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

Change-Id: Ia2c5b71c6849b5306169097223726e8afd37223f

Merge vk-gl-cts/vulkan-cts-1.2.0 into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Ic06c88a4f8c25b5664a1c1fb87df125694e5b11f

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.2.0

Change-Id: I29e9c508144c70edd96aec27c6fde1e32f1c4149

Merge vk-gl-cts/vulkan-cts-1.1.4 into vk-gl-cts/vulkan-cts-1.1.5

Change-Id: Id21fae15bb0fad00a22e2d5798f958e6640804e2

Merge vk-gl-cts/vulkan-cts-1.1.3 into vk-gl-cts/vulkan-cts-1.1.4

Change-Id: I58c329c990cacdb992c4c9be0b95cb65aa469a80

VkPhysicalDeviceVulkan1{1,2}Features in framework

Use VkPhysicalDeviceVulkan11Features and
VkPhysicalDeviceVulkan12Features in DefaultDevice chain.
This change also adds Context::isDeviceFunctionalitySupported and
Context::isInstanceFunctionalitySupported methods that checks if
extension are available and if proper feature bit are set (when
needed).

Components: Framework

VK-GL-CTS issue: 2005

Affects:
dEQP-VK.*

Change-Id: I81cf677fe46ae5bd870fde37253e97bdfcc4ae3a

Update Amber

Updated Amber to get support for new image comparison features.

Components: Framework
Change-Id: I9b750912dd0324395a5b8c43f6317f341121e9e9

Merge vk-gl-cts/vulkan-cts-1.2.0 into vk-gl-cts/vulkan-cts-next-dev

Change-Id: I32be1e098131623df73a3c70ed5fc7fac2f9e3a6

Fix incorrect assumption for pipeline cache hit

The test incorrectly assumes that adding a pipeline creation flag bit
VK_PIPELINE_CREATE_ALLOW_DERIVATIVES_BIT to only one pipeline
still should cause a cache hit. Fixed the problem by adding the
VK_PIPELINE_CREATE_ALLOW_DERIVATIVES_BIT flag bit to both pipelines.

Affects:
dEQP-VK.pipeline.creation_feedback.compute_tests.compute_stage

Components: Vulkan
VK-GL-CTS issue: 2068

Change-Id: Idb963219a4e894eee6a0bb3b1fc20797d5b0323e

Merge vk-gl-cts/vulkan-cts-1.1.6 into vk-gl-cts/master

Change-Id: Id91cbdafbce603e6cd681035b8a535e3aaa294dd

Add tests for uploading depth/stencil

Components: Vulkan

Affects:
dEQP-VK.api.copy_and_blit.core.buffer_to_depthstencil.*
dEQP-VK.api.copy_and_blit.dedicated_allocation.buffer_to_depthstencil.*

VK-GL-CTS issue: 1888
Change-Id: Id2153ef90014175f5b52811c80a905164e491735

Enable validation layers for custom devices and instances

Honor the --deqp-validation=enable command line option in tests using
custom devices and instances.

Affected tests:
dEQP-VK.api.buffer.*
dEQP-VK.api.device_init.*
dEQP-VK.api.external.memory.*
dEQP-VK.api.info.*
dEQP-VK.api.object_management.*
dEQP-VK.api.version_check.*
dEQP-VK.compute.basic.*
dEQP-VK.compute.device_group.*
dEQP-VK.device_group.*
dEQP-VK.image.swapchain_mutable.*
dEQP-VK.memory.allocation.*
dEQP-VK.memory.device_group_allocation.*
dEQP-VK.multiview.*
dEQP-VK.pipeline.push_descriptor.*
dEQP-VK.protected_memory.*
dEQP-VK.robustness.*
dEQP-VK.sparse_resources.*
dEQP-VK.synchronization.basic.semaphore.*
dEQP-VK.synchronization.cross_instance.*
dEQP-VK.synchronization.internally_synchronized_objects.*
dEQP-VK.synchronization.op.multi_queue.*
dEQP-VK.synchronization.smoke.*
dEQP-VK.synchronization.win32_keyed_mutex.*
dEQP-VK.wsi.*
dEQP-VK.ycbcr.format.*

Components: Vulkan
VK-GL-CTS issue: 1831

Change-Id: I569f2362cfbc2446ba099d38da48d58be451ebeb

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.2.0

Change-Id: I3cb15875a0283934251bddc35af68cab1c3394c5

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.1.6

Change-Id: Ic9de3ec00dbd38b2106f1c497059e40d4d361297

Fix waitSemaphoreCount value in signal ordering tests

VK-GL-CTS Issue: 2059
Components: Vulkan
Affects: dEQP-VK.synchronization.signal_order.*timeline_semaphore.*

Change-Id: I25b86f1ae2b8e3650e061f9f2e345147c01419c6

Fix external memory import

Fix adds checking for flag
VK_EXTERNAL_MEMORY_FEATURE_DEDICATED_ONLY_BIT
when external memory is being imported.

Affects:
dEQP-VK.synchronization.signal_order.*

Components: Vulkan
VK-GL-CTS issue: 2063

Change-Id: Iebc72ba0ed89ba3ca2b83a5b97d6b24ab7871edb

Fixes VS num instances assumption

Components: Vulkan
VK-GL-CTS Issue: 2022

Affects:
dEQP-VK.spirv_assembly.instruction.*

Change-Id: Ibb0a17d8843c03cd1cf15d2c5af1f547481e8452

Fix LAYOUT_UNDEFINED transitions in descriptor_indexing tests

Commit 7b7ba4e0 caused the UNDEFINED->COLOR layout transition to be applied four
times, corrupting 3/4 of the image. Change the tests to do the layout transition
once, right after the framebuffer is created.

VK-GL-CTS Issue: 2064
Components: Vulkan
Affects: dEQP-VK.descriptor_indexing.*

Change-Id: Ie599ba76ba75af6fee1f4163bd96ea2bd6e4c145

Fix name shadowing warning in SurfaceCountersError

This is a simple warning fix that should not affect any test results,
related to the previous solution to the issue below.

Components: Vulkan
VK-GL-CTS issue: 1683

Change-Id: Ifa702330768f87e25d73f7489a306b4d4a1bbfb3

Fix build issues

Change-Id: I0b0da0af71d3d91e38eddf8e9c60be27b4d6ab11

dEQP-VK.graphicsfuzz.switch-if-discard

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.switch-if-discard

Change-Id: I055c4d205c0a6ce66289a1249dd8a025d4e01aa6

Reduce number of created pipelines in compute subgroups tests

Some drivers are taking too much time to execute compute subgroups
tests, mostly because they need to build N times the same pipeline
where the only difference is the local size values passed via
specialization constants.

As there are drivers that don't take advantage of pipeline derivatives,
reducing the number of pipeline creations helps to reduce execution time.

Affected tests:

   dEQP-VK.subgroups.*compute*

Components: Vulkan
VK-GL-CTS issue: 2019

Change-Id: Ie3ccbc53390bdba691c16d51c173b0a8a16119d9

Added script to list changed tests

Added script which reads git log and lists changed tests since specified
version or date.

Affects: none

Change-Id: I64e6403ed9a3bd9531c01ddb0007139950e4dda0
Components: Framework

Add tests for counters with command buffer scope

Component: Vulkan

VK-GL-CTS Issue: 1972

Affects: dEQP-VK.query_pool.performance_query.*

Change-Id: Ibf14aaef5b76bf2b2f5dc1f78779eb577c126e23

Add tests for VK_EXT_blend_operation_advanced

There at three types of tests:

* Ops: they check that all the defined operations work
for a different set of source and destination colors, iterating over
(non)premultiplied colors and uncorrelate/disjoint/conjoint alpha
values.  The tests run on different number of color attachment as well.

* Independent: it checks that color attachments with different depth
operations do not affect the others.

* Coherent: it checks if (non)coherent blend operations work as
expected. It creates two pipelines with 2 different advanced blend ops,
first draw clears the pixel with a specified destination color and do
the blending; the second draw reuses the result of the first one as
destination color and do blending on top. In the case of noncoherent
blending, it adds a memory dependency between both advanced blend
operations.

Added tests:

   dEQP-VK.pipeline.blend_operation_advanced.*

Components: Vulkan
VK-GL-CTS issue: 1673

Change-Id: I4417e6a22e0c2882849f8bf59233c266c60e45c6

Add tests for sparse images using YCbCr formats

Sparse image tests include YCbCr image formats including multiplanar
formats and a group of previously ommited non-YCbCr image formats.

Affects tests:
dEQP-VK.sparse_resources.image_sparse_binding.*
dEQP-VK.sparse_resources.device_group_image_sparse_binding.*
dEQP-VK.sparse_resources.aligned_mip_size.*
dEQP-VK.sparse_resources.image_block_shapes.*
dEQP-VK.sparse_resources.image_sparse_residency.*
dEQP-VK.sparse_resources.mipmap_sparse_residency.*
dEQP-VK.sparse_resources.image_sparse_memory_aliasing.*
dEQP-VK.sparse_resources.shader_intrinsics.*
dEQP-VK.ycbcr.*

New tests:
dEQP-VK.sparse_resources.image_sparse_binding.*
dEQP-VK.sparse_resources.device_group_image_sparse_binding.*
dEQP-VK.sparse_resources.aligned_mip_size.*
dEQP-VK.sparse_resources.image_block_shapes.*
dEQP-VK.sparse_resources.image_sparse_residency.*
dEQP-VK.sparse_resources.mipmap_sparse_residency.*
dEQP-VK.sparse_resources.image_sparse_memory_aliasing.*
dEQP-VK.sparse_resources.shader_intrinsics.*
dEQP-VK.ycbcr.storage_image_write.*

Components: Vulkan Framework

VK-GL-CTS issue: 1419
VK-GL-CTS issue: 1084

Change-Id: I3e335840efbbabc2fc320629f5a165becdbb4f26

Handle LPCWSTR in framework

Components: Framework

VK-GL-CTS issue: 1703

Affects:
dEQP-VK.*

Change-Id: I2061de4eac657aa9453bc3143306d1670eb271aa

Reenable checking for core ballot functions

All the subgroup ballot tests were mistakenly created as extension
tests, so none of them were testing the core functionality.

Components: Vulkan
Affect: dEQP-VK.subgroups.ballot.framebuffer.*

Change-Id: I2876ed3e04c831e90d7d605019cdcfd8ecb8e375

Use buildOptions consistently

Backport the changes to buildOptions. This makes it easier to change the
build options for all the tests at the same time.

Components: Vulkan
Affects: dEQP-VK.subgroups.ballot_broadcast.*

Change-Id: Idf75d3fac002da85e800634078c11f1797ba4812

Whitelist Vulkan CTS 1.2.0.1

Components: Vulkan

Change-Id: Ic95049203b246c17b535d86fcaad2ee431c6b751

Merge vk-gl-cts/vulkan-cts-1.1.6 into vk-gl-cts/master

Change-Id: Iae396c7a296389d754c462f1047329a70d619a7a

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.2.0

Change-Id: I5955793c43464b3046180ca90c1a64fa0e9379b3

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.1.6

Change-Id: Ib3cab2c6bc13dee32caf2403b9dc52492f660082

Merge vk-gl-cts/vulkan-cts-1.1.4 into vk-gl-cts/vulkan-cts-1.1.5

Change-Id: I2aa94658d6c8b6543fd9c4899f7da481ba1120a1

Merge vk-gl-cts/vulkan-cts-1.1.3 into vk-gl-cts/vulkan-cts-1.1.4

Change-Id: I0954d6799d7b0e4a527cd0fbbdcf61237ca7e9a7

Remove float16 derivatives tests

SPIR-V restricts derivatives to 32-floating point scalars and vectors
only.

Component: Vulkan

VK-GL-CTS Issue: 1931

Affects:
dEQP-VK.spirv_assembly.instruction.graphics.float16.derivative_*.*

Change-Id: Icc5341e697b797045080cc55428dbc2c02d71def
(cherry picked from commit 436d6dd5e7a0c0811bf1b6059542d1cdbccb9566)

Remove float16 derivatives tests

SPIR-V restricts derivatives to 32-floating point scalars and vectors
only.

Component: Vulkan

VK-GL-CTS Issue: 1931

Affects:
dEQP-VK.spirv_assembly.instruction.graphics.float16.derivative_*.*

Change-Id: Icc5341e697b797045080cc55428dbc2c02d71def

Make texture aniso basic self tests less sensitive

The tests cases first check with a fuzzy diff that the images are more
or less the same.

The second diff then checks with a standard image comparison + tolerance
that the images are not identical.

Some of IMG's implementations produce a closer image than the test is
currently expecting - we need to lower the tolerance from 0.05 to 0.02
in order for the diff algorithm to detect that the images are
different 'enough'.

Though there is no actual spec requirement, it is 'generally going to
be the case', for any implementation, that the images will have some
amount of difference. It is potentially useful to still have this test
inplace to check againsty aniso being accidently disabled in the tested
implementation, rather than remove it altogether.

Affects:
dEQP-VK.texture.filtering_anisotropy.basic.*

Components: Vulkan

VK-GL-CTS Issue: 2047

Change-Id: Ia9148e12172ac1c4757f5007e57b3a2b3a3434eb

Relaxes line rasterization tolerances

Components: Framework
VK-GL-CTS Issue: 2033

Affects:
dEQP-VK.rasterization.*

Change-Id: Ib41552b1bb6d9aa2b25d0cbc6f02d51ce997f4e3

Revert "Fixes VS num instances assumption"

This reverts commit bac8df251bb6103d7844bfeacbb29bf14011b7cb.

Change-Id: I60aa83d860b0b11ec7fc58fb97806635a1eebdca

Remove tests that check non-zero values for reservedMustBeZero on all APIs

Although initially it was agreed to keep them for the APIs that still
include that field as reserved, and only remove them when the API
defines the field (baseInstance), in the end it was agreed to remove
it completely, as it is testing an undefined behaviour.

Components: OpenGL, OpenGL ES

VK-GL-CTS issue: 2029

Affects:
KHR-GL40.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL40.draw_indirect.misc-reservedMustBeZero-elements
KHR-GL41.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL41.draw_indirect.misc-reservedMustBeZero-elements
KHR-GLES31.core.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GLES31.core.draw_indirect.misc-reservedMustBeZero-elements

Change-Id: Iafe3e711107da818a392ebcdfe50379cf3790d17

Merge "Merge vk-gl-cts/vulkan-cts-1.1.6 into vk-gl-cts/master"

Replace imageAtomicExchange with imageAtomicCompSwap in vertex shader

Recently the test was changed in order to replace a imageAtomicAdd for
a imageAtomicExchange, as the test can't assume that the vertex shader
will be executed exactly one per vertex shader (see VK-GL-CTS issue
1910).

But imageAtomicExchange is also problematic because in some
architecture/driver, specially tile-based, the test can't also assume
that all the vertex shader executions would finish before
rasterization starts (so fragment shader execution). So when using
imageAtomicExchange could happend that a vertex shader execution would
override the value that the fragment shader already started to update.

Due this two restrictions, this patch changes the atomic operation on
the vertex shader to imageAtomicCompSwap, that sets the value of the
image at the vertex shader only if it is the initial value (so
zero). That would prevent any possible execution when the
rasterization already started to override the value of such image.

This patch also adds a comment on the code explaining the situation,
to avoid any possible future change to a simpler atomic operation by
mistake.

Components: OpenGL ES

VK-GL-CTS issue: 1997

Affects:
KHR-GLES31.core.shader_image_load_store.advanced-allStages-oneImage

Change-Id: Icbc34662f33839a1ced9f728cee45a1198e4b495

Fix use of invalid HALF_FLOAT enum

GLES 2.0 supports GL_HALF_FLOAT_OES via the OES_texture_half_float
extension. When the enum GL_HALF_FLOAT was added in GLES 3.0,
unfortunately it did not use the same enum value as GL_HALF_FLOAT_OES.
Technically, GL_HALF_FLOAT is not supported on a GLES 2.0 implementation
even with OES_texture_half_float. Fix GLES 2.0 tests to always use
GL_HALF_FLOAT_OES to avoid the issue.

Affects:

KHR-GLES2.core.internalformat.texture2d.*_half_float_*
KHR-GLES3*.core.internalformat.texture2d.*_half_float_*

Components: OpenGL

VK-GL-CTS issue: 1923

Change-Id: I9b43091382341798a0d62b9a4ec09153e915dc5c

Merge vk-gl-cts/vulkan-cts-1.1.6 into vk-gl-cts/master

Change-Id: I2df30e7152844ece359dd9d4bfb90eca990a1bd8

Generate a non-zero exit code on test failure

If there are any failed tests, or the test run was not
compelte, the process will now generate EXIT_FAILURE (1)
instead of EXIT_SUCCESS (0).

Affects:

dEQP-VK.*

Components: Framework

VK-GL-CTS issue: 1982

Change-Id: I061d9a3b6703e1f23c48ce8a31e943744f915c92
(cherry picked from commit cfa72a70ef8371eab29da003dbb4d932adba189a)

Generate a non-zero exit code on test failure

If there are any failed tests, or the test run was not
compelte, the process will now generate EXIT_FAILURE (1)
instead of EXIT_SUCCESS (0).

Affects:

dEQP-VK.*

Components: Framework

VK-GL-CTS issue: 1982

Change-Id: I061d9a3b6703e1f23c48ce8a31e943744f915c92
(cherry picked from commit cfa72a70ef8371eab29da003dbb4d932adba189a)

Generate a non-zero exit code on test failure

If there are any failed tests, or the test run was not
compelte, the process will now generate EXIT_FAILURE (1)
instead of EXIT_SUCCESS (0).

Affects:

dEQP-VK.*

Components: Framework

VK-GL-CTS issue: 1982

Change-Id: I061d9a3b6703e1f23c48ce8a31e943744f915c92

dEQP-VK.graphicsfuzz.always-false-if-with-discard-return

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.always-false-if-with-discard-return

Change-Id: Ib49084f32085d7a620e305637a8e75e062d7ad83

Postpone .amber file parsing

All .amber files were parsed during test indexing, so they were parsed
even if none of the tests were executed. This wasn't a big performance
hit yet, but we're likely to increase the use of amber.

Due to some chicken-and-egg issues part of the capability check was
moved to the start of iterate, but since that throws InternalError if
the check fails this shouldn't be a performance issue.

The tests themselves have not changed, just the order of when to parse
the file.

Affects: None

Components: Vulkan, Framework

VK-GL-CTS issue: 899

Change-Id: Iacdfd9a61d439e3f5dec952b2426386a2eb3451c

Test VkCopyDescriptorSet with inline uniform blocks

Use VkCopyDescriptorSet to copy data between inline uniform blocks
during vkUpdateDescriptorSets if VK_EXT_inline_uniform_block is
supported by the implementation.

New tests:
dEQP-VK.binding_model.descriptor_copy.*.inline_uniform_block_*
dEQP-VK.binding_model.descriptor_copy.*.mix_3
dEQP-VK.binding_model.descriptor_copy.*.mix_array1

Components: Vulkan
VK-GL-CTS issue: 2012

Change-Id: I3898f5c2f236ef62a392409791a870633b572e12

Add tests for VK_EXT_display_surface_counter

This commit adds tests for VK_EXT_display_surface_counter that try to
verify vkGetPhysicalDeviceSurfaceCapabilities2EXT returns the same set
of capabilities as vkGetPhysicalDeviceSurfaceCapabilitiesKHR and the
supportedSurfaceCounters field has coherent values for display surfaces
and non-display surfaces.

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

New tests:
dEQP-VK.wsi.*.surface.query_surface_counters
dEQP-VK.wsi.display.surface_counters

Components: Vulkan
VK-GL-CTS issue: 1683

Change-Id: I7754411c7b458201e9a7f5e8a99dfada388124d7

dEQP-VK.graphicsfuzz.switch-with-empty-if-false

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.switch-with-empty-if-false

Change-Id: I42790dbee5975b301126a268c801c6a616aab721

dEQP-VK.graphicsfuzz.unreachable-return-in-loop

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.unreachable-return-in-loop

Change-Id: I339de39006859f5396dc7397b722323f56a2222d

dEQP-VK.graphicsfuzz.disc-and-add-in-func-in-loop

Components: Vulkan

New Tests:

dEQP-VK.graphicsfuzz.disc-and-add-in-func-in-loop

Change-Id: I87cbd5e5f9b0d27f9d875a1ca31d3c7aa12cb433

Disentangle storage and arithmetic from conversions

Remove the requirement for 16-bit storage from float16 conversions
tests. Add f16 -> f32 and f32 -> f16 tests using only 16-bit storage.

Component: Vulkan

VK-GL-CTS Issue: 1931

Affects:
dEQP-VK.spirv_assembly.instruction.*.fconvert.*float16*
dEQP-VK.spirv_assembly.instruction.*.convertftou.*float16*
dEQP-VK.spirv_assembly.instruction.*.convertftos.*float16*
dEQP-VK.spirv_assembly.instruction.*.convertutof.*float16*
dEQP-VK.spirv_assembly.instruction.*.convertstof.*float16*

Change-Id: I993bdb7a6313cf19ce59b5fa43645f5168ea3cbc

Whitelist VK_KHR_shader_subgroup_extended_types

Components: Vulkan

Change-Id: Ie68fd0fd2606aac29ddf1e1a0c6668504a42a91a

Merge vk-gl-cts/vulkan-cts-1.1.6 into vk-gl-cts/master

Change-Id: I7a8febada6ebf7ad2de0f9727a5efbbcc729c3d0

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.2.0

Change-Id: I01ffb4e6195d7754314ec4a72be0c03f595b6e02

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.1.6

Change-Id: I3b96cb8cde6cbfc585d717a94f2164739887726f

Merge vk-gl-cts/vulkan-cts-1.1.4 into vk-gl-cts/vulkan-cts-1.1.5

Change-Id: Iea7706e036758d8a648ef7938b3d08e21321a087

Use the correct imageType for image support checks.

Components: Vulkan

Affects:
dEQP-VK.api.copy_and_blit.core.blit_image.all_formats.depth_stencil.{1d,3d}*

VK-GL-CTS issue: 2050

Change-Id: If73d9ff3e026ae4c32fd00fcc546734eebf9990e

Fix wrong wait stage mask

The wait stage mask wasn't accounting for the
stages the readers actually operate in, hence
being effectless.
Track the union of the reader's stage masks
and use that to block execution on the semaphore.

Affects:
dEQP-VK.synchronization.signal_order.*

Components: Vulkan
VK-GL-CTS issue: 2044

Change-Id: Ia4d5c7bcb9dd6eec25e7245dc41754acccd1b995

Fix in shadercacheLoad

There is a slim chance that function shadercacheLoad() can pass a null
pointer into the std::string constructor, causing an exception. This can
happen in the case there is a mismatch between the calculated hash and
the one stored in the cache file.

This commit makes the code more robust, handling this case as a miss.

Components: Vulkan

VK-GL-CTS issue: 2052

Change-Id: I84a382dec51f487314304ec9057cefdff22920c3

Fixed mandatory caselist for parallel CTS run

Duplicate tests listed in deqp-fraction-mandatory-caselist-file
command line option are now skipped during test tree parsing,
when parallel CTS run is configured.

Components: Framework

VK-GL-CTS issue: 2053

Change-Id: I4ce2b6f33fd22c7ddc3aba0adbe9f7e761237e51

Merge vk-gl-cts/vulkan-cts-1.2.0 into vk-gl-cts/vulkan-cts-next-dev

Change-Id: Ic6e8b920be9a0ff25a8765b7c11d57248fb400a3

Merge vk-gl-cts/vulkan-cts-1.1.6 into vk-gl-cts/master

Change-Id: If68c0cb9516bb0bb1f441c9fde8bfd5456cee76d

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

Relax deviceName and deviceInfo string comparison

Change deMemCmp() to strncmp() for two char arrays in
VkPhysicalDeviceVulkan12Properties struct, so that we don't
compare bytes after null termination character.

Affects:
dEQP-VK.api.info.vulkan1p2.properties

Components: Vulkan

VK-GL-CTS issue: 2040

Change-Id: I312575c5b29705c1cb2010f2a2d247055680bf97

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.2.0

Change-Id: Ie09cc3cac8462a7669c3fa0c25edb40d8e05d062

Remove superfluous call to createCustomDevice

Components: Vulkan

Change-Id: I7982ff1203b8eaeccfdd4b3a6b9bc3ab0b688721

Merge vk-gl-cts/vulkan-cts-1.1.5 into vk-gl-cts/vulkan-cts-1.1.6

Change-Id: Ife971c6779bf35094b5a72a87abaa47a6901530b

Merge vk-gl-cts/vulkan-cts-1.1.4 into vk-gl-cts/vulkan-cts-1.1.5

Change-Id: Ibb2b736bbe1c62688044aa21dc75205f6fa32f1c

Merge vk-gl-cts/vulkan-cts-1.1.3 into vk-gl-cts/vulkan-cts-1.1.4

Change-Id: I79fdf16146cf659aeae510fb6fb2ad72f641ba3e

Merge vk-gl-cts/vulkan-cts-1.1.2 into vk-gl-cts/vulkan-cts-1.1.3

Change-Id: Ia0d88c0631e427241c7795931a834d9baf43e5a3

RelaxedPrecision tests for image extend operands

Where the decoration makes sense (that is, for images with fewer than 16
bits per channel), add test that use the RelaxedPrecision decoration.

Components: Vulkan
VK-GL-CTS issue: 2038
New tests: dEQP-VK.image.extend_operands_spirv1p4.*_relaxed_*

Change-Id: I26e42a1d7bb807616ac50ce461dd37a55effd6f8
(cherry picked from commit aaf3d22fa09cc800af076dd2d47124dd1659bcd1)

RelaxedPrecision tests for image extend operands

Where the decoration makes sense (that is, for images with fewer than 16
bits per channel), add test that use the RelaxedPrecision decoration.

Components: Vulkan
VK-GL-CTS issue: 2038
New tests: dEQP-VK.image.extend_operands_spirv1p4.*_relaxed_*

Change-Id: I26e42a1d7bb807616ac50ce461dd37a55effd6f8

UConvert tests for signed int operand type

OpUConvert where the operand type is a signed integer type.

Add tests:
 * dEQP-VK.spirv_assembly.instruction.compute.uconvert.int*
 * dEQP-VK.spirv_assembly.instruction.graphics.uconvert.int*

Affects:
 * dEQP-VK.spirv_assembly.instruction.compute.uconvert.*
 * dEQP-VK.spirv_assembly.instruction.compute.uconvert.*

Components: Vulkan

VK-GL-CTS issue: 1876

Change-Id: I0cee4fa43fab995ef20dcc30454d910ce371bb6e
(cherry picked from commit b0eda2776125d76c3ef3df394d6c224271f90f95)

Add tests for SignExtend with 'unsigned' types

SPIR-V wants to allow 'unsigned' as meaning 'unknown-signedness', in
which case a signed access will have an unsigned SampledType and the
SignExtend operand. Add tests for this combination.

Components: Vulkan
VK-GL-CTS issue: 2038
Affects: dEQP-VK.image.extend_operands_spirv1p4.*

Change-Id: Icd2ea0f41fcaab36d03ea3cff919449db3bd469a
(cherry picked from commit b76f0d042c3587221daa77fef55b2642204a7ca3)

Add tests for SignExtend with 'unsigned' types

SPIR-V wants to allow 'unsigned' as meaning 'unknown-signedness', in
which case a signed access will have an unsigned SampledType and the
SignExtend operand. Add tests for this combination.

Components: Vulkan
VK-GL-CTS issue: 2038
Affects: dEQP-VK.image.extend_operands_spirv1p4.*

Change-Id: Icd2ea0f41fcaab36d03ea3cff919449db3bd469a

UConvert tests for signed int operand type

OpUConvert where the operand type is a signed integer type.

Add tests:
 * dEQP-VK.spirv_assembly.instruction.compute.uconvert.int*
 * dEQP-VK.spirv_assembly.instruction.graphics.uconvert.int*

Affects:
 * dEQP-VK.spirv_assembly.instruction.compute.uconvert.*
 * dEQP-VK.spirv_assembly.instruction.compute.uconvert.*

Components: Vulkan

VK-GL-CTS issue: 1876

Change-Id: I0cee4fa43fab995ef20dcc30454d910ce371bb6e

Skip test with no display plane is available

Similarly to dEQP-VK.wsi.display.get_display_properties that skips
when no display available.

Components: Vulkan
Affects: dEQP-VK.wsi.display.get_display_plane_capabilities2

Change-Id: I46af976a9ba3e77e250504d0164dec6ac4cdd607
(cherry picked from commit fd915c2d0ec7b544b670192447223061fe4b6cf3)

Skip test with no display plane is available

Similarly to dEQP-VK.wsi.display.get_display_properties that skips
when no display available.

Components: Vulkan
Affects: dEQP-VK.wsi.display.get_display_plane_capabilities2

Change-Id: I46af976a9ba3e77e250504d0164dec6ac4cdd607
(cherry picked from commit fd915c2d0ec7b544b670192447223061fe4b6cf3)

Skip test with no display plane is available

Similarly to dEQP-VK.wsi.display.get_display_properties that skips
when no display available.

Components: Vulkan
Affects: dEQP-VK.wsi.display.get_display_plane_capabilities2

Change-Id: I46af976a9ba3e77e250504d0164dec6ac4cdd607

Fix subnormal floats in extended image usage tests

When transcoding images in the extended image usage bit tests, Inf and
NaN were correctly taken into account and removed from input data.
However, subnormal floats were only being taken into account in the
32-bit float case.

The Vulkan spec regarding the precision and operation of SPIR-V
instructions mentions subnormal values may be flushed to zero in any
shader instruction, including the shader code to transcode images, and
have to be taken into account for half floats too.

Afffected tests:
dEQP-VK.image.extended_usage_bit.*

Components: Vulkan
VK-GL-CTS issue: 1747

Change-Id: I74a5d27b40f773f8d85e86a0a5d04eb0b4f16194
(cherry picked from commit 300b27043f8c93dcce4b3c788b33688ab59867af)

Fix subnormal floats in extended image usage tests

When transcoding images in the extended image usage bit tests, Inf and
NaN were correctly taken into account and removed from input data.
However, subnormal floats were only being taken into account in the
32-bit float case.

The Vulkan spec regarding the precision and operation of SPIR-V
instructions mentions subnormal values may be flushed to zero in any
shader instruction, including the shader code to transcode images, and
have to be taken into account for half floats too.

Afffected tests:
dEQP-VK.image.extended_usage_bit.*

Components: Vulkan
VK-GL-CTS issue: 1747

Change-Id: I74a5d27b40f773f8d85e86a0a5d04eb0b4f16194

Remove uses of strstr in vktSpvAsmFloatControlsTests.cpp

Add flags to detect usage of types and/or constants using bit
flags instead of expensive string lookup operations (like strstr).

Affects:
 * dEQP-VK.spirv_assembly.instruction.compute.float_controls.*
 * dEQP-VK.spirv_assembly.instruction.graphics.float_controls.*

Components: vulkan

VK-GL-CTS issue: 1927

Change-Id: I9635c81c056a8c8d83a832eddf2e371d1fcec885
(cherry picked from commit 4e2e58f4f53df69a295182a837f385ea8f8d1929)

Remove uses of strstr in vktSpvAsmFloatControlsTests.cpp

Add flags to detect usage of types and/or constants using bit
flags instead of expensive string lookup operations (like strstr).

Affects:
 * dEQP-VK.spirv_assembly.instruction.compute.float_controls.*
 * dEQP-VK.spirv_assembly.instruction.graphics.float_controls.*

Components: vulkan

VK-GL-CTS issue: 1927

Change-Id: I9635c81c056a8c8d83a832eddf2e371d1fcec885

Remove tests that check non-zero values for reservedMustBeZero after GL 4.2

draw_indirect.misc-reservedMustBeZero-arrays and
draw_indirect.misc-reservedMustBeZero-elements tests that the driver
doesn't crash or emit any error if the reserved field at
DrawArraysIndirectCommand and DrawEelementsIndirectCommand is
different to zero.

But that field become defined on GL 4.2 (it became baseInstance), so
after 4.2 those tests doesn't make sense.

Components: OpenGL

VK-GL-CTS issue: 2023

Affects:
KHR-GL42.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL42.draw_indirect.misc-reservedMustBeZero-elements
KHR-GL43.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL43.draw_indirect.misc-reservedMustBeZero-elements
KHR-GL44.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL44.draw_indirect.misc-reservedMustBeZero-elements
KHR-GL45.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL45.draw_indirect.misc-reservedMustBeZero-elements
KHR-GL46.draw_indirect.misc-reservedMustBeZero-arrays
KHR-GL46.draw_indirect.misc-reservedMustBeZero-elements

Change-Id: I6c9872efd4f161179da157f6f5f26731a2032372

Set swap interval to 0

By default, when swapping front and back frame buffers, this is in sync
with v-blank, which prevents tearing.

So this means the tests does not run at is maximum allowed speed, but it
is restricted somewhat by the attached screen.

When running in a window and X realizes the is no monitor attached, the
server's Present code will vsync at 1 FPS, which artificially slow down
the tests a ton.

To avoid this issue, and also because we want to run things as fast as
possible, set the swap interval to 0, which basically deactivate the
sync.

Test: cmake .. -DDEQP_TARGET=x11_egl

Change-Id: I04bcb814fb289d58f8788e90f5118db0b7175777

Limit the apiVersion of the instance within the CTS framework

This CL allows older CTS tests to run on implementations that
support a newer instance version not supported by this version
of CTS.

The dEQP-VK.api.version_check.version test was modified to
fail if this version of CTS is run on an implementation
it does not support. This is to stop old CTS from passing
on newer implementations, where the newer CTS should have
been used.

Affects:

dEQP-VK.api.version_check.version
dEQP-VK.*

Components: Vulkan

VK-GL-CTS issue: 1979

Change-Id: I97cd17b0aae9b3cba7c2e000d21636ad37c1fd80

Fix SEMAPHORE_TYPE_TIMELINE enum

Components: Vulkan

Change-Id: Id4949383bed578436e321c621a8ff495157da895

Merge vk-gl-cts/vulkan-cts-1.1.3 into vk-gl-cts/vulkan-cts-1.1.4

Change-Id: I6597defb9bda8c8e6393033bf444befd4b9301fb

Fix relaxed precision and no perspective dEQP tests

In the RelaxedPrecision case, change epsilon to 2e-3 for a 16-bit float.
In the NoPerspective case, change epsilon to 3e-7 for a 32-bit float.

The epsilon is now relative to the correct value.

Affects:
dEQP-VK.spirv_assembly.instruction.graphics.cross_stage.interface_blocks.*

Components: Vulkan

VK-GL-CTS issue: 1818

Change-Id: Id21382ce4b1cb58458ec3fb521630ab75e1910a3

Update glslang

Affects: dEQP-VK.binding_model.buffer_device_address.*

Components: Vulkan

VK-GL-CTS issue: 2046

Change-Id: I60e3579df3bb3f81732fec0c46803a97fc00953a

Fixed flags used to select queue in dEQP-VK.synchronization.op.multi_queue

getDefaultQueue takes a VkQueueFlags argument and returns an appropriate VkQueue.
This function was being called with inappropriate arguments
causing the wrong queue to be returned. This could ultimately lead to work
being submitted to a queue which lacked support for it (e.g. graphics work
on a sparse queue).

Affects:

dEQP-VK.synchronization.op.multi_queue*

Components: Vulkan

VK-GL-CTS issue: 2043

Change-Id: I7887d1b4da8c916aed032433c6634c99b723f285
(cherry picked from commit 1bd32d3221dd65087de14ae061f798b78ac7ce0d)

Fixed flags used to select queue in dEQP-VK.synchronization.op.multi_queue

getDefaultQueue takes a VkQueueFlags argument and returns an appropriate VkQueue.
This function was being called with inappropriate arguments
causing the wrong queue to be returned. This could ultimately lead to work
being submitted to a queue which lacked support for it (e.g. graphics work
on a sparse queue).

Affects:

dEQP-VK.synchronization.op.multi_queue*

Components: Vulkan

VK-GL-CTS issue: 2043

Change-Id: I7887d1b4da8c916aed032433c6634c99b723f285

Fix semaphore creation in smoke tests

The test was calling createSemaphore instead of createSemaphoreType,
meaning that the semaphoreType value would incorrectly be interpreted
as a VkSemaphoreCreateFlags.

Additionally, the VkTimelineSemaphoreSubmitInfo struct was not being
added into the pNext chain of the submits. This probably went unnoticed
because the first bug could cause the test to use binary semaphores
even if semaphoreType was TIMELINE.

Affects:

dEQP-VK.synchronization.smoke.binary_semaphores
dEQP-VK.synchronization.smoke.timeline_semaphores

Components: Vulkan

VK-GL-CTS issue: 2041

Change-Id: I6936013b7f5a7e25f4af1fb4776d4d7273243d7a
(cherry picked from commit a8238c0aa023c121f79f09c6565875e235292f27)

Fix semaphore creation in smoke tests

The test was calling createSemaphore instead of createSemaphoreType,
meaning that the semaphoreType value would incorrectly be interpreted
as a VkSemaphoreCreateFlags.

Additionally, the VkTimelineSemaphoreSubmitInfo struct was not being
added into the pNext chain of the submits. This probably went unnoticed
because the first bug could cause the test to use binary semaphores
even if semaphoreType was TIMELINE.

Affects:

dEQP-VK.synchronization.smoke.binary_semaphores
dEQP-VK.synchronization.smoke.timeline_semaphores

Components: Vulkan

VK-GL-CTS issue: 2041

Change-Id: I6936013b7f5a7e25f4af1fb4776d4d7273243d7a

Split descriptor set indexing render pass workload

The render pass workload for descriptor set indexing tests should be
split into multiple submissions to prevent timeouts observed in tests
for the VK_DESCRIPTOR_TYPE_SAMPLER,
VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
VK_DESCRIPTOR_TYPE_INPUT_ATTACHMENT,
VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER and
VK_DESCRIPTOR_TYPE_SAMPLED_IMAGE descriptor types.

Affects: dEQP-VK.descriptor_indexing*

Components: Vulkan

VK-GL-CTS Issue: 2026

Change-Id: I3b4876ec204fe1c1ebf174e06328653f0a5532d3