rt2x00.serialmonkey.com

Hello everyone,

A well-known problem in setting up a real ad-hoc network refers to the BSSID cell splliting problem, which I'm experiencing now. I'm setting up a network with a couple of nodes, some of them use usb dongle from Edimax, EW-7318USg. I'm not fully sure that they are causing the cell split, but as far as I have researched, the only way to fix this is by setting up a static BSSID and deactivating the BSSID selection algorithm in the driver.

So, I have tried:

# iwconfig rausb0 bssid <static MAC>

But nothing changed. I'm using the legacy rt73 driver. Can anyone clarify if this is supported or not in this driver, or if there is a patch for this functionality?

thanks in advance,


- Breno

[moved to correct forum, you state you are using rt73 legacy driver, so you should post on rt73 legacy forum]
[closed duplicate thread, don't start new topcics when you didn't get an answer on your first topic]

_________________
Regards,
Ivo van Doorn
Project Administrator
http://rt2x00.serialmonkey.com

I assume you really meanIf not, could you try that and see what happens? (Make sure its a "locally administrated" address.)

Thanks,

_________________
Yr Hmbl Obt Svt & c
Bryan - In favor of Big Oil. Big fan of General Grievous.

Hello Vern, thanks for replying.

I have tried iwconfig rausb0 ap <static MAC>, but nothing happens. I've been investigating this, and now I'm pretty sure that when I'm using realatively homogeneous adapters - such as several Edimax dongles and a single one from RealTek, the network is fairly stable. Turning on one of our Nokia N810 tablets turns everything into a mess - all other nodes start to get confused about what BSSID to choose, and then they split the whole network. The Nokias use a prism chipset, btw.

It's interesting to note that everyone is in-range in our tests. I have read several texts referring to the BSSID splitting problem in scenarios where nodes are distant or out-of-range of each other, and then get in trouble to syncronize their BSSIDs when getting in close range. In either case, setting a static BSSID and deactivating the automatic selection algorithm would solve the problem (that's what mostly everyone is doing in practical mesh networking today). But up to now only some Atheros chipsets and drivers seem to support this.

I was just trying to confirm for sure if the rt73 drivers or the new rt2x00 drivers support or do not support setting up a static BSSID, but I think the results from the tests attest to no support. If so, I'd like to know if it is possible to introduce this support as a patch or plugin to the driver in the future.

regards,

- Breno

For rt2x00 the "iwconfig wlan0 ap <bssid>" is not only supported it is mandatory to start an association attempt.

_________________
Regards,
Ivo van Doorn
Project Administrator
http://rt2x00.serialmonkey.com

Hello all,

Thank you a lot for the information! I should then really try out the the rt2x00 drivers.

One final question: is there any plans to make the rt2x00 a standalone driver? I mean, to decouple it from the kernel, and provide a module which can be plugged into any (compatible) kernel?

regards, best wishes

- Breno

you mean the compat-wireless package from linux-wireless.org?

_________________
Regards,
Ivo van Doorn
Project Administrator
http://rt2x00.serialmonkey.com

Hi brenojac,

The attached patch should let you specify a BSSID in adhoc mode using the "iwconfig wlan0 ap <bssid>" command. You can return to the default behavior by specifying a bssid of all zeros.

Could you apply it to a vanilla copy of the lates CVS and see what happens?

Thanks,

_________________
Yr Hmbl Obt Svt & c
Bryan - In favor of Big Oil. Big fan of General Grievous.

What? You just taking my word for this?

Y'all wuz warned.

(Hope this stuff works.)

_________________
Yr Hmbl Obt Svt & c
Bryan - In favor of Big Oil. Big fan of General Grievous.