>> And second, I supped the kernel sources yesturday (Feb 14) and rebuilt the
>> kernel... now when I boot up, I can get to single user mode, but if I try
>> to boot into multi-user mode, I get messages about sd0g not being
>> configured (sd1g and sd2g give me the same problem.).  Next I see messages
>> about needing to run FSCK by hand (I assume because it can't read the
>> non-configured sd0,1, and 2)
>>
>> Then, it locks up.  No debugger, no prompt, no messages, no nutin.
>
>The scsi drives are coming up numbered the same as before?

Yup.  Same numbers, sd0,1, and 2.

>
>> If I go into single user mode and put the old kernel back (backups are
>> wonderfull!) and then reboot, viola... everything works again.
>
>You overwrote a good kernel with a test kernel? Tisk, tisk! My root
>has about 5 kernels scattered around at the moment. One works, one
>is netbsd (a link to whatever's running so ifconfig works), and the
>others are various debug kernels.

Perhaps this isn't clear... I *DID* keep my last kernel around, with a
different name, of course.  In fact, I have about 4 lying around.  So
there! ;-)

>
>> If it matters, this is running on a Mac IIx w/ 8mb of ram, and three
>>harddrives
>> (one for / and /usr, one for /var, and one for random data...  The root
>> drive has 3 partitions (1 mac, 1 / and 1 /usr) and the other 2 just have 1
>> partition each.)
>>
>> Any thoughts?
>
>Is disklabel staticly linked? The partition mounting code was recently
>changed, so you might be getting partitons in new places. The problem
>might be that you seem to have a /usr partiton, and most commands need
>the shared libraries in /usr/lib. :-(

Yeah, I'm begining to think I may move / and /usr to one partition because
of this.  For /usr/bin and /usr/sbin, I moved everything to /bin2 and
/sbin2 and then set up symbolic links in /usr, both in the mount point
directory (off of /) and in the /usr partition itself.  I guess the same
approach would work for /usr/lib, since the links "under" the partition
"disappear" when /usr is mounted, but this is making my root partition
pretty ugly!

>disklabel should be able to tell you were the partitions are placing
>themselves. Then a quick fix to /etc/fstab, and you're done.

I may take a look at trying this, but further tests I ran last night tell
me its something more fundamental than a change in partition names... I
tried to run fsck on / (sd0a) and got the same error.  It was already
mounted as read only, else I couldn't have gotten to single user mode, but
fsck couldn't find /dev/rsd0a either.

I did sup everything last night and try with a newer kernel, but got
similar errors (Different messages, though... I was too tired to write them
down).  I'll crank out the sources I have for netbsd 1.1 somewhere and see
what has changed in the SCSI code.  If I can get things working by going
back to the 1.1 sources from the original release, I'll post the changes I
made so someone can help me determine if it's a problem with some funky
hard drive I have or just the phase of the Moon.

And if that doesn't work, I'll post the text of the actual errors instead.

>
>Take care,
>
>Bill


--
Aaron Scott Magill                                             amagill@uiuc.edu
-------------------------------------------------------------------------------
}{  "I have SEEN evil!  I have SEEN horror!  I have seen the unholy maggots  }{
}{    which feast in the dark recesses of the human soul!  I have seen all   }{
}{  this, officer, but until today, I had never seen... YOU!" - Gomez Addams }{
-------------------------------------------------------------------------------