Re: [babel] Work to do (IHU TLV AE 0)

[Denis Ovsienko <[e-mail]>]

---- On Sun, 25 Mar 2018 02:30:26 +0100 Juliusz Chroboczek  wrote ---- 
>Hello to all, 
> 
>We've had a productive IETF, and therefore we've got a lot of work to do. 
>My understanding of the current situation. 
> 
> 
>1. draft-ietf-babel-rfc6126bis 
> 
>This has been in last call, like, forever. We're still waiting for IESG 
>review, let's please make it happen. Donald, Russ, could you please poke 
>the responsible AD? 
> 
>(I'm still not happy about Unicast Hellos, but I understand I'm in the 
>minority, so I'll shut up. Er, no, I'll keep complaining loudly, but 
>I won't block the process.) 

There is one more important thing to check before setting 6126bis in stone. The current revision of 6126bis says an IHU TLV may omit Address (AE = 0). The current revision of 7298bis needs Address meaningfully set so that the receiving node can pick the TS/PC sub-TLV from the IHU TLV and translate that to the age threshold in units of seconds. If there is no Address field but there is a TS/PC sub-TLV and 7298bis tries to use that and somebody swaps the packets between two point-to-point links, I wonder if that makes another attack possible.

So far the implied destination address doesn't seem to be an obviously safe design. I currently see a few ways of getting rid of this edge case:
1. 6126bis and 7298bis coordinate to have IHU AE 1 or 3 when 7298bis is enabled
2. 6126bis simply mandates Address in IHU (AE = 1 or 3)
3. 7298bis does destination address protection too for all packets (7298 did only source, as 6126 is multicast-only).

I am still considering those options, feedback is welcome. One thing I wanted to confirm is whether point-to-point links in Babel are always meant to have addresses or they can be address-less.

>3. draft-ovsienko-babel-rfc7298bis 
> 
>Denis has produced a -00. I believe that Toke is interested both in 
>reviewing the draft and implementing it, and so am I. The subject of 
>adoption was raised at the meeting, and nobody objected. So: 
> 
> 3.1 is anyone opposed to adopting this draft? 
> 3.2 Denis, do you want to produce a new revision before adoption, or are 
> we welcome to adopt? 

OK to go into further work as it is, I have a rough plan of remaining changes, but cannot make everything in one quick go right now.

-- 

    Denis Ovsienko