> 
> ...just tell DDB to "continue" ... it will then proceed with
> the standard panic procedure, including dumping a core file.
> 

   Thanks. However, when I try this I still seem to get the
wrong stack. I get:

(kgdb) where
#0  0xf8028914 in mi_switch ()
#1  0xf8028284 in bpendtsleep ()
#2  0xf80c1184 in scheduler ()
#3  0xf801cfe8 in main ()
(kgdb) 

Whereas the fault actually occurred during arplookup(), called
via a soft interrupt. Perhaps I need to change process context
within kgdb?

Simon