[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAJnrk1YB5SSetrKLMZz3HySdMxu4N50_rmbvOAtu-Vmv+tdcOQ@mail.gmail.com>
Date: Tue, 17 Jan 2023 09:20:56 -0800
From: Joanne Koong <joannelkoong@...il.com>
To: Tariq Toukan <ttoukan.linux@...il.com>
Cc: Alexei Starovoitov <alexei.starovoitov@...il.com>,
Kumar Kartikeya Dwivedi <memxor@...il.com>,
bpf <bpf@...r.kernel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Martin KaFai Lau <martin.lau@...nel.org>,
Andrii Nakryiko <andrii@...nel.org>,
Alexei Starovoitov <ast@...nel.org>,
Network Development <netdev@...r.kernel.org>,
Kernel Team <Kernel-team@...com>,
Jakub Kicinski <kuba@...nel.org>,
Toke Høiland-Jørgensen <toke@...hat.com>,
Gal Pressman <gal@...dia.com>, Tariq Toukan <tariqt@...dia.com>
Subject: Re: [PATCH bpf-next v7 0/3] Add skb + xdp dynptrs
On Sun, Jan 15, 2023 at 4:41 AM Tariq Toukan <ttoukan.linux@...il.com> wrote:
>
> Hi,
>
> I'm reviving this thread, following the discussion here:
> https://lore.kernel.org/bpf/87fscjakba.fsf@toke.dk/
>
> On 21/10/2022 4:19, Alexei Starovoitov wrote:
> > On Thu, Oct 20, 2022 at 6:15 PM Joanne Koong <joannelkoong@...il.com> wrote:
> >>
> >> This patchset is the 2nd in the dynptr series. The 1st can be found here [0].
> >>
> >> This patchset adds skb and xdp type dynptrs, which have two main benefits for
> >> packet parsing:
> >> * allowing operations on sizes that are not statically known at
> >> compile-time (eg variable-sized accesses).
> >> * more ergonomic and less brittle iteration through data (eg does not need
> >> manual if checking for being within bounds of data_end)
> >>
> >> When comparing the differences in runtime for packet parsing without dynptrs
> >> vs. with dynptrs for the more simple cases, there is no noticeable difference.
> >> For the more complex cases where lengths are non-statically known at compile
> >> time, there can be a significant speed-up when using dynptrs (eg a 2x speed up
> >> for cls redirection). Patch 3 contains more details as well as examples of how
> >> to use skb and xdp dynptrs.
> >
> > Before proceeding with this patchset I think we gotta resolve the
> > issues with dynptr-s that Kumar found.
>
> Just to make sure I'm following: The issues that are discussed here?
> https://lore.kernel.org/all/CAP01T74icBDXOM=ckxYVPK90QLcU4n4VRBjON_+v74dQwJfZvw@mail.gmail.com/
>
> What is the current status of dynptrs?
> Any updates since October?
> Do we have any agreement or a plan for this?
Hi Tariq,
The current status of dynptrs is blocked on two things: 1) the fixes
by Kumar in [1] landing upstream and 2) a bigger question of whether
bpf development should proceed with kfuncs or helpers (thread in [2]).
We had a bpf office hour session last Thursday about whether helpers
should be frozen (#2), but did not come to a conclusion in the time
allotted. As a follow-up to the office hour, Toke and David Vernet, I
believe, are working on a document outlining how kfunc stability will
work. After this document is sent out, I think we will probably have
another office hour session or discussion upstream about which
direction to take.
[1] https://lore.kernel.org/bpf/20230101083403.332783-1-memxor@gmail.com/
[2] https://lore.kernel.org/bpf/20221208015434.ervz6q5j7bb4jt4a@macbook-pro-6.dhcp.thefacebook.com/t/#u
>
> Regards,
> Tariq
Powered by blists - more mailing lists