Re: No scan results from wl1271-sdio

[Luciano Coelho <luciano.coelho@nokia.com>]

On Tue, 2010-11-23 at 13:51 -0500, ext Josh Watts wrote:
> On Tue, 2010-11-23 at 19:51 +0200, Luciano Coelho wrote:
> > Hi Josh,
> > 
> > On Tue, 2010-11-23 at 10:34 -0500, ext Josh Watts wrote:
> > > I've been trying to get the wl1271-sdio driver to work, and so far I've
> > > managed to get the device to start up, but am unable to see any access
> > > points. I'm using the latest head from Luciano's wl12xx tree, and for
> > > firmware I'm using the files committed to linux-firmware by Shahar Levi
> > > (wl1271-fw.bin version 6.1.0.0.343). I'm wondering if there's maybe some
> > > minor component or setting that I'm missing (like radio-enable, or
> > > perhaps limit to 802.11a, etc)?
> > 
> > If you got that far, the scan "should" work. ;) My main guess is that
> > your device is not calibrated properly (the NVS that Shahar provided is
> > a "default" one and may work, but the results will vary wildly depending
> > on your RF).
> > 
> > Which device are you using? If you bought a end-user device, you may
> > find the appropriate NVS lying somewhere in the filesystem.  If you're
> > working on a new device, you will need to get it calibrated.
> > Unfortunately we don't have a publicly available application to do that
> > yet (but there's one under work).
> 
> We are using the TiWi-R1 from LS Research. Connected to the U.Fl
> connector is a custom antenna board with a ceramic chip on it. Is this
> calibration specific to the TiWi-R1 itself, or to the entire system
> including antenna?

There are two main parts in the NVS.  One is more or less per design.
Meaning that one type of board needs certain parameters.  This is the so
called "Radio Parameters" in the NVS.  Only the manufacturer or someone
with appropriate knowledge and equipment can provide you with that (I
have no clue about the details, because I'm RF illiterate).  I suggest
you ask this from LS Research, since the probably have that already.

The other part is per device.  Each individual device needs to be
calibrated.  You don't need any special equipment for that, the firmware
can perform all the steps needed.  But you need an application that
sends the commands to the firmware and reads the results (via
NL80211_TESTMODE).


> If my particular implementation requires calibration, what kind of
> expertise and equipment would we need to perform it? (I myself am no RF
> engineer ;-) If not, it seems I should get in contact with someone at LS
> Research to see about obtaining the proper NVS file?

The radio parameters part, as far as I know, is more important than the
per-device calibration.  But you won't be able to achieve good
throughput and you'll see lots of retransmissions if you don't calibrate
it properly.


> > > With all debug information turned on in the driver, I see plenty of
> > > communication going back and forth w/o any (noticeable) errors, but I
> > > don't see any sign of magic strings such as any of the ESSIDs I expect
> > > to see in the area.
> > 
> > Please send the debug information to us and we may be able to figure
> > something out from the logs.  You probably don't need DEBUG_SDIO,
> > because that one will dump a *lot* of stuff that will probably not help
> > here...
> 
> See attached (wlan0.log.20101123_1317 201.4KiB) for the debug log (copy
> of /var/log/messages) sans DEBUG_SDIO.

I'll take a look at it now...


-- 
Cheers,
Luca.