On Jul 19, 2010, at 4:54 PM, Robert Cragie wrote:

> I'm still confused by this. Earlier, Zach said that in fact we should not be 
> using GP16 in the src IP address of the NS. I totally agree.

Sorry, that was a temporary brain-short (which I corrected). In nd-10 the NS 
src IP address must be the address being registered, and the SLLAO the LLA 
corresponding to that address. In the case of ZigBee IP, that would be the GP16 
and the 16-bit MAC address. 

The main reason for using the address being registered as the NS src IP address 
is related to security. We would like to preserve compatibility with SeND for 
more general applicability of this mechanism in the future. 

I am not totally caught up on the 6lowpan and ZigBee IP mailing lists (just 
came back from vacation yesterday), but to throw a few possible options out 
here:

a. Do things as in nd-10 and require more specialized code for this, this was 
the conclusion the authors came when closing ticket #87. Most 6lowpan 
implementations are specialized stacks anyways - is this really an issue?

b. Move the address being registered to the ARO and add a flag (breaking SeND 
compatibility). 

c. Allow use of the 64-bit MAC address in the SLLAO of the NS even when the IP 
src address is a tentative GP16. This doesn't really solve the resolution 
problem worrying Dario and Robert though.

I will add a bullet point about this to the ND presentation for the WG meeting 
tomorrow.

Zach   

-- 
Zach Shelby, Chief Nerd, Sensinode Ltd.
http://zachshelby.org  - My blog "On the Internet of Things"
http://6lowpan.net - My book "6LoWPAN: The Wireless Embedded Internet"
Mobile: +358 40 7796297

_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan

Reply via email to