Obtained from: email@example.com
* kame/kame/bgpd/dump.c (dump_if_rtable): was added to dump more information about each inteface including local addresses. print_ifrt_dump() was removed according to this change.
* freebsd3/sys/netinet/tcp_subr.c (tcp_newtcpcb): always initialize inp_ip_ttl in tcp_newtcpcb() in order to avoid zero ttl when matching an incoming IPv4-maaped IPv6 address. This fix was in response to a problem report from firstname.lastname@example.org.
zero ttl when matching an IPv4-maaped IPv6 address. (in response to a problem report from email@example.com)
- more documented how each parameter works
try to print IPv4 NLRI (not tested as we have no trace!)
tcp_pullup(), as it degraded stability and made some wrong assumptions (see below, Nov 6).
addresses in transport mode case. If you define protocol/mode//level then kernel get addresses from IP header to match SPD. XXX should be considered the relative between inner and outer IP addresses when packet will launch from own interface by tunnel mode.
we will revisit it later.
this makes arp[0:1] to work again ("0:1" was mistakingly considered as IPv6 address). Uses very ugly regex...
XXX more compact regex for IPv6 address?
(like "moga" in "out ipsec ah/transport/moga/use")
add more debug info on parse error (print "yytext" on error message)
(without it we have no clue about scopeid and will have hard time setting up routing table with route(8))
which ensures mbuf alignment match with what tcp_input() assumes. the routine is kind of ugly because, in most architectures, MHLEN < ip + tcp + tcp opt with IPv6.
intermediate extension headers. see comment for the reasons.
IPv6 now works on awi.