> JT> options		PCI_CONF_MODE=x
> 
> No dice, it just blows more chunks than it did before.
> 
> However, the following hack to mainbus.c coupled with a config line
> 	pci1 at mainbus0 bus 1
> did get it to probe the AHC devices (2 controllers) and detect the disk
> drives.  So, it does seem to have a second primary PCI bus.

That isn't a second primary PCI bus, that's a secondary PCI bus
configured via one of the system chipset chips, rather than a plain
old PPB.

The folks at Cal Tech ran into a similar problem on their server.

I don't have your dmesg output handy (did you send it?), but
undoubtedly it included at least one PCI Host Bus Bridge.  It could be
that a driver for that devices is where the logic to do the secondary
setup belongs, or it could be another device in the system.
Unfortunately, I was never able to find datasheets on the various
chipset chips in the Cal Tech folks' machine.  Hopefully you'll have
better lucking finding datasheets for yours.


> Can someone suggest a more correct way to probe multiple primary PCI buses?

Again, this is _not_ a second primary PCI bus.  That's a relatively
well-defined term (meaning: a PCI bus with bus number zero), and this
isn't one of them.


chris