lkml.org 
[lkml]   [2011]   [Mar]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH,RFC] perf: panic due to inclied cpu context task_ctx value
On Wed, Mar 30, 2011 at 11:26:45PM +0200, Peter Zijlstra wrote:
> On Wed, 2011-03-30 at 18:37 +0200, Oleg Nesterov wrote:
> > > +static void perf_sched_events_dec(void)
> > > +{
> > > + jump_label_dec(&perf_sched_events_in);
> > > + JUMP_LABEL(&perf_sched_events_in, no_sync);
> > > + synchronize_sched();
> > > +no_sync:
> > > + jump_label_dec(&perf_sched_events_out);
> > > +}
> >
> > Nice! I didn't realize we can simply use JUMP_LABEL() directly and then
> > the code doesn't depend on HAVE_JUMP_LABEL.
>
> Yeah, avoids having to sprinkle tons of #ifdef goo around ;-)
>
> Anyway how about we do the partial revert below that should get us back
> to a working kernel and is a nice small patch to send -stable wards.
>
> After that we can try and be clever with clearing ->task_ctx from things
> like remove_from_context and the like.
>
>
> ---
> include/linux/perf_event.h | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
> index 311b4dc..04d75a8 100644
> --- a/include/linux/perf_event.h
> +++ b/include/linux/perf_event.h
> @@ -1086,7 +1086,7 @@ void perf_event_task_sched_out(struct task_struct *task, struct task_struct *nex
> {
> perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0);
>
> - COND_STMT(&perf_sched_events, __perf_event_task_sched_out(task, next));
> + __perf_event_task_sched_out(task, next);
> }
>
> extern void perf_event_mmap(struct vm_area_struct *vma);
>
works for me, passed my tests ;)

jirka


\
 
 \ /
  Last update: 2011-03-31 12:35    [W:0.120 / U:0.120 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site