gh-120326: Include intrin.h on Windows

[Eclips4]

• Issue: Python 3.13 beta 2 build fails on Windows when using both --experimental-jit and --disable-gil #120326

[Eclips4]

@cdgriffith can you confirm that this resolves the issue?

[cdgriffith]

@Eclips4 The build completes, and produces a 3.14 executable (assuming this will also be merged into 3.13).

Thank you for fast fix!

[vstinner]

LGTM, I just left minor suggestions.

[Eclips4]

Thanks for the review, Victor.

[vstinner]

LGTM

[vstinner]

Oh wait, now I recall vaguely a C++20 module issue if the include is done inside a extern "C" { ... }. Would it be possible to move the include to Python.h, and add a comment explaining why it's needed, mention at least one function which is used?

Also, is this include only needed for Free Threading, or for JIT compiler, or both?

[vstinner]

Thanks, I enabled auto-merge.

For example, I see that _Py_ThreadId() uses:

#if defined(_MSC_VER) && defined(_M_X64) tid = __readgsqword(48); 

[Eclips4]

@vstinner I've disabled automerge due to failure of "Tests / Windows (free-threading) / build and test (x86) (pull_request)" job (see https://github.com/python/cpython/actions/runs/9468440435/attempts/1?pr=120329) . I've re-run it and seems it will pass (UPD: Yes, it is).

I don't know if it is related to this pull request, but it is definitely something strange, and I am not sure we can merge it now. However, I cannot reproduce the failure from the job locally on my Windows setup :(

[miss-islington-app]

Thanks @Eclips4 for the PR, and @vstinner for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13.
🐍🍒⛏🤖

[bedevere-app]

GH-120363 is a backport of this pull request to the 3.13 branch.

[vstinner]

@vstinner I've disabled automerge due to failure of "Tests / Windows (free-threading) / build and test (x86) (pull_request)" job (see https://github.com/python/cpython/actions/runs/9468440435/attempts/1?pr=120329) . I've re-run it and seems it will pass (UPD: Yes, it is).

I don't see how an include can trigger this issue. It's strange since the job passed when re-run.

[vstinner]

Oh, the x86 job failed again on the commit checks: https://github.com/python/cpython/actions/runs/9470417218/job/26091279722

Run .\python.bat -m test.pythoninfo Running Debug|Win32 interpreter... Error: Process completed with exit code 1. 

[vstinner]

I built Python with Free Threading on Windows for x86: I failed to reproduce the issue.

I ran PCbuild\build.bat -e -d -p x86 --disable-gil to build Python, and then I ran python.bat -m test.pythoninfo.

[vstinner]

Windows x86 Free Threading failed 2x in a row, and then succeeded 5x in a row. This bug is strange. I don't know how to debug it since I cannot reproduce it locally.

[miss-islington-app]

Thanks @Eclips4 for the PR, and @vstinner for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13.
🐍🍒⛏🤖

[bedevere-app]

GH-120414 is a backport of this pull request to the 3.13 branch.