Wifi can't connect at all

[hiciu]

I tried on 2007.02 and 2008.08, always getting this:

root@om-gta02:~# iwconfig eth0 essid my_ssid
 Error for wireless request "Set ESSID" (8B1A) :
     SET failed on device eth0 ; Invalid argument.
root@om-gta02:~# iwconfig eth0 essid any
root@om-gta02:~# dmesg|tail
usb0: full speed config #1: 500 mA, Ethernet Gadget, using CDC Ethernet
ADDRCONF(NETDEV_CHANGE): usb0: link becomes ready
usb0: no IPv6 routers present
Alignment trap: phone-kit (1441) PC=0x000114bc Instr=0xe5963054 Address=0xbec55683 FSR 0x013
Alignment trap: phone-kit (1441) PC=0x00012f40 Instr=0xe5980038 Address=0xbec55667 FSR 0x013
Alignment trap: phone-kit (1441) PC=0x000114bc Instr=0xe5963054 Address=0xbec55683 FSR 0x013
Alignment trap: phone-kit (1441) PC=0x000114bc Instr=0xe5963054 Address=0xbec55683 FSR 0x013
Alignment trap: phone-kit (1441) PC=0x000114bc Instr=0xe5963054 Address=0xbec55683 FSR 0x013
eth0: no IPv6 routers present
AR6000 disconnectedAR6000 disconnectedAR6000 disconnected
root@om-gta02:~#  

It strange, because iwlist scan works. Logs says imo nothing about this.

My kernel: ​http://buildhost.openmoko.org/daily/freerunner/200808/20080829/uImage-om-gta02-latest.bin
logread after iwconfig essid my_essid: ​http://wklej.org/txt/1763/
dmesg after iwconfig essid my_essid: ​http://wklej.org/txt/1764/

[hiciu]

("Bug Filing Policy" says to write tickets in this way)
Kernel: ​http://buildhost.openmoko.org/daily/freerunner/200808/20080829/uImage-om-gta02-latest.bin (and other kernels too)
How to reproduce:

iwconfig eth0 essid my_ssid

Results:

 Error for wireless request "Set ESSID" (8B1A) :
     SET failed on device eth0 ; Invalid argument.

Expected:
Wifi connect's to accesspoint.
Reproducible:
Always.

[kalavan]

Error appears only when trying to set ESSID one-character long. Longer ESSID names are set properly.

[werner]

Indeed, there's some confusion in the Atheros stack about how the ESSID is
passed and stored. It also gets confused if the ESSID has a length of 32
bytes. This might be because the format of the data passed in the ioctl has
changed from including the final \0 to not including it anymore.

I don't want to change this before having a better understanding of what's
happening with the ESSID inside the stack. So I'd recommend, as a
work-around, to avoid ESSIDs of length 1 or anything larger than 31 bytes.