From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1750760AbVLBVid (ORCPT ); Fri, 2 Dec 2005 16:38:33 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750746AbVLBVic (ORCPT ); Fri, 2 Dec 2005 16:38:32 -0500 Received: from mail.kroah.org ([69.55.234.183]:31905 "EHLO perch.kroah.org") by vger.kernel.org with ESMTP id S1750760AbVLBVic (ORCPT ); Fri, 2 Dec 2005 16:38:32 -0500 Date: Fri, 2 Dec 2005 13:38:11 -0800 From: Greg KH To: Jeff Garzik Cc: Linux Kernel , Matthew Wilcox , Andi Kleen , tom.l.nguyen@intel.com Subject: Re: Status of PCI domain support? Message-ID: <20051202213811.GA3948@suse.de> References: <4390B752.90006@pobox.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4390B752.90006@pobox.com> User-Agent: Mutt/1.5.11 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Dec 02, 2005 at 04:06:26PM -0500, Jeff Garzik wrote: > > The lack of PCI domain support on x86-64 prevents me from seeing the > following devices on the PCI bus: > > 61:04.0 RAID bus controller: Silicon Image, Inc. Adaptec AAR-1210SA SATA > HostRAID Controller (rev 02) > 61:06.0 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X > Fusion-MPT Dual Ultra320 SCSI (rev 07) > 61:06.1 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X > Fusion-MPT Dual Ultra320 SCSI (rev 07) > > I get the following errors, unless I disable the BIOS option 'ACPI bus > segmentation', which is enabled by default (and will be enabled on most > future machines): > > Dec 2 10:40:18 localhost kernel: ACPI: PCI Root Bridge [PCI1] (0001:40) > Dec 2 10:40:18 localhost kernel: PCI: Multiple domains not supported > Dec 2 10:40:18 localhost kernel: ACPI: PCI Root Bridge [PCI2] (0002:80) > Dec 2 10:40:18 localhost kernel: PCI: Multiple domains not supported > > Full machine info was just posted in another message, subject > "[2.6.15-rc4] oops in acpiphp". > > Is this a bug? Is this expected behavior (not implemented yet)? A > jump-start on tracking this down would be appreciated. Based on the logic in arch/i386/pci/acpi.c (which is shared with x86-64), it's just not implemented yet at all. I don't know of anyone currently working on adding this support to this code, does anyone else? thanks, greg k-h