>> 	any interesting values on netstat -sn?
>Ah, an excellent thought! The ip "with data size < data length" one is
>incrementing for every packet that comes back via the GRE. I wonder if
>I screwed up the patch somehow? I've appended a diff; perhaps you could
>look it over and see if I missed something.

>+	m_adj(m, hlen);
> 	m->m_pkthdr.len -= hlen;	<-- maybe this line is redundant.

itojun