[TTI] Constify BasicTTIImplBase::thisT() (NFCI) (#136575)

The main change is making `thisT` method `const`, the rest of the
changes is fixing compilation errors (*).

(*) There are two tricky methods, `getVectorInstrCost()` and
`getIntImmCost()`.
They have several overloads; some of these overloads are typically
pulled in to derived classes using the `using` directive, and then
hidden by methods in the derived class.
The compiler does not complain if the hiding methods are not marked as
`const`, which means that clients will use the methods from the base
class. If after this change your target fails cost model tests, this
must be the reason. To resolve the issue you need  to make all hiding
overloads `const`. See the second commit in this PR.

Pull Request: #136575

Author: s-barannikov

llvm/include/llvm/Analysis/TargetTransformInfoImpl.h

@@ -47,6 +47,8 @@ class TargetTransformInfoImplBase {
 
   const DataLayout &getDataLayout() const { return DL; }
 
+  // FIXME: It looks like this implementation is dead. All clients appear to
+  //  use the (non-const) version from `TargetTransformInfoImplCRTPBase`.
   InstructionCost getGEPCost(Type *PointeeType, const Value *Ptr,
                              ArrayRef<const Value *> Operands, Type *AccessType,
                              TTI::TargetCostKind CostKind) const {

llvm/include/llvm/CodeGen/BasicTTIImpl.h

@@ -371,7 +371,7 @@ bool AArch64TTIImpl::shouldMaximizeVectorBandwidth(
 /// Calculate the cost of materializing a 64-bit value. This helper
 /// method might only calculate a fraction of a larger immediate. Therefore it
 /// is valid to return a cost of ZERO.
-InstructionCost AArch64TTIImpl::getIntImmCost(int64_t Val) {
+InstructionCost AArch64TTIImpl::getIntImmCost(int64_t Val) const {
   // Check if the immediate can be encoded within an instruction.
   if (Val == 0 || AArch64_AM::isLogicalImmediate(Val, 64))
     return 0;
@@ -386,8 +386,9 @@ InstructionCost AArch64TTIImpl::getIntImmCost(int64_t Val) {
 }
 
 /// Calculate the cost of materializing the given constant.
-InstructionCost AArch64TTIImpl::getIntImmCost(const APInt &Imm, Type *Ty,
-                                              TTI::TargetCostKind CostKind) {
+InstructionCost
+AArch64TTIImpl::getIntImmCost(const APInt &Imm, Type *Ty,
+                              TTI::TargetCostKind CostKind) const {
   assert(Ty->isIntegerTy());
 
   unsigned BitSize = Ty->getPrimitiveSizeInBits();
@@ -577,7 +578,7 @@ static InstructionCost getHistogramCost(const IntrinsicCostAttributes &ICA) {
 
 InstructionCost
 AArch64TTIImpl::getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
-                                      TTI::TargetCostKind CostKind) {
+                                      TTI::TargetCostKind CostKind) const {
   // The code-generator is currently not able to handle scalable vectors
   // of <vscale x 1 x eltty> yet, so return an invalid cost to avoid selecting
   // it. This change will be removed when code-generation for these types is
@@ -2806,7 +2807,7 @@ AArch64TTIImpl::getRegisterBitWidth(TargetTransformInfo::RegisterKind K) const {
 
 bool AArch64TTIImpl::isWideningInstruction(Type *DstTy, unsigned Opcode,
                                            ArrayRef<const Value *> Args,
-                                           Type *SrcOverrideTy) {
+                                           Type *SrcOverrideTy) const {
   // A helper that returns a vector type from the given type. The number of
   // elements in type Ty determines the vector width.
   auto toVectorTy = [&](Type *ArgTy) {
@@ -2903,7 +2904,7 @@ bool AArch64TTIImpl::isWideningInstruction(Type *DstTy, unsigned Opcode,
 //   trunc i16 (lshr (add %x, %y), 1) -> i8
 //
 bool AArch64TTIImpl::isExtPartOfAvgExpr(const Instruction *ExtUser, Type *Dst,
-                                        Type *Src) {
+                                        Type *Src) const {
   // The source should be a legal vector type.
   if (!Src->isVectorTy() || !TLI->isTypeLegal(TLI->getValueType(DL, Src)) ||
       (Src->isScalableTy() && !ST->hasSVE2()))
@@ -2948,7 +2949,7 @@ InstructionCost AArch64TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
                                                  Type *Src,
                                                  TTI::CastContextHint CCH,
                                                  TTI::TargetCostKind CostKind,
-                                                 const Instruction *I) {
+                                                 const Instruction *I) const {
   int ISD = TLI->InstructionOpcodeToISD(Opcode);
   assert(ISD && "Invalid opcode");
   // If the cast is observable, and it is used by a widening instruction (e.g.,
@@ -3619,7 +3620,7 @@ InstructionCost AArch64TTIImpl::getExtractWithExtendCost(unsigned Opcode,
 
 InstructionCost AArch64TTIImpl::getCFInstrCost(unsigned Opcode,
                                                TTI::TargetCostKind CostKind,
-                                               const Instruction *I) {
+                                               const Instruction *I) const {
   if (CostKind != TTI::TCK_RecipThroughput)
     return Opcode == Instruction::PHI ? 0 : 1;
   assert(CostKind == TTI::TCK_RecipThroughput && "unexpected CostKind");
@@ -3630,7 +3631,7 @@ InstructionCost AArch64TTIImpl::getCFInstrCost(unsigned Opcode,
 InstructionCost AArch64TTIImpl::getVectorInstrCostHelper(
     unsigned Opcode, Type *Val, TTI::TargetCostKind CostKind, unsigned Index,
     bool HasRealUse, const Instruction *I, Value *Scalar,
-    ArrayRef<std::tuple<Value *, User *, int>> ScalarUserAndIdx) {
+    ArrayRef<std::tuple<Value *, User *, int>> ScalarUserAndIdx) const {
   assert(Val->isVectorTy() && "This must be a vector type");
 
   if (Index != -1U) {
@@ -3802,7 +3803,7 @@ InstructionCost AArch64TTIImpl::getVectorInstrCostHelper(
 InstructionCost AArch64TTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
                                                    TTI::TargetCostKind CostKind,
                                                    unsigned Index, Value *Op0,
-                                                   Value *Op1) {
+                                                   Value *Op1) const {
   bool HasRealUse =
       Opcode == Instruction::InsertElement && Op0 && !isa<UndefValue>(Op0);
   return getVectorInstrCostHelper(Opcode, Val, CostKind, Index, HasRealUse);
@@ -3826,7 +3827,7 @@ InstructionCost AArch64TTIImpl::getVectorInstrCost(const Instruction &I,
 
 InstructionCost AArch64TTIImpl::getScalarizationOverhead(
     VectorType *Ty, const APInt &DemandedElts, bool Insert, bool Extract,
-    TTI::TargetCostKind CostKind, ArrayRef<Value *> VL) {
+    TTI::TargetCostKind CostKind, ArrayRef<Value *> VL) const {
   if (isa<ScalableVectorType>(Ty))
     return InstructionCost::getInvalid();
   if (Ty->getElementType()->isFloatingPointTy())
@@ -3840,8 +3841,7 @@ InstructionCost AArch64TTIImpl::getScalarizationOverhead(
 InstructionCost AArch64TTIImpl::getArithmeticInstrCost(
     unsigned Opcode, Type *Ty, TTI::TargetCostKind CostKind,
     TTI::OperandValueInfo Op1Info, TTI::OperandValueInfo Op2Info,
-    ArrayRef<const Value *> Args,
-    const Instruction *CxtI) {
+    ArrayRef<const Value *> Args, const Instruction *CxtI) const {
 
   // The code-generator is currently not able to handle scalable vectors
   // of <vscale x 1 x eltty> yet, so return an invalid cost to avoid selecting
@@ -4171,7 +4171,7 @@ InstructionCost AArch64TTIImpl::getAddressComputationCost(Type *Ty,
 InstructionCost AArch64TTIImpl::getCmpSelInstrCost(
     unsigned Opcode, Type *ValTy, Type *CondTy, CmpInst::Predicate VecPred,
     TTI::TargetCostKind CostKind, TTI::OperandValueInfo Op1Info,
-    TTI::OperandValueInfo Op2Info, const Instruction *I) {
+    TTI::OperandValueInfo Op2Info, const Instruction *I) const {
   // TODO: Handle other cost kinds.
   if (CostKind != TTI::TCK_RecipThroughput)
     return BaseT::getCmpSelInstrCost(Opcode, ValTy, CondTy, VecPred, CostKind,
@@ -4284,7 +4284,7 @@ bool AArch64TTIImpl::prefersVectorizedAddressing() const {
 InstructionCost
 AArch64TTIImpl::getMaskedMemoryOpCost(unsigned Opcode, Type *Src,
                                       Align Alignment, unsigned AddressSpace,
-                                      TTI::TargetCostKind CostKind) {
+                                      TTI::TargetCostKind CostKind) const {
   if (useNeonVector(Src))
     return BaseT::getMaskedMemoryOpCost(Opcode, Src, Alignment, AddressSpace,
                                         CostKind);
@@ -4331,7 +4331,7 @@ static unsigned getSVEGatherScatterOverhead(unsigned Opcode,
 
 InstructionCost AArch64TTIImpl::getGatherScatterOpCost(
     unsigned Opcode, Type *DataTy, const Value *Ptr, bool VariableMask,
-    Align Alignment, TTI::TargetCostKind CostKind, const Instruction *I) {
+    Align Alignment, TTI::TargetCostKind CostKind, const Instruction *I) const {
   if (useNeonVector(DataTy) || !isLegalMaskedGatherScatter(DataTy))
     return BaseT::getGatherScatterOpCost(Opcode, DataTy, Ptr, VariableMask,
                                          Alignment, CostKind, I);
@@ -4371,7 +4371,7 @@ InstructionCost AArch64TTIImpl::getMemoryOpCost(unsigned Opcode, Type *Ty,
                                                 unsigned AddressSpace,
                                                 TTI::TargetCostKind CostKind,
                                                 TTI::OperandValueInfo OpInfo,
-                                                const Instruction *I) {
+                                                const Instruction *I) const {
   EVT VT = TLI->getValueType(DL, Ty, true);
   // Type legalization can't handle structs
   if (VT == MVT::Other)
@@ -4980,7 +4980,7 @@ bool AArch64TTIImpl::isLegalToVectorizeReduction(
 InstructionCost
 AArch64TTIImpl::getMinMaxReductionCost(Intrinsic::ID IID, VectorType *Ty,
                                        FastMathFlags FMF,
-                                       TTI::TargetCostKind CostKind) {
+                                       TTI::TargetCostKind CostKind) const {
   // The code-generator is currently not able to handle scalable vectors
   // of <vscale x 1 x eltty> yet, so return an invalid cost to avoid selecting
   // it. This change will be removed when code-generation for these types is
@@ -5005,7 +5005,7 @@ AArch64TTIImpl::getMinMaxReductionCost(Intrinsic::ID IID, VectorType *Ty,
 }
 
 InstructionCost AArch64TTIImpl::getArithmeticReductionCostSVE(
-    unsigned Opcode, VectorType *ValTy, TTI::TargetCostKind CostKind) {
+    unsigned Opcode, VectorType *ValTy, TTI::TargetCostKind CostKind) const {
   std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(ValTy);
   InstructionCost LegalizationCost = 0;
   if (LT.first > 1) {
@@ -5032,7 +5032,7 @@ InstructionCost AArch64TTIImpl::getArithmeticReductionCostSVE(
 InstructionCost
 AArch64TTIImpl::getArithmeticReductionCost(unsigned Opcode, VectorType *ValTy,
                                            std::optional<FastMathFlags> FMF,
-                                           TTI::TargetCostKind CostKind) {
+                                           TTI::TargetCostKind CostKind) const {
   // The code-generator is currently not able to handle scalable vectors
   // of <vscale x 1 x eltty> yet, so return an invalid cost to avoid selecting
   // it. This change will be removed when code-generation for these types is
@@ -5207,8 +5207,9 @@ AArch64TTIImpl::getMulAccReductionCost(bool IsUnsigned, Type *ResTy,
   return BaseT::getMulAccReductionCost(IsUnsigned, ResTy, VecTy, CostKind);
 }
 
-InstructionCost AArch64TTIImpl::getSpliceCost(VectorType *Tp, int Index,
-                                              TTI::TargetCostKind CostKind) {
+InstructionCost
+AArch64TTIImpl::getSpliceCost(VectorType *Tp, int Index,
+                              TTI::TargetCostKind CostKind) const {
   static const CostTblEntry ShuffleTbl[] = {
       { TTI::SK_Splice, MVT::nxv16i8,  1 },
       { TTI::SK_Splice, MVT::nxv8i16,  1 },
@@ -5340,7 +5341,7 @@ InstructionCost AArch64TTIImpl::getPartialReductionCost(
 InstructionCost AArch64TTIImpl::getShuffleCost(
     TTI::ShuffleKind Kind, VectorType *Tp, ArrayRef<int> Mask,
     TTI::TargetCostKind CostKind, int Index, VectorType *SubTp,
-    ArrayRef<const Value *> Args, const Instruction *CxtI) {
+    ArrayRef<const Value *> Args, const Instruction *CxtI) const {
   std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(Tp);
 
   // If we have a Mask, and the LT is being legalized somehow, split the Mask

llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp

@@ -61,7 +61,7 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
 
   bool isWideningInstruction(Type *DstTy, unsigned Opcode,
                              ArrayRef<const Value *> Args,
-                             Type *SrcOverrideTy = nullptr);
+                             Type *SrcOverrideTy = nullptr) const;
 
   // A helper function called by 'getVectorInstrCost'.
   //
@@ -75,7 +75,7 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
   InstructionCost getVectorInstrCostHelper(
       unsigned Opcode, Type *Val, TTI::TargetCostKind CostKind, unsigned Index,
       bool HasRealUse, const Instruction *I = nullptr, Value *Scalar = nullptr,
-      ArrayRef<std::tuple<Value *, User *, int>> ScalarUserAndIdx = {});
+      ArrayRef<std::tuple<Value *, User *, int>> ScalarUserAndIdx = {}) const;
 
 public:
   explicit AArch64TTIImpl(const AArch64TargetMachine *TM, const Function &F)
@@ -99,9 +99,9 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
   /// @{
 
   using BaseT::getIntImmCost;
-  InstructionCost getIntImmCost(int64_t Val);
+  InstructionCost getIntImmCost(int64_t Val) const;
   InstructionCost getIntImmCost(const APInt &Imm, Type *Ty,
-                                TTI::TargetCostKind CostKind);
+                                TTI::TargetCostKind CostKind) const;
   InstructionCost getIntImmCostInst(unsigned Opcode, unsigned Idx,
                                     const APInt &Imm, Type *Ty,
                                     TTI::TargetCostKind CostKind,
@@ -131,7 +131,7 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
   }
 
   InstructionCost getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
-                                        TTI::TargetCostKind CostKind);
+                                        TTI::TargetCostKind CostKind) const;
 
   std::optional<Instruction *> instCombineIntrinsic(InstCombiner &IC,
                                                     IntrinsicInst &II) const;
@@ -173,30 +173,32 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
 
   InstructionCost getMaskedMemoryOpCost(unsigned Opcode, Type *Src,
                                         Align Alignment, unsigned AddressSpace,
-                                        TTI::TargetCostKind CostKind);
+                                        TTI::TargetCostKind CostKind) const;
 
   InstructionCost getGatherScatterOpCost(unsigned Opcode, Type *DataTy,
                                          const Value *Ptr, bool VariableMask,
                                          Align Alignment,
                                          TTI::TargetCostKind CostKind,
-                                         const Instruction *I = nullptr);
+                                         const Instruction *I = nullptr) const;
 
-  bool isExtPartOfAvgExpr(const Instruction *ExtUser, Type *Dst, Type *Src);
+  bool isExtPartOfAvgExpr(const Instruction *ExtUser, Type *Dst,
+                          Type *Src) const;
 
   InstructionCost getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src,
                                    TTI::CastContextHint CCH,
                                    TTI::TargetCostKind CostKind,
-                                   const Instruction *I = nullptr);
+                                   const Instruction *I = nullptr) const;
 
   InstructionCost getExtractWithExtendCost(unsigned Opcode, Type *Dst,
                                            VectorType *VecTy, unsigned Index);
 
   InstructionCost getCFInstrCost(unsigned Opcode, TTI::TargetCostKind CostKind,
-                                 const Instruction *I = nullptr);
+                                 const Instruction *I = nullptr) const;
 
   InstructionCost getVectorInstrCost(unsigned Opcode, Type *Val,
                                      TTI::TargetCostKind CostKind,
-                                     unsigned Index, Value *Op0, Value *Op1);
+                                     unsigned Index, Value *Op0,
+                                     Value *Op1) const;
 
   /// \param ScalarUserAndIdx encodes the information about extracts from a
   /// vector with 'Scalar' being the value being extracted,'User' being the user
@@ -213,20 +215,21 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
 
   InstructionCost getMinMaxReductionCost(Intrinsic::ID IID, VectorType *Ty,
                                          FastMathFlags FMF,
-                                         TTI::TargetCostKind CostKind);
+                                         TTI::TargetCostKind CostKind) const;
 
-  InstructionCost getArithmeticReductionCostSVE(unsigned Opcode,
-                                                VectorType *ValTy,
-                                                TTI::TargetCostKind CostKind);
+  InstructionCost
+  getArithmeticReductionCostSVE(unsigned Opcode, VectorType *ValTy,
+                                TTI::TargetCostKind CostKind) const;
 
   InstructionCost getSpliceCost(VectorType *Tp, int Index,
-                                TTI::TargetCostKind CostKind);
+                                TTI::TargetCostKind CostKind) const;
 
   InstructionCost getArithmeticInstrCost(
       unsigned Opcode, Type *Ty, TTI::TargetCostKind CostKind,
       TTI::OperandValueInfo Op1Info = {TTI::OK_AnyValue, TTI::OP_None},
       TTI::OperandValueInfo Op2Info = {TTI::OK_AnyValue, TTI::OP_None},
-      ArrayRef<const Value *> Args = {}, const Instruction *CxtI = nullptr);
+      ArrayRef<const Value *> Args = {},
+      const Instruction *CxtI = nullptr) const;
 
   InstructionCost getAddressComputationCost(Type *Ty, ScalarEvolution *SE,
                                             const SCEV *Ptr);
@@ -236,17 +239,17 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
       TTI::TargetCostKind CostKind,
       TTI::OperandValueInfo Op1Info = {TTI::OK_AnyValue, TTI::OP_None},
       TTI::OperandValueInfo Op2Info = {TTI::OK_AnyValue, TTI::OP_None},
-      const Instruction *I = nullptr);
+      const Instruction *I = nullptr) const;
 
   TTI::MemCmpExpansionOptions enableMemCmpExpansion(bool OptSize,
                                                     bool IsZeroCmp) const;
   bool useNeonVector(const Type *Ty) const;
 
-  InstructionCost
-  getMemoryOpCost(unsigned Opcode, Type *Src, MaybeAlign Alignment,
-                  unsigned AddressSpace, TTI::TargetCostKind CostKind,
-                  TTI::OperandValueInfo OpInfo = {TTI::OK_AnyValue, TTI::OP_None},
-                  const Instruction *I = nullptr);
+  InstructionCost getMemoryOpCost(
+      unsigned Opcode, Type *Src, MaybeAlign Alignment, unsigned AddressSpace,
+      TTI::TargetCostKind CostKind,
+      TTI::OperandValueInfo OpInfo = {TTI::OK_AnyValue, TTI::OP_None},
+      const Instruction *I = nullptr) const;
 
   InstructionCost getCostOfKeepingLiveOverCall(ArrayRef<Type *> Tys);
 
@@ -423,9 +426,10 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
     return ST->hasSVE();
   }
 
-  InstructionCost getArithmeticReductionCost(unsigned Opcode, VectorType *Ty,
-                                             std::optional<FastMathFlags> FMF,
-                                             TTI::TargetCostKind CostKind);
+  InstructionCost
+  getArithmeticReductionCost(unsigned Opcode, VectorType *Ty,
+                             std::optional<FastMathFlags> FMF,
+                             TTI::TargetCostKind CostKind) const;
 
   InstructionCost getExtendedReductionCost(unsigned Opcode, bool IsUnsigned,
                                            Type *ResTy, VectorType *ValTy,
@@ -441,13 +445,13 @@ class AArch64TTIImpl : public BasicTTIImplBase<AArch64TTIImpl> {
                                  TTI::TargetCostKind CostKind, int Index,
                                  VectorType *SubTp,
                                  ArrayRef<const Value *> Args = {},
-                                 const Instruction *CxtI = nullptr);
+                                 const Instruction *CxtI = nullptr) const;
 
   InstructionCost getScalarizationOverhead(VectorType *Ty,
                                            const APInt &DemandedElts,
                                            bool Insert, bool Extract,
                                            TTI::TargetCostKind CostKind,
-                                           ArrayRef<Value *> VL = {});
+                                           ArrayRef<Value *> VL = {}) const;
 
   /// Return the cost of the scaling factor used in the addressing
   /// mode represented by AM for this target, for a load/store