tls@rek.tjls.com said:
> It could be some oddity of the IBM bridge and the BIOS not knowing how
> to configure it, but I'd think modern x86 BIOSes could cope... 

If the IBM bridges are still what they were about 2 yaers ago - they
are really strange. There was a problem with the granularity of the
address ranges passed to the subordinate buses -- I don't remember
the details, but I've seen different BIOSes despare about it.
And: even if the BIOS managed to deal with the configuration the
bridges locked up the bus from time to time if there was
concurrent access from both sides.
Ordering issues on PCI are a hairy matter (and got some refinement
in newer PCI specs), so I'd really make sure to get an up-to date
PCI-PCI bridge into my system.

best regards
Matthias