I assume I'm the only one this is happening to, but the details to
fix it are beyond me for now...

I tried upgrading my multiprocessor SS20 from 1.6W to 1.6ZC (and
now 1.6ZD) and have gotten a very repeatable panic (every time),
and never once gotten a core dump.

DDBs trace command tells me:  (no serial console, so it's copied
off the screen.  I'll copy all the characters if needed.)

cpu_debugger
mem_access_fault_4m+0x11c
cpu_switchto+0x194
netbsd:snapshot+0x40

It appears to happen just before the kernel prints out the scsibus0
information, and just after doing the IPsec stuff.  I was about to
guess that it was because my machine doesn't like to create or
switch threads like all the other sparcs out there, but my 1.6W
kernel has [scsibus0] as PID 2, so [swapper] and init should be
doing that already, right?  (Would my config file help anyone in
assisting to track this down?)

So... Any hints on how to get a working kernel from -current?

Jon