Sporadic failure of P0660R10_jthread_and_cv_any

[StephanTLavavej]

After updating to Python 3.9, I'm seeing sporadic failures in P0660R10_jthread_and_cv_any. I can repro them with x64 (haven't tried x86 yet) and they don't seem to be specific to any particular configuration. They're pretty common and both cl and clang-cl can fail. They're always asserting on this line:

STL/tests/std/tests/P0660R10_jthread_and_cv_any/test.cpp

Line 219 in e7d5113

assert(cv.wait_for(lck, move(token), forever, [&] { return b; }) == true);

[CaseyCarter]

Passing move(token) to cv.wait_for twice:

STL/tests/std/tests/P0660R10_jthread_and_cv_any/test.cpp

Lines 182 to 183 in e7d5113

	assert(cv.wait(lck, move(token), [] { return true; }) == true);
	assert(cv.wait(lck, move(token), [&] { return b; }) == true);

STL/tests/std/tests/P0660R10_jthread_and_cv_any/test.cpp

Lines 200 to 201 in e7d5113

	assert(cv.wait_until(lck, move(token), infinity, [] { return true; }) == true);
	assert(cv.wait_until(lck, move(token), infinity, [&] { return b; }) == true);

STL/tests/std/tests/P0660R10_jthread_and_cv_any/test.cpp

Lines 218 to 219 in e7d5113

	assert(cv.wait_for(lck, move(token), forever, [] { return true; }) == true);
	assert(cv.wait_for(lck, move(token), forever, [&] { return b; }) == true);

is not a bug, but it's smelly enough to deserve a comment.

[CaseyCarter]

Does master...CaseyCarter:absolute_time fix it?

(FWIW: I've been using python 3.9 for about a week without issue, but I almost never perform full test runs locally.)

[StephanTLavavej]

@CaseyCarter Yes! I ran x86 and x64 10 times each, and they all passed.

(My machine seems to be exceptionally good at finding sporadic multithreading bugs.)