Monitor interface for P2P

Ecco (Inhyang) Park eccopark at broadcom.com
Fri May 27 23:55:04 EDT 2011


Hi Jouni
Thanks you for the response.

I understand what are you worried about.

We will find out the good solution to be acceptable for P2P.

After making the patch for that, I will update you.

Regards,
Ecco

-----Original Message-----
From: Jouni Malinen [mailto:j at w1.fi] 
Sent: Thursday, May 26, 2011 2:08 PM
To: Ecco (Inhyang) Park
Cc: hostap at lists.shmoo.com; Neeraj Kumar Garg; SK (Sang-Ku) Lee; Lin (Sunnyvale) Ma; Henry Ptasinski; Greg Goldman
Subject: Re: Monitor interface for P2P

On Thu, May 26, 2011 at 10:44:48AM -0700, Ecco (Inhyang) Park wrote:
> > What do you mean with "p2p interface" and how would you receive (and
> > send) management frames using it?
> 
> I mean p2p-eth0-x for p2p virtual interface.
> We can receive management frames using NL80211_CMD_FRAME, but need to modify some codes in driver_nl80211.c

OK, that sounds fine. I would assume this would be used only for P2P
Action frames in that specific group while other P2P (Public) Action
frames would be delivered similarly on the first interface (eth0 in this
case?). Frames to (and from) both the P2P Device Address and one or more
P2P Interface Addresses could be received/sent when a group is running.

> Yes, I already comment out in wpa_driver_nl80211_hapd_send_eapol to prevent sending EAPOL frame through monitor interface.
> Instead of that, we can use p2p virtual interface(p2p-eth0-x) for sending EAPOL frame.

OK. Though, this needs to be done dynamically at runtime based on driver
capabilities.

> Do you mean that we need driver capability in cfg80211 to check whether the driver support monitor interface or not ?

Yes, or well, some other flag indicating that monitor mode is not needed
in AP mode would probably be better way of handling this. In other
words, if this new flag is not indicated, the old mechanism (monitor
mode) will be used. If it is indicated, nl80211 commands would be used
instead.

> In my opinion, we can define some flags like MONITOR_SUPPORT  in Makefile of wpa_supplicant to check monitor capability at compile time.

No, that's not acceptable. The same binary needs to work both with
drivers that use monitor interface and with drivers that don't.

-- 
Jouni Malinen                                            PGP id EFC895FA




More information about the HostAP mailing list