[ocaml5-issue] MinGW+MSVC 5.5.0+trunk may fail `Lin Dynlink stress test with Domain` with code 0xc0000374

[jmid]

On #575 I just observed a new error on MSVC 5.5.0+trunk which I haven't seen before:
https://github.com/ocaml-multicore/multicoretests/actions/runs/20234491076/job/58085432534?pr=575#step:16:1325

File "src/dynlink/dune", line 14, characters 7-16:
14 |  (name lin_tests)
            ^^^^^^^^^
(cd _build/default/src/dynlink && .\lin_tests.exe --verbose)
Command exited with code 0xc0000374.
[ ]    0    0    0    0 / 1000     0.0s Lin Dynlink stress test with Domain

Since this is the first occurrence - and only so far on MSVC and trunk I suspect it is caused by a recently merged PR... 🤔

[jmid]

On the merge to main of #577 this issue did not trigger, which indicates that it is not a consistent error (I don't think it has been fixed).

[jmid]

This triggered on the weekly run, but this time on the MinGW trunk workflow:
https://github.com/ocaml-multicore/multicoretests/actions/runs/20422260638/job/58676117371

random seed: 263767059
generated error fail pass / total     time test name

[ ]    0    0    0    0 /  100     0.0s negative Lin Dynlink test with Domain
[✓]    1    0    1    0 /  100     1.0s negative Lin Dynlink test with Domain

File "src/dynlink/dune", line 14, characters 7-16:
14 |  (name lin_tests)
            ^^^^^^^^^
(cd _build/default/src/dynlink && .\lin_tests.exe --verbose)
Command exited with code 0xc0000374.
[ ]    0    0    0    0 / 1000     0.0s Lin Dynlink stress test with Domain

Interestingly, this started triggering on MinGW too after ocaml/ocaml#13416 was merged, causing MSVC and MinGW to share the same thread implementation.