[clang][bytecode] Implement __builtin_reduce_and (#118289)

Author: tbaederr

clang/lib/AST/ByteCode/InterpBuiltin.cpp

@@ -1700,7 +1700,7 @@ static bool interp__builtin_vector_reduce(InterpState &S, CodePtr OpPC,
   PrimType ElemT = *S.getContext().classify(ElemType);
   unsigned NumElems = Arg.getNumElems();
 
-  INT_TYPE_SWITCH(ElemT, {
+  INT_TYPE_SWITCH_NO_BOOL(ElemT, {
     T Result = Arg.atIndex(0).deref<T>();
     unsigned BitWidth = Result.bitWidth();
     for (unsigned I = 1; I != NumElems; ++I) {
@@ -1723,6 +1723,9 @@ static bool interp__builtin_vector_reduce(InterpState &S, CodePtr OpPC,
                                 Elem.toAPSInt(OverflowBits)));
           return false;
         }
+
+      } else if (ID == Builtin::BI__builtin_reduce_and) {
+        (void)T::bitAnd(Result, Elem, BitWidth, &Result);
       } else {
         llvm_unreachable("Unhandled vector reduce builtin");
       }
@@ -2206,6 +2209,7 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const Function *F,
 
   case Builtin::BI__builtin_reduce_add:
   case Builtin::BI__builtin_reduce_mul:
+  case Builtin::BI__builtin_reduce_and:
     if (!interp__builtin_vector_reduce(S, OpPC, Frame, F, Call))
       return false;
     break;

clang/test/AST/ByteCode/builtin-functions.cpp

@@ -1064,6 +1064,22 @@ namespace ReduceMul {
   static_assert(__builtin_reduce_mul((vector4ulong){~0ULL, 1, 1, 2}) == ~0ULL - 1);
 }
 
+namespace ReduceAnd {
+  static_assert(__builtin_reduce_and((vector4char){}) == 0);
+  static_assert(__builtin_reduce_and((vector4char){(char)0x11, (char)0x22, (char)0x44, (char)0x88}) == 0);
+  static_assert(__builtin_reduce_and((vector4short){(short)0x1111, (short)0x2222, (short)0x4444, (short)0x8888}) == 0);
+  static_assert(__builtin_reduce_and((vector4int){(int)0x11111111, (int)0x22222222, (int)0x44444444, (int)0x88888888}) == 0);
+#if __INT_WIDTH__ == 32
+  static_assert(__builtin_reduce_and((vector4long){(long long)0x1111111111111111L, (long long)0x2222222222222222L, (long long)0x4444444444444444L, (long long)0x8888888888888888L}) == 0L);
+  static_assert(__builtin_reduce_and((vector4char){(char)-1, (char)~0x22, (char)~0x44, (char)~0x88}) == 0x11);
+  static_assert(__builtin_reduce_and((vector4short){(short)~0x1111, (short)-1, (short)~0x4444, (short)~0x8888}) == 0x2222);
+  static_assert(__builtin_reduce_and((vector4int){(int)~0x11111111, (int)~0x22222222, (int)-1, (int)~0x88888888}) == 0x44444444);
+  static_assert(__builtin_reduce_and((vector4long){(long long)~0x1111111111111111L, (long long)~0x2222222222222222L, (long long)~0x4444444444444444L, (long long)-1}) == 0x8888888888888888L);
+  static_assert(__builtin_reduce_and((vector4uint){0x11111111U, 0x22222222U, 0x44444444U, 0x88888888U}) == 0U);
+  static_assert(__builtin_reduce_and((vector4ulong){0x1111111111111111UL, 0x2222222222222222UL, 0x4444444444444444UL, 0x8888888888888888UL}) == 0L);
+#endif
+}
+
 namespace BuiltinMemcpy {
   constexpr int simple() {
     int a = 12;