> -----Original Message----- > From: Arend Van Spriel [mailto:arend.vanspr...@broadcom.com] > Sent: Sunday, September 18, 2016 23:00 > To: Luca Coelho <l...@coelho.fi>; johan...@sipsolutions.net > Cc: linux-wireless@vger.kernel.org; Beker, Ayala <ayala.be...@intel.com>; > Otcheretianski, Andrei <andrei.otcheretian...@intel.com>; Grumbach, > Emmanuel <emmanuel.grumb...@intel.com>; Coelho, Luciano > <luciano.coe...@intel.com> > Subject: Re: [PATCH v2 6/9] cfg80211: Provide an API to report NAN function > termination > > On 18-9-2016 21:56, Arend Van Spriel wrote: > > On 16-9-2016 10:33, Luca Coelho wrote: > >> From: Ayala Beker <ayala.be...@intel.com> > >> > >> Provide a function that reports NAN DE function termination. The > >> function may be terminated due to one of the following reasons: user > >> request, ttl expiration or failure. > >> If the NAN instance is tied to the owner, the notification will be > >> sent to the socket that started the NAN interface only > > > > So the driver is supposed to use this function from the .rm_nan_func > > callback (or .del_nan_func). How should the driver use this together > > with cfg80211_free_nan_func() function. > > Hit Send button too fast. Would it make sense to free the nan func implicitly > in cfg80211_nan_func_terminated() function or would there be reasons to > keep it.
Currently when the NAN interface is stopped (ieee80211_do_stop) the termination notification can't be sent, since ieee80211_vif_to_wdev will return NULL after SDATA_STATE_RUNNING bit is cleared. I guess it's similar not only for mac80211, and in case of interface removal we don't really want to send notifications for each function. > > Regards, > Arend
