From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263104AbUCXIae (ORCPT ); Wed, 24 Mar 2004 03:30:34 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263105AbUCXIae (ORCPT ); Wed, 24 Mar 2004 03:30:34 -0500 Received: from ns.virtualhost.dk ([195.184.98.160]:41601 "EHLO virtualhost.dk") by vger.kernel.org with ESMTP id S263104AbUCXI3s (ORCPT ); Wed, 24 Mar 2004 03:29:48 -0500 Date: Wed, 24 Mar 2004 09:29:46 +0100 From: Jens Axboe To: Mike Christie Cc: Linux Kernel Subject: Re: [PATCH] catch error when completing bio pairs Message-ID: <20040324082945.GF3377@suse.de> References: <405FE501.6030704@cs.wisc.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <405FE501.6030704@cs.wisc.edu> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Mar 22 2004, Mike Christie wrote: > A couple of drivers can sometimes fail the first > segments in a bio then requeue the rest of the request. In this > situation, if the last part of the bio completes successfully > bio_pair_end_* will miss that the beginging of the bio had > failed becuase they just return one when bi_size is not yet > zero. The attached patch moves the error value test before > the bi_size to catch the above case. Yup that's a bug, patch is correct. -- Jens Axboe