GPIO: do not overwrite interrupt handler

[bugadani]

Another idea I'd want some feedback on. Closes #2009 I hope.

Why not keep this in Io::new? I intend to move the Pins out into Peripherals so that users don't need to create Io - unless they want to handle user interrupts. That is a bigger change, though.

[jessebraham]

I'm not sure I understand these changes, it seems to solve one problem while creating another. As far as I can tell, if you do not want to bind the interrupt handler you can no longer call esp_hal::init()? This will result in a large amount of boilerplate required for initializing the WDTs, clocks, etc. in this case.

[bugadani]

In which case do you explicitly not want to bind any interrupt handlers?

[jessebraham]

In which case do you explicitly not want to bind any interrupt handlers?

When using e.g. RTIC and using hardware tasks involving GPIO. This was the reason the Io::new_no_bind_interrupt() constructor was added in the first place. I've confirmed that using this branch and calling esp_hal::init() causes the sw_and_hw example to stop functioning as expected.

[bugadani]

RTIC seems to be working for me now, please take another look. If the user doesn't want our handler, they will be able to disable it using the interrupt APIs. We are now detecting whether the peripheral interrupt is bound to any CPU interrupt, so it shouldn't be possible for us to overwrite any early bound interrupt handlers. Later the users can swap ours with anything they want using the esp_hal::interrupt APIs.

[jessebraham]

Can confirm things are working as expected again, thanks for taking care of that!

[bjoernQ]

LGTM