MADWIFI

Hello Everyone,


I am having the issue Madwifi driver + AR2413 chipset transmitting higher
 power at lower band(<2.4012 Ghz) when there are no AP. Verified this by
 connecting spectrum analyzer and seeing the transmitted power.

 I am using the chipset AR2413 and madwifi driver svn revision 4022 and
linux 2.6.24.3. More
 details of

 lspci
 03:0f.0 Ethernet controller: Atheros Communications, Inc. AR5005G
 802.11abg NIC (rev 01)
        Subsystem: Atheros Communications, Inc. Compex Wireless 802.11 b/g
 MiniPCI Adapter, Rev A1 [WLM54G]
        Flags: bus master, medium devsel, latency 96, IRQ 20
        Memory at dfce0000 (32-bit, non-prefetchable) [size=64K]
        Capabilities: [44] Power Management version 4

-----------------------

 dmesg when the module is loaded
 MadWifi: ath_attach: Switching rfkill capability off.
 wifi0: Atheros AR2413 chip found (MAC 7.8, PHY 2112A 4.5, Radio 5.6)
 ath_pci: wifi0: Atheros 2413: mem=0xdfce0000, irq=20


 -------------------------

 I am using wpa_supplicant, the wpa_supplicant.conf settings are
 ctrl_interface=/var/run
 ctrl_interface_group=<groupname>
 update_config=1

 ----------------------------
 Since this is operated in germany, I am setting the regdomain to 0x37 in
 EEPROM of chip using ath_info utility
 ath_info -f -w dfce0000 regdomain=0x37

 Then loading the driver using countrycode settings
 modprobe ath_pci countrycode=276

 strating the wpa_supplicant
 ----------------------------------
 Note there are on access point available and when we measure the transmit
 power it shows 10dbm higher in lower band(<channel1(2.4012Ghz), this is
 violating the german ETSI regulations . As soon as access point/ssid
 becomes available the power level goes down and is within germany
 regulation.

 When does the maxtxpower gets set ? I am not sure if the driver is
 initializing all the values when I load the madwifi driver with germany
 country-code or it does later when it find a AP during scan. Any solution
 to this would help greatly, this blocking from product being passed for
 certification.



When is the maxtxpower set? As we see the txpower go down when it finds any
AP otherwise it very high at lower band(<2.4012Ghz)



-- 
Thanks and Regards,
Soumya

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

soumya metukul a écrit :
>
> Note there are on access point available and when we measure the
> transmit power it shows 10dbm higher in lower
> band(<channel1(2.4012Ghz), this is violating the german ETSI
> regulations . As soon as access point/ssid becomes available the
> power level goes down and is within germany regulation.
>
> When does the maxtxpower gets set ? I am not sure if the driver is
> initializing all the values when I load the madwifi driver with
> germany country-code or it does later when it find a AP during
> scan. Any solution to this would help greatly, this blocking from
> product being passed for certification.
Hello,

I do not fully understand your problem. On what frequency are you
doing your measures? What TX power do you expect? What TX power do
get? How did you connect the spectrum analyzer to your system
(splitter, cable, and so on)?

Regards,
Benoit

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkqFSi8ACgkQOR6EySwP7oI1uACg+hY0RjloW7TiLHXPx1pB6keU
UFQAoJj8f6C0k9CMEpdB7vYhLmKdMj4f
=fqyx
-----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

soumya metukul a écrit :
> I am attaching the graph, which can give more clear picture. As you
> can see from graph 2, the transmitted power in lowerband (<2.4012)
> goes beyond 28dbm, I am not sure how do we get this noise. Also the
> test environment is in chamber where there are no access point
> available and I think I driver is doing a scan which causes the
> transmission power to go so high. But as soon as we make Access
> point available as you can see form graph 2 the entire transmission
>  power drops down and also lower band noise is not seen and its in
> compliance. Are there any initialization of max txpower which
> driver does during scan when it finds the Access point.
>
> I really appreciate your help. I am new to madwifi driver and
> trying to understand what the driver is doing during scan. Also
> when I load the madwifi driver with country code(modprobe ath_pci
> countrycode=276), when does it initialise the maxtxpower settings
> and from when does it take effect?
>
> The spectrum analyzer is connected through cable. WE are scanning
> for frequency 2.37 to 2.5, 2.435 being the center frequency.
>
>
> Thanks, Soumya

I have seen your graphs [BTW, most spectrum analyzer have an option to
either do a nice screencopy or make a .PNG file, your scanned copy is
pretty ugly].
- From your graph, I still don't see the problem. 2412 MHz is a 802.11
channel, not 2401.2 (are you mixing things up?). Moreover, your TX
power is -28dBm on the first graph and -60dBm (the second file has 2
graph, I take for granted that one of those graph is the same as in
the first file...). What is the threshold for DE regulation?

I don't have a spectrum analyzer at hand so I cannot help much and I
just know that TX power is a very complex things.

Regards,
Benoit
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkqF02MACgkQOR6EySwP7oIAkACeLdvsSjWHQOQ1Jyx2zK1nI/kv
a3sAoORozIHWlbHmFehRlsttz0VD9kqo
=nlL2
-----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Andrea Tarasconi a écrit :
> Dear Benoit,
>
> the Atheros HAL has more than 100 bugs and there are at least 3
> bugs related to wrongly TX power handling.
>
> For that reason we decided 3 years ago to make our own Atheros HAL.
> This is the only way to make professional devices as our Essentia
> Wifless ESS are.
>
> For more details, visit our web site http://www.wifless.eu.
>
> Kind regards, Andrea.

We all know that Atheros HAL has bugs. The only problem with the
Atheros HAL is that its closed source and as such, cannot be fixed (or
worst, I can be fixed only at Atheros discretion).
So, I don't see the benefit from using another closed source like you
seem to propose...

There is something else I don't understand. Atheros is selling their
software along their hardware, but none of them is "certified". Ie
their software do not meet any compliance and their hardware is
produce by ODM, that again, do not check for any compliance (or if
they do, they do it with a special software that you don't have so it
voids the usefulness of such tests). For me, that is all wasted time.

Hopefully, the ath9k initiative, lead by Atheros itself, seems to be
the way to go.

Regards,
Benoit
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkqJxIcACgkQOR6EySwP7oLl5ACg8fyJBdUWWFxdEb46aSHXccS3
7TYAn2Gt9J9Rtl9guBhjVG9/G4u00DPs
=RS23
-----END PGP SIGNATURE-----