netdev - Re: Mass udp flow reboot linux with RealTek RTL-8169 Gigabit

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] [day] [month] [year] [list]

Date:	Thu, 10 Mar 2011 13:08:56 +0100
From:	Francois Romieu <romieu@...zoreil.com>
To:	Seblu <seblu@...lu.net>
Cc:	Eric Dumazet <eric.dumazet@...il.com>,
	lkml <linux-kernel@...r.kernel.org>,
	netdev <netdev@...r.kernel.org>, Ivan Vecera <ivecera@...hat.com>
Subject: Re: Mass udp flow reboot linux with RealTek RTL-8169 Gigabit

Seblu <seblu@...lu.net> :
[...]
> I catched the following trace during my previous torture session.
> Maybe it can help.

It's the usual r8169 TX timeout watchdog.

[...]
> > Can you apply the two attached patches on top of the previous ones and
> > give it a try ? The debug should not be too verbose if things are stationary
> > enough.
> 2.6.38-rc7 with your 2 previous patch change the game. No reboot. No
> strange message in dmesg.

?

"strange message" as :
[ ] netdev watchdog messages
[ ] 0001 0001 0001 0001 (or similar) message
[ ] net_ratelimit message

> But some sent packets are lost from some host. Example:
[...]
> This is maybe normal under stress, card discard packet after all.

It seems so. 0.08% packet loss. 10 ~ 20kpps (right ?). Sample at 0.1 Hz (ping).

[...]
> I've a serial cable and a second computer, but my first computer
> doesn't have a com port. Is it then possible?

Hardly. Forget it for now.

> Do you need more test?

1. 2.6.38-rc7 without the patches
2. 2.6.38-rc7 with the r8169.c driver of 2.6.38-rc5, without the patches
3. current setup + pktgen. Lower the packet size as long as it increases
   the sender's pps.

I do not understand why the bug would be gone if it was in the r8169
proper.

-- 
Ueimor
--
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