[AArch64] Removed the narrow load merging code in the ld/st optimizer.

[mcrosier]

Summary

This feature has been disabled for some time now due to reports of performance regressions. Since no one has complained and we've moved on to 4.0 I'd like to remove this code permanently.

This should be a NFC for all AArch64 subtargets.

The FeatureMergeNarrowLd used to control merging of both narrow loads and narrow zero stores (which was a bit misleading). I've renamed it to FeatureMergeNarrowSt to be more inline with what's actually happening.

Currently, narrow zero stores are only merged for Kryo and Cortex-A57. I suspect this would benefit other subtargets, but I'll leave that to the respective owners to decide.

Chad

[junbuml]

LGTM. Just minor inline comments.
Thanks Chad for handling this.

lib/Target/AArch64/AArch64.td

64–67

FeatureMergeNarrowZeroSt should be the right name.

lib/Target/AArch64/AArch64LoadStoreOptimizer.cpp

40–41

NumNarrowLoadsPromoted should be removed

542–544

isPromotableZeroStoreOpcode(MachineInstr &MI) is used only in here.
By using isPromotableZeroStoreOpcode(MI.getOpcode()), we can remove :

static bool isPromotableZeroStoreOpcode(MachineInstr &MI) {

return isPromotableZeroStoreOpcode(MI.getOpcode());

}

1491

EnableNarrowStOpt -> EnableNarrowZeroStOpt could be better.

1705–1706

enableNarrowStOpt -> EnableNarrowZeroStOpt

lib/Target/AArch64/AArch64Subtarget.h

74

MergeNarrowStores -> MergeNarrowZeroStores

182

mergeNarrowStores -> mergeNarrowZeroStores

[mcrosier]

Address Jun's comments.

[junbuml]

LGTM. Thanks Chad!

lib/Target/AArch64/AArch64LoadStoreOptimizer.cpp

1450

Now, mergeNarrowInsns () is used only for narrow zero store merge. So, "mergeNarrowZeroStore" seems to be the right name.

lib/Target/AArch64/AArch64.td

def FeatureStrictAlign : SubtargetFeature<"strict-align",
                                           "StrictAlign", "true",
                                           "Disallow all unaligned memory "
                                           "access">;
 
 def FeatureReserveX18 : SubtargetFeature<"reserve-x18", "ReserveX18", "true",
                                          "Reserve X18, making it unavailable "
                                          "as a GPR">;
 
-def FeatureMergeNarrowLd : SubtargetFeature<"merge-narrow-ld",
-                                            "MergeNarrowLoads", "true",
-                                            "Merge narrow load instructions">;
+def FeatureMergeNarrowZeroSt : SubtargetFeature<"merge-narrow-zero-st",
+                                                "MergeNarrowZeroStores", "true",
+                                                "Merge narrow zero store "
+                                                "instructions">;
 
 def FeatureUseAA : SubtargetFeature<"use-aa", "UseAA", "true",
                                     "Use alias analysis during codegen">;
 
 def FeatureBalanceFPOps : SubtargetFeature<"balance-fp-ops", "BalanceFPOps",
     "true",
     "balance mix of odd and even D-registers for fp multiply(-accumulate) ops">;
 

 
 def ProcA57     : SubtargetFeature<"a57", "ARMProcFamily", "CortexA57",
                                    "Cortex-A57 ARM processors", [
                                    FeatureBalanceFPOps,
                                    FeatureCRC,
                                    FeatureCrypto,
                                    FeatureCustomCheapAsMoveHandling,
                                    FeatureFPARMv8,
-                                   FeatureMergeNarrowLd,
+                                   FeatureMergeNarrowZeroSt,
                                    FeatureNEON,
                                    FeaturePerfMon,
                                    FeaturePostRAScheduler,
                                    FeaturePredictableSelectIsExpensive
                                    ]>;
 
 def ProcA72     : SubtargetFeature<"a72", "ARMProcFamily", "CortexA72",
                                    "Cortex-A72 ARM processors", [
def ProcExynosM2 : SubtargetFeature<"exynosm2", "ARMProcFamily", "ExynosM1",
                                      FeatureZCZeroing]>;
 
 def ProcKryo    : SubtargetFeature<"kryo", "ARMProcFamily", "Kryo",
                                    "Qualcomm Kryo processors", [
                                    FeatureCRC,
                                    FeatureCrypto,
                                    FeatureCustomCheapAsMoveHandling,
                                    FeatureFPARMv8,
-                                   FeatureMergeNarrowLd,
+                                   FeatureMergeNarrowZeroSt,
                                    FeatureNEON,
                                    FeaturePerfMon,
                                    FeaturePostRAScheduler,
                                    FeaturePredictableSelectIsExpensive,
                                    FeatureZCZeroing
                                    ]>;
 
 def ProcVulcan  : SubtargetFeature<"vulcan", "ARMProcFamily", "Vulcan",

lib/Target/AArch64/AArch64LoadStoreOptimizer.cpp

 using namespace llvm;
 
 #define DEBUG_TYPE "aarch64-ldst-opt"
 
 STATISTIC(NumPairCreated, "Number of load/store pair instructions generated");
 STATISTIC(NumPostFolded, "Number of post-index updates folded");
 STATISTIC(NumPreFolded, "Number of pre-index updates folded");
 STATISTIC(NumUnscaledPairCreated,
           "Number of load/store from unscaled generated");
-STATISTIC(NumNarrowLoadsPromoted, "Number of narrow loads promoted");
 STATISTIC(NumZeroStoresPromoted, "Number of narrow zero stores promoted");
 STATISTIC(NumLoadsFromStoresPromoted, "Number of loads from stores promoted");
 
 // The LdStLimit limits how far we search for load/store pairs.
 static cl::opt<unsigned> LdStLimit("aarch64-load-store-scan-limit",
                                    cl::init(20), cl::Hidden);
 
 // The UpdateLimit limits how far we search for update instructions when we form
 // pre-/post-index instructions.
 static cl::opt<unsigned> UpdateLimit("aarch64-update-scan-limit", cl::init(100),
                                      cl::Hidden);
 
-static cl::opt<bool> EnableNarrowLdMerge("enable-narrow-ld-merge", cl::Hidden,
-                                         cl::init(false),
-                                         cl::desc("Enable narrow load merge"));
-
 #define AARCH64_LOAD_STORE_OPT_NAME "AArch64 load / store optimization pass"
 
 namespace {
 
 typedef struct LdStPairFlags {
   // If a matching instruction is found, MergeForward is set to true if the
   // merge is to remove the first instruction and replace the second with
   // a pair-wise insn, and false if the reverse is true.
struct AArch64LoadStoreOpt : public MachineFunctionPass {
   bool isMatchingUpdateInsn(MachineInstr &MemMI, MachineInstr &MI,
                             unsigned BaseReg, int Offset);
 
   // Merge a pre- or post-index base register update into a ld/st instruction.
   MachineBasicBlock::iterator
   mergeUpdateInsn(MachineBasicBlock::iterator I,
                   MachineBasicBlock::iterator Update, bool IsPreIdx);
 
-  // Find and merge foldable ldr/str instructions.
-  bool tryToMergeLdStInst(MachineBasicBlock::iterator &MBBI);
+  // Find and merge zero store instructions.
+  bool tryToMergeZeroStInst(MachineBasicBlock::iterator &MBBI);
 
   // Find and pair ldr/str instructions.
   bool tryToPairLdStInst(MachineBasicBlock::iterator &MBBI);
 
   // Find and promote load instructions which read directly from store.
   bool tryToPromoteLoadFromStore(MachineBasicBlock::iterator &MBBI);
 
-  bool optimizeBlock(MachineBasicBlock &MBB, bool enableNarrowLdOpt);
+  bool optimizeBlock(MachineBasicBlock &MBB, bool EnableNarrowZeroStOpt);
 
   bool runOnMachineFunction(MachineFunction &Fn) override;
 
   MachineFunctionProperties getRequiredProperties() const override {
     return MachineFunctionProperties().set(
         MachineFunctionProperties::Property::NoVRegs);
   }
 
   StringRef getPassName() const override { return AARCH64_LOAD_STORE_OPT_NAME; }
 };
 char AArch64LoadStoreOpt::ID = 0;
 } // namespace
 
 INITIALIZE_PASS(AArch64LoadStoreOpt, "aarch64-ldst-opt",
                 AARCH64_LOAD_STORE_OPT_NAME, false, false)
 
-static unsigned getBitExtrOpcode(MachineInstr &MI) {
-  switch (MI.getOpcode()) {
-  default:
-    llvm_unreachable("Unexpected opcode.");
-  case AArch64::LDRBBui:
-  case AArch64::LDURBBi:
-  case AArch64::LDRHHui:
-  case AArch64::LDURHHi:
-    return AArch64::UBFMWri;
-  case AArch64::LDRSBWui:
-  case AArch64::LDURSBWi:
-  case AArch64::LDRSHWui:
-  case AArch64::LDURSHWi:
-    return AArch64::SBFMWri;
-  }
-}
-
 static bool isNarrowStore(unsigned Opc) {
   switch (Opc) {
   default:
     return false;
   case AArch64::STRBBui:
   case AArch64::STURBBi:
   case AArch64::STRHHui:
   case AArch64::STURHHi:
     return true;
   }
 }
 
-static bool isNarrowLoad(unsigned Opc) {
-  switch (Opc) {
-  default:
-    return false;
-  case AArch64::LDRHHui:
-  case AArch64::LDURHHi:
-  case AArch64::LDRBBui:
-  case AArch64::LDURBBi:
-  case AArch64::LDRSHWui:
-  case AArch64::LDURSHWi:
-  case AArch64::LDRSBWui:
-  case AArch64::LDURSBWi:
-    return true;
-  }
-}
-
-static bool isNarrowLoad(MachineInstr &MI) {
-  return isNarrowLoad(MI.getOpcode());
-}
-
-static bool isNarrowLoadOrStore(unsigned Opc) {
-  return isNarrowLoad(Opc) || isNarrowStore(Opc);
-}
-
 // Scaling factor for unscaled load or store.
 static int getMemScale(MachineInstr &MI) {
   switch (MI.getOpcode()) {
   default:
     llvm_unreachable("Opcode has unknown scale!");
   case AArch64::LDRBBui:
   case AArch64::LDURBBi:
   case AArch64::LDRSBWui:
static unsigned getMatchingNonSExtOpcode(unsigned Opc,
   case AArch64::LDRWui:
   case AArch64::LDURWi:
   case AArch64::LDRXui:
   case AArch64::LDURXi:
   case AArch64::STRSui:
   case AArch64::STURSi:
   case AArch64::LDRSui:
   case AArch64::LDURSi:
-  case AArch64::LDRHHui:
-  case AArch64::LDURHHi:
-  case AArch64::LDRBBui:
-  case AArch64::LDURBBi:
     return Opc;
   case AArch64::LDRSWui:
     return AArch64::LDRWui;
   case AArch64::LDURSWi:
     return AArch64::LDURWi;
-  case AArch64::LDRSBWui:
-    return AArch64::LDRBBui;
-  case AArch64::LDRSHWui:
-    return AArch64::LDRHHui;
-  case AArch64::LDURSBWi:
-    return AArch64::LDURBBi;
-  case AArch64::LDURSHWi:
-    return AArch64::LDURHHi;
   }
 }
 
 static unsigned getMatchingWideOpcode(unsigned Opc) {
   switch (Opc) {
   default:
     llvm_unreachable("Opcode has no wide equivalent!");
   case AArch64::STRBBui:
     return AArch64::STRHHui;
   case AArch64::STRHHui:
     return AArch64::STRWui;
   case AArch64::STURBBi:
     return AArch64::STURHHi;
   case AArch64::STURHHi:
     return AArch64::STURWi;
   case AArch64::STURWi:
     return AArch64::STURXi;
   case AArch64::STRWui:
     return AArch64::STRXui;
-  case AArch64::LDRHHui:
-  case AArch64::LDRSHWui:
-    return AArch64::LDRWui;
-  case AArch64::LDURHHi:
-  case AArch64::LDURSHWi:
-    return AArch64::LDURWi;
-  case AArch64::LDRBBui:
-  case AArch64::LDRSBWui:
-    return AArch64::LDRHHui;
-  case AArch64::LDURBBi:
-  case AArch64::LDURSBWi:
-    return AArch64::LDURHHi;
   }
 }
 
 static unsigned getMatchingPairOpcode(unsigned Opc) {
   switch (Opc) {
   default:
     llvm_unreachable("Opcode has no pairwise equivalent!");
   case AArch64::STRSui:
int UnscaledStOffset = TII->isUnscaledLdSt(StoreInst)
                              : getLdStOffsetOp(StoreInst).getImm() * StoreSize;
   int UnscaledLdOffset = TII->isUnscaledLdSt(LoadInst)
                              ? getLdStOffsetOp(LoadInst).getImm()
                              : getLdStOffsetOp(LoadInst).getImm() * LoadSize;
   return (UnscaledStOffset <= UnscaledLdOffset) &&
          (UnscaledLdOffset + LoadSize <= (UnscaledStOffset + StoreSize));
 }
 
-static bool isPromotableZeroStoreOpcode(unsigned Opc) {
-  return isNarrowStore(Opc) || Opc == AArch64::STRWui || Opc == AArch64::STURWi;
-}
-
-static bool isPromotableZeroStoreOpcode(MachineInstr &MI) {
-  return isPromotableZeroStoreOpcode(MI.getOpcode());
-}
-
 static bool isPromotableZeroStoreInst(MachineInstr &MI) {
-  return (isPromotableZeroStoreOpcode(MI)) &&
+  unsigned Opc = MI.getOpcode();
+  return (Opc == AArch64::STRWui || Opc == AArch64::STURWi ||
+          isNarrowStore(Opc)) &&
          getLdStRegOp(MI).getReg() == AArch64::WZR;
 }
 
 MachineBasicBlock::iterator
 AArch64LoadStoreOpt::mergeNarrowInsns(MachineBasicBlock::iterator I,
                                       MachineBasicBlock::iterator MergeMI,
                                       const LdStPairFlags &Flags) {
+  assert(isPromotableZeroStoreInst(*I) && isPromotableZeroStoreInst(*MergeMI) &&
+         "Expected promotable zero stores.");
+
   MachineBasicBlock::iterator NextI = I;
   ++NextI;
   // If NextI is the second of the two instructions to be merged, we need
   // to skip one further. Either way we merge will invalidate the iterator,
   // and we don't need to scan the new instruction, as it's a pairwise
   // instruction, which we're not considering for further action anyway.
   if (NextI == MergeMI)
     ++NextI;
AArch64LoadStoreOpt::mergeNarrowInsns(MachineBasicBlock::iterator I,
 
   int OffsetImm = getLdStOffsetOp(*RtMI).getImm();
   // Change the scaled offset from small to large type.
   if (IsScaled) {
     assert(((OffsetImm & 1) == 0) && "Unexpected offset to merge");
     OffsetImm /= 2;
   }
 
+  // Construct the new instruction.
   DebugLoc DL = I->getDebugLoc();
   MachineBasicBlock *MBB = I->getParent();
-  if (isNarrowLoad(Opc)) {
-    MachineInstr *RtNewDest = &*(MergeForward ? I : MergeMI);
-    // When merging small (< 32 bit) loads for big-endian targets, the order of
-    // the component parts gets swapped.
-    if (!Subtarget->isLittleEndian())
-      std::swap(RtMI, Rt2MI);
-    // Construct the new load instruction.
-    MachineInstr *NewMemMI, *BitExtMI1, *BitExtMI2;
-    NewMemMI =
-        BuildMI(*MBB, InsertionPoint, DL, TII->get(getMatchingWideOpcode(Opc)))
-            .addOperand(getLdStRegOp(*RtNewDest))
-            .addOperand(BaseRegOp)
-            .addImm(OffsetImm)
-            .setMemRefs(I->mergeMemRefsWith(*MergeMI));
-    (void)NewMemMI;
-
-    DEBUG(
-        dbgs()
-        << "Creating the new load and extract. Replacing instructions:\n    ");
-    DEBUG(I->print(dbgs()));
-    DEBUG(dbgs() << "    ");
-    DEBUG(MergeMI->print(dbgs()));
-    DEBUG(dbgs() << "  with instructions:\n    ");
-    DEBUG((NewMemMI)->print(dbgs()));
-
-    int Width = getMemScale(*I) == 1 ? 8 : 16;
-    int LSBLow = 0;
-    int LSBHigh = Width;
-    int ImmsLow = LSBLow + Width - 1;
-    int ImmsHigh = LSBHigh + Width - 1;
-    MachineInstr *ExtDestMI = &*(MergeForward ? MergeMI : I);
-    if ((ExtDestMI == Rt2MI) == Subtarget->isLittleEndian()) {
-      // Create the bitfield extract for high bits.
-      BitExtMI1 =
-          BuildMI(*MBB, InsertionPoint, DL, TII->get(getBitExtrOpcode(*Rt2MI)))
-              .addOperand(getLdStRegOp(*Rt2MI))
-              .addReg(getLdStRegOp(*RtNewDest).getReg())
-              .addImm(LSBHigh)
-              .addImm(ImmsHigh);
-      // Create the bitfield extract for low bits.
-      if (RtMI->getOpcode() == getMatchingNonSExtOpcode(RtMI->getOpcode())) {
-        // For unsigned, prefer to use AND for low bits.
-        BitExtMI2 = BuildMI(*MBB, InsertionPoint, DL, TII->get(AArch64::ANDWri))
-                        .addOperand(getLdStRegOp(*RtMI))
-                        .addReg(getLdStRegOp(*RtNewDest).getReg())
-                        .addImm(ImmsLow);
-      } else {
-        BitExtMI2 =
-            BuildMI(*MBB, InsertionPoint, DL, TII->get(getBitExtrOpcode(*RtMI)))
-                .addOperand(getLdStRegOp(*RtMI))
-                .addReg(getLdStRegOp(*RtNewDest).getReg())
-                .addImm(LSBLow)
-                .addImm(ImmsLow);
-      }
-    } else {
-      // Create the bitfield extract for low bits.
-      if (RtMI->getOpcode() == getMatchingNonSExtOpcode(RtMI->getOpcode())) {
-        // For unsigned, prefer to use AND for low bits.
-        BitExtMI1 = BuildMI(*MBB, InsertionPoint, DL, TII->get(AArch64::ANDWri))
-                        .addOperand(getLdStRegOp(*RtMI))
-                        .addReg(getLdStRegOp(*RtNewDest).getReg())
-                        .addImm(ImmsLow);
-      } else {
-        BitExtMI1 =
-            BuildMI(*MBB, InsertionPoint, DL, TII->get(getBitExtrOpcode(*RtMI)))
-                .addOperand(getLdStRegOp(*RtMI))
-                .addReg(getLdStRegOp(*RtNewDest).getReg())
-                .addImm(LSBLow)
-                .addImm(ImmsLow);
-      }
-
-      // Create the bitfield extract for high bits.
-      BitExtMI2 =
-          BuildMI(*MBB, InsertionPoint, DL, TII->get(getBitExtrOpcode(*Rt2MI)))
-              .addOperand(getLdStRegOp(*Rt2MI))
-              .addReg(getLdStRegOp(*RtNewDest).getReg())
-              .addImm(LSBHigh)
-              .addImm(ImmsHigh);
-    }
-    (void)BitExtMI1;
-    (void)BitExtMI2;
-
-    DEBUG(dbgs() << "    ");
-    DEBUG((BitExtMI1)->print(dbgs()));
-    DEBUG(dbgs() << "    ");
-    DEBUG((BitExtMI2)->print(dbgs()));
-    DEBUG(dbgs() << "\n");
-
-    // Erase the old instructions.
-    I->eraseFromParent();
-    MergeMI->eraseFromParent();
-    return NextI;
-  }
-  assert(isPromotableZeroStoreInst(*I) && isPromotableZeroStoreInst(*MergeMI) &&
-         "Expected promotable zero store");
-
-  // Construct the new instruction.
   MachineInstrBuilder MIB;
   MIB = BuildMI(*MBB, InsertionPoint, DL, TII->get(getMatchingWideOpcode(Opc)))
             .addReg(isNarrowStore(Opc) ? AArch64::WZR : AArch64::XZR)
             .addOperand(BaseRegOp)
             .addImm(OffsetImm)
             .setMemRefs(I->mergeMemRefsWith(*MergeMI));
   (void)MIB;
 
-  DEBUG(dbgs() << "Creating wider load/store. Replacing instructions:\n    ");
+  DEBUG(dbgs() << "Creating wider store. Replacing instructions:\n    ");
   DEBUG(I->print(dbgs()));
   DEBUG(dbgs() << "    ");
   DEBUG(MergeMI->print(dbgs()));
   DEBUG(dbgs() << "  with instruction:\n    ");
   DEBUG(((MachineInstr *)MIB)->print(dbgs()));
   DEBUG(dbgs() << "\n");
 
   // Erase the old instructions.
static bool areCandidatesToMergeOrPair(MachineInstr &FirstMI, MachineInstr &MI,
   assert(IsValidLdStrOpc &&
          "Given Opc should be a Load or Store with an immediate");
   // OpcA will be the first instruction in the pair.
   if (NonSExtOpc == getMatchingNonSExtOpcode(OpcB, &PairIsValidLdStrOpc)) {
     Flags.setSExtIdx(NonSExtOpc == (unsigned)OpcA ? 1 : 0);
     return true;
   }
 
-  // If the second instruction isn't even a load/store, bail out.
+  // If the second instruction isn't even a mergable/pairable load/store, bail
+  // out.
   if (!PairIsValidLdStrOpc)
     return false;
 
-  // FIXME: We don't support merging narrow loads/stores with mixed
-  // scaled/unscaled offsets.
-  if (isNarrowLoadOrStore(OpcA) || isNarrowLoadOrStore(OpcB))
+  // FIXME: We don't support merging narrow stores with mixed scaled/unscaled
+  // offsets.
+  if (isNarrowStore(OpcA) || isNarrowStore(OpcB))
     return false;
 
   // Try to match an unscaled load/store with a scaled load/store.
   return TII->isUnscaledLdSt(OpcA) != TII->isUnscaledLdSt(OpcB) &&
          getMatchingPairOpcode(OpcA) == getMatchingPairOpcode(OpcB);
 
   // FIXME: Can we also match a mixed sext/zext unscaled/scaled pair?
 }
if (findMatchingStore(MBBI, LdStLimit, StoreI)) {
     // pain, so we let the merge routine tell us what the next instruction
     // is after it's done mucking about.
     MBBI = promoteLoadFromStore(MBBI, StoreI);
     return true;
   }
   return false;
 }
 
-// Find narrow loads that can be converted into a single wider load with
-// bitfield extract instructions.  Also merge adjacent zero stores into a wider
-// store.
-bool AArch64LoadStoreOpt::tryToMergeLdStInst(
+// Merge adjacent zero stores into a wider store.
+bool AArch64LoadStoreOpt::tryToMergeZeroStInst(
     MachineBasicBlock::iterator &MBBI) {
-  assert((isNarrowLoad(*MBBI) || isPromotableZeroStoreOpcode(*MBBI)) &&
-         "Expected narrow op.");
+  assert(isPromotableZeroStoreInst(*MBBI) && "Expected narrow store.");
   MachineInstr &MI = *MBBI;
   MachineBasicBlock::iterator E = MI.getParent()->end();
 
   if (!TII->isCandidateToMergeOrPair(MI))
     return false;
 
-  // For promotable zero stores, the stored value should be WZR.
-  if (isPromotableZeroStoreOpcode(MI) &&
-      getLdStRegOp(MI).getReg() != AArch64::WZR)
-    return false;
-
   // Look ahead up to LdStLimit instructions for a mergable instruction.
   LdStPairFlags Flags;
   MachineBasicBlock::iterator MergeMI =
       findMatchingInsn(MBBI, Flags, LdStLimit, /* FindNarrowMerge = */ true);
   if (MergeMI != E) {
-    if (isNarrowLoad(MI)) {
-      ++NumNarrowLoadsPromoted;
-    } else if (isPromotableZeroStoreInst(MI)) {
-      ++NumZeroStoresPromoted;
-    }
+
     // Keeping the iterator straight is a pain, so we let the merge routine tell
     // us what the next instruction is after it's done mucking about.
     MBBI = mergeNarrowInsns(MBBI, MergeMI, Flags);
     return true;
   }
   return false;
 }
 
 // Find loads and stores that can be merged into a single load or store pair
 // instruction.
 bool AArch64LoadStoreOpt::tryToPairLdStInst(MachineBasicBlock::iterator &MBBI) {
if (Paired != E) {
     // us what the next instruction is after it's done mucking about.
     MBBI = mergePairedInsns(MBBI, Paired, Flags);
     return true;
   }
   return false;
 }
 
 bool AArch64LoadStoreOpt::optimizeBlock(MachineBasicBlock &MBB,
-                                        bool enableNarrowLdOpt) {
+                                        bool EnableNarrowZeroStOpt) {
   bool Modified = false;
   // Four tranformations to do here:
   // 1) Find loads that directly read from stores and promote them by
   //    replacing with mov instructions. If the store is wider than the load,
   //    the load will be replaced with a bitfield extract.
   //      e.g.,
   //        str w1, [x0, #4]
   //        ldrh w2, [x0, #6]
case AArch64::LDURXi: {
         Modified = true;
         break;
       }
       ++MBBI;
       break;
     }
     }
   }
-  // 2) Find narrow loads that can be converted into a single wider load
-  //    with bitfield extract instructions.
-  //      e.g.,
-  //        ldrh w0, [x2]
-  //        ldrh w1, [x2, #2]
-  //        ; becomes
-  //        ldr w0, [x2]
-  //        ubfx w1, w0, #16, #16
-  //        and w0, w0, #ffff
-  //
-  //    Also merge adjacent zero stores into a wider store.
+  // 2) Merge adjacent zero stores into a wider store.
   //      e.g.,
   //        strh wzr, [x0]
   //        strh wzr, [x0, #2]
   //        ; becomes
   //        str wzr, [x0]
+  //      e.g.,
+  //        str wzr, [x0]
+  //        str wzr, [x0, #4]
+  //        ; becomes
+  //        str xzr, [x0]
   for (MachineBasicBlock::iterator MBBI = MBB.begin(), E = MBB.end();
-       enableNarrowLdOpt && MBBI != E;) {
-    MachineInstr &MI = *MBBI;
-    unsigned Opc = MI.getOpcode();
-    if (isPromotableZeroStoreOpcode(Opc) ||
-        (EnableNarrowLdMerge && isNarrowLoad(Opc))) {
-      if (tryToMergeLdStInst(MBBI)) {
+       EnableNarrowZeroStOpt && MBBI != E;) {
+    if (isPromotableZeroStoreInst(*MBBI)) {
+      if (tryToMergeZeroStInst(MBBI)) {
         Modified = true;
       } else
         ++MBBI;
     } else
       ++MBBI;
   }
 
   // 3) Find loads and stores that can be merged into a single load or store
bool AArch64LoadStoreOpt::runOnMachineFunction(MachineFunction &Fn) {
 
   // Resize the modified and used register bitfield trackers.  We do this once
   // per function and then clear the bitfield each time we optimize a load or
   // store.
   ModifiedRegs.resize(TRI->getNumRegs());
   UsedRegs.resize(TRI->getNumRegs());
 
   bool Modified = false;
-  bool enableNarrowLdOpt =
-    Subtarget->mergeNarrowLoads() && !Subtarget->requiresStrictAlign();
+  bool enableNarrowZeroStOpt =
+      Subtarget->mergeNarrowStores() && !Subtarget->requiresStrictAlign();
   for (auto &MBB : Fn)
-    Modified |= optimizeBlock(MBB, enableNarrowLdOpt);
+    Modified |= optimizeBlock(MBB, enableNarrowZeroStOpt);
 
   return Modified;
 }
 
 // FIXME: Do we need/want a pre-alloc pass like ARM has to try to keep
 // loads and stores near one another?
 
 // FIXME: When pairing store instructions it's very possible for this pass to

lib/Target/AArch64/AArch64Subtarget.h

protected:
   // HasZeroCycleRegMove - Has zero-cycle register mov instructions.
   bool HasZeroCycleRegMove = false;
 
   // HasZeroCycleZeroing - Has zero-cycle zeroing instructions.
   bool HasZeroCycleZeroing = false;
 
   // StrictAlign - Disallow unaligned memory accesses.
   bool StrictAlign = false;
-  bool MergeNarrowLoads = false;
+  bool MergeNarrowZeroStores = false;
   bool UseAA = false;
   bool PredictableSelectIsExpensive = false;
   bool BalanceFPOps = false;
   bool CustomAsCheapAsMove = false;
   bool UsePostRAScheduler = false;
   bool Misaligned128StoreIsSlow = false;
   bool AvoidQuadLdStPairs = false;
   bool UseAlternateSExtLoadCVTF32Pattern = false;
public:
   bool requiresStrictAlign() const { return StrictAlign; }
 
   bool isX18Reserved() const { return ReserveX18; }
   bool hasFPARMv8() const { return HasFPARMv8; }
   bool hasNEON() const { return HasNEON; }
   bool hasCrypto() const { return HasCrypto; }
   bool hasCRC() const { return HasCRC; }
   bool hasRAS() const { return HasRAS; }
-  bool mergeNarrowLoads() const { return MergeNarrowLoads; }
+  bool mergeNarrowStores() const { return MergeNarrowZeroStores; }
   bool balanceFPOps() const { return BalanceFPOps; }
   bool predictableSelectIsExpensive() const {
     return PredictableSelectIsExpensive;
   }
   bool hasCustomCheapAsMoveHandling() const { return CustomAsCheapAsMove; }
   bool isMisaligned128StoreSlow() const { return Misaligned128StoreIsSlow; }
   bool avoidQuadLdStPairs() const { return AvoidQuadLdStPairs; }
   bool useAlternateSExtLoadCVTF32Pattern() const {

test/CodeGen/AArch64/arm64-narrow-ldst-merge.ll

-; RUN: llc < %s -mtriple aarch64--none-eabi -mcpu=cortex-a57 -verify-machineinstrs -enable-narrow-ld-merge=true | FileCheck %s --check-prefix=CHECK --check-prefix=LE
-; RUN: llc < %s -mtriple aarch64_be--none-eabi -mcpu=cortex-a57 -verify-machineinstrs -enable-narrow-ld-merge=true | FileCheck %s --check-prefix=CHECK --check-prefix=BE
-; RUN: llc < %s -mtriple aarch64--none-eabi -mcpu=kryo -verify-machineinstrs -enable-narrow-ld-merge=true | FileCheck %s --check-prefix=CHECK --check-prefix=LE
-
-; CHECK-LABEL: Ldrh_merge
-; CHECK-NOT: ldrh
-; CHECK: ldr [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xffff
-; CHECK-DAG: lsr [[HI_PART:w[0-9]+]], [[NEW_DEST]], #16
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i16 @Ldrh_merge(i16* nocapture readonly %p) {
-  %1 = load i16, i16* %p, align 2
-  %arrayidx2 = getelementptr inbounds i16, i16* %p, i64 1
-  %2 = load i16, i16* %arrayidx2, align 2
-  %add = sub nuw nsw i16 %1, %2
-  ret i16 %add
-}
-
-; CHECK-LABEL: Ldurh_merge
-; CHECK-NOT: ldurh
-; CHECK: ldur [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xffff
-; CHECK-DAG: lsr  [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i16 @Ldurh_merge(i16* nocapture readonly %p)  {
-entry:
-  %arrayidx = getelementptr inbounds i16, i16* %p, i64 -2
-  %0 = load i16, i16* %arrayidx
-  %arrayidx3 = getelementptr inbounds i16, i16* %p, i64 -1
-  %1 = load i16, i16* %arrayidx3
-  %add = sub nuw nsw i16 %0, %1
-  ret i16 %add
-}
-
-; CHECK-LABEL: Ldrh_4_merge
-; CHECK-NOT: ldrh
-; CHECK: ldp [[WORD1:w[0-9]+]], [[WORD2:w[0-9]+]], [x0]
-; CHECK-DAG: and [[WORD1LO:w[0-9]+]], [[WORD1]], #0xffff
-; CHECK-DAG: lsr [[WORD1HI:w[0-9]+]], [[WORD1]], #16
-; CHECK-DAG: and [[WORD2LO:w[0-9]+]], [[WORD2]], #0xffff
-; CHECK-DAG: lsr [[WORD2HI:w[0-9]+]], [[WORD2]], #16
-; LE-DAG: sub [[TEMP1:w[0-9]+]], [[WORD1HI]], [[WORD1LO]]
-; BE-DAG: sub [[TEMP1:w[0-9]+]], [[WORD1LO]], [[WORD1HI]]
-; LE: udiv [[TEMP2:w[0-9]+]], [[TEMP1]], [[WORD2LO]]
-; BE: udiv [[TEMP2:w[0-9]+]], [[TEMP1]], [[WORD2HI]]
-; LE: sub w0, [[TEMP2]], [[WORD2HI]]
-; BE: sub w0, [[TEMP2]], [[WORD2LO]]
-define i16 @Ldrh_4_merge(i16* nocapture readonly %P) {
-  %arrayidx = getelementptr inbounds i16, i16* %P, i64 0
-  %l0 = load i16, i16* %arrayidx
-  %arrayidx2 = getelementptr inbounds i16, i16* %P, i64 1
-  %l1 = load i16, i16* %arrayidx2
-  %arrayidx7 = getelementptr inbounds i16, i16* %P, i64 2
-  %l2 = load i16, i16* %arrayidx7
-  %arrayidx12 = getelementptr inbounds i16, i16* %P, i64 3
-  %l3 = load i16, i16* %arrayidx12
-  %add4 = sub nuw nsw i16 %l1, %l0
-  %add9 = udiv i16 %add4, %l2
-  %add14 = sub nuw nsw i16 %add9, %l3
-  ret i16 %add14
-}
-
-; CHECK-LABEL: Ldrsh_merge
-; CHECK: ldr [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: asr [[LO_PART:w[0-9]+]], [[NEW_DEST]], #16
-; CHECK-DAG: sxth [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-
-define i32 @Ldrsh_merge(i16* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i16, i16* %p, i64 4
-  %tmp = load i16, i16* %add.ptr0
-  %add.ptr = getelementptr inbounds i16, i16* %p, i64 5
-  %tmp1 = load i16, i16* %add.ptr
-  %sexttmp = sext i16 %tmp to i32
-  %sexttmp1 = sext i16 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp1, %sexttmp
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldrsh_zsext_merge
-; CHECK: ldr [[NEW_DEST:w[0-9]+]]
-; LE-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xffff
-; LE-DAG: asr [[HI_PART:w[0-9]+]], [[NEW_DEST]], #16
-; BE-DAG: sxth [[LO_PART:w[0-9]+]], [[NEW_DEST]]
-; BE-DAG: lsr [[HI_PART:w[0-9]+]], [[NEW_DEST]], #16
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldrsh_zsext_merge(i16* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i16, i16* %p, i64 4
-  %tmp = load i16, i16* %add.ptr0
-  %add.ptr = getelementptr inbounds i16, i16* %p, i64 5
-  %tmp1 = load i16, i16* %add.ptr
-  %sexttmp = zext i16 %tmp to i32
-  %sexttmp1 = sext i16 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldrsh_szext_merge
-; CHECK: ldr [[NEW_DEST:w[0-9]+]]
-; LE-DAG: sxth [[LO_PART:w[0-9]+]], [[NEW_DEST]]
-; LE-DAG: lsr [[HI_PART:w[0-9]+]], [[NEW_DEST]], #16
-; BE-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xffff
-; BE-DAG: asr [[HI_PART:w[0-9]+]], [[NEW_DEST]], #16
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldrsh_szext_merge(i16* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i16, i16* %p, i64 4
-  %tmp = load i16, i16* %add.ptr0
-  %add.ptr = getelementptr inbounds i16, i16* %p, i64 5
-  %tmp1 = load i16, i16* %add.ptr
-  %sexttmp = sext i16 %tmp to i32
-  %sexttmp1 = zext i16 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldrb_merge
-; CHECK: ldrh [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xff
-; CHECK-DAG: ubfx [[HI_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldrb_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 2
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 3
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = zext i8 %tmp to i32
-  %sexttmp1 = zext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldrsb_merge
-; CHECK: ldrh [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: sxtb [[LO_PART:w[0-9]+]], [[NEW_DEST]]
-; CHECK-DAG: sbfx [[HI_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldrsb_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 2
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 3
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = sext i8 %tmp to i32
-  %sexttmp1 = sext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldrsb_zsext_merge
-; CHECK: ldrh [[NEW_DEST:w[0-9]+]]
-; LE-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xff
-; LE-DAG: sbfx [[HI_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; BE-DAG: sxtb [[LO_PART:w[0-9]+]], [[NEW_DEST]]
-; BE-DAG: ubfx [[HI_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldrsb_zsext_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 2
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 3
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = zext i8 %tmp to i32
-  %sexttmp1 = sext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldrsb_szext_merge
-; CHECK: ldrh [[NEW_DEST:w[0-9]+]]
-; LE-DAG: sxtb [[LO_PART:w[0-9]+]], [[NEW_DEST]]
-; LE-DAG: ubfx [[HI_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; BE-DAG: and [[LO_PART:w[0-9]+]], [[NEW_DEST]], #0xff
-; BE-DAG: sbfx [[HI_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldrsb_szext_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 2
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 3
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = sext i8 %tmp to i32
-  %sexttmp1 = zext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldursh_merge
-; CHECK: ldur [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: asr  [[LO_PART:w[0-9]+]], [[NEW_DEST]], #16
-; CHECK-DAG: sxth [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldursh_merge(i16* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i16, i16* %p, i64 -1
-  %tmp = load i16, i16* %add.ptr0
-  %add.ptr = getelementptr inbounds i16, i16* %p, i64 -2
-  %tmp1 = load i16, i16* %add.ptr
-  %sexttmp = sext i16 %tmp to i32
-  %sexttmp1 = sext i16 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldursh_zsext_merge
-; CHECK: ldur [[NEW_DEST:w[0-9]+]]
-; LE-DAG: lsr  [[LO_PART:w[0-9]+]], [[NEW_DEST]], #16
-; LE-DAG: sxth [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; BE-DAG: asr  [[LO_PART:w[0-9]+]], [[NEW_DEST]], #16
-; BE-DAG: and [[HI_PART:w[0-9]+]], [[NEW_DEST]], #0xffff
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldursh_zsext_merge(i16* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i16, i16* %p, i64 -1
-  %tmp = load i16, i16* %add.ptr0
-  %add.ptr = getelementptr inbounds i16, i16* %p, i64 -2
-  %tmp1 = load i16, i16* %add.ptr
-  %sexttmp = zext i16 %tmp to i32
-  %sexttmp1 = sext i16 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldursh_szext_merge
-; CHECK: ldur [[NEW_DEST:w[0-9]+]]
-; LE-DAG: asr  [[LO_PART:w[0-9]+]], [[NEW_DEST]], #16
-; LE-DAG: and [[HI_PART:w[0-9]+]], [[NEW_DEST]], #0xffff
-; BE-DAG: lsr  [[LO_PART:w[0-9]+]], [[NEW_DEST]], #16
-; BE-DAG: sxth [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldursh_szext_merge(i16* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i16, i16* %p, i64 -1
-  %tmp = load i16, i16* %add.ptr0
-  %add.ptr = getelementptr inbounds i16, i16* %p, i64 -2
-  %tmp1 = load i16, i16* %add.ptr
-  %sexttmp = sext i16 %tmp to i32
-  %sexttmp1 = zext i16 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldurb_merge
-; CHECK: ldurh [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: ubfx  [[LO_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; CHECK-DAG: and [[HI_PART:w[0-9]+]], [[NEW_DEST]], #0xff
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldurb_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 -1
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 -2
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = zext i8 %tmp to i32
-  %sexttmp1 = zext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldursb_merge
-; CHECK: ldurh [[NEW_DEST:w[0-9]+]]
-; CHECK-DAG: sbfx [[LO_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; CHECK-DAG: sxtb [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldursb_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 -1
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 -2
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = sext i8 %tmp to i32
-  %sexttmp1 = sext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldursb_zsext_merge
-; CHECK: ldurh [[NEW_DEST:w[0-9]+]]
-; LE-DAG: ubfx [[LO_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; LE-DAG: sxtb [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; BE-DAG: sbfx [[LO_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; BE-DAG: and [[HI_PART:w[0-9]+]], [[NEW_DEST]], #0xff
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldursb_zsext_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 -1
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 -2
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = zext i8 %tmp to i32
-  %sexttmp1 = sext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
-
-; CHECK-LABEL: Ldursb_szext_merge
-; CHECK: ldurh [[NEW_DEST:w[0-9]+]]
-; LE-DAG: sbfx [[LO_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; LE-DAG: and [[HI_PART:w[0-9]+]], [[NEW_DEST]], #0xff
-; BE-DAG: ubfx [[LO_PART:w[0-9]+]], [[NEW_DEST]], #8, #8
-; BE-DAG: sxtb [[HI_PART:w[0-9]+]], [[NEW_DEST]]
-; LE: sub {{w[0-9]+}}, [[LO_PART]], [[HI_PART]]
-; BE: sub {{w[0-9]+}}, [[HI_PART]], [[LO_PART]]
-define i32 @Ldursb_szext_merge(i8* %p) nounwind {
-  %add.ptr0 = getelementptr inbounds i8, i8* %p, i64 -1
-  %tmp = load i8, i8* %add.ptr0
-  %add.ptr = getelementptr inbounds i8, i8* %p, i64 -2
-  %tmp1 = load i8, i8* %add.ptr
-  %sexttmp = sext i8 %tmp to i32
-  %sexttmp1 = zext i8 %tmp1 to i32
-  %add = sub nsw i32 %sexttmp, %sexttmp1
-  ret i32 %add
-}
+; RUN: llc < %s -mtriple aarch64--none-eabi -mcpu=cortex-a57 -verify-machineinstrs | FileCheck %s --check-prefix=CHECK --check-prefix=LE
+; RUN: llc < %s -mtriple aarch64_be--none-eabi -mcpu=cortex-a57 -verify-machineinstrs | FileCheck %s --check-prefix=CHECK --check-prefix=BE
+; RUN: llc < %s -mtriple aarch64--none-eabi -mcpu=kryo -verify-machineinstrs | FileCheck %s --check-prefix=CHECK --check-prefix=LE
 
 ; CHECK-LABEL: Strh_zero
 ; CHECK: str wzr
 define void @Strh_zero(i16* nocapture %P, i32 %n) {
 entry:
   %idxprom = sext i32 %n to i64
   %arrayidx = getelementptr inbounds i16, i16* %P, i64 %idxprom
   store i16 0, i16* %arrayidx
   %add = add nsw i32 %n, 1
   %idxprom1 = sext i32 %add to i64
   %arrayidx2 = getelementptr inbounds i16, i16* %P, i64 %idxprom1
   store i16 0, i16* %arrayidx2
   ret void
 }
 
 ; CHECK-LABEL: Strh_zero_4