homepage

The ffi_prep_args function in libffi_msvc/ffi.c needs the attached patch to handle 64-bit parameters with the correct padding.
Without this patch, garbage may appear in the top part of 64-bit arguments as the values are not zeroed out by the memcpy.
I'm not 100% sure who 'owns' this file - is there an upstream project that should get the patch rather than Python? If not, I see no reason not to check this in now, even though it doesn't seem to repro with the VC10 builds.

Nosying the other Windows guys - seems like the ctypes nosy list is inactive, and this only affects MSVC.

This patch also resolves this test issue:
[ 68/390/2] test_ttk_textonly
test test_ttk_textonly crashed -- Traceback (most recent call last):
 File "D:\...\lib\test\regrtest.py", line 1280, in runtest_inner
 test_runner()
 File "D:\...\lib\test\test_ttk_textonly.py", line 14, in test_main
 *runtktests.get_tests(gui=False, packages=['test_ttk']))
 File "D:\...\lib\tkinter\test\runtktests.py", line 65, in get_tests
 for module in get_tests_modules(gui=gui, packages=packages):
 File "D:\...\lib\tkinter\test\runtktests.py", line 50, in get_tests_modules
 "tkinter.test")
 File "D:\...\lib\importlib\__init__.py", line 109, in import_module
 return _bootstrap._gcd_import(name[level:], package, level)
 File "<frozen importlib._bootstrap>", line 2208, in _gcd_import
 File "<frozen importlib._bootstrap>", line 2191, in _find_and_load
 File "<frozen importlib._bootstrap>", line 2180, in _find_and_load_unlocked
 File "<frozen importlib._bootstrap>", line 1149, in _load_unlocked
 File "<frozen importlib._bootstrap>", line 1420, in exec_module
 File "<frozen importlib._bootstrap>", line 321, in _call_with_frames_removed
 File "D:\...\lib\tkinter\test\test_ttk\test_extensions.py", line 8, in <module>
 requires('gui')
 File "D:\...\lib\test\support\__init__.py", line 492, in requires
 if resource == 'gui' and not _is_gui_available():
 File "D:\...\lib\test\support\__init__.py", line 436, in _is_gui_available
 raise ctypes.WinError()
OSError: [WinError 0] The operation completed successfully.
I'm not entirely sure when/if I should just commit the fix without any reviews, but this issue needs to be fixed, so eventually I'll probably just do it :)

I don't know near enough to comment on this one. If you feel confident in it, I'd say go ahead.

Likewise.

New changeset 4c81b1846e14 by Steve Dower in branch 'default':
Issue #22733: MSVC ffi_prep_args doesn't handle 64-bit arguments properly
https://hg.python.org/cpython/rev/4c81b1846e14 

New changeset 24f4569a1308 by Steve Dower in branch 'default':
Issue #22733: Added NEWS item
https://hg.python.org/cpython/rev/24f4569a1308 

Maybe worth backporting to 3.3 and up? Especially if #20160 is to be backported as they seem to be related.