| View previous topic :: View next topic |
| Author |
Message |
charles17
Advocate
Joined: 02 Mar 2008
Posts: 3686
|
 Posted: Tue Sep 14, 2010 5:24 pm Post subject: RFKILL now makes Trouble with Baselayout-2 |
 |
|
After kernel upgrade from 2.5.29-r5 to 2.6.32-r7 RFKILL seems to be broken.
With gentoo-sources-2.6.29-r5 on my laptop (Compaq nc6320) the HW RFKILL button properly switched radio off an on.
After kernel upgrade to gentoo-sources-2.6.32-r7 I can use it to switch radio off. But then the button is dead and doesn't do anything. To bring network up again I have to use "/etc/init.d/net.wlan0 restart" instead.
I know there has been changes in the kernel configuration and in my 2.6.32-r7 kernel I have | Code: | grep -e 'RFKILL\|IWL .conf
CONFIG_RFKILL=y
CONFIG_RFKILL_LEDS=y
CONFIG_RFKILL_INPUT=y
CONFIG_IWLWIFI=m
CONFIG_IWLWIFI_LEDS=y
# CONFIG_IWLWIFI_SPECTRUM_MEASUREMENT is not set
# CONFIG_IWLWIFI_DEBUG is not set
# CONFIG_IWLAGN is not set
CONFIG_IWL3945=m
CONFIG_IWL3945_SPECTRUM_MEASUREMENT=y
|
Is there a kernel bug with RFKILL or do I need to modify something else except the kernel itself?
Any help highly appreciated
Last edited by charles17 on Sat May 07, 2011 3:58 pm; edited 2 times in total |
|
| Back to top |
|
 |
DONAHUE
Watchman
Joined: 09 Dec 2006
Posts: 7651
Location: Goose Creek SC
|
| Posted: Tue Sep 14, 2010 6:56 pm Post subject: |
|
|
| ? |
|
| Back to top |
|
|
charles17
Advocate
Joined: 02 Mar 2008
Posts: 3686
|
| Posted: Wed Sep 15, 2010 9:06 am Post subject: |
|
|
Thanks for the hint. I've tried it, but emerging net-wireless/rfkill did not help. The hardware RFKILL button is still dead after switch off.
Anyway, net-wireless/rfkill is masked unstable.
Turning back to the old 2.6.29-r5 kernel solves the problem. But how could I upgrade without having this rfkill problem? |
|
| Back to top |
|
|
DONAHUE
Watchman
Joined: 09 Dec 2006
Posts: 7651
Location: Goose Creek SC
|
| Posted: Wed Sep 15, 2010 1:23 pm Post subject: |
|
|
see if this helps
based on this I would suspect 2.5.29-r5 throws an acpi event and 2.6.32-r7 does not |
|
| Back to top |
|
|
charles17
Advocate
Joined: 02 Mar 2008
Posts: 3686
|
| Posted: Wed Sep 15, 2010 4:06 pm Post subject: |
|
|
Not sure if this is an acpi event. I have something in the messages when switching off | Code: | Sep 13 18:00:09 server acpid: client 5122[0:0] has disconnected
Sep 13 18:00:31 server kernel: iwl3945 0000:08:00.0: Card state received: HW:Kill SW:On
Sep 13 18:00:31 server kernel: wlan0: deauthenticating from 00:1c:4a:46:4d:b0 by local choice (reason=3)
Sep 13 18:00:31 server kernel: iwl3945 0000:08:00.0: Error sending REPLY_LEDS_CMD: enqueue_hcmd failed: -5
Sep 13 18:00:31 server dhcpcd[5979]: wlan0: carrier lost
Sep 13 18:00:31 server wpa_cli: interface wlan0 DISCONNECTED
Sep 13 18:00:32 server fetchmail[6188]: terminated with signal 15
Sep 13 18:00:32 server ntpd[6325]: Terminating
Sep 13 18:00:32 server dhcpcd[5979]: wlan0: received SIGTERM, stopping
|
with HW:KILL in the second line kernel: iwl3945 0000:08:00.0: Card state received: HW:Kill SW:On.
But there is no such event when I press the button again. I have to manually restart "/etc/init.d/wlan0 restart".
The workaround by AM088 seems not to be applicable for me since I don't have /etc/init.d /wpa_supplicant. Instead wpa_supplicant will be started when wlan0 comes up.
Any solution without having to create additional scripts and udev rules? |
|
| Back to top |
|
|
DONAHUE
Watchman
Joined: 09 Dec 2006
Posts: 7651
Location: Goose Creek SC
|
| Posted: Wed Sep 15, 2010 4:57 pm Post subject: |
|
|
So the fixed script which may work with both soft and hard blocks and should be for your situation is:
Suggest | Code: | | nano /etc/udev/rfkill.sh |
fill it with: | Quote: | #!/bin/bash
if [ "${RFKILL_STATE}" == 1 ]; then
/etc/init.d/net.wlan0 start
else # state is 0 or 2
/etc/init.d/net.wlan0 stop
fi |
make it executable | Code: | | chmod 775 /etc/udev/rfkill.sh |
reboot and test the button.
if fails: error messages?
use button to kill wlan0
| Code: | | rfkill unblock wifi |
push button. wlan0 start? if not:
| Code: | | rfkill unblock wifi |
wlan0 start? |
|
| Back to top |
|
|
charles17
Advocate
Joined: 02 Mar 2008
Posts: 3686
|
| Posted: Wed Sep 15, 2010 6:35 pm Post subject: |
|
|
Now it works, even without reboot | Code: | KERNEL[1284574130.440528] change /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/ieee80211/phy0/rfkill0 (rfkill)
KERNEL[1284574131.373197] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::radio (leds)
KERNEL[1284574131.373322] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::assoc (leds)
UDEV [1284574131.373503] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::radio (leds)
KERNEL[1284574131.373717] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::RX (leds)
KERNEL[1284574131.373857] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::TX (leds)
UDEV [1284574131.373876] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::RX (leds)
UDEV [1284574131.374410] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::assoc (leds)
UDEV [1284574131.374711] add /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/leds/iwl-phy0::TX (leds)
UDEV [1284574131.853079] change /devices/pci0000:00/0000:00:1c.0/0000:08:00.0/ieee80211/phy0/rfkill0 (rfkill)
|
Thanks a lot for your help. |
|
| Back to top |
|
|
DONAHUE
Watchman
Joined: 09 Dec 2006
Posts: 7651
Location: Goose Creek SC
|
| Posted: Wed Sep 15, 2010 6:41 pm Post subject: |
|
|
Well done. Pls edit title of your first post to include [Solved] for the next guy.
Thanks go to AM088. |
|
| Back to top |
|
|
aCOSwt
Bodhisattva
Joined: 19 Oct 2007
Posts: 2537
Location: Hilbert space
|
| Posted: Thu Sep 16, 2010 1:30 pm Post subject: Re: [solved] RFKILL broken in gentoo-sources-2.6.32-r7 |
|
|
| charles17 wrote: | | After kernel upgrade from 2.5.29-r5 to 2.6.32-r7 |
I apologize for going off topic however I am interested in the facts governing a decision to upgrade from 2.5.29 to 2.6.32 and not to a newer version. |
|
| Back to top |
|
|
charles17
Advocate
Joined: 02 Mar 2008
Posts: 3686
|
| Posted: Thu Sep 16, 2010 5:52 pm Post subject: |
|
|
The kernel's CONFIG_RFKILL seems to be broken in 2.6.32-r7. But you can use that kernel with the workaround provided in this thread.
Disabling CONFIG_RFKILL would enable the buttons capabilities but only very unreliably. So you'd better have CONFIG_RFKILL=y and use the mentioned udev rule. It works. |
|
| Back to top |
|
|
charles17
Advocate
Joined: 02 Mar 2008
Posts: 3686
|
| Posted: Sat May 07, 2011 3:56 pm Post subject: RFKILL makes Trouble with Baselayout-2 |
|
|
Now, 8 months later and after migration to openrc and baselayout-2, I am getting troubles again.
Each time on booting the system hangs with a one minute delay waiting for uevents being processed. The message is | Code: | udevadm settle - timeout of 60 seconds reached, the event queue contains
/sys/devices/pci0000:00/0000:00:1c.0/0000:08:00.0/ieee80211/phy0/rfkill0
/sys/devices/pci0000:00/0000:00:1c.0/0000:08:00.0/ieee80211/phy0/rfkill0
/sys/devices/pci0000:00/0000:00:1c.0/0000:08:00.0/ieee80211/phy0/rfkill0 |
Somehow, baselayout-2 does no longer like my existing rule and script
| /etc/udev/rules.d/40-rfkill.rules wrote: | SUBSYSTEM=="rfkill", ATTR{type}=="wlan", RUN+="/etc/udev/rfkill.sh"
|
| /etc/udev/rfkill.sh wrote: | #!/bin/bash
if [ "${RFKILL_STATE}" == 1 ]; then
/etc/init.d/net.wlan0 start
rc
else # state is 0 or 2
/etc/init.d/net.wlan0 stop
fi |
Disabling above rule would clearly solve the timeout delay on boot, so I am almost sure something in the script is incompatible with the new Baselayout-2.
No clue how to adjust for the new boot process. Could you help again, please? |
|
| Back to top |
|
|
|
Networking & Security
