[SKIP SOF-TEST] Add new file: app/overlays/repro-build.conf

[marc-hb]

Passing options one by one in .github/workflows/zephyr.yml was becoming
unwieldy. Reduces Windows/Linux duplication. This also makes local
testing easier; less typing and guess work.

This should also help with #9116.

[marc-hb]

@andyross can you please help fix these (totally unrelated) sparse warning?

/zep_workspace/sof/src/audio/google/google_rtc_audio_processing.c:529:76: warning: incorrect type in argument 1 (different address spaces)
/zep_workspace/sof/src/audio/google/google_rtc_audio_processing.c:557:40: warning: incorrect type in assignment (different address spaces)
/zep_workspace/sof/src/audio/google/google_rtc_audio_processing.c:558:39: warning: incorrect type in assignment (different address spaces)
/zep_workspace/sof/src/audio/google/google_rtc_audio_processing.c:807:17: warning: incorrect type in argument 1 (different address spaces)

git blame points at you but this code isn't new; so I don't know why this just showed up recently. Someone changed some CONFIG_ somewhere?

Last passing daily test: https://github.com/thesofproject/sof/actions/runs/9606358628

First failing daily test 5 days ago: https://github.com/thesofproject/sof/actions/runs/9621445375

cc: @eddy1021

[andyross]

@andyross can you please help fix these (totally unrelated) sparse warning?

Looking right now. Are there docs on running sparse locally?

FWIW that code did change: it used to be getting the big block from the heap (where it would have been returned as a cached pointer) and now it's doing the conversion itself from a static pointer into .data, which is uncached. Never thought about the sparse interaction until this error showed up, but it's likely real enough and needs an appropriate cast/whitelist/whatever.

[marc-hb]

Looking right now. Are there docs on running sparse locally?

Better than documentation: the specific command can be found in .github/workflows/sparse-zephyr.yml. No CI secret for this one!

As you submitted fixup #9265, I assume you already found it.

[marc-hb]

A single "updownmixer" test failed in https://sof-ci.01.org/sof-pr-viewer/#/build/PR9264/build14068938, all other tests passed so this PR is clearly not to blame.

EDIT: for some unknown reason this was re-run and https://sof-ci.01.org/sof-pr-viewer/#/build/PR9264/build14069829 is now green.

[lyakh]

for my understanding: this alone won't be enough to resolve the difference between Windows and Linux SOF builds when using LLEXT as in https://github.com/thesofproject/sof/actions/runs/9550440753/job/26323025134?pr=9116 ?
EDIT: just saw #9116 (comment) - got the idea. Maybe even CONFIG_MODULES=n