Age | Commit message (Collapse) | Author | Files | Lines |
|
We want to be able to run atomic counter tests even when the DSA
entry points are unavailable (e.g. in GLES).
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
v2: Fix some formatting
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
Tests that check frontbuffer contents after drawing may fail because
the X server touch the front contents between drawing and reading. In those
cases, there's typically a pending expose event. So check for that
situation and rerun the test.
This fixes sporadic failures with glx-swap-copy when run with a compositor and
a swap-interval of zero. It would also have fixed gl-1.0@swapbuffers-behavior,
but a workaround (at least for dri3) was pushed in
"df56af2b"
(gl-1.0/swapbuffers-behavior: Try avoid reading from real front v2), but
that test-specific workaround could be reverted after this commit.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Eric Anholt: <eric@anholt.net>
|
|
f() is deprecated unprototyped declaration use f(void) instead.
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
Passes on intel beignet, intel CPU, CUDA OpenCL, clover(carrizo)
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
|
|
The existing tests only test that shaders receive the query result.
The new tests also check for earlier pipeline stages.
|
|
|
|
There doesn't seem to be a reason for not running those tests
concurrently. Looks like they were just missed.
|
|
Test the interaction of gl_SampleMask and gl_SampleMaskIn with different
sample shading rates. It used to fail on radeonsi.
|
|
This test causes an assert to be thrown on i965 currently.
|
|
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
These both use 20 counters so check against the limits before running.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
It's undefined (including crashing) behaviour to use atomic counters
without having bound enough storage for them. This fixes this test
to bind the storage for the counters it uses.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
gfx9 added v_mad_mix_f32, v_mad_mixlo_f16,
and v_mad_mixhi_f16 instructions. Make sure the
conversion to/from f16 is folded into this
instruction and it works.
These aren't great since they need more test
values, and generating half results is kind
of a pain from any other tool. The perfect
values used don't really stress the conversions,
but this as at least enough to make sure the
encoding is correct.
v2: Fix backslash alignments
Reviewed-by: Jan Veselu <jan.vesely@rutgers.edu>
|
|
v2: add enum test for GL_QUERY_COUNTER_BITS_ARB
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
|
|
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
clc allows max long type
Fixes program@execute@calls on CUDA OpenCL
Fixes: e408ce1f2bff23121670a8206258c80bb3d9befd (cl: Add tests for function calls)
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
This test compares rendering to the window system framebuffer with
rendering to an FBO, and it does so by rendering rotated quads. There
can be slight differences in rounding if the Y axis is flipped between
those buffers causing rendering differences which are permitted by GL.
Change the test to render axis-aligned quads in each viewport instead.
This should still catch any error with flipped Ys, and there are no
rounding issues since pixels are fully covered by quads.
Reviewed-by: Brian Paul <brianp@vmware.com>
|
|
ldexp(0.5, -1028) will be different from 0.0 on implementations that
support denorms. Simply crank up the exponents to exclude the possibility
of denorms.
|
|
v2: Fix xor local test
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
This is ported from a vulkan cts test showing the same brokenness.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
|
|
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|
|
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
|
|
Passes on ROCm, I haven't tried clover recently. Last
time I did it errored because the AsmParser wasn't properly
initialized.
v2: Fix non-unique test names, Wrap noinline in unguarded macro,
use prettier test names, use device_regex (effectively restricting to ROCm)
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
Tolerance was added for the tests a while ago, but it looks like it was
forgotten for the nearest_biased test (the nearest one has it).
Also, for the linear test, add tolerance too when comparing x and y lodq
results - the values should probably be the same mostly, however it's possible
(due to interpolation inaccuracies) to get values just below 0 or above 3, in
which case they will get clamped. (Could just do a clamp instead of allowing
tolerance I suppose, but some tolerance might be allowed in any case there
too.)
This is required for llvmpipe (with a in-progress change) to pass.
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
|
|
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
Following the pattern, if there ever were a OpenGL 4.6 CTS, it would
probably have tests named like this.
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
|
|
Trivial. ".50." is clearly a typo, it should have been ".50"
This fixes test failures in some environments.
Tested-by: Mark Janes <mark.a.janes@intel.com>
|
|
v2: "The size of each element in the mask must match the size of each element in the result."
v3: fixup python style (six. iteritems, enumerate, itertools.product, ...)
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
|
|
Fix Coverity unchecked return value defects.
CID: 743618
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
|
|
Fixes: f6d4e22d8b8eef1fbf5c7a234f520ae013a3e0d5 ("cl: Add support for OCL 2.0")
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Tested-by: Vinson Lee <vlee@freedesktop.org>
|
|
Fix Coverity unchecked return value defect.
CID: 1415101
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
|
|
GL_MAX_PROGRAM_TEXTURE_GATHER_COMPONENTS_ARB = 1
Expected results were with respect to
GL_MAX_PROGRAM_TEXTURE_GATHER_COMPONENTS_ARB = 4. This patch fixes
it for GL_MAX_PROGRAM_TEXTURE_GATHER_COMPONENTS_ARB = 1 as well.
Reviewed-by: Brian Paul <brianp@vmware.com>
|
|
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
|
|
Color everything so that it appears like 4 horizontal stripes (black, red,
yellow, white) to make sure everything is where it should be.
|
|
|
|
From GLGL 4.5 spec, 5.11 "Out-of-Bounds Accesses":
"In the subsections described above for array, vector, matrix and
structure accesses, any out-of-bounds access produced undefined
behavior. However, if robust buffer access is enabled via the
OpenGL API, such accesses will be bound within the memory extent
of the active program."
These tests perform writes to an unsized array, however some of
these writes can be out-of-bounds and the result is undefined. As
the tests are focused on in-bound access and they don't enable
robust buffer access, then this patch removes the out-of-bounds writes
because we cannot assume they will be ignored.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Arthur Huillet <ahuillet@nvidia.com>
|
|
v2: simplify
mark local storage volatile
v3: python style
Passes on beignet(IVB), clover(carrizo, iceland), and intel CPU
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
|
|
See bugzilla https://bugs.freedesktop.org/show_bug.cgi?id=101941
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
|
|
We found that if we swapped the order of the "vertex" and "attr"
attributes in the VS that the test would fail with Mesa (but not
NVIDIA).
This patch updates the program to test both declaration orders.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101941
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
|
|
All switch cases did the same thing. Spotted by Eric Anholt.
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|
|
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|
|
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|
|
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|
|
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|
|
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
|