[PATCH] slab: cache_reap(): further reduction in interrupt holdoff
authorChristoph Lameter <clameter@engr.sgi.com>
Wed, 22 Mar 2006 08:09:05 +0000 (00:09 -0800)
committerLinus Torvalds <torvalds@g5.osdl.org>
Wed, 22 Mar 2006 15:54:05 +0000 (07:54 -0800)
commit35386e3b0f876bf194982f48f027af0c216499ce
tree9fb79545f6797422f5cf666b0e65a4bb1882ee1d
parent248a0301e703cbf781aa02a91bcfc6da75870dd7
[PATCH] slab: cache_reap(): further reduction in interrupt holdoff

cache_reap takes the l3->list_lock (disabling interrupts) unconditionally
and then does a few checks and maybe does some cleanup.  This patch makes
cache_reap() only take the lock if there is work to do and then the lock is
taken and released for each cleaning action.

The checking of when to do the next reaping is done without any locking and
becomes racy.  Should not matter since reaping can also be skipped if the
slab mutex cannot be acquired.

The same is true for the touched processing.  If we get this wrong once in
awhile then we will mistakenly clean or not clean the shared cache.  This
will impact performance slightly.

Note that the additional drain_array() function introduced here will fall
out in a subsequent patch since array cleaning will now be very similar
from all callers.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
mm/slab.c