Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751139AbZG3BtV (ORCPT ); Wed, 29 Jul 2009 21:49:21 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750809AbZG3BtU (ORCPT ); Wed, 29 Jul 2009 21:49:20 -0400 Received: from mga14.intel.com ([143.182.124.37]:55417 "EHLO mga14.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750776AbZG3BtU (ORCPT ); Wed, 29 Jul 2009 21:49:20 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.43,291,1246863600"; d="scan'208";a="170380126" Date: Thu, 30 Jul 2009 09:49:17 +0800 From: Wu Fengguang To: Martin Bligh Cc: Chad Talbott , "linux-kernel@vger.kernel.org" , "linux-mm@kvack.org" , Michael Rubin , Andrew Morton , "sandeen@redhat.com" , Michael Davidson Subject: Re: Bug in kernel 2.6.31, Slow wb_kupdate writeout Message-ID: <20090730014917.GB7326@localhost> References: <1786ab030907281211x6e432ba6ha6afe9de73f24e0c@mail.gmail.com> <33307c790907281449k5e8d4f6cib2c93848f5ec2661@mail.gmail.com> <33307c790907290015m1e6b5666x9c0014cdaf5ed08@mail.gmail.com> <20090729114322.GA9335@localhost> <33307c790907291719r2caf7914xb543877464ba6fc2@mail.gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <33307c790907291719r2caf7914xb543877464ba6fc2@mail.gmail.com> 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 Content-Length: 1353 Lines: 38 On Thu, Jul 30, 2009 at 08:19:34AM +0800, Martin Bligh wrote: > BTW, can you explain this code at the bottom of generic_sync_sb_inodes > for me? > > if (wbc->nr_to_write <= 0) { > wbc->more_io = 1; > break; > } > > I don't understand why we are setting more_io here? AFAICS, more_io > means there's more stuff to write ... I would think we'd set this if > nr_to_write was > 0 ? That's true: wbc.nr_to_write will always be set to MAX_WRITEBACK_PAGES by wb_writeback() before entering generic_sync_sb_inodes(). So wbc.nr_to_write <=0 indicates we are interrupted by the quota and should revisit generic_sync_sb_inodes() to check for more io (which will _normally_ find more dirty pages to write). > Or just have the section below brought up above this > break check and do: > > if (!list_empty(&sb->s_more_io) || !list_empty(&sb->s_io)) > wbc->more_io = 1; > > Am I just misunderstanding the intent of more_io ? It should be OK. I agree on the change if it makes the logic more straightforward. Thanks, Fengguang -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/