ether_resolvemulti() doesn't expire routes

[Bruce M Simpson]

I noticed that if I do this:-
# route add -net 224.0.0.0/4 -iface xl0 -expire 3000

The resultant cloned routes don't get given a lifetime, i.e. they're
totally static and remain in the route table for the lifetime of the kernel.

Either multicast designated receivers or IGMP aware routers are the two
possible deployment scenarios I can think of which might want to hold on
to multicast routes for a long period of time - but these same machines
will probably be running mrouted, rather than using a simple primary-interface
technique as above to leverage link-layer multicast.

What's to stop a malicious user from writing a program which fills the
routing table up with multicast routes which aren't actually used by the box?

Would it not be a good idea to age these routes and thus prevent them from
polluting the routing table in this case?

BMS