gh-145446: Add critical section in functools module for PyDict_Next

[bkap123]

Added a critical section whenever PyDict_Next is called in Module/_functoolsmodule.c in the free-threaded build.

Additionally, I had to manually remove the lock created by a critical section in case of an early return due to an error.

• Issue: Missing critical sections for PyDict_Next calls in _functoolsmodule.c #145446

[vstinner]

LGTM

[colesbury]

You don't need to lock keyword arguments dictionaries. They're not exposed to other threads and cannot be modified concurrently.

This is true even if you splat a dict: foo(**mydict). The called function gets a copy of mydict.

[bedevere-app]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

[bkap123]

I have made the requested changes; please review again

[bedevere-app]

Thanks for making the requested changes!

@vstinner, @colesbury: please review the changes made to this pull request.

[colesbury]

I don't think the remaining critical sections are needed either.

[bkap123]

Okay, after thinking about it I see what you mean. I can close the PR if the critical sections are not needed.

[bkap123]

@colesbury After further reflection, I actually think that you need the critical section in the remaining cases. For example, without the critical section, this code results in a data race:

from functools import partial 
from threading import Thread

p = partial(lambda: None)

def f():
    for _ in range(100):
        repr(p)

def g():
    for i in range(100):
        p.keywords[f"{i}"] = i


t1 = Thread(target=f)
t2 = Thread(target=g)

t1.start()
t2.start()

t1.join()
t2.join()

I attached the full TSan report here although it is quite long and I don't think it is that illuminating.

However, the only remaining cases are for the kw entry of the partial object. As such, a better solution might be to transition this kw dict to a frozendict (see issue #145478).

Let me know if you think it is worth continuing this PR or if we should close it.