der Mouse wrote:
> What problem?  What MI need does monitoring or varying the speed of a
> cpu - the same or another - address?

As I understand it, the ability to vary the speed of a CPU is a
desirable feature regardless of the architecture, to be provided where
the architecture supports it. i386 is by no means the only architecture
which supports changing the CPU frequency and voltage, and if we are to
provide such features we should not have to reinvent the wheel for every
architecture which has that capability.

> On at least some MP platforms, also, the answer is "you don't".  I'd
> even hazard a guess that that's the answer for most of them, though
> admittedly that is a guess.  Certainly on the MicroVAX-II that's the
> answer, and I think it's the answer on at least some sun4m SPARCs.

I agree that on most platforms the answer is likely to be "you don't".
But that's not the point. There's benefit to be had from implementing
this in a generic MI way if at least two ports support this. I would
even go as far as to say that the benefit is still to be had even if
only one ports supports it, and there is a reasonable expectation that
the future will bring another one.

Obviously the implementation of the API for ports which don't support
the feature will be trivial.

-- 
Gillette - the best a man can forget