https://bugs.winehq.org/show_bug.cgi?id=50549
--- Comment #3 from Henri Verbeet hverbeet@gmail.com --- Created attachment 69227 --> https://bugs.winehq.org/attachment.cgi?id=69227 patch
The attached patch seems like the most straightforward solution, does that work? However, note that I didn't look much into the original issue the relevant commits were supposed to solve, so I might be missing some obvious reason for why this approach wasn't taken in the first place.
I don't believe either d9c9b5ea614780340708cf18423a10704d8587e1 or 2d4dd4252b0cf6526b3cc8194cce642b16eb12f6 introduced any new tests, but the attached patch doesn't introduce any failures in the vulkan-1 tests at least.
What I find most surprising about this bug though, is that it wasn't caught by the tests. The wined3d Vulkan renderer is not enabled by default, but I can also trivially reproduce this issue with the d3d12 tests, and the d3d12 part of the dxgi tests. Seemingly that's not the case for any of the people who signed off on this patch, and it might be worthwhile to figure out why that is.