On Mon, 15 Mar 1999, Neil A. Carson wrote:

> basic ASID support, but it's too much of a bodge to be useful. My
> original discussion was meant to be CPU-indepenadant though :)

In that case, it's not really CPU-independent because depending on the
cache architecture you might need to handle writebacks to pages that are
being invalidated, possible copyin/copyout to address spaces of different 
processes, calling PROM routines that may overlap either the user address
space or the kernel address space, or even setups where the kernel address
space overlaps user address space.  Different architectures need to deal
with different aspects of these issues, and although individual solutions
are possible, I don't think you can really come up with a single generic
solution for all possible machines, let alone the ones that are currently
supported.

=========================================================================
Eduardo Horvath				eeh@one-o.com
	"I need to find a pithy new quote." -- me