problems with Wireless event: new AP (00:00:00:00:00:00)

[Jouni Malinen]

On Tue, May 17, 2005 at 05:55:51PM +0200, Sebastian Weitzel wrote:

> (jkm:)
> >For me, this is not an association with a bogus AP BSSID, but a
> >disassociation from the current AP. If the driver is generating this
> >event in a case where it is not actually being disassociated/
> >deauthenticated, the driver should be fixed..

> Maybe http://article.gmane.org/gmane.linux.drivers.hostap/10487 
> allready tried to patch this issue. I applied the patch and retested. 

No, that is a completely different issue.

> It still dont solves the problem. As you can see the ifindex differs 
> which should lead to reject this event, but does not.
> 
> 
> ifi->ifi_index=6
> drv->ifindex(8) != ifi->ifi_index(6), reject event

?? this was rejected..

> fromlen=12 addr=00:00:00:00:00:00
> ifi->ifi_index=8
> Wireless event: cmd=0x8b15 len=20
> Wireless event: new AP: 00:00:00:00:00:00

This is another event, isn't it?

> I dont know yet why the wireless event still leads to blacklisting. 

Because the driver is reporting a disassociation!

I took a closer look at madwifi and what is happening (at least in my
test case) is that the association is completed successfully, but
Atheros mac reports a beacon miss and the driver interprets this as the
AP has disappeared, which is not really the case here. This was
relatively easy to reproduce at least with long beacon interval (1000
TU).

This is starting to look like the most commonly reported problem with
madwifi driver* when using wpa_supplicant and I would really appreciate
it if someone would take a closer look at what is happening with beacon
miss events. Either the hardware is configured incorrectly or there are
bogus bmiss events. Either way, the driver should filter these out and
not report disassociation. If the time sync for beacon miss cannot be
fixed, the driver could, e.g., use probe request to verify whether the
AP is still in range before processing the beacon miss event.

*) see, e.g., http://hostap.epitest.fi/bugz/show_bug.cgi?id=44

-- 
Jouni Malinen                                            PGP id EFC895FA