On 4 Sep 1998, Rasmus Wiman wrote:

> >1.) The toolchain (gcc, as, ld, etc.) shipped with NetBSD Amiga knows how
> >    to create a.out m68k binaries. It doesn't know anything about other
> >    CPUs or file formats (PPC uses ELF).
> 
> Since a PPCP (or whatever that platform is called) port is in the works, the
> tools (some of them at least) should already be there. But this was one part I
> wasn't sure of. I have been thinking of compiling an i386 kernel the day I
> have a HCN (Home Computer Network, I just love inventing silly TLA:s ;) )
> instead of downloading everything. Now I realise it just won't work.
> 
	You can make a powerpc/elf crosscompiler, and in fact NetBSD is in
	the process of switching to egcs (though there are some issues
	with m68k cpus), and it will make crosscompiling even easier.

	We currently have three powerpc ports in the tree, OpenFirmware
	based machines, Power macintoshes, and beboxes, so there is
	certainly some code that could be used for an amigappc port :)

> >2.) Even if it would now it wouldn't help because as the Amiga kernel
> >    sources are really CPU specific, a small part of them is even written
> >    in assembler.
> 
> Is everything CPU specific, or is most of it "just" machine specific? 
> 
	It varies, wherever possible code is made machine independent, and
	between the m68k amiga and other powerpc ports there should be
	a fair amount you could reuse.

> >To really get NetBSD running on a PPC board you will need to create a
> >*complete* new NetBSD port - call it "amippc" - based on the generic
> >PPC architecture.
> 
> How different is a PPC Amiga from a normal one? Is it just an Amiga with a
> totally different CPU (I know how the PPC instruction set looks like, I once
> read the 601 reference manual briefly when I saw it in a library) or is it to
> be considered a completely new machine? 
> 
	I'm afraid I can't help on this one, never having seen a ppc amiga
	:)

> >Afterwards you'll have to rework all the Zorro II/III drivers to be machine
> >indepent because the CPUs are quite different.
> 
> I thought some of the work was already done, but that may be wrong. My
> thought was that the main thing that changes is anything written in
> assembler or MMU dependent, and C functions called with explicitly
> registerized parameters. Of course, things that rely on the CPU for
> timing would have to be rewritten, but I was unsure if anyone had
> wriiten such code on the Amiga since F/A 18 Interceptor. ;)
> 
	Its certainly a very nice project for someone willing to spend the
	time and effort to work on! :)

		David/absolute

	  -=-  Maybe your misinterpretation of my actions is 
	      in conflict with your misconception of who I am  -=-