[Karmic Lucid Maverick] Atheros AR5211 - ath5k driver connection and performance issues

Architecture: i386
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: brice 1449 F.... pulseaudio
CRDA: Error: [Errno 2] No such file or directory
Card0.Amixer.info:
 Card hw:0 'I82801DBICH4'/'Intel 82801DB-ICH4 with AD1981B at irq 11'
   Mixer name : 'Analog Devices AD1981B'
   Components : 'AC97a:41445374'
   Controls : 26
   Simple ctrls : 18
DistroRelease: Ubuntu 9.10
HibernationDevice: RESUME=UUID=f5dfae80-7439-4d40-890b-fc06c78c51c8
InstallationMedia: Ubuntu 9.10 "Karmic Koala" - Release i386 (20091026.1)
Lsusb:
 Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
MachineType: IBM 2373G1G
Package: linux (not installed)
PccardctlIdent:
 Socket 0:
   no product info available
 Socket 1:
   no product info available
PccardctlStatus:
 Socket 0:
   no card
 Socket 1:
   no card
ProcCmdLine: BOOT_IMAGE=/boot/vmlinuz-2.6.31-14-generic root=UUID=8ee24573-73d0-498d-babd-af66746a8ef4 ro elevator=deadline hpet=force quiet splash
ProcEnviron:
 SHELL=/bin/bash
 LANG=en_DK.UTF-8
ProcVersionSignature: Ubuntu 2.6.31-14.48-generic
RelatedPackageVersions:
 linux-backports-modules-2.6.31-14-generic 2.6.31-14.16
 linux-firmware 1.24
Uname: Linux 2.6.31-14-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare
XsessionErrors:
 (gnome-settings-daemon:1493): GLib-CRITICAL **: g_propagate_error: assertion `src != NULL' failed
 (gnome-settings-daemon:1493): GLib-CRITICAL **: g_propagate_error: assertion `src != NULL' failed
 (nautilus:1554): Eel-CRITICAL **: eel_preferences_get_boolean: assertion `preferences_is_initialized ()' failed
 (polkit-gnome-authentication-agent-1:1563): GLib-CRITICAL **: g_once_init_leave: assertion `initialization_value != 0' failed
dmi.bios.date: 02/27/2006
dmi.bios.vendor: IBM
dmi.bios.version: 1RETDOWW (3.20 )
dmi.board.name: 2373G1G
dmi.board.vendor: IBM
dmi.board.version: Not Available
dmi.chassis.asset.tag: No Asset Information
dmi.chassis.type: 10
dmi.chassis.vendor: IBM
dmi.chassis.version: Not Available
dmi.modalias: dmi:bvnIBM:bvr1RETDOWW(3.20):bd02/27/2006:svnIBM:pn2373G1G:pvrThinkPadT40p:rvnIBM:rn2373G1G:rvrNotAvailable:cvnIBM:ct10:cvrNotAvailable:
dmi.product.name: 2373G1G
dmi.product.version: ThinkPad T40p
dmi.sys.vendor: IBM

Attached log files via apport-collect as requested after steps 1 and 2.

I can't connect at all to my home WPA-Personal AP. I could not get any better results by getting closer to the signal. AFAIK, I was using madwifi under Jaunty on the same hardware, and despite a slow rate that I thought was due to the old 802.11b adapter, it was reliable. I have a rt2870sta USB dongle which had a similar symptom (association to the AP failing); installing linux-backports-modules fixed the problem for it, but not for the built-in adapter.

The same hardware at the exact same distance from the AP works just fine in Win7. If I had no USB dongle, this would be a showstopper for me.

I found some other connection bugs with unencrypted networks. Have marked those as dups of this one. Also removed the Thinkpad reference as there are Toshiba and other hardware but all links back to ath5k.

Architecture: i386
ArecordDevices:
 **** List of CAPTURE Hardware Devices ****
 card 0: Intel [HDA Intel], device 0: AD198x Analog [AD198x Analog]
   Subdevices: 1/1
   Subdevice #0: subdevice #0
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: ubuntu 3305 F.... knotify4
                      ubuntu 3342 F.... kmix
CRDA: Error: [Errno 2] No such file or directory
Card0.Amixer.info:
 Card hw:0 'Intel'/'HDA Intel at 0xee240000 irq 17'
   Mixer name : 'Analog Devices AD1981'
   Components : 'HDA:11d41981,17aa2025,00100200'
   Controls : 20
   Simple ctrls : 11
DistroRelease: Ubuntu 9.10
LiveMediaBuild: Kubuntu 9.10 "Karmic Koala" - Release Candidate i386 (20091020.2)
MachineType: LENOVO 9459W57
Package: linux (not installed)
PccardctlIdent:
 Socket 0:
   no product info available
PccardctlStatus:
 Socket 0:
   no card
ProcCmdLine: BOOT_IMAGE=/casper/vmlinuz file=/cdrom/preseed/kubuntu.seed boot=casper initrd=/casper/initrd.lz quiet splash -- debian-installer/language=da console-setup/layoutcode?=dk
ProcEnviron:
 SHELL=/bin/bash
 LANG=da_DK.UTF-8
 LANGUAGE=
ProcVersionSignature: Ubuntu 2.6.31-14.48-generic
RelatedPackageVersions:
 linux-backports-modules-2.6.31-14-generic N/A
 linux-firmware 1.24
RfKill:
 0: phy0: Wireless LAN
  Soft blocked: no
  Hard blocked: no
Uname: Linux 2.6.31-14-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare
dmi.bios.date: 05/28/2008
dmi.bios.vendor: LENOVO
dmi.bios.version: 7CETD2WW (2.22 )
dmi.board.name: 9459W57
dmi.board.vendor: LENOVO
dmi.board.version: Not Available
dmi.chassis.asset.tag: No Asset Information
dmi.chassis.type: 10
dmi.chassis.vendor: LENOVO
dmi.chassis.version: Not Available
dmi.modalias: dmi:bvnLENOVO:bvr7CETD2WW(2.22):bd05/28/2008:svnLENOVO:pn9459W57:pvrThinkPadR60:rvnLENOVO:rn9459W57:rvrNotAvailable:cvnLENOVO:ct10:cvrNotAvailable:
dmi.product.name: 9459W57
dmi.product.version: ThinkPad R60
dmi.sys.vendor: LENOVO

The apport-collect above was collected using the Karmic RC Live CD and default ath5k module from the linux-image. Connectionspeed was checked several times using the ISPs reliable internet speed test, and the result was less than 256Kbit/ps downspeed and very unstable - link was checked to more than 2 Mbit/ps with an other Kubuntu 8.04 with the default ath_pci module loaded, this was done in between the Karmic RC tests - The boxes was connected to the AP one at a time, and no other access for other client boxes was allowed.

The LBM was tested on an new installed Karmic RC fra the same live CD and on the same box, and was updatet with all updates including the the LBM. The test results was even worse than with the default ath5k module.

A test on an enterprise Cisco WiFi network was done too with the same box, and the connection speed was even less than with the default module and wireless stack.
Compiling an ath-pci module from source (Ver. 10.0.5.6), and loading the module on the other hand showed the expected connection speed on the enterprise network, which is approximately 18.000 Kbit/ps (18 Mbps effective on a 54G network).

I will try to go on to step 3 later this night, and report back if I succeed.

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

Could this be related:

Apr 8 11:23:25 teej-laptop kernel: [ 1038.540099] No probe response from AP 00:1f:9f:eb:32:d9 after 500ms, disconnecting.

?

Also sorry about status change, clicked wrong button, fixed it back to Triaged.

https://bugzilla.kernel.org/show_bug.cgi?id=13892 from a duplicate seems more helpful

Changed upstream to more relevant bug report

Teej, have you tried to change channel on your AP ?, I can see your are running channel 11 (2,462 Mhz). Some atheros adapters have seems to have thees problems with channels above 6 (2,437 Mhz) - see my comment in #419 above.

I tried changing channels. It did nothing for the AR5211 I have. Even with the latest bleeding-edge compat-wireless.

@Christiansen: Already tried that with no effect.

I installed the madwifi drivers from "Hardware Drivers" and rebooted and the adapter was showing in iwconfig, but could not connect either through the Network Manager applet or manually with iwconfig. BUT...removing madwifi and rebooting again gained some results.
I can now connect to my AP but it is very intermittent and drops out as above but can browse for about 2 minutes now! Getting there...

For me, changing to lucid has worsened the performance. The zero-drops of connection speed (and I mean zero) are more frequent (like two times every minute or something like that) and the speed overall is worse as well.

Are we supposed to install the lucid backport package? I tried and so far it seems the signal is displayed to be a bit stronger (like 8-10 %) than with pure lucid but the actual performance looks unchanged worse than in karmic with karmic backports....

I have been running Lucid for a couple of weeks and are seeing hugely degraded performance. Backports in Karmic was OK.

Bug #461419 I am wondering if this bug have some other bearing on the performance ? (Could be completely wrong though)

I have just installed the backports module... seems a little faster

Is there any movement here ?

I am having ok performance but the Thinkpad hangs every day ..... Would be nice to get to the bottom of this bug as it has been around for a good couple of months.now. Can anyone help? I am willing to investigate if I can be guided?

Thanks!

I have to ask; since the first bug was filed, has there only been one card fixed? Am I the only one finding this absolutely pathetic?

Ok, this card that got fixed has seen quite an improvement (I know because I have that card) but what about the other very common card AR5211. That card has been completely broken for years. It barely connects with ath_pci (and of course that driver causes kernel panics) and has been removed from Ubuntu and ath5k doesn't connect at all even with the computer next to the router. Both ath5k and ath_pci are so bad that I even resorted to ndiswrapper, which performed better but caused even more frequent lockups than ath_pci.

IBM Thinkpad T40-43 are very common laptops and the fact that AR5211 does not work is quite a bad situation as it's embedded in many of these laptops. Laptops which many people use, after they have bought a new one, to test Linux for the first time.

Can anyone answer why there seems to be no work done on that card? Is Atheros not interested in fixing the driver for that card or what? It would be nice to know because then I would know if I should throw that piece of junk AR5211 away and buy an Intel card, or not.

Best regards
Julius
A recently happy owner of a T60 (AR5212)
but a long frustrated owner of a T40p (AR5211)

I am feeling a bit stupid not having tried this before.
My Packard Bell with Atheros ARBXB63 (see previously submitted apport info) would not associate with the accesspoint. The accesspoint was set on channel 8. I changed the accesspoint to 9 and now it works.

Some extra info : The same laptop booted on Vista ( I know, i know) does not have this problem. So is not the hardware, it has to have something to do with the driver.
Anyway.. my problem is solved, I hope this is helpfull to someone.

I was having random and frequent freezes/crashes with 2.6.32-22-generic lucid and network manager on my t41p with ar5212 chipset.
I removed network manager and installed wicd. That seems to have got rid of freezes crashes. But then it would connect fine at home with a wap2 ap but not at a friends place with a normal non encrypted ap. It would just not see any wireless network.

Going to command line and doing the following as root fixed it.
# to view available wireless
iwlist wlan0 scan
# to connect to a an ap ssid1
iwconfig wlan0 essid ssid1
# to get ip address
dhclient3 wlan0 &

The upstream bug report has not been updated for over 6 weeks. To ensure that this is fixed, can ALL affected users please make sure that you have selected "Affects me too" at the top of the bug report, underneath the bug title. Also if you have any debugging information, add it and add a note to the upstream tracker as this should help push the bugfix process through.

On another note, I can also confirm that this affects several other Linux distros, namely Mandriva, Fedora, Sabayon, Gentoo, Linux Mint, etc, on both i386 and amd64, so the sooner this is fixed the better for everyone :)

I would suggest that users affected first try out one of the latest compat-wireless module before marking this bug, as there has been a lot of changes to the ath5k module in the last month or so. And despite completely silence from the devs on the bug rapports, this has solved at least the problems I was experiencing - using the compat-wireless-2010-05-22.tar.bz2 and running ./scripts/driver-select ath5k before compiling.

Hence the reason for the wording "affected users". If you don't have any more problems then you're not affected ;)

Your absolutely right, and maybe some other users aren't too, if they give latest compat-wireless modules a go - hence "I would suggest that users affected first try out one of the latest compat-wireless module before marking this bug". Sorry if that is to hard to get, that there would then be a solution upstream, that could be used as an solution/update to ubuntu.

Installing the latest compat-wireless package (2010-05-27) solved the issues on my Thinkpad T61 (AR5212, 1014:058a) running a stock Lucid with kernel 2.6.32-22-generic #33.

Thinkpad X41 Tablet
Installed the compat-wireless package (from end of May). It now takes minutes instead of seconds to connect to the wireless network (and therefore several manual retrials) - error message is simple try1-3 and then timeout.

After a few minutes the connection suddenly works usually and then the dmesg output gets really strange with several cfg80211: World regulatory domain updated: - messages, I don't understand and never saw before.

The good part is, the connection is stable (at least the applet does not indicate a reconnection) but the connection still comes to a halt sometimes (once or twice a minute usually for a few seconds) - it feels better, anyway, thanks for that work!

System-upgrades (like today with open-office-updates) lead to time-outs in the update-GUI nontheless. So still not entirely there but it is finally at least getting a bit better since the connection does not drop out anymore.

I attached an example of the dmesg log.

This is a patch that was in the upstream bug report which appears to fix the problem for the users that have tested it. Leann, is there any chance we can get this into Ubuntu before Maverick is released?

I second that. The patch fixed my problem as well, even though I only used it to disable L0S, not L1. Recent compat-wireless doesn't compile for the latest lucid kernel (due to the wext refactoring thing), but with 2.6.35-4-generic from maverick it works perfectly.

Hi Teej,

I see this patch has been submitted to the upstream mailing list for consideration already:

http://thread.gmane.org/gmane.linux.kernel.wireless.general/51512

It's however still undergoing modification and review it seems and has not yet been applied. It's the preferred policy of the Ubuntu kernel team for patches to be accepted upstream first and then be pulled back into Ubuntu. Otherwise maintaining out of tree patches causes quite a bit of a maintenance burden. I'll try to keep an eye on the progress the patch makes, but feel free to ping me if you notice it is accepted upstream before I do. Thanks.

That's brilliant thanks for looking into this Leann, very much
appreciated. I'll keep an eye on the upstream status too :)

A fix was committed upstream yesterday: https://bugzilla.kernel.org/show_bug.cgi?id=13892#c27

The patch noted in the above comment #495 was included in the upstream stable 2.6.35.4 kernel. The Ubuntu Maverick kernel was recently rebased to the 2.6.35.4 stable kernel and uploaded:

https://edge.launchpad.net/ubuntu/+source/linux/2.6.35-20.29/

Please test this newest kernel once it's propagated through the archive. For now I'm marking this bug Fix Released. If you are still experiencing issues after installing this 2.6.35-20.29 kernel, could I request that you please open a new bug report rather than continuing to dog pile on this one. Thanks.

Unfortunately this hasn't fixed the issue I was having, but I know this is FAR too crowded now to make any difference. I will have another bug filed properly for the issue.
Thank you Leann for all your help :)