Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Ice939f27162062bcc5405315432a2732ebbe8afb

Remove old Android build scripts

Change-Id: I72128ec3cda3ddc78442861af9ae9768510bef25

Add new Android build and install scripts

This change adds new Android build and install scripts under
scripts/android. Key improvements over old ones are:

 * Build no longer relies on ant or 'android project' tools.

 * Native code build leverages scripts/build code which should fix
   incremental builds and improve compatibility.

 * Build script error reporting should be much better.

 * Final APK is now built incrementally which should enable much faster
   incremental builds once asset copy targets are fixed in main build.

This work required some changes to common code:

 * Android cross-compile toolchain is set up by including
   targets/android/ndk-r11.cmake before project() in the main
   CMakeLists.txt instead of using -DCMAKE_TOOLCHAIN_FILE. CMake native
   toolchain file support seems incredbly buggy and configuring
   toolchain in regular build files seems to be much more robust.

 * scripts/build/config.py now finds CMake automatically on OS X.

 * New HostInfo class has been added into scripts/build/config.py.

Components: AOSP, Framework

Change-Id: I4b5b78c0d4d3aff248887ba5ced0c91081e24e6b

Create dedicated Vulkan devices for SPIR-V assembly tests

This CL changes SPIR-V pipeline tests to use dedicated Vulkan
devices instead of the default one. So we can enable a special
extension on request.

Component: Vulkan

Affects: dEQP-VK.spirv_assembly.*

Change-Id: I3572fbc6807ba8938c7a9d9b3c82954887a5a36f

Fix multiview tests to properly throw NotSupported

Fix issue introduced in 61d205: do not query the
VkPhysicalDeviceMultiviewPropertiesKHX when VK_KHX_multiview extension is not
supported.

Either the above structure or the getter (defined in
VK_KHR_get_physical_device_properties2 extension, required for multiview) may be
unsupported leading to undefined behavior.

Affects: dEQP-VK.multiview.index.tesellation_shader.max_multi_view_view_count
Components: Vulkan
VK-GL-CTS issue: 400

Change-Id: I6a6f8411455483dca56788766ca4b62d90da7888

Use alias tag to check if ext and core funcs are same

OpenGL API registry commands have <alias> tag that can be used to
determine if extension function can be treated as an alias to a core
entry point.

This change updates scripts to take advantage of alias tag to correctly
determine if ext and core entry points can be folded into one. This
fixes issues with EXT_direct_state_access where behavior has changed
between EXT and core versions.

VK-GL-CTS issue: 368
Components: Framework
Change-Id: I6addfdca02c778b7a44db1cb81d5a88aae6cf273

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Ia594773daf04996cc73dd552b2887cc1f018a371

Improve GLSL source program support

Old GLSL source program support in Vulkan framework was built on top
of glu::ProgramSources. This had two problems:

1) There was no obvious way to extend/annotate sources with SPIR-V
-specific information such as target SPIR-V version.

2) Most of glu::ProgramSources, such as attribute bindings or TF
configuration, was completely ignored.

This change introduces new vk::GlslSource type that replaces
glu::ProgramSources in vk::SourceCollections. vk::GlslSource contains
shader source strings as well as build options (vk::GlslBuildOptions).

vk::GlslBuildOptions doesn't currently contain anything useful but can
be later extended to support selecting target SPIR-V version for
example.

The change is mostly transparent to existing code, except:

 * vkSpirvProgram.hpp was somewhat unnecessarily including
tcuTestLog.hpp and changing that to a forward declaration required
fixing includes in several test code files.

  * vktShaderExecutor.cpp has been extended to allow specifying build
options.

Components: Vulkan

Change-Id: I69a6cd55ea91215585515b9d41860d818cbc97ae

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I18f38e8eb61e9460bdf6455965ce0db8f42e2be8

Add tests for FragCoord interaction with multisampling

Components: Vulkan
VK-GL-CTS issue: 89

Test: fragcoord_msaa, frontface, fragdepth, clipping
Change-Id: Ibf342e63f224b88fe0b0cffde59b10b34afd1faf

Remove vktWsiTests.cpp.orig

VK-GL-CTS issue: 402

Change-Id: Ib72fcd8b8ff39013f6cd3093ee9160be320be2b0

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Ifeb12ec2b673f413c9c1b5d78339575ee40a8b0d

Add tests for VK_KHR_display

Implemented coverage tests for functions:
 * vkGetPhysicalDeviceDisplayPropertiesKHR
 * vkGetPhysicalDeviceDisplayPlanePropertiesKHR
 * vkGetDisplayPlaneSupportedDisplaysKHR
 * vkGetDisplayModePropertiesKHR
 * vkCreateDisplayModeKHR
 * vkGetDisplayPlaneCapabilitiesKHR
 * vkCreateDisplayPlaneSurfaceKHR

Components: Vulkan

VK_GL_CTS Issue: #155

Change-Id: Ia16ab268f6c8c001fbde64f5663f195078dd9aa4

Add test for FragDepth shader built-in

Components: Vulkan

VK-GL-CTS issue: 93

Test: fragdepth, frontface, clipping
Change-Id: I531e39a58dd6b577108bff14b2206d7bf9d9cd4f

Fix - checkSparseSupport missing image format check

checkSparseSupport does not check if sparse operations are supported
for image format for example for depth/stencil formats.
This fix extends checkSparseSupport with this additional check.

Affected tests:
dEQP-VK.glsl.texture_functions.textureoffset.sparse_sampler2darrayshadow_vertex
dEQP-VK.glsl.texture_functions.textureoffset.sparse_sampler2darrayshadow_fragment

Change-Id: Ib4ca34ede1b5efd382ed6491b88e10baac173895

Fix image.qualifiers.restrict tests not running without optional capability

Tests dEQP-VK.image.qualifiers* were passsing previously, but since
645b5c6d they require shaderStorageImageReadWithoutFormat capability.
This was not intended - this change restores the test to previous
behavior.

VK-GL-CTS issue: 345
Components: Vulkan
Affects: dEQP-VK.image.qualifiers.restrict.*

Change-Id: Ie3e70b72a9f3b3b0700c3fed898c30fd1f6ae290

Fix regression from c35c2901: spirv_assembly crashes on non-extended VK

Fix spirv_assembly test unconditionally calling
vkGetPhysicalDeviceFeatures2KHR.

This change adds check for VK_KHR_get_physical_device_properties2
extension and considers all related features disabled - if the
extension is not supported.

VK-GL-CTS issue: 393
Component: Vulkan
Affects: dEQP-VK.spirv_assembly.*

Change-Id: I6e42d05f9e051cf5424c60abe1736f471d2f04a1

Fix PrimitiveID variable as input of TesEval stage

Change-Id: I43d66d391520efc238d2012bf5b7c65c8307a5b7

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Id8d24b792b609193c85664d970c313981378feca

Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master

Change-Id: I3e4105213b99ff48e1f733a68f07821e8e5ba422

Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2

makeFramebufferWithoutAttachments creates invalid framebuffer

Current behaviour of setting width, height, and layers to zero is
invalid. The validity language specifies that width, height, and
layers must be > 0.

The same function exists in the utils files for sparse_resources
and synchronization tests, I've removed the unused copies.

Affects:

dEQP-VK.tessellation*

Components: Vulkan

VK-GL-CTS issue: 265

(cherry picked from commit bce45b549d21f068a75d005fe55bd87460b5df29)

Change-Id: I7e6196944b1197ca3ce2d16ff17623e6e9bcfae7

makeFramebufferWithoutAttachments creates invalid framebuffer

Current behaviour of setting width, height, and layers to zero is
invalid. The validity language specifies that width, height, and
layers must be > 0.

The same function exists in the utils files for sparse_resources
and synchronization tests, I've removed the unused copies.

Affects:

dEQP-VK.tessellation*

Components: Vulkan

VK-GL-CTS issue: 265

Change-Id: I201377ec5064d84eb3beb959ac5acf72e687c0b2

Fix view mask in multiview tests; Add maxMultiviewViewCount cases

Test was using 10 views while min required view count is 6.
Also added tests for max_multi_view_view_count

Renamed dEQP-VK.multiview.*.1_2_4_8_16_32_64_128_256_512
to dEQP-VK.multiview.*.1_2_4_8_16_32

Affects:
dEQP-VK.multiview.*.1_2_4_8_16_32

New testcases:
dEQP-VK.multiview.*.max_multi_view_view_count

Components: Vulkan
VK-GL-CTS issue: 366

Change-Id: Ic97f0f3db52c79fb0c3bf3e14018c66fcfe2a071

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: If924d0d298e0a61a245580584c51d58e3cd4fd59

Remove wrong image barrier from multisample_shader_builtin tests

Spec says:
"If vkCmdPipelineBarrier is called within a render pass instance,
the oldLayout and newLayout members of an element of
pImageMemoryBarriers must be equal"

The `multisample_shader_builtin` tries to use barrier for changing
the layout of the image used as render target and input attachment.
The barrier is invalid and, given the existing subpass dependencies,
not needed at all.

VK-GL-CTS Issue: 386
Component: Vulkan
Affects: dEQP-VK.pipeline.multisample_shader_builtin.*

Change-Id: I23bf1de5d39a5d786a6bf863a429f8d7e97f495d

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I13917698fb6cd8fcd4c9f8d64deaef1879c303ad

Merge vk-gl-cts/opengl-es-cts-3.2.3 into vk-gl-cts/master

Change-Id: I50888b10fa13d6a25938031ec4dc4547f59f1427

Add explicit mustpass for GL

Components: OpenGL
Affects: GL mustpass

Change-Id: I3fc78bd507e912e933770cebc73497df8e6e063e

Added CTS_ARB_shader_atomic_counter_ops specification proposal

affects:

GL45-CTS.shader_atomic_counter_ops_tests.*

components: OpenGL

VK-GL-CTS issue: 175

Change-Id: I6c878cdf77bf163531eba51008aabc3718adb322

Added CTS_ARB_shader_atomic_counter_ops tests

Added CTS_ARB_shader_atomic_counter_ops cases:
ShaderAtomicCounterOpsAdditionSubstractionTestCase
ShaderAtomicCounterOpsMinMaxTestCase
ShaderAtomicCounterOpsBitwiseTestCase
ShaderAtomicCounterOpsExchangeTestCase

affects:

GL45-CTS.shader_atomic_counter_ops_tests.*

components: OpenGL

VK-GL-CTS issue: 175

Change-Id: I6a8205f97f3cf22aa701a09b2b0222c9373413ba

Added CTS_ARB_shader_group_vote specification proposal

affects:

GL45-CTS.shader_group_vote_tests.*

components: OpenGL

VK-GL-CTS issue: 211

Change-Id: I4aecf0789e0da32e37125a582ff95eaf228fce94

Added CTS_EXT_polygon_offset_clamp specification proposal

affects:

GL45-CTS.polygon_offset_clamp_tests.*

components: OpenGL

VK-GL-CTS issue: 304

Change-Id: I99deb3ea1db90f36baf2a1592c7b06cae630b947

Cleanup PER_VERTEX declarations for specialization

This change harmonizes all of the PER_VERTEX declarations in the shader
specialization map so that they all use the same naming convention, and
have the same code formatting style. It also corrects some values that
were incorrect or missing for some tests.

Fixes:

KHR-GL44.geometry_shader.api.program_pipeline_vs_gs_capture
KHR-GL44.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_MaxPatchVertices_Position_PointSize
KHR-GL45.geometry_shader.api.program_pipeline_vs_gs_capture
KHR-GL45.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_MaxPatchVertices_Position_PointSize

Affects:

KHR-GLES31.core.geometry_shader.api.incompatible_draw_call_mode
KHR-GLES31.core.geometry_shader.api.insufficient_emitted_vertices
KHR-GLES31.core.geometry_shader.api.program_pipeline_vs_gs_capture
KHR-GLES31.core.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_MaxPatchVertices_Position_PointSize
KHR-GLES31.core.tessellation_shader.single.xfb_captures_data_from_correct_stage
KHR-GLES32.core.geometry_shader.api.incompatible_draw_call_mode
KHR-GLES32.core.geometry_shader.api.insufficient_emitted_vertices
KHR-GLES32.core.geometry_shader.api.program_pipeline_vs_gs_capture
KHR-GLES32.core.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_MaxPatchVertices_Position_PointSize
KHR-GLES32.core.tessellation_shader.single.xfb_captures_data_from_correct_stage
KHR-GL44.geometry_shader.api.incompatible_draw_call_mode
KHR-GL44.geometry_shader.api.insufficient_emitted_vertices
KHR-GL44.geometry_shader.api.program_pipeline_vs_gs_capture
KHR-GL44.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_MaxPatchVertices_Position_PointSize
KHR-GL44.tessellation_shader.single.xfb_captures_data_from_correct_stage
KHR-GL45.geometry_shader.api.incompatible_draw_call_mode
KHR-GL45.geometry_shader.api.insufficient_emitted_vertices
KHR-GL45.geometry_shader.api.program_pipeline_vs_gs_capture
KHR-GL45.tessellation_shader.tessellation_control_to_tessellation_evaluation.gl_MaxPatchVertices_Position_PointSize
KHR-GL45.tessellation_shader.single.xfb_captures_data_from_correct_stage

Component: OpenGL

VK-GL-CTS issue: 364

Change-Id: I1eb6d81b6de95a1f247615afe0ebc9dd99bddb1a

Add SPIR-V function to assembly graphics pipelines

* Add SPIR-V Function to determine the first execution thread in
graphics pipeline

* Shaders can use this function to avoid write conflicts by parallel
shader invocations

Affects: dEQP-VK.spirv_assembly.*graphics*.*

Component: Vulkan

Change-Id: I40ab73f185629dd65179bd73d8a878776b202b5f

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Ie8458116080a936455e0f14f0df50e8779646e93

Add tests for storage buffer unsized array length

Verifies that the length observed in the shader is correct, when binding
a portion of a buffer. Covers both the case where an explicit length of
the binding is given, and when VK_WHOLE_SIZE is used.

VK-GL-CTS issue: 449
Components: Vulkan

Change-Id: I9e30b2b56e4de888bb02fef3de1370fcd106c845

Add render pass multisample resolve tests

- Check that attachment components are resolved consistently
- Check that attachments are resolved consistently
- Check that integer formats always resolve to one of the samples
- Check that empty sample mask doesn't change values
- Check that all possible sample combinations of 1.0 and 0.0 values
  resolve to 0.5 on average.

VK-GL-CTS issues: 96
Components: Vulkan

Change-Id: I18b92ed713175cd935651575f4e48d9fd207ef00

Add multisample renderpass input attachment tests

VK-GL-CTS issues: 96
Components: Vulkan

Change-Id: I9ea87736c82148515da3c2d95301e6302be4257d

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I970a132b57dfe50bed70307c8264454decaab3a5

Merge aosp/deqp-dev into vk-gl-cts/master

Change-Id: Ib50d43dd6d3ba38a146e5266f1036fd27beb279b
Components: AOSP

Fix GCC 6.3 warnings in SRGB write control tests

Change-Id: I6f0f2054a50288a1c65bbd9693e9382d8f69bbcd

ES verification script reports EGL runs as errors

EGL is not recognizes as mandatory config by the verification script.
This change fixes it.

Components: OpenGL
Affects: submission verification

VK-GL-CTS issue: 382

Change-Id: I9262508e40966f5f63f235fb8ec077289de2f220

Ensure expected error in copytexsubimage2d_incomplete_framebuffer
am: 89bb39d249

Change-Id: I43f1e866fb5b932720623e55b1ee2aa2f14bb1d2

Update KC CTS reference

Affects:
GTF-GLES2.gtf.GL2ExtensionTests.egl_create_context.egl_create_context

Components: OpenGL

VK-GL-CTS issue: 353

Change-Id: I95dd480216a685e79c5dc936f1ee59c4eda714aa

Ensure expected error in copytexsubimage2d_incomplete_framebuffer

The GLES2 version of the copytexsubimage2d_incomplete_framebuffer test could
potentially throw INVALID_OPERATION instead of expected INVALID_FRAMEBUFFER_OPERATION
due to the cubemap texture not being defined by a previous TexImage2D.
This change re-uses the GLES3 version of the same test and ensures that
INVALID_FRAMEBUFFER_OPERATION is thrown.

Affects:
dEQP-GLES2.functional.negative_api.texture.copytexsubimage2d_incomplete_framebuffer

Change-Id: I7b3f770ff31c73d770a53f6b6bdae3ea4a37b63a
Components: AOSP
Google bug: 37497541

Exclude dEQP-GLES2.functional.negative_api.texture.copytexsubimage2d_incomplete_framebuffer

Affects:
dEQP-GLES2.functional.negative_api.texture.copytexsubimage2d_incomplete_framebuffer

Components: OpenGL

VK-GL-CTS issue: 380

Change-Id: Ie30686686d9c2829e23f86b3ff1292ddcb2614fe

Add negative tests for implicit conversions

Implicit conversion are forbidden in GLES
unless GL_EXT_shader_implicit_conversions is supported.
This change adds negative tests for implicit conversions.

GL_EXT_shader_implicit_conversions is specified against
ESSL 3.10, therefore it should be fine to add the negative
tests to dEQP-GLES2/dEQP-GLES3

Components: AOSP

Affects:
dEQP-GLES2.functional.shaders.invalid_implicit_conversions.*
dEQP-GLES3.functional.shaders.invalid_implicit_conversions.*

VK-GL-CTS issue: 142

Change-Id: I3afb0a5f2074868f882644ad830e4159747e4b14

Exclude dEQP-GLES2.functional.flush_finish.flush

Affects:
dEQP-GLES2.functional.flush_finish.flush

Components: OpenGL

VK-GL-CTS issue: 375

Change-Id: I1751a07886650114734ed656c30f754cb470b1c1

Exclude dEQP-GLES2.functional.polygon_offset.*render_with_units

Affects:

dEQP-GLES2.functional.polygon_offset.default_render_with_units
dEQP-GLES2.functional.polygon_offset.fixed16_render_with_units

Components: OpenGL

VK-GL-CTS issue: 374

Change-Id: I51b635df479215e5a5886e0a2ff23ec8363b1acf

Exclude few dEQP-GLES2.functional.*.nested_structs_arrays tests

Affects:

dEQP-GLES2.functional.uniform_api.info_query.nested_structs_arrays.mat4_mat2_vertex
dEQP-GLES2.functional.uniform_api.info_query.nested_structs_arrays.mat4_mat2_both
dEQP-GLES2.functional.uniform_api.value.initial.get_uniform.nested_structs_arrays.mat4_mat2_vertex
dEQP-GLES2.functional.uniform_api.value.initial.get_uniform.nested_structs_arrays.mat4_mat2_both
dEQP-GLES2.functional.uniform_api.value.initial.render.nested_structs_arrays.mat4_mat2_vertex
dEQP-GLES2.functional.uniform_api.value.initial.render.nested_structs_arrays.mat4_mat2_both
dEQP-GLES2.functional.uniform_api.value.assigned.by_pointer.get_uniform.nested_structs_arrays.mat4_mat2_vertex
dEQP-GLES2.functional.uniform_api.value.assigned.by_pointer.get_uniform.nested_structs_arrays.mat4_mat2_both
dEQP-GLES2.functional.uniform_api.value.assigned.by_pointer.render.nested_structs_arrays.mat4_mat2_vertex
dEQP-GLES2.functional.uniform_api.value.assigned.by_pointer.render.nested_structs_arrays.mat4_mat2_both

Components: OpenGL

VK-GL-CTS issue: 373

Change-Id: I343b177810bc37d6199c0135a591e2148f687ebd

Exclude dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond*

Affects:

dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_float
dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_vec2
dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_vec3
dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_vec4
dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_mat2
dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_mat3
dEQP-GLES2.functional.attribute_location.bind_aliasing.max_cond_mat4

Components: OpenGL

VK-GL-CTS issue: 372

Change-Id: I74de43935fc3ead5008ce5f31439716f5600ef35

DEPTH_STENCIL_OES as tex format requires OES_depth_texture
am: 8290443470

Change-Id: I84e9f7f9e732a2b5c728dd2cd35a68198d091fad

DEPTH_STENCIL_OES as tex format requires OES_depth_texture

Affects:

dEQP-GLES2.functional.fbo.completeness.*
dEQP-GLES3.functional.fbo.completeness.*

Bug: 36727598
Change-Id: Ic79408c21b80ea458baecf79c042e2694e72e0b2

eglGetFrameTimestamps: Use reserved enum values.
am: 84574b5c7b

Change-Id: I00ff6beadd8056729afb3f607efb4786b4287fbb

eglGetFrameTimestamps: Allow optional present support.
am: afb7428641

Change-Id: Ie0bdecf04dca30dfe9aa6e3e600c81317df15b7d

eglGetFrameTimestamps: pending vs. invalid timestamps.
am: b4e217e4a1

Change-Id: Ie88254e6829e2f5526240ad05a7d6b3e4010b3c8

eglGetFrameTimestamps: Use reserved enum values.

Test: --deqp-case=dEQP-EGL*get_frame_timestamps*

eglGetFrameTimestamps: Allow optional present support.

Test: --deqp-case=dEQP-EGL*get_frame_timestamps*

Change-Id: I5cfb334966f7f2794fe0043e1809de324a6e3d0e

eglGetFrameTimestamps: pending vs. invalid timestamps.

Test: --deqp-case=dEQP-EGL*get_frame_timestamps*

Change-Id: I27718995c82748b55bc53e4d78a7f572e16a45ab

Exclude dEQP-GLES2.functional.fragment_ops.depth_stencil.stencil_ops.*

Affects: dEQP-GLES2.functional.fragment_ops.depth_stencil.stencil_ops.*

Components: OpenGL

VK-GL-CTS issue: 370

Change-Id: Ifc25a189fa4ebf6cd36d0f07ec0af1c6f329b24a

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I13e7247a063ccdb13170804d92daf458dae5ab5c

Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master

Change-Id: I396972ad31a87e7e88b69c05595393af3f990540

Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2

Merge vk-gl-cts/vulkan-cts-1.0.0 into vk-gl-cts/vulkan-cts-1.0.1

Change-Id: Iff109ea0dba048599d3deb4d8f42a7cac8b62283

Fix typo in dE-IT.build_info.de_endianness_consistent name

Components: AOSP

Change-Id: Ie2cb1335834c010251021bac6b3882a37ede3f75

Merge "Try harder to defeat GLSL compiler dead-code optimizations" into nougat-cts-dev am: aeef6dd8f5 am: a4ddab16dd
am: 8ca5b73a02

Change-Id: I278dbd93f410e6880a0a9d5b10d0c1268fa43f26

Merge "Try harder to defeat GLSL compiler dead-code optimizations" into nougat-cts-dev am: aeef6dd8f5
am: a4ddab16dd

Change-Id: I18fe6a8ba3ba328513afdf1f5e5d32db6de32373

Merge "Try harder to defeat GLSL compiler dead-code optimizations" into nougat-cts-dev
am: aeef6dd8f5

Change-Id: I8300511b30c7bd03885a6cb019970852f7e72693

Merge "Try harder to defeat GLSL compiler dead-code optimizations" into nougat-cts-dev

Try harder to defeat GLSL compiler dead-code optimizations

A number of CTS tests generate shaders like like:

    #version 310 es

    buffer TargetInterface
    {
            highp float target;
    };

    highp vec4 readInputs()
    {
            highp vec4 retValue = vec4(0.0);
            retValue += vec4(float(target));
            return retValue;
    }

    void writeOutputs(in highp vec4 dummyValue)
    {
            target = float(dummyValue.y);
    }

    void main()
    {
            writeOutputs(readInputs());
    }

After various common optimizations this becomes:

    buffer TargetInterface
    {
            highp float target;
    };

    void main()
    {
            target = target;
    }

In the absence of memoryBarrier() or qualifiers on the buffer, there is
no guarantee about the order of writes to the buffer.  Since this write
is not guaranteed to be visible either on the GPU or the CPU, we
eliminate it.  Since there is no access to target in the shader, we
report GL_REFERENCED_BY_FRAGMENT_SHADER = GL_FALSE.  The tests expect
GL_TRUE.

The vectored versions of this test swizzle the value read from the
buffer before writing it back.  These writes are not eliminated.

Adding a uniform instead of a literal constant also prevents the reads
and writes of the SSBO from being eliminated.

v2: Ignore the uniform named "zero" in
ResourceListTestCase::verifyResourceList.  The alternative was to add
zero to the resource list, but that required making small changes
(mostly removing const) from over a dozen places in the code.  This
slightly hacky, but localized, change seemed better.

v3: Various coding standards fixes suggested by Alexander Galazin and
Pyry.  Add getDummyZeroUniformName to query name of the zero uniform and
a lot more documentation.  Both suggested by Pyry.

The following tests are affected:

- dEQP-GLES31.functional.program_interface_query.buffer_variable.random.6
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.compute.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.separable_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_fragment_only_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_geo_fragment_only_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_geo_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_tess_fragment_only_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_tess_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_tess_geo_fragment_only_fragment.unnamed_block.float
- dEQP-GLES31.functional.program_interface_query.buffer_variable.referenced_by.vertex_tess_geo_fragment.unnamed_block.float

Change-Id: I867ad32476269ac1272c09672be0a6d6fe37e31e
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://cvs.khronos.org/bugzilla/show_bug.cgi?id=16114
Google bug: 37228062

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 issues: 107, 108, 109
Components: Vulkan

Change-Id: I25355801c273cf34c6db7ee24f876cfc3b6932d0

Merge "Allow robustness tests on GLES 3.0"
am: ab3013ef10

Change-Id: If45ff3202fd0a0683567d879d188cae2833ceece

Merge "Allow robustness tests on GLES 3.0"

Allow robustness tests on GLES 3.0

Allow selected robustness tests run on GLES 3.0 instead of always
requiring 3.1.

Affects: dEQP-EGL.functional.robustness.*
Bug: 36428911
Test: dEQP-EGL.functional.robustness.* on Pixel XL

Change-Id: I78f543dcd667b5121f61150b082fd4a4a9c14b23

Add CTS_ARB_indirect_parameters specification proposal

This is CTS_ARB_indirect_parameters specification proposal.

components: OpenGL

VK-GL-CTS issue: 194

Change-Id: I3c59cedfe75f4c757bae3fbfb2a6b3759f5903a9

Add CTS_ARB_indirect_parameters implementation

This is CTS_ARB_indirect_parameters tests implementation

Affects:

GL45-CTS.indirect_parameters_tests.*

Components: OpenGL, Framework

VK-GL-CTS issue: 194

Change-Id: Ibddff6daf485a9d673cf1b09554c01b8b83958e1

copy_image test: do not require compressed formats on CM_ARRAY

Spec:
"An INVALID_OPERATION error is generated by TexImage3D if internalformat
is one of the EAC, ETC2, or RGTC compressed formats and either border is non-zero,
or target is not TEXTURE_2D_ARRAY."

Affects:
GL44-CTS.copy_image.incompatible_formats_compression

Components: OpenGL

VK-GL-CTS issue: 352

Change-Id: I77cbc09c2383d7d055776e1665f25b8c6051ee7d

Redeclare built-in variables in separable shaders

Change adds redeclaration of the built-in gl_perVertex struct
before using any of its member variables. Per spec, it is
necessary to do so for separable shaders.

Affects:

KHR-GL45.shader_image_size.* (46 tests)

Component: OpenGL

VK-GL-CTS issue: 343

Change-Id: I53372e9d98ee96201ab7f760e7ed16d78832f283

Allow PROVOKING_VERTEX for LAYER_PROVOKING_VERTEX

OpenGL differs from ES in that it allows GL_PROVOKING_VERTEX as a valid
response to a query about GL_LAYER_PROVOKING_VERTEX. In this case, the
test needs to then retrieve and use the value of GL_PROVOKING_VERTEX.

Affects:

KHR-GLES31.core.geometry_shader.layered_rendering.layered_rendering
KHR-GLES32.core.geometry_shader.layered_rendering.layered_rendering
KHR-GL44.geometry_shader.layered_rendering.layered_rendering
KHR-GL45.geometry_shader.layered_rendering.layered_rendering

(without the mustpass list patch applied, KHR-GLxx.* are GLxx-CTS.*)

Components: OpenGL
VK-GL-CTS issue: 341

Change-Id: I3fbd87f1ec47ee08f6584c4be27b41e80bc4e750

Set nearest filtering in GL CopyImage tests.

glCopyImageSubData requires textures to be mipmap complete.  These
tests create integer textures, but left the default min filter of
GL_NEAREST_MIPMAP_LINEAR or GL_LINEAR (for rectangle textures) in
place.  The completeness rules say that integer textures cannot be
mipmap complete with linear filtering.

Affects:
- GL45-CTS.copy_image.smoke_test
- GL45-CTS.copy_image.incompatible_formats
- GL45-CTS.copy_image.incompatible_formats_compression

Components: OpenGL
Khronos bugzilla: 16224
VK-GL-CTS issue: 330

Change-Id: I9554402ce348eb7dda63bf01b0548f68374d334a

Enable rasterizer discard for fma_precision tests

The gpu_shader5.fma_precision_* tests do not write to gl_Position in the
vertex shader as the only useful output from the tests is from transform
feedback. This leaves the gl_Position value undefined, causing
inconsistent results in our regression tests. Other similar tests (e.g.
fma_accuracy) do enable rasterizer discard.

Affects:

KHR-GLES31.core.gpu_shader5.fma_precision_float
KHR-GLES31.core.gpu_shader5.fma_precision_vec2
KHR-GLES31.core.gpu_shader5.fma_precision_vec3
KHR-GLES31.core.gpu_shader5.fma_precision_vec4
KHR-GLES32.core.gpu_shader5.fma_precision_float
KHR-GLES32.core.gpu_shader5.fma_precision_vec2
KHR-GLES32.core.gpu_shader5.fma_precision_vec3
KHR-GLES32.core.gpu_shader5.fma_precision_vec4
KHR-GLESEXT.gpu_shader5.fma_precision_float
KHR-GLESEXT.gpu_shader5.fma_precision_vec2
KHR-GLESEXT.gpu_shader5.fma_precision_vec3
KHR-GLESEXT.gpu_shader5.fma_precision_vec4

Components: OpenGL

VK-GL-CTS issue: 324

Change-Id: I955cf1bd7a31c1e67ecb613ecd0cedb58c8befe2

Fix unaligned XFB stores in GL44-CTS.gpu_shader_fp64.built_in_functions test

This change aligns all values in result buffer passed through transform
feedback to component size. This is done by aligning each variable in
result, so as per-vertex stride.

Spec:
"The results of appending a
    varying variable to a transform feedback buffer are undefined if any
    component of that variable would be written at an offset not aligned to
    the size of the component."

Components: OpenGL
VK-GL-CTS issue: 263

Affects test:
GL44-CTS.gpu_shader_fp64.built_in_functions

Change-Id: I74a02b161c06d8fb2d755d0fcf6211677a003e17

DirectStateAccessTexturesTests: fixes for textures_storage_errors

This commit fixes two situations while testing
TextureStorage2DMultisample and TextureStorage3DMultisample:

1. While testing a wrong target, the test was expecting
   INVALID_ENUM. But, from OpenGL 4.5 spec, section 8.19
   Immutable-Format Texture Images:

   "An INVALID_ENUM error is generated by TexStorage2DMultisample if
    target is not TEXTURE_2D_MULTISAMPLE or
    PROXY_TEXTURE_2D_MULTISAMPLE.

    An INVALID_OPERATION error is generated by
    TextureStorage2DMultisample if the effective target is not
    TEXTURE_2D_MULTISAMPLE."

2. While testing samples > MAX_SAMPLES, the test was expecting
   INVALID_VALUE. But from OpenGL 4.5 spec, section 8.19
   Immutable-Format Texture:

   "Calling TexStorage2DMultisample is equivalent to calling
    TexImage2DMultisample with the equivalently named parameters set
    to the same values, except that the resulting texture has
    immutable format."

    so going back to OpenGL 4.5 spec, section 8.8 Multisample
    Textures:

   "An INVALID_OPERATION error is generated if samples is greater than
    the maximum number of samples supported for this target and
    internalformat.  The maximum number of samples supported can be
    determined by calling GetInternalformativ with a pname of SAMPLES
    (see section 22.3)."

Affects:
GL45-CTS.direct_state_access.textures_storage_errors

Components: OpenGL

VK-GL-CTS issue: 183

Change-Id: I7484afba1bbadb1ab6ad8fc1a9e4a455ec2f84f5

Update GLES mustpass after AOSP merge

Components: OpenGL

Change-Id: Ie89739862c0d5bf7995f341cb626b1a5ddfb2d5b

Merge vulkan-cts-1.0.2 into aosp/master
am: 55da06144c

Change-Id: I6f5bb189fa1db31bc18acc50340113dd359da968

Merge vk-gl-cts/master into aosp/deqp-dev

Merge "x11_egl: search for glesv1_cm, toggle DEQP_SUPPORT_GLES1 on if found"
am: 89c3844ccb

Change-Id: I173509e62c9d905f6dcf4b5163c98d2ba82e7473

Merge vulkan-cts-1.0.2 into aosp/master

Contains following fixes that have been made after vulkan-cts-1.0.2.2:

 * Fix - uniform buffer incorrectly dereferenced
 * Use vkGetInstanceProcAddr(NULL) to load platform funcs
 * Imgtec Waiver for texture_gather*cube*depth32f*
 * Add VK_KHR_incremental_present to the list of allowed device
extensions
 * Use -std=c++03 with GCC and clang
 * Fix GCC 6.3 warnings in vulkan-cts-1.0.2
 * Improve check_build_sanity.py
 * Relax image verification in anisotropic filtering tests
 * Respect maxColorAttachments in image tests
 * Fix SPIR-V generation in dEQP-VK.spirv_assembly
 * Fix - depth/stencil images mandatory only for VK_IMAGE_TYPE_2D
 * Add missing barrier in ssbo layout tests

This merge doesn't match any Vulkan CTS release tag.

Bug: 36899783
Bug: 36817508
Change-Id: I5dff02835e5296e1c25ba67675c06e2261e5c476

Merge "x11_egl: search for glesv1_cm, toggle DEQP_SUPPORT_GLES1 on if found"

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I33dbb2baeb33661b2b8f872e8529ec280b0494b2

Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master

Change-Id: I17256fdcfe30d879548307bf5feb61a3ecd4ff39

Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2

Change-Id: Iaedfbb3579c55d33ebfc415197a67b8145725f4f

Fix - uniform buffer incorrectly dereferenced

Uniform buffer is incorrectly dereferenced in
ssbo layout tests.

Affected tests: dEQP-VK.ssbo.layout.*

Change-Id: I5831ad75f558f8992bed6bdc48b1833d2ee13d26

Use vkGetInstanceProcAddr(NULL) to load platform funcs

Vulkan specification says that only vkGetInstanceProcAddr() has to be
loaded using platform-specific means. Other 'platform'-class functions,
listed below, can be loaded using vkGetInstanceProcAddr(NULL, pName).

Following functions are loaded with vkGetInstanceProcAddr():
 * vkEnumerateInstanceExtensionProperties
 * vkEnumerateInstanceLayerProperties
 * vkCreateInstance

Change-Id: Iab0657bf93d3a36e4f4dda4a1f1550024db0890e

Do not re-init GL functions after context reset
am: 72c880b4c9

Change-Id: Ibe9564fccb61261dffba6f5fdefcccad755b31ba

Do not re-init GL functions after context reset

SharedContextResetCase was calling initGLFunctions() using a context
belonging to a share group where context reset has occurred.
initGLFunctions() calls glGetError() internally and at that point it
would return GL_CONTEXT_LOST.

Affects:

dEQP-EGL.functional.robustness.reset_context.shaders.infinite_loop.shared_context_status.vertex
dEQP-EGL.functional.robustness.reset_context.shaders.infinite_loop.shared_context_status.fragment
dEQP-EGL.functional.robustness.reset_context.shaders.infinite_loop.shared_context_status.vertex_and_fragment
dEQP-EGL.functional.robustness.reset_context.shaders.infinite_loop.shared_context_status.compute

Bug: 37077017
Change-Id: I27ae0951565760dee0bcc5040bb02a45c8f568f3

Merge vk-gl-cts/master into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I4e62724262b5530bfba97bc5b948f0f0f65bb711

Merge vk-gl-cts/vulkan-cts-1.0.2 into vk-gl-cts/master

Change-Id: I8acaa616704915c8cb487ff9be2445b6b9ba26be

Merge vk-gl-cts/vulkan-cts-1.0.1 into vk-gl-cts/vulkan-cts-1.0.2

Change-Id: I2614d3a7ce451d47418e7c850a7c2c372f2189a9