In some mail from itojun@iijlab.net, sie said:
> 
> >Should we be thinking of adding a knob to control how frequently
> >NetBSD sends out ICMP unreachables ?  Below I've got a patch which
> >I just hacked up (untested :) which provides a knob in sysctl to
> >control how often to actually send an ICMP error back.  I've added
> >a patch which uses the same knob for ICMP6 (an ICMP error is an
> >ICMP error, be it v4 or v6).
> 
> 	I think ICMPv6 part is not necessary, because KAME code hs 
> 	rate limiting code for ICMPv6 errors.  This is mandatory from
> 	RFC2463 page 5.  see sys/netinet6/icmp6.c:icmp6_ratelimit().
> 	sysctl knob is net.inet6.icmp6.errratelimit.  I think you are doing
> 	exactly the same thing.

Right.  In that case, the name for inet4 should be the same as that for
inet6.

Darren