Replace uses of floating point log2 when we want an integer result from
an integer operand.  This avoids concerns about accuracy of floating
point library functions.

callGraph.push_front -> emplace_front to fix UBSAN

Add support for GL_EXT_shared_memory_block

UBSAN rightly complains on `push_front` here:

    glslang/MachineIndependent/localintermediate.h:100:8: runtime error: load of value 160, which is not a valid value for type 'bool'
    #0 in glslang::TCall::TCall(glslang::TCall&&) glslang/MachineIndependent/localintermediate.h:100
    #1 in void __gnu_cxx::new_allocator<std::_List_node<glslang::TCall> >::construct<glslang::TCall, glslang::TCall>(glslang::TCall*, glslang::TCall&&) /usr/include/c++/10/ext/new_allocator.h:150
    #2 in void std::allocator_traits<std::allocator<std::_List_node<glslang::TCall> > >::construct<glslang::TCall, glslang::TCall>(std::allocator<std::_List_node<glslang::TCall> >&, glslang::TCall*, glslang::TCall&&) /usr/include/c++/10/bits/alloc_traits.h:512
    #3 in std::_List_node<glslang::TCall>* std::__cxx11::list<glslang::TCall, std::allocator<glslang::TCall> >::_M_create_node<glslang::TCall>(glslang::TCall&&) (...)
    #4 in void std::__cxx11::list<glslang::TCall, std::allocator<glslang::TCall> >::_M_insert<glslang::TCall>(std::_List_iterator<glslang::TCall>, glslang::TCall&&) /usr/include/c++/10/bits/stl_list.h:1911
    #5 in std::__cxx11::list<glslang::TCall, std::allocator<glslang::TCall> >::push_front(glslang::TCall&&) /usr/include/c++/10/bits/stl_list.h:1167
    #6 in glslang::TIntermediate::addToCallGraph(TInfoSink&, std::__cxx11::basic_string<char, std::char_traits<char>, glslang::pool_allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, glslang::pool_allocator<char> > const&) glslang/MachineIndependent/Intermediate.cpp:2860

What happens here:

1. TCall's bool fields are not initialized on construction.
2. `push_front` move the `TCall` passed into it.
3. The move constructor copies unitialized bool, which may have an
   out-of-range value.

What this fix does:

Calls `emplace_back` to ensure no copy/move constructor is called.

Fixes #2222
Refs #2112

Uses SPV_KHR_workgroup_memory_explicit_layout.  Note that if
GL_EXT_scalar_block_layout is enabled, Workgroup blocks can also use
scalar layout.

Also update the test expectations regarding validation accordingly and
the in-tree spirv.hpp copy.

If GL_EXT_scalar_block_layout is requested by the shader, set the
option to allow scalar blocks in the SPIR-V validator.

Fix the existing tests using scalar layout to not expect "Validation
failed".

Fixes #2400.

Add missing capability when QueueFamily scope is used

Also, if this capability is added and the memory model is not
otherwise enabled by pragma, enable it as part of postprocessing.

Implement GL_EXT_null_initializer

Adds null initializer syntax (empty braces)
Allows null initialization of shared variables

Fix missing enum values in the C-interface #2507

license-checker.cfg: Update rules

Add CMake support for Fuchsia

The license classification rules in github.com/google/licensecheck have been updated.

Fix option replacement when flags are empty

Prevent HLSL input under OpenGL semantics

Fix debugInfo test to target vulkan1.1 as intended

Fix #2497.

Fixes #2494

For the scenario when cmake is executed from MSVC with empty CXX flags.
Signed-off-by: Denis Zalevskiy <denis.zalevskiy@basemark.com>

Fix build with android ndk r16b

Fix tess factor cut and paste error

Fixes #2476

Don't compile round() to roundEven in DX9 compatibility mode

Fixed compile warning in reflection.cpp for ENABLE_HLSL = 0 build. [-Wunused-parameter]

Update SPIRV-Tools known-good: validate OpImageRead result type

See #2486

Pick up https://github.com/KhronosGroup/SPIRV-Tools/pull/4072
which validates the result type of OpImageRead.

This also requires test expectation updates.

See #2486 tracks the needed upate to Glslang code generation.

Issue #2483

According to GLSL spec the prototype is:
        uint rayQueryGetIntersectionInstanceShaderBindingTableRecordOffsetEXT(rayQueryEXT q, bool committed);

but that was incorrectly getting translated to SPIRV as an `int`, and this was
causing SPIR-V validation errors when used.

Added explicit testing for the return types of all the builtin functions in GL_EXT_ray_query

Update README to avoid googletest breakage

Additional fix for #2480