I was playing with rdump today between my pc532 and Sun. to see if I
could figure out why rdump'ing is general is so damn slow.  What I
came up with is a way to hang NetBSD on the pc532 solidly.

One invocation of "dump 0fb pc532:/dev/null 126 /" or two sequential
invocations of "dump 0fb pc532:/dev/null 64 /"; I ran a script which
looped doing "vmstat; pstat -T; netstat -m; iostat" looking for
anything abnormal, but no dice.  My kernel has DDB, but BREAK did
nothing, which either means the system is looping at splhigh(), or the
hardware has locked up.

This seems (superficially at least) similar to the problem Philip
Zimmerman has been seeing (his pc532 is on a much busier net than my
little 3-node network at home).

-phil