Bug 12809

Summary: iozone regression with 2.6.29-rc6
Product: Memory Management Reporter: Rafael J. Wysocki (rjw)
Component: OtherAssignee: Andrew Morton (akpm)
Status: CLOSED CODE_FIX    
Severity: normal CC: ming.m.lin
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.29-rc6 Subsystem:
Regression: Yes Bisected commit-id:
Bug Depends on:    
Bug Blocks: 12398    

Description Rafael J. Wysocki 2009-03-02 13:48:07 UTC
Subject    : iozone regression with 2.6.29-rc6
Submitter  : Lin Ming <ming.m.lin@intel.com>
Date       : 2009-02-27 9:13
References : http://marc.info/?l=linux-kernel&m=123572630504360&w=4
Handled-By : Wu Fengguang <fengguang.wu@intel.com>

This entry is being used for tracking a regression from 2.6.28.  Please don't
close it until the problem is fixed in the mainline.

Caused by:

commit 1cf6e7d83bf334cc5916137862c920a97aabc018
Author: Nick Piggin <npiggin@suse.de>
Date:   Wed Feb 18 14:48:18 2009 -0800

    mm: task dirty accounting fix

    Cc: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
    Signed-off-by: Nick Piggin <npiggin@suse.de>
    Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
    Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
    Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Comment 1 Rafael J. Wysocki 2009-03-02 13:48:35 UTC
First-Bad-Commit : 1cf6e7d83bf334cc5916137862c920a97aabc018
Comment 2 Rafael J. Wysocki 2009-03-07 13:37:05 UTC
On Thursday 05 March 2009, Wu Fengguang wrote:
> On Tue, Mar 03, 2009 at 09:25:59PM +0200, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a report
> > of recent regressions.
> > 
> > The following bug entry is on the current list of known regressions
> > from 2.6.28.  Please verify if it still should be listed and let me know
> > (either way).
> > 
> > 
> > Bug-Entry   : http://bugzilla.kernel.org/show_bug.cgi?id=12809
> > Subject             : iozone regression with 2.6.29-rc6
> > Submitter   : Lin Ming <ming.m.lin@intel.com>
> > Date                : 2009-02-27 9:13 (5 days old)
> 
> Following Peter's idea of relaxing writeback throttling, Lin Ming
> and I tried the idea of totally disabling writeback throttling by
> doing benchmarks with the following parameters:
> 
>         echo 60 > /proc/sys/vm/dirty_ratio
>         echo 50 > /proc/sys/vm/dirty_background_ratio
> 
> The result is encouraging: the iozone performance is restored to the
> level of 2.6.29-rc5!
> 
> We'll continue to evaluate the dirty/throttling numbers with Nick's patch.
Comment 3 Rafael J. Wysocki 2009-03-15 03:33:19 UTC
References : http://lkml.org/lkml/2009/3/14/265
Comment 4 Rafael J. Wysocki 2009-04-06 22:26:55 UTC
On Tuesday 07 April 2009, Wu Fengguang wrote:
> On Tue, Apr 07, 2009 at 06:15:09AM +0800, Wu Fengguang wrote:
> > On Tue, Apr 07, 2009 at 03:05:38AM +0800, Rafael J. Wysocki wrote:
> > > This message has been generated automatically as a part of a report
> > > of recent regressions.
> > > 
> > > The following bug entry is on the current list of known regressions
> > > from 2.6.28.  Please verify if it still should be listed and let me know
> > > (either way).
> > > 
> > > 
> > > Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=12809
> > > Subject           : iozone regression with 2.6.29-rc6
> > > Submitter : Lin Ming <ming.m.lin@intel.com>
> > > Date              : 2009-02-27 9:13 (39 days old)
> > > First-Bad-Commit:
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=1cf6e7d83bf334cc5916137862c920a97aabc018
> > > References        :
> http://marc.info/?l=linux-kernel&m=123572630504360&w=4
> > >             http://lkml.org/lkml/2009/3/14/265
> > > Handled-By        : Wu Fengguang <fengguang.wu@intel.com>
> > > 
> > 
> > This bug could be closed.
> > 
> > The performance regression is triggered by commit 1cf6e7d83bf3(mm:
> > task dirty accounting fix). Which has been reverted in 2.6.29.
> 
> Sorry for the mistake, commit 1cf6e7d83bf3 was _not_ reverted.
> So 2.6.29 is expected to have this regression.. 
> 
> > For 2.6.30, commit 1cf6e7d83bf33(mm: task dirty accounting fix) and
> > 1b5e62b42b55(writeback: double the dirty thresholds) were just merged
> > together. The latter one restores the iozone performance. (However
> > might deteriorate some writeback slowness bugs..)