GH-130727: test_wmi_query_error test is flaky

[chris-eibl]

I can reproduce the the "invalid handle" errors which #130727 (comment) under high load (using "Stress CPU" of CPU-Z) running

python -m test test_wmi -m test.test_wmi.WmiTests.test_wmi_query_repeated -F

The reason is, that the main thread calls CloseHandle on the still running thread in case of a WAIT_TIMEOUT, and then the thread dies with

Exception thrown at 0x00007FFE1A35144A (ntdll.dll) in python_d.exe: 0xC0000008: An invalid handle was specified.

and call stack

 	ntdll.dll!00007ffe1a35144a()	Unknown
 	KernelBase.dll!00007ffe17cd0905()	Unknown
>	_wmi_d.pyd!_query_thread(void * param) Line 204	C++

where

    CloseHandle(data->writePipe); /Line 204
    return (DWORD)hr;

when I debug it in Visual Studio.

This has already been anticipated by #125315 (comment).

Another manifestation of this race results in

======================================================================
FAIL: test_wmi_query_repeated (test.test_wmi.WmiTests.test_wmi_query_repeated)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "R:\buildarea\3.14.ambv-bb-win11.bigmem\build\Lib\test\test_wmi.py", line 41, in test_wmi_query_repeated
    self.test_wmi_query_os_version()
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
  File "R:\buildarea\3.14.ambv-bb-win11.bigmem\build\Lib\test\test_wmi.py", line 30, in test_wmi_query_os_version
    self.assertEqual(1, len(r))
    ~~~~~~~~~~~~~~~~^^^^^^^^^^^
AssertionError: 1 != 2

which sometimes occur on the build bots, e.g. https://buildbot.python.org/#/builders/1799/builds/20.

Copying the _query_data struct in the wmi thread fixes both issues.

• Issue: [Windows] test_wmi: test_wmi_query_error test is flaky #130727

[chris-eibl]

#125315, I went with the former. In that case, I think it doesn't need a news entry, since it is just fixing a flaky test, but I'd happily blurb it if it deems appropriate.

[zooba]

@chris-eibl Yeah, give it a NEWS entry, since it's a change to the runtime rather than the tests, and so users might see it out in the wild and wonder about the changed behaviour.

Before merging, I'd want to confirm that there's no intentional modification of that struct that is supposed to be picked up by the other thread. I suspect not, but I haven't read the code to double-check.

[miss-islington-app]

Thanks @zooba for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13.
🐍🍒⛏🤖

[miss-islington-app]

Thanks @zooba for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14.
🐍🍒⛏🤖

[bedevere-app]

3.13 branch.

[bedevere-app]

3.14 branch.