From mboxrd@z Thu Jan 1 00:00:00 1970 From: Matthew Wilcox Subject: Re: [PATCH]: PCI Error Recovery: Symbios SCSI device driver Date: Wed, 1 Nov 2006 21:56:34 -0700 Message-ID: <20061102045634.GB31830@parisc-linux.org> References: <20061020180510.GN6537@austin.ibm.com> <20061031185506.GE26964@parisc-linux.org> <20061031231334.GR6360@austin.ibm.com> <20061102044633.GB23840@colo.lackof.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20061102044633.GB23840@colo.lackof.org> Sender: linux-kernel-owner@vger.kernel.org To: Grant Grundler Cc: Linas Vepstas , linux-scsi@vger.kernel.org, linux-pci@atrey.karlin.mff.cuni.cz, linux-kernel@vger.kernel.org List-Id: linux-scsi@vger.kernel.org On Wed, Nov 01, 2006 at 09:46:33PM -0700, Grant Grundler wrote: > ISTR some chipsets return 0 or the most recent data on the bus > when INB/INW master-abort. Maybe this an ISA bus behavior? It's not a master-abort; it won't get as far as the device. Documentation/pci-error-recovery.txt says reads return 0xffffffff. > Or is config space access the only space which behaves this way > for master abort on PCI? > I'm looking at drivers/pci/probe.c:pci_scan_device(). As the comment says, the boards which do this are broken. It's highly unlikely those boards will support error isolation and recovery ;-)