mm, slab: faster active and free stats
authorGreg Thelen <gthelen@google.com>
Tue, 13 Dec 2016 00:41:41 +0000 (16:41 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 13 Dec 2016 02:55:06 +0000 (18:55 -0800)
commitf728b0a5d72ae99c446f933912914a61254c03b6
treee35973494392c657faece3a8191adb5816e86d7f
parente70954fd6d4b469517fd906ef1c33310e90ef9f0
mm, slab: faster active and free stats

Reading /proc/slabinfo or monitoring slabtop(1) can become very
expensive if there are many slab caches and if there are very lengthy
per-node partial and/or free lists.

Commit 07a63c41fa1f ("mm/slab: improve performance of gathering slabinfo
stats") addressed the per-node full lists which showed a significant
improvement when no objects were freed.  This patch has the same
motivation and optimizes the remainder of the usecases where there are
very lengthy partial and free lists.

This patch maintains per-node active_slabs (full and partial) and
free_slabs rather than iterating the lists at runtime when reading
/proc/slabinfo.

When allocating 100GB of slab from a test cache where every slab page is
on the partial list, reading /proc/slabinfo (includes all other slab
caches on the system) takes ~247ms on average with 48 samples.

As a result of this patch, the same read takes ~0.856ms on average.

[rientjes@google.com: changelog]
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1611081505240.13403@chino.kir.corp.google.com
Signed-off-by: Greg Thelen <gthelen@google.com>
Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/slab.c
mm/slab.h