On May  9, 10:51am, Patrick Vervoorn wrote:
> Like I mentioned, I've tried other kernels: the 940427 sun-lamp one and the
> Warp + IDE test kernel from Michael Hitch I found at ftp.coe.montana.edu
> (which probably would have had other problems since I also have an IDE disk
> connected (IDE sd0 <-> SCSI sd0). Anyhow, this last one also failed but
> this time it jumped into some sort of boot-monitor, which I didn't quite
> understand... :)

  My kernel is compiled with the debugger.  If you add the -S option to
loadbsd, the kernel symbols should be available to the debugger.  Then
when your system panics, you can enter the command "trace" and get a
stack trace.  This should show what routines had been called at that
point.  That should give a clue as to what is going on.

> My guess is something goes very wrong when the kernel finds a tape device
> it doesn't recognize; it seems to jump to some bogus address...

  I suspect it's trying to call a routine that hasn't been set up yet,
so the address is NULL.  I can't think of any thing in the tape driver
that should cause that.

> If anyone has any ideas, or if you need more info, let me know... I would
> very much like to backup NetBSD, before something goes very wrong... :)

  One problem with the "GENERIC" tape handling now is that it assumes
a 512-byte fixed length record, so it can't handle variable length
record devices.

  The stack trace back with symbols should give me a good idea of what
is happening, so give that a try.

Michael

-- 
Michael L. Hitch			INTERNET:  osymh@montana.edu
Computer Consultant			BITNET:  OSYMH@MTSUNIX1.BITNET
Office of Systems and Computing Services
Montana State University	Bozeman, MT	USA

------------------------------------------------------------------------------