RFC:new superio proposal

[Eric W. Biederman]

Bari Ari <bari at onelabs.com> writes:

> ron minnich wrote:
> 
> >On Tue, 13 May 2003, Bari Ari wrote:
> >
> >
> >> The new super i/o structure of freebios2 should also account for these
> >> changes.
> >>
> >
> > any suggestions on how this should look? I can see a
> > src/superio/<vendor>/microprocessor, to which you pass all these parameters,
> > and which talks to the microprocessor to set them up. Would that be
> > sufficient? What new parameters would it need?
> >
> >
> Besides just setting a few bits in registers to config the super i/o, one now
> has the option/ability to reload the binary firmware into the flash memory of
> the super i/o to customize the board. The binary may be 64KB for some cpu based
> super i/o's.
> 
> Some examples of cpu based super i/o's
> 
> http://www.renesas.com/eng/products/mpumcu/16bit/h8s/2100/index.html

Actually if you look up the original keyboard controller it was
also a cpu.  Though a very limited one.

If you look you can find cpus in some of the strangest places.

As for the configuration process it is largely irrelevant.  The
important infrastructure piece is the call graph and some small hooks
to allow the same drivers to be used in different hardware
configurations.

Once your driver gets control it can do whatever makes sense.
And that 64K blob I suspect is something you can just setup
to be compiled in.  If that is a reasonable way to do things.

Eric