[ESP32-C6] I2C writes get stuck, since commit f9203d - affects 0.23.{01} and latest main

[lure23]

Bug description

I came back to a project of mine, using the VL53L5CX distance sensor, and after updating esp-hal changes to 0.23.0, it always got stuck in I2C operations.

I narrowed down the reason to the commit in the title.

To Reproduce

A bit lengthy instructions (hw and a driver download with click-through are needed) are here.

Expected behavior

I can continue developing the code, using latest esp-hal releases.

Actual behavour

I am stuck at 0.22.0 release.

Environment

• Target device: ESP32-C6
• Crate name and version: esp-hal 0.23.0, 0.23.1 and latest main (bb1ad1)

[bjoernQ]

I happen to own one of those ToF sensors and have a very simple example using it here: https://github.com/bjoernQ/vl53l5-c2rust

That one still works fine for me. (in the repo it's using a C3, but verified it's works the same on C6)

Your code is unfortunately huge - if you can reduce it a minimal reproducer would help a lot.

The referenced commit belongs to #2544 which basically just "turns off" the peripheral when it gets dropped (just to give some direction to look into)

Hard to say if that commit actually breaks something for you or if it just reveals a problem lurking in the code already before (not saying it's the case)

[lure23]

Thanks @bjoernQ

I'll likely need to carry on with what works for me (0.22), but:

That one still works fine for me. (in the repo it's using a C3, but verified it's works the same on C6)

It's good to know stuff works elsewhere (with the same sensor).

Your code is unfortunately huge - if you can reduce it a minimal reproducer would help a lot.

That made me realize it... might be. 😬 The size is certainly not the goal - on the contrary. The aim is to bridge the gap for a newcomer to get sensors working, but... if it cannot be used as a replication... perhaps something is wrong.

Ideally, it'd be something akin to Arduino in easy-of-embracing.

The referenced commit belongs to #2544 which basically just "turns off" the peripheral when it gets dropped (just to give some direction to look into)

My code keeps the same I2c all through the example, so no dropping. Weird.

I'm looking through the commit, the current (main) code, and your working code. Perhaps I spot something.. In the mean time, I'll leave this issue open, if it's okay.

Hard to say if that commit actually breaks something for you or if it just reveals a problem lurking in the code already before (not saying it's the case)

I agree. Just learning.