On 12:28 Wed 16 Nov , Shubhrajyoti wrote:
> On Wednesday 16 November 2011 04:07 PM, oskar.andero@xxxxxxxxxxxxxxxx
> wrote:
> > On 19:29 Tue 15 Nov , Dmitry Torokhov wrote:
> >> On Tue, Nov 15, 2011 at 01:53:52PM +0100, oskar.andero@xxxxxxxxxxxxxxxx wrote:
> >>> On 10:43 Tue 15 Nov , Dmitry Torokhov wrote:
> >>>> On Tue, Nov 15, 2011 at 10:26:00AM +0100, oskar.andero@xxxxxxxxxxxxxxxx wrote:
> >>>>> From: Courtney Cavin <courtney.cavin@xxxxxxxxxxxxxxxx>
> >>>>> +
> >>>>> +#ifdef CONFIG_PM_SLEEP
> >>>>> +static int gp2a_suspend(struct device *dev)
> >>>>> +{
> >>>>> + struct i2c_client *client = to_i2c_client(dev);
> >>>>> + struct gp2a_data *dt = i2c_get_clientdata(client);
> >>>>> + int error;
> >>>>> +
> >>>>> + mutex_lock(&dt->device->mutex);
> >>>>> +
> >>>>> + if (dt->device->users) {
> >>>>> + if (device_may_wakeup(&client->dev)) {
> >>>>> + enable_irq_wake(client->irq);
> >>>> I think this part should happen regardless of whether device has users,
> >>>> only non wakeup source case needs it.
> >>> Hmm.. why would one want to enable irq_wake when there are no users?
> >>> Wouldn't this cause the device to wakeup at every irq and report an
> >>> switch event that no one listens to?
> >> You are suspending the system and want to have this device as a wakeup
> >> source. Note: not wake up _device_ at every IRQ but wake up the whole
> >> _system_ when device generates an IRQ while system is asleep.
> >> It does not matter whether there are users for the events; you
> >> want the system to wake up.
> >>
> >> At least this is the usual semantics.
> > I see you point. However, the way we use the proximity sensor we can only wake up
> > the system when we have an application that is actually interested in the change.
> > This is for power save reasons.
> > If we use the "usual semantic", we would wake up the system at every proximity
> > detection regardlessly. For instance, I wouldn't want to wake up a cell phone laying
> > on the desk when I put my hand over it. That would hurt the battery time.
> Even in that case it shouldn't harm the power if no user is there the
> close would have been called
> and you should be in the shut down mode? Am I missing something?
You are right - the chip will only generate interrupts when it's opened, so the case
I described shouldn't actually be a problem. However, does it make sense to enable irq_wake
for a device that will never generate interrupts?
Also, how does this correspond to the semantics that Dmitry described, especially:
> >> It does not matter whether there are users for the events; you
> >> want the system to wake up.
-Oskar
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
