From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756967AbZEMFwf (ORCPT ); Wed, 13 May 2009 01:52:35 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755891AbZEMFwS (ORCPT ); Wed, 13 May 2009 01:52:18 -0400 Received: from hera.kernel.org ([140.211.167.34]:49954 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756139AbZEMFwR (ORCPT ); Wed, 13 May 2009 01:52:17 -0400 Message-ID: <4A0A6003.8060800@kernel.org> Date: Wed, 13 May 2009 14:52:03 +0900 From: Tejun Heo User-Agent: Thunderbird 2.0.0.19 (X11/20081227) MIME-Version: 1.0 To: Jens Axboe CC: Kiyoshi Ueda , Boaz Harrosh , FUJITA Tomonori , linux-kernel@vger.kernel.org, benh@kernel.crashing.org Subject: Re: [PATCH] swim3: use blk_end_request instead of blk_update_request References: <1242127787-29842-1-git-send-email-fujita.tomonori@lab.ntt.co.jp> <1242127787-29842-3-git-send-email-fujita.tomonori@lab.ntt.co.jp> <4A098B8D.10107@panasas.com> <4A099562.7020805@kernel.org> <4A0A1941.2010105@ct.jp.nec.com> <20090513054901.GW4140@kernel.dk> In-Reply-To: <20090513054901.GW4140@kernel.dk> X-Enigmail-Version: 0.95.7 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.0 (hera.kernel.org [127.0.0.1]); Wed, 13 May 2009 05:52:06 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Jens Axboe wrote: >>> Yeah, conversion was easy that way. I think it's sane to have >>> blk_update_request() exported as long as request internal tinkering is >>> kept in block layer proper. >> blk_update_request() is needed for request-based dm to keep the request >> completion ordering in bottom-up, although request-based dm is not >> in upstream yet. >> >> Jens, please keep blk_update_request() exported. > > I did, I applied the swim3 patches yesterday as well. I don't think the patch is correct. If it calls blk_end_request_all(), it should also clear the current request which the patch doesn't. Also, given that the driver doesn't support partially failing the request, I think it's correct to fail segment-by-segment to avoid merged request failure affects unrelated bios. Thanks. -- tejun