RE: FW: New Version Notification for draft-rafiee-6man-cga-attack-00.txt

["Hosnieh Rafiee" <[e-mail]>]

>forgive neophyte questions: the attacker, using this hash-collision
mechanism, has effectively gained the right to subvert *one* CGA identified
host, right?

Yes, true.

>and, they still have to achieve path injection, spoofed source, TCP/IP
sequencing &c, in order to exploit this attack, against this one host.
right?

Yes, This is actually what I am trying to say and all of this is because of
the combination of NDP with CGA and forgetting some steps in verifications.

>and, the cost of that one host, and its location, is not of the attackers
chosing: they don't know in advance, that they can gain entry to any
specific CGA, so they cannot direct the attack against a known CGA endpoint:
they opportunisitically can attack *SOME* CGA, by virtue of the hash
collision: its not clear if its a high- or a low- value endpoint.

That is true. You are the attacker, join network and eavesdrop and now want
to forge the identity of the nodes. CGA claimed to make it harder but it is
not as hard as we thought before.

Smile,
Hosnieh