Created attachment 29522 [details]
Stock iwlagn driver showing repeating errors and a final crash with systrace
After few hours connected to a WNDR3700 AP running DD-WRT I no longer had access to internet. I then decided to disconnect from the AP using nm-applet and reconnect to it. I got prompted for the password, which was already entered but when pressing OK I was unable to reconnect. Looking at the kernel.log I can see a lot of
"kernel: iwlagn 0000:02:00.0: Received BA when not expected"
and toward the end of the file there is a pretty severe crash containing two system traces. This happened to me twice yesterday.
This is not linked only to WNDR3700 AP only. I have had the problem with other brand/model or wifi access points. Furthermore, it takes much more time to crash when connected to 54G AP compared to 54N AP.
The above crash was using default iwlagn driver found in the 2.6.35 kernel running on an Archlinux install (64-bit). The corresponding kernel log is attached under kernel-default_driver-crash.log.
The I got a recommendation to try to switch to compat-wireless. It didn't crash during the day when connected to a 54G AP but I received a lot of "iwlagn 0000:02:00.0: Received BA when not expected". The kernel log for this session can be found under kernel-compat-54G.log.
Finally coming back home, I connected to my WNDR3700 and downloaded an Ubuntu CD. Connection died on (no IP traffic possible) while showing associated to AP. I had to reboot in order to be back online. Again this is with compat-wireless enabled and corresponding kernel.log is under kernel-compat-WNDR3700-crash.log
Created attachment 29532 [details]
Compat-wireless enabled, no crash on 54G but lots of repeating errors
Created attachment 29542 [details]
Compat-wireless enabled, lots of errors and crash (connection lost) but no systrace
event log in dmesg indicate that is a miss alignment between the processors of the ucode and the radio.
This is the same bug#16691 just show up on different devices.
The quick W/A is disable 11n, we are working on release new uCode now.
I see that there is a new linux-firmware 20100911 out there. Is it supposed to fix the problem?
Are you running the latest firmware available for your device?
If so, does the issue persist with the latest kernels (i.e. 2.6.38 or later)?
Closing due to lack of response...