mm: reduce the amount of work done when updating min_free_kbytes
authorMel Gorman <mgorman@suse.de>
Tue, 10 Jan 2012 23:07:14 +0000 (15:07 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 11 Jan 2012 00:30:42 +0000 (16:30 -0800)
commit938929f14cb595f43cd1a4e63e22d36cab1e4a1f
tree54c23d02494c05d13cc6a6ffb327cc1fb03e72fd
parent937a94c9db30a818baa5e2c09dbf4589251355c3
mm: reduce the amount of work done when updating min_free_kbytes

When min_free_kbytes is updated, some pageblocks are marked
MIGRATE_RESERVE.  Ordinarily, this work is unnoticable as it happens early
in boot but on large machines with 1TB of memory, this has been reported
to delay boot times, probably due to the NUMA distances involved.

The bulk of the work is due to calling calling pageblock_is_reserved() an
unnecessary amount of times and accessing far more struct page metadata
than is necessary.  This patch significantly reduces the amount of work
done by setup_zone_migrate_reserve() improving boot times on 1TB machines.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/page_alloc.c