Eduardo Lima Mitev [Wed, 8 Mar 2017 08:42:25 +0000 (09:42 +0100)]
preprocessor: Undefining __VERSION__ on GLSL 3.30 is not an error
OpenGL Shading Language 4.5 spec, section '3.3 Preprocessor',
page 12 states:
"By convention, all macro names containing two consecutive
underscores ( __ ) are reserved for use by underlying
software layers. Defining such a name in a shader does not
itself result in an error, but may result in unintended
behaviors that stem from having multiple definitions of the
same name."
There is no explicit mention to #undef, but the consensus is that
defining or undefining are both subject to the same restrictions
in this context.
Affects:
GL45-CTS.shaders.preprocessor.definitions.undefine_version_fragment
GL45-CTS.shaders.preprocessor.definitions.undefine_version_vertex
VK-GL-CTS issue: 206
Components: OpenGL
Change-Id: I36c9ada736d7704767e403b18f21d7343dff1aa2
Alejandro Piñeiro [Thu, 2 Mar 2017 17:49:50 +0000 (18:49 +0100)]
Skip FRAMEBUFFER_UNSUPPORTED subtests on DSA textures_buffer
From OpenGL 4.5 spec, section 9.4.2 "Whole Framebuffer Completeness":
"Although the GL defines a wide variety of internal formats for
framebuffer-attachable images, such as texture images and
renderbuffer images, some implementations may not support
rendering to particular combinations of internal formats. If the
combination of formats of the images attached to a framebuffer
object are not supported by the implementation, then the
framebuffer is not complete under the clause labeled
FRAMEBUFFER_UNSUPPORTED .
Implementations are required to support certain combinations of
framebuffer internal formats as described under “Required
Framebuffer Formats” in sec- tion 9.4.3."
That means that if the combination being tested returns
FRAMEBUFFER_UNSUPPORTED, it is still allowed by OpenGL spec. For those
cases, the test should skip instead of fail.
Affects:
GL45-CTS.direct_state_access.textures_buffer
Components: OpenGL
VK-GL-CTS: 201
Change-Id: Ia272a975e8417837637cd47b621b91306d1aa567
Eduardo Lima Mitev [Mon, 6 Mar 2017 14:45:35 +0000 (15:45 +0100)]
CullDistanceTests/coverage: Fix a couple issues in the shaders
There are two specific problems in the shaders of this test, both
preventing the fragment shader from receiving any fragment (no geometry
is renderer), and thus the gl_maxCullDistances value read back from the
texture is always zero.
The first issue is in the geometry shader. The 4 emitted vertices have
a 0.0 normal, the quad is not visible.
The second issue is in the tessellation evaluation shader, where the
value of gl_Position is not explicitly set, so the point is rendered
at an undefined location.
Components: OpenGL
Affects:
GL45-CTS.cull_distance.coverage
VL-GL-CTS issue: 197
Change-Id: I145894192fe9504808666e2c2fcd6cf53dbb03df
Juan A. Suarez Romero [Tue, 28 Feb 2017 10:05:46 +0000 (10:05 +0000)]
Fix GL45-CTS.cull_distance.functional test
This commit fixes several issues in this test:
- Coordinate X conversion from float in [0, 1] range to uint in range
[0, m_to_with] is wrong. Same for coordinate Y.
- Offset drawing point for primitive point is wrong.
- Generated vertex shader for the case of dynamic index writes with
culling but not clipping fails to compile, due "unsized array index must
be constant". This happens because we make reference in the code to the
clipdistance array, which is unsized. This problem also happens with
clipping but not culling.
- Program fails to link when using tessellation shaders, due
"definitions of interface block `gl_PerVertex' do not match". This
happens because tessellation control shader redefines gl_PerVertex, but
tessellation evaluation shader don't.
Components: OpenGL
VK-GL-CTS issue: 191
Affects:
GL45-CTS.cull_distance.functional
Change-Id: I5143b45caa5a0490f58628326b44c4373d5c8305
Adam Czupryna [Mon, 30 Jan 2017 07:04:50 +0000 (08:04 +0100)]
Add CTS_ARB_sparse_texture_clamp tests
Added CTS_ARB_sparse_texture_clamp cases:
ShaderExtensionTestCase
SparseTextureClampLookupResidencyTestCase
SparseTextureClampLookupColorTestCase
affects:
GL45-CTS.sparse_texture_tests.*
GL45-CTS.sparse_texture2_tests.*
components: OpenGL, Framework
VK-GL-CTS issue: 60
Change-Id: I1d312fe7fd82d5de74926d189f1e6ec6ce910ce2
Pyry Haulos [Thu, 9 Mar 2017 23:41:19 +0000 (15:41 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: Ic82729c2a686a720ac8c2775b8a2dc92a25989b1
Pyry Haulos [Wed, 8 Mar 2017 21:05:00 +0000 (13:05 -0800)]
Refactor vktShaderExecutor
ShaderExecutor Vulkan port had several issues:
* Lifetime of ShaderExecutor was designed to extend from TestCase to
TestInstance. This was done in order to allow generating shader sources at
TestCase and later reuse same object to do actual execution when Vulkan
device was available. Result was total lifetime mess and leak of Vulkan
objects beyond TestInstance lifetime.
* Whole UniformSetup system was badly designed mess that provided
functionality for just a few tests.
Following cleanups and fixes have been made:
* ShaderExecutor objects are now expected to be live only when Vulkan
device is available, i.e. only within vkt::TestInstance. Shader source
generation part has been moved into a separate utility.
* UniformSetup mess is removed and replaced with much leaner and
extensible way to supply additional resources. Descriptor set 1 is
reserved for test case specific resources and those can be provided by
supplying custom descriptor layout and descriptor set. Tests using old
UniformSetup system have been adjusted accordingly.
There is still room for additional improvements, such as moving much of
pipeline construction from ShaderExecutor::execute() into constructors,
but these can be done in follow-up CLs.
Affects following test groups:
dEQP-VK.glsl.builtin.*
dEQP-VK.glsl.opaque_type_indexing.
dEQP-VK.texture.explicit_lod.*
VK-GL-CTS issue: 182
Components: Vulkan
Change-Id: Icfc45a36cb4651f39f12608842ef347a55353d84
Pyry Haulos [Mon, 13 Feb 2017 21:11:53 +0000 (13:11 -0800)]
Improve precision handling in texture.explicit_lod
This change improves precision requirement handling in
dEQP-VK.texture.explicit_lod test group by adding separate controls for
format conversion and filtering math.
Changed tests:
dEQP-VK.texture.explicit_lod.*
Components: Vulkan
Change-Id: Id64da1d63d87d02c73c757a3f2d5d50f2fbc19c0
(cherry picked from commit
4269dc0c1e513c550d3d45c16f2760450dfcef5c)
Pyry Haulos [Thu, 9 Mar 2017 23:01:35 +0000 (15:01 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: I56bcaca8068dd1922af827f35253319b1106d5dd
Pyry Haulos [Thu, 9 Mar 2017 23:01:33 +0000 (15:01 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2
Change-Id: I14e2b7bbfe3940ace0e6ae9b661f2a56a49edf98
Pyry Haulos [Thu, 9 Mar 2017 23:01:31 +0000 (15:01 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.0 into vk-gl-cts/vulkan-cts-1.0.1
Change-Id: Ica55df93dd12ea2c6460ec7c6fbb1383e6d38f4d
Jason Ekstrand [Thu, 26 Jan 2017 22:27:48 +0000 (14:27 -0800)]
Add yflip versions of tessellation winding order tests
Changed: dEQP-VK.tessellation.winding.triangles_*
Components: Vulkan
Change-Id: I3e0abc8d25132e129cb2979b1fce985ca4a60dc4
Arkadiusz Sarwa [Mon, 13 Feb 2017 09:24:26 +0000 (10:24 +0100)]
Improve precision handling in tests with FragCoord
The tests for gl_FragCoord are not accurate enough, these tests pass
event for invalid driver. The comparison is made on RGBA8 surfaces,
it is not accurate enough for the tests.
The fix makes comparison more precise for the tests.
Modified tests:
dEQP-VK.glsl.builtin_var.simple.fragcoord_xyz
dEQP-VK.glsl.builtin_var.simple.fragcoord_w
Components: Vulkan
VK-GL-CTS issue: 77
Change-Id: I34ba77e696524215468202ef3fe77d1284e3e3df
Alexander Galazin [Wed, 8 Mar 2017 08:36:19 +0000 (09:36 +0100)]
Add Khronos mustpass v3.2.3
Change-Id: I7dafffd7928cee4653ac9073ecde18dba5a7a43a
Maciej Jesionowski [Wed, 8 Mar 2017 13:15:55 +0000 (14:15 +0100)]
Fix a compilation error on GCC 5.4
Components: Vulkan
Change-Id: I6c6681d16613f18b2a3192368927f6d176f63a64
Alexander Galazin [Mon, 5 Dec 2016 09:25:27 +0000 (10:25 +0100)]
Take snapshot of the AOSP master mustpass and use it in GL CTS
Components: OpenGL
VK-GL-CTS issue: 23
Change-Id: I9ebb11fd8c69f1b697bb08ac7351033c9b06ef94
Pyry Haulos [Wed, 8 Mar 2017 23:26:36 +0000 (15:26 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: Ice81ca8779a7f7cfea5d8fab0e4b48fe6e603ec2
Pyry Haulos [Wed, 8 Mar 2017 23:10:14 +0000 (15:10 -0800)]
Merge vk-gl-cts/aosp-master into vk-gl-cts/master
Components: AOSP, Framework
Change-Id: I0de6207abfc1c990a85fcf977d78ed1103c52a1b
Alexander Galazin [Thu, 23 Feb 2017 12:01:18 +0000 (13:01 +0100)]
Use glReadnPixels only if KHR_robustness is supported
glReadnPixels is introduced in KHR_robustness/GLES 3.2.
Use of this function in GLES 3.1 tests will result in segfault
on implementations that don't support KHR_robustness.
Components: AOSP, Framework
Affects: dEQP-GLES31.functional.debug.negative_coverage.callbacks.buffer.read_pixels_format_mismatch
VK-GL-CTS issue: 166
Change-Id: Idbe8734f7d82882ea99c75a8b74d693ac6dd831f
Graeme Leese [Thu, 2 Mar 2017 15:08:20 +0000 (15:08 +0000)]
Fix indirect draw line offset calculation
The GL_LINES test was correctly avoiding undefined rasterizer tie-break
conditions but other tests that draw lines were not. Include the fix in
the other tests.
Affects:
KHR-GLES31.core.draw_indirect.basic-mode-draw*-line*_adjacency
Components: OpenGL
VK-GL-CTS issue: 193
Change-Id: I58862469b0bae3b5f968bad8586adc7cad6ad7c7
Treehugger Robot [Mon, 6 Mar 2017 19:54:49 +0000 (19:54 +0000)]
Merge "Modify EGL robustness tests to use VBOs"
John Richardson [Fri, 3 Mar 2017 12:01:08 +0000 (12:01 +0000)]
Modify EGL robustness tests to use VBOs
Remove use of client-side vertex arrays and use buffer objects instead.
Delete buffers on teardown and in destructor.
Move test description logs to start log file.
Affected tests:
- dEQP-EGL.functional.robustness.reset_context.*
Change-Id: I9432258f30d6c7e02362c0b2f6c36fc3694bd72d
Pyry Haulos [Fri, 3 Mar 2017 16:44:08 +0000 (08:44 -0800)]
Fix "null" ID in sRGB decode tests
Use 0 as "null" GL object ID as -1 is potentially valid ID.
Change-Id: I95f9f0c300d5b6e95c0c7c81e29fd28ac9ca3b82
Adam Czupryna [Wed, 1 Feb 2017 14:07:45 +0000 (15:07 +0100)]
Add CTS_ARB_fragment_shader_interlock specification proposal
This is CTS_ARB_fragment_shader_interlock specification proposal
components: OpenGL
VK-GL-CTS issue: 61
Change-Id: I0a7827e87c807d1044d7434ac1e9833530be3cde
Pyry Haulos [Thu, 2 Mar 2017 20:40:08 +0000 (12:40 -0800)]
Fix narrowing conversion warning
Change-Id: I07b20837e5dac2f6a88504b582732a7db10cf5c8
Mark Adams [Thu, 2 Mar 2017 15:19:49 +0000 (10:19 -0500)]
Modify single_tessellation_stage to check for GL_NV_gpu_shader5
That extension permits the use of a tessellation control shader
without a tessellation evaluation shader. Affected tests:
dEQP-GLES31.functional.debug.negative_coverage.*.tessellation.single_tessellation_stage
Maciej Jesionowski [Thu, 2 Mar 2017 11:51:39 +0000 (12:51 +0100)]
Fix a compilation error on MSVC12
There's a problem with uint32_t, so better just use delib defines.
Components: Vulkan
Change-Id: I589ac7f220a7e651ed31910d16c643866314f681
Pyry Haulos [Wed, 1 Mar 2017 15:53:48 +0000 (07:53 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: I453e9f9eedc6e07d785d054b69c6822c0a9e80b1
Pyry Haulos [Wed, 1 Mar 2017 15:53:46 +0000 (07:53 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2
Change-Id: I1a17ff8600abeab32614212bdab2e4646e50bc25
Pyry Haulos [Wed, 1 Mar 2017 15:53:44 +0000 (07:53 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.0 into vk-gl-cts/vulkan-cts-1.0.1
Change-Id: I9eec4dd1fa88a85374eba8d0851b00a28c50b4ac
John Richardson [Wed, 1 Feb 2017 15:00:53 +0000 (15:00 +0000)]
Add new robustness extension tests
New tests for KHR_robustness extension. Attempts to provoke context
reset scenarios and test GL state after a reset has occurred.
New test group:
- dEQP-EGL.functional.robustness.*
New tests:
- create_context.query_robust_access
- create_context.no_reset_notification
- create_context.lose_context_on_reset
- reset_context.shaders.infinite_loop.* (various shaders)
- reset_context.shaders.out_of_bounds.* (various resource types)
- reset_context.fixed_function_pipeline.* (index and vertex buffers)
- negative_context.invalid_robust_context_creation
- negative_context.invalid_robust_shared_context_creation
Note: Test groups ending in *_non_robust are for internal use only.
Change-Id: I91d7e3e95ec52f908dbaa1f49c56dcd94d60fd8f
Mika Isojärvi [Wed, 22 Feb 2017 21:25:59 +0000 (13:25 -0800)]
Fix tessellation support checks in negative binding tests
Bug:
35618877
Test: Run the tests on couple of devices supporting tessellation
Change-Id: Ieca6a28a57a278a07ff688f760eb7da807e66d49
Pyry Haulos [Mon, 27 Feb 2017 19:26:45 +0000 (11:26 -0800)]
Merge Vulkan CTS 1.0.2.1 into AOSP
Change-Id: Ib0aaf12ec60902ebc31d00ad02291bd3d1cd7fae
Pyry Haulos [Thu, 23 Feb 2017 22:11:27 +0000 (14:11 -0800)]
Fix bug in reallocation callback validation
vk::validateAllocationCallbacks() didn't correctly handle a case where
reallocate(NULL, size) returned previously seen pointer value.
VK-GL-CTS issue: 176
Components: Vulkan
Change-Id: I0822f8e4176d20d7bf296cd85f5c5223e189d250
Igor Ostrowski [Mon, 27 Feb 2017 12:47:12 +0000 (13:47 +0100)]
Fix input attachment layout in renderpass tests
Input attachment layout was hardcoded for descriptor set update.
Added input attachment layout getter.
Affects:
dEQP-VK.renderpass.*
Components: Vulkan
VK-GL-CTS issue: 189
Change-Id: I2083c7b0cf4c68072f98d369f6c3089c6eb2b184
Pyry Haulos [Mon, 27 Feb 2017 06:12:52 +0000 (22:12 -0800)]
Merge vk-gl-cts-public/master into vk-gl-cts/master
Pyry Haulos [Mon, 27 Feb 2017 05:36:38 +0000 (21:36 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: I9d823bd148d6767786e6ad28f19fc381070d149f
Graeme Leese [Thu, 23 Feb 2017 16:56:10 +0000 (16:56 +0000)]
Fix extension support checks in negative api tests
Add checks for tessellation and geometry shader support for negative
shader storage and primitive bounding box tests.
Move all uses of cubemap array textures inside the extension checks in
the texture tests.
Bug:
29835218
Change-Id: I6051fff231285ed7d9c71015c2bb11a9fc7b909b
Signed-off-by: Pierre Couillaud <pierre@broadcom.com>
Alexander Galazin [Fri, 24 Feb 2017 20:47:42 +0000 (21:47 +0100)]
Fetch gl.xml, egl.xml from Github repos
SVN repository containing gl.xml and egl.xml is no longer updated.
The new locations for these files are OpenGL-Registry and
EGL-Registry repositories on Github.
Adjust fetching scripts to use the new locations.
Components: Framework
Affects: headers generation
Change-Id: Ic091a46c636d446b6a4751bd3b797bcd19f90014
Alejandro Piñeiro [Wed, 15 Feb 2017 12:48:19 +0000 (13:48 +0100)]
Fix [Named]FramebufferTextureLayer tests
The test was failing because it was reusing the texture ID used while
testing FramebufferTexture. But that is a GL_TEXTURE_2D, that is a
wrong target for [Named]FramebufferTextureLayer. So for the other
subtests, that were checking things like invalid attachment, etc,
there was two possible errors, and mesa driver raised first the one
related to a wrong target.
This patch reuses a texture ID with a valid target, being also used to
test [Named]FramebufferTextureLayer
For what it is worth, [Named]FramebufferTextureLayer with a wrong
target is also tested.
Affects:
GL45-CTS.direct_state_access.framebuffers_texture_attachment_errors
Components: OpenGL
VK-GL-CTS issue: 173
Change-Id: I388e1593e67d8708433d0f4763c716a4f3f03067
Lionel Landwerlin [Fri, 16 Dec 2016 13:17:02 +0000 (13:17 +0000)]
Make vertex shader boilerplate to set gl_Position
If we don't generate any fragment from the previous stages, the
fragment shader might not be run. Hence invalid results for
GL45-CTS.shader_image_load_store.multiple-uniforms in the fragment
stage.
Components: OpenGL
Affects: GL45-CTS.shader_image_load_store.multiple-uniforms
VK-GL-CTS issue: 30
Change-Id: I6342e27ffd51725879fe1519d81122f41cb81eef
Alexander Galazin [Fri, 24 Feb 2017 17:12:13 +0000 (18:12 +0100)]
Merge vk-gl-cts/opengl-es-cts-3.2.2 into vk-gl-cts/master
Change-Id: I039aa2432d455394b254e0556a1ab63f02fba35a
Iago Toral Quiroga [Tue, 21 Feb 2017 11:26:10 +0000 (12:26 +0100)]
Fix post-depth coverage shader test case
The tests uses 4 different versions of GLSL shader code for
the fragment shader that are supposed to be assigned to the
variables m_fragShader{1..4} that are then used in the 4
test cases implemented. However, the last shader code is not
assigned to m_fragShader4, instead it immediately overwrites
m_fragShader3. The test would then fail as it attempts to build
a shader program with an empty string for the fragment shader
when it uses m_fragShader4.
Components: OpenGL
VK-GL-CTS issue: 181
Affects:
GL45-CTS.post_depth_coverage_tests.PostDepthShader
Change-Id: I49474f04668de2037b4afc5570082bd4110f4fab
Alejandro Piñeiro [Mon, 20 Feb 2017 12:23:16 +0000 (13:23 +0100)]
DirectStateAccessTexturesTests: fix textures_parameters_errors
Right now GL45-CTS.direct_state_access.textures_parameter_errors
expects INVALID_OPERATION when using a wrong target.
From OpenGL 4.5 spec, section 8.11.2, Texture Parameter Queries:
"An INVALID_ENUM error is generated if the effective target is not
one of the texture targets described above."
The test even has a comment about expecting a INVALID_ENUM, so this
seems a typo when calling CheckErrorAndLog
Affects:
GL45-CTS.direct_state_access.textures_parameter_errors
Components: OpenGL
VK-GL-CTS issue: 180
Change-Id: Id82559090ccefa5713b0dd01d0ef26fdf4860ed3
Nicolai Hähnle [Fri, 17 Feb 2017 17:45:03 +0000 (18:45 +0100)]
Use glInvalidateBuffer[Sub]Data correctly
These function take a buffer name, not a bind point.
Modules: OpenGL
Affects:
GL45-CTS.sparse_buffer_tests.BufferStorageTest
VK-GL-CTS issue: 178
Change-Id: Iffdecfb5dd738ea82073ccb1b8771ca411f8f3b8
Nicolai Hähnle [Fri, 17 Feb 2017 15:38:02 +0000 (16:38 +0100)]
Fix shaders using reserved keywords incorrectly
"input" is a reserved keyword in GLSL 1.40 (and other versions,
for that matter); this test should never have passed for any
correct implementation.
"unsigned int" should be "uint".
Affects:
GL45-CTS.sparse_buffer_tests.BufferStorageTest
Modules: OpenGL
VK-GL-CTS Issue: 177
Change-Id: I68bba7548225ada7949182a1239d423eff15fc35
Eduardo Lima Mitev [Wed, 15 Feb 2017 11:37:31 +0000 (12:37 +0100)]
GetTextureSubImageTests: Remove m_texture_rectangle_compressed
GetTextureSubImage::Errors test attempts to setup a compressed texture
using GL_TEXTURE_RECTANGLE as target, during test preparation. This is
explicitly forbidden by the spec.
From OpenGL 4.5 PDF, section "8.7. COMPRESSED TEXTURE IMAGES", page 214:
"An INVALID_ENUM error is generated if the target parameter to any
of the CompressedTexImagenD commands is TEXTURE_RECTANGLE or
PROXY_TEXTURE_RECTANGLE ."
This patch removes the m_texture_rectangle_compressed texture from the
test, and all the associated test sub-cases.
Affects: GL45-CTS.get_texture_sub_image.errors_test
VK-GL-CTS issue: 169
Components: OpenGL
Change-Id: Ia3c2008cb948662f064dbc13fcd4d61c7d2caebf
Andres Gomez [Tue, 20 Dec 2016 17:25:46 +0000 (19:25 +0200)]
Unsized array of arrays in uniform block
From the GL_ARB_arrays_of_arrays spec:
"For unsized arrays, only the outermost dimension can be lacking a
size. A type that includes an unknown array size cannot be formed
into an array until it gets an explicit size."
This is later further detailed as per OpenGL® Shading Language 4.40.
From page 3 (page 9 of the PDF) of the GLSL 4.00 v9 spec:
"* Bug 11987: Allow array-of-array constructors and initialized
variables to have any dimension unsized (not just the outer
dimension) and get their size from the content of the constructor
or initializer."
From page 37 (page 43 of the PDF) of the GLSL 4.00 v9 spec:
" vec4 a[][] = { vec4[2](vec4(0.0), vec4(1.0)), // okay, size to
vec4[2](vec4(0.0), vec4(1.0)), // a[3][2]
vec4[2](vec4(0.0), vec4(1.0)) };"
From page 38 (page 44 of the PDF) of the GLSL 4.00 v9 spec:
"For implicitly-sized or run-time-sized arrays, only the outermost
dimension can be lacking a size. A type that includes an unknown
array size cannot be formed into an array until it gets an explicit
size, except for shader storage blocks where the only unsized array
member is the last member of the block."
Hence, in the absence of a constructor/initializer, unsized arrays are
only allowed for the outermost dimension of an AoA.
Affects:
GL43-CTS.arrays_of_arrays_gl.InteractionUniformBuffers3
Components: OpenGL
VK-GL-CTS issue: 146
Change-Id: Icbfe49c16e8ce7ffd55b92adede90405ac8e5f26
Pyry Haulos [Fri, 24 Feb 2017 16:52:34 +0000 (16:52 +0000)]
Merge "Drop MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS to 2048"
Andres Gomez [Tue, 20 Dec 2016 17:25:46 +0000 (19:25 +0200)]
Unsized array of arrays in shader storage block
From the GL_ARB_arrays_of_arrays spec:
"For unsized arrays, only the outermost dimension can be lacking a
size. A type that includes an unknown array size cannot be formed
into an array until it gets an explicit size."
This is later further detailed as per OpenGL® Shading Language 4.40.
From page 3 (page 9 of the PDF) of the GLSL 4.00 v9 spec:
"* Bug 11987: Allow array-of-array constructors and initialized
variables to have any dimension unsized (not just the outer
dimension) and get their size from the content of the constructor
or initializer."
From page 37 (page 43 of the PDF) of the GLSL 4.00 v9 spec:
" vec4 a[][] = { vec4[2](vec4(0.0), vec4(1.0)), // okay, size to
vec4[2](vec4(0.0), vec4(1.0)), // a[3][2]
vec4[2](vec4(0.0), vec4(1.0)) };"
From page 38 (page 44 of the PDF) of the GLSL 4.00 v9 spec:
"For implicitly-sized or run-time-sized arrays, only the outermost
dimension can be lacking a size. A type that includes an unknown
array size cannot be formed into an array until it gets an explicit
size, except for shader storage blocks where the only unsized array
member is the last member of the block."
Hence, in the absence of a constructor/initializer, unsized arrays are
only allowed for the outermost dimension of an AoA.
Affects:
GL43-CTS.arrays_of_arrays_gl.InteractionStorageBuffers3
Components: OpenGL
VK-GL-CTS issue: 145
Change-Id: Ic751b3ff52a84e3aee604552edf9441211890a11
Andres Gomez [Tue, 20 Dec 2016 16:57:57 +0000 (18:57 +0200)]
Unsized array of arrays in uniform variable
From the GL_ARB_arrays_of_arrays spec:
"For unsized arrays, only the outermost dimension can be lacking a
size. A type that includes an unknown array size cannot be formed
into an array until it gets an explicit size."
This is later further detailed as per OpenGL® Shading Language 4.40.
From page 3 (page 9 of the PDF) of the GLSL 4.00 v9 spec:
"* Bug 11987: Allow array-of-array constructors and initialized
variables to have any dimension unsized (not just the outer
dimension) and get their size from the content of the constructor
or initializer."
From page 37 (page 43 of the PDF) of the GLSL 4.00 v9 spec:
" vec4 a[][] = { vec4[2](vec4(0.0), vec4(1.0)), // okay, size to
vec4[2](vec4(0.0), vec4(1.0)), // a[3][2]
vec4[2](vec4(0.0), vec4(1.0)) };"
From page 38 (page 44 of the PDF) of the GLSL 4.00 v9 spec:
"For implicitly-sized or run-time-sized arrays, only the outermost
dimension can be lacking a size. A type that includes an unknown
array size cannot be formed into an array until it gets an explicit
size, except for shader storage blocks where the only unsized array
member is the last member of the block."
Hence, in the absence of a constructor/initializer, unsized arrays are
only allowed for the outermost dimension of an AoA.
Affects:
GL43-CTS.arrays_of_arrays_gl.InteractionUniforms2
Components: OpenGL
VK-GL-CTS issue: 144
Change-Id: Icd3e710f1bb69b35355a5c4172dc8839b9f98c82
Alejandro Piñeiro [Fri, 3 Feb 2017 17:33:02 +0000 (18:33 +0100)]
BufferStorageTests: do not unmap with MapOwner if not needed
This test uses the following method to call MapRange on a buffer:
Buffer::MapOwner Buffer::MapRange(glw::GLintptr offset,
glw::GLsizeiptr length,
glw::GLenum access)
It returns a MapOwner, that unmaps the buffer when it is destroyed.
For some tests, it is called like this:
{ /* start test */
{
buffer.MapRange(<params>);
} /* freed here */
/* Do stuff with the buffer. Assumes to be still mapped */
/* Checks the buffer */
} /* end test */
The problem with this approach is that as the Mapowner returned by
MapRange is not used, it is destroyed (on the previous code snippet at
"freed here"), and the buffer is unmapped unintentionally.
This patch gets out the call to MapRange, so it is not unmapped until
the end (on the code snippet, at "end test").
Fixes the following tests (only Mesa Intel driver tested):
GL45-CTS.buffer_storage.get_buffer_parameter
GL45-CTS.buffer_storage.map_persistent_buffer_sub_data
GL45-CTS.buffer_storage.map_persistent_read_pixels
Components: OpenGL
VK-GL-CTS issue: 148
Change-Id: I676eeaf0808e9bb19617bde7d658c14859f1357d
Piers Daniell [Fri, 24 Feb 2017 16:05:36 +0000 (09:05 -0700)]
Merge pull request #30 from BigBigos/master
Fix GL45-CTS.pipeline_statistics_query_tests_ARB.functional_default_qo_values
Pyry Haulos [Thu, 23 Feb 2017 20:50:22 +0000 (12:50 -0800)]
Drop MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS to 2048
GL_MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS minmax has been changed to
2048 (from 4096) in OpenGL ES 3.2 spec in response to bug 12823.
Bug:
35619900
Change-Id: I59d75d23df02d339bf7094a8e73d3a24d2e785be
Maciej Jesionowski [Tue, 17 Jan 2017 15:36:33 +0000 (16:36 +0100)]
Add render to mipmaps tests
This test repeatedly draws into the same image using consecutive
mip levels as color and dept/stencil attachments. All mip levels
and array layers are exercised.
Modified tests:
dEQP-VK.pipeline.render_to_image.*
New tests:
dEQP-VK.pipeline.render_to_image.*.mipmap.*
Components: Vulkan
Vulkan CTS issue: 487, 535
Change-Id: I7e4b3c6978d069c0accc377e9ab6e9d05e8e3cc6
Maciej Jesionowski [Tue, 17 Jan 2017 15:36:33 +0000 (16:36 +0100)]
Add max size attachments tests
This test creates very big color, and depth/stencil attachments
in all possible dimensions and layers.
Original tests in the affected group were moved to *.small.*
subgroup.
Modified tests:
dEQP-VK.pipeline.render_to_image.*.small.*
New tests:
dEQP-VK.pipeline.render_to_image.*.huge.*
Components: Vulkan
Vulkan CTS issue: 451, 535
Change-Id: I1e3bd1f7a657ff837b830bbada539a1240b1fd04
Iago Toral Quiroga [Tue, 10 Jan 2017 11:50:23 +0000 (12:50 +0100)]
Add framebuffer attachment size tests
New tests:
- dEQP-VK.pipeline.framebuffer_attachment.*
VK-GL-CTS issue: 72
Components: Vulkan
Change-Id: Iab4c077d60471633688476a87570e779d56bcda8
Mark Adams [Wed, 22 Feb 2017 22:03:36 +0000 (17:03 -0500)]
Modify tessellation_not_active to check for GL_NV_gpu_shader5
That extension permits use of GL_PATCHES for stages other than
tessellation stages. Affected test:
dEQP-GLES31.functional.debug.negative_coverage.get_error.tessellation.tessellation_not_active
Bug
35670519
Treehugger Robot [Wed, 22 Feb 2017 19:44:27 +0000 (19:44 +0000)]
Merge "Fix whitespace issues"
John Richardson [Mon, 20 Feb 2017 11:26:04 +0000 (11:26 +0000)]
Add non-seperable TCS-but-no-TES subcase and fix draw time check subcase
- Add a non-separable TCS-but-no-TES (new subcase).
Section 7.3 of the OpenGL ES 3.2 specification says:
"Linking can fail for [...] any of the following reasons:
* program contains an object to form a tessellation evaluation
shader [...] and [...] the program is not separable and
contains no object to form a tessellation control shader"
Add a test to ensure the implementation does raise a link error.
- Make program separable in TES-but-no-TCS draw time check test (fix)
Since the test is trying to check for a draw time error, it should use
a separable program - otherwise we may hit a link error.
Changes made to test case:
dEQP-GLES31.functional.debug.negative_coverage.*.tessellation.single_tessellation_stage
Bug:
35354738
Change-Id: I902106864059cfbe7d718da0d0527284de04a915
Pyry Haulos [Wed, 22 Feb 2017 18:05:57 +0000 (10:05 -0800)]
Fix whitespace issues
Change-Id: I31cd4e4d3e113192fce0bccd20fd8fa14d182620
Brian Anderson [Sun, 22 Jan 2017 00:03:19 +0000 (16:03 -0800)]
Add GetFrameTimestamps tests
Add tests for EGL_ANDROID_get_frame_timestamps.
Verifies:
1) Frame IDs properly identify frames.
2) The frame timestamp ordering is correct
3) The compositor timing values are reasonable.
4) All timestamps from the most recently swapped frame can
be retrieved without additional swaps.
Test: --deqp-case=dEQP-EGL*get_frame_timestamps*
Change-Id: I393a3bc79f019951a2fddb83760b6e1d919663b6
Treehugger Robot [Tue, 21 Feb 2017 18:52:36 +0000 (18:52 +0000)]
Merge "Fix multiple resource leaks in new srgb_texture_decode tests"
Kalle Raita [Tue, 21 Feb 2017 17:43:45 +0000 (17:43 +0000)]
Merge "Reduce read pixel calls in copy image tests"
Kalle Raita [Tue, 21 Feb 2017 17:43:25 +0000 (17:43 +0000)]
Merge "Restructure copy image test iterations"
Mark Adams [Fri, 17 Feb 2017 19:14:18 +0000 (14:14 -0500)]
Fix multiple resource leaks in new srgb_texture_decode tests
There were multiple issues here, ranging from never freeing resources,
to overwriting resources, to only freeing them at class destruction
instead of deinit time. This was resulting in hitting OOM on platforms
without tons of memory.
Bug:
35457323
Change-Id: I492177bb9e71bdbc7efc9d3a7c8a34bef35c709e
Boris Zanin [Tue, 21 Feb 2017 11:56:07 +0000 (12:56 +0100)]
Build fix (uint32_t used instead of deUint32)
It looks like uint32_t is not used within the CTS.
Replacing it with deUint32 allows build to pass.
Components: Vulkan
VK-GL-CTS Issue #66
Change-Id: Ia23f3fb7fe87ec21349c46df52c7b9f83dd14a32
Pyry Haulos [Fri, 17 Feb 2017 20:21:40 +0000 (20:21 +0000)]
Merge changes Icb29d2da,Ida906e21
* changes:
Add a non-separable TCS-but-no-TES subcase.
Make program separable in TCS-but-no-TES draw time check test.
Boris Zanin [Mon, 13 Feb 2017 09:47:28 +0000 (10:47 +0100)]
Workaround SSL cert issues, enable python 3
* The script command line switch --insecure added to bypass SSL
certificate validity checks. Though bypassing is not recommended,
sometimes the issue can be resolved on the remote side only,
which we have no control over. For example please see
https://sourceforge.net/p/forge/site-support/14336/
The parameter is supported for pythons 2.7.9 and 3.4.3 and newer,
due to older versions of urlopen do not support context parameter
* Minor changes allow to run fetch_sources.py under both v2 and v3
pythons. Changes required due to python 3 prohibits statement-style
usage for print and forces print statements to be converted into
functions. Also urlopen() is in library urllib.request in python 3
Components: Framework
VK-GL-CTS Issue #64
Change-Id: If4354332ea19878f8f72352e2df5cac9ee254ac6
Marcin Kańtoch [Fri, 10 Feb 2017 09:43:52 +0000 (10:43 +0100)]
Fix incorrectly defined image memory barrier
In ResolveImageToImage::iterate() method a memory barrier is defined
to change the layout of source image. By mistake destination image
parameters are used to define subresource range. It causes validation
errors as in some cases the subresource range extends beyond
the dimensions of the source image.
The fix uses source image parameters to define the memory barrier.
Modified tests:
dEQP-VK.api.copy_and_blit.resolve_image.*
Components: Vulkan
VK-GL-CTS issue: 160
Change-Id: I0ef72fbc99b127129441b7c66b43afd7b5a1e0ef
Nicolai Hähnle [Thu, 2 Feb 2017 15:18:57 +0000 (16:18 +0100)]
Move DrawElementsBaseVertexTests to GL40 package
These tests make use of tessellation shaders, which were introduced
in 4.0.
The "using deqp::TestCaseGroup" is added to fix compilation:
gl3cTestPackages.cpp refers to glcts::TestCaseGroup. Similar using
directives are spread throughout the code base, one of them is in
esextcDrawElementsBaseVertexTests.hpp, which is no longer included.
Affects:
GL45-CTS.draw_elements_base_vertex_tests.*
Components: OpenGL
VK-GL-CTS issue: 28
Change-Id: I7fd2d2a641532ee4ac44ad2d71544ce08b9b69ec
Jason Ekstrand [Tue, 24 Jan 2017 01:44:17 +0000 (17:44 -0800)]
dEQP-VK: Enable b10g10r11 and e5b9g9r9 blitting tests
This enables 620 new tests for blitting to/from e5b9g9r9 and b10g10r10.
Looking through vulkan.gpuinfo.org, most implementations support the
BLIT_SRC_BIT on these formats. I have a branch of our driver that
supports BLIT_DST_BIT for these formats and the tests all work fine.
Change-Id: I964abad8c44d3d7cc52a79bbb73ee6b9833845d6
Pyry Haulos [Thu, 16 Feb 2017 19:24:18 +0000 (11:24 -0800)]
Fix dethread build for OS X and iOS
Change-Id: I6fb2f2af89d7568bd473597a0d413c3987e457ae
Pyry Haulos [Fri, 27 Jan 2017 17:02:12 +0000 (09:02 -0800)]
Fix atan2() precision tests and add to mustpass
atan2() had a bug where tests were rounding codomain -PI..PI inwards,
not outwards, causing it to reject perfectly valid roundings of constant
PI.
Changed tests:
dEQP-VK.glsl.builtin.precision.atan2.*
Change-Id: Ib75da103944f773dd819c459b9aa1257c58ca921
Components: Vulkan
Petros Bantolas [Fri, 4 Nov 2016 15:12:52 +0000 (15:12 +0000)]
Test descriptor_update_template, push_descriptor
Extended the existing binding_model tests with 3 new variants,
'with_template', 'with_push_template' and 'with_push'.
The least invasive method to support push descriptor updates without
template was to persist the DescriptorSetUpdateBuilder object in the
test instance classes, so that it can be referenced when building the
command buffer.
Change-Id: I731fa41c5ff99ab1d390d77d7aabb4424e091e35
(cherry picked from commit
7f5ba3914c1d9033dd4f896ef6fcaa22f5c5e038)
Alexander Galazin [Tue, 14 Feb 2017 12:24:14 +0000 (13:24 +0100)]
Fix coding error in protocol detection
* add 'return' statement to detectProtocol
* call detectProtocol as a class member not as a global function
Components: Framework
Affects: KC CTS checkout
Change-Id: Id226166d0c35f6d10b23307e5cdec0fbd29102e8
Michael Hadley [Wed, 1 Feb 2017 12:48:08 +0000 (12:48 +0000)]
Add sRGB skip decode test
New Tests:
- deqp-case=dEQP-GLES31.functional.srgb_texture_decode.*
Change-Id: I0257e63b15ee0334efad985411d8070ddecdd51a
Kalle Raita [Thu, 19 Jan 2017 22:15:58 +0000 (14:15 -0800)]
Reduce read pixel calls in copy image tests
Reduce glReadPixels calls in the copy image tests by rendering multiple
texture sub-images for each read back. Significant improvement for
certain platforms, no impact on others.
Bug:
33965234
Test: Run tests on Pixel C and Pixel XL, Check that old N9 results still
have same failures. Spot checks with 240x200 target surface.
Change-Id: I133c56b7855a664f164f1eb0b1307259462e098f
Kalle Raita [Tue, 24 Jan 2017 19:47:17 +0000 (11:47 -0800)]
Restructure copy image test iterations
Packing test iterations so that work per buffer swap is maximized
without running the risk of hitting per-iteration timeout. Yields
execution time drops of roughly 50% on at least 2 GPU architectures.
Bug:
33965234
Test: Copy image tests on multiple devices and L MR1 N9 for detection capability.
Change-Id: I7de5231da0d8a1d2bd22ec3741f1481e2ebd5475
Mika Isojärvi [Wed, 8 Feb 2017 00:37:51 +0000 (16:37 -0800)]
Add create functions for common Vulkan types.
These functions simplify creating the Vulkan objects by taking the
arguments directly instead of passing through struct. Multiple test
cases already reimplement these functions.
Change-Id: I09f6b0163b5aa2198617230767cd072185afe45c
Pyry Haulos [Mon, 13 Feb 2017 21:11:53 +0000 (13:11 -0800)]
Improve precision handling in texture.explicit_lod
This change improves precision requirement handling in
dEQP-VK.texture.explicit_lod test group by adding separate controls for
format conversion and filtering math.
Changed tests:
dEQP-VK.texture.explicit_lod.*
Components: Vulkan
Change-Id: Id64da1d63d87d02c73c757a3f2d5d50f2fbc19c0
Pyry Haulos [Mon, 13 Feb 2017 19:16:33 +0000 (11:16 -0800)]
Start new mustpass for Vulkan CTS 1.0.3
Vulkan CTS 1.0.2 mustpass is now archived under mustpass/1.0.2 and new
mustpass can be found under mustpass/1.0.3.
Components: Vulkan
Change-Id: I7c073dc7436a434558ba1710e01988377a8d686c
Kenneth Graunke [Sat, 11 Feb 2017 06:45:39 +0000 (22:45 -0800)]
Add a non-separable TCS-but-no-TES subcase.
"Linking can fail for [...] any of the following reasons:
* program contains an object to form a tessellation control
shader [...] and [...] the program is not separable and
contains no object to form a tessellation evaluation shader"
Add a test to ensure the implementation does raise a link error.
The new subcase is part of:
dEQP-GLES31.functional.debug.negative_coverage.*.tessellation.single_tessellation_stage
Change-Id: Icb29d2da50b2c327bfd2196af9be56bb7d8644f9
Kenneth Graunke [Sat, 11 Feb 2017 05:30:50 +0000 (21:30 -0800)]
Make program separable in TCS-but-no-TES draw time check test.
Section 7.3 of the OpenGL ES 3.2 specification says:
"Linking can fail for [...] any of the following reasons:
* program contains an object to form a tessellation control
shader [...] and [...] the program is not separable and
contains no object to form a tessellation evaluation shader"
Since the test is trying to check for a draw time error, it should use
a separable program - otherwise we may hit a link error.
Fixes a bug in:
dEQP-GLES31.functional.debug.negative_coverage.*.tessellation.single_tessellation_stage
Change-Id: Ida906e21c3e3534fd2a8c7bac0334083881bdb19
Pyry Haulos [Mon, 13 Feb 2017 19:06:24 +0000 (11:06 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: I0fc80b05b1d3a795fff26f6c971e57064d224478
Pyry Haulos [Mon, 13 Feb 2017 18:42:12 +0000 (10:42 -0800)]
Update Android CTS Vulkan mustpass
Following commits modify Vulkan CTS 1.0.2 mustpass:
*
fbaba83dbee101b3096ce33f21ae398efee3fa67
*
c09596ed4b8d17f057f77a42475d59c1ac126797
This change simply updates Android version of mustpass to match Khronos
Vulkan CTS 1.0.2 mustpass.
Change-Id: I79e2696ec4c457d11b59e03cf2a73977ab98d8d5
Components: AOSP
Pyry Haulos [Tue, 31 Jan 2017 00:02:32 +0000 (16:02 -0800)]
Update Android CTS Vulkan mustpass to match 1.0.2
This CL updates Android CTS mustpass for Vulkan to match Khronos Vulkan
CTS 1.0.2.0.
Change-Id: Ia2f82b5f6675c67940679eb98f59510f264d62bf
(cherry picked from commit
15f342274521c1be1032e993dbfd4a46b878f337)
Pyry Haulos [Mon, 13 Feb 2017 18:34:40 +0000 (10:34 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master
Change-Id: Ie75f95c0972b43843398ca6d68fd39a538f23761
Pyry Haulos [Mon, 13 Feb 2017 18:34:37 +0000 (10:34 -0800)]
Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2
Change-Id: Ida077bae420c265f4a92d78d68b0f1c9b8d1c85c
Iago Toral Quiroga [Fri, 10 Feb 2017 10:35:52 +0000 (11:35 +0100)]
Write gl_PointSize value when it is undefined
gl_PointSize is undefined unless expicitly written to. The only
exception to this rule is the case when
shaderTessellationAndGeometryPointSize is disabled, in which case it
defaults to 1.0 for these stages.
Since we don't have access to the physical device features at shader
compile time, implement two variants of the tessellation evaluation
shader when point mode is enabled, one that writes gl_PointSize and
one that doesn't and decide the one we need at run time based on
whether the feature is enabled or not.
Components: Vulkan
Fixes VK-GL-CTS issue: 159
Affects:
dEQP-VK.tessellation.primitive_discard.*point_mode
Change-Id: Idadbc70a704f00ed601ff65afeb2a3edfbeb3f19
Kenneth Graunke [Thu, 2 Feb 2017 18:40:02 +0000 (10:40 -0800)]
Fix a few bugs in a EGL_KHR_no_config_context test.
The dEQP-EGL.functional.create_context.no_config test had the eglBindAPI
check backwards - so it was skipping APIs assuming they failed when they
actually succeeded.
Secondly, it failed to actually set the test result to "pass" if all
cases passed, which led to assertion failures:
Assertion `testResult != QP_TEST_RESULT_LAST' failed.
Change-Id: I7a20467012fd6c174f79679bbf640917c327574c
Iago Toral Quiroga [Thu, 2 Feb 2017 09:09:53 +0000 (10:09 +0100)]
Fix crashes in dEQP-VK.glsl.matrix.inverse.*
This is a regression caused by altering the value of the GRID_SIZE
in commit
2ba66fa2, which results in the inverse() tests generating
input matrices which determinant is zero, and thus, cannot be
inverted, which makes the tests hit an assertion.
This patch adds a mechanism for shader tests to specify a grid size
different from the default. This allows shader testst that are
sensitive to this value, such as matrix inverse tests, to define
their own requirements.
Fixes VK-GL-CTS isssue: 66
Components: Vulkan
Affects:
dEQP-VK.glsl.matrix.inverse.*
Change-Id: I31a34c2d156ee5b98d2de1b10ac4cec1a90d480e
Antia Puentes [Tue, 17 Jan 2017 20:30:31 +0000 (21:30 +0100)]
geometry_shader/api: Fix linking error
The tests redeclared the built-in 'gl_PerVertex' interface block in the
Vertex shader when run in a GL context. However, they did not redeclare
'in gl_PerVertex gl_in[]' and they used it, leading to an undesirable
linking error which is not what the test wished to check.
The patch adds the 'in gl_PerVertex gl_in[]' redeclaration to the
Geometry shader when the tests are run under GL.
From the GLSL 4.5 specification, "7.1 Built-In Language Variables"
(page 130):
"If multiple shaders using members of a built-in block belonging to
the same interface are linked together in the same program, they
must all redeclare the built-in block in the same way, as described
in section 4.3.9 “Interface Blocks” for interface-block matching, or
a link-time error will result. It will also be a link-time error if
some shaders in a program redeclare a specific built-in interface
block while another shader in that program does not redeclare that
interface block yet still uses a member of that interface block. If
a built-in block interface is formed across shaders in different
programs, the shaders must all redeclare the built-in block in the
same way (as described for a single program), or the values passed
along the interface are undefined."
VK-GL-CTS Issue: 130
Components: OpenGL
Affects:
GL44-CTS.geometry_shader.api.insufficient_emitted_vertices
GL44-CTS.geometry_shader.api.incompatible_draw_call_mode
GL45-CTS.geometry_shader.api.insufficient_emitted_vertices
GL45-CTS.geometry_shader.api.incompatible_draw_call_mode
Change-Id: I7d5f54b287b778e46376ef6ac9effcd90778705c
Nicolai Hähnle [Thu, 2 Feb 2017 15:43:40 +0000 (16:43 +0100)]
Take GL_MAX_IMAGE_SAMPLES into account
GL45-CTS.shader_texture_image_samples_tests.image_functional_test
tests imageSamples(), but binds images without checking whether their
number of samples are within the bounds allowed by
GL_MAX_IMAGE_SAMPLES.
This change brings the test in line with the shader_image_size_test,
which also skips unsupported sample counts.
Affects:
GL45-CTS.shader_texture_image_samples_tests.image_functional_test
Components: OpenGL
VK-GL-CTS issue: 29
Change-Id: Ia945a0f6472317f9014d4f820572eb006743450a
Adam Czupryna [Mon, 19 Dec 2016 15:07:42 +0000 (16:07 +0100)]
Add CTS_ARB_sparse_texture_clamp test specification proposal
This is CTS_ARB_sparse_texture_clamp test specification proposal.
components: OpenGL
VK-GL-CTS issue: 60
Change-Id: Ic11f1752e7f875d2cf55ad3fa4bf954c8a714fe3
Daniel Koch [Thu, 8 Dec 2016 07:24:40 +0000 (02:24 -0500)]
Enable GL support for Android build
- Add GL45Activity which creates a GL4.5 core activity
- Enable android build scripts to take 'gl' as argument for
-glcts-gtf-target
- update documentation for Android building and running
Components: Framework, AOSP, OpenGL
Change-Id: I2231adac800ac641287f157c712a25d9ddd66853
Mika Isojrvi [Thu, 9 Feb 2017 22:44:03 +0000 (22:44 +0000)]
Limit changes by xor to upper 8 bits in mixed atomic tests am:
6bc3c7a634
am:
eef2e716e7
Change-Id: Icadf536794c067056e66f6a2041fbb937952f35a
Mika Isojrvi [Thu, 9 Feb 2017 22:41:59 +0000 (22:41 +0000)]
Limit changes by xor to upper 8 bits in mixed atomic tests
am:
6bc3c7a634
Change-Id: I354ee074ad57d53c2b4b0a84d8bc02a02912f19d
Pyry Haulos [Wed, 11 Jan 2017 23:11:26 +0000 (15:11 -0800)]
Android: Warn about inconsistent OpSMod and OpSRem
Change-Id: Ib0e32c55f9236e4f77ad461341329cce41c22e0a
Igor Ostrowski [Mon, 2 Jan 2017 16:41:30 +0000 (17:41 +0100)]
Add copy image to image and blit image tests with various image layouts. Refactor copy and blit tests generation.
Tests took into account only optimal layout cases. Added general and mixed ones.
Structure of all copy and blit tests was too flat. Lowered granularity of test groups.
New tests:
- dEQP-VK.api.copy_and_blit.image_to_image.all_formats.*optimal_optimal*
- dEQP-VK.api.copy_and_blit.image_to_image.all_formats.*optimal_general*
- dEQP-VK.api.copy_and_blit.image_to_image.all_formats.*general_optimal*
- dEQP-VK.api.copy_and_blit.image_to_image.all_formats.*general_general*
- dEQP-VK.api.copy_and_blit.blit_image.all_formats.*optimal_optimal*
- dEQP-VK.api.copy_and_blit.blit_image.all_formats.*optimal_general*
- dEQP-VK.api.copy_and_blit.blit_image.all_formats.*general_optimal*
- dEQP-VK.api.copy_and_blit.blit_image.all_formats.*general_general*
Affected tests:
- dEQP-VK.api.copy_and_blit.*
Components: Vulkan
Change-Id: I5724bb7c0a3ccb2e36319a20671761a9b16fe4ed