Beagleboard, Ubuntu Wireless Issues


I’ve been trying to connect a wireless card (TRENDNET TEW-624UB) but i’ve been having been unsuccessful.

When I plug the wireless card into my desktop (Ubuntu 9.04), it works without any problems. When I connect the card to my beagleboard through a powered high speed hub, usb0 becomes available to ifconfig. Problem is, iwconfig claims that there are no wireless extensions available on that interface.

After poking around, I found the rt2870 driver is used on my desktop. The firmware for this chipset is included in linux-firmware and I did include this when I build my image.

Upon some further searching, I’ve found that in my desktop setup, /lib/modules/2.6.28-12-generic/kernel/drivers/net/wireless/ has a folder rt2x00 which appears to contain the shared library the kernel would use. The similar directory on my bb only has drivers for zd1211rw.

And I’m stuck. I’m very new to alot of this but I understand “ko”'s are shared libraries, thus copying them from my x86 distro to my ARM distro should not work (right?). I would appreciate any and all advice or suggestions on what to do next.

I have the same problem with a bluetooth card but I figure that if I can figure out one of them, a similar solution will (hopefully) work for the other.

Many thanks,
Christopher Prevoe

Hi Christopher,

What kernel version are you running?

With 2.6.29-oer3x the rt2x00 modules are built-in so you won't see the
*.ko's in the directory:


However they may have the same issue as the zd1211's and have to be
moved out as external modules.

Does it make a difference if you plug the device into the USB port
after bootup/login? (thus allowing the file system with the firmware
to be accessible during hotpluging)


Hi Robert,

Thanks for your reply. I'm using 2.6.29-oer36 so you may be right
about them being built in. I've tried plugging in the device both
before and after boot up with no difference.

I'd like to amend something in my previous email, in the best case,
the card shows up as USB0; but most of the time it does not show up as
anything. I just did a set of tests to try to figure out when it does
show up as usb0 and I could not get it to load as usb0 at all.

Are there any benefits to having the drivers built into the kernel
instead of as modules? I don't understand the difference functionally;
why not just build them all as modules?

The following are my test results from this morning. I have saved the
kernel panic dump/stacktrace/log (what do you actually call these)
from ttyS2 so I have 3 of them available on request.

Many thanks,
Christopher Prevoe

P.S. 130609 - bb usb wireless tests

All trials done through ttyS2 (serial) and a Dynex powered usb 2.0
high speed hub
uname -a: beagleboard #2 Thu Jun 4 23:24:08 UTC 2009 armv7l
To ensure the hub is not the issues, a 500mb flash drive is plugged
into the hub and it showed up as /dev/sda every time
It is notable that the hub provides power to all devices even when the
hub is not plugged into a computer (bb or desktop)

Trial 1: Device plugged into hub at boot up
- lsusb does show the device as present
- ifconfig -a does not show the device at all
- Removing and replacing the device has no effect.

Trial 2: Device not plugged in at boot, hub plugged in at boot, once
booted, log in then use hub
- lsusb shows the device as present
- ifconfig -a does not show the device at all
- Removing the device then replacing it has no effect
- Note: unplugging usbhub from bb resulted in a kernel panic; I saved
the serial output on that one but have no clue what to do with it

Trial 3: Boot system without USB hub plugged in, boot, login, then
plug usb hub in
- lsusb shows the device as present
- ifconfig -a does not show the device at all
- Removing the device then replacing the device has no effect.
- Note: unplugged usbhub from bb again but no kernel panic occurred

Trial 4: Plug usbhub into 'desktop'. Desktop establishes a wireless
connection fine. Boot and login to bb, then unplug hub from desktop
and plug into bb
- Kernel panic - saved

Trial 5: Retry trial 4 two more times
- Kernel panic - seems the kernel panic in trial 4 is reproducable.