Re: hit a KASan bug related to Perf during stress test

[Oleg Nesterov]

On 10/24, Peter Zijlstra wrote:
>
> On Mon, Oct 24, 2016 at 01:15:27PM +0200, Oleg Nesterov wrote:
> >
> > --- x/kernel/events/core.c
> > +++ x/kernel/events/core.c
> > @@ -1257,7 +1257,7 @@ static u32 perf_event_pid(struct perf_ev
> >  	if (event->parent)
> >  		event = event->parent;
> >
> > -	return task_tgid_nr_ns(p, event->ns);
> > +	return pid_alive(p) ? task_tgid_nr_ns(p, event->ns) : 0;
> >  }
> >
> >  static u32 perf_event_tid(struct perf_event *event, struct task_struct *p)
>
> Should we do the same for perf_event_tid() and report -1 as the pid/tid
> in the !alive case? -1 should be an obvious invalid pid since we limit
> the pid-space to less than 32 bits.

task_pid_nr_ns() is always safe, it calls __task_pid_nr_ns(). But yes,
it can return zero if called after exit_notify() and/or release_task().

And while zero is not a valid pid too, I guess it can be confused with
the idle thread's "pid" ?

Oleg.