[PATCH 11/12] mesh: Send peering open frame again if beacon from listen state peer is received

Bob Copeland me at bobcopeland.com
Tue Nov 11 08:26:21 EST 2014


On Sun, Nov 09, 2014 at 06:36:12AM +0000, Nishikawa, Kenzoh wrote:
> Using the existence of a sta_info as a flag of peering success is the
> root cause of this problem.
> 
> Peering should be managed by the state machine independently.
> A sta_info should be related to the existence of the node, the beacon
> of the node independently.
> So the existence of a sta_info should not be related to the peering success.
>
> I think the original patch and the patch to be committed to the
> open80211s can solve the problem.

It does feel kind of kludgy to have to delete the station when we
know it's there just to make the MPM happy, so I suppose I'm not
opposed to the proposed kernel patch, especially as it does at least
filter on user_mpm and plink state, giving userspace the option to
not care.  We'll need to check how authsae handles new peer candidate
events if it already knows about the station or if auth is in progress.

In any case, I guess the kernel patch needs to be submitted upstream
for discussion.

The submitted mesh patch needs to be cleaned up a bit to remove the
unrelated changes, though.

-- 
Bob Copeland %% http://bobcopeland.com/


More information about the HostAP mailing list