[DXIL] exp, any, lerp, & rcp Intrinsic Lowering (#84526)

This change implements lowering for #70076, #70100, #70072, & #70102 
`CGBuiltin.cpp` - - simplify `lerp` intrinsic
`IntrinsicsDirectX.td` - simplify `lerp` intrinsic
`SemaChecking.cpp` - remove unnecessary check
`DXILIntrinsicExpansion.*` - add intrinsic to instruction expansion
cases
`DXILOpLowering.cpp` - make sure `DXILIntrinsicExpansion` happens first
`DirectX.h` - changes to support new pass
`DirectXTargetMachine.cpp` - changes to support new pass

Why `any`, and `lerp` as instruction expansion just for DXIL?
- SPIR-V there is an
[OpAny](https://registry.khronos.org/SPIR-V/specs/unified1/SPIRV.html#OpAny)
- SPIR-V has a GLSL lerp extension via
[Fmix](https://registry.khronos.org/SPIR-V/specs/1.0/GLSL.std.450.html#FMix)

Why `exp` instruction expansion?
- We have an `exp2` opcode and `exp` reuses that opcode. So instruction
expansion is a convenient way to do preprocessing.
- Further SPIR-V has a GLSL exp extension via
[Exp](https://registry.khronos.org/SPIR-V/specs/1.0/GLSL.std.450.html#Exp)
and
[Exp2](https://registry.khronos.org/SPIR-V/specs/1.0/GLSL.std.450.html#Exp2)

Why `rcp` as instruction expansion?
This one is a bit of the odd man out and might have to move to
`cgbuiltins` when we better understand SPIRV requirements. However I
included it because it seems like [fast math mode has an AllowRecip
flag](https://registry.khronos.org/SPIR-V/specs/unified1/SPIRV.html#_fp_fast_math_mode)
which lets you compute the reciprocal without performing the division.
We don't have that in DXIL so thought to include it.

clang/include/clang/AST/Type.h

@@ -2244,6 +2244,7 @@ class alignas(TypeAlignment) Type : public ExtQualsTypeCommonBase {
   bool isFloatingType() const;     // C99 6.2.5p11 (real floating + complex)
   bool isHalfType() const;         // OpenCL 6.1.1.1, NEON (IEEE 754-2008 half)
   bool isFloat16Type() const;      // C11 extension ISO/IEC TS 18661
+  bool isFloat32Type() const;
   bool isBFloat16Type() const;
   bool isFloat128Type() const;
   bool isIbm128Type() const;
@@ -7452,6 +7453,10 @@ inline bool Type::isFloat16Type() const {
   return isSpecificBuiltinType(BuiltinType::Float16);
 }
 
+inline bool Type::isFloat32Type() const {
+  return isSpecificBuiltinType(BuiltinType::Float);
+}
+
 inline bool Type::isBFloat16Type() const {
   return isSpecificBuiltinType(BuiltinType::BFloat16);
 }

clang/include/clang/Basic/Builtins.td

@@ -4598,7 +4598,7 @@ def HLSLRcp : LangBuiltin<"HLSL_LANG"> {
 
 def HLSLRSqrt : LangBuiltin<"HLSL_LANG"> {
   let Spellings = ["__builtin_hlsl_elementwise_rsqrt"];
-  let Attributes = [NoThrow, Const, CustomTypeChecking];
+  let Attributes = [NoThrow, Const];
   let Prototype = "void(...)";
 }
 

clang/lib/CodeGen/CGBuiltin.cpp

@@ -18021,38 +18021,11 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
     Value *X = EmitScalarExpr(E->getArg(0));
     Value *Y = EmitScalarExpr(E->getArg(1));
     Value *S = EmitScalarExpr(E->getArg(2));
-    llvm::Type *Xty = X->getType();
-    llvm::Type *Yty = Y->getType();
-    llvm::Type *Sty = S->getType();
-    if (!Xty->isVectorTy() && !Yty->isVectorTy() && !Sty->isVectorTy()) {
-      if (Xty->isFloatingPointTy()) {
-        auto V = Builder.CreateFSub(Y, X);
-        V = Builder.CreateFMul(S, V);
-        return Builder.CreateFAdd(X, V, "dx.lerp");
-      }
-      llvm_unreachable("Scalar Lerp is only supported on floats.");
-    }
-    // A VectorSplat should have happened
-    assert(Xty->isVectorTy() && Yty->isVectorTy() && Sty->isVectorTy() &&
-           "Lerp of vector and scalar is not supported.");
-
-    [[maybe_unused]] auto *XVecTy =
-        E->getArg(0)->getType()->getAs<VectorType>();
-    [[maybe_unused]] auto *YVecTy =
-        E->getArg(1)->getType()->getAs<VectorType>();
-    [[maybe_unused]] auto *SVecTy =
-        E->getArg(2)->getType()->getAs<VectorType>();
-    // A HLSLVectorTruncation should have happend
-    assert(XVecTy->getNumElements() == YVecTy->getNumElements() &&
-           XVecTy->getNumElements() == SVecTy->getNumElements() &&
-           "Lerp requires vectors to be of the same size.");
-    assert(XVecTy->getElementType()->isRealFloatingType() &&
-           XVecTy->getElementType() == YVecTy->getElementType() &&
-           XVecTy->getElementType() == SVecTy->getElementType() &&
-           "Lerp requires float vectors to be of the same type.");
+    if (!E->getArg(0)->getType()->hasFloatingRepresentation())
+      llvm_unreachable("lerp operand must have a float representation");
     return Builder.CreateIntrinsic(
-        /*ReturnType=*/Xty, Intrinsic::dx_lerp, ArrayRef<Value *>{X, Y, S},
-        nullptr, "dx.lerp");
+        /*ReturnType=*/X->getType(), Intrinsic::dx_lerp,
+        ArrayRef<Value *>{X, Y, S}, nullptr, "dx.lerp");
   }
   case Builtin::BI__builtin_hlsl_elementwise_frac: {
     Value *Op0 = EmitScalarExpr(E->getArg(0));

clang/lib/Sema/SemaChecking.cpp

@@ -5234,10 +5234,6 @@ bool CheckVectorElementCallArgs(Sema *S, CallExpr *TheCall) {
                            TheCall->getArg(1)->getEndLoc());
         retValue = true;
       }
-
-      if (!retValue)
-        TheCall->setType(VecTyA->getElementType());
-
       return retValue;
     }
   }
@@ -5251,11 +5247,12 @@ bool CheckVectorElementCallArgs(Sema *S, CallExpr *TheCall) {
   return true;
 }
 
-bool CheckAllArgsHaveFloatRepresentation(Sema *S, CallExpr *TheCall) {
-  QualType ExpectedType = S->Context.FloatTy;
+bool CheckArgsTypesAreCorrect(
+    Sema *S, CallExpr *TheCall, QualType ExpectedType,
+    llvm::function_ref<bool(clang::QualType PassedType)> Check) {
   for (unsigned i = 0; i < TheCall->getNumArgs(); ++i) {
     QualType PassedType = TheCall->getArg(i)->getType();
-    if (!PassedType->hasFloatingRepresentation()) {
+    if (Check(PassedType)) {
       if (auto *VecTyA = PassedType->getAs<VectorType>())
         ExpectedType = S->Context.getVectorType(
             ExpectedType, VecTyA->getNumElements(), VecTyA->getVectorKind());
@@ -5268,6 +5265,26 @@ bool CheckAllArgsHaveFloatRepresentation(Sema *S, CallExpr *TheCall) {
   return false;
 }
 
+bool CheckAllArgsHaveFloatRepresentation(Sema *S, CallExpr *TheCall) {
+  auto checkAllFloatTypes = [](clang::QualType PassedType) -> bool {
+    return !PassedType->hasFloatingRepresentation();
+  };
+  return CheckArgsTypesAreCorrect(S, TheCall, S->Context.FloatTy,
+                                  checkAllFloatTypes);
+}
+
+bool CheckFloatOrHalfRepresentations(Sema *S, CallExpr *TheCall) {
+  auto checkFloatorHalf = [](clang::QualType PassedType) -> bool {
+    clang::QualType BaseType =
+        PassedType->isVectorType()
+            ? PassedType->getAs<clang::VectorType>()->getElementType()
+            : PassedType;
+    return !BaseType->isHalfType() && !BaseType->isFloat32Type();
+  };
+  return CheckArgsTypesAreCorrect(S, TheCall, S->Context.FloatTy,
+                                  checkFloatorHalf);
+}
+
 void SetElementTypeAsReturnType(Sema *S, CallExpr *TheCall,
                                 QualType ReturnType) {
   auto *VecTyA = TheCall->getArg(0)->getType()->getAs<VectorType>();
@@ -5295,21 +5312,27 @@ bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
       return true;
     break;
   }
-  case Builtin::BI__builtin_hlsl_elementwise_isinf: {
-    if (checkArgCount(*this, TheCall, 1))
-      return true;
+  case Builtin::BI__builtin_hlsl_elementwise_rcp: {
     if (CheckAllArgsHaveFloatRepresentation(this, TheCall))
       return true;
-    SetElementTypeAsReturnType(this, TheCall, this->Context.BoolTy);
+    if (PrepareBuiltinElementwiseMathOneArgCall(TheCall))
+      return true;
     break;
   }
   case Builtin::BI__builtin_hlsl_elementwise_rsqrt:
-  case Builtin::BI__builtin_hlsl_elementwise_rcp:
   case Builtin::BI__builtin_hlsl_elementwise_frac: {
-    if (CheckAllArgsHaveFloatRepresentation(this, TheCall))
+    if (CheckFloatOrHalfRepresentations(this, TheCall))
+      return true;
+    if (PrepareBuiltinElementwiseMathOneArgCall(TheCall))
+      return true;
+    break;
+  }
+  case Builtin::BI__builtin_hlsl_elementwise_isinf: {
+    if (CheckFloatOrHalfRepresentations(this, TheCall))
       return true;
     if (PrepareBuiltinElementwiseMathOneArgCall(TheCall))
       return true;
+    SetElementTypeAsReturnType(this, TheCall, this->Context.BoolTy);
     break;
   }
   case Builtin::BI__builtin_hlsl_lerp: {
@@ -5319,7 +5342,7 @@ bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
       return true;
     if (SemaBuiltinElementwiseTernaryMath(TheCall))
       return true;
-    if (CheckAllArgsHaveFloatRepresentation(this, TheCall))
+    if (CheckFloatOrHalfRepresentations(this, TheCall))
       return true;
     break;
   }

clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl

@@ -1,27 +1,5 @@
 // RUN: %clang_cc1 -finclude-default-header -x hlsl -triple dxil-pc-shadermodel6.3-library %s -fnative-half-type -emit-llvm -disable-llvm-passes -o - | FileCheck %s
 
-
-
-// CHECK-LABEL: builtin_lerp_half_scalar
-// CHECK: %3 = fsub double %conv1, %conv
-// CHECK: %4 = fmul double %conv2, %3
-// CHECK: %dx.lerp = fadd double %conv, %4
-// CHECK: %conv3 = fptrunc double %dx.lerp to half
-// CHECK: ret half %conv3
-half builtin_lerp_half_scalar (half p0) {
-  return __builtin_hlsl_lerp ( p0, p0, p0 );
-}
-
-// CHECK-LABEL: builtin_lerp_float_scalar
-// CHECK: %3 = fsub double %conv1, %conv
-// CHECK: %4 = fmul double %conv2, %3
-// CHECK: %dx.lerp = fadd double %conv, %4
-// CHECK: %conv3 = fptrunc double %dx.lerp to float
-// CHECK: ret float %conv3
-float builtin_lerp_float_scalar ( float p0) {
-  return __builtin_hlsl_lerp ( p0, p0, p0 );
-}
-
 // CHECK-LABEL: builtin_lerp_half_vector
 // CHECK: %dx.lerp = call <3 x half> @llvm.dx.lerp.v3f16(<3 x half> %0, <3 x half> %1, <3 x half> %2)
 // CHECK: ret <3 x half> %dx.lerp

clang/test/CodeGenHLSL/builtins/lerp.hlsl

@@ -6,51 +6,46 @@
 // RUN:   dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes \
 // RUN:   -o - | FileCheck %s --check-prefixes=CHECK,NO_HALF
 
-// NATIVE_HALF: %3 = fsub half %1, %0
-// NATIVE_HALF: %4 = fmul half %2, %3
-// NATIVE_HALF: %dx.lerp = fadd half %0, %4
+
+// NATIVE_HALF: %dx.lerp = call half @llvm.dx.lerp.f16(half %0, half %1, half %2)
 // NATIVE_HALF: ret half %dx.lerp
-// NO_HALF: %3 = fsub float %1, %0
-// NO_HALF: %4 = fmul float %2, %3
-// NO_HALF: %dx.lerp = fadd float %0, %4
+// NO_HALF: %dx.lerp = call float @llvm.dx.lerp.f32(float %0, float %1, float %2)
 // NO_HALF: ret float %dx.lerp
 half test_lerp_half(half p0) { return lerp(p0, p0, p0); }
 
 // NATIVE_HALF: %dx.lerp = call <2 x half> @llvm.dx.lerp.v2f16(<2 x half> %0, <2 x half> %1, <2 x half> %2)
 // NATIVE_HALF: ret <2 x half> %dx.lerp
 // NO_HALF: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
 // NO_HALF: ret <2 x float> %dx.lerp
-half2 test_lerp_half2(half2 p0, half2 p1) { return lerp(p0, p0, p0); }
+half2 test_lerp_half2(half2 p0) { return lerp(p0, p0, p0); }
 
 // NATIVE_HALF: %dx.lerp = call <3 x half> @llvm.dx.lerp.v3f16(<3 x half> %0, <3 x half> %1, <3 x half> %2)
 // NATIVE_HALF: ret <3 x half> %dx.lerp
 // NO_HALF: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %2)
 // NO_HALF: ret <3 x float> %dx.lerp
-half3 test_lerp_half3(half3 p0, half3 p1) { return lerp(p0, p0, p0); }
+half3 test_lerp_half3(half3 p0) { return lerp(p0, p0, p0); }
 
 // NATIVE_HALF: %dx.lerp = call <4 x half> @llvm.dx.lerp.v4f16(<4 x half> %0, <4 x half> %1, <4 x half> %2)
 // NATIVE_HALF: ret <4 x half> %dx.lerp
 // NO_HALF: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2)
 // NO_HALF: ret <4 x float> %dx.lerp
-half4 test_lerp_half4(half4 p0, half4 p1) { return lerp(p0, p0, p0); }
+half4 test_lerp_half4(half4 p0) { return lerp(p0, p0, p0); }
 
-// CHECK: %3 = fsub float %1, %0
-// CHECK: %4 = fmul float %2, %3
-// CHECK: %dx.lerp = fadd float %0, %4
+// CHECK: %dx.lerp = call float @llvm.dx.lerp.f32(float %0, float %1, float %2)
 // CHECK: ret float %dx.lerp
-float test_lerp_float(float p0, float p1) { return lerp(p0, p0, p0); }
+float test_lerp_float(float p0) { return lerp(p0, p0, p0); }
 
 // CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
 // CHECK: ret <2 x float> %dx.lerp
-float2 test_lerp_float2(float2 p0, float2 p1) { return lerp(p0, p0, p0); }
+float2 test_lerp_float2(float2 p0) { return lerp(p0, p0, p0); }
 
 // CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %2)
 // CHECK: ret <3 x float> %dx.lerp
-float3 test_lerp_float3(float3 p0, float3 p1) { return lerp(p0, p0, p0); }
+float3 test_lerp_float3(float3 p0) { return lerp(p0, p0, p0); }
 
 // CHECK: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2)
 // CHECK: ret <4 x float> %dx.lerp
-float4 test_lerp_float4(float4 p0, float4 p1) { return lerp(p0, p0, p0); }
+float4 test_lerp_float4(float4 p0) { return lerp(p0, p0, p0); }
 
 // CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %splat.splat, <2 x float> %1, <2 x float> %2)
 // CHECK: ret <2 x float> %dx.lerp

clang/test/SemaHLSL/BuiltIns/frac-errors.hlsl

@@ -25,3 +25,15 @@ float2 builtin_frac_int2_to_float2_promotion(int2 p1) {
   return __builtin_hlsl_elementwise_frac(p1);
   // expected-error@-1 {{passing 'int2' (aka 'vector<int, 2>') to parameter of incompatible type '__attribute__((__vector_size__(2 * sizeof(float)))) float' (vector of 2 'float' values)}}
 }
+
+// builtins are variadic functions and so are subject to DefaultVariadicArgumentPromotion
+half builtin_frac_half_scalar (half p0) {
+  return __builtin_hlsl_elementwise_frac (p0);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}
+
+float builtin_frac_float_scalar ( float p0) {
+  return __builtin_hlsl_elementwise_frac (p0);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}
+

clang/test/SemaHLSL/BuiltIns/isinf-errors.hlsl

@@ -25,3 +25,14 @@ bool2 builtin_isinf_int2_to_float2_promotion(int2 p1) {
   return __builtin_hlsl_elementwise_isinf(p1);
   // expected-error@-1 {{passing 'int2' (aka 'vector<int, 2>') to parameter of incompatible type '__attribute__((__vector_size__(2 * sizeof(float)))) float' (vector of 2 'float' values)}}
 }
+
+// builtins are variadic functions and so are subject to DefaultVariadicArgumentPromotion
+half builtin_isinf_half_scalar (half p0) {
+  return __builtin_hlsl_elementwise_isinf (p0);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}
+
+float builtin_isinf_float_scalar ( float p0) {
+  return __builtin_hlsl_elementwise_isinf (p0);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}

clang/test/SemaHLSL/BuiltIns/lerp-errors.hlsl

@@ -92,5 +92,18 @@ float builtin_lerp_int_to_float_promotion(float p0, int p1) {
 
 float4 test_lerp_int4(int4 p0, int4 p1, int4 p2) {
   return __builtin_hlsl_lerp(p0, p1, p2);
-   // expected-error@-1 {{1st argument must be a floating point type (was 'int4' (aka 'vector<int, 4>'))}}
-}
+  // expected-error@-1 {{1st argument must be a floating point type (was 'int4' (aka 'vector<int, 4>'))}}
+}
+
+// note: DefaultVariadicArgumentPromotion --> DefaultArgumentPromotion has already promoted to double
+// we don't know anymore that the input was half when __builtin_hlsl_lerp is called so we default to float
+// for expected type
+half builtin_lerp_half_scalar (half p0) {
+  return __builtin_hlsl_lerp ( p0, p0, p0 );
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}
+
+float builtin_lerp_float_scalar ( float p0) {
+  return __builtin_hlsl_lerp ( p0, p0, p0 );
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}

clang/test/SemaHLSL/BuiltIns/rsqrt-errors.hlsl

@@ -25,3 +25,14 @@ float2 builtin_rsqrt_int2_to_float2_promotion(int2 p1) {
   return __builtin_hlsl_elementwise_rsqrt(p1);
   // expected-error@-1 {{passing 'int2' (aka 'vector<int, 2>') to parameter of incompatible type '__attribute__((__vector_size__(2 * sizeof(float)))) float' (vector of 2 'float' values)}}
 }
+
+// builtins are variadic functions and so are subject to DefaultVariadicArgumentPromotion
+half builtin_rsqrt_half_scalar (half p0) {
+  return __builtin_hlsl_elementwise_rsqrt (p0);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}
+
+float builtin_rsqrt_float_scalar ( float p0) {
+  return __builtin_hlsl_elementwise_rsqrt (p0);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}

llvm/include/llvm/IR/IntrinsicsDirectX.td

@@ -33,9 +33,7 @@ def int_dx_isinf :
     DefaultAttrsIntrinsic<[LLVMScalarOrSameVectorWidth<0, llvm_i1_ty>],
     [llvm_anyfloat_ty]>;
 
-def int_dx_lerp :
-    Intrinsic<[LLVMScalarOrSameVectorWidth<0, LLVMVectorElementType<0>>],
-    [llvm_anyvector_ty, LLVMScalarOrSameVectorWidth<0, LLVMVectorElementType<0>>,LLVMScalarOrSameVectorWidth<0, LLVMVectorElementType<0>>],
+def int_dx_lerp : Intrinsic<[LLVMMatchType<0>], [llvm_anyfloat_ty, LLVMMatchType<0>,LLVMMatchType<0>], 
     [IntrNoMem, IntrWillReturn] >;
 
 def int_dx_imad : DefaultAttrsIntrinsic<[llvm_anyint_ty], [LLVMMatchType<0>, LLVMMatchType<0>, LLVMMatchType<0>]>;

llvm/lib/Target/DirectX/CMakeLists.txt

@@ -19,6 +19,7 @@ add_llvm_target(DirectXCodeGen
   DirectXSubtarget.cpp
   DirectXTargetMachine.cpp
   DXContainerGlobals.cpp
+  DXILIntrinsicExpansion.cpp
   DXILMetadata.cpp
   DXILOpBuilder.cpp
   DXILOpLowering.cpp