Open randombit opened 4 months ago
Possibly related: https://bugs.python.org/issue33725
Oh smoking gun in the multiprocessing docs:
Changed in version 3.8: On macOS, the spawn start method is now the default. The fork start method should be considered unsafe as it can lead to crashes of the subprocess.
Oh but we're running on Python 3.12 in CI ...
Recurred again in https://github.com/randombit/botan/actions/runs/9965253706/job/27535202682 so #4216 was not the fix
Tried running the CLI tests repeatedly with multiple threads on my M2 mac w/ macOS 14.5 and Python 3.12.4. No luck. :(
I've seen this twice now:
https://github.com/randombit/botan/actions/runs/8614981865/job/23609543455?pr=3990
The "Ran 232 tests" line is the last thing the script prints before it returns, so it looks like Python is crashing with a segfault on exit. Maybe a bug in
multiprocessing
affecting Mac on Aarch64?