bpo-46709: fix race conditions in unittest/test_brake

[sobolevn]

Affected tests:

• ./python.exe -m test -m unittest.test.test_break.TestBreakDefaultIntHandler.testInterruptCaught test_unittest -F
• ./python.exe -m test -m unittest.test.test_break.TestBreakDefaultIntHandler.testSecondInterrupt test_unittest -F
• ./python.exe -m test -m unittest.test.test_break.TestBreakDefaultIntHandler.testTwoResults test_unittest -F
• ./python.exe -m test -m unittest.test.test_break.TestBreakDefaultIntHandler.testHandlerReplacedButCalled test_unittest -F

Before:

After (manually terminated):

I simply re-arranged assert statements to make this pass: https://docs.python.org/3/library/signal.html#execution-of-python-signal-handlers

Maybe there's something more clever to do instead? 🙂

https://bugs.python.org/issue46709

[kumaraditya303]

The tests were passing before so I guess it might have be caused due to a missing eval breaker in ceval.c in specialised bytecode to check for signals. Perhaps you can git bisect to see if this is the case?

[sobolevn]

The tests were passing before

They are still passing. Take a look:

But, they are flaky.

it might have be caused due to a missing eval breaker in ceval.c in specialised bytecode to check for signals. Perhaps you can git bisect to see if this is the case?

I will take a look if there were recent changes. Thanks for the idea! 👍

[sobolevn]

git bisect shows that this is the first failing commit: 3163e68

» git bisect good
3163e68c342434db37c69669017f96a4bb2d5f13 is the first bad commit
commit 3163e68c342434db37c69669017f96a4bb2d5f13
Author: Ken Jin <[email protected]>
Date:   Wed Oct 20 07:16:36 2021 +0800

    [bpo-44525](https://bugs.python.org/issue44525): Specialize ``CALL_FUNCTION`` for C function calls (GH-26934)

 Include/internal/pycore_code.h                     |   1 +
 Include/opcode.h                                   |  51 +++----
 Lib/opcode.py                                      |   5 +
 .../2021-06-28-22-23-59.[bpo-44525](https://bugs.python.org/issue44525).sSvUKG.rst       |  10 ++
 Python/ceval.c                                     | 147 +++++++++++++++++++++
 Python/opcode_targets.h                            |  54 ++++----
 Python/specialize.c                                | 147 +++++++++++++++++++++
 7 files changed, 365 insertions(+), 50 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Core and Builtins/2021-06-28-22-23-59.[bpo-44525](https://bugs.python.org/issue44525).sSvUKG.rst

CC @Fidget-Spinner

[sobolevn]

I think the proper fix would be adding CHECK_EVAL_BREAKER(); into specialized CALL_NO_KW_ opcodes. Testing this out 🤔

[sobolevn]

Duplicate of https://github.com/python/cpython/pull/30826/files