Re: G5 powermacs (was: ofwboot.xcf problem...)

[Havard Eidnes <he%NetBSD.org@localhost>]

>> Let me go off on a little tangent: what is the state of the support
>> for the G5 PowerMacs?  Are there any "gotchas"?
>
> Based on reading the archive, there's a gsoc project with a SF repo,
> there's code in head & the effort has seen activity twice. Most recently
> April 2013, which was very fruitful in relation to patches.

One of my coworkers had a G5 PowerMac standing idle (by the looks
of it a 2GHz dual-core w/superdrive variant), and I've been
permitted to lay my grubby hands on it, something I've now done.

> I have a first gen iMac G5, what's in head will not boot on it, blowing
> up at the boot loader stage. Requiring a power cycle.

The G5 PowerMac I tested with didn't require a power cycle, at
least OpenFirmware was still responsive.  But it also blows up
while running ofwboot.xcf while trying to boot the 7.1 bits, like
so:

  Apple PowerMac11,2 5.2.7f1 BootROM built on 09/30/05 at 15:31:03
  Copyright 1994-2005 Apple Computer, Inc.
  All Rights Reserved

  Welcome to Open firmware, the system time and date is: 06/04/2017 13:00:47

  ...

  0 > boot cd:\ofwboot.xcf netbsd.macppc load-size=11516 adler32=9b3aeb2b

  loading XCOFF

  tsize=eec0 dsize=260 bsize=2750 entry=e00000
  SECTIONS:
  .eh_fram 00000074 00000074 00002218 0000014c
  .text    00e00000 00e00000 0000eec0 00002370
  .data    00e0f000 00e0f000 00000260 00011230
  .bss     00e0f260 00e0f260 00002750 00000000
  .gnu.att 00000000 00000000 00000010 00011490
  .ident   00000000 00000000 00000076 000114a0
  loading .text, done..
  loading .data, done..
  clearing .bss, done...
  >> NetBSD/macppc OpenFirmware Boot, Revision 1.12
  5926884
  Invalid memory access at  %SRR0: 00000000.00e0bed8  %SRR1: 10000000.00003030
   ok
  0 >

Trying to look at that instruction with gdb in 7.1 gives:

Reading symbols from ofwboot.xcf...I'm sorry, Dave, I can't do that.  Symbol format `aixcoff-rs6000' unknown.
(gdb) i targ
Local exec file:
        `/usr/mdec/ofwboot.xcf', file type aixcoff-rs6000.
        Entry point: 0xe00000
        0x00000074 - 0x0000228c is .eh_fram
        0x00e00000 - 0x00e0eec0 is .text
        0x00e0f000 - 0x00e0f260 is .data
        0x00e0f260 - 0x00e119b0 is .bss
(gdb) x/i 0xe0bed8
Segmentation fault
granny-smith: {2} 

Well, so much for that.
Doing the same with ofwboot.elf gives:

(gdb) i target
Symbols from "/usr/mdec/ofwboot.elf".
Local exec file:
        `/usr/mdec/ofwboot.elf', file type elf32-powerpc.
        Entry point: 0xe00000
        0x00e00000 - 0x00e0d7f4 is .text
        0x00e0d7f4 - 0x00e0eeaf is .rodata
        0x00e0eeb0 - 0x00e110c8 is .eh_frame
        0x00e110c8 - 0x00e112fd is .data
        0x00e11300 - 0x00e11322 is .sdata
        0x00e11324 - 0x00e1136d is .sbss
        0x00e11370 - 0x00e13a70 is .bss
(gdb) 
(gdb) x/i 0xe0bed8
   0xe0bed8:    b       0xe0ac50
(gdb)
(gdb) x/3i 0xe0bed0
   0xe0bed0:    li      r10,18
   0xe0bed4:    stw     r10,0(r31)
   0xe0bed8:    b       0xe0ac50
(gdb) 

That all looks quite innocent, unless it halts on the "next"
instruction and the contents of r31 makes it unhappy?

I'll do a re-build of the 7.1 release bits and look at the
unstripped ofwboot.mrg so that I can get symbols...


My build host is working on producing -current bits for macppc
(and -current is the target for this testing anyway), we'll see
how that goes when they're ready and I'm where the machine is...

Regards,

- Håvard