Age | Commit message (Collapse) | Author | Files | Lines |
|
PlainExecTest but only call the __init__ function of the Test
baseclass. I haven't really figured why this is and also not really
what command I should pass to PlainExecTest.__init__, so just
replicated the init code for now.
|
|
v2: Also add all the boilerplate needed to handle the new test status
in summaries. For the color I've opted for two shades of light blue,
they nicely stick out from the current selection.
|
|
I bit gross, we probably need more fine-grained limits for
thrashing/swapping tests vs. others. But at least a start.
|
|
i-g-t tests can take a long time, and for a bunch of reasons (bad
tuning on disparate set of platforms, stuck in the kernel which often
can be recovered by interrupting with a signal, ...) that sometimes
extends to a good approximation of forever.
Hence this adds a per-test timeout value and a bit of infrastructure
to adjust the results. Test results adjusting is done after calling
interpretResult so that we don't have to replicate this all over the
place. This might need to be adjusted for the piglit-native subtest
stuff, but otoh igt is a bit special with it's crazy long-running
tests. So I think we can fix this once it's actually needed.
The default timeout is None, so this is purely opt-in.
Note on the implementation: SIG_ALARM doesn't exists on Windows and
stackoverflow overwhelmingly recommended to go with this thread-based
approach here. But only tested on my Linux box here.
|
|
I so failed at my python setup when doing this way back (less than a
year ago).
Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
|
|
An awful lot of tests rely on in-kernel checks to detect issues. So
we better make them fail properly.
Note that the dmesg check relies on dmesg timestamps to work correctly.
If you don't have them it won't work.
Cc: Paulo Zanoni <przanoni@gmail.com>
|
|
i-g-t has a bunch of testcase that cause informational output. In
addition I usually enable drm debugging on my machines, so the
kms tests cause massive amounts of spew.
I've thought whether an option would be useful, but then I didn't
really see any reason why we should ever care about debug/info
messages. Hence the default.
Cc: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
|
|
Note: as of Mesa 5cb80f0, tests "interface-vs-array-to-fs-unnamed" and
"interface-vs-unnamed-to-fs-array" are known to fail.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Several geometry shader tests were improperly added to all.tests with
names like "spec/glsl-1.50/glsl-1.50-geometry-foo" instead of
"spec/glsl-1.50/execution/geometry/foo". This patch renames the tests
to follow our usual piglit conventions.
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
|
|
And fix some warnings.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=71513
|
|
Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
|
|
Compiled regexes have to be freed with regfree.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
|
|
This patch exercises the same corner case as
https://bugs.freedesktop.org/show_bug.cgi?id=70368.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
These tests use check_link: true, which means that they each need to
specify a complete set of geometry shader layout qualifiers in order
to pass.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Validated using the NVIDIA proprietary driver for Linux (version
313.18).
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Validated using the NVIDIA proprietary driver for Linux (version
313.18).
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
When the current program has both a geometry shader and a vertex
shader, transform feedback needs to link with the geometry shader.
This test verifies that the type and size of the varyings captured by
transform feedback match their declarations in the geometry shader,
even if there are vertex shader outputs with the same names and
different types.
The test exercises one built-in shader output (gl_ClipDistance) and
one user-defined shader output.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
The test was correct, but not for the reasons stated in the comment.
The reason a sized VS output interface array should fail to match an
unsized GS input interface array is because geometry shader inputs
require an extra level of array indexing to match corresponding vertex
shader outputs.
This patch renames the test, and updates the comment, to reflect
what's actually being tested.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Note: as of Mesa 5cb80f0, test "intrastage-interface-unnamed-array" is
known to fail.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
When fast color clears are added to i965, there is a danger that they
may cause this test to stop reproducing bug 53077. To avoid that, add
a command-line option which causes the test to use a clear color that
isn't fast color clearable.
As a result of this patch, a new piglit test is added:
"spec/EXT_framebuffer_multisample/alpha-blending slow_cc". As of Mesa
commit fd05ede, this test is known to fail on Ivy Bridge and Haswell.
These failures are already covered by bug 53077.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
|
|
Tests rendering into a single framebuffer surface with multiple viewports
via a geometry shader. Scissoring is used to restrict quads to a smaller
area on the surface. Confirm that each area of the surface delineated by
a scissor rectangle for viewport index renders the correct color. Both
indexed scissor tests and indexed scissor enables are used. Geometry
shader is used to expand a single rectangle primitve to N rectangles.
Tested on Nvidia Quadro 600, test passes.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Draws quads into a single framebuffer surface with multiple viewports
via a geometry shader. Each different viewport has a different DepthRange.
The fragment shader outputs a color based on the depthRange, z, and viewport
index. Confirm that each quad has the correct color and therefore has
the correct depthRange for that viewport index.
Tested on Nvidia Quadro 600, test passes.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Tests rendering into a single framebuffer surface with multiple viewports
via a geometry shader. Confirm that each area of the surface delineated by
a viewport renders the correct color. Use multiple draw calls to replicate
geometry rather than the geometry shader.
Tested on Nvidia Quadro 600, test passes.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Tested GLenums are GL_MAX_VIEWPORT_DIMS, GL_MAX_VIEWPORTS,
GL_VIEWPORT_SUBPIXEL_BITS, GL_VIEWPORT_BOUNDS_RANGE, GL_LAYER_PROVOKING_VERTEX,
GL_VIEWPORT_INDEX_PROVOKING_VERTEX.
Tested on Nvidia Quadro 600, test passes.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Test valid and invalid queries using glGetFLoati_v and glGetDoublei_v.
Also test correct queries of GL_SCISSOR_TEST default value and settable values.
Tested on Nvidia Quadro 600 and all tests pass.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Tests for the validity of Viewport bounds, Depth Range bounds and
Scissor Box bounds with viewport arrays (0 to GL_MAX_VIEWPORTS-1).
"Bounds" refer to the bounding rectangle or range (eg x, y, width, height).
Tested with Nvidia Quadro 600. All tests pass except Scissor Box with a
negative height fails to return a gl error.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Tests GL_ARB_viewport_array validity for indices. Use both valid and
invalid parameters (index, first, count) for these new API entry points:
glScissorArrayv, glScissorIndexed, glScissorIndexedv.
Also test SCISSOR_TEST indices with glEnablei and others.
Tested on Nvidia Quadro 600 all test cases pass except for invalid
indices with glIsEnabledi(GL_SCISSOR_TEST,...).
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Tests GL_ARB_viewport_array validity for indices. Use both valid and
invalid parameters (index, first, count) for all these new API entry points:
glDepthRangeArrayv, glDepthRangeIndexed, glGetDoublei_v.
Tested on Nvidia Quadro 600 all tests pass.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Tests GL_ARB_viewport_array regarding the validity for the indices.
Use both valid and invalid parameters (index, first, count) for these new
API entry points:
glViewportArrayv, glViewportIndexedf, glViewportIndexedfv, glGetFloati_v.
Also test that writing to an invalid viewport index for Viewport, DepthRange,
Scissor Box, Scissor Test does not modify any of the state for the valid
range of indices.
Tested on Nvidia Quadro 600 all tests pass.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Jon Ashburn <jon@lunarg.com>
|
|
Before, the test always drew green rects for all the target buffers.
Now we draw a different color into each buffer to be more thorough.
Plus, replace 16 with MAX_TARGETS, add some comments, etc.
Note: the test now fails with Mesa's swrast because swrast errantly
writes the same color to all render targets.
|
|
This passes w/ Mesa but crashes NVIDIA's driver.
|
|
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
Fixes "Uninitialized scalar field" reported by Coverity.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Scripted with sed with fix ups to
- not - actually writes .w component, so probe rgba.
- reflect - writes const 1.0 .w component, so probe rgb.
- cosh - writes const 1.0 .w component, and used probe rgb with 4
components, so drop the .w from the probe.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Also basically a revert of 6533d757 to make the cosh test match.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Working with paths is hard, and we don't always do it right. Instead of
trying to be perfect, just use python's os.path module to do all of the
hard work for us
Signed-off-by: Dylan Baker <baker.dylan.c@gmail.com>
|
|
Specifically for glean tests setting certain environment variables are
needed to run a certain subtest in glean. This patch prints that in the
HTML results.
Signed-off-by: Dylan Baker <baker.dylan.c@gmail.com>
|
|
For i965/gen7 hardware, the driver has to go to extra work to make
gl_PrimitiveID work in the case where there is no geometry shader. So
we need a separate test to make sure that works properly.
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
In the OpenGL 3.3 specification, this changed from INVALID_VALUE to
INVALID_OPERATION.
According to Ian Romanick, the GL specification provides the guidance
that INVALID_VALUE should be generated for "Numeric argument out of
range". Since FALSE is in range for a boolean, INVALID_VALUE is not
appropriate, and the change to INVALID_OPERATION was meant as a
correction.
See:
http://lists.freedesktop.org/archives/mesa-dev/2013-July/042509.html
Most implementations support 3.3, so they likely follow the new code.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: Ian Romanick <idr@freedesktop.org>
Cc: Fredrik Höglund <fredrik@kde.org>
Cc: Corey Richardson <corey@octayn.net>
|
|
All the hiz tests requested a 400x400 window. Instead, don't specify
window size and let Piglit choose it. This significantly speeds up their
execution time.
Signed-off-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Signed-off-by: Dylan Baker <baker.dylan.c@gmail.com>
|
|
|
|
|