rt2x00.serialmonkey.com

When I try and bring the interface up the first time I get this:

This is with no usb devices plugged in so I'm not entirely sure where those messages are coming from! If I try bringing the interface down and then up again I get:

This is on my eee with ralink rt2860 and the latest git snapshot, rt2800pci module loaded. If you want anything more specific do let me know.
Tom

The interface now comes up and down without error. However, "iwlist scan wlan0" shows no results, nor does the card associate with an AP. Dmesg is showing more or less the same:

Then when bringing up the card with ifconfig the bottom four lines are repeated.

Nothing is getting received or transmitted according to ifconfig.

A new error has emerged! On bringing up the interface now, I get this message in dmesg:

As far as I can tell the card is still not transmitting or receiving.
Appreciate your work!
Tom

Hi,

I just starting trying to get the RT2860 in my Eee Box B202 running (goal: work as access point) some days ago. I read through all information that looked like being potentially useful and finally got to the same issue as pezboytom just described here. Although not being a kernel developer, but having seen (and created) a bit of C code years ago (actually I'm a Java developer), I looked at parts of the code as far as I could, just using common sense, to see whether I could at least narrow the region where something goes wrong. Added some more debugging statements to parts of the code, actually related to the "startup" of the device and loading of the firmware (which I had to configure correctly first), but didn't really succeed.

Just for the case that something of this still might be interesting, here are some of details I got from my "tests":
- rt2x00lib_request_firmware was finally OK, i.e. it returns 0
- The firmware loaded (it has been put into the kernel, so it's not loaded from the file system by now) has 8192 bytes, which matches the size of the original firmware file, so seems to be OK.
- rt2x00dev->ops->lib->load_firmwarert2x00dev, rt2x00dev->fw->data, rt2x00dev->fw->size) returns 0 = OK, too.
- Same for rt2x00lib_start(...), rt2x00lib_initialize(...) and rt2x00lib_start(...)



The only thing I'm not really sure about is that some of my "traces" were not written, i.e. the respective code was not run. Looking closer at the code around, which is _request_firmware(...) in drivers/base/firmware_class.c, I found that that function "quickly" returns with 0 if it finds the requested firmware in the kernel itself ("built-in"), so there is some code which is not executed then, and it doesn't look absolutely obvious to me that all of it is really only relevant if the firmware is read from the user space (I don't know exactly how that works, that's why I chose the built-in firmware by now - please enlighten me here ).

@Ivo: Can you tell whether
1) there might be any relation at all (maybe it's a completely wrong location where to look) and whether
2) some of that code should actually be called - in that case, please give me some hint how I can test with "external" firmware, i.e. which component is responsible for providing it at runtime. Just some basic hint so I can continue to collect the required information.

If I can help in any way to solve this, please let me know. As I mentioned, I don't have experience with kernel programming, but I can do some easy modifications like adding tracing code or slightly modifying existing behavior, if that helps.

Thanks and best regards
Heiko

rt2x00 doesn't work with built-in firmware, it will always need the external firmware file to operate.
Other then that there is no point in following the error codes from the rt2x00 functions, they will all return 0. The problem is the hardware register initialization and either there are some missing initialization calls or they are not in the correct order. THe best thing is comparing the initialization with the original Ralink drivers to see if there are critical differences.

The behaviour of the card has changed with the latest snapshot from git. Trying to bring the interface up I now get this output:
The led on my eeepc goes out at this point and it becomes impossible to switch back on without restarting. Upon trying to remove the module rt2800pci, the system crashes completely.

Any progress with "master mode"? Managed to start the card with hostapd (it is in master mode, and there is mon.wlan0) but i can't see the network from my laptop or connect to "hidden ssid". All this with WPA2 enc, but the problem is also there without any enc. Any suggestions? PS using 2860 pci card.

How did you manage that? With standart Ralink RT2860_Linux_STA drivers?

No with the driver from git. With original(ralink) driver it works perfect as client, but from what i know master mode is not supported Unfortunately i need it to work as Access point

Shock horror, does this mean that the changes get made to git without them coming up on the web interface? I was under the impression that there had been no update since 23rd may and that snapshot crashes my card completely!

Does anyone know if this driver is now working in managed mode?

From the git web:


I tried them with my SparkLAN WPIR-501. Just to be precise it's in fact a RT2800pci, RT2860 MAC/BBP and RT2820 RF.
Before the new release, when modprobing rt2800pci (staging) I used to get lots of this (and the card wouldn't work).

Now, when modprobing the drivers:

Which seems alright to me, but I'm no expert.
Then when I try to put it in master mode with iwconfig, this happens. (Is that normal? )

With that being said, hostapd *seems* to start just fine and ends up saying:

I've yet to test if it really works since I'm remote right now, but iwconfig doesn't show anything wrong (maybe except the levels, but I didn't connect any antennas for the time being).

Oh, and also, dmesg still throws


Btw, many thanks to IvD for all the work on the drivers. Also thanks to Mattias, Mark, Felix and Xose for the latest patch work. Sweet piece of code guys.

I'll give it a try tomorrow. The thing with iwconfing wlanx mode master + err is normal. Only way to get it in master mode is hostapd.

For what its worth, I just tried the rt2800pci drivers. It appears that I cannot send wireless packets, either in managed mode or AP mode.

AP mode with hostapd, when I tcpdump the monitor interface, I see iwlist scans from my test client, and see SSID responses. However, tcpdump and monitor from the client never sees the response.

Managed mode, I can do iwlist $dev scan sometimes and see SSIDS. However, I can't connect to even an open AP.

FWIW, both work much better with a rt2800usb-based adapter I had kicking around.

Anything I can do to help get this going?

Is there any way to compile the rt2800pci driver without recompiling the entire kernel from your git?
BTW: compat-wireless does not include it in its latest "bleeding edge" version (2009-07-14).

No.



That is because compat-wireless is based on the upstream wireless-testing.git and not rt2x00.git.