1 files changed, 67 insertions, 21 deletions

diff --git a/llvm/lib/Target/ARM/ARMTargetTransformInfo.h b/llvm/lib/Target/ARM/ARMTargetTransformInfo.h
index 880588adfdfdd..7bf6de4bffe07 100644
--- a/llvm/lib/Target/ARM/ARMTargetTransformInfo.h
+++ b/llvm/lib/Target/ARM/ARMTargetTransformInfo.h
@@ -38,6 +38,16 @@ class ScalarEvolution;
 class Type;
 class Value;
 
+namespace TailPredication {
+  enum Mode {
+    Disabled = 0,
+    EnabledNoReductions,
+    Enabled,
+    ForceEnabledNoReductions,
+    ForceEnabled
+  };
+}
+
 class ARMTTIImpl : public BasicTTIImplBase<ARMTTIImpl> {
   using BaseT = BasicTTIImplBase<ARMTTIImpl>;
   using TTI = TargetTransformInfo;
@@ -47,13 +57,13 @@ class ARMTTIImpl : public BasicTTIImplBase<ARMTTIImpl> {
   const ARMSubtarget *ST;
   const ARMTargetLowering *TLI;
 
-  // Currently the following features are excluded from InlineFeatureWhitelist.
+  // Currently the following features are excluded from InlineFeaturesAllowed.
   // ModeThumb, FeatureNoARM, ModeSoftFloat, FeatureFP64, FeatureD32
   // Depending on whether they are set or unset, different
   // instructions/registers are available. For example, inlining a callee with
   // -thumb-mode in a caller with +thumb-mode, may cause the assembler to
   // fail if the callee uses ARM only instructions, e.g. in inline asm.
-  const FeatureBitset InlineFeatureWhitelist = {
+  const FeatureBitset InlineFeaturesAllowed = {
       ARM::FeatureVFP2, ARM::FeatureVFP3, ARM::FeatureNEON, ARM::FeatureThumb2,
       ARM::FeatureFP16, ARM::FeatureVFP4, ARM::FeatureFPARMv8,
       ARM::FeatureFullFP16, ARM::FeatureFP16FML, ARM::FeatureHWDivThumb,
@@ -93,11 +103,8 @@ public:
 
   bool enableInterleavedAccessVectorization() { return true; }
 
-  bool shouldFavorBackedgeIndex(const Loop *L) const {
-    if (L->getHeader()->getParent()->hasOptSize())
-      return false;
-    return ST->isMClass() && ST->isThumb2() && L->getNumBlocks() == 1;
-  }
+  bool shouldFavorBackedgeIndex(const Loop *L) const;
+  bool shouldFavorPostInc() const;
 
   /// Floating-point computation using ARMv8 AArch32 Advanced
   /// SIMD instructions remains unchanged from ARMv7. Only AArch64 SIMD
@@ -113,9 +120,10 @@ public:
                             Type *Ty);
 
   using BaseT::getIntImmCost;
-  int getIntImmCost(const APInt &Imm, Type *Ty);
+  int getIntImmCost(const APInt &Imm, Type *Ty, TTI::TargetCostKind CostKind);
 
-  int getIntImmCostInst(unsigned Opcode, unsigned Idx, const APInt &Imm, Type *Ty);
+  int getIntImmCostInst(unsigned Opcode, unsigned Idx, const APInt &Imm,
+                        Type *Ty, TTI::TargetCostKind CostKind);
 
   /// @}
 
@@ -153,31 +161,57 @@ public:
     return ST->getMaxInterleaveFactor();
   }
 
-  bool isLegalMaskedLoad(Type *DataTy, MaybeAlign Alignment);
+  bool isProfitableLSRChainElement(Instruction *I);
+
+  bool isLegalMaskedLoad(Type *DataTy, Align Alignment);
 
-  bool isLegalMaskedStore(Type *DataTy, MaybeAlign Alignment) {
+  bool isLegalMaskedStore(Type *DataTy, Align Alignment) {
     return isLegalMaskedLoad(DataTy, Alignment);
   }
 
-  bool isLegalMaskedGather(Type *Ty, MaybeAlign Alignment);
+  bool isLegalMaskedGather(Type *Ty, Align Alignment);
 
-  bool isLegalMaskedScatter(Type *Ty, MaybeAlign Alignment) { return false; }
+  bool isLegalMaskedScatter(Type *Ty, Align Alignment) {
+    return isLegalMaskedGather(Ty, Alignment);
+  }
 
   int getMemcpyCost(const Instruction *I);
 
-  int getShuffleCost(TTI::ShuffleKind Kind, Type *Tp, int Index, Type *SubTp);
+  int getShuffleCost(TTI::ShuffleKind Kind, VectorType *Tp, int Index,
+                     VectorType *SubTp);
 
   bool useReductionIntrinsic(unsigned Opcode, Type *Ty,
                              TTI::ReductionFlags Flags) const;
 
   bool shouldExpandReduction(const IntrinsicInst *II) const {
-    return false;
+    switch (II->getIntrinsicID()) {
+    case Intrinsic::experimental_vector_reduce_v2_fadd:
+    case Intrinsic::experimental_vector_reduce_v2_fmul:
+      // We don't have legalization support for ordered FP reductions.
+      if (!II->getFastMathFlags().allowReassoc())
+        return true;
+      // Can't legalize reductions with soft floats.
+      return TLI->useSoftFloat() || !TLI->getSubtarget()->hasFPRegs();
+
+    case Intrinsic::experimental_vector_reduce_fmin:
+    case Intrinsic::experimental_vector_reduce_fmax:
+      // Can't legalize reductions with soft floats, and NoNan will create
+      // fminimum which we do not know how to lower.
+      return TLI->useSoftFloat() || !TLI->getSubtarget()->hasFPRegs() ||
+             !II->getFastMathFlags().noNaNs();
+
+    default:
+      // Don't expand anything else, let legalization deal with it.
+      return false;
+    }
   }
 
   int getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src,
+                       TTI::TargetCostKind CostKind,
                        const Instruction *I = nullptr);
 
   int getCmpSelInstrCost(unsigned Opcode, Type *ValTy, Type *CondTy,
+                         TTI::TargetCostKind CostKind,
                          const Instruction *I = nullptr);
 
   int getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index);
@@ -187,6 +221,7 @@ public:
 
   int getArithmeticInstrCost(
       unsigned Opcode, Type *Ty,
+      TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput,
       TTI::OperandValueKind Op1Info = TTI::OK_AnyValue,
       TTI::OperandValueKind Op2Info = TTI::OK_AnyValue,
       TTI::OperandValueProperties Opd1PropInfo = TTI::OP_None,
@@ -195,13 +230,20 @@ public:
       const Instruction *CxtI = nullptr);
 
   int getMemoryOpCost(unsigned Opcode, Type *Src, MaybeAlign Alignment,
-                      unsigned AddressSpace, const Instruction *I = nullptr);
+                      unsigned AddressSpace,
+                      TTI::TargetCostKind CostKind,
+                      const Instruction *I = nullptr);
 
-  int getInterleavedMemoryOpCost(unsigned Opcode, Type *VecTy, unsigned Factor,
-                                 ArrayRef<unsigned> Indices, unsigned Alignment,
-                                 unsigned AddressSpace,
-                                 bool UseMaskForCond = false,
-                                 bool UseMaskForGaps = false);
+  int getInterleavedMemoryOpCost(
+      unsigned Opcode, Type *VecTy, unsigned Factor, ArrayRef<unsigned> Indices,
+      Align Alignment, unsigned AddressSpace,
+      TTI::TargetCostKind CostKind = TTI::TCK_SizeAndLatency,
+      bool UseMaskForCond = false, bool UseMaskForGaps = false);
+
+  unsigned getGatherScatterOpCost(unsigned Opcode, Type *DataTy,
+                                  const Value *Ptr, bool VariableMask,
+                                  Align Alignment, TTI::TargetCostKind CostKind,
+                                  const Instruction *I = nullptr);
 
   bool isLoweredToCall(const Function *F);
   bool isHardwareLoopProfitable(Loop *L, ScalarEvolution &SE,
@@ -217,6 +259,10 @@ public:
   void getUnrollingPreferences(Loop *L, ScalarEvolution &SE,
                                TTI::UnrollingPreferences &UP);
 
+  bool emitGetActiveLaneMask() const;
+
+  void getPeelingPreferences(Loop *L, ScalarEvolution &SE,
+                             TTI::PeelingPreferences &PP);
   bool shouldBuildLookupTablesForConstant(Constant *C) const {
     // In the ROPI and RWPI relocation models we can't have pointers to global
     // variables or functions in constant data, so don't convert switches to