Hi Tony,
Reply inline.

> Technically you would comply with the RFCs, but the FFFE value was
> negotiated with the IEEE to avoid colliding with their eventual conversion
> to EUI-64. If you pick other values for the actual mac you need to talk to
> the IEEE, but if it is only the IPv6 address the bits will work until an
> implementation gets a full EUI-64 and doesn't want you claiming its
> autoconfigured address.

=> I presume that the "u" bit i.e. universal/local bit that indicates the scope
will have its significance in any IEEE EUI-64 identifier. In that case all
IEEE EUI-64 identifiers assigned to IEEE 802.3 MAC interfaces will
have "0" in their
"u" bit and will result in "global" scoped interface-ids. Since my
implementation
generates only "local"scoped interface-ids, I don't think I'll be claiming
any other interface's autoconf address.

My implementation will anyway have to change if and when they move to
EUI-64. Probably I'll then have to hash it to 48-bits. But, for now, my hash
function is the NULL function.

> If you really want an implementation specific value, don't bother copying
> the mac, hash it then put in your filler.
=> I did consider hashing but due to the unique requirements of my environment,
I've decided to specifically have the MAC address in the Interface-ID.

Thanks & Regards
Vijay


On Thu, Jun 25, 2009 at 4:49 AM, Tony Hain<[email protected]> wrote:
> Technically you would comply with the RFCs, but the FFFE value was
> negotiated with the IEEE to avoid colliding with their eventual conversion
> to EUI-64. If you pick other values for the actual mac you need to talk to
> the IEEE, but if it is only the IPv6 address the bits will work until an
> implementation gets a full EUI-64 and doesn't want you claiming its
> autoconfigured address. Other than that you will simply confuse people that
> read documents that state the EUI-48 transform is supposed to be FFFE.
>
> If you really want an implementation specific value, don't bother copying
> the mac, hash it then put in your filler.
>
> Tony
>
>> -----Original Message-----
>> From: [email protected] [mailto:[email protected]] On Behalf Of
>> Vijayrajan ranganathan
>> Sent: Tuesday, June 23, 2009 10:09 PM
>> To: [email protected]
>> Subject: Implementation specific Interface-ID
>>
>> Hi Everyone,
>>
>> I have an ethernet interface for which I am defining the Interface-ID
>> in a different manner.
>>
>> For an ethernet interface with MAC "34-56-78-9A-BC-DE", I am defining
>> the Interface-ID to be "34-56-78-xy-zw-9A-BC-DE" instead of
>> "36-56-78-FF-FE-9A-BC-DE"
>>
>> where x,y,z,w are my implementation-defined octets.
>>
>> Note:
>> 1. I am fine with a locally-unique Interface-ID
>> 2. The solicited-node multicast address i'll generate for a given
>> unicast/anycast address based on this Interface-ID is the same as
>> that for RFC 2464 based Interface-ID.
>>
>>
>> Assuming I make this change in an IPv6 stack which conforms to RFCs
>>
>> 2464 - Transmission of IPv6 Packets over Ethernet Networks
>> 3513 - Internet Protocol Version 6 (IPv6) Addressing Architecture
>> 4861 - Neighbor Discovery for IP version 6 (IPv6)
>> 4862 - IPv6 Stateless Address Autoconfiguration
>> 3810 - Multicast Listener Discovery Version 2 (MLDv2) for IPv6
>>
>> After this change, will I still conform to these RFCs??
>>
>>
>> Thanks & Regards
>> Vijay
>> --------------------------------------------------------------------
>> IETF IPv6 working group mailing list
>> [email protected]
>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
>> --------------------------------------------------------------------
>
>
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to