Re: [0/4] NUMA-Q: remove PCI bus number mangling

[Greg KH]

On Tue, Nov 12, 2002 at 03:58:24PM -0800, William Lee Irwin III wrote:
> On Tue, Nov 12, 2002 at 03:53:49PM -0800, Martin J. Bligh wrote:
> >> Right, I'm not against the sysdata thing, seems like a much better way
> >> to do it in general (what I did was a quick hack). Was just confused
> >> by the global bus number assertion, but if we use the sysdata stuff,
> >> it's all a non-issue ;-)
> > 
> On Tue, Nov 12, 2002 at 02:59:37PM -0800, William Lee Irwin III wrote:
> > Non-issue for merging...
> > The pain isn't over yet. =(
> > Core PCI code is assuming unique bus numbers in several places.
> 
> Okay, an attempt to remedy this world-breaking braindamage with the
> fewest lines of code:
> 
> This alters PCI bus number "clash" detection to compare ->sysdata in
> addition to the numbers. The bus number is not a unique identifier.

Um, why not?

And what would /sys/bus/pci/devices look like if you allow the same
identifiers for different devices?  :)

thanks,

greg k-h