Re: [RFC][PATCH 03/11] sched: Robustify preemption leak checks

[Peter Zijlstra <peterz@infradead.org>]

On Tue, Sep 29, 2015 at 11:07:34AM -0400, Steven Rostedt wrote:
> On Tue, 29 Sep 2015 11:28:28 +0200
> Peter Zijlstra <peterz@infradead.org> wrote:

> > --- a/kernel/sched/core.c
> > +++ b/kernel/sched/core.c
> > @@ -2960,8 +2960,10 @@ static inline void schedule_debug(struct
> >  	 * schedule() atomically, we ignore that path. Otherwise whine
> >  	 * if we are scheduling when we should not.
> >  	 */
> > -	if (unlikely(in_atomic_preempt_off() && prev->state != TASK_DEAD))
> > +	if (unlikely(in_atomic_preempt_off() && prev->state != TASK_DEAD)) {
> >  		__schedule_bug(prev);
> > +		preempt_count_set(PREEMPT_DISABLED);
> > +	}
> 
> Of course, if this was not a preemption leak, but something that called
> schedule within a preempt_disable()/preempt_enable() section, when it
> returns, preemption will be enabled, right?

Indeed.. But it ensures only the task that incorrectly called schedule()
gets screwed and not everybody else.

This is most important on x86 which has a per cpu preempt_count that is
not saved/restored (after this series). So if you schedule with an
invalid (!2*PREEMPT_DISABLE_OFFSET) preempt_count the next task is
messed up too.

Enforcing this invariant limits the borkage to just the one task.