[support] RAs leaking into adjacent wifi networks

[Georgopoulos, Panagiotis]

Hello again,


> -----Original Message-----
> From: Arnaud Ebalard [mailto:arno at natisbad.org]
> Sent: 03 September 2009 16:46
> To: Georgopoulos, Panagiotis
> Cc: support at jules.nautilus6.org
> Subject: Re: [support] RAs leaking into adjacent wifi networks
> 
> Hi,
> 
> "Georgopoulos, Panagiotis" <panos at comp.lancs.ac.uk> writes:
> 
> >> Are AP1 and AP2 on close channels (like 6 and 7)
> >
> > The sta interface is connected to AP1 (channel 2) but hears RAs from
> AP2
> > (channel 5). When I bring AP2 down, it hears RAs from AP3 which
> operates in
> > channel 7.
> 
> very weird.
> 
> > I might try to use channels that are far away, e.g. 1, 7, and 13
> > to see what happens.
> 
> yes, even if AP3 is already 5 channels away from AP1.
> 
> >>and have you by (lack of) luck put the interface of the MR in
> promiscuous
> > mode?
> >
> > Hm, that is interesting. I thought of that before, and I realized
> that I
> > have not explicitly configured the interface to work in promiscuous
> mode,
> > however I did at some point enable promiscuous mode in wireshark. I
> am not
> > quite sure how this affect the operation of the madwifi drivers, but
> in
> > order to debug my mobility testing I have to use wireshark and
> capture
> > traffic in promiscuous mode...:-/
> 
> Hit by the Heisenberg uncertainty principle? ;-)
> 

Hm, I would say it follows the chaotic theory ;-) 

Btw, I have disabled the promiscuous mode setting in wireshark for the last
couple of hours and I have not seen that (ie. hearing RAs from adjacent
networks) happening again... let's see how it goes though. 


> >> Just a guess though. You should definitely ask on madwifi ml. If you
> >> happen to find the root of the issue, I'll be interested.
> >>
> >
> > No responses on the madwifi mailing list... I can inform you if I
> find
> > something. Are you interested in this because you have experienced
> this
> > behaviour yourself or is it just a generic interest ? ;-)
> 
> generic interest.
> 
> > Is it something that you hear first time?
> 
> Yes, but it sounds like a nice bug/side effect.

Hm, I think you would avoid the word "nice" if you were in my shoes ;-)


> Once, I had the reverse version of that kind of bug where an broadcom
> ethernet interface had to be in promiscuous mode to receive multicast
> traffic: it worked fine while debugging things with tcpdump but no more
> when stopping tcpdump.
> 

This is what I call nice:-P

Thanks for your reply,
Cheers,
Panos