]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
sched/numa: Complete scanning of partial VMAs regardless of PID activity
authorMel Gorman <mgorman@techsingularity.net>
Tue, 10 Oct 2023 08:31:42 +0000 (09:31 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 4 Oct 2024 14:29:22 +0000 (16:29 +0200)
commitcb7846df6b4f347fa6bd27aaff83db8c144bea74
tree47f84815973b3caf704933a869ff20d732e028bb
parent7f01977665d700924f24d9113d3fe99a96d0f47b
sched/numa: Complete scanning of partial VMAs regardless of PID activity

[ Upstream commit b7a5b537c55c088d891ae554103d1b281abef781 ]

NUMA Balancing skips VMAs when the current task has not trapped a NUMA
fault within the VMA. If the VMA is skipped then mm->numa_scan_offset
advances and a task that is trapping faults within the VMA may never
fully update PTEs within the VMA.

Force tasks to update PTEs for partially scanned PTEs. The VMA will
be tagged for NUMA hints by some task but this removes some of the
benefit of tracking PID activity within a VMA. A follow-on patch
will mitigate this problem.

The test cases and machines evaluated did not trigger the corner case so
the performance results are neutral with only small changes within the
noise from normal test-to-test variance. However, the next patch makes
the corner case easier to trigger.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Raghavendra K T <raghavendra.kt@amd.com>
Link: https://lore.kernel.org/r/20231010083143.19593-6-mgorman@techsingularity.net
Stable-dep-of: f22cde4371f3 ("sched/numa: Fix the vma scan starving issue")
Signed-off-by: Sasha Levin <sashal@kernel.org>
include/linux/sched/numa_balancing.h
include/trace/events/sched.h
kernel/sched/fair.c