From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753949AbZCOH4a (ORCPT ); Sun, 15 Mar 2009 03:56:30 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751358AbZCOH4U (ORCPT ); Sun, 15 Mar 2009 03:56:20 -0400 Received: from mga14.intel.com ([143.182.124.37]:61197 "EHLO mga14.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751031AbZCOH4T (ORCPT ); Sun, 15 Mar 2009 03:56:19 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.38,365,1233561600"; d="scan'208";a="120379903" Date: Sun, 15 Mar 2009 15:55:52 +0800 From: Wu Fengguang To: Linus Torvalds Cc: "Rafael J. Wysocki" , Linux Kernel Mailing List , Kernel Testers List , Andrew Morton , "Lin, Ming M" , Nick Piggin , Peter Zijlstra Subject: Re: [Bug #12809] iozone regression with 2.6.29-rc6 Message-ID: <20090315075552.GA6015@localhost> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sun, Mar 15, 2009 at 08:27:08AM +0800, Linus Torvalds wrote: > > > On Sat, 14 Mar 2009, Rafael J. Wysocki wrote: > > > > 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 > > Date : 2009-02-27 9:13 (16 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 > > Handled-By : Wu Fengguang > > I suspect that I should just raise the default dirty limits. Wu reported > that it fixed the regression, and while he picked some rather high > percentages, I think we could certainly raise the rather aggressive > default ones. > > After all, those default percentages were picked (a) with the old dirty > logic and (b) largely at random and (c) designed to be aggressive. In > particular, that (a) means that having fixed some of the dirty accounting, > maybe the real bug is now that it was always too aggressive, just hidden > by an accounting issue. I second that. 1) The _real_ dirty threshold used to be large. 2) It is a _real_ regression. It impacts real user experiences. So when introducing Nick's correct-dirty-accounting patch, we'd better increase the dirty thresholds correspondingly. > If we raised the default ratio from 5/10 to 10/20, what happens to the > iozone regression? Maybe tomorrow. Ling Ming? In general we should not cater the thresholds for one specific workload. But this is a case of _regression_, and it would be better to raise the bars above it. Thanks, Fengguang From mboxrd@z Thu Jan 1 00:00:00 1970 From: Wu Fengguang Subject: Re: [Bug #12809] iozone regression with 2.6.29-rc6 Date: Sun, 15 Mar 2009 15:55:52 +0800 Message-ID: <20090315075552.GA6015@localhost> References: Mime-Version: 1.0 Return-path: Content-Disposition: inline In-Reply-To: Sender: kernel-testers-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Linus Torvalds Cc: "Rafael J. Wysocki" , Linux Kernel Mailing List , Kernel Testers List , Andrew Morton , "Lin, Ming M" , Nick Piggin , Peter Zijlstra On Sun, Mar 15, 2009 at 08:27:08AM +0800, Linus Torvalds wrote: > > > On Sat, 14 Mar 2009, Rafael J. Wysocki wrote: > > > > 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 > > Date : 2009-02-27 9:13 (16 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 > > Handled-By : Wu Fengguang > > I suspect that I should just raise the default dirty limits. Wu reported > that it fixed the regression, and while he picked some rather high > percentages, I think we could certainly raise the rather aggressive > default ones. > > After all, those default percentages were picked (a) with the old dirty > logic and (b) largely at random and (c) designed to be aggressive. In > particular, that (a) means that having fixed some of the dirty accounting, > maybe the real bug is now that it was always too aggressive, just hidden > by an accounting issue. I second that. 1) The _real_ dirty threshold used to be large. 2) It is a _real_ regression. It impacts real user experiences. So when introducing Nick's correct-dirty-accounting patch, we'd better increase the dirty thresholds correspondingly. > If we raised the default ratio from 5/10 to 10/20, what happens to the > iozone regression? Maybe tomorrow. Ling Ming? In general we should not cater the thresholds for one specific workload. But this is a case of _regression_, and it would be better to raise the bars above it. Thanks, Fengguang