Re: [RFC][PATCH 5/5] sched: Reduce ttwu rq->lock contention

[Oleg Nesterov <oleg@redhat.com>]

On 12/18, Yong Zhang wrote:
>
> On Sat, Dec 18, 2010 at 2:15 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> >
> > static int
> > try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags)
> > {
> >        unsigned long flags;
> >        int cpu, ret = 0;
> >
> >        smp_wmb();
> >        raw_spin_lock_irqsave(&p->pi_lock, flags);
> >
> >        if (!(p->state & state))
> >                goto unlock;
> >
> >        ret = 1; /* we qualify as a proper wakeup now */
>
> Could below happen in this __window__?
>
> p is going through wake_event

I don't think this can happen with wait_event/wake_up/etc,
wait_queue_head_t->lock adds the necessary synchronization.

But, in general,

> and it first set TASK_UNINTERRUPTIBLE,
> then waker see that and above if (!(p->state & state)) passed.
> But at this time condition == true for p, and p return to run and
> intend to sleep:
>           p->state == XXX;
>           sleep;
>
> then we could wake up a process which has wrong state, no?

I think this is possible, and this is possible whatever we do.
Afaics, this patch changes nothing in this sense. Consider:

	set_current_state(TASK_INTERRUPTIBLE);

	set_current_state(TASK_UNINTERRUPTIBLE);
	schedule();

wake_up_state(TASK_INTERRUPTIBLE) in between can in fact wakeup
this task in TASK_UNINTERRUPTIBLE state.

I do not think this is the problem. The user of wake_up_process()
should take care and write the correct code ;) And in any case,
any wait-event-like code should handle the spurious wakeups
correctly.

Or I missed your point?

Oleg.