apollo4p_blue_kxr_evb: Multiple samples/tests fails to build in CI due to warnings

[aescolar]

apollo4p_blue_kxr_evb: Multiple samples/tests fails to build in CI due to warnings

[aaronyegx]

@RichardSWheatley I see you are looking at the issue (maybe another similar ticket) and have some fix in the HAL?

[pillo79]

Previous ticket for the same issue: #72775

[aescolar]

Just a question for reference, how come this failure wasn't noticed until months after the hal addition that causes this failure?

The CI run in a PR is a small subset of tests. What tests on what targets is run is based on what files are changed. Sometimes the subset does not cover enough and some issue gets in.
Other PRs which later affect a test may trigger the test being built/run on the affected targets revealing the issue.
The runs in main merge jobs are bigger, but still, do not cover everything. Though when something is detected there it tends to be reported quite soon though.
We also have a weekly job running on a lot of tests x targets combinations, but due to the amount of tests failures are too normal, so only when a pattern is clear it is seen and reported. That is what led to this issue being opened.

[aescolar]

CC @aaronyegx @AlessandroLuo @RichardSWheatley

[RichardSWheatley]

I have an AmbiqHal branch updated to include the folders. Let's speak at Ambiq to determine best path.

[nashif]

I have an AmbiqHal branch updated to include the folders. Let's speak at Ambiq to determine best path.

any updates?

[RichardSWheatley]

It's on the list.

[aescolar]

Lowering priority after the board has been disabled in CI

[RichardSWheatley]

Previous ticket for the same issue: #72775

Just a question for reference, how come this failure wasn't noticed until months after the hal addition that causes this failure?

[RichardSWheatley]

Just a question for reference, how come this failure wasn't noticed until months after the hal addition that causes this failure?

The CI run in a PR is a small subset of tests. What tests on what targets is run is based on what files are changed. Sometimes the subset does not cover enough and some issue gets in.

Other PRs which later affect a test may trigger the test being built/run on the affected targets revealing the issue.

The runs in main merge jobs are bigger, but still, do not cover everything. Though when something is detected there it tends to be reported quite soon though.

We also have a weekly job running on a lot of tests x targets combinations, but due to the amount of tests failures are too normal, so only when a pattern is clear it is seen and reported. That is what led to this issue being opened.

Did you see the PR my colleague made for this?
It's a short term fix only until we have time to address the issue.

PR #74232

If you approve, I will approve.

[aescolar]

#74275 ?
That one is not building yet

[RichardSWheatley]

#74275 ?

That one is not building yet

They need to add HAL changes to support the kconfig changes.

I am asking about whether that will work for you.

[aescolar]

I am asking about whether that will work for you.

If the tests pass again, sure. I only disabled it to unblock CI for others.