As Chris G. Demetriou wrote...
> 
> > > If there's no support for a system, a kernel won't run on it (for very
> > > long at least; it'll print out a message that the system is
> > > unsupported, then halt).  You can't really do much more than that.
> > > "Support" for a system includes all of the low-level code which
> > > controls how it's I/O bus is attached, how interrupts are configured,
> > > how the console is configured, etc.
> > 
> > But are all machines completely different in this respect, or is there
> > some common ground? I have not studied this subject, mind you, but I'm
> > just curious.
> 
> I've yet to see a machine with a one systype that was "sufficiently
> close" to any machine of another systype that the same code would come

OK, so I was just too optimistic. I was thinking of e.g. a 2100
versus a 2100A (A shifted from EISA -> PCI slots). 

> close to functioning correctly on it.  (Where "close to functioning
> correctly" means "gets as far as finishing autoconfig or to a single
> user prompt.")
> 
> Some parts of the code are in fact the same, e.g. all machines with
> 2117[12] chipsets can use the same 'cia' code, etc.  However, even
> among workstation models, the interrupt glue in particular seems very
> prone to change from machine to machine.  The different families of
> server machines seem to consistently have different I/O busses, too.

Ah, that's DEC like I know 'm ;-)

_     ____________________________________________________________________
 |   / o / /  _  Bulte  email: wilko@yedi.iaf.nl http://www.tcja.nl/~wilko
 |/|/ / / /( (_) Arnhem, The Netherlands - Do, or do not. There is no 'try'
------------------  Support your local daemons: run FreeBSD Unix  -----Yoda