[PATCH 4.12 21/27] sched/numa: Override part of migrate_degrades_locality() when idle balancing

From: Greg Kroah-Hartman
Date: Mon Jul 10 2017 - 13:11:40 EST

4.12-stable review patch. If anyone has any objections, please let me know.


From: Rik van Riel <riel@xxxxxxxxxx>

commit 739294fb03f590401bbd7faa6d31a507e3ffada5 upstream.

Several tests in the NAS benchmark seem to run a lot slower with
NUMA balancing enabled, than with NUMA balancing disabled. The
slower run time corresponds with increased idle time.

Overriding the final test of migrate_degrades_locality (but still
doing the other NUMA tests first) seems to improve performance
of those benchmarks.

Reported-by: Jirka Hladky <jhladky@xxxxxxxxxx>
Signed-off-by: Rik van Riel <riel@xxxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Mel Gorman <mgorman@xxxxxxx>
Cc: Mike Galbraith <efault@xxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: linux-kernel@xxxxxxxxxxxxxxx
Link: http://lkml.kernel.org/r/20170623165530.22514-2-riel@xxxxxxxxxx
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
Signed-off-by: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>

kernel/sched/fair.c | 4 ++++
1 file changed, 4 insertions(+)

--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6661,6 +6661,10 @@ static int migrate_degrades_locality(str
if (dst_nid == p->numa_preferred_nid)
return 0;

+ /* Leaving a core idle is often worse than degrading locality. */
+ if (env->idle != CPU_NOT_IDLE)
+ return -1;
if (numa_group) {
src_faults = group_faults(p, src_nid);
dst_faults = group_faults(p, dst_nid);