> I have been having the same problem on similar hardware. I built a
> kernel with a symbol table and loaded it a while ago. Its taken a 
> while to break, but it did so last night. The break is just like
> all the others we have had. The crash info is below, I'll hang onto
> the dump so I can answer questions :-). 
> 
> NetBSD 1.3.2 (WEBCACHE) #5: Mon Oct 12 10:15:52 BST 1998
>     root@webcache2:/usr/src/sys/arch/i386/compile/WEBCACHE
> 
> fatal page fault in supervisor mode
> trap type 6 code 0 eip f0180a14 cs 8 eflags 10282 cr2 deadbef7 cpl c0000000
> panic: trap
> syncing disks... 230 230 135 37 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 giving up

No, this is clearly a different bug.  Note that cr2 contains the
address that was faulted on.  In George's case, it was clearly a null
pointer dereference.  In your case it's clearly a case of using
already freed memory.

Could you provide me with the kernel and core dump so that I can
analyze them?