Unable to pass traffic - signal/antenna issue?

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=4559


15-01-2008 05:33:41

Now that I'm able to associate with my Gigabyte GN-WBKG using the rt2500usb driver (thread link[/urlmkzawa9q]), I'm trying to pass traffic over the interface, but am almost completely unsuccesful at doing so. I have seen several threads referencing signal strength issues, rate issues, and antenna tuning issues, all of which seem to be inter-related, but I haven't seen any solution arise. What I'm hoping to do it describe the behavior that I am seeing and get an updated status on the issues involved. It should be noted that I'm working on migrating from the rt2570 driver, so some comparisions will be done against that (working) driver.

I'm able to bring up the interface and associate, but almost immediately after associating watch the bitrate fluctuate between 11, 2 and 1 Mbps before finally settling on 0. Running `iwconfig wlan0 rate 11M` doesn't seem to have any effect.

Viewing the output of `ifconfig` shows that some packets are being transmitted and received, but the count is much lower than would be expected, even for just trying to obtain a DHCP lease. I've tested this by manually setting an IP address and route, and trying to ping the AP with `ping -f $IP -c 500`. The ping output reports 500 packets sent, with an error every so often. The ifconfig output, however, only reports an increase of 5-15 packets in the TX count.

The Link Quality statistic is much different under rt2500usb than values used for rt2570 (this is on a desktop and the physical location is always the same). Link Quality under rt2570 is 69-75 (-74 to -77 dBm), whereas with rt2500usb it is 47-48 (-75 to -79 dBm), but when the bitrate drops to zero, it shows Link Quality dropping to around 28 and Signal Level is (-73 to -80 dBm). The link quality reading is consistent whether read when associated, or as the output of a scan.

Finally, at some point after the rate hits zero and the interface appears dead, it is made official by the following message from the kernel, and removing the RUNNING flag from the ifconfig output
wlan0 No ProbeResp from current AP 00022d20e6a6 - assume out of range

An aside that I have noticed while doing testing is that after associating with the rt2500usb driver and having the interface fail as described above, subsequently loading and running the rt2570 driver works, but only for a short duration (~5-10 minutes). I'm presuming that registers are left set in some incorrect state and the rt2570 driver doesn't properly initialize them. This may or may not help indicate possible issues, just throwing it out there as a data point.

Any status updates, ETAs, or requests for additional data are plenty welcome - thanks for your help and work so far.


15-01-2008 19:00:51

It is most likely an antenna issue, the only to debug that is by checking register settings.

I have attached a script to deump all relevant registers from debugfs.
If you could use it to create a dump of the non-working setup and a working setup with the same device I can analyze the data and try to find the problem.

For a working setup you'll most likely need the legacy driver compiled with the "make debugfs" option.


15-01-2008 19:45:45

The dumps will be similar to the dumps I provided when diagnosing the "Association timed out" issue here[/urlljac7n6e].

If you'd like I can recreate the new dumps after being associated this evening (do you need the CSR register contents, and if so has the legacy driver been fixed?).


15-01-2008 19:56:46

The log for rt2x00 will look differently since now the association has kicked in as well as "better" link tuning results. This also affects the antenna selection. So a new dump is very usefull.

I have just found a bug in the rt2570 legacy driver which caused the invalid csr reading. )


16-01-2008 02:28:52

Okay, here are some new dumps. Please let me know what you find out or if you need more information.


19-01-2008 18:44:22

Any ideas on this or other testing I can do?


19-01-2008 19:23:12

Not at the moment.
I have looked into the debug traces, and the antenna setup is the same for all versions. So apparently that is not the issue.

Could you try using wireshark to see if the frames are getting out (and only the response is missing).
And if they aren't getting out of the device, could you use the TX/RX frame dumping in debugfs and check what rt2500usb reports as outgoing frame?


26-01-2008 13:04:46

It does not seem the packets leave the card.
If I watch the traffic using wireshark, I recieve taffic from the router and another client on the network.
To be sure I ran airodump-ng with a laptop while trying to communicate through the buggy card. There's no traffic recognized.
I am running 2.6.24 now and I don't seem to have a frame/dump entry on debugfs.

Edit Ok, I just read it's in the git. I will give it a shot now.


26-01-2008 14:02:11

I captured the dump using rt2wtap and opened it with wireshark.
I just get a list of packets, their time and nothing else.
And a note from wireshark

user encap not hanlded DLT=147, check you Preferences->Protocols->DLT_USER[/quote1zqn2oyy]

I'll check with this dissector now.


26-01-2008 15:47:09

I now have the readable dump opened in Wireshark.
What I see is a lot of association requests and malformed packets.
Tell me how I can help.