Implement the `unpack_u8u16` HLSL Function

[farzonl]

• Implement unpack_u8u16 clang builtin,
• Link unpack_u8u16 clang builtin with hlsl_intrinsics.h
• Add sema checks for unpack_u8u16 to CheckHLSLBuiltinFunctionCall in SemaChecking.cpp
• Add codegen for unpack_u8u16 to EmitHLSLBuiltinExpr in CGBuiltin.cpp
• Add codegen tests to clang/test/CodeGenHLSL/builtins/unpack_u8u16.hlsl
• Add sema tests to clang/test/SemaHLSL/BuiltIns/unpack_u8u16-errors.hlsl
• Create the int_dx_unpack_u8u16 intrinsic in IntrinsicsDirectX.td
• Create the DXILOpMapping of int_dx_unpack_u8u16 to 219 in DXIL.td
• Create the unpack_u8u16.ll and unpack_u8u16_errors.ll tests in llvm/test/CodeGen/DirectX/
• Create the int_spv_unpack_u8u16 intrinsic in IntrinsicsSPIRV.td
• In SPIRVInstructionSelector.cpp create the unpack_u8u16 lowering and map it to int_spv_unpack_u8u16 in SPIRVInstructionSelector::selectIntrinsic.
• Create SPIR-V backend test case in llvm/test/CodeGen/SPIRV/hlsl-intrinsics/unpack_u8u16.ll

DirectX

DXIL Opcode	DXIL OpName	Shader Model	Shader Stages
219	Unpack4x8	6.6	()

SPIR-V

OpUConvert:

Description:

Convert unsigned width. This is either a truncate or a zero extend.

Result Type must be a scalar or vector of integer type,
whose Signedness operand is 0.

Unsigned Value must be a scalar or vector of integer
type. It must have the same number of components as Result
Type. The component width must not equal the component width in Result
Type.

Results are computed per component.

Word Count	Opcode	Results	Operands
4	113	<id> Result Type	Result <id>	<id> Unsigned Value

Test Case(s)

Example 1

//dxc unpack_u8u16_test.hlsl -T lib_6_8 -enable-16bit-types -O0

export uint16_t4 fn(uint8_t4_packed p1) {
    return unpack_u8u16(p1);
}

HLSL:

Syntax

uint16_t<4> unpack_u8u16(p32u8 pk);

Type Description

Name	Template Type	Component Type	Size
ret	vector	uint16_t	4
pk	scalar	uint8_t4_packed	1

Minimum Shader Model

This function is supported in the following shader models.

Shader Model	Supported
Shader Model 6.6 and higher shader models	yes

Shader Stages

See also

• Intrinsic Functions (DirectX HLSL)
• See Unpack Intrinsics

[Alexander-Johnston]

Happy to pick this up.