>> > > > "Optimization", from what I remember, seems to have been largely a
>> > > > question of bumping up the buffer size, wasn't it?
>> > > Ah, now that I look, bpf in -current is lagging behind output from
>> > > tcpdump.org.  Who's the maintainer of libpcap/tcpdump things in NetBSD ?

	in what sense?  there's basically no difference.

>> > Since when does tcpdump.org maintain bpf?
>> Since BPF is part of libpcap.
>libpcap is a client of bpf. it is not the bpf code in the kernel. Or
>am I gravely mistaken here somehow?

	libpcap includes userland evaluator for BPF code, and the BPF
	interpreter in the kenrel is a subset of userland evaluator.  therefore,
	you can say that tcpdump.org is responsible for BPF evaluator.

itojun