Dear Mathis,

Thanks for adding and correcting me.

Let me answer again to your two questions in first email

1-Solicited NA (EARO): In direct reply to an NS (source ≠ ::), sent unicast
with S = 1.

2-Unsolicited (“STATUS=Moved”) NA (EARO): Sent multicast BY 6LR to its
vicinity for stale-state cleanup with S = 0. Every NA (EARO), whether
solicited or not, carries the original 6LN’s ROVR to identify which binding
is being confirmed or purged.

Standards Respected:

- S-bit per RFC 4861/6775/8505/9685
- EARO format and semantics per RFC 8505


Regards,


AR

On Fri, Apr 25, 2025, 14:25 Mathis Marion <[email protected]> wrote:

> On 24/04/2025 18:48, Adnan Rashid wrote:
> > CAUTION: This email originated from outside of the organization. Do not
> click links or open attachments unless you recognize the sender and know
> the content is safe.
> >
> > Hi Mathis,
> >
>
> Hello Adnan,
>
> > You are in a right place to ask this question. I will try to answer your
> questions as per my best knowledge.
> >
> > 1-RFC 9685 did not update RFC6775.
> > 2- ROVR always in NA(EARO),  whether it's unicast or multicast. So, all
> one hop nodes know who was the actual source of NS(EARO).
>
> In a multicast NA(EARO) packet for registration refresh, if we put the
> source EUI-64 of a NS(EARO) in the ROVR, then the NA(EARO) will be
> processed by a single neighbor: the one who owns the EUI-64 and
> initially sent a NS(EARO). However the purpose of sending the
> "Registration Refresh Request" in multicast seems to be to query all of
> the neighbors at the same time.
>
> > 3- S-bit will not set in NA(EARO) whether it's unicast or multicast. If
> I am not wrong, I think it is for NUD, but please check RFC 4861.
>
>  From my understanding:
>
>    - In classic NDP (RFC 4861), the S bit is set in all unicast NA.
>      These unicast NA can be responses to NUD probes (unicast NS), but
>      also responses to multicast NS for address resolution.
>    - In 6LoWPAN NDP (RFC 6775), the goal is to remove multicast NS for
>      address resolution, so hosts are expected to register their address
>      to their router. In this case the role of the NS and NA is quite
>      different: NS(ARO) is used to advertise an address and NA(ARO)
>      serves as an acknowledgement. I believe it makes sense to set the S
>      bit in these NA(ARO) since they are still "solicited" (ie. a
>      response to a NS). This is not stated explicitly but I would assume
>      this from classic NDP.
>    - In RFC 9685, the NA(EARO) gains a new purpose with the registration
>      refresh procedure. In this instance the multicast NA(EARO) is
>      spontaneously sent by a router, so it makes sense to clear the S
>      bit (ie. the packet is unsolicited).
>
> >
> > I hope my answers correctly address your questions.
> >
>
> Thank you for sharing your knowledge.
>
> >
> > Regards,
> >
> >
> > AR
> >
> > On Thu, Apr 24, 2025, 14:34 Mathis Marion <mathis.marion=
> [email protected] <mailto:[email protected]>> wrote:
> >
> >     Hello,
> >
> >     This is my first time writing on a IETF mailing list, so I hope that
> it
> >     is the right place to ask such questions. If not, feel free to
> redirect
> >     me to the right place.
> >
> >     First, a bit of context: I am working on the Wi-SUN protocol stack
> >     which uses the 6LoWPAN adaptations mechanisms for IPv6 Neighbor
> >     Discovery. In this context, the "Registration Ownership Verifier"
> >     (ROVR) field of the "Extended Address Registration Option" (EARO) is
> >     always an EUI-64.
> >
> >     RFC 9685 defines a new mechanism "Registration Refresh Request" to
> >     force transmission of NS(ARO) packets from neighboring hosts: a
> router
> >     sends a NA(EARO) packet in multicast with a dedicated EARO status
> code.
> >
> >     My question is: in this special NA(EARO) multicast packet, what value
> >     should be put in the ROVR field of the EARO? This packet is clearly
> not
> >     destined to any specific neighbor so it does not make sense to fill
> it
> >     with a neighbor EUI-64. Putting the router EUI-64 is also unexpected
> >     since the router is not registering its own address. My last guess
> >     would be to use a broadcast EUI-64 as the ROVR to indicate that the
> ARO
> >     is not destined to any particular node.
> >
> >     Note that on reception of a NA(ARO), hosts must check the EUI-64
> field,
> >     and drop the packet if the value is not the host EUI-64:
> >
> >               RFC 6775 5.5.2. Processing a Neighbor Advertisement
> >
> >          In addition to the normal validation of an NA and its options,
> the
> >          ARO (if present) is verified as follows.  [...] If the EUI-64
> field
> >          does not match the EUI-64 of the interface, the option is
> silently
> >          ignored.
> >
> >     I did not find any exception to this rule in RFC 8505 nor RFC 9685.
> Do
> >     we need to consider the "Solicited" (S) bit of the NA to act
> >     differently?
> >
> >     Thank you for your time and consideration.
> >
> >     Mathis Marion
> >     Silicon Laboratories
> >
> >     _______________________________________________
> >     6lo mailing list -- [email protected] <mailto:[email protected]>
> >     To unsubscribe send an email to [email protected] <mailto:
> [email protected]>
> >
>
>
_______________________________________________
6lo mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to