[PATCH 00/20] wpa_s: p2p: support nl80211 P2P_DEVICE interface

[Arend van Spriel]

On 05/19/2013 09:08 PM, Jouni Malinen wrote:
> On Thu, May 16, 2013 at 03:28:15PM +0200, Arend van Spriel wrote:
>> This patch series adds support for the nl80211 P2P_DEVICE interface
>> concept. It has been based partly on the patches sent by David Spinadel.
>> This series differs from that David's series in the fact that the
>> P2P Device interface is represented by a struct wpa_supplicant instance.
>>
>> The code changes has been tested with one peer supporting P2P_DEVICE and
>> one peer with a so-called dummy net device, ie. p2p0. Peers found
>> each other using p2p_find and connection was established with p2p_connect
>> using pbc method.
>
> Thanks. I started applying commits, but had to stop with that pretty
> quickly since this broke number of things in my tests (including hostapd
> build and wpa_supplicant start).. I think I managed to fix most of the
> issues, but this still does not work in my hwsim tests.
>
> I did quite a bit of cleanup and re-ordering of commits etc. to avoid
> breaking things in the middle of series. It is possible that I may have
> broken something here, but I didn't find what it could be. As such, I
> did not push any of this into the master branch yet, but you can find
> the latest version I tested with in the devel branch. It would be
> helpful if you could verify whether it still works for you. It should be
> noted that I did not get the unmodified versions working either (after
> fixing the TDLS startup issue with a one-liner), so I'm not completely
> sure what to expect from these changes with mac80211_hwsim.

Thanks, Jouni

I will have a look at the devel branch. Will not be until tuesday though 
due to holiday. What would be needed to run your hwsim tests if I want 
to look into that.

> One thing that came up, though, is that this is likely to break upper
> layer expectations since this changes default behavior by creating a new
> control interface for the P2P Device functionality. This will likely
> need to be made configurable to avoid that kind of default behavior
> change. This could be either build or runtime configuration, I guess,
> but I have not yet went through all the use cases so don't know what
> would be the best approach.

Ok. I put p2p_mgmt flag in wpa_supplicant struct, which could be used to 
skip creating a control interface for it if it is preferred to keep the 
behaviour the same. I forgot to mention, but there may be more stuff 
that is not really needed for this type of interface (eg. eapol related 
stuff) that I left as is.

Gr. AvS