[HLSL] Implementation of the elementwise fmod builtin

This change add the elementwise fmod builtin to support HLSL function 'fmod' in clang for #99118
Builtins.td           - add the fmod builtin
CGBuiltin.cpp         - lower the builtin to llvm FRem instruction
hlsl_intrinsics.h     - add the fmod api
SemaChecking.cpp      - add type checks for builtin
SemaHLSL.cpp          - add HLSL type checks for builtin

clang/docs/LanguageExtensions.rst  - add the builtin in *Elementwise Builtins*
clang/docs/ReleaseNotes.rst        - announce the builtin

Author: lizhengxing

clang/docs/LanguageExtensions.rst

@@ -700,6 +700,8 @@ Unless specified otherwise operation(±0) = ±0 and operation(±infinity) = ±in
  T __builtin_elementwise_canonicalize(T x)   return the platform specific canonical encoding                  floating point types
                                              of a floating-point number
  T __builtin_elementwise_copysign(T x, T y)  return the magnitude of x with the sign of y.                    floating point types
+ T __builtin_elementwise_fmod(T x, T y)      return The floating-point remainder of (x/y) whose sign          floating point types
+                                             matches the sign of x.
  T __builtin_elementwise_max(T x, T y)       return x or y, whichever is larger                               integer and floating point types
  T __builtin_elementwise_min(T x, T y)       return x or y, whichever is smaller                              integer and floating point types
  T __builtin_elementwise_add_sat(T x, T y)   return the sum of x and y, clamped to the range of               integer types

clang/docs/ReleaseNotes.rst

@@ -123,6 +123,8 @@ C++ Language Changes
 
 - Add ``__builtin_elementwise_popcount`` builtin for integer types only.
 
+- Add ``__builtin_elementwise_fmod`` builtin for floating point types only.
+
 - The builtin type alias ``__builtin_common_type`` has been added to improve the
   performance of ``std::common_type``.
 

clang/include/clang/Basic/Builtins.td

@@ -1328,6 +1328,12 @@ def ElementwisePopcount : Builtin {
   let Prototype = "void(...)";
 }
 
+def ElementwiseFmod : Builtin {
+  let Spellings = ["__builtin_elementwise_fmod"];
+  let Attributes = [NoThrow, Const, CustomTypeChecking];
+  let Prototype = "void(...)";
+}
+
 def ElementwisePow : Builtin {
   let Spellings = ["__builtin_elementwise_pow"];
   let Attributes = [NoThrow, Const, CustomTypeChecking];

clang/lib/CodeGen/CGBuiltin.cpp

@@ -2878,7 +2878,8 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
     case Builtin::BI__builtin_fmodf:
     case Builtin::BI__builtin_fmodf16:
     case Builtin::BI__builtin_fmodl:
-    case Builtin::BI__builtin_fmodf128: {
+    case Builtin::BI__builtin_fmodf128:
+    case Builtin::BI__builtin_elementwise_fmod: {
       CodeGenFunction::CGFPOptionsRAII FPOptsRAII(*this, E);
       Value *Arg1 = EmitScalarExpr(E->getArg(0));
       Value *Arg2 = EmitScalarExpr(E->getArg(1));

clang/lib/Headers/hlsl/hlsl_intrinsics.h

@@ -913,6 +913,39 @@ float3 floor(float3);
 _HLSL_BUILTIN_ALIAS(__builtin_elementwise_floor)
 float4 floor(float4);
 
+//===----------------------------------------------------------------------===//
+// fmod builtins
+//===----------------------------------------------------------------------===//
+
+/// \fn T fmod(T x, T y)
+/// \brief Returns the linear interpolation of x to y.
+/// \param x [in] The dividend.
+/// \param y [in] The divisor.
+///
+/// Return the floating-point remainder of the x parameter divided by the y parameter.
+
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+half fmod(half, half);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+half2 fmod(half2, half2);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+half3 fmod(half3, half3);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+half4 fmod(half4, half4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+float fmod(float, float);
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+float2 fmod(float2, float2);
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+float3 fmod(float3, float3);
+_HLSL_BUILTIN_ALIAS(__builtin_elementwise_fmod)
+float4 fmod(float4, float4);
+
 //===----------------------------------------------------------------------===//
 // frac builtins
 //===----------------------------------------------------------------------===//

clang/lib/Sema/SemaChecking.cpp

@@ -2755,6 +2755,7 @@ Sema::CheckBuiltinFunctionCall(FunctionDecl *FDecl, unsigned BuiltinID,
 
   // These builtins restrict the element type to floating point
   // types only, and take in two arguments.
+  case Builtin::BI__builtin_elementwise_fmod:
   case Builtin::BI__builtin_elementwise_pow: {
     if (BuiltinElementwiseMath(TheCall))
       return ExprError();

clang/lib/Sema/SemaHLSL.cpp

@@ -1771,6 +1771,7 @@ bool SemaHLSL::CheckBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
   case Builtin::BI__builtin_elementwise_exp:
   case Builtin::BI__builtin_elementwise_exp2:
   case Builtin::BI__builtin_elementwise_floor:
+  case Builtin::BI__builtin_elementwise_fmod:
   case Builtin::BI__builtin_elementwise_log:
   case Builtin::BI__builtin_elementwise_log2:
   case Builtin::BI__builtin_elementwise_log10:

clang/test/CodeGen/builtins-elementwise-math.c

@@ -607,6 +607,26 @@ void test_builtin_elementwise_popcount(si8 vi1, si8 vi2,
   si = __builtin_elementwise_popcount(si);
 }
 
+void test_builtin_elementwise_fmod(float f1, float f2, double d1, double d2,
+                                      float4 vf1, float4 vf2) {
+
+  // CHECK-LABEL: define void @test_builtin_elementwise_fmod(
+  // CHECK:      [[F1:%.+]] = load float, ptr %f1.addr, align 4
+  // CHECK:      [[F2:%.+]] = load float, ptr %f2.addr, align 4
+  // CHECK-NEXT:  frem float [[F1]], [[F2]]
+  f2 = __builtin_elementwise_fmod(f1, f2);
+
+  // CHECK:      [[D1:%.+]] = load double, ptr %d1.addr, align 8
+  // CHECK:      [[D2:%.+]] = load double, ptr %d2.addr, align 8
+  // CHECK-NEXT: frem double [[D1]], [[D2]]
+  d2 = __builtin_elementwise_fmod(d1, d2);
+
+  // CHECK:      [[VF1:%.+]] = load <4 x float>, ptr %vf1.addr, align 16
+  // CHECK:      [[VF2:%.+]] = load <4 x float>, ptr %vf2.addr, align 16
+  // CHECK-NEXT: frem <4 x float> [[VF1]], [[VF2]]
+  vf2 = __builtin_elementwise_fmod(vf1, vf2);
+}
+
 void test_builtin_elementwise_pow(float f1, float f2, double d1, double d2,
                                       float4 vf1, float4 vf2) {
 

clang/test/CodeGen/strictfp-elementwise-bulitins.cpp

@@ -306,3 +306,14 @@ float4 strict_elementwise_fma(float4 a, float4 b, float4 c) {
 float4 strict_elementwise_pow(float4 a, float4 b) {
   return __builtin_elementwise_pow(a, b);
 }
+
+// CHECK-LABEL: define dso_local noundef <4 x float> @_Z23strict_elementwise_fmodDv4_fS_
+// CHECK-SAME: (<4 x float> noundef [[A:%.*]], <4 x float> noundef [[B:%.*]]) local_unnamed_addr #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <4 x float> @llvm.experimental.constrained.frem.v4f32(<4 x float> [[A]], <4 x float> [[B]],
+// CHECK-SAME:    metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR4]]
+// CHECK-NEXT:    ret <4 x float> [[TMP0]]
+//
+float4 strict_elementwise_fmod(float4 a, float4 b) {
+  return __builtin_elementwise_fmod(a, b);
+}

clang/test/CodeGenHLSL/builtins/fmod.hlsl

@@ -0,0 +1,77 @@
+// DirectX target:
+//
+// ---------- Native Half support test -----------
+//
+// 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:   -DFNATTRS=noundef -DTYPE=half
+
+//
+// ---------- No Native Half support test -----------
+//
+// 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 \
+// RUN:   -DFNATTRS=noundef -DTYPE=float
+
+
+// Spirv target:
+//
+// ---------- Native Half support test -----------
+//
+// 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:   -DFNATTRS="spir_func noundef" -DTYPE=half
+
+//
+// ---------- No Native Half support test -----------
+//
+// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple \
+// RUN:   spirv-unknown-vulkan-compute %s -emit-llvm -disable-llvm-passes \
+// RUN:   -o - | FileCheck %s \
+// RUN:   -DFNATTRS="spir_func noundef" -DTYPE=float
+
+
+
+// CHECK: define [[FNATTRS]] [[TYPE]] @
+// CHECK: %fmod = frem [[TYPE]]
+// CHECK: ret [[TYPE]] %fmod
+half test_fmod_half(half p0, half p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] <2 x [[TYPE]]> @
+// CHECK: %fmod = frem <2 x [[TYPE]]>
+// CHECK: ret <2 x [[TYPE]]> %fmod
+half2 test_fmod_half2(half2 p0, half2 p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] <3 x [[TYPE]]> @
+// CHECK: %fmod = frem <3 x [[TYPE]]>
+// CHECK: ret <3 x [[TYPE]]> %fmod
+half3 test_fmod_half3(half3 p0, half3 p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] <4 x [[TYPE]]> @
+// CHECK: %fmod = frem <4 x [[TYPE]]>
+// CHECK: ret <4 x [[TYPE]]> %fmod
+half4 test_fmod_half4(half4 p0, half4 p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] float @
+// CHECK: %fmod = frem float
+// CHECK: ret float %fmod
+float test_fmod_float(float p0, float p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] <2 x float> @
+// CHECK: %fmod = frem <2 x float>
+// CHECK: ret <2 x float> %fmod
+float2 test_fmod_float2(float2 p0, float2 p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] <3 x float> @
+// CHECK: %fmod = frem <3 x float>
+// CHECK: ret <3 x float> %fmod
+float3 test_fmod_float3(float3 p0, float3 p1) { return fmod(p0, p1); }
+
+// CHECK: define [[FNATTRS]] <4 x float> @
+// CHECK: %fmod = frem <4 x float>
+// CHECK: ret <4 x float> %fmod
+float4 test_fmod_float4(float4 p0, float4 p1) { return fmod(p0, p1); }
+

clang/test/Sema/builtins-elementwise-math.c

@@ -538,6 +538,32 @@ void test_builtin_elementwise_popcount(int i, float f, double d, float4 v, int3
   // expected-error@-1 {{assigning to 'int3' (vector of 3 'int' values) from incompatible type 'unsigned3' (vector of 3 'unsigned int' values)}}
 }
 
+void test_builtin_elementwise_fmod(int i, short s, double d, float4 v, int3 iv, unsigned3 uv, int *p) {
+  i = __builtin_elementwise_fmod(p, d);
+  // expected-error@-1 {{arguments are of different types ('int *' vs 'double')}}
+
+  struct Foo foo = __builtin_elementwise_fmod(i, i);
+  // expected-error@-1 {{1st argument must be a floating point type (was 'int')}}
+
+  i = __builtin_elementwise_fmod(i);
+  // expected-error@-1 {{too few arguments to function call, expected 2, have 1}}
+
+  i = __builtin_elementwise_fmod();
+  // expected-error@-1 {{too few arguments to function call, expected 2, have 0}}
+
+  i = __builtin_elementwise_fmod(i, i, i);
+  // expected-error@-1 {{too many arguments to function call, expected 2, have 3}}
+
+  i = __builtin_elementwise_fmod(v, iv);
+  // expected-error@-1 {{arguments are of different types ('float4' (vector of 4 'float' values) vs 'int3' (vector of 3 'int' values))}}
+
+  i = __builtin_elementwise_fmod(uv, iv);
+  // expected-error@-1 {{arguments are of different types ('unsigned3' (vector of 3 'unsigned int' values) vs 'int3' (vector of 3 'int' values))}}
+
+  i = __builtin_elementwise_fmod(d, v);
+  // expected-error@-1 {{arguments are of different types ('double' vs 'float4' (vector of 4 'float' values))}}
+}
+
 void test_builtin_elementwise_pow(int i, short s, double d, float4 v, int3 iv, unsigned3 uv, int *p) {
   i = __builtin_elementwise_pow(p, d);
   // expected-error@-1 {{arguments are of different types ('int *' vs 'double')}}
@@ -562,7 +588,6 @@ void test_builtin_elementwise_pow(int i, short s, double d, float4 v, int3 iv, u
 
 }
 
-
 void test_builtin_elementwise_roundeven(int i, float f, double d, float4 v, int3 iv, unsigned u, unsigned4 uv) {
 
   struct Foo s = __builtin_elementwise_roundeven(f);

clang/test/SemaCXX/builtins-elementwise-math.cpp

@@ -255,6 +255,14 @@ void test_builtin_elementwise_fma() {
   static_assert(!is_const<decltype(__builtin_elementwise_fma(c, c, c))>::value);
 }
 
+void test_builtin_elementwise_fmod() {
+  const double a = 2;
+  double b = 1;
+  static_assert(!is_const<decltype(__builtin_elementwise_fmod(a, b))>::value);
+  static_assert(!is_const<decltype(__builtin_elementwise_fmod(b, a))>::value);
+  static_assert(!is_const<decltype(__builtin_elementwise_fmod(a, a))>::value);
+}
+
 void test_builtin_elementwise_pow() {
   const double a = 2;
   double b = 1;

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

@@ -0,0 +1,22 @@
+
+// RUN: %clang_cc1 -finclude-default-header -triple dxil-pc-shadermodel6.6-library %s -fnative-half-type -emit-llvm-only -disable-llvm-passes -verify -verify-ignore-unexpected
+
+float builtin_bool_to_float_type_promotion(bool p1, bool p2) {
+  return __builtin_elementwise_fmod(p1, p2);
+  // expected-error@-1 {{1st argument must be a vector, integer or floating point type (was 'bool')}}
+}
+
+float2 builtin_fmod_int2_to_float2_promotion(int2 p1, int2 p2) {
+  return __builtin_elementwise_fmod(p1, p2);
+  // expected-error@-1 {{1st argument must be a floating point type (was 'int2' (aka 'vector<int, 2>'))}}
+}
+
+half builtin_fmod_double_type (double p0, double p1) {
+  return __builtin_elementwise_fmod(p0, p1);
+  // expected-error@-1 {{passing 'double' to parameter of incompatible type 'float'}}
+}
+
+half builtin_fmod_double2_type (double2 p0, double2 p1) {
+  return __builtin_elementwise_fmod(p0, p1);
+  // expected-error@-1 {{passing 'double2' (aka 'vector<double, 2>') to parameter of incompatible type '__attribute__((__vector_size__(2 * sizeof(float)))) float' (vector of 2 'float' values)}}
+}