linux-kernel - [PATCH] kswapd: avoid unnecessary rebalance after an unsuccessful balancing

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 29 Jul 2011 23:23:10 +0800
From:	Alex Shi <alex.shi@...el.com>
To:	linux-mm@...ck.org, P@...igBrady.com
Cc:	mgorman@...e.de, linux-kernel@...r.kernel.org, andrea@...share.com,
	tim.c.chen@...el.com, shaohua.li@...el.com,
	akpm@...ux-foundation.org, riel@...hat.com, luto@....edu
Subject: [PATCH] kswapd: avoid unnecessary rebalance after an unsuccessful balancing

In commit 215ddd66, Mel Gorman said kswapd is better to sleep after a
unsuccessful balancing if there is tighter reclaim request pending in
the balancing. In this scenario, the 'order' and 'classzone_idx'
that are checked for tighter request judgment is incorrect, since they
aren't the one kswapd should read from new pgdat, but the last time pgdat
value for just now balancing. Then kswapd will skip try_to_sleep func
and rebalance the last pgdat request. It's not our expected behavior.

So, I added new variables to distinguish the returned order/classzone_idx
from last balancing, that can resolved above issue in that scenario.

I tested the patch on our LKP system with swap-cp/fio mmap randrw
benchmarks. The performance has no change.

Padraig Brady, would you like to test this patch for your scenario.

Signed-off-by: Alex Shi <alex.shi@...el.com>
Reviewed-by:  Tim Chen <tim.c.chen@...ux.intel.com>
---
 mm/vmscan.c |   11 ++++++++---
 1 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index eb7bcce..6380674 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2792,7 +2792,9 @@ static void kswapd_try_to_sleep(pg_data_t *pgdat, int order, int classzone_idx)
 static int kswapd(void *p)
 {
 	unsigned long order, new_order;
+	unsigned balanced_order;
 	int classzone_idx, new_classzone_idx;
+	int balanced_classzone_idx;
 	pg_data_t *pgdat = (pg_data_t*)p;
 	struct task_struct *tsk = current;
 
@@ -2823,7 +2825,9 @@ static int kswapd(void *p)
 	set_freezable();
 
 	order = new_order = 0;
+	balanced_order = 0;
 	classzone_idx = new_classzone_idx = pgdat->nr_zones - 1;
+	balanced_classzone_idx = classzone_idx;
 	for ( ; ; ) {
 		int ret;
 
@@ -2832,7 +2836,7 @@ static int kswapd(void *p)
 		 * new request of a similar or harder type will succeed soon
 		 * so consider going to sleep on the basis we reclaimed at
 		 */
-		if (classzone_idx >= new_classzone_idx && order == new_order) {
+		if (balanced_classzone_idx >= new_classzone_idx && balanced_order == new_order) {
 			new_order = pgdat->kswapd_max_order;
 			new_classzone_idx = pgdat->classzone_idx;
 			pgdat->kswapd_max_order =  0;
@@ -2847,7 +2851,7 @@ static int kswapd(void *p)
 			order = new_order;
 			classzone_idx = new_classzone_idx;
 		} else {
-			kswapd_try_to_sleep(pgdat, order, classzone_idx);
+			kswapd_try_to_sleep(pgdat, balanced_order, balanced_classzone_idx);
 			order = pgdat->kswapd_max_order;
 			classzone_idx = pgdat->classzone_idx;
 			new_order = order;
@@ -2866,7 +2870,8 @@ static int kswapd(void *p)
 		 */
 		if (!ret) {
 			trace_mm_vmscan_kswapd_wake(pgdat->node_id, order);
-			order = balance_pgdat(pgdat, order, &classzone_idx);
+			balanced_classzone_idx = classzone_idx;
+			balanced_order = balance_pgdat(pgdat, order, &balanced_classzone_idx);
 		}
 	}
 	return 0;
-- 
1.6.3.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/