http://bugs.winehq.org/show_bug.cgi?id=49055 --- Comment #5 from Stian Low <wineryyyyy@gmail.com> --- (In reply to realmaitreal from comment #4)
So CSMT is not the variable -- it fails with csmt on and off. The renderer is: GL fails deterministically, Vulkan is clean. On GL it is the same four cases every time, the ones where the render target is an additional swapchain's back buffer and a Present has happened.
gl failing vs vulkan seems to be the opposite of d3d10core.c/d3d11.c tests cases I recently reported for MR-10567: - https://gitlab.winehq.org/wine/wine/-/merge_requests/10567#note_148492 Sorry if gitlab struggles to load such a bloated mess of a merge. I tried suggesting splitting but it is what it is now. Here's the relevant comments from MR-10567 note_148492:
Also the reason for vulkan tests failing has been determined: vk_blitter_clear_rendertargets delays blitting for ClearRenderTargetView which may cause some order of operations discrepancy which leads to srgb correction not being applied properly. I'll have to take a closer look how to properly fix it.
Forcing attachment_count > 0 for vk_blitter_clear_rendertargets forces immediate blitting and allows all vulkan tests to pass for srgb ClearRenderTargetView.
Seems similar to deterministic bug you describe which also happens regardless of CSMT but I've only encountered for vulkan but not gl so far which seems to handled blitting immediately but I'll double check if it also may delay blitting under some circumstances. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.