sched/fair: Make select_idle_cpu() more aggressive
authorPeter Zijlstra <peterz@infradead.org>
Wed, 1 Mar 2017 10:24:35 +0000 (11:24 +0100)
committerIngo Molnar <mingo@kernel.org>
Thu, 2 Mar 2017 07:50:17 +0000 (08:50 +0100)
commit4c77b18cf8b7ab37c7d5737b4609010d2ceec5f0
tree9810db6cc4cfc6ab80aa0704d9bef7d9634497fb
parent4977ab6e92e267afe9d8f78438c3db330ca8434c
sched/fair: Make select_idle_cpu() more aggressive

Kitsunyan reported desktop latency issues on his Celeron 887 because
of commit:

  1b568f0aabf2 ("sched/core: Optimize SCHED_SMT")

... even though his CPU doesn't do SMT.

The effect of running the SMT code on a !SMT part is basically a more
aggressive select_idle_cpu(). Removing the avg condition fixed things
for him.

I also know FB likes this test gone, even though other workloads like
having it.

For now, take it out by default, until we get a better idea.

Reported-by: kitsunyan <kitsunyan@inbox.ru>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Mike Galbraith <umgwanakikbuti@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/fair.c
kernel/sched/features.h