> I think of OpenBoot (among other things) as a mechanism that allows a
> kernel to access a device in the absence of a native device driver
> (perhaphs I'm wrong).  When looking at devices for a new target, going
> down the path of implemting the open boot interface (assuming the rest
> was there) would be easier then initially worrying about native devices.

As far as I know that isn't the case, because it probably doesn't take
interrupts into account.  I believe the Sparc OpenBOOT device drivers
are all polled.