From mboxrd@z Thu Jan 1 00:00:00 1970 From: Michael Reed Subject: Re: [PATCH] OOPS due to clearing eh_action prior to aborting eh command Date: Wed, 07 Dec 2005 20:31:56 -0600 Message-ID: <43979B1C.60400@sgi.com> References: <43975A8C.2030208@sgi.com> <1133993204.3303.46.camel@mulgrave> <43975F24.4040904@sgi.com> <1133998497.3898.4.camel@mulgrave> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from omx2-ext.sgi.com ([192.48.171.19]:41682 "EHLO omx2.sgi.com") by vger.kernel.org with ESMTP id S932673AbVLHCcA (ORCPT ); Wed, 7 Dec 2005 21:32:00 -0500 In-Reply-To: <1133998497.3898.4.camel@mulgrave> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: James Bottomley Cc: linux-scsi@vger.kernel.org James Bottomley wrote: > On Wed, 2005-12-07 at 16:16 -0600, Michael Reed wrote: >> FWIW, the situation I'm reporting isn't done after timeout, it's >> the scsi error handler calling the LLDD abort routine, which actually >> aborts the command and completes it. The eh_action is cleared before >> the abort, violating what appears to be the accepted protocol of having >> the LLDD complete aborted commands. >> >> Am I missing something in your comment? (It wouldn't surprise me!) > > The timeout race still exists, though. Both fixes would solve both > problems. It would just be easier to follow if the logic of the error > handler done follows that of the normal done function (i.e. don't do > anything after the command times out). "Ah ha", he said. Further examination of the code clarifies what you were suggesting. I'll see what I can work up. Thanks, Mike