From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: RE: [PATCH ] scsi-misc-2.6: File System going into read-only mode Date: Fri, 13 Nov 2009 14:55:22 -0600 Message-ID: <1258145722.5965.55.camel@mulgrave.site> References: <4F2B1A2459C7AD4D96A23CE911C352CB1E9150B4EE@CHN-HCLT-EVS07.HCLT.CORP.HCL.IN> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: Received: from cantor.suse.de ([195.135.220.2]:35902 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932271AbZKMUzY (ORCPT ); Fri, 13 Nov 2009 15:55:24 -0500 In-Reply-To: <4F2B1A2459C7AD4D96A23CE911C352CB1E9150B4EE@CHN-HCLT-EVS07.HCLT.CORP.HCL.IN> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: "Penchala Narasimha Reddy Chilakala, ERS-HCLTech" Cc: "'linux-scsi@vger.kernel.org'" , James Bottomley , ServeRAID Driver On Thu, 2009-11-12 at 13:58 +0530, Penchala Narasimha Reddy Chilakala, ERS-HCLTech wrote: > Hi James, > > Thank you very much for your suggestion. > > I feel that we need to work out for a better solution which would not > create any issues. We have to go ahead with the patch as it is tested > and working as expected (with out any issues) even though we are > thinking there might be some racy issue until we come out with a > better solution than the current one Sure, I can wait for a better solution. > I am not quite ok with the fix because of the following reasons: > > 1. I feel that calling the dev->management_fib_count-- > statement in one place, that is in dpcsup.c (Whenever interrupt comes > from arcraid controller, the corresponding ISR route will be called > and the value will be decremented, if the response is for the > requested management fib) is good instead of calling the statement in > two places, which are dpcsup.c and commctrl.c. Um, so that's the same mechanism you implemented for this undone fib problem: setting the ->done flag as mediator. > 2. Moving the dev->management_fib_count++ from aac_fib_send () > function to aac_do_ioctl () function did not work well during our > testing (This was suggested by you in one of the earlier mails). That doesn't give me a lot to go on. However, I'm not particularly bothered whose solution it is, just that we have a solution that isn't demonstrably racy. James