Wireless flaky on Acer Aspire 5100 after installing Natty

well now I can't even make it FAIL reliably! On this boot the wireless again came up pretty quickly.

When it is failing I get lots of timeouts in syslog. Usually network manager doesn't even try for several minutes and then when it does, there will be dozens of failures. In this case it only failed for two sets of three tries. Which to me is wonderful...

Apr 20 08:51:19 AcerAspire5100 kernel: [ 51.128794] wlan0: direct probe to c0:3f:0e:b9:f3:b2 (try 1/3)
Apr 20 08:51:20 AcerAspire5100 kernel: [ 51.328049] wlan0: direct probe to c0:3f:0e:b9:f3:b2 (try 2/3)
Apr 20 08:51:20 AcerAspire5100 kernel: [ 51.528046] wlan0: direct probe to c0:3f:0e:b9:f3:b2 (try 3/3)
Apr 20 08:51:20 AcerAspire5100 kernel: [ 51.728056] wlan0: direct probe to c0:3f:0e:b9:f3:b2 timed out
Apr 20 08:51:32 AcerAspire5100 kernel: [ 63.446709] wlan0: direct probe to c0:3f:0e:b9:f3:b2 (try 1/3)
Apr 20 08:51:32 AcerAspire5100 kernel: [ 63.644035] wlan0: direct probe to c0:3f:0e:b9:f3:b2 (try 2/3)
Apr 20 08:51:32 AcerAspire5100 kernel: [ 63.844283] wlan0: direct probe to c0:3f:0e:b9:f3:b2 (try 3/3)
Apr 20 08:51:32 AcerAspire5100 kernel: [ 64.044054] wlan0: direct probe to c0:3f:0e:b9:f3:b2 timed out
Apr 20 08:51:43 AcerAspire5100 kernel: [ 74.851447] wlan0: authenticate with c0:3f:0e:b9:f3:b2 (try 1)
Apr 20 08:51:43 AcerAspire5100 kernel: [ 74.852944] wlan0: authenticated
Apr 20 08:51:43 AcerAspire5100 kernel: [ 74.852973] wlan0: associate with c0:3f:0e:b9:f3:b2 (try 1)
Apr 20 08:51:43 AcerAspire5100 kernel: [ 74.855423] wlan0: RX AssocResp from c0:3f:0e:b9:f3:b2 (capab=0x411 status=0 aid=4)
Apr 20 08:51:43 AcerAspire5100 kernel: [ 74.855429] wlan0: associated

A quick note regarding the acer-wmi driver and your hotkey.

That driver should support your machine, but it won't load automatically due to a typo (incorrect capitalization) in the wmi guid within the driver. This can be easily fixed. Your logs show that it loaded successfully (I'm assuming you loaded it manually in this instance), but then I see some errors from the AML parser from execution of one of the wmi methods. That's seems likely to be a bios problem, but I'm not sure right now what the implications are.

You won't see the Acer wmi hotkeys device in lsinput because the wmi event guid the acer-wmi driver uses doesn't exist on your machine. The hotkeys may be reported on your AT keyboard device, and some may generate ACPI events and no key events. To check, use lsinput to find your AT keyboard device, then run:

  sudo /lib/udev/keymap -i input/event<n>

where input/event<n> corresponds to the device you found with lsinput. While this is running, press your wifi hotkey. Then do the same while running:

  acpi_listen

Thanks!

cold-started laptop from complete shutdown and the wireless didn't come right up. I let it try to connect for a few seconds and it failed and asked me to reiterate password. Just after that moment (Apr 20 09:01:53 in log) I un-checked "Enable Wireless" in the network manager applet. Then a few minutes later (Apr 20 09:03:27 in log) I checked "Enable Wireless" in the network manager applet, and the network connected successfully after a short delay.

Attaching excerpt of kern.log from this boot until after successful connection.

@Seth I just saw your update so I will go through that next.

Attaching syslog for same time period as kern.log above

I have NOT modprobe acer-wmi yet (though it doesn't seem to change lsinput)

merle@AcerAspire5100:~$ sudo lsinput
[sudo] password for merle:
/dev/input/event0
   bustype : BUS_HOST
   vendor : 0x0
   product : 0x5
   version : 0
   name : "Lid Switch"
   phys : "PNP0C0D/button/input0"
   bits ev : EV_SYN EV_SW

/dev/input/event1
   bustype : BUS_HOST
   vendor : 0x0
   product : 0x1
   version : 0
   name : "Power Button"
   phys : "PNP0C0C/button/input0"
   bits ev : EV_SYN EV_KEY

/dev/input/event2
   bustype : BUS_HOST
   vendor : 0x0
   product : 0x3
   version : 0
   name : "Sleep Button"
   phys : "PNP0C0E/button/input0"
   bits ev : EV_SYN EV_KEY

/dev/input/event3
   bustype : BUS_HOST
   vendor : 0x0
   product : 0x1
   version : 0
   name : "Power Button"
   phys : "LNXPWRBN/button/input0"
   bits ev : EV_SYN EV_KEY

/dev/input/event4
   bustype : BUS_I8042
   vendor : 0x1
   product : 0x1
   version : 43841
   name : "AT Translated Set 2 keyboard"
   phys : "isa0060/serio0/input0"
   bits ev : EV_SYN EV_KEY EV_MSC EV_LED EV_REP

/dev/input/event5
   bustype : BUS_HOST
   vendor : 0x0
   product : 0x6
   version : 0
   name : "Video Bus"
   phys : "LNXVIDEO/video/input0"
   bits ev : EV_SYN EV_KEY

/dev/input/event6
   bustype : BUS_I8042
   vendor : 0x2
   product : 0x7
   version : 4785
   name : "SynPS/2 Synaptics TouchPad"
   phys : "isa0060/serio4/input0"
   bits ev : EV_SYN EV_KEY EV_ABS

merle@AcerAspire5100:~$

had lots of trouble with kemap -- it immediately starts scrolling hundreds of lines so I missed the "press Esc" message and had to increase the terminal buffer just to find it... also pressing the wlan button shuts off the radio and I have to manually toggle network manager so it will successfully reconnect

$ sudo /lib/udev/keymap -i input/event4

scan code: 0xD5 key code: wlan

(again I have not yet done modprobe acer-wmi )

I see no messages from acpi_listen at all, either before or after modprobe acer-wmi

@Tommy:

It look like all your hotkeys are coming in on the AT keyboard device anyway, so it won't really matter whether or not you load the acer-wmi driver. The bios doesn't appear to be using wmi to report any key events, so you'll never see any difference in the lsinput output or the key events being generated when you load the acer-wmi driver. The hotkey is generating the appropriate key event, so nothing is broken there. I'd still be interested to know if there's any output from acpi_listen when you press the hotkey.

To clarify a comment you made on bug #710738 -- you said that pressing the wifi hotkey disabled wifi, but you had to use the network manager panel applet to re-enable it. Can I infer from this that the wifi hotkey fails to enable wifi when it is disabled? Does the behavior differ with or without the acer-wmi module loaded?

Sorry, I know the hotkey is the less important issue in this bug report, I just wanted to close the loop on the conversation we started on the other bug while it's fresh on my mind.

Whoops, looks like you added the acpi_listen information while I was typing :) Thanks!

I have not been able to discern any effect of loading acer-wmi except for the new ACPI Exception and ACPI Error messages in syslog and kern.log

The workaround protocol I have developed for dealing with the wireless:

1) Boot the laptop & log in
2) Look at the Network Manager wireless indicator.
2a) If indicator is "hollow" then the wireless is not going to even attempt to connect (no ESSIDs listed) but that may change spontaneously sometime within the next few hundred seconds.
2b) If indicator has pulsing waves then it is trying to connect, but it may try to establish the auto connection and in a few seconds will ask for the credentials of the access point.
2c) If the indicator has pulsing waves SOMETIMES it will connect after just a few seconds.

If either 2a or 2b failures are detected, it seems to help to un-check "Enable Wireless" in network manager, wait a while (a minute?) and check it again. After maybe 30-60 seconds it will usually switch from the hollow to pulsing waves and successfully connect.

I usually leave the wireless power switch alone. If you press it, the light goes off. Press it again and the light comes back on. HOWEVER after turning wireless back on, if you do not subsequently toggle the "Enable Wireless" setting off then on in Network Manager, the wireless connection will fail anyway and will not successfully establish a connection to the access point.

SO if I flip the wireless power off and on, I need to be sure to do the same with Network Manager.

I hope this description is helpful in describing what I see.

I could have worded one part better --

2b) If the indicator has pulsing waves then it is trying to auto connect to the access point, but it may fail to connect and ask for the credentials. Even if I re-type the credentials the access point will usually not connect at this point, and it works best to disable and re-enable wireless using Network Manager.

OK never mind (maybe). I just spent several hours trying all sorts of things to make the Acer 5100 laptop connect wirelessly. I turned on and off all the settings, rebooted, logged in and out, removed and restored the ath5k nohwcrypt=1 setting, waved a dead chicken, etc. and nothing worked...

UNTIL I opened a little netbook nearby (running Ubuntu 10.10, if that matters) and woke it from sleep and connected it to the AP. As soon as the netbook was up and running the Acer 5100 connected too, with no further intervention on my part.

I don't know why or how.

I don't know that I can point fingers at the wireless radio signal from the netbook because my cell phone has been connected to the same AP all day, and it has been in my shirt pocket the whole time, too.

Weird.

That is pretty strange. Is it reproducible?

I'd also suggest testing the latest mainline kernel build. Instructions for installing mainline builds are available at:

https://wiki.ubuntu.com/KernelMainlineBuilds

so far I haven't reproduced it, but I haven't been here much today. I did let the machine boot and sit tonight and it connected on its own after about 20 minutes. I think that may be its "typical" behavior.

I am somewhat confused by how to use the mainline builds despite reading through that page. If I download the version linked as the "latest" http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/current/ -- at this moment it contains an amd64 kernel version "2.6.39.999" so wouldn't I need to compile my own modules? I found a page http://kernel.ubuntu.com/~kernel-ppa/info/kernel-version-map.html -- this says I would want to match the mainline build closest to what I already have, which (if I am reading it right) it looks like would be 2.6.38x so I could use the modules I already have.

I have compiled kernels before but that's obviously not what you meant.

I may have a few minutes to tinker with this tomorrow. Thanks for your advice...

more circumstantial evidence this morning of a link between the sytems; I started AcerAspire5100 at 06:42:05. It finished all its ordinary messages in syslog/kern.log by 06:42:21 (last messages from rtkit-daemon about supervised threads) but no wireless connection. After a minute or so I woke the netbook from sleep (sorry I didn't notice the exact time) but after the netbook's usual delay connecting, within a few seconds AcerAspire5100 had kern.log messages at 06:45:33 saying wlan0: direct probe to [MAC address], two sets of three tries and timeouts, and then authenticate with the AP at 06:45:57.

The netbook has a Broadcom 4313 wireless chip, if that matters.

Apparently my module concern was unfounded. Sorry for the lack of understanding.

I looked back and found the last daily mainline build that includes i386 and installed it... 2.6.39-999-generic #2001104191714

On first boot, the behavior IS different under this kernel. For one thing, it attempted to connect immediately (unsuccessfully, but at least it attempted). I will try some things over a few reboots, including sleeping this netbook during the next few attempts and documenting what I see. If you can point out some things I should look for specifically (do you look at kern.log or syslog or both or neither)

sorry I was wrong. I had the netbook going so it was trying to connect. The behavior is NOT different under the mainline kernel.

It is fairly consistent... with the netbook off, when I toggle Network Manager's "Enable Wireless" off and then on, I get a series of messages in syslog:

Apr 22 07:39:35 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 3 -> 2 (reason 0)
Apr 22 07:39:35 AcerAspire5100 NetworkManager[781]: <info> (wlan0): deactivating device (reason: 0).
Apr 22 07:39:35 AcerAspire5100 NetworkManager[781]: <info> (wlan0): taking down device.
Apr 22 07:40:02 AcerAspire5100 NetworkManager[781]: <info> (wlan0): bringing up device.
Apr 22 07:40:02 AcerAspire5100 kernel: [ 125.661418] ADDRCONF(NETDEV_UP): wlan0: link is not ready
Apr 22 07:40:02 AcerAspire5100 NetworkManager[781]: <info> (wlan0): supplicant interface state: starting -> ready
Apr 22 07:40:02 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 2 -> 3 (reason 42)
Apr 22 07:40:03 AcerAspire5100 wpa_supplicant[874]: Failed to initiate AP scan.
Apr 22 07:41:14 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 3 -> 2 (reason 0)
Apr 22 07:41:14 AcerAspire5100 NetworkManager[781]: <info> (wlan0): deactivating device (reason: 0).
Apr 22 07:41:14 AcerAspire5100 NetworkManager[781]: <info> (wlan0): taking down device.
Apr 22 07:41:30 AcerAspire5100 NetworkManager[781]: <info> (wlan0): bringing up device.
Apr 22 07:41:30 AcerAspire5100 kernel: [ 213.378302] ADDRCONF(NETDEV_UP): wlan0: link is not ready
Apr 22 07:41:30 AcerAspire5100 NetworkManager[781]: <info> (wlan0): supplicant interface state: starting -> ready
Apr 22 07:41:30 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 2 -> 3 (reason 42)
Apr 22 07:41:31 AcerAspire5100 wpa_supplicant[874]: Failed to initiate AP scan.

(this was two toggle attempts with nebook asleep)

After connecting the netbook, here's what happens in syslog:

Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0) starting connection 'Auto aureola'
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 3 -> 4 (reason 0)
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0) Stage 1 of 5 (Device Prepare) scheduled...
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0) Stage 1 of 5 (Device Prepare) started...
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0) Stage 2 of 5 (Device Configure) scheduled...
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0) Stage 1 of 5 (Device Prepare) complete.
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0) Stage 2 of 5 (Device Configure) starting...
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 4 -> 5 (reason 0)
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> Activation (wlan0/wireless): access point 'Auto aureola' has security, but secrets are required.
Apr 22 07:43:03 AcerAspire5100 NetworkManager[781]: <info> (wlan0): device state change: 5 -> 6 (reason 0)
Apr 22 07:43:03 AcerAspire5100 Net...

trying some different kernels from the mainline archive

2.6.35-02063512 (20110411 maverick) -- won't boot? (didn't try hard)

2.6.36-02063601 (20101123 natty) -- boots with difficulty and network came right up (without netbook assistance)

will load some other kernels and test a few major releases to see if I can find the approximate location of the regression... ?

OK ... I have just about run out of time this am so I will summarize my testing so far.... let me know if you want me to try any particular Ubuntu kernels.

version 2.6.39-999.201104191714_i386 --from http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/2011-04-19-natty/
 this kernel behaves pretty much like the current kernel (newer kernels in this directory didn't have i386). Boots perfectly, network NOT connected immediately.

I started using kernels from http://kernel.ubuntu.com/~kernel-ppa/mainline/?C=M;O=A

02063512.201104011012 -- (latest Maverick kernel) NO BOOT (hangs with garbled display)
02063601.201011231330 -- boots w/minor issues, network comes up immediately
020637.201101050908 -- boots w/minor issues, network comes up immediately
020638.201103151303 -- NO BOOT (hangs with fan on high)
02063801.201103241244 -- boots perfectly, Network NOT connected immediately.

(I'm running now on 2.6.36-1 to confirm. It came up fine.)

I am out of time now but it looks like the next things to try will be 2.6.37-1 through 2.3.37-6 to see if I can find the regression

I really need to get to work today... but I tested two more I had already downloaded....

02063706.201103281005 -- boots perfectly, network comes up immediately
02063804.201104221009 -- boots perfectly, network NOT connected immediately

SO far none of the 2.6.37 kernels exhibit the problems, and all of the 2.6.38 ones do

Wow, thanks for all the testing. You've given me a lot of good information to work with.

I'm travelling this weekend, so I'm not really going to have a chance to go through all of it until Monday. If I can't get any clues as to what's happening from the logs we can try a kernel bisection, but if the regression is between 2.6.37.x and 2.6.38-rc1 the bisection could be difficult.

I had a look at the changelog and yes there were lots of entries for ath5k (assuming that may be the problem) for .rc1.

And as far as I can tell all of them have the problem, though it is harder to get .rc1 and .rc2 up; if I were a tougher CLI jockey I would set up a test case using terminal utilities and try it via a console.

020638rc1.201101192228 -- boots badly (1)(2). Network scans but not connected immediately
020638rc2.201101220905 -- boots badly(1)(2). Network scans but not connected immediately
020638rc3.201102010912 -- boots OK after message(1). Network not connected immediately

(1):message: /proc/device-tree can't find root (on console)
(2): video terrible but switching consoles gets X to redraw

This may not be germane to this issue, but my netbook developed an issue when I installed Maverick 10.10 that I initially thought was related to what I have been seeing on the AcerAspire5100. When I wake the netbook from sleep, it takes about 30 seconds before it even starts scanning for the local access point. It connects reliably, it just waits that many seconds to even start. The netbook did NOT show this behavior under Lucid 10.04 (it scanned and connected as soon as I woke it from sleep).

I have not yet tried to boot Natty on the netbook, but obviously I will be very sensitive to any network issues on it too. I am mentioning it only because it is mysterious too. Also since the netbook has a Broadcom chip I wonder if the actual culprit on the AcerAspire5100 may be something OTHER than ath5k.

From the Acer5100-20110420-syslog attachment I see that the probe requests to the access point appear to be repeatedly timing out. On the successful connection the authentication (obviously) gets past that point, but I don't see anything that suggests why that attempt was successful and previous attempts failed. I'm not really familiar with the protocol, so it's taking me a while to sift through the data.

Does that log represent the behavior you're typically seeing when trying to connect? Do you have the same trouble with all access points, or is it only with that particular one?

Sorry, we are just winding down after an afternoon and evening of tornadoes. Sadly they are predicted again for tomorrow.

I usually only connect to the one access point, and haven't compared the connection failures until this experience.

There are others in the neighborhood, sometimes unsecured, and a public access point (which may be secured by WEP) at a restaurant a couple of blocks away. If it would be helpful to compare the logs when connecting to different types of access points I could easily set up an unsecured one and/or connect to the public one and have a look at the logs, but that will be tomorrow at the earliest. Or should I stick to different WPA connections?

Reading the kernel wireless debugging information, it looks like they prefer seeing wpa_supplicant scripts and logs. I started reading the man page but I haven't started tinkering with it yet. Let me know if that sort of information would be better for your purposes so we can rule out any Network Manager weirdness.

The "best" kernel for this hardware has been http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.37.6-natty/

> merle@AcerAspire5100:~$ uname -a
> Linux AcerAspire5100 2.6.37-02063706-generic #201103281005 SMP Mon Mar 28 11:33:42 UTC 2011 i686 athlon i386 GNU/Linux

I am attaching an extract of the syslog when booting from this kernel but otherwise using current Natty software.

I would say this represents ideal wireless behavior as the link is active "immediately" upon login. This is to the same access point I always use (Netgear WNR3500L if that matters).

I just checked to be sure and I still am using the file /etc/modprobe.d/atheros-workaround.conf containing:

options ath5k nohwcrypt=1

@Tommy,

The most helpful log for this problem is the syslog, as it includes the logs from the kernel, NetworkManager, and wpa_supplicant all intermixed to get an idea of the order in which things are happening between the separate components. Looking at the kernel log you just attached I don't see the same "direct probe" time-outs as in the original logs you attached. The successful connection in those logs also seems to skip the "direct probe" stage. I'm currently looking at the code to try and determine why that stage would or wouldn't be run. I don't see it happening with my machines (none of which use the ath5k driver).

I asked about connecting to other APs out of curiosity, as your problems could be related to how your AP is set up (e.g. hidden SSID, mac address filtering, etc). You don't have to go out of your way to find another AP to connect to, but if you have an opportunity to try it I'd be interested to hear the result.

We could try bisecting between 2.6.37 and 2.6.38-rc1, but it could be difficult as that period represents the "merge window" when the bulk of the new development work for the 2.6.38 kernel release was being merged. That means that there is a lot of potential for instability. So I'd like to do a little more digging before trying that.

Does removing hwcrypt=1 make any difference on the 2.6.37 kernels?

@Tommy, I think it would be helpful to collect verbose logs from wpa_supplicant. To do this, modify /usr/share/dbus-1/system-services/fi.epitest.hostap.WPASupplicant.service. You should have a line like:

  Exec=/sbin/wpa_supplicant -u -s

Add -dd to the arguments, e.g.

  Exec=/sbin/wpa_supplicant -u -s -dd

Reboot and attach /var/log/syslog for a problematic connection attempt. It might also be helpful to attach syslog for a good connection with the 2.6.37.6 kernel as well if you get a chance. You can remove the -dd after collecting the logs.

Thanks!

regular (non-verbose) syslog with /etc/modprobe.d/atheros-workaround.conf renamed to /etc/modprobe.d/atheros-workaround.conf.disabled -- I saw no difference from the previous behavior on this 2.6.37.6 kernel but maybe there's a difference in the syslog

also just for comparison I let it connect to the regular access point then connected it to a non-encrypted access point called nowhere_man that has no WAN connection then connected it back to the regular access point.

Verbosity up. This time when I booted it picked the nowhere_man AP (probably because it's right next to me) so I just switched once to the regular access point. Because this is verbose you also get some info about some of the other APs in my neighborhood.

verbosity up. kernel 2.6.38-8 kernel. nohwcrypt=1

I booted under this kernel (sorry I don't remember ... is this the stock kernel?) and the system immediately connected to the nowhere_man unsecured AP with no WAN. I switched to the regular AP and let it fail, ask for credentials, typed credentials, I think it failed again and I canceled and it reconnected to the nowhere_man unsecured AP again.

Let me know if you need more focused logs showing the WPA AP only or the unsecured AP only (but this time I would connect it to the WAN) or ...

I have a better idea of what's going on, but still not a good picture of what might have changed to cause your regression.

Let's go ahead and try a bisect. I've restricted it to a couple of subdirectories of the kernel source tree, so that should limit the number of builds that are required. Basically, I'll provide you with a build, then you install, test, and let me know whether that build was good or bad. Then I provide another build, etc., until we've identified the change that introduced your issue.

You can get the first build (version 2.6.37-020637.0001) at:

  http://people.canonical.com/~sforshee/lp767192/

That build (2.6.37-020637.0001) doesn't appear to boot completely. It hangs with a garbled display similar to what I saw with the Maverick kernel 2.6.35.12-02063512.201104011012 I mentioned in comment #20.

I don't see any logs written for the two boot attempts on that kernel. Does that make sense? The disk activity stops just after the abbreviated boot splash that this model seems to do. (Grub selection followed by black screen with disk activity, followed by a second or two of Ubuntu logo with the red 'boot complete' dots, then in the case of this kernel the screen hangs with blocky solid colored irregular stripes.)

I have tried a couple of boot attempts, though some of the previous kernels I tried put similar garbage on the screen and switching virtual consoles got X to redraw. HOWEVER Since the disk activity stops and it doesn't respond to Ctrl-Alt-F1 etc. I presume it is really hung.

That's why I was hesitant to try bisection. Stability is suspect in the merge window :(

Let's try one more. I've uploaded version 2.6.37-020637.0002 to the same location.

If you can't boot this one, you might also try editing /etc/default/grub and commenting out the GRUB_CMDLINE_LINUX_DEFAULT line, then running 'sudo update-grub2' and booting up the bisect kernel. That _might_ give some output when you boot that could tell us why it hangs.

It hung, with a slightly more beautiful screen... ;-)

I have tried several reboots watching for things. Of course several screenfulls scroll off, and when it hangs there are some messages -- the LAST message says
stopping save udev log and update rules [ OK ]

but way up under the fsck message at the top of the screen, it says

fb: conflictign fb hw usage radeondrmfb vs VESA VGA -- removing ge

also a couple of lines down

Cache read/write disabled: /ssy/kernel/security/apparmor/features interface file missing. (Kernel needs AppArmor 2.4 compatibility patch.)

sorry typo in "fb: conflicting fb hw usage radeondrmfb vs VESA VGA -- removing ge" but "ge" IS where the text ended.

I don't know whether or not the "conflicting fb" message is causing your boot problems, but we can try to work around it. Add this line:

GRUB_GFXPAYLOAD_LINUX=text

to /etc/default/grub, re-run 'sudo update-grub2', and see if you can boot the bisect kernels after that. This should prevent the vesafb driver from trying to use the framebuffer.

Wonderful! The 020637.002 kernel booted fine with the GRUB_GFXPAYLOAD_LINUX=text command,

HOWEVER the network exhibits its characteristic problem with "wlan0: direct probe" messages etc.

so I will try the .001 kernel from earlier and report back.

The 020637.001 kernel booted fine with the GRUB command.

HOWEVER the network is even less functional -- there don't seem to be any ath5k messages in the syslog at all on this boot.

Could you attach syslog from the 020637.0001 kernel? Thanks!

sorry for forgetting the syslog... I should know better by now!

in case you need the 2.6.37.0002 kernel's syslog

You don't need to attach syslog for every bisection run, only if something weird happens like having absolutely no wireless.

I don't know what happened in 0001, but we can ignore it for now. It will only become an issue if the problem is with a nearby commit. And the apparmor messages are nothing to worry about.

The next kernel is building, should be ready before too much longer.

Build 020637.0003 is now available.

The 020637.0003 booted but the wireless wasn't available. It didn't even try to connect. syslog attached

Nothing about the device in the logs. What happens if you try loading the module manually, 'sudo modprobe ath5k'?

yes after the modprobe, the wireless came up, tried to connect and failed. The connection icon pulsed a little differently from usual as it was trying to connect... attaching the log.

The log still shows the direct probe issue. Build 020637.0004 is uploaded for you to test.

Build 020637.0004 booted fine and the wireless came up AND connected immediately.

Do you need anything from the logs? I have not tried booting with GFXPAYLOAD and nohwcrypt parameters back to default.

I just booted 020637.0004 with my /etc/modprobe.d/atheros-workaround.conf renamed to atheros-workaround.conf.disabled and the wireless came up AND connected immediately.

I presume that's all I need to do to keep the modprobe parameter from loading; I haven't been able to confirm that.

I know it's not germane to this but the boot is also successful with the GRUB_GFXPAYLOAD_LINUX=text line commented out in /etc/default/grub (I did run update-grub)

never mind; subsequent boots didn't work so I had to reenable GRUB_GFXPAYLOAD_LINUX=text to get 020637.0004 to boot.

You can just leave the GRUB_GFXPAYLOAD_LINUX parameter in place while we do the test, because it will likely be a problem with all the kernels.

Probably the only thing I'd be interested to know from your syslog is whether or not you have any of the 'direct probe' messages. It sounds like this kernel is good though, so I went ahead and started the next build.

no direct probe messages in syslog!

Build 020637.0005 is uploaded.

020637.0005 booted fine, network came up quickly, AND no direct probe messages in syslog...

Build 020637.0006 is uploaded. We're getting close!

020637.006 booted fine, network came up quickly, AND no direct probe messages in syslog.

(I am experiencing some problem with the touchpad & controlling Unity but it could be humidity; we have flood warnings all around us.)

Build 020637.0007 is uploaded.

020637.0007 boots but the wireless fails to connect to WPA with direct probe messages in syslog. Let me know if I need to capture any specific info.

Build 020637.0008 is uploaded.

020637.0008 booted well and the network came up immediately. There were no direct probe messages in syslog, though it had one additional state change in the syslog before connecting so I am attaching it. (It probably isn't significant, as the link is working fine.)

the filename is wrong, but the contents ARE of the .0008 syslog.

Okay, the bisection has identified this as the offending commit:

  commit 08ca944eb240b2299e743c76b43fbc7c2dd251de
  Author: Helmut Schaa <email address hidden>
  Date: Tue Nov 30 12:19:34 2010 +0100

      mac80211: Minor optimization in ieee80211_rx_h_data

However I'm skeptical that this is the correct commit. I think it limited the bisection more than I had intended and thus identified the wrong commit.

So we'll probably need to do another bisect. The good news is that I have a range of 20-30 commits (almost all ath5k stuff), so it should go pretty quickly. I'll post a new build shortly.

Build 020637.0009 is uploaded.

020637.0009 booted OK and the network came up really fast. No direct probe messages, and FEWER state changes than before in syslog. I won't attach it though unless you need it.

Build 020637.0010 is uploaded.

booted 020637.0010 but the network failed with direct probe messages in syslog

Build 020637.0011 is uploaded.

booted 020637.0011 and network came up quickly with no direct probe messages.

(It had the additional state change again but I presume that isn't relevant.)

Build 020637.0012 is uploaded.

booted 020637.0012 and network came up quickly with no direct probe messages in syslog

And build 020637.0013 is uploaded. This one should tell us the (real) commit that introduces your problem.

Booted 020637.0013 and the network came up quickly with no direct probe messages in syslog.

The bisect identified this as the first bad commit.

  commit 8aec7af99b1e4594c4bb9e1c48005e6111f97e8e
  Author: Nick Kossifidis <email address hidden>
  Date: Tue Nov 23 21:39:28 2010 +0200

      ath5k: Support synth-only channel change for AR2413/AR5413

I've looked it over, but without being familiar with the hardware it's hard for me to tell how these changes lead to your problem. So I'm afraid I might have to subject you to some more test kernels ;) These kernels will be based on the Ubuntu kernel so you should be able to remove any hacks you had to put in place for testing the kernels while we were bisecting. The first build is available in the same directory, version 2.6.38-9.43~lp767192v201105031700.

Thanks for all your help testing these kernels!

booted 2.6.38-9.43 and the network came up quickly, no direct probe in syslog; I had already removed the ath5k nohwcrypt=1 hack, and for this boot I removed the GRUB_GFXPAYLOAD_LINUX=text line in /etc/default/grub so it was not in effect either.

SO for this particular issue, this one looks good... I will boot another time or two to be sure.

I could attach the syslog but I don't think it shows anything interesting.

Here's what lshw says about the wireless interface: it IS an AR2413 as mentioned in the commit:

  *-network:1
       description: Wireless interface
       product: AR2413 802.11bg NIC
       vendor: Atheros Communications Inc.
       physical id: 2
       bus info: pci@0000:06:02.0
       logical name: wlan0
       version: 01
       serial: 00:19:7d:1c:82:d3
       width: 32 bits
       clock: 33MHz
       capabilities: pm bus_master cap_list ethernet physical wireless
       configuration: broadcast=yes driver=ath5k driverversion=2.6.38-9-generic firmware=N/A ip=192.168.1.32 latency=168 link=yes maxlatency=28 mingnt=10 multicast=yes wireless=IEEE 802.11bg
       resources: irq:22 memory:f0310000-f031ffff

btw I Googled the guid on that commit and found someone else bisecting around it back in March to find a problem... http://<email address hidden>/msg18985.html

Thanks for the pointers. I'm pretty close to sending our findings on to the maintainer of the driver, but there's one more thing we should test first. We need to verify that it was the change I made in that last build that fixed your problem and not some other update in the kernel.

Please test the kernel that's currently in the proposed archive, which will be identical to the one I just gave you aside from the changes I made. You can either enable the proposed archive using these instructions (note that this will give you all proposed natty updates, not just the kernel).

  https://wiki.ubuntu.com/Testing/EnableProposed

Or you can download the debs directly using these links.

  https://launchpad.net/ubuntu/+archive/primary/+files/linux-headers-2.6.38-9-generic_2.6.38-9.43_i386.deb
  https://launchpad.net/ubuntu/+archive/primary/+files/linux-headers-2.6.38-9_2.6.38-9.43_all.deb
  https://launchpad.net/ubuntu/+archive/primary/+files/linux-image-2.6.38-9-generic_2.6.38-9.43_i386.deb

booted 2.6.38-9.43 (proposed; from links abobe) and it booted great but it was one of the LESS functional kernels for the wireless.

I booted and logged in, and Network manager's icon was hollow. I gave it a few seconds then I did a

modprobe ath5k

I gave it a few seconds with no response

I turned off networking in network manager, and when I turned it back on, it still didn't show a connection. The log complains about an unmanaged device; attached.

I did NOT try opening the netbook or connecting to an unsecured AP

booted 2.6.38-9.43 (proposed; from links abobe) and it booted great but it was one of the LESS functional kernels for the wireless.

I booted and logged in, and Network manager's icon was hollow. I gave it a few seconds then I did a

modprobe ath5k

I gave it a few seconds with no response

I turned off networking in network manager, and when I turned it back on, it still didn't show a connection. The log complains about an unmanaged device; attached.

I did NOT try opening the netbook or connecting to an unsecured AP

I reinstalled and booted 2.6.38-9.43~lp767192v201105031700 with no changes to any other files and again wireless came up perfectly. The totally uninteresting syslog is attached.

The results with 2.6.38-9.43 are odd. I browsed the changelog and nothing stood out as potentially causing this change on the surface. Did you boot more than once to check that it wasn't a fluke?

There was a new stable kernel this week that has some ahteros fixes. I don't think any will fix your problem, but it might be worth trying if you get a chance. It's available at

  http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.38.5-natty/

I'll start a dialog with the maintainers to see if they can provide an appropriate fix.

I couldn't remember; maybe I did only boot it once before, and maybe I was not being patient enough. Sorry.

Today I reinstalled 2.6.38-9.43 proposed. I booted twice, and on the second boot I waited a few minutes, and the network ultimately came up without any modprobes or network-manager toggles. The syslog shows the expected direct probe messages. Log attached.

The network seems to be fully functional, just slow to come up. I haven't tested its throughput or anything like that.

I will boot one more time on this one and will also try the other kernel you mentioned.

the next boot of 2.6.38-9.43proposed took ten minutes before it started direct-probing (even after turning the netbook on after about seven minutes) and the network connection failed.

I installed mainline 2.6.38-02063805-generic #201105030911 and after about a two minute wait it started direct probing and the network came up. (The netbook happened to already be on.)

These failures are more consistent with the expected flakiness.

I am capturing the logs if you want them.

on the next boot of mainline 2.6.38-02063805-generic #201105030911 I could not get the wireless network to come up at all. Numerous direct probe attempts and failures. I deactivated and reactivated the network, and I had had the netbook up, but it still failed. Again I kept the syslog if you need it.

It won't be necessary to supply the logs.

I've sent a message to the maintainers, so we'll see what kind of response we get from them.

I'm afraid I've been asked to have you test yet another kernel, to see if the bug is fixed in the latest 2.6.39 release candidate. That one isn't yet available in our normal mainline builds location, so I've built one and uploaded it to:

  http://people.canonical.com/~sforshee/lp767192/linux-2.6.39-rc6/

If things still aren't working with this kernel, please enable ath5k reset debugging. I think this command should do it:

  sudo sh -c "echo reset > /sys/kernel/debug/ieee80211/phy0/ath5k/debug"

But without having the hardware I'm not completely sure about the path, so if it fails you might poke around in /sys/kernel/debug/ieee80211 to find the file (note that /sys/kernel/debug can only be accessed as root). Please disable wireless, run that command, re-enable wireless, and then capture syslog and dmesg showing the failed connection attempts.

Thanks!

/sys/kernel/debug/ieee80211/phy0 directory contains:

root@AcerAspire5100:/sys/kernel/debug/ieee80211/phy0# ls -l
total 0
-r-------- 1 root root 0 2011-05-04 18:29 channel_type
-r--r--r-- 1 root root 0 2011-05-04 18:29 fragmentation_threshold
-r-------- 1 root root 0 2011-05-04 18:29 frequency
-r--r--r-- 1 root root 0 2011-05-04 18:29 ht40allow_map
drwxr-xr-x 4 root root 0 2011-05-04 18:29 keys
-r--r--r-- 1 root root 0 2011-05-04 18:29 long_retry_limit
drwxr-xr-x 3 root root 0 2011-05-04 18:29 netdev:wlan0
-r-------- 1 root root 0 2011-05-04 18:29 noack
-r-------- 1 root root 0 2011-05-04 18:29 power
-r-------- 1 root root 0 2011-05-04 18:29 queues
drwxr-xr-x 2 root root 0 2011-05-04 18:29 rc
--w------- 1 root root 0 2011-05-04 18:29 reset
-r--r--r-- 1 root root 0 2011-05-04 18:29 rts_threshold
-r--r--r-- 1 root root 0 2011-05-04 18:29 short_retry_limit
drwxr-xr-x 2 root root 0 2011-05-04 18:29 statistics
-r-------- 1 root root 0 2011-05-04 18:29 total_ps_buffered
-r-------- 1 root root 0 2011-05-04 18:29 tsf
-r-------- 1 root root 0 2011-05-04 18:29 uapsd_max_sp_len
-r-------- 1 root root 0 2011-05-04 18:29 uapsd_queues
-r-------- 1 root root 0 2011-05-04 18:29 user_power
-r-------- 1 root root 0 2011-05-04 18:29 wep_iv
root@AcerAspire5100:/sys/kernel/debug/ieee80211/phy0# cd netdev\:wlan0/

root@AcerAspire5100:/sys/kernel/debug/ieee80211/phy0/netdev:wlan0# ls -l
total 0
-r-------- 1 root root 0 2011-05-04 18:29 aid
-r-------- 1 root root 0 2011-05-04 18:29 ave_beacon
-r-------- 1 root root 0 2011-05-04 18:29 bssid
-r-------- 1 root root 0 2011-05-04 18:29 channel_type
-r-------- 1 root root 0 2011-05-04 18:29 drop_unencrypted
-r-------- 1 root root 0 2011-05-04 18:29 flags
-r-------- 1 root root 0 2011-05-04 18:29 last_beacon
-r-------- 1 root root 0 2011-05-04 18:29 rc_rateidx_mask_2ghz
-r-------- 1 root root 0 2011-05-04 18:29 rc_rateidx_mask_5ghz
-rw------- 1 root root 0 2011-05-04 18:29 smps
-r-------- 1 root root 0 2011-05-04 18:29 state
drwxr-xr-x 3 root root 0 2011-05-04 18:29 stations
--w------- 1 root root 0 2011-05-04 18:29 tkip_mic_test
root@AcerAspire5100:/sys/kernel/debug/ieee80211/phy0/netdev:wlan0#

could it be that I should do it the other way around? In other words I could try this:

sudo sh -c "echo debug > /sys/kernel/debug/ieee80211/phy0/reset"

maybe...

This was the first boot so I will try booting a couple of times and if it still looks the same I will give that one a try .

sudo sh -c "echo debug > /sys/kernel/debug/ieee80211/phy0/reset"

didn't seem to do anything.

I don't see a debug file anywhere. This (2007) email implies I should see it, as long as the define was set when the driver was compiled...

http://<email address hidden>/msg00266.html

Yeah, I missed that there was a config option I needed to enable. I put a new build with that option enabled in the same directory, version 2.6.39-020639rc6.201105050227. Sorry for the mistake.

been using 2.6.39-020639rc6.201105050227

it does have /sys/kernel/debug/ieee80211/phy0/ath5k/debug

I have been tinkering with this for awhile now; you asked for dmesg and syslog, but I could never see any additional messages EITHER place no matter what I did to the debug file. It turns out all the additional messages are going to kern.log. So I will grab some or all of the kern.log along with the syslog so you can look at the time stamps or whatever.

ALSO for whatever reason the wireless has never completely failed this evening -- it waits a minute or two, does its direct probe a couple of times and then comes up. SO I haven't yet captured the total failure mode.

I was wrong about the messages; they were showing up in kern.log AND syslog as expected but I think I have too many logs for the system log viewer application to handle smoothly.

I am attaching a syslog for 2.6.39-020639rc6.201105050227, I stopped networking, wrote reset to /sys/kernel/debug/ieee80211/phy0/ath5k/debug started networking again... I believe I did some of those more than once.

The network failed to start scanning until after I opened the netbook nearby whereupon the laptop connected to the AP.

 I have a syslog where the network fails to even scan (not included here). I will try a couple more times to get a syslog where it scans but fails to connect.

dmesg for 2.6.39-020639rc6.201105050227 as requested earlier

I haven't been able to get this kernel to start a scan and then fail to connect to the access point. It will sit there a LONG time before it tries, though. With the attached syslog the I let it sit quite a long time before I finally opened the netbook nearby and it connected then.

this one isn't really any different than the previous one, but maybe it more clearly shows the difficulty because of the delay

I guess I never attached it. As I said it isn't very interesting except that I let it sit a half an hour before nudging it to connect.network comes up eventually

I was finally able to observe the "scan but never connect" failure mode.

I booted, set debug as quickly as I could (though the network was already scanning) and waited for it to fail. Then I turned networking off and then on and it scanned and connected.

I've been asked to have you attempt scanning and connecting to an open AP using command-line tools. You'll need to do the following somewhere where an open AP is available. Please capture the output of all commands and attach it here.

1. Disable networking in the notification area
2. Run 'sudo ifconfig wlan0 up'
3. Run 'sudo iw dev wlan0 scan'
4. Run 'sudo iw dev wlan0 connect -w <SSID>', using the SSID of the open AP
5. Run 'sudo iw dev wlan0 disconnect' to disconnect if #4 was successful.

What we want to see is data from when either the scan or connectoin fails, but if you can't ever get a failure that's useful information too. Please capture dmesg following your attempt(s) to connect with iw and attach it here as well.

Thanks!

I created a guest network on the same AP without any security and I can start and stop it as needed. I have plenty of different APs so I can plug in the one I used the other day and connect it to the WAN if needed.

I had to apt-get install iw for these commands ... but I'm not seeing much feedback in the terminal. I guess it doesn't have much to say on its own.

I think I have the routine now and I will capture a few sessions later this afternoon. I'm also discovering that the dmesg I have been looking at in the system log viewer does NOT change when there is more to see in dmesg so I will write the cli dmesg to a file from now on. (Any dmesg info I posted earlier was captured from the system log viewer.)

booted 2.6.39-020639rc6.201105050227

captured a few complete failures and copied dmesg, syslog & terminal commands into .tar.gz

booted 2.6.39-020639rc6.201105050227 again

captured more failures but was a lot more persistent until I was able to get the network to come up

and copied dmesg, syslog & terminal commands into .tar.gz

I wanted to drag out the other unit and set it up again to try a connection to a different unsecured AP (just in case I am seeing a quirk with this one) but I cannot get to it due to family crises. Let me know if I should look for anything in particular when I do.

plugged in wrt54gl AP running a recent version of Tomato (my usual AP is a Netgear WNR3500L running stock firmware).

booted 2.6.39-020639rc6.201105050227

went through the procedure:

1. Disable networking in the notification area
2. Run 'sudo ifconfig wlan0 up'
3. Run 'sudo iw dev wlan0 scan'
4. Run 'sudo iw dev wlan0 connect -w <SSID>', using the SSID of the open AP
5. Run 'sudo iw dev wlan0 disconnect' to disconnect if #4 was successful.

(I repeated step 3 until it showed something.)

In this run step 4 worked (the laptop connected to the open AP) on the first attempt.

plugged in wrt54gl AP running a recent version of Tomato (my usual AP is a Netgear WNR3500L running stock firmware).

booted 2.6.39-020639rc6.201105050227

went through the procedure:

1. Disable networking in the notification area
2. Run 'sudo ifconfig wlan0 up'
3. Run 'sudo iw dev wlan0 scan'
4. Run 'sudo iw dev wlan0 connect -w <SSID>', using the SSID of the open AP
5. Run 'sudo iw dev wlan0 disconnect' to disconnect if #4 was successful.

(I repeated step 3 until it showed something.)

In this run step 4 worked (the laptop connected to the open AP) on the first attempt.

I'm so glad to see progress being made on this issue. This is all new to me but if you need someone else to help test, i'm on an acer aspire one zg5 with the 5001.

sorry for the delay, but I captured one the other day where it never did successfully connect to the open AP.

same booted 2.6.39-020639rc6.201105050227 kernel

same kernel, booted today and let it sit several hours and everything worked perfectly.

It seems like maybe you're only able to connect after a successful scan. Can you run the steps from comment #104 after running 'sudo sh -c "echo reset > /sys/kernel/debug/ieee80211/phy0/ath5k/debug"' with the kernel from comment #93? Ideally I'd like to see some unsuccessful scan/connnect attempts followed by a successful one. Please do not enable wireless in network-manager while running the test. Thanks!

I don't know if this info helps. I have the same problem with my aspire 5110. When I'm close to the router there is no problem to find or connect to my network. But when I try and connect from the next room the problem starts. Sometimes I see the network (with full signal strength) but cant connect but most of the time no networks comes upp all. Running Lubuntu 11.04 2.6.38-8-generic

per comment 114 I am trying some more, same kernel. I was completely unable to connect.

in this attempt apparently it was time to purge my logs so syslog is truncated.

per comment 114 this time I got good logs

ALSO this time I tried for a few minutes then I got tired of no connection so after awhile I disconnected the nowhere_man AP from the router and plugged it in just a few feet away from the laptop. It connected easily.

IMPORTANT -- the only difference with the good connect here was moving the access point within a few feet of the laptop.

per comment 114 I kept the access point nearby and ran the procedure again just to show what it's like

apparently it's like shooting 80211b in a barrel

The upstream maintainer made a patch to disable by default the changes that are causing the problem. I've uploaded a build with the patch. Please verify that it fixes the issue. The more people who can test this build the better.

The new build is available at:

  http://people.canonical.com/~sforshee/lp767192/linux-2.6.39-rc6.201105171845/

Please report back here with your results. Thanks!

installed 2.6.39-020639rc6-generic #201105171845

booted perfectly; wireless came up perfectly.

so far I have booted twice, same result.

When i try to install the new build the software center shows it as being available for upgrade but when i click on it nothing happens. Good to see more progress for those who actually know what they're doing though.

upon further inspection, if i goto installed software it does show as being installed but still wireless doesn't work on this acer aspire one zg5 with the 5001 hardware.

@ethos_dacapo: Be sure you reboot after installing the kernel .deb files. After rebooting you can check what kernel you're using by opening a terminal and running 'cat /proc/version_signature'. If you're running the right kernel it should say something along the lines of 'Ubuntu 2.6.39-020639rc6.201105171845-generic 2.6.39-rc6'.

Hi,
I have Acer Aspire 5100 Model BL51 with
Ethernet controller: Atheros Communications Inc. AR2413 802.11bg NIC (rev 01)

After the new kernel everything looks working fine.
Wireless interface comes up at the boot and connects without waiting, without any problem.

fhm@H2O:~$ uname -a
Linux H2O 2.6.39-020639rc6-generic #201105171845 SMP Tue May 17 18:46:57 UTC 2011 i686 athlon i386 GNU/Linux

[ 37.389468] wlan0: authenticate with 00:25:9c:54:44:25 (try 1)
[ 37.390973] wlan0: authenticated
[ 37.391000] wlan0: associate with 00:25:9c:54:44:25 (try 1)
[ 37.395682] wlan0: RX AssocResp from 00:25:9c:54:44:25 (capab=0x431 status=0 aid=2)
[ 37.395688] wlan0: associated

Best regards.

I did reboot my computer originally with no success. Here is the output of the command you posted. Btw it should be 'cat /proc/version' now.

cat /proc/version
Linux version 2.6.39-020639rc6-generic (root@tangerine) (gcc version 4.6.1 20110428 (prerelease) (Ubuntu 4.6.0-6ubuntu1) ) #201105050227 SMP Thu May 5 01:28:51 UTC 2011

Hope this helps. Thanks

@ethos_dacapo: be sure you downloaded and installed all THREE files listed here that end in .deb:

http://people.canonical.com/~sforshee/lp767192/linux-2.6.39-rc6.201105171845/

If you installed them by double-clicking them it may take additional attempts because one will refuse to install until the others have been installed.

The fastest way to install all three at once is to move them into a directory without any other .deb files and use the Terminal command (in that directory)

sudo dpkg -i *.deb

@ethos_dacapo: /proc/version_signature was what I meant, but I forgot it wouldn't be present in this build. That file shows that you have the right kernel installed though. I did a quick search for your netbook and I suspect that it has different wireless hardware than what's being addressed here. Can you attach the output of 'lspci -vnvn' so I can check?

@Tommy, @Fehmi, thanks for testing the proposed fix.

@Seth Forshee

lspci -vnvn
00:00.0 Host bridge [0600]: Intel Corporation Mobile 945GME Express Memory Controller Hub [8086:27ac] (rev 03)
 Subsystem: Acer Incorporated [ALI] Device [1025:015b]
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort+ >SERR- <PERR- INTx-
 Latency: 0
 Capabilities: <access denied>
 Kernel driver in use: agpgart-intel

00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GME Express Integrated Graphics Controller [8086:27ae] (rev 03) (prog-if 00 [VGA controller])
 Subsystem: Acer Incorporated [ALI] Device [1025:015b]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 0
 Interrupt: pin A routed to IRQ 16
 Region 0: Memory at 58380000 (32-bit, non-prefetchable) [size=512K]
 Region 1: I/O ports at 60c0 [size=8]
 Region 2: Memory at 40000000 (32-bit, prefetchable) [size=256M]
 Region 3: Memory at 58400000 (32-bit, non-prefetchable) [size=256K]
 Expansion ROM at <unassigned> [disabled]
 Capabilities: <access denied>
 Kernel driver in use: i915
 Kernel modules: intelfb, i915

00:02.1 Display controller [0380]: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller [8086:27a6] (rev 03)
 Subsystem: Acer Incorporated [ALI] Device [1025:015b]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 0
 Region 0: Memory at 58300000 (32-bit, non-prefetchable) [size=512K]
 Capabilities: <access denied>

00:1b.0 Audio device [0403]: Intel Corporation N10/ICH 7 Family High Definition Audio Controller [8086:27d8] (rev 02)
 Subsystem: Acer Incorporated [ALI] Device [1025:015b]
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 0
 Interrupt: pin A routed to IRQ 45
 Region 0: Memory at 58440000 (64-bit, non-prefetchable) [size=16K]
 Capabilities: <access denied>
 Kernel driver in use: HDA Intel
 Kernel modules: snd-hda-intel

00:1c.0 PCI bridge [0604]: Intel Corporation N10/ICH 7 Family PCI Express Port 1 [8086:27d0] (rev 02) (prog-if 00 [Normal decode])
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 0
 Bus: primary=00, secondary=01, subordinate=01, sec-latency=0
 I/O behind bridge: 00005000-00005fff
 Memory behind bridge: 57300000-582fffff
 Prefetchable memory behind bridge: 0000000050000000-0000000050ffffff
 Secondary status: 66MHz- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- <SERR- <PERR-
 BridgeCtl: Parity- SERR- NoISA- VGA- MAbort- >Reset- FastB2B-
  PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERR...

@ethos_dacapo, your wireless is an ath5k but not identical. If the fix works for others but not for you (as appears to be the case so far) you likely have a separate problem. Your best bet is to open a new bug by running 'ubuntu-bug linux' in a terminal (be sure you've booted back to your original kernel before reporting the bug, you can get a menu to select which kernel to boot with by holding down the left shift key when you boot your machine).

Ok thanks Seth. I submitted the bug https://bugs.launchpad.net/ubuntu/+source/linux/+bug/784768

installed 2.6.39-020639rc6-generic #201105171845
No problem connection to my wireless network.

I would like to test the kernel provided in comment #119 but I'm on amd64 architecture (I'm on Acer TravelMate 5510). Is it possible to have the kernel compiled in such version?
Thanks.

@Sandro: I posted some amd64 debs. Thanks for testing!

Hi Seth, thanks for your .deb! I installed and booted them. The first time the boot blocked in the black screen after grub and before the ubuntu logo with the five dots. The second and third time the boot was successful but it took almost three minutes to complete! Most of time was spent in the screen with ubuntu logo and five dots.
I couldn't test the wireless because at this moment I'm not at home: I only see the problem when I'm at home and try to connect from a room different from the one the router is in (-->it only connects if I'm VERY close to the router ).
I'll try in the evening.

Booted 3 times at home: first time booted quickly, second time hang with black screen, third time booted quickly. Wireless was always OK! So the patch works!!!
If someone will try to reintroduce and patch the changes that caused the problem I can help in testing those patches.

@ Seth: Per comments #129-130 i reported the bug affecting my version of the ath5k. Its going on 16 days since i filed it but nobody has responded other than the initial confirmation. Is this normal or did i miss something? At this point i forget what its like not to be tethered to an ethernet cable on this netbook. lol

Installed the three .deb files on a Acer aspire 5101, and it fixed network issues on this card:
06:02.0 Ethernet controller: Atheros Communications Inc. AR2413 802.11bg NIC (rev 01)

Will this be merged into kernels in main repositories?

@toffyrn: Of course the plan is to integrate this patch into the natty kernel. In this case it is an upstream patch, but the upstream developer has been slow about getting the patch into the upstream kernel, and that's what has been holding up this fix. This is finally progressing. It's too late to get this into the next natty kernel update, but I expect it will be in the next update after that.

A fix for this has been released in oneiric. I'm working on the SRU for natty.

SRU Justification

Impact: Fast channel switching causes problems with scanning and associating for many ath5k users.

Fix: Backport of upstream commit which disables fast channel switching by default and adds a module parameter to enable it.

Test case: Tested by affected users on LP #767192.

The fix for this issue is now available in the proposed kernel. Please verify that the issue has been fixed. Instructions for enabling proposed updates are located at https://wiki.ubuntu.com/Testing/EnableProposed. Thanks!

Please test the kernel as requested in comment #141 and update this bug with the results. If the problem is solved, change the tag 'verification-needed-natty' to 'verification-done-natty'.

If verification is not done by one week from today, this fix will be dropped from the source code, and this bug will be closed.

Installed all the linux-image-2.6.38-11 packages and dependencies from natty-proposed and booted from it. The wireless network came up immediately on this Acer Aspire 5100-3949. I will boot a few more times to test but on the first try it looks good.

P.S.:
Herton R. Krzesinski wrote :
> If verification is not done by one week from today, this fix will be dropped from the source code, and this bug will be closed.

Harsh! I made sure to test this before leaving for vacation, though I don't understand the reason for the dire warning. Surely a multiply-confirmed bug wouldn't be dropped completely?!?

I can confirm that wifi works with kernel 2.6.38-11.
Acer 5110. Ubuntu 11.04 amd64.

Tested on Acer TravelMate 5510 (amd64 architecture): worked perfectly!

This bug was fixed in the package linux - 2.6.38-11.48

---------------
linux (2.6.38-11.48) natty-proposed; urgency=low

  [Herton R. Krzesinski]

  * Release Tracking Bug
    - LP: #818175

  [ Upstream Kernel Changes ]

  * Revert "HID: magicmouse: ignore 'ivalid report id' while switching
    modes"
    - LP: #814250

linux (2.6.38-11.47) natty-proposed; urgency=low

  [Steve Conklin]

  * Release Tracking Bug
    - LP: #811180

  [ Keng-Yu Lin ]

  * SAUCE: Revert: "dell-laptop: Toggle the unsupported hardware
    killswitch"
    - LP: #775281

  [ Ming Lei ]

  * SAUCE: fix yama_ptracer_del lockdep warning
    - LP: #791019

  [ Stefan Bader ]

  * SAUCE: Re-enable RODATA for i386 virtual
    - LP: #809838

  [ Tim Gardner ]

  * [Config] Add grub-efi as a recommended bootloader for server and
    generic
    - LP: #800910
  * SAUCE: rtl8192se: Force a build for a 2.6/3.0 kernel
    - LP: #805494

  [ Upstream Kernel Changes ]

  * Revert "bridge: Forward reserved group addresses if !STP"
    - LP: #793702
  * Fix up ABI directory
  * bonding: Incorrect TX queue offset, CVE-2011-1581
    - LP: #792312
    - CVE-2011-1581
  * fs/partitions/efi.c: corrupted GUID partition tables can cause kernel
    oops
    - LP: #795418
    - CVE-2011-1577
  * usbnet/cdc_ncm: add missing .reset_resume hook
    - LP: #793892
  * ath5k: Disable fast channel switching by default
    - LP: #767192
  * mm: vmscan: correctly check if reclaimer should schedule during
    shrink_slab
    - LP: #755066
  * mm: vmscan: correct use of pgdat_balanced in sleeping_prematurely
    - LP: #755066
  * ALSA: hda - Use LPIB for ATI/AMD chipsets as default
    - LP: #741825
  * ALSA: hda - Enable snoop bit for AMD controllers
    - LP: #741825
  * ALSA: hda - Enable sync_write workaround for AMD generically
    - LP: #741825
  * cpuidle: menu: fixed wrapping timers at 4.294 seconds
    - LP: #774947
  * drm/i915: Fix gen6 (SNB) missed BLT ring interrupts.
    - LP: #761065
  * USB: ehci: remove structure packing from ehci_def
    - LP: #791552
  * drm/i915: disable PCH ports if needed when disabling a CRTC
    - LP: #791752
  * kmemleak: Do not return a pointer to an object that kmemleak did not
    get
    - LP: #793702
  * kmemleak: Initialise kmemleak after debug_objects_mem_init()
    - LP: #793702
  * Fix _OSC UUID in pcc-cpufreq
    - LP: #793702
  * CPU hotplug, re-create sysfs directory and symlinks
    - LP: #793702
  * Fix memory leak in cpufreq_stat
    - LP: #793702
  * net: recvmmsg: Strip MSG_WAITFORONE when calling recvmsg
    - LP: #793702
  * ftrace: Only update the function code on write to filter files
    - LP: #793702
  * qla2xxx: Fix hang during driver unload when vport is active.
    - LP: #793702
  * qla2xxx: Fix virtual port failing to login after chip reset.
    - LP: #793702
  * qla2xxx: Fix vport delete hang when logins are outstanding.
    - LP: #793702
  * powerpc/kdump64: Don't reference freed memory as pacas
    - LP: #793702
  * powerpc/kexec: Fix memory corruption from unallocated slaves
    - LP: #793702
  * x86, cpufeature: Fix cpuid leaf 7 feature detection
    - LP: #793702
  * ath9k_hw: do noise floor calibration only on required chain...