> > Looks better, but > > > > > +static struct work_struct rfkill_any_work; > > > > At least on module exit you need to cancel this work. > > It is cancelled in rfkill_any_led_trigger_unregister(). It seemed > fitting to do it this way as rfkill_any_work is initialized in > rfkill_any_led_trigger_register(). And if CONFIG_RFKILL_LEDS=n, > rfkill_any_work is neither initialized nor scheduled, so we should be > good as well. Am I missing something? No, I just missed that, sorry. Applied the patch now. johannes
