[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAN+4W8iSA0Y8iEvYg79=CTNvwkQB5qs_F3vjE7vep-eHR01oJw@mail.gmail.com>
Date: Wed, 21 Jun 2023 09:01:15 +0100
From: Lorenz Bauer <lmb@...valent.com>
To: Kuniyuki Iwashima <kuniyu@...zon.com>
Cc: andrii@...nel.org, ast@...nel.org, bpf@...r.kernel.org,
daniel@...earbox.net, davem@...emloft.net, dsahern@...nel.org,
edumazet@...gle.com, haoluo@...gle.com, hemanthmalla@...il.com,
joe@...d.net.nz, john.fastabend@...il.com, jolsa@...nel.org,
kpsingh@...nel.org, kuba@...nel.org, linux-kernel@...r.kernel.org,
linux-kselftest@...r.kernel.org, martin.lau@...ux.dev,
mykolal@...com, netdev@...r.kernel.org, pabeni@...hat.com,
sdf@...gle.com, shuah@...nel.org, song@...nel.org,
willemdebruijn.kernel@...il.com, yhs@...com
Subject: Re: [PATCH bpf-next v2 3/6] net: remove duplicate reuseport_lookup functions
On Tue, Jun 20, 2023 at 7:31 PM Kuniyuki Iwashima <kuniyu@...zon.com> wrote:
>
> Good point. This is based on an assumption that all SO_REUSEPORT
> sockets have the same score, which is wrong for two corner cases
> if reuseport_has_conns() == true :
>
> 1) SO_INCOMING_CPU is set
> -> selected sk might have +1 score
>
> 2) BPF prog returns ESTABLISHED and/or SO_INCOMING_CPU sk
> -> selected sk will have more than 8
>
> Using the old score could trigger more lookups depending on the
> order that sockets are created.
So the result will still be correct, but it's less performant? Happy
to fix a perf regression, but if the result is incorrect this might
need a separate fix?
Best
Lorenz
Powered by blists - more mailing lists