From: George Beshers <gbeshers@redhat.com>
Date: Tue, 29 Jul 2008 11:34:40 -0400
Subject: [mm] drain_node_page: drain pages in batch units
Message-id: 488F3890.60909@redhat.com
O-Subject: [RHEL5.3 PATCH] drain_node_page(): Drain pages in batch units
Bugzilla: 442179
RH-Acked-by: Larry Woodman <lwoodman@redhat.com>

[PATCH] drain_node_page(): Drain pages in batch units

BZ442179.
http://git2.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=bc4ba393c007248f76c05945abb7b7b892cdd1cc
patch against 2.6.18-99.

Tested both at SGI and by me on ia64/altix.  Brew build is
http://porkchop.redhat.com/brewroot/scratch/gbeshers/task_1413055/
for further testing.

drain_node_pages() currently drains the complete pageset of all pages.  If
there are a large number of pages in the queues then we may hold off
interrupts for too long.

Duplicate the method used in free_hot_cold_page.  Only drain pcp->batch
pages at one time.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Please ACK or comment.

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 45e6c43..46f8b2f 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -642,9 +642,16 @@ void drain_node_pages(int nodeid)
 
 			pcp = &pset->pcp[i];
 			if (pcp->count) {
+				int to_drain;
+
 				local_irq_save(flags);
-				free_pages_bulk(zone, pcp->count, &pcp->list, 0);
-				pcp->count = 0;
+				if (pcp->count >= pcp->batch)
+					to_drain = pcp->batch;
+				else
+					to_drain = pcp->count;
+				free_pages_bulk(zone, to_drain, &pcp->list, 0);
+				pcp->count -= to_drain;
+
 				touch_softlockup_watchdog();
 				local_irq_restore(flags);
 			}