Hi,
On Fri, Feb 23, 2018 at 06:04:22PM +0800, Jeffy Chen wrote:
> On 02/13/2018 06:13 AM, Brian Norris wrote:
> > > >
> > > > if (bdata->gpiod) {
> > > >+ int active_low = gpiod_is_active_low(bdata->gpiod);
> > > >+
> > > > if (button->debounce_interval) {
> > > > error = gpiod_set_debounce(bdata->gpiod,
> > > > button->debounce_interval *
> > > > 1000);
> > > >@@ -568,6 +572,16 @@ static int gpio_keys_setup_key(struct
> > > >platform_device *pdev,
> > > > isr = gpio_keys_gpio_isr;
> > > > irqflags = IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING;
> > > >
> > > >+ switch (button->wakeup_event_action) {
> > > >+ case EV_ACT_ASSERTED:
> > > >+ bdata->wakeup_trigger_type = active_low ?
> > > >+ IRQF_TRIGGER_FALLING :
> > > >IRQF_TRIGGER_RISING;
> > > >+ break;
> > > >+ case EV_ACT_DEASSERTED:
> > > >+ bdata->wakeup_trigger_type = active_low ?
> > > >+ IRQF_TRIGGER_RISING :
> > > >IRQF_TRIGGER_FALLING;
> > > >+ break;
> > What about EV_ACT_ANY? And default case? I think for ANY, we're OK
> > letting suspend/resume not reconfigure the trigger type, but maybe a
> > comment here to note that?
> right, will add comment in the next version.
> >
> > > >+ }
> > > > } else {
> > What about the 'else' case? Shouldn't we try to handle that?
> i think the else case is for irq key, which would generate down and up
> events in one irq, so it would use the same trigger type for all these 3
> cases.
Not necessarily. It uses whatever trigger was provided in
platform/DT/etc. data. You could retrieve that with
irq_get_trigger_type() and try to interpret that. Or you could just
outlaw using that combination (e.g., in the binding documentation).
Brian
> i'll add comment in the next version too.
> >
> > Brian
> >
>
>
