From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933653Ab1AMUiV (ORCPT ); Thu, 13 Jan 2011 15:38:21 -0500 Received: from thunk.org ([69.25.196.29]:52601 "EHLO thunker.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752092Ab1AMUiO (ORCPT ); Thu, 13 Jan 2011 15:38:14 -0500 Date: Thu, 13 Jan 2011 15:38:09 -0500 From: "Ted Ts'o" To: Jason Wessel Cc: "Luis R. Rodriguez" , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: Power plug off / on - EXT4-fs (sda1): re-mounted. Opts: errors=remount-ro,commit=600 Message-ID: <20110113203809.GB31800@thunk.org> Mail-Followup-To: Ted Ts'o , Jason Wessel , "Luis R. Rodriguez" , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org References: <4D2F5EFE.6060407@windriver.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4D2F5EFE.6060407@windriver.com> User-Agent: Mutt/1.5.20 (2009-06-14) X-SA-Exim-Connect-IP: X-SA-Exim-Mail-From: tytso@thunk.org 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 Thu, Jan 13, 2011 at 02:22:22PM -0600, Jason Wessel wrote: > > That is interesting indeed. While I am running a substantially > older kernel ( 2.6.35-24-generic #42-Ubuntu 10.10 ). I was seeing a > consistent problem where if you pulled the power cord while the > system was under heavy I/O load, you could no longer run sync. The > jbd2 process just kept on running forever continuously. Shutdown > was impossible to because vfs unmount blocked (hold power key for 4+ > seconds...). Yeah, there were two separate bugs that have been addressed recently; both were in the generic VFS and writeback code. One was a fix to do more efficient forced writeouts at umount time. The other was a fix so that if new dirty pages are continuously being created (by having processes always writing more pages, those dastards :-), to make sync stop by only having it write the pages that were dirty at the time when the sync was initiated. - Ted