http://bugs.winehq.org/show_bug.cgi?id=11188
--- Comment #122 from Claudio sick_soul@yahoo.it 2009-01-24 17:35:32 --- (In reply to comment #119)
I was able to reproduce this error with WC3 demo and current git, I just had to play a little to get a second level. This bug is caused by race condition, it means that it only occurs sometime in more or less random manner, when threads execute in specific order in time.
I am going to test your patch heavily.
In the meantime, the reverse regression test with the 'cla9' saved game/testcase gave this result:
commit 4f74de5b366275ea522e269d29d2011a7b153e9e Author: Aleksey Bragin aleksey@reactos.org Date: Wed Dec 24 06:04:25 2008 +0400
ntdll: Fix buffer overread in RtlNumberOfSetBits.
with 4f74de5b366275ea522e269d29d2011a7b153e9e^ (the predecessor) the save attempt consistently fails. Applying 4f74de5b366275ea522e269d29d2011a7b153e9e makes the save succeed.
Repeatingly applying and reverting said commit causes the cla9 testcase to succeed and fail respectively, in a way to suggest that this change has at least an influence on this bug.
Can you put this information in correlation to the race condition?