Ramon van Handel wrote:
>
> Kevin Lawton wrote:
> > Nevertheless, there's something that we need to be aware for
> > _any_ technique which doesn't defer to monitor (ring0) space to handle
> > virtualization.
>
> I want to run prescanned code in ring0, actually.
> I don't quite see the problems here (it requires
> to be more careful with page permissions, that's
> all.) Am I missing something?
Yeah, it can not be our end-all architecture, because running at
ring0 is quite dangerous. We need something that works reasonably
well at ring3, with perhaps a config option that changes behaviour
to work at ring0 for the adventurous types.
Admin types will want a way to turn off (perhaps compile out) the
ability to use ring0, so they don't expose themselves to huge
security risk or computer downtimes from crashes. But, while
still having reasonable performance.
I'm working on typing up my thoughts on a quasi dynamic translation
strategy, now. I think this should give us the boost we need, along
with many other general monitor optimizations we haven't yet done.
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Kevin Lawton [EMAIL PROTECTED]
MandrakeSoft, Inc. Plex86 developer
http://www.linux-mandrake.com/ http://www.plex86.org/
