https://bugs.winehq.org/show_bug.cgi?id=46099
--- Comment #17 from Greg Smith codedonewell@gmail.com --- Created attachment 62875 --> https://bugs.winehq.org/attachment.cgi?id=62875 Test to highlight the bug
This test starts a group of "load" threads (CPU count * 4) whose purpose is only to burden the CPU - theoretically making the crucial race condition more likely. It then starts 2 threads that wait and wake (all) on the same address. The only difference is that the second thread skips the first wait. The test fails if any wait takes longer than 10 seconds to be notified. Purer versions of this test mean bigger wait times. The failure occurs quite quickly on my 2 core 4 thread processor, although if you have a high thread to core ratio, the test may not fail due to insufficient load.