Issue3247 - Compiler optimization level causing displayio instability

[DavePutz]

Testing the demo spirit level script for the Clue showed that with the -O2 optimization from PR #3190 the display was not stable. This PR reverts the optimization back to -Os.

[dhalbert]

Testing the demo spirit level script for the Clue showed that with the -O2 optimization from PR #3190 the display was not stable. This PR reverts the optimization back to -Os.

In the past this kind of behavior (usually -Og vs -Os) has typically meant that something that should be volatile is not, or that there's some uninitialized.

[DavePutz]

@dhalbert Are there any tools to help track that sort of thing down? I think safest for now is to take the relatively small performance hit and revert the optimization.

[dhalbert]

We have no such tools now. Could you file a bug re displayio and optimization level? Thanks! I think we should assume this is a bug, not a compiler problem.

[deshipu]

I wonder if you could run valgrind on the unix build?

[DavePutz]

Closing, fixed by PR #3281. Tested OK with a Clue and Circuitplayground Bluefruit with an attached Gizmo display.

[dhalbert]

Great, thank you for testing!