[Bug 53216] Re: BUG() report in getnstimeofday: soft lockup

[Jerome Lacoste]

On 10/11/06, Hervé Fache <herve at lucidia.net> wrote:
>
> are you saying that:
> 15:44:55.147148 sched_setscheduler(21264, SCHED_FIFO, { 2 }) = 0 <0.000027
> >
> is the last thing in the trace before:
> 15:45:34.548964 select(40, [3 4 5 7 39], [], [], {0, 496177}
> ?


Not there but here:

15:44:47.164609 sched_get_priority_max(SCHED_FIFO) = 99 <0.000008>
15:44:47.164665 sched_setscheduler(21263, SCHED_FIFO, { 99 }) = 0 <7.981930>
15:44:55.146712 mmap2(NULL, 8392704, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb3adb000 < 0.000022>

The sched_setscheduler takes almost 8 seconds to return.

IOW, is it this particular call (the first one, of course) that locks
> the kernel? That would pinpoint it quite well!


Whether it is a cause of a result of the hang I cannot say (it may be
something else causing the hang). But there's probably some relationship
between the hang and that call. Maybe a shared lock or something else.

Note that the load always sky-rockets after the hang.


> I shall try it myself.


Good luck. Take enough space on your partition. Skype takes around 120M of
strace logs per hour on my machine without particular activity.

J

-- 
BUG() report in getnstimeofday: soft lockup
https://launchpad.net/bugs/53216