linux-kernel - Re: [PATCH 30/34] mm: vmscan: Do not force kswapd to scan small targets

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

Message-ID: <20120725214428.GA6502@kroah.com>
Date:	Wed, 25 Jul 2012 14:44:28 -0700
From:	Greg KH <gregkh@...uxfoundation.org>
To:	Mel Gorman <mgorman@...e.de>
Cc:	Stable <stable@...r.kernel.org>, Linux-MM <linux-mm@...ck.org>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 30/34] mm: vmscan: Do not force kswapd to scan small
 targets

On Wed, Jul 25, 2012 at 10:35:08PM +0100, Mel Gorman wrote:
> On Wed, Jul 25, 2012 at 12:59:48PM -0700, Greg KH wrote:
> > On Mon, Jul 23, 2012 at 02:38:43PM +0100, Mel Gorman wrote:
> > > commit ad2b8e601099a23dffffb53f91c18d874fe98854 upstream - WARNING: this is a substitute patch.
> > > 
> > > Stable note: Not tracked in Bugzilla. This is a substitute for an
> > > 	upstream commit addressing a completely different issue that
> > > 	accidentally contained an important fix. The workload this patch
> > > 	helps was memcached when IO is started in the background. memcached
> > > 	should stay resident but without this patch it gets swapped more
> > > 	than it should. Sometimes this manifests as a drop in throughput
> > > 	but mostly it was observed through /proc/vmstat.
> > > 
> > > Commit [246e87a9: memcg: fix get_scan_count() for small targets] was
> > > meant to fix a problem whereby small scan targets on memcg were ignored
> > > causing priority to raise too sharply. It forced scanning to take place
> > > if the target was small, memcg or kswapd.
> > > 
> > > >From the time it was introduced it cause excessive reclaim by kswapd
> > > with workloads being pushed to swap that previously would have stayed
> > > resident. This was accidentally fixed by commit [ad2b8e60: mm: memcg:
> > > remove optimization of keeping the root_mem_cgroup LRU lists empty] but
> > > that patchset is not suitable for backporting.
> > > 
> > > The original patch came with no information on what workloads it benefits
> > > but the cost of it is obvious in that it forces scanning to take place
> > > on lists that would otherwise have been ignored such as small anonymous
> > > inactive lists. This patch partially reverts 246e87a9 so that small lists
> > > are not force scanned which means that IO-intensive workloads with small
> > > amounts of anonymous memory will not be swapped.
> > > 
> > > Signed-off-by: Mel Gorman <mgorman@...e.de>
> > > ---
> > >  mm/vmscan.c |    3 ---
> > >  1 file changed, 3 deletions(-)
> > 
> > I don't understand this patch.  The original
> > ad2b8e601099a23dffffb53f91c18d874fe98854 commit touched the file
> > mm/memcontrol.c and seemed to do something quite different from what you
> > have done below.
> > 
> 
> The main problem is I'm an idiot and "missed" when copying&paste and followed
> through with the mistake. The actual commit of interest was the one after it
> [b95a2f2d: mm: vmscan: convert global reclaim to per-memcg LRU lists]
> 
> That patch has this hunk in it
> 
> @@ -1886,7 +1886,7 @@ static void get_scan_count(struct mem_cgroup_zone *mz, struct scan_control *sc,
>          * latencies, so it's better to scan a minimum amount there as
>          * well.
>          */
> -       if (current_is_kswapd())
> +       if (current_is_kswapd() && mz->zone->all_unreclaimable)
>                 force_scan = true;
>         if (!global_reclaim(sc))
>                 force_scan = true;
> 
> This change makes it very difficult for kswapd to force scan which was
> the fix I was interested in but the series is not suitable for backport.
> This has changed again since in 3.5-rc1 due to commit [90126375: mm/vmscan:
> push lruvec pointer into get_scan_count()] where this check became
> 
> 	if (current_is_kswapd() && zone->all_unreclaimable)
> 
> Superficially that looks ok to backport, but it's not due to a subtle
> difference in how zone is looked up in the new context.
> 
> Can you use this patch as a replacement? It is functionally much closer
> to what happens upstream while still backporting the actual fix of
> interest.

Yes, that makes more sense as that is what the patch you included does
:)

I'll go queue it up now, thanks for the backport.

greg k-h
--
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/