http://bugs.winehq.org/show_bug.cgi?id=26098
--- Comment #1 from Juan Lang juan_lang@yahoo.com 2011-02-14 17:05:14 CST --- (In reply to comment #0)
Invalid read of size 4 at CryptHashData (crypt.c:1753) by test_hashes (rsaenh.c:571)
This is expected. The provider in which the hash is created was destroyed before the hash, invalidating the hash.
Invalid read of size 4 at CryptDestroyHash (crypt.c:884) by test_hashes (rsaenh.c:577)
Likewise with this one.
12 bytes in 1 blocks are definitely lost 368 bytes in 1 blocks are definitely lost
The leaks are likewise related.
It could be that a better reference counting mechanism could be devised to avoid such errors, but it's not clear that apps would benefit as a result. I think a suppression might well be the appropriate response here.