Hi Elwyn, thanks for the clarification. Responses in-line. >Tim Enos wrote: >>> From: Su Thunder <[EMAIL PROTECTED]> >>> Date: 2006/09/26 Tue AM 10:53:59 CDT >>> To: ipv6@ietf.org >>> Subject: A question about Source Address of IP Field of ND Packets(RFC2461) >>> >> >> >>> The Source Address of Router Solicitation Message is specified in section >>> 4.1 as follows: >>> IP Fields: >>> >>> Source Address >>> An IP address assigned to the sending interface, or >>> the unspecified address if no address is assigned >>> to the sending interface. >>> My question is, why not specify whether the Source Address is link-local >>> address or other type when it is not the unspecified address?And when to >>> use which type?Or it must be link-local address as the one of the Router >>> Advertisement Message? >>> The same problem occurs in the same field of Neighbor Solicitation >>> Message,Neighbor Advertisement Message. >>> >> >> Unless there is no address assigned to a sending interface (in which case >> the SA would be ::), I can't think of a reason for any ND message to require >> the SA be anything but link-local in scope. >> >For neighbour solicitations: >Please read section 7.2.2 of RFC2461 which explains when it is desirable >to use another (generally a global scope address) of the sending >interfaces' assigned addresses rather than the link local address.
Not always (sorry for hair-splitting), but generally yes. Section 7.2.2. beginning with: "If the source address of the packet prompting the solicitation is the same as one of the addresses assigned to the outgoing interface, that address SHOULD be placed in the IP Source Address of the outgoing solicitation..." It >helps get the neighbour caches at both ends set up and the association >of link layer and IP addresses determined in one round trip.BTW this >sometimes means that NS/ND packets contravene the usual rules on >source/destination addresses matching - see the end of section 2.3 in >the updated version of RFC2461 (draft-ietf-ipv6-2461bis-08.txt). Yes, it does say "It is possible in some cases for hosts to use a source address of a larger scope than the destination address in the IPv6 header.". (Parenthetically, I realize my need to more carefully review drafts even though they are "merely" works-in-progress.) > >For router solicitations: >The usual case is that a node sending a router solicitation will only >have link local addresses at the stage in configuration when the >soliictation is sent, so it doesn't have a choice. Nevertheless, it IS link-local... even "worse" it could have no address (::) at all, in which case of course an SLLAO MUST NOT be used. As with neighbour >solitctation using a real address rather tha the unspecified address >allows address resolution to be piggybacked onto the single round trip. Of course, when possible. Good point. The real address might or might not be link-local. > >Regards, >Elwyn > >> As a general rule (and perhaps going beyond the intended scope of the >> original e-mail), RFC 3484 section 5 provides guidance for source address >> selection. Further, section 10.1 of that same spec provides what are IMO >> good examples thereof. >> >> Best Regards, >> >> Tim Enos >> Rom 8:28 -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
