experimental jit causes infinite loops on 3.13 branch

tacaswell

Bug report

Bug description:

Attempting to install cython on the 3.13 branch fails with what looks like an infinite loop that sits there and burns CPU forever.

I have bisected this to #124266 which makes no sense to me, but I'm pretty confident about it as observed behavior.

bisect log:

Details

git bisect log
git bisect start
# status: waiting for both good and bad commits
# good: [60403a5409ff2c3f3b07dd2ca91a7a3e096839c7] Python 3.13.0
git bisect good 60403a5409ff2c3f3b07dd2ca91a7a3e096839c7
# status: waiting for bad commit, 1 good commit known
# bad: [ff044ed8004e31c1896ca641b81b13ab04e92837] [3.13] gh-124295: Add translation tests for argparse (GH-124803) (GH-126046)
git bisect bad ff044ed8004e31c1896ca641b81b13ab04e92837
# bad: [1279be610dd22fc116998a60c0f9a08fa18386ea] [3.13] gh-123133: clarify p=0 case for "f" and "e" formatting types (GH-125426) (#125428)
git bisect bad 1279be610dd22fc116998a60c0f9a08fa18386ea
# bad: [7bc99dd49ed4cebe4795cc7914c4231209b2aa4b] [3.13] Tee of tee was not producing n independent iterators (gh-123884) (gh-125081)
git bisect bad 7bc99dd49ed4cebe4795cc7914c4231209b2aa4b
# bad: [db3ccd8b62c7ff9f1ce5a7d1b8e7d68eed09517f] [3.13] gh-53780: Ignore the first "--" (double dash) between an option and command in argparse (GH-124275) (GH-125073)
git bisect bad db3ccd8b62c7ff9f1ce5a7d1b8e7d68eed09517f
# bad: [11d4b54b88063efbeb8968263dfa77dcfb816f8c] [3.13] gh-116850: Fix argparse for namespaces with not directly writable dict (GH-124667) (GH-124757)
git bisect bad 11d4b54b88063efbeb8968263dfa77dcfb816f8c
# bad: [167d8d2f072505a85c6db72fdaf4edffd6d17f3a] [3.13] gh-63143: Fix parsing mutually exclusive arguments in argparse (GH-124307) (GH-124418)
git bisect bad 167d8d2f072505a85c6db72fdaf4edffd6d17f3a
# good: [80ba17a3dd383882622cf303a2f2fb19cbaa0ee6] [3.13] gh-112804: Clamping timeout value for _PySemaphore_PlatformWait (gh-124914) (gh-124991)
git bisect good 80ba17a3dd383882622cf303a2f2fb19cbaa0ee6
# bad: [63870162f4d80359025762aec802d46ab0ecdb6b] [3.13] gh-81691: Fix handling of multiple "--" (double dashes) in argparse (GH-124233) (GH-124266)
git bisect bad 63870162f4d80359025762aec802d46ab0ecdb6b
# good: [6fe746d702bfeeafbde1e9df2b1b5e061808fbb5] [3.13] gh-122392: IDLE - Fix overlapping lines in browsers (GH-122392) (GH-124975) (#125061)
git bisect good 6fe746d702bfeeafbde1e9df2b1b5e061808fbb5
# first bad commit: [63870162f4d80359025762aec802d46ab0ecdb6b] [3.13] gh-81691: Fix handling of multiple "--" (double dashes) in argparse (GH-124233) (GH-124266)

The Python invocation it is failing on from setuptools is :

/home/tcaswell/.virtualenvs/bisect/bin/python3 -u -c
exec(compile('''
# This is <pip-setuptools-caller> -- a caller that pip uses to run setup.py
#
# - It imports setuptools before invoking setup.py, to enable projects that directly
#   import from `distutils.core` to work with newer packaging standards.
# - It provides a clear error message when setuptools is not installed.
# - It sets `sys.argv[0]` to the underlying `setup.py`, when invoking `setup.py` so
#   setuptools doesn't think the script is `-c`. This avoids the following warning:
#     manifest_maker: standard file '-c' not found".
# - It generates a shim setup.py, for handling setup.cfg-only projects.
import os, sys, tokenize

try:
    import setuptools
except ImportError as error:
    print(
        "ERROR: Can not execute `setup.py` since setuptools is not available in "
        "the build environment.",
        file=sys.stderr,
    )
    sys.exit(1)

__file__ = %r
sys.argv[0] = __file__

if os.path.exists(__file__):
    filename = __file__
    with tokenize.open(__file__) as f:
        setup_py_code = f.read()
else:
    filename = "<auto-generated setuptools caller>"
    setup_py_code = "from setuptools import setup; setup()"

exec(compile(setup_py_code, filename, "exec"))
''' % ('/home/tcaswell/source/p/cython/cython/setup.py',), "<pip-setuptools-caller>", "exec")) bdist_wheel -d /tmp/pip-wheel-l3necg1_~

I have not succeeded in getting traceback out of this yet.

I'm pretty confident in the bisect on the 3.13 branch, turning the jit off works fine on the both the tip of the branch and 6387016
the initial commit on main that was backported fails to build cython for other reasons (see GH-125868: Fix STORE_ATTR_WITH_HINT specialization #125876)
the current main branch works fine both with and without the jit

attn @da-woods @scoder

CPython versions tested on:

3.13, CPython main branch

Operating systems tested on:

Linux