ofwboot.xcf, kernel problems with new GCC?

["John D. Baker" <jdbaker%mylinuxisp.com@localhost>]

I finally had opportunity to build a fresh -current for macppc
(20110805) and have a go at booting my PowerMac G4 (AGP Graphics)
machine.

Since it's experimental, I've only ever net-booted and it has worked
just fine.

Putting all the new files in place, the machine fails to load and
execute "ofwboot.xcf".  Transcribed from the ofw display:

----------
DEFAULT CATCH!, code=300 at    %SRR0: ff83e948    %SRR1: 00003030
CLIENT: <client mac> <client IP>
SERVER: <server mac> <server IP>
Transfer FILE: ofwboot.xcf \
TFTP-actual=e455 TFTP-adler32=ea65794e load-size=e455 adler32=ea65794e

tsize=e040 dsize=260 bsize=2730 entry=e00000
SECTIONS:
.text    00e00000 00e00000 0000e040 00000130
.data    00e0f000 00e0f000 00000260 0000e170
.bss     00e0f260 0030f260 00002730 00000000
.gnu.att 00000000 00000000 00000014 0000e3d0
.ident   00000000 00000000 00000071 0000e3e4
loading .text, done..
loading .data, done..
clearing .bss, done..

Apple PowerMac3,1 4.2.8f1 BootROM built on 10/11/01 at 14:12:47
Copyright 1994-2001 Apple Computer, Inc.
All Rights Reserve.

Welcome to Open Firmware, the system time and date is 04:06:15 01/01/1904

To continue booting, type "mac-boot" and press return.
To shut down, type "shut-down" and press return.
 ok
0 >
----------

The order is a little wierd because my OF boot-command is set to:

  " screen" output boot

So it tries to boot right away and doesn't print the OF banner unless
and until it fails.

Thinking there might be a pram configuration issue (after an abortive
attempt to load MacOS X 10.5), I pressed the reset switch on the main
board after an earlier attempt so that's why the OF time/date is off.

Repeating the boot attempt from the console fails similarly, but
with a message of:

----------
[...]
loading .text

CLAIM failed
 ok
0 >
----------



I copied an "ofwboot.xcf" from an 5.99.51 build and it loads and
runs just fine, but the new GENERIC kernel promptly panics.

It performs a backtrace immediately, but the console font is quite
large and I don't know if I've actally got everything.  What I see is:

----------
0
            r1=0x96c5a0 cr=0x24024022 xer=0x20000000 ctr=0x101af8
0x0096c5a0: at 0xd4784dac
0x0096c5b0: at ahc_pci_attach+0x1e4
0x0096c600: at config_attach_loc+0x1f0
0x0096c630: at pci_probe_device+0x4c0
0x0096c6f0: at pci_enumerate_bus+0x208
0x0096c730: at pcirescan+0x24
0x0096c740: at pciattach+0x204
0x0096c770: at config_attach_loc+0x0x1f0
0x0096c7a0: at ppbattach+0x134
0x0096c900: at config_attach_loc+0x1f0
0x0096c930: at pci_probe_device+0x4c0
0x0096c9f0: at pci_enumerate_bus+0x208
0x0096ca30: at pcirescan+0x24
0x0096ca40: at pciattach+0x204
0x0096ca70: at config_attach_loc+0x1f0
0x0096caa0: at uninorth_attach+0x2e0
0x0096cbd0: at config_attach_loc+0x1f0
0x0096cc00: at mainbus_attach+0x13c
0x0096ccd0: at config_attach_loc+0x1f0
0x0096cd00: at cpu_configure+0x158
0x0096cec0: at main+0x1c8
0x0096cf10: at 0x1000a0
db>
----------

It is nice to have working USB keyboard in ddb.


The start of the backtrace indicates something with the "ahc" driver.
The machine does have an Adaptec 2930 SCSI card in it.  I'll take it
out and see if the behavior changes.

--
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com    OpenBSD            FreeBSD
| X  No HTML/proprietary data in email.   BSD just sits there and works!
|/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645