http://bugs.winehq.org/show_bug.cgi?id=15323
--- Comment #11 from Paul "TBBle" Hampson Paul.Hampson@Pobox.com 2009-03-01 00:13:59 --- Created an attachment (id=19723) --> (http://bugs.winehq.org/attachment.cgi?id=19723) Demonstration of failure case in the kernel32 Wine unit test
I've attached a patch against current git HEAD which demonstrates the failure case of warpatch.bin. (It replaces the 'thread' test in kernel32 with three tests plus one in the START_TEST routine, and the relevant test is "Thread ran before creator did".)
On Windows (only tested on WinXP) it always succeeds. Under Wine on my machine, it generally fails if wineserver is not running, and generally succeeds if wineserver is running. But not always, in both cases.
Anyway, I'm going to post this patch to wine-devel for discussion, but as far as I can tell there's nothing documented that specifies either for or against this behaviour, but this is the bug in a nutshell.