Still trying...

(And for those of you who wonder, no, I did not expect the following to boot)

I stuck 'netbsd' at the base of a linux-pmac source tree and named it
vmlinux, and then made vmlinux.coff with what linux uses as a boot loader.
This time, the bootloader ran, the bootloader printed its output to
console, it (I think) actually jumped to the entry point of the kernel -
and then died, but this time with 0xFFF00300 - Data Storage Interrupt.

I have a few more ideas to try to isolate the problem - a couple of pause()
calls to see how far we're getting, etc. - but I am guessing there is
something fundamentally different about the boot code.  I'll make a nice
little .tgz of linux-pmac/arch/ppc and put it up on the web for anyone who
wants to take a look at it.

Apparently, linuxppc does indeed use OF to boot - as nearly as I can tell,
at least.  I also have a feeling I'm miscompiling ofwboot.xcoff for some
reason - and what is boot.ppc that people keep mentioning?

The linuxppc boot code has frankly confused the heck out of me, combined
with my knowledge of OF.  What values are assumed to be in which registers
when the boot code is called?  Also, if anyone can figure this out staring
at linuxppc, why does it seem to me that when coffboot() calls (*sa)(a1,
a2, prom) the routine that gets invoked is start(something, something,
(*openfirmware)(), something, something)?  Are those two parameters
appearing in the assembly and somehow having me miss them, or do I have the
entry points wrong?

Etc.


Any goodies or badies I produce will end up in http://www.drow.net/netbsd .

Dan

---------------------------------------------------------------------
| Dan Jacobowitz                 | drow@drow.net                    |
|  Administrator Extraordinaire  | Web site coming someday          |
| My opinions are my own -       | Day Job: http://www.wwwcomm.com  |
| My mistakes are someone else's |                                  |
---------------------------------------------------------------------