GitLab

Add the following tests:
    hiz-stencil-read-fbo-d0-s8.c
    hiz-stencil-read-fbo-d24-s8.c
    hiz-stencil-read-fbo-d24s8.c
    hiz-stencil-read-window-depth0.c
    hiz-stencil-read-window-depth1.c

These tests check that stencil reads work correctly when rendering to
various framebuffer configurations, including FBO's and window
framebuffers.
Signed-off-by: Chad Versace <chad.versace@intel.com>

Add functions:
    piglit_probe_pixel_stencil
    piglit_probe_rect_stencil
Signed-off-by: Chad Versace <chad.versace@intel.com>

Add the following tests:
    hiz-stencil-test-fbo-d0-s8
    hiz-stencil-test-fbo-d24-s8
    hiz-stencil-test-fbo-d24s8
    hiz-stencil-test-window-depth0
    hiz-stencil-test-window-depth1

These tests check that stencil test works correctly when rendering to
various framebuffer configurations, including FBO's and window
framebuffers.
Signed-off-by: Chad Versace <chad.versace@intel.com>

to match OpenGL 4.0 Compatiblity, page 449:

If the value of FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE is NONE, no
framebuffer is bound to target. In this case querying pname FRAMEBUFFER_-
ATTACHMENT_OBJECT_NAME will return zero, and all other queries will generate
an INVALID_OPERATION error.

Reviewed-by: Chad Versace <chad@chad-versace.us>

Add config_attribs field to egl_test so that tests have to pass
their wanted attributes for the EGL config selection. Existing
tests were modified to pass the attributes.
Reviewed-by: Chad Versace <chad@chad-versace.us>

Tapani Pälli authored May 26, 2011 and Brian Paul committed May 26, 2011

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

Tapani Pälli authored May 24, 2011 and Brian Paul committed May 25, 2011

Queries the value for GL_MAX_VERTEX_ATTRIBS and uses that as index
to set a value with various functions. GL specification states that
GL_INVALID_VALUE should occur if index >= GL_MAX_VERTEX_ATTRIBS.

Refactored (no shader needed for this) and cleaned up the code + added
missing copyright header.
Signed-off-by: Brian Paul <brianp@vmware.com>

Kenneth Graunke authored Jan 23, 2011 and Emma Anholt committed May 25, 2011

The Intel team is all using the closed-source GLES2 suite internally
for additional regression testing beyond what piglit provides.  This
integration gets our test suites together so we get better coverage in
our daily testing.  The list of tests is computed automatically from
the tests present in the tree surrounding the GTF binary pointed at by
./bin/GTF symlink, if present.

Reviewed-by: Chad Versace <chad@chad-versace.us>

Reviewed-by: Chad Versace <chad@chad-versace.us>

Reviewed-by: Chad Versace <chad@chad-versace.us>

Tapani Pälli authored May 24, 2011 and Brian Paul committed May 24, 2011

GL_GEOMETRY_SHADER is only defined if USE_OPENGL is defined
and GL/glext.h is included (not defined in GLES2/gl2ext.h).
Signed-off-by: Brian Paul <brianp@vmware.com>

When reading in test run results we were mistakenly not decoding
escape sequences inside arrays.  This resulted in occasional extra
backslashes in lists such as "errors" and "errors_ignored".
Reviewed-by: Chad Versace <chad@chad-versace.us>

Less worrying console output when it passes.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

This may help distinguish an un-loaded varying read from just reading
the wrong varying.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

I was confused by the undefined garbage showing up.  Also, in my
failure mode, some pixels would come out OK.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

Exercises an issue reloading LLVM enabled DRI drivers on Fedora 15.
Reviewed-by: Brian Paul <brianp@vmware.com>

X11 is already linked above for all targets.

or when piglit_link_simple_program() fails.  Callers can then check
the return value to see if compiling/linking worked.

There are no overloads of outerProduct that take any of these types as
parameters.  Currently the bvec tests compile (thereby failing) on
Mesa.

The previous batch tested non-constant parameters to outerProduct.
All of those shaders should execute the outerProduct code on the GPU.
This batch adds the same set of tests but using constant parameters to
outerProduct.  These are likely executed in the compiler on the CPU
during a constant folding pass.

Testing all of the possible paths seems to make sense.

There are implicit conversions from ivec to vec, so these are supposed
to work just like the vec tests.

... by splitting it into separate probe functions for the color and depth
buffer (hiz_probe_color_buffer() and hiz_probe_depth_buffer()).  By
separating the probe functions, we can write tests that focus on just one
feature.

Also, make the probe functions more general by adding a function parameter
that specifies the expected probe values.
Signed-off-by: Chad Versace <chad.versace@intel.com>

This function will soon cease to be useful, since more complex hiz tests
will each have their own drawing routine.

The function body has been copied into hiz_run_test_depth_test_common(),
which contains the common functionality needed by
hiz_run_test_depth_test_{fbo,window}().
Signed-off-by: Chad Versace <chad.versace@intel.com>

... in order to impose naming consistency with
hiz_run_test_depth_test_window() and future test functions.
Signed-off-by: Chad Versace <chad.versace@intel.com>