[TOPIC-GPIO] proposed workaround to distinguish interrupt disable from default no-change

[pabigot]

Fixes #19552.

[zephyrbot]

All checks are passing now.

checkpatch (informational only, not a failure)

-:101: WARNING:LONG_LINE: line over 80 characters
#101: FILE: include/drivers/gpio.h:640:
+	mode = (enum gpio_int_mode)(flags & (GPIO_INT_EDGE | GPIO_INT_DISABLE | GPIO_INT_ENABLE));

- total: 0 errors, 1 warnings, 118 lines checked

NOTE: For some of the reported defects, checkpatch may be able to
      mechanically convert to the typical style using --fix or --fix-inplace.

Your patch has style problems, please review.

NOTE: Ignored message types: AVOID_EXTERNS BRACES CONFIG_EXPERIMENTAL CONST_STRUCT DATE_TIME FILE_PATH_CHANGES MINMAX NETWORKING_BLOCK_COMMENT_STYLE PRINTK_WITHOUT_KERN_LEVEL SPLIT_STRING VOLATILE

NOTE: If any of the errors are false positives, please report
      them to the maintainers.

Tip: The bot edits this comment instead of posting a new one, so you can check the comment's history to see earlier messages.

[mnkp]

This is a well prepared PR, it does solve the problem with interrupt handling that we have. My only concern is that the approach is not quite clean. It's applicable as a temporary workaround but not as a long term solution. In the drivers one can find code like this one:

	if (mode != GPIO_INT_MODE_DISABLED) {
		/* Enable the interrupt. */
		...
	}

Which will be confusing to any new user reading the code. It's because part of the states encoded by enum gpio_int_mode are handled within this API and part by the driver. For the solution to be permanent / long term all of the states should be handled directly by the driver.

I would like to propose another approach that doesn't require adding GPIO_INT_MODE_UNCHANGED state. In z_impl_gpio_pin_interrupt_configure we need to remove the condition if (mode != GPIO_INT_MODE_UNCHANGED) { reverting back to the original implementation. Then we need to add the following assert (next to the existing one):

	__ASSERT((flags & (GPIO_INT_DISABLE | GPIO_INT_ENABLE)) != 0),
		 "Interrupt must be either enabled or disabled");

This way we keep the original behavior. gpio_pin_interrupt_configure function will either enable or disable interrupt. The don't modify behavior is in any case not useful. The gpio_pin_configure function will not accidentally disable interrupts since now empty and non empty interrupt configuration flags can be distinguished. The driver code will not require any updates and will still read well.

[pabigot]

Had this PR been addressed promptly we might not have a bunch of drivers that can't distinguish between interrupt disable and no-change states.

I've made the change as requested.