Kernel: (drm-intel-nightly) git-ec45c7
Bug detailed description:
It’s not a regression, firstly test ogl-samples.
- build: https://github.com/g-truc/ogl-samples
It’s core dump.
-"OpenGL Error(GL_INVALID_OPERATION): initTexture"
-ogl-samples/framework/test.cpp:412: bool test::checkError(const char*) const: Assertion `0' failed.
Another test case: Gl-330-texture-rect too.
The program calls glGenerateMipmaps on a RGBA8UI texture, which Anuj disallowed in:
Author: Anuj Phogat <email@example.com>
Date: Fri Nov 2 10:47:33 2012 -0700
mesa: Generate invalid operation in glGenerateMipMap for integer textures
Khronos has reached a conclusion and disallowed following texture formats in
(a) ASTC textures
(b) integer internal formats (e.g., RGBA8UI, RG16I)
(c) textures with stencil formats (e.g., STENCIL_INDEX8)
(d) textures with packed depth/stencil formats (e.g, DEPTH24_STENCIL8)
Note: This is a candidate for stable branches.
Signed-off-by: Anuj Phogat <firstname.lastname@example.org>
Reviewed-by: Brian Paul <email@example.com>
After that, the program asserts that there is no error...and dies.
Reading the Khronos bug, it sounds like the final resolution was to do what Anuj implemented. However, I see zero text to that effect in the OpenGL 4.3 or even 4.4 specs. It sounds like they instead adopted a "it ought to work" resolution.
Reverting Anuj's commit fixes this demo. Ian, do you think we should clarify with Khronos, or how would you like to proceed?
(In reply to comment #1)
> Another test case: Gl-330-texture-rect too.
That's an unrelated failure. I've split that issue out into a new bug, #76496.
This bug remains about gl-320-texture-integer.
Mesa: User error: GL_INVALID_OPERATION in glGenerateMipmap(invalid internal format GL_RGBA8UI)
I've made a comment in the Khronos bug about this. Looks like it was forgotten.