sched/numa: Reduce conflict between fbq_classify_rq() and migration
authorRik van Riel <riel@redhat.com>
Fri, 15 May 2015 02:59:36 +0000 (22:59 -0400)
committerIngo Molnar <mingo@kernel.org>
Tue, 19 May 2015 06:39:19 +0000 (08:39 +0200)
commitc1ceac6276e4ee12e4129afd380db10fae0db7df
tree401ed45d2621e110c5c4699c317224066e6d87ab
parent80ed87c8a9ca0cad7ca66cf3bbdfb17559a66dcf
sched/numa: Reduce conflict between fbq_classify_rq() and migration

It is possible for fbq_classify_rq() to indicate that a CPU has tasks that
should be moved to another NUMA node, but for migrate_improves_locality
and migrate_degrades_locality to not identify those tasks.

This patch always gives preference to preferred node evaluations, and
only checks the number of faults when evaluating moves between two
non-preferred nodes on a larger NUMA system.

On a two node system, the number of faults is never evaluated. Either
a task is about to be pulled off its preferred node, or migrated onto
it.

Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: mgorman@suse.de
Link: http://lkml.kernel.org/r/20150514225936.35b91717@annuminas.surriel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/fair.c