https://bugs.winehq.org/show_bug.cgi?id=39794
--- Comment #3 from Nikolay Sivov bunglehead@gmail.com --- The idea is correct, patch itself is wrong though, and mixes two separate issues. Second part is handled by drivers already, except for last error, which we don't necessary want to be set.
Also I agree, there's no reason to split it from bug 33140.
I just sent a test for initial problem https://www.winehq.org/pipermail/wine-patches/2015-December/145178.html.