April 9, 2026
11:39 a.m.
On Tue Apr 7 17:33:31 2026 +0000, Matteo Bruni wrote:
It doesn't look like mip levels should be checked here. I tried to tweak the new test from the previous commit to use a different mip count and it still passed. I only tested d3dx10_43 32-bit; hopefully the behavior isn't version or bitness-specific. Can you give me a diff to test? Any changes I make to this test like:
diff --git a/dlls/d3dx10_43/tests/d3dx10.c b/dlls/d3dx10_43/tests/d3dx10.c
index 5a22240d37b..dd8b5b848b6 100644
--- a/dlls/d3dx10_43/tests/d3dx10.c
+++ b/dlls/d3dx10_43/tests/d3dx10.c
@@ -1852,7 +1852,7 @@ static const struct test_image_load_info
{
dds_dxt10_4_4, sizeof(dds_dxt10_4_4),
{
- D3DX10_DEFAULT, D3DX10_DEFAULT, D3DX10_DEFAULT, D3DX10_DEFAULT, 1, (D3D10_USAGE)D3DX10_DEFAULT,
+ D3DX10_DEFAULT, D3DX10_DEFAULT, D3DX10_DEFAULT, D3DX10_DEFAULT, 2, (D3D10_USAGE)D3DX10_DEFAULT,
D3DX10_DEFAULT, D3DX10_DEFAULT, D3DX10_DEFAULT, D3DX10_DEFAULT, 7, D3DX10_DEFAULT
},
S_OK, D3D10_RESOURCE_DIMENSION_TEXTURE2D,
Results in failure on native, both 32 and 64 bit tests on both d3dx10_43 and d3dx10_33. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10513#note_135710