[Clang] VectorExprEvaluator::VisitCallExpr - add x86 __builtin_ia32_select* constexpr handling

[RKSimon]

The following builtins need to be handled in constexpr to allow us to constexpr the avx512 predicated intrinsics:

 X86::BI__builtin_ia32_selectb_128:
 X86::BI__builtin_ia32_selectb_256:
 X86::BI__builtin_ia32_selectb_512:
 X86::BI__builtin_ia32_selectw_128:
 X86::BI__builtin_ia32_selectw_256:
 X86::BI__builtin_ia32_selectw_512:
 X86::BI__builtin_ia32_selectd_128:
 X86::BI__builtin_ia32_selectd_256:
 X86::BI__builtin_ia32_selectd_512:
 X86::BI__builtin_ia32_selectq_128:
 X86::BI__builtin_ia32_selectq_256:
 X86::BI__builtin_ia32_selectq_512:
 X86::BI__builtin_ia32_selectph_128:
 X86::BI__builtin_ia32_selectph_256:
 X86::BI__builtin_ia32_selectph_512:
 X86::BI__builtin_ia32_selectpbf_128:
 X86::BI__builtin_ia32_selectpbf_256:
 X86::BI__builtin_ia32_selectpbf_512:
 X86::BI__builtin_ia32_selectps_128:
 X86::BI__builtin_ia32_selectps_256:
 X86::BI__builtin_ia32_selectps_512:
 X86::BI__builtin_ia32_selectpd_128:
 X86::BI__builtin_ia32_selectpd_256:
 X86::BI__builtin_ia32_selectpd_512:

The expansion isn't too difficult:

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 58 to 76 in a9dacb1

	static Value *getMaskVecValue(CodeGenFunction &CGF, Value *Mask,
	unsigned NumElts) {
	auto *MaskTy = llvm::FixedVectorType::get(
	CGF.Builder.getInt1Ty(),
	cast<IntegerType>(Mask->getType())->getBitWidth());
	Value *MaskVec = CGF.Builder.CreateBitCast(Mask, MaskTy);
	// If we have less than 8 elements, then the starting mask was an i8 and
	// we need to extract down to the right number of elements.
	if (NumElts < 8) {
	int Indices[4];
	for (unsigned i = 0; i != NumElts; ++i)
	Indices[i] = i;
	MaskVec = CGF.Builder.CreateShuffleVector(
	MaskVec, MaskVec, ArrayRef(Indices, NumElts), "extract");
	}
	return MaskVec;
	}

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 210 to 222 in a9dacb1

	static Value *EmitX86Select(CodeGenFunction &CGF,
	Value *Mask, Value *Op0, Value *Op1) {
	// If the mask is all ones just return first argument.
	if (const auto *C = dyn_cast<Constant>(Mask))
	if (C->isAllOnesValue())
	return Op0;
	Mask = getMaskVecValue(
	CGF, Mask, cast<llvm::FixedVectorType>(Op0->getType())->getNumElements());
	return CGF.Builder.CreateSelect(Mask, Op0, Op1);
	}

ExprConstant.cpp already has precedent for handling target specific builtins (the x86 BMI intrinsics), so hopefully this isn't controversial.

[llvmbot]

@llvm/issue-subscribers-backend-x86

Author: Simon Pilgrim (RKSimon)

The following builtins need to be handled in constexpr to allow us to constexpr the avx512 predicated intrinsics: ``` X86::BI__builtin_ia32_selectb_128: X86::BI__builtin_ia32_selectb_256: X86::BI__builtin_ia32_selectb_512: X86::BI__builtin_ia32_selectw_128: X86::BI__builtin_ia32_selectw_256: X86::BI__builtin_ia32_selectw_512: X86::BI__builtin_ia32_selectd_128: X86::BI__builtin_ia32_selectd_256: X86::BI__builtin_ia32_selectd_512: X86::BI__builtin_ia32_selectq_128: X86::BI__builtin_ia32_selectq_256: X86::BI__builtin_ia32_selectq_512: X86::BI__builtin_ia32_selectph_128: X86::BI__builtin_ia32_selectph_256: X86::BI__builtin_ia32_selectph_512: X86::BI__builtin_ia32_selectpbf_128: X86::BI__builtin_ia32_selectpbf_256: X86::BI__builtin_ia32_selectpbf_512: X86::BI__builtin_ia32_selectps_128: X86::BI__builtin_ia32_selectps_256: X86::BI__builtin_ia32_selectps_512: X86::BI__builtin_ia32_selectpd_128: X86::BI__builtin_ia32_selectpd_256: X86::BI__builtin_ia32_selectpd_512: ```

The expansion isn't too difficult:

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 58 to 76 in a9dacb1

	static Value *getMaskVecValue(CodeGenFunction &CGF, Value *Mask,
	unsigned NumElts) {
	auto *MaskTy = llvm::FixedVectorType::get(
	CGF.Builder.getInt1Ty(),
	cast<IntegerType>(Mask->getType())->getBitWidth());
	Value *MaskVec = CGF.Builder.CreateBitCast(Mask, MaskTy);
	// If we have less than 8 elements, then the starting mask was an i8 and
	// we need to extract down to the right number of elements.
	if (NumElts < 8) {
	int Indices[4];
	for (unsigned i = 0; i != NumElts; ++i)
	Indices[i] = i;
	MaskVec = CGF.Builder.CreateShuffleVector(
	MaskVec, MaskVec, ArrayRef(Indices, NumElts), "extract");
	}
	return MaskVec;
	}

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 210 to 222 in a9dacb1

	static Value *EmitX86Select(CodeGenFunction &CGF,
	Value *Mask, Value *Op0, Value *Op1) {
	// If the mask is all ones just return first argument.
	if (const auto *C = dyn_cast<Constant>(Mask))
	if (C->isAllOnesValue())
	return Op0;
	Mask = getMaskVecValue(
	CGF, Mask, cast<llvm::FixedVectorType>(Op0->getType())->getNumElements());
	return CGF.Builder.CreateSelect(Mask, Op0, Op1);
	}

ExprConstant.cpp already has precedent for handling target specific builtins (the x86 BMI intrinsics), so hopefully this isn't controversial.

[llvmbot]

Hi!

This issue may be a good introductory issue for people new to working on LLVM. If you would like to work on this issue, your first steps are:

1. Check that no other contributor is working on this issue. If someone is assigned to the issue or claimed to be working on it, ping the person. After one week without a response, the assignee may be changed.
2. Leave a comment indicating that you are working on the issue, or just create a pull request after following the steps below. Mention this issue in the description of the pull request.
3. Fix the issue locally.
4. Run the test suite locally. Remember that the subdirectories under test/ create fine-grained testing targets, so you can e.g. use make check-clang-ast to only run Clang's AST tests.
5. Create a Git commit.
6. Run git clang-format HEAD~1 to format your changes.
7. Open a pull request to the upstream repository on GitHub. Detailed instructions can be found in GitHub's documentation. Mention this issue in the description of the pull request.

If you have any further questions about this issue, don't hesitate to ask via a comment in the thread below.

[llvmbot]

@llvm/issue-subscribers-good-first-issue

Author: Simon Pilgrim (RKSimon)

The following builtins need to be handled in constexpr to allow us to constexpr the avx512 predicated intrinsics: ``` X86::BI__builtin_ia32_selectb_128: X86::BI__builtin_ia32_selectb_256: X86::BI__builtin_ia32_selectb_512: X86::BI__builtin_ia32_selectw_128: X86::BI__builtin_ia32_selectw_256: X86::BI__builtin_ia32_selectw_512: X86::BI__builtin_ia32_selectd_128: X86::BI__builtin_ia32_selectd_256: X86::BI__builtin_ia32_selectd_512: X86::BI__builtin_ia32_selectq_128: X86::BI__builtin_ia32_selectq_256: X86::BI__builtin_ia32_selectq_512: X86::BI__builtin_ia32_selectph_128: X86::BI__builtin_ia32_selectph_256: X86::BI__builtin_ia32_selectph_512: X86::BI__builtin_ia32_selectpbf_128: X86::BI__builtin_ia32_selectpbf_256: X86::BI__builtin_ia32_selectpbf_512: X86::BI__builtin_ia32_selectps_128: X86::BI__builtin_ia32_selectps_256: X86::BI__builtin_ia32_selectps_512: X86::BI__builtin_ia32_selectpd_128: X86::BI__builtin_ia32_selectpd_256: X86::BI__builtin_ia32_selectpd_512: ```

The expansion isn't too difficult:

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 58 to 76 in a9dacb1

	static Value *getMaskVecValue(CodeGenFunction &CGF, Value *Mask,
	unsigned NumElts) {
	auto *MaskTy = llvm::FixedVectorType::get(
	CGF.Builder.getInt1Ty(),
	cast<IntegerType>(Mask->getType())->getBitWidth());
	Value *MaskVec = CGF.Builder.CreateBitCast(Mask, MaskTy);
	// If we have less than 8 elements, then the starting mask was an i8 and
	// we need to extract down to the right number of elements.
	if (NumElts < 8) {
	int Indices[4];
	for (unsigned i = 0; i != NumElts; ++i)
	Indices[i] = i;
	MaskVec = CGF.Builder.CreateShuffleVector(
	MaskVec, MaskVec, ArrayRef(Indices, NumElts), "extract");
	}
	return MaskVec;
	}

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 210 to 222 in a9dacb1

	static Value *EmitX86Select(CodeGenFunction &CGF,
	Value *Mask, Value *Op0, Value *Op1) {
	// If the mask is all ones just return first argument.
	if (const auto *C = dyn_cast<Constant>(Mask))
	if (C->isAllOnesValue())
	return Op0;
	Mask = getMaskVecValue(
	CGF, Mask, cast<llvm::FixedVectorType>(Op0->getType())->getNumElements());
	return CGF.Builder.CreateSelect(Mask, Op0, Op1);
	}

ExprConstant.cpp already has precedent for handling target specific builtins (the x86 BMI intrinsics), so hopefully this isn't controversial.

[llvmbot]

@llvm/issue-subscribers-clang-frontend

Author: Simon Pilgrim (RKSimon)

The following builtins need to be handled in constexpr to allow us to constexpr the avx512 predicated intrinsics: ``` X86::BI__builtin_ia32_selectb_128: X86::BI__builtin_ia32_selectb_256: X86::BI__builtin_ia32_selectb_512: X86::BI__builtin_ia32_selectw_128: X86::BI__builtin_ia32_selectw_256: X86::BI__builtin_ia32_selectw_512: X86::BI__builtin_ia32_selectd_128: X86::BI__builtin_ia32_selectd_256: X86::BI__builtin_ia32_selectd_512: X86::BI__builtin_ia32_selectq_128: X86::BI__builtin_ia32_selectq_256: X86::BI__builtin_ia32_selectq_512: X86::BI__builtin_ia32_selectph_128: X86::BI__builtin_ia32_selectph_256: X86::BI__builtin_ia32_selectph_512: X86::BI__builtin_ia32_selectpbf_128: X86::BI__builtin_ia32_selectpbf_256: X86::BI__builtin_ia32_selectpbf_512: X86::BI__builtin_ia32_selectps_128: X86::BI__builtin_ia32_selectps_256: X86::BI__builtin_ia32_selectps_512: X86::BI__builtin_ia32_selectpd_128: X86::BI__builtin_ia32_selectpd_256: X86::BI__builtin_ia32_selectpd_512: ```

The expansion isn't too difficult:

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 58 to 76 in a9dacb1

	static Value *getMaskVecValue(CodeGenFunction &CGF, Value *Mask,
	unsigned NumElts) {
	auto *MaskTy = llvm::FixedVectorType::get(
	CGF.Builder.getInt1Ty(),
	cast<IntegerType>(Mask->getType())->getBitWidth());
	Value *MaskVec = CGF.Builder.CreateBitCast(Mask, MaskTy);
	// If we have less than 8 elements, then the starting mask was an i8 and
	// we need to extract down to the right number of elements.
	if (NumElts < 8) {
	int Indices[4];
	for (unsigned i = 0; i != NumElts; ++i)
	Indices[i] = i;
	MaskVec = CGF.Builder.CreateShuffleVector(
	MaskVec, MaskVec, ArrayRef(Indices, NumElts), "extract");
	}
	return MaskVec;
	}

llvm-project/clang/lib/CodeGen/TargetBuiltins/X86.cpp

Lines 210 to 222 in a9dacb1

	static Value *EmitX86Select(CodeGenFunction &CGF,
	Value *Mask, Value *Op0, Value *Op1) {
	// If the mask is all ones just return first argument.
	if (const auto *C = dyn_cast<Constant>(Mask))
	if (C->isAllOnesValue())
	return Op0;
	Mask = getMaskVecValue(
	CGF, Mask, cast<llvm::FixedVectorType>(Op0->getType())->getNumElements());
	return CGF.Builder.CreateSelect(Mask, Op0, Op1);
	}

ExprConstant.cpp already has precedent for handling target specific builtins (the x86 BMI intrinsics), so hopefully this isn't controversial.