netdev - Re: [PATCH] skbuff: Fix a race between coalescing and releasing SKBs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAKhg4tKT=LqohZi1RHhY1_dJjFbO6WvwZJRtUDM5rrZ7Xuqskw@mail.gmail.com>
Date:   Fri, 7 Apr 2023 08:45:53 +0800
From:   Liang Chen <liangchen.linux@...il.com>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     Ilias Apalodimas <ilias.apalodimas@...aro.org>,
        Eric Dumazet <edumazet@...gle.com>,
        Alexander Duyck <alexander.duyck@...il.com>, hawk@...nel.org,
        davem@...emloft.net, pabeni@...hat.com, netdev@...r.kernel.org
Subject: Re: [PATCH] skbuff: Fix a race between coalescing and releasing SKBs

On Thu, Apr 6, 2023 at 10:59 PM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Thu, 6 Apr 2023 19:54:23 +0800 Liang Chen wrote:
> > > > Same feeling on my side.
> > > > I prefer not trying to merge mixed pp_recycle skbs "just because we
> > > > could" at the expense
> > > > of adding more code in a fast path.
> > >
> > > +1 here.  The intention of recycling was to affect the normal path as
> > > less as possible.  On top of that, we've some amount of race
> > > conditions over the years, trying to squeeze more performance with
> > > similar tricks.  I'd much rather be safe here, since recycling by
> > > itself is a great performance boost
> >
> > Sorry, I didn't check my inbox before sending out the v2 patch.
>
> I can discard v2 from patchwork, let's continue the conversation
> here.
>
> > Yeah, It is a bit complicated as we expected. The patch is sent out.
> > Please take a look to see if it is the way to go, or We should stay
> > with the current patch for simplicity reasons. Thanks!
>
> Sounds like you know what Eric and Ilias agreed with, I'm a bit
> confused.. are we basically going back to v1? (hopefully with coding
> style fixed)

Sorry for making the confusion. I just felt the v2 patch complicated
the code a bit.  I am not quite sure if the performance gain justifies
the added complexity, and would really appreciate it if you could make
the decision on which patch to pick up. Thanks a lot!