On Sat, 25 Jul 1998, Tsubai Masanari wrote:

> >I think all those characters are coming out of the xcoff loader, not the
> >linuxppc booter.
> 
> I understand.
> (but the machine I tested boots...  it has 384MB memory, OF-2.0f1)

??? Weird. I thought I'd run into the "64 MB" problem the linuxppc folks
have seen. But if you've booted a G3 with 384 MB, then the problem's
elsewhere. What kind of machine was it? Maybe there's a bug in one of the
drivers which isn't tickled on your system but is on mine.

> >But they both die with a DEFAULT CATCH code @ FFF03000 error (the %SRR0
> >and %SRR1 numbers differ). Or is that the address of the catcher, not the
> >code which caused the problem?
> 
> ...0300 (I think 0300, not 3000) means DSI trap (failed to translate data
> address).  SRR0 is the address.

Ahhhh. Then this error is not exactly the linuxppc one. It is for a
different address (SRR0 value). Is this basically an MMU fault?

> >Also, I can configure a kernel with any pci device, can't I? I'm thinking
> >of getting a 10/100 MB Ethernet card, and I think NetBSD already supports
> >it.
> 
> Yes, perhaps you can if you make a kernel from the newest sources.  I
> tested "ne" and "ep" (3c905).

Hmmm. I think I'll look and see how much PCI SCSI controllers are. :-)
Then I'd get around the no-ide-or-mesh problem. :-) Or I might just finish
my dissertation and then not worry about it. :-)

Take care,

Bill