On Wed, Nov 05, 1997 at 10:50:46AM +0100, Anders Magnusson wrote:
> > 
> > Here's the output of a panic:
> > 
> > creating runtime link editor directory cache.
> > checking for core dump...
> > ock=0 36:52, fs=/idt:ev v=1 0xa904, bl
> > panicc: blksfreesd: nfrst:eIeing f ree frag
> > syncing disks...10 10 9 2 2 2 ...a bunch of twos...2 2 2 giving up
> > 
> > 
> > Note that the garbled output in these messages is not a typing error, nor do
> > I normally have any garbled output on the screen; it's normally just fine.
> > 
> This sounds like you have partitioned your disk with swap overlapping
> anothe filesystem or something like that. It may also be because you
> are having your root partition as e (why?) which probably may not work.
> Or to express it like this: On the root disk I would recommend to have
> root on a partition, swap on b and the whole disk on c. Anything else
> may not work at all.

I would suspect that you were right, except that I was seeing the same
behaviour when I was booted off of the miniroot tape, and I checked the
disklabel output for ra0 and it's *exactly* like the one used in the example
in the installation instructions (on http://vaxine.bitcon.no), i.e. none of
the partitions are overlapping.

Does BSD attach some significance to the "a" partition being root that other
UNIXes do not?  (I'm relatively new to BSD; other than administering a
couple of SunOS 4 machines when I got here, most of my administration
experience has been with Solaris 2, Linux, & VMS, and I know neither Solaris
nor Linux care where the partitions are.)  At any rate, I *can* boot up off
of the "e" partition provided I use the boot loader from the tape.  

The crashes occur whether or not I have swap turned on.  (There is no swap
in my fstab; I figured 13mb would probably be enough to let me boot up
without it.)

On a whim, I just looked at the top of my disk.  It has 30 bad sectors
mapped out.  Does the fact that NetBSD doesn't have badblocks support mean I
needed to map those sectors out when I built the filesystems?

At any rate, I think I'm going to try writing my own rc script that starts
up just the stuff I know I need.  (i.e. sets up network interfaces, checks
mounts, etc.)  If that doesn't work out, I'll repartition the disks.

Heck, I suppose I could even boot single-user from the e partition (which I
can do now), then mount the "a" partition on "/mnt" and install the stuff I
need there, then reboot off of the "a" partition.  I'll give it a try.

Thanks for the info everyone.

- Tom
-- 
Tom Guptill                     Department of Physics and Astronomy
UNIX SA                         University of Rochester  Rochester, NY USA
t g p t @ p a s . r o c h e s t e r . e d u