dEQP-VK: Add SPV_KHR_16bit_storage at shaders using 16bit

All shaders stages that use VK_KHR_16bit_storage capabilities
need to enable the SPV_KHR_16bit_storage extension at the SPIR-V
assembly.

A recent update of the SPIRV-Tools checks for this error in the
SPIR-V code. This has raised several link error in the execution
of the VK_KHR_16bit_storage tests.

This patch enables the extension at the assembly code of the compute
shaders tests and the pass-through stages of the input-output
graphics tests.

Also includes newline at OpExtension "SPV_KHR_16bit_storage" uses.

Components: Vulkan
VK-GL-CTS issue: 454

Affects:
* dEQP-VK.spirv_assembly.instruction.graphics.16bit_storage.input_output.*
* dEQP-VK.spirv_assembly.instruction.compute.16bit_storage.*

Change-Id: I29b7e8706eba54b5f931f33442c85613d99cd113

Allow denormalized input values to be flushed to zeros

Components: Vulkan
VK-GL-CTS issue: 452

Affects:
dEQP-VK.spirv_assembly.instruction.graphics.16bit_storage.*
dEQP-VK.spirv_assembly.instruction.compute.16bit_storage.*

Change-Id: I2ab9fe88cd603157e477c0132c6db3421b1e440f

Fix for crash in createDeviceWithExtensions

createDeviceWithExtensions function is crashing if extensions vector
does not contain any elements. This simple fix prevents the crash
from happening.

Affects: dEQP-VK.spirv_assembly.*

Components: Vulkan

Change-Id: If4e5149df8d00b25b6ee97a235757d44dccd60d1

Fix incorrect sType in cross instance tests

The tests create a VkExternalImageFormatPropertiesKHX struct with
sType VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTERNAL_IMAGE_FORMAT_INFO_KHX,
should be VK_STRUCTURE_TYPE_EXTERNAL_IMAGE_FORMAT_PROPERTIES_KHX.

Components: Vulkan
VK-GL-CTS issue: 446

Affects:
dEQP-VK.synchronization.cross_instance.*

Change-Id: Ibc1d28d56f495f34f24f9baaf3ca7a79cc5fe415

Check unsupported image formats in cross instance tests

vkGetPhysicalDeviceImageFormatProperties2KHR can return
VK_ERROR_FORMAT_NOT_SUPPORTED, in which case the tests
shouldn't fail and be marked as unsupported instead.

Components: Vulkan
VK-GL-CTS issue: 431

Affects:
dEQP-VK.synchronization.cross_instance.*

Change-Id: I0b2c1f49f26f2516d0610c14d285e0b11c72a762

Test VK_KHR_dedicated_allocation mem requirement queries

Add tests:
 * dEQP-VK.memory.requirements.dedicated_allocation.buffer.*
 * dEQP-VK.memory.requirements.dedicated_allocation.image.*

Components: Vulkan

VK-GL-CTS issue: 391

Change-Id: I9510464987e33b395a5ac20c2fdb9129eefd91da

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

Change-Id: I93bd873bad46a33a3e6a86971f343981b188b1b3

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

Change-Id: I0a9fdded2cbfd25f1e2a1c73da97b2d8b8431962

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

Change-Id: Ib63e665cb64989f8c5f0c53c9d229e5f17172d62

Allow pop-free clipping for lines

Following test checks has been relaxed:
 * dEQP-VK.rasterization.primitives.lines_wide
 * dEQP-VK.rasterization.primitives.line_strip_wide
 * dEQP-VK.rasterization.primitives_multisample_2_bit.lines_wide
 * dEQP-VK.rasterization.primitives_multisample_4_bit.lines_wide
 * dEQP-VK.rasterization.primitives_multisample_8_bit.lines_wide

Components: Vulkan

VK-GL-CTS issue: 421

Change-Id: I21c04015d36ae55c20ef1c1dda8e2d1fee471221

Check 'drawIndirectFirstInstance' feature to run indirect draw tests

Component: Vulkan
VK-GL-CTS issue: 413

Change-Id: I13a86721ec76ac50af55e0730fd50e3d7ce159d8

dEQP-VK: Fix an incorrect layout transition

The tests upload data to the resolve destination using CopyImageToBuffer
prior to doing a resolve to a chunk out of the middle of the image. This
lets the test validate the resolve to a subarea works correctly.
However, they were also transitioning the destination from UNDEFINED to
TRANSFER_DST_OPTIMAL between uploading the data and doing the resolve.
This transition destroys the contents of the image making verifying that
contents incorrect.  The correct transition is from TRANSFER_DST_OPTIMAL
to TRANSFER_DST_OPTIMAL.

This affects the following tests:

 - dEQP-VK.api.copy_and_blit.resolve_image.partial*
 - dEQP-VK.api.copy_and_blit.resolve_image.with_regions*

Change-Id: I9336195d7977d7dd2ffb057525134846b10e8010

Fixed invalid usage of VkBufferImageCopy struct

The offset member of VkBufferImageCopy was violating the
valid usage. The offset is now calculated using lowest
common denominator between the pixelSize and 4.

Affects:

dEQP-VK.api.image_clearing.*

Components: Vulkan

VK-GL-CTS issue: 422

Change-Id: I1ef07e06f790a9ed403941b1d94c7840485cec3e

Build fix for X11-less Linux builds

The combination of recent refactor on release branch and
merge to master resulted in XinitThreads being called from
code that may be compiled even if X11 API is not available.

Components: Framework

VK-GL-CTS Issue: 450

Change-Id: Ia7d1d23eabd1b222d5039868c1fb762e1d42f01f

Update KC CTS ref

Components: OpenGL

VK-GL-CTS issue: 440

Affects:
GL4*-CTS.gtf42.GL3Tests.texture_storage.texture_storage_texture_levels

Change-Id: Id0b338bb74d3b1923744556527b45f3e8db8db38

Use 1.0 for shadow texture expected result

Change the texture clamp test to use 1.0 as the expected value
for shadow texture.

It's now required to pass <format> to generateExpectedResult
Note that the path using compute shader is doing an imageLoad on
depth texture so is not using a depth compare

Affects:

KHR-GL45.sparse_texture_clamp_tests.*

Components: OpenGL

VK-GL-CTS issue: 444

Change-Id: I94c79abe1b9b932471d1d8abdb8819ee3d7cc947

Skip uncommitted depth/stencil test validation

The test in UncommittedDepthStencil expects the stencil buffer to be
unmodified because the shader is doing an uncommitted sparse fetch

This is incorrect, the depth/stencil should be unmodified only if
the depth/stencil buffer are uncommitted and it's not the case,
they are fully-committed non-sparse depth/stencil buffer

This change skip uncommitted depth/stencil test validation

Affects:

KHR-GL45.sparse_texture2_tests.*

Components: OpenGL

VK-GL-CTS issue: 439

Change-Id: I909937081f7058d0fbc57469626ab3d34d43486a

MultiView extended tests

Affects:
dEQP-VK.multiview.*

New testcases:
dEQP-VK.multiview.input_attachments.*
dEQP-VK.multiview.instanced.*
dEQP-VK.multiview.input_instance.*
dEQP-VK.multiview.draw_indirect.*
dEQP-VK.multiview.clear_attachments.*
dEQP-VK.multiview.secondary_cmd_buffer.*

Components: Vulkan

VK_GL_CTS Issue: #70

Authors: Iago Toral Quiroga, Arkadiusz Sarwa

Change-Id: If45159d462186380cf3e8e1b6cd70c38af153762

Add test for VK_KHR_get_memory_requirements2

Update tests:
 * dEQP-VK.memory.requirements.buffer.*
 * dEQP-VK.memory.requirements.image.*

Rename tests (add .core subgroup):
 * dEQP-VK.memory.requirements.core.buffer.*
 * dEQP-VK.memory.requirements.core.image.*

Add tests:
 * dEQP-VK.memory.requirements.extended.buffer.*
 * dEQP-VK.memory.requirements.extended.image.*

Components: Vulkan

VK-GL-CTS issue: 391

Change-Id: I307bbf361d36d5f277ff70ce3204d67f8eb59871

Fix sparse cubemap array tests issues

Add cubeMapArrayRefZDef, a new shader token used as <compare>
parameters on some shadow cube map array texture instructions

Add <CUBE_REFZ_DEF> to some FunctionToken argument list definitions.
It will be replaced by the value of cubeMapArrayRefZDef.

Copy the depth reference value to s.cubeMapArrayRefZDef when we
generate the shader tokens

Change the way cube map array texture coordinate are computed in
the shader to support array with more than 6 layers

Specify that textureGradClampARB on shadow cubemap array is not
supported

Generate cube map array texture coordinate attributes as a vec4
with different coordinate for each layer

Affects:

KHR-GL45.sparse_texture2_tests.*
KHR-GL45.sparse_texture_clamp_tests.*

Components: OpenGL

VK-GL-CTS issue: 445

Change-Id: Ia9fb2240396ab4a34dd697d0b96985da35dfe802

Skip height validation on TEXTURE_1D_ARRAY

In verifyTexStorageInvalidValueErrors the test validate that a
GL_INVALID_VALUE error is returned by the GL when the height
parameters of the texStorage call is invalid.

The height is ignored by the texStorage call for 1d array, so the
error is never generated.

We should skip this validation for 1d array

Affects:

KHR-GL45.sparse_texture_tests.*

Components: OpenGL

VK-GL-CTS issue: 437

Change-Id: Ibe93db467deb560f5394b4b658dfa3f32a953709

Fix texPageCommitment alignment test

In verifyInvalidValueErrors, the test wants to validate that
texPageCommitment returns a GL_INVALID_VALUE error when the page
depth offset is not a multiple of pageSizeZ.
However if pageSizeZ equals 1, any z offset will be a multiple of
pageSizeZ and the test should not return an error.

Logic is added to make sure there's really an alignment error

Affects:

KHR-GL45.sparse_texture_tests.*

Components: OpenGL

VK-GL-CTS issue: 438

Change-Id: If7a00dee071c1dafabbda608f7cdbb14d1eb56b3

textureClamp texcoord with shadow 1d uses vec3

This change modifies the current shader generation for textureClamp
subtests.
s.coordType and s.sizeDef will now use vec3 for sampler1DArrayShadow
and sampler1DShadow

Affects:

KHR-GL45.sparse_texture_clamp_tests.*

Components: OpenGL

VK-GL-CTS issue: 443

Change-Id: I025061981877a62b6eed6bcc6c783f6fdc68c7d3

Initialize FunctionToken for every iteration

The change initializes the variable inside the loop so it gets
re-initialized at every iteration.

This help avoid issues where changes accumulates at every iteration

Affects:

KHR-GL45.sparse_texture_clamp_tests.*

Components: OpenGL

VK-GL-CTS issue: 442

Change-Id: I10eac411a03b3eff5139026ac004dc2df0c025f0

Set uncommited region access expected data to 0

In the GL_TEXTURE_CUBE_MAP path, the expected data is initialized
to 0xff. This seems incorrect.
The test will only fecth the uncommitted region of the texture which
should return 0.

This change fixes this issue and initializes the data to 0

Affects:

KHR-GL45.sparse_texture2_tests.*
KHR-GL45.sparse_texture_clamp_tests.*

Components: OpenGL

VK-GL-CTS issue: 441

Change-Id: Ic00cdfde24248d1640fe458c8f0a547d7d174e3d

Return an error if the pagesize index is invalid

In verifyTexStorageVirtualPageSizeIndexError, the test call
texStorage with a VIRTUAL_PAGE_SIZE_INDEX texture parameter set
to NUM_VIRTUAL_PAGE_SIZES + 1.

However the verifyError after the texStorage expect GL_NO_ERROR.
It should be GL_INVALID_OPERATION

Affects:

KHR-GL45.sparse_texture_tests.*

Components: OpenGL

VK-GL-CTS issue: 436

Change-Id: I43d02dcbee4bbe2c81feb4a17e9fde7386375656

Fix pack alignment issue on odd window width

The test allocates a buffer of width*height elements and expects the
readPixels to only write to this memory.

This is not correct if the window width is not a multiple of 4

Adding a gl.pixelStorei(GL_PACK_ALIGNMENT, 1) before the readPixels
fixes the issue

Affects:

KHR-GL45.indirect_parameters_tests.MultiDrawArraysIndirectCount

Components: OpenGL

VK-GL-CTS issue: 433

Change-Id: I7f8b92af7b5a27a6770e7cde23e51df30508a3c0

Remove OUT_OF_MEMORY coverage of buffer api

The OpenGL spec leaves much of the behaviour around OUT_OF_MEMORY
errors undefined, which makes it difficult to test for them. Regardless
of whether the error is generated or not, the GL context is by
definition left in an undefined state, which could invalidate any
subsequent tests' results.

This change removes OUT_OF_MEMORY test coverage.

VK-GL-CTS issue 435

Component: OpenGL

Affects:

KHR-GL45.direct_state_access.buffers_errors

Change-Id: I6436d4d72a0a7b7cd08336895e400d85bea8da15

Added CTS_ARB_shader_ballot specification proposal

affects:

GL45-CTS.shader_ballot_tests.*

components: OpenGL

VK-GL-CTS issue: 143

Change-Id: Ida320ac444665f09edd86d6a69aae9af7515a237

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

Change-Id: I7b3bddd72baec1544475f3320aca2a13a59ffa3c

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

Change-Id: I95690b5479602db484d2f89bdc711a683afe96b4

Fix uniform location use in robust_buffer_access

Common code will try to set the texture at uniform location 0, so change
the ES shader to use location 0.

Affects: KHR-GLES32.robust.robust_buffer_access_behavior.texel_fetch

Components: OpenGL
VK-GL-CTS Issue: 416

Change-Id: I6e4e291b33072b16e1ab73eb8592afd00687adc7

Remove dEQP-GLES31.*.vertex_array.*_invalid_map

The tests expects GL_INVALID_OPERATION to be generated if
a non-zero buffer object name is bound to an enabled array
or the element array and the buffer object's data store
is currently mapped.

Such behavior is not required by the specification.

Components: OpenGL

Affects:
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_elements_base_vertex_invalid_map
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_elements_instanced_base_vertex_invalid_map
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_range_elements_base_vertex_invalid_map

VK-GL-CTS issue: 429

Change-Id: Ib7e9c95dde9b86f227b3d014210a2fe732315f31

Remove unused function from es31fSRGBDecodeTests.cpp

Unused function causes compiler warnings to appear with recent GCC
versions.

Change-Id: I941c579994acbc9146df86acf5b9da9ee85341f5
(cherry picked from commit 1ee59ff98641dc0be7511e03e37b5843ff7eed82)

Fix texture/sampler mapping in sRGB tests

sRGB decode tests assume that an API side texture with index i
is connected with a shader sampler variable uTexture with index i.

This relation was however incorrectly established, e.g.
the tests were mapping texture0 to a sampler with location 0, while
such sampler was not necessarily uTexture0.

This change fixes the mapping and makes the relation more obvious.

Components: AOSP
Affects:
dEQP-GLES31.functional.srgb_texture_decode.skip_decode.*.toggled
dEQP-GLES31.functional.srgb_texture_decode.skip_decode.*.conversion_gpu

Google bug: 38257749
VK-GL-CTS issue: 428

Change-Id: I16508f6033b31583b1c5973b70a610dd59e78e8f
(cherry picked from commit c8efb45eadba4a910ab5770b2207582cf1bb7ca0)

Adjust out of surface verification for wide lines

Wide lines rasterization tests may draw lines out of
the allocated surfaces. Such situation causes verification
failures.
This change adjust the verification algorithm to ignore
missing pixels for out of surface rendering.

Components: Framework

Affects:
dEQP-GLES2.functional.rasterization.primitives.lines_wide
dEQP-GLES2.functional.rasterization.primitives.line_strip_wide
dEQP-GLES2.functional.rasterization.primitives.line_loop_wide

VK-GL-CTS issue: 427

Change-Id: Ia70f33fb6daad0d3774ae9edbd438bd14296f1d3
(cherry picked from commit c065847203ba5c1ec0bba76fe06e0b7a17cbc6af)

Ensure vertex shader presence if a geometry shader exists

Certain tests verify that INVALID_OPERATION is generated
if a geometry shader is active and mode is incompatible with
the input primitive type of the geometry shader in the currently
installed program object. However these tests have no vertex shader and
this is the real reason why INVALID_OPERATION is generated.
This change ensures that the vertex shader is present.

Components: AOSP

VK-GL-CTS issue: 426

Affects:
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_elements_base_vertex_primitive_mode_mismatch
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_elements_instanced_base_vertex_primitive_mode_mismatch
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_range_elements_base_vertex_primitive_mode_mismatch

Change-Id: I36bcee8892e4f641feadb0b444a53ed35273b727
(cherry picked from commit f5f087e9bd571b5aee3b810957af7051cd82d2c0)

Fix DrawRangeElementsBaseVertex tests with negative basevertex

DrawRangeElementsBaseVertex tests with negative basevertex call
glDrawRangeElementsBaseVertex with start = 0 and basevertex = -1.
According to the spec: If no element array buffer is bound, the vertex ID of
the ith element transferred is indices[i] + basevertex...
If the vertex ID is larger than the maximum value representable by type,
it should behave as if the calculation were upconverted to 32-bit unsigned
integers (with wrapping on overflow conditions).
Behavior of DrawElementsOneInstance is undefined if the vertex ID is negative
for any element, and should be handled as described in section 6.4.

This change ensures that indexMin + baseVertex will be zero or positive value.

Components: AOSP

VK-GL-CTS issue: 425

Affects:
dEQP-GLES31.functional.draw_base_vertex.draw_range_elements_base_vertex.base_vertex.index_neg_byte
dEQP-GLES31.functional.draw_base_vertex.draw_range_elements_base_vertex.base_vertex.index_neg_short
dEQP-GLES31.functional.draw_base_vertex.draw_range_elements_base_vertex.base_vertex.index_neg_int

Change-Id: I07a9e6bc60d4340784d400b5087bb8fff01f0e57
(cherry picked from commit e490e2f2158aa9d3aec7d8e385ca3995fefc9ffd)

Fix negative CompressedTexImage2D tests

Certain tests expect GL_INVALID_VAULE for
glCompressedTexImage2D(GL_TEXTURE_2D, -1, 0x8b90, 0, 0, 0, 0, 0).

0x8b90 is a palette format PALETTE4_RGB8 and
the OES_compressed_paletted_texture extension allows
negative values for the level parameter:
    Accepted by the <level> parameter of CompressedTexImage2D
        Zero and negative values.  |level| + 1 determines the number of
        mip levels defined for the paletted texture.

This change fixes the tests in presence of
the OES_compressed_paletted_texture extension.

Components: AOSP
Google bug: 38109626
VK-GL-CTS issue: 424

Affects:
dEQP-GLES2.functional.negative_api.texture.compressedteximage2d_neg_level_tex2d
dEQP-GLES2.functional.negative_api.texture.compressedteximage2d_neg_level_cube

Change-Id: Ia37e3fecaf3547abfb209803378114043c63a243
(cherry picked from commit 9de005f0355e389f1782543f34e0fb3c65a0a700)

Correct the test conditions for sRGB_write_control

The extension is not part of ES3.2, so the not-supported version of the
tests should be run on ES3.2 implementations without the
GL_EXT_sRGB_write_control extension.

VK-GL-CTS Issue: 389
Components: AOSP

Affects:
dEQP-GLES31.functional.fbo.srgb_write_control.framebuffer_srgb_enabled
dEQP-GLES31.functional.fbo.srgb_write_control.framebuffer_srgb_unsupported_enum

Change-Id: I42a531aa7d1ad38f7a85cde1a5da659f780ec5d5
(cherry picked from commit 2d987e3b1e3c564ddc197124b1c7689aed060b8e)

Update KC CTS ref

Components: OpenGL

VK-GL-CTS issue: 430

Affects:
GTF-GL45.gtf30.GL3Tests.framebuffer_srgb.framebuffer_srgb_draw

Change-Id: I01b5d5bb9ae37b24a44427c1a7cd735f98e60675

Add tests for VK_KHR_storage_buffer_storage_class

This change extends following groups to cover the new StorageBuffer
decoration:

 * dEQP-VK.glsl.opaque_type_indexing
 * dEQP-VK.spirv_assembly.compute.opatomic

VK-GL-CTS issue: 357
Components: Vulkan

Change-Id: I283a39f4e92f1be6597d843c67debc3ae28c9f22

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

Change-Id: I7d89570a93664a95736113863921985ee1af9d6c

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

Change-Id: Ie841d0a435d1d587450b7c2d57307b37b3295bbb

Refactor X11 and wayland platforms to run alongside each other

This adds a capability for running Vulkan WSI ane EGL tests on both
X11 and wayland compositors from one binary built in 'default' target.

The change replaces X11 & wayland tcu::Platform-s with one common
'linux' platform, that is parametrized by DEQP_SUPPORT_
(X11|GLX|XCB|WAYLAND) at compile time.

This new platform supports creating WSI displays and EGL contexts
for each of compositors enabled at build-time. To choose EGL compositor
use --deqp-egl-display-type=x11 or Wayland.

No functional change is made to 'wayland' target: it is still possible
to build Wayland-only binary using that target.

VK-GL-CTS issue: 83
Components: Framework
Affects: dEQP-VK.wsi.* and all tests using EGL.

Change-Id: I1d8725d1329e1071ac545f765c374d2d7d5a29df

Fix compute shader in robustness tests to produce an output

The compute shader only writes to shared variables, which means
that it doesn't produce any useful output. Shader compiler
backends can see this and mark all uniform variables declared
by the shader as "inactive" and remove them, which would defeat
the purpose of the 'getnuniform' test and lead to a failure.
Fix this by making the compute shader write to a shader storage
buffer object instead, so that uniforms remain active for being
involved in generating a shader output.

Component: OpenGL
VK-GL-CTS issue: 415

Affects:
KHR-GL45.robustness.getnuniform

Change-Id: I67b3fd8eb3a63012b6c46713e17c86d51b42d82e

Tie TCS XFB to Desktop GL + NV_gpu_shader5

This change ties the ability to read PATCHES data directly from a
tessellation control shader via transform feedback to the presence of
NV_gpu_shader5 and a desktop context. Any other use of transform
feedback from a tessellation control shader is considered an error.

As part of this change, we also revert a change that used the wrong
notation for consuming PATCHES data with transform feedback.

(Reverts part of 1ab254234c4f6c45036ba68f67980858e0ef5001)

Affects:

KHR-GLES31.core.tessellation_shader.single.xfb_captures_data_from_correct_stage
KHR-GLES32.core.tessellation_shader.single.xfb_captures_data_from_correct_stage
KHR-GL44.tessellation_shader.single.xfb_captures_data_from_correct_stage
KHR-GL45.tessellation_shader.single.xfb_captures_data_from_correct_stage

Component: OpenGL

VK-GL-CTS issue: 365

Change-Id: Ie5619f9146448c54906401d2805de676000b34e8

Remove copy_image special path for rgb9e5/rg11b10

Copy_image is similar to a memcpy. When analyzing copy_image result,
the test have a special path for rgb9e5 and rg11b10.
Instead of using a memcmp, it converts each component to float and
then compare values.

This could be problematic, when data from other formats is
reinterpreted as rgb9e5/rg11b10, the result can be INF or NAN.

Doing a float equal comparison on 2 NAN values will return false,
so we can have 2 identical 32-bits values that are considered
different by the test because they are reinterpreted as float.

This change remove the special path for rgb9e5 and rg11b10.
I don't think it is required since copy_image should be like a
memcpy, a memcmp should be enough to verify the result

Affects:

KHR-GL45.copy_image.functional

Components: OpenGL

VK-GL-CTS issue: 404

Change-Id: Ifc7f7b4c8d62692e9364510ebc4e4162468e9568

KHR-GL44.khr_debug: create texture before calling glObjectLabel

The test does not call glBindTexture after glGenTextures, so
texture object is not created.

This causes later glObjectLabel GL error checks to fail due to wrong
 GL error codes.

Affects: KHR-GL44.khr_debug.*
Components: OpenGL
VK-GL-CTS Issue: 410

Change-Id: I9036264b2c549c58a5a80fb1ec89e1fdaae4b6b7

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

Change-Id: Ia3866a3a0df54013e3f62f0b1263b9b5085bc548

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

Change-Id: I305e5144dee58d10e2284e1824d9d4683a9c67d8

Fix Vulkan null driver

Vulkan platform was changed to use vkGetInstanceProcAddr(NULL, pName)
for loading all platform-level functions, except vkGetInstanceProcAddr
itself, in commit bbd3dce44fafaa4b1952b63564022f19b3a65ca3. Null
driver was not adjusted for that change and that resulted calling into
null function pointer early in the initialization code.

Additionally, null driver now advertises HOST_COHERENT memory.

Change-Id: I65c8b37722950de08e337d7df95aa46a0e826c7b

Fix extension check in ES3.2 BoundingBox test

The test is still using the suffixed name to access gl_BoundingBox[EXT],
which means it still needs to enable the extension.

Affects: dEQP-GLES31.functional.android_extension_pack.shaders.es32.
extension_directive.ext_primitive_bounding_box

Components: AOSP

VK-GL-CTS issue: 423

Change-Id: I652e59b6ca29b1d237028eac57fc860731378c35
(cherry picked from commit abef7613781dfc4b07b8a37cd17ab5fe3dfe0c01)

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

Change-Id: I538752108415ebe9a68ca04289e2d3048083a77e

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

Change-Id: Icf7defc7279b1b0cf9f6b13498de3d1456239e9c

Update vulkan.h.in to 1.0.49 and strip unused exts

This change updates vulkan.h.in to match header version 49.
Vendor-specific extensions and KHX extensions that are not covered by CTS
are removed.

gen_framework.py now auto-detects function class from its first argument,
except for vkGet*ProcAddr, which require special handling.

Components: Vulkan

Change-Id: If6e1d959091a9064585bf29ef9983d03649ce513

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

Change-Id: I6f2d23322dbd2b82afcf6439c8f4c62891745a43

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

Change-Id: Iee00ded195e3178e685470b19df25a9b352ddfb5

Remove unused function from es31fSRGBDecodeTests.cpp
am: 1ee59ff986

Change-Id: I0bf132f2d1a09e66768bddc5818906aa0e066967

Remove unused function from es31fSRGBDecodeTests.cpp

Unused function causes compiler warnings to appear with recent GCC
versions.

Change-Id: I941c579994acbc9146df86acf5b9da9ee85341f5

Fix texture/sampler mapping in sRGB tests
am: c8efb45ead

Change-Id: I23849cbcae67248ac5ca55f443f226d6b1e4389e

Add tests for the VK_KHR_16bit_storage extension

* Four capabilities
  * StorageUniform16
  * StorageUniformBufferBlock16
  * StorageInputOutput16
  * StoragePushConstant16
* 16bit -> 32bit and 32bit -> 16bit
* integers and floats
* vectors and matrices of integers and floats

Component: Vulkan

Change-Id: I01f1273ce0e50f6dd3e8bfa8b13550bff0b6907d

Fix texture/sampler mapping in sRGB tests

sRGB decode tests assume that an API side texture with index i
is connected with a shader sampler variable uTexture with index i.

This relation was however incorrectly established, e.g.
the tests were mapping texture0 to a sampler with location 0, while
such sampler was not necessarily uTexture0.

This change fixes the mapping and makes the relation more obvious.

Components: AOSP
Affects:
dEQP-GLES31.functional.srgb_texture_decode.skip_decode.*.toggled
dEQP-GLES31.functional.srgb_texture_decode.skip_decode.*.conversion_gpu

Google bug: 38257749

Change-Id: I16508f6033b31583b1c5973b70a610dd59e78e8f

Added CTS_ARB_shader_ballot tests

Added CTS_ARB_shader_ballot cases:
ShaderBallotAvailabilityTestCase
ShaderBallotBitmasksTestCase
ShaderBallotFunctionBallotTestCase
ShaderBallotFunctionReadTestCase

affects:

GL45-CTS.shader_ballot_tests.*

components: OpenGL

VK-GL-CTS issue: 143

Change-Id: I28c724a84adab7e40edb542e70bc019cde0a0156

Fix VS2013 build by avoiding using non-standard C extension

Avoid using C++14 0b prefix.

Affects:
 * GL45-CTS.shader_atomic_counter_ops_tests.*

Components: OpenGL

VK-GL-CTS issue: 175

Change-Id: Ia500db76e1112933ad3471e9ac29423ccda81d4f

Fix VS2013 build by avoiding using non-standard C extension

Avoid using C++14 0b prefix.

Affects:
 * GL45-CTS.shader_atomic_counter_ops_tests.*

Components: OpenGL

VK-GL-CTS issue: 175

Change-Id: Ia500db76e1112933ad3471e9ac29423ccda81d4f

Added CTS_ARB_shader_group_vote tests

Added CTS_ARB_shader_group_vote cases:
ShaderGroupVoteAvailabilityTestCase
ShaderGroupVoteAllInvocationsTestCase
ShaderGroupVoteAnyInvocationTestCase
ShaderGroupVoteAllInvocationsEqualTestCase

affects:

GL45-CTS.shader_group_vote_tests.*

components: OpenGL

VK-GL-CTS issue: 211

Change-Id: I479b09277d8386c41b007075465f8ea383fb7f0a

resolve merge conflicts of 633ab6f3 to deqp-dev

Change-Id: I879838e31c3fc6bbceb102c9b2720a2372360c1a

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

Change-Id: Ic72d76a608170ff3297d112699914a9529d565bb

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

Change-Id: Icd63416e0133bca1c1580e3c36b950caa7b1bc23

Import fixes from Android Vulkan CTS 1.0.2 branch

 * Android copy of mustpass is updated to exclude waived tests.

 * Add missing file to Android.mk

 * Add -fwrapv and -Wno-delete-non-virtual-dtor to Android.mk

Change-Id: I9cdb0b0490dbad4134ad085216c3969cda40747a

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

Change-Id: Ib5a98bc9b3c1781a6d4a6ca5eaabe74aee42badf

Remove tcuWin32GLES3Platform.*

GLES3 wrapper is no longer supported so these are clearly broken.

Components: Framework

Change-Id: I656b70aecc5f0380d99b471a516d271f5b820836

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

Change-Id: I2d372c38240ffa856509c479b4f24b03ca95649c

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

Change-Id: Idd304bc2401d429f8a7153fef43c3af5169a5425

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

Change-Id: I94c771f530e058c9ac30185e2708fd0ac6fccf36

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

Change-Id: I11828b81ea4c9e67dc879dd98a9cbf91341de2c9

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

Change-Id: I6245604cf5ce44a2c4d9c064efb65b058d8f24eb

Add more alternatives to reflect() precision tests

Intel/Mesa started to generate these in some scenarios recently.

Components: Vulkan
Vk-GL-CTS Issue: 392

Affects:
dEQP-VK.glsl.builtin.precision.reflect.*

Change-Id: Ic943cf14d531e392c03adf18605e6330267be4a1

Check sampler_mirror_clamp_to_edge extension in tex filter tests

Components: Vulkan
VK-GL-CTS issue: 384

Affects:
dEQP-VK.texture.filtering.*.combinations.*_mirror_clamp_to_edge_*

Change-Id: Ie829b9247cd374e6c34ac70885e9e4421d284d1e

Do not report warning if OOM testing is incomplete

object_management.alloc_callback_fail tests are trying to exhaustively
test all OOM paths in vkCreate* functions. However, to keep test run-
time reasonable, they have a fixed maximum allocation count and if
that is exceeded report a warning since OOM testing was not
exhaustive.

Since an implementation that requires more allocations than the limit
is valid, this commit changes the tests to report clean Pass instead
of CompatibilityWarning. Note about incomplete OOM testing is included
in the test log contents instead.

Affects:

dEQP-VK.api.object_management.alloc_callback_fail.*

VK-GL-CTS issue: 399
Components: Vulkan

Change-Id: I6890321546ae42f2646b8653007172d4b8affb8a

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

Fix extension check in ES3.2 BoundingBox test

The test is still using the suffixed name to access gl_BoundingBox[EXT],
which means it still needs to enable the extension.

Affects: dEQP-GLES31.functional.android_extension_pack.shaders.es32.
extension_directive.ext_primitive_bounding_box

Components: AOSP

Change-Id: I652e59b6ca29b1d237028eac57fc860731378c35

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

Change-Id: I6d1646f1e77cdf1855b88abee9ab8e76df5af25c

Update Vulkan CTS to version 1.0.2.3

New tests:

 * Add tests for VK_KHR_shared_presentable_image
 * Add tests for VK_KHR_get_surface_capabilities2

Fixes:

 * makeFramebufferWithoutAttachments creates invalid framebuffer
 * Fix typo in dE-IT.build_info.de_endianness_consistent name
 * 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

Bug: 38260702
Change-Id: I1c00d4b63c339ba69990b64a5bb225c661957e4c

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

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

Change-Id: Ifb187a8bff7ec809e78fb53da5c84807e1569843

Check for INVALID_OPERATION from TextureStorage

When the effective target differs from the glTextureStorage* function
used, GL should generate an GL_INVALID_OPERATION error instead of
GL_INVALID_ENUM.

Affects:

KHR-GL45.direct_state_access.textures_storage_errors

Components: OpenGL

Issue: 407

Change-Id: I8cbfaeee7ce6bb7519cf2df9a255bfcfb3c84165

Ignore performance messages in KHR_debug tests

The KHR-GL45.khr_debug.receiveing_messages checks that only expected
messages are recorded so it should ignore spurious performance
messages that can be injected by drivers and mess up the expected
message counts.

Components: OpenGL
VK-GL-CTS issue: 394

Affects:
KHR-GL45.khr_debug.receiveing_messages

Change-Id: Iee24cb5d6de3d2c6311dc96da312540512dc2173

Fix texture_size_promotion snorm/RGB10_A2UI issue

Read color clamping was not considered in the snorm expected value,
so the change clamp the expected value

The alpha component in RGB10_A2UI didn't have enough bits to hold
an expected value of 4. So we change the expected alpha value to 1

Affects:

KHR-GL45.texture_size_promotion.functional

Components: OpenGL

VK-GL-CTS issue: 395

Change-Id: I920eb187e558faf9557f886be0269ea0c927918b

Adjust cull_distance position on pixel center

The primitive position is on the pixel edge, it should be on the
pixel center to be rendered at the expected location

Affects:

KHR-GL45.cull_distance.functional

Components: OpenGL

VK-GL-CTS issue: 396

Change-Id: I6f604c559894a4d9b1220a65935eb59b8b8e53e6

Fix DrawElements indice to use a NULL offset

We are using the address of an array as indice parameters
to a DrawElements call.

The address will be used as an offset to the element buffer,
and it's not what we want.

Use NULL instead

Affects:

KHR-GL45.enhanced_layouts.vertex_attrib_locations

Components: OpenGL

VK-GL-CTS issue: 397

Change-Id: Ic6be8925b990107e342f609726730ab51a33da9b

Use correct format for integer DSA buffer clears

Affects:

KHR-GL45.direct_state_access.buffers_clear

Components: OpenGL

VK-GL-CTS issue: 405

Change-Id: I86f2c463a5bcc86bd08188e42a285673ffb058d4

shader_integer_mix is unsupported prior to 4.5

This change adds logic to make sure shader_integer_mix.prototypes
with shader version prior to 4.5 is marked unsupported

It adds a new shaders45 group of tests to make sure core glsl 4.5
supports the extension

It adds the test shader_integer_mix.prototypes-negative to make sure
glsl versions that doesn't support the feature are returning an error

Finally it updates the mustpass list with the new tests

Affects:

KHR-GLES3.shaders.shader_integer_mix.prototypes-negative
KHR-GLES31.core.shader_integer_mix.prototypes-negative
KHR-GLES32.shaders.shader_integer_mix.prototypes-negative
KHR-GL33.shaders.shader_integer_mix.prototypes
KHR-GL33.shaders.shader_integer_mix.prototypes-negative
KHR-GL40.shaders.shader_integer_mix.prototypes
KHR-GL40.shaders.shader_integer_mix.prototypes-negative
KHR-GL41.shaders.shader_integer_mix.prototypes
KHR-GL41.shaders.shader_integer_mix.prototypes-negative
KHR-GL42.shaders.shader_integer_mix.prototypes
KHR-GL42.shaders.shader_integer_mix.prototypes-negative
KHR-GL43.shaders.shader_integer_mix.prototypes
KHR-GL43.shaders.shader_integer_mix.prototypes-negative
KHR-GL44.shaders.shader_integer_mix.prototypes
KHR-GL44.shaders.shader_integer_mix.prototypes-negative
KHR-GL45.shaders.shader_integer_mix.prototypes
KHR-GL45.shaders.shader_integer_mix.prototypes-negative
KHR-GL45.shaders45.shader_integer_mix.*

Components: OpenGL

VK-GL-CTS issue: 406

Change-Id: I574bec1a8d5aca0e4ec52b8df96abf45828b527a

ARB_texture_filter_minmax tests should check if the extension is supported

Component: OpenGL
Vk-GL-CTS issue: 409

Affects:
KHR-GL45.texture_filter_minmax_tests.*

Change-Id: Ic1e63db507afb7a25f30b3309ba06fce2889cb8a

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

Expand dEQP-VK.glsl.opaque_type_indexing shader types

This change expands dEQP-VK.glsl.opaque_type_indexing group to cover
geometry and tessellation shaders.

VK-GL-CTS issue: 357
Components: Vulkan

Change-Id: I3a89e3d2d59b826b078504cc6831254e839bb741

Correct the test conditions for sRGB_write_control

The extension is not part of ES3.2, so the not-supported version of the
tests should be run on ES3.2 implementations without the
GL_EXT_sRGB_write_control extension.

VK-GL-CTS Issue: 389
Components: AOSP

Affects:
dEQP-GLES31.functional.fbo.srgb_write_control.framebuffer_srgb_enabled
dEQP-GLES31.functional.fbo.srgb_write_control.framebuffer_srgb_unsupported_enum

Change-Id: I42a531aa7d1ad38f7a85cde1a5da659f780ec5d5

Adjust out of surface verification for wide lines

Wide lines rasterization tests may draw lines out of
the allocated surfaces. Such situation causes verification
failures.
This change adjust the verification algorithm to ignore
missing pixels for out of surface rendering.

Components: Framework

Affects:
dEQP-GLES2.functional.rasterization.primitives.lines_wide
dEQP-GLES2.functional.rasterization.primitives.line_strip_wide
dEQP-GLES2.functional.rasterization.primitives.line_loop_wide

Change-Id: Ia70f33fb6daad0d3774ae9edbd438bd14296f1d3

Ensure vertex shader presence if a geometry shader exists

Certain tests verify that INVALID_OPERATION is generated
if a geometry shader is active and mode is incompatible with
the input primitive type of the geometry shader in the currently
installed program object. However these tests have no vertex shader and
this is the real reason why INVALID_OPERATION is generated.
This change ensures that the vertex shader is present.

Components: AOSP

Affects:
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_elements_base_vertex_primitive_mode_mismatch
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_elements_instanced_base_vertex_primitive_mode_mismatch
dEQP-GLES31.functional.debug.negative_coverage.get_error.vertex_array.draw_range_elements_base_vertex_primitive_mode_mismatch

Change-Id: I36bcee8892e4f641feadb0b444a53ed35273b727