FYI: rudimentary loader for ppcbug-based systems

[Peter Grehan]

Hi Rafal,

> What do you think is the right way of adding support in the kernel for another firmware 
> but within the same architecture (powerpc in this case)? Should we just have a number of 
> #ifdefs where appropriate, or maybe we should think about separating some bits and defining
> clean interfaces to common parts? Eventually I intend to let my work be merged with existing
> powerpc code in the tree and that's why I'd like to know what course to take. For now I'll 
> just be using conditional compilation flags.

 Clean interfaces would be the best. The ultimate is a single kernel that could
run on all combinations of processor and firmware, the true GENERIC, but I'm not 
sure if that's possible.

 Here's a quick summary of the OFW dependencies in powerpc/powerpc/* -

locore.S - the entry point assumes the OFW calling convention

clock.c - the time-of-day and CPU's timebase frequency are obtained from OFW

machdep.c - the shutdown/halt routines call into OFW.

nexus.c - heavily OFW dependent. Any other platform will require their own version
	of this.

openpic.c - The OFW code could be split out of this to allow the OpenPIC code to
	be used on other platforms

pmap.c - Not as bad as it looks. The setting up of the BAT array is specific to
 	the h/w layout of Macs. OFW is used to determine the memory layout, and
	the PTE entries for OFW are not required on non-OFW systems.

later,

Peter.