[Prism54-users] iwspy functionality w/ prism54

Jean Tourrilhes jt@hpl.hp.com
Mon, 1 Mar 2004 09:42:17 -0800


On Sat, Feb 28, 2004 at 02:40:20AM -0500, brf@sailor.bunzy.net wrote:
> On Sat, Feb 28, 2004 at 12:04:08AM -0500, Luis R. Rodriguez wrote:
> > Attention *Jean*
> > 
> > On Tue, Feb 24, 2004 at 10:42:50PM -0600, Bob Farmer wrote:
> > > Since there is currently no iwspy functionality in the Prism driver, we 
> > > (myself and/or colleagues) are considering implementing it ourselves. 
> > > We require the ability to collect link quality stats on the individual 
> > > nodes in an ad hoc network.
> > > 
> > > However, there does not seem to be any clue in the driver as to how one 
> > > would obtain the link quality data.  It appears that the driver is 
> > > obtaining link quality stats in AP mode by querying it via the BSSID, 
> > > but I'm guessing that wouldn't help for ad hoc mode.  Does anyone know 
> > > how one would go about obtaining the data?  
> > 
> > I'd check other drivers and see what they are doing. If its about
> > intercepting 802.11 Ad-Hoc packets and generating wireless statistics
> > for each node I think that then this is generic. Why would it be
> > driver-specific? Unless of course, I've missed the whole concept of
> > iwpsy.
> 
> I've looked at the Orinoco driver (which has working iwspy support) and it
> appears to get a "quality" field with each frame it receives, and then it
> keeps track of the latest quality on a per-MAC basis (for each MAC being
> "spied")...  In the prism54 driver, I didn't see evidence of any such info
> currently being pulled (or even available at all) for each frame or MAC.
> 
> The driver-specific part is getting the signal/noise info.  That's where I 
> was hoping you might be able to offer clues.

	Correct. I believe that each Rx frame contains some signal and
noise info. Intersil had this feature in PrismII, so I would expect to
find it in Prism54 as well. I know that some driver also export some
signal and noise info through the monitor interface, so that's another
area where to look at.
	Also, remember that translating the raw signal level value to
dBm is the responsability of the driver.
	Note that the iwspy driver interface has dramatically changed
in WE-15, in WE-15 there is much less to do in the driver. Of course,
the Orinoco driver doesn't have the new code, so you may want to check
the Aironet or HostAP driver.

> > I'm not sure if we can legally contribute to a GPL driver if we sign an
> > NDA. Of course, I'm not sure since I don't even know what the NDA looks
> > like.
> 
> Ugh.

	Depend how the NDA is written. I've seen NDA made on purpose
the write GPL driver (i.e. so just protecting the docs themselves).

> Thanks,
> Bob Farmer

	Jean