On Thursday 06 August 2009, Daniel Mack wrote:

On Wed, Aug 05, 2009 at 10:15:52PM +0200, pHilipp Zabel wrote:

quoted

On Wed, Aug 5, 2009 at 8:29 PM, Daniel Mack[off-list ref] wrote:

quoted

-static int gpio_keys_resume(struct platform_device *pdev)
+static int gpio_keys_resume(struct device *dev)
 {
-       struct gpio_keys_platform_data *pdata = pdev->dev.platform_data;
+       struct gpio_keys_platform_data *pdata = dev->platform_data;
       int i;

-       if (device_may_wakeup(&pdev->dev)) {
+       if (device_may_wakeup(dev)) {
               for (i = 0; i < pdata->nbuttons; i++) {
                       struct gpio_keys_button *button = &pdata->buttons[i];
                       if (button->wakeup) {

@@ -251,19 +251,27 @@ static int gpio_keys_resume(struct platform_device *pdev)

       return 0;
 }
+
+static struct dev_pm_ops gpio_keys_pm_ops = {
+       .suspend        = gpio_keys_suspend,
+       .freeze         = gpio_keys_suspend,
+       .resume         = gpio_keys_resume,
+       .thaw           = gpio_keys_resume,

I'm not sure I understand hibernation correctly, but isn't
.freeze/.thaw about saving state and halting/resuming the device
operation only?

It is.

quoted

It seems to me that enabling system wakeup functionality should go
into .poweroff. (See <linux/pm.h>)

That's correct.

And .restore() plays the role of ".resume() from hibernation".

Frankly, I'm not sure either. But luckily, the linux-pm list is copied
so someone can elborate on this :)

HTH

Best,
Rafael