Benjamin Herrenschmidt wrote:
BootX version is 1.2.2. As for PCI cards, yes, I have three:
1. Atto "Express PCI" SCSI card. Has a SYM53c875 chip. No other markings.
2. Apple iX 3D video board. iXMICRO chip set. TI TVP3030 chip and
iXMICRO Twin Turbo 9135-388.
3. Apple Fast Ethernet 10/100Base-T Card.
Yes, well... the problem with BootX is that it takes over MacOS,
which means that MacOS already initialized those cards and have
drivers using them. Those cards may be bus-mastering to memory
(that is writing to memory locations that the MacOS driver has
previously allocated for that).
When Linux takes over via BootX, it doesn't (cannot) know what
the MacOS driver was doing, it starts using the whole memory for
itself, but if a piece of hardware is still writing to that memory,
it will possibly damage some information or code that linux is
putting in that same place, thus leading to all sorts of strange
crashes.
BootX contains a bunch of hacks to "stop" a list of well known
PCI cards from writing to memory, afaik, the Atto symbios card
is part of that (provided it's the one Apple bundled in std.
with the G3 back them, if not, it may be the culprit we are
looking for). The ethernet card may be the cause too.
Hmmm. The Atto board was, I think, originally bundled with the box.
My wife originally ordered it for use at her work, and I think it came
with all this stuff standard. I will ask anyway. The ethernet board,
well, (shuffles feet, stares at floor) was originally in the box too,
but I'd removed it a year or so ago, while the box was still running
MacOS. I only put it back in while I had the box apart this evening to
find out what boards were in it. I found that the empty hole was letting
in quite a lot of dust, so I wanted to block it up with something. And
its own Ethernet board seemed ideal for the purpose. Thus it probably
plays no part in the current problem.
But...
Now when I try and boot into the Linux installer, I always get a crash:
Machine check in kernel mode. (regs at c01f5ea0)
Caused by (from srr1): Unknown values in srr1
NIP: C015E3CC XER: 00000000 LR: C0163014 REGS: c01f5ea0 TRAP: 0200
MSR: 00001030 [IR, DR, ME]
TASK = c01f41e0[0] 'swapper' mm->pgd c01f2000 Last syscall: 112
GPR00: 000000FF C01F5F50 C01F41E0 C053B000 00008000 C01F5FF0 FFFFFF9D
F3000020
GPR08: C01F3414 FE010014 00010000 FE000000 00000000 0040A0A8 00000000
00000000
GPR16: 00000000 00000000 20000002 00000000 00001032 001F5FE0 00000000
C0004154
GPR24: C0006518 00000000 40000000 0040A000 0000002E C01F5FF0 00009032
C053B000
Call backtrace:
00000000 C0163014 C00064A0 C000CEFC C0006558 C0004154 C0021030
C0006604 C0006630 C0003EC4 C0220000 C021C8CC C0003CF0
Kernel panic: machine check
In swapper task - not syncing
Rebooting in 180 seconds..
(sorry about the [EMAIL PROTECTED]&*()_+=- word wrap)
Before, I only got this one occasionally; now I get it every time.
It happens immediately after the two SCSI disks are checked, so it's
probably related to the SCSI controller. I'll see if I can connect the
drives to the other SCSI controller and remove the Atto board
(temporarily at least). But that'll have to wait till tomorrow; it's
getting late.
Thanks for your help.
.....Ron
--
Ron Murray ([EMAIL PROTECTED])
http://www.rjmx.net/~ron
GPG Public Key Fingerprint: F2C1 FC47 5EF7 0317 133C D66B 8ADA A3C4
D86C 74DE
