OTOH,
oDAD (RFC 4429) has it that:
* Never using an Optimistic Address as the source address of a Router
Solicitation with a SLLAO. Another address, or the unspecified
address, may be used, or the RS may be sent without a SLLAO.
What the new ND is doing is making a superOptimistic address out of
EUI-64 based addresses that would bypass even oDAD's limitations.
For addresses for which there is no way to be absolutely sure that
there's no dup, it still makes sense to apply oDAD.
That will be more efficient if the draft enforces that the routers place
the SLLA option in their RAs, because otherwise, the optimistic device
has no way to learn the router LLA.
Pascal
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On
> Behalf Of Erik Nordmark
> Sent: Sunday, May 30, 2010 3:14 PM
> To: Tetsuya Murakami
> Cc: [email protected]
> Subject: Re: [6lowpan] SLLA option in NS
>
> On 05/26/10 03:21 PM, Tetsuya Murakami wrote:
> > Hi All,
> >
> > According to the current 6lowpan-nd draft, I have one question.
> >
> > The draft mentioned NS must include SLLA option in order to inform
the
> > link-layer address of the host. However when the router receives RS
> > message from the host, the router responds RA to the host. At that
> > time, the router needs to know the link-layer address of the host in
> > order to send out RA to the host. So, I think SLLA option of the
host
> > must be included in RS instead of NS.
>
> We should make it clear that the SLLA is required in the RS.
> FWIW the ability to omit them in RFC4861 is for the case of sending a
RS
> before DAD has completed, in which case RFC4861 allows an unspecified
> source address in the RS. We don't need that for 6lowpan.
>
> But we still need SLLA in the NS, since it is the SLLA plus the ARO
that are
> used to form the registration with the router.
>
> > According to the current ND draft, after receiving RS message, the
> > router sends NS message in order to solve the link layer address of
> > the host by using solicited node multicast address. In order to
> > eliminate the multicast packet as much as possible, I recommend to
> > include SLLA option in RS instead of NS.
>
> I don't think there is anything in 6lowpan-nd-09 that says that the
router
> should multicast a NS in this case. (RFC 4861 implementations might do
that,
> or might instead multicast the RA to all-nodes.) Do you have a
pointer to
> specific text in 6lowpan-nd-09 that says this, or that isn't
sufficiently clear on
> never sending multicast NS?
>
> Erik
> _______________________________________________
> 6lowpan mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/6lowpan
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
