From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261220AbVFDBxw (ORCPT ); Fri, 3 Jun 2005 21:53:52 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261221AbVFDBxw (ORCPT ); Fri, 3 Jun 2005 21:53:52 -0400 Received: from fire.osdl.org ([65.172.181.4]:62095 "EHLO smtp.osdl.org") by vger.kernel.org with ESMTP id S261220AbVFDBxu (ORCPT ); Fri, 3 Jun 2005 21:53:50 -0400 Date: Fri, 3 Jun 2005 18:55:40 -0700 (PDT) From: Linus Torvalds To: Andreas Koch cc: linux-pci@atrey.karlin.mff.cuni.cz, linux-kernel@vger.kernel.org, gregkh@suse.de Subject: Re: PROBLEM: Devices behind PCI Express-to-PCI bridge not mapped In-Reply-To: <20050604013311.GA30151@erebor.esa.informatik.tu-darmstadt.de> Message-ID: References: <20050603232828.GA29860@erebor.esa.informatik.tu-darmstadt.de> <20050604013311.GA30151@erebor.esa.informatik.tu-darmstadt.de> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sat, 4 Jun 2005, Andreas Koch wrote: > > As you suspected, it wasn't a panacea: The kernel now panics, with a > call chain of > > ... > pcibios_init() > pci_assign_unassigned_resources() > pci_bus_assign_resources() > pci_setup_bridge() > > I can collect more specific info if necessary. It would be nice to know exactly what it is that panics, I could well imagine that it's something like the "bus->self" that ends up being NULL for the root bus or something silly like that, simply because x86 has never needed to use these functions. If so, it migth be as easy as just skipping buses that don't have bridge device associated with them, but this would require that you try to debug the oops a bit to figure out where it is.. Greg, do you have any PCI Express hw? Although I suspect that the pci_assign_unassigned_resources() problem probably happens on any PC, I could try it on my laptop. Linus