linux-kernel - Re: mmotm woes, mainly compaction

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 for Android: free password hash cracker in your pocket

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

Date:	Tue, 12 Apr 2016 11:31:08 +0200
From:	Vlastimil Babka <vbabka@...e.cz>
To:	Hugh Dickins <hughd@...gle.com>, Michal Hocko <mhocko@...nel.org>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Michal Hocko <mhocko@...e.com>, linux-kernel@...r.kernel.org,
	linux-mm@...ck.org, Joonsoo Kim <iamjoonsoo.kim@....com>
Subject: Re: mmotm woes, mainly compaction

On 04/12/2016 09:18 AM, Hugh Dickins wrote:
> 3. /proc/sys/vm/stat_refresh warns nr_isolated_anon and nr_isolated_file
>     go increasingly negative under compaction: which would add delay when
>     should be none, or no delay when should delay.  putback_movable_pages()
>     decrements the NR_ISOLATED counts which acct_isolated() increments,
>     so isolate_migratepages_block() needs to acct before putback in that
>     special case, and isolate_migratepages_range() can always do the acct
>     itself, leaving migratepages putback to caller like most other places.

Sigh, looks like I notoriously suck at the nr_isolated_* accounting. The
isolate_migratepages_range() is also due to my 3.18 commit. Back then,
Joonsoo caught the problem for compaction side, but CMA issue remains.
Sorry and thanks.

----8<----
>From 8aa6e765d4f931718386e13290d43348e34f0e76 Mon Sep 17 00:00:00 2001
From: Hugh Dickins <hughd@...gle.com>
Date: Tue, 12 Apr 2016 11:14:49 +0200
Subject: [PATCH] mm, cma: prevent nr_isolated_* counters from going negative

/proc/sys/vm/stat_refresh warns nr_isolated_anon and nr_isolated_file
go increasingly negative under compaction: which would add delay when
should be none, or no delay when should delay. The bug in compaction was
due to a recent mmotm patch, but much older instance of the bug was also
noticed in isolate_migratepages_range() which is used for CMA and
gigantic hugepage allocations.

The bug is caused by putback_movable_pages() in an error path decrementing
the isolated counters without them being previously incremented by
acct_isolated(). Fix isolate_migratepages_range() by removing the error-path
putback, thus reaching acct_isolated() with migratepages still isolated, and
leaving putback to caller like most other places do.

[vbabka@...e.cz: expanded the changelog]
Fixes: edc2ca612496 ("mm, compaction: move pageblock checks up from isolate_migratepages_range()")
Cc: stable@...r.kernel.org #3.18+
Cc: Joonsoo Kim <iamjoonsoo.kim@....com>
Signed-off-by: Hugh Dickins <hughd@...gle.com>
Signed-off-by: Vlastimil Babka <vbabka@...e.cz>
---
 mm/compaction.c | 10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/mm/compaction.c b/mm/compaction.c
index ab649fba3d88..c0be38f634d3 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -930,16 +930,8 @@ isolate_migratepages_range(struct compact_control *cc, unsigned long start_pfn,
 		pfn = isolate_migratepages_block(cc, pfn, block_end_pfn,
 							ISOLATE_UNEVICTABLE);
 
-		/*
-		 * In case of fatal failure, release everything that might
-		 * have been isolated in the previous iteration, and signal
-		 * the failure back to caller.
-		 */
-		if (!pfn) {
-			putback_movable_pages(&cc->migratepages);
-			cc->nr_migratepages = 0;
+		if (!pfn)
 			break;
-		}
 
 		if (cc->nr_migratepages == COMPACT_CLUSTER_MAX)
 			break;
-- 
2.8.1