GitLab

This better matches the interface of glGetStringi, which should be used
for GL >= 3.0.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>

Fix incorrect texture height (16, not 15).
Fix incorrect width, height for glCopyTexSubImage2D() test.
Add two new SubImage tests for non-existant mipmap levels.

Check for expected OpenGL errors related to texture compression.  For
example, glCompresedTexSubImage2D() offset/size values must be multiples
of four.

This test could be expanded to cover other compression formats in the future.

Also, this test contains some texture compression helper functions that could
be made into piglit utility functions.

Verified with NVIDIA's driver.  One error found in Mesa, to be fixed soon.

Create an image buffer with red, green, blue, white quadrants.
This code was extracted out of the piglit_rgbw_texture() function.
The piglit_rgbw_texture() function is then rewritten to use the new
piglit_rgbw_image() function.

1. It doesn't use sampler2DRect, so don't require ARB_texture_rectangle
2. Remove pointless glFinish() before piglit_present_results()
3. Use Piglit utility functions for compiling/linking shaders.
4. Print the description of what you're seeing in *non*-automatic mode.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Brian Paul <brianp@vmware.com>

Reviewed-by: Brian Paul <brianp@vmware.com>

This tests every possible DEPTH_TEXTURE_MODE with a random smattering of
EXT_texture_swizzle modes.  While it could be more exhaustive (or use a
more logically ordered layout), it does catch the bugs I wanted to
detect.

v2: Fix probing bug, minor cleanups based on Brian's feedback.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>

This was originally generated by a script until I realised the script
was actually nearly just the contents of the files (there really is
that little overlap), so drop the script and just push the files.

This should test (non-textureSize ones)

      gvec4 texture(gsamplerCubeArray sampler, vec4 coord
                    [, float bias])
      gvec4 textureLod(gsamplerCubeArray sampler, vec4 coord, float lod)
      float texture(samplerCubeArrayShadow sampler, vec4 coord, float compare)
      gvec4 textureGrad(gsamplerCubeArray sampler, vec4 coord,
                        vec3 ddx, vec3 ddy);

work at the compiler level.
Signed-off-by: Dave Airlie <airlied@redhat.com>

Fix const correctness warnings.  Remove unused label.  Use program name
in error messages.
Signed-off-by: Dave Airlie <airlied@gmail.com>

Signed-off-by: Dave Airlie <airlied@redhat.com>

This tests the glFramebufferTextureLayer interface.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This tests cubemap array shadow support, these are different as we have
to pass the compare value not in the texture coord but in a separate attribute.

Tested against NVIDIA 295 series.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

test the getintegerv addition from the spec.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This tests teximage3d returns invalid values for width != height
and depth not divisible by 6.
Signed-off-by: Dave Airlie <airlied@redhat.com>

This creates a cube map array of 2 cubes, and tests the upload via
glTexImage3D and rendering of each layer happens correctly.

note it needs to redefine the common cubemap stuff as they are in the
wrong order for cubemap arrays.

v2: fixup Brian's review
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This one is wierd since you give glTexImage3D layers * 6, but
textureSize reports back layers on its own.

Tested and passes on nvidia 295.59.
Signed-off-by: Dave Airlie <airlied@redhat.com>

This reflects the order they are requires for ARB_texture_cube_map_array,

I've tested that this causes no regressions here.
Signed-off-by: Dave Airlie <airlied@redhat.com>

Fixes Solaris Studio build.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50802Signed-off-by: Vinson Lee <vlee@freedesktop.org>

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

Should fix shader_runner build failure.

This introduces OpenCL testing to piglit under tests/all_cl.tests.

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>

Reviewed-by: Brian Paul <brianp@vmware.com>

The version number is taken from the GLSL version requirement, if there is one.

This is part of the effort to make version handling more flexible for GLES.

v2 (Paul Berry <stereotype441@gmail.com>): use strstr() to check
whether a #version directive is already present, in case it isn't at
the start of the shader.
Reviewed-by: Paul Berry <stereotype441@gmail.com>

Previously, the test made the erroneous assumption that if the size of
a texture is 2^n, then that texture has n miplevels.  This was
wrong--the texture has n+1 miplevels, ranging in size from 2^n to 2^0.
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>

Reviewed-by: Brian Paul <brianp@vmware.com>

It wasn't doing anything.
Tested-by: Andreas Boll <andreas.boll.dev@gmail.com>

It takes 10 minutes for llvmpipe to finish this test. With this patch,
it takes 20 seconds.

Before:

Tested 104 GL_REPLACE combinations
Tested 7540 GL_ADD combinations
Tested 7540 GL_ADD_SIGNED_EXT combinations
Tested 7540 GL_MODULATE combinations
Tested 5957 GL_INTERPOLATE_EXT combinations
Tested 7540 GL_DOT3_RGB_EXT combinations
Tested 7540 GL_DOT3_RGBA_EXT combinations
Tested 47662 GL_MODULATE_ADD_ATI combinations
Tested 47662 GL_MODULATE_SIGNED_ADD_ATI combinations
Tested 47662 GL_MODULATE_SUBTRACT_ATI combinations
Tested 218 multitexture combinations
Tested 8 crossbar combinations

After:

Tested 17 GL_REPLACE combinations
Tested 1237 GL_ADD combinations
Tested 1237 GL_ADD_SIGNED_EXT combinations
Tested 1237 GL_MODULATE combinations
Tested 978 GL_INTERPOLATE_EXT combinations
Tested 1237 GL_DOT3_RGB_EXT combinations
Tested 1237 GL_DOT3_RGBA_EXT combinations
Tested 7825 GL_MODULATE_ADD_ATI combinations
Tested 7825 GL_MODULATE_SIGNED_ADD_ATI combinations
Tested 7825 GL_MODULATE_SUBTRACT_ATI combinations
Tested 35 multitexture combinations
Tested 8 crossbar combinations
Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Andreas Boll <andreas.boll.dev@gmail.com>

before with r600g:

real    0m24.506s
user    0m22.393s
sys     0m1.812s

after with r600g

real    0m0.747s
user    0m0.616s
sys     0m0.100s

before with llvmpipe

real    4m28.217s
user    4m27.061s
sys     0m1.020s

after with llvmpipe

real    0m7.358s
user    0m7.224s
sys     0m0.112s

Check that all the common unsized internal formats are illegal.
And check that some common sized formats are accepted.

Per issue 17 of the extension spec, unsized internalFormat values are
not allowed.  Change GL_RGBA to GL_RGBA8 and add some extra error
checks.

The test passes on NVIDIA now.

We might not have enough memory for a texture that large.
Use width or height = 1 instead.
Reviewed-by: José Fonseca <jfonseca@vmware.com>

The ARB_explicit_attrib_location tests layout-12.frag and layout-13.frag
were incorrect: they both used index == 2 (which is invalid) and
expected the compile to succeed.  Although the extension spec doesn't
state anything about generating an error on indices other than 0 or 1,
the GLSL 4.30 spec clarifies this as a compile error.  This matches the
behavior of the API call defined in ARB_blend_func_extended,
glBindFragDataLocationIndexed, which generates INVALID_VALUE for indices
other than 0 or 1.

This patch changes layout-13.frag to use index = 1 so it will actually
test the intended behavior.

layout-12.frag was just bogus: the comment said "Even though the
specified locations overlap [...]" but the test only had one variable.
This patch turns it into a negative test to ensure index == 2 is a
compile error.

This patch also adds layout-14.frag, a test to check that index < 0 is
also a compile error.