[Clang] [Sema] Fix bug in _Complex float+int arithmetic (#83063)

C23 6.3.1.8 ‘Usual arithmetic conversions’ p1 states (emphasis mine): 
> Otherwise, if the corresponding real type of either operand is
`float`, the other operand is converted, *without change of type
domain*, to a type whose corresponding real type is `float`.

‘type domain’ here refers to `_Complex` vs real (i.e. non-`_Complex`);
there is another clause that states the same for `double`.

Consider the following code:
```c++
_Complex float f;
int x;
f / x;
```

After talking this over with @AaronBallman, we came to the conclusion
that `x` should be converted to `float` and *not* `_Complex float` (that
is, we should perform a division of `_Complex float / float`, and *not*
`_Complex float / _Complex float`; the same also applies to `-+*`). This
was already being done correctly for cases where `x` was already a
`float`; it’s just mixed `_Complex float`+`int` operations that
currently suffer from this problem.

This pr removes the extra `FloatingRealToComplex` conversion that we
were erroneously inserting and adds some tests to make sure we’re
actually doing `_Complex float / float` and not `_Complex float /
_Complex float` (and analogously for `double` and `-+*`).

The only exception here is `float / _Complex float`, which calls a
library function (`__divsc3`) that takes 4 `float`s, so we end up having
to convert the `float` to a `_Complex float` after all (and analogously
for `double`); I don’t believe there is a way around this.

Lastly, we were also missing tests for `_Complex` arithmetic at compile
time, so this adds some tests for that as well.

Author: Sirraide

clang/docs/ReleaseNotes.rst

@@ -278,6 +278,13 @@ Bug Fixes in This Version
 - Clang now correctly generates overloads for bit-precise integer types for
   builtin operators in C++. Fixes #GH82998.
 
+- When performing mixed arithmetic between ``_Complex`` floating-point types and integers,
+  Clang now correctly promotes the integer to its corresponding real floating-point
+  type only rather than to the complex type (e.g. ``_Complex float / int`` is now evaluated
+  as ``_Complex float / float`` rather than ``_Complex float / _Complex float``), as mandated
+  by the C standard. This significantly improves codegen of `*` and `/` especially.
+  Fixes (`#31205 <https://github.com/llvm/llvm-project/issues/31205>`_).
+
 Bug Fixes to Compiler Builtins
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 

clang/lib/Sema/SemaExpr.cpp

@@ -1099,12 +1099,13 @@ ExprResult Sema::DefaultVariadicArgumentPromotion(Expr *E, VariadicCallType CT,
   return E;
 }
 
-/// Converts an integer to complex float type.  Helper function of
+/// Convert complex integers to complex floats and real integers to
+/// real floats as required for complex arithmetic. Helper function of
 /// UsualArithmeticConversions()
 ///
 /// \return false if the integer expression is an integer type and is
-/// successfully converted to the complex type.
-static bool handleIntegerToComplexFloatConversion(Sema &S, ExprResult &IntExpr,
+/// successfully converted to the (complex) float type.
+static bool handleComplexIntegerToFloatConversion(Sema &S, ExprResult &IntExpr,
                                                   ExprResult &ComplexExpr,
                                                   QualType IntTy,
                                                   QualType ComplexTy,
@@ -1114,8 +1115,6 @@ static bool handleIntegerToComplexFloatConversion(Sema &S, ExprResult &IntExpr,
   if (IntTy->isIntegerType()) {
     QualType fpTy = ComplexTy->castAs<ComplexType>()->getElementType();
     IntExpr = S.ImpCastExprToType(IntExpr.get(), fpTy, CK_IntegralToFloating);
-    IntExpr = S.ImpCastExprToType(IntExpr.get(), ComplexTy,
-                                  CK_FloatingRealToComplex);
   } else {
     assert(IntTy->isComplexIntegerType());
     IntExpr = S.ImpCastExprToType(IntExpr.get(), ComplexTy,
@@ -1160,11 +1159,11 @@ static QualType handleComplexFloatConversion(Sema &S, ExprResult &Shorter,
 static QualType handleComplexConversion(Sema &S, ExprResult &LHS,
                                         ExprResult &RHS, QualType LHSType,
                                         QualType RHSType, bool IsCompAssign) {
-  // if we have an integer operand, the result is the complex type.
-  if (!handleIntegerToComplexFloatConversion(S, RHS, LHS, RHSType, LHSType,
+  // Handle (complex) integer types.
+  if (!handleComplexIntegerToFloatConversion(S, RHS, LHS, RHSType, LHSType,
                                              /*SkipCast=*/false))
     return LHSType;
-  if (!handleIntegerToComplexFloatConversion(S, LHS, RHS, LHSType, RHSType,
+  if (!handleComplexIntegerToFloatConversion(S, LHS, RHS, LHSType, RHSType,
                                              /*SkipCast=*/IsCompAssign))
     return RHSType;
 

clang/test/CodeGen/complex-math-mixed.c

@@ -0,0 +1,146 @@
+// RUN: %clang_cc1 %s -O0 -emit-llvm -triple x86_64-unknown-unknown -o - | FileCheck %s --check-prefix=X86
+// RUN: %clang_cc1 %s -O0 -triple x86_64-unknown-unknown -fsyntax-only -ast-dump | FileCheck %s --check-prefix=AST
+
+// Check that for 'F _Complex + int' (F = real floating-point type), we emit an
+// implicit cast from 'int' to 'F', but NOT to 'F _Complex' (i.e. that we do
+// 'F _Complex + F', NOT 'F _Complex + F _Complex'), and likewise for -/*.
+
+// AST-NOT: FloatingRealToComplex
+
+float _Complex add_float_ci(float _Complex a, int b) {
+  // X86-LABEL: @add_float_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fadd float {{.*}}, [[I]]
+  // X86-NOT: fadd
+  return a + b;
+}
+
+float _Complex add_float_ic(int a, float _Complex b) {
+  // X86-LABEL: @add_float_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fadd float [[I]]
+  // X86-NOT: fadd
+  return a + b;
+}
+
+float _Complex sub_float_ci(float _Complex a, int b) {
+  // X86-LABEL: @sub_float_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fsub float {{.*}}, [[I]]
+  // X86-NOT: fsub
+  return a - b;
+}
+
+float _Complex sub_float_ic(int a, float _Complex b) {
+  // X86-LABEL: @sub_float_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fsub float [[I]]
+  // X86: fneg
+  // X86-NOT: fsub
+  return a - b;
+}
+
+float _Complex mul_float_ci(float _Complex a, int b) {
+  // X86-LABEL: @mul_float_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fmul float {{.*}}, [[I]]
+  // X86: fmul float {{.*}}, [[I]]
+  // X86-NOT: fmul
+  return a * b;
+}
+
+float _Complex mul_float_ic(int a, float _Complex b) {
+  // X86-LABEL: @mul_float_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fmul float [[I]]
+  // X86: fmul float [[I]]
+  // X86-NOT: fmul
+  return a * b;
+}
+
+float _Complex div_float_ci(float _Complex a, int b) {
+  // X86-LABEL: @div_float_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: fdiv float {{.*}}, [[I]]
+  // X86: fdiv float {{.*}}, [[I]]
+  // X86-NOT: @__divsc3
+  return a / b;
+}
+
+// There is no good way of doing this w/o converting the 'int' to a complex
+// number, so we expect complex division here.
+float _Complex div_float_ic(int a, float _Complex b) {
+  // X86-LABEL: @div_float_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to float
+  // X86: call {{.*}} @__divsc3(float {{.*}} [[I]], float noundef 0.{{0+}}e+00, float {{.*}}, float {{.*}})
+  return a / b;
+}
+
+double _Complex add_double_ci(double _Complex a, int b) {
+  // X86-LABEL: @add_double_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fadd double {{.*}}, [[I]]
+  // X86-NOT: fadd
+  return a + b;
+}
+
+double _Complex add_double_ic(int a, double _Complex b) {
+  // X86-LABEL: @add_double_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fadd double [[I]]
+  // X86-NOT: fadd
+  return a + b;
+}
+
+double _Complex sub_double_ci(double _Complex a, int b) {
+  // X86-LABEL: @sub_double_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fsub double {{.*}}, [[I]]
+  // X86-NOT: fsub
+  return a - b;
+}
+
+double _Complex sub_double_ic(int a, double _Complex b) {
+  // X86-LABEL: @sub_double_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fsub double [[I]]
+  // X86: fneg
+  // X86-NOT: fsub
+  return a - b;
+}
+
+double _Complex mul_double_ci(double _Complex a, int b) {
+  // X86-LABEL: @mul_double_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fmul double {{.*}}, [[I]]
+  // X86: fmul double {{.*}}, [[I]]
+  // X86-NOT: fmul
+  return a * b;
+}
+
+double _Complex mul_double_ic(int a, double _Complex b) {
+  // X86-LABEL: @mul_double_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fmul double [[I]]
+  // X86: fmul double [[I]]
+  // X86-NOT: fmul
+  return a * b;
+}
+
+double _Complex div_double_ci(double _Complex a, int b) {
+  // X86-LABEL: @div_double_ci
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: fdiv double {{.*}}, [[I]]
+  // X86: fdiv double {{.*}}, [[I]]
+  // X86-NOT: @__divdc3
+  return a / b;
+}
+
+// There is no good way of doing this w/o converting the 'int' to a complex
+// number, so we expect complex division here.
+double _Complex div_double_ic(int a, double _Complex b) {
+  // X86-LABEL: @div_double_ic
+  // X86: [[I:%.*]] = sitofp i32 {{%.*}} to double
+  // X86: call {{.*}} @__divdc3(double {{.*}} [[I]], double noundef 0.{{0+}}e+00, double {{.*}}, double {{.*}})
+  return a / b;
+}

clang/test/CodeGen/volatile.cpp

@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 -O2 -triple=x86_64-unknown-linux-gnu -emit-llvm %s -o -  | FileCheck %s -check-prefix CHECK
+// RUN: %clang_cc1 -O2 -triple=x86_64-unknown-linux-gnu -emit-llvm %s -o -  | FileCheck %s
 struct agg 
 {
 int a ;
@@ -10,34 +10,32 @@ _Complex float cf;
 int volatile vol =10;
 void f0() {
     const_cast<volatile _Complex float &>(cf) = const_cast<volatile _Complex float&>(cf) + 1;
-//  CHECK: %cf.real = load volatile float, ptr @cf
-//  CHECK: %cf.imag = load volatile float, ptr getelementptr
-//  CHECK: %add.r = fadd float %cf.real, 1.000000e+00
-//  CHECK: %add.i = fadd float %cf.imag, 0.000000e+00
-//  CHECK: store volatile float %add.r
-//  CHECK: store volatile float %add.i, ptr getelementptr
+//  CHECK: [[Re1:%.*]] = load volatile float, ptr @cf
+//  CHECK: [[Im1:%.*]] = load volatile float, ptr getelementptr
+//  CHECK: [[Add1:%.*]] = fadd float [[Re1]], 1.000000e+00
+//  CHECK: store volatile float [[Add1]], ptr @cf
+//  CHECK: store volatile float [[Im1]], ptr getelementptr
       static_cast<volatile _Complex float &>(cf) = static_cast<volatile _Complex float&>(cf) + 1;
-//  CHECK: %cf.real1 = load volatile float, ptr @cf
-//  CHECK: %cf.imag2 = load volatile float, ptr getelementptr
-//  CHECK: %add.r3 = fadd float %cf.real1, 1.000000e+00
-//  CHECK: %add.i4 = fadd float %cf.imag2, 0.000000e+00
-//  CHECK: store volatile float %add.r3, ptr @cf
-//  CHECK: store volatile float %add.i4, ptr getelementptr
+//  CHECK: [[Re2:%.*]] = load volatile float, ptr @cf
+//  CHECK: [[Im2:%.*]] = load volatile float, ptr getelementptr
+//  CHECK: [[Add2:%.*]] = fadd float [[Re2]], 1.000000e+00
+//  CHECK: store volatile float [[Add2]], ptr @cf
+//  CHECK: store volatile float [[Im2]], ptr getelementptr
     const_cast<volatile  int  &>(a.a) = const_cast<volatile int &>(t.a) ;
-//  CHECK: %0 = load volatile i32, ptr @t
-//  CHECK: store volatile i32 %0, ptr @a
+//  CHECK: [[I1:%.*]] = load volatile i32, ptr @t
+//  CHECK: store volatile i32 [[I1]], ptr @a
     static_cast<volatile  int  &>(a.b) = static_cast<volatile int  &>(t.a) ;
-//  CHECK: %1 = load volatile i32, ptr @t
-//  CHECK: store volatile i32 %1, ptr getelementptr
+//  CHECK: [[I2:%.*]] = load volatile i32, ptr @t
+//  CHECK: store volatile i32 [[I2]], ptr getelementptr
     const_cast<volatile int&>(vt) = const_cast<volatile int&>(vt) + 1;
-//  CHECK: %2 = load volatile i32, ptr @vt
-//  CHECK: %add = add nsw i32 %2, 1
-//  CHECK: store volatile i32 %add, ptr @vt
+//  CHECK: [[I3:%.*]] = load volatile i32, ptr @vt
+//  CHECK: [[Add3:%.*]] = add nsw i32 [[I3]], 1
+//  CHECK: store volatile i32 [[Add3]], ptr @vt
      static_cast<volatile int&>(vt) = static_cast<volatile int&>(vt) + 1;
-//  CHECK: %3 = load volatile i32, ptr @vt
-//  CHECK: %add5 = add nsw i32 %3, 1
-//  CHECK: store volatile i32 %add5, ptr @vt
+//  CHECK: [[I4:%.*]] = load volatile i32, ptr @vt
+//  CHECK: [[Add4:%.*]] = add nsw i32 [[I4]], 1
+//  CHECK: store volatile i32 [[Add4]], ptr @vt
     vt = const_cast<int&>(vol);
-//  %4 = load i32, ptr @vol
-//  store i32 %4, ptr @vt
+//  [[I5:%.*]] = load i32, ptr @vol
+//  store i32 [[I5]], ptr @vt
 }

clang/test/Sema/complex-arithmetic.c

@@ -0,0 +1,115 @@
+// RUN: %clang_cc1 -verify %s
+// expected-no-diagnostics
+
+// This tests evaluation of _Complex arithmetic at compile time.
+
+#define APPROX_EQ(a, b) (                             \
+  __builtin_fabs(__real (a) - __real (b)) < 0.0001 && \
+  __builtin_fabs(__imag (a) - __imag (b)) < 0.0001    \
+)
+
+#define EVAL(a, b) _Static_assert(a == b, "")
+#define EVALF(a, b) _Static_assert(APPROX_EQ(a, b), "")
+
+// _Complex float + _Complex float
+void a() {
+  EVALF((2.f + 3i) + (4.f + 5i), 6.f + 8i);
+  EVALF((2.f + 3i) - (4.f + 5i), -2.f - 2i);
+  EVALF((2.f + 3i) * (4.f + 5i), -7.f + 22i);
+  EVALF((2.f + 3i) / (4.f + 5i), 0.5609f + 0.0487i);
+
+  EVALF((2. + 3i) + (4. + 5i), 6. + 8i);
+  EVALF((2. + 3i) - (4. + 5i), -2. - 2i);
+  EVALF((2. + 3i) * (4. + 5i), -7. + 22i);
+  EVALF((2. + 3i) / (4. + 5i), .5609 + .0487i);
+}
+
+// _Complex int + _Complex int
+void b() {
+  EVAL((2 + 3i) + (4 + 5i), 6 + 8i);
+  EVAL((2 + 3i) - (4 + 5i), -2 - 2i);
+  EVAL((2 + 3i) * (4 + 5i), -7 + 22i);
+  EVAL((8 + 30i) / (4 + 5i), 4 + 1i);
+}
+
+// _Complex float + float
+void c() {
+  EVALF((2.f + 4i) + 3.f, 5.f + 4i);
+  EVALF((2.f + 4i) - 3.f, -1.f + 4i);
+  EVALF((2.f + 4i) * 3.f, 6.f + 12i);
+  EVALF((2.f + 4i) / 2.f, 1.f + 2i);
+
+  EVALF(3.f + (2.f + 4i), 5.f + 4i);
+  EVALF(3.f - (2.f + 4i), 1.f - 4i);
+  EVALF(3.f * (2.f + 4i), 6.f + 12i);
+  EVALF(3.f / (2.f + 4i), .3f - 0.6i);
+
+  EVALF((2. + 4i) + 3., 5. + 4i);
+  EVALF((2. + 4i) - 3., -1. + 4i);
+  EVALF((2. + 4i) * 3., 6. + 12i);
+  EVALF((2. + 4i) / 2., 1. + 2i);
+
+  EVALF(3. + (2. + 4i), 5. + 4i);
+  EVALF(3. - (2. + 4i), 1. - 4i);
+  EVALF(3. * (2. + 4i), 6. + 12i);
+  EVALF(3. / (2. + 4i), .3 - 0.6i);
+}
+
+// _Complex int + int
+void d() {
+  EVAL((2 + 4i) + 3, 5 + 4i);
+  EVAL((2 + 4i) - 3, -1 + 4i);
+  EVAL((2 + 4i) * 3, 6 + 12i);
+  EVAL((2 + 4i) / 2, 1 + 2i);
+
+  EVAL(3 + (2 + 4i), 5 + 4i);
+  EVAL(3 - (2 + 4i), 1 - 4i);
+  EVAL(3 * (2 + 4i), 6 + 12i);
+  EVAL(20 / (2 + 4i), 2 - 4i);
+}
+
+// _Complex float + int
+void e() {
+  EVALF((2.f + 4i) + 3, 5.f + 4i);
+  EVALF((2.f + 4i) - 3, -1.f + 4i);
+  EVALF((2.f + 4i) * 3, 6.f + 12i);
+  EVALF((2.f + 4i) / 2, 1.f + 2i);
+
+  EVALF(3 + (2.f + 4i), 5.f + 4i);
+  EVALF(3 - (2.f + 4i), 1.f - 4i);
+  EVALF(3 * (2.f + 4i), 6.f + 12i);
+  EVALF(3 / (2.f + 4i), .3f - 0.6i);
+
+  EVALF((2. + 4i) + 3, 5. + 4i);
+  EVALF((2. + 4i) - 3, -1. + 4i);
+  EVALF((2. + 4i) * 3, 6. + 12i);
+  EVALF((2. + 4i) / 2, 1. + 2i);
+
+  EVALF(3 + (2. + 4i), 5. + 4i);
+  EVALF(3 - (2. + 4i), 1. - 4i);
+  EVALF(3 * (2. + 4i), 6. + 12i);
+  EVALF(3 / (2. + 4i), .3 - 0.6i);
+}
+
+// _Complex int + float
+void f() {
+  EVALF((2 + 4i) + 3.f, 5.f + 4i);
+  EVALF((2 + 4i) - 3.f, -1.f + 4i);
+  EVALF((2 + 4i) * 3.f, 6.f + 12i);
+  EVALF((2 + 4i) / 2.f, 1.f + 2i);
+
+  EVALF(3.f + (2 + 4i), 5.f + 4i);
+  EVALF(3.f - (2 + 4i), 1.f - 4i);
+  EVALF(3.f * (2 + 4i), 6.f + 12i);
+  EVALF(3.f / (2 + 4i), .3f - 0.6i);
+
+  EVALF((2 + 4i) + 3., 5. + 4i);
+  EVALF((2 + 4i) - 3., -1. + 4i);
+  EVALF((2 + 4i) * 3., 6. + 12i);
+  EVALF((2 + 4i) / 2., 1. + 2i);
+
+  EVALF(3. + (2 + 4i), 5. + 4i);
+  EVALF(3. - (2 + 4i), 1. - 4i);
+  EVALF(3. * (2 + 4i), 6. + 12i);
+  EVALF(3. / (2 + 4i), .3 - 0.6i);
+}