Ingo Molnar af6d596fd6 sched: prevent divide by zero error in cpu_avg_load_per_task, update
Regarding the bug addressed in:

  4cd4262: sched: prevent divide by zero error in cpu_avg_load_per_task

Linus points out that the fix is not complete:

> There's nothing that keeps gcc from deciding not to reload
> rq->nr_running.
>
> Of course, in _practice_, I don't think gcc ever will (if it decides
> that it will spill, gcc is likely going to decide that it will
> literally spill the local variable to the stack rather than decide to
> reload off the pointer), but it's a valid compiler optimization, and
> it even has a name (rematerialization).
>
> So I suspect that your patch does fix the bug, but it still leaves the
> fairly unlikely _potential_ for it to re-appear at some point.
>
> We have ACCESS_ONCE() as a macro to guarantee that the compiler
> doesn't rematerialize a pointer access. That also would clarify
> the fact that we access something unsafe outside a lock.

So make sure our nr_running value is immutable and cannot change
after we check it for nonzero.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-11-29 20:45:15 +01:00
..
2008-10-22 07:37:17 +02:00
2008-10-16 11:21:30 -07:00
2008-03-30 14:18:41 -07:00
2008-09-14 16:25:35 +02:00
2008-11-19 18:49:58 -08:00
2008-07-28 14:37:38 +02:00
2007-07-16 09:05:50 -07:00
2008-08-13 12:55:10 +02:00
2008-10-28 11:19:07 +01:00
2008-09-02 19:21:40 -07:00
2008-10-20 08:52:39 -07:00
2008-10-21 15:59:53 +02:00
2008-11-18 15:08:56 +01:00
2008-11-11 11:57:22 +01:00
2008-11-05 10:30:14 +01:00
2007-07-16 09:05:50 -07:00
2008-11-06 08:41:56 +01:00
2008-10-16 11:21:31 -07:00
2008-02-06 10:41:02 -08:00
2008-11-19 18:49:57 -08:00
2008-07-25 10:53:45 -07:00
2008-10-16 11:21:47 -07:00
2008-10-16 11:21:31 -07:00