Comment # 10 on bug 1004052 from Takashi Iwai

Thanks, this clarifies more.  It seems like some inconsistency in rfkill state.
Could you check the rfkill state at the boot time after WiFi disabled, and
after pressing the WiFi key (but not working WiFi), and after pressing again
(with working WiFi)?  Run like below at each time:

  % grep -H . /sys/class/rfkill/*/state


BTW, the atkbd warning must come from the wrong keymap via udev.
In /usr/lib/udev/hwdb.d/60-keyboard.hwdb, there is an entry:

evdev:atkbd:dmi:bvn*:bvr*:bd*:svnDell*:pn*
....
 KEYBOARD_KEY_88=unknown                                # Fn+F2 Turn On/Off
Wireless - handled in hardware

However, this is overridden for Latitude by the following:

evdev:atkbd:dmi:bvn*:bvr*:bd*:svnDell*:pnLatitude*:pvr*
evdev:atkbd:dmi:bvn*:bvr*:bd*:svnDell*:pnPrecision*:pvr*
 KEYBOARD_KEY_88=!                                      # wireless switch

So it generates a key even although it should be ignored.

That said, for "Unknown key pressed" messages, we need to fix udev hwdb instead
of kernel.