Re: [Madwifi-devel] Stuck beacon
Status: Beta
Brought to you by:
otaku
From: Benoit P. <ben...@fr...> - 2009-05-20 21:24:14
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Derek Smithies a écrit : > On Fri, 15 May 2009, Sven-Ola Tücke wrote: > >> madwifi has a history of stuck beacons as we all know. It has >> something to do >> with the IBSS-merge feature active in HAL or Hardware. > Not so. Stuck beacons are reported in AP mode also. > Right, same here (ie, stuck beacons happens in AP mode so it's not related to all the IBSS supposed mess). My own partial theory is close to what Derek says : - - every beacon interval, a beacon must be sent - - due to CSMA/CA rules, the beacon might not be sent at all. At this stage, CSMA/CA can be triggered by Derek's continuous stuff (which i never tested, I tested with a signal generator blasting a single carrier....) - - since the beacon is not sent, it's still in the HW queue somehow - - after some times, we end up with lots of beacons pending, which triggers the beacon stuck tasklet and says : "Stuck beacon; resetting..." I've read the 802.11 specs on that points and it's not clear in my mind. So I don't know what exact behavior we should expect. For sure, it might be possible to have a condition where beacons cannot be sent at the PHY.... Moreover, there is a side consequence if a madwifi client is connected to the AP. The client's own vision is that it is missing beacons and so, at some points, the beacon miss will be falsely triggered! One odd behavior, two very bad side effects! Regards, Benoit -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkoUdOQACgkQOR6EySwP7oLwQwCgx7Xvs9SDeT0UJBoUy//HNNd2 hLAAnj0uo0ebIk+gyodz2E2wTjEFRgsY =ZVVY -----END PGP SIGNATURE----- |