[PATCH v2 00/20] mesh support for wpa_supplicant

Jouni Malinen j
Sun Nov 16 08:21:00 PST 2014


On Fri, Oct 24, 2014 at 09:19:41AM -0400, Bob Copeland wrote:
> Sure, makes sense.  That sequence I posted comes from Figure 11-4 in
> 802.11-2012, where AP has gone from Nothing -> Confirmed (due to "Com"
> event) which does "1(0),2" i.e. send tr#1 + tr#2 as part of the
> transition.  It looks like mesh case will be a bit different than
> AP<->STA in any case, so I guess the question is, would it be best to:
> 
>  - modify the SAE state machine to do:
> 
>      if AP<->STA, transition to Committed on tr#1 reception
>      without sending tr#2, and from Committed state to Accepted on
>      tr#2 reception (sending tr#2).

The SAE finite state machine does not actually say "transmit" for
1(0),2; it says "generate".. Anyway, I don't think this state machine
was designed with infrastructure BSS case in mind. It might be useful to
discuss that in the IEEE 802.11 TGmc (maintenance group) to see if the
AP<->STA case should be handled separately or at least noted in the
state machine design.

It is relatively straightforward to handle these additional state
changes in the same state machine implementation, so I added such
operations based on hapd->conf->mesh. This seems to work fine now for
both mesh and non-mesh cases.
 
-- 
Jouni Malinen                                            PGP id EFC895FA



More information about the Hostap mailing list