http://bugs.winehq.org/show_bug.cgi?id=20919
--- Comment #5 from Jörg Höhle hoehle@users.sourceforge.net 2010-05-19 11:14:04 --- winmm/wave.ok does not hang either, although it appears to do. I let it run for several hours at max CPU and it eventually run to completion.
Thanks to Dan's env patch to kernel32/process.c[*], I could launch the test with WINETEST_REPORT_SUCCESS=1 and see that it actually did something: almost continuously, the last line printed before a pause was
... xyz succeeded wave.c:621: Test succeeded wait ~5 seconds, then proceed with wave.c:660: Test succeeded ... more successes until running into line 621 again line 660 is WaveOutOpen. Something seems to cause Wine to spend a lot of time inside WaveOutOpen when running under Valgrind - perhaps all the ACM "try and find a driver for that format" machinery?
[*] isn't there a bug with that patch attached somewhere? I don't mean bug #14359, comment #0.