From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: Major qla2xxx regression on sparc64 Date: Wed, 18 Apr 2007 13:13:41 -0700 (PDT) Message-ID: <20070418.131341.129061984.davem@davemloft.net> References: <20070416200857.GH10822@andrew-vasquezs-computer.local> <20070418171346.GA16471@infradead.org> <20070418172802.GD12708@andrew-vasquezs-computer.local> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Return-path: Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:56639 "EHLO sunset.davemloft.net" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S2992490AbXDRUNi (ORCPT ); Wed, 18 Apr 2007 16:13:38 -0400 In-Reply-To: <20070418172802.GD12708@andrew-vasquezs-computer.local> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: andrew.vasquez@qlogic.com Cc: hch@infradead.org, linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, James.Bottomley@SteelEye.com, ema@debian.org From: Andrew Vasquez Date: Wed, 18 Apr 2007 10:28:02 -0700 > On Wed, 18 Apr 2007, Christoph Hellwig wrote: > > > I don't think a module option is a good idea at this point. The problem > > is you broke some so far perfectly working setups, which is not okay. > > The only first step can be printing a really big warning. After this > > has been in for a while (at lest half a year) we can make it a non-default > > option or turn if off completely in case the warning never triggered in > > practice. > > > > The only resonable thing for 2.6.21 is to put in David's patch, possible > > with an even more drastic warning when the rom is invalid and there's > > no prom-fallback available. > > > > Note that I expect Sun put in the invalid ROM intentionally, as we have > > similar cases with other cards that have totally messed up ROMs in > > Sun-branded versions. Personally I think that's an utterly bad decision > > from Sun's side, but we'll have to live with this. > > Fine. I'll rework an alternate patch for the 2.6.22 timeframe... We need to fix things now for 2.6.21 and the 2.6.x -stable branches because users have unusable systems currently. If it's just a time issue I can work on and push the patch, especially since I have the means to test things here.