[PATCH v2 00/20] mesh support for wpa_supplicant

[Bob Copeland]

On Sun, Oct 05, 2014 at 10:47:48PM +0300, Jouni Malinen wrote:
> I still need to review the details of SAE changes since those can affect
> non-mesh cases. In addition, it would be nice to get somewhat more
> description commit messages for couple of the commits, like these:
> 
>     nl80211: Process mesh point interface like station interface

Please drop.

>     nl80211: Do not reset mesh point interface type on deinit

Please drop.  I dug up the original commit that added this line and its
commit log didn't make sense to me.

>     nl80211: Reject deauth/disassoc for mesh interface

The original commit log for that patch (which got squashed somewhere
along the way):

    Author: Chun-Yeow Yeoh <yeohchunyeow at gmail.com>
    Date:   Fri Apr 12 17:54:48 2013 +0800

    nl80211: deauth and disassoc are not relevant to mesh

    sta_info call these to send out the disassoc and deauth frame
    which are both not relevent to mesh. So don't send them.

    Signed-off-by: Chun-Yeow Yeoh <yeohchunyeow at gmail.com>

>     nl80211: Do not set zero STA capability (mesh)

I think the intent here was actually to avoid nl80211_set_station() failing
when capability was specified at all for mesh interfaces, as with supp rates.
So perhaps it should go under if (!params->set) {} conditional instead.

>     nl80211: Do not try to change supported rates

Maybe something like:

Linux prohibits station updates from modifying the supported rateset
except for TDLS interfaces.  When marking a station authenticated
(as in mesh), avoid passing supported rates.

[cfg80211_check_station_change() is the responsible function, FWIW]

>     nl80211: set_key updates for mesh
>     nl80211: Add WPA_IF_MESH (TODO: why?)
>     write SAE key_mgmt to config

To be continued...

-- 
Bob Copeland %% www.bobcopeland.com