Re: Question on IPv6 over IB

Vivek Kashyap <[email protected]> Thu, 13 Jan 2005 15:54:53 -0800 (PST)
Newsgroups gmane.ietf.ipoib
Message-ID <Pine.LNX.4.44.0501131547280.12924-100000@dyn319548.beaverton.ibm.com>
On Thu, 13 Jan 2005, Roland Dreier wrote:

>     Anand> However, if you look at the linux ipv6 ndisc
>     Anand> implementation, the pad is "appended" to the 20-byte
>     Anand> hardware address for infiniband and not "prepended". Could
>     Anand> someone comment on this?
>  
> Yes, it seems the Linux stack assumes that all link address types
> behave the same way as RFC 2464 IPv6-over-Ethernet, namely that
> padding in the ndisc option comes at the end of the link address.  If

Isn't the length of the option already 8 when used on Ethernet (type+length+
48bit)? Then it wouldn't require the padding.

> I'm reading the code in FreeBSD's nd6_ns_output() correctly, it makes
> the same assumption.
> 
> It would seem to make the most sense to update the IPoIB draft to
> state that link address padding goes at the end of the ndisc option.

Will it not align the local address at 2 byte boundary as against 4?

> 
>  - Roland
> 
> _______________________________________________
> IPoverIB mailing list
> [email protected]
> https://www1.ietf.org/mailman/listinfo/ipoverib
> 
>