>   I'd heard that PCI officially specs some variation of the Sun
> sbus forth as official byte code for device BIOSes. I do not know how
> this and I20 interacts.

The spec you are probably meaning is IEEE-1275, the "IEEE Standard for Boot
(Initialization Configuration) Firmware: Core Requirements and Practices".
also known as OpenFirmware.

This is indeed a slight modification of Sun's OpenBoot firmware.  The idea
is to place some byte code in the ROMs on add-in cards so they can be used
on different processors.

Now the PCI spec allows for inclusion of code for different processors in
one ROM on PCI cards.  It explicitly specifies two values for the processor
type field in the header, namely x86 and OpenFirmware code.

And no, I don't know either how this may or may not interfere with I2O.

>   Back to serial device stuff: I would welcome pointers to work done
> to rethink the serial device stuff. I read the plan9 papers a long
> time ago, I'm going to reread some of that stuff, and the v8 streams
> stuff if I can find it. 

I've been wondering for a while what the reasoning behind the default mode
of cooked for opening serial lines is...   (Yes, I know that there once
were reasons for that, but theose are long gone, not?).
--
ws@TooLs.DE     (Wolfgang Solfrank, TooLs GmbH) 	+49-228-985800