Add support for I2C on STM32F7 family and on STM32F723E-DISCO board

[aurel32]

This patch series adds support for I2C bus of the STM32F7 family, enables it on the STM32F723E-DISCO board, and finally fixes an issue found while testing the I2C support in interrupt mode.

I have tested it using by communicating with the WM8994 chip on the STM32F723E-DISCO board.

[codecov-io]

Codecov Report

Merging #8716 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #8716   +/-   ##
=======================================
  Coverage   52.45%   52.45%           
=======================================
  Files         200      200           
  Lines       25121    25121           
  Branches     5244     5244           
=======================================
  Hits        13176    13176           
  Misses       9823     9823           
  Partials     2122     2122

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a159447...4dbcb0d. Read the comment docs.

[ydamigos]

Looks good. I am just blocking it because PR #7626 also introduces the NACK fix.

@aurel32 could you coordinate with @vaussard on this?

[aurel32]

@ydamigos, thanks for the pointer. Given there is already a PR opened about that issue, I believe it's better to discuss it there. In addition the fix is now conflicting with the changes introduced by PR#5224, and it more work is needed to understand how to handle a NACK in slave mode. I'll therefore just drop that commit from the pull request.

[aurel32]

And I have just added an additional cleanup commit to make @ulfalizer happy.

[aurel32]

@ydamigos, there was a third PR with a fix for this issue, PR #5224. Given it has just been merged, i confirm that just removing the commit from my PR was the right thing to do.

[erwango]

One comment.

Otherwise, could you update help message for I2C_STM32_V2. Message is not quite clear but we should be explicit that drivers now supports F7 (while previously it was 'compatible': I agree this is subtle :-))

[erwango]

Should be 'n' by default.

[ydamigos]

@erwango We enable buses on board level when there is a peripheral (e.g. WM8994 or touch panel in this case) that needs them, aren't we?

[erwango]

@ydamigos : we should configure them, and this is what is done here. But activation should always be done in application and not default.
Aim is that several developers on the same board know what is the base components for the board so they only enable what they need. But they don't have to bother disabling what other developers may have enabled meanwhile.

[ydamigos]

@erwango We need to clean up board _defconfig files then.

[erwango]

@ydamigos : yes, there is some work ahead! I raised #7151 earlier, waiting #7176 to be reviewed and used a reference (which was locked waiting for new SDK version and which I hope should get some feedbacks soon)

[aurel32]

@erwango: I have just done the requested changes, i.e. I updated the Kconfig.stm32 help message to mark F7 as supported, and removed CONFIG_I2C=y from the board defconfig.

[aurel32]

@erwango, I have just pushed a new version to solve the conflict introduced by PR #8663 and to add the new clock properties introduced in PR #8759.

[dbkinder]

Doc changes LGTM, thanks.

[aurel32]

I have just pushed a new version to solve the conflict due the merge of USB OTG FS support. There is no code change beside solving conflicts.

[aurel32]

I have just pushed a new version to solve the conflict due the merge of USB OTG HS support. There is no code change beside solving conflicts.