> > For instance, am I correct in assuming that the function ofwr_init()
> > exists to get around a buggy firmware implementation that fails to
> > ``save and restore the first location of each interrupt that it wants to
> > "take over".' [1]'?
> 
> I'll answer my own question :-)
> 
> 	No, it also gets around the lack of support for real mode openfirmware.

Yes, this is the main reason for this ofwr_init stuff.  It saves the complete
state of the MMU immediately after the boot, so it can switch back to that
state on every call to any OpenFirmware routine.

> Can I suggest that the two be separated.  The second (save of the
> interrupt vectors) being done *after* VM has been turned off.

Note that this is ONLY done for the MMU trap vectors.  I imagine this as part
of the MMU state, so I cannot see any advantage in separating these two.
--
ws@TooLs.DE     (Wolfgang Solfrank, TooLs GmbH) 	+49-228-985800