[ofw] RFC for DHCPv6 over IPoIB

[Alex Naslednikov]

Hello Fab,
I asked this question because the analogy with IPv4, where we totally replaced the CID field, according to the predefined pattern.
For this reason, I wonder if there is a predefined patterns also for DHCPv6. 

>From the RFC:
A DUID consists of a two-octet type code represented in network byte
   order, followed by a variable number of octets that make up the
   actual identifier.  A DUID can be no more than 128 octets long (not
   including the type code).  The following types are currently defined:

      1        Link-layer address plus time [DUID-LLT]
      2        Vendor-assigned unique ID based on Enterprise Number [DUID-EN]
      3        Link-layer address [DUID-LLT]

For LL and LLT, is it really enough to replace link address by our GUID, or should we apply some additional transformations ?
For EN - what is the pattern ?

-----Original Message-----
From: Fab Tillier [mailto:ftillier at microsoft.com] 
Sent: Thursday, November 04, 2010 7:57 PM
To: Alex Naslednikov; ofw at lists.openfabrics.org
Cc: Liran Liss
Subject: RE: RFC for DHCPv6 over IPoIB

Hi Alex,

Alex Naslednikov wrote on Thu, 4 Nov 2010 at 09:53:40

> Hello, We started to implement IPv6 support, but there is a problem with
> DHCP message. More correct, there is a problem that DUID (DHCP
> identifier that replaces old CID field) does not defined well for IPoIB.
> Does somebody know the answer how DUID should be defined ?

What is the problem with the DUID field?  It seems you would only need to filter it if using DUID-LLT or DUIT-LL, in which case you modify the HW type and the link layer address.  What does the DUID look like as sent by Windows?

-Fab