Fix GS atomics for gl4 buffer storage tests

GS could be invoked more than once for each input primitive.
This should be accounted for.

Components: OpenGL

Affects: KHR-GL4*.buffer_storage.map_persistent_draw

VK-GL-CTS issue: 3786

Change-Id: Ic42cd972b8cbc1ced62de0f33b7f831bc4ed8ab6

Merge vk-gl-cts/opengl-cts-4.6.0 into vk-gl-cts/opengl-cts-4.6.1

Change-Id: I02e16a1b389011005aebf279c6e689ca0a457ad5

Fix Zlib depedency

Use newer version of Zlib as older is no longer available to download from the zlib.net page

Components: Framework

Affects:

None

VK-GL-CTS Issue: 3587

Change-Id: Ie166dc897b0d1318c1ae73434bfa19687b579691
(cherry picked from commit ec1804831b654ac55bd2a7a5dd27a556afe05030)

fix GL46 vector construction sizing

the vector here is 4 rows of 2 components, but std::vector.size()
returns the total vector length (8), so divide by 2 to get the correct
length here and avoid out-of-bounds reads

Affects:

KHR-GL46.shaders30.glsl_constructors*

Components: OpenGL
VK-GL-CTS Issue: 3538

Change-Id: I8a07957722bf8b49fdcee14fe109d284f8cb9e27

Merge vk-gl-cts/opengl-cts-4.6.0 into vk-gl-cts/opengl-cts-4.6.1

Change-Id: I04ec68c36522b8082694e1f3f849ffbdafd34ea7

fix invalid buffer bind size in KHR-GL46.sparse_buffer_tests.BufferStorageTest

m_sparse_bo_data_size is the size of the sparse data, but this may exceed
the value returned by GL_MAX_UNIFORM_BLOCK_SIZE, which (should) cause
the test to fail on any driver where the sparse data size is larger than
the max uniform block size

instead, use the size of the data being passed to the shaders directly
so that this can never be too large

Affects:
KHR-GL46.sparse_buffer_tests.BufferStorageTest

Components: OpenGL
VK-GL-CTS Issue: 3468

Change-Id: Ie9d3de5760773e9b1b3b24e8af77976e7a54036c

Merge vk-gl-cts/opengl-cts-4.6.0 into vk-gl-cts/opengl-cts-4.6.1

Change-Id: I7ea3c530d47c2ebcc1c8c93b30f6e177ec5fb4e7

Fix compilation issues with ShaderLibraryCase assert

This assert was causing issues with some compilers/options.

Components: Framework, OpenGL

VK-GL-CTS issue: 3399

Fixes: a590d9ab4 ("Add check for ES compatibility extensions with Desktop OpenGL")
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Change-Id: I070d82e7e9c0a4aafbda21cabd18a2d37b5b2cf0

Query default FBO Depth, Stencil only when format requires

Earlier commit removed DSA usage but introduced a problem with
framebuffer parameter type and size queries in some configs.

Without DSA, we get errors if querying for attachment that does
not exist in framebuffer. Patch introduces a check to make sure
we do not query for separate depth/stencil attachment if that
would not exist in format.

VK-GL-CTS issue: 3092

Affects:
KHR-*.internalformat.renderbuffer.*

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Change-Id: I65ba7f74e5abd3a5bff17cd74ca4abcdf3f0ab53

Fix OpenGL implicit conversions tests

This change removes following tests:

   dEQP-GL45.functional.shaders.implicit_conversions.es31.*
   dEQP-GL45.functional.shaders.implicit_conversions.es32.*

The problem with these tests was generated data for the tests
required OpenGL ES extension. Instead we import new generated
data for gl45 and utilize that creating following tests for
OpenGL 4.5:

   dEQP-GL45.functional.shaders.implicit_conversions.gl45.*

Affects:
   dEQP-GL45.functional.shaders.implicit_conversions.es31.*
   dEQP-GL45.functional.shaders.implicit_conversions.es32.*
   dEQP-GL45.functional.shaders.implicit_conversions.gl45.*

Components: OpenGL
VK-GL-CTS issue: 3399

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Change-Id: Ib2b6bc35183f05a351b5a33e98a93ef0cdf7fdc6

Fix ARB_sparse_texture commit tests texture allocation

GL_NUM_SPARSE_LEVELS_ARB default value is 0, so the orginal code
will always set mState.levels to 0 which failed the following
glTexStorage* call.

In fact we can only know the meaningful GL_NUM_SPARSE_LEVELS_ARB
value after glTexStorage* which tells width/height/depth of the
texture. So check it before glTexStorage* is not right.

Affects:

KHR-GL46.sparse_texture_tests.SparseDSATextureCommitment
KHR-GL46.sparse_texture_tests.SparseTextureCommitment
KHR-GL46.sparse_texture2_tests.UncommittedRegionsAccess
KHR-GL46.sparse_texture2_tests.SparseTexture2Lookup

Components: OpenGL
Change-Id: I8ebedbb954ace56a49003d828179ac6225fe0b49

Add check for ES compatibility extensions with Desktop OpenGL

Affects:
   dEQP-GL45.functional.shaders.functions.es31.*
   dEQP-GL45.functional.shaders.arrays.es31.*
   dEQP-GL45.functional.shaders.implicit_conversions.es31.*
   dEQP-GL45.functional.shaders.uniform_block.es31.*

Components: Framework, OpenGL

VK-GL-CTS issue: 3399

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Change-Id: Ife97fee3d2e5c433050142e9ba9e330bfe3cbde9

Merge vk-gl-cts/opengl-cts-4.6.0 into vk-gl-cts/opengl-cts-4.6.1

Change-Id: I5dbe2d36f399ebb391471963192be42243fa8111

Improve max image samples check in KHR-GL46.multi_bind test

The test uses fillMSTexture method to fill a shader image
using imageStore()-s. However, the test does not check if
the number of samples in the image is less than MAX_IMAGE_SAMPLES
- what may lead to invalid access on store, when MAX_IMAGE_SAMPLES
is 1.

It looks like the test needs at least 2 samples in the multisampled
image, so this change switches the texture target to non-multisampled
when MAX_IMAGE_SAMPLES is less than 2.

According to the spec:
"""
Invalid image stores will have no effect. (Invalid image atomics will
not update any texture bound to the image unit and will return zero.(...)
An access is considered invalid if: (...)
• the image has more samples than the implementation-dependent value of
  MAX_IMAGE_SAMPLES.
"""

Components: OpenGL

VK-GL-CTS Issue: 3402

Affects: KHR-GL46.multi_bind.dispatch_bind_image_textures

Change-Id: I9915849f1cd3c708ae86deee40f1b94ec75d5fea

Enable the dEQP-GL45 package on Android

When the AOSP ES tests were added to the GL 4.5 mustpass packages
they were only added for non-Android, resulting in an empty configuration
when run on Android.

Affects:

dEQP-GL45.*

Components: OpenGL
VK-GL-CTS issue: 3314

Change-Id: I62e72cb507bf74cf5e267f35ae5e754bd8f63656

Remove direct state access usage from RenderbufferCase

Driver might not support GL_[ARB|EXT]_direct_state_access.

Affects:
KHR-GL*internalformat.renderbuffer.*

Components: OpenGL
VK-GL-CTS issue: 3092

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Change-Id: I89e50d51100fc45de159ce21191a38f3c4f2de49

Retry lesser FBO sample counts if UNSUPPORTED

On some of our implementations we return GL_FRAMEBUFFER_UNSUPPORTED
when the no-attachment maximum width, height and samples are all
used in combination. This is because the combination goes beyond
some of our hardware limits. It would be fine to use either
maximum width/height or maximum samples, but not all
together.

Since GL_FRAMEBUFFER_UNSUPPORTED is a valid FBO status for
implementation-defined reasons, the test should recognize this
and try again with lesser limits until the FBO is complete.

Affects:

dEQP-GL45.functional.fbo.no_attachments.maximums.all

Components: OpenGL

VK-GL-CTS issue: 2973

Change-Id: I8db2e9f802b3f4d1ee59bbfbfa1935bbe3cc6d00

Move two GL groups to single config

Move enhanced_layouts and arrays_of_arrays_gl to the single config group
since these tests do not depend on the default framebuffer

Components: OpenGL

VK-GL-CTS issue: 2969

Removed tests:
KHR-GL*.arrays_of_arrays_gl.*
KHR-GL*.enhanced_layouts.*

Added tests:
KHR-Single-GL*.arrays_of_arrays_gl.*
KHR-Single-GL*.enhanced_layouts.*

Change-Id: I2ae554c0d92c9959d71232052c3f177265a715a7

Change GLSL->SPIR-V mapping of notEqual to OpFUnordNotEqual not OpFOrdNotEqual

The glslang compiler was changed in https://github.com/KhronosGroup/glslang/pull/2260
to generate OpFUnordNotEqual rather than OpFOrdNotEqual for a!=b.

Without this change the
KHR-GL46.gl_spirv.spirv_glsl_to_spirv_builtin_functions_test
test generates an InternalError because it can't find the
expected mapping.

Affects:

KHR-GL46.gl_spirv.spirv_glsl_to_spirv_builtin_functions_test

Change-Id: Ieda8dc02c0be90a7985b15ca28dbefb1a45cc7c1
Components: OpenGL

Fix the default GL_IMAGE_BINDING_FORMAT for desktop OpenGL

The default texture format for desktop OpenGL is GL_R8 not GL_R32UI,
like it is for OpenGL ES.

See table 23.45 on page 627 of the OpenGL 4.6 core spec vs
table 21.33 on page 479 of the OpenGL ES 3.2 spec.

Affects:

KHR-GL46.shader_image_load_store.*

Change-Id: I1f5f8be422ddb358fba6df6bc450d3da702467bc
Components: OpenGL

Check resource limits properly

The tests were not checking the resource limits per stage when
creating the shaders, which could create potential issues on
some drivers. Specially the random generated tests.

Affected tests:

   dEQP-VK.ssbo.layout.*

Components: Vulkan
VK-GL-CTS issue: 2953

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Change-Id: I7509dce38b849a0dd26f9e47d7f74e637930d897

Capture early errors on instance creation

This commit makes CTS pass a VkDebugReportCallbackCreateInfoEXT
structure in the pNext chain of VkInstanceCreateInfo when creating the
default and other custom instances if validation is enabled from the
command line.

To achieve this, the DebugReportRecorder class has been uncoupled from
the instance and it's no longer responsible of owning the debug report
callback object (VkDebugReportCallbackEXT) that can be created using an
instance. That responsibility now falls into the default device or
custom instance wrapping objects.

This makes it possible to use the DebugReportRecorder in a
VkDebugReportCallbackCreateInfoEXT structure chained when creating
instances before they actually exist, and errors reported during the
instance creation process are handled by the recorder together with
other errors like validation errors.

In addition, the enumerate_devices_alloc_leak test has been modified to
use a custom instance (so it has validation enabled if requested) and
the API tooling info tests have been changed to use the custom instance
creation wrapper function createCustomInstanceFromContext().

Custom instance creation has been improved to detect the lack of
installed validation layers in order to avoid crashes.

Due to these changes affecting the default device and how validation is
enabled when requested, all tests are marked as affected.

Affected tests:
dEQP-VK.*

Components: Vulkan
VK-GL-CTS issue: 2952

Change-Id: I27b78e3cb101142416477ee2460283f917f58a4e

Add depth execution mode tests

This CL adds tests for DepthLess, DepthGreater and DepthUnchanged
execution modes.

New tests:
dEQP-VK.spirv_assembly.instruction.graphics.execution_mode.*

Components: Vulkan

VK-GL-CTS issue: 2735

Change-Id: Ib17d8f04b1cd254b85dfad2c779cb84b5e9fde9d

Merge vk-gl-cts/opengl-es-cts-3.2.7 into vk-gl-cts/master

Change-Id: I5de40715011c884f0da041ed927dd418549e7685

Merge vk-gl-cts/opengl-es-cts-3.2.6 into vk-gl-cts/opengl-es-cts-3.2.7

Change-Id: Ib0f4c996967d629009113c3d17b543fc9218def0

Test instance matrix update after command recording

This commit modifies ray query and ray tracing direction length tests so
the instance transform matrix in the top level acceleration structure is
sometimes updated after the AS build commands have been recorded in the
command buffer.

Affected tests:
dEQP-VK.ray_tracing_pipeline.direction_length.*
dEQP-VK.ray_tracing_pipeline.inside_aabbs.*
dEQP-VK.ray_query.direction_length.*
dEQP-VK.ray_query.inside_aabbs.*

Components: Vulkan
VK-GL-CTS issue: 2881

Change-Id: I9051783443e3775e20ccaa0eab32c5bc2f21d165

Test null miss shader group handles work

New test that creates a miss shader binding table filled with zeros and
checks everything works as expected.

New tests:
dEQP-VK.ray_tracing_pipeline.misc.null_miss

Components: Vulkan
VK-GL-CTS issue: 2929

Change-Id: Iead899ff0e6371a77d102901c8a8cc7583de49c8

Test tracing rays from inside AABBs

Test tracing rays from inside AABBs with different Tmin and Tmax values,
origins and directions to verify they always generate a hit due to the
ray origin being located inside the AABB.

New tests:
dEQP-VK.ray_query.inside_aabbs.*
dEQP-VK.ray_tracing_pipeline.inside_aabbs.*

Components: Vulkan
VK-GL-CTS issue: 2928

Change-Id: I6b0a7c3a3bbe68cd2bc3b670c6dabcfa4374d473

Fix wait semaphore count value

Components: Vulkan
VK-GL-CTS issue: 2949
Fixes: a47665a6e "Generalize synchronization tests"

Affected tests:

   dEQP-VK.synchronization.op.single_queue.binary_semaphore.*

Change-Id: I14bb0d79b67e444b20683a766b769ce3a94a3222
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>

Fix internal error for fragment stage in RT tests

Previous CL added missing vertex pipeline stores check,
but forgot to do a passthough for fragment stage, which
is causing an internal error.

Affected tests:
dEQP-VK.ray_query.watertightness.*.frag.*
dEQP-VK.ray_query.builtin.*.frag.*
dEQP-VK.ray_query.advanced.*.frag.*
dEQP-VK.binding_model.descriptor_update.acceleration_structure.ray_query.*frag*

Components: Vulkan
VK-GL-CTS issue: 2948

Change-Id: Id8b7a6c13bf8acda3b0fd58e2afd3f61709ed295

Add GL_ARB_separate_shader_objects requirement

Components: OpenGL

Affects: KHR*.shader_subroutine.subroutines_with_separate_shader_objects

VK-GL-CTS public issue: 271

Change-Id: I021776f105ec2ddf310deaf21827e5ce151784af

Raise glsl version to 420 for compute shaders

Components: OpenGL

Affects:
KHR*.pipeline_statistics_query_tests_ARB.*
KHR*.cull_distance.*
KHR*.texture_gather.*

VK-GL-CTS public issue: 257

Change-Id: Ic777e7a821cbab9e6b71e266591352aea2fdd81d

Raise glsl version to 410 core for shaders with GL_ARB_shader_atomic_counters

Components: OpenGL

Affects: KHR*.shader_subroutine.*

VK-GL-CTS public issue: 270

Change-Id: Ifed270991a1b18fb88514d6f0b96f0802498fb56

Remove shaders.switch from the Khronos mustpasses

Components: OpenGL

VK-GL-CTS issue: 2540

Affects: KHR*.shaders.switch.*

Change-Id: I41867656e67adc5077527a6d78791b06cee52c2f

Add a test for VK_EXT_global_priority_query

Components: Vulkan

VK-GL-CTS Issue: 2840

New tests: dEQP-VK.api.device_init.create_device_global_priority_query

Change-Id: I5245711e6db62f5fac06c146e868dfa4e88ab918
(cherry picked from commit 9c2b7bd961ac5f0394011930be9a4b58be37f217)

Update Vulkan headers

Components: Vulkan

Change-Id: I357b284202a389ca269891938630e95edab57124

Fix validation errors in ASTC decode mode tests

The tests were trying to compare the raw contents of each ASTC
block contents between the tested and reference images via UINT and
SINT image views, but the type of the image view must be compatible
with the image format.

This CL changes the test to compare individual sampled texel
values instead of raw ASTC texel block values. The generated
images now contain valid ASTC blocks instead of random values.

Affects:
dEQP-VK.image.astc_decode_mode.*

Components: Vulkan

VK-GL-CTS issue: 2820

Change-Id: I38113725692f754c5de4393b5fad5d333cf99533

Merge vk-gl-cts/vulkan-cts-1.2.6 into vk-gl-cts/master

Change-Id: I070c88e5936fddc054050484c28fd892c6fb8a3a

Merge vk-gl-cts/vulkan-cts-1.2.5 into vk-gl-cts/vulkan-cts-1.2.6

Change-Id: Id8117502d12cca36e3a414da3df7b8a3bb661cf5

Static code analysis fix for bounding box test

After closer analysis turns out the mistake in the code
was using xmin twice; changed to use xmin and xmax.

Affects:
dEQP-GLES31.functional.primitive_bounding_box.call_order.*

Change-Id: I307126c3558e4b8700d3ed808e24c7d85bcb50dc
Components: OpenGL

Add timer query handle reuse test for opengl es

This change adds base class for disjoint_timer_query tests
as well as test for handle reuse which was ported from the
GL4+ version.

New test: KHR-GLESEXT.disjoint_timer_query.handle_reuse

Components: OpenGL
VK-GL-CTS issue: 2911

Change-Id: I44421813a5de510e52fb6e1109fb2863856351c6

Invalid VkPipelineRasterizationLineStateCreateInfoEXT chained

Some tests get the structure empty (and implementations most
likely ignore it), other tests get the structure with fields
inadequate to the test.

Update tests:
 * dEQP-VK.rasterization.*

Components: Vulkan

VK-GL-CTS issue: 2942

Change-Id: Ie485ad55266a53e64c74eab3db31a51d781c9d57

Fix vertex count calculation in XFB query tests

The primitive to vertex count calculation for
VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY was
`(primCount + 4u) * 2u` when it should be `primCount * 2u + 4u`.

This change affects the bufferSize parameter passed to the tests.
The vertex count returned by this function is part of the test name
so the affected test cases are also renamed.

Affects:
dEQP-VK.transform_feedback.simple.query*
dEQP-VK.transform_feedback.simple.host_query_reset*

Components: Vulkan

VK-GL-CTS issue: 2934

Change-Id: I59e95a3240c6c45f8caf0dd4dc5b61c593059304

Set the recursion depth for chit and miss cases

In the new dEQP-VK.binding_model.descriptor_update.acceleration_structure.ray_tracing.*
group of tests the "chit" and "miss" variants the shaders
generate additional rays, which means their recursion depth needs to be
greater than 1. In both cases "2" covers it.

Affects:

dEQP-VK.binding_model.descriptor_update.acceleration_structure.*

Components: Vulkan

VK-GL-CTS issue: 2924

Change-Id: I7acb1eda10f0ad6c7f2754deaa8451c317b4841b

Add depth-stencil tests that do vkCmdCopyImage() with both image aspects

Added tests:

   dEQP-VK.api.copy_and_blit.*_depth_stencil_aspects*

Components: Vulkan
VK-GL-CTS issue: 2926

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Change-Id: I9673674035d7a3fbd1947819305ea9dcb8dede6a

Test for fragment shading rate in FragCoord.zw

Expansion of cases in pixel consistency test with varying shading rate
applied using zw fragment coordinates.

VK-GL-CTS issue: 2850

Components: Vulkan

Affects:
dEQP-VK.fragment_shading_rate.pixel_consistency.*

Change-Id: I8ba67a603a8b8b3b53f85207b5adae3bbc9ba51b

Add tests for vkCmdCopyQueryPoolResults and pipeline statistics

Added tests:

   dEQP-VK.query_pool.statistics_query.*cmdcopyquerypoolresults*

Modified tests:

   dEQP-VK.query_pool.statistics_query.multiple_queries.*

Renamed tests:
   dEQP-VK.query_pool.statistics_query.*dstoffset* to
   dEQP-VK.query_pool.statistics_query.*dstoffset_cmdcopyquerypoolresults*

Components: Vulkan
VK-GL-CTS issue: 2656

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Change-Id: I9cb6f4671f772834ca43f24a342323a42266c581

Update glslang, SPIR-V Tools/Headers

Components: Framework

Change-Id: I88d7d3fb789c667261db76b143979cb7e7a87f30

Check return code for vkGetAndroidHardwareBufferPropertiesANDROID

Components: Vulkan

VK-GL-CTS Issue: 2944

Affects: dEQP-VK.api.external.memory.android_hardware_buffer.*

Change-Id: I2da2fe29ae34bd78c3bf2d4e37b002ef58828c12

Add script for comparing dEQP SwiftShader runs

This submit adds a new script to vk-gl-cts/scripts that checks
for differences/regression in two different deqp-vk runs using regres and SwiftShader.
The script outputs a log of found differences from
the two different runs.

Change-Id: I9bfcc74e581c939a7b541fb7bb2a7141f0b9aeed

Merge vk-gl-cts/github-master into vk-gl-cts/master

Change-Id: Ia2f1289fc2d08e1238e0278067d1cf884e1b8287

Merge pull request #258 from asimiklit/fix/arrayed-tests

Disable explicit location tests for `arrayed` block members

Merge vk-gl-cts/vulkan-cts-1.2.6 into vk-gl-cts/master

Change-Id: I7627d6a95d9d8fccff9da037851bd3f11eddc7b5

Update OpenGL mustpass

Components: OpenGL

Change-Id: I0ecc800bfa213c4567354afc5c5cc7c0b96dd00b

Fix several issues in swapchain tests

This commit fixes the following issues found in the swapchain tests:

- The initial image layout in some render passes did not match the
  actual image layout when starting the render pass.

- image_swapchain_create_info tests were using VK_KHR_bind_memory2
  functions without activating the extension on the device or making
  sure it was core in the used Vulkan API version.

- Many devices were being created with the wrong instance, using the
  context instance instead of the custom instance created for the tests.

- Several tests using VK_SHARING_MODE_CONCURRENT did not check for the
  availability of two queues or were always using one, which is invalid.

- VkDeviceGroupSwapchainCreateInfoKHR had a wrong sType.

- Swapchain resize tests were calling vkBeginCommandBuffer() on active
  command buffers without waiting for them to complete execution.

- Typos and other minor issues.

Affected tests:
dEQP-VK.wsi.*.swapchain.*

Components: Vulkan
VK-GL-CTS issue: 2941

Change-Id: I54d5f46cbcb76617260439a59bf3aef4028309bb

Add AOSP tests to the GL4.5 mustpass

Components: OpenGL

Change-Id: I947a22b7e1f74f343e3b9eed2866a122756ef260

Merge vk-gl-cts/vulkan-cts-1.2.5 into vk-gl-cts/vulkan-cts-1.2.6

Change-Id: Ia66e6fb7a5444ffd1f1a5f14ef456c5129e0fd8f

Fix surfaceless platform for Android

Fix Vulkan library path on Android and fix config selection when
glu::RenderConfig::DONT_CARE is used.

Affected tests:
dEQP-VK.*

Components: Framework
Change-Id: I8809dd691e144223991a693aa9b90c03cd829a63

Check supportedAlpha is never zero

According to the spec, the value of supportedAlpha must contain at least
one valid VkDisplayPlaneAlphaFlagBitsKHR bit.

Affected tests:
dEQP-VK.wsi.display.get_display_plane_capabilities

Components: Vulkan
VK-GL-CTS issue: 2718

Change-Id: I4bdba857c8571d38a41bf8d69b10acc2bb5065ee

Do not clamp ambiguous snorms values on vkCmdCopy*() functions

As per Vulkan WG agreed, these functions perform a bit-for-bit copy
of the values, even if the result is unrepresentable.

Reverts:

* "Fixes sparse resource SNORM tests"
  -> 00e745999e771df2a604bf24d85594a69ad441fe

* "copies_and_blitting: fix ambiguous signed normalized 1.0 value"
  -> bce4ec47056a0473ccce9cf7951023da80c3b745

Affects:
dEQP-VK.api.copy_and_blit.*.image_to_image.all_formats.color.*.r32_sfloat.a8b8g8r8_snorm_pack32*
dEQP-VK.api.copy_and_blit.*.image_to_image.all_formats.color.*.r32_sfloat.r8g8b8a8_snorm.*
dEQP-VK.sparse_resources.image_sparse_binding.*snorm*
dEQP-VK.sparse_resources.mipmap_sparse_residency.*snorm*

VK-GL-CTS issue: 2917
Componentes: Vulkan

Change-Id: I5878379aa78e68ac7452a83c1a64abf01c9472c8
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>

Add writing to invalid multisample indices tests

In these new tests, we check that writes to invalid sample indices are
correctly discarded.

Components: Vulkan

VK-GL-CTS Issue: 2310

New tests:
dEQP-VK.texture.multisample.invalid_sample_index.*

Change-Id: I1feb6f8d44c6cdc526eb5b3ab1d045cd3a181b17

Add GL_EXT_texture_sRGB_RG8 support

Affects:

dEQP-GLES3.functional.texture.format.sized.*.srgb_rg8*
dEQP-GLES3.functional.texture.filtering.*.formats.srgb_rg8*
dEQP-GLES31.functional.texture.filtering.cube_array.formats.srg8*
dEQP-GLES31.functional.texture.format.sized.cube_array.srgb_rg8*
dEQP-GLES31.functional.srgb_texture_decode.skip_decode.srg8.*

Components: OpenGL, AOSP

Change-Id: I5194f06dbfedc375ade3e8b4fc2e84aa65868df1

Add a batch of GraphicsFuzz coverage tests

This commit adds a batch of GraphicsFuzz coverage tests.

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.cov-cast-float-to-int-and-back
dEQP-VK.graphicsfuzz.cov-clamp-vector-element-ceil-negative
dEQP-VK.graphicsfuzz.cov-clear-yz-inside-condition
dEQP-VK.graphicsfuzz.cov-conditional-discard-inside-loop
dEQP-VK.graphicsfuzz.cov-continue-break-discard-return-in-loop
dEQP-VK.graphicsfuzz.cov-cosh-clamped-to-one
dEQP-VK.graphicsfuzz.cov-double-if-true-in-loop
dEQP-VK.graphicsfuzz.cov-for-loop-min-increment-array-element
dEQP-VK.graphicsfuzz.cov-if-true-discard-in-do-while-never-reached
dEQP-VK.graphicsfuzz.cov-increment-float-in-loop-abs
dEQP-VK.graphicsfuzz.cov-increment-vector-component-with-matrix-copy
dEQP-VK.graphicsfuzz.cov-matching-conditions-break
dEQP-VK.graphicsfuzz.cov-matching-if-always-true-inside-loop
dEQP-VK.graphicsfuzz.cov-rcp-negative-int
dEQP-VK.graphicsfuzz.cov-return-after-first-iteration

Change-Id: I0dd986b855b4ce4c14fe653cfb069f1d038f6cdc

Add a glMemoryBarrier between buffer write and read

The tests' behavior is like this:
1. Use transform feedback to write to one buffer
2. Set this buffer to texture, and then to image
3. glDispatchCompute -> use ImageLoad to read from the image
We have to add a glMemoryBarrier between step 1 and 3.

Affects:
KHR-GLES3*.core.texture_buffer.texture_buffer_operations_transform_feedback

Components: OpenGL

VK-GL-CTS issue: 2936

Change-Id: If711469a6cbf4ae1df6851e4006d937760e37ca7

Add permutations of different cases on existing test

A test for copying a partial image with npot dimensions,
differing format from source to destination and clearing
of image was previously added. This commit adds different variations
of said test.

New tests:

dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_pot_diff_format_clear
dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_npot_diff_format_noclear
dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_pot_diff_format_noclear
dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_npot_same_format_clear
dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_pot_same_format_clear
dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_npot_same_format_noclear
dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_pot_same_format_noclear

Affects:

dEQP-VK.api.copy_and_blit.copy_commands2.image_to_image.simple_tests.partial_image_npot_diff_format_clear

Components: Vulkan

VK-GL-CTS issue: 2683

Change-Id: Ib82339d66e451e34a633e45f50ed8981db21731a

Add multiple statistics query tests

Add tests for different query operations:

* Multiple statistics query flags enabled but only input assembly
  vertices and primitives together with vertex or fragment shader
  invocations.
* Partial queries.
* Add tests with/without wait flag.
* Getting query values with vkCmdCopyQueryPoolResults() and
vkGetQueryPoolResults()
  - Usage of dstOffset in vkCmdCopyQueryPoolResults().

Components: vulkan, framework
VK-GL-CTS issue: 2909

Added tests:

   dEQP-VK.query_pool.statistics_query.multiple_queries.*

Change-Id: I2dd5412e20268c7997a870dd33a1f3c0a1779600
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>

Use arrays of pointers with non-identity matrices

Ray tracing tests were lacking coverage for using non-identity matrices
while instanceData.arrayOfPointers is set to true. This commit modifies
ray direction tests, notable for using non-identity matrices, so every
other case uses arrays of pointers for both instance data and the
ppGeometries build geometry info structure when creating the top level
acceleration structure.

Affected tests:
dEQP-VK.ray_tracing_pipeline.direction_length.*
dEQP-VK.ray_query.direction_length.*

Components: Vulkan
VK-GL-CTS issue: 2881

Change-Id: I6464673a3d62f2b77ae3ad428de09d30be123ea7

Remove empty subtest

Removed subtest function definitions and empty (always passing,
test not run) implementation for integer_constant_expression.

Affects:
KHR-GLES31.core.layout_binding.sampler2D_layout_binding_texture_FragmentShader

Components: OpenGL, OpenGL ES
VK-GL-CTS issue: 2232

Change-Id: I42fa99f09895d00adeb25123e26c4dbaefeb498f

Lower vector constructor test shader version

Changed shaders in GLSL vector constructor tests
to require version tag matching the OpenGL 3.0
specification.

Affects:

KHR-GL30.shaders30.glsl_constructors.*

Components: OpenGL

VK-GL-CTS issue: 2892

Change-Id: I2fdff3de489bfa20115270e542104d06aeb65b0f

Test framebuffer completeness when layered rendering

Added OpenGL ES tests for rendering to layered (cube map) framebuffer.

VK-GL-CTS issue: 2801

Components: OpenGL
Change-Id: I1cede4806af33520612b7335d553df7da119520e

Merge vk-gl-cts/vulkan-cts-1.2.6 into vk-gl-cts/master

Change-Id: Ie5c97143a71df2572ed345e6bddf2937e81e4573

Merge vk-gl-cts/vulkan-cts-1.2.5 into vk-gl-cts/vulkan-cts-1.2.6

Change-Id: I693041c49bff85c924472119584f938c958ea7eb

Allow false positive hits from build-in AABBs intersector

This extends the fix from 7ff1f9cd to 'mixed' tests, which
also use AABBs.

This change allows the test to execute any hit shader,
in case of AABB geometry is used and the ray falsely 'intersects'
it.

This is needed, as the test does not calculate the proper
intersection in 'intersection' stage - it uses a shader that always
reports TRUE. Hence the test relies on build-in AABB
intersector, which may report false positives according to the
specification.

VK-GL-CTS Issue: 2848

Components: Vulkan

Affects: dEQP-VK.ray_tracing_pipeline.build.*mix*

Change-Id: Ia962992df866b68f84714dcb6429a1a90499d9b6

Merge vk-gl-cts/vulkan-cts-1.2.4 into vk-gl-cts/vulkan-cts-1.2.5

Change-Id: I5a38303ad269d39786584512a105e0e4b636eca1

Merge vk-gl-cts/vulkan-cts-1.2.3 into vk-gl-cts/vulkan-cts-1.2.4

Change-Id: I2182f671aa88657e150c9d987713b93920377b23

Merge vk-gl-cts/vulkan-cts-1.2.2 into vk-gl-cts/vulkan-cts-1.2.3

Change-Id: I6a05c3d94da3f1f6397b029997d6d43db2439548

Merge vk-gl-cts/vulkan-cts-1.2.1 into vk-gl-cts/vulkan-cts-1.2.2

Change-Id: Ib9e1ec8639e95a5bda7c361e1db102b5f4f94194

Fixes ycbcr copy tests using NaNs

Removes NaNs from the source images

Components: Vulkan
VK-GL-CTS issue: 2772

Affects:
dEQP-VK.ycbcr.copy.*

Change-Id: I52d279633e081ecce0ef93681c4d199296de6107

Merge vk-gl-cts/vulkan-cts-1.2.6 into vk-gl-cts/master

Change-Id: I8f74d90e51fc450a85239f6c65fcc792e8e370b6

Merge vk-gl-cts/vulkan-cts-1.2.5 into vk-gl-cts/vulkan-cts-1.2.6

Change-Id: If63af4526f72fc1f214b60c4c9d891718bf2b440

Merge vk-gl-cts/vulkan-cts-1.2.4 into vk-gl-cts/vulkan-cts-1.2.5

Change-Id: I768331416f61a0a76c6bff504e0fedd545080f22

Merge vk-gl-cts/vulkan-cts-1.2.3 into vk-gl-cts/vulkan-cts-1.2.4

Change-Id: I9ac48fcdfe6e253139ccde50371b9dbfa31ce437

Merge vk-gl-cts/vulkan-cts-1.2.2 into vk-gl-cts/vulkan-cts-1.2.3

Change-Id: I9a085c6867ee119951a839b0d5eecb7fcf351029

Merge vk-gl-cts/vulkan-cts-1.2.1 into vk-gl-cts/vulkan-cts-1.2.2

Change-Id: I575ea750c7a89ec9f8306537a4bc3f7fbb64a619

Fix precision issue in ShaderMatrixTest

According to the Spec, the compiler is free to promote computation
precisions, and such promotion depends on the specific shader
content or compiler.

When tests use the lowp or mediump precision in the shaders, It
can't confirm the final computation precision. Now the precision
of the shader inputs is highp, if the real computing precision in
GPU HW is lowp or mediump. It can easily get the comparison failure.
So it's better to adjust the precision of the inputs as the expected
precision in the specific tests.

Affects:

dEQP-GLES3.functional.shaders.matrix.*

Components: OpenGL

VK-GL-CTS issue: 2863

Change-Id: I30af699da12bf88596e7046922b2f1f00d5d3f06

Remove unused check

This check is useless here. Remove it.

Also a year later there was a specification issue #1222 "Should
requiresDedicatedAllocation imply requiresDedicatedAllocation?"

That clarified this question and specification now contains
following paragraph totally invalidating this check:

When the implementation sets requiresDedicatedAllocation to VK_TRUE,
it must also set prefersDedicatedAllocation to VK_TRUE.

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

Components: Vulkan

VK-GL-CTS issue: 391

VK-GL-CTS public issue: 267

Change-Id: I557d3c0fac752ec2f42db7e6dcd78787a0c69a26

Add a batch of GraphicsFuzz coverage tests

This commit adds a batch of GraphicsFuzz coverage tests.

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.cov-atan-trunc-vec4
dEQP-VK.graphicsfuzz.cov-bitwise-inverse-uniform-condition
dEQP-VK.graphicsfuzz.cov-clamp-loop-limit-increment-float-array
dEQP-VK.graphicsfuzz.cov-clamp-lower-limit-from-always-false
dEQP-VK.graphicsfuzz.cov-function-vec2-never-discard
dEQP-VK.graphicsfuzz.cov-global-loop-counter-float-accumulate-matrix
dEQP-VK.graphicsfuzz.cov-int-div-round-to-zero
dEQP-VK.graphicsfuzz.cov-loop-abs-multiply-offset
dEQP-VK.graphicsfuzz.cov-loop-increment-or-divide-by-loop-index
dEQP-VK.graphicsfuzz.cov-matrix-double-transpose
dEQP-VK.graphicsfuzz.cov-nested-loops-global-loop-counter-do-while-accumulate-float
dEQP-VK.graphicsfuzz.cov-set-output-color-function-call-nested-loop
dEQP-VK.graphicsfuzz.cov-texel-double-negation
dEQP-VK.graphicsfuzz.cov-unused-matrix-copy-inside-loop
dEQP-VK.graphicsfuzz.cov-vector-illegal-index-never-executed

Change-Id: I26425024458bb5e080647194e90a87cc7e8512d8

Merge vk-gl-cts/opengl-es-cts-3.2.7 into vk-gl-cts/master

Change-Id: I37c6ded86ee2b7242e2146185b7683147fcdf234

Add XFB tests for VK_EXT_provoking_vertex

Test that the positions of provoking vertices are preserved in
transform feedback when transformFeedbackPreservesProvokingVertex
is advertised.

New tests:
dEQP-VK.rasterization.provoking_vertex.transform_feedback.*

Affects:
dEQP-VK.rasterization.provoking_vertex.*

Components: Vulkan

VK-GL-CTS issue: 2615

Change-Id: Id9cc1884487d204f1993aab0c8a0f618f32eb7c4
(cherry picked from commit 16663f3d6416027a00decc9810a4cfbd1d0457eb)

Add tests for VK_EXT_provoking_vertex

Replace VkScript-based provoking vertex tests with one file that tests
both the default Vulkan provoking vertex convention and different
provoking vertex modes provided by VK_EXT_provoking_vertex.

New tests:
dEQP-VK.rasterization.provoking_vertex.default.*
dEQP-VK.rasterization.provoking_vertex.first.*
dEQP-VK.rasterization.provoking_vertex.last.*
dEQP-VK.rasterization.provoking_vertex.per_pipeline.*

Affects:
dEQP-VK.rasterization.provoking_vertex.*

Components: Vulkan

VK-GL-CTS issue: 2092

Change-Id: I472c8c15ca1c0ad51a2dcc16ef9fa3d5f6ba5fa7
(cherry picked from commit 2d23e0e1a180c4dc8b5bcde8a4713527b7ef50b9)

Static code analysis fixes for GL cts

This change contains multiple static code analysis fixes,
including uninitialized member variables, initialization
order and buffer overruns.

In theory these changes shouldn't change the tests, but
due to the nature of the changes, some behavior changes
may occur.

Due to the large number of files changed, the affects
mask is over-wide.

Affects:
dEQP-GL*
KHR-*

Change-Id: I57518c38fca9775691f02edfc0cbeee28294bc1e
Components: Framework, OpenGL

Add tests for multidimensional atomic counter array

VK-GL-CTS Issue: 1916

New tests:

KHR-GLES31.core.shader_atomic_counters.advanced-usage-multidim-array*

Components: OpenGL ES
Change-Id: Ic3013a5be424238afca0f20e16ef94d3076acfdb

Revert "Remove too long ray_tracing.build tests"

This reverts commit 6e14c25f5d049cb0dadebdd8a48392721b8a4d0b.

This reinstates long ray_tracing.build tests

VK-GL-CTS Issue: 2896

Components: Vulkan

Change-Id: I1b0ecaa3eeb63f5ca82d7bffe790979e21d08b7c
Affects: dEQP-VK.ray_tracing_pipeline.build.*

Add missing vertex pipeline stores check with ray queries

Several tests related to ray queries were performing image stores from
vertex pipeline shader stages without checking support for vertex
pipeline stores and atomics support.

Affected tests:
dEQP-VK.ray_query.watertightness.*.vert.*
dEQP-VK.ray_query.watertightness.*.tesc.*
dEQP-VK.ray_query.watertightness.*.tese.*
dEQP-VK.ray_query.watertightness.*.geom.*
dEQP-VK.ray_query.builtin.*.vert.*
dEQP-VK.ray_query.builtin.*.tesc.*
dEQP-VK.ray_query.builtin.*.tese.*
dEQP-VK.ray_query.builtin.*.geom.*
dEQP-VK.ray_query.advanced.*.vert.*
dEQP-VK.ray_query.advanced.*.tesc.*
dEQP-VK.ray_query.advanced.*.tese.*
dEQP-VK.ray_query.advanced.*.geom.*
dEQP-VK.binding_model.descriptor_update.acceleration_structure.ray_query.*vert*
dEQP-VK.binding_model.descriptor_update.acceleration_structure.ray_query.*tesc*
dEQP-VK.binding_model.descriptor_update.acceleration_structure.ray_query.*tese*
dEQP-VK.binding_model.descriptor_update.acceleration_structure.ray_query.*geom*

Components: Vulkan
VK-GL-CTS issue: 2907

Change-Id: I1d13c227bb52dfad393c9b52681083598bff2f80

Add vkCmdCopyQueryPoolResults() tests with dstOffset != 0

Components: Vulkan
VK-GL-CTS issue: 2909

Added tests:

   dEQP-VK.query_pool.*dstoffset*

Change-Id: Ib8b7173a58f370f0f7bb48a5d6a7de07621be096

VK_VALVE_mutable_descriptor_type tests

This commit adds a new group of tests to check the
VK_VALVE_mutable_descriptor_type extension. The extension allows
applications to use a new type of descriptor that doesn't have a fixed
type and can mutate among several actual descriptor types.

In addition, the extension also allows creating descriptor pools and
sets that reside entirely in host memory. They can be used as the source
for descriptor copies but cannot be bound in a command buffer.

The new tests mix mutable descriptors with nonmutable ones, arrays of
them, descriptor writes, descriptor copies, host-only and normal pool
and sets, update-after-bind, different shader stages, etc.

New tests:
dEQP-VK.binding_model.mutable_descriptor.*

Components: Framework, Vulkan
VK-GL-CTS issue: 2852

Change-Id: I19b00c2e0f56ce547a615cb39d50965f513c74db

Add new multi-planar format tests

Adds new multi-planar format tests with per-plane views using compatible
formats taken from the table "Compatible Formats".

New tests:
dEQP-VK.ycbcr.plane_view.*_compatible_format_*

Affects:
dEQP-VK.ycbcr.plane_view.*

Components: Vulkan

VK-GL-CTS issue: 1630

Change-Id: Ib9a75e77990fe5c817a20b605a265f5ad326597a

Update Vulkan headers

Components: Vulkan

Change-Id: I0256daab1e697718a061db6eff68eb01a1cb6196

Disable explicit location tests for `arrayed` block members

According to the OpenGL 4.5 and Open GLES 3.2 specs these tests
aren't correct:

From Section 4.4.1 (Input Layout Qualifiers) of the GLSL 4.50 spec:
  "For some blocks declared as arrays, the location can only be applied
   at the block level: When a block is declared as an array where
   additional locations are needed for each member for each block array
   element, it is a compile-time error to specify locations on the block
   members. That is, when locations would be under specified by applying
   them on block members, they are not allowed on block members. For
   arrayed interfaces (those generally having an extra level of
   arrayness due to interface expansion), the outer array is stripped
   before applying this rule"

From Section 4.4.1 (Input Layout Qualifiers) of the GLSL ES 3.20 spec
  "If an input is declared as an array of blocks, excluding per-vertex-arrays
   as required for tessellation, it is an error to declare a member of
   the block with a location qualifier"

From Section 1.1.3 (Changes from GLSL ES 3.2 revision 3) of the GLSL ES 3.20 spec:
  "Arrayed blocks cannot have layout location qualifiers on members"

Here is commit related to it https://github.com/KhronosGroup/VK-GL-CTS/commit/b1e32845c7a95cf2f4b5cd59f7ddaa3c655da626
dEQP-GLES31.functional.program_interface_query.program_output.location.interface_blocks.out.block_array.var_array_explicit_location
dEQP-GLES31.functional.program_interface_query.program_output.location.interface_blocks.out.block_array.var_explicit_location
dEQP-GLES31.functional.program_interface_query.program_output.location.interface_blocks.out.block_array.var_struct_explicit_location
dEQP-GLES31.functional.program_interface_query.program_output.location.interface_blocks.patch_out.block_array.var_array_explicit_location
dEQP-GLES31.functional.program_interface_query.program_output.location.interface_blocks.patch_out.block_array.var_explicit_location
dEQP-GLES31.functional.program_interface_query.program_output.location.interface_blocks.patch_out.block_array.var_struct_explicit_location

dEQP-GLES31.functional.program_interface_query.program_input.location.interface_blocks.in.block_array.var_array_explicit_location
dEQP-GLES31.functional.program_interface_query.program_input.location.interface_blocks.in.block_array.var_explicit_location
dEQP-GLES31.functional.program_interface_query.program_input.location.interface_blocks.in.block_array.var_struct_explicit_location
dEQP-GLES31.functional.program_interface_query.program_input.location.interface_blocks.patch_in.block_array.var_array_explicit_location
dEQP-GLES31.functional.program_interface_query.program_input.location.interface_blocks.patch_in.block_array.var_explicit_location
dEQP-GLES31.functional.program_interface_query.program_input.location.interface_blocks.patch_in.block_array.var_struct_explicit_location

Signed-off-by: Andrii Simiklit <andrii.simiklit@globallogic.com>

Portability: handle vertex input stride

KHR_portability_subset provides minVertexInputBindingStrideAlignment
property that indicates the minimum alignment for vertex input strides.
This change fixes all cases where vertex attribute offsets exceeded
vertex buffer stride on MacOS.

Note that minVertexInputBindingStrideAlignment has value of 4 on MacOS
and becouse of that check for this was not added to all 250 places in
cts code that use VkVertexInputBindingDescription. All those places
currently use stride that is multiple of 4. Check was only added to
haderrender, shaderexecutor and in spirv_assembly module.

See:
https://github.com/KhronosGroup/MoltenVK/issues/1285

Components: Vulkan

Affects:
dEQP-VK.*

Change-Id: Ia7d74b9950e8af88c2d7a8462a830c3a6469538f