doc: GPIO_INT config option dependencies aren't clear

[dbkinder]

As reported on the dev mailing list, there's a dependency between GPIO_INT_EDGE and GPIO_INT that's not clear. Can this dependency be added to the Kconfig file?

I don’t understand what the documentation says for the flags:

GPIO_INT
   GPIO pin to trigger interrupt.

GPIO_INT_EDGE
   Do Edge trigger.

So, what I’m concluding is that you need to specify GPIO_INT to enable interrupts on the pin and then another flag like GPIO_INT_EDGE to say when in the signal processing you want the interrupt to occur? If so... makes sense (now). I thought saying GPIO_INT_EDGE was enough to enable interrupts and to say they should occur on edge. My bad.

[iha]

I agree. But I think it more intuitive to remove the flag to enable GPIO_INT in the documentation and leave it implicit:

/** @cond INTERNAL_HIDDEN */
/** GPIO pin to trigger interrupt. */
#define GPIO_INT                (1 << 1)
** @endcond */
 
/** GPIO pin trigger on level low or falling edge. */
#define GPIO_INT_ACTIVE_LOW     ((0 << 2) | GPIO_INT)
 
/** GPIO pin trigger on level high or rising edge. */
#define GPIO_INT_ACTIVE_HIGH    ((1 << 2) | GPIO_INT)
.
.
.

And using Kconfig to set this flag can be a bit annoying. I had to do something like this to set 16 GPIO ports for lack of API option.

[mnkp]

Should be fixed by #16648 (not yet on master). Only the configuration flags which are supposed to be used by the application code have doxygen documentation. All of these compound flags explicitly include GPIO_INT_ENABLE.

[carlescufi]

This has been addressed in the GPIO rework