Subject: RE: Boot problem, Mac-68k, 1.5.2 release
To: 'dakidd' <dakidd@sonic.net>
From: Dan Willson <Dan.Willson@VPUADV.UAB.EDU>
List: port-mac68k
Date: 05/06/2002 15:29:23
FWIW, I got the same error message on an LC 475 this afternoon after
installing 1.5.2 (upgraded to full 68040, 8 MB of RAM, 1.2 gig internal SCSI
drive, used latest installer/booter from NetBSD FTP site, turned off VM in
Mac OS, booted in 1-bit video with extensions off, etc.). I had just
finished an install without errors and was booting into single-user mode
with the latest Booter -- had the LC video hack checkbox enabled.

I found this description of the error in groups.google.com:
Leo Weppelman (leo@wau.mis.ah.nl) wrote this on 03/15/2000
in response to a thread titled "panic: cnopen: cn_tab->cn_dev == NODEV"
"I added this panic to prevent weird kernel/
behaviour when misconfiguring the console device. The most probable cause
of this panic is the failure to properly attach a console device. Either
none of the devices did match/attach or you did mess up the config file."

I also found mention on groups.google.com that the LC video hack option is
really unnecessary and has caused a few folks problems (they said that not
using the LC video hack option allowed their Mac to boot into NetBSD). There
was also mention of newer kernels not requiring the use of the video hack.
Anyone successful in installing 1.5.2 from scratch on a similar machine who
can verify this? I can't wait to get home and find out for myself, but it
would be cool if someone could clue me in and save me some time. Thanks in
advance!

Dan Willson
ponyboy@uab.edu



-----Original Message-----
From: dakidd [mailto:dakidd@sonic.net]
Sent: Monday, May 06, 2002 1:13 AM
To: port-mac68k@netbsd.org
Subject: Boot problem, Mac-68k, 1.5.2 release


Greetings, folks.

New NetBSD user trying to come online here, but running into trouble.

System details:
PowerBook 145 that's had radical brain-surgery - It's been various
machines through the years, but nowdays, gestalt (and every other method
I know to throw at it) says that it's a PowerBook 170.

8 megs physical RAM
256 Meg HD (SCSI, of course) at ID 0
Nothing "exotic" attached to it or stuffed into it (other than the
motherboard/daughterboard that make it a 170 instead of the 145 the case
claims it to be)

Running System 7.5.5 before the NetBSD Install. Now running a severely
gutted (to fit the OS, installer, booter, and a carefully chosen small
handful of useful MacOS items onto a 32Meg partition) version of 7.5.5.

Drive has been low-level formatted, and partitioned to 32 Megs Mac, 32
Megs NetBSD Swap, and the rest of the drive NetBSD Root/Usr. Partition
map is 0: MacOS Driver partition, 1: MacOS volume, 2: Root/Usr
partition, 3: Swap partition.

mkfs v1.47 applied to set up and format the partitions appropriately.
No errors reported during the process.

NetBSD/Mac68k Installer (Not sure of version - grabbed the one symlinked
to on the NetBSD Site) used to install the following tarballs, all of
which were downloaded (very carefully using binary mode FTP) from links
to the 1.5.2 release on the NetBSD.org site:

base.tgz
kern.tgz
etc.tgz
comp.tgz
text.tgz
man.tgz
misc.tgz

Basically, a full install, minus the games.tgz tarball, and the Xwindows
stuff.

All tarballs were downloaded to another Mac, the drive they were stored on
was mounted to the Powerbook via the chooser, then they installed via my
localtalk network.

Installer "build devices" command was invoked, and made no complaints.

So the hard part SHOULD be over, right?
Apparently not...

Mac/NetBSD Booter 1.11.3g:

Booting options:
Kernel location "BSD Device", Kernel name netbsd, no partition name
specified (left blank, as found)
Root SCSI ID 0, enabled
RAM size autodetect checked. (correctly fills in 8 MB)
Auto-set GMT bias checked. (correctly fills in -420 minutes)
"Single User" checked.

Serial options:
Serial Boot Echo unchecked. Localtalk connected to Printer port checked.
Open port before booting checked. Serial Console unchecked.
All other settings left at default values.

Monitors options:
All default values (They seem completely sane for this machine)

Startup Options:
All default values, except "Log debugging to file" checked, and the file
specified as "bootlog" on the Mac Desktop

Options saved.

"Boot Now" selected.
Time passes, and lots of stuff rolls past -
many Env: lines, none of which appear to be insane, followed after a few
seconds by the BSD Copyright notice, then lots of alphabet soup, some of
which I at least sort of understand. The stuff I do understand looks
reasonably sane. (but due to my inexperience with *nix/BSD, could be
COMPLETELY whacked out, and I just wouldn't recognize it as being bad)

The last few lines of the boot process (as far as it gets) look like this:

boot device: sd0
root on sd0a dumps on sd0b
PRAM time does not appear to have been read correctly.
PRAM: 0x83da4f80, macos_boottime: 0x3cd55d17.
root file system type: ffs
panic: cnopen: cn_tab->cn_dev == NODEV

Stopped in init at    _cpu_Debugger+0x6:   unlk   a6
db>

At this point, the boot attempt hangs.

Things I've checked/tried:
The machine has never (to my knowledge) had RAMDoubler or similar
installed. VM is off via the memory control panel. (Machine HAS been
rebooted since turning off VM) 32 bit addressing is on in the memory
control panel (Again, rebooted since verifying the setting). This machine
only does 1 bit video (and the display looks perfectly fine, up until the
panic/drop to debugger)

Doing a "t" command from the db> prompt shows the following:
_cpu_Debugger(ffffffff,0,7cd640,1cef8c,7e5dac) + 6
_panic(1af47d,0,7e5df0,6901e,0) + 60
_cnopen(0,3,2000,7cd640,0) + 22
_spec_open(7e514) + 136
_vn_open(7e5ee8,3,0) + 31a
_sys_open(7cd640,7e5f88,7e5f80) + 92
_syscall(5) + 144
_trap() + e
db>

I recognize some of the routine names (trap, syscall, sys_open are fairly
obvious) but don't yet know enough about the debugger (I assume I'm seeing
the debugger...) to do anything more useful than the stack trace (The "t"
command *IS* a stack trace, right?) without some guidance.

Trying a "continue" command at the debugger prompt gets me:
syncing disks... done
NetBSD/mac68k does not trust itself to update the RTC on shutdown.

dumping to dev 4,1 offset 47630
dump 1 2 3 4 5 6 7 succeeded
rebooting
<Startup bong, followed by normal MacOS startup sequence>

Problem is 100% repeatable, and the results don't vary at all (so far as I
can see)

Anybody got any ideas for me?

If this message looks familiar, you probably saw a slightly less detailed
vaersion of it posted to comp.unix.bsd.netbsd.misc. Most of it is a
copy-paste from there, with some added details, when I got the bright idea
to
try posting my troubles here.


Don Bruder -  dakidd@sonic.net      <--- Preferred Email - unmunged
I will choose a path that's clear: I will choose Free Will! - N. Peart