lkml.org 
[lkml]   [2010]   [Nov]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 1/2] mm: page allocator: Adjust the per-cpu counter threshold when memory is low
On Fri, Nov 26, 2010 at 11:06:19AM -0500, Kyle McMartin wrote:
> On Wed, Oct 27, 2010 at 09:47:35AM +0100, Mel Gorman wrote:
> ><snip>
> > To ensure that kswapd wakes up, a safe version of zone_watermark_ok()
> > is introduced that takes a more accurate reading of NR_FREE_PAGES when
> > called from wakeup_kswapd, when deciding whether it is really safe to go
> > back to sleep in sleeping_prematurely() and when deciding if a zone is
> > really balanced or not in balance_pgdat(). We are still using an expensive
> > function but limiting how often it is called.
> ><snip>
> > Reported-by: Shaohua Li <shaohua.li@intel.com>
> > Signed-off-by: Mel Gorman <mel@csn.ul.ie>
>
> Hi Mel,
>
> I notice these aren't flagged for stable, should they be? (They fairly
> trivially apply and compile on 2.6.36 barring the trace_ points which
> changed.)

They were not flagged for stable because they were performance rather than
function bugs that affected a limited number of machines. Should that decision
be revisited?

> I've got a few bug reports against .36/.37 where kswapd has
> been sleeping for 60s+.
>

I do not believe these patches would affect kswapd sleeping for 60s.

> I built them some kernels with these patches, but haven't heard back yet
> as to whether it fixes things for them.
>
> Thanks for any insight,

Can you point me at a relevant bugzilla entry or forward me the bug report
and I'll take a look?

--
Mel Gorman
Part-time Phd Student Linux Technology Center
University of Limerick IBM Dublin Software Lab


\
 
 \ /
  Last update: 2010-11-29 10:59    [from the cache]
©2003-2011 Jasper Spaans