Update @llvm.powi to handle different int sizes for the exponent

This can be seen as a follow up to commit 0ee439b,
that changed the second argument of __powidf2, __powisf2 and
__powitf2 in compiler-rt from si_int to int. That was to align with
how those runtimes are defined in libgcc.
One thing that seem to have been missing in that patch was to make
sure that the rest of LLVM also handle that the argument now depends
on the size of int (not using the si_int machine mode for 32-bit).
When using __builtin_powi for a target with 16-bit int clang crashed.
And when emitting libcalls to those rtlib functions, typically when
lowering @llvm.powi), the backend would always prepare the exponent
argument as an i32 which caused miscompiles when the rtlib was
compiled with 16-bit int.

The solution used here is to use an overloaded type for the second
argument in @llvm.powi. This way clang can use the "correct" type
when lowering __builtin_powi, and then later when emitting the libcall
it is assumed that the type used in @llvm.powi matches the rtlib
function.

One thing that needed some extra attention was that when vectorizing
calls several passes did not support that several arguments could
be overloaded in the intrinsics. This patch allows overload of a
scalar operand by adding hasVectorInstrinsicOverloadedScalarOpd, with
an entry for powi.

Differential Revision: https://reviews.llvm.org/D99439

Author: bjope

clang/lib/CodeGen/CGBuiltin.cpp

@@ -2946,10 +2946,21 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
 
   case Builtin::BI__builtin_powi:
   case Builtin::BI__builtin_powif:
-  case Builtin::BI__builtin_powil:
-    return RValue::get(emitBinaryMaybeConstrainedFPBuiltin(
-        *this, E, Intrinsic::powi, Intrinsic::experimental_constrained_powi));
+  case Builtin::BI__builtin_powil: {
+    llvm::Value *Src0 = EmitScalarExpr(E->getArg(0));
+    llvm::Value *Src1 = EmitScalarExpr(E->getArg(1));
 
+    if (Builder.getIsFPConstrained()) {
+      CodeGenFunction::CGFPOptionsRAII FPOptsRAII(*this, E);
+      Function *F = CGM.getIntrinsic(Intrinsic::experimental_constrained_powi,
+                                     Src0->getType());
+      return RValue::get(Builder.CreateConstrainedFPCall(F, { Src0, Src1 }));
+    }
+
+    Function *F = CGM.getIntrinsic(Intrinsic::powi,
+                                   { Src0->getType(), Src1->getType() });
+    return RValue::get(Builder.CreateCall(F, { Src0, Src1 }));
+  }
   case Builtin::BI__builtin_isgreater:
   case Builtin::BI__builtin_isgreaterequal:
   case Builtin::BI__builtin_isless:

clang/test/CodeGen/avr-builtins.c

@@ -104,3 +104,24 @@ unsigned long long byteswap64(unsigned long long x) {
 
 // CHECK: define{{.*}} i64 @byteswap64
 // CHECK: i64 @llvm.bswap.i64(i64
+
+double powi(double x, int y) {
+  return __builtin_powi(x, y);
+}
+
+// CHECK: define{{.*}} float @powi
+// CHECK: float @llvm.powi.f32.i16(float %0, i16 %1)
+
+float powif(float x, int y) {
+    return __builtin_powif(x, y);
+}
+
+// CHECK: define{{.*}} float @powif
+// CHECK: float @llvm.powi.f32.i16(float %0, i16 %1)
+
+long double powil(long double x, int y) {
+    return __builtin_powil(x, y);
+}
+
+// CHECK: define{{.*}} float @powil
+// CHECK: float @llvm.powi.f32.i16(float %0, i16 %1)

clang/test/CodeGen/math-builtins.c

@@ -133,12 +133,12 @@ void foo(double *d, float f, float *fp, long double *l, int *i, const char *c) {
 
   __builtin_powi(f,f);        __builtin_powif(f,f);       __builtin_powil(f,f);
 
-// NO__ERRNO: declare double @llvm.powi.f64(double, i32) [[READNONE_INTRINSIC]]
-// NO__ERRNO: declare float @llvm.powi.f32(float, i32) [[READNONE_INTRINSIC]]
-// NO__ERRNO: declare x86_fp80 @llvm.powi.f80(x86_fp80, i32) [[READNONE_INTRINSIC]]
-// HAS_ERRNO: declare double @llvm.powi.f64(double, i32) [[READNONE_INTRINSIC]]
-// HAS_ERRNO: declare float @llvm.powi.f32(float, i32) [[READNONE_INTRINSIC]]
-// HAS_ERRNO: declare x86_fp80 @llvm.powi.f80(x86_fp80, i32) [[READNONE_INTRINSIC]]
+// NO__ERRNO: declare double @llvm.powi.f64.i32(double, i32) [[READNONE_INTRINSIC]]
+// NO__ERRNO: declare float @llvm.powi.f32.i32(float, i32) [[READNONE_INTRINSIC]]
+// NO__ERRNO: declare x86_fp80 @llvm.powi.f80.i32(x86_fp80, i32) [[READNONE_INTRINSIC]]
+// HAS_ERRNO: declare double @llvm.powi.f64.i32(double, i32) [[READNONE_INTRINSIC]]
+// HAS_ERRNO: declare float @llvm.powi.f32.i32(float, i32) [[READNONE_INTRINSIC]]
+// HAS_ERRNO: declare x86_fp80 @llvm.powi.f80.i32(x86_fp80, i32) [[READNONE_INTRINSIC]]
 
   /* math */
   __builtin_acos(f);       __builtin_acosf(f);      __builtin_acosl(f); __builtin_acosf128(f);

clang/test/CodeGen/msp430-builtins.c

@@ -0,0 +1,37 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// RUN: %clang_cc1 -triple msp430-unknown-unknown -O3 -emit-llvm -o- %s | FileCheck %s
+// REQUIRES: msp430-registered-target
+
+_Static_assert(sizeof(int) == 2, "Assumption failed");
+_Static_assert(sizeof(long) == 4, "Assumption failed");
+_Static_assert(sizeof(long long) == 8, "Assumption failed");
+_Static_assert(sizeof(float) == 4, "Assumption failed");
+_Static_assert(sizeof(double) == 8, "Assumption failed");
+_Static_assert(sizeof(long double) == 8, "Assumption failed");
+
+// CHECK-LABEL: @powif(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call float @llvm.powi.f32.i16(float [[X:%.*]], i16 [[Y:%.*]])
+// CHECK-NEXT:    ret float [[TMP0]]
+//
+float powif(float x, int y) {
+    return __builtin_powif(x, y);
+}
+
+// CHECK-LABEL: @powi(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call double @llvm.powi.f64.i16(double [[X:%.*]], i16 [[Y:%.*]])
+// CHECK-NEXT:    ret double [[TMP0]]
+//
+double powi(double x, int y) {
+  return __builtin_powi(x, y);
+}
+
+// CHECK-LABEL: @powil(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call double @llvm.powi.f64.i16(double [[X:%.*]], i16 [[Y:%.*]])
+// CHECK-NEXT:    ret double [[TMP0]]
+//
+long double powil(long double x, int y) {
+    return __builtin_powil(x, y);
+}

llvm/docs/LangRef.rst

@@ -13647,13 +13647,16 @@ This is an overloaded intrinsic. You can use ``llvm.powi`` on any
 floating-point or vector of floating-point type. Not all targets support
 all types however.
 
+Generally, the only supported type for the exponent is the one matching
+with the C type ``int``.
+
 ::
 
-      declare float     @llvm.powi.f32(float  %Val, i32 %power)
-      declare double    @llvm.powi.f64(double %Val, i32 %power)
-      declare x86_fp80  @llvm.powi.f80(x86_fp80  %Val, i32 %power)
-      declare fp128     @llvm.powi.f128(fp128 %Val, i32 %power)
-      declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128  %Val, i32 %power)
+      declare float     @llvm.powi.f32.i32(float  %Val, i32 %power)
+      declare double    @llvm.powi.f64.i16(double %Val, i16 %power)
+      declare x86_fp80  @llvm.powi.f80.i32(x86_fp80  %Val, i32 %power)
+      declare fp128     @llvm.powi.f128.i32(fp128 %Val, i32 %power)
+      declare ppc_fp128 @llvm.powi.ppcf128.i32(ppc_fp128  %Val, i32 %power)
 
 Overview:
 """""""""

llvm/include/llvm/Analysis/VectorUtils.h

@@ -317,6 +317,11 @@ bool isTriviallyVectorizable(Intrinsic::ID ID);
 /// Identifies if the vector form of the intrinsic has a scalar operand.
 bool hasVectorInstrinsicScalarOpd(Intrinsic::ID ID, unsigned ScalarOpdIdx);
 
+/// Identifies if the vector form of the intrinsic has a scalar operand that has
+/// an overloaded type.
+bool hasVectorInstrinsicOverloadedScalarOpd(Intrinsic::ID ID,
+                                            unsigned ScalarOpdIdx);
+
 /// Returns intrinsic ID for call.
 /// For the input call instruction it finds mapping intrinsic and returns
 /// its intrinsic ID, in case it does not found it return not_intrinsic.

llvm/include/llvm/CodeGen/ISDOpcodes.h

@@ -851,8 +851,8 @@ enum NodeType {
   STRICT_FP_TO_FP16,
 
   /// Perform various unary floating-point operations inspired by libm. For
-  /// FPOWI, the result is undefined if if the integer operand doesn't fit
-  /// into 32 bits.
+  /// FPOWI, the result is undefined if if the integer operand doesn't fit into
+  /// sizeof(int).
   FNEG,
   FABS,
   FSQRT,

llvm/include/llvm/IR/Intrinsics.td

@@ -652,7 +652,7 @@ let IntrProperties = [IntrNoMem, IntrSpeculatable, IntrWillReturn] in {
   // rounding mode. LLVM purposely does not model changes to the FP
   // environment so they can be treated as readnone.
   def int_sqrt : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>]>;
-  def int_powi : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>, llvm_i32_ty]>;
+  def int_powi : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>, llvm_anyint_ty]>;
   def int_sin  : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>]>;
   def int_cos  : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>]>;
   def int_pow  : DefaultAttrsIntrinsic<[llvm_anyfloat_ty],

llvm/lib/Analysis/VectorUtils.cpp

@@ -114,6 +114,16 @@ bool llvm::hasVectorInstrinsicScalarOpd(Intrinsic::ID ID,
   }
 }
 
+bool llvm::hasVectorInstrinsicOverloadedScalarOpd(Intrinsic::ID ID,
+                                                  unsigned ScalarOpdIdx) {
+  switch (ID) {
+  case Intrinsic::powi:
+    return (ScalarOpdIdx == 1);
+  default:
+    return false;
+  }
+}
+
 /// Returns intrinsic ID for call.
 /// For the input call instruction it finds mapping intrinsic and returns
 /// its ID, in case it does not found it return not_intrinsic.

llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp

@@ -4044,6 +4044,17 @@ void SelectionDAGLegalize::ConvertNodeToLibcall(SDNode *Node) {
                                     Exponent));
       break;
     }
+    unsigned Offset = Node->isStrictFPOpcode() ? 1 : 0;
+    bool ExponentHasSizeOfInt =
+        DAG.getLibInfo().getIntSize() ==
+        Node->getOperand(1 + Offset).getValueType().getSizeInBits();
+    if (!ExponentHasSizeOfInt) {
+      // If the exponent does not match with sizeof(int) a libcall to
+      // RTLIB::POWI would use the wrong type for the argument.
+      DAG.getContext()->emitError("POWI exponent does not match sizeof(int)");
+      Results.push_back(DAG.getUNDEF(Node->getValueType(0)));
+      break;
+    }
     ExpandFPLibCall(Node, LC, Results);
     break;
   }