[PATCH v2 1/2] sched/loadavg: Avoid loadavg spikes caused by delayed NO_HZ accounting

[Matt Fleming <matt@xxxxxxxxxxxxxxxxxxx>]

If we crossed a sample window while in NO_HZ we will add LOAD_FREQ to
the pending sample window time on exit, setting the next update not
one window into the future, but two.

This situation on exiting NO_HZ is described by:

  this_rq->calc_load_update < jiffies < calc_load_update

In this scenario, what we should be doing is:

  this_rq->calc_load_update = calc_load_update		     [ next window ]

But what we actually do is:

  this_rq->calc_load_update = calc_load_update + LOAD_FREQ   [ next+1 window ]

This has the effect of delaying load average updates for potentially
up to ~9seconds.

This can result in huge spikes in the load average values due to
per-cpu uninterruptible task counts being out of sync when accumulated
across all CPUs.

It's safe to update the per-cpu active count if we wake between sample
windows because any load that we left in 'calc_load_idle' will have
been zero'd when the idle load was folded in calc_global_load().

This issue is easy to reproduce before,

  commit 9d89c257dfb9 ("sched/fair: Rewrite runnable load and utilization average tracking")

just by forking short-lived process pipelines built from ps(1) and
grep(1) in a loop. I'm unable to reproduce the spikes after that
commit, but the bug still seems to be present from code review.

Fixes: commit 5167e8d ("sched/nohz: Rewrite and fix load-avg computation -- again")
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Mike Galbraith <umgwanakikbuti@xxxxxxxxx>
Cc: Morten Rasmussen <morten.rasmussen@xxxxxxx>
Cc: Vincent Guittot <vincent.guittot@xxxxxxxxxx>
Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
Cc: <stable@xxxxxxxxxxxxxxx> # v3.5+
Signed-off-by: Matt Fleming <matt@xxxxxxxxxxxxxxxxxxx>
---
 kernel/sched/loadavg.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Changes in v2:

 - Folded in Peter's suggestion for how to fix this.

 - Tried to clairfy the changelog based on feedback from Peter and
   Frederic

diff --git a/kernel/sched/loadavg.c b/kernel/sched/loadavg.c
index a2d6eb71f06b..ec91fcc09bfe 100644
--- a/kernel/sched/loadavg.c
+++ b/kernel/sched/loadavg.c
@@ -201,8 +201,9 @@ void calc_load_exit_idle(void)
 	struct rq *this_rq = this_rq();
 
 	/*
-	 * If we're still before the sample window, we're done.
+	 * If we're still before the pending sample window, we're done.
 	 */
+	this_rq->calc_load_update = calc_load_update;
 	if (time_before(jiffies, this_rq->calc_load_update))
 		return;
 
@@ -211,7 +212,6 @@ void calc_load_exit_idle(void)
 	 * accounted through the nohz accounting, so skip the entire deal and
 	 * sync up for the next window.
 	 */
-	this_rq->calc_load_update = calc_load_update;
 	if (time_before(jiffies, this_rq->calc_load_update + 10))
 		this_rq->calc_load_update += LOAD_FREQ;
 }
-- 
2.10.0