[CIR] introduce CIR floating-point types (llvm#385)

This PR adds a dedicated `cir.float` type for representing
floating-point types. There are several issues linked to this PR: llvm#5,
llvm#78, and llvm#90.

clang/include/clang/CIR/Dialect/IR/CIRAttrs.td

@@ -215,6 +215,33 @@ def IntAttr : CIR_Attr<"Int", "int", [TypedAttrInterface]> {
   let hasCustomAssemblyFormat = 1;
 }
 
+//===----------------------------------------------------------------------===//
+// FPAttr
+//===----------------------------------------------------------------------===//
+
+def FPAttr : CIR_Attr<"FP", "fp", [TypedAttrInterface]> {
+  let summary = "An attribute containing a floating-point value";
+  let description = [{
+    An fp attribute is a literal attribute that represents a floating-point
+    value of the specified floating-point type.
+  }];
+  let parameters = (ins AttributeSelfTypeParameter<"">:$type, "APFloat":$value);
+  let builders = [
+    AttrBuilderWithInferredContext<(ins "Type":$type,
+                                        "const APFloat &":$value), [{
+      return $_get(type.getContext(), type, value);
+    }]>,
+  ];
+  let extraClassDeclaration = [{
+    static FPAttr getZero(mlir::Type type);
+  }];
+  let genVerifyDecl = 1;
+
+  let assemblyFormat = [{
+    `<` custom<FloatLiteral>($value, ref($type)) `>`
+  }];
+}
+
 //===----------------------------------------------------------------------===//
 // ConstPointerAttr
 //===----------------------------------------------------------------------===//

clang/include/clang/CIR/Dialect/IR/CIROps.td

@@ -2615,8 +2615,8 @@ def IterEndOp : CIR_Op<"iterator_end"> {
 
 class UnaryFPToFPBuiltinOp<string mnemonic>
     : CIR_Op<mnemonic, [Pure, SameOperandsAndResultType]> {
-  let arguments = (ins AnyFloat:$src);
-  let results = (outs AnyFloat:$result);
+  let arguments = (ins CIR_AnyFloat:$src);
+  let results = (outs CIR_AnyFloat:$result);
   let summary = "libc builtin equivalent ignoring "
                 "floating point exceptions and errno";
   let assemblyFormat = "$src `:` type($src) attr-dict";

clang/include/clang/CIR/Dialect/IR/CIRTypes.h

@@ -16,6 +16,7 @@
 #include "mlir/IR/BuiltinAttributes.h"
 #include "mlir/IR/Types.h"
 #include "mlir/Interfaces/DataLayoutInterfaces.h"
+#include "clang/CIR/Interfaces/CIRFPTypeInterface.h"
 
 #include "clang/CIR/Interfaces/ASTAttrInterfaces.h"
 

clang/include/clang/CIR/Dialect/IR/CIRTypes.td

@@ -15,15 +15,18 @@
 
 include "clang/CIR/Dialect/IR/CIRDialect.td"
 include "clang/CIR/Interfaces/ASTAttrInterfaces.td"
+include "clang/CIR/Interfaces/CIRFPTypeInterface.td"
 include "mlir/Interfaces/DataLayoutInterfaces.td"
 include "mlir/IR/AttrTypeBase.td"
+include "mlir/IR/EnumAttr.td"
 
 //===----------------------------------------------------------------------===//
 // CIR Types
 //===----------------------------------------------------------------------===//
 
-class CIR_Type<string name, string typeMnemonic, list<Trait> traits = []> :
-    TypeDef<CIR_Dialect, name, traits> {
+class CIR_Type<string name, string typeMnemonic, list<Trait> traits = [],
+               string baseCppClass = "::mlir::Type">
+    : TypeDef<CIR_Dialect, name, traits, baseCppClass> {
   let mnemonic = typeMnemonic;
 }
 
@@ -94,6 +97,37 @@ def SInt16 : SInt<16>;
 def SInt32 : SInt<32>;
 def SInt64 : SInt<64>;
 
+//===----------------------------------------------------------------------===//
+// FloatType
+//===----------------------------------------------------------------------===//
+
+class CIR_FloatType<string name, string mnemonic> 
+    : CIR_Type<name, mnemonic,
+          [
+            DeclareTypeInterfaceMethods<DataLayoutTypeInterface>,
+            DeclareTypeInterfaceMethods<CIRFPTypeInterface>,
+          ]> {}
+
+def CIR_Single : CIR_FloatType<"Single", "float"> {
+  let summary = "CIR single-precision float type";
+  let description = [{
+    Floating-point type that represents the `float` type in C/C++. Its
+    underlying floating-point format is the IEEE-754 binary32 format.
+  }];
+}
+
+def CIR_Double : CIR_FloatType<"Double", "double"> {
+  let summary = "CIR double-precision float type";
+  let description = [{
+    Floating-point type that represents the `double` type in C/C++. Its
+    underlying floating-point format is the IEEE-754 binar64 format.
+  }];
+}
+
+// Constraints
+
+def CIR_AnyFloat: AnyTypeOf<[CIR_Single, CIR_Double]>;
+
 //===----------------------------------------------------------------------===//
 // PointerType
 //===----------------------------------------------------------------------===//
@@ -318,7 +352,7 @@ def CIR_StructType : Type<CPred<"$_self.isa<::mlir::cir::StructType>()">,
 
 def CIR_AnyType : AnyTypeOf<[
   CIR_IntType, CIR_PointerType, CIR_BoolType, CIR_ArrayType, CIR_VectorType,
-  CIR_FuncType, CIR_VoidType, CIR_StructType, CIR_ExceptionInfo, AnyFloat,
+  CIR_FuncType, CIR_VoidType, CIR_StructType, CIR_ExceptionInfo, CIR_AnyFloat,
 ]>;
 
 #endif // MLIR_CIR_DIALECT_CIR_TYPES

clang/include/clang/CIR/Interfaces/CIRFPTypeInterface.h

@@ -0,0 +1,22 @@
+//===- CIRFPTypeInterface.h - Interface for CIR FP types -------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===---------------------------------------------------------------------===//
+//
+// Defines the interface to generically handle CIR floating-point types.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef CLANG_INTERFACES_CIR_CIR_FPTYPEINTERFACE_H
+#define CLANG_INTERFACES_CIR_CIR_FPTYPEINTERFACE_H
+
+#include "mlir/IR/Types.h"
+#include "llvm/ADT/APFloat.h"
+
+/// Include the tablegen'd interface declarations.
+#include "clang/CIR/Interfaces/CIRFPTypeInterface.h.inc"
+
+#endif // CLANG_INTERFACES_CIR_CIR_FPTYPEINTERFACE_H

clang/include/clang/CIR/Interfaces/CIRFPTypeInterface.td

@@ -0,0 +1,52 @@
+//===- CIRFPTypeInterface.td - CIR FP Interface Definitions -----*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef MLIR_CIR_INTERFACES_CIR_FP_TYPE_INTERFACE
+#define MLIR_CIR_INTERFACES_CIR_FP_TYPE_INTERFACE
+
+include "mlir/IR/OpBase.td"
+
+def CIRFPTypeInterface : TypeInterface<"CIRFPTypeInterface"> {
+  let description = [{
+    Contains helper functions to query properties about a floating-point type.
+  }];
+  let cppNamespace = "::mlir::cir";
+
+  let methods = [
+    InterfaceMethod<[{
+        Returns the bit width of this floating-point type.
+      }],
+      /*retTy=*/"unsigned",
+      /*methodName=*/"getWidth",
+      /*args=*/(ins),
+      /*methodBody=*/"",
+      /*defaultImplementation=*/[{
+          return llvm::APFloat::semanticsSizeInBits($_type.getFloatSemantics());
+        }]
+    >,
+    InterfaceMethod<[{
+        Return the mantissa width.
+      }],
+      /*retTy=*/"unsigned",
+      /*methodName=*/"getFPMantissaWidth",
+      /*args=*/(ins),
+      /*methodBody=*/"",
+      /*defaultImplementation=*/[{
+          return llvm::APFloat::semanticsPrecision($_type.getFloatSemantics());
+        }]
+    >,
+    InterfaceMethod<[{
+        Return the float semantics of this floating-point type.
+      }],
+      /*retTy=*/"const llvm::fltSemantics &",
+      /*methodName=*/"getFloatSemantics"
+    >,
+  ];
+}
+
+#endif // MLIR_CIR_INTERFACES_CIR_FP_TYPE_INTERFACE

clang/include/clang/CIR/Interfaces/CMakeLists.txt

@@ -20,6 +20,15 @@ function(add_clang_mlir_op_interface interface)
   add_dependencies(mlir-generic-headers MLIR${interface}IncGen)
 endfunction()
 
+function(add_clang_mlir_type_interface interface)
+  set(LLVM_TARGET_DEFINITIONS ${interface}.td)
+  mlir_tablegen(${interface}.h.inc -gen-type-interface-decls)
+  mlir_tablegen(${interface}.cpp.inc -gen-type-interface-defs)
+  add_public_tablegen_target(MLIR${interface}IncGen)
+  add_dependencies(mlir-generic-headers MLIR${interface}IncGen)
+endfunction()
+
 add_clang_mlir_attr_interface(ASTAttrInterfaces)
 add_clang_mlir_op_interface(CIROpInterfaces)
 add_clang_mlir_op_interface(CIRLoopOpInterface)
+add_clang_mlir_type_interface(CIRFPTypeInterface)

clang/lib/CIR/CodeGen/CIRGenBuilder.h

@@ -224,8 +224,10 @@ class CIRGenBuilderTy : public CIRBaseBuilderTy {
   mlir::TypedAttr getZeroInitAttr(mlir::Type ty) {
     if (ty.isa<mlir::cir::IntType>())
       return mlir::cir::IntAttr::get(ty, 0);
-    if (ty.isa<mlir::FloatType>())
-      return mlir::FloatAttr::get(ty, 0.0);
+    if (auto fltType = ty.dyn_cast<mlir::cir::SingleType>())
+      return mlir::cir::FPAttr::getZero(fltType);
+    if (auto fltType = ty.dyn_cast<mlir::cir::DoubleType>())
+      return mlir::cir::FPAttr::getZero(fltType);
     if (auto arrTy = ty.dyn_cast<mlir::cir::ArrayType>())
       return getZeroAttr(arrTy);
     if (auto ptrTy = ty.dyn_cast<mlir::cir::PointerType>())
@@ -256,12 +258,13 @@ class CIRGenBuilderTy : public CIRBaseBuilderTy {
     if (const auto boolVal = attr.dyn_cast<mlir::cir::BoolAttr>())
       return !boolVal.getValue();
 
-    if (const auto fpVal = attr.dyn_cast<mlir::FloatAttr>()) {
+    if (auto fpAttr = attr.dyn_cast<mlir::cir::FPAttr>()) {
+      auto fpVal = fpAttr.getValue();
       bool ignored;
       llvm::APFloat FV(+0.0);
-      FV.convert(fpVal.getValue().getSemantics(),
-                 llvm::APFloat::rmNearestTiesToEven, &ignored);
-      return FV.bitwiseIsEqual(fpVal.getValue());
+      FV.convert(fpVal.getSemantics(), llvm::APFloat::rmNearestTiesToEven,
+                 &ignored);
+      return FV.bitwiseIsEqual(fpVal);
     }
 
     if (const auto structVal = attr.dyn_cast<mlir::cir::ConstStructAttr>()) {
@@ -348,23 +351,21 @@ class CIRGenBuilderTy : public CIRBaseBuilderTy {
   }
   bool isInt(mlir::Type i) { return i.isa<mlir::cir::IntType>(); }
 
-  mlir::FloatType getLongDouble80BitsTy() const {
-    return typeCache.LongDouble80BitsTy;
-  }
+  mlir::Type getLongDouble80BitsTy() const { llvm_unreachable("NYI"); }
 
   /// Get the proper floating point type for the given semantics.
-  mlir::FloatType getFloatTyForFormat(const llvm::fltSemantics &format,
-                                      bool useNativeHalf) const {
+  mlir::Type getFloatTyForFormat(const llvm::fltSemantics &format,
+                                 bool useNativeHalf) const {
     if (&format == &llvm::APFloat::IEEEhalf()) {
       llvm_unreachable("IEEEhalf float format is NYI");
     }
 
     if (&format == &llvm::APFloat::BFloat())
       llvm_unreachable("BFloat float format is NYI");
     if (&format == &llvm::APFloat::IEEEsingle())
-      llvm_unreachable("IEEEsingle float format is NYI");
+      return typeCache.FloatTy;
     if (&format == &llvm::APFloat::IEEEdouble())
-      llvm_unreachable("IEEEdouble float format is NYI");
+      return typeCache.DoubleTy;
     if (&format == &llvm::APFloat::IEEEquad())
       llvm_unreachable("IEEEquad float format is NYI");
     if (&format == &llvm::APFloat::PPCDoubleDouble())
@@ -491,9 +492,9 @@ class CIRGenBuilderTy : public CIRBaseBuilderTy {
   }
 
   bool isSized(mlir::Type ty) {
-    if (ty.isIntOrFloat() ||
-        ty.isa<mlir::cir::PointerType, mlir::cir::StructType,
-               mlir::cir::ArrayType, mlir::cir::BoolType, mlir::cir::IntType>())
+    if (ty.isa<mlir::cir::PointerType, mlir::cir::StructType,
+               mlir::cir::ArrayType, mlir::cir::BoolType, mlir::cir::IntType,
+               mlir::cir::CIRFPTypeInterface>())
       return true;
     assert(0 && "Unimplemented size for type");
     return false;

clang/lib/CIR/CodeGen/CIRGenExprConst.cpp

@@ -1705,7 +1705,9 @@ mlir::Attribute ConstantEmitter::tryEmitPrivate(const APValue &Value,
       assert(0 && "not implemented");
     else {
       mlir::Type ty = CGM.getCIRType(DestType);
-      return builder.getFloatAttr(ty, Init);
+      assert(ty.isa<mlir::cir::CIRFPTypeInterface>() &&
+             "expected floating-point type");
+      return CGM.getBuilder().getAttr<mlir::cir::FPAttr>(ty, Init);
     }
   }
   case APValue::Array: {

clang/lib/CIR/CodeGen/CIRGenExprScalar.cpp

@@ -165,9 +165,11 @@ class ScalarExprEmitter : public StmtVisitor<ScalarExprEmitter, mlir::Value> {
   }
   mlir::Value VisitFloatingLiteral(const FloatingLiteral *E) {
     mlir::Type Ty = CGF.getCIRType(E->getType());
+    assert(Ty.isa<mlir::cir::CIRFPTypeInterface>() &&
+           "expect floating-point type");
     return Builder.create<mlir::cir::ConstantOp>(
         CGF.getLoc(E->getExprLoc()), Ty,
-        Builder.getFloatAttr(Ty, E->getValue()));
+        Builder.getAttr<mlir::cir::FPAttr>(Ty, E->getValue()));
   }
   mlir::Value VisitCharacterLiteral(const CharacterLiteral *E) {
     mlir::Type Ty = CGF.getCIRType(E->getType());
@@ -1227,7 +1229,7 @@ mlir::Value ScalarExprEmitter::buildSub(const BinOpInfo &Ops) {
       llvm_unreachable("NYI");
 
     assert(!UnimplementedFeature::cirVectorType());
-    if (Ops.LHS.getType().isa<mlir::FloatType>()) {
+    if (Ops.LHS.getType().isa<mlir::cir::CIRFPTypeInterface>()) {
       CIRGenFunction::CIRGenFPOptionsRAII FPOptsRAII(CGF, Ops.FPFeatures);
       return Builder.createFSub(Ops.LHS, Ops.RHS);
     }
@@ -1701,20 +1703,20 @@ mlir::Value ScalarExprEmitter::buildScalarCast(
       llvm_unreachable("NYI: signed bool");
     if (CGF.getBuilder().isInt(DstTy)) {
       CastKind = mlir::cir::CastKind::bool_to_int;
-    } else if (DstTy.isa<mlir::FloatType>()) {
+    } else if (DstTy.isa<mlir::cir::CIRFPTypeInterface>()) {
       CastKind = mlir::cir::CastKind::bool_to_float;
     } else {
       llvm_unreachable("Internal error: Cast to unexpected type");
     }
   } else if (CGF.getBuilder().isInt(SrcTy)) {
     if (CGF.getBuilder().isInt(DstTy)) {
       CastKind = mlir::cir::CastKind::integral;
-    } else if (DstTy.isa<mlir::FloatType>()) {
+    } else if (DstTy.isa<mlir::cir::CIRFPTypeInterface>()) {
       CastKind = mlir::cir::CastKind::int_to_float;
     } else {
       llvm_unreachable("Internal error: Cast to unexpected type");
     }
-  } else if (SrcTy.isa<mlir::FloatType>()) {
+  } else if (SrcTy.isa<mlir::cir::CIRFPTypeInterface>()) {
     if (CGF.getBuilder().isInt(DstTy)) {
       // If we can't recognize overflow as undefined behavior, assume that
       // overflow saturates. This protects against normal optimizations if we
@@ -1724,7 +1726,7 @@ mlir::Value ScalarExprEmitter::buildScalarCast(
       if (Builder.getIsFPConstrained())
         llvm_unreachable("NYI");
       CastKind = mlir::cir::CastKind::float_to_int;
-    } else if (DstTy.isa<mlir::FloatType>()) {
+    } else if (DstTy.isa<mlir::cir::CIRFPTypeInterface>()) {
       // TODO: split this to createFPExt/createFPTrunc
       return Builder.createFloatingCast(Src, DstTy);
     } else {

clang/lib/CIR/CodeGen/CIRGenModule.cpp

@@ -133,11 +133,10 @@ CIRGenModule::CIRGenModule(mlir::MLIRContext &context,
 
   // TODO: HalfTy
   // TODO: BFloatTy
-  FloatTy = builder.getF32Type();
-  DoubleTy = builder.getF64Type();
+  FloatTy = ::mlir::cir::SingleType::get(builder.getContext());
+  DoubleTy = ::mlir::cir::DoubleType::get(builder.getContext());
   // TODO(cir): perhaps we should abstract long double variations into a custom
   // cir.long_double type. Said type would also hold the semantics for lowering.
-  LongDouble80BitsTy = builder.getF80Type();
 
   // TODO: PointerWidthInBits
   PointerAlignInBytes =

clang/lib/CIR/CodeGen/CIRGenTypeCache.h

@@ -37,7 +37,8 @@ struct CIRGenTypeCache {
   // mlir::Type HalfTy, BFloatTy;
   // TODO(cir): perhaps we should abstract long double variations into a custom
   // cir.long_double type. Said type would also hold the semantics for lowering.
-  mlir::FloatType FloatTy, DoubleTy, LongDouble80BitsTy;
+  mlir::cir::SingleType FloatTy;
+  mlir::cir::DoubleType DoubleTy;
 
   /// int
   mlir::Type UIntTy;