On Sat, 31 Jul 1999, Daniel R. Killoran,Ph.D. wrote:

>  Panic switch: PC is at 0x37c2
>  Stopped at      _Debugger+0x6:	unlk  a6
>  db> ps
>  PID       PPID       PGRP       UID   S   FLAGS   COMMAND   WAIT
>  0          -1          0         0    3 0X20204    swapper   scsipi_
>  db> t
>  _Debugger(139e4e,37c2,1eae54,358a,0) + 6
>  _nmihand(0,ffffffff,0,0,0) + 26
>  _lev7intr(?)
>  _mi_switch(2004,2200,0,8,6) + 12
>  _tsleep(58c000,11,11c690,0) + 162
>  _scsipi_execute_xs(58c000) + ac
>  _scsi_scsipi_cmd(584ac0,1eaef2,6,0,0) + 9a
>  _scsi_prevent(584ac0,1,180) + 44
>  _sdopen(409,0,6000,0) + e8
>  _sdsize(409,2004,ffff2200,1e84c8,60b00000) + 58
>  _cpu_dumpconf(0,0) + 46
>  uvmfault(0x15f860,0x1ed000,0,0,0x1)->0x1

I'm guessing now, that the trouble starts in cpu_dumpconf(), where it
tries to set up the dump partition. One work-around that comes
immediately to mind is to bypass the guts of that routine altogether,
by changing the type of your swap partition to anything else. Either
way, it's starting to look like a bug.

There seems to be nothing out of the ordinary with the installer
output you posted separately. I hate to say this, but one thing that's
worked in the past is to do a low-level format and then lay all the
partitions down in one shot. I thought all the disklabel problems were
fixed in 1.4, but the Mac partition scheme is so free-form, it is not
inconceivable that you could have somehow constructed a partition
table which confuses the fake disklabel routines. Re-formatting is
worth a shot.