rfkill switch on Intel 4965AGN wireless does not work in intrepid

[Everett Guerny]

I'm using a Fujitsu T2010 laptop currently running Intrepid Alpha 5.

condition 1: If I boot into the LiveCD with the hardware wireless switch enabled (wireless on) from the beginning, WiFi works as expected.

condition 2: If I boot into the LiveCD with the hardware wireless switch disabled (wireless off) and then switch it on, network-manager does not detect any nearby wireless networks, and manually entering a known working one does not result in a connection being established. I've also waited a number of minutes after enabling wireless, as I know that it sometimes takes a few minutes for wireless networks to appear in the network-manager-applet.

From my dmesg (condition 2):

[ 97.650227] iwlagn: Intel(R) Wireless WiFi Link AGN driver for Linux, 1.3.27ks
[ 97.650233] iwlagn: Copyright(c) 2003-2008 Intel Corporation
[ 97.651622] iwlagn: Detected Intel Wireless WiFi Link 4965AGN REV=0x4

So it looks like the wireless card is detected and whatnot, so I'm guessing it might be a network-manager problem.

I'm not familiar with managing wireless connections from the command line to troubleshoot this outside of the scope of network-manager, but if necessary, please let me know if there's any further information I can provide to help better diagnose this problem.

By the way, this hardware works without difficulty in Hardy.

[Alexander Sack (asac)]

please reproduce and attach your complete syslog taken _after_ reproducing. Thanks!

[Everett Guerny (everett)]

Thank you for the response, Alexander, but I have a couple questions...

1) Is syslog something different than my dmesg output? I see a few syslog packages, but none are installed by default, and I'm using a LiveCD.

2) The manner in which you're asking me to reproduce this... is this what I described as "condition 2" in the original bug report? (I'm having some difficulty parsing the "_after_" in your reply.)

Thanks again, and sorry for my lack of knowledge...

[matejcik (matejcik)]

i see this on MSI PR200, and it's unrelated to the network-manager, actually the -rfkill switch- is broken.
(worked fine in 2.6.24, but i can't confirm that it works in hardy with 2.6.27, because the debs from kernel.ubuntu.com don't have the appropriate firmware with them and so the wifi doesn't work at all)

goes like this:

1. take a machine, killall NetworkManager for good measure, rmmod iwlagn and switch its killswitch to off (LED responds just fine)

2.
# modprobe iwlagn
Sep 15 14:28:07 limetka kernel: [ 2842.166586] iwlagn: Intel(R) Wireless WiFi Link AGN driver for Linux, 1.3.27ks
Sep 15 14:28:07 limetka kernel: [ 2842.166595] iwlagn: Copyright(c) 2003-2008 Intel Corporation
Sep 15 14:28:07 limetka kernel: [ 2842.166742] iwlagn 0000:02:00.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18
Sep 15 14:28:07 limetka kernel: [ 2842.166796] iwlagn: Detected Intel Wireless WiFi Link 4965AGN REV=0x4
Sep 15 14:28:07 limetka kernel: [ 2842.205510] iwlagn: Tunable channels: 11 802.11bg, 13 802.11a channels
# ifconfig wlan0 up
Sep 15 14:28:40 limetka kernel: [ 2874.924355] iwlagn 0000:02:00.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18
Sep 15 14:28:40 limetka kernel: [ 2874.924782] firmware: requesting iwlwifi-4965-2.ucode
Sep 15 14:28:40 limetka kernel: [ 2874.931693] iwlagn: Radio disabled by HW RF Kill switch
Sep 15 14:28:40 limetka kernel: [ 2874.950597] ADDRCONF(NETDEV_UP): wlan0: link is not ready
# cat /sys/class/rfkill/rfkill?/state
2

at this point, you can play with the killswitch for as long as you like, it won't matter to the card and it will stay offline. only output is along the lines of :

Sep 15 14:29:53 limetka kernel: [ 2947.914303] atkbd.c: Unknown key pressed (translated set 2, code 0xe2 on isa0060/serio0).
Sep 15 14:29:53 limetka kernel: [ 2947.914314] atkbd.c: Use 'setkeycodes e062 <keycode>' to make it known.
Sep 15 14:29:53 limetka kernel: [ 2947.915542] atkbd.c: Unknown key released (translated set 2, code 0xe2 on isa0060/serio0).
Sep 15 14:29:53 limetka kernel: [ 2947.915553] atkbd.c: Use 'setkeycodes e062 <keycode>' to make it known.
Sep 15 14:29:53 limetka kernel: [ 2947.916402] atkbd.c: Unknown key pressed (translated set 2, code 0xf6 on isa0060/serio0).
Sep 15 14:29:53 limetka kernel: [ 2947.916412] atkbd.c: Use 'setkeycodes e076 <keycode>' to make it known.
Sep 15 14:29:53 limetka kernel: [ 2947.917742] atkbd.c: Unknown key released (translated set 2, code 0xf6 on isa0060/serio0).
Sep 15 14:29:53 limetka kernel: [ 2947.917751] atkbd.c: Use 'setkeycodes e076 <keycode>' to make it known.

(yes, two different keys - the killswitch key on PR200 actually turns on bluetooth and wifi simultaneously, one of the scancodes is for wifi and the other for BT, i don't know which is which)

3. rmmod iwlagn, set killswitch to on
# modprobe iwlagn
Sep 15 14:32:28 limetka kernel: [ 3102.793984] iwlagn: Intel(R) Wireless WiFi Link AGN driver for Linux, 1.3.27ks
Sep 15 14:32:28 limetka kernel: [ 3102.793997] iwlagn: Copyright(c) 2003-2008 Intel Corporation
Sep 15 14:32:28 limetka kernel: [ 3102.794899] iwlagn 0000:02:00.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18
Sep 15 14:32:28 limetka kernel: [ 3102.794966] iwlagn: Detected Intel Wirele...

[matejcik (matejcik)]

(i've been trying to remove network-manager from the bug and reassign it to kernel, but sadly i don't know how to do that :e/ please somebody fix this bugreport)

[Leann Ogasawara (leannogasawara)]

@matejcik - care to test using the latest Intrepid Alpha LiveCD - http://www.ubuntu.com/testing

@Everett - can you do the following:

1) as you describe in condition 2, boot with wireless off
2) open a terminal and do 'tail -n0 -f /var/log/syslog > syslog.log'
3) enable wireless (ie flip your kill switch)
4) capture the dmesg output by doing 'dmesg > dmesg.log'
5) attach both dmesg.log and syslog.log to this report

Thanks.

[matejcik (matejcik)]

@leann - this -is- the latest intrepid alpha with full updates

[matejcik (matejcik)]

to clarify:
- works fine in hardy/2.6.24 from latest updates
- can't test in hardy/2.6.27r3, because wifi doesn't work in there
- fails in intrepid/2.6.27 (alpha5)

[Everett Guerny (everett)]

Thank you for the instructions, Leann. Here is the syslog...

[Everett Guerny (everett)]

...and here is my dmesg.

[junk123 (junk123-deactivatedaccount-deactivatedaccount)]

I also have the same problem, worked fine in Hardy, upgraded to Intrepid (updated tonight) and I have no wireless connection.

[Fernando Miguel (fernandomiguel)]

daily LiveCD

$ uname -a
Linux ubuntu 2.6.27-4-generic #1 SMP Mon Sep 22 04:40:15 UTC 2008 x86_64 GNU/Linux

[Fernando Miguel (fernandomiguel)]

Relevant lines:

Sep 29 13:24:12 ubuntu kernel: [ 37.177957] iwlagn 0000:04:00.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18
Sep 29 13:24:12 ubuntu kernel: [ 37.178042] iwlagn 0000:04:00.0: restoring config space at offset 0x1 (was 0x100102, writing 0x100106)
Sep 29 13:24:12 ubuntu kernel: [ 37.178212] firmware: requesting iwlwifi-4965-2.ucode
Sep 29 13:24:12 ubuntu kernel: [ 37.195939] iwlagn: Radio disabled by HW RF Kill switch

[Brian Murray (brian-murray)]

Possibly related to bug 193970 if the different drivers share code.

[Fernando Miguel (fernandomiguel)]

Although this may look like a dupe (and it is in terms of problem) #193970 seems to deal with hardy and backported kernels, while this one is focused on Intrepid and 2.6.27.

But do as i find best, according to your workflow.

[Everett Guerny (everett)]

This is still the case in the Intrepid Beta.

Interestingly enough, I noticed that booting the laptop with the wireless switch off, turning the switch on, suspending the system and then powering it back on will get wireless working, just as if the system was originally booted with the switch on.

[Everett Guerny (everett)]

As a temporary workaround (found in a reply to bug 193970), the following seems to get wireless working for me:

sudo modprobe -r iwl4965 && sudo modprobe iwl4965

[matejcik (matejcik)]

...yes, but then you need to do it again when you want to switch it off...

[Frits Jalvingh (fjalvingh)]

Same problem here, on an Acer Aspire 9920. This always boots with the kill switch ON (killed); the only way to get a connection is to first toggle the switch and then removing and adding the iwlagn module. This worked properly on Hardy.
What has changed is that with Hardy I got messages about "scan code xxx" in the dmesg log when I toggled the kill switch. In Intrepid nothing is visible when I toggle the switch.

[nickleus (nick-humphrey)]

thanks to Everett Guerny for the workaround:
sudo modprobe -r iwl4965 && sudo modprobe iwl4965

that worked for me. i'm on an acer travelmate 6292 and wireless has stopped working after moving to intrepid. is there a fix for this?

[nickleus (nick-humphrey)]

i also installed linux-restricted-modules right before this so i don't know if that was a requirement as well?

[Leann Ogasawara (leannogasawara)]

Hi Guys,

I believe this bug should be resolved with a patch that was applied for bug 193970:

https://bugs.edge.launchpad.net/ubuntu/+bug/193970/comments/76

The kernel which contains this patch is now available in the intrepid-proposed repository:

https://bugs.edge.launchpad.net/ubuntu/+bug/193970/comments/85

"Please see https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed."

Please let us know your results. Thanks.

[matejcik (matejcik)]

If i load the module with killswitch disabled (wifi on), it appears to work; however, networkmanager appears to have some trouble deciding when to show the networks and when not.

If the killswitch is enabled (wifi off) when the module is loaded, it doesn't help, it is still impossible to turn the wifi on.

also, no messages in dmesg about rfkill flip - maybe it happens "silently" and networkmanager is not notified? when i turn the wifi off, dmesg says:
[ 770.852065] wlan0: No ProbeResp from current AP 00:1a:1e:a6:f9:c0 - assume out of range
which doesn't really make that much sense ;e)

[Everett Guerny (everett)]

Thank you for the update, Leann.

That said, using 2.6.27-8-generic in Intrepid, I'm still experiencing the same issue as originally reported (and as described by matejcik a few minutes ago).

[Leann Ogasawara (leannogasawara)]

Hi Guys,

I apologize, but I slightly jumped the gun here after seeing Martin's post. You'll want to test the the 2.6.27-10.20 kernel which I promise is making it's way into intrepid-proposed. Please keep and eye out for it. Thanks and sorry again for any inconvenience or confusion.

[Everett Guerny (everett)]

Looks like this works now with the -10 kernel. Thank you for the updates, Leann!

Do you know if it is within the scope of this bug that enabling the kill switch after having been connected to an access point, networkmanager still thinks the wireless is available and spins for a bit before asking for a passphrase for the last network to which one was connected?

[Andy Whitcroft (apw)]

On Wed, Nov 26, 2008 at 01:36:46AM -0000, Everett Guerny wrote:
> Looks like this works now with the -10 kernel. Thank you for the
> updates, Leann!
>
> Do you know if it is within the scope of this bug that enabling the kill
> switch after having been connected to an access point, networkmanager
> still thinks the wireless is available and spins for a bit before asking
> for a passphrase for the last network to which one was connected?

@Everett Guerny -- I think the network manager confusion when rfkill is
enabled is very likely being sorted under bug #267875. This bug itself
is very likely effectivly a duplicate of bug #193970, the fix for which
is in the 2.6.26-10.20 kernel.

I can confirm that this kernel is now officially being offered by
update-manager.

[Launchpad Janitor (janitor)]

Per a decision made by the Ubuntu Kernel Team, bugs will longer be assigned to the ubuntu-kernel-team in Launchpad as part of the bug triage process. The ubuntu-kernel-team is being unassigned from this bug report. Refer to https://wiki.ubuntu.com/KernelTeamBugPolicies for more information. Thanks.