linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
To: Rik van Riel <riel@redhat.com>,
	Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: kosaki.motohiro@jp.fujitsu.com, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org
Subject: Re: [patch 05/19] split LRU lists into anon & file sets
Date: Wed, 30 Jan 2008 17:57:54 +0900	[thread overview]
Message-ID: <20080130175439.1AFD.KOSAKI.MOTOHIRO@jp.fujitsu.com> (raw)
In-Reply-To: <20080130121152.1AF1.KOSAKI.MOTOHIRO@jp.fujitsu.com>

Hi Rik, Lee

I found number of scan pages calculation bug.

1. wrong calculation order

	ap *= rotate_sum / (zone->recent_rotated_anon + 1);

   when recent_rotated_anon = 100 and recent_rotated_file = 0,
   
     rotate_sum / (zone->recent_rotated_anon + 1)
   = 100 / 101
   = 0

   at that time, ap become 0.

2. wrong fraction omission

	nr[l] = zone->nr_scan[l] * percent[file] / 100;

	when percent is very small,
	nr[l] become 0.

Test Result:
(1) $ ./hackbench 150 process 1000
(2) # sync; echo 3 > /proc/sys/vm/drop_caches
    $ dd if=tmp10G of=/dev/null
    $ ./hackbench 150 process 1000

rvr-split-lru + revert patch of previous mail
 	(1) 83.014
	(2) 717.009

rvr-split-lru + revert patch of previous mail + below patch
	(1) 61.965
	(2) 85.444 !!


Now, We got 1000% performance improvement against 2.6.24-rc8-mm1 :)



- kosaki


Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>

---
 mm/vmscan.c |   11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

Index: b/mm/vmscan.c
===================================================================
--- a/mm/vmscan.c	2008-01-30 15:22:10.000000000 +0900
+++ b/mm/vmscan.c	2008-01-30 16:03:28.000000000 +0900
@@ -1355,7 +1355,7 @@ static void get_scan_ratio(struct zone *
 	 *               anon + file       rotate_sum
 	 */
 	ap = (anon_prio * anon) / (anon + file + 1);
-	ap *= rotate_sum / (zone->recent_rotated_anon + 1);
+	ap = (ap * rotate_sum) / (zone->recent_rotated_anon + 1);
 	if (ap == 0)
 		ap = 1;
 	else if (ap > 100)
@@ -1363,7 +1363,7 @@ static void get_scan_ratio(struct zone *
 	percent[0] = ap;
 
 	fp = (file_prio * file) / (anon + file + 1);
-	fp *= rotate_sum / (zone->recent_rotated_file + 1);
+	fp = (fp * rotate_sum) / (zone->recent_rotated_file + 1);
 	if (fp == 0)
 		fp = 1;
 	else if (fp > 100)
@@ -1402,6 +1402,7 @@ static unsigned long shrink_zone(int pri
 
 	for_each_reclaimable_lru(l) {
 		if (scan_global_lru(sc)) {
+			unsigned long nr_max_scan;
 			int file = is_file_lru(l);
 			/*
 			 * Add one to nr_to_scan just to make sure that the
@@ -1409,7 +1410,11 @@ static unsigned long shrink_zone(int pri
 			 */
 			zone->nr_scan[l] += (zone_page_state(zone,
 				NR_INACTIVE_ANON + l) >> priority) + 1;
-			nr[l] = zone->nr_scan[l] * percent[file] / 100;
+			nr[l] = (zone->nr_scan[l] * percent[file] / 100) + 1;
+			nr_max_scan = zone_page_state(zone, NR_INACTIVE_ANON+l);
+			if (nr[l] > nr_max_scan)
+				nr[l] = nr_max_scan;
+
 			if (nr[l] >= sc->swap_cluster_max)
 				zone->nr_scan[l] = 0;
 			else



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  reply	other threads:[~2008-01-30  8:57 UTC|newest]

Thread overview: 74+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-01-08 20:59 [patch 00/19] VM pageout scalability improvements Rik van Riel
2008-01-08 20:59 ` [patch 01/19] move isolate_lru_page() to vmscan.c Rik van Riel
2008-01-08 22:03   ` Christoph Lameter
2008-01-08 20:59 ` [patch 02/19] free swap space on swap-in/activation Rik van Riel
2008-01-08 22:10   ` Christoph Lameter
2008-01-08 20:59 ` [patch 03/19] define page_file_cache() function Rik van Riel
2008-01-08 22:18   ` Christoph Lameter
2008-01-08 22:28     ` Rik van Riel
2008-01-09  4:26       ` KAMEZAWA Hiroyuki
2008-01-08 20:59 ` [patch 04/19] Use an indexed array for LRU variables Rik van Riel
2008-01-08 20:59 ` [patch 05/19] split LRU lists into anon & file sets Rik van Riel
2008-01-08 22:22   ` Christoph Lameter
2008-01-08 22:36     ` Rik van Riel
2008-01-08 22:42       ` Christoph Lameter
2008-01-09  2:45         ` Rik van Riel
2008-01-09  4:41   ` KAMEZAWA Hiroyuki
2008-01-10  2:21     ` Balbir Singh
2008-01-10  2:36       ` KAMEZAWA Hiroyuki
2008-01-10  3:26         ` Balbir Singh
2008-01-10  4:23           ` KAMEZAWA Hiroyuki
2008-01-10  2:28   ` KAMEZAWA Hiroyuki
2008-01-10  2:37     ` Rik van Riel
2008-01-11  3:59   ` KOSAKI Motohiro
2008-01-11 15:37     ` Rik van Riel
2008-01-11  6:24   ` KOSAKI Motohiro
2008-01-11 15:42     ` Rik van Riel
2008-01-11 15:59       ` Lee Schermerhorn
2008-01-11 16:15         ` Rik van Riel
2008-01-11 19:51           ` Lee Schermerhorn
2008-01-11 15:50     ` Lee Schermerhorn
2008-01-11 16:06       ` Rik van Riel
2008-01-11  7:35   ` KOSAKI Motohiro
2008-01-11 15:46     ` Rik van Riel
2008-01-14 23:57       ` KOSAKI Motohiro
2008-01-30  3:25   ` KOSAKI Motohiro
2008-01-30  8:57     ` KOSAKI Motohiro [this message]
2008-01-30 14:29       ` Lee Schermerhorn
2008-01-31  1:17         ` KOSAKI Motohiro
2008-01-31 10:48           ` Rik van Riel
2008-01-31 10:59             ` KOSAKI Motohiro
2008-02-07  0:35       ` Rik van Riel
2008-02-07  1:20         ` KOSAKI Motohiro
2008-02-07  1:36           ` Rik van Riel
2008-01-08 20:59 ` [patch 06/19] SEQ replacement for anonymous pages Rik van Riel
2008-01-08 20:59 ` [patch 07/19] (NEW) add some sanity checks to get_scan_ratio Rik van Riel
2008-01-09  4:16   ` KAMEZAWA Hiroyuki
2008-01-09 12:53     ` Rik van Riel
2008-01-08 20:59 ` [patch 08/19] add newly swapped in pages to the inactive list Rik van Riel
2008-01-08 22:28   ` Christoph Lameter
2008-01-08 20:59 ` [patch 09/19] (NEW) more aggressively use lumpy reclaim Rik van Riel
2008-01-08 22:30   ` Christoph Lameter
2008-01-14 15:28     ` Mel Gorman
2008-01-08 20:59 ` [patch 10/19] No Reclaim LRU Infrastructure Rik van Riel
2008-01-11  4:36   ` KOSAKI Motohiro
2008-01-11 15:43     ` Lee Schermerhorn
2008-01-15  0:06       ` KOSAKI Motohiro
2008-01-08 20:59 ` [patch 11/19] Non-reclaimable page statistics Rik van Riel
2008-01-08 20:59 ` [patch 12/19] scan noreclaim list for reclaimable pages Rik van Riel
2008-01-08 20:59 ` [patch 13/19] ramfs pages are non-reclaimable Rik van Riel
2008-01-08 20:59 ` [patch 14/19] SHM_LOCKED pages are nonreclaimable Rik van Riel
2008-01-08 20:59 ` [patch 15/19] non-reclaimable mlocked pages Rik van Riel
2008-01-08 20:59 ` [patch 16/19] mlock vma pages under mmap_sem held for read Rik van Riel
2008-01-08 20:59 ` [patch 17/19] handle mlocked pages during map/unmap and truncate Rik van Riel
2008-01-08 20:59 ` [patch 18/19] account mlocked pages Rik van Riel
2008-01-11 12:51   ` Balbir Singh
2008-01-13  5:18     ` Rik van Riel
2008-01-08 20:59 ` [patch 19/19] cull non-reclaimable anon pages from the LRU at fault time Rik van Riel
2008-01-10  4:39 ` [patch 00/19] VM pageout scalability improvements Mike Snitzer
2008-01-10 15:41   ` Rik van Riel
2008-01-10 16:08     ` Mike Snitzer
2008-01-11 10:41 ` Balbir Singh
2008-01-11 15:38   ` Rik van Riel
2008-01-11 11:47 ` Balbir Singh
2008-01-16  6:17 ` rvr split LRU minor regression ? KOSAKI Motohiro

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080130175439.1AFD.KOSAKI.MOTOHIRO@jp.fujitsu.com \
    --to=kosaki.motohiro@jp.fujitsu.com \
    --cc=Lee.Schermerhorn@hp.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=riel@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox