From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934543AbYEUMe2 (ORCPT ); Wed, 21 May 2008 08:34:28 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1762057AbYEUMeQ (ORCPT ); Wed, 21 May 2008 08:34:16 -0400 Received: from www.church-of-our-saviour.ORG ([69.25.196.31]:46388 "EHLO thunker.thunk.org" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1761994AbYEUMeP (ORCPT ); Wed, 21 May 2008 08:34:15 -0400 Date: Wed, 21 May 2008 08:32:55 -0400 From: Theodore Tso To: Pavel Machek Cc: Chris Mason , Andrew Morton , Eric Sandeen , Andi Kleen , linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org Subject: Re: [PATCH 0/4] (RESEND) ext3[34] barrier changes Message-ID: <20080521123255.GG8581@mit.edu> Mail-Followup-To: Theodore Tso , Pavel Machek , Chris Mason , Andrew Morton , Eric Sandeen , Andi Kleen , linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org References: <482DDA56.6000301@redhat.com> <20080518211140.b29bee30.akpm@linux-foundation.org> <200805191316.27551.chris.mason@oracle.com> <200805191439.36577.chris.mason@oracle.com> <20080521112224.GD5028@ucw.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20080521112224.GD5028@ucw.cz> User-Agent: Mutt/1.5.15+20070412 (2007-04-11) X-SA-Exim-Connect-IP: X-SA-Exim-Mail-From: tytso@mit.edu X-SA-Exim-Scanned: No (on thunker.thunk.org); SAEximRunCond expanded to false Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, May 21, 2008 at 01:22:25PM +0200, Pavel Machek wrote: > > Ok, Andrew, is this enough to get barrier patch applied and stop > corrupting data in default config, or do you want some more testing? It is for me; I think we have to enable barriers for ext3/4, and then work to improve the overhead in ext4/jbd2. It's probably true that the vast majority of systems don't run under conditions similar to what Chris used to demonstrate the problem, but the default has to be filesystem safety. People who are sure they are extremely unlikely to run into this problem can turn barriers off (and I suspect they won't see that much difference in the most common workloads anyway). - Ted