Created attachment 43532 [details] /var/log/messages during suspend-resume with killswitch state changing If my notebook is booted or resumed with an enabled wifi killswitch (wifi off), disabling the killswitch does not turn on wifi. However, this is dependent on the state of the killswitch during shutdown or suspend, or whether the killswitch state changed during shutdown/suspend. Rules: start from a working system (Wifi Connected, killswitch off). Shutdown and restart with switch in test position. Once restarted, disable killswitch and record result – does WiFi work? Shutdown-restart Killswitch state Shutdown ON Shutdown OFF Startup ON Wifi fails to turn on Wifi fails to turn on Startup OFF OK OK Suspend-resume Killswitch state Suspend ON Suspend OFF Resume ON OK Wifi fails to turn on Resume OFF Wifi fails to turn on OK During a session: if wifi is working, the killswitch will work properly during the session.
What kernel version are you using?
Right now, 2.6.37 compiled from mainline. This bug has been around for a while though - since at least 2.6.31 (which I reported downstream, ages ago).
After you turn-off the rfkill, does "ifconfig wlan0 up" bring the device up?
John, I tried suspending with RF kill ON and resuming with RF kill OFF, then tried ifconfig wlan0 up and got this output SIOCSIFFLAGS: Operation not possible due to RF-kill
not sure the following patch help commit 3dd823e6b86407aed1a025041d8f1df77e43a9c8 Author: Don Fry <donald.h.fry@intel.com> Date: Sun Feb 6 09:29:45 2011 -0800 iwlagn: Re-enable RF_KILL interrupt when down With commit 554d1d027b19265c4aa3f718b3126d2b86e09a08 only one RF_KILL interrupt will be seen by the driver when the interface is down. Re-enable the interrupt when it occurs to see all transitions. Wey
Wey, I checked out commit 3dd823e6b86407aed1a025041d8f1df77e43a9c8 and compiled a kernel but it does not fix the issue. Thanks, Michael
ok, thansk for testing so just want to be sure, it is for 4965 and you are using 2.6.37 kernel Thanks Wey
Wey, lspci reports: 06:00.0 Network controller: Intel Corporation PRO/Wireless 4965 AG or AGN [Kedron] Network Connection (rev 61) And the kernel at that commit was v2.6.37 Michael
Just tested 2.6.39-rc1 and the first situation is fixed: If I boot with killswitch enabled (radio off), I can now enable radio by disabling killswitch.
1) Shutdown-restart Killswitch state Shutdown ON Shutdown OFF Startup ON Wifi fails to turn on Wifi fails to turn on Startup OFF OK OK 2) Suspend-resume Killswitch state Suspend ON Suspend OFF Resume ON OK Wifi fails to turn on Resume OFF Wifi fails to turn on OK how about suspend-resume? Thanks Wey
Wey, Unfortunately I just discovered that suspend-resume is broken in .39-rc1, so I can't test it; there is another bug to be fixed first... (specifically, resume hangs).
is the WiFi NIC hang the system, or something elae? Thanks Wey
Something else I think, syslog says: Mar 31 16:54:34 doris kernel: [ 657.300081] sky2 0000:07:00.0: eth0: disabling interface Mar 31 16:54:35 doris kernel: [ 657.589885] PM: Syncing filesystems ... done. Mar 31 16:54:35 doris kernel: [ 657.594636] PM: Preparing system for mem sleep Mar 31 16:55:36 doris kernel: imklog 4.2.0, log source = /proc/kmsg started. Mar 31 16:55:36 doris rsyslogd: [origin software="rsyslogd" swVersion="4.2.0" x-pid="661" x-info="http://www.rsyslog.com"] (re)start Mar 31 16:55:36 doris rsyslogd: rsyslogd's groupid changed to 103 Mar 31 16:55:36 doris rsyslogd: rsyslogd's userid changed to 101 Mar 31 16:55:36 doris rsyslogd-2039: Could no open output file '/dev/xconsole' [try http://www.rsyslog.com/e/2039 ]
Wey, I think it's the display driver: https://bugzilla.kernel.org/show_bug.cgi?id=32692 Once that is fixed I can do more testing on this bug.
np, we encounter the same problem here :-) Thanks Wey
It seems that there is another bug, after 32692, that causes similar failure to resume, which I am tracking down with bisect... will let you know when I've found it.
OK, looks like this bug is fixed in 2.6.39-rc3, though I had to patch a bug in sony-laptop to get resume to work at all. Hopefully Mattia's patch will make it into -rc4 so you can test it too. I can now change kill switch state during suspend to ram and enable my WiFi if it was disabled during boot.