On Tue, Sep 23, 2008 at 9:18 PM, Georg Lukas <georg@boerde.de> wrote:
Hi Ignacio,

* Ignacio García Pérez <iggarpe@gmail.com> [2008-09-23 20:59]:
> At some point in time, nodes C, D and E become no longer visible from nodes
> A and B. Wiring my laptop to either node D or E I discover that actually
> they are up and that they can only see each other. Everything else seems
> normal.

That sounds like a typical IBSS split problem. You can analyze it by
using tools like horst[1] or ibss-merge[2] on an additional machine
running in monitor mode.

I'll give them a try and report back tomorrow.

Also, iwconfig might show different Cell IDs in its output.

Please try the attached patch, best on top of madwifi-hal-testing (it is
already part of madwifi-dfs and might get integrated into madwifi-openwrt).
There is still one cheesy thing in MadWifi regarding IBSS merges, which
might cause such behaviour. So please test and report back if it works.

Thanks. I'll try tomorrow.

> And here comes the fun: *if I set the wireless interface of any node in
> promiscuous mode *(by launching tcpdump)* it starts receiving packets from
> the nodes in the other group.

If it is an IBSS split problem, that is a bug - only monitor interfaces
should receive frames not intended for the same BSS. Though it is rather
a minor one.

Any ideas on how I can check if that's a bug?

> wlan0     IEEE 802.11g  ESSID:"test_mesh_1"  Nickname:"node_250"
>           Mode:Ad-Hoc  Frequency:2.452 GHz  Cell: FA:BA:00:00:00:01
>           Bit Rate:0 kb/s   Tx-Power:19 dBm   Sensitivity=1/1
>           Retry:off   RTS thr:off   Fragment thr:off
>           Encryption key:????-????-????-????-????-????-??   Security
> mode:restricted
>           Power Management:off
>           Link Quality=45/70  Signal level=-51 dBm  Noise level=-96 dBm
>           Rx invalid nwid:212  Rx invalid crypt:0  Rx invalid frag:0
>           Tx excessive retries:0  Invalid misc:0   Missed beacon:0*

wlan0 is not very typical for madwifi.

I know, but my system is supposed to be able to use several types of WiFi cards and I tried to use an uniform naming scheme, so wlan0 is always the first card.

Now that you mention the subject, I've always though that the naming scheme for atheros devices would not affect at all functionality, but I've noticed (at least in 0.9.4) that in monitor mode there are differences (for example in the received power value) depending just on the naming of the interface. It is quite confusing IMHO.
Also, the Cell ID
(FA:BA:00:00:00:01) looks very suspicious.

I think I mentioned that one of the things I tried to fix the "group split" problem was to manually set the cell ID in all nodes using "iwconfig wlan0 ap", and that capture is from a setup in which I had tried that. I just made up that cell ID ("faba" is a kind of bean in the local language :-)


[1] http://br1.einfach.org/horst
[2] http://ivs.cs.uni-magdeburg.de/EuK/mitarbeiter/glukas/ibss-merge.scapy
|| http://op-co.de ++  GCS/CM d? s: a-- C+++ UL+++ !P L+++ E--- W++  ++
|| gpg: 0x962FD2DE ||  N++ o? K- w---() O M V? PS+ PE-- Y+ PGP++ t*  ||
|| Ge0rG: euIRCnet ||  5 X+ R tv b+(+++) DI+(+++) D+ G e* h! r* !y+  ||
++ IRCnet OFTC OPN ||________________________________________________||

Version: GnuPG v1.4.9 (GNU/Linux)


This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
Madwifi-users mailing list