[xdp-hints] Re: [PATCH bpf-next v4 01/15] bpf: Document XDP RX metadata

["Toke Høiland-Jørgensen" <toke@redhat.com>]

Stanislav Fomichev <sdf@google.com> writes:

> On Wed, Dec 14, 2022 at 3:46 PM Toke Høiland-Jørgensen <toke@redhat.com> wrote:
>>
>> Stanislav Fomichev <sdf@google.com> writes:
>>
>> > On Wed, Dec 14, 2022 at 2:34 AM Toke Høiland-Jørgensen <toke@redhat.com> wrote:
>> >>
>> >> Stanislav Fomichev <sdf@google.com> writes:
>> >>
>> >> > On Tue, Dec 13, 2022 at 8:37 AM David Vernet <void@manifault.com> wrote:
>> >> >>
>> >> >> On Mon, Dec 12, 2022 at 06:35:51PM -0800, Stanislav Fomichev wrote:
>> >> >> > Document all current use-cases and assumptions.
>> >> >> >
>> >> >> > Cc: John Fastabend <john.fastabend@gmail.com>
>> >> >> > Cc: David Ahern <dsahern@gmail.com>
>> >> >> > Cc: Martin KaFai Lau <martin.lau@linux.dev>
>> >> >> > Cc: Jakub Kicinski <kuba@kernel.org>
>> >> >> > Cc: Willem de Bruijn <willemb@google.com>
>> >> >> > Cc: Jesper Dangaard Brouer <brouer@redhat.com>
>> >> >> > Cc: Anatoly Burakov <anatoly.burakov@intel.com>
>> >> >> > Cc: Alexander Lobakin <alexandr.lobakin@intel.com>
>> >> >> > Cc: Magnus Karlsson <magnus.karlsson@gmail.com>
>> >> >> > Cc: Maryam Tahhan <mtahhan@redhat.com>
>> >> >> > Cc: xdp-hints@xdp-project.net
>> >> >> > Cc: netdev@vger.kernel.org
>> >> >> > Signed-off-by: Stanislav Fomichev <sdf@google.com>
>> >> >> > ---
>> >> >> >  Documentation/bpf/xdp-rx-metadata.rst | 90 +++++++++++++++++++++++++++
>> >> >> >  1 file changed, 90 insertions(+)
>> >> >> >  create mode 100644 Documentation/bpf/xdp-rx-metadata.rst
>> >> >> >
>> >> >> > diff --git a/Documentation/bpf/xdp-rx-metadata.rst b/Documentation/bpf/xdp-rx-metadata.rst
>> >> >> > new file mode 100644
>> >> >> > index 000000000000..498eae718275
>> >> >> > --- /dev/null
>> >> >> > +++ b/Documentation/bpf/xdp-rx-metadata.rst
>> >> >>
>> >> >> I think you need to add this to Documentation/bpf/index.rst. Or even
>> >> >> better, maybe it's time to add an xdp/ subdirectory and put all docs
>> >> >> there? Don't want to block your patchset from bikeshedding on this
>> >> >> point, so for now it's fine to just put it in
>> >> >> Documentation/bpf/index.rst until we figure that out.
>> >> >
>> >> > Maybe let's put it under Documentation/networking/xdp-rx-metadata.rst
>> >> > and reference form Documentation/networking/index.rst? Since it's more
>> >> > relevant to networking than the core bpf?
>> >> >
>> >> >> > @@ -0,0 +1,90 @@
>> >> >> > +===============
>> >> >> > +XDP RX Metadata
>> >> >> > +===============
>> >> >> > +
>> >> >> > +XDP programs support creating and passing custom metadata via
>> >> >> > +``bpf_xdp_adjust_meta``. This metadata can be consumed by the following
>> >> >> > +entities:
>> >> >>
>> >> >> Can you add a couple of sentences to this intro section that explains
>> >> >> what metadata is at a high level?
>> >> >
>> >> > I'm gonna copy-paste here what I'm adding, feel free to reply back if
>> >> > still unclear. (so we don't have to wait another week to discuss the
>> >> > changes)
>> >> >
>> >> > XDP programs support creating and passing custom metadata via
>> >> > ``bpf_xdp_adjust_meta``. The metadata can contain some extra information
>> >> > about the packet: timestamps, hash, vlan and tunneling information, etc.
>> >> > This metadata can be consumed by the following entities:
>> >>
>> >> This is not really accurate, though? The metadata area itself can
>> >> contain whatever the XDP program wants it to, and I think you're
>> >> conflating the "old" usage for arbitrary storage with the driver-kfunc
>> >> metadata support.
>> >>
>> >> I think we should clear separate the two: the metadata area is just a
>> >> place to store data (and is not consumed by the stack, except that
>> >> TC-BPF programs can access it), and the driver kfuncs are just a general
>> >> way to get data out of the drivers (and has nothing to do with the
>> >> metadata area, you can just get the data into stack variables).
>> >>
>> >> While it would be good to have a documentation of the general metadata
>> >> area stuff somewhere, I don't think it necessarily have to be part of
>> >> this series, so maybe just stick to documenting the kfuncs?
>> >
>> > Maybe I can reword to something like below?
>> >
>> > The metadata can be used to store some extra information about the
>> > packet timestamps, hash, vlan and tunneling information, etc.
>> >
>> > This way we are not actually defining what it is, but hinting about
>> > how it's commonly used?
>>
>> Sent another reply to the original patch with some comments that are
>> hopefully a bit more constructive :)
>
> Thanks, everything makes sense, will incorporate. I'll also try to
> share the patches privately with you sometime tomorrow maybe; not
> super comfortable sending them out with a bunch of changes on top of
> your authorship without some kind of ack from you :-)

OK, sounds good. Tomorrow (Friday) is my last day before the holidays,
but happy to look things over before I sign off :)

-Toke