Re: [gentoo-user] Network card failing under load

[Kerin Millar]

Datty wrote:
> Hi all,
>
> I have an issue with a remote server that keeps losing access to the
> network. I've managed to replicate the issue by pulling a large file
> from the server at full speed (100mbit). It will cut out during the file
> copy and the log will show the following:
>
> Oct 17 17:47:26 eric kernel: WARNING: at net/sched/sch_generic.c:256
> dev_watchdog+0x251/0x260()
> Oct 17 17:47:26 eric kernel: Hardware name: KS
> Oct 17 17:47:26 eric kernel: NETDEV WATCHDOG: eth0 (sis900): transmit
> queue 0 timed out
> Oct 17 17:47:26 eric kernel: Modules linked in: tun
> Oct 17 17:47:26 eric kernel: Pid: 0, comm: swapper/0 Not tainted
> 3.4.9-gentoo-gnu-dat #7
> Oct 17 17:47:26 eric kernel: Call Trace:
> Oct 17 17:47:26 eric kernel: <IRQ>  [<ffffffff810370db>] ?
> warn_slowpath_common+0x7b/0xc0
> Oct 17 17:47:26 eric kernel: [<ffffffff810371d5>] ?
> warn_slowpath_fmt+0x45/0x50
> Oct 17 17:47:26 eric kernel: [<ffffffff81217f81>] ? timerqueue_del+0x31/0x90
> Oct 17 17:47:26 eric kernel: [<ffffffff8131a791>] ? dev_watchdog+0x251/0x260
> Oct 17 17:47:26 eric kernel: [<ffffffff81042936>] ?
> run_timer_softirq+0x126/0x250
> Oct 17 17:47:26 eric kernel: [<ffffffff8131a540>] ? qdisc_reset+0x40/0x40
> Oct 17 17:47:26 eric kernel: [<ffffffff810573ad>] ?
> hrtimer_run_queues+0x5d/0x150
> Oct 17 17:47:26 eric kernel: [<ffffffff8103d178>] ? __do_softirq+0x98/0x120
> Oct 17 17:47:26 eric kernel: [<ffffffff8141e3fc>] ? call_softirq+0x1c/0x30
> Oct 17 17:47:26 eric kernel: [<ffffffff8100d065>] ? do_softirq+0x65/0xa0
> Oct 17 17:47:26 eric kernel: [<ffffffff8103d47e>] ? irq_exit+0x8e/0xb0
> Oct 17 17:47:26 eric kernel: [<ffffffff81023ce8>] ?
> smp_apic_timer_interrupt+0x68/0xa0
> Oct 17 17:47:26 eric kernel: [<ffffffff8141dc8a>] ?
> apic_timer_interrupt+0x6a/0x70
> Oct 17 17:47:26 eric kernel: <EOI>  [<ffffffff81060435>] ?
> sched_clock_local+0x15/0x80
> Oct 17 17:47:26 eric kernel: [<ffffffff81012c78>] ? mwait_idle+0x58/0x80
> Oct 17 17:47:26 eric kernel: [<ffffffff810136b6>] ? cpu_idle+0x56/0xa0
> Oct 17 17:47:26 eric kernel: [<ffffffff815b4ae9>] ? start_kernel+0x2c6/0x2d1
> Oct 17 17:47:26 eric kernel: [<ffffffff815b459a>] ?
> repair_env_string+0x5c/0x5c
> Oct 17 17:47:26 eric kernel: [<ffffffff815b441e>] ?
> x86_64_start_kernel+0x104/0x113
> Oct 17 17:47:26 eric kernel: ---[ end trace 50e815f551bee4ac ]---
> Oct 17 17:47:26 eric kernel: eth0: Transmit timeout, status 00000005
> 00000000
> Oct 17 17:47:34 eric kernel: eth0: Transmit timeout, status 00000005
> 00000000
> Oct 17 17:47:42 eric kernel: eth0: Transmit timeout, status 00000005
> 00000000
> Oct 17 17:47:50 eric kernel: eth0: Transmit timeout, status 00000005
> 00000000
> Oct 17 17:47:58 eric kernel: eth0: Transmit timeout, status 00000005
> 00000000
>
> As it is remote I have to hard reboot it at this point to get anything
> more out of it, something that I'm not too keen on doing too often.
>
>  From that I can work out there seems to be an issue with the sis900
> card but where to go from here I'm not sure on.
>
> Any help would be greatly appreciated

I would suggest that you upgrade to 3.4.14 then, if it still occurs, 
re-open the following bug (which appears to be related):

https://bugzilla.kernel.org/show_bug.cgi?id=20092

Elsewhere, this was reported as a regression between Ubuntu 9.10 and 
10.04 which might help to pinpoint where things went wrong in the 
mainline kernel. Not that Ubuntu's lame bug resolution process has shed 
any further light on the matter. See also:

http://www.alsvartr.de/?p=1104

--Kerin