Folks, draft-gont-opsec-ipv6-implications-on-ipv4-nets has been adopted as an IETF opsec wg item (please see: <http://tools.ietf.org/html/draft-ietf-opsec-ipv6-implications-on-ipv4-nets>) I was thinking about discussing the following scenario, that I came up with a few days ago: A dual-stacked user (v6 enabled by default) "visits" an IPv4-only network, and establish his VPN with his office (for "mitigating" sniffing attacks, etc.). A local attacker sends forged ICMPv6 RAs, thus triggering IPv6 configuration at the victim nodes. If any of the remote nodes the victim is trying to "visit" is IPv6-enabled, then it's possible/likely that the IPv6 destination address will be used over the IPv4 one. in which case the victim will send his traffic on the local network, as opposed to "through the VPN". Assuming the VPN product does not disable local v6 support, and that the VPN does not provide IPv6 connectivity (*), this attack vector could prove to be an interesting one ("unexpected", to some extent). (*) even then, this attack might still work. Thoughts? P.S.: Comments on the current version of the aforementioned Internet-Draft will be welcome, too. And yeah, our Twitter is @SI6Networks... Thanks! Best regards, -- Fernando Gont SI6 Networks e-mail: fgont@xxxxxxxxxxxxxxx PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492
