Questions about WPS

[Jouni Malinen]

On Tue, Jan 05, 2010 at 07:52:44AM -0800, Sam Leffler wrote:
> On Wed, Dec 30, 2009 at 2:51 AM, Marcel Holtmann <marcel at holtmann.org>wrote:
> > Initial testing with some of the WPS capable access points here also
> > reveals that they don't really bother to update the IE correctly.
> >
> > So telling if WPS is actually usable or if WPA/RSN has been manually
> > configured is kinda hard.

> I thought updating the IE was part of the spec and ap's that don't can't
> pass certification--but it's been a while and I'm not sure I read the final
> version.  Perhaps it got moved to a manual step (i.e. you manually bring the
> ap out of this mode).

The IE must be changed for some use cases (e.g., PBC), but not all.
Unfortunately, a large number of workarounds is also needed with
deployed implementations, so this is not as simple as one would hope..

wpa_supplicant will need to track the IE changes, but I'm not fully sure
how much the D-Bus API clients would really need to know (and they
certainly should not be required to know about all the silly
workarounds that are needed to make this work).

wpa_gui does use couple of additional signal, i.e., active-PBC-AP-found,
active-PIN-AP-found, active-WPS-AP-found ("active" meaning selected
registrar attribute == TRUE in this case). This information could be
parsed from the WPS IE(s) (concatenated version of them, i.e., just the
TLVs), but it would probably be better to remove the need from the D-Bus
API users to parse the TLVs (they could obviously still do it based on
the full IE data if really needed for some special cases).

-- 
Jouni Malinen                                            PGP id EFC895FA