Re: [PATCH,RFC] perf: panic due to inclied cpu context task_ctx value

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Sat, 2011-03-26 at 18:35 +0100, Oleg Nesterov wrote:
> On 03/26, Oleg Nesterov wrote:
> >
> > On 03/26, Peter Zijlstra wrote:
> > >
> > > diff --git a/kernel/perf_event.c b/kernel/perf_event.c
> > > index c75925c..e9e4e35 100644
> > > --- a/kernel/perf_event.c
> > > +++ b/kernel/perf_event.c
> > > @@ -1073,6 +1073,8 @@ event_sched_out(struct perf_event *event,
> > >  	if (!is_software_event(event))
> > >  		cpuctx->active_oncpu--;
> > >  	ctx->nr_active--;
> > > +	if (!ctx->nr_active && cpuctx->task_ctx == ctx)
> > > +		cpuctx->task_ctx = NULL;
> >
> > If we clear cpuctx->task_ctx, we should also clear ctx->is_active.

Right.

> > Otherwise I can't see any problem, but I do not understand this code
> > enough.
> 
> but probably we also need update_context_time().

It looks like event_sched_out() relies on up-to-date ctx->time through
perf_event_time() and most call-paths leading to event_sched_out() do
indeed seem to update the ctx time, all except the move_group branch in
perf_event_open() afaict.

Sadly the reproducer doesn't seem to trigger the issue at all, its still
running on a plain -tip kernel.