Re: [PATCH RFC] time,signal: protect resource use statistics with seqlock

[Rik van Riel <riel@redhat.com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 08/15/2014 12:49 PM, Oleg Nesterov wrote:

> Just in case... Yes, sure, "seqlock_t stats_lock" is more scalable.
> Just I do not know it's worth the trouble.

If we don't know whether it is worth the trouble, it is probably best
to stick to a well-known generic locking algorithm, instead of brewing
our own and trying to maintain it.

I have fixed the other locking issue you pointed out, Oleg.

Now to see if this change to cputime_adjust does the trick :)

+++ b/kernel/sched/cputime.c
@@ -605,9 +605,12 @@ static void cputime_adjust(struct task_cputime *curr,
 	 * If the tick based count grows faster than the scheduler one,
 	 * the result of the scaling may go backward.
 	 * Let's enforce monotonicity.
+	 * Atomic exchange protects against concurrent cputime_adjust.
 	 */
- -	prev->stime = max(prev->stime, stime);
- -	prev->utime = max(prev->utime, utime);
+	while (stime > (rtime = ACCESS_ONCE(prev->stime)))
+		cmpxchg(&prev->stime, rtime, stime);
+	while (utime > (rtime = ACCESS_ONCE(prev->utime)))
+		cmpxchg(&prev->utime, rtime, utime);

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1

iQEcBAEBAgAGBQJT7kJ4AAoJEM553pKExN6Do/oH/2lA5X/CrVuhOLBK1sVq3kRh
gGiOTT9pDQZH1wwafVNHKWaro3T/s9GNqemgvgt4UiKbjFeYkaOycHp1cuntJj8j
Wk8zNnWBOuGqqcSxzk1Duco3CByxshLNXxuYJfpdkdEXPqRyvURAOL58pxSybZzh
E6lT747ntFJu3GIbfC6Ta3q58pWLpVrhWlvonhSaqat6tOvlzo4MKiJxz3SbT6i0
cCpmQ5p/JoQ5+IUEbTOZYbE2bK2y5tSrMggAFwKWLB3/0zJm1h4+2Q/5PenCX59X
VDFmaOJLkNxGcVXg8x87itvqzfq/LkvDtwl9tTJmA5ccG37MPvM3803XF5OWVo0=
=aKES
-----END PGP SIGNATURE-----