co_instrumentation and co_monitoring Unexpectedly Set

Bug report

Bug description:

I ran into the following problematic case with gh-133221 (which had to be reverted):

import sys
import _testinternalcapi

def foo():
    pass

def trace_temporarily():
    sys._getframe().f_trace_opcodes = True
    sys.settrace(print)
    sys.settrace(None)
    sys._getframe().f_trace_opcodes = False
trace_temporarily()

_testinternalcapi.verify_stateless_code(foo.__code__)
foo()
_testinternalcapi.verify_stateless_code(foo.__code__)  # fails

(See https://github.com/python/cpython/pull/133221#issuecomment-2853467104.)

I would expect co_instrumentation and co_monitoring to have remained NULL.

CC @markshannon

CPython versions tested on:

CPython main branch

Operating systems tested on:

No response

Reproduisez l’exemple fourni sur CPython main branch, en vous concentrant sur sys.settrace, f_trace_opcodes et _testinternalcapi.verify_stateless_code. Suivez l’évolution de l’état de co_instrumentation et co_monitoring pendant le tracing temporaire, et confirmez le fix en vous assurant qu’ils restent tous deux NULL et que la vérification réussit après l’exécution de foo().