Add step builtins and step HLSL function to DirectX and SPIR-V backend (#106471)

This PR adds the step intrinsic and an HLSL function that uses it.
The SPIRV backend is also implemented.

Used #102683 as a reference.
Fixes #99157

Author: bob80905

clang/include/clang/Basic/Builtins.td

@@ -4763,6 +4763,7 @@ def HLSLSaturate : LangBuiltin<"HLSL_LANG"> {
   let Prototype = "void(...)";
 }
 
+
 def HLSLSelect : LangBuiltin<"HLSL_LANG"> {
   let Spellings = ["__builtin_hlsl_select"];
   let Attributes = [NoThrow, Const];
@@ -4775,6 +4776,12 @@ def HLSLSign : LangBuiltin<"HLSL_LANG"> {
   let Prototype = "void(...)";
 }
 
+def HLSLStep: LangBuiltin<"HLSL_LANG"> {
+  let Spellings = ["__builtin_hlsl_step"];
+  let Attributes = [NoThrow, Const];
+  let Prototype = "void(...)";
+}
+
 // Builtins for XRay.
 def XRayCustomEvent : Builtin {
   let Spellings = ["__xray_customevent"];

clang/lib/CodeGen/CGBuiltin.cpp

@@ -18861,6 +18861,16 @@ case Builtin::BI__builtin_hlsl_elementwise_isinf: {
 
     return SelectVal;
   }
+  case Builtin::BI__builtin_hlsl_step: {
+    Value *Op0 = EmitScalarExpr(E->getArg(0));
+    Value *Op1 = EmitScalarExpr(E->getArg(1));
+    assert(E->getArg(0)->getType()->hasFloatingRepresentation() &&
+           E->getArg(1)->getType()->hasFloatingRepresentation() &&
+           "step operands must have a float representation");
+    return Builder.CreateIntrinsic(
+        /*ReturnType=*/Op0->getType(), CGM.getHLSLRuntime().getStepIntrinsic(),
+        ArrayRef<Value *>{Op0, Op1}, nullptr, "hlsl.step");
+  }
   case Builtin::BI__builtin_hlsl_wave_get_lane_index: {
     return EmitRuntimeCall(CGM.CreateRuntimeFunction(
         llvm::FunctionType::get(IntTy, {}, false), "__hlsl_wave_get_lane_index",

clang/lib/CodeGen/CGHLSLRuntime.h

@@ -81,6 +81,7 @@ class CGHLSLRuntime {
   GENERATE_HLSL_INTRINSIC_FUNCTION(Rsqrt, rsqrt)
   GENERATE_HLSL_INTRINSIC_FUNCTION(Saturate, saturate)
   GENERATE_HLSL_INTRINSIC_FUNCTION(Sign, sign)
+  GENERATE_HLSL_INTRINSIC_FUNCTION(Step, step)
   GENERATE_HLSL_INTRINSIC_FUNCTION(ThreadId, thread_id)
   GENERATE_HLSL_INTRINSIC_FUNCTION(FDot, fdot)
   GENERATE_HLSL_INTRINSIC_FUNCTION(SDot, sdot)

clang/lib/Headers/hlsl/hlsl_intrinsics.h

@@ -1717,6 +1717,39 @@ float3 sqrt(float3);
 _HLSL_BUILTIN_ALIAS(__builtin_elementwise_sqrt)
 float4 sqrt(float4);
 
+//===----------------------------------------------------------------------===//
+// step builtins
+//===----------------------------------------------------------------------===//
+
+/// \fn T step(T x, T y)
+/// \brief Returns 1 if the x parameter is greater than or equal to the y
+/// parameter; otherwise, 0. vector. \param x [in] The first floating-point
+/// value to compare. \param y [in] The first floating-point value to compare.
+///
+/// Step is based on the following formula: (x >= y) ? 1 : 0
+
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+half step(half, half);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+half2 step(half2, half2);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+half3 step(half3, half3);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+half4 step(half4, half4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+float step(float, float);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+float2 step(float2, float2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+float3 step(float3, float3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_step)
+float4 step(float4, float4);
+
 //===----------------------------------------------------------------------===//
 // tan builtins
 //===----------------------------------------------------------------------===//

clang/lib/Sema/SemaHLSL.cpp

@@ -1747,6 +1747,18 @@ bool SemaHLSL::CheckBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
     SetElementTypeAsReturnType(&SemaRef, TheCall, getASTContext().IntTy);
     break;
   }
+  case Builtin::BI__builtin_hlsl_step: {
+    if (SemaRef.checkArgCount(TheCall, 2))
+      return true;
+    if (CheckFloatOrHalfRepresentations(&SemaRef, TheCall))
+      return true;
+
+    ExprResult A = TheCall->getArg(0);
+    QualType ArgTyA = A.get()->getType();
+    // return type is the same as the input type
+    TheCall->setType(ArgTyA);
+    break;
+  }
   // Note these are llvm builtins that we want to catch invalid intrinsic
   // generation. Normal handling of these builitns will occur elsewhere.
   case Builtin::BI__builtin_elementwise_bitreverse: {

clang/test/CodeGenHLSL/builtins/step.hlsl

@@ -0,0 +1,84 @@
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   dxil-pc-shadermodel6.3-library %s -fnative-half-type \
+// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s \
+// RUN:   --check-prefixes=CHECK,NATIVE_HALF \
+// RUN:   -DFNATTRS=noundef -DTARGET=dx
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes \
+// RUN:   -o - | FileCheck %s --check-prefixes=CHECK,NO_HALF \
+// RUN:   -DFNATTRS=noundef -DTARGET=dx
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   spirv-unknown-vulkan-compute %s -fnative-half-type \
+// RUN:   -emit-llvm -disable-llvm-passes -o - | FileCheck %s \
+// RUN:   --check-prefixes=CHECK,NATIVE_HALF \
+// RUN:   -DFNATTRS="spir_func noundef" -DTARGET=spv
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   spirv-unknown-vulkan-compute %s -emit-llvm -disable-llvm-passes \
+// RUN:   -o - | FileCheck %s --check-prefixes=CHECK,NO_HALF \
+// RUN:   -DFNATTRS="spir_func noundef" -DTARGET=spv
+
+// NATIVE_HALF: define [[FNATTRS]] half @
+// NATIVE_HALF: call half @llvm.[[TARGET]].step.f16(half
+// NO_HALF: call float @llvm.[[TARGET]].step.f32(float
+// NATIVE_HALF: ret half
+// NO_HALF: ret float
+half test_step_half(half p0, half p1)
+{
+    return step(p0, p1);
+}
+// NATIVE_HALF: define [[FNATTRS]] <2 x half> @
+// NATIVE_HALF: call <2 x half> @llvm.[[TARGET]].step.v2f16(<2 x half>
+// NO_HALF: call <2 x float> @llvm.[[TARGET]].step.v2f32(<2 x float>
+// NATIVE_HALF: ret <2 x half> %hlsl.step
+// NO_HALF: ret <2 x float> %hlsl.step
+half2 test_step_half2(half2 p0, half2 p1)
+{
+    return step(p0, p1);
+}
+// NATIVE_HALF: define [[FNATTRS]] <3 x half> @
+// NATIVE_HALF: call <3 x half> @llvm.[[TARGET]].step.v3f16(<3 x half>
+// NO_HALF: call <3 x float> @llvm.[[TARGET]].step.v3f32(<3 x float>
+// NATIVE_HALF: ret <3 x half> %hlsl.step
+// NO_HALF: ret <3 x float> %hlsl.step
+half3 test_step_half3(half3 p0, half3 p1)
+{
+    return step(p0, p1);
+}
+// NATIVE_HALF: define [[FNATTRS]] <4 x half> @
+// NATIVE_HALF: call <4 x half> @llvm.[[TARGET]].step.v4f16(<4 x half>
+// NO_HALF: call <4 x float> @llvm.[[TARGET]].step.v4f32(<4 x float>
+// NATIVE_HALF: ret <4 x half> %hlsl.step
+// NO_HALF: ret <4 x float> %hlsl.step
+half4 test_step_half4(half4 p0, half4 p1)
+{
+    return step(p0, p1);
+}
+
+// CHECK: define [[FNATTRS]] float @
+// CHECK: call float @llvm.[[TARGET]].step.f32(float
+// CHECK: ret float
+float test_step_float(float p0, float p1)
+{
+    return step(p0, p1);
+}
+// CHECK: define [[FNATTRS]] <2 x float> @
+// CHECK: %hlsl.step = call <2 x float> @llvm.[[TARGET]].step.v2f32(
+// CHECK: ret <2 x float> %hlsl.step
+float2 test_step_float2(float2 p0, float2 p1)
+{
+    return step(p0, p1);
+}
+// CHECK: define [[FNATTRS]] <3 x float> @
+// CHECK: %hlsl.step = call <3 x float> @llvm.[[TARGET]].step.v3f32(
+// CHECK: ret <3 x float> %hlsl.step
+float3 test_step_float3(float3 p0, float3 p1)
+{
+    return step(p0, p1);
+}
+// CHECK: define [[FNATTRS]] <4 x float> @
+// CHECK: %hlsl.step = call <4 x float> @llvm.[[TARGET]].step.v4f32(
+// CHECK: ret <4 x float> %hlsl.step
+float4 test_step_float4(float4 p0, float4 p1)
+{
+    return step(p0, p1);
+}

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

@@ -0,0 +1,31 @@
+// RUN: %clang_cc1 -finclude-default-header -triple dxil-pc-shadermodel6.6-library %s -fnative-half-type -disable-llvm-passes -verify -verify-ignore-unexpected
+
+void test_too_few_arg()
+{
+  return __builtin_hlsl_step();
+  // expected-error@-1 {{too few arguments to function call, expected 2, have 0}}
+}
+
+void test_too_many_arg(float2 p0)
+{
+  return __builtin_hlsl_step(p0, p0, p0);
+  // expected-error@-1 {{too many arguments to function call, expected 2, have 3}}
+}
+
+bool builtin_bool_to_float_type_promotion(bool p1)
+{
+  return __builtin_hlsl_step(p1, p1);
+  // expected-error@-1 {passing 'bool' to parameter of incompatible type 'float'}}
+}
+
+bool builtin_step_int_to_float_promotion(int p1)
+{
+  return __builtin_hlsl_step(p1, p1);
+  // expected-error@-1 {{passing 'int' to parameter of incompatible type 'float'}}
+}
+
+bool2 builtin_step_int2_to_float2_promotion(int2 p1)
+{
+  return __builtin_hlsl_step(p1, 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)}}
+}

llvm/include/llvm/IR/IntrinsicsDirectX.td

@@ -87,7 +87,7 @@ def int_dx_umad : DefaultAttrsIntrinsic<[llvm_anyint_ty], [LLVMMatchType<0>, LLV
 def int_dx_normalize : DefaultAttrsIntrinsic<[LLVMMatchType<0>], [llvm_anyfloat_ty]>;
 def int_dx_rcp  : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>]>;
 def int_dx_rsqrt  : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>]>;
-
 def int_dx_wave_is_first_lane : DefaultAttrsIntrinsic<[llvm_i1_ty], [], [IntrConvergent]>;
 def int_dx_sign : DefaultAttrsIntrinsic<[LLVMScalarOrSameVectorWidth<0, llvm_i32_ty>], [llvm_any_ty]>;
+def int_dx_step : DefaultAttrsIntrinsic<[LLVMMatchType<0>], [llvm_anyfloat_ty, LLVMMatchType<0>]>;
 }

llvm/include/llvm/IR/IntrinsicsSPIRV.td

@@ -67,6 +67,7 @@ let TargetPrefix = "spv" in {
   def int_spv_normalize : DefaultAttrsIntrinsic<[LLVMMatchType<0>], [llvm_anyfloat_ty]>;
   def int_spv_rsqrt : DefaultAttrsIntrinsic<[LLVMMatchType<0>], [llvm_anyfloat_ty]>;
   def int_spv_saturate : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>]>;
+  def int_spv_step : DefaultAttrsIntrinsic<[LLVMMatchType<0>], [LLVMMatchType<0>, llvm_anyfloat_ty]>;
   def int_spv_fdot :
     DefaultAttrsIntrinsic<[LLVMVectorElementType<0>],
     [llvm_anyfloat_ty, LLVMScalarOrSameVectorWidth<0, LLVMVectorElementType<0>>],

llvm/lib/Target/DirectX/DXILIntrinsicExpansion.cpp

@@ -50,6 +50,7 @@ static bool isIntrinsicExpansion(Function &F) {
   case Intrinsic::dx_sdot:
   case Intrinsic::dx_udot:
   case Intrinsic::dx_sign:
+  case Intrinsic::dx_step:
     return true;
   }
   return false;
@@ -322,6 +323,28 @@ static Value *expandPowIntrinsic(CallInst *Orig) {
   return Exp2Call;
 }
 
+static Value *expandStepIntrinsic(CallInst *Orig) {
+
+  Value *X = Orig->getOperand(0);
+  Value *Y = Orig->getOperand(1);
+  Type *Ty = X->getType();
+  IRBuilder<> Builder(Orig);
+
+  Constant *One = ConstantFP::get(Ty->getScalarType(), 1.0);
+  Constant *Zero = ConstantFP::get(Ty->getScalarType(), 0.0);
+  Value *Cond = Builder.CreateFCmpOLT(Y, X);
+
+  if (Ty != Ty->getScalarType()) {
+    auto *XVec = dyn_cast<FixedVectorType>(Ty);
+    One = ConstantVector::getSplat(
+        ElementCount::getFixed(XVec->getNumElements()), One);
+    Zero = ConstantVector::getSplat(
+        ElementCount::getFixed(XVec->getNumElements()), Zero);
+  }
+
+  return Builder.CreateSelect(Cond, Zero, One);
+}
+
 static Intrinsic::ID getMaxForClamp(Type *ElemTy,
                                     Intrinsic::ID ClampIntrinsic) {
   if (ClampIntrinsic == Intrinsic::dx_uclamp)
@@ -433,8 +456,9 @@ static bool expandIntrinsic(Function &F, CallInst *Orig) {
   case Intrinsic::dx_sign:
     Result = expandSignIntrinsic(Orig);
     break;
+  case Intrinsic::dx_step:
+    Result = expandStepIntrinsic(Orig);
   }
-
   if (Result) {
     Orig->replaceAllUsesWith(Result);
     Orig->eraseFromParent();

llvm/lib/Target/SPIRV/SPIRVInstructionSelector.cpp

@@ -263,6 +263,9 @@ class SPIRVInstructionSelector : public InstructionSelector {
   bool selectSpvThreadId(Register ResVReg, const SPIRVType *ResType,
                          MachineInstr &I) const;
 
+  bool selectStep(Register ResVReg, const SPIRVType *ResType,
+                  MachineInstr &I) const;
+
   bool selectUnmergeValues(MachineInstr &I) const;
 
   Register buildI32Constant(uint32_t Val, MachineInstr &I,
@@ -1710,6 +1713,25 @@ bool SPIRVInstructionSelector::selectSign(Register ResVReg,
   return Result;
 }
 
+bool SPIRVInstructionSelector::selectStep(Register ResVReg,
+                                          const SPIRVType *ResType,
+                                          MachineInstr &I) const {
+
+  assert(I.getNumOperands() == 4);
+  assert(I.getOperand(2).isReg());
+  assert(I.getOperand(3).isReg());
+  MachineBasicBlock &BB = *I.getParent();
+
+  return BuildMI(BB, I, I.getDebugLoc(), TII.get(SPIRV::OpExtInst))
+      .addDef(ResVReg)
+      .addUse(GR.getSPIRVTypeID(ResType))
+      .addImm(static_cast<uint32_t>(SPIRV::InstructionSet::GLSL_std_450))
+      .addImm(GL::Step)
+      .addUse(I.getOperand(2).getReg())
+      .addUse(I.getOperand(3).getReg())
+      .constrainAllUses(TII, TRI, RBI);
+}
+
 bool SPIRVInstructionSelector::selectBitreverse(Register ResVReg,
                                                 const SPIRVType *ResType,
                                                 MachineInstr &I) const {
@@ -2468,6 +2490,8 @@ bool SPIRVInstructionSelector::selectIntrinsic(Register ResVReg,
         .addUse(GR.getSPIRVTypeID(ResType))
         .addUse(GR.getOrCreateConstInt(3, I, IntTy, TII));
   }
+  case Intrinsic::spv_step:
+    return selectStep(ResVReg, ResType, I);
   default: {
     std::string DiagMsg;
     raw_string_ostream OS(DiagMsg);