[BUG] kernel panic during shutdown in run_timer_softirq()

[BUG] kernel panic during shutdown in run_timer_softirq()

[Knut Petersen]

[-- Attachment #1: Type: text/plain, Size: 1534 bytes --]

Hi everybody!

Since about July I observe occasional kernel panics  happening only during system shutdown on two systems.

Hardware: mobos: both AOpen i915GMm-hfs mobos, cpus: Pentium-M Dothan / Banias, mem: 2GB
Although the stack traces differ, there is one thing in common: CS:EIP is in run_timer_softirq().

A first report was posted to lkml on 2013-07-10: http://www.gossamer-threads.com/lists/linux/kernel/1744892?#1744892
jpg of the stack trace: http://www.gossamer-threads.com/lists/engine?do=post_attachment;postatt_id=57017;list=linux

There were several reports of kernel panics with run_timer_softirq() in the call_trace.

@Grant: You reported a kernel panic during shutdown for kernel 3.7.5, but the jpg is not accessible.
Could you please verify if your report is related?

The problem is still present in yesterdays git master, see attached stacktrace.
A shortened transcription of that stacktrace:

Bug: unable to handle kernel paging request at ...
[...]
Call Trace:
__do_softirq+[...]
irq_exit+[...]
smp_apic_timer_interrupt+[...]
apic_timer_interrupt+[...]
?default_idle+[...]
?clocksource_mark_unstable+[...]
?default_idle+[...]
arch_cpu_idle+[...]
cpu_startup_entry+[...]
rest_init+[...]
start_kernel+[...]
i386_start_kernel+[...]
[...]
EIP: run_timer_softirq+[...]
Kernel panic panic - not syncing: Fatal exception in interrupt

Any idea how to debug that problem? Insert some debugging code and constantly rebooting one of the
machines would be an option, but I don´t have a clue what to insert ....

cu,
  Knut

[-- Attachment #2: shutdown-trace-2013-09-13.pdf --]
[-- Type: application/pdf, Size: 331076 bytes --]

Re: [BUG] kernel panic during shutdown in run_timer_softirq()

[Frederic Weisbecker]

On Fri, Sep 13, 2013 at 12:01:56PM +0200, Knut Petersen wrote:
> Hi everybody!
> 
> Since about July I observe occasional kernel panics  happening only during system shutdown on two systems.
> 
> Hardware: mobos: both AOpen i915GMm-hfs mobos, cpus: Pentium-M Dothan / Banias, mem: 2GB
> Although the stack traces differ, there is one thing in common: CS:EIP is in run_timer_softirq().
> 
> A first report was posted to lkml on 2013-07-10: http://www.gossamer-threads.com/lists/linux/kernel/1744892?#1744892
> jpg of the stack trace: http://www.gossamer-threads.com/lists/engine?do=post_attachment;postatt_id=57017;list=linux
> 
> There were several reports of kernel panics with run_timer_softirq() in the call_trace.
> 
> @Grant: You reported a kernel panic during shutdown for kernel 3.7.5, but the jpg is not accessible.
> Could you please verify if your report is related?
> 
> The problem is still present in yesterdays git master, see attached stacktrace.
> A shortened transcription of that stacktrace:
> 
> Bug: unable to handle kernel paging request at ...
> [...]
> Call Trace:
> __do_softirq+[...]
> irq_exit+[...]
> smp_apic_timer_interrupt+[...]
> apic_timer_interrupt+[...]
> ?default_idle+[...]
> ?clocksource_mark_unstable+[...]
> ?default_idle+[...]
> arch_cpu_idle+[...]
> cpu_startup_entry+[...]
> rest_init+[...]
> start_kernel+[...]
> i386_start_kernel+[...]
> [...]
> EIP: run_timer_softirq+[...]
> Kernel panic panic - not syncing: Fatal exception in interrupt
> 
> Any idea how to debug that problem? Insert some debugging code and constantly rebooting one of the
> machines would be an option, but I don´t have a clue what to insert ....

Adding Thomas, in case he has a clue.

Thanks.