Re: intermittent problem with skge driver/hardware

Previous thread: [PATCH RESEND] pata_mpc52xx: driver needs BMDMA by Wolfram Sang on Wednesday, December 22, 2010 - 8:50 am. (2 messages)

Next thread: [PATCH] KEYS: Don't call up_write() if __key_link_begin() returns an error by David Howells on Wednesday, December 22, 2010 - 9:24 am. (1 message)
From: Thomas Fjellstrom
Date: Wednesday, December 22, 2010 - 8:56 am

I've been getting a strange issue, where traffic over one of the built in skge
nics in my server will just die. No traffic can make it over, but it'll claim
the device is up:

dmesg:
[85007.362349] skge 0000:05:06.0: PCI error cmd=0x7 status=0x22b0
[85007.362735] skge 0000:05:06.0: unable to clear error (so ignoring them)
[85076.960083] ------------[ cut here ]------------
[85076.960963] WARNING: at net/sched/sch_generic.c:258 dev_watchdog+0xfc/0x19b()
[85076.961799] Hardware name: GA-MA790FXT-UD5P
[85076.962654] NETDEV WATCHDOG: eth2 (skge): transmit queue 0 timed out
[85076.963520] Modules linked in: tun ip6table_filter ip6_tables iptable_filter ip_tables x_tables powernow_k8 mperf cpufreq_conservative cpufreq_stats cpufreq_userspace cpufreq_$
[85076.970411] Pid: 0, comm: kworker/0:1 Not tainted 2.6.36.1+ #5
[85076.971401] Call Trace:
[85076.971406]  <IRQ>  [<ffffffff8103687e>] ? warn_slowpath_common+0x78/0x8c
[85076.971426]  [<ffffffff81036931>] ? warn_slowpath_fmt+0x45/0x4a
[85076.971434]  [<ffffffff8122b866>] ? netif_tx_lock+0x3d/0x64
[85076.971442]  [<ffffffff8122b989>] ? dev_watchdog+0xfc/0x19b
[85076.971450]  [<ffffffff8104097a>] ? cascade+0x60/0x7a
[85076.971458]  [<ffffffff81026bda>] ? check_preempt_curr+0x1a/0x31
[85076.971466]  [<ffffffff8104232e>] ? run_timer_softirq+0x1c2/0x284
[85076.971475]  [<ffffffff8109cece>] ? perf_event_task_tick+0x6a/0x185
[85076.971483]  [<ffffffff8122b88d>] ? dev_watchdog+0x0/0x19b
[85076.971493]  [<ffffffff8103bec4>] ? __do_softirq+0xde/0x19e
[85076.971501]  [<ffffffff81059f8a>] ? tick_dev_program_event+0x33/0xf0
[85076.971510]  [<ffffffff8100384c>] ? call_softirq+0x1c/0x28
[85076.971517]  [<ffffffff81004c01>] ? do_softirq+0x31/0x63
[85076.971525]  [<ffffffff8103bd4b>] ? irq_exit+0x36/0x79
[85076.971534]  [<ffffffff81017b4b>] ? smp_apic_timer_interrupt+0x87/0x95
[85076.971541]  [<ffffffff81003313>] ? apic_timer_interrupt+0x13/0x20
[85076.971545]  <EOI>  [<ffffffff8100942c>] ? default_idle+0x36/0x4c
[85076.971557]  [<ffffffff8100940c>] ? ...
From: Thomas Fjellstrom
Date: Thursday, December 23, 2010 - 8:36 am

I should make a quick correction, the skge adapter is not built in, rather its 
a plain old D-Link GbE pci card. If it happens again, I'll try one of my 
spares (bought several D-Link cards prior to upgrading all my hardware, so 
most of them are redundant now, as the boards all have GbE built in).

If anyone else knows what might be the cause, please let me know.

-- 
Thomas Fjellstrom
thomas@fjellstrom.ca
--

Previous thread: [PATCH RESEND] pata_mpc52xx: driver needs BMDMA by Wolfram Sang on Wednesday, December 22, 2010 - 8:50 am. (2 messages)

Next thread: [PATCH] KEYS: Don't call up_write() if __key_link_begin() returns an error by David Howells on Wednesday, December 22, 2010 - 9:24 am. (1 message)