My assumption is that SEND WG will adapt to the changes that RFC2462bis
will make. So if we tweak 2462 to require configuration of a link-local
address per suffix, SEND-CGA nodes will do this. This will then prevent
(well, detect) collision as discussed above.
Well, strictly speaking, a (pure) SEND node cannot configure a link-local
address with such a suffix, since the resulting address will not be a
SEND address. The CGA test will fail.
Let me clarify:
1. Say that a (pure, strict) SEND node configures a new global address P::A, where A is computed by the CGA algorithm: A = cga(public key, prefix P, other info)
2. If a SEND node now would like to configure a link local address fe80::A, it (strictly speaking) cannot do that, since for fe80 it MUST use a different suffix, L, where L = cga(public key, prefix fe80, other info).
Putting that aside, a SEND node could well *defend* the address fe80::A for DAD/DIID purposes, but it would never actually use it.
--Pekka Nikander
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------