http://bugs.winehq.org/show_bug.cgi?id=33307
--- Comment #25 from Kirill K. Smirnov lich@math.spbu.ru 2013-04-22 22:22:16 CDT --- This code works too: 964: memcpy_unaligned( buffer, DIR_Windows, len * sizeof(WCHAR) ); 965: memcpy( buffer + len, default_syswow64W, sizeof(default_syswow64W) ); 966: DIR_SysWow64 = buffer;
It seems gcc incorrectly optimizes bulk of memcpy() function calls one after another, not just single memcpy().