Bill,
	I know I just sent this to you, but I've been meaning to repost this to 
lkml anyway.  Here's something I wanted to add to the generic topology 
infrastructure for a while.  pcibus_to_node()

Have a look at this patch, and see if it might be useful to you, ok?

Cheers!

-Matt

William Lee Irwin III wrote:
> On Tue, Nov 12, 2002 at 04:37:46AM -0800, William Lee Irwin III wrote:
> 
>>This fixes a longstanding bug with respect to bridge handling as well as
>>a Linux PCI faux pas, namely an attempt to support PCI domains with bus
>>number mangling.
>>The end result is that bridges off of quad 0 now work, and the code now
>>follows Linux PCI conventions.
>>[1/4] NUMA-Q: use sysdata as quad numbers in pci_scan_bus()"
>>[2/4] NUMA-Q: fetch quad numbers from struct pci_bus"
>>[3/4] NUMA-Q: use quad numbers passed to low-level config cycles"
>>[4/4] NUMA-Q: remove last traces of bus number mangling"
> 
> 
> Follow on #1:
> 
> [5/4] NUMA-Q: use "quad" instead of node in arch/i386/pci/numa.c
> 
> 
> This renames all nonessential uses of the word "quad" in
> arch/i386/pci/numa.c with "node".
> 
>  numa.c |   24 ++++++++++++------------
>  1 files changed, 12 insertions(+), 12 deletions(-)
 >
 > <snip>
 >