Re: cvs for powermac work

[Bill Studenmund <skippy%macro.stanford.edu@localhost>]

On Thu, 5 Feb 1998, SUNAGAWA Keiki wrote:

> Dan Jacobowitz <youngdrow%geocities.com@localhost>  wrote:
> 
> >> Kazuki Sakamoto and I talked about it yesterday, and got
> >> same conclusion.  We also thought that dev/powerpc should
> >> contains processor specific code like m68k or mips,
> >> dev/{bebox,powermac,etc.} should contain architecture
> >> specific ones.

I agree that this is should be an eventual goal. I like the idea Wolfgang
mentioned of sys/cpu/CPU_NAME and sys/arch/COMPUTER_NAME. Mainly as we
still need a home for the Open-Firmware PPC port.

> Dan> Thoughts from the list?
> 
> Not only.  Working on these weeks, I thought that though
> Apple changed the bus architecture from NuBus to PCI, they
> didn't the other hardware disign concept much at the time of
> Macintosh Quadra or so.

Right. I've poored over the CHRP specs, and the MacIO chip (whatever it
was they called it) is the same as the 8500 I/O chip, and I think is the
same as the 7100 I/O device, except that the latter's NuBus.

Even if the 7100 is not the same, it's other influence would be the Quadra
840 & 660, both of which port-mac68k supports.

So CHRP should give us a good idea of what the I/O's like on PCI macs, and
NuBus PowerMacs can either learn from the PCI ones, or from the last 68k
macs. Aslo, I think our mkLinux friends know a lot about the 7100's. :-)

> Dan> I personally think this would be a WONDERFUL idea, but
> Dan> perhaps too much work.  Any beboxers with a comment on
> Dan> that?
> 
> Kazuki is bebox's port master.
> 
> Dan> I think it's an idea to hold on to, but perhaps to
> Dan> worry about later on.
> 
> I agree.  It's no urgent issue, something in the future.

It certainly can wait, but isn't it just a matter of moving files in the
CVS repository, and changing all the #include's to point to the right
place?

Take care,

Bill