linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mel Gorman <mgorman@suse.de>
To: Ben Hutchings <ben@decadent.org.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	linux-kernel@vger.kernel.org, stable@vger.kernel.org,
	torvalds@linux-foundation.org, akpm@linux-foundation.org,
	alan@lxorguk.ukuu.org.uk, Dave Chinner <dchinner@redhat.com>,
	Al Viro <viro@zeniv.linux.org.uk>
Subject: Re: [ 13/40] vmscan: shrinker->nr updates race and go wrong
Date: Mon, 30 Jul 2012 10:06:45 +0100	[thread overview]
Message-ID: <20120730090645.GI612@suse.de> (raw)
In-Reply-To: <1343593770.4642.11.camel@deadeye.wl.decadent.org.uk>

On Sun, Jul 29, 2012 at 09:29:30PM +0100, Ben Hutchings wrote:
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -251,17 +251,29 @@ unsigned long shrink_slab(struct shrink_
> >  		unsigned long total_scan;
> [...]
> > +		total_scan += delta;
> > +		if (total_scan < 0) {
> [...]
> 
> This condition is never true since total_scan is unsigned.
> I think 3.0.y needs this as well:
> 
> commit 635697c663f38106063d5659f0cf2e45afcd4bb5
> Author: Konstantin Khlebnikov <khlebnikov@openvz.org>
> Date:   Thu Dec 8 14:33:51 2011 -0800
> 
>     vmscan: fix initial shrinker size handling
> 

You're right, thanks for pointing it out as I missed it. Greg, this commit
is already in 3.2 but can you pick it up for 3.0-stable please?

---8<---
vmscan: fix initial shrinker size handling

commit 635697c663f38106063d5659f0cf2e45afcd4bb5 upstream.

Stable note: The commit [acf92b48: vmscan: shrinker->nr updates race and
	go wrong] aimed to reduce excessive reclaim of slab objects but
	had bug in how it treated shrinker functions that returned -1.

A shrinker function can return -1, means that it cannot do anything
without a risk of deadlock.  For example prune_super() does this if it
cannot grab a superblock refrence, even if nr_to_scan=0.  Currently we
interpret this -1 as a ULONG_MAX size shrinker and evaluate `total_scan'
according to this.  So the next time around this shrinker can cause
really big pressure.  Let's skip such shrinkers instead.

Also make total_scan signed, otherwise the check (total_scan < 0) below
never works.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
---
 mm/vmscan.c |    9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index e0afff3..bd8a540 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -248,12 +248,16 @@ unsigned long shrink_slab(struct shrink_control *shrink,
 
 	list_for_each_entry(shrinker, &shrinker_list, list) {
 		unsigned long long delta;
-		unsigned long total_scan;
-		unsigned long max_pass;
+		long total_scan;
+		long max_pass;
 		int shrink_ret = 0;
 		long nr;
 		long new_nr;
 
+		max_pass = do_shrinker_shrink(shrinker, shrink, 0);
+		if (max_pass <= 0)
+			continue;
+
 		/*
 		 * copy the current shrinker scan count into a local variable
 		 * and zero it so that other concurrent shrinker invocations
@@ -264,7 +268,6 @@ unsigned long shrink_slab(struct shrink_control *shrink,
 		} while (cmpxchg(&shrinker->nr, nr, 0) != nr);
 
 		total_scan = nr;
-		max_pass = do_shrinker_shrink(shrinker, shrink, 0);
 		delta = (4 * nr_pages_scanned) / shrinker->seeks;
 		delta *= max_pass;
 		do_div(delta, lru_pages + 1);

  reply	other threads:[~2012-07-30  9:06 UTC|newest]

Thread overview: 54+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-07-26 21:14 [ 00/40] 3.0.39-stable review Greg KH
2012-07-26 21:29 ` [ 01/40] cifs: always update the inode cache with the results from a FIND_* Greg Kroah-Hartman
2012-07-26 21:29   ` [ 02/40] ntp: Fix STA_INS/DEL clearing bug Greg Kroah-Hartman
2012-07-26 21:29   ` [ 03/40] mm: fix lost kswapd wakeup in kswapd_stop() Greg Kroah-Hartman
2012-07-26 21:29   ` [ 04/40] MIPS: Properly align the .data..init_task section Greg Kroah-Hartman
2012-07-26 21:29   ` [ 05/40] UBIFS: fix a bug in empty space fix-up Greg Kroah-Hartman
2012-07-26 21:29   ` [ 06/40] dm raid1: fix crash with mirror recovery and discard Greg Kroah-Hartman
2012-07-26 21:29   ` [ 07/40] mm/vmstat.c: cache align vm_stat Greg Kroah-Hartman
2012-07-26 21:29   ` [ 08/40] mm: memory hotplug: Check if pages are correctly reserved on a per-section basis Greg Kroah-Hartman
2012-07-26 21:29   ` [ 09/40] mm: reduce the amount of work done when updating min_free_kbytes Greg Kroah-Hartman
2012-07-26 21:29   ` [ 10/40] mm: vmscan: fix force-scanning small targets without swap Greg Kroah-Hartman
2012-07-26 21:29   ` [ 11/40] vmscan: clear ZONE_CONGESTED for zone with good watermark Greg Kroah-Hartman
2012-07-26 21:29   ` [ 12/40] vmscan: add shrink_slab tracepoints Greg Kroah-Hartman
2012-07-26 21:29   ` [ 13/40] vmscan: shrinker->nr updates race and go wrong Greg Kroah-Hartman
2012-07-29 20:29     ` Ben Hutchings
2012-07-30  9:06       ` Mel Gorman [this message]
2012-07-30 15:41         ` Greg Kroah-Hartman
2012-07-26 21:29   ` [ 14/40] vmscan: reduce wind up shrinker->nr when shrinker cant do work Greg Kroah-Hartman
2012-07-26 21:29   ` [ 15/40] vmscan: limit direct reclaim for higher order allocations Greg Kroah-Hartman
2012-07-26 21:29   ` [ 16/40] vmscan: abort reclaim/compaction if compaction can proceed Greg Kroah-Hartman
2012-07-26 21:29   ` [ 17/40] mm: compaction: trivial clean up in acct_isolated() Greg Kroah-Hartman
2012-07-26 21:29   ` [ 18/40] mm: change isolate mode from #define to bitwise type Greg Kroah-Hartman
2012-07-26 21:29   ` [ 19/40] mm: compaction: make isolate_lru_page() filter-aware Greg Kroah-Hartman
2012-07-26 21:29   ` [ 20/40] mm: zone_reclaim: " Greg Kroah-Hartman
2012-07-26 21:29   ` [ 21/40] mm: migration: clean up unmap_and_move() Greg Kroah-Hartman
2012-07-26 21:29   ` [ 22/40] mm: compaction: allow compaction to isolate dirty pages Greg Kroah-Hartman
2012-07-26 21:29   ` [ 23/40] mm: compaction: determine if dirty pages can be migrated without blocking within ->migratepage Greg Kroah-Hartman
2012-07-26 21:29   ` [ 24/40] mm: page allocator: do not call direct reclaim for THP allocations while compaction is deferred Greg Kroah-Hartman
2012-07-26 21:29   ` [ 25/40] mm: compaction: make isolate_lru_page() filter-aware again Greg Kroah-Hartman
2012-07-26 21:29   ` [ 26/40] kswapd: avoid unnecessary rebalance after an unsuccessful balancing Greg Kroah-Hartman
2012-07-26 21:29   ` [ 27/40] kswapd: assign new_order and new_classzone_idx after wakeup in sleeping Greg Kroah-Hartman
2012-07-26 21:29   ` [ 28/40] mm: compaction: introduce sync-light migration for use by compaction Greg Kroah-Hartman
2012-07-26 21:29   ` [ 29/40] mm: vmscan: when reclaiming for compaction, ensure there are sufficient free pages available Greg Kroah-Hartman
2012-07-26 21:29   ` [ 30/40] mm: vmscan: do not OOM if aborting reclaim to start compaction Greg Kroah-Hartman
2012-07-26 21:29   ` [ 31/40] mm: vmscan: check if reclaim should really abort even if compaction_ready() is true for one zone Greg Kroah-Hartman
2012-07-26 21:29   ` [ 32/40] vmscan: promote shared file mapped pages Greg Kroah-Hartman
2012-07-26 21:29   ` [ 33/40] vmscan: activate executable pages after first usage Greg Kroah-Hartman
2012-07-26 21:29   ` [ 34/40] mm/vmscan.c: consider swap space when deciding whether to continue reclaim Greg Kroah-Hartman
2012-07-26 21:29   ` [ 35/40] mm: test PageSwapBacked in lumpy reclaim Greg Kroah-Hartman
2012-07-26 21:29   ` [ 36/40] mm: vmscan: convert global reclaim to per-memcg LRU lists Greg Kroah-Hartman
2012-07-30  0:25     ` Ben Hutchings
2012-07-30 15:29       ` Greg Kroah-Hartman
2012-07-26 21:29   ` [ 37/40] cpusets: avoid looping when storing to mems_allowed if one node remains set Greg Kroah-Hartman
2012-07-26 21:29   ` [ 38/40] cpusets: stall when updating mems_allowed for mempolicy or disjoint nodemask Greg Kroah-Hartman
2012-07-26 21:29   ` [ 39/40] cpuset: mm: reduce large amounts of memory barrier related damage v3 Greg Kroah-Hartman
2012-07-27 15:08     ` Herton Ronaldo Krzesinski
2012-07-27 15:23       ` Mel Gorman
2012-07-27 19:01         ` Greg Kroah-Hartman
2012-07-28  5:02           ` Herton Ronaldo Krzesinski
2012-07-28 10:26             ` Mel Gorman
2012-07-30 15:39               ` Greg Kroah-Hartman
2012-07-30 15:37             ` Greg Kroah-Hartman
2012-07-30 15:38               ` Greg Kroah-Hartman
2012-07-26 21:29   ` [ 40/40] mm/hugetlb: fix warning in alloc_huge_page/dequeue_huge_page_vma Greg Kroah-Hartman

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=20120730090645.GI612@suse.de \
    --to=mgorman@suse.de \
    --cc=akpm@linux-foundation.org \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=ben@decadent.org.uk \
    --cc=dchinner@redhat.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=viro@zeniv.linux.org.uk \
    /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;
as well as URLs for NNTP newsgroup(s).