Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751731Ab0HSPwl (ORCPT ); Thu, 19 Aug 2010 11:52:41 -0400 Received: from hera.kernel.org ([140.211.167.34]:50613 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751445Ab0HSPwj (ORCPT ); Thu, 19 Aug 2010 11:52:39 -0400 Message-ID: <4C6D5257.9040305@kernel.org> Date: Thu, 19 Aug 2010 17:48:39 +0200 From: Tejun Heo User-Agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.2.8) Gecko/20100802 Thunderbird/3.1.2 MIME-Version: 1.0 To: Christoph Hellwig CC: "Darrick J. Wong" , "Ted Ts'o" , Mingming Cao , Ric Wheeler , linux-ext4 , linux-kernel , Keith Mannthey , Mingming Cao Subject: Re: [RFC v3] ext4: Combine barrier requests coming from fsync References: <4BE02C45.6010608@redhat.com> <1273002566.3755.10.camel@mingming-laptop> <20100629205102.GM15515@tux1.beaverton.ibm.com> <20100805164008.GH2901@thunk.org> <20100805164504.GI2901@thunk.org> <20100806070424.GD2109@tux1.beaverton.ibm.com> <20100809195324.GG2109@tux1.beaverton.ibm.com> <20100809210723.GA28171@infradead.org> <20100816161433.GK2109@tux1.beaverton.ibm.com> <20100819020734.GL2109@tux1.beaverton.ibm.com> <20100819085349.GA8782@infradead.org> <4C6CF6BC.3030804@kernel.org> In-Reply-To: <4C6CF6BC.3030804@kernel.org> X-Enigmail-Version: 1.1.1 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (hera.kernel.org [127.0.0.1]); Thu, 19 Aug 2010 15:52:22 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1067 Lines: 26 Hello, again. On 08/19/2010 11:17 AM, Tejun Heo wrote: > What was the configuration? If dm was involved, both md and dm can > only process single flush request at a time. Supporing multiple > flushes in flight wouldn't be too difficult. It's just the way things > are implemented with the previous barrier implementation. It can > definitely be improved. Oh, I just realized that the current bio-based dm implementation doesn't allow other bio's to be processed if a flush is in progress. So, it's not about not being able to handle multiple flushes. The queue just stalls while flush is in progress and because it also waits for flush completion by waiting for all commands in progress to finish. It basically ends up draining and stalling everything. Shouldn't be too hard to make it better. Thanks. -- tejun -- 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/