Re: [PATCH net-next 00/13] selftests: net: migrate AF_XDP test suite over to net

[Maciej Fijalkowski <maciej.fijalkowski@intel.com>]

On Tue, Oct 06, 2026 at 03:04:08PM -0700, Stanislav Fomichev wrote:
> On 10/06, Maciej Fijalkowski wrote:
> > On Mon, Oct 05, 2026 at 11:06:52AM -0700, Stanislav Fomichev wrote:
> > > On 10/01, Maciej Fijalkowski wrote:

[...]

> > > > 
> > > > 
> > > > Known issues:
> > > > - Every case pays for process start-up, XDP attach and detach and, on
> > > >   hardware, its remote commands. We used to configure resources once
> > > >   and then execute the whole test suite; it doesn't seem to be
> > > >   CI-friendly and it is preferred to have each case's resource
> > > >   management separated; that on the other hand increases the
> > > >   execution time of the whole test suite.
> > > 
> > > [..]
> > > 
> > > > - The XDP programs redirect every packet, so the link under test must
> > > >   carry no other traffic. SSH and the control channel go to the
> > > >   REMOTE_ARGS host, which has to be reached over another link.
> > > 
> > > Will this work on NIPA?
> > 
> > Yeah good that you're bringing this up, I see NIPA has a e810 setup within
> > same machine which is not what i tested on my side. This means the
> > assumption/requirement of having isolated link under test has to be lifted
> > as e810 cards on NIPA will carry management traffic via same link. I'll
> > add XDP prog logic as you point out, thanks.
> > 
> > I also hit the ice bug after connecting interfaces within single machine
> > which was hiding from me throughout whole local testing, during RSS update
> > where we only want to touch indirection table, symmetric-xor hashing was
> > being turned on which caused later rss operations to fail; I'll post a fix
> > to iwl-net and include some heads-up to v2.
> > 
> > > 
> > > I took a quick pass, nothing pops us for me. The only thing I'm not sure
> > > is the bpftool dependency (whether we need to build it or there is
> > > something on the system).
> > 
> > We need it due to skeleton usage and other change was also utilizing it so
> > I thought it would be acceptable.
> 
> No, no, I'm not questioning the need, just not sure whether we need to build
> it in the selftest makefiles (like bpf selftests do) or it's ok to use
> the system one (if there is one).

We could build it within makefiles as a bootstrap. How about that? I am
not aware how much of an effort would be to provide system installs of
bpftool on CI runners, so maybe this smaller build would be acceptable?