sched: Fix nohz load accounting -- again!
authorPeter Zijlstra <peterz@infradead.org>
Thu, 1 Mar 2012 14:04:46 +0000 (15:04 +0100)
committerIngo Molnar <mingo@elte.hu>
Mon, 12 Mar 2012 19:43:17 +0000 (20:43 +0100)
commitc308b56b5398779cd3da0f62ab26b0453494c3d4
treeb2d90bd6f8ebbbeaae97f4ef319f7e22fcc77436
parent8e3fabfde445a872c8aec2296846badf24d7c8b4
sched: Fix nohz load accounting -- again!

Various people reported nohz load tracking still being wrecked, but Doug
spotted the actual problem. We fold the nohz remainder in too soon,
causing us to loose samples and under-account.

So instead of playing catch-up up-front, always do a single load-fold
with whatever state we encounter and only then fold the nohz remainder
and play catch-up.

Reported-by: Doug Smythies <dsmythies@telus.net>
Reported-by: LesÃ…=82aw Kope=C4=87 <leslaw.kopec@nasza-klasa.pl>
Reported-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-4v31etnhgg9kwd6ocgx3rxl8@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/sched/core.c