From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1425333AbcBRITf (ORCPT ); Thu, 18 Feb 2016 03:19:35 -0500 Received: from mx2.suse.de ([195.135.220.15]:39684 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1424759AbcBRITe (ORCPT ); Thu, 18 Feb 2016 03:19:34 -0500 Date: Thu, 18 Feb 2016 09:19:30 +0100 From: Johannes Thumshirn To: Sudip Mukherjee Cc: One Thousand Gnomes , Adaptec OEM Raid Solutions , "Martin K. Petersen" , linux-kernel@vger.kernel.org, linux-scsi@vger.kernel.org Subject: Re: [PATCH v2] [SCSI] dpt_i2o: use proper pci driver Message-ID: <20160218081930.GM18134@c203.arch.suse.de> References: <1455711614-5185-1-git-send-email-sudipm.mukherjee@gmail.com> <20160217142732.6dcad24c@lxorguk.ukuu.org.uk> <56C496AB.2070200@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <56C496AB.2070200@gmail.com> User-Agent: Mutt/1.5.24 (2015-08-30) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Feb 17, 2016 at 09:20:03PM +0530, Sudip Mukherjee wrote: > On Wednesday 17 February 2016 07:57 PM, One Thousand Gnomes wrote: > >On Wed, 17 Feb 2016 17:50:14 +0530 > >Sudip Mukherjee wrote: > > > >>This is a pci device but was not done in the usual way a pci driver is > >>done. Convert the driver into a proper pci driver. > > > >This looks completely wrong. Please read the I2O 1.5 specification > >document before playing with that code. You can't simply convert it to a > >standard PCI driver as all the IOPs are supposed to be detected and then > >the correct initialization sequence executed across the set of IOPs. IOPs > >are allowed to talk to one another and the system table binds it all > >together. > > Yes, thanks for letting me know about the spec. It is saying > "The host locates each IOP, adds it > to the system configuration table, and initializes the IOP’s outbound > message queue. The host then > provides each IOP with a list of all IOPs and the physical location of their > inbound message queue." > > > > >If you do hot plug then you need to follow the specification and do > >all the extra notifications. Unless you've got multiple FC909/FC920 > >fibechannel cards or similar to test with I would just leave this well > >alone. > > point noted. > > >Your original simple patch is *MUCH* safer in this specific case. > > The original patch is already having NACK from Johannes. So i better leave > this code here. Nah, wasn't a hard NACK. I'm OK with the original code, just was reluctant to the #ifdefs. I agree I didn't know the I2O specifics and thought a "normal" PCI driver fitting nicely in the driver model would be the way to go. Please resend your v1 with my Reviewed-by > > regards > sudip -- Johannes Thumshirn Storage jthumshirn@suse.de +49 911 74053 689 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) Key fingerprint = EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850