| commit | da574ed9f320cd8d768096931f5d51ac7e984242 | |
| author | Henri Verbeet <hverbeet@codeweavers.com> | |
| Thu, 14 Sep 2023 14:29:48 +0000 (14 16:29 +0200) | ||
| committer | Alexandre Julliard <julliard@winehq.org> | |
| Wed, 20 Sep 2023 20:15:29 +0000 (20 22:15 +0200) | ||
| tree | a70929f3a348dddc72c91590bc4f302c811a3a6f | treesnapshot (tar.gz zip) |
| parent | fa4be54777517c93c6058cc23efdb63fbca37b35 | commitdiff |
wined3d: Don't validate the frontbuffer's DRAWABLE location in wined3d_swapchain_resize_buffers() when NO3D is set.
In general, NO3D surfaces are always in SYSMEM. More importantly though,
validating DRAWABLE will evict SYSMEM, but a subsequent
wined3d_texture_no3d_load_location() won't recreate it because it's always
supposed to be current. This fixes a regression introduced by commit
ebaa0a9426864cfa6e36955f26ff4c66c1d5af76, and exposed by running the ddraw
tests with the "no3d" renderer.
In general, NO3D surfaces are always in SYSMEM. More importantly though,
validating DRAWABLE will evict SYSMEM, but a subsequent
wined3d_texture_no3d_load_location() won't recreate it because it's always
supposed to be current. This fixes a regression introduced by commit
ebaa0a9426864cfa6e36955f26ff4c66c1d5af76, and exposed by running the ddraw
tests with the "no3d" renderer.
| dlls/wined3d/swapchain.c | diffblobblamehistory |