http://bugs.winehq.org/show_bug.cgi?id=18900
--- Comment #28 from Alexandre Julliard julliard@winehq.org 2009-06-15 03:21:49 --- (In reply to comment #27)
I'm fascinated to know: what was the linkage between the commit my bisect turned up and the underlying bug that's fixed by the patch? The commit seems to just explicitly add some alignment fields in a structure; I'm baffled as to how this could lead to duplicates in the object wait list!
It doesn't, the duplicates have always been there. Since we are accessing invalid data, the crash may behave differently because of changes in the memory layout, but the bug was already present before that commit.