Re: G4 support?

["Brian C. Grayson" <bgrayson%marvin.ece.utexas.edu@localhost>]

On Wed, Jun 30, 1999 at 08:30:28PM +0200, Ignatios Souvatzis wrote:
> On Tue, Jun 29, 1999 at 11:25:04PM -0700, Erik E. Fair wrote:
> > I didn't even know the G4 was in production yet.

  Frankly, it may still be in pre-production/first-silicon, for
all I know.  Or in small samples or whatever.  I don't think it's
in full-blown crank'em-out mode.

> > If you've got some G4 systems with OpenBoot Firmware, why not try to
> > netboot a netbsd kernel on one and let us know how it blows up?

  I believe netbooting is out of the question:  by default, I
believe the G4s can not mount from other machines (policy), and
I do not have root on any machines.  The firewall would prevent
using an external NFS server.  And I don't know if we
have any spare disks handy.  I do have some old 127MB IDE disks
(non-UltraDMA) at home that I could bring in, if the IDE
controller won't choke on such ancient disks.

  I guess best bet would be a floppy-based test boot.  Can Mac
drives boot off of an IBM-format disk?  (Yes, I'm showing my
lack of Mac knowledge here.  But ask me anything about the
Apple ][ disk encoding and I can probably answer it!)  I just put
the 1.4 boot image on an IBM floppy, and will try it out on
Thursday if I get a chance and an idle G4.

  If that works, I can bring in two 127MB IDE disks, one blank, and
the other containing an FFS with the binary stuff (sets, etc.),
and do a sysinst install from disk1's files to disk0.  (I'd be
setting disk1 up from an x86 -- will "newfs -B be" do everything
I need?)

  If what I've written above would Not Work, and someone would
be willing to walk me through everything, or point me to the
appropriate docs (www.netbsd.org still seems down for me),
that'd be helpful.

> > Of course, if you can shake loose a G4 system
> > for a NetBSD developer to work with, that would be pretty cool too...

  I think a lot of the stuff is too much
proprietary/in-development to let it get into third-party
hands.  But I'm a NetBSD developer.  Just not a powerpc one, so
far.  :)  I'm pretty sure my boss would be perfectly happy with
me hacking in my spare time, and giving away resulting code, but
letting others play around on the boxes would probably be a
no-no.

> > The other thing coming down the pipe at us with the G4, as I understood it
> > from Apple at the WWDC in May is MP machines - assuming that the backside
> > cache bus is still there, how does the G4 handle cache coherency?

  I don't know -- most of my work at Moto is not on G4, so I
don't know much about it.  And even if I find out, I'm not sure
how much of that info is confidential at this point.  But...

> hopefully in a way which is supporting the PowerPC architecture definition,
> which means it will magically work if our code was correct from the beginning.

  ... I highly doubt that G4 would violate the spec, so yes, if our
code is correct, I expect things will work fine.  (One thing I've
noticed about the whole PPC thing is that it seems to match the
NetBSD philosophy of "Get it right the first time" -- witness
AltiVec's cleanness vs. MMX + SSE ugliness!)

  I may try to convince my manager to allow me to work on
NetBSD-powerpc bring-up for G4 (if I don't do it, I don't think
anyone else will until it ships, at what point one of y'all could
do it).  But as it is, I'm already backlogged with work, so I
don't know if I'd be able to do much if it didn't Just Work.

  Random question:  does XFree86 work on ppc?  I've never seen
anyone kick G3 or G4 boxes running Lin*x into X, and don't know
if there's a good reason.

  Many thanks in advance.

  Brian (speaking solely for myself, not Motorola or Intel for that matter!)