When enabling C++17, LLVM uses std::shared_mutex instead of its own
implementation. On macOS it requires version 10.12 (Sierra), while
Chrome must still support 10.11. However, we don't need LLVM as Reactor
backend, except for ARM64, but Mac devices with such CPU have a more
recent OS version.

llvm-10.0/llvm/include/llvm/Support/RWMutex.h:100:8: error: 'shared_mutex' is unavailable: introduced in macOS 10.12

Bug: b/174843857
Change-Id: Ie24095aa17d22072cfb9d655b87252cbea5209d0
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/52030Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>

The legacy PixelProcessor and VertexProcessor classes for OpenGL ES
contain member fields with types like float4, which require 16-byte
alignment. This isn't guaranteed by the compiler until C++17. On macOS,
this also requires OS support, which was added in version 10.14
(Mojave), while Chrome still has to support macOS 10.11.

src/Renderer/PixelProcessor.cpp:75:18: error: aligned deallocation function of type 'void (void *, std::align_val_t) noexcept' is only available on macOS 10.14 or newer

Overriding new and delete for these classes allows us to use our custom
allocator which guarantees alignment. Note that the Renderer class
already used the same approach. Also note that this hasn't caused issues
before because these fields aren't actually accessed by instructions
which demand alignment. However, it's still good for performance and to
align with the intent (pun intended).

Bug: b/174843857
Change-Id: Ia5de5f6fe67a4f54805cdde3fed565c30b9318a8
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/52029Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>

Updates build files to specify using the C++17 standard.

Add a usage of the C++17 exclusive [[maybe_unused]] attribute as a test.
Also unconditionally use <filesystem> in Reactor unit tests.

C++17 deprecates things like std::unary_function<>, which were present
in Subzero's LLVM headers, but the functionality was unused so it has
been deleted.

Bug: b/174843857
Bug: b/155971541
Change-Id: Ib167726535399b0b12751a3d93feaa9db762cd5d
Signed-off-by: Nicolas Capens <capn@google.com>
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51948Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>

This extension allows users to pass depthstencil attachments that
vkCmdResolveImage will resolve to in addition to its other resolve
operations. Only the mandatory resolve modes
"VK_RESOLVE_MODE_SAMPLE_ZERO_BIT" and "VK_RESOLVE_MODE_NONE" are
supported.

It's trivial to support both independent resolve modes since we must
resolve depth and stencil attachments separately due to how depth and
stencil attachments are stored internally.

Change-Id: I0f8ff7cddca5f5acbac1d991b11f0a4447956784
Bug: b/167558951
Test: dEQP-VK.renderpass2.depth_stencil_resolve.*
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/48808Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>

This will allow us to reuse this code across unit tests and benchmarks.

Bug: b/176981107
Change-Id: Ie5ea3708cb959dbd0189658b42f01bad998ae94a
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51869Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>

* Move utility classes into separate hpp/cpp files
* Move utility functions into Util.hpp/cpp

Bug: b/176981107
Change-Id: If9046f12080f1eb09378586673d871867a836a1b
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51868
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>

The SPIR-V 'EarlyFragmentTests' execution mode requires performing
additional per-fragment tests prior to fragment shader execution,
even if depth values computed by the shader would otherwise affect
them: https://www.khronos.org/registry/vulkan/specs/1.1/html/vkspec.html#shaders-fragment-earlytest

Bug: b/140302392
Change-Id: If5d836a35ee9cfc758243b34fad1222d1b0f1085
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51888
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

This CL adds a Posix-based implementation for the
VK_KHR_external_memory_fd Vulkan extension, using the SysV
shm_open() / shm_unlink() API to create the shared
memory region.

The Linux backend still uses memfd-based regions which
are easier to create, and because Android does not provide
shm_open() intentionally [1].

[1] https://android.googlesource.com/platform/ndk/+/4e159d95ebf23b5f72bb707b0cb1518ef96b3d03/docs/system/libc/SYSV-IPC.TXT

Bug: b/140419396
Tests: dEQP-VK.*
Change-Id: Ibbb23c3af59e81f76e41a0e71281f6d1a8b07c01
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/41408Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Yilong Li <liyl@google.com>
Commit-Queue: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>

* Identify members that are owning handles with a comment
* Reorder these members in creation order
* Reorder explicit destruction of these members in reverse order in destructors
* Replace raw pointers with unique_ptrs, but call reset explicitly in destructors to maintain reverse destruction order
* Define VULKAN_HPP_NO_NODISCARD_WARNINGS to future-proof for C++17 and up
* Remove vestigial vk::UniqueShaderModule and vk::UniquePipeline usage

Bug: b/176981107
Change-Id: Ia090be98c2f8ed47701dfc92388547985e5c57b4
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51829Tested-by: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

Reactor backend: Subzero

Change-Id: I0e14ea808524eea31d4c12aa501cd2e98e445521
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51768Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>

It is legal for a shader to not declare BuiltInPosition.
This cl adds logic to use 0 for all position components
when the position builtin is not present and then keeps
executing the rest of the shader normally.

Bug: b/176161380
Change-Id: I1b4637b813d3d30958e2db655d299603d625c20a
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51808
Commit-Queue: Alexis Hétu <sugoi@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>
Presubmit-Ready: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

This filename is output in the emitted asm files when
ENABLE_RR_EMIT_ASM_FILE is enabled, but on Windows, it was always
"<unknown>". Fixed this by splitting out the name by backslash as well.

Bug: b/174358505
Change-Id: Ie5386fa50fd8a6aa43a48bc6fa5fcf987936aacf
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51750
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

Found these leaks using AppVerifier on Windows.

Bug: b/176981107
Change-Id: I89cbfc02e6adf0ea2a54cf80501fbbb3d0867964
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51748Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>

Reactor backend: LLVM

Change-Id: I4590357d7c1eac21c570cbcbdeffa0f4de6d70cd
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51208Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

This change was originally started by capn@, which added texture
sampling to the Triangle benchmarks. I finished up this change with the
following changes and fixes:

* Implement a correct fill texture to white using copyBufferToImage.
* Add Buffer class to wrap vk::Buffer and vk::DeviceMemory.
* Add support for using a VK_EXT_headless_surface Window, which is the
default on non-Windows platforms for now. On Windows, can be forced to
use this by defining USE_HEADLESS_SURFACE = 1.
* Fix out of bounds access because renderPassBeginInfo.clearValueCount
was set to 2 instead of 1.
* Fix illegal access to destructed ArrayRef in call to
commandBuffers[i].bindVertexBuffers.
* Make it build and run on Linux (fix compilation errors and use
aforementioned headless surface mode)

Bug: b/176981107
Change-Id: If82496315e77a062c024766863a1ca01e6b424f9
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51668
Kokoro-Result: Antonio Maiorano <amaiorano@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

This implements the extension defined here:
https://www.khronos.org/registry/vulkan/specs/1.2-extensions/man/html/VK_EXT_headless_surface.html

This extension is useful for tests/benchmarks as it provides a way to
create a platform-agnostic Surface with Swapchain, etc., instead of
having to create an non-visible framebuffer.

Bug: b/176981107
Change-Id: Iad1e55ed7a4e32cebf7577d9d88a9aed4b6f7a20
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51589Tested-by: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

Eventually, this should probably be fetched from the
gralloc buffer metadata using
Gralloc4::Get(android::gralloc4::MetadataType_Dataspace)

Bug: b/147316305
Bug: b/175132241
Test: launch Cuttlefish w/ SwANGLE
Test: cts -m CtsGraphicsTestCases -t
      MediaVulkanGpuTest#testMediaImportAndRendering
Change-Id: Ie991380d2d27e7a540ffe37cfc81109618f2d507
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51688
Presubmit-Ready: Jason Macnak <natsu@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Jason Macnak <natsu@google.com>
Commit-Queue: Jason Macnak <natsu@google.com>

Bug: b/176981107
Change-Id: Ifab91e61ceb74c5b6caf62b8f9f1abbd8fef9e01
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/45848Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Commit-Queue: Nicolas Capens <nicolascapens@google.com>

When the homogeneous w coordinate is near 0, the projected x or y
coordinates (after viewport and subpixel scaling) may be larger than
what's representable by 32-bit signed integers. On x86 this produces
the "indefinite" integer value 0x80000000. This negative value is
returned even for positive inputs. This causes the back-face culling
calculations to reject triangles which are in fact visible.

This change addresses that by introducing the rr::RoundIntClamped()
function, which guarantees that very large positive floating-point
values are converted to a positive integer value near INT_MAX.

Note that the Vulkan spec states that "[the determination whether the
triangle is back-facing or front-facing] is made based on the sign of
the (clipped or unclipped) polygon’s area computed in framebuffer
coordinates." Thus the alternative of performing culling using the
unprojected coordinates would have to take negative viewport dimensions
into account, and may not produce the same results as sub-pixel
snapped coordinates.

Adjust back-face culling of wireframe triangles to use the same
calculations as solid triangles.

Bug: b/177382194
Change-Id: Ice8493d4cfd164638f091bf5a39b5396d65458e2
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51648Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

Members of classes that are declared as static constexpr are not defined
as part of their declaration. Thus they must be defined elsewhere. And
they can't be defined in the header file, or there will be multiple
definitions for each file that includes the header.

So define these contexprs in Constants.cpp

Bug: b/177220369
Change-Id: I9ffc63867948bc72061ddb8b0b586a46d840196c
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51628
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>

This extension allows the host to reset query pools without the use of
vkCmdResetQueryPool.

Bug: b/169602655
Tests: dEQP-VK.query_pool.occlusion_query.*
Change-Id: I20a121d30ff5c39dc9df0a803c3f054ff217e5dc
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51309
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

This extension allows users to query and specify different usages for
the main image and the stencil. SwiftShader does not use an image's
usage for non-debug purposes outside of
vkGetPhysicalDeviceImageFormatProperties2. This applies to separate
stencil usage as well.

Bug: b/176098507
Tests: dEQP-VK.pipeline.sampler.separate_stencil_usage.*
Tests: dEQP-VK.renderpass.suballocation.multisample.separate_stencil_usage.*
Tests: dEQP-VK.renderpass2.suballocation.multisample.separate_stencil_usage.*
Change-Id: Iea74a9fa0de8cdcc51bdc613cdc0754e0f327fd5
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51368Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

It only contained the now trivial OpcodeName() implementation. Since the
SpirvShaderInstructions.cpp file already contains other SPIR-V opcode
utility functions, move it there.

Also make this function return a pointer to a constant char string,
to avoid needless std::string construction.

Bug: b/145351270
Change-Id: Iacfc887566b0a9057f37391ad52d43cdca912308
Signed-off-by: Nicolas Capens <capn@google.com>
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51448
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

It's only used for compilation, so don't hang on to it for the entire
lifetime of the routine. Likewise the dylib reference can be local.

Bug: b/177024837
Change-Id: I261e877f983aa3ec52ef61dc6976da75a4fd5efb
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51588Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>

It's only used for mangling the names of our generated functions during
Routine construction, so don't hang on to it for the entire lifetime of
the routine.

Also, don't set Function attributes during routine compilation, set them
at creation.

Bug: b/177024837
Change-Id: I974b10430bced5d59f51e54426f7ee0aa8c96ca2
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51569
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

The LLVMContext object is used for caching LLVM types, eliminating
duplicate constants, and things like handling diagnostics information.
Previously we created one for the JITBuilder as well as the JITRoutine.
While having multiple ones is mostly benign from a correctness point of
view, it adds unnecessary overhead.

This change reuses the context created for the JITBuilder object (which
is used during IR construction), for the JITRoutine (which uses it for
actual compilation).

Bug: b/177024837
Change-Id: Iefbbd7fbeb53e67ab41914fad471dd94d9755311
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51568
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

Clang 9 escalates a warning about an unsigned integer being compared
against a signed integer into an error.

Fixed by making an integer literal be unsigned.

Bug: b/152777669
Change-Id: I38c61ac0837f756c73b8664f7774719a94353fac
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51530Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

Produced by Clang 10's -Wmissing-field-initializers.

Bug: b/152777669
Change-Id: Ib6cda930945b1b46e5530f2aef7bbc9ee07ab9df
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51549
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Commit-Queue: Nicolas Capens <nicolascapens@google.com>

Bug: b/176909700
Change-Id: I5cecfbdfc4b5fc28180b1421c36ce94813605bd5
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51508
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Nicolas Caramelli <caramelli.devel@gmail.com>
Commit-Queue: Nicolas Caramelli <caramelli.devel@gmail.com>

While SwiftShader does not support this extension, it is core in Vulkan
1.2. So we must properly process the structures it adds.

Bug: b/176819536
Change-Id: Ib4e0334d9224581b2541f89b356e7bef5f138d30
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51490
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Sean Risser <srisser@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Commit-Queue: Sean Risser <srisser@google.com>

Clang 10's -Wimplicit-int-float-conversion complains that
MAX = 2147483647 cannot be exactly represented as a float.

Use an explicit conversion to suppress the warning locally. Note that
this value's usage to produce a random boolean doesn't critically depend
on exact conversion.

Bug: b/152777669
Change-Id: I1136ca16dcb842b97f4a76a6ed3e1c3333e814f8
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51488Tested-by: Nicolas Capens <nicolascapens@google.com>
Commit-Queue: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

Bug: documentation
Change-Id: I708546425c3fee85844884ed5dfc8f52af1f1830
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51428
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

VkPhysicalDeviceVulkan12Properties was added to the core spec by Vulkan
1.2. It allows users to query the following properties simultaneously:
    * VkPhysicalDeviceDriverProperties
    * VkPhysicalDeviceFloatControlsProperties
    * VkPhysicalDeviceDescriptorIndexingProperties
    * VkPhysicalDeviceDepthStencilResolveProperties
    * VkPhysicalDeviceSamplerFilterMinmaxProperties
    * VkPhysicalDeviceTimelineSemaphoreProperties
    * and framebufferIntegerColorSampleCounts

All these structs are a part of the core spec and must be supported.

Bug: b/176498010
Change-Id: I82b06f219d2ee091139964ee8b19f05e1c64e3a8
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51408
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>

This change deprecates rr::RcpSqrt_pp with rr::RcpSqrt. As with Rcp,
RcpSqrt computes the result using Newton-Rhapson if it's faster and the
initial approximation intrinsic is available on the current target.
Currently, only LLVM on Intel will use NR for RelaxedPrecision. Note
that passing in Precision::Relaxed will produce a faster, but less
precise reciprocal sqrt.

Also made it so that SprivShader instruction GLSLstd450InverseSqrt now
invokes RcpSqrt(x, Precision::Full) instead of performing 1/sqrt(x).
Note that the Vulkan spec states that inversesqrt()'s precision is 2
ULP, and sqrt()'s precision is inherited from 1.0 / inversesqrt();
however, our rr::Sqrt is implemented in terms of x86's sqrt intrinsic on
x86, or as calls to sqrt from Math.h.

Bug: b/169760262
Change-Id: I65ba9a64d1db934c523dda11c1a2c186059d220b
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51268
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>

This change deprecates rr::Rcp_pp with rr::Rcp, which makes sure to
correctly compute the reciprocal using the Newton-Rhapson refinement
only if the current target supports the required instrinsic, otherwise
using 1 / x. Currently, only LLVM on Intel will use NR. Note that
passing in Precision::Relaxed will produce a faster, but less precise
reciprocal.

Also removed PixelProgram::linearToSRGB as it's unused.

Bug: b/169760262
Bug: b/149574741
Change-Id: I4a2f943aa60116c4397d7a8ae18583a260824788
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/50648Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>

ReactorBenchmarks.cpp was accidentally moved to the wrong location as
part of swiftshader-cl/50388.

Bug: b/173404759
Change-Id: I4c0f836ddb7729d6d4409fe757d384cadb5b48b3
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51289Tested-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

These were used by older versions of the Android CTS and are no longer
present in dEQP.

Change-Id: I0c66aed16470697b6c925dfe222f26aeb439848f
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51008Tested-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

VkBuffers backed by AHBs do not require dedicated allocations.
As such, update SupportsAllocateInfo() to report supported as
long as a VkImportAndroidHardwareBufferInfoANDROID or a
VkExportMemoryAllocateInfo with AHB bit was found.

Updates allocateAndroidHardwareBuffer() to treat the lack of a
dedicated buffer or image as a request to allocate a blob AHB.

Bug: b/147316305
Bug: b/169796031
Test: cts -m CtsNativeHardwareTestCases
Test: cts -m CtsDeqpTestCases -t
  dEQP-VK.api.external.memory.android_hardware_buffer.*
Change-Id: Ica2782b4570bd4d4c1e5585b696b40314a267b2f
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51248
Presubmit-Ready: Jason Macnak <natsu@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Jason Macnak <natsu@google.com>
Commit-Queue: Jason Macnak <natsu@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

UniformBufferStandardLayout changes the alignment rules for uniform
buffers. This doesn't change SwiftShader's behavior, and is treated as
always enabled for the purposes of validating inputs to SPIRV-Tools.

Bug: b/169604082
Tests: dEQP-VK.ubo.*
Change-Id: I0ec59ceb0bf6cc487f37e292a04e875268c37185
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/49868Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

Remove the requirement that Subzero must not diverge from the upstream
repo.

Recommend using the MAKE_BUILD_PARALLEL_LEVEL environment variable to
change the default parallelism.

Fixes: b/166244497
Fixes: b/170565506
Change-Id: I00029e392282edd650712f031079073482365af3
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/51188
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>