From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jesse Barnes Subject: Re: SCSI QLA not working on latest *-mm SN2 Date: Wed, 22 Sep 2004 17:33:03 -0400 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <200409221733.03938.jbarnes@engr.sgi.com> References: <20040921225405.GC147518@sgi.com> <200409211917.31817.jbarnes@engr.sgi.com> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from omx3-ext.sgi.com ([192.48.171.20]:44996 "EHLO omx3.sgi.com") by vger.kernel.org with ESMTP id S267508AbUIVVd2 (ORCPT ); Wed, 22 Sep 2004 17:33:28 -0400 In-Reply-To: <200409211917.31817.jbarnes@engr.sgi.com> Content-Disposition: inline List-Id: linux-scsi@vger.kernel.org To: Jeremy Higdon Cc: Andrew Vasquez , Matthew Wilcox , James Bottomley , Grant Grundler , pj@sgi.com, SCSI Mailing List , mdr@cthulhu.engr.sgi.com, jeremy@cthulhu.engr.sgi.com, djh@cthulhu.engr.sgi.com, Andrew Morton On Tuesday, September 21, 2004 7:17 pm, Jesse Barnes wrote: > On Tuesday, September 21, 2004 6:54 pm, Jeremy Higdon wrote: > > On Tue, Sep 21, 2004 at 06:43:49PM -0400, Jesse Barnes wrote: > > > On Tuesday, September 21, 2004 6:39 pm, Jeremy Higdon wrote: > > > > On Tue, Sep 21, 2004 at 06:36:36PM -0400, Jesse Barnes wrote: > > > > > Shouldn't we do a config space read and *then* start the delay, > > > > > which will only delay for as long as it takes to reset the card on > > > > > a 33 MHz bus? > > > > > > > > The config space read has the same problem that the mmio space read > > > > does. The chip does not respond. > > > > > > Right, but config space reads are supposed to soft fail no matter what. > > > So we'll get back all ones, but we'll also know that the write has been > > > received by the device. We can start the delay regardless of the value > > > the read returns. > > > > I thought you said that config space read failures generate an MCA. > > Or are you going to fix that? > > That's the plan. I'll come up with a patch. On second thought Andrew (Vasquez), can not add config space reads quite yet (just stick with generous udelay() calls)? It'll probably take me a couple of days to put together a patch that fixes config space reads on sn2, and I'd rather not have qla2xxx cause MCAs in the meantime. Thanks, Jesse