netdev - Re: GRO + splice panics in 3.7.0-rc5

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]

Date:	Fri, 16 Nov 2012 07:39:05 +0100
From:	Willy Tarreau <w@....eu>
To:	Eric Dumazet <eric.dumazet@...il.com>
Cc:	netdev@...r.kernel.org
Subject: Re: GRO + splice panics in 3.7.0-rc5

Hi Eric,

On Thu, Nov 15, 2012 at 03:49:04PM -0800, Eric Dumazet wrote:
> On Thu, 2012-11-15 at 23:28 +0100, Willy Tarreau wrote:
> > Hello,
> > 
> > I was just about to make a quick comparison between LRO and GRO in
> > 3.7.0-rc5 to see if LRO still had the big advantage I've always observed,
> > but I failed the test because as soon as I enable LRO + splice, the kernel
> > panics and reboots.
> > 
> > I could not yet manage to catch the panic output, I could just reliably
> > reproduce it, it crashes instantly.
> > 
> > All I can say at the moment is the following :
> >   - test consist in forwarding HTTP traffic between two NICs via haproxy
> >   - driver used was myri10ge
> >   - LRO + recv+send : OK
> >   - LRO + splice : OK
> >   - GRO + recv+send : OK
> >   - GRO + splice : panic
> >   - no such problem was observed in 3.6.6 so I think this is a recent
> >     regression.
> > 
> > I'll go back digging for more information, but as I'm used to often see
> > Eric suggest the right candidates for reverting, I wanted to report the
> > issue here in case there are easy ones to try first.
> 
> Hi Willy
> 
> Nothing particular comes to mind, there were a lot of recent changes
> that could trigger this kind of bug.

OK, no problem.

> A stack trace would be useful of course ;)

I'll arrange to get one, then to bisect. I'll also try to reproduce with
the onboard NIC (sky2) to see if this is something related to the driver
itself. I can't promise anything before the end of the week-end, though.

Thanks !
Willy

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html