drivers: i2c: stm32: Disable suspend to idle during the transaction

[coran21]

Suspend-to-idle stops I2C module clocks, which must remain active during the transaction.

Waiting for semaphore inside stm32_i2c_msg_write and stm32_i2c_msg_read triggers suspend to idle when pm is enabled, which results in disabled clocks, timeout, and error.

[erwango]

I think the proposal is correct, but highlights the complexity of PM subsystem.
I'd like PM subsystem maintainer to comment on this case.

[coran21]

@ceolin
Could you look at the suggested change?

[github-actions]

This pull request has been marked as stale because it has been open (more than) 60 days with no activity. Remove the stale label or add a comment saying that you would like to have the label removed otherwise this pull request will automatically be closed in 14 days. Note, that you can always re-open a closed pull request at any time.

[ceolin]

@coran21 @erwango Sorry for the late reply, I think I was not in the review list and have missed the notification. I tend to go over prs that have me listed as reviewer.

[erwango]

@coran21 Please rebase

[coran21]

@coran21 Please rebase

@erwango
I want to rebase, but I'm unsure how to do that correctly because the last time I rebased the code from Zephyr:main into my local branch, it pushed all the new commits into the PR, and I had to close it.

I wrote you a couple messages asking for help:

It looks like there were some changes in the actual file which resulted in conflict. Could you help me how I can update my local branch fix_stm32_i2c_disable_suspend with these changes so that I can push the changes again? I know that I should do some kind of rebase, but it happened to me a few times that I did a rebase and force-pushed all the previous commits, which were rebased into the pull request, and it was a mess.
In Zephyr documentation is that I should do this while I have checked out my branch with changes:

git fetch --all
git rebase --ignore-whitespace upstream/main
And do interactive rebase of my commit, is that true? I'm really not sure what to do if there is a conflict in my commit.

@erwango, Could you help me with that? How could I update my local branch, which has conflicts with Zephyr:main, so that it won't push all the Zephyr commits that were merged after this commit after I do force-push to this PR branch?

I would really appreciate any help with this.

[erwango]

I would really appreciate any help with this.

If there is only one commit, what you can do is:

• create a new branch temp_i2c_rebase
• check it out on top of current main
• cherry pick your commit from branch fix_stm32_i2c_disable_suspend to the new branch
• resolve the conflict
• once you're done, rename fix_stm32_i2c_disable_suspend to fix_stm32_i2c_disable_suspend_old
• rename temp_i2c_rebase to fix_stm32_i2c_disable_suspend
• force push the new stuff

[coran21]

@erwango That was exactly what I was looking for, than you!

[Oleh-Kravchenko]

Doesn't fix issue for STM32F1X.
This patch works for me #58929 (comment)

*not really work, but reduce reproducibility a lot :-/