gpio: add wakeup source support

[RomainPelletant]

• Update GPIO API with wakeup source read function
• Update GPIO Kconfig to add experimental support
• Add a sample with silabs gecko implementation

Fixes #74272

[bjarki-andreasen]

Have you considered the existing function gpio_get_pending_int()? This is intended to be used to check if any interrupt is pending, which matches the GPIO_INT_WAKEUP trigger used to configure the GPIO wakeup.

The function you are adding here does have an additional feature, being that it can return a mask of which specific GPIO/GPIOs triggered the interrupt, which I think could be a useful addition. I would agree with adding a new API (which would deprecate gpio_get_pending_int()) which provides an optional pin mask argument:

int gpio_get_pending_interrupts(const struct device *dev, gpio_port_pins_t *mask);

Notice the shift away from including wakeup_source in the API, since this can be used for any interrupts, whatever their trigger. Also using the gpio_port_pins_t for the mask instead of uint32_t

Do note that many GPIO controller's don't store which pin triggered a wakeup interrupt, only which port, so the mask argument has to be optional, as in, if the mask arg it is not NULL, and the GPIO controller can't specify the actual pin that has triggered the interrupt, the API should return -EINVAL or something :)

[RomainPelletant]

@bjarki-andreasen you are right. Support for EM4 is specific to current sleep mode: I am thinking about how could we manage that.
Thanks for your message