It appears that our Piglit tests that hit glGenerateMipmap always create the cube map texture with Texture Storage, so Num Layers always gets set to 6. I will make a simple test this afternoon for non texture storage cube maps. On Mar 12, 2015 12:33 PM, "Laura Ekstrand" <la...@jlekstrand.net> wrote:
> This reverts commit 1ee000a0b6737d6c140d4f07b6044908b8ebfdc7. > Failures with the GLES3 conformance suite and Synmark2 OGLHdrBloom revealed > that this commit might be in error. A look at the offended test in GLES3 > conformance suite, NPOT gen mipmap, suggests that NumLayers may not > actually > always be 6 for a cube complete texture with target = GL_TEXTURE_CUBE_MAP. > > Extensive testing with Piglit prior to patch review and upstreaming did not > reveal this problem because, in the few Piglit tests that test for cube > completeness, NumLayers = 6. So it appears that perhaps existing test > coverage in Piglit is inadequate. > --- > src/mesa/main/texobj.c | 4 ---- > 1 file changed, 4 deletions(-) > > diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c > index a99b108..e018ab9 100644 > --- a/src/mesa/main/texobj.c > +++ b/src/mesa/main/texobj.c > @@ -879,10 +879,6 @@ _mesa_cube_level_complete(const struct > gl_texture_object *texObj, > if (texObj->Target != GL_TEXTURE_CUBE_MAP) > return GL_FALSE; > > - /* Make sure we have enough image planes for a cube map. */ > - if (texObj->NumLayers < 6) > - return GL_FALSE; > - > if ((level < 0) || (level >= MAX_TEXTURE_LEVELS)) > return GL_FALSE; > > -- > 2.1.0 > >
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
