GitLab

Added VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT at image creation. According to
issue 17 of EXT_external_objects:

Vulkan applications will be required to set
the mutable format bit when creating an  image that will alias with
an OpenGL texture on an OpenGL implementation that supports
ARB_texture_view, OES_texture_view, EXT_texture_view, or OpenGL 4.3
and above.
Signed-off-by: Eleni Maria Stea <elene.mst@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

If vk-image-display and vk-image-overwrite fail the tests will return
without swap buffers taking place first. Made sure that each test calls
swap buffers before returning any results.
Signed-off-by: Eleni Maria Stea <elene.mst@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

Signed-off-by: cheyang <cheyang@bytedance.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <!438>

This fixes validation error VUID-vkCmdDraw-None-04007 thrown when
running tests with validation layer.

Error is thrown because pipeline is setup with a binding description
but vkCmdBindVertexBuffers is not called. Fix is to skip setting up
binding description when vbo is not being used by the test.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Eleni Maria Stea <estea@igalia.com>
Part-of: <!464>

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <!451>

Add a test which renders to a texture, exports it, imports it, and
checks its contents. Demonstrates a bug in iris.

Reproduces mesa#2678

.

v2: Require for GL_OES_EGL_image_external texture_for_egl_image().
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <!460>

None of the tests require this.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Eleni Maria Stea <estea@igalia.com>
Part-of: <!458>

Calling piglit_link_check_status without checking return value

Reported-by: Coverity Scan
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <!251>

Calling piglit_link_check_status without checking return value.

Reported-by: Coverity Scan
Fixes: 13fbb844

 ("sso: Combine 20 vertex shaders and 20 fragment shaders in various ways.")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <!247>

We only need four vertices to draw a quad
with GL_TRIANGLE_STRIP primitive.

Reviewed-by: Eric Anholt eric@anholt.net
Part-of: <!409>

See mesa!6564

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <!377>

Test is iterating through levels and expects fbo to be complete on each
iteration. According to spec, fbo is expected complete only if all
non-base levels are mipmap complete.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!455>

In this test fbo is expected to be complete with 2 levels, however
driver calculated max level with this size is bigger. Set max level
so that fbo is considered mipmap complete.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!455>

Ensure that the "new" functions are only added when the extension is
enabled.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!457>

These (previously) had issues in glsl_to_tgsi. Add tests to demonstrate
the issue as well as prevent it from recurring.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!457>

While the extension is written *against* the GLSL 4.40 spec, it in no
way requires GLSL 4.40. Reduce the version to 4.30 -- going lower
requires more complicated extension enables.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!452>

khr_texture_compression_astc can be exposed on opengl es 2.0, and
looking at the dependencies, it doesn't require you to have explicit
support for texture 3d or texture 2d array. So skipping fully the tex
image 3d checks if no texture 3d is supported, and skipping the 2d
array target if that one is not supported.

v2: test_compressed_teximg_3d: handle the case of having cube maps
    support but note texture 3d, and move the target/extension support
    to that method (Eric)
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!204>

This is not really fixing anything, but it is a nice good to have when
using the test to guide the development work on that extension.
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!204>

It's better to probe the whole rectangle than the middle pixel in this
test, in order to make sure that all pixels have been properly
overwritten by OpenGL.
Signed-off-by: Eleni Maria Stea <estea@igalia.com>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>

Large enough group sizes require SIMD32 code generation on recent
Intel hardware, which has regressed several times in master in
combination with 64-bit types.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111566

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

Exhibits the following bugs:
- glBlitFramebuffer when src fbo uses a texture view.
  See mesa#4034


- glBlitFramebuffer when dst fbo uses a texture view
- glReadPixels from a fbo using a texture view
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!447>

Drivers may perform optimizations around glClear calls. iris in
particular, can skip redundant clear calls.

This test demonstrates an iris driver bug in which the skipping
optimization is incorrectly affected by a conditional clear.
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!421>

Section 10.4 (Drawing Commands Using Vertex Arrays) of the OpenGL 4.6
compatibility profile spec says:

    The index of the draw (i in the above pseudo-code) may be read by a
    vertex shader as gl_DrawID, as described in section 11.1.3.9.

Based on this, I infer that count=0 entries should increment gl_DrawID.

This test passed before 85b6ba136bd ("st/mesa: implement
Driver.DrawGallium callbacks"), but it fails after.

Issues: mesa#4045

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!444>

Passes on i965, but fails on Iris with:

src/gallium/auxiliary/util/u_inlines.h:341:pipe_buffer_map_range: Assertion `length' failed.
Aborted (core dumped)
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!444>

Nixos makes this an error by default, so let's just fix it.
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!449>

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!448>

barrier() is not enough to ensure that all atomic writes
are complete, the recent GLSL 4.60 spec says:

"A barrier() affects control flow but only synchronizes memory
 accesses to shared variables and tessellation control output
 variables. For other memory accesses, it does not ensure that
 values written by one invocation prior to a given static instance
 of barrier() can be safely read by other invocations after their
 call to the same static instance of barrier(). To achieve this
 requires the use of both barrier() and a memory barrier."

A memoryBarrierAtomicCounter() before barrier() will ensure that
in each invocation the memory write is completed.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!445>

Test requires GL_ARB_gpu_shader5 in fragment shader therefor test
should have GL_ARB_gpu_shader5 in [require] section.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!446>

GL_MAX_GEOMETRY_SHADER_INVOCATIONS doesn't exist in GL3.2 only with
GL_ARB_gpu_shader5
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!440>

Otherwise we may have aligned_size less than chunk_size
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!441>

This test clear depth of a fbo with attached depth texture then
pass this texture to compute shader which expects certain depth value.
CS verifies all texels of depth texture and returns an amount of
texels which depth value differs from expected value.

Mesa Issue: mesa#3883

Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Signed-off-by: Andrii Simiklit <andrii.simiklit@globallogic.com>

We would, depending on stack state, try to link in a garbage program.
Fixes flaky result on softpipe.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!439>

As with fixed-function, fog should not affect the alpha channel.
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!437>

in the words of @ajax

:
The mode's already there, just not wired up for some reason.

but this is -g because -f is already taken and I hate uppercase opts
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <!432>

Apparently we were lacking tests going through all the ops and mods,
as I noticed when trying to replace st_atifs_to_tgsi.
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <!434>

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <!435>

v2:
  - Added more typos.

Signed-off-by: Andres Gomez <agomez@igalia.com
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <!416>

We were adding the images only when the comparison failed. We would
rather have the available information in any case.
Signed-off-by: Andres Gomez <agomez@igalia.com>
Acked-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <!426>

Before, we were just replacing its name with the checksum. We don't
want to lose the original name so we just append the ckecksum.
Signed-off-by: Andres Gomez <agomez@igalia.com>
Acked-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <!426>

Text in XML cannot include certain characters. The XML 1.0
Recommendation defines a valid Char falling in this range:

"

  [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] |
               [#x10000-#x10FFFF] /* any Unicode character, excluding
                                     the surrogate blocks, FFFE, and
                                     FFFF. */
"

This solves this kind of error:

"
  Traceback (most recent call last):
    File "./piglit", line 170, in <module>
      main()
    File "./piglit", line 166, in main
      sys.exit(runner(args))
    File "/piglit/framework/exceptions.py", line 49, in _inner
      return func(*args, **kwargs)
    File "/piglit/framework/programs/summary.py", line 271, in aggregate
      use_compression = backends.write(results, outfile)
    File "/piglit/framework/backends/__init__.py", line 201, in write
      return writer(results, file_path)
    File "/piglit/framework/backends/junit.py", line 483, in write_results
      writer(f, k, v)
    File "/piglit/framework/backends/junit.py", line 200, in __call__
      out.text = mola
    File "src/lxml/etree.pyx", line 1018, in lxml.etree._Element.text.__set__
    File "src/lxml/apihelpers.pxi", line 710, in lxml.etree._setNodeText
    File "src/lxml/apihelpers.pxi", line 698, in lxml.etree._createTextNode
    File "src/lxml/apihelpers.pxi", line 1493, in lxml.etree._utf8
  ValueError: All strings must be XML compatible: Unicode or ASCII, no NULL bytes or control characters
"

v2:
  - Speed up by compiling the regex before hand and move the logic to
    a specific function.
Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <!425>