I'm not sure if it's the best way of fixing the bug. I was trying to set compilation options to workaround the problem but it didn't work well (probably due to my limited knowledge in this area).
The patch was originally sent in !5276.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/6668
This is my last set of patches changing DDS pixel format handling.
--
v2: d3dx9: Add support for more DDS pixel formats.
d3dx9: Rework conversion to/from D3DFORMAT from/to DDS pixel format.
d3dx9: Validate the size of DDS files containing packed pixel formats.
d3dx9/tests: Add file size validation tests for DDS files containing packed pixel formats.
https://gitlab.winehq.org/wine/wine/-/merge_requests/6653
Making all the windows handled by the driver to be either GL/VK client surfaces, or top-level windows. This avoids leaking host windows into the Win32 space, and makes it possible to get rid of some remaining NtUserMapWindowPoints in `map_event_coords` for mouse input.
--
v7: winex11: Get rid of the now unnecessary foreign windows.
winex11: Generate relative ConfigureNotify on parent ConfigureNotify events.
winex11: Use the new host windows to register foreign window events.
winex11: Keep track of the host windows children window rects.
winex11: Keep track of the host windows relative rects.
winex11: Keep track of the host window children of interest.
winex11: Create host windows recursively up to root_window.
winex11: Introduce a new struct host_window for host-only windows.
https://gitlab.winehq.org/wine/wine/-/merge_requests/6649