[rt2x00-users] Low link quality / or disconnect after scan with RT3572 and Xirrus AP

[Ronald Wahl]

Hi,

I have some strange issues with an RT3572 based WLAN USB Stick and my 
Xirrus AP. Maybe one of you can help me debugging this:

My platform:

Embedded platform using AT91SAM9G20
Full speed USB (i.e. 12 MBit USB link speed)
WLAN Stick is an ASUS USB N-53 based on Ralink RT3572
The AP is an Xirrus XN4.
The Linux kernel version is 3.2.5. I'm using the nl80211 API to access 
the kernel in in wpa_supplicant.

The problem:

The quality of the WLAN connection is initially usable (RSSI: -50db) and 
after a scan it drops immediately to low values (RSSI: -75db or lower) 
or the STA is disconnected altogether immediately. This happens in the 
5GHz and the 2.4GHz bands. The 5GHz band tends to just drop the quality 
and the 2.4GHz band often just disconnects but this is probably just a 
minor detail. After the supplicant reconnects everything is back to 
normal. Also if I restart the scan when quality is low the link quality 
is back to normal during this scan but drops again after the scan 
finishes. A precondition to this is some medium traffic on the 
connection (in my case an snmp bulkwalk in an endless loop against the STA).

This issue happens only on my the embedded platform which is a lot 
slower than a regular PC (even if I use an USB 1.1 hub to slow down the 
USB link). I can reproduce this only with my Xirrus XN4 AP and not with 
an  Cisco Aironet 1142 AP.

Also the scan times are much longer on the embedded platform - for each 
channel it takes almost a second to scan. So a full scan took 37 seconds.

For testing purposes I applied all the RT2X00 patches from wireless-next 
branch before the VCO patches because there were some changes rearding 
the RT3572 chip. But this did not change anything on the behavior.

I also tried using the wext driver instead og nl80211 in wpa_supplicant 
but this does also not change anything.

An interesting observation might be also that it looks like that the 
station does not send packets anymore to the AP because after some time 
the station is disconnected by the AP because of inactivity and also the 
bulkwalk that I have running does only timeout after the scan.

I suspect that because of the long scan time something times out and 
some driver/chip state is not restored that is necessary for regular 
operation. Might this be true?

Any ideas how to debug / fix this?

Thanks,
ron

PS: I'm not in the list so keep me in Cc:, please. Thx!

-- 
Ronald Wahl - ronald.wahl at raritan.com - Phone +49 375271349-0 Fax -99
Raritan Deutschland GmbH, Kornmarkt 7, 08056 Zwickau, Germany
USt-IdNr. DE813094160, Steuer-Nr. 227/117/01749
Amtsgericht Chemnitz HRB 23605
Geschäftsführung: Stuart Hopper, Burkhard Wessler