed
>  Mahameed <saeedm@mellanox.com>, Maciej Fijalkowski
>  <maciej.fijalkowski@intel.com>
> Subject: [xdp-hints] Re: [PATCH bpf-next v7 17/18] selftests/bpf: Use
>  AF_INET for TX in xdp_metadata
> 
> On 11/15, Larysa Zaremba wrote:
> > The easiest way to simulate stripped VLAN tag in veth is to send a packet
> > from VLAN interface, attached to veth. Unfortunately, this approach is
> > incompatible with AF_XDP on TX side, because VLAN interfaces do not have
> > such feature.
> > 
> > Replace AF_XDP packet generation with sending the same datagram via
> > AF_INET socket.
> > 
> > This does not change the packet contents or hints values with one notable
> > exception: rx_hash_type, which previously was expected to be 0, now is
> > expected be at least XDP_RSS_TYPE_L4.
> 
> Btw, I've been thinking a bit about how we can make this test work for both
> your VLANs and my upcoming af_xdp tx side. And seems like the best
> way, probably, is to have two tx paths exercised: veth and af_xdp.
> For veth, we'll verify everything+vlans, for af_xdp we'll verify
> everything except the vlans.
> 
> Originally I was assuming that I'll switch this part back to af_xdp, but
> I don't think having tx vlan offload makes sense (because af_xdp
> userspace can just prepare the correct header from the start).
> 
> So if you're doing a respin, maybe see if we can keep af_xdp tx part
> but make it skip the vlans verification?
> 
> generate_packet_af_xdp();
> verify_xsk_metadata(/*verify_vlans=*/false);
> geenrate_packet_veth();
> verify_xsk_metadata(/*verify_vlans=*/true);
> 
> ?

Sounds good to me, will do so.