From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1763468AbYBHD0M (ORCPT ); Thu, 7 Feb 2008 22:26:12 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1760768AbYBHDZ5 (ORCPT ); Thu, 7 Feb 2008 22:25:57 -0500 Received: from mx1.redhat.com ([66.187.233.31]:54469 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758147AbYBHDZ4 (ORCPT ); Thu, 7 Feb 2008 22:25:56 -0500 Date: Fri, 8 Feb 2008 03:24:23 +0000 From: Alasdair G Kergon To: Andrew Morton Cc: Linus Torvalds , Brian Wood , Daniel Walker , Herbert Xu , Jonathan Brassow , "Jun'ichi Nomura" , Milan Broz , Paul Jimenez , "Robert P. J. Day" , Vasily Averin , dm-devel@redhat.com, linux-kernel@vger.kernel.org Subject: Re: [git pull] device-mapper update for 2.6.25 Message-ID: <20080208032423.GX8075@agk.fab.redhat.com> Mail-Followup-To: Andrew Morton , Linus Torvalds , Brian Wood , Daniel Walker , Herbert Xu , Jonathan Brassow , Jun'ichi Nomura , Milan Broz , Paul Jimenez , "Robert P. J. Day" , Vasily Averin , dm-devel@redhat.com, linux-kernel@vger.kernel.org References: <20080208024532.GC24157@agk.fab.redhat.com> <20080207190447.b4dcfe0c.akpm@linux-foundation.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20080207190447.b4dcfe0c.akpm@linux-foundation.org> User-Agent: Mutt/1.4.1i Organization: Red Hat UK Ltd. Registered in England and Wales, number 03798903. Registered Office: Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE. Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Feb 07, 2008 at 07:04:47PM -0800, Andrew Morton wrote: > On Fri, 8 Feb 2008 02:45:32 +0000 Alasdair G Kergon wrote: > > > dm: move deferred bio flushing to workqueue > > To my unpracticed eye this one looks fairly fundamental. Unfortunately the > changelog: > > Add a single-thread workqueue for each mapped device and move flushing > of the lists of pushback and deferred bios to this new workqueue. > > tells us nothing of the reasons for this change. (It should have done so!) > > It _looks_ pretty unscalable. Perhaps these deferred BIOs don't happen > very often? Indeed - it's a rare occurrence - happens when the dm device is suspended/resumed which usually means it was being reconfigured by something in userspace. It's the first patch of a series that will provide barrier support (slated for 2.6.26). The appearance of all these workqueues in dm goes back to this old change: md-dm-reduce-stack-usage-with-stacked-block-devices.patch Alasdair -- agk@redhat.com