[issue45200] test_multiprocessing_fork: test_get() fails with timeout

STINNER Victor added the comment: > https://github.com/python/cpython/pull/28344/checks?check_run_id=3605759743 That's an Address Sanitizer job. Pablo: would it be possible to make the Address Sanitizer not mandatory on pull requests, until this libasan race condition is fixed? --

[issue45200] test_multiprocessing_fork: test_get() fails with timeout

STINNER Victor added the comment: When the test hangs, a background Python process starts to use 100% of my CPU. This process is stuck in _thread.start_new_thread(). I opened the process in gdb. (gdb) thread 1 [Switching to thread 1 (Thread 0x7fece33615c0 (LWP 132301))] #0

[issue45200] test_multiprocessing_fork: test_get() fails with timeout

STINNER Victor added the comment: > I'm able to reproduce the issue in a reliable way using attached bisect10 > file and the command: I'm talking about a Python built with Address Sanitizer: $ ./configure --with-address-sanitizer --without-pymalloc $ make --

[issue45200] test_multiprocessing_fork: test_get() fails with timeout

STINNER Victor added the comment: I'm able to reproduce the issue in a reliable way using attached bisect10 file and the command: $ ./python -m test test_multiprocessing_fork --timeout=40 -v --matchfile=bisect10 (...) test_fork (test.test_multiprocessing_fork.WithProcessesTestQueue) ... ok

[issue45200] test_multiprocessing_fork: test_get() fails with timeout

STINNER Victor added the comment: Sometimes, WithProcessesTestProcess.test_error_on_stdio_flush_1() hangs. -- ___ Python tracker ___

[issue45200] test_multiprocessing_fork: test_get() fails with timeout

Change by STINNER Victor : -- title: test_multiprocessing_fork failws with timeout -> test_multiprocessing_fork: test_get() fails with timeout ___ Python tracker ___