(started early Sep with ESP core crypto engine rewrite). perform m_pullup2 equivalent on IPv6 ESP transport mode, to avoid non-continuous mbuf. (this is not necessary for m_pulldown case) KAME PR 283. Thanks to TAHI guys and firstname.lastname@example.org.
* kame/sys/netinet6/ip6_forward.c (ip6_forward): if the incoming interface equals to the outgoing one, and the link attached to the interface is point-to-point, then immediately drop the packet and send an ICMPv6 destination unreachable error message. This is currently experimental and needs the PROHIBIT_P2PREDIRECT kernel compilation option.
discard the packet and return an icmp6 dst unreach error. (experimental, enabled with PROHIBIT_P2PREDIRECT only)
old behavior (relay telnet session) does not seem like a sensible default - the behavior is historical. now faithd will die with usage on no argument. WARNING: you may need to change startup scripts, if you rely upon old "no argument" behavior.
be a common default. it may break backward compatibility of startup scripts.
…r layer protocol between two nodes.
* kame/kame/pim6sd/pim6_proto.c (receive_pim6_join_prune): improved the algorithm of random delay calculation; the result of the previous algorithm was always 0 in spite of using random(). This fix was based on a comment from Toshiaki.Nakatsu@fujixerox.co.jp.
based on a comment from Toshiaki.Nakatsu@fujixerox.co.jp.
* kame/sys/netinet6/ip6_mroute.c (pim6_input): added validation for the IP version of inner packets. 2000-09-10 JINMEI, Tatuya <email@example.com> * kame/kame/pim6sd/pim6_proto.c (receive_pim6_register): improved robustness: - added validation for the length of register messages. - added validation for the IP version of inner packets, according to the recent discussion in the IETF pim ML.
…(i.e. IPv6). (according to clarfication of recent discussion in the IETF pim ML)