Drew Northup wrote:
> > >just iding until a hardware clock runs out we don't
> > >want it to just be sitting there eating up cpu
> > cycles
> > >that could be used elsewhere--so why not trap "nop"
> > >and thus allow the clock cycles to be put to good
> > use
> > >while redirecting any wait states into a virtual
> > >hardware timer inside of the monitor?
> >
> > You're confused with HLT, here.
> Hmmmmmm, ok. Back to the manual........
No problem. BTW, this is exactly what we can
do with the HLT instruction, to recover CPU resources
when the guest is spinning its wheels. Also
with jmp -2. Or for that matter, we could detect
known calls or addresses and do the same. (guest
specific).
I'd like to be able to leave a VM running, and
dynamically the resources get returned to the
host when nothing is going on. Saves on
boot time. :^)
-Kevin
