test: arm: irq: Add overlays files for Infineon boards

[npal-cy]

Infineon: board: Add CONFIG_GPIO to defconfigs

Add CONFIG_GPIO from defconfigs for Infineon boards.

Revert pull/81377, which affect some ble samples which
used GPIO.

Signed-off-by: Nazar Palamar [email protected]

test: arm: irq: Add overlays files for Infineon boards

Changed interrupt priority for GPIO, default 6 is not suitable for
for the ZERO_LATENCY_IRQS function used in this test.
used in this test.

FIX CI: https://github.com/zephyrproject-rtos/zephyr/actions/runs/11524969451/job/32695832470?pr=80025
Signed-off-by: Nazar Palamar [email protected]

[mmahadevan108]

I had expressed my concerns about deleting GPIO from the defconfig #81377 (comment)
However you had mentioned that change was needed for the zero latency use-case.
Can you please explain what you are changing here?

[thedjnK]

This was literally removed not even 1 day ago 7e1b00d why was this removed only to be re-added?

[npal-cy]

I had expressed my concerns about deleting GPIO from the defconfig #81377 (comment)
However you had mentioned that change was needed for the zero latency use-case.
Can you please explain what you are changing here?

Yes this changes was fixed zero latency use-case, but i found it affected some sample(#81435). So before 4.0 release it safe to fix locally only zero latency by adding overlay files rather then add overlays to a lot set of samples...(see my closed PR #81435)

[npal-cy]

This was literally removed not even 1 day ago 7e1b00d why was this removed only to be re-added?

see my comment above.

[dkalowsk]

Looking at this right now we are marking this for backport to 4.0.1. We're not comfortable dismissing the change request comment here based upon the answer provided.

[thedjnK]

Revert pull/81377, which affect some ble samples which used GPIO.

If a sample needs GPIO then the sample should enable it

[npal-cy]

If a sample needs GPIO then the sample should enable it

Hi thedjnK, need some clarification here.

1. If board has some LED/buttons defined in board DTS, should we mandatory enable CONFIG_GPIO or not?

2. "If a sample needs GPIO then the sample should enable..." I came to the same conclusion when I disabled GPIO from the board defconfig (7e1b00d). but in samples: bluetooth: update prj.conf (added CONFIG_GPIO) #81435 was explained that GPIO should be enabled by board-specific conf files (samples: bluetooth: update prj.conf (added CONFIG_GPIO) #81435 (comment)).
   I have decided to revert 7e1b00d. Otherwise, I will have to go over all the samples and check if they use GPIO. If so, I will need to add the board-specific configuration files to enable board.

[thedjnK]

If board has some LED/buttons defined in board DTS, should we mandatory enable CONFIG_GPIO or not?

It should be in the defconfig yes

"If a sample needs GPIO then the sample should enable..." I came to the same conclusion when I disabled GPIO from the board defconfig (7e1b00d). but in #81435 was explained that GPIO should be enabled by board-specific conf files (#81435 (comment)).

@Thalley seems like there is a problem with the samples, there should be #ifdef's for the GPIOs, one has it correct: https://github.com/npal-cy/zephyr-1/blob/35cd59ce4cee79aec8cb4956bb01d417c731d87e/samples/bluetooth/peripheral_hr/src/main.c#L122 but this is wrong because it assumes GPIO support is always enabled: https://github.com/npal-cy/zephyr-1/blob/35cd59ce4cee79aec8cb4956bb01d417c731d87e/samples/bluetooth/mesh/src/board.c#L14 a board might have buttons defined but if CONFIG_GPIO=n then the buttons shouldn't even be looked at

[thedjnK]

Changes here are OK but (outside scope of this PR) Bluetooth samples should be fixed so that having GPIO disabled does not cause build errors

[Thalley]

If board has some LED/buttons defined in board DTS, should we mandatory enable CONFIG_GPIO or not?

It should be in the defconfig yes

"If a sample needs GPIO then the sample should enable..." I came to the same conclusion when I disabled GPIO from the board defconfig (7e1b00d). but in #81435 was explained that GPIO should be enabled by board-specific conf files (#81435 (comment)).

@Thalley seems like there is a problem with the samples, there should be #ifdef's for the GPIOs, one has it correct: https://github.com/npal-cy/zephyr-1/blob/35cd59ce4cee79aec8cb4956bb01d417c731d87e/samples/bluetooth/peripheral_hr/src/main.c#L122 but this is wrong because it assumes GPIO support is always enabled: https://github.com/npal-cy/zephyr-1/blob/35cd59ce4cee79aec8cb4956bb01d417c731d87e/samples/bluetooth/mesh/src/board.c#L14 a board might have buttons defined but if CONFIG_GPIO=n then the buttons shouldn't even be looked at

Yeah, I got aware of this issue. I haven't touched any of those samples, nor am I maintainer of them, so I would suggest to create GH issue for them and assign to @PavelVPV and @alwa-nordic (unless they want to do themselves). None of the ISO/LE audio samples have this issue as far as I am aware