>If anyone is interested I'll code this change.

>>   I'll vote for that.
>>
>>   The odd thing is, it doesn't even seem to be a broadcast. I can watch
>>my other NetBSD machines complain on their consoles about bad calls to
>>bootparam, and they don't happen at the same time. It seems to try every
>>single IP address one by one, waiting for it to time out before moving
>>on to the next. If it were broadcasting, the real bootparam server would
>>respond right away and there wouldn't be a five-minute delay before it
>>finally gets around to trying .99.

  ..maybe a bit of both. I usually install netbsd nfs from a linux box.
When I gnarled things up, and had reason to monitor traffic on the linux'
eth0 for bootpd packets, these do indeed appear as broadcasts, but me
thinks the sun only considers answers on a per host/per timeout period
basis, and all things being equal, this should be starting from the
lowermost member of the connected network and continue on up thru the
bitmask until the netmask boundary is reached..??..that matches the
precepts of the routing statement.

   I've always usually had my bootparamd/rarpd/nfsd living at the bottom
of class-c (.1), I recall once believing for days that something was
broken bad when I was trying the same operation from host .212 

      Cheers!