Hi Chris,

> 	They were showing up as DNS not resolving errors from my mail
> server, but I could resolve it fine from home so I don't know who's
> malfunction it is.

John Klos has stepped up and we're working out an arrangement for me to
move to his hosting service. Many thanks, John!

> > Could you tell if it was related? Can you send a bt from ddb?
> 
> 	Probably has to do with running a current kernel on a RC4
> userland.

I have been assured beyond all doubt that if this is the case, this is a
bug, period. -current kernels can run an earlier userland, but -current
userlands have to have -current kernels.

> 	The other macio only seems to have one channel enabled and it's
> the internal modem (if installed)

Ok, makes sense.

> 	If it means anything, there are physically two Heathrow chips on
> the motherboard that are actually labeled Heathrow1 and Heathrow2. I
> grabbed a Mainstreet motherboard and scanned it for you :) Grackle and
> the ROM are on the CPU daughtercard.
> 
> http://user.dtcc.edu/~ctribo/mainstreet_mb.jpg

Actually, this picture is great. I wasn't sure it was two separate I/O
controllers or one with two PCI device IDs. Both chips have the same PCI
vendor and product ID, but their "compatible" property in Open Firmware
is different. The primary I/O is listed as "Heathrow" compatible and
does not contain any interrupt information. The secondary I/O, for the
internal modem and the mediabay, is listed as "Gatwick" and contains an
interrupt. No device under Gatwick has an interrupt property.

> 	There's a lot of info about the non-Powerbook implementation of
> Mac-IO in here:
> 
> http://mail-index.netbsd.org/port-macppc/2004/11/12/0003.html

You know, Michael mentioned it in reference to mesh, so I didn't really
associate it with this. I've downloaded it and I will review it. Thanks
for bringing this to my attention!

tim