Merge llvm 3.5.0 release from ^/vendor/llvm/dist, resolve conflicts, and

preserve our customizations, where necessary.

96 files changed, 9778 insertions, 5992 deletions

diff --git a/contrib/llvm/lib/Target/ARM/A15SDOptimizer.cpp b/contrib/llvm/lib/Target/ARM/A15SDOptimizer.cpp
index 3e805a2b68bc..92eaf9e1c9b3 100644
--- a/contrib/llvm/lib/Target/ARM/A15SDOptimizer.cpp
+++ b/contrib/llvm/lib/Target/ARM/A15SDOptimizer.cpp
@@ -24,36 +24,30 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "a15-sd-optimizer"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
-#include "ARMSubtarget.h"
-#include "ARMISelLowering.h"
-#include "ARMTargetMachine.h"
-
-#include "llvm/ADT/SmallPtrSet.h"
+#include "ARMBaseRegisterInfo.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/MachineInstr.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
-#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
-#include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetRegisterInfo.h"
-
 #include <set>
 
 using namespace llvm;
 
+#define DEBUG_TYPE "a15-sd-optimizer"
+
 namespace {
   struct A15SDOptimizer : public MachineFunctionPass {
     static char ID;
     A15SDOptimizer() : MachineFunctionPass(ID) {}
 
-    virtual bool runOnMachineFunction(MachineFunction &Fn);
+    bool runOnMachineFunction(MachineFunction &Fn) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM A15 S->D optimizer";
     }
 
@@ -97,7 +91,7 @@ namespace {
     unsigned createImplicitDef(MachineBasicBlock &MBB,
                                MachineBasicBlock::iterator InsertBefore,
                                DebugLoc DL);
-    
+
     //
     // Various property checkers
     //
@@ -165,7 +159,7 @@ unsigned A15SDOptimizer::getPrefSPRLane(unsigned SReg) {
   if (!MI) return ARM::ssub_0;
   MachineOperand *MO = MI->findRegisterDefOperand(SReg);
 
-  assert(MO->isReg() && "Non register operand found!");
+  assert(MO->isReg() && "Non-register operand found!");
   if (!MO) return ARM::ssub_0;
 
   if (MI->isCopy() && usesRegClass(MI->getOperand(1),
@@ -227,9 +221,9 @@ void A15SDOptimizer::eraseInstrWithNoUses(MachineInstr *MI) {
           IsDead = false;
           break;
         }
-        for (MachineRegisterInfo::use_iterator II = MRI->use_begin(Reg),
-                            EE = MRI->use_end();
-                            II != EE; ++II) {
+        for (MachineRegisterInfo::use_instr_iterator
+             II = MRI->use_instr_begin(Reg), EE = MRI->use_instr_end();
+             II != EE; ++II) {
           // We don't care about self references.
           if (&*II == Def)
             continue;
@@ -266,7 +260,7 @@ unsigned A15SDOptimizer::optimizeSDPattern(MachineInstr *MI) {
       if (DPRMI && SPRMI) {
         // See if the first operand of this insert_subreg is IMPLICIT_DEF
         MachineInstr *ECDef = elideCopies(DPRMI);
-        if (ECDef != 0 && ECDef->isImplicitDef()) {
+        if (ECDef && ECDef->isImplicitDef()) {
           // Another corner case - if we're inserting something that is purely
           // a subreg copy of a DPR, just use that DPR.
 
@@ -327,8 +321,7 @@ unsigned A15SDOptimizer::optimizeSDPattern(MachineInstr *MI) {
       return optimizeAllLanesPattern(MI, MI->getOperand(0).getReg());
   }
 
-  assert(0 && "Unhandled update pattern!");
-  return 0;
+  llvm_unreachable("Unhandled update pattern!");
 }
 
 // Return true if this MachineInstr inserts a scalar (SPR) value into
@@ -355,10 +348,10 @@ MachineInstr *A15SDOptimizer::elideCopies(MachineInstr *MI) {
   if (!MI->isFullCopy())
     return MI;
   if (!TRI->isVirtualRegister(MI->getOperand(1).getReg()))
-    return NULL;
+    return nullptr;
   MachineInstr *Def = MRI->getVRegDef(MI->getOperand(1).getReg());
   if (!Def)
-    return NULL;
+    return nullptr;
   return elideCopies(Def);
 }
 
@@ -442,7 +435,7 @@ A15SDOptimizer::createDupLane(MachineBasicBlock &MBB,
                          Out)
                    .addReg(Reg)
                    .addImm(Lane));
- 
+
   return Out;
 }
 
@@ -608,7 +601,7 @@ bool A15SDOptimizer::runOnInstruction(MachineInstr *MI) {
   //   * INSERT_SUBREG: * If the SPR value was originally in another DPR/QPR
   //                      lane, and the other lane(s) of the DPR/QPR register
   //                      that we are inserting in are undefined, use the
-  //                      original DPR/QPR value. 
+  //                      original DPR/QPR value.
   //                    * Otherwise, fall back on the same stategy as COPY.
   //
   //   * REG_SEQUENCE:  * If all except one of the input operands are
@@ -654,7 +647,7 @@ bool A15SDOptimizer::runOnInstruction(MachineInstr *MI) {
       unsigned DPRDefReg = MI->getOperand(0).getReg();
       for (MachineRegisterInfo::use_iterator I = MRI->use_begin(DPRDefReg),
              E = MRI->use_end(); I != E; ++I)
-        Uses.push_back(&I.getOperand());
+        Uses.push_back(&*I);
 
       // We can optimize this.
       unsigned NewReg = optimizeSDPattern(MI);
@@ -700,7 +693,7 @@ bool A15SDOptimizer::runOnMachineFunction(MachineFunction &Fn) {
       MI != ME;) {
       Modified |= runOnInstruction(MI++);
     }
- 
+
   }
 
   for (std::set<MachineInstr *>::iterator I = DeadInstr.begin(),
diff --git a/contrib/llvm/lib/Target/ARM/ARM.h b/contrib/llvm/lib/Target/ARM/ARM.h
index 80e5f37eb086..55df29c1499a 100644
--- a/contrib/llvm/lib/Target/ARM/ARM.h
+++ b/contrib/llvm/lib/Target/ARM/ARM.h
@@ -15,19 +15,19 @@
 #ifndef TARGET_ARM_H
 #define TARGET_ARM_H
 
-#include "MCTargetDesc/ARMBaseInfo.h"
-#include "MCTargetDesc/ARMMCTargetDesc.h"
-#include "llvm/Support/DataTypes.h"
-#include "llvm/Target/TargetMachine.h"
+#include "llvm/Support/CodeGen.h"
 
 namespace llvm {
 
 class ARMAsmPrinter;
 class ARMBaseTargetMachine;
 class FunctionPass;
+class ImmutablePass;
 class JITCodeEmitter;
 class MachineInstr;
 class MCInst;
+class TargetLowering;
+class TargetMachine;
 
 FunctionPass *createARMISelDag(ARMBaseTargetMachine &TM,
                                CodeGenOpt::Level OptLevel);
@@ -43,6 +43,7 @@ FunctionPass *createARMGlobalMergePass(const TargetLowering* tli);
 FunctionPass *createARMConstantIslandPass();
 FunctionPass *createMLxExpansionPass();
 FunctionPass *createThumb2ITBlockPass();
+FunctionPass *createARMOptimizeBarriersPass();
 FunctionPass *createThumb2SizeReductionPass();
 
 /// \brief Creates an ARM-specific Target Transformation Info pass.
diff --git a/contrib/llvm/lib/Target/ARM/ARM.td b/contrib/llvm/lib/Target/ARM/ARM.td
index 36e5680ca4e0..7916ccc180c8 100644
--- a/contrib/llvm/lib/Target/ARM/ARM.td
+++ b/contrib/llvm/lib/Target/ARM/ARM.td
@@ -73,6 +73,11 @@ def FeatureCrypto : SubtargetFeature<"crypto", "HasCrypto", "true",
 def FeatureCRC : SubtargetFeature<"crc", "HasCRC", "true",
                           "Enable support for CRC instructions">;
 
+// Cyclone has preferred instructions for zeroing VFP registers, which can
+// execute in 0 cycles.
+def FeatureZCZeroing : SubtargetFeature<"zcz", "HasZeroCycleZeroing", "true",
+                                        "Has zero-cycle zeroing instructions">;
+
 // Some processors have FP multiply-accumulate instructions that don't
 // play nicely with other VFP / NEON instructions, and it's generally better
 // to just not use them.
@@ -179,7 +184,14 @@ def ProcA5      : SubtargetFeature<"a5", "ARMProcFamily", "CortexA5",
                                    "Cortex-A5 ARM processors",
                                    [FeatureSlowFPBrcc, FeatureHasSlowFPVMLx,
                                     FeatureVMLxForwarding, FeatureT2XtPk,
-                                    FeatureTrustZone]>;
+                                    FeatureTrustZone, FeatureMP]>;
+def ProcA7      : SubtargetFeature<"a7", "ARMProcFamily", "CortexA7",
+                                   "Cortex-A7 ARM processors",
+                                   [FeatureSlowFPBrcc, FeatureHasSlowFPVMLx,
+                                    FeatureVMLxForwarding, FeatureT2XtPk,
+                                    FeatureVFP4, FeatureMP,
+                                    FeatureHWDiv, FeatureHWDivARM,
+                                    FeatureTrustZone, FeatureVirtualization]>;
 def ProcA8      : SubtargetFeature<"a8", "ARMProcFamily", "CortexA8",
                                    "Cortex-A8 ARM processors",
                                    [FeatureSlowFPBrcc, FeatureHasSlowFPVMLx,
@@ -198,6 +210,15 @@ def ProcSwift   : SubtargetFeature<"swift", "ARMProcFamily", "Swift",
                                     FeatureHWDivARM, FeatureAvoidPartialCPSR,
                                     FeatureAvoidMOVsShOp,
                                     FeatureHasSlowFPVMLx, FeatureTrustZone]>;
+def ProcA12     : SubtargetFeature<"a12", "ARMProcFamily", "CortexA12",
+                                   "Cortex-A12 ARM processors",
+                                   [FeatureVMLxForwarding,
+                                    FeatureT2XtPk, FeatureVFP4,
+                                    FeatureHWDiv, FeatureHWDivARM,
+                                    FeatureAvoidPartialCPSR,
+                                    FeatureVirtualization,
+                                    FeatureTrustZone]>;
+
 
 // FIXME: It has not been determined if A15 has these features.
 def ProcA15      : SubtargetFeature<"a15", "ARMProcFamily", "CortexA15",
@@ -227,6 +248,26 @@ def ProcR5      : SubtargetFeature<"r5", "ARMProcFamily", "CortexR5",
                                     FeatureAvoidPartialCPSR,
                                     FeatureT2XtPk]>;
 
+// FIXME: krait has currently the same features as A9
+// plus VFP4 and hardware division features.
+def ProcKrait   : SubtargetFeature<"krait", "ARMProcFamily", "Krait",
+                                   "Qualcomm ARM processors",
+                                   [FeatureVMLxForwarding,
+                                    FeatureT2XtPk, FeatureFP16,
+                                    FeatureAvoidPartialCPSR,
+                                    FeatureTrustZone,
+                                    FeatureVFP4,
+                                    FeatureHWDiv,
+                                    FeatureHWDivARM]>;
+
+
+def FeatureAPCS  : SubtargetFeature<"apcs", "TargetABI", "ARM_ABI_APCS",
+                                   "Use the APCS ABI">;
+
+def FeatureAAPCS : SubtargetFeature<"aapcs", "TargetABI", "ARM_ABI_AAPCS",
+                                   "Use the AAPCS ABI">;
+
+
 class ProcNoItin<string Name, list<SubtargetFeature> Features>
  : Processor<Name, NoItineraries, Features>;
 
@@ -296,6 +337,10 @@ def : ProcessorModel<"cortex-a5",   CortexA8Model,
                                     [ProcA5, HasV7Ops, FeatureNEON, FeatureDB,
                                      FeatureVFP4, FeatureDSPThumb2,
                                      FeatureHasRAS, FeatureAClass]>;
+def : ProcessorModel<"cortex-a7",   CortexA8Model,
+                                    [ProcA7, HasV7Ops, FeatureNEON, FeatureDB,
+                                     FeatureDSPThumb2, FeatureHasRAS,
+                                     FeatureAClass]>;
 def : ProcessorModel<"cortex-a8",   CortexA8Model,
                                     [ProcA8, HasV7Ops, FeatureNEON, FeatureDB,
                                      FeatureDSPThumb2, FeatureHasRAS,
@@ -308,11 +353,26 @@ def : ProcessorModel<"cortex-a9-mp", CortexA9Model,
                                     [ProcA9, HasV7Ops, FeatureNEON, FeatureDB,
                                      FeatureDSPThumb2, FeatureMP,
                                      FeatureHasRAS, FeatureAClass]>;
+
+// FIXME: A12 has currently the same Schedule model as A9
+def : ProcessorModel<"cortex-a12", CortexA9Model,
+                                    [ProcA12, HasV7Ops, FeatureNEON, FeatureDB,
+                                     FeatureDSPThumb2, FeatureMP,
+                                     FeatureHasRAS, FeatureAClass]>;
+
 // FIXME: A15 has currently the same ProcessorModel as A9.
 def : ProcessorModel<"cortex-a15",   CortexA9Model,
                                     [ProcA15, HasV7Ops, FeatureNEON, FeatureDB,
                                      FeatureDSPThumb2, FeatureHasRAS,
                                      FeatureAClass]>;
+
+// FIXME: krait has currently the same Schedule model as A9
+def : ProcessorModel<"krait",       CortexA9Model,
+                                    [ProcKrait, HasV7Ops,
+                                     FeatureNEON, FeatureDB,
+                                     FeatureDSPThumb2, FeatureHasRAS,
+                                     FeatureAClass]>;
+
 // FIXME: R5 has currently the same ProcessorModel as A8.
 def : ProcessorModel<"cortex-r5",   CortexA8Model,
                                     [ProcR5, HasV7Ops, FeatureDB,
@@ -347,6 +407,13 @@ def : ProcNoItin<"cortex-a57",      [ProcA57, HasV8Ops, FeatureAClass,
                                     FeatureDB, FeatureFPARMv8,
                                     FeatureNEON, FeatureDSPThumb2]>;
 
+// Cyclone is very similar to swift
+def : ProcessorModel<"cyclone",     SwiftModel,
+                                    [ProcSwift, HasV8Ops, HasV7Ops,
+                                     FeatureCrypto, FeatureFPARMv8,
+                                     FeatureDB,FeatureDSPThumb2,
+                                     FeatureHasRAS, FeatureZCZeroing]>;
+
 //===----------------------------------------------------------------------===//
 // Register File Description
 //===----------------------------------------------------------------------===//
@@ -363,17 +430,6 @@ include "ARMInstrInfo.td"
 
 def ARMInstrInfo : InstrInfo;
 
-
-//===----------------------------------------------------------------------===//
-// Assembly printer
-//===----------------------------------------------------------------------===//
-// ARM Uses the MC printer for asm output, so make sure the TableGen
-// AsmWriter bits get associated with the correct class.
-def ARMAsmWriter : AsmWriter {
-  string AsmWriterClassName  = "InstPrinter";
-  bit isMCAsmWriter = 1;
-}
-
 //===----------------------------------------------------------------------===//
 // Declare the target which we are implementing
 //===----------------------------------------------------------------------===//
@@ -381,6 +437,4 @@ def ARMAsmWriter : AsmWriter {
 def ARM : Target {
   // Pull in Instruction Info:
   let InstructionSet = ARMInstrInfo;
-
-  let AssemblyWriters = [ARMAsmWriter];
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.cpp b/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.cpp
index e79f88d4b6f1..28d2610c3903 100644
--- a/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.cpp
@@ -12,10 +12,8 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "asm-printer"
 #include "ARMAsmPrinter.h"
 #include "ARM.h"
-#include "ARMBuildAttrs.h"
 #include "ARMConstantPoolValue.h"
 #include "ARMFPUName.h"
 #include "ARMMachineFunctionInfo.h"
@@ -26,13 +24,13 @@
 #include "MCTargetDesc/ARMMCExpr.h"
 #include "llvm/ADT/SetVector.h"
 #include "llvm/ADT/SmallString.h"
-#include "llvm/Assembly/Writer.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/MachineJumpTableInfo.h"
 #include "llvm/CodeGen/MachineModuleInfoImpls.h"
-#include "llvm/DebugInfo.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DataLayout.h"
+#include "llvm/IR/DebugInfo.h"
+#include "llvm/IR/Mangler.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Type.h"
 #include "llvm/MC/MCAsmInfo.h"
@@ -45,80 +43,19 @@
 #include "llvm/MC/MCSectionMachO.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSymbol.h"
+#include "llvm/Support/ARMBuildAttributes.h"
+#include "llvm/Support/COFF.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ELF.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/TargetRegistry.h"
 #include "llvm/Support/raw_ostream.h"
-#include "llvm/Target/Mangler.h"
 #include "llvm/Target/TargetMachine.h"
 #include <cctype>
 using namespace llvm;
 
-/// EmitDwarfRegOp - Emit dwarf register operation.
-void ARMAsmPrinter::EmitDwarfRegOp(const MachineLocation &MLoc,
-                                   bool Indirect) const {
-  const TargetRegisterInfo *RI = TM.getRegisterInfo();
-  if (RI->getDwarfRegNum(MLoc.getReg(), false) != -1) {
-    AsmPrinter::EmitDwarfRegOp(MLoc, Indirect);
-    return;
-  }
-  assert(MLoc.isReg() && !Indirect &&
-         "This doesn't support offset/indirection - implement it if needed");
-  unsigned Reg = MLoc.getReg();
-  if (Reg >= ARM::S0 && Reg <= ARM::S31) {
-    assert(ARM::S0 + 31 == ARM::S31 && "Unexpected ARM S register numbering");
-    // S registers are described as bit-pieces of a register
-    // S[2x] = DW_OP_regx(256 + (x>>1)) DW_OP_bit_piece(32, 0)
-    // S[2x+1] = DW_OP_regx(256 + (x>>1)) DW_OP_bit_piece(32, 32)
-
-    unsigned SReg = Reg - ARM::S0;
-    bool odd = SReg & 0x1;
-    unsigned Rx = 256 + (SReg >> 1);
-
-    OutStreamer.AddComment("DW_OP_regx for S register");
-    EmitInt8(dwarf::DW_OP_regx);
-
-    OutStreamer.AddComment(Twine(SReg));
-    EmitULEB128(Rx);
-
-    if (odd) {
-      OutStreamer.AddComment("DW_OP_bit_piece 32 32");
-      EmitInt8(dwarf::DW_OP_bit_piece);
-      EmitULEB128(32);
-      EmitULEB128(32);
-    } else {
-      OutStreamer.AddComment("DW_OP_bit_piece 32 0");
-      EmitInt8(dwarf::DW_OP_bit_piece);
-      EmitULEB128(32);
-      EmitULEB128(0);
-    }
-  } else if (Reg >= ARM::Q0 && Reg <= ARM::Q15) {
-    assert(ARM::Q0 + 15 == ARM::Q15 && "Unexpected ARM Q register numbering");
-    // Q registers Q0-Q15 are described by composing two D registers together.
-    // Qx = DW_OP_regx(256+2x) DW_OP_piece(8) DW_OP_regx(256+2x+1)
-    // DW_OP_piece(8)
-
-    unsigned QReg = Reg - ARM::Q0;
-    unsigned D1 = 256 + 2 * QReg;
-    unsigned D2 = D1 + 1;
-
-    OutStreamer.AddComment("DW_OP_regx for Q register: D1");
-    EmitInt8(dwarf::DW_OP_regx);
-    EmitULEB128(D1);
-    OutStreamer.AddComment("DW_OP_piece 8");
-    EmitInt8(dwarf::DW_OP_piece);
-    EmitULEB128(8);
-
-    OutStreamer.AddComment("DW_OP_regx for Q register: D2");
-    EmitInt8(dwarf::DW_OP_regx);
-    EmitULEB128(D2);
-    OutStreamer.AddComment("DW_OP_piece 8");
-    EmitInt8(dwarf::DW_OP_piece);
-    EmitULEB128(8);
-  }
-}
+#define DEBUG_TYPE "asm-printer"
 
 void ARMAsmPrinter::EmitFunctionBodyEnd() {
   // Make sure to terminate any constant pools that were at the end
@@ -145,12 +82,13 @@ void ARMAsmPrinter::EmitXXStructor(const Constant *CV) {
   const GlobalValue *GV = dyn_cast<GlobalValue>(CV->stripPointerCasts());
   assert(GV && "C++ constructor pointer was not a GlobalValue!");
 
-  const MCExpr *E = MCSymbolRefExpr::Create(getSymbol(GV),
-                                            (Subtarget->isTargetDarwin()
-                                             ? MCSymbolRefExpr::VK_None
-                                             : MCSymbolRefExpr::VK_ARM_TARGET1),
+  const MCExpr *E = MCSymbolRefExpr::Create(GetARMGVSymbol(GV,
+                                                           ARMII::MO_NO_FLAG),
+                                            (Subtarget->isTargetELF()
+                                             ? MCSymbolRefExpr::VK_ARM_TARGET1
+                                             : MCSymbolRefExpr::VK_None),
                                             OutContext);
-  
+
   OutStreamer.EmitValue(E, Size);
 }
 
@@ -161,7 +99,28 @@ bool ARMAsmPrinter::runOnMachineFunction(MachineFunction &MF) {
   AFI = MF.getInfo<ARMFunctionInfo>();
   MCP = MF.getConstantPool();
 
-  return AsmPrinter::runOnMachineFunction(MF);
+  SetupMachineFunction(MF);
+
+  if (Subtarget->isTargetCOFF()) {
+    bool Internal = MF.getFunction()->hasInternalLinkage();
+    COFF::SymbolStorageClass Scl = Internal ? COFF::IMAGE_SYM_CLASS_STATIC
+                                            : COFF::IMAGE_SYM_CLASS_EXTERNAL;
+    int Type = COFF::IMAGE_SYM_DTYPE_FUNCTION << COFF::SCT_COMPLEX_TYPE_SHIFT;
+
+    OutStreamer.BeginCOFFSymbolDef(CurrentFnSym);
+    OutStreamer.EmitCOFFSymbolStorageClass(Scl);
+    OutStreamer.EmitCOFFSymbolType(Type);
+    OutStreamer.EndCOFFSymbolDef();
+  }
+
+  // Have common code print out the function header with linkage info etc.
+  EmitFunctionHeader();
+
+  // Emit the rest of the function body.
+  EmitFunctionBody();
+
+  // We didn't modify anything.
+  return false;
 }
 
 void ARMAsmPrinter::printOperand(const MachineInstr *MI, int OpNum,
@@ -206,25 +165,16 @@ void ARMAsmPrinter::printOperand(const MachineInstr *MI, int OpNum,
     else if ((Modifier && strcmp(Modifier, "hi16") == 0) ||
              (TF & ARMII::MO_HI16))
       O << ":upper16:";
-    O << *getSymbol(GV);
+    O << *GetARMGVSymbol(GV, TF);
 
     printOffset(MO.getOffset(), O);
     if (TF == ARMII::MO_PLT)
       O << "(PLT)";
     break;
   }
-  case MachineOperand::MO_ExternalSymbol: {
-    O << *GetExternalSymbolSymbol(MO.getSymbolName());
-    if (TF == ARMII::MO_PLT)
-      O << "(PLT)";
-    break;
-  }
   case MachineOperand::MO_ConstantPoolIndex:
     O << *GetCPISymbol(MO.getIndex());
     break;
-  case MachineOperand::MO_JumpTableIndex:
-    O << *GetJTISymbol(MO.getIndex());
-    break;
   }
 }
 
@@ -232,16 +182,18 @@ void ARMAsmPrinter::printOperand(const MachineInstr *MI, int OpNum,
 
 MCSymbol *ARMAsmPrinter::
 GetARMJTIPICJumpTableLabel2(unsigned uid, unsigned uid2) const {
+  const DataLayout *DL = TM.getDataLayout();
   SmallString<60> Name;
-  raw_svector_ostream(Name) << MAI->getPrivateGlobalPrefix() << "JTI"
+  raw_svector_ostream(Name) << DL->getPrivateGlobalPrefix() << "JTI"
     << getFunctionNumber() << '_' << uid << '_' << uid2;
   return OutContext.GetOrCreateSymbol(Name.str());
 }
 
 
 MCSymbol *ARMAsmPrinter::GetARMSJLJEHLabel() const {
+  const DataLayout *DL = TM.getDataLayout();
   SmallString<60> Name;
-  raw_svector_ostream(Name) << MAI->getPrivateGlobalPrefix() << "SJLJEH"
+  raw_svector_ostream(Name) << DL->getPrivateGlobalPrefix() << "SJLJEH"
     << getFunctionNumber();
   return OutContext.GetOrCreateSymbol(Name.str());
 }
@@ -311,7 +263,7 @@ bool ARMAsmPrinter::PrintAsmOperand(const MachineInstr *MI, unsigned OpNum,
       if (ARM::GPRPairRegClass.contains(RegBegin)) {
         const TargetRegisterInfo *TRI = MF->getTarget().getRegisterInfo();
         unsigned Reg0 = TRI->getSubReg(RegBegin, ARM::gsub_0);
-        O << ARMInstPrinter::getRegisterName(Reg0) << ", ";;
+        O << ARMInstPrinter::getRegisterName(Reg0) << ", ";
         RegBegin = TRI->getSubReg(RegBegin, ARM::gsub_1);
       }
       O << ARMInstPrinter::getRegisterName(RegBegin);
@@ -446,8 +398,22 @@ bool ARMAsmPrinter::PrintAsmMemoryOperand(const MachineInstr *MI,
   return false;
 }
 
+static bool isThumb(const MCSubtargetInfo& STI) {
+  return (STI.getFeatureBits() & ARM::ModeThumb) != 0;
+}
+
+void ARMAsmPrinter::emitInlineAsmEnd(const MCSubtargetInfo &StartInfo,
+                                     const MCSubtargetInfo *EndInfo) const {
+  // If either end mode is unknown (EndInfo == NULL) or different than
+  // the start mode, then restore the start mode.
+  const bool WasThumb = isThumb(StartInfo);
+  if (!EndInfo || WasThumb != isThumb(*EndInfo)) {
+    OutStreamer.EmitAssemblerFlag(WasThumb ? MCAF_Code16 : MCAF_Code32);
+  }
+}
+
 void ARMAsmPrinter::EmitStartOfAsmFile(Module &M) {
-  if (Subtarget->isTargetDarwin()) {
+  if (Subtarget->isTargetMachO()) {
     Reloc::Model RelocM = TM.getRelocationModel();
     if (RelocM == Reloc::PIC_ || RelocM == Reloc::DynamicNoPIC) {
       // Declare all the text sections up front (before the DWARF sections
@@ -468,7 +434,7 @@ void ARMAsmPrinter::EmitStartOfAsmFile(Module &M) {
       // Now any user defined text sections from function attributes.
       for (Module::iterator F = M.begin(), e = M.end(); F != e; ++F)
         if (!F->isDeclaration() && !F->hasAvailableExternallyLinkage())
-          TextSections.insert(TLOFMacho.SectionForGlobal(F, Mang, TM));
+          TextSections.insert(TLOFMacho.SectionForGlobal(F, *Mang, TM));
       // Now the coalescable sections.
       TextSections.insert(TLOFMacho.getTextCoalSection());
       TextSections.insert(TLOFMacho.getConstTextCoalSection());
@@ -480,23 +446,30 @@ void ARMAsmPrinter::EmitStartOfAsmFile(Module &M) {
       if (RelocM == Reloc::DynamicNoPIC) {
         const MCSection *sect =
           OutContext.getMachOSection("__TEXT", "__symbol_stub4",
-                                     MCSectionMachO::S_SYMBOL_STUBS,
+                                     MachO::S_SYMBOL_STUBS,
                                      12, SectionKind::getText());
         OutStreamer.SwitchSection(sect);
       } else {
         const MCSection *sect =
           OutContext.getMachOSection("__TEXT", "__picsymbolstub4",
-                                     MCSectionMachO::S_SYMBOL_STUBS,
+                                     MachO::S_SYMBOL_STUBS,
                                      16, SectionKind::getText());
         OutStreamer.SwitchSection(sect);
       }
       const MCSection *StaticInitSect =
         OutContext.getMachOSection("__TEXT", "__StaticInit",
-                                   MCSectionMachO::S_REGULAR |
-                                   MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS,
+                                   MachO::S_REGULAR |
+                                   MachO::S_ATTR_PURE_INSTRUCTIONS,
                                    SectionKind::getText());
       OutStreamer.SwitchSection(StaticInitSect);
     }
+
+    // Compiling with debug info should not affect the code
+    // generation.  Ensure the cstring section comes before the
+    // optional __DWARF secion. Otherwise, PC-relative loads would
+    // have to use different instruction sequences at "-g" in order to
+    // reach global data in the same object file.
+    OutStreamer.SwitchSection(getObjFileLowering().getCStringSection());
   }
 
   // Use unified assembler syntax.
@@ -507,9 +480,32 @@ void ARMAsmPrinter::EmitStartOfAsmFile(Module &M) {
     emitAttributes();
 }
 
+static void
+emitNonLazySymbolPointer(MCStreamer &OutStreamer, MCSymbol *StubLabel,
+                         MachineModuleInfoImpl::StubValueTy &MCSym) {
+  // L_foo$stub:
+  OutStreamer.EmitLabel(StubLabel);
+  //   .indirect_symbol _foo
+  OutStreamer.EmitSymbolAttribute(MCSym.getPointer(), MCSA_IndirectSymbol);
+
+  if (MCSym.getInt())
+    // External to current translation unit.
+    OutStreamer.EmitIntValue(0, 4/*size*/);
+  else
+    // Internal to current translation unit.
+    //
+    // When we place the LSDA into the TEXT section, the type info
+    // pointers need to be indirect and pc-rel. We accomplish this by
+    // using NLPs; however, sometimes the types are local to the file.
+    // We need to fill in the value for the NLP in those cases.
+    OutStreamer.EmitValue(
+        MCSymbolRefExpr::Create(MCSym.getPointer(), OutStreamer.getContext()),
+        4 /*size*/);
+}
+
 
 void ARMAsmPrinter::EmitEndOfAsmFile(Module &M) {
-  if (Subtarget->isTargetDarwin()) {
+  if (Subtarget->isTargetMachO()) {
     // All darwin targets use mach-o.
     const TargetLoweringObjectFileMachO &TLOFMacho =
       static_cast<const TargetLoweringObjectFileMachO &>(getObjFileLowering());
@@ -523,27 +519,9 @@ void ARMAsmPrinter::EmitEndOfAsmFile(Module &M) {
       // Switch with ".non_lazy_symbol_pointer" directive.
       OutStreamer.SwitchSection(TLOFMacho.getNonLazySymbolPointerSection());
       EmitAlignment(2);
-      for (unsigned i = 0, e = Stubs.size(); i != e; ++i) {
-        // L_foo$stub:
-        OutStreamer.EmitLabel(Stubs[i].first);
-        //   .indirect_symbol _foo
-        MachineModuleInfoImpl::StubValueTy &MCSym = Stubs[i].second;
-        OutStreamer.EmitSymbolAttribute(MCSym.getPointer(),MCSA_IndirectSymbol);
-
-        if (MCSym.getInt())
-          // External to current translation unit.
-          OutStreamer.EmitIntValue(0, 4/*size*/);
-        else
-          // Internal to current translation unit.
-          //
-          // When we place the LSDA into the TEXT section, the type info
-          // pointers need to be indirect and pc-rel. We accomplish this by
-          // using NLPs; however, sometimes the types are local to the file.
-          // We need to fill in the value for the NLP in those cases.
-          OutStreamer.EmitValue(MCSymbolRefExpr::Create(MCSym.getPointer(),
-                                                        OutContext),
-                                4/*size*/);
-      }
+
+      for (auto &Stub : Stubs)
+        emitNonLazySymbolPointer(OutStreamer, Stub.first, Stub.second);
 
       Stubs.clear();
       OutStreamer.AddBlankLine();
@@ -551,17 +529,11 @@ void ARMAsmPrinter::EmitEndOfAsmFile(Module &M) {
 
     Stubs = MMIMacho.GetHiddenGVStubList();
     if (!Stubs.empty()) {
-      OutStreamer.SwitchSection(getObjFileLowering().getDataSection());
+      OutStreamer.SwitchSection(TLOFMacho.getNonLazySymbolPointerSection());
       EmitAlignment(2);
-      for (unsigned i = 0, e = Stubs.size(); i != e; ++i) {
-        // L_foo$stub:
-        OutStreamer.EmitLabel(Stubs[i].first);
-        //   .long _foo
-        OutStreamer.EmitValue(MCSymbolRefExpr::
-                              Create(Stubs[i].second.getPointer(),
-                                     OutContext),
-                              4/*size*/);
-      }
+
+      for (auto &Stub : Stubs)
+        emitNonLazySymbolPointer(OutStreamer, Stub.first, Stub.second);
 
       Stubs.clear();
       OutStreamer.AddBlankLine();
@@ -574,6 +546,28 @@ void ARMAsmPrinter::EmitEndOfAsmFile(Module &M) {
     // generates code that does this, it is always safe to set.
     OutStreamer.EmitAssemblerFlag(MCAF_SubsectionsViaSymbols);
   }
+
+  // Emit a .data.rel section containing any stubs that were created.
+  if (Subtarget->isTargetELF()) {
+    const TargetLoweringObjectFileELF &TLOFELF =
+      static_cast<const TargetLoweringObjectFileELF &>(getObjFileLowering());
+
+    MachineModuleInfoELF &MMIELF = MMI->getObjFileInfo<MachineModuleInfoELF>();
+
+    // Output stubs for external and common global variables.
+    MachineModuleInfoELF::SymbolListTy Stubs = MMIELF.GetGVStubList();
+    if (!Stubs.empty()) {
+      OutStreamer.SwitchSection(TLOFELF.getDataRelSection());
+      const DataLayout *TD = TM.getDataLayout();
+
+      for (auto &stub: Stubs) {
+        OutStreamer.EmitLabel(stub.first);
+        OutStreamer.EmitSymbolValue(stub.second.getPointer(),
+                                    TD->getPointerSize(0));
+      }
+      Stubs.clear();
+    }
+  }
 }
 
 //===----------------------------------------------------------------------===//
@@ -611,28 +605,33 @@ static ARMBuildAttrs::CPUArch getArchForCPU(StringRef CPU,
 }
 
 void ARMAsmPrinter::emitAttributes() {
-  MCTargetStreamer &TS = OutStreamer.getTargetStreamer();
+  MCTargetStreamer &TS = *OutStreamer.getTargetStreamer();
   ARMTargetStreamer &ATS = static_cast<ARMTargetStreamer &>(TS);
 
   ATS.switchVendor("aeabi");
 
   std::string CPUString = Subtarget->getCPUString();
 
-  if (CPUString != "generic")
+  // FIXME: remove krait check when GNU tools support krait cpu
+  if (CPUString != "generic" && CPUString != "krait")
     ATS.emitTextAttribute(ARMBuildAttrs::CPU_name, CPUString);
 
   ATS.emitAttribute(ARMBuildAttrs::CPU_arch,
                     getArchForCPU(CPUString, Subtarget));
 
-  if (Subtarget->isAClass()) {
-    ATS.emitAttribute(ARMBuildAttrs::CPU_arch_profile,
-                      ARMBuildAttrs::ApplicationProfile);
-  } else if (Subtarget->isRClass()) {
-    ATS.emitAttribute(ARMBuildAttrs::CPU_arch_profile,
-                      ARMBuildAttrs::RealTimeProfile);
-  } else if (Subtarget->isMClass()){
-    ATS.emitAttribute(ARMBuildAttrs::CPU_arch_profile,
-                      ARMBuildAttrs::MicroControllerProfile);
+  // Tag_CPU_arch_profile must have the default value of 0 when "Architecture
+  // profile is not applicable (e.g. pre v7, or cross-profile code)".
+  if (Subtarget->hasV7Ops()) {
+    if (Subtarget->isAClass()) {
+      ATS.emitAttribute(ARMBuildAttrs::CPU_arch_profile,
+                        ARMBuildAttrs::ApplicationProfile);
+    } else if (Subtarget->isRClass()) {
+      ATS.emitAttribute(ARMBuildAttrs::CPU_arch_profile,
+                        ARMBuildAttrs::RealTimeProfile);
+    } else if (Subtarget->isMClass()) {
+      ATS.emitAttribute(ARMBuildAttrs::CPU_arch_profile,
+                        ARMBuildAttrs::MicroControllerProfile);
+    }
   }
 
   ATS.emitAttribute(ARMBuildAttrs::ARM_ISA_use, Subtarget->hasARMOps() ?
@@ -673,6 +672,20 @@ void ARMAsmPrinter::emitAttributes() {
       ATS.emitFPU(ARM::VFPV2);
   }
 
+  if (TM.getRelocationModel() == Reloc::PIC_) {
+    // PIC specific attributes.
+    ATS.emitAttribute(ARMBuildAttrs::ABI_PCS_RW_data,
+                      ARMBuildAttrs::AddressRWPCRel);
+    ATS.emitAttribute(ARMBuildAttrs::ABI_PCS_RO_data,
+                      ARMBuildAttrs::AddressROPCRel);
+    ATS.emitAttribute(ARMBuildAttrs::ABI_PCS_GOT_use,
+                      ARMBuildAttrs::AddressGOT);
+  } else {
+    // Allow direct addressing of imported data for all other relocation models.
+    ATS.emitAttribute(ARMBuildAttrs::ABI_PCS_GOT_use,
+                      ARMBuildAttrs::AddressDirect);
+  }
+
   // Signal various FP modes.
   if (!TM.Options.UnsafeFPMath) {
     ATS.emitAttribute(ARMBuildAttrs::ABI_FP_denormal, ARMBuildAttrs::Allowed);
@@ -687,10 +700,10 @@ void ARMAsmPrinter::emitAttributes() {
     ATS.emitAttribute(ARMBuildAttrs::ABI_FP_number_model,
                       ARMBuildAttrs::AllowIEE754);
 
-  // FIXME: add more flags to ARMBuildAttrs.h
+  // FIXME: add more flags to ARMBuildAttributes.h
   // 8-bytes alignment stuff.
-  ATS.emitAttribute(ARMBuildAttrs::ABI_align8_needed, 1);
-  ATS.emitAttribute(ARMBuildAttrs::ABI_align8_preserved, 1);
+  ATS.emitAttribute(ARMBuildAttrs::ABI_align_needed, 1);
+  ATS.emitAttribute(ARMBuildAttrs::ABI_align_preserved, 1);
 
   // ABI_HardFP_use attribute to indicate single precision FP.
   if (Subtarget->isFPOnlySP())
@@ -709,11 +722,39 @@ void ARMAsmPrinter::emitAttributes() {
   if (Subtarget->hasMPExtension())
       ATS.emitAttribute(ARMBuildAttrs::MPextension_use, ARMBuildAttrs::AllowMP);
 
-  if (Subtarget->hasDivide()) {
-    // Check if hardware divide is only available in thumb2 or ARM as well.
-    ATS.emitAttribute(ARMBuildAttrs::DIV_use,
-      Subtarget->hasDivideInARMMode() ? ARMBuildAttrs::AllowDIVExt :
-                                        ARMBuildAttrs::AllowDIVIfExists);
+  // Hardware divide in ARM mode is part of base arch, starting from ARMv8.
+  // If only Thumb hwdiv is present, it must also be in base arch (ARMv7-R/M).
+  // It is not possible to produce DisallowDIV: if hwdiv is present in the base
+  // arch, supplying -hwdiv downgrades the effective arch, via ClearImpliedBits.
+  // AllowDIVExt is only emitted if hwdiv isn't available in the base arch;
+  // otherwise, the default value (AllowDIVIfExists) applies.
+  if (Subtarget->hasDivideInARMMode() && !Subtarget->hasV8Ops())
+      ATS.emitAttribute(ARMBuildAttrs::DIV_use, ARMBuildAttrs::AllowDIVExt);
+
+  if (MMI) {
+    if (const Module *SourceModule = MMI->getModule()) {
+      // ABI_PCS_wchar_t to indicate wchar_t width
+      // FIXME: There is no way to emit value 0 (wchar_t prohibited).
+      if (auto WCharWidthValue = cast_or_null<ConstantInt>(
+              SourceModule->getModuleFlag("wchar_size"))) {
+        int WCharWidth = WCharWidthValue->getZExtValue();
+        assert((WCharWidth == 2 || WCharWidth == 4) &&
+               "wchar_t width must be 2 or 4 bytes");
+        ATS.emitAttribute(ARMBuildAttrs::ABI_PCS_wchar_t, WCharWidth);
+      }
+
+      // ABI_enum_size to indicate enum width
+      // FIXME: There is no way to emit value 0 (enums prohibited) or value 3
+      //        (all enums contain a value needing 32 bits to encode).
+      if (auto EnumWidthValue = cast_or_null<ConstantInt>(
+              SourceModule->getModuleFlag("min_enum_size"))) {
+        int EnumWidth = EnumWidthValue->getZExtValue();
+        assert((EnumWidth == 1 || EnumWidth == 4) &&
+               "Minimum enum width must be 1 or 4 bytes");
+        int EnumBuildAttr = EnumWidth == 1 ? 1 : 2;
+        ATS.emitAttribute(ARMBuildAttrs::ABI_enum_size, EnumBuildAttr);
+      }
+    }
   }
 
   if (Subtarget->hasTrustZone() && Subtarget->hasVirtualization())
@@ -729,28 +770,6 @@ void ARMAsmPrinter::emitAttributes() {
   ATS.finishAttributeSection();
 }
 
-void ARMAsmPrinter::emitARMAttributeSection() {
-  // <format-version>
-  // [ <section-length> "vendor-name"
-  // [ <file-tag> <size> <attribute>*
-  //   | <section-tag> <size> <section-number>* 0 <attribute>*
-  //   | <symbol-tag> <size> <symbol-number>* 0 <attribute>*
-  //   ]+
-  // ]*
-
-  if (OutStreamer.hasRawTextSupport())
-    return;
-
-  const ARMElfTargetObjectFile &TLOFELF =
-    static_cast<const ARMElfTargetObjectFile &>
-    (getObjFileLowering());
-
-  OutStreamer.SwitchSection(TLOFELF.getAttributesSection());
-
-  // Format version
-  OutStreamer.EmitIntValue(0x41, 1);
-}
-
 //===----------------------------------------------------------------------===//
 
 static MCSymbol *getPICLabel(const char *Prefix, unsigned FunctionNumber,
@@ -765,36 +784,57 @@ static MCSymbolRefExpr::VariantKind
 getModifierVariantKind(ARMCP::ARMCPModifier Modifier) {
   switch (Modifier) {
   case ARMCP::no_modifier: return MCSymbolRefExpr::VK_None;
-  case ARMCP::TLSGD:       return MCSymbolRefExpr::VK_ARM_TLSGD;
-  case ARMCP::TPOFF:       return MCSymbolRefExpr::VK_ARM_TPOFF;
-  case ARMCP::GOTTPOFF:    return MCSymbolRefExpr::VK_ARM_GOTTPOFF;
-  case ARMCP::GOT:         return MCSymbolRefExpr::VK_ARM_GOT;
-  case ARMCP::GOTOFF:      return MCSymbolRefExpr::VK_ARM_GOTOFF;
+  case ARMCP::TLSGD:       return MCSymbolRefExpr::VK_TLSGD;
+  case ARMCP::TPOFF:       return MCSymbolRefExpr::VK_TPOFF;
+  case ARMCP::GOTTPOFF:    return MCSymbolRefExpr::VK_GOTTPOFF;
+  case ARMCP::GOT:         return MCSymbolRefExpr::VK_GOT;
+  case ARMCP::GOTOFF:      return MCSymbolRefExpr::VK_GOTOFF;
   }
   llvm_unreachable("Invalid ARMCPModifier!");
 }
 
-MCSymbol *ARMAsmPrinter::GetARMGVSymbol(const GlobalValue *GV) {
-  bool isIndirect = Subtarget->isTargetDarwin() &&
-    Subtarget->GVIsIndirectSymbol(GV, TM.getRelocationModel());
-  if (!isIndirect)
-    return getSymbol(GV);
+MCSymbol *ARMAsmPrinter::GetARMGVSymbol(const GlobalValue *GV,
+                                        unsigned char TargetFlags) {
+  if (Subtarget->isTargetMachO()) {
+    bool IsIndirect = (TargetFlags & ARMII::MO_NONLAZY) &&
+      Subtarget->GVIsIndirectSymbol(GV, TM.getRelocationModel());
+
+    if (!IsIndirect)
+      return getSymbol(GV);
 
-  // FIXME: Remove this when Darwin transition to @GOT like syntax.
-  MCSymbol *MCSym = GetSymbolWithGlobalValueBase(GV, "$non_lazy_ptr");
-  MachineModuleInfoMachO &MMIMachO =
-    MMI->getObjFileInfo<MachineModuleInfoMachO>();
-  MachineModuleInfoImpl::StubValueTy &StubSym =
-    GV->hasHiddenVisibility() ? MMIMachO.getHiddenGVStubEntry(MCSym) :
-    MMIMachO.getGVStubEntry(MCSym);
-  if (StubSym.getPointer() == 0)
-    StubSym = MachineModuleInfoImpl::
-      StubValueTy(getSymbol(GV), !GV->hasInternalLinkage());
-  return MCSym;
+    // FIXME: Remove this when Darwin transition to @GOT like syntax.
+    MCSymbol *MCSym = getSymbolWithGlobalValueBase(GV, "$non_lazy_ptr");
+    MachineModuleInfoMachO &MMIMachO =
+      MMI->getObjFileInfo<MachineModuleInfoMachO>();
+    MachineModuleInfoImpl::StubValueTy &StubSym =
+      GV->hasHiddenVisibility() ? MMIMachO.getHiddenGVStubEntry(MCSym)
+                                : MMIMachO.getGVStubEntry(MCSym);
+    if (!StubSym.getPointer())
+      StubSym = MachineModuleInfoImpl::StubValueTy(getSymbol(GV),
+                                                   !GV->hasInternalLinkage());
+    return MCSym;
+  } else if (Subtarget->isTargetCOFF()) {
+    assert(Subtarget->isTargetWindows() &&
+           "Windows is the only supported COFF target");
+
+    bool IsIndirect = (TargetFlags & ARMII::MO_DLLIMPORT);
+    if (!IsIndirect)
+      return getSymbol(GV);
+
+    SmallString<128> Name;
+    Name = "__imp_";
+    getNameWithPrefix(Name, GV);
+
+    return OutContext.GetOrCreateSymbol(Name);
+  } else if (Subtarget->isTargetELF()) {
+    return getSymbol(GV);
+  }
+  llvm_unreachable("unexpected target");
 }
 
 void ARMAsmPrinter::
 EmitMachineConstantPoolValue(MachineConstantPoolValue *MCPV) {
+  const DataLayout *DL = TM.getDataLayout();
   int Size = TM.getDataLayout()->getTypeAllocSize(MCPV->getType());
 
   ARMConstantPoolValue *ACPV = static_cast<ARMConstantPoolValue*>(MCPV);
@@ -803,7 +843,7 @@ EmitMachineConstantPoolValue(MachineConstantPoolValue *MCPV) {
   if (ACPV->isLSDA()) {
     SmallString<128> Str;
     raw_svector_ostream OS(Str);
-    OS << MAI->getPrivateGlobalPrefix() << "_LSDA_" << getFunctionNumber();
+    OS << DL->getPrivateGlobalPrefix() << "_LSDA_" << getFunctionNumber();
     MCSym = OutContext.GetOrCreateSymbol(OS.str());
   } else if (ACPV->isBlockAddress()) {
     const BlockAddress *BA =
@@ -811,7 +851,11 @@ EmitMachineConstantPoolValue(MachineConstantPoolValue *MCPV) {
     MCSym = GetBlockAddressSymbol(BA);
   } else if (ACPV->isGlobalValue()) {
     const GlobalValue *GV = cast<ARMConstantPoolConstant>(ACPV)->getGV();
-    MCSym = GetARMGVSymbol(GV);
+
+    // On Darwin, const-pool entries may get the "FOO$non_lazy_ptr" mangling, so
+    // flag the global as MO_NONLAZY.
+    unsigned char TF = Subtarget->isTargetMachO() ? ARMII::MO_NONLAZY : 0;
+    MCSym = GetARMGVSymbol(GV, TF);
   } else if (ACPV->isMachineBasicBlock()) {
     const MachineBasicBlock *MBB = cast<ARMConstantPoolMBB>(ACPV)->getMBB();
     MCSym = MBB->getSymbol();
@@ -827,7 +871,7 @@ EmitMachineConstantPoolValue(MachineConstantPoolValue *MCPV) {
                             OutContext);
 
   if (ACPV->getPCAdjustment()) {
-    MCSymbol *PCLabel = getPICLabel(MAI->getPrivateGlobalPrefix(),
+    MCSymbol *PCLabel = getPICLabel(DL->getPrivateGlobalPrefix(),
                                     getFunctionNumber(),
                                     ACPV->getLabelId(),
                                     OutContext);
@@ -929,10 +973,10 @@ void ARMAsmPrinter::EmitJump2Table(const MachineInstr *MI) {
   for (unsigned i = 0, e = JTBBs.size(); i != e; ++i) {
     MachineBasicBlock *MBB = JTBBs[i];
     const MCExpr *MBBSymbolExpr = MCSymbolRefExpr::Create(MBB->getSymbol(),
-                                                      OutContext);
+                                                          OutContext);
     // If this isn't a TBB or TBH, the entries are direct branch instructions.
     if (OffsetWidth == 4) {
-      OutStreamer.EmitInstruction(MCInstBuilder(ARM::t2B)
+      EmitToStreamer(OutStreamer, MCInstBuilder(ARM::t2B)
         .addExpr(MBBSymbolExpr)
         .addImm(ARMCC::AL)
         .addReg(0));
@@ -966,7 +1010,7 @@ void ARMAsmPrinter::EmitUnwindingInstruction(const MachineInstr *MI) {
   assert(MI->getFlag(MachineInstr::FrameSetup) &&
       "Only instruction which are involved into frame setup code are allowed");
 
-  MCTargetStreamer &TS = OutStreamer.getTargetStreamer();
+  MCTargetStreamer &TS = *OutStreamer.getTargetStreamer();
   ARMTargetStreamer &ATS = static_cast<ARMTargetStreamer &>(TS);
   const MachineFunction &MF = *MI->getParent()->getParent();
   const TargetRegisterInfo *RegInfo = MF.getTarget().getRegisterInfo();
@@ -1030,7 +1074,8 @@ void ARMAsmPrinter::EmitUnwindingInstruction(const MachineInstr *MI) {
       RegList.push_back(SrcReg);
       break;
     }
-    ATS.emitRegSave(RegList, Opc == ARM::VSTMDDB_UPD);
+    if (MAI->getExceptionHandlingType() == ExceptionHandling::ARM)
+      ATS.emitRegSave(RegList, Opc == ARM::VSTMDDB_UPD);
   } else {
     // Changes of stack / frame pointer.
     if (SrcReg == ARM::SP) {
@@ -1075,20 +1120,22 @@ void ARMAsmPrinter::EmitUnwindingInstruction(const MachineInstr *MI) {
       }
       }
 
-      if (DstReg == FramePtr && FramePtr != ARM::SP)
-        // Set-up of the frame pointer. Positive values correspond to "add"
-        // instruction.
-        ATS.emitSetFP(FramePtr, ARM::SP, -Offset);
-      else if (DstReg == ARM::SP) {
-        // Change of SP by an offset. Positive values correspond to "sub"
-        // instruction.
-        ATS.emitPad(Offset);
-      } else {
-        MI->dump();
-        llvm_unreachable("Unsupported opcode for unwinding information");
+      if (MAI->getExceptionHandlingType() == ExceptionHandling::ARM) {
+        if (DstReg == FramePtr && FramePtr != ARM::SP)
+          // Set-up of the frame pointer. Positive values correspond to "add"
+          // instruction.
+          ATS.emitSetFP(FramePtr, ARM::SP, -Offset);
+        else if (DstReg == ARM::SP) {
+          // Change of SP by an offset. Positive values correspond to "sub"
+          // instruction.
+          ATS.emitPad(Offset);
+        } else {
+          // Move of SP to a register.  Positive values correspond to an "add"
+          // instruction.
+          ATS.emitMovSP(DstReg, -Offset);
+        }
       }
     } else if (DstReg == ARM::SP) {
-      // FIXME: .movsp goes here
       MI->dump();
       llvm_unreachable("Unsupported opcode for unwinding information");
     }
@@ -1099,13 +1146,13 @@ void ARMAsmPrinter::EmitUnwindingInstruction(const MachineInstr *MI) {
   }
 }
 
-extern cl::opt<bool> EnableARMEHABI;
-
 // Simple pseudo-instructions have their lowering (with expansion to real
 // instructions) auto-generated.
 #include "ARMGenMCPseudoLowering.inc"
 
 void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
+  const DataLayout *DL = TM.getDataLayout();
+
   // If we just ended a constant pool, mark it as such.
   if (InConstantPool && MI->getOpcode() != ARM::CONSTPOOL_ENTRY) {
     OutStreamer.EmitDataRegion(MCDR_DataRegionEnd);
@@ -1113,7 +1160,8 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   }
 
   // Emit unwinding stuff for frame-related instructions
-  if (EnableARMEHABI && MI->getFlag(MachineInstr::FrameSetup))
+  if (Subtarget->isTargetEHABICompatible() &&
+       MI->getFlag(MachineInstr::FrameSetup))
     EmitUnwindingInstruction(MI);
 
   // Do any auto-generated pseudo lowerings.
@@ -1133,7 +1181,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   case ARM::t2LEApcrel: {
     // FIXME: Need to also handle globals and externals
     MCSymbol *CPISymbol = GetCPISymbol(MI->getOperand(1).getIndex());
-    OutStreamer.EmitInstruction(MCInstBuilder(MI->getOpcode() ==
+    EmitToStreamer(OutStreamer, MCInstBuilder(MI->getOpcode() ==
                                               ARM::t2LEApcrel ? ARM::t2ADR
                   : (MI->getOpcode() == ARM::tLEApcrel ? ARM::tADR
                      : ARM::ADR))
@@ -1150,7 +1198,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     MCSymbol *JTIPICSymbol =
       GetARMJTIPICJumpTableLabel2(MI->getOperand(1).getIndex(),
                                   MI->getOperand(2).getImm());
-    OutStreamer.EmitInstruction(MCInstBuilder(MI->getOpcode() ==
+    EmitToStreamer(OutStreamer, MCInstBuilder(MI->getOpcode() ==
                                               ARM::t2LEApcrelJT ? ARM::t2ADR
                   : (MI->getOpcode() == ARM::tLEApcrelJT ? ARM::tADR
                      : ARM::ADR))
@@ -1164,7 +1212,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   // Darwin call instructions are just normal call instructions with different
   // clobber semantics (they clobber R9).
   case ARM::BX_CALL: {
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::MOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::MOVr)
       .addReg(ARM::LR)
       .addReg(ARM::PC)
       // Add predicate operands.
@@ -1173,19 +1221,19 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       // Add 's' bit operand (always reg0 for this)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::BX)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::BX)
       .addReg(MI->getOperand(0).getReg()));
     return;
   }
   case ARM::tBX_CALL: {
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tMOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tMOVr)
       .addReg(ARM::LR)
       .addReg(ARM::PC)
       // Add predicate operands.
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tBX)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tBX)
       .addReg(MI->getOperand(0).getReg())
       // Add predicate operands.
       .addImm(ARMCC::AL)
@@ -1193,7 +1241,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     return;
   }
   case ARM::BMOVPCRX_CALL: {
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::MOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::MOVr)
       .addReg(ARM::LR)
       .addReg(ARM::PC)
       // Add predicate operands.
@@ -1202,7 +1250,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       // Add 's' bit operand (always reg0 for this)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::MOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::MOVr)
       .addReg(ARM::PC)
       .addReg(MI->getOperand(0).getReg())
       // Add predicate operands.
@@ -1213,7 +1261,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     return;
   }
   case ARM::BMOVPCB_CALL: {
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::MOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::MOVr)
       .addReg(ARM::LR)
       .addReg(ARM::PC)
       // Add predicate operands.
@@ -1222,10 +1270,12 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       // Add 's' bit operand (always reg0 for this)
       .addReg(0));
 
-    const GlobalValue *GV = MI->getOperand(0).getGlobal();
-    MCSymbol *GVSym = getSymbol(GV);
+    const MachineOperand &Op = MI->getOperand(0);
+    const GlobalValue *GV = Op.getGlobal();
+    const unsigned TF = Op.getTargetFlags();
+    MCSymbol *GVSym = GetARMGVSymbol(GV, TF);
     const MCExpr *GVSymExpr = MCSymbolRefExpr::Create(GVSym, OutContext);
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::Bcc)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::Bcc)
       .addExpr(GVSymExpr)
       // Add predicate operands.
       .addImm(ARMCC::AL)
@@ -1239,33 +1289,28 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     TmpInst.addOperand(MCOperand::CreateReg(MI->getOperand(0).getReg()));
 
     unsigned TF = MI->getOperand(1).getTargetFlags();
-    bool isPIC = TF == ARMII::MO_LO16_NONLAZY_PIC;
     const GlobalValue *GV = MI->getOperand(1).getGlobal();
-    MCSymbol *GVSym = GetARMGVSymbol(GV);
+    MCSymbol *GVSym = GetARMGVSymbol(GV, TF);
     const MCExpr *GVSymExpr = MCSymbolRefExpr::Create(GVSym, OutContext);
-    if (isPIC) {
-      MCSymbol *LabelSym = getPICLabel(MAI->getPrivateGlobalPrefix(),
-                                       getFunctionNumber(),
-                                       MI->getOperand(2).getImm(), OutContext);
-      const MCExpr *LabelSymExpr= MCSymbolRefExpr::Create(LabelSym, OutContext);
-      unsigned PCAdj = (Opc == ARM::MOVi16_ga_pcrel) ? 8 : 4;
-      const MCExpr *PCRelExpr =
-        ARMMCExpr::CreateLower16(MCBinaryExpr::CreateSub(GVSymExpr,
-                                  MCBinaryExpr::CreateAdd(LabelSymExpr,
+
+    MCSymbol *LabelSym = getPICLabel(DL->getPrivateGlobalPrefix(),
+                                     getFunctionNumber(),
+                                     MI->getOperand(2).getImm(), OutContext);
+    const MCExpr *LabelSymExpr= MCSymbolRefExpr::Create(LabelSym, OutContext);
+    unsigned PCAdj = (Opc == ARM::MOVi16_ga_pcrel) ? 8 : 4;
+    const MCExpr *PCRelExpr =
+      ARMMCExpr::CreateLower16(MCBinaryExpr::CreateSub(GVSymExpr,
+                                      MCBinaryExpr::CreateAdd(LabelSymExpr,
                                       MCConstantExpr::Create(PCAdj, OutContext),
-                                          OutContext), OutContext), OutContext);
+                                      OutContext), OutContext), OutContext);
       TmpInst.addOperand(MCOperand::CreateExpr(PCRelExpr));
-    } else {
-      const MCExpr *RefExpr= ARMMCExpr::CreateLower16(GVSymExpr, OutContext);
-      TmpInst.addOperand(MCOperand::CreateExpr(RefExpr));
-    }
 
     // Add predicate operands.
     TmpInst.addOperand(MCOperand::CreateImm(ARMCC::AL));
     TmpInst.addOperand(MCOperand::CreateReg(0));
     // Add 's' bit operand (always reg0 for this)
     TmpInst.addOperand(MCOperand::CreateReg(0));
-    OutStreamer.EmitInstruction(TmpInst);
+    EmitToStreamer(OutStreamer, TmpInst);
     return;
   }
   case ARM::MOVTi16_ga_pcrel:
@@ -1277,32 +1322,27 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     TmpInst.addOperand(MCOperand::CreateReg(MI->getOperand(1).getReg()));
 
     unsigned TF = MI->getOperand(2).getTargetFlags();
-    bool isPIC = TF == ARMII::MO_HI16_NONLAZY_PIC;
     const GlobalValue *GV = MI->getOperand(2).getGlobal();
-    MCSymbol *GVSym = GetARMGVSymbol(GV);
+    MCSymbol *GVSym = GetARMGVSymbol(GV, TF);
     const MCExpr *GVSymExpr = MCSymbolRefExpr::Create(GVSym, OutContext);
-    if (isPIC) {
-      MCSymbol *LabelSym = getPICLabel(MAI->getPrivateGlobalPrefix(),
-                                       getFunctionNumber(),
-                                       MI->getOperand(3).getImm(), OutContext);
-      const MCExpr *LabelSymExpr= MCSymbolRefExpr::Create(LabelSym, OutContext);
-      unsigned PCAdj = (Opc == ARM::MOVTi16_ga_pcrel) ? 8 : 4;
-      const MCExpr *PCRelExpr =
+
+    MCSymbol *LabelSym = getPICLabel(DL->getPrivateGlobalPrefix(),
+                                     getFunctionNumber(),
+                                     MI->getOperand(3).getImm(), OutContext);
+    const MCExpr *LabelSymExpr= MCSymbolRefExpr::Create(LabelSym, OutContext);
+    unsigned PCAdj = (Opc == ARM::MOVTi16_ga_pcrel) ? 8 : 4;
+    const MCExpr *PCRelExpr =
         ARMMCExpr::CreateUpper16(MCBinaryExpr::CreateSub(GVSymExpr,
                                    MCBinaryExpr::CreateAdd(LabelSymExpr,
                                       MCConstantExpr::Create(PCAdj, OutContext),
                                           OutContext), OutContext), OutContext);
       TmpInst.addOperand(MCOperand::CreateExpr(PCRelExpr));
-    } else {
-      const MCExpr *RefExpr= ARMMCExpr::CreateUpper16(GVSymExpr, OutContext);
-      TmpInst.addOperand(MCOperand::CreateExpr(RefExpr));
-    }
     // Add predicate operands.
     TmpInst.addOperand(MCOperand::CreateImm(ARMCC::AL));
     TmpInst.addOperand(MCOperand::CreateReg(0));
     // Add 's' bit operand (always reg0 for this)
     TmpInst.addOperand(MCOperand::CreateReg(0));
-    OutStreamer.EmitInstruction(TmpInst);
+    EmitToStreamer(OutStreamer, TmpInst);
     return;
   }
   case ARM::tPICADD: {
@@ -1312,12 +1352,12 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // This adds the address of LPC0 to r0.
 
     // Emit the label.
-    OutStreamer.EmitLabel(getPICLabel(MAI->getPrivateGlobalPrefix(),
+    OutStreamer.EmitLabel(getPICLabel(DL->getPrivateGlobalPrefix(),
                           getFunctionNumber(), MI->getOperand(2).getImm(),
                           OutContext));
 
     // Form and emit the add.
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tADDhirr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tADDhirr)
       .addReg(MI->getOperand(0).getReg())
       .addReg(MI->getOperand(0).getReg())
       .addReg(ARM::PC)
@@ -1333,12 +1373,12 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // This adds the address of LPC0 to r0.
 
     // Emit the label.
-    OutStreamer.EmitLabel(getPICLabel(MAI->getPrivateGlobalPrefix(),
+    OutStreamer.EmitLabel(getPICLabel(DL->getPrivateGlobalPrefix(),
                           getFunctionNumber(), MI->getOperand(2).getImm(),
                           OutContext));
 
     // Form and emit the add.
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::ADDrr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::ADDrr)
       .addReg(MI->getOperand(0).getReg())
       .addReg(ARM::PC)
       .addReg(MI->getOperand(1).getReg())
@@ -1364,7 +1404,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // a PC-relative address at the ldr instruction.
 
     // Emit the label.
-    OutStreamer.EmitLabel(getPICLabel(MAI->getPrivateGlobalPrefix(),
+    OutStreamer.EmitLabel(getPICLabel(DL->getPrivateGlobalPrefix(),
                           getFunctionNumber(), MI->getOperand(2).getImm(),
                           OutContext));
 
@@ -1382,7 +1422,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     case ARM::PICLDRSB: Opcode = ARM::LDRSB; break;
     case ARM::PICLDRSH: Opcode = ARM::LDRSH; break;
     }
-    OutStreamer.EmitInstruction(MCInstBuilder(Opcode)
+    EmitToStreamer(OutStreamer, MCInstBuilder(Opcode)
       .addReg(MI->getOperand(0).getReg())
       .addReg(ARM::PC)
       .addReg(MI->getOperand(1).getReg())
@@ -1419,7 +1459,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   }
   case ARM::t2BR_JT: {
     // Lower and emit the instruction itself, then the jump table following it.
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tMOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tMOVr)
       .addReg(ARM::PC)
       .addReg(MI->getOperand(0).getReg())
       // Add predicate operands.
@@ -1432,7 +1472,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   }
   case ARM::t2TBB_JT: {
     // Lower and emit the instruction itself, then the jump table following it.
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::t2TBB)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::t2TBB)
       .addReg(ARM::PC)
       .addReg(MI->getOperand(0).getReg())
       // Add predicate operands.
@@ -1447,7 +1487,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   }
   case ARM::t2TBH_JT: {
     // Lower and emit the instruction itself, then the jump table following it.
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::t2TBH)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::t2TBH)
       .addReg(ARM::PC)
       .addReg(MI->getOperand(0).getReg())
       // Add predicate operands.
@@ -1474,7 +1514,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // Add 's' bit operand (always reg0 for this)
     if (Opc == ARM::MOVr)
       TmpInst.addOperand(MCOperand::CreateReg(0));
-    OutStreamer.EmitInstruction(TmpInst);
+    EmitToStreamer(OutStreamer, TmpInst);
 
     // Make sure the Thumb jump table is 4-byte aligned.
     if (Opc == ARM::tMOVr)
@@ -1504,7 +1544,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // Add predicate operands.
     TmpInst.addOperand(MCOperand::CreateImm(ARMCC::AL));
     TmpInst.addOperand(MCOperand::CreateReg(0));
-    OutStreamer.EmitInstruction(TmpInst);
+    EmitToStreamer(OutStreamer, TmpInst);
 
     // Output the data for the jump table itself
     EmitJumpTable(MI);
@@ -1513,7 +1553,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   case ARM::BR_JTadd: {
     // Lower and emit the instruction itself, then the jump table following it.
     // add pc, target, idx
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::ADDrr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::ADDrr)
       .addReg(ARM::PC)
       .addReg(MI->getOperand(0).getReg())
       .addReg(MI->getOperand(1).getReg())
@@ -1530,7 +1570,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   case ARM::TRAP: {
     // Non-Darwin binutils don't yet support the "trap" mnemonic.
     // FIXME: Remove this special case when they do.
-    if (!Subtarget->isTargetDarwin()) {
+    if (!Subtarget->isTargetMachO()) {
       //.long 0xe7ffdefe @ trap
       uint32_t Val = 0xe7ffdefeUL;
       OutStreamer.AddComment("trap");
@@ -1549,7 +1589,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   case ARM::tTRAP: {
     // Non-Darwin binutils don't yet support the "trap" mnemonic.
     // FIXME: Remove this special case when they do.
-    if (!Subtarget->isTargetDarwin()) {
+    if (!Subtarget->isTargetMachO()) {
       //.short 57086 @ trap
       uint16_t Val = 0xdefe;
       OutStreamer.AddComment("trap");
@@ -1573,14 +1613,14 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     unsigned ValReg = MI->getOperand(1).getReg();
     MCSymbol *Label = GetARMSJLJEHLabel();
     OutStreamer.AddComment("eh_setjmp begin");
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tMOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tMOVr)
       .addReg(ValReg)
       .addReg(ARM::PC)
       // Predicate.
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tADDi3)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tADDi3)
       .addReg(ValReg)
       // 's' bit operand
       .addReg(ARM::CPSR)
@@ -1590,7 +1630,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tSTRi)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tSTRi)
       .addReg(ValReg)
       .addReg(SrcReg)
       // The offset immediate is #4. The operand value is scaled by 4 for the
@@ -1600,7 +1640,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tMOVi8)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tMOVi8)
       .addReg(ARM::R0)
       .addReg(ARM::CPSR)
       .addImm(0)
@@ -1609,13 +1649,13 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addReg(0));
 
     const MCExpr *SymbolExpr = MCSymbolRefExpr::Create(Label, OutContext);
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tB)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tB)
       .addExpr(SymbolExpr)
       .addImm(ARMCC::AL)
       .addReg(0));
 
     OutStreamer.AddComment("eh_setjmp end");
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tMOVi8)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tMOVi8)
       .addReg(ARM::R0)
       .addReg(ARM::CPSR)
       .addImm(1)
@@ -1639,7 +1679,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     unsigned ValReg = MI->getOperand(1).getReg();
 
     OutStreamer.AddComment("eh_setjmp begin");
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::ADDri)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::ADDri)
       .addReg(ValReg)
       .addReg(ARM::PC)
       .addImm(8)
@@ -1649,7 +1689,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       // 's' bit operand (always reg0 for this).
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::STRi12)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::STRi12)
       .addReg(ValReg)
       .addReg(SrcReg)
       .addImm(4)
@@ -1657,7 +1697,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::MOVi)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::MOVi)
       .addReg(ARM::R0)
       .addImm(0)
       // Predicate.
@@ -1666,7 +1706,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       // 's' bit operand (always reg0 for this).
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::ADDri)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::ADDri)
       .addReg(ARM::PC)
       .addReg(ARM::PC)
       .addImm(0)
@@ -1677,7 +1717,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addReg(0));
 
     OutStreamer.AddComment("eh_setjmp end");
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::MOVi)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::MOVi)
       .addReg(ARM::R0)
       .addImm(1)
       // Predicate.
@@ -1694,7 +1734,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // bx $scratch
     unsigned SrcReg = MI->getOperand(0).getReg();
     unsigned ScratchReg = MI->getOperand(1).getReg();
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::LDRi12)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::LDRi12)
       .addReg(ARM::SP)
       .addReg(SrcReg)
       .addImm(8)
@@ -1702,7 +1742,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::LDRi12)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::LDRi12)
       .addReg(ScratchReg)
       .addReg(SrcReg)
       .addImm(4)
@@ -1710,7 +1750,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::LDRi12)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::LDRi12)
       .addReg(ARM::R7)
       .addReg(SrcReg)
       .addImm(0)
@@ -1718,7 +1758,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::BX)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::BX)
       .addReg(ScratchReg)
       // Predicate.
       .addImm(ARMCC::AL)
@@ -1733,7 +1773,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     // bx $scratch
     unsigned SrcReg = MI->getOperand(0).getReg();
     unsigned ScratchReg = MI->getOperand(1).getReg();
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tLDRi)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tLDRi)
       .addReg(ScratchReg)
       .addReg(SrcReg)
       // The offset immediate is #8. The operand value is scaled by 4 for the
@@ -1743,14 +1783,14 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tMOVr)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tMOVr)
       .addReg(ARM::SP)
       .addReg(ScratchReg)
       // Predicate.
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tLDRi)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tLDRi)
       .addReg(ScratchReg)
       .addReg(SrcReg)
       .addImm(1)
@@ -1758,7 +1798,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tLDRi)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tLDRi)
       .addReg(ARM::R7)
       .addReg(SrcReg)
       .addImm(0)
@@ -1766,7 +1806,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
       .addImm(ARMCC::AL)
       .addReg(0));
 
-    OutStreamer.EmitInstruction(MCInstBuilder(ARM::tBX)
+    EmitToStreamer(OutStreamer, MCInstBuilder(ARM::tBX)
       .addReg(ScratchReg)
       // Predicate.
       .addImm(ARMCC::AL)
@@ -1778,7 +1818,7 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
   MCInst TmpInst;
   LowerARMMachineInstrToMCInst(MI, TmpInst, *this);
 
-  OutStreamer.EmitInstruction(TmpInst);
+  EmitToStreamer(OutStreamer, TmpInst);
 }
 
 //===----------------------------------------------------------------------===//
@@ -1787,6 +1827,8 @@ void ARMAsmPrinter::EmitInstruction(const MachineInstr *MI) {
 
 // Force static initialization.
 extern "C" void LLVMInitializeARMAsmPrinter() {
-  RegisterAsmPrinter<ARMAsmPrinter> X(TheARMTarget);
-  RegisterAsmPrinter<ARMAsmPrinter> Y(TheThumbTarget);
+  RegisterAsmPrinter<ARMAsmPrinter> X(TheARMLETarget);
+  RegisterAsmPrinter<ARMAsmPrinter> Y(TheARMBETarget);
+  RegisterAsmPrinter<ARMAsmPrinter> A(TheThumbLETarget);
+  RegisterAsmPrinter<ARMAsmPrinter> B(TheThumbBETarget);
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.h b/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.h
index de72e063e0d5..7c103c6763f2 100644
--- a/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.h
+++ b/contrib/llvm/lib/Target/ARM/ARMAsmPrinter.h
@@ -10,14 +10,16 @@
 #ifndef ARMASMPRINTER_H
 #define ARMASMPRINTER_H
 
-#include "ARM.h"
-#include "ARMTargetMachine.h"
+#include "ARMSubtarget.h"
 #include "llvm/CodeGen/AsmPrinter.h"
-#include "llvm/Support/Compiler.h"
+#include "llvm/Target/TargetMachine.h"
 
 namespace llvm {
 
+class ARMFunctionInfo;
 class MCOperand;
+class MachineConstantPool;
+class MachineOperand;
 
 namespace ARM {
   enum DW_ISA {
@@ -45,37 +47,41 @@ class LLVM_LIBRARY_VISIBILITY ARMAsmPrinter : public AsmPrinter {
   bool InConstantPool;
 public:
   explicit ARMAsmPrinter(TargetMachine &TM, MCStreamer &Streamer)
-    : AsmPrinter(TM, Streamer), AFI(NULL), MCP(NULL), InConstantPool(false) {
-      Subtarget = &TM.getSubtarget<ARMSubtarget>();
-    }
+    : AsmPrinter(TM, Streamer), AFI(nullptr), MCP(nullptr),
+      InConstantPool(false) {
+    Subtarget = &TM.getSubtarget<ARMSubtarget>();
+  }
 
-  virtual const char *getPassName() const LLVM_OVERRIDE {
+  const char *getPassName() const override {
     return "ARM Assembly / Object Emitter";
   }
 
   void printOperand(const MachineInstr *MI, int OpNum, raw_ostream &O,
-                    const char *Modifier = 0);
+                    const char *Modifier = nullptr);
+
+  bool PrintAsmOperand(const MachineInstr *MI, unsigned OpNum,
+                       unsigned AsmVariant, const char *ExtraCode,
+                       raw_ostream &O) override;
+  bool PrintAsmMemoryOperand(const MachineInstr *MI, unsigned OpNum,
+                             unsigned AsmVariant, const char *ExtraCode,
+                             raw_ostream &O) override;
 
-  virtual bool PrintAsmOperand(const MachineInstr *MI, unsigned OpNum,
-                               unsigned AsmVariant, const char *ExtraCode,
-                               raw_ostream &O) LLVM_OVERRIDE;
-  virtual bool PrintAsmMemoryOperand(const MachineInstr *MI, unsigned OpNum,
-                                     unsigned AsmVariant, const char *ExtraCode,
-                                     raw_ostream &O) LLVM_OVERRIDE;
+  void emitInlineAsmEnd(const MCSubtargetInfo &StartInfo,
+                        const MCSubtargetInfo *EndInfo) const override;
 
   void EmitJumpTable(const MachineInstr *MI);
   void EmitJump2Table(const MachineInstr *MI);
-  virtual void EmitInstruction(const MachineInstr *MI) LLVM_OVERRIDE;
-  virtual bool runOnMachineFunction(MachineFunction &F) LLVM_OVERRIDE;
+  void EmitInstruction(const MachineInstr *MI) override;
+  bool runOnMachineFunction(MachineFunction &F) override;
 
-  virtual void EmitConstantPool() LLVM_OVERRIDE {
+  void EmitConstantPool() override {
     // we emit constant pools customly!
   }
-  virtual void EmitFunctionBodyEnd() LLVM_OVERRIDE;
-  virtual void EmitFunctionEntryLabel() LLVM_OVERRIDE;
-  virtual void EmitStartOfAsmFile(Module &M) LLVM_OVERRIDE;
-  virtual void EmitEndOfAsmFile(Module &M) LLVM_OVERRIDE;
-  virtual void EmitXXStructor(const Constant *CV) LLVM_OVERRIDE;
+  void EmitFunctionBodyEnd() override;
+  void EmitFunctionEntryLabel() override;
+  void EmitStartOfAsmFile(Module &M) override;
+  void EmitEndOfAsmFile(Module &M) override;
+  void EmitXXStructor(const Constant *CV) override;
 
   // lowerOperand - Convert a MachineOperand into the equivalent MCOperand.
   bool lowerOperand(const MachineOperand &MO, MCOperand &MCOp);
@@ -84,9 +90,6 @@ private:
   // Helpers for EmitStartOfAsmFile() and EmitEndOfAsmFile()
   void emitAttributes();
 
-  // Helper for ELF .o only
-  void emitARMAttributeSection();
-
   // Generic helper used to emit e.g. ARMv5 mul pseudos
   void EmitPatchedInstruction(const MachineInstr *MI, unsigned TargetOpc);
 
@@ -97,13 +100,9 @@ private:
                                    const MachineInstr *MI);
 
 public:
-  /// EmitDwarfRegOp - Emit dwarf register operation.
-  virtual void EmitDwarfRegOp(const MachineLocation &MLoc, bool Indirect) const
-      LLVM_OVERRIDE;
-
-  virtual unsigned getISAEncoding() LLVM_OVERRIDE {
+  unsigned getISAEncoding() override {
     // ARM/Darwin adds ISA to the DWARF info for each function.
-    if (!Subtarget->isTargetDarwin())
+    if (!Subtarget->isTargetMachO())
       return 0;
     return Subtarget->isThumb() ?
       ARM::DW_ISA_ARM_thumb : ARM::DW_ISA_ARM_arm;
@@ -115,13 +114,12 @@ private:
 
   MCSymbol *GetARMSJLJEHLabel() const;
 
-  MCSymbol *GetARMGVSymbol(const GlobalValue *GV);
+  MCSymbol *GetARMGVSymbol(const GlobalValue *GV, unsigned char TargetFlags);
 
 public:
   /// EmitMachineConstantPoolValue - Print a machine constantpool value to
   /// the .s file.
-  virtual void
-    EmitMachineConstantPoolValue(MachineConstantPoolValue *MCPV) LLVM_OVERRIDE;
+  void EmitMachineConstantPoolValue(MachineConstantPoolValue *MCPV) override;
 };
 } // end namespace llvm
 
diff --git a/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
index 658af8380d8d..0288db91dcbf 100644
--- a/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
@@ -32,16 +32,19 @@
 #include "llvm/IR/Function.h"
 #include "llvm/IR/GlobalValue.h"
 #include "llvm/MC/MCAsmInfo.h"
+#include "llvm/MC/MCExpr.h"
 #include "llvm/Support/BranchProbability.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 
+using namespace llvm;
+
+#define DEBUG_TYPE "arm-instrinfo"
+
 #define GET_INSTRINFO_CTOR_DTOR
 #include "ARMGenInstrInfo.inc"
 
-using namespace llvm;
-
 static cl::opt<bool>
 EnableARM3Addr("enable-arm-3-addr-conv", cl::Hidden,
                cl::desc("Enable ARM 2-addr to 3-addr conv"));
@@ -100,14 +103,15 @@ ARMBaseInstrInfo::ARMBaseInstrInfo(const ARMSubtarget& STI)
 
 // Use a ScoreboardHazardRecognizer for prepass ARM scheduling. TargetInstrImpl
 // currently defaults to no prepass hazard recognizer.
-ScheduleHazardRecognizer *ARMBaseInstrInfo::
-CreateTargetHazardRecognizer(const TargetMachine *TM,
-                             const ScheduleDAG *DAG) const {
+ScheduleHazardRecognizer *
+ARMBaseInstrInfo::CreateTargetHazardRecognizer(const TargetSubtargetInfo *STI,
+                                               const ScheduleDAG *DAG) const {
   if (usePreRAHazardRecognizer()) {
-    const InstrItineraryData *II = TM->getInstrItineraryData();
+    const InstrItineraryData *II =
+        &static_cast<const ARMSubtarget *>(STI)->getInstrItineraryData();
     return new ScoreboardHazardRecognizer(II, DAG, "pre-RA-sched");
   }
-  return TargetInstrInfo::CreateTargetHazardRecognizer(TM, DAG);
+  return TargetInstrInfo::CreateTargetHazardRecognizer(STI, DAG);
 }
 
 ScheduleHazardRecognizer *ARMBaseInstrInfo::
@@ -125,14 +129,14 @@ ARMBaseInstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI,
   // FIXME: Thumb2 support.
 
   if (!EnableARM3Addr)
-    return NULL;
+    return nullptr;
 
   MachineInstr *MI = MBBI;
   MachineFunction &MF = *MI->getParent()->getParent();
   uint64_t TSFlags = MI->getDesc().TSFlags;
   bool isPre = false;
   switch ((TSFlags & ARMII::IndexModeMask) >> ARMII::IndexModeShift) {
-  default: return NULL;
+  default: return nullptr;
   case ARMII::IndexModePre:
     isPre = true;
     break;
@@ -144,10 +148,10 @@ ARMBaseInstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI,
   // operation.
   unsigned MemOpc = getUnindexedOpcode(MI->getOpcode());
   if (MemOpc == 0)
-    return NULL;
+    return nullptr;
 
-  MachineInstr *UpdateMI = NULL;
-  MachineInstr *MemMI = NULL;
+  MachineInstr *UpdateMI = nullptr;
+  MachineInstr *MemMI = nullptr;
   unsigned AddrMode = (TSFlags & ARMII::AddrModeMask);
   const MCInstrDesc &MCID = MI->getDesc();
   unsigned NumOps = MCID.getNumOperands();
@@ -169,7 +173,7 @@ ARMBaseInstrInfo::convertToThreeAddress(MachineFunction::iterator &MFI,
       if (ARM_AM::getSOImmVal(Amt) == -1)
         // Can't encode it in a so_imm operand. This transformation will
         // add more than 1 instruction. Abandon!
-        return NULL;
+        return nullptr;
       UpdateMI = BuildMI(MF, MI->getDebugLoc(),
                          get(isSub ? ARM::SUBri : ARM::ADDri), WBReg)
         .addReg(BaseReg).addImm(Amt)
@@ -273,8 +277,8 @@ ARMBaseInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
                                 MachineBasicBlock *&FBB,
                                 SmallVectorImpl<MachineOperand> &Cond,
                                 bool AllowModify) const {
-  TBB = 0;
-  FBB = 0;
+  TBB = nullptr;
+  FBB = nullptr;
 
   MachineBasicBlock::iterator I = MBB.end();
   if (I == MBB.begin())
@@ -283,7 +287,7 @@ ARMBaseInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
 
   // Walk backwards from the end of the basic block until the branch is
   // analyzed or we give up.
-  while (isPredicated(I) || I->isTerminator()) {
+  while (isPredicated(I) || I->isTerminator() || I->isDebugValue()) {
 
     // Flag to be raised on unanalyzeable instructions. This is useful in cases
     // where we want to clean up on the end of the basic block before we bail
@@ -331,12 +335,12 @@ ARMBaseInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
            I->isReturn())) {
       // Forget any previous condition branch information - it no longer applies.
       Cond.clear();
-      FBB = 0;
+      FBB = nullptr;
 
       // If we can modify the function, delete everything below this
       // unconditional branch.
       if (AllowModify) {
-        MachineBasicBlock::iterator DI = llvm::next(I);
+        MachineBasicBlock::iterator DI = std::next(I);
         while (DI != MBB.end()) {
           MachineInstr *InstToDelete = DI;
           ++DI;
@@ -405,7 +409,7 @@ ARMBaseInstrInfo::InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
   assert((Cond.size() == 2 || Cond.size() == 0) &&
          "ARM branch conditions have two components!");
 
-  if (FBB == 0) {
+  if (!FBB) {
     if (Cond.empty()) { // Unconditional branch?
       if (isThumb)
         BuildMI(&MBB, DL, get(BOpc)).addMBB(TBB).addImm(ARMCC::AL).addReg(0);
@@ -535,6 +539,22 @@ bool ARMBaseInstrInfo::isPredicable(MachineInstr *MI) const {
   return true;
 }
 
+namespace llvm {
+template <> bool IsCPSRDead<MachineInstr>(MachineInstr *MI) {
+  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI->getOperand(i);
+    if (!MO.isReg() || MO.isUndef() || MO.isUse())
+      continue;
+    if (MO.getReg() != ARM::CPSR)
+      continue;
+    if (!MO.isDead())
+      return false;
+  }
+  // all definitions of CPSR are dead
+  return true;
+}
+}
+
 /// FIXME: Works around a gcc miscompilation with -fstrict-aliasing.
 LLVM_ATTRIBUTE_NOINLINE
 static unsigned getNumJTEntries(const std::vector<MachineJumpTableEntry> &JT,
@@ -559,15 +579,10 @@ unsigned ARMBaseInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
   // If this machine instr is an inline asm, measure it.
   if (MI->getOpcode() == ARM::INLINEASM)
     return getInlineAsmLength(MI->getOperand(0).getSymbolName(), *MAI);
-  if (MI->isLabel())
-    return 0;
   unsigned Opc = MI->getOpcode();
   switch (Opc) {
-  case TargetOpcode::IMPLICIT_DEF:
-  case TargetOpcode::KILL:
-  case TargetOpcode::PROLOG_LABEL:
-  case TargetOpcode::EH_LABEL:
-  case TargetOpcode::DBG_VALUE:
+  default:
+    // pseudo-instruction sizes are zero.
     return 0;
   case TargetOpcode::BUNDLE:
     return getInstBundleLength(MI);
@@ -611,7 +626,7 @@ unsigned ARMBaseInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
       MI->getOperand(NumOps - (MI->isPredicable() ? 3 : 2));
     unsigned JTI = JTOP.getIndex();
     const MachineJumpTableInfo *MJTI = MF->getJumpTableInfo();
-    assert(MJTI != 0);
+    assert(MJTI != nullptr);
     const std::vector<MachineJumpTableEntry> &JT = MJTI->getJumpTables();
     assert(JTI < JT.size());
     // Thumb instructions are 2 byte aligned, but JT entries are 4 byte
@@ -630,9 +645,6 @@ unsigned ARMBaseInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
       ++NumEntries;
     return NumEntries * EntrySize + InstSize;
   }
-  default:
-    // Otherwise, pseudo-instruction sizes are zero.
-    return 0;
   }
 }
 
@@ -1242,7 +1254,8 @@ static unsigned duplicateCPV(MachineFunction &MF, unsigned &CPI) {
     static_cast<ARMConstantPoolValue*>(MCPE.Val.MachineCPVal);
 
   unsigned PCLabelId = AFI->createPICLabelUId();
-  ARMConstantPoolValue *NewCPV = 0;
+  ARMConstantPoolValue *NewCPV = nullptr;
+
   // FIXME: The below assumes PIC relocation model and that the function
   // is Thumb mode (t1 or t2). PCAdjustment would be 8 for ARM mode PIC, and
   // zero for non-PIC in ARM or Thumb. The callers are all of thumb LDR
@@ -1325,10 +1338,11 @@ bool ARMBaseInstrInfo::produceSameValue(const MachineInstr *MI0,
       Opcode == ARM::t2LDRpci_pic ||
       Opcode == ARM::tLDRpci ||
       Opcode == ARM::tLDRpci_pic ||
-      Opcode == ARM::MOV_ga_dyn ||
+      Opcode == ARM::LDRLIT_ga_pcrel ||
+      Opcode == ARM::LDRLIT_ga_pcrel_ldr ||
+      Opcode == ARM::tLDRLIT_ga_pcrel ||
       Opcode == ARM::MOV_ga_pcrel ||
       Opcode == ARM::MOV_ga_pcrel_ldr ||
-      Opcode == ARM::t2MOV_ga_dyn ||
       Opcode == ARM::t2MOV_ga_pcrel) {
     if (MI1->getOpcode() != Opcode)
       return false;
@@ -1340,10 +1354,11 @@ bool ARMBaseInstrInfo::produceSameValue(const MachineInstr *MI0,
     if (MO0.getOffset() != MO1.getOffset())
       return false;
 
-    if (Opcode == ARM::MOV_ga_dyn ||
+    if (Opcode == ARM::LDRLIT_ga_pcrel ||
+        Opcode == ARM::LDRLIT_ga_pcrel_ldr ||
+        Opcode == ARM::tLDRLIT_ga_pcrel ||
         Opcode == ARM::MOV_ga_pcrel ||
         Opcode == ARM::MOV_ga_pcrel_ldr ||
-        Opcode == ARM::t2MOV_ga_dyn ||
         Opcode == ARM::t2MOV_ga_pcrel)
       // Ignore the PC labels.
       return MO0.getGlobal() == MO1.getGlobal();
@@ -1534,7 +1549,7 @@ bool ARMBaseInstrInfo::isSchedulingBoundary(const MachineInstr *MI,
     return false;
 
   // Terminators and labels can't be scheduled around.
-  if (MI->isTerminator() || MI->isLabel())
+  if (MI->isTerminator() || MI->isPosition())
     return true;
 
   // Treat the start of the IT block as a scheduling boundary, but schedule
@@ -1650,10 +1665,10 @@ ARMBaseInstrInfo::commuteInstruction(MachineInstr *MI, bool NewMI) const {
     ARMCC::CondCodes CC = getInstrPredicate(MI, PredReg);
     // MOVCC AL can't be inverted. Shouldn't happen.
     if (CC == ARMCC::AL || PredReg != ARM::CPSR)
-      return NULL;
+      return nullptr;
     MI = TargetInstrInfo::commuteInstruction(MI, NewMI);
     if (!MI)
-      return NULL;
+      return nullptr;
     // After swapping the MOVCC operands, also invert the condition.
     MI->getOperand(MI->findFirstPredOperandIdx())
       .setImm(ARMCC::getOppositeCondition(CC));
@@ -1669,35 +1684,36 @@ static MachineInstr *canFoldIntoMOVCC(unsigned Reg,
                                       const MachineRegisterInfo &MRI,
                                       const TargetInstrInfo *TII) {
   if (!TargetRegisterInfo::isVirtualRegister(Reg))
-    return 0;
+    return nullptr;
   if (!MRI.hasOneNonDBGUse(Reg))
-    return 0;
+    return nullptr;
   MachineInstr *MI = MRI.getVRegDef(Reg);
   if (!MI)
-    return 0;
+    return nullptr;
   // MI is folded into the MOVCC by predicating it.
   if (!MI->isPredicable())
-    return 0;
+    return nullptr;
   // Check if MI has any non-dead defs or physreg uses. This also detects
   // predicated instructions which will be reading CPSR.
   for (unsigned i = 1, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
     // Reject frame index operands, PEI can't handle the predicated pseudos.
     if (MO.isFI() || MO.isCPI() || MO.isJTI())
-      return 0;
+      return nullptr;
     if (!MO.isReg())
       continue;
     // MI can't have any tied operands, that would conflict with predication.
     if (MO.isTied())
-      return 0;
+      return nullptr;
     if (TargetRegisterInfo::isPhysicalRegister(MO.getReg()))
-      return 0;
+      return nullptr;
     if (MO.isDef() && !MO.isDead())
-      return 0;
+      return nullptr;
   }
   bool DontMoveAcrossStores = true;
-  if (!MI->isSafeToMove(TII, /* AliasAnalysis = */ 0, DontMoveAcrossStores))
-    return 0;
+  if (!MI->isSafeToMove(TII, /* AliasAnalysis = */ nullptr,
+                        DontMoveAcrossStores))
+    return nullptr;
   return MI;
 }
 
@@ -1732,14 +1748,14 @@ MachineInstr *ARMBaseInstrInfo::optimizeSelect(MachineInstr *MI,
   if (!DefMI)
     DefMI = canFoldIntoMOVCC(MI->getOperand(1).getReg(), MRI, this);
   if (!DefMI)
-    return 0;
+    return nullptr;
 
   // Find new register class to use.
   MachineOperand FalseReg = MI->getOperand(Invert ? 2 : 1);
   unsigned       DestReg  = MI->getOperand(0).getReg();
   const TargetRegisterClass *PreviousClass = MRI.getRegClass(FalseReg.getReg());
   if (!MRI.constrainRegClass(DestReg, PreviousClass))
-    return 0;
+    return nullptr;
 
   // Create a new predicated version of DefMI.
   // Rfalse is the first use.
@@ -1857,12 +1873,22 @@ void llvm::emitARMRegPlusImmediate(MachineBasicBlock &MBB,
   }
 }
 
-bool llvm::tryFoldSPUpdateIntoPushPop(MachineFunction &MF,
-                                      MachineInstr *MI,
+static bool isAnySubRegLive(unsigned Reg, const TargetRegisterInfo *TRI,
+                      MachineInstr *MI) {
+  for (MCSubRegIterator Subreg(Reg, TRI, /* IncludeSelf */ true);
+       Subreg.isValid(); ++Subreg)
+    if (MI->getParent()->computeRegisterLiveness(TRI, *Subreg, MI) !=
+        MachineBasicBlock::LQR_Dead)
+      return true;
+  return false;
+}
+bool llvm::tryFoldSPUpdateIntoPushPop(const ARMSubtarget &Subtarget,
+                                      MachineFunction &MF, MachineInstr *MI,
                                       unsigned NumBytes) {
   // This optimisation potentially adds lots of load and store
   // micro-operations, it's only really a great benefit to code-size.
-  if (!MF.getFunction()->hasFnAttribute(Attribute::MinSize))
+  if (!MF.getFunction()->getAttributes().hasAttribute(
+          AttributeSet::FunctionIndex, Attribute::MinSize))
     return false;
 
   // If only one register is pushed/popped, LLVM can use an LDR/STR
@@ -1911,7 +1937,6 @@ bool llvm::tryFoldSPUpdateIntoPushPop(MachineFunction &MF,
   for (int i = MI->getNumOperands() - 1; i >= RegListIdx; --i)
     RegList.push_back(MI->getOperand(i));
 
-  MachineBasicBlock *MBB = MI->getParent();
   const TargetRegisterInfo *TRI = MF.getRegInfo().getTargetRegisterInfo();
   const MCPhysReg *CSRegs = TRI->getCalleeSavedRegs(&MF);
 
@@ -1932,9 +1957,11 @@ bool llvm::tryFoldSPUpdateIntoPushPop(MachineFunction &MF,
     // registers live within the function we might clobber a return value
     // register; the other way a register can be live here is if it's
     // callee-saved.
+    // TODO: Currently, computeRegisterLiveness() does not report "live" if a
+    // sub reg is live. When computeRegisterLiveness() works for sub reg, it
+    // can replace isAnySubRegLive().
     if (isCalleeSavedRegister(CurReg, CSRegs) ||
-        MBB->computeRegisterLiveness(TRI, CurReg, MI) !=
-            MachineBasicBlock::LQR_Dead) {
+        isAnySubRegLive(CurReg, TRI, MI)) {
       // VFP pops don't allow holes in the register list, so any skip is fatal
       // for our transformation. GPR pops do, so we should just keep looking.
       if (IsVFPPushPop)
@@ -2159,7 +2186,7 @@ static bool isSuitableForMask(MachineInstr *&MI, unsigned SrcReg,
       // Walk down one instruction which is potentially an 'and'.
       const MachineInstr &Copy = *MI;
       MachineBasicBlock::iterator AND(
-        llvm::next(MachineBasicBlock::iterator(MI)));
+        std::next(MachineBasicBlock::iterator(MI)));
       if (AND == MI->getParent()->end()) return false;
       MI = AND;
       return isSuitableForMask(MI, Copy.getOperand(0).getReg(),
@@ -2235,9 +2262,10 @@ optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg, unsigned SrcReg2,
   // Masked compares sometimes use the same register as the corresponding 'and'.
   if (CmpMask != ~0) {
     if (!isSuitableForMask(MI, SrcReg, CmpMask, false) || isPredicated(MI)) {
-      MI = 0;
-      for (MachineRegisterInfo::use_iterator UI = MRI->use_begin(SrcReg),
-           UE = MRI->use_end(); UI != UE; ++UI) {
+      MI = nullptr;
+      for (MachineRegisterInfo::use_instr_iterator
+           UI = MRI->use_instr_begin(SrcReg), UE = MRI->use_instr_end();
+           UI != UE; ++UI) {
         if (UI->getParent() != CmpInstr->getParent()) continue;
         MachineInstr *PotentialAND = &*UI;
         if (!isSuitableForMask(PotentialAND, SrcReg, CmpMask, true) ||
@@ -2261,17 +2289,17 @@ optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg, unsigned SrcReg2,
   // One is MI, the other is a SUB instruction.
   // For CMPrr(r1,r2), we are looking for SUB(r1,r2) or SUB(r2,r1).
   // For CMPri(r1, CmpValue), we are looking for SUBri(r1, CmpValue).
-  MachineInstr *Sub = NULL;
+  MachineInstr *Sub = nullptr;
   if (SrcReg2 != 0)
     // MI is not a candidate for CMPrr.
-    MI = NULL;
+    MI = nullptr;
   else if (MI->getParent() != CmpInstr->getParent() || CmpValue != 0) {
     // Conservatively refuse to convert an instruction which isn't in the same
     // BB as the comparison.
     // For CMPri, we need to check Sub, thus we can't return here.
     if (CmpInstr->getOpcode() == ARM::CMPri ||
        CmpInstr->getOpcode() == ARM::t2CMPri)
-      MI = NULL;
+      MI = nullptr;
     else
       return false;
   }
@@ -2947,7 +2975,7 @@ ARMBaseInstrInfo::getNumMicroOps(const InstrItineraryData *ItinData,
         break;
       }
       return UOps;
-    } else if (Subtarget.isCortexA8()) {
+    } else if (Subtarget.isCortexA8() || Subtarget.isCortexA7()) {
       if (NumRegs < 4)
         return 2;
       // 4 registers would be issued: 2, 2.
@@ -2984,7 +3012,7 @@ ARMBaseInstrInfo::getVLDMDefCycle(const InstrItineraryData *ItinData,
     return ItinData->getOperandCycle(DefClass, DefIdx);
 
   int DefCycle;
-  if (Subtarget.isCortexA8()) {
+  if (Subtarget.isCortexA8() || Subtarget.isCortexA7()) {
     // (regno / 2) + (regno % 2) + 1
     DefCycle = RegNo / 2 + 1;
     if (RegNo % 2)
@@ -3025,7 +3053,7 @@ ARMBaseInstrInfo::getLDMDefCycle(const InstrItineraryData *ItinData,
     return ItinData->getOperandCycle(DefClass, DefIdx);
 
   int DefCycle;
-  if (Subtarget.isCortexA8()) {
+  if (Subtarget.isCortexA8() || Subtarget.isCortexA7()) {
     // 4 registers would be issued: 1, 2, 1.
     // 5 registers would be issued: 1, 2, 2.
     DefCycle = RegNo / 2;
@@ -3059,7 +3087,7 @@ ARMBaseInstrInfo::getVSTMUseCycle(const InstrItineraryData *ItinData,
     return ItinData->getOperandCycle(UseClass, UseIdx);
 
   int UseCycle;
-  if (Subtarget.isCortexA8()) {
+  if (Subtarget.isCortexA8() || Subtarget.isCortexA7()) {
     // (regno / 2) + (regno % 2) + 1
     UseCycle = RegNo / 2 + 1;
     if (RegNo % 2)
@@ -3099,7 +3127,7 @@ ARMBaseInstrInfo::getSTMUseCycle(const InstrItineraryData *ItinData,
     return ItinData->getOperandCycle(UseClass, UseIdx);
 
   int UseCycle;
-  if (Subtarget.isCortexA8()) {
+  if (Subtarget.isCortexA8() || Subtarget.isCortexA7()) {
     UseCycle = RegNo / 2;
     if (UseCycle < 2)
       UseCycle = 2;
@@ -3236,8 +3264,7 @@ static const MachineInstr *getBundledDefMI(const TargetRegisterInfo *TRI,
   Dist = 0;
 
   MachineBasicBlock::const_iterator I = MI; ++I;
-  MachineBasicBlock::const_instr_iterator II =
-    llvm::prior(I.getInstrIterator());
+  MachineBasicBlock::const_instr_iterator II = std::prev(I.getInstrIterator());
   assert(II->isInsideBundle() && "Empty bundle?");
 
   int Idx = -1;
@@ -3276,7 +3303,7 @@ static const MachineInstr *getBundledUseMI(const TargetRegisterInfo *TRI,
 
   if (Idx == -1) {
     Dist = 0;
-    return 0;
+    return nullptr;
   }
 
   UseIdx = Idx;
@@ -3290,7 +3317,7 @@ static int adjustDefLatency(const ARMSubtarget &Subtarget,
                             const MachineInstr *DefMI,
                             const MCInstrDesc *DefMCID, unsigned DefAlign) {
   int Adjust = 0;
-  if (Subtarget.isCortexA8() || Subtarget.isLikeA9()) {
+  if (Subtarget.isCortexA8() || Subtarget.isLikeA9() || Subtarget.isCortexA7()) {
     // FIXME: Shifter op hack: no shift (i.e. [r +/- r]) or [r + r << 2]
     // variants are one cycle cheaper.
     switch (DefMCID->getOpcode()) {
@@ -3591,7 +3618,8 @@ ARMBaseInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
                                   UseMCID, UseIdx, UseAlign);
 
   if (Latency > 1 &&
-      (Subtarget.isCortexA8() || Subtarget.isLikeA9())) {
+      (Subtarget.isCortexA8() || Subtarget.isLikeA9() ||
+       Subtarget.isCortexA7())) {
     // FIXME: Shifter op hack: no shift (i.e. [r +/- r]) or [r + r << 2]
     // variants are one cycle cheaper.
     switch (DefMCID.getOpcode()) {
@@ -4333,6 +4361,29 @@ breakPartialRegDependency(MachineBasicBlock::iterator MI,
   MI->addRegisterKilled(DReg, TRI, true);
 }
 
+void ARMBaseInstrInfo::getUnconditionalBranch(
+    MCInst &Branch, const MCSymbolRefExpr *BranchTarget) const {
+  if (Subtarget.isThumb())
+    Branch.setOpcode(ARM::tB);
+  else if (Subtarget.isThumb2())
+    Branch.setOpcode(ARM::t2B);
+  else
+    Branch.setOpcode(ARM::Bcc);
+
+  Branch.addOperand(MCOperand::CreateExpr(BranchTarget));
+  Branch.addOperand(MCOperand::CreateImm(ARMCC::AL));
+  Branch.addOperand(MCOperand::CreateReg(0));
+}
+
+void ARMBaseInstrInfo::getTrap(MCInst &MI) const {
+  if (Subtarget.isThumb())
+    MI.setOpcode(ARM::tTRAP);
+  else if (Subtarget.useNaClTrap())
+    MI.setOpcode(ARM::TRAPNaCl);
+  else
+    MI.setOpcode(ARM::TRAP);
+}
+
 bool ARMBaseInstrInfo::hasNOP() const {
   return (Subtarget.getFeatureBits() & ARM::HasV6T2Ops) != 0;
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.h b/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.h
index 93e59647d220..b8d675806b25 100644
--- a/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMBaseInstrInfo.h
@@ -14,7 +14,7 @@
 #ifndef ARMBASEINSTRUCTIONINFO_H
 #define ARMBASEINSTRUCTIONINFO_H
 
-#include "ARM.h"
+#include "MCTargetDesc/ARMBaseInfo.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/SmallSet.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
@@ -42,37 +42,37 @@ public:
   // if there is not such an opcode.
   virtual unsigned getUnindexedOpcode(unsigned Opc) const =0;
 
-  virtual MachineInstr *convertToThreeAddress(MachineFunction::iterator &MFI,
-                                              MachineBasicBlock::iterator &MBBI,
-                                              LiveVariables *LV) const;
+  MachineInstr *convertToThreeAddress(MachineFunction::iterator &MFI,
+                                      MachineBasicBlock::iterator &MBBI,
+                                      LiveVariables *LV) const override;
 
   virtual const ARMBaseRegisterInfo &getRegisterInfo() const = 0;
   const ARMSubtarget &getSubtarget() const { return Subtarget; }
 
   ScheduleHazardRecognizer *
-  CreateTargetHazardRecognizer(const TargetMachine *TM,
-                               const ScheduleDAG *DAG) const;
+  CreateTargetHazardRecognizer(const TargetSubtargetInfo *STI,
+                               const ScheduleDAG *DAG) const override;
 
   ScheduleHazardRecognizer *
   CreateTargetPostRAHazardRecognizer(const InstrItineraryData *II,
-                                     const ScheduleDAG *DAG) const;
+                                     const ScheduleDAG *DAG) const override;
 
   // Branch analysis.
-  virtual bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
-                             MachineBasicBlock *&FBB,
-                             SmallVectorImpl<MachineOperand> &Cond,
-                             bool AllowModify = false) const;
-  virtual unsigned RemoveBranch(MachineBasicBlock &MBB) const;
-  virtual unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
-                                MachineBasicBlock *FBB,
-                                const SmallVectorImpl<MachineOperand> &Cond,
-                                DebugLoc DL) const;
-
-  virtual
-  bool ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const;
+  bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
+                     MachineBasicBlock *&FBB,
+                     SmallVectorImpl<MachineOperand> &Cond,
+                     bool AllowModify = false) const override;
+  unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
+  unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
+                        MachineBasicBlock *FBB,
+                        const SmallVectorImpl<MachineOperand> &Cond,
+                        DebugLoc DL) const override;
+
+  bool
+  ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const override;
 
   // Predication support.
-  bool isPredicated(const MachineInstr *MI) const;
+  bool isPredicated(const MachineInstr *MI) const override;
 
   ARMCC::CondCodes getPredicate(const MachineInstr *MI) const {
     int PIdx = MI->findFirstPredOperandIdx();
@@ -80,76 +80,73 @@ public:
                       : ARMCC::AL;
   }
 
-  virtual
   bool PredicateInstruction(MachineInstr *MI,
-                            const SmallVectorImpl<MachineOperand> &Pred) const;
+                    const SmallVectorImpl<MachineOperand> &Pred) const override;
 
-  virtual
   bool SubsumesPredicate(const SmallVectorImpl<MachineOperand> &Pred1,
-                         const SmallVectorImpl<MachineOperand> &Pred2) const;
+                   const SmallVectorImpl<MachineOperand> &Pred2) const override;
 
-  virtual bool DefinesPredicate(MachineInstr *MI,
-                                std::vector<MachineOperand> &Pred) const;
+  bool DefinesPredicate(MachineInstr *MI,
+                        std::vector<MachineOperand> &Pred) const override;
 
-  virtual bool isPredicable(MachineInstr *MI) const;
+  bool isPredicable(MachineInstr *MI) const override;
 
   /// GetInstSize - Returns the size of the specified MachineInstr.
   ///
   virtual unsigned GetInstSizeInBytes(const MachineInstr* MI) const;
 
-  virtual unsigned isLoadFromStackSlot(const MachineInstr *MI,
-                                       int &FrameIndex) const;
-  virtual unsigned isStoreToStackSlot(const MachineInstr *MI,
-                                      int &FrameIndex) const;
-  virtual unsigned isLoadFromStackSlotPostFE(const MachineInstr *MI,
-                                             int &FrameIndex) const;
-  virtual unsigned isStoreToStackSlotPostFE(const MachineInstr *MI,
-                                            int &FrameIndex) const;
+  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+                               int &FrameIndex) const override;
+  unsigned isStoreToStackSlot(const MachineInstr *MI,
+                              int &FrameIndex) const override;
+  unsigned isLoadFromStackSlotPostFE(const MachineInstr *MI,
+                                     int &FrameIndex) const override;
+  unsigned isStoreToStackSlotPostFE(const MachineInstr *MI,
+                                    int &FrameIndex) const override;
 
-  virtual void copyPhysReg(MachineBasicBlock &MBB,
-                           MachineBasicBlock::iterator I, DebugLoc DL,
-                           unsigned DestReg, unsigned SrcReg,
-                           bool KillSrc) const;
+  void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator I,
+                   DebugLoc DL, unsigned DestReg, unsigned SrcReg,
+                   bool KillSrc) const override;
 
-  virtual void storeRegToStackSlot(MachineBasicBlock &MBB,
-                                   MachineBasicBlock::iterator MBBI,
-                                   unsigned SrcReg, bool isKill, int FrameIndex,
-                                   const TargetRegisterClass *RC,
-                                   const TargetRegisterInfo *TRI) const;
+  void storeRegToStackSlot(MachineBasicBlock &MBB,
+                           MachineBasicBlock::iterator MBBI,
+                           unsigned SrcReg, bool isKill, int FrameIndex,
+                           const TargetRegisterClass *RC,
+                           const TargetRegisterInfo *TRI) const override;
 
-  virtual void loadRegFromStackSlot(MachineBasicBlock &MBB,
-                                    MachineBasicBlock::iterator MBBI,
-                                    unsigned DestReg, int FrameIndex,
-                                    const TargetRegisterClass *RC,
-                                    const TargetRegisterInfo *TRI) const;
+  void loadRegFromStackSlot(MachineBasicBlock &MBB,
+                            MachineBasicBlock::iterator MBBI,
+                            unsigned DestReg, int FrameIndex,
+                            const TargetRegisterClass *RC,
+                            const TargetRegisterInfo *TRI) const override;
 
-  virtual bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const;
+  bool expandPostRAPseudo(MachineBasicBlock::iterator MI) const override;
 
-  virtual void reMaterialize(MachineBasicBlock &MBB,
-                             MachineBasicBlock::iterator MI,
-                             unsigned DestReg, unsigned SubIdx,
-                             const MachineInstr *Orig,
-                             const TargetRegisterInfo &TRI) const;
+  void reMaterialize(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
+                     unsigned DestReg, unsigned SubIdx,
+                     const MachineInstr *Orig,
+                     const TargetRegisterInfo &TRI) const override;
 
-  MachineInstr *duplicate(MachineInstr *Orig, MachineFunction &MF) const;
+  MachineInstr *duplicate(MachineInstr *Orig,
+                          MachineFunction &MF) const override;
 
-  MachineInstr *commuteInstruction(MachineInstr*, bool=false) const;
+  MachineInstr *commuteInstruction(MachineInstr*,
+                                   bool=false) const override;
 
   const MachineInstrBuilder &AddDReg(MachineInstrBuilder &MIB, unsigned Reg,
                                      unsigned SubIdx, unsigned State,
                                      const TargetRegisterInfo *TRI) const;
 
-  virtual bool produceSameValue(const MachineInstr *MI0,
-                                const MachineInstr *MI1,
-                                const MachineRegisterInfo *MRI) const;
+  bool produceSameValue(const MachineInstr *MI0, const MachineInstr *MI1,
+                        const MachineRegisterInfo *MRI) const override;
 
   /// areLoadsFromSameBasePtr - This is used by the pre-regalloc scheduler to
   /// determine if two loads are loading from the same base address. It should
   /// only return true if the base pointers are the same and the only
   /// differences between the two addresses is the offset. It also returns the
   /// offsets by reference.
-  virtual bool areLoadsFromSameBasePtr(SDNode *Load1, SDNode *Load2,
-                                       int64_t &Offset1, int64_t &Offset2)const;
+  bool areLoadsFromSameBasePtr(SDNode *Load1, SDNode *Load2, int64_t &Offset1,
+                               int64_t &Offset2) const override;
 
   /// shouldScheduleLoadsNear - This is a used by the pre-regalloc scheduler to
   /// determine (in conjunction with areLoadsFromSameBasePtr) if two loads
@@ -159,83 +156,86 @@ public:
   /// from the common base address. It returns true if it decides it's desirable
   /// to schedule the two loads together. "NumLoads" is the number of loads that
   /// have already been scheduled after Load1.
-  virtual bool shouldScheduleLoadsNear(SDNode *Load1, SDNode *Load2,
-                                       int64_t Offset1, int64_t Offset2,
-                                       unsigned NumLoads) const;
-
-  virtual bool isSchedulingBoundary(const MachineInstr *MI,
-                                    const MachineBasicBlock *MBB,
-                                    const MachineFunction &MF) const;
-
-  virtual bool isProfitableToIfCvt(MachineBasicBlock &MBB,
-                                   unsigned NumCycles, unsigned ExtraPredCycles,
-                                   const BranchProbability &Probability) const;
-
-  virtual bool isProfitableToIfCvt(MachineBasicBlock &TMBB,
-                                   unsigned NumT, unsigned ExtraT,
-                                   MachineBasicBlock &FMBB,
-                                   unsigned NumF, unsigned ExtraF,
-                                   const BranchProbability &Probability) const;
-
-  virtual bool isProfitableToDupForIfCvt(MachineBasicBlock &MBB,
-                                         unsigned NumCycles,
-                                         const BranchProbability
-                                         &Probability) const {
+  bool shouldScheduleLoadsNear(SDNode *Load1, SDNode *Load2,
+                               int64_t Offset1, int64_t Offset2,
+                               unsigned NumLoads) const override;
+
+  bool isSchedulingBoundary(const MachineInstr *MI,
+                            const MachineBasicBlock *MBB,
+                            const MachineFunction &MF) const override;
+
+  bool isProfitableToIfCvt(MachineBasicBlock &MBB,
+                           unsigned NumCycles, unsigned ExtraPredCycles,
+                           const BranchProbability &Probability) const override;
+
+  bool isProfitableToIfCvt(MachineBasicBlock &TMBB, unsigned NumT,
+                           unsigned ExtraT, MachineBasicBlock &FMBB,
+                           unsigned NumF, unsigned ExtraF,
+                           const BranchProbability &Probability) const override;
+
+  bool isProfitableToDupForIfCvt(MachineBasicBlock &MBB, unsigned NumCycles,
+                          const BranchProbability &Probability) const override {
     return NumCycles == 1;
   }
 
-  virtual bool isProfitableToUnpredicate(MachineBasicBlock &TMBB,
-                                         MachineBasicBlock &FMBB) const;
+  bool isProfitableToUnpredicate(MachineBasicBlock &TMBB,
+                                 MachineBasicBlock &FMBB) const override;
 
   /// analyzeCompare - For a comparison instruction, return the source registers
   /// in SrcReg and SrcReg2 if having two register operands, and the value it
   /// compares against in CmpValue. Return true if the comparison instruction
   /// can be analyzed.
-  virtual bool analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
-                              unsigned &SrcReg2, int &CmpMask,
-                              int &CmpValue) const;
+  bool analyzeCompare(const MachineInstr *MI, unsigned &SrcReg,
+                      unsigned &SrcReg2, int &CmpMask,
+                      int &CmpValue) const override;
 
   /// optimizeCompareInstr - Convert the instruction to set the zero flag so
   /// that we can remove a "comparison with zero"; Remove a redundant CMP
   /// instruction if the flags can be updated in the same way by an earlier
   /// instruction such as SUB.
-  virtual bool optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg,
-                                    unsigned SrcReg2, int CmpMask, int CmpValue,
-                                    const MachineRegisterInfo *MRI) const;
+  bool optimizeCompareInstr(MachineInstr *CmpInstr, unsigned SrcReg,
+                            unsigned SrcReg2, int CmpMask, int CmpValue,
+                            const MachineRegisterInfo *MRI) const override;
 
-  virtual bool analyzeSelect(const MachineInstr *MI,
-                             SmallVectorImpl<MachineOperand> &Cond,
-                             unsigned &TrueOp, unsigned &FalseOp,
-                             bool &Optimizable) const;
+  bool analyzeSelect(const MachineInstr *MI,
+                     SmallVectorImpl<MachineOperand> &Cond,
+                     unsigned &TrueOp, unsigned &FalseOp,
+                     bool &Optimizable) const override;
 
-  virtual MachineInstr *optimizeSelect(MachineInstr *MI, bool) const;
+  MachineInstr *optimizeSelect(MachineInstr *MI, bool) const override;
 
   /// FoldImmediate - 'Reg' is known to be defined by a move immediate
   /// instruction, try to fold the immediate into the use instruction.
-  virtual bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
-                             unsigned Reg, MachineRegisterInfo *MRI) const;
+  bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
+                     unsigned Reg, MachineRegisterInfo *MRI) const override;
 
-  virtual unsigned getNumMicroOps(const InstrItineraryData *ItinData,
-                                  const MachineInstr *MI) const;
+  unsigned getNumMicroOps(const InstrItineraryData *ItinData,
+                          const MachineInstr *MI) const override;
 
-  virtual
   int getOperandLatency(const InstrItineraryData *ItinData,
                         const MachineInstr *DefMI, unsigned DefIdx,
-                        const MachineInstr *UseMI, unsigned UseIdx) const;
-  virtual
+                        const MachineInstr *UseMI,
+                        unsigned UseIdx) const override;
   int getOperandLatency(const InstrItineraryData *ItinData,
                         SDNode *DefNode, unsigned DefIdx,
-                        SDNode *UseNode, unsigned UseIdx) const;
+                        SDNode *UseNode, unsigned UseIdx) const override;
 
   /// VFP/NEON execution domains.
   std::pair<uint16_t, uint16_t>
-  getExecutionDomain(const MachineInstr *MI) const;
-  void setExecutionDomain(MachineInstr *MI, unsigned Domain) const;
+  getExecutionDomain(const MachineInstr *MI) const override;
+  void setExecutionDomain(MachineInstr *MI, unsigned Domain) const override;
 
   unsigned getPartialRegUpdateClearance(const MachineInstr*, unsigned,
-                                        const TargetRegisterInfo*) const;
+                                      const TargetRegisterInfo*) const override;
   void breakPartialRegDependency(MachineBasicBlock::iterator, unsigned,
-                                 const TargetRegisterInfo *TRI) const;
+                                 const TargetRegisterInfo *TRI) const override;
+
+  void
+  getUnconditionalBranch(MCInst &Branch,
+                         const MCSymbolRefExpr *BranchTarget) const override;
+
+  void getTrap(MCInst &MI) const override;
+
   /// Get the number of addresses by LDM or VLDM or zero for unknown.
   unsigned getNumLDMAddresses(const MachineInstr *MI) const;
 
@@ -264,24 +264,27 @@ private:
                         const MCInstrDesc &UseMCID,
                         unsigned UseIdx, unsigned UseAlign) const;
 
-  unsigned getPredicationCost(const MachineInstr *MI) const;
+  unsigned getPredicationCost(const MachineInstr *MI) const override;
 
   unsigned getInstrLatency(const InstrItineraryData *ItinData,
                            const MachineInstr *MI,
-                           unsigned *PredCost = 0) const;
+                           unsigned *PredCost = nullptr) const override;
 
   int getInstrLatency(const InstrItineraryData *ItinData,
-                      SDNode *Node) const;
+                      SDNode *Node) const override;
 
   bool hasHighOperandLatency(const InstrItineraryData *ItinData,
                              const MachineRegisterInfo *MRI,
                              const MachineInstr *DefMI, unsigned DefIdx,
-                             const MachineInstr *UseMI, unsigned UseIdx) const;
+                             const MachineInstr *UseMI,
+                             unsigned UseIdx) const override;
   bool hasLowDefLatency(const InstrItineraryData *ItinData,
-                        const MachineInstr *DefMI, unsigned DefIdx) const;
+                        const MachineInstr *DefMI,
+                        unsigned DefIdx) const override;
 
   /// verifyInstruction - Perform target specific instruction verification.
-  bool verifyInstruction(const MachineInstr *MI, StringRef &ErrInfo) const;
+  bool verifyInstruction(const MachineInstr *MI,
+                         StringRef &ErrInfo) const override;
 
 private:
   /// Modeling special VFP / NEON fp MLA / MLS hazards.
@@ -417,7 +420,8 @@ void emitThumbRegPlusImmediate(MachineBasicBlock &MBB,
 /// NumBytes. This can save a few bytes per function in code-size, but
 /// obviously generates more memory traffic. As such, it only takes
 /// effect in functions being optimised for size.
-bool tryFoldSPUpdateIntoPushPop(MachineFunction &MF, MachineInstr *MI,
+bool tryFoldSPUpdateIntoPushPop(const ARMSubtarget &Subtarget,
+                                MachineFunction &MF, MachineInstr *MI,
                                 unsigned NumBytes);
 
 /// rewriteARMFrameIndex / rewriteT2FrameIndex -
diff --git a/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.cpp
index 8717dc0cde90..32b5f4aa2942 100644
--- a/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.cpp
@@ -38,20 +38,29 @@
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetOptions.h"
 
+#define DEBUG_TYPE "arm-register-info"
+
 #define GET_REGINFO_TARGET_DESC
 #include "ARMGenRegisterInfo.inc"
 
 using namespace llvm;
 
 ARMBaseRegisterInfo::ARMBaseRegisterInfo(const ARMSubtarget &sti)
-  : ARMGenRegisterInfo(ARM::LR, 0, 0, ARM::PC), STI(sti),
-    FramePtr((STI.isTargetDarwin() || STI.isThumb()) ? ARM::R7 : ARM::R11),
-    BasePtr(ARM::R6) {
+    : ARMGenRegisterInfo(ARM::LR, 0, 0, ARM::PC), STI(sti), BasePtr(ARM::R6) {
+  if (STI.isTargetMachO()) {
+    if (STI.isTargetDarwin() || STI.isThumb1Only())
+      FramePtr = ARM::R7;
+    else
+      FramePtr = ARM::R11;
+  } else if (STI.isTargetWindows())
+    FramePtr = ARM::R11;
+  else // ARM EABI
+    FramePtr = STI.isThumb() ? ARM::R7 : ARM::R11;
 }
 
-const uint16_t*
+const MCPhysReg*
 ARMBaseRegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
-  const uint16_t *RegList = (STI.isTargetIOS() && !STI.isAAPCS_ABI())
+  const MCPhysReg *RegList = (STI.isTargetIOS() && !STI.isAAPCS_ABI())
                                 ? CSR_iOS_SaveList
                                 : CSR_AAPCS_SaveList;
 
@@ -107,7 +116,7 @@ ARMBaseRegisterInfo::getThisReturnPreservedMask(CallingConv::ID CC) const {
   // should return NULL
   if (CC == CallingConv::GHC)
     // This is academic becase all GHC calls are (supposed to be) tail calls
-    return NULL;
+    return nullptr;
   return (STI.isTargetIOS() && !STI.isAAPCS_ABI())
     ? CSR_iOS_ThisReturn_RegMask : CSR_AAPCS_ThisReturn_RegMask;
 }
@@ -173,7 +182,7 @@ ARMBaseRegisterInfo::getPointerRegClass(const MachineFunction &MF, unsigned Kind
 const TargetRegisterClass *
 ARMBaseRegisterInfo::getCrossCopyRegClass(const TargetRegisterClass *RC) const {
   if (RC == &ARM::CCRRegClass)
-    return 0;  // Can't copy CCR registers.
+    return nullptr;  // Can't copy CCR registers.
   return RC;
 }
 
@@ -408,6 +417,11 @@ emitLoadConstPool(MachineBasicBlock &MBB,
     .setMIFlags(MIFlags);
 }
 
+bool ARMBaseRegisterInfo::mayOverrideLocalAssignment() const {
+  // The native linux build hits a downstream codegen bug when this is enabled.
+  return STI.isTargetDarwin();
+}
+
 bool ARMBaseRegisterInfo::
 requiresRegisterScavenging(const MachineFunction &MF) const {
   return true;
@@ -590,10 +604,8 @@ materializeFrameBaseRegister(MachineBasicBlock *MBB,
     AddDefaultCC(MIB);
 }
 
-void
-ARMBaseRegisterInfo::resolveFrameIndex(MachineBasicBlock::iterator I,
-                                       unsigned BaseReg, int64_t Offset) const {
-  MachineInstr &MI = *I;
+void ARMBaseRegisterInfo::resolveFrameIndex(MachineInstr &MI, unsigned BaseReg,
+                                            int64_t Offset) const {
   MachineBasicBlock &MBB = *MI.getParent();
   MachineFunction &MF = *MBB.getParent();
   const ARMBaseInstrInfo &TII =
@@ -765,3 +777,60 @@ ARMBaseRegisterInfo::eliminateFrameIndex(MachineBasicBlock::iterator II,
     MI.getOperand(FIOperandNum).ChangeToRegister(ScratchReg, false, false,true);
   }
 }
+
+bool ARMBaseRegisterInfo::shouldCoalesce(MachineInstr *MI,
+                                  const TargetRegisterClass *SrcRC,
+                                  unsigned SubReg,
+                                  const TargetRegisterClass *DstRC,
+                                  unsigned DstSubReg,
+                                  const TargetRegisterClass *NewRC) const {
+  auto MBB = MI->getParent();
+  auto MF = MBB->getParent();
+  const MachineRegisterInfo &MRI = MF->getRegInfo();
+  // If not copying into a sub-register this should be ok because we shouldn't
+  // need to split the reg.
+  if (!DstSubReg)
+    return true;
+  // Small registers don't frequently cause a problem, so we can coalesce them.
+  if (NewRC->getSize() < 32 && DstRC->getSize() < 32 && SrcRC->getSize() < 32)
+    return true;
+
+  auto NewRCWeight =
+              MRI.getTargetRegisterInfo()->getRegClassWeight(NewRC);
+  auto SrcRCWeight =
+              MRI.getTargetRegisterInfo()->getRegClassWeight(SrcRC);
+  auto DstRCWeight =
+              MRI.getTargetRegisterInfo()->getRegClassWeight(DstRC);
+  // If the source register class is more expensive than the destination, the
+  // coalescing is probably profitable.
+  if (SrcRCWeight.RegWeight > NewRCWeight.RegWeight)
+    return true;
+  if (DstRCWeight.RegWeight > NewRCWeight.RegWeight)
+    return true;
+
+  // If the register allocator isn't constrained, we can always allow coalescing
+  // unfortunately we don't know yet if we will be constrained.
+  // The goal of this heuristic is to restrict how many expensive registers
+  // we allow to coalesce in a given basic block.
+  auto AFI = MF->getInfo<ARMFunctionInfo>();
+  auto It = AFI->getCoalescedWeight(MBB);
+
+  DEBUG(dbgs() << "\tARM::shouldCoalesce - Coalesced Weight: "
+    << It->second << "\n");
+  DEBUG(dbgs() << "\tARM::shouldCoalesce - Reg Weight: "
+    << NewRCWeight.RegWeight << "\n");
+
+  // This number is the largest round number that which meets the criteria:
+  //  (1) addresses PR18825
+  //  (2) generates better code in some test cases (like vldm-shed-a9.ll)
+  //  (3) Doesn't regress any test cases (in-tree, test-suite, and SPEC)
+  // In practice the SizeMultiplier will only factor in for straight line code
+  // that uses a lot of NEON vectors, which isn't terribly common.
+  unsigned SizeMultiplier = MBB->size()/100;
+  SizeMultiplier = SizeMultiplier ? SizeMultiplier : 1;
+  if (It->second < NewRCWeight.WeightLimit * SizeMultiplier) {
+    It->second += NewRCWeight.RegWeight;
+    return true;
+  }
+  return false;
+}
diff --git a/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.h b/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.h
index e28fff68f4e2..833d3f218480 100644
--- a/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMBaseRegisterInfo.h
@@ -14,7 +14,7 @@
 #ifndef ARMBASEREGISTERINFO_H
 #define ARMBASEREGISTERINFO_H
 
-#include "ARM.h"
+#include "MCTargetDesc/ARMBaseInfo.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 
 #define GET_REGINFO_HEADER
@@ -42,7 +42,7 @@ static inline bool isARMArea1Register(unsigned Reg, bool isIOS) {
     case R4:  case R5:  case R6:  case R7:
     case LR:  case SP:  case PC:
       return true;
-    case R8:  case R9:  case R10: case R11:
+    case R8:  case R9:  case R10: case R11: case R12:
       // For iOS we want r7 and lr to be next to each other.
       return !isIOS;
     default:
@@ -53,7 +53,7 @@ static inline bool isARMArea1Register(unsigned Reg, bool isIOS) {
 static inline bool isARMArea2Register(unsigned Reg, bool isIOS) {
   using namespace ARM;
   switch (Reg) {
-    case R8: case R9: case R10: case R11:
+    case R8: case R9: case R10: case R11: case R12:
       // iOS has this second area.
       return isIOS;
     default:
@@ -100,8 +100,9 @@ protected:
 
 public:
   /// Code Generation virtual methods...
-  const uint16_t *getCalleeSavedRegs(const MachineFunction *MF = 0) const;
-  const uint32_t *getCallPreservedMask(CallingConv::ID) const;
+  const MCPhysReg *
+  getCalleeSavedRegs(const MachineFunction *MF = nullptr) const override;
+  const uint32_t *getCallPreservedMask(CallingConv::ID) const override;
   const uint32_t *getNoPreservedMask() const;
 
   /// getThisReturnPreservedMask - Returns a call preserved mask specific to the
@@ -113,48 +114,51 @@ public:
   /// Should return NULL in the case that the calling convention does not have
   /// this property
   const uint32_t *getThisReturnPreservedMask(CallingConv::ID) const;
-  
-  BitVector getReservedRegs(const MachineFunction &MF) const;
 
-  const TargetRegisterClass*
-  getPointerRegClass(const MachineFunction &MF, unsigned Kind = 0) const;
-  const TargetRegisterClass*
-  getCrossCopyRegClass(const TargetRegisterClass *RC) const;
+  BitVector getReservedRegs(const MachineFunction &MF) const override;
 
-  const TargetRegisterClass*
-  getLargestLegalSuperClass(const TargetRegisterClass *RC) const;
+  const TargetRegisterClass *
+  getPointerRegClass(const MachineFunction &MF,
+                     unsigned Kind = 0) const override;
+  const TargetRegisterClass *
+  getCrossCopyRegClass(const TargetRegisterClass *RC) const override;
+
+  const TargetRegisterClass *
+  getLargestLegalSuperClass(const TargetRegisterClass *RC) const override;
 
   unsigned getRegPressureLimit(const TargetRegisterClass *RC,
-                               MachineFunction &MF) const;
+                               MachineFunction &MF) const override;
 
   void getRegAllocationHints(unsigned VirtReg,
                              ArrayRef<MCPhysReg> Order,
                              SmallVectorImpl<MCPhysReg> &Hints,
                              const MachineFunction &MF,
-                             const VirtRegMap *VRM) const;
+                             const VirtRegMap *VRM) const override;
 
   void UpdateRegAllocHint(unsigned Reg, unsigned NewReg,
-                          MachineFunction &MF) const;
+                          MachineFunction &MF) const override;
 
-  virtual bool avoidWriteAfterWrite(const TargetRegisterClass *RC) const;
+  bool avoidWriteAfterWrite(const TargetRegisterClass *RC) const override;
 
   bool hasBasePointer(const MachineFunction &MF) const;
 
   bool canRealignStack(const MachineFunction &MF) const;
-  bool needsStackRealignment(const MachineFunction &MF) const;
-  int64_t getFrameIndexInstrOffset(const MachineInstr *MI, int Idx) const;
-  bool needsFrameBaseReg(MachineInstr *MI, int64_t Offset) const;
+  bool needsStackRealignment(const MachineFunction &MF) const override;
+  int64_t getFrameIndexInstrOffset(const MachineInstr *MI,
+                                   int Idx) const override;
+  bool needsFrameBaseReg(MachineInstr *MI, int64_t Offset) const override;
   void materializeFrameBaseRegister(MachineBasicBlock *MBB,
                                     unsigned BaseReg, int FrameIdx,
-                                    int64_t Offset) const;
-  void resolveFrameIndex(MachineBasicBlock::iterator I,
-                         unsigned BaseReg, int64_t Offset) const;
-  bool isFrameOffsetLegal(const MachineInstr *MI, int64_t Offset) const;
+                                    int64_t Offset) const override;
+  void resolveFrameIndex(MachineInstr &MI, unsigned BaseReg,
+                         int64_t Offset) const override;
+  bool isFrameOffsetLegal(const MachineInstr *MI,
+                          int64_t Offset) const override;
 
   bool cannotEliminateFrame(const MachineFunction &MF) const;
 
   // Debug information queries.
-  unsigned getFrameRegister(const MachineFunction &MF) const;
+  unsigned getFrameRegister(const MachineFunction &MF) const override;
   unsigned getBaseRegister() const { return BasePtr; }
 
   bool isLowRegister(unsigned Reg) const;
@@ -164,25 +168,33 @@ public:
   /// specified immediate.
   virtual void emitLoadConstPool(MachineBasicBlock &MBB,
                                  MachineBasicBlock::iterator &MBBI,
-                                 DebugLoc dl,
-                                 unsigned DestReg, unsigned SubIdx,
-                                 int Val,
-                                 ARMCC::CondCodes Pred = ARMCC::AL,
+                                 DebugLoc dl, unsigned DestReg, unsigned SubIdx,
+                                 int Val, ARMCC::CondCodes Pred = ARMCC::AL,
                                  unsigned PredReg = 0,
                                  unsigned MIFlags = MachineInstr::NoFlags)const;
 
   /// Code Generation virtual methods...
-  virtual bool requiresRegisterScavenging(const MachineFunction &MF) const;
+  bool mayOverrideLocalAssignment() const override;
+
+  bool requiresRegisterScavenging(const MachineFunction &MF) const override;
+
+  bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const override;
 
-  virtual bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const;
+  bool requiresFrameIndexScavenging(const MachineFunction &MF) const override;
 
-  virtual bool requiresFrameIndexScavenging(const MachineFunction &MF) const;
+  bool requiresVirtualBaseRegisters(const MachineFunction &MF) const override;
 
-  virtual bool requiresVirtualBaseRegisters(const MachineFunction &MF) const;
+  void eliminateFrameIndex(MachineBasicBlock::iterator II,
+                           int SPAdj, unsigned FIOperandNum,
+                           RegScavenger *RS = nullptr) const override;
 
-  virtual void eliminateFrameIndex(MachineBasicBlock::iterator II,
-                                   int SPAdj, unsigned FIOperandNum,
-                                   RegScavenger *RS = NULL) const;
+  /// \brief SrcRC and DstRC will be morphed into NewRC if this returns true
+  bool shouldCoalesce(MachineInstr *MI,
+                      const TargetRegisterClass *SrcRC,
+                      unsigned SubReg,
+                      const TargetRegisterClass *DstRC,
+                      unsigned DstSubReg,
+                      const TargetRegisterClass *NewRC) const override;
 };
 
 } // end namespace llvm
diff --git a/contrib/llvm/lib/Target/ARM/ARMBuildAttrs.h b/contrib/llvm/lib/Target/ARM/ARMBuildAttrs.h
deleted file mode 100644
index b16d4ef54b6d..000000000000
--- a/contrib/llvm/lib/Target/ARM/ARMBuildAttrs.h
+++ /dev/null
@@ -1,170 +0,0 @@
-//===-- ARMBuildAttrs.h - ARM Build Attributes ------------------*- C++ -*-===//
-//
-//                     The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-//
-// This file contains enumerations and support routines for ARM build attributes
-// as defined in ARM ABI addenda document (ABI release 2.08).
-//
-//===----------------------------------------------------------------------===//
-
-#ifndef __TARGET_ARMBUILDATTRS_H__
-#define __TARGET_ARMBUILDATTRS_H__
-
-namespace llvm {
-namespace ARMBuildAttrs {
-
-  enum SpecialAttr {
-    // This is for the .cpu asm attr. It translates into one or more
-    // AttrType (below) entries in the .ARM.attributes section in the ELF.
-    SEL_CPU
-  };
-
-  enum AttrType {
-    // Rest correspond to ELF/.ARM.attributes
-    File                      = 1,
-    Section                   = 2,
-    Symbol                    = 3,
-    CPU_raw_name              = 4,
-    CPU_name                  = 5,
-    CPU_arch                  = 6,
-    CPU_arch_profile          = 7,
-    ARM_ISA_use               = 8,
-    THUMB_ISA_use             = 9,
-    VFP_arch                  = 10,
-    WMMX_arch                 = 11,
-    Advanced_SIMD_arch        = 12,
-    PCS_config                = 13,
-    ABI_PCS_R9_use            = 14,
-    ABI_PCS_RW_data           = 15,
-    ABI_PCS_RO_data           = 16,
-    ABI_PCS_GOT_use           = 17,
-    ABI_PCS_wchar_t           = 18,
-    ABI_FP_rounding           = 19,
-    ABI_FP_denormal           = 20,
-    ABI_FP_exceptions         = 21,
-    ABI_FP_user_exceptions    = 22,
-    ABI_FP_number_model       = 23,
-    ABI_align8_needed         = 24,
-    ABI_align8_preserved      = 25,
-    ABI_enum_size             = 26,
-    ABI_HardFP_use            = 27,
-    ABI_VFP_args              = 28,
-    ABI_WMMX_args             = 29,
-    ABI_optimization_goals    = 30,
-    ABI_FP_optimization_goals = 31,
-    compatibility             = 32,
-    CPU_unaligned_access      = 34,
-    FP_HP_extension           = 36,
-    ABI_FP_16bit_format       = 38,
-    MPextension_use           = 42, // was 70, 2.08 ABI
-    DIV_use                   = 44,
-    nodefaults                = 64,
-    also_compatible_with      = 65,
-    T2EE_use                  = 66,
-    conformance               = 67,
-    Virtualization_use        = 68,
-    MPextension_use_old       = 70
-  };
-
-  // Magic numbers for .ARM.attributes
-  enum AttrMagic {
-    Format_Version  = 0x41
-  };
-
-  // Legal Values for CPU_arch, (=6), uleb128
-  enum CPUArch {
-    Pre_v4       = 0,
-    v4       = 1,   // e.g. SA110
-    v4T      = 2,   // e.g. ARM7TDMI
-    v5T      = 3,   // e.g. ARM9TDMI
-    v5TE     = 4,   // e.g. ARM946E_S
-    v5TEJ    = 5,   // e.g. ARM926EJ_S
-    v6       = 6,   // e.g. ARM1136J_S
-    v6KZ     = 7,   // e.g. ARM1176JZ_S
-    v6T2     = 8,   // e.g. ARM1156T2F_S
-    v6K      = 9,   // e.g. ARM1136J_S
-    v7       = 10,  // e.g. Cortex A8, Cortex M3
-    v6_M     = 11,  // e.g. Cortex M1
-    v6S_M    = 12,  // v6_M with the System extensions
-    v7E_M    = 13,  // v7_M with DSP extensions
-    v8       = 14   // v8, AArch32
-  };
-
-  enum CPUArchProfile { // (=7), uleb128
-    Not_Applicable = 0, // pre v7, or cross-profile code
-    ApplicationProfile = (0x41), // 'A' (e.g. for Cortex A8)
-    RealTimeProfile = (0x52), // 'R' (e.g. for Cortex R4)
-    MicroControllerProfile = (0x4D), // 'M' (e.g. for Cortex M3)
-    SystemProfile = (0x53) // 'S' Application or real-time profile
-  };
-
-  // The following have a lot of common use cases
-  enum {
-    Not_Allowed = 0,
-    Allowed = 1,
-
-    // Tag_ARM_ISA_use (=8), uleb128
-
-    // Tag_THUMB_ISA_use, (=9), uleb128
-    AllowThumb32 = 2, // 32-bit Thumb (implies 16-bit instructions)
-
-    // Tag_FP_arch (=10), uleb128 (formerly Tag_VFP_arch = 10)
-    AllowFPv2  = 2, // v2 FP ISA permitted (implies use of the v1 FP ISA)
-    AllowFPv3A = 3, // v3 FP ISA permitted (implies use of the v2 FP ISA)
-    AllowFPv3B = 4, // v3 FP ISA permitted, but only D0-D15, S0-S31
-    AllowFPv4A = 5, // v4 FP ISA permitted (implies use of v3 FP ISA)
-    AllowFPv4B = 6, // v4 FP ISA was permitted, but only D0-D15, S0-S31
-    AllowFPARMv8A = 7, // Use of the ARM v8-A FP ISA was permitted
-    AllowFPARMv8B = 8, // Use of the ARM v8-A FP ISA was permitted, but only D0-D15, S0-S31
-
-    // Tag_WMMX_arch, (=11), uleb128
-    AllowWMMXv1 = 1,  // The user permitted this entity to use WMMX v1
-    AllowWMMXv2 = 2,  // The user permitted this entity to use WMMX v2
-
-    // Tag_Advanced_SIMD_arch, (=12), uleb128
-    AllowNeon = 1, // SIMDv1 was permitted
-    AllowNeon2 = 2, // SIMDv2 was permitted (Half-precision FP, MAC operations)
-    AllowNeonARMv8 = 3, // ARM v8-A SIMD was permitted
-
-    // Tag_ABI_FP_denormal, (=20), uleb128
-    PreserveFPSign = 2, // sign when flushed-to-zero is preserved
-
-    // Tag_ABI_FP_number_model, (=23), uleb128
-    AllowRTABI = 2,  // numbers, infinities, and one quiet NaN (see [RTABI])
-    AllowIEE754 = 3, // this code to use all the IEEE 754-defined FP encodings
-
-    // Tag_ABI_HardFP_use, (=27), uleb128
-    HardFPImplied = 0, // FP use should be implied by Tag_FP_arch
-    HardFPSinglePrecision = 1, // Single-precision only
-
-    // Tag_ABI_VFP_args, (=28), uleb128
-    BaseAAPCS = 0,
-    HardFPAAPCS = 1,
-
-    // Tag_FP_HP_extension, (=36), uleb128
-    AllowHPFP = 1, // Allow use of Half Precision FP
-
-    // Tag_MPextension_use, (=42), uleb128
-    AllowMP = 1, // Allow use of MP extensions
-
-    // Tag_DIV_use, (=44), uleb128
-    AllowDIVIfExists = 0, // Allow hardware divide if available in arch, or no info exists.
-    DisallowDIV = 1, // Hardware divide explicitly disallowed
-    AllowDIVExt = 2, // Allow hardware divide as optional architecture extension above
-                     // the base arch specified by Tag_CPU_arch and Tag_CPU_arch_profile.
-
-    // Tag_Virtualization_use, (=68), uleb128
-    AllowTZ = 1,
-    AllowVirtualization = 2,
-    AllowTZVirtualization = 3
-  };
-
-} // namespace ARMBuildAttrs
-} // namespace llvm
-
-#endif // __TARGET_ARMBUILDATTRS_H__
diff --git a/contrib/llvm/lib/Target/ARM/ARMCallingConv.h b/contrib/llvm/lib/Target/ARM/ARMCallingConv.h
index 4f94ad2403d3..dc41c1c14bbb 100644
--- a/contrib/llvm/lib/Target/ARM/ARMCallingConv.h
+++ b/contrib/llvm/lib/Target/ARM/ARMCallingConv.h
@@ -28,7 +28,7 @@ namespace llvm {
 static bool f64AssignAPCS(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
                           CCValAssign::LocInfo &LocInfo,
                           CCState &State, bool CanFail) {
-  static const uint16_t RegList[] = { ARM::R0, ARM::R1, ARM::R2, ARM::R3 };
+  static const MCPhysReg RegList[] = { ARM::R0, ARM::R1, ARM::R2, ARM::R3 };
 
   // Try to get the first register.
   if (unsigned Reg = State.AllocateReg(RegList, 4))
@@ -71,10 +71,10 @@ static bool CC_ARM_APCS_Custom_f64(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
 static bool f64AssignAAPCS(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
                            CCValAssign::LocInfo &LocInfo,
                            CCState &State, bool CanFail) {
-  static const uint16_t HiRegList[] = { ARM::R0, ARM::R2 };
-  static const uint16_t LoRegList[] = { ARM::R1, ARM::R3 };
-  static const uint16_t ShadowRegList[] = { ARM::R0, ARM::R1 };
-  static const uint16_t GPRArgRegs[] = { ARM::R0, ARM::R1, ARM::R2, ARM::R3 };
+  static const MCPhysReg HiRegList[] = { ARM::R0, ARM::R2 };
+  static const MCPhysReg LoRegList[] = { ARM::R1, ARM::R3 };
+  static const MCPhysReg ShadowRegList[] = { ARM::R0, ARM::R1 };
+  static const MCPhysReg GPRArgRegs[] = { ARM::R0, ARM::R1, ARM::R2, ARM::R3 };
 
   unsigned Reg = State.AllocateReg(HiRegList, ShadowRegList, 2);
   if (Reg == 0) {
@@ -123,8 +123,8 @@ static bool CC_ARM_AAPCS_Custom_f64(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
 
 static bool f64RetAssign(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
                          CCValAssign::LocInfo &LocInfo, CCState &State) {
-  static const uint16_t HiRegList[] = { ARM::R0, ARM::R2 };
-  static const uint16_t LoRegList[] = { ARM::R1, ARM::R3 };
+  static const MCPhysReg HiRegList[] = { ARM::R0, ARM::R2 };
+  static const MCPhysReg LoRegList[] = { ARM::R1, ARM::R3 };
 
   unsigned Reg = State.AllocateReg(HiRegList, LoRegList, 2);
   if (Reg == 0)
@@ -160,6 +160,105 @@ static bool RetCC_ARM_AAPCS_Custom_f64(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
                                    State);
 }
 
+static const uint16_t SRegList[] = { ARM::S0,  ARM::S1,  ARM::S2,  ARM::S3,
+                                     ARM::S4,  ARM::S5,  ARM::S6,  ARM::S7,
+                                     ARM::S8,  ARM::S9,  ARM::S10, ARM::S11,
+                                     ARM::S12, ARM::S13, ARM::S14,  ARM::S15 };
+static const uint16_t DRegList[] = { ARM::D0, ARM::D1, ARM::D2, ARM::D3,
+                                     ARM::D4, ARM::D5, ARM::D6, ARM::D7 };
+static const uint16_t QRegList[] = { ARM::Q0, ARM::Q1, ARM::Q2, ARM::Q3 };
+
+// Allocate part of an AAPCS HFA or HVA. We assume that each member of the HA
+// has InConsecutiveRegs set, and that the last member also has
+// InConsecutiveRegsLast set. We must process all members of the HA before
+// we can allocate it, as we need to know the total number of registers that
+// will be needed in order to (attempt to) allocate a contiguous block.
+static bool CC_ARM_AAPCS_Custom_HA(unsigned &ValNo, MVT &ValVT, MVT &LocVT,
+                                   CCValAssign::LocInfo &LocInfo,
+                                   ISD::ArgFlagsTy &ArgFlags, CCState &State) {
+  SmallVectorImpl<CCValAssign> &PendingHAMembers = State.getPendingLocs();
+  // AAPCS HFAs must have 1-4 elements, all of the same type
+  assert(PendingHAMembers.size() < 8);
+  if (PendingHAMembers.size() > 0)
+    assert(PendingHAMembers[0].getLocVT() == LocVT);
+
+  // Add the argument to the list to be allocated once we know the size of the
+  // HA
+  PendingHAMembers.push_back(
+      CCValAssign::getPending(ValNo, ValVT, LocVT, LocInfo));
+
+  if (ArgFlags.isInConsecutiveRegsLast()) {
+    assert(PendingHAMembers.size() > 0 && PendingHAMembers.size() <= 8 &&
+           "Homogeneous aggregates must have between 1 and 4 members");
+
+    // Try to allocate a contiguous block of registers, each of the correct
+    // size to hold one member.
+    const uint16_t *RegList;
+    unsigned NumRegs;
+    switch (LocVT.SimpleTy) {
+    case MVT::i32:
+    case MVT::f32:
+      RegList = SRegList;
+      NumRegs = 16;
+      break;
+    case MVT::f64:
+      RegList = DRegList;
+      NumRegs = 8;
+      break;
+    case MVT::v2f64:
+      RegList = QRegList;
+      NumRegs = 4;
+      break;
+    default:
+      llvm_unreachable("Unexpected member type for HA");
+      break;
+    }
+
+    unsigned RegResult =
+        State.AllocateRegBlock(RegList, NumRegs, PendingHAMembers.size());
+
+    if (RegResult) {
+      for (SmallVectorImpl<CCValAssign>::iterator It = PendingHAMembers.begin();
+           It != PendingHAMembers.end(); ++It) {
+        It->convertToReg(RegResult);
+        State.addLoc(*It);
+        ++RegResult;
+      }
+      PendingHAMembers.clear();
+      return true;
+    }
+
+    // Register allocation failed, fall back to the stack
+
+    // Mark all VFP regs as unavailable (AAPCS rule C.2.vfp)
+    for (unsigned regNo = 0; regNo < 16; ++regNo)
+      State.AllocateReg(SRegList[regNo]);
+
+    unsigned Size = LocVT.getSizeInBits() / 8;
+    unsigned Align = Size;
+
+    if (LocVT.SimpleTy == MVT::v2f64 || LocVT.SimpleTy == MVT::i32) {
+      // Vectors are always aligned to 8 bytes. If we've seen an i32 here
+      // it's because it's been split from a larger type, also with align 8.
+      Align = 8;
+    }
+
+    for (auto It : PendingHAMembers) {
+      It.convertToMem(State.AllocateStack(Size, Align));
+      State.addLoc(It);
+
+      // Only the first member needs to be aligned.
+      Align = 1;
+    }
+
+    // All pending members have now been allocated
+    PendingHAMembers.clear();
+  }
+
+  // This will be allocated by the last member of the HA
+  return true;
+}
+
 } // End llvm namespace
 
 #endif
diff --git a/contrib/llvm/lib/Target/ARM/ARMCallingConv.td b/contrib/llvm/lib/Target/ARM/ARMCallingConv.td
index 9bea4b2d68e9..526089b01b6f 100644
--- a/contrib/llvm/lib/Target/ARM/ARMCallingConv.td
+++ b/contrib/llvm/lib/Target/ARM/ARMCallingConv.td
@@ -64,6 +64,13 @@ def FastCC_ARM_APCS : CallingConv<[
   CCIfType<[f64], CCAssignToReg<[D0, D1, D2, D3, D4, D5, D6, D7]>>,
   CCIfType<[f32], CCAssignToReg<[S0, S1, S2, S3, S4, S5, S6, S7, S8,
                                  S9, S10, S11, S12, S13, S14, S15]>>,
+
+  // CPRCs may be allocated to co-processor registers or the stack - they
+  // may never be allocated to core registers. 
+  CCIfType<[f32], CCAssignToStackWithShadow<4, 4, [Q0, Q1, Q2, Q3]>>,
+  CCIfType<[f64], CCAssignToStackWithShadow<8, 4, [Q0, Q1, Q2, Q3]>>,
+  CCIfType<[v2f64], CCAssignToStackWithShadow<16, 4, [Q0, Q1, Q2, Q3]>>,
+
   CCDelegateTo<CC_ARM_APCS>
 ]>;
 
@@ -114,10 +121,11 @@ def CC_ARM_AAPCS_Common : CallingConv<[
   CCIfType<[i32], CCIf<"ArgFlags.getOrigAlign() != 8",
                        CCAssignToReg<[R0, R1, R2, R3]>>>,
 
-  CCIfType<[i32], CCIfAlign<"8", CCAssignToStackWithShadow<4, 8, R3>>>,
-  CCIfType<[i32, f32], CCAssignToStack<4, 4>>,
-  CCIfType<[f64], CCAssignToStack<8, 8>>,
-  CCIfType<[v2f64], CCAssignToStack<16, 8>>
+  CCIfType<[i32], CCIfAlign<"8", CCAssignToStackWithShadow<4, 8, [R0, R1, R2, R3]>>>,
+  CCIfType<[i32], CCAssignToStackWithShadow<4, 4, [R0, R1, R2, R3]>>,
+  CCIfType<[f32], CCAssignToStackWithShadow<4, 4, [Q0, Q1, Q2, Q3]>>,
+  CCIfType<[f64], CCAssignToStackWithShadow<8, 8, [Q0, Q1, Q2, Q3]>>,
+  CCIfType<[v2f64], CCAssignToStackWithShadow<16, 8, [Q0, Q1, Q2, Q3]>>
 ]>;
 
 def RetCC_ARM_AAPCS_Common : CallingConv<[
@@ -166,6 +174,9 @@ def CC_ARM_AAPCS_VFP : CallingConv<[
   CCIfType<[v1i64, v2i32, v4i16, v8i8, v2f32], CCBitConvertToType<f64>>,
   CCIfType<[v2i64, v4i32, v8i16, v16i8, v4f32], CCBitConvertToType<v2f64>>,
 
+  // HFAs are passed in a contiguous block of registers, or on the stack
+  CCIfConsecutiveRegs<CCCustom<"CC_ARM_AAPCS_Custom_HA">>,
+
   CCIfType<[v2f64], CCAssignToReg<[Q0, Q1, Q2, Q3]>>,
   CCIfType<[f64], CCAssignToReg<[D0, D1, D2, D3, D4, D5, D6, D7]>>,
   CCIfType<[f32], CCAssignToReg<[S0, S1, S2, S3, S4, S5, S6, S7, S8,
diff --git a/contrib/llvm/lib/Target/ARM/ARMCodeEmitter.cpp b/contrib/llvm/lib/Target/ARM/ARMCodeEmitter.cpp
index 568ca858c4d2..5fb6ebfeaae1 100644
--- a/contrib/llvm/lib/Target/ARM/ARMCodeEmitter.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMCodeEmitter.cpp
@@ -12,10 +12,10 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "jit"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMConstantPoolValue.h"
+#include "ARMMachineFunctionInfo.h"
 #include "ARMRelocations.h"
 #include "ARMSubtarget.h"
 #include "ARMTargetMachine.h"
@@ -40,6 +40,8 @@
 #endif
 using namespace llvm;
 
+#define DEBUG_TYPE "jit"
+
 STATISTIC(NumEmitted, "Number of machine instructions emitted");
 
 namespace {
@@ -57,7 +59,7 @@ namespace {
     bool IsPIC;
     bool IsThumb;
 
-    void getAnalysisUsage(AnalysisUsage &AU) const {
+    void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.addRequired<MachineModuleInfo>();
       MachineFunctionPass::getAnalysisUsage(AU);
     }
@@ -65,10 +67,10 @@ namespace {
     static char ID;
   public:
     ARMCodeEmitter(TargetMachine &tm, JITCodeEmitter &mce)
-      : MachineFunctionPass(ID), JTI(0),
+      : MachineFunctionPass(ID), JTI(nullptr),
         II((const ARMBaseInstrInfo *)tm.getInstrInfo()),
         TD(tm.getDataLayout()), TM(tm),
-        MCE(mce), MCPEs(0), MJTEs(0),
+        MCE(mce), MCPEs(nullptr), MJTEs(nullptr),
         IsPIC(TM.getRelocationModel() == Reloc::PIC_), IsThumb(false) {}
 
     /// getBinaryCodeForInstr - This function, generated by the
@@ -76,9 +78,9 @@ namespace {
     /// machine instructions.
     uint64_t getBinaryCodeForInstr(const MachineInstr &MI) const;
 
-    bool runOnMachineFunction(MachineFunction &MF);
+    bool runOnMachineFunction(MachineFunction &MF) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM Machine Code Emitter";
     }
 
@@ -207,8 +209,6 @@ namespace {
       const { return 0; }
     unsigned getThumbAddrModeRegRegOpValue(const MachineInstr &MI, unsigned Op)
       const { return 0; }
-    unsigned getT2AddrModeImm12OpValue(const MachineInstr &MI, unsigned Op)
-      const { return 0; }
     unsigned getT2AddrModeImm8OpValue(const MachineInstr &MI, unsigned Op)
       const { return 0; }
     unsigned getT2Imm8s4OpValue(const MachineInstr &MI, unsigned Op)
@@ -219,8 +219,6 @@ namespace {
       const { return 0; }
     unsigned getT2AddrModeImm8OffsetOpValue(const MachineInstr &MI, unsigned Op)
       const { return 0; }
-    unsigned getT2AddrModeImm12OffsetOpValue(const MachineInstr &MI,unsigned Op)
-      const { return 0; }
     unsigned getT2AddrModeSORegOpValue(const MachineInstr &MI, unsigned Op)
       const { return 0; }
     unsigned getT2SORegOpValue(const MachineInstr &MI, unsigned Op)
@@ -238,10 +236,6 @@ namespace {
       const { return 0; }
     unsigned getBitfieldInvertedMaskOpValue(const MachineInstr &MI,
                                             unsigned Op) const { return 0; }
-    unsigned getSsatBitPosValue(const MachineInstr &MI,
-                                unsigned Op) const { return 0; }
-    uint32_t getLdStmModeOpValue(const MachineInstr &MI, unsigned OpIdx)
-      const {return 0; }
     uint32_t getLdStSORegOpValue(const MachineInstr &MI, unsigned OpIdx)
       const { return 0; }
 
@@ -270,8 +264,6 @@ namespace {
       return 0;
     }
 
-    uint32_t getAddrMode2OpValue(const MachineInstr &MI, unsigned OpIdx)
-      const { return 0;}
     uint32_t getAddrMode2OffsetOpValue(const MachineInstr &MI, unsigned OpIdx)
       const { return 0;}
     uint32_t getPostIdxRegOpValue(const MachineInstr &MI, unsigned OpIdx)
@@ -282,8 +274,6 @@ namespace {
       const { return 0; }
     uint32_t getAddrModeThumbSPOpValue(const MachineInstr &MI, unsigned Op)
       const { return 0; }
-    uint32_t getAddrModeSOpValue(const MachineInstr &MI, unsigned Op)
-      const { return 0; }
     uint32_t getAddrModeISOpValue(const MachineInstr &MI, unsigned Op)
       const { return 0; }
     uint32_t getAddrModePCOpValue(const MachineInstr &MI, unsigned Op)
@@ -385,7 +375,7 @@ bool ARMCodeEmitter::runOnMachineFunction(MachineFunction &MF) {
 
   Subtarget = &TM.getSubtarget<ARMSubtarget>();
   MCPEs = &MF.getConstantPool()->getConstants();
-  MJTEs = 0;
+  MJTEs = nullptr;
   if (MF.getJumpTableInfo()) MJTEs = &MF.getJumpTableInfo()->getJumpTables();
   IsPIC = TM.getRelocationModel() == Reloc::PIC_;
   IsThumb = MF.getInfo<ARMFunctionInfo>()->isThumbFunction();
@@ -866,7 +856,8 @@ void ARMCodeEmitter::emitPseudoInstruction(const MachineInstr &MI) {
     }
     break;
   }
-  case TargetOpcode::PROLOG_LABEL:
+  case TargetOpcode::CFI_INSTRUCTION:
+    break;
   case TargetOpcode::EH_LABEL:
     MCE.emitLabel(MI.getOperand(0).getMCSymbol());
     break;
diff --git a/contrib/llvm/lib/Target/ARM/ARMConstantIslandPass.cpp b/contrib/llvm/lib/Target/ARM/ARMConstantIslandPass.cpp
index cff5ce27bca6..ce264eeef90f 100644
--- a/contrib/llvm/lib/Target/ARM/ARMConstantIslandPass.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMConstantIslandPass.cpp
@@ -13,7 +13,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-cp-islands"
 #include "ARM.h"
 #include "ARMMachineFunctionInfo.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
@@ -36,6 +35,8 @@
 #include <algorithm>
 using namespace llvm;
 
+#define DEBUG_TYPE "arm-cp-islands"
+
 STATISTIC(NumCPEs,       "Number of constpool entries");
 STATISTIC(NumSplit,      "Number of uncond branches inserted");
 STATISTIC(NumCBrFixed,   "Number of cond branches fixed");
@@ -266,9 +267,9 @@ namespace {
     static char ID;
     ARMConstantIslands() : MachineFunctionPass(ID) {}
 
-    virtual bool runOnMachineFunction(MachineFunction &MF);
+    bool runOnMachineFunction(MachineFunction &MF) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM constant island placement and branch shortening pass";
     }
 
@@ -569,10 +570,10 @@ static bool BBHasFallthrough(MachineBasicBlock *MBB) {
   // Get the next machine basic block in the function.
   MachineFunction::iterator MBBI = MBB;
   // Can't fall off end of function.
-  if (llvm::next(MBBI) == MBB->getParent()->end())
+  if (std::next(MBBI) == MBB->getParent()->end())
     return false;
 
-  MachineBasicBlock *NextBB = llvm::next(MBBI);
+  MachineBasicBlock *NextBB = std::next(MBBI);
   for (MachineBasicBlock::succ_iterator I = MBB->succ_begin(),
        E = MBB->succ_end(); I != E; ++I)
     if (*I == NextBB)
@@ -593,7 +594,7 @@ ARMConstantIslands::CPEntry
     if (CPEs[i].CPEMI == CPEMI)
       return &CPEs[i];
   }
-  return NULL;
+  return nullptr;
 }
 
 /// getCPELogAlign - Returns the required alignment of the constant pool entry
@@ -917,7 +918,7 @@ MachineBasicBlock *ARMConstantIslands::splitBlockBeforeInstr(MachineInstr *MI) {
                      CompareMBBNumbers);
   MachineBasicBlock* WaterBB = *IP;
   if (WaterBB == OrigBB)
-    WaterList.insert(llvm::next(IP), NewBB);
+    WaterList.insert(std::next(IP), NewBB);
   else
     WaterList.insert(IP, OrigBB);
   NewWaterList.insert(OrigBB);
@@ -1102,7 +1103,7 @@ bool ARMConstantIslands::decrementCPEReferenceCount(unsigned CPI,
   assert(CPE && "Unexpected!");
   if (--CPE->RefCount == 0) {
     removeDeadCPEMI(CPEMI);
-    CPE->CPEMI = NULL;
+    CPE->CPEMI = nullptr;
     --NumCPEs;
     return true;
   }
@@ -1135,7 +1136,7 @@ int ARMConstantIslands::findInRangeCPEntry(CPUser& U, unsigned UserOffset)
     if (CPEs[i].CPEMI == CPEMI)
       continue;
     // Removing CPEs can leave empty entries, skip
-    if (CPEs[i].CPEMI == NULL)
+    if (CPEs[i].CPEMI == nullptr)
       continue;
     if (isCPEntryInRange(UserMI, UserOffset, CPEs[i].CPEMI, U.getMaxDisp(),
                      U.NegOk)) {
@@ -1188,7 +1189,7 @@ bool ARMConstantIslands::findAvailableWater(CPUser &U, unsigned UserOffset,
     return false;
 
   unsigned BestGrowth = ~0u;
-  for (water_iterator IP = prior(WaterList.end()), B = WaterList.begin();;
+  for (water_iterator IP = std::prev(WaterList.end()), B = WaterList.begin();;
        --IP) {
     MachineBasicBlock* WaterBB = *IP;
     // Check if water is in range and is either at a lower address than the
@@ -1249,7 +1250,7 @@ void ARMConstantIslands::createNewWater(unsigned CPUserIndex,
     if (isOffsetInRange(UserOffset, CPEOffset, U)) {
       DEBUG(dbgs() << "Split at end of BB#" << UserMBB->getNumber()
             << format(", expected CPE offset %#x\n", CPEOffset));
-      NewMBB = llvm::next(MachineFunction::iterator(UserMBB));
+      NewMBB = std::next(MachineFunction::iterator(UserMBB));
       // Add an unconditional branch from UserMBB to fallthrough block.  Record
       // it for branch lengthening; this new branch will not get out of range,
       // but if the preceding conditional branch is out of range, the targets
@@ -1317,11 +1318,10 @@ void ARMConstantIslands::createNewWater(unsigned CPUserIndex,
   ++MI;
   unsigned CPUIndex = CPUserIndex+1;
   unsigned NumCPUsers = CPUsers.size();
-  MachineInstr *LastIT = 0;
+  MachineInstr *LastIT = nullptr;
   for (unsigned Offset = UserOffset+TII->GetInstSizeInBytes(UserMI);
        Offset < BaseInsertOffset;
-       Offset += TII->GetInstSizeInBytes(MI),
-       MI = llvm::next(MI)) {
+       Offset += TII->GetInstSizeInBytes(MI), MI = std::next(MI)) {
     assert(MI != UserMBB->end() && "Fell off end of block");
     if (CPUIndex < NumCPUsers && CPUsers[CPUIndex].MI == MI) {
       CPUser &U = CPUsers[CPUIndex];
@@ -1393,7 +1393,7 @@ bool ARMConstantIslands::handleConstantPoolUser(unsigned CPUserIndex) {
       NewWaterList.insert(NewIsland);
 
     // The new CPE goes before the following block (NewMBB).
-    NewMBB = llvm::next(MachineFunction::iterator(WaterBB));
+    NewMBB = std::next(MachineFunction::iterator(WaterBB));
 
   } else {
     // No water found.
@@ -1405,7 +1405,7 @@ bool ARMConstantIslands::handleConstantPoolUser(unsigned CPUserIndex) {
     // next iteration for constant pools, but in this context, we don't want
     // it.  Check for this so it will be removed from the WaterList.
     // Also remove any entry from NewWaterList.
-    MachineBasicBlock *WaterBB = prior(MachineFunction::iterator(NewMBB));
+    MachineBasicBlock *WaterBB = std::prev(MachineFunction::iterator(NewMBB));
     IP = std::find(WaterList.begin(), WaterList.end(), WaterBB);
     if (IP != WaterList.end())
       NewWaterList.erase(WaterBB);
@@ -1443,7 +1443,7 @@ bool ARMConstantIslands::handleConstantPoolUser(unsigned CPUserIndex) {
 
   // Increase the size of the island block to account for the new entry.
   BBInfo[NewIsland->getNumber()].Size += Size;
-  adjustBBOffsetsAfter(llvm::prior(MachineFunction::iterator(NewIsland)));
+  adjustBBOffsetsAfter(std::prev(MachineFunction::iterator(NewIsland)));
 
   // Finally, change the CPI in the instruction operand to be ID.
   for (unsigned i = 0, e = UserMI->getNumOperands(); i != e; ++i)
@@ -1492,7 +1492,7 @@ bool ARMConstantIslands::removeUnusedCPEntries() {
       for (unsigned j = 0, ee = CPEs.size(); j != ee; ++j) {
         if (CPEs[j].RefCount == 0 && CPEs[j].CPEMI) {
           removeDeadCPEMI(CPEs[j].CPEMI);
-          CPEs[j].CPEMI = NULL;
+          CPEs[j].CPEMI = nullptr;
           MadeChange = true;
         }
       }
@@ -1592,7 +1592,7 @@ ARMConstantIslands::fixupConditionalBr(ImmBranch &Br) {
 
   ++NumCBrFixed;
   if (BMI != MI) {
-    if (llvm::next(MachineBasicBlock::iterator(MI)) == prior(MBB->end()) &&
+    if (std::next(MachineBasicBlock::iterator(MI)) == std::prev(MBB->end()) &&
         BMI->getOpcode() == Br.UncondBr) {
       // Last MI in the BB is an unconditional branch. Can we simply invert the
       // condition and swap destinations:
@@ -1622,7 +1622,7 @@ ARMConstantIslands::fixupConditionalBr(ImmBranch &Br) {
     MBB->back().eraseFromParent();
     // BBInfo[SplitBB].Offset is wrong temporarily, fixed below
   }
-  MachineBasicBlock *NextBB = llvm::next(MachineFunction::iterator(MBB));
+  MachineBasicBlock *NextBB = std::next(MachineFunction::iterator(MBB));
 
   DEBUG(dbgs() << "  Insert B to BB#" << DestBB->getNumber()
                << " also invert condition and change dest. to BB#"
@@ -1845,7 +1845,7 @@ bool ARMConstantIslands::optimizeThumb2JumpTables() {
   // FIXME: After the tables are shrunk, can we get rid some of the
   // constantpool tables?
   MachineJumpTableInfo *MJTI = MF->getJumpTableInfo();
-  if (MJTI == 0) return false;
+  if (!MJTI) return false;
 
   const std::vector<MachineJumpTableEntry> &JT = MJTI->getJumpTables();
   for (unsigned i = 0, e = T2JumpTables.size(); i != e; ++i) {
@@ -1971,7 +1971,7 @@ bool ARMConstantIslands::reorderThumb2JumpTables() {
   bool MadeChange = false;
 
   MachineJumpTableInfo *MJTI = MF->getJumpTableInfo();
-  if (MJTI == 0) return false;
+  if (!MJTI) return false;
 
   const std::vector<MachineJumpTableEntry> &JT = MJTI->getJumpTables();
   for (unsigned i = 0, e = T2JumpTables.size(); i != e; ++i) {
@@ -2013,11 +2013,11 @@ adjustJTTargetBlockForward(MachineBasicBlock *BB, MachineBasicBlock *JTBB) {
   // try to move it; otherwise, create a new block following the jump
   // table that branches back to the actual target. This is a very simple
   // heuristic. FIXME: We can definitely improve it.
-  MachineBasicBlock *TBB = 0, *FBB = 0;
+  MachineBasicBlock *TBB = nullptr, *FBB = nullptr;
   SmallVector<MachineOperand, 4> Cond;
   SmallVector<MachineOperand, 4> CondPrior;
   MachineFunction::iterator BBi = BB;
-  MachineFunction::iterator OldPrior = prior(BBi);
+  MachineFunction::iterator OldPrior = std::prev(BBi);
 
   // If the block terminator isn't analyzable, don't try to move the block
   bool B = TII->AnalyzeBranch(*BB, TBB, FBB, Cond);
@@ -2033,7 +2033,7 @@ adjustJTTargetBlockForward(MachineBasicBlock *BB, MachineBasicBlock *JTBB) {
     // Update numbering to account for the block being moved.
     MF->RenumberBlocks();
     ++NumJTMoved;
-    return NULL;
+    return nullptr;
   }
 
   // Create a new MBB for the code after the jump BB.
diff --git a/contrib/llvm/lib/Target/ARM/ARMConstantPoolValue.h b/contrib/llvm/lib/Target/ARM/ARMConstantPoolValue.h
index 7ae7bf46f19d..c7a84155bd48 100644
--- a/contrib/llvm/lib/Target/ARM/ARMConstantPoolValue.h
+++ b/contrib/llvm/lib/Target/ARM/ARMConstantPoolValue.h
@@ -103,12 +103,12 @@ public:
   bool isLSDA() const { return Kind == ARMCP::CPLSDA; }
   bool isMachineBasicBlock() const{ return Kind == ARMCP::CPMachineBasicBlock; }
 
-  virtual unsigned getRelocationInfo() const { return 2; }
+  unsigned getRelocationInfo() const override { return 2; }
 
-  virtual int getExistingMachineCPValue(MachineConstantPool *CP,
-                                        unsigned Alignment);
+  int getExistingMachineCPValue(MachineConstantPool *CP,
+                                unsigned Alignment) override;
 
-  virtual void addSelectionDAGCSEId(FoldingSetNodeID &ID);
+  void addSelectionDAGCSEId(FoldingSetNodeID &ID) override;
 
   /// hasSameValue - Return true if this ARM constpool value can share the same
   /// constantpool entry as another ARM constpool value.
@@ -120,7 +120,7 @@ public:
       this->Modifier == A->Modifier;
   }
 
-  virtual void print(raw_ostream &O) const;
+  void print(raw_ostream &O) const override;
   void print(raw_ostream *O) const { if (O) print(*O); }
   void dump() const;
 };
@@ -164,16 +164,16 @@ public:
   const GlobalValue *getGV() const;
   const BlockAddress *getBlockAddress() const;
 
-  virtual int getExistingMachineCPValue(MachineConstantPool *CP,
-                                        unsigned Alignment);
+  int getExistingMachineCPValue(MachineConstantPool *CP,
+                                unsigned Alignment) override;
 
   /// hasSameValue - Return true if this ARM constpool value can share the same
   /// constantpool entry as another ARM constpool value.
-  virtual bool hasSameValue(ARMConstantPoolValue *ACPV);
+  bool hasSameValue(ARMConstantPoolValue *ACPV) override;
 
-  virtual void addSelectionDAGCSEId(FoldingSetNodeID &ID);
+  void addSelectionDAGCSEId(FoldingSetNodeID &ID) override;
 
-  virtual void print(raw_ostream &O) const;
+  void print(raw_ostream &O) const override;
   static bool classof(const ARMConstantPoolValue *APV) {
     return APV->isGlobalValue() || APV->isBlockAddress() || APV->isLSDA();
   }
@@ -198,16 +198,16 @@ public:
 
   const char *getSymbol() const { return S.c_str(); }
 
-  virtual int getExistingMachineCPValue(MachineConstantPool *CP,
-                                        unsigned Alignment);
+  int getExistingMachineCPValue(MachineConstantPool *CP,
+                                unsigned Alignment) override;
 
-  virtual void addSelectionDAGCSEId(FoldingSetNodeID &ID);
+  void addSelectionDAGCSEId(FoldingSetNodeID &ID) override;
 
   /// hasSameValue - Return true if this ARM constpool value can share the same
   /// constantpool entry as another ARM constpool value.
-  virtual bool hasSameValue(ARMConstantPoolValue *ACPV);
+  bool hasSameValue(ARMConstantPoolValue *ACPV) override;
 
-  virtual void print(raw_ostream &O) const;
+  void print(raw_ostream &O) const override;
 
   static bool classof(const ARMConstantPoolValue *ACPV) {
     return ACPV->isExtSymbol();
@@ -234,16 +234,16 @@ public:
 
   const MachineBasicBlock *getMBB() const { return MBB; }
 
-  virtual int getExistingMachineCPValue(MachineConstantPool *CP,
-                                        unsigned Alignment);
+  int getExistingMachineCPValue(MachineConstantPool *CP,
+                                unsigned Alignment) override;
 
-  virtual void addSelectionDAGCSEId(FoldingSetNodeID &ID);
+  void addSelectionDAGCSEId(FoldingSetNodeID &ID) override;
 
   /// hasSameValue - Return true if this ARM constpool value can share the same
   /// constantpool entry as another ARM constpool value.
-  virtual bool hasSameValue(ARMConstantPoolValue *ACPV);
+  bool hasSameValue(ARMConstantPoolValue *ACPV) override;
 
-  virtual void print(raw_ostream &O) const;
+  void print(raw_ostream &O) const override;
 
   static bool classof(const ARMConstantPoolValue *ACPV) {
     return ACPV->isMachineBasicBlock();
diff --git a/contrib/llvm/lib/Target/ARM/ARMExpandPseudoInsts.cpp b/contrib/llvm/lib/Target/ARM/ARMExpandPseudoInsts.cpp
index 3e62b649e0dc..51d3dbb5bd8e 100644
--- a/contrib/llvm/lib/Target/ARM/ARMExpandPseudoInsts.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMExpandPseudoInsts.cpp
@@ -14,21 +14,25 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-pseudo"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMBaseRegisterInfo.h"
+#include "ARMConstantPoolValue.h"
 #include "ARMMachineFunctionInfo.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
+#include "llvm/CodeGen/MachineInstrBundle.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/IR/GlobalValue.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/raw_ostream.h" // FIXME: for debug only. remove!
 #include "llvm/Target/TargetFrameLowering.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "arm-pseudo"
+
 static cl::opt<bool>
 VerifyARMPseudo("verify-arm-pseudo-expand", cl::Hidden,
                 cl::desc("Verify machine code after expanding ARM pseudos"));
@@ -44,9 +48,9 @@ namespace {
     const ARMSubtarget *STI;
     ARMFunctionInfo *AFI;
 
-    virtual bool runOnMachineFunction(MachineFunction &Fn);
+    bool runOnMachineFunction(MachineFunction &Fn) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM pseudo instruction expansion pass";
     }
 
@@ -343,7 +347,7 @@ static const NEONLdStTableEntry *LookupNEONLdSt(unsigned Opcode) {
     std::lower_bound(NEONLdStTable, NEONLdStTable + NumEntries, Opcode);
   if (I != NEONLdStTable + NumEntries && I->PseudoOpc == Opcode)
     return I;
-  return NULL;
+  return nullptr;
 }
 
 /// GetDSubRegs - Get 4 D subregisters of a Q, QQ, or QQQQ register,
@@ -479,6 +483,8 @@ void ARMExpandPseudo::ExpandVST(MachineBasicBlock::iterator &MBBI) {
 
   if (SrcIsKill && !SrcIsUndef) // Add an implicit kill for the super-reg.
     MIB->addRegisterKilled(SrcReg, TRI, true);
+  else if (!SrcIsUndef)
+    MIB.addReg(SrcReg, RegState::Implicit); // Add implicit uses for src reg.
   TransferImpOps(MI, MIB, MIB);
 
   // Transfer memoperands.
@@ -604,12 +610,45 @@ void ARMExpandPseudo::ExpandVTBL(MachineBasicBlock::iterator &MBBI,
   MIB.addOperand(MI.getOperand(OpIdx++));
   MIB.addOperand(MI.getOperand(OpIdx++));
 
-  if (SrcIsKill)  // Add an implicit kill for the super-reg.
-    MIB->addRegisterKilled(SrcReg, TRI, true);
+  // Add an implicit kill and use for the super-reg.
+  MIB.addReg(SrcReg, RegState::Implicit | getKillRegState(SrcIsKill));
   TransferImpOps(MI, MIB, MIB);
   MI.eraseFromParent();
 }
 
+static bool IsAnAddressOperand(const MachineOperand &MO) {
+  // This check is overly conservative.  Unless we are certain that the machine
+  // operand is not a symbol reference, we return that it is a symbol reference.
+  // This is important as the load pair may not be split up Windows.
+  switch (MO.getType()) {
+  case MachineOperand::MO_Register:
+  case MachineOperand::MO_Immediate:
+  case MachineOperand::MO_CImmediate:
+  case MachineOperand::MO_FPImmediate:
+    return false;
+  case MachineOperand::MO_MachineBasicBlock:
+    return true;
+  case MachineOperand::MO_FrameIndex:
+    return false;
+  case MachineOperand::MO_ConstantPoolIndex:
+  case MachineOperand::MO_TargetIndex:
+  case MachineOperand::MO_JumpTableIndex:
+  case MachineOperand::MO_ExternalSymbol:
+  case MachineOperand::MO_GlobalAddress:
+  case MachineOperand::MO_BlockAddress:
+    return true;
+  case MachineOperand::MO_RegisterMask:
+  case MachineOperand::MO_RegisterLiveOut:
+    return false;
+  case MachineOperand::MO_Metadata:
+  case MachineOperand::MO_MCSymbol:
+    return true;
+  case MachineOperand::MO_CFIIndex:
+    return false;
+  }
+  llvm_unreachable("unhandled machine operand type");
+}
+
 void ARMExpandPseudo::ExpandMOV32BitImm(MachineBasicBlock &MBB,
                                         MachineBasicBlock::iterator &MBBI) {
   MachineInstr &MI = *MBBI;
@@ -620,10 +659,14 @@ void ARMExpandPseudo::ExpandMOV32BitImm(MachineBasicBlock &MBB,
   bool DstIsDead = MI.getOperand(0).isDead();
   bool isCC = Opcode == ARM::MOVCCi32imm || Opcode == ARM::t2MOVCCi32imm;
   const MachineOperand &MO = MI.getOperand(isCC ? 2 : 1);
+  bool RequiresBundling = STI->isTargetWindows() && IsAnAddressOperand(MO);
   MachineInstrBuilder LO16, HI16;
 
   if (!STI->hasV6T2Ops() &&
       (Opcode == ARM::MOVi32imm || Opcode == ARM::MOVCCi32imm)) {
+    // FIXME Windows CE supports older ARM CPUs
+    assert(!STI->isTargetWindows() && "Windows on ARM requires ARMv7+");
+
     // Expand into a movi + orr.
     LO16 = BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(ARM::MOVi), DstReg);
     HI16 = BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(ARM::ORRri))
@@ -660,17 +703,29 @@ void ARMExpandPseudo::ExpandMOV32BitImm(MachineBasicBlock &MBB,
     .addReg(DstReg, RegState::Define | getDeadRegState(DstIsDead))
     .addReg(DstReg);
 
-  if (MO.isImm()) {
+  switch (MO.getType()) {
+  case MachineOperand::MO_Immediate: {
     unsigned Imm = MO.getImm();
     unsigned Lo16 = Imm & 0xffff;
     unsigned Hi16 = (Imm >> 16) & 0xffff;
     LO16 = LO16.addImm(Lo16);
     HI16 = HI16.addImm(Hi16);
-  } else {
+    break;
+  }
+  case MachineOperand::MO_ExternalSymbol: {
+    const char *ES = MO.getSymbolName();
+    unsigned TF = MO.getTargetFlags();
+    LO16 = LO16.addExternalSymbol(ES, TF | ARMII::MO_LO16);
+    HI16 = HI16.addExternalSymbol(ES, TF | ARMII::MO_HI16);
+    break;
+  }
+  default: {
     const GlobalValue *GV = MO.getGlobal();
     unsigned TF = MO.getTargetFlags();
     LO16 = LO16.addGlobalAddress(GV, MO.getOffset(), TF | ARMII::MO_LO16);
     HI16 = HI16.addGlobalAddress(GV, MO.getOffset(), TF | ARMII::MO_HI16);
+    break;
+  }
   }
 
   LO16->setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
@@ -678,6 +733,9 @@ void ARMExpandPseudo::ExpandMOV32BitImm(MachineBasicBlock &MBB,
   LO16.addImm(Pred).addReg(PredReg);
   HI16.addImm(Pred).addReg(PredReg);
 
+  if (RequiresBundling)
+    finalizeBundle(MBB, &*LO16, &*MBBI);
+
   TransferImpOps(MI, LO16, HI16);
   MI.eraseFromParent();
 }
@@ -869,10 +927,16 @@ bool ARMExpandPseudo::ExpandMI(MachineBasicBlock &MBB,
     }
     case ARM::tTPsoft:
     case ARM::TPsoft: {
-      MachineInstrBuilder MIB =
-        BuildMI(MBB, MBBI, MI.getDebugLoc(),
-                TII->get(Opcode == ARM::tTPsoft ? ARM::tBL : ARM::BL))
-        .addExternalSymbol("__aeabi_read_tp", 0);
+      MachineInstrBuilder MIB;
+      if (Opcode == ARM::tTPsoft)
+        MIB = BuildMI(MBB, MBBI, MI.getDebugLoc(),
+                      TII->get( ARM::tBL))
+              .addImm((unsigned)ARMCC::AL).addReg(0)
+              .addExternalSymbol("__aeabi_read_tp", 0);
+      else
+        MIB = BuildMI(MBB, MBBI, MI.getDebugLoc(),
+                      TII->get( ARM::BL))
+              .addExternalSymbol("__aeabi_read_tp", 0);
 
       MIB->setMemRefs(MI.memoperands_begin(), MI.memoperands_end());
       TransferImpOps(MI, MIB, MIB);
@@ -900,10 +964,61 @@ bool ARMExpandPseudo::ExpandMI(MachineBasicBlock &MBB,
       return true;
     }
 
-    case ARM::MOV_ga_dyn:
+    case ARM::LDRLIT_ga_abs:
+    case ARM::LDRLIT_ga_pcrel:
+    case ARM::LDRLIT_ga_pcrel_ldr:
+    case ARM::tLDRLIT_ga_abs:
+    case ARM::tLDRLIT_ga_pcrel: {
+      unsigned DstReg = MI.getOperand(0).getReg();
+      bool DstIsDead = MI.getOperand(0).isDead();
+      const MachineOperand &MO1 = MI.getOperand(1);
+      const GlobalValue *GV = MO1.getGlobal();
+      bool IsARM =
+          Opcode != ARM::tLDRLIT_ga_pcrel && Opcode != ARM::tLDRLIT_ga_abs;
+      bool IsPIC =
+          Opcode != ARM::LDRLIT_ga_abs && Opcode != ARM::tLDRLIT_ga_abs;
+      unsigned LDRLITOpc = IsARM ? ARM::LDRi12 : ARM::tLDRpci;
+      unsigned PICAddOpc =
+          IsARM
+              ? (Opcode == ARM::LDRLIT_ga_pcrel_ldr ? ARM::PICADD : ARM::PICLDR)
+              : ARM::tPICADD;
+
+      // We need a new const-pool entry to load from.
+      MachineConstantPool *MCP = MBB.getParent()->getConstantPool();
+      unsigned ARMPCLabelIndex = 0;
+      MachineConstantPoolValue *CPV;
+
+      if (IsPIC) {
+        unsigned PCAdj = IsARM ? 8 : 4;
+        ARMPCLabelIndex = AFI->createPICLabelUId();
+        CPV = ARMConstantPoolConstant::Create(GV, ARMPCLabelIndex,
+                                              ARMCP::CPValue, PCAdj);
+      } else
+        CPV = ARMConstantPoolConstant::Create(GV, ARMCP::no_modifier);
+
+      MachineInstrBuilder MIB =
+          BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(LDRLITOpc), DstReg)
+            .addConstantPoolIndex(MCP->getConstantPoolIndex(CPV, 4));
+      if (IsARM)
+        MIB.addImm(0);
+      AddDefaultPred(MIB);
+
+      if (IsPIC) {
+        MachineInstrBuilder MIB =
+          BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(PICAddOpc))
+            .addReg(DstReg, RegState::Define | getDeadRegState(DstIsDead))
+            .addReg(DstReg)
+            .addImm(ARMPCLabelIndex);
+
+        if (IsARM)
+          AddDefaultPred(MIB);
+      }
+
+      MI.eraseFromParent();
+      return true;
+    }
     case ARM::MOV_ga_pcrel:
     case ARM::MOV_ga_pcrel_ldr:
-    case ARM::t2MOV_ga_dyn:
     case ARM::t2MOV_ga_pcrel: {
       // Expand into movw + movw. Also "add pc" / ldr [pc] in PIC mode.
       unsigned LabelId = AFI->createPICLabelUId();
@@ -912,14 +1027,11 @@ bool ARMExpandPseudo::ExpandMI(MachineBasicBlock &MBB,
       const MachineOperand &MO1 = MI.getOperand(1);
       const GlobalValue *GV = MO1.getGlobal();
       unsigned TF = MO1.getTargetFlags();
-      bool isARM = (Opcode != ARM::t2MOV_ga_pcrel && Opcode!=ARM::t2MOV_ga_dyn);
-      bool isPIC = (Opcode != ARM::MOV_ga_dyn && Opcode != ARM::t2MOV_ga_dyn);
+      bool isARM = Opcode != ARM::t2MOV_ga_pcrel;
       unsigned LO16Opc = isARM ? ARM::MOVi16_ga_pcrel : ARM::t2MOVi16_ga_pcrel;
       unsigned HI16Opc = isARM ? ARM::MOVTi16_ga_pcrel :ARM::t2MOVTi16_ga_pcrel;
-      unsigned LO16TF = isPIC
-        ? ARMII::MO_LO16_NONLAZY_PIC : ARMII::MO_LO16_NONLAZY;
-      unsigned HI16TF = isPIC
-        ? ARMII::MO_HI16_NONLAZY_PIC : ARMII::MO_HI16_NONLAZY;
+      unsigned LO16TF = TF | ARMII::MO_LO16;
+      unsigned HI16TF = TF | ARMII::MO_HI16;
       unsigned PICAddOpc = isARM
         ? (Opcode == ARM::MOV_ga_pcrel_ldr ? ARM::PICLDR : ARM::PICADD)
         : ARM::tPICADD;
@@ -927,16 +1039,11 @@ bool ARMExpandPseudo::ExpandMI(MachineBasicBlock &MBB,
                                          TII->get(LO16Opc), DstReg)
         .addGlobalAddress(GV, MO1.getOffset(), TF | LO16TF)
         .addImm(LabelId);
-      MachineInstrBuilder MIB2 = BuildMI(MBB, MBBI, MI.getDebugLoc(),
-                                         TII->get(HI16Opc), DstReg)
+
+      BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(HI16Opc), DstReg)
         .addReg(DstReg)
         .addGlobalAddress(GV, MO1.getOffset(), TF | HI16TF)
         .addImm(LabelId);
-      if (!isPIC) {
-        TransferImpOps(MI, MIB1, MIB2);
-        MI.eraseFromParent();
-        return true;
-      }
 
       MachineInstrBuilder MIB3 = BuildMI(MBB, MBBI, MI.getDebugLoc(),
                                          TII->get(PICAddOpc))
@@ -1032,33 +1139,6 @@ bool ARMExpandPseudo::ExpandMI(MachineBasicBlock &MBB,
       MI.eraseFromParent();
       return true;
     }
-    case ARM::VDUPfqf:
-    case ARM::VDUPfdf:{
-      unsigned NewOpc = Opcode == ARM::VDUPfqf ? ARM::VDUPLN32q :
-        ARM::VDUPLN32d;
-      MachineInstrBuilder MIB =
-        BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(NewOpc));
-      unsigned OpIdx = 0;
-      unsigned SrcReg = MI.getOperand(1).getReg();
-      unsigned Lane = TRI->getEncodingValue(SrcReg) & 1;
-      unsigned DReg = TRI->getMatchingSuperReg(SrcReg,
-                            Lane & 1 ? ARM::ssub_1 : ARM::ssub_0,
-                            &ARM::DPR_VFP2RegClass);
-      // The lane is [0,1] for the containing DReg superregister.
-      // Copy the dst/src register operands.
-      MIB.addOperand(MI.getOperand(OpIdx++));
-      MIB.addReg(DReg);
-      ++OpIdx;
-      // Add the lane select operand.
-      MIB.addImm(Lane);
-      // Add the predicate operands.
-      MIB.addOperand(MI.getOperand(OpIdx++));
-      MIB.addOperand(MI.getOperand(OpIdx++));
-
-      TransferImpOps(MI, MIB, MIB);
-      MI.eraseFromParent();
-      return true;
-    }
 
     case ARM::VLD2q8Pseudo:
     case ARM::VLD2q16Pseudo:
@@ -1253,7 +1333,7 @@ bool ARMExpandPseudo::ExpandMBB(MachineBasicBlock &MBB) {
 
   MachineBasicBlock::iterator MBBI = MBB.begin(), E = MBB.end();
   while (MBBI != E) {
-    MachineBasicBlock::iterator NMBBI = llvm::next(MBBI);
+    MachineBasicBlock::iterator NMBBI = std::next(MBBI);
     Modified |= ExpandMI(MBB, MBBI);
     MBBI = NMBBI;
   }
diff --git a/contrib/llvm/lib/Target/ARM/ARMFPUName.def b/contrib/llvm/lib/Target/ARM/ARMFPUName.def
index 9a1bbe703d99..1fef3b3bc5e2 100644
--- a/contrib/llvm/lib/Target/ARM/ARMFPUName.def
+++ b/contrib/llvm/lib/Target/ARM/ARMFPUName.def
@@ -28,5 +28,6 @@ ARM_FPU_NAME("neon", NEON)
 ARM_FPU_NAME("neon-vfpv4", NEON_VFPV4)
 ARM_FPU_NAME("neon-fp-armv8", NEON_FP_ARMV8)
 ARM_FPU_NAME("crypto-neon-fp-armv8", CRYPTO_NEON_FP_ARMV8)
+ARM_FPU_NAME("softvfp", SOFTVFP)
 
 #undef ARM_FPU_NAME
diff --git a/contrib/llvm/lib/Target/ARM/ARMFastISel.cpp b/contrib/llvm/lib/Target/ARM/ARMFastISel.cpp
index a4004f32db37..e2d90cd9306c 100644
--- a/contrib/llvm/lib/Target/ARM/ARMFastISel.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMFastISel.cpp
@@ -14,11 +14,12 @@
 //===----------------------------------------------------------------------===//
 
 #include "ARM.h"
-#include "ARMBaseInstrInfo.h"
+#include "ARMBaseRegisterInfo.h"
 #include "ARMCallingConv.h"
 #include "ARMConstantPoolValue.h"
+#include "ARMISelLowering.h"
+#include "ARMMachineFunctionInfo.h"
 #include "ARMSubtarget.h"
-#include "ARMTargetMachine.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/CodeGen/Analysis.h"
@@ -30,18 +31,18 @@
 #include "llvm/CodeGen/MachineMemOperand.h"
 #include "llvm/CodeGen/MachineModuleInfo.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
+#include "llvm/IR/CallSite.h"
 #include "llvm/IR/CallingConv.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/DerivedTypes.h"
+#include "llvm/IR/GetElementPtrTypeIterator.h"
 #include "llvm/IR/GlobalVariable.h"
 #include "llvm/IR/Instructions.h"
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Operator.h"
-#include "llvm/Support/CallSite.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/ErrorHandling.h"
-#include "llvm/Support/GetElementPtrTypeIterator.h"
 #include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetLowering.h"
 #include "llvm/Target/TargetMachine.h"
@@ -73,11 +74,12 @@ namespace {
      }
   } Address;
 
-class ARMFastISel : public FastISel {
+class ARMFastISel final : public FastISel {
 
   /// Subtarget - Keep a pointer to the ARMSubtarget around so that we can
   /// make the right decision when generating code for different targets.
   const ARMSubtarget *Subtarget;
+  Module &M;
   const TargetMachine &TM;
   const TargetInstrInfo &TII;
   const TargetLowering &TLI;
@@ -91,6 +93,7 @@ class ARMFastISel : public FastISel {
     explicit ARMFastISel(FunctionLoweringInfo &funcInfo,
                          const TargetLibraryInfo *libInfo)
     : FastISel(funcInfo, libInfo),
+      M(const_cast<Module&>(*funcInfo.Fn->getParent())),
       TM(funcInfo.MF->getTarget()),
       TII(*TM.getInstrInfo()),
       TLI(*TM.getTargetLowering()) {
@@ -102,8 +105,6 @@ class ARMFastISel : public FastISel {
 
     // Code from FastISel.cpp.
   private:
-    unsigned FastEmitInst_(unsigned MachineInstOpcode,
-                           const TargetRegisterClass *RC);
     unsigned FastEmitInst_r(unsigned MachineInstOpcode,
                             const TargetRegisterClass *RC,
                             unsigned Op0, bool Op0IsKill);
@@ -120,10 +121,6 @@ class ARMFastISel : public FastISel {
                              const TargetRegisterClass *RC,
                              unsigned Op0, bool Op0IsKill,
                              uint64_t Imm);
-    unsigned FastEmitInst_rf(unsigned MachineInstOpcode,
-                             const TargetRegisterClass *RC,
-                             unsigned Op0, bool Op0IsKill,
-                             const ConstantFP *FPImm);
     unsigned FastEmitInst_rri(unsigned MachineInstOpcode,
                               const TargetRegisterClass *RC,
                               unsigned Op0, bool Op0IsKill,
@@ -132,22 +129,15 @@ class ARMFastISel : public FastISel {
     unsigned FastEmitInst_i(unsigned MachineInstOpcode,
                             const TargetRegisterClass *RC,
                             uint64_t Imm);
-    unsigned FastEmitInst_ii(unsigned MachineInstOpcode,
-                             const TargetRegisterClass *RC,
-                             uint64_t Imm1, uint64_t Imm2);
-
-    unsigned FastEmitInst_extractsubreg(MVT RetVT,
-                                        unsigned Op0, bool Op0IsKill,
-                                        uint32_t Idx);
 
     // Backend specific FastISel code.
   private:
-    virtual bool TargetSelectInstruction(const Instruction *I);
-    virtual unsigned TargetMaterializeConstant(const Constant *C);
-    virtual unsigned TargetMaterializeAlloca(const AllocaInst *AI);
-    virtual bool tryToFoldLoadIntoMI(MachineInstr *MI, unsigned OpNo,
-                                     const LoadInst *LI);
-    virtual bool FastLowerArguments();
+    bool TargetSelectInstruction(const Instruction *I) override;
+    unsigned TargetMaterializeConstant(const Constant *C) override;
+    unsigned TargetMaterializeAlloca(const AllocaInst *AI) override;
+    bool tryToFoldLoadIntoMI(MachineInstr *MI, unsigned OpNo,
+                             const LoadInst *LI) override;
+    bool FastLowerArguments() override;
   private:
   #include "ARMGenFastISel.inc"
 
@@ -176,8 +166,6 @@ class ARMFastISel : public FastISel {
 
     // Utility routines.
   private:
-    unsigned constrainOperandRegClass(const MCInstrDesc &II, unsigned OpNum,
-                                      unsigned Op);
     bool isTypeLegal(Type *Ty, MVT &VT);
     bool isLoadTypeLegal(Type *Ty, MVT &VT);
     bool ARMEmitCmp(const Value *Src1Value, const Value *Src2Value,
@@ -201,6 +189,8 @@ class ARMFastISel : public FastISel {
     unsigned ARMSelectCallOp(bool UseReg);
     unsigned ARMLowerPICELF(const GlobalValue *GV, unsigned Align, MVT VT);
 
+    const TargetLowering *getTargetLowering() { return TM.getTargetLowering(); }
+
     // Call handling routines.
   private:
     CCAssignFn *CCAssignFnForCall(CallingConv::ID CC,
@@ -293,32 +283,6 @@ ARMFastISel::AddOptionalDefs(const MachineInstrBuilder &MIB) {
   return MIB;
 }
 
-unsigned ARMFastISel::constrainOperandRegClass(const MCInstrDesc &II,
-                                               unsigned Op, unsigned OpNum) {
-  if (TargetRegisterInfo::isVirtualRegister(Op)) {
-    const TargetRegisterClass *RegClass =
-        TII.getRegClass(II, OpNum, &TRI, *FuncInfo.MF);
-    if (!MRI.constrainRegClass(Op, RegClass)) {
-      // If it's not legal to COPY between the register classes, something
-      // has gone very wrong before we got here.
-      unsigned NewOp = createResultReg(RegClass);
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
-                              TII.get(TargetOpcode::COPY), NewOp).addReg(Op));
-      return NewOp;
-    }
-  }
-  return Op;
-}
-
-unsigned ARMFastISel::FastEmitInst_(unsigned MachineInstOpcode,
-                                    const TargetRegisterClass* RC) {
-  unsigned ResultReg = createResultReg(RC);
-  const MCInstrDesc &II = TII.get(MachineInstOpcode);
-
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg));
-  return ResultReg;
-}
-
 unsigned ARMFastISel::FastEmitInst_r(unsigned MachineInstOpcode,
                                      const TargetRegisterClass *RC,
                                      unsigned Op0, bool Op0IsKill) {
@@ -329,12 +293,12 @@ unsigned ARMFastISel::FastEmitInst_r(unsigned MachineInstOpcode,
   // for this instruction.
   Op0 = constrainOperandRegClass(II, Op0, 1);
   if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addReg(Op0, Op0IsKill * RegState::Kill));
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II,
+                            ResultReg).addReg(Op0, Op0IsKill * RegState::Kill));
   } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                    .addReg(Op0, Op0IsKill * RegState::Kill));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                    TII.get(TargetOpcode::COPY), ResultReg)
                    .addReg(II.ImplicitDefs[0]));
   }
@@ -354,14 +318,15 @@ unsigned ARMFastISel::FastEmitInst_rr(unsigned MachineInstOpcode,
   Op1 = constrainOperandRegClass(II, Op1, 2);
 
   if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addReg(Op0, Op0IsKill * RegState::Kill)
-                   .addReg(Op1, Op1IsKill * RegState::Kill));
+    AddOptionalDefs(
+        BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II, ResultReg)
+            .addReg(Op0, Op0IsKill * RegState::Kill)
+            .addReg(Op1, Op1IsKill * RegState::Kill));
   } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                    .addReg(Op0, Op0IsKill * RegState::Kill)
                    .addReg(Op1, Op1IsKill * RegState::Kill));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                            TII.get(TargetOpcode::COPY), ResultReg)
                    .addReg(II.ImplicitDefs[0]));
   }
@@ -383,16 +348,17 @@ unsigned ARMFastISel::FastEmitInst_rrr(unsigned MachineInstOpcode,
   Op2 = constrainOperandRegClass(II, Op1, 3);
 
   if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addReg(Op0, Op0IsKill * RegState::Kill)
-                   .addReg(Op1, Op1IsKill * RegState::Kill)
-                   .addReg(Op2, Op2IsKill * RegState::Kill));
+    AddOptionalDefs(
+        BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II, ResultReg)
+            .addReg(Op0, Op0IsKill * RegState::Kill)
+            .addReg(Op1, Op1IsKill * RegState::Kill)
+            .addReg(Op2, Op2IsKill * RegState::Kill));
   } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                    .addReg(Op0, Op0IsKill * RegState::Kill)
                    .addReg(Op1, Op1IsKill * RegState::Kill)
                    .addReg(Op2, Op2IsKill * RegState::Kill));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                            TII.get(TargetOpcode::COPY), ResultReg)
                    .addReg(II.ImplicitDefs[0]));
   }
@@ -410,39 +376,15 @@ unsigned ARMFastISel::FastEmitInst_ri(unsigned MachineInstOpcode,
   // for this instruction.
   Op0 = constrainOperandRegClass(II, Op0, 1);
   if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addReg(Op0, Op0IsKill * RegState::Kill)
-                   .addImm(Imm));
+    AddOptionalDefs(
+        BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II, ResultReg)
+            .addReg(Op0, Op0IsKill * RegState::Kill)
+            .addImm(Imm));
   } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                    .addReg(Op0, Op0IsKill * RegState::Kill)
                    .addImm(Imm));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
-                           TII.get(TargetOpcode::COPY), ResultReg)
-                   .addReg(II.ImplicitDefs[0]));
-  }
-  return ResultReg;
-}
-
-unsigned ARMFastISel::FastEmitInst_rf(unsigned MachineInstOpcode,
-                                      const TargetRegisterClass *RC,
-                                      unsigned Op0, bool Op0IsKill,
-                                      const ConstantFP *FPImm) {
-  unsigned ResultReg = createResultReg(RC);
-  const MCInstrDesc &II = TII.get(MachineInstOpcode);
-
-  // Make sure the input operand is sufficiently constrained to be legal
-  // for this instruction.
-  Op0 = constrainOperandRegClass(II, Op0, 1);
-  if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addReg(Op0, Op0IsKill * RegState::Kill)
-                   .addFPImm(FPImm));
-  } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
-                   .addReg(Op0, Op0IsKill * RegState::Kill)
-                   .addFPImm(FPImm));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                            TII.get(TargetOpcode::COPY), ResultReg)
                    .addReg(II.ImplicitDefs[0]));
   }
@@ -462,16 +404,17 @@ unsigned ARMFastISel::FastEmitInst_rri(unsigned MachineInstOpcode,
   Op0 = constrainOperandRegClass(II, Op0, 1);
   Op1 = constrainOperandRegClass(II, Op1, 2);
   if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addReg(Op0, Op0IsKill * RegState::Kill)
-                   .addReg(Op1, Op1IsKill * RegState::Kill)
-                   .addImm(Imm));
+    AddOptionalDefs(
+        BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II, ResultReg)
+            .addReg(Op0, Op0IsKill * RegState::Kill)
+            .addReg(Op1, Op1IsKill * RegState::Kill)
+            .addImm(Imm));
   } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                    .addReg(Op0, Op0IsKill * RegState::Kill)
                    .addReg(Op1, Op1IsKill * RegState::Kill)
                    .addImm(Imm));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                            TII.get(TargetOpcode::COPY), ResultReg)
                    .addReg(II.ImplicitDefs[0]));
   }
@@ -485,58 +428,25 @@ unsigned ARMFastISel::FastEmitInst_i(unsigned MachineInstOpcode,
   const MCInstrDesc &II = TII.get(MachineInstOpcode);
 
   if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                   .addImm(Imm));
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II,
+                            ResultReg).addImm(Imm));
   } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                    .addImm(Imm));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                            TII.get(TargetOpcode::COPY), ResultReg)
                    .addReg(II.ImplicitDefs[0]));
   }
   return ResultReg;
 }
 
-unsigned ARMFastISel::FastEmitInst_ii(unsigned MachineInstOpcode,
-                                      const TargetRegisterClass *RC,
-                                      uint64_t Imm1, uint64_t Imm2) {
-  unsigned ResultReg = createResultReg(RC);
-  const MCInstrDesc &II = TII.get(MachineInstOpcode);
-
-  if (II.getNumDefs() >= 1) {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
-                    .addImm(Imm1).addImm(Imm2));
-  } else {
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
-                    .addImm(Imm1).addImm(Imm2));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
-                            TII.get(TargetOpcode::COPY),
-                            ResultReg)
-                    .addReg(II.ImplicitDefs[0]));
-  }
-  return ResultReg;
-}
-
-unsigned ARMFastISel::FastEmitInst_extractsubreg(MVT RetVT,
-                                                 unsigned Op0, bool Op0IsKill,
-                                                 uint32_t Idx) {
-  unsigned ResultReg = createResultReg(TLI.getRegClassFor(RetVT));
-  assert(TargetRegisterInfo::isVirtualRegister(Op0) &&
-         "Cannot yet extract from physregs");
-
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
-                          DL, TII.get(TargetOpcode::COPY), ResultReg)
-                  .addReg(Op0, getKillRegState(Op0IsKill), Idx));
-  return ResultReg;
-}
-
 // TODO: Don't worry about 64-bit now, but when this is fixed remove the
 // checks from the various callers.
 unsigned ARMFastISel::ARMMoveToFPReg(MVT VT, unsigned SrcReg) {
   if (VT == MVT::f64) return 0;
 
   unsigned MoveReg = createResultReg(TLI.getRegClassFor(VT));
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(ARM::VMOVSR), MoveReg)
                   .addReg(SrcReg));
   return MoveReg;
@@ -546,7 +456,7 @@ unsigned ARMFastISel::ARMMoveToIntReg(MVT VT, unsigned SrcReg) {
   if (VT == MVT::i64) return 0;
 
   unsigned MoveReg = createResultReg(TLI.getRegClassFor(VT));
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(ARM::VMOVRS), MoveReg)
                   .addReg(SrcReg));
   return MoveReg;
@@ -572,9 +482,8 @@ unsigned ARMFastISel::ARMMaterializeFP(const ConstantFP *CFP, MVT VT) {
       Opc = ARM::FCONSTS;
     }
     unsigned DestReg = createResultReg(TLI.getRegClassFor(VT));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc),
-                            DestReg)
-                    .addImm(Imm));
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                            TII.get(Opc), DestReg).addImm(Imm));
     return DestReg;
   }
 
@@ -582,20 +491,20 @@ unsigned ARMFastISel::ARMMaterializeFP(const ConstantFP *CFP, MVT VT) {
   if (!Subtarget->hasVFP2()) return false;
 
   // MachineConstantPool wants an explicit alignment.
-  unsigned Align = TD.getPrefTypeAlignment(CFP->getType());
+  unsigned Align = DL.getPrefTypeAlignment(CFP->getType());
   if (Align == 0) {
     // TODO: Figure out if this is correct.
-    Align = TD.getTypeAllocSize(CFP->getType());
+    Align = DL.getTypeAllocSize(CFP->getType());
   }
   unsigned Idx = MCP.getConstantPoolIndex(cast<Constant>(CFP), Align);
   unsigned DestReg = createResultReg(TLI.getRegClassFor(VT));
   unsigned Opc = is64bit ? ARM::VLDRD : ARM::VLDRS;
 
   // The extra reg is for addrmode5.
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc),
-                          DestReg)
-                  .addConstantPoolIndex(Idx)
-                  .addReg(0));
+  AddOptionalDefs(
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(Opc), DestReg)
+          .addConstantPoolIndex(Idx)
+          .addReg(0));
   return DestReg;
 }
 
@@ -612,7 +521,7 @@ unsigned ARMFastISel::ARMMaterializeInt(const Constant *C, MVT VT) {
     const TargetRegisterClass *RC = isThumb2 ? &ARM::rGPRRegClass :
       &ARM::GPRRegClass;
     unsigned ImmReg = createResultReg(RC);
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(Opc), ImmReg)
                     .addImm(CI->getZExtValue()));
     return ImmReg;
@@ -626,7 +535,7 @@ unsigned ARMFastISel::ARMMaterializeInt(const Constant *C, MVT VT) {
     if (UseImm) {
       unsigned Opc = isThumb2 ? ARM::t2MVNi : ARM::MVNi;
       unsigned ImmReg = createResultReg(TLI.getRegClassFor(MVT::i32));
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                               TII.get(Opc), ImmReg)
                       .addImm(Imm));
       return ImmReg;
@@ -640,24 +549,25 @@ unsigned ARMFastISel::ARMMaterializeInt(const Constant *C, MVT VT) {
   unsigned DestReg = createResultReg(TLI.getRegClassFor(VT));
 
   // MachineConstantPool wants an explicit alignment.
-  unsigned Align = TD.getPrefTypeAlignment(C->getType());
+  unsigned Align = DL.getPrefTypeAlignment(C->getType());
   if (Align == 0) {
     // TODO: Figure out if this is correct.
-    Align = TD.getTypeAllocSize(C->getType());
+    Align = DL.getTypeAllocSize(C->getType());
   }
   unsigned Idx = MCP.getConstantPoolIndex(C, Align);
 
   if (isThumb2)
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(ARM::t2LDRpci), DestReg)
                     .addConstantPoolIndex(Idx));
-  else
+  else {
     // The extra immediate is for addrmode2.
     DestReg = constrainOperandRegClass(TII.get(ARM::LDRcp), DestReg, 0);
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(ARM::LDRcp), DestReg)
                     .addConstantPoolIndex(Idx)
                     .addImm(0));
+  }
 
   return DestReg;
 }
@@ -673,37 +583,36 @@ unsigned ARMFastISel::ARMMaterializeGV(const GlobalValue *GV, MVT VT) {
     (const TargetRegisterClass*)&ARM::GPRRegClass;
   unsigned DestReg = createResultReg(RC);
 
-  // FastISel TLS support on non-Darwin is broken, punt to SelectionDAG.
+  // FastISel TLS support on non-MachO is broken, punt to SelectionDAG.
   const GlobalVariable *GVar = dyn_cast<GlobalVariable>(GV);
   bool IsThreadLocal = GVar && GVar->isThreadLocal();
-  if (!Subtarget->isTargetDarwin() && IsThreadLocal) return 0;
+  if (!Subtarget->isTargetMachO() && IsThreadLocal) return 0;
 
   // Use movw+movt when possible, it avoids constant pool entries.
-  // Darwin targets don't support movt with Reloc::Static, see
-  // ARMTargetLowering::LowerGlobalAddressDarwin.  Other targets only support
-  // static movt relocations.
-  if (Subtarget->useMovt() &&
-      Subtarget->isTargetDarwin() == (RelocM != Reloc::Static)) {
+  // Non-darwin targets only support static movt relocations in FastISel.
+  if (Subtarget->useMovt(*FuncInfo.MF) &&
+      (Subtarget->isTargetMachO() || RelocM == Reloc::Static)) {
     unsigned Opc;
+    unsigned char TF = 0;
+    if (Subtarget->isTargetMachO())
+      TF = ARMII::MO_NONLAZY;
+
     switch (RelocM) {
     case Reloc::PIC_:
       Opc = isThumb2 ? ARM::t2MOV_ga_pcrel : ARM::MOV_ga_pcrel;
       break;
-    case Reloc::DynamicNoPIC:
-      Opc = isThumb2 ? ARM::t2MOV_ga_dyn : ARM::MOV_ga_dyn;
-      break;
     default:
       Opc = isThumb2 ? ARM::t2MOVi32imm : ARM::MOVi32imm;
       break;
     }
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc),
-                            DestReg).addGlobalAddress(GV));
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                            TII.get(Opc), DestReg).addGlobalAddress(GV, 0, TF));
   } else {
     // MachineConstantPool wants an explicit alignment.
-    unsigned Align = TD.getPrefTypeAlignment(GV->getType());
+    unsigned Align = DL.getPrefTypeAlignment(GV->getType());
     if (Align == 0) {
       // TODO: Figure out if this is correct.
-      Align = TD.getTypeAllocSize(GV->getType());
+      Align = DL.getTypeAllocSize(GV->getType());
     }
 
     if (Subtarget->isTargetELF() && RelocM == Reloc::PIC_)
@@ -722,18 +631,18 @@ unsigned ARMFastISel::ARMMaterializeGV(const GlobalValue *GV, MVT VT) {
     MachineInstrBuilder MIB;
     if (isThumb2) {
       unsigned Opc = (RelocM!=Reloc::PIC_) ? ARM::t2LDRpci : ARM::t2LDRpci_pic;
-      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc), DestReg)
-        .addConstantPoolIndex(Idx);
+      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(Opc),
+                    DestReg).addConstantPoolIndex(Idx);
       if (RelocM == Reloc::PIC_)
         MIB.addImm(Id);
       AddOptionalDefs(MIB);
     } else {
       // The extra immediate is for addrmode2.
       DestReg = constrainOperandRegClass(TII.get(ARM::LDRcp), DestReg, 0);
-      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(ARM::LDRcp),
-                    DestReg)
-        .addConstantPoolIndex(Idx)
-        .addImm(0);
+      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                    TII.get(ARM::LDRcp), DestReg)
+                .addConstantPoolIndex(Idx)
+                .addImm(0);
       AddOptionalDefs(MIB);
 
       if (RelocM == Reloc::PIC_) {
@@ -741,7 +650,7 @@ unsigned ARMFastISel::ARMMaterializeGV(const GlobalValue *GV, MVT VT) {
         unsigned NewDestReg = createResultReg(TLI.getRegClassFor(VT));
 
         MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
-                                          DL, TII.get(Opc), NewDestReg)
+                                          DbgLoc, TII.get(Opc), NewDestReg)
                                   .addReg(DestReg)
                                   .addImm(Id);
         AddOptionalDefs(MIB);
@@ -754,15 +663,15 @@ unsigned ARMFastISel::ARMMaterializeGV(const GlobalValue *GV, MVT VT) {
     MachineInstrBuilder MIB;
     unsigned NewDestReg = createResultReg(TLI.getRegClassFor(VT));
     if (isThumb2)
-      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                     TII.get(ARM::t2LDRi12), NewDestReg)
             .addReg(DestReg)
             .addImm(0);
     else
-      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(ARM::LDRi12),
-                    NewDestReg)
-            .addReg(DestReg)
-            .addImm(0);
+      MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                    TII.get(ARM::LDRi12), NewDestReg)
+                .addReg(DestReg)
+                .addImm(0);
     DestReg = NewDestReg;
     AddOptionalDefs(MIB);
   }
@@ -802,10 +711,12 @@ unsigned ARMFastISel::TargetMaterializeAlloca(const AllocaInst *AI) {
   // This will get lowered later into the correct offsets and registers
   // via rewriteXFrameIndex.
   if (SI != FuncInfo.StaticAllocaMap.end()) {
+    unsigned Opc = isThumb2 ? ARM::t2ADDri : ARM::ADDri;
     const TargetRegisterClass* RC = TLI.getRegClassFor(VT);
     unsigned ResultReg = createResultReg(RC);
-    unsigned Opc = isThumb2 ? ARM::t2ADDri : ARM::ADDri;
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    ResultReg = constrainOperandRegClass(TII.get(Opc), ResultReg, 0);
+
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(Opc), ResultReg)
                             .addFrameIndex(SI->second)
                             .addImm(0));
@@ -841,7 +752,7 @@ bool ARMFastISel::isLoadTypeLegal(Type *Ty, MVT &VT) {
 // Computes the address to get to an object.
 bool ARMFastISel::ARMComputeAddress(const Value *Obj, Address &Addr) {
   // Some boilerplate from the X86 FastISel.
-  const User *U = NULL;
+  const User *U = nullptr;
   unsigned Opcode = Instruction::UserOp1;
   if (const Instruction *I = dyn_cast<Instruction>(Obj)) {
     // Don't walk into other basic blocks unless the object is an alloca from
@@ -889,11 +800,11 @@ bool ARMFastISel::ARMComputeAddress(const Value *Obj, Address &Addr) {
            i != e; ++i, ++GTI) {
         const Value *Op = *i;
         if (StructType *STy = dyn_cast<StructType>(*GTI)) {
-          const StructLayout *SL = TD.getStructLayout(STy);
+          const StructLayout *SL = DL.getStructLayout(STy);
           unsigned Idx = cast<ConstantInt>(Op)->getZExtValue();
           TmpOffset += SL->getElementOffset(Idx);
         } else {
-          uint64_t S = TD.getTypeAllocSize(GTI.getIndexedType());
+          uint64_t S = DL.getTypeAllocSize(GTI.getIndexedType());
           for (;;) {
             if (const ConstantInt *CI = dyn_cast<ConstantInt>(Op)) {
               // Constant-offset addressing.
@@ -979,7 +890,7 @@ void ARMFastISel::ARMSimplifyAddress(Address &Addr, MVT VT, bool useAM3) {
       (const TargetRegisterClass*)&ARM::GPRRegClass;
     unsigned ResultReg = createResultReg(RC);
     unsigned Opc = isThumb2 ? ARM::t2ADDri : ARM::ADDri;
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(Opc), ResultReg)
                             .addFrameIndex(Addr.Base.FI)
                             .addImm(0));
@@ -1130,7 +1041,7 @@ bool ARMFastISel::ARMEmitLoad(MVT VT, unsigned &ResultReg, Address &Addr,
   if (allocReg)
     ResultReg = createResultReg(RC);
   assert (ResultReg > 255 && "Expected an allocated virtual register.");
-  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                                     TII.get(Opc), ResultReg);
   AddLoadStoreOperands(VT, Addr, MIB, MachineMemOperand::MOLoad, useAM3);
 
@@ -1138,7 +1049,7 @@ bool ARMFastISel::ARMEmitLoad(MVT VT, unsigned &ResultReg, Address &Addr,
   // load.  Now we must move from the GRP to the FP register.
   if (needVMOV) {
     unsigned MoveReg = createResultReg(TLI.getRegClassFor(MVT::f32));
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(ARM::VMOVSR), MoveReg)
                     .addReg(ResultReg));
     ResultReg = MoveReg;
@@ -1180,7 +1091,7 @@ bool ARMFastISel::ARMEmitStore(MVT VT, unsigned SrcReg, Address &Addr,
         (const TargetRegisterClass*)&ARM::GPRRegClass);
       unsigned Opc = isThumb2 ? ARM::t2ANDri : ARM::ANDri;
       SrcReg = constrainOperandRegClass(TII.get(Opc), SrcReg, 1);
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                               TII.get(Opc), Res)
                       .addReg(SrcReg).addImm(1));
       SrcReg = Res;
@@ -1227,7 +1138,7 @@ bool ARMFastISel::ARMEmitStore(MVT VT, unsigned SrcReg, Address &Addr,
       // Unaligned stores need special handling. Floats require word-alignment.
       if (Alignment && Alignment < 4) {
         unsigned MoveReg = createResultReg(TLI.getRegClassFor(MVT::i32));
-        AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+        AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                                 TII.get(ARM::VMOVRS), MoveReg)
                         .addReg(SrcReg));
         SrcReg = MoveReg;
@@ -1252,7 +1163,7 @@ bool ARMFastISel::ARMEmitStore(MVT VT, unsigned SrcReg, Address &Addr,
 
   // Create the base instruction, then add the operands.
   SrcReg = constrainOperandRegClass(TII.get(StrOpc), SrcReg, 0);
-  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                                     TII.get(StrOpc))
                             .addReg(SrcReg);
   AddLoadStoreOperands(VT, Addr, MIB, MachineMemOperand::MOStore, useAM3);
@@ -1363,9 +1274,9 @@ bool ARMFastISel::SelectBranch(const Instruction *I) {
         return false;
 
       unsigned BrOpc = isThumb2 ? ARM::t2Bcc : ARM::Bcc;
-      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(BrOpc))
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(BrOpc))
       .addMBB(TBB).addImm(ARMPred).addReg(ARM::CPSR);
-      FastEmitBranch(FBB, DL);
+      FastEmitBranch(FBB, DbgLoc);
       FuncInfo.MBB->addSuccessor(TBB);
       return true;
     }
@@ -1376,7 +1287,7 @@ bool ARMFastISel::SelectBranch(const Instruction *I) {
       unsigned TstOpc = isThumb2 ? ARM::t2TSTri : ARM::TSTri;
       unsigned OpReg = getRegForValue(TI->getOperand(0));
       OpReg = constrainOperandRegClass(TII.get(TstOpc), OpReg, 0);
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                               TII.get(TstOpc))
                       .addReg(OpReg).addImm(1));
 
@@ -1387,10 +1298,10 @@ bool ARMFastISel::SelectBranch(const Instruction *I) {
       }
 
       unsigned BrOpc = isThumb2 ? ARM::t2Bcc : ARM::Bcc;
-      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(BrOpc))
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(BrOpc))
       .addMBB(TBB).addImm(CCMode).addReg(ARM::CPSR);
 
-      FastEmitBranch(FBB, DL);
+      FastEmitBranch(FBB, DbgLoc);
       FuncInfo.MBB->addSuccessor(TBB);
       return true;
     }
@@ -1398,7 +1309,7 @@ bool ARMFastISel::SelectBranch(const Instruction *I) {
              dyn_cast<ConstantInt>(BI->getCondition())) {
     uint64_t Imm = CI->getZExtValue();
     MachineBasicBlock *Target = (Imm == 0) ? FBB : TBB;
-    FastEmitBranch(Target, DL);
+    FastEmitBranch(Target, DbgLoc);
     return true;
   }
 
@@ -1414,8 +1325,10 @@ bool ARMFastISel::SelectBranch(const Instruction *I) {
   // the one-bit value left in the virtual register.
   unsigned TstOpc = isThumb2 ? ARM::t2TSTri : ARM::TSTri;
   CmpReg = constrainOperandRegClass(TII.get(TstOpc), CmpReg, 0);
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(TstOpc))
-                  .addReg(CmpReg).addImm(1));
+  AddOptionalDefs(
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(TstOpc))
+          .addReg(CmpReg)
+          .addImm(1));
 
   unsigned CCMode = ARMCC::NE;
   if (FuncInfo.MBB->isLayoutSuccessor(TBB)) {
@@ -1424,9 +1337,9 @@ bool ARMFastISel::SelectBranch(const Instruction *I) {
   }
 
   unsigned BrOpc = isThumb2 ? ARM::t2Bcc : ARM::Bcc;
-  BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(BrOpc))
+  BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(BrOpc))
                   .addMBB(TBB).addImm(CCMode).addReg(ARM::CPSR);
-  FastEmitBranch(FBB, DL);
+  FastEmitBranch(FBB, DbgLoc);
   FuncInfo.MBB->addSuccessor(TBB);
   return true;
 }
@@ -1436,8 +1349,8 @@ bool ARMFastISel::SelectIndirectBr(const Instruction *I) {
   if (AddrReg == 0) return false;
 
   unsigned Opc = isThumb2 ? ARM::tBRIND : ARM::BX;
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc))
-                  .addReg(AddrReg));
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                          TII.get(Opc)).addReg(AddrReg));
 
   const IndirectBrInst *IB = cast<IndirectBrInst>(I);
   for (unsigned i = 0, e = IB->getNumSuccessors(); i != e; ++i)
@@ -1470,7 +1383,7 @@ bool ARMFastISel::ARMEmitCmp(const Value *Src1Value, const Value *Src2Value,
       const APInt &CIVal = ConstInt->getValue();
       Imm = (isZExt) ? (int)CIVal.getZExtValue() : (int)CIVal.getSExtValue();
       // For INT_MIN/LONG_MIN (i.e., 0x80000000) we need to use a cmp, rather
-      // then a cmn, because there is no way to represent 2147483648 as a 
+      // then a cmn, because there is no way to represent 2147483648 as a
       // signed 32-bit int.
       if (Imm < 0 && Imm != (int)0x80000000) {
         isNegativeImm = true;
@@ -1542,11 +1455,11 @@ bool ARMFastISel::ARMEmitCmp(const Value *Src1Value, const Value *Src2Value,
   SrcReg1 = constrainOperandRegClass(II, SrcReg1, 0);
   if (!UseImm) {
     SrcReg2 = constrainOperandRegClass(II, SrcReg2, 1);
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
                     .addReg(SrcReg1).addReg(SrcReg2));
   } else {
     MachineInstrBuilder MIB;
-    MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+    MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, II)
       .addReg(SrcReg1);
 
     // Only add immediate for icmp as the immediate for fcmp is an implicit 0.0.
@@ -1558,7 +1471,7 @@ bool ARMFastISel::ARMEmitCmp(const Value *Src1Value, const Value *Src2Value,
   // For floating point we need to move the result to a comparison register
   // that we can then use for branches.
   if (Ty->isFloatTy() || Ty->isDoubleTy())
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(ARM::FMSTAT)));
   return true;
 }
@@ -1586,7 +1499,7 @@ bool ARMFastISel::SelectCmp(const Instruction *I) {
   Constant *Zero = ConstantInt::get(Type::getInt32Ty(*Context), 0);
   unsigned ZeroReg = TargetMaterializeConstant(Zero);
   // ARMEmitCmp emits a FMSTAT when necessary, so it's always safe to use CPSR.
-  BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(MovCCOpc), DestReg)
+  BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(MovCCOpc), DestReg)
           .addReg(ZeroReg).addImm(1)
           .addImm(ARMPred).addReg(ARM::CPSR);
 
@@ -1606,7 +1519,7 @@ bool ARMFastISel::SelectFPExt(const Instruction *I) {
   if (Op == 0) return false;
 
   unsigned Result = createResultReg(&ARM::DPRRegClass);
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(ARM::VCVTDS), Result)
                   .addReg(Op));
   UpdateValueMap(I, Result);
@@ -1625,7 +1538,7 @@ bool ARMFastISel::SelectFPTrunc(const Instruction *I) {
   if (Op == 0) return false;
 
   unsigned Result = createResultReg(&ARM::SPRRegClass);
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(ARM::VCVTSD), Result)
                   .addReg(Op));
   UpdateValueMap(I, Result);
@@ -1670,9 +1583,8 @@ bool ARMFastISel::SelectIToFP(const Instruction *I, bool isSigned) {
   else return false;
 
   unsigned ResultReg = createResultReg(TLI.getRegClassFor(DstVT));
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc),
-                          ResultReg)
-                  .addReg(FP));
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                          TII.get(Opc), ResultReg).addReg(FP));
   UpdateValueMap(I, ResultReg);
   return true;
 }
@@ -1697,9 +1609,8 @@ bool ARMFastISel::SelectFPToI(const Instruction *I, bool isSigned) {
 
   // f64->s32/u32 or f32->s32/u32 both need an intermediate f32 reg.
   unsigned ResultReg = createResultReg(TLI.getRegClassFor(MVT::f32));
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opc),
-                          ResultReg)
-                  .addReg(Op));
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+                          TII.get(Opc), ResultReg).addReg(Op));
 
   // This result needs to be in an integer register, but the conversion only
   // takes place in fp-regs.
@@ -1746,8 +1657,10 @@ bool ARMFastISel::SelectSelect(const Instruction *I) {
 
   unsigned CmpOpc = isThumb2 ? ARM::t2CMPri : ARM::CMPri;
   CondReg = constrainOperandRegClass(TII.get(CmpOpc), CondReg, 0);
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(CmpOpc))
-                  .addReg(CondReg).addImm(0));
+  AddOptionalDefs(
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(CmpOpc))
+          .addReg(CondReg)
+          .addImm(0));
 
   unsigned MovCCOpc;
   const TargetRegisterClass *RC;
@@ -1765,12 +1678,20 @@ bool ARMFastISel::SelectSelect(const Instruction *I) {
   if (!UseImm) {
     Op2Reg = constrainOperandRegClass(TII.get(MovCCOpc), Op2Reg, 1);
     Op1Reg = constrainOperandRegClass(TII.get(MovCCOpc), Op1Reg, 2);
-    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(MovCCOpc), ResultReg)
-    .addReg(Op2Reg).addReg(Op1Reg).addImm(ARMCC::NE).addReg(ARM::CPSR);
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(MovCCOpc),
+            ResultReg)
+        .addReg(Op2Reg)
+        .addReg(Op1Reg)
+        .addImm(ARMCC::NE)
+        .addReg(ARM::CPSR);
   } else {
     Op1Reg = constrainOperandRegClass(TII.get(MovCCOpc), Op1Reg, 1);
-    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(MovCCOpc), ResultReg)
-    .addReg(Op1Reg).addImm(Imm).addImm(ARMCC::EQ).addReg(ARM::CPSR);
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(MovCCOpc),
+            ResultReg)
+        .addReg(Op1Reg)
+        .addImm(Imm)
+        .addImm(ARMCC::EQ)
+        .addReg(ARM::CPSR);
   }
   UpdateValueMap(I, ResultReg);
   return true;
@@ -1859,7 +1780,7 @@ bool ARMFastISel::SelectBinaryIntOp(const Instruction *I, unsigned ISDOpcode) {
   unsigned ResultReg = createResultReg(&ARM::GPRnopcRegClass);
   SrcReg1 = constrainOperandRegClass(TII.get(Opc), SrcReg1, 1);
   SrcReg2 = constrainOperandRegClass(TII.get(Opc), SrcReg2, 2);
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(Opc), ResultReg)
                   .addReg(SrcReg1).addReg(SrcReg2));
   UpdateValueMap(I, ResultReg);
@@ -1901,7 +1822,7 @@ bool ARMFastISel::SelectBinaryFPOp(const Instruction *I, unsigned ISDOpcode) {
   if (Op2 == 0) return false;
 
   unsigned ResultReg = createResultReg(TLI.getRegClassFor(VT.SimpleTy));
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(Opc), ResultReg)
                   .addReg(Op1).addReg(Op2));
   UpdateValueMap(I, ResultReg);
@@ -2013,7 +1934,7 @@ bool ARMFastISel::ProcessCallArgs(SmallVectorImpl<Value*> &Args,
 
   // Issue CALLSEQ_START
   unsigned AdjStackDown = TII.getCallFrameSetupOpcode();
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(AdjStackDown))
                   .addImm(NumBytes));
 
@@ -2058,9 +1979,8 @@ bool ARMFastISel::ProcessCallArgs(SmallVectorImpl<Value*> &Args,
 
     // Now copy/store arg to correct locations.
     if (VA.isRegLoc() && !VA.needsCustom()) {
-      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(TargetOpcode::COPY),
-              VA.getLocReg())
-        .addReg(Arg);
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+              TII.get(TargetOpcode::COPY), VA.getLocReg()).addReg(Arg);
       RegArgs.push_back(VA.getLocReg());
     } else if (VA.needsCustom()) {
       // TODO: We need custom lowering for vector (v2f64) args.
@@ -2072,7 +1992,7 @@ bool ARMFastISel::ProcessCallArgs(SmallVectorImpl<Value*> &Args,
       assert(VA.isRegLoc() && NextVA.isRegLoc() &&
              "We only handle register args!");
 
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                               TII.get(ARM::VMOVRRD), VA.getLocReg())
                       .addReg(NextVA.getLocReg(), RegState::Define)
                       .addReg(Arg));
@@ -2099,7 +2019,7 @@ bool ARMFastISel::FinishCall(MVT RetVT, SmallVectorImpl<unsigned> &UsedRegs,
                              unsigned &NumBytes, bool isVarArg) {
   // Issue CALLSEQ_END
   unsigned AdjStackUp = TII.getCallFrameDestroyOpcode();
-  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                           TII.get(AdjStackUp))
                   .addImm(NumBytes).addImm(0));
 
@@ -2116,7 +2036,7 @@ bool ARMFastISel::FinishCall(MVT RetVT, SmallVectorImpl<unsigned> &UsedRegs,
       MVT DestVT = RVLocs[0].getValVT();
       const TargetRegisterClass* DstRC = TLI.getRegClassFor(DestVT);
       unsigned ResultReg = createResultReg(DstRC);
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                               TII.get(ARM::VMOVDRR), ResultReg)
                       .addReg(RVLocs[0].getLocReg())
                       .addReg(RVLocs[1].getLocReg()));
@@ -2137,7 +2057,8 @@ bool ARMFastISel::FinishCall(MVT RetVT, SmallVectorImpl<unsigned> &UsedRegs,
       const TargetRegisterClass* DstRC = TLI.getRegClassFor(CopyVT);
 
       unsigned ResultReg = createResultReg(DstRC);
-      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(TargetOpcode::COPY),
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+              TII.get(TargetOpcode::COPY),
               ResultReg).addReg(RVLocs[0].getLocReg());
       UsedRegs.push_back(RVLocs[0].getLocReg());
 
@@ -2214,15 +2135,15 @@ bool ARMFastISel::SelectRet(const Instruction *I) {
     // Avoid a cross-class copy. This is very unlikely.
     if (!SrcRC->contains(DstReg))
       return false;
-    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(TargetOpcode::COPY),
-            DstReg).addReg(SrcReg);
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+            TII.get(TargetOpcode::COPY), DstReg).addReg(SrcReg);
 
     // Add register to return instruction.
     RetRegs.push_back(VA.getLocReg());
   }
 
   unsigned RetOpc = isThumb2 ? ARM::tBX_RET : ARM::BX_RET;
-  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                                     TII.get(RetOpc));
   AddOptionalDefs(MIB);
   for (unsigned i = 0, e = RetRegs.size(); i != e; ++i)
@@ -2243,8 +2164,9 @@ unsigned ARMFastISel::getLibcallReg(const Twine &Name) {
   EVT LCREVT = TLI.getValueType(GVTy);
   if (!LCREVT.isSimple()) return 0;
 
-  GlobalValue *GV = new GlobalVariable(Type::getInt32Ty(*Context), false,
-                                       GlobalValue::ExternalLinkage, 0, Name);
+  GlobalValue *GV = new GlobalVariable(M, Type::getInt32Ty(*Context), false,
+                                       GlobalValue::ExternalLinkage, nullptr,
+                                       Name);
   assert(GV->getType() == GVTy && "We miscomputed the type for the global!");
   return ARMMaterializeGV(GV, LCREVT.getSimpleVT());
 }
@@ -2295,7 +2217,7 @@ bool ARMFastISel::ARMEmitLibcall(const Instruction *I, RTLIB::Libcall Call) {
     if (!isTypeLegal(ArgTy, ArgVT)) return false;
 
     ISD::ArgFlagsTy Flags;
-    unsigned OriginalAlignment = TD.getABITypeAlignment(ArgTy);
+    unsigned OriginalAlignment = DL.getABITypeAlignment(ArgTy);
     Flags.setOrigAlign(OriginalAlignment);
 
     Args.push_back(Op);
@@ -2320,7 +2242,7 @@ bool ARMFastISel::ARMEmitLibcall(const Instruction *I, RTLIB::Libcall Call) {
   // Issue the call.
   unsigned CallOpc = ARMSelectCallOp(EnableARMLongCalls);
   MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
-                                    DL, TII.get(CallOpc));
+                                    DbgLoc, TII.get(CallOpc));
   // BL / BLX don't take a predicate, but tBL / tBLX do.
   if (isThumb2)
     AddDefaultPred(MIB);
@@ -2348,7 +2270,7 @@ bool ARMFastISel::ARMEmitLibcall(const Instruction *I, RTLIB::Libcall Call) {
 }
 
 bool ARMFastISel::SelectCall(const Instruction *I,
-                             const char *IntrMemName = 0) {
+                             const char *IntrMemName = nullptr) {
   const CallInst *CI = cast<CallInst>(I);
   const Value *Callee = CI->getCalledValue();
 
@@ -2428,7 +2350,7 @@ bool ARMFastISel::SelectCall(const Instruction *I,
     if (Arg == 0)
       return false;
 
-    unsigned OriginalAlignment = TD.getABITypeAlignment(ArgTy);
+    unsigned OriginalAlignment = DL.getABITypeAlignment(ArgTy);
     Flags.setOrigAlign(OriginalAlignment);
 
     Args.push_back(*i);
@@ -2461,7 +2383,7 @@ bool ARMFastISel::SelectCall(const Instruction *I,
   // Issue the call.
   unsigned CallOpc = ARMSelectCallOp(UseReg);
   MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
-                                    DL, TII.get(CallOpc));
+                                    DbgLoc, TII.get(CallOpc));
 
   unsigned char OpFlags = 0;
 
@@ -2578,7 +2500,7 @@ bool ARMFastISel::SelectIntrinsicCall(const IntrinsicInst &I) {
     unsigned Depth = cast<ConstantInt>(I.getOperand(0))->getZExtValue();
     while (Depth--) {
       DestReg = createResultReg(RC);
-      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+      AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                               TII.get(LdrOpc), DestReg)
                       .addReg(SrcReg).addImm(0));
       SrcReg = DestReg;
@@ -2635,7 +2557,7 @@ bool ARMFastISel::SelectIntrinsicCall(const IntrinsicInst &I) {
     return SelectCall(&I, "memset");
   }
   case Intrinsic::trap: {
-    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(
       Subtarget->useNaClTrap() ? ARM::TRAPNaCl : ARM::TRAP));
     return true;
   }
@@ -2788,7 +2710,7 @@ unsigned ARMFastISel::ARMEmitIntExt(MVT SrcVT, unsigned SrcReg, MVT DestVT,
     unsigned ImmEnc = ImmIsSO ? ARM_AM::getSORegOpc(ShiftAM, Imm) : Imm;
     bool isKill = 1 == Instr;
     MachineInstrBuilder MIB = BuildMI(
-        *FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(Opcode), ResultReg);
+        *FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(Opcode), ResultReg);
     if (setsCPSR)
       MIB.addReg(ARM::CPSR, RegState::Define);
     SrcReg = constrainOperandRegClass(TII.get(Opcode), SrcReg, 1 + setsCPSR);
@@ -2866,7 +2788,7 @@ bool ARMFastISel::SelectShift(const Instruction *I,
   unsigned ResultReg = createResultReg(&ARM::GPRnopcRegClass);
   if(ResultReg == 0) return false;
 
-  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                                     TII.get(Opc), ResultReg)
                             .addReg(Reg1);
 
@@ -3027,7 +2949,7 @@ unsigned ARMFastISel::ARMLowerPICELF(const GlobalValue *GV,
   // Load value.
   if (isThumb2) {
     DestReg1 = constrainOperandRegClass(TII.get(ARM::t2LDRpci), DestReg1, 0);
-    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
                             TII.get(ARM::t2LDRpci), DestReg1)
                     .addConstantPoolIndex(Idx));
     Opc = UseGOTOFF ? ARM::t2ADDrr : ARM::t2LDRs;
@@ -3035,7 +2957,7 @@ unsigned ARMFastISel::ARMLowerPICELF(const GlobalValue *GV,
     // The extra immediate is for addrmode2.
     DestReg1 = constrainOperandRegClass(TII.get(ARM::LDRcp), DestReg1, 0);
     AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
-                            DL, TII.get(ARM::LDRcp), DestReg1)
+                            DbgLoc, TII.get(ARM::LDRcp), DestReg1)
                     .addConstantPoolIndex(Idx).addImm(0));
     Opc = UseGOTOFF ? ARM::ADDrr : ARM::LDRrs;
   }
@@ -3051,7 +2973,7 @@ unsigned ARMFastISel::ARMLowerPICELF(const GlobalValue *GV,
   DestReg1 = constrainOperandRegClass(TII.get(Opc), DestReg1, 1);
   GlobalBaseReg = constrainOperandRegClass(TII.get(Opc), GlobalBaseReg, 2);
   MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
-                                    DL, TII.get(Opc), DestReg2)
+                                    DbgLoc, TII.get(Opc), DestReg2)
                             .addReg(DestReg1)
                             .addReg(GlobalBaseReg);
   if (!UseGOTOFF)
@@ -3125,7 +3047,8 @@ bool ARMFastISel::FastLowerArguments() {
     // Without this, EmitLiveInCopies may eliminate the livein if its only
     // use is a bitcast (which isn't turned into an instruction).
     unsigned ResultReg = createResultReg(RC);
-    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(TargetOpcode::COPY),
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+            TII.get(TargetOpcode::COPY),
             ResultReg).addReg(DstReg, getKillRegState(true));
     UpdateValueMap(I, ResultReg);
   }
@@ -3141,7 +3064,7 @@ namespace llvm {
     const ARMSubtarget *Subtarget = &TM.getSubtarget<ARMSubtarget>();
     // Thumb2 support on iOS; ARM support on iOS, Linux and NaCl.
     bool UseFastISel = false;
-    UseFastISel |= Subtarget->isTargetIOS() && !Subtarget->isThumb1Only();
+    UseFastISel |= Subtarget->isTargetMachO() && !Subtarget->isThumb1Only();
     UseFastISel |= Subtarget->isTargetLinux() && !Subtarget->isThumb();
     UseFastISel |= Subtarget->isTargetNaCl() && !Subtarget->isThumb();
 
@@ -3153,6 +3076,6 @@ namespace llvm {
       TM.Options.NoFramePointerElim = true;
       return new ARMFastISel(funcInfo, libInfo);
     }
-    return 0;
+    return nullptr;
   }
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMFeatures.h b/contrib/llvm/lib/Target/ARM/ARMFeatures.h
index dafc4b3a82bd..e191a3c779f1 100644
--- a/contrib/llvm/lib/Target/ARM/ARMFeatures.h
+++ b/contrib/llvm/lib/Target/ARM/ARMFeatures.h
@@ -1,4 +1,4 @@
-//===-- ARMFeatures.h - Checks for ARM instruction features ------*- C++ -*-===//
+//===-- ARMFeatures.h - Checks for ARM instruction features -----*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -14,37 +14,27 @@
 #ifndef TARGET_ARM_FEATURES_H
 #define TARGET_ARM_FEATURES_H
 
-#include "ARM.h"
+#include "MCTargetDesc/ARMMCTargetDesc.h"
 
 namespace llvm {
 
 template<typename InstrType> // could be MachineInstr or MCInst
-inline bool isV8EligibleForIT(InstrType *Instr, int BLXOperandIndex = 0) {
+bool IsCPSRDead(InstrType *Instr);
+
+template<typename InstrType> // could be MachineInstr or MCInst
+inline bool isV8EligibleForIT(InstrType *Instr) {
   switch (Instr->getOpcode()) {
   default:
     return false;
   case ARM::tADC:
   case ARM::tADDi3:
   case ARM::tADDi8:
-  case ARM::tADDrSPi:
   case ARM::tADDrr:
   case ARM::tAND:
   case ARM::tASRri:
   case ARM::tASRrr:
   case ARM::tBIC:
-  case ARM::tCMNz:
-  case ARM::tCMPi8:
-  case ARM::tCMPr:
   case ARM::tEOR:
-  case ARM::tLDRBi:
-  case ARM::tLDRBr:
-  case ARM::tLDRHi:
-  case ARM::tLDRHr:
-  case ARM::tLDRSB:
-  case ARM::tLDRSH:
-  case ARM::tLDRi:
-  case ARM::tLDRr:
-  case ARM::tLDRspi:
   case ARM::tLSLri:
   case ARM::tLSLrr:
   case ARM::tLSRri:
@@ -56,6 +46,24 @@ inline bool isV8EligibleForIT(InstrType *Instr, int BLXOperandIndex = 0) {
   case ARM::tROR:
   case ARM::tRSB:
   case ARM::tSBC:
+  case ARM::tSUBi3:
+  case ARM::tSUBi8:
+  case ARM::tSUBrr:
+    // Outside of an IT block, these set CPSR.
+    return IsCPSRDead(Instr);
+  case ARM::tADDrSPi:
+  case ARM::tCMNz:
+  case ARM::tCMPi8:
+  case ARM::tCMPr:
+  case ARM::tLDRBi:
+  case ARM::tLDRBr:
+  case ARM::tLDRHi:
+  case ARM::tLDRHr:
+  case ARM::tLDRSB:
+  case ARM::tLDRSH:
+  case ARM::tLDRi:
+  case ARM::tLDRr:
+  case ARM::tLDRspi:
   case ARM::tSTRBi:
   case ARM::tSTRBr:
   case ARM::tSTRHi:
@@ -63,21 +71,17 @@ inline bool isV8EligibleForIT(InstrType *Instr, int BLXOperandIndex = 0) {
   case ARM::tSTRi:
   case ARM::tSTRr:
   case ARM::tSTRspi:
-  case ARM::tSUBi3:
-  case ARM::tSUBi8:
-  case ARM::tSUBrr:
   case ARM::tTST:
     return true;
 // there are some "conditionally deprecated" opcodes
   case ARM::tADDspr:
+  case ARM::tBLXr:
     return Instr->getOperand(2).getReg() != ARM::PC;
   // ADD PC, SP and BLX PC were always unpredictable,
   // now on top of it they're deprecated
   case ARM::tADDrSP:
   case ARM::tBX:
     return Instr->getOperand(0).getReg() != ARM::PC;
-  case ARM::tBLXr:
-    return Instr->getOperand(BLXOperandIndex).getReg() != ARM::PC;
   case ARM::tADDhirr:
     return Instr->getOperand(0).getReg() != ARM::PC &&
            Instr->getOperand(2).getReg() != ARM::PC;
diff --git a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp
index d32bdbc58939..a67b3600c4fe 100644
--- a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.cpp
@@ -14,15 +14,18 @@
 #include "ARMFrameLowering.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMBaseRegisterInfo.h"
+#include "ARMConstantPoolValue.h"
 #include "ARMMachineFunctionInfo.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/CodeGen/MachineModuleInfo.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/RegisterScavenging.h"
 #include "llvm/IR/CallingConv.h"
 #include "llvm/IR/Function.h"
+#include "llvm/MC/MCContext.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Target/TargetOptions.h"
 
@@ -36,6 +39,10 @@ static MachineBasicBlock::iterator
 skipAlignedDPRCS2Spills(MachineBasicBlock::iterator MI,
                         unsigned NumAlignedDPRCS2Regs);
 
+ARMFrameLowering::ARMFrameLowering(const ARMSubtarget &sti)
+    : TargetFrameLowering(StackGrowsDown, sti.getStackAlignment(), 0, 4),
+      STI(sti) {}
+
 /// hasFP - Return true if the specified function should have a dedicated frame
 /// pointer register.  This is true if the function has variable sized allocas
 /// or if frame pointer elimination is disabled.
@@ -84,7 +91,7 @@ ARMFrameLowering::canSimplifyCallFramePseudos(const MachineFunction &MF) const {
 
 static bool isCSRestore(MachineInstr *MI,
                         const ARMBaseInstrInfo &TII,
-                        const uint16_t *CSRegs) {
+                        const MCPhysReg *CSRegs) {
   // Integer spill area is handled with "pop".
   if (isPopOpcode(MI->getOpcode())) {
     // The first two operands are predicates. The last two are
@@ -129,24 +136,47 @@ static void emitSPUpdate(bool isARM, MachineBasicBlock &MBB,
                        MIFlags, Pred, PredReg);
 }
 
+static int sizeOfSPAdjustment(const MachineInstr *MI) {
+  assert(MI->getOpcode() == ARM::VSTMDDB_UPD);
+  int count = 0;
+  // ARM and Thumb2 push/pop insts have explicit "sp, sp" operands (+
+  // pred) so the list starts at 4.
+  for (int i = MI->getNumOperands() - 1; i >= 4; --i)
+    count += 8;
+  return count;
+}
+
+static bool WindowsRequiresStackProbe(const MachineFunction &MF,
+                                      size_t StackSizeInBytes) {
+  const MachineFrameInfo *MFI = MF.getFrameInfo();
+  if (MFI->getStackProtectorIndex() > 0)
+    return StackSizeInBytes >= 4080;
+  return StackSizeInBytes >= 4096;
+}
+
 void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   MachineBasicBlock &MBB = MF.front();
   MachineBasicBlock::iterator MBBI = MBB.begin();
   MachineFrameInfo  *MFI = MF.getFrameInfo();
   ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
+  MachineModuleInfo &MMI = MF.getMMI();
+  MCContext &Context = MMI.getContext();
+  const TargetMachine &TM = MF.getTarget();
+  const MCRegisterInfo *MRI = Context.getRegisterInfo();
   const ARMBaseRegisterInfo *RegInfo =
-    static_cast<const ARMBaseRegisterInfo*>(MF.getTarget().getRegisterInfo());
+    static_cast<const ARMBaseRegisterInfo*>(TM.getRegisterInfo());
   const ARMBaseInstrInfo &TII =
-    *static_cast<const ARMBaseInstrInfo*>(MF.getTarget().getInstrInfo());
+    *static_cast<const ARMBaseInstrInfo*>(TM.getInstrInfo());
   assert(!AFI->isThumb1OnlyFunction() &&
          "This emitPrologue does not support Thumb1!");
   bool isARM = !AFI->isThumbFunction();
-  unsigned Align = MF.getTarget().getFrameLowering()->getStackAlignment();
+  unsigned Align = TM.getFrameLowering()->getStackAlignment();
   unsigned ArgRegsSaveSize = AFI->getArgRegsSaveSize(Align);
   unsigned NumBytes = MFI->getStackSize();
   const std::vector<CalleeSavedInfo> &CSI = MFI->getCalleeSavedInfo();
   DebugLoc dl = MBBI != MBB.end() ? MBBI->getDebugLoc() : DebugLoc();
   unsigned FramePtr = RegInfo->getFrameRegister(MF);
+  int CFAOffset = 0;
 
   // Determine the sizes of each callee-save spill areas and record which frame
   // belongs to which callee-save spill areas.
@@ -159,22 +189,46 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   if (MF.getFunction()->getCallingConv() == CallingConv::GHC)
     return;
 
-  // Allocate the vararg register save area. This is not counted in NumBytes.
-  if (ArgRegsSaveSize)
+  // Allocate the vararg register save area.
+  if (ArgRegsSaveSize) {
     emitSPUpdate(isARM, MBB, MBBI, dl, TII, -ArgRegsSaveSize,
                  MachineInstr::FrameSetup);
+    CFAOffset -= ArgRegsSaveSize;
+    unsigned CFIIndex = MMI.addFrameInst(
+        MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+    BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+        .addCFIIndex(CFIIndex);
+  }
 
-  if (!AFI->hasStackFrame()) {
-    if (NumBytes != 0)
-      emitSPUpdate(isARM, MBB, MBBI, dl, TII, -NumBytes,
+  if (!AFI->hasStackFrame() &&
+      (!STI.isTargetWindows() || !WindowsRequiresStackProbe(MF, NumBytes))) {
+    if (NumBytes - ArgRegsSaveSize != 0) {
+      emitSPUpdate(isARM, MBB, MBBI, dl, TII, -(NumBytes - ArgRegsSaveSize),
                    MachineInstr::FrameSetup);
+      CFAOffset -= NumBytes - ArgRegsSaveSize;
+      unsigned CFIIndex = MMI.addFrameInst(
+          MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
     return;
   }
 
+  // Determine spill area sizes.
   for (unsigned i = 0, e = CSI.size(); i != e; ++i) {
     unsigned Reg = CSI[i].getReg();
     int FI = CSI[i].getFrameIdx();
     switch (Reg) {
+    case ARM::R8:
+    case ARM::R9:
+    case ARM::R10:
+    case ARM::R11:
+    case ARM::R12:
+      if (STI.isTargetDarwin()) {
+        GPRCS2Size += 4;
+        break;
+      }
+      // fallthrough
     case ARM::R0:
     case ARM::R1:
     case ARM::R2:
@@ -188,18 +242,6 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
         FramePtrSpillFI = FI;
       GPRCS1Size += 4;
       break;
-    case ARM::R8:
-    case ARM::R9:
-    case ARM::R10:
-    case ARM::R11:
-    case ARM::R12:
-      if (Reg == FramePtr)
-        FramePtrSpillFI = FI;
-      if (STI.isTargetIOS())
-        GPRCS2Size += 4;
-      else
-        GPRCS1Size += 4;
-      break;
     default:
       // This is a DPR. Exclude the aligned DPRCS2 spills.
       if (Reg == ARM::D8)
@@ -210,18 +252,21 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   }
 
   // Move past area 1.
-  MachineBasicBlock::iterator LastPush = MBB.end(), FramePtrPush;
+  MachineBasicBlock::iterator LastPush = MBB.end(), GPRCS1Push, GPRCS2Push,
+      DPRCSPush;
   if (GPRCS1Size > 0)
-    FramePtrPush = LastPush = MBBI++;
+    GPRCS1Push = LastPush = MBBI++;
 
   // Determine starting offsets of spill areas.
   bool HasFP = hasFP(MF);
-  unsigned DPRCSOffset  = NumBytes - (GPRCS1Size + GPRCS2Size + DPRCSSize);
+  unsigned DPRCSOffset  = NumBytes - (ArgRegsSaveSize + GPRCS1Size
+                                      + GPRCS2Size + DPRCSSize);
   unsigned GPRCS2Offset = DPRCSOffset + DPRCSSize;
   unsigned GPRCS1Offset = GPRCS2Offset + GPRCS2Size;
   int FramePtrOffsetInPush = 0;
   if (HasFP) {
-    FramePtrOffsetInPush = MFI->getObjectOffset(FramePtrSpillFI) + GPRCS1Size;
+    FramePtrOffsetInPush = MFI->getObjectOffset(FramePtrSpillFI)
+                           + GPRCS1Size + ArgRegsSaveSize;
     AFI->setFramePtrSpillOffset(MFI->getObjectOffset(FramePtrSpillFI) +
                                 NumBytes);
   }
@@ -230,13 +275,12 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   AFI->setDPRCalleeSavedAreaOffset(DPRCSOffset);
 
   // Move past area 2.
-  if (GPRCS2Size > 0) {
-    LastPush = MBBI++;
-  }
+  if (GPRCS2Size > 0)
+    GPRCS2Push = LastPush = MBBI++;
 
   // Move past area 3.
   if (DPRCSSize > 0) {
-    LastPush = MBBI++;
+    DPRCSPush = MBBI;
     // Since vpush register list cannot have gaps, there may be multiple vpush
     // instructions in the prologue.
     while (MBBI->getOpcode() == ARM::VSTMDDB_UPD)
@@ -254,11 +298,60 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
   } else
     NumBytes = DPRCSOffset;
 
+  if (STI.isTargetWindows() && WindowsRequiresStackProbe(MF, NumBytes)) {
+    uint32_t NumWords = NumBytes >> 2;
+
+    if (NumWords < 65536)
+      AddDefaultPred(BuildMI(MBB, MBBI, dl, TII.get(ARM::t2MOVi16), ARM::R4)
+                     .addImm(NumWords)
+                     .setMIFlags(MachineInstr::FrameSetup));
+    else
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::t2MOVi32imm), ARM::R4)
+        .addImm(NumWords)
+        .setMIFlags(MachineInstr::FrameSetup);
+
+    switch (TM.getCodeModel()) {
+    case CodeModel::Small:
+    case CodeModel::Medium:
+    case CodeModel::Default:
+    case CodeModel::Kernel:
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::tBL))
+        .addImm((unsigned)ARMCC::AL).addReg(0)
+        .addExternalSymbol("__chkstk")
+        .addReg(ARM::R4, RegState::Implicit)
+        .setMIFlags(MachineInstr::FrameSetup);
+      break;
+    case CodeModel::Large:
+    case CodeModel::JITDefault:
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::t2MOVi32imm), ARM::R12)
+        .addExternalSymbol("__chkstk")
+        .setMIFlags(MachineInstr::FrameSetup);
+
+      BuildMI(MBB, MBBI, dl, TII.get(ARM::tBLXr))
+        .addImm((unsigned)ARMCC::AL).addReg(0)
+        .addReg(ARM::R12, RegState::Kill)
+        .addReg(ARM::R4, RegState::Implicit)
+        .setMIFlags(MachineInstr::FrameSetup);
+      break;
+    }
+
+    AddDefaultCC(AddDefaultPred(BuildMI(MBB, MBBI, dl, TII.get(ARM::t2SUBrr),
+                                        ARM::SP)
+                                .addReg(ARM::SP, RegState::Define)
+                                .addReg(ARM::R4, RegState::Kill)
+                                .setMIFlags(MachineInstr::FrameSetup)));
+    NumBytes = 0;
+  }
+
+  unsigned adjustedGPRCS1Size = GPRCS1Size;
   if (NumBytes) {
     // Adjust SP after all the callee-save spills.
-    if (tryFoldSPUpdateIntoPushPop(MF, LastPush, NumBytes)) {
-      if (LastPush == FramePtrPush)
+    if (tryFoldSPUpdateIntoPushPop(STI, MF, LastPush, NumBytes)) {
+      if (LastPush == GPRCS1Push) {
         FramePtrOffsetInPush += NumBytes;
+        adjustedGPRCS1Size += NumBytes;
+        NumBytes = 0;
+      }
     } else
       emitSPUpdate(isARM, MBB, MBBI, dl, TII, -NumBytes,
                    MachineInstr::FrameSetup);
@@ -275,17 +368,138 @@ void ARMFrameLowering::emitPrologue(MachineFunction &MF) const {
       AFI->setShouldRestoreSPFromFP(true);
   }
 
+  if (adjustedGPRCS1Size > 0) {
+    CFAOffset -= adjustedGPRCS1Size;
+    unsigned CFIIndex = MMI.addFrameInst(
+        MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+    MachineBasicBlock::iterator Pos = ++GPRCS1Push;
+    BuildMI(MBB, Pos, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+        .addCFIIndex(CFIIndex);
+    for (const auto &Entry : CSI) {
+      unsigned Reg = Entry.getReg();
+      int FI = Entry.getFrameIdx();
+      switch (Reg) {
+      case ARM::R8:
+      case ARM::R9:
+      case ARM::R10:
+      case ARM::R11:
+      case ARM::R12:
+        if (STI.isTargetDarwin())
+          break;
+        // fallthrough
+      case ARM::R0:
+      case ARM::R1:
+      case ARM::R2:
+      case ARM::R3:
+      case ARM::R4:
+      case ARM::R5:
+      case ARM::R6:
+      case ARM::R7:
+      case ARM::LR:
+        CFIIndex = MMI.addFrameInst(MCCFIInstruction::createOffset(
+            nullptr, MRI->getDwarfRegNum(Reg, true), MFI->getObjectOffset(FI)));
+        BuildMI(MBB, Pos, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+            .addCFIIndex(CFIIndex);
+        break;
+      }
+    }
+  }
+
   // Set FP to point to the stack slot that contains the previous FP.
   // For iOS, FP is R7, which has now been stored in spill area 1.
   // Otherwise, if this is not iOS, all the callee-saved registers go
   // into spill area 1, including the FP in R11.  In either case, it
   // is in area one and the adjustment needs to take place just after
   // that push.
-  if (HasFP)
-    emitRegPlusImmediate(!AFI->isThumbFunction(), MBB, ++FramePtrPush, dl, TII,
+  if (HasFP) {
+    emitRegPlusImmediate(!AFI->isThumbFunction(), MBB, GPRCS1Push, dl, TII,
                          FramePtr, ARM::SP, FramePtrOffsetInPush,
                          MachineInstr::FrameSetup);
+    if (FramePtrOffsetInPush) {
+      CFAOffset += FramePtrOffsetInPush;
+      unsigned CFIIndex = MMI.addFrameInst(MCCFIInstruction::createDefCfa(
+          nullptr, MRI->getDwarfRegNum(FramePtr, true), CFAOffset));
+      BuildMI(MBB, GPRCS1Push, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
 
+    } else {
+      unsigned CFIIndex =
+          MMI.addFrameInst(MCCFIInstruction::createDefCfaRegister(
+              nullptr, MRI->getDwarfRegNum(FramePtr, true)));
+      BuildMI(MBB, GPRCS1Push, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
+  }
+
+  if (GPRCS2Size > 0) {
+    MachineBasicBlock::iterator Pos = ++GPRCS2Push;
+    if (!HasFP) {
+      CFAOffset -= GPRCS2Size;
+      unsigned CFIIndex = MMI.addFrameInst(
+          MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+      BuildMI(MBB, Pos, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
+    for (const auto &Entry : CSI) {
+      unsigned Reg = Entry.getReg();
+      int FI = Entry.getFrameIdx();
+      switch (Reg) {
+      case ARM::R8:
+      case ARM::R9:
+      case ARM::R10:
+      case ARM::R11:
+      case ARM::R12:
+        if (STI.isTargetDarwin()) {
+          unsigned DwarfReg =  MRI->getDwarfRegNum(Reg, true);
+          unsigned Offset = MFI->getObjectOffset(FI);
+          unsigned CFIIndex = MMI.addFrameInst(
+              MCCFIInstruction::createOffset(nullptr, DwarfReg, Offset));
+          BuildMI(MBB, Pos, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+              .addCFIIndex(CFIIndex);
+        }
+        break;
+      }
+    }
+  }
+
+  if (DPRCSSize > 0) {
+    // Since vpush register list cannot have gaps, there may be multiple vpush
+    // instructions in the prologue.
+    do {
+      MachineBasicBlock::iterator Push = DPRCSPush++;
+      if (!HasFP) {
+        CFAOffset -= sizeOfSPAdjustment(Push);
+        unsigned CFIIndex = MMI.addFrameInst(
+            MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+        BuildMI(MBB, DPRCSPush, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+            .addCFIIndex(CFIIndex);
+      }
+    } while (DPRCSPush->getOpcode() == ARM::VSTMDDB_UPD);
+
+    for (const auto &Entry : CSI) {
+      unsigned Reg = Entry.getReg();
+      int FI = Entry.getFrameIdx();
+      if ((Reg >= ARM::D0 && Reg <= ARM::D31) &&
+          (Reg < ARM::D8 || Reg >= ARM::D8 + AFI->getNumAlignedDPRCS2Regs())) {
+        unsigned DwarfReg = MRI->getDwarfRegNum(Reg, true);
+        unsigned Offset = MFI->getObjectOffset(FI);
+        unsigned CFIIndex = MMI.addFrameInst(
+            MCCFIInstruction::createOffset(nullptr, DwarfReg, Offset));
+        BuildMI(MBB, DPRCSPush, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+            .addCFIIndex(CFIIndex);
+      }
+    }
+  }
+
+  if (NumBytes) {
+    if (!HasFP) {
+      CFAOffset -= NumBytes;
+      unsigned CFIIndex = MMI.addFrameInst(
+          MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
+  }
 
   if (STI.isTargetELF() && hasFP(MF))
     MFI->setOffsetAdjustment(MFI->getOffsetAdjustment() -
@@ -378,11 +592,11 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF,
     return;
 
   if (!AFI->hasStackFrame()) {
-    if (NumBytes != 0)
-      emitSPUpdate(isARM, MBB, MBBI, dl, TII, NumBytes);
+    if (NumBytes - ArgRegsSaveSize != 0)
+      emitSPUpdate(isARM, MBB, MBBI, dl, TII, NumBytes - ArgRegsSaveSize);
   } else {
     // Unwind MBBI to point to first LDR / VLDRD.
-    const uint16_t *CSRegs = RegInfo->getCalleeSavedRegs(&MF);
+    const MCPhysReg *CSRegs = RegInfo->getCalleeSavedRegs(&MF);
     if (MBBI != MBB.begin()) {
       do {
         --MBBI;
@@ -392,7 +606,8 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF,
     }
 
     // Move SP to start of FP callee save spill area.
-    NumBytes -= (AFI->getGPRCalleeSavedArea1Size() +
+    NumBytes -= (ArgRegsSaveSize +
+                 AFI->getGPRCalleeSavedArea1Size() +
                  AFI->getGPRCalleeSavedArea2Size() +
                  AFI->getDPRCalleeSavedAreaSize());
 
@@ -430,7 +645,8 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF,
                                  ARM::SP)
             .addReg(FramePtr));
       }
-    } else if (NumBytes && !tryFoldSPUpdateIntoPushPop(MF, MBBI, NumBytes))
+    } else if (NumBytes &&
+               !tryFoldSPUpdateIntoPushPop(STI, MF, MBBI, NumBytes))
         emitSPUpdate(isARM, MBB, MBBI, dl, TII, NumBytes);
 
     // Increment past our save areas.
@@ -453,7 +669,7 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF,
     // Jump to label or value in register.
     if (RetOpcode == ARM::TCRETURNdi) {
       unsigned TCOpcode = STI.isThumb() ?
-               (STI.isTargetIOS() ? ARM::tTAILJMPd : ARM::tTAILJMPdND) :
+               (STI.isTargetMachO() ? ARM::tTAILJMPd : ARM::tTAILJMPdND) :
                ARM::TAILJMPd;
       MachineInstrBuilder MIB = BuildMI(MBB, MBBI, dl, TII.get(TCOpcode));
       if (JumpTarget.isGlobal())
@@ -473,7 +689,7 @@ void ARMFrameLowering::emitEpilogue(MachineFunction &MF,
         addReg(JumpTarget.getReg(), RegState::Kill);
     }
 
-    MachineInstr *NewMI = prior(MBBI);
+    MachineInstr *NewMI = std::prev(MBBI);
     for (unsigned i = 1, e = MBBI->getNumOperands(); i != e; ++i)
       NewMI->addOperand(MBBI->getOperand(i));
 
@@ -598,7 +814,7 @@ void ARMFrameLowering::emitPushInst(MachineBasicBlock &MBB,
     unsigned LastReg = 0;
     for (; i != 0; --i) {
       unsigned Reg = CSI[i-1].getReg();
-      if (!(Func)(Reg, STI.isTargetIOS())) continue;
+      if (!(Func)(Reg, STI.isTargetDarwin())) continue;
 
       // D-registers in the aligned area DPRCS2 are NOT spilled here.
       if (Reg >= ARM::D8 && Reg < ARM::D8 + NumAlignedDPRCS2Regs)
@@ -644,6 +860,11 @@ void ARMFrameLowering::emitPushInst(MachineBasicBlock &MBB,
       AddDefaultPred(MIB);
     }
     Regs.clear();
+
+    // Put any subsequent vpush instructions before this one: they will refer to
+    // higher register numbers so need to be pushed first in order to preserve
+    // monotonicity.
+    --MI;
   }
 }
 
@@ -671,7 +892,7 @@ void ARMFrameLowering::emitPopInst(MachineBasicBlock &MBB,
     bool DeleteRet = false;
     for (; i != 0; --i) {
       unsigned Reg = CSI[i-1].getReg();
-      if (!(Func)(Reg, STI.isTargetIOS())) continue;
+      if (!(Func)(Reg, STI.isTargetDarwin())) continue;
 
       // The aligned reloads from area DPRCS2 are not inserted here.
       if (Reg >= ARM::D8 && Reg < ARM::D8 + NumAlignedDPRCS2Regs)
@@ -727,6 +948,10 @@ void ARMFrameLowering::emitPopInst(MachineBasicBlock &MBB,
       AddDefaultPred(MIB);
     }
     Regs.clear();
+
+    // Put any subsequent vpop instructions after this one: they will refer to
+    // higher register numbers so need to be popped afterwards.
+    ++MI;
   }
 }
 
@@ -858,7 +1083,7 @@ static void emitAlignedDPRCS2Spills(MachineBasicBlock &MBB,
   }
 
   // The last spill instruction inserted should kill the scratch register r4.
-  llvm::prior(MI)->addRegisterKilled(ARM::R4, TRI);
+  std::prev(MI)->addRegisterKilled(ARM::R4, TRI);
 }
 
 /// Skip past the code inserted by emitAlignedDPRCS2Spills, and return an
@@ -968,7 +1193,7 @@ static void emitAlignedDPRCS2Restores(MachineBasicBlock &MBB,
                    .addReg(ARM::R4).addImm(2*(NextReg-R4BaseReg)));
 
   // Last store kills r4.
-  llvm::prior(MI)->addRegisterKilled(ARM::R4, TRI);
+  std::prev(MI)->addRegisterKilled(ARM::R4, TRI);
 }
 
 bool ARMFrameLowering::spillCalleeSavedRegisters(MachineBasicBlock &MBB,
@@ -1036,12 +1261,9 @@ bool ARMFrameLowering::restoreCalleeSavedRegisters(MachineBasicBlock &MBB,
 static unsigned GetFunctionSizeInBytes(const MachineFunction &MF,
                                        const ARMBaseInstrInfo &TII) {
   unsigned FnSize = 0;
-  for (MachineFunction::const_iterator MBBI = MF.begin(), E = MF.end();
-       MBBI != E; ++MBBI) {
-    const MachineBasicBlock &MBB = *MBBI;
-    for (MachineBasicBlock::const_iterator I = MBB.begin(),E = MBB.end();
-         I != E; ++I)
-      FnSize += TII.GetInstSizeInBytes(I);
+  for (auto &MBB : MF) {
+    for (auto &MI : MBB)
+      FnSize += TII.GetInstSizeInBytes(&MI);
   }
   return FnSize;
 }
@@ -1054,21 +1276,21 @@ static unsigned estimateRSStackSizeLimit(MachineFunction &MF,
                                          const TargetFrameLowering *TFI) {
   const ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
   unsigned Limit = (1 << 12) - 1;
-  for (MachineFunction::iterator BB = MF.begin(),E = MF.end(); BB != E; ++BB) {
-    for (MachineBasicBlock::iterator I = BB->begin(), E = BB->end();
-         I != E; ++I) {
-      for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
-        if (!I->getOperand(i).isFI()) continue;
+  for (auto &MBB : MF) {
+    for (auto &MI : MBB) {
+      for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+        if (!MI.getOperand(i).isFI())
+          continue;
 
         // When using ADDri to get the address of a stack object, 255 is the
         // largest offset guaranteed to fit in the immediate offset.
-        if (I->getOpcode() == ARM::ADDri) {
+        if (MI.getOpcode() == ARM::ADDri) {
           Limit = std::min(Limit, (1U << 8) - 1);
           break;
         }
 
         // Otherwise check the addressing mode.
-        switch (I->getDesc().TSFlags & ARMII::AddrModeMask) {
+        switch (MI.getDesc().TSFlags & ARMII::AddrModeMask) {
         case ARMII::AddrMode3:
         case ARMII::AddrModeT2_i8:
           Limit = std::min(Limit, (1U << 8) - 1);
@@ -1205,7 +1427,7 @@ ARMFrameLowering::processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
 
   // Don't spill FP if the frame can be eliminated. This is determined
   // by scanning the callee-save registers to see if any is used.
-  const uint16_t *CSRegs = RegInfo->getCalleeSavedRegs(&MF);
+  const MCPhysReg *CSRegs = RegInfo->getCalleeSavedRegs(&MF);
   for (unsigned i = 0; CSRegs[i]; ++i) {
     unsigned Reg = CSRegs[i];
     bool Spilled = false;
@@ -1220,7 +1442,7 @@ ARMFrameLowering::processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
     if (Spilled) {
       NumGPRSpills++;
 
-      if (!STI.isTargetIOS()) {
+      if (!STI.isTargetDarwin()) {
         if (Reg == ARM::LR)
           LRSpilled = true;
         CS1Spilled = true;
@@ -1242,7 +1464,7 @@ ARMFrameLowering::processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
         break;
       }
     } else {
-      if (!STI.isTargetIOS()) {
+      if (!STI.isTargetDarwin()) {
         UnspilledCS1GPRs.push_back(Reg);
         continue;
       }
@@ -1317,6 +1539,10 @@ ARMFrameLowering::processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
 
     if (hasFP(MF)) {
       MRI.setPhysRegUsed(FramePtr);
+      auto FPPos = std::find(UnspilledCS1GPRs.begin(), UnspilledCS1GPRs.end(),
+                             FramePtr);
+      if (FPPos != UnspilledCS1GPRs.end())
+        UnspilledCS1GPRs.erase(FPPos);
       NumGPRSpills++;
     }
 
@@ -1444,3 +1670,370 @@ eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
   MBB.erase(I);
 }
 
+/// Get the minimum constant for ARM that is greater than or equal to the
+/// argument. In ARM, constants can have any value that can be produced by
+/// rotating an 8-bit value to the right by an even number of bits within a
+/// 32-bit word.
+static uint32_t alignToARMConstant(uint32_t Value) {
+  unsigned Shifted = 0;
+
+  if (Value == 0)
+      return 0;
+
+  while (!(Value & 0xC0000000)) {
+      Value = Value << 2;
+      Shifted += 2;
+  }
+
+  bool Carry = (Value & 0x00FFFFFF);
+  Value = ((Value & 0xFF000000) >> 24) + Carry;
+
+  if (Value & 0x0000100)
+      Value = Value & 0x000001FC;
+
+  if (Shifted > 24)
+      Value = Value >> (Shifted - 24);
+  else
+      Value = Value << (24 - Shifted);
+
+  return Value;
+}
+
+// The stack limit in the TCB is set to this many bytes above the actual
+// stack limit.
+static const uint64_t kSplitStackAvailable = 256;
+
+// Adjust the function prologue to enable split stacks. This currently only
+// supports android and linux.
+//
+// The ABI of the segmented stack prologue is a little arbitrarily chosen, but
+// must be well defined in order to allow for consistent implementations of the
+// __morestack helper function. The ABI is also not a normal ABI in that it
+// doesn't follow the normal calling conventions because this allows the
+// prologue of each function to be optimized further.
+//
+// Currently, the ABI looks like (when calling __morestack)
+//
+//  * r4 holds the minimum stack size requested for this function call
+//  * r5 holds the stack size of the arguments to the function
+//  * the beginning of the function is 3 instructions after the call to
+//    __morestack
+//
+// Implementations of __morestack should use r4 to allocate a new stack, r5 to
+// place the arguments on to the new stack, and the 3-instruction knowledge to
+// jump directly to the body of the function when working on the new stack.
+//
+// An old (and possibly no longer compatible) implementation of __morestack for
+// ARM can be found at [1].
+//
+// [1] - https://github.com/mozilla/rust/blob/86efd9/src/rt/arch/arm/morestack.S
+void ARMFrameLowering::adjustForSegmentedStacks(MachineFunction &MF) const {
+  unsigned Opcode;
+  unsigned CFIIndex;
+  const ARMSubtarget *ST = &MF.getTarget().getSubtarget<ARMSubtarget>();
+  bool Thumb = ST->isThumb();
+
+  // Sadly, this currently doesn't support varargs, platforms other than
+  // android/linux. Note that thumb1/thumb2 are support for android/linux.
+  if (MF.getFunction()->isVarArg())
+    report_fatal_error("Segmented stacks do not support vararg functions.");
+  if (!ST->isTargetAndroid() && !ST->isTargetLinux())
+    report_fatal_error("Segmented stacks not supported on this platform.");
+
+  MachineBasicBlock &prologueMBB = MF.front();
+  MachineFrameInfo *MFI = MF.getFrameInfo();
+  MachineModuleInfo &MMI = MF.getMMI();
+  MCContext &Context = MMI.getContext();
+  const MCRegisterInfo *MRI = Context.getRegisterInfo();
+  const ARMBaseInstrInfo &TII =
+      *static_cast<const ARMBaseInstrInfo*>(MF.getTarget().getInstrInfo());
+  ARMFunctionInfo *ARMFI = MF.getInfo<ARMFunctionInfo>();
+  DebugLoc DL;
+
+  uint64_t StackSize = MFI->getStackSize();
+
+  // Do not generate a prologue for functions with a stack of size zero
+  if (StackSize == 0)
+    return;
+
+  // Use R4 and R5 as scratch registers.
+  // We save R4 and R5 before use and restore them before leaving the function.
+  unsigned ScratchReg0 = ARM::R4;
+  unsigned ScratchReg1 = ARM::R5;
+  uint64_t AlignedStackSize;
+
+  MachineBasicBlock *PrevStackMBB = MF.CreateMachineBasicBlock();
+  MachineBasicBlock *PostStackMBB = MF.CreateMachineBasicBlock();
+  MachineBasicBlock *AllocMBB = MF.CreateMachineBasicBlock();
+  MachineBasicBlock *GetMBB = MF.CreateMachineBasicBlock();
+  MachineBasicBlock *McrMBB = MF.CreateMachineBasicBlock();
+
+  for (MachineBasicBlock::livein_iterator i = prologueMBB.livein_begin(),
+                                          e = prologueMBB.livein_end();
+       i != e; ++i) {
+    AllocMBB->addLiveIn(*i);
+    GetMBB->addLiveIn(*i);
+    McrMBB->addLiveIn(*i);
+    PrevStackMBB->addLiveIn(*i);
+    PostStackMBB->addLiveIn(*i);
+  }
+
+  MF.push_front(PostStackMBB);
+  MF.push_front(AllocMBB);
+  MF.push_front(GetMBB);
+  MF.push_front(McrMBB);
+  MF.push_front(PrevStackMBB);
+
+  // The required stack size that is aligned to ARM constant criterion.
+  AlignedStackSize = alignToARMConstant(StackSize);
+
+  // When the frame size is less than 256 we just compare the stack
+  // boundary directly to the value of the stack pointer, per gcc.
+  bool CompareStackPointer = AlignedStackSize < kSplitStackAvailable;
+
+  // We will use two of the callee save registers as scratch registers so we
+  // need to save those registers onto the stack.
+  // We will use SR0 to hold stack limit and SR1 to hold the stack size
+  // requested and arguments for __morestack().
+  // SR0: Scratch Register #0
+  // SR1: Scratch Register #1
+  // push {SR0, SR1}
+  if (Thumb) {
+    AddDefaultPred(BuildMI(PrevStackMBB, DL, TII.get(ARM::tPUSH)))
+        .addReg(ScratchReg0).addReg(ScratchReg1);
+  } else {
+    AddDefaultPred(BuildMI(PrevStackMBB, DL, TII.get(ARM::STMDB_UPD))
+                   .addReg(ARM::SP, RegState::Define).addReg(ARM::SP))
+        .addReg(ScratchReg0).addReg(ScratchReg1);
+  }
+
+  // Emit the relevant DWARF information about the change in stack pointer as
+  // well as where to find both r4 and r5 (the callee-save registers)
+  CFIIndex =
+      MMI.addFrameInst(MCCFIInstruction::createDefCfaOffset(nullptr, -8));
+  BuildMI(PrevStackMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createOffset(
+      nullptr, MRI->getDwarfRegNum(ScratchReg1, true), -4));
+  BuildMI(PrevStackMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createOffset(
+      nullptr, MRI->getDwarfRegNum(ScratchReg0, true), -8));
+  BuildMI(PrevStackMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+
+  // mov SR1, sp
+  if (Thumb) {
+    AddDefaultPred(BuildMI(McrMBB, DL, TII.get(ARM::tMOVr), ScratchReg1)
+                      .addReg(ARM::SP));
+  } else if (CompareStackPointer) {
+    AddDefaultPred(BuildMI(McrMBB, DL, TII.get(ARM::MOVr), ScratchReg1)
+                      .addReg(ARM::SP)).addReg(0);
+  }
+
+  // sub SR1, sp, #StackSize
+  if (!CompareStackPointer && Thumb) {
+    AddDefaultPred(
+        AddDefaultCC(BuildMI(McrMBB, DL, TII.get(ARM::tSUBi8), ScratchReg1))
+            .addReg(ScratchReg1).addImm(AlignedStackSize));
+  } else if (!CompareStackPointer) {
+    AddDefaultPred(BuildMI(McrMBB, DL, TII.get(ARM::SUBri), ScratchReg1)
+                      .addReg(ARM::SP).addImm(AlignedStackSize)).addReg(0);
+  }
+
+  if (Thumb && ST->isThumb1Only()) {
+    unsigned PCLabelId = ARMFI->createPICLabelUId();
+    ARMConstantPoolValue *NewCPV = ARMConstantPoolSymbol::Create(
+        MF.getFunction()->getContext(), "__STACK_LIMIT", PCLabelId, 0);
+    MachineConstantPool *MCP = MF.getConstantPool();
+    unsigned CPI = MCP->getConstantPoolIndex(NewCPV, MF.getAlignment());
+
+    // ldr SR0, [pc, offset(STACK_LIMIT)]
+    AddDefaultPred(BuildMI(GetMBB, DL, TII.get(ARM::tLDRpci), ScratchReg0)
+                      .addConstantPoolIndex(CPI));
+
+    // ldr SR0, [SR0]
+    AddDefaultPred(BuildMI(GetMBB, DL, TII.get(ARM::tLDRi), ScratchReg0)
+                      .addReg(ScratchReg0).addImm(0));
+  } else {
+    // Get TLS base address from the coprocessor
+    // mrc p15, #0, SR0, c13, c0, #3
+    AddDefaultPred(BuildMI(McrMBB, DL, TII.get(ARM::MRC), ScratchReg0)
+                     .addImm(15)
+                     .addImm(0)
+                     .addImm(13)
+                     .addImm(0)
+                     .addImm(3));
+
+    // Use the last tls slot on android and a private field of the TCP on linux.
+    assert(ST->isTargetAndroid() || ST->isTargetLinux());
+    unsigned TlsOffset = ST->isTargetAndroid() ? 63 : 1;
+
+    // Get the stack limit from the right offset
+    // ldr SR0, [sr0, #4 * TlsOffset]
+    AddDefaultPred(BuildMI(GetMBB, DL, TII.get(ARM::LDRi12), ScratchReg0)
+                      .addReg(ScratchReg0).addImm(4 * TlsOffset));
+  }
+
+  // Compare stack limit with stack size requested.
+  // cmp SR0, SR1
+  Opcode = Thumb ? ARM::tCMPr : ARM::CMPrr;
+  AddDefaultPred(BuildMI(GetMBB, DL, TII.get(Opcode))
+                    .addReg(ScratchReg0)
+                    .addReg(ScratchReg1));
+
+  // This jump is taken if StackLimit < SP - stack required.
+  Opcode = Thumb ? ARM::tBcc : ARM::Bcc;
+  BuildMI(GetMBB, DL, TII.get(Opcode)).addMBB(PostStackMBB)
+       .addImm(ARMCC::LO)
+       .addReg(ARM::CPSR);
+
+
+  // Calling __morestack(StackSize, Size of stack arguments).
+  // __morestack knows that the stack size requested is in SR0(r4)
+  // and amount size of stack arguments is in SR1(r5).
+
+  // Pass first argument for the __morestack by Scratch Register #0.
+  //   The amount size of stack required
+  if (Thumb) {
+    AddDefaultPred(AddDefaultCC(BuildMI(AllocMBB, DL, TII.get(ARM::tMOVi8),
+                                        ScratchReg0)).addImm(AlignedStackSize));
+  } else {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::MOVi), ScratchReg0)
+                      .addImm(AlignedStackSize)).addReg(0);
+  }
+  // Pass second argument for the __morestack by Scratch Register #1.
+  //   The amount size of stack consumed to save function arguments.
+  if (Thumb) {
+    AddDefaultPred(
+        AddDefaultCC(BuildMI(AllocMBB, DL, TII.get(ARM::tMOVi8), ScratchReg1))
+            .addImm(alignToARMConstant(ARMFI->getArgumentStackSize())));
+  } else {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::MOVi), ScratchReg1)
+                   .addImm(alignToARMConstant(ARMFI->getArgumentStackSize())))
+                   .addReg(0);
+  }
+
+  // push {lr} - Save return address of this function.
+  if (Thumb) {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::tPUSH)))
+        .addReg(ARM::LR);
+  } else {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::STMDB_UPD))
+                   .addReg(ARM::SP, RegState::Define)
+                   .addReg(ARM::SP))
+        .addReg(ARM::LR);
+  }
+
+  // Emit the DWARF info about the change in stack as well as where to find the
+  // previous link register
+  CFIIndex =
+      MMI.addFrameInst(MCCFIInstruction::createDefCfaOffset(nullptr, -12));
+  BuildMI(AllocMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createOffset(
+        nullptr, MRI->getDwarfRegNum(ARM::LR, true), -12));
+  BuildMI(AllocMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+
+  // Call __morestack().
+  if (Thumb) {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::tBL)))
+        .addExternalSymbol("__morestack");
+  } else {
+    BuildMI(AllocMBB, DL, TII.get(ARM::BL))
+        .addExternalSymbol("__morestack");
+  }
+
+  // pop {lr} - Restore return address of this original function.
+  if (Thumb) {
+    if (ST->isThumb1Only()) {
+      AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::tPOP)))
+                     .addReg(ScratchReg0);
+      AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::tMOVr), ARM::LR)
+                     .addReg(ScratchReg0));
+    } else {
+      AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::t2LDR_POST))
+                     .addReg(ARM::LR, RegState::Define)
+                     .addReg(ARM::SP, RegState::Define)
+                     .addReg(ARM::SP)
+                     .addImm(4));
+    }
+  } else {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::LDMIA_UPD))
+                   .addReg(ARM::SP, RegState::Define)
+                   .addReg(ARM::SP))
+      .addReg(ARM::LR);
+  }
+
+  // Restore SR0 and SR1 in case of __morestack() was called.
+  // __morestack() will skip PostStackMBB block so we need to restore
+  // scratch registers from here.
+  // pop {SR0, SR1}
+  if (Thumb) {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::tPOP)))
+      .addReg(ScratchReg0)
+      .addReg(ScratchReg1);
+  } else {
+    AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(ARM::LDMIA_UPD))
+                   .addReg(ARM::SP, RegState::Define)
+                   .addReg(ARM::SP))
+      .addReg(ScratchReg0)
+      .addReg(ScratchReg1);
+  }
+
+  // Update the CFA offset now that we've popped
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createDefCfaOffset(nullptr, 0));
+  BuildMI(AllocMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+
+  // bx lr - Return from this function.
+  Opcode = Thumb ? ARM::tBX_RET : ARM::BX_RET;
+  AddDefaultPred(BuildMI(AllocMBB, DL, TII.get(Opcode)));
+
+  // Restore SR0 and SR1 in case of __morestack() was not called.
+  // pop {SR0, SR1}
+  if (Thumb) {
+    AddDefaultPred(BuildMI(PostStackMBB, DL, TII.get(ARM::tPOP)))
+      .addReg(ScratchReg0)
+      .addReg(ScratchReg1);
+  } else {
+    AddDefaultPred(BuildMI(PostStackMBB, DL, TII.get(ARM::LDMIA_UPD))
+                   .addReg(ARM::SP, RegState::Define)
+                   .addReg(ARM::SP))
+      .addReg(ScratchReg0)
+      .addReg(ScratchReg1);
+  }
+
+  // Update the CFA offset now that we've popped
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createDefCfaOffset(nullptr, 0));
+  BuildMI(PostStackMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+
+  // Tell debuggers that r4 and r5 are now the same as they were in the
+  // previous function, that they're the "Same Value".
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createSameValue(
+      nullptr, MRI->getDwarfRegNum(ScratchReg0, true)));
+  BuildMI(PostStackMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+  CFIIndex = MMI.addFrameInst(MCCFIInstruction::createSameValue(
+      nullptr, MRI->getDwarfRegNum(ScratchReg1, true)));
+  BuildMI(PostStackMBB, DL, TII.get(TargetOpcode::CFI_INSTRUCTION))
+      .addCFIIndex(CFIIndex);
+
+  // Organizing MBB lists
+  PostStackMBB->addSuccessor(&prologueMBB);
+
+  AllocMBB->addSuccessor(PostStackMBB);
+
+  GetMBB->addSuccessor(PostStackMBB);
+  GetMBB->addSuccessor(AllocMBB);
+
+  McrMBB->addSuccessor(GetMBB);
+
+  PrevStackMBB->addSuccessor(McrMBB);
+
+#ifdef XDEBUG
+  MF.verify();
+#endif
+}
diff --git a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h
index d95a2cb9fd42..709afbcdc681 100644
--- a/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h
+++ b/contrib/llvm/lib/Target/ARM/ARMFrameLowering.h
@@ -14,8 +14,6 @@
 #ifndef ARM_FRAMEINFO_H
 #define ARM_FRAMEINFO_H
 
-#include "ARM.h"
-#include "ARMSubtarget.h"
 #include "llvm/Target/TargetFrameLowering.h"
 
 namespace llvm {
@@ -26,38 +24,36 @@ protected:
   const ARMSubtarget &STI;
 
 public:
-  explicit ARMFrameLowering(const ARMSubtarget &sti)
-    : TargetFrameLowering(StackGrowsDown, sti.getStackAlignment(), 0, 8),
-      STI(sti) {
-  }
+  explicit ARMFrameLowering(const ARMSubtarget &sti);
 
   /// emitProlog/emitEpilog - These methods insert prolog and epilog code into
   /// the function.
-  void emitPrologue(MachineFunction &MF) const;
-  void emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const;
+  void emitPrologue(MachineFunction &MF) const override;
+  void emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const override;
 
   bool spillCalleeSavedRegisters(MachineBasicBlock &MBB,
                                  MachineBasicBlock::iterator MI,
                                  const std::vector<CalleeSavedInfo> &CSI,
-                                 const TargetRegisterInfo *TRI) const;
+                                 const TargetRegisterInfo *TRI) const override;
 
   bool restoreCalleeSavedRegisters(MachineBasicBlock &MBB,
-                                   MachineBasicBlock::iterator MI,
-                                   const std::vector<CalleeSavedInfo> &CSI,
-                                   const TargetRegisterInfo *TRI) const;
+                                  MachineBasicBlock::iterator MI,
+                                  const std::vector<CalleeSavedInfo> &CSI,
+                                  const TargetRegisterInfo *TRI) const override;
 
-  bool hasFP(const MachineFunction &MF) const;
-  bool hasReservedCallFrame(const MachineFunction &MF) const;
-  bool canSimplifyCallFramePseudos(const MachineFunction &MF) const;
+  bool hasFP(const MachineFunction &MF) const override;
+  bool hasReservedCallFrame(const MachineFunction &MF) const override;
+  bool canSimplifyCallFramePseudos(const MachineFunction &MF) const override;
   int getFrameIndexReference(const MachineFunction &MF, int FI,
-                             unsigned &FrameReg) const;
-  int ResolveFrameIndexReference(const MachineFunction &MF,
-                                 int FI,
+                             unsigned &FrameReg) const override;
+  int ResolveFrameIndexReference(const MachineFunction &MF, int FI,
                                  unsigned &FrameReg, int SPAdj) const;
-  int getFrameIndexOffset(const MachineFunction &MF, int FI) const;
+  int getFrameIndexOffset(const MachineFunction &MF, int FI) const override;
 
   void processFunctionBeforeCalleeSavedScan(MachineFunction &MF,
-                                            RegScavenger *RS) const;
+                                            RegScavenger *RS) const override;
+
+  void adjustForSegmentedStacks(MachineFunction &MF) const override;
 
  private:
   void emitPushInst(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
@@ -71,10 +67,10 @@ public:
                    bool(*Func)(unsigned, bool),
                    unsigned NumAlignedDPRCS2Regs) const;
 
-  virtual void eliminateCallFramePseudoInstr(
-                                    MachineFunction &MF,
-                                    MachineBasicBlock &MBB,
-                                    MachineBasicBlock::iterator MI) const;
+  void
+  eliminateCallFramePseudoInstr(MachineFunction &MF,
+                                MachineBasicBlock &MBB,
+                                MachineBasicBlock::iterator MI) const override;
 };
 
 } // End llvm namespace
diff --git a/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.cpp b/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.cpp
index c69d313fd9ce..0885c4e5721a 100644
--- a/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.cpp
@@ -57,7 +57,7 @@ ARMHazardRecognizer::getHazardType(SUnit *SU, int Stalls) {
           (LastMCID.TSFlags & ARMII::DomainMask) == ARMII::DomainGeneral) {
         MachineBasicBlock::iterator I = LastMI;
         if (I != LastMI->getParent()->begin()) {
-          I = llvm::prior(I);
+          I = std::prev(I);
           DefMI = &*I;
         }
       }
@@ -77,7 +77,7 @@ ARMHazardRecognizer::getHazardType(SUnit *SU, int Stalls) {
 }
 
 void ARMHazardRecognizer::Reset() {
-  LastMI = 0;
+  LastMI = nullptr;
   FpMLxStalls = 0;
   ScoreboardHazardRecognizer::Reset();
 }
@@ -95,7 +95,7 @@ void ARMHazardRecognizer::EmitInstruction(SUnit *SU) {
 void ARMHazardRecognizer::AdvanceCycle() {
   if (FpMLxStalls && --FpMLxStalls == 0)
     // Stalled for 4 cycles but still can't schedule any other instructions.
-    LastMI = 0;
+    LastMI = nullptr;
   ScoreboardHazardRecognizer::AdvanceCycle();
 }
 
diff --git a/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.h b/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.h
index e1dcec3d1cc8..a8198e26703e 100644
--- a/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.h
+++ b/contrib/llvm/lib/Target/ARM/ARMHazardRecognizer.h
@@ -35,13 +35,13 @@ public:
   ARMHazardRecognizer(const InstrItineraryData *ItinData,
                       const ScheduleDAG *DAG)
     : ScoreboardHazardRecognizer(ItinData, DAG, "post-RA-sched"),
-      LastMI(0) {}
+      LastMI(nullptr) {}
 
-  virtual HazardType getHazardType(SUnit *SU, int Stalls);
-  virtual void Reset();
-  virtual void EmitInstruction(SUnit *SU);
-  virtual void AdvanceCycle();
-  virtual void RecedeCycle();
+  HazardType getHazardType(SUnit *SU, int Stalls) override;
+  void Reset() override;
+  void EmitInstruction(SUnit *SU) override;
+  void AdvanceCycle() override;
+  void RecedeCycle() override;
 };
 
 } // end namespace llvm
diff --git a/contrib/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp b/contrib/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp
index 6d9b18877f72..38547cfae2e0 100644
--- a/contrib/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMISelDAGToDAG.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-isel"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMTargetMachine.h"
@@ -32,12 +31,13 @@
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
-#include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetLowering.h"
 #include "llvm/Target/TargetOptions.h"
 
 using namespace llvm;
 
+#define DEBUG_TYPE "arm-isel"
+
 static cl::opt<bool>
 DisableShifterOp("disable-shifter-op", cl::Hidden,
   cl::desc("Disable isel of shifter-op"),
@@ -60,24 +60,26 @@ enum AddrMode2Type {
 };
 
 class ARMDAGToDAGISel : public SelectionDAGISel {
-  ARMBaseTargetMachine &TM;
-
   /// Subtarget - Keep a pointer to the ARMSubtarget around so that we can
   /// make the right decision when generating code for different targets.
   const ARMSubtarget *Subtarget;
 
 public:
-  explicit ARMDAGToDAGISel(ARMBaseTargetMachine &tm,
-                           CodeGenOpt::Level OptLevel)
-    : SelectionDAGISel(tm, OptLevel), TM(tm),
-      Subtarget(&TM.getSubtarget<ARMSubtarget>()) {
+  explicit ARMDAGToDAGISel(ARMBaseTargetMachine &tm, CodeGenOpt::Level OptLevel)
+      : SelectionDAGISel(tm, OptLevel) {}
+
+  bool runOnMachineFunction(MachineFunction &MF) override {
+    // Reset the subtarget each time through.
+    Subtarget = &MF.getTarget().getSubtarget<ARMSubtarget>();
+    SelectionDAGISel::runOnMachineFunction(MF);
+    return true;
   }
 
-  virtual const char *getPassName() const {
+  const char *getPassName() const override {
     return "ARM Instruction Selection";
   }
 
-  virtual void PreprocessISelDAG();
+  void PreprocessISelDAG() override;
 
   /// getI32Imm - Return a target constant of type i32 with the specified
   /// value.
@@ -85,7 +87,7 @@ public:
     return CurDAG->getTargetConstant(Imm, MVT::i32);
   }
 
-  SDNode *Select(SDNode *N);
+  SDNode *Select(SDNode *N) override;
 
 
   bool hasNoVMLxHazardUse(SDNode *N) const;
@@ -253,13 +255,10 @@ private:
 
   SDNode *SelectConcatVector(SDNode *N);
 
-  SDNode *SelectAtomic(SDNode *N, unsigned Op8, unsigned Op16, unsigned Op32, unsigned Op64);
-
   /// SelectInlineAsmMemoryOperand - Implement addressing mode selection for
   /// inline asm expressions.
-  virtual bool SelectInlineAsmMemoryOperand(const SDValue &Op,
-                                            char ConstraintCode,
-                                            std::vector<SDValue> &OutOps);
+  bool SelectInlineAsmMemoryOperand(const SDValue &Op, char ConstraintCode,
+                                    std::vector<SDValue> &OutOps) override;
 
   // Form pairs of consecutive R, S, D, or Q registers.
   SDNode *createGPRPairNode(EVT VT, SDValue V0, SDValue V1);
@@ -401,7 +400,7 @@ void ARMDAGToDAGISel::PreprocessISelDAG() {
     N1 = CurDAG->getNode(ISD::SHL, SDLoc(N1), MVT::i32,
                          N1, CurDAG->getConstant(TZ, MVT::i32));
     CurDAG->UpdateNodeOperands(N, N0, N1);
-  }  
+  }
 }
 
 /// hasNoVMLxHazardUse - Return true if it's desirable to select a FP MLA / MLS
@@ -414,8 +413,8 @@ bool ARMDAGToDAGISel::hasNoVMLxHazardUse(SDNode *N) const {
   if (!CheckVMLxHazard)
     return true;
 
-  if (!Subtarget->isCortexA8() && !Subtarget->isCortexA9() &&
-      !Subtarget->isSwift())
+  if (!Subtarget->isCortexA7() && !Subtarget->isCortexA8() &&
+      !Subtarget->isCortexA9() && !Subtarget->isSwift())
     return true;
 
   if (!N->hasOneUse())
@@ -425,8 +424,8 @@ bool ARMDAGToDAGISel::hasNoVMLxHazardUse(SDNode *N) const {
   if (Use->getOpcode() == ISD::CopyToReg)
     return true;
   if (Use->isMachineOpcode()) {
-    const ARMBaseInstrInfo *TII =
-      static_cast<const ARMBaseInstrInfo*>(TM.getInstrInfo());
+    const ARMBaseInstrInfo *TII = static_cast<const ARMBaseInstrInfo *>(
+        CurDAG->getTarget().getInstrInfo());
 
     const MCInstrDesc &MCID = TII->get(Use->getMachineOpcode());
     if (MCID.mayStore())
@@ -534,8 +533,7 @@ bool ARMDAGToDAGISel::SelectAddrModeImm12(SDValue N,
     }
 
     if (N.getOpcode() == ARMISD::Wrapper &&
-        !(Subtarget->useMovt() &&
-                     N.getOperand(0).getOpcode() == ISD::TargetGlobalAddress)) {
+        N.getOperand(0).getOpcode() != ISD::TargetGlobalAddress) {
       Base = N.getOperand(0);
     } else
       Base = N;
@@ -702,8 +700,7 @@ AddrMode2Type ARMDAGToDAGISel::SelectAddrMode2Worker(SDValue N,
       Base = CurDAG->getTargetFrameIndex(FI,
                                          getTargetLowering()->getPointerTy());
     } else if (N.getOpcode() == ARMISD::Wrapper &&
-               !(Subtarget->useMovt() &&
-                 N.getOperand(0).getOpcode() == ISD::TargetGlobalAddress)) {
+               N.getOperand(0).getOpcode() != ISD::TargetGlobalAddress) {
       Base = N.getOperand(0);
     }
     Offset = CurDAG->getRegister(0, MVT::i32);
@@ -963,8 +960,7 @@ bool ARMDAGToDAGISel::SelectAddrMode5(SDValue N,
       Base = CurDAG->getTargetFrameIndex(FI,
                                          getTargetLowering()->getPointerTy());
     } else if (N.getOpcode() == ARMISD::Wrapper &&
-               !(Subtarget->useMovt() &&
-                 N.getOperand(0).getOpcode() == ISD::TargetGlobalAddress)) {
+               N.getOperand(0).getOpcode() != ISD::TargetGlobalAddress) {
       Base = N.getOperand(0);
     }
     Offset = CurDAG->getTargetConstant(ARM_AM::getAM5Opc(ARM_AM::add, 0),
@@ -1141,8 +1137,7 @@ ARMDAGToDAGISel::SelectThumbAddrModeImm5S(SDValue N, unsigned Scale,
 
   if (!CurDAG->isBaseWithConstantOffset(N)) {
     if (N.getOpcode() == ARMISD::Wrapper &&
-        !(Subtarget->useMovt() &&
-          N.getOperand(0).getOpcode() == ISD::TargetGlobalAddress)) {
+        N.getOperand(0).getOpcode() != ISD::TargetGlobalAddress) {
       Base = N.getOperand(0);
     } else {
       Base = N;
@@ -1278,8 +1273,7 @@ bool ARMDAGToDAGISel::SelectT2AddrModeImm12(SDValue N,
     }
 
     if (N.getOpcode() == ARMISD::Wrapper &&
-               !(Subtarget->useMovt() &&
-                 N.getOperand(0).getOpcode() == ISD::TargetGlobalAddress)) {
+        N.getOperand(0).getOpcode() != ISD::TargetGlobalAddress) {
       Base = N.getOperand(0);
       if (Base.getOpcode() == ISD::TargetConstantPool)
         return false;  // We want to select t2LDRpci instead.
@@ -1412,7 +1406,7 @@ bool ARMDAGToDAGISel::SelectT2AddrModeSoReg(SDValue N,
 
 bool ARMDAGToDAGISel::SelectT2AddrModeExclusive(SDValue N, SDValue &Base,
                                                 SDValue &OffImm) {
-  // This *must* succeed since it's used for the irreplacable ldrex and strex
+  // This *must* succeed since it's used for the irreplaceable ldrex and strex
   // instructions.
   Base = N;
   OffImm = CurDAG->getTargetConstant(0, MVT::i32);
@@ -1449,7 +1443,7 @@ SDNode *ARMDAGToDAGISel::SelectARMIndexedLoad(SDNode *N) {
   LoadSDNode *LD = cast<LoadSDNode>(N);
   ISD::MemIndexedMode AM = LD->getAddressingMode();
   if (AM == ISD::UNINDEXED)
-    return NULL;
+    return nullptr;
 
   EVT LoadedVT = LD->getMemoryVT();
   SDValue Offset, AMOpc;
@@ -1515,14 +1509,14 @@ SDNode *ARMDAGToDAGISel::SelectARMIndexedLoad(SDNode *N) {
     }
   }
 
-  return NULL;
+  return nullptr;
 }
 
 SDNode *ARMDAGToDAGISel::SelectT2IndexedLoad(SDNode *N) {
   LoadSDNode *LD = cast<LoadSDNode>(N);
   ISD::MemIndexedMode AM = LD->getAddressingMode();
   if (AM == ISD::UNINDEXED)
-    return NULL;
+    return nullptr;
 
   EVT LoadedVT = LD->getMemoryVT();
   bool isSExtLd = LD->getExtensionType() == ISD::SEXTLOAD;
@@ -1549,7 +1543,7 @@ SDNode *ARMDAGToDAGISel::SelectT2IndexedLoad(SDNode *N) {
         Opcode = isPre ? ARM::t2LDRB_PRE : ARM::t2LDRB_POST;
       break;
     default:
-      return NULL;
+      return nullptr;
     }
     Match = true;
   }
@@ -1563,7 +1557,7 @@ SDNode *ARMDAGToDAGISel::SelectT2IndexedLoad(SDNode *N) {
                                   MVT::Other, Ops);
   }
 
-  return NULL;
+  return nullptr;
 }
 
 /// \brief Form a GPRPair pseudo register from a pair of GPR regs.
@@ -1708,10 +1702,10 @@ static bool isVSTfixed(unsigned Opc)
   case ARM::VST1d16wb_fixed : return true;
   case ARM::VST1d32wb_fixed : return true;
   case ARM::VST1d64wb_fixed : return true;
-  case ARM::VST1q8wb_fixed : return true; 
-  case ARM::VST1q16wb_fixed : return true; 
-  case ARM::VST1q32wb_fixed : return true; 
-  case ARM::VST1q64wb_fixed : return true; 
+  case ARM::VST1q8wb_fixed : return true;
+  case ARM::VST1q16wb_fixed : return true;
+  case ARM::VST1q32wb_fixed : return true;
+  case ARM::VST1q64wb_fixed : return true;
   case ARM::VST1d64TPseudoWB_fixed : return true;
   case ARM::VST1d64QPseudoWB_fixed : return true;
   case ARM::VST2d8wb_fixed : return true;
@@ -1785,7 +1779,7 @@ SDNode *ARMDAGToDAGISel::SelectVLD(SDNode *N, bool isUpdating, unsigned NumVecs,
   SDValue MemAddr, Align;
   unsigned AddrOpIdx = isUpdating ? 1 : 2;
   if (!SelectAddrMode6(N, N->getOperand(AddrOpIdx), MemAddr, Align))
-    return NULL;
+    return nullptr;
 
   SDValue Chain = N->getOperand(0);
   EVT VT = N->getValueType(0);
@@ -1904,7 +1898,7 @@ SDNode *ARMDAGToDAGISel::SelectVLD(SDNode *N, bool isUpdating, unsigned NumVecs,
   ReplaceUses(SDValue(N, NumVecs), SDValue(VLd, 1));
   if (isUpdating)
     ReplaceUses(SDValue(N, NumVecs + 1), SDValue(VLd, 2));
-  return NULL;
+  return nullptr;
 }
 
 SDNode *ARMDAGToDAGISel::SelectVST(SDNode *N, bool isUpdating, unsigned NumVecs,
@@ -1918,7 +1912,7 @@ SDNode *ARMDAGToDAGISel::SelectVST(SDNode *N, bool isUpdating, unsigned NumVecs,
   unsigned AddrOpIdx = isUpdating ? 1 : 2;
   unsigned Vec0Idx = 3; // AddrOpIdx + (isUpdating ? 2 : 1)
   if (!SelectAddrMode6(N, N->getOperand(AddrOpIdx), MemAddr, Align))
-    return NULL;
+    return nullptr;
 
   MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
   MemOp[0] = cast<MemIntrinsicSDNode>(N)->getMemOperand();
@@ -2064,7 +2058,7 @@ SDNode *ARMDAGToDAGISel::SelectVLDSTLane(SDNode *N, bool IsLoad,
   unsigned AddrOpIdx = isUpdating ? 1 : 2;
   unsigned Vec0Idx = 3; // AddrOpIdx + (isUpdating ? 2 : 1)
   if (!SelectAddrMode6(N, N->getOperand(AddrOpIdx), MemAddr, Align))
-    return NULL;
+    return nullptr;
 
   MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
   MemOp[0] = cast<MemIntrinsicSDNode>(N)->getMemOperand();
@@ -2169,7 +2163,7 @@ SDNode *ARMDAGToDAGISel::SelectVLDSTLane(SDNode *N, bool IsLoad,
   ReplaceUses(SDValue(N, NumVecs), SDValue(VLdLn, 1));
   if (isUpdating)
     ReplaceUses(SDValue(N, NumVecs + 1), SDValue(VLdLn, 2));
-  return NULL;
+  return nullptr;
 }
 
 SDNode *ARMDAGToDAGISel::SelectVLDDup(SDNode *N, bool isUpdating,
@@ -2180,7 +2174,7 @@ SDNode *ARMDAGToDAGISel::SelectVLDDup(SDNode *N, bool isUpdating,
 
   SDValue MemAddr, Align;
   if (!SelectAddrMode6(N, N->getOperand(1), MemAddr, Align))
-    return NULL;
+    return nullptr;
 
   MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
   MemOp[0] = cast<MemIntrinsicSDNode>(N)->getMemOperand();
@@ -2252,7 +2246,7 @@ SDNode *ARMDAGToDAGISel::SelectVLDDup(SDNode *N, bool isUpdating,
   ReplaceUses(SDValue(N, NumVecs), SDValue(VLdDup, 1));
   if (isUpdating)
     ReplaceUses(SDValue(N, NumVecs + 1), SDValue(VLdDup, 2));
-  return NULL;
+  return nullptr;
 }
 
 SDNode *ARMDAGToDAGISel::SelectVTBL(SDNode *N, bool IsExt, unsigned NumVecs,
@@ -2291,7 +2285,7 @@ SDNode *ARMDAGToDAGISel::SelectVTBL(SDNode *N, bool IsExt, unsigned NumVecs,
 SDNode *ARMDAGToDAGISel::SelectV6T2BitfieldExtractOp(SDNode *N,
                                                      bool isSigned) {
   if (!Subtarget->hasV6T2Ops())
-    return NULL;
+    return nullptr;
 
   unsigned Opc = isSigned
     ? (Subtarget->isThumb() ? ARM::t2SBFX : ARM::SBFX)
@@ -2304,7 +2298,7 @@ SDNode *ARMDAGToDAGISel::SelectV6T2BitfieldExtractOp(SDNode *N,
 
       // The immediate is a mask of the low bits iff imm & (imm+1) == 0
       if (And_imm & (And_imm + 1))
-        return NULL;
+        return nullptr;
 
       unsigned Srl_imm = 0;
       if (isOpcWithIntImmediate(N->getOperand(0).getNode(), ISD::SRL,
@@ -2324,7 +2318,7 @@ SDNode *ARMDAGToDAGISel::SelectV6T2BitfieldExtractOp(SDNode *N,
             SDValue Ops[] = { N->getOperand(0).getOperand(0),
                               CurDAG->getTargetConstant(LSB, MVT::i32),
                               getAL(CurDAG), Reg0, Reg0 };
-            return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops, 5);
+            return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops);
           }
 
           // ARM models shift instructions as MOVsi with shifter operand.
@@ -2334,17 +2328,17 @@ SDNode *ARMDAGToDAGISel::SelectV6T2BitfieldExtractOp(SDNode *N,
                                       MVT::i32);
           SDValue Ops[] = { N->getOperand(0).getOperand(0), ShOpc,
                             getAL(CurDAG), Reg0, Reg0 };
-          return CurDAG->SelectNodeTo(N, ARM::MOVsi, MVT::i32, Ops, 5);
+          return CurDAG->SelectNodeTo(N, ARM::MOVsi, MVT::i32, Ops);
         }
 
         SDValue Ops[] = { N->getOperand(0).getOperand(0),
                           CurDAG->getTargetConstant(LSB, MVT::i32),
                           CurDAG->getTargetConstant(Width, MVT::i32),
-          getAL(CurDAG), Reg0 };
-        return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops, 5);
+                          getAL(CurDAG), Reg0 };
+        return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops);
       }
     }
-    return NULL;
+    return nullptr;
   }
 
   // Otherwise, we're looking for a shift of a shift
@@ -2358,16 +2352,16 @@ SDNode *ARMDAGToDAGISel::SelectV6T2BitfieldExtractOp(SDNode *N,
       unsigned Width = 32 - Srl_imm - 1;
       int LSB = Srl_imm - Shl_imm;
       if (LSB < 0)
-        return NULL;
+        return nullptr;
       SDValue Reg0 = CurDAG->getRegister(0, MVT::i32);
       SDValue Ops[] = { N->getOperand(0).getOperand(0),
                         CurDAG->getTargetConstant(LSB, MVT::i32),
                         CurDAG->getTargetConstant(Width, MVT::i32),
                         getAL(CurDAG), Reg0 };
-      return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops, 5);
+      return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops);
     }
   }
-  return NULL;
+  return nullptr;
 }
 
 /// Target-specific DAG combining for ISD::XOR.
@@ -2386,10 +2380,10 @@ SDNode *ARMDAGToDAGISel::SelectABSOp(SDNode *N){
   EVT VT = N->getValueType(0);
 
   if (Subtarget->isThumb1Only())
-    return NULL;
+    return nullptr;
 
   if (XORSrc0.getOpcode() != ISD::ADD || XORSrc1.getOpcode() != ISD::SRA)
-    return NULL;
+    return nullptr;
 
   SDValue ADDSrc0 = XORSrc0.getOperand(0);
   SDValue ADDSrc1 = XORSrc0.getOperand(1);
@@ -2400,13 +2394,13 @@ SDNode *ARMDAGToDAGISel::SelectABSOp(SDNode *N){
   unsigned Size = XType.getSizeInBits() - 1;
 
   if (ADDSrc1 == XORSrc1 && ADDSrc0 == SRASrc0 &&
-      XType.isInteger() && SRAConstant != NULL &&
+      XType.isInteger() && SRAConstant != nullptr &&
       Size == SRAConstant->getZExtValue()) {
     unsigned Opcode = Subtarget->isThumb2() ? ARM::t2ABS : ARM::ABS;
     return CurDAG->SelectNodeTo(N, Opcode, VT, ADDSrc0);
   }
 
-  return NULL;
+  return nullptr;
 }
 
 SDNode *ARMDAGToDAGISel::SelectConcatVector(SDNode *N) {
@@ -2418,44 +2412,12 @@ SDNode *ARMDAGToDAGISel::SelectConcatVector(SDNode *N) {
   return createDRegPairNode(VT, N->getOperand(0), N->getOperand(1));
 }
 
-SDNode *ARMDAGToDAGISel::SelectAtomic(SDNode *Node, unsigned Op8,
-                                      unsigned Op16,unsigned Op32,
-                                      unsigned Op64) {
-  // Mostly direct translation to the given operations, except that we preserve
-  // the AtomicOrdering for use later on.
-  AtomicSDNode *AN = cast<AtomicSDNode>(Node);
-  EVT VT = AN->getMemoryVT();
-
-  unsigned Op;
-  SDVTList VTs = CurDAG->getVTList(AN->getValueType(0), MVT::Other);
-  if (VT == MVT::i8)
-    Op = Op8;
-  else if (VT == MVT::i16)
-    Op = Op16;
-  else if (VT == MVT::i32)
-    Op = Op32;
-  else if (VT == MVT::i64) {
-    Op = Op64;
-    VTs = CurDAG->getVTList(MVT::i32, MVT::i32, MVT::Other);
-  } else
-    llvm_unreachable("Unexpected atomic operation");
-
-  SmallVector<SDValue, 6> Ops;
-  for (unsigned i = 1; i < AN->getNumOperands(); ++i)
-      Ops.push_back(AN->getOperand(i));
-
-  Ops.push_back(CurDAG->getTargetConstant(AN->getOrdering(), MVT::i32));
-  Ops.push_back(AN->getOperand(0)); // Chain moves to the end
-
-  return CurDAG->SelectNodeTo(Node, Op, VTs, &Ops[0], Ops.size());
-}
-
 SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
   SDLoc dl(N);
 
   if (N->isMachineOpcode()) {
     N->setNodeId(-1);
-    return NULL;   // Already selected.
+    return nullptr;   // Already selected.
   }
 
   switch (N->getOpcode()) {
@@ -2477,19 +2439,21 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
   case ISD::Constant: {
     unsigned Val = cast<ConstantSDNode>(N)->getZExtValue();
     bool UseCP = true;
-    if (Subtarget->hasThumb2())
+    if (Subtarget->useMovt(*MF))
       // Thumb2-aware targets have the MOVT instruction, so all immediates can
       // be done with MOV + MOVT, at worst.
-      UseCP = 0;
+      UseCP = false;
     else {
       if (Subtarget->isThumb()) {
-        UseCP = (Val > 255 &&                          // MOV
-                 ~Val > 255 &&                         // MOV + MVN
-                 !ARM_AM::isThumbImmShiftedVal(Val));  // MOV + LSL
+        UseCP = (Val > 255 &&                                  // MOV
+                 ~Val > 255 &&                                 // MOV + MVN
+                 !ARM_AM::isThumbImmShiftedVal(Val) &&         // MOV + LSL
+                 !(Subtarget->hasV6T2Ops() && Val <= 0xffff)); // MOVW
       } else
-        UseCP = (ARM_AM::getSOImmVal(Val) == -1 &&     // MOV
-                 ARM_AM::getSOImmVal(~Val) == -1 &&    // MVN
-                 !ARM_AM::isSOImmTwoPartVal(Val));     // two instrs.
+        UseCP = (ARM_AM::getSOImmVal(Val) == -1 &&             // MOV
+                 ARM_AM::getSOImmVal(~Val) == -1 &&            // MVN
+                 !ARM_AM::isSOImmTwoPartVal(Val) &&            // two instrs.
+                 !(Subtarget->hasV6T2Ops() && Val <= 0xffff)); // MOVW
     }
 
     if (UseCP) {
@@ -2499,7 +2463,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                       getTargetLowering()->getPointerTy());
 
       SDNode *ResNode;
-      if (Subtarget->isThumb1Only()) {
+      if (Subtarget->isThumb()) {
         SDValue Pred = getAL(CurDAG);
         SDValue PredReg = CurDAG->getRegister(0, MVT::i32);
         SDValue Ops[] = { CPIdx, Pred, PredReg, CurDAG->getEntryNode() };
@@ -2517,7 +2481,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                        Ops);
       }
       ReplaceUses(SDValue(N, 0), SDValue(ResNode, 0));
-      return NULL;
+      return nullptr;
     }
 
     // Other cases are autogenerated.
@@ -2531,14 +2495,14 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     if (Subtarget->isThumb1Only()) {
       SDValue Ops[] = { TFI, CurDAG->getTargetConstant(0, MVT::i32),
                         getAL(CurDAG), CurDAG->getRegister(0, MVT::i32) };
-      return CurDAG->SelectNodeTo(N, ARM::tADDrSPi, MVT::i32, Ops, 4);
+      return CurDAG->SelectNodeTo(N, ARM::tADDrSPi, MVT::i32, Ops);
     } else {
       unsigned Opc = ((Subtarget->isThumb() && Subtarget->hasThumb2()) ?
                       ARM::t2ADDri : ARM::ADDri);
       SDValue Ops[] = { TFI, CurDAG->getTargetConstant(0, MVT::i32),
                         getAL(CurDAG), CurDAG->getRegister(0, MVT::i32),
                         CurDAG->getRegister(0, MVT::i32) };
-      return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops, 5);
+      return CurDAG->SelectNodeTo(N, Opc, MVT::i32, Ops);
     }
   }
   case ISD::SRL:
@@ -2565,10 +2529,10 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
         SDValue Reg0 = CurDAG->getRegister(0, MVT::i32);
         if (Subtarget->isThumb()) {
           SDValue Ops[] = { V, V, ShImmOp, getAL(CurDAG), Reg0, Reg0 };
-          return CurDAG->SelectNodeTo(N, ARM::t2ADDrs, MVT::i32, Ops, 6);
+          return CurDAG->SelectNodeTo(N, ARM::t2ADDrs, MVT::i32, Ops);
         } else {
           SDValue Ops[] = { V, V, Reg0, ShImmOp, getAL(CurDAG), Reg0, Reg0 };
-          return CurDAG->SelectNodeTo(N, ARM::ADDrsi, MVT::i32, Ops, 7);
+          return CurDAG->SelectNodeTo(N, ARM::ADDrsi, MVT::i32, Ops);
         }
       }
       if (isPowerOf2_32(RHSV+1)) {  // 2^n-1?
@@ -2581,10 +2545,10 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
         SDValue Reg0 = CurDAG->getRegister(0, MVT::i32);
         if (Subtarget->isThumb()) {
           SDValue Ops[] = { V, V, ShImmOp, getAL(CurDAG), Reg0, Reg0 };
-          return CurDAG->SelectNodeTo(N, ARM::t2RSBrs, MVT::i32, Ops, 6);
+          return CurDAG->SelectNodeTo(N, ARM::t2RSBrs, MVT::i32, Ops);
         } else {
           SDValue Ops[] = { V, V, Reg0, ShImmOp, getAL(CurDAG), Reg0, Reg0 };
-          return CurDAG->SelectNodeTo(N, ARM::RSBrsi, MVT::i32, Ops, 7);
+          return CurDAG->SelectNodeTo(N, ARM::RSBrsi, MVT::i32, Ops);
         }
       }
     }
@@ -2699,7 +2663,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     }
   }
   case ISD::LOAD: {
-    SDNode *ResNode = 0;
+    SDNode *ResNode = nullptr;
     if (Subtarget->isThumb() && Subtarget->hasThumb2())
       ResNode = SelectT2IndexedLoad(N);
     else
@@ -2746,13 +2710,13 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     }
     ReplaceUses(SDValue(N, 0),
                 SDValue(Chain.getNode(), Chain.getResNo()));
-    return NULL;
+    return nullptr;
   }
   case ARMISD::VZIP: {
     unsigned Opc = 0;
     EVT VT = N->getValueType(0);
     switch (VT.getSimpleVT().SimpleTy) {
-    default: return NULL;
+    default: return nullptr;
     case MVT::v8i8:  Opc = ARM::VZIPd8; break;
     case MVT::v4i16: Opc = ARM::VZIPd16; break;
     case MVT::v2f32:
@@ -2772,7 +2736,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     unsigned Opc = 0;
     EVT VT = N->getValueType(0);
     switch (VT.getSimpleVT().SimpleTy) {
-    default: return NULL;
+    default: return nullptr;
     case MVT::v8i8:  Opc = ARM::VUZPd8; break;
     case MVT::v4i16: Opc = ARM::VUZPd16; break;
     case MVT::v2f32:
@@ -2792,7 +2756,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     unsigned Opc = 0;
     EVT VT = N->getValueType(0);
     switch (VT.getSimpleVT().SimpleTy) {
-    default: return NULL;
+    default: return nullptr;
     case MVT::v8i8:  Opc = ARM::VTRNd8; break;
     case MVT::v4i16: Opc = ARM::VTRNd16; break;
     case MVT::v2f32:
@@ -2873,7 +2837,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                          ARM::VLD1q16wb_fixed,
                                          ARM::VLD1q32wb_fixed,
                                          ARM::VLD1q64wb_fixed };
-    return SelectVLD(N, true, 1, DOpcodes, QOpcodes, 0);
+    return SelectVLD(N, true, 1, DOpcodes, QOpcodes, nullptr);
   }
 
   case ARMISD::VLD2_UPD: {
@@ -2884,7 +2848,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     static const uint16_t QOpcodes[] = { ARM::VLD2q8PseudoWB_fixed,
                                          ARM::VLD2q16PseudoWB_fixed,
                                          ARM::VLD2q32PseudoWB_fixed };
-    return SelectVLD(N, true, 2, DOpcodes, QOpcodes, 0);
+    return SelectVLD(N, true, 2, DOpcodes, QOpcodes, nullptr);
   }
 
   case ARMISD::VLD3_UPD: {
@@ -2951,7 +2915,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                          ARM::VST1q16wb_fixed,
                                          ARM::VST1q32wb_fixed,
                                          ARM::VST1q64wb_fixed };
-    return SelectVST(N, true, 1, DOpcodes, QOpcodes, 0);
+    return SelectVST(N, true, 1, DOpcodes, QOpcodes, nullptr);
   }
 
   case ARMISD::VST2_UPD: {
@@ -2962,7 +2926,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     static const uint16_t QOpcodes[] = { ARM::VST2q8PseudoWB_fixed,
                                          ARM::VST2q16PseudoWB_fixed,
                                          ARM::VST2q32PseudoWB_fixed };
-    return SelectVST(N, true, 2, DOpcodes, QOpcodes, 0);
+    return SelectVST(N, true, 2, DOpcodes, QOpcodes, nullptr);
   }
 
   case ARMISD::VST3_UPD: {
@@ -3027,13 +2991,16 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
     default:
       break;
 
+    case Intrinsic::arm_ldaexd:
     case Intrinsic::arm_ldrexd: {
-      SDValue MemAddr = N->getOperand(2);
       SDLoc dl(N);
       SDValue Chain = N->getOperand(0);
-
+      SDValue MemAddr = N->getOperand(2);
       bool isThumb = Subtarget->isThumb() && Subtarget->hasThumb2();
-      unsigned NewOpc = isThumb ? ARM::t2LDREXD :ARM::LDREXD;
+
+      bool IsAcquire = IntNo == Intrinsic::arm_ldaexd;
+      unsigned NewOpc = isThumb ? (IsAcquire ? ARM::t2LDAEXD : ARM::t2LDREXD)
+                                : (IsAcquire ? ARM::LDAEXD : ARM::LDREXD);
 
       // arm_ldrexd returns a i64 value in {i32, i32}
       std::vector<EVT> ResTys;
@@ -3083,9 +3050,9 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
         ReplaceUses(SDValue(N, 1), Result);
       }
       ReplaceUses(SDValue(N, 2), OutChain);
-      return NULL;
+      return nullptr;
     }
-
+    case Intrinsic::arm_stlexd:
     case Intrinsic::arm_strexd: {
       SDLoc dl(N);
       SDValue Chain = N->getOperand(0);
@@ -3111,7 +3078,9 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
       Ops.push_back(CurDAG->getRegister(0, MVT::i32));
       Ops.push_back(Chain);
 
-      unsigned NewOpc = isThumb ? ARM::t2STREXD : ARM::STREXD;
+      bool IsRelease = IntNo == Intrinsic::arm_stlexd;
+      unsigned NewOpc = isThumb ? (IsRelease ? ARM::t2STLEXD : ARM::t2STREXD)
+                                : (IsRelease ? ARM::STLEXD : ARM::STREXD);
 
       SDNode *St = CurDAG->getMachineNode(NewOpc, dl, ResTys, Ops);
       // Transfer memoperands.
@@ -3127,7 +3096,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                            ARM::VLD1d32, ARM::VLD1d64 };
       static const uint16_t QOpcodes[] = { ARM::VLD1q8, ARM::VLD1q16,
                                            ARM::VLD1q32, ARM::VLD1q64};
-      return SelectVLD(N, false, 1, DOpcodes, QOpcodes, 0);
+      return SelectVLD(N, false, 1, DOpcodes, QOpcodes, nullptr);
     }
 
     case Intrinsic::arm_neon_vld2: {
@@ -3135,7 +3104,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                            ARM::VLD2d32, ARM::VLD1q64 };
       static const uint16_t QOpcodes[] = { ARM::VLD2q8Pseudo, ARM::VLD2q16Pseudo,
                                            ARM::VLD2q32Pseudo };
-      return SelectVLD(N, false, 2, DOpcodes, QOpcodes, 0);
+      return SelectVLD(N, false, 2, DOpcodes, QOpcodes, nullptr);
     }
 
     case Intrinsic::arm_neon_vld3: {
@@ -3198,7 +3167,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                            ARM::VST1d32, ARM::VST1d64 };
       static const uint16_t QOpcodes[] = { ARM::VST1q8, ARM::VST1q16,
                                            ARM::VST1q32, ARM::VST1q64 };
-      return SelectVST(N, false, 1, DOpcodes, QOpcodes, 0);
+      return SelectVST(N, false, 1, DOpcodes, QOpcodes, nullptr);
     }
 
     case Intrinsic::arm_neon_vst2: {
@@ -3206,7 +3175,7 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
                                            ARM::VST2d32, ARM::VST1q64 };
       static uint16_t QOpcodes[] = { ARM::VST2q8Pseudo, ARM::VST2q16Pseudo,
                                      ARM::VST2q32Pseudo };
-      return SelectVST(N, false, 2, DOpcodes, QOpcodes, 0);
+      return SelectVST(N, false, 2, DOpcodes, QOpcodes, nullptr);
     }
 
     case Intrinsic::arm_neon_vst3: {
@@ -3320,91 +3289,6 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
 
   case ISD::CONCAT_VECTORS:
     return SelectConcatVector(N);
-
-  case ISD::ATOMIC_LOAD:
-    if (cast<AtomicSDNode>(N)->getMemoryVT() == MVT::i64)
-      return SelectAtomic(N, 0, 0, 0, ARM::ATOMIC_LOAD_I64);
-    else
-      break;
-
-  case ISD::ATOMIC_STORE:
-    if (cast<AtomicSDNode>(N)->getMemoryVT() == MVT::i64)
-      return SelectAtomic(N, 0, 0, 0, ARM::ATOMIC_STORE_I64);
-    else
-      break;
-
-  case ISD::ATOMIC_LOAD_ADD:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_ADD_I8,
-                        ARM::ATOMIC_LOAD_ADD_I16,
-                        ARM::ATOMIC_LOAD_ADD_I32,
-                        ARM::ATOMIC_LOAD_ADD_I64);
-  case ISD::ATOMIC_LOAD_SUB:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_SUB_I8,
-                        ARM::ATOMIC_LOAD_SUB_I16,
-                        ARM::ATOMIC_LOAD_SUB_I32,
-                        ARM::ATOMIC_LOAD_SUB_I64);
-  case ISD::ATOMIC_LOAD_AND:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_AND_I8,
-                        ARM::ATOMIC_LOAD_AND_I16,
-                        ARM::ATOMIC_LOAD_AND_I32,
-                        ARM::ATOMIC_LOAD_AND_I64);
-  case ISD::ATOMIC_LOAD_OR:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_OR_I8,
-                        ARM::ATOMIC_LOAD_OR_I16,
-                        ARM::ATOMIC_LOAD_OR_I32,
-                        ARM::ATOMIC_LOAD_OR_I64);
-  case ISD::ATOMIC_LOAD_XOR:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_XOR_I8,
-                        ARM::ATOMIC_LOAD_XOR_I16,
-                        ARM::ATOMIC_LOAD_XOR_I32,
-                        ARM::ATOMIC_LOAD_XOR_I64);
-  case ISD::ATOMIC_LOAD_NAND:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_NAND_I8,
-                        ARM::ATOMIC_LOAD_NAND_I16,
-                        ARM::ATOMIC_LOAD_NAND_I32,
-                        ARM::ATOMIC_LOAD_NAND_I64);
-  case ISD::ATOMIC_LOAD_MIN:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_MIN_I8,
-                        ARM::ATOMIC_LOAD_MIN_I16,
-                        ARM::ATOMIC_LOAD_MIN_I32,
-                        ARM::ATOMIC_LOAD_MIN_I64);
-  case ISD::ATOMIC_LOAD_MAX:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_MAX_I8,
-                        ARM::ATOMIC_LOAD_MAX_I16,
-                        ARM::ATOMIC_LOAD_MAX_I32,
-                        ARM::ATOMIC_LOAD_MAX_I64);
-  case ISD::ATOMIC_LOAD_UMIN:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_UMIN_I8,
-                        ARM::ATOMIC_LOAD_UMIN_I16,
-                        ARM::ATOMIC_LOAD_UMIN_I32,
-                        ARM::ATOMIC_LOAD_UMIN_I64);
-  case ISD::ATOMIC_LOAD_UMAX:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_LOAD_UMAX_I8,
-                        ARM::ATOMIC_LOAD_UMAX_I16,
-                        ARM::ATOMIC_LOAD_UMAX_I32,
-                        ARM::ATOMIC_LOAD_UMAX_I64);
-  case ISD::ATOMIC_SWAP:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_SWAP_I8,
-                        ARM::ATOMIC_SWAP_I16,
-                        ARM::ATOMIC_SWAP_I32,
-                        ARM::ATOMIC_SWAP_I64);
-  case ISD::ATOMIC_CMP_SWAP:
-    return SelectAtomic(N,
-                        ARM::ATOMIC_CMP_SWAP_I8,
-                        ARM::ATOMIC_CMP_SWAP_I16,
-                        ARM::ATOMIC_CMP_SWAP_I32,
-                        ARM::ATOMIC_CMP_SWAP_I64);
   }
 
   return SelectCode(N);
@@ -3425,7 +3309,8 @@ SDNode *ARMDAGToDAGISel::SelectInlineAsm(SDNode *N){
   // them into a GPRPair.
 
   SDLoc dl(N);
-  SDValue Glue = N->getGluedNode() ? N->getOperand(NumOps-1) : SDValue(0,0);
+  SDValue Glue = N->getGluedNode() ? N->getOperand(NumOps-1)
+                                   : SDValue(nullptr,0);
 
   SmallVector<bool, 8> OpChanged;
   // Glue node will be appended late.
@@ -3507,7 +3392,7 @@ SDNode *ARMDAGToDAGISel::SelectInlineAsm(SDNode *N){
       // Update the original glue user.
       std::vector<SDValue> Ops(GU->op_begin(), GU->op_end()-1);
       Ops.push_back(T1.getValue(1));
-      CurDAG->UpdateNodeOperands(GU, &Ops[0], Ops.size());
+      CurDAG->UpdateNodeOperands(GU, Ops);
       GU = T1.getNode();
     }
     else {
@@ -3554,11 +3439,10 @@ SDNode *ARMDAGToDAGISel::SelectInlineAsm(SDNode *N){
   if (Glue.getNode())
     AsmNodeOperands.push_back(Glue);
   if (!Changed)
-    return NULL;
+    return nullptr;
 
   SDValue New = CurDAG->getNode(ISD::INLINEASM, SDLoc(N),
-      CurDAG->getVTList(MVT::Other, MVT::Glue), &AsmNodeOperands[0],
-                        AsmNodeOperands.size());
+      CurDAG->getVTList(MVT::Other, MVT::Glue), AsmNodeOperands);
   New->setNodeId(-1);
   return New.getNode();
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp b/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp
index f3be81877e52..a76531a3869a 100644
--- a/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -12,9 +12,7 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-isel"
 #include "ARMISelLowering.h"
-#include "ARM.h"
 #include "ARMCallingConv.h"
 #include "ARMConstantPoolValue.h"
 #include "ARMMachineFunctionInfo.h"
@@ -38,29 +36,26 @@
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/GlobalValue.h"
+#include "llvm/IR/IRBuilder.h"
 #include "llvm/IR/Instruction.h"
 #include "llvm/IR/Instructions.h"
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/Type.h"
 #include "llvm/MC/MCSectionMachO.h"
 #include "llvm/Support/CommandLine.h"
+#include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/MathExtras.h"
-#include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetOptions.h"
 #include <utility>
 using namespace llvm;
 
+#define DEBUG_TYPE "arm-isel"
+
 STATISTIC(NumTailCalls, "Number of tail calls");
 STATISTIC(NumMovwMovt, "Number of GAs materialized with movw + movt");
 STATISTIC(NumLoopByVals, "Number of loops generated for byval arguments");
 
-// This option should go away when tail calls fully work.
-static cl::opt<bool>
-EnableARMTailCalls("arm-tail-calls", cl::Hidden,
-  cl::desc("Generate tail calls (TEMPORARY OPTION)."),
-  cl::init(false));
-
 cl::opt<bool>
 EnableARMLongCalls("arm-long-calls", cl::Hidden,
   cl::desc("Generate calls via indirect call instructions"),
@@ -87,7 +82,7 @@ namespace {
 }
 
 // The APCS parameter registers.
-static const uint16_t GPRArgRegs[] = {
+static const MCPhysReg GPRArgRegs[] = {
   ARM::R0, ARM::R1, ARM::R2, ARM::R3
 };
 
@@ -160,25 +155,26 @@ void ARMTargetLowering::addQRTypeForNEON(MVT VT) {
   addTypeForNEON(VT, MVT::v2f64, MVT::v4i32);
 }
 
-static TargetLoweringObjectFile *createTLOF(TargetMachine &TM) {
-  if (TM.getSubtarget<ARMSubtarget>().isTargetDarwin())
+static TargetLoweringObjectFile *createTLOF(const Triple &TT) {
+  if (TT.isOSBinFormatMachO())
     return new TargetLoweringObjectFileMachO();
-
+  if (TT.isOSWindows())
+    return new TargetLoweringObjectFileCOFF();
   return new ARMElfTargetObjectFile();
 }
 
 ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
-    : TargetLowering(TM, createTLOF(TM)) {
+    : TargetLowering(TM, createTLOF(Triple(TM.getTargetTriple()))) {
   Subtarget = &TM.getSubtarget<ARMSubtarget>();
   RegInfo = TM.getRegisterInfo();
   Itins = TM.getInstrItineraryData();
 
   setBooleanVectorContents(ZeroOrNegativeOneBooleanContent);
 
-  if (Subtarget->isTargetIOS()) {
+  if (Subtarget->isTargetMachO()) {
     // Uses VFP for Thumb libfuncs if available.
     if (Subtarget->isThumb() && Subtarget->hasVFP2() &&
-        Subtarget->hasARMOps()) {
+        Subtarget->hasARMOps() && !TM.Options.UseSoftFloat) {
       // Single-precision floating-point arithmetic.
       setLibcallName(RTLIB::ADD_F32, "__addsf3vfp");
       setLibcallName(RTLIB::SUB_F32, "__subsf3vfp");
@@ -254,172 +250,134 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
   }
 
   // These libcalls are not available in 32-bit.
-  setLibcallName(RTLIB::SHL_I128, 0);
-  setLibcallName(RTLIB::SRL_I128, 0);
-  setLibcallName(RTLIB::SRA_I128, 0);
-
-  if (Subtarget->isAAPCS_ABI() && !Subtarget->isTargetDarwin()) {
-    // Double-precision floating-point arithmetic helper functions
-    // RTABI chapter 4.1.2, Table 2
-    setLibcallName(RTLIB::ADD_F64, "__aeabi_dadd");
-    setLibcallName(RTLIB::DIV_F64, "__aeabi_ddiv");
-    setLibcallName(RTLIB::MUL_F64, "__aeabi_dmul");
-    setLibcallName(RTLIB::SUB_F64, "__aeabi_dsub");
-    setLibcallCallingConv(RTLIB::ADD_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::DIV_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::MUL_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SUB_F64, CallingConv::ARM_AAPCS);
-
-    // Double-precision floating-point comparison helper functions
-    // RTABI chapter 4.1.2, Table 3
-    setLibcallName(RTLIB::OEQ_F64, "__aeabi_dcmpeq");
-    setCmpLibcallCC(RTLIB::OEQ_F64, ISD::SETNE);
-    setLibcallName(RTLIB::UNE_F64, "__aeabi_dcmpeq");
-    setCmpLibcallCC(RTLIB::UNE_F64, ISD::SETEQ);
-    setLibcallName(RTLIB::OLT_F64, "__aeabi_dcmplt");
-    setCmpLibcallCC(RTLIB::OLT_F64, ISD::SETNE);
-    setLibcallName(RTLIB::OLE_F64, "__aeabi_dcmple");
-    setCmpLibcallCC(RTLIB::OLE_F64, ISD::SETNE);
-    setLibcallName(RTLIB::OGE_F64, "__aeabi_dcmpge");
-    setCmpLibcallCC(RTLIB::OGE_F64, ISD::SETNE);
-    setLibcallName(RTLIB::OGT_F64, "__aeabi_dcmpgt");
-    setCmpLibcallCC(RTLIB::OGT_F64, ISD::SETNE);
-    setLibcallName(RTLIB::UO_F64,  "__aeabi_dcmpun");
-    setCmpLibcallCC(RTLIB::UO_F64,  ISD::SETNE);
-    setLibcallName(RTLIB::O_F64,   "__aeabi_dcmpun");
-    setCmpLibcallCC(RTLIB::O_F64,   ISD::SETEQ);
-    setLibcallCallingConv(RTLIB::OEQ_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UNE_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OLT_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OLE_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OGE_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OGT_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UO_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::O_F64, CallingConv::ARM_AAPCS);
-
-    // Single-precision floating-point arithmetic helper functions
-    // RTABI chapter 4.1.2, Table 4
-    setLibcallName(RTLIB::ADD_F32, "__aeabi_fadd");
-    setLibcallName(RTLIB::DIV_F32, "__aeabi_fdiv");
-    setLibcallName(RTLIB::MUL_F32, "__aeabi_fmul");
-    setLibcallName(RTLIB::SUB_F32, "__aeabi_fsub");
-    setLibcallCallingConv(RTLIB::ADD_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::DIV_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::MUL_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SUB_F32, CallingConv::ARM_AAPCS);
-
-    // Single-precision floating-point comparison helper functions
-    // RTABI chapter 4.1.2, Table 5
-    setLibcallName(RTLIB::OEQ_F32, "__aeabi_fcmpeq");
-    setCmpLibcallCC(RTLIB::OEQ_F32, ISD::SETNE);
-    setLibcallName(RTLIB::UNE_F32, "__aeabi_fcmpeq");
-    setCmpLibcallCC(RTLIB::UNE_F32, ISD::SETEQ);
-    setLibcallName(RTLIB::OLT_F32, "__aeabi_fcmplt");
-    setCmpLibcallCC(RTLIB::OLT_F32, ISD::SETNE);
-    setLibcallName(RTLIB::OLE_F32, "__aeabi_fcmple");
-    setCmpLibcallCC(RTLIB::OLE_F32, ISD::SETNE);
-    setLibcallName(RTLIB::OGE_F32, "__aeabi_fcmpge");
-    setCmpLibcallCC(RTLIB::OGE_F32, ISD::SETNE);
-    setLibcallName(RTLIB::OGT_F32, "__aeabi_fcmpgt");
-    setCmpLibcallCC(RTLIB::OGT_F32, ISD::SETNE);
-    setLibcallName(RTLIB::UO_F32,  "__aeabi_fcmpun");
-    setCmpLibcallCC(RTLIB::UO_F32,  ISD::SETNE);
-    setLibcallName(RTLIB::O_F32,   "__aeabi_fcmpun");
-    setCmpLibcallCC(RTLIB::O_F32,   ISD::SETEQ);
-    setLibcallCallingConv(RTLIB::OEQ_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UNE_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OLT_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OLE_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OGE_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::OGT_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UO_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::O_F32, CallingConv::ARM_AAPCS);
-
-    // Floating-point to integer conversions.
-    // RTABI chapter 4.1.2, Table 6
-    setLibcallName(RTLIB::FPTOSINT_F64_I32, "__aeabi_d2iz");
-    setLibcallName(RTLIB::FPTOUINT_F64_I32, "__aeabi_d2uiz");
-    setLibcallName(RTLIB::FPTOSINT_F64_I64, "__aeabi_d2lz");
-    setLibcallName(RTLIB::FPTOUINT_F64_I64, "__aeabi_d2ulz");
-    setLibcallName(RTLIB::FPTOSINT_F32_I32, "__aeabi_f2iz");
-    setLibcallName(RTLIB::FPTOUINT_F32_I32, "__aeabi_f2uiz");
-    setLibcallName(RTLIB::FPTOSINT_F32_I64, "__aeabi_f2lz");
-    setLibcallName(RTLIB::FPTOUINT_F32_I64, "__aeabi_f2ulz");
-    setLibcallCallingConv(RTLIB::FPTOSINT_F64_I32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOUINT_F64_I32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOSINT_F64_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOUINT_F64_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOSINT_F32_I32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOUINT_F32_I32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOSINT_F32_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPTOUINT_F32_I64, CallingConv::ARM_AAPCS);
-
-    // Conversions between floating types.
-    // RTABI chapter 4.1.2, Table 7
-    setLibcallName(RTLIB::FPROUND_F64_F32, "__aeabi_d2f");
-    setLibcallName(RTLIB::FPEXT_F32_F64,   "__aeabi_f2d");
-    setLibcallCallingConv(RTLIB::FPROUND_F64_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::FPEXT_F32_F64, CallingConv::ARM_AAPCS);
-
-    // Integer to floating-point conversions.
-    // RTABI chapter 4.1.2, Table 8
-    setLibcallName(RTLIB::SINTTOFP_I32_F64, "__aeabi_i2d");
-    setLibcallName(RTLIB::UINTTOFP_I32_F64, "__aeabi_ui2d");
-    setLibcallName(RTLIB::SINTTOFP_I64_F64, "__aeabi_l2d");
-    setLibcallName(RTLIB::UINTTOFP_I64_F64, "__aeabi_ul2d");
-    setLibcallName(RTLIB::SINTTOFP_I32_F32, "__aeabi_i2f");
-    setLibcallName(RTLIB::UINTTOFP_I32_F32, "__aeabi_ui2f");
-    setLibcallName(RTLIB::SINTTOFP_I64_F32, "__aeabi_l2f");
-    setLibcallName(RTLIB::UINTTOFP_I64_F32, "__aeabi_ul2f");
-    setLibcallCallingConv(RTLIB::SINTTOFP_I32_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UINTTOFP_I32_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SINTTOFP_I64_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UINTTOFP_I64_F64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SINTTOFP_I32_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UINTTOFP_I32_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SINTTOFP_I64_F32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UINTTOFP_I64_F32, CallingConv::ARM_AAPCS);
-
-    // Long long helper functions
-    // RTABI chapter 4.2, Table 9
-    setLibcallName(RTLIB::MUL_I64,  "__aeabi_lmul");
-    setLibcallName(RTLIB::SHL_I64, "__aeabi_llsl");
-    setLibcallName(RTLIB::SRL_I64, "__aeabi_llsr");
-    setLibcallName(RTLIB::SRA_I64, "__aeabi_lasr");
-    setLibcallCallingConv(RTLIB::MUL_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SDIV_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UDIV_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SHL_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SRL_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SRA_I64, CallingConv::ARM_AAPCS);
-
-    // Integer division functions
-    // RTABI chapter 4.3.1
-    setLibcallName(RTLIB::SDIV_I8,  "__aeabi_idiv");
-    setLibcallName(RTLIB::SDIV_I16, "__aeabi_idiv");
-    setLibcallName(RTLIB::SDIV_I32, "__aeabi_idiv");
-    setLibcallName(RTLIB::SDIV_I64, "__aeabi_ldivmod");
-    setLibcallName(RTLIB::UDIV_I8,  "__aeabi_uidiv");
-    setLibcallName(RTLIB::UDIV_I16, "__aeabi_uidiv");
-    setLibcallName(RTLIB::UDIV_I32, "__aeabi_uidiv");
-    setLibcallName(RTLIB::UDIV_I64, "__aeabi_uldivmod");
-    setLibcallCallingConv(RTLIB::SDIV_I8, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SDIV_I16, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SDIV_I32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::SDIV_I64, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UDIV_I8, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UDIV_I16, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UDIV_I32, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::UDIV_I64, CallingConv::ARM_AAPCS);
-
-    // Memory operations
-    // RTABI chapter 4.3.4
-    setLibcallName(RTLIB::MEMCPY,  "__aeabi_memcpy");
-    setLibcallName(RTLIB::MEMMOVE, "__aeabi_memmove");
-    setLibcallName(RTLIB::MEMSET,  "__aeabi_memset");
-    setLibcallCallingConv(RTLIB::MEMCPY, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::MEMMOVE, CallingConv::ARM_AAPCS);
-    setLibcallCallingConv(RTLIB::MEMSET, CallingConv::ARM_AAPCS);
+  setLibcallName(RTLIB::SHL_I128, nullptr);
+  setLibcallName(RTLIB::SRL_I128, nullptr);
+  setLibcallName(RTLIB::SRA_I128, nullptr);
+
+  if (Subtarget->isAAPCS_ABI() && !Subtarget->isTargetMachO() &&
+      !Subtarget->isTargetWindows()) {
+    static const struct {
+      const RTLIB::Libcall Op;
+      const char * const Name;
+      const CallingConv::ID CC;
+      const ISD::CondCode Cond;
+    } LibraryCalls[] = {
+      // Double-precision floating-point arithmetic helper functions
+      // RTABI chapter 4.1.2, Table 2
+      { RTLIB::ADD_F64, "__aeabi_dadd", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::DIV_F64, "__aeabi_ddiv", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::MUL_F64, "__aeabi_dmul", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SUB_F64, "__aeabi_dsub", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Double-precision floating-point comparison helper functions
+      // RTABI chapter 4.1.2, Table 3
+      { RTLIB::OEQ_F64, "__aeabi_dcmpeq", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::UNE_F64, "__aeabi_dcmpeq", CallingConv::ARM_AAPCS, ISD::SETEQ },
+      { RTLIB::OLT_F64, "__aeabi_dcmplt", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::OLE_F64, "__aeabi_dcmple", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::OGE_F64, "__aeabi_dcmpge", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::OGT_F64, "__aeabi_dcmpgt", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::UO_F64,  "__aeabi_dcmpun", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::O_F64,   "__aeabi_dcmpun", CallingConv::ARM_AAPCS, ISD::SETEQ },
+
+      // Single-precision floating-point arithmetic helper functions
+      // RTABI chapter 4.1.2, Table 4
+      { RTLIB::ADD_F32, "__aeabi_fadd", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::DIV_F32, "__aeabi_fdiv", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::MUL_F32, "__aeabi_fmul", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SUB_F32, "__aeabi_fsub", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Single-precision floating-point comparison helper functions
+      // RTABI chapter 4.1.2, Table 5
+      { RTLIB::OEQ_F32, "__aeabi_fcmpeq", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::UNE_F32, "__aeabi_fcmpeq", CallingConv::ARM_AAPCS, ISD::SETEQ },
+      { RTLIB::OLT_F32, "__aeabi_fcmplt", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::OLE_F32, "__aeabi_fcmple", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::OGE_F32, "__aeabi_fcmpge", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::OGT_F32, "__aeabi_fcmpgt", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::UO_F32,  "__aeabi_fcmpun", CallingConv::ARM_AAPCS, ISD::SETNE },
+      { RTLIB::O_F32,   "__aeabi_fcmpun", CallingConv::ARM_AAPCS, ISD::SETEQ },
+
+      // Floating-point to integer conversions.
+      // RTABI chapter 4.1.2, Table 6
+      { RTLIB::FPTOSINT_F64_I32, "__aeabi_d2iz",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOUINT_F64_I32, "__aeabi_d2uiz", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOSINT_F64_I64, "__aeabi_d2lz",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOUINT_F64_I64, "__aeabi_d2ulz", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOSINT_F32_I32, "__aeabi_f2iz",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOUINT_F32_I32, "__aeabi_f2uiz", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOSINT_F32_I64, "__aeabi_f2lz",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPTOUINT_F32_I64, "__aeabi_f2ulz", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Conversions between floating types.
+      // RTABI chapter 4.1.2, Table 7
+      { RTLIB::FPROUND_F64_F32, "__aeabi_d2f", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::FPEXT_F32_F64,   "__aeabi_f2d", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Integer to floating-point conversions.
+      // RTABI chapter 4.1.2, Table 8
+      { RTLIB::SINTTOFP_I32_F64, "__aeabi_i2d",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UINTTOFP_I32_F64, "__aeabi_ui2d", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SINTTOFP_I64_F64, "__aeabi_l2d",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UINTTOFP_I64_F64, "__aeabi_ul2d", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SINTTOFP_I32_F32, "__aeabi_i2f",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UINTTOFP_I32_F32, "__aeabi_ui2f", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SINTTOFP_I64_F32, "__aeabi_l2f",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UINTTOFP_I64_F32, "__aeabi_ul2f", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Long long helper functions
+      // RTABI chapter 4.2, Table 9
+      { RTLIB::MUL_I64, "__aeabi_lmul", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SHL_I64, "__aeabi_llsl", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SRL_I64, "__aeabi_llsr", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SRA_I64, "__aeabi_lasr", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Integer division functions
+      // RTABI chapter 4.3.1
+      { RTLIB::SDIV_I8,  "__aeabi_idiv",     CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SDIV_I16, "__aeabi_idiv",     CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SDIV_I32, "__aeabi_idiv",     CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::SDIV_I64, "__aeabi_ldivmod",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UDIV_I8,  "__aeabi_uidiv",    CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UDIV_I16, "__aeabi_uidiv",    CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UDIV_I32, "__aeabi_uidiv",    CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::UDIV_I64, "__aeabi_uldivmod", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+
+      // Memory operations
+      // RTABI chapter 4.3.4
+      { RTLIB::MEMCPY,  "__aeabi_memcpy",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::MEMMOVE, "__aeabi_memmove", CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+      { RTLIB::MEMSET,  "__aeabi_memset",  CallingConv::ARM_AAPCS, ISD::SETCC_INVALID },
+    };
+
+    for (const auto &LC : LibraryCalls) {
+      setLibcallName(LC.Op, LC.Name);
+      setLibcallCallingConv(LC.Op, LC.CC);
+      if (LC.Cond != ISD::SETCC_INVALID)
+        setCmpLibcallCC(LC.Op, LC.Cond);
+    }
+  }
+
+  if (Subtarget->isTargetWindows()) {
+    static const struct {
+      const RTLIB::Libcall Op;
+      const char * const Name;
+      const CallingConv::ID CC;
+    } LibraryCalls[] = {
+      { RTLIB::FPTOSINT_F32_I64, "__stoi64", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::FPTOSINT_F64_I64, "__dtoi64", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::FPTOUINT_F32_I64, "__stou64", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::FPTOUINT_F64_I64, "__dtou64", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::SINTTOFP_I64_F32, "__i64tos", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::SINTTOFP_I64_F64, "__i64tod", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::UINTTOFP_I64_F32, "__u64tos", CallingConv::ARM_AAPCS_VFP },
+      { RTLIB::UINTTOFP_I64_F64, "__u64tod", CallingConv::ARM_AAPCS_VFP },
+    };
+
+    for (const auto &LC : LibraryCalls) {
+      setLibcallName(LC.Op, LC.Name);
+      setLibcallCallingConv(LC.Op, LC.CC);
+    }
   }
 
   // Use divmod compiler-rt calls for iOS 5.0 and later.
@@ -438,8 +396,6 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
     addRegisterClass(MVT::f32, &ARM::SPRRegClass);
     if (!Subtarget->isFPOnlySP())
       addRegisterClass(MVT::f64, &ARM::DPRRegClass);
-
-    setTruncStoreAction(MVT::f64, MVT::f32, Expand);
   }
 
   for (unsigned VT = (unsigned)MVT::FIRST_VECTOR_VALUETYPE;
@@ -451,6 +407,13 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
     setLoadExtAction(ISD::SEXTLOAD, (MVT::SimpleValueType)VT, Expand);
     setLoadExtAction(ISD::ZEXTLOAD, (MVT::SimpleValueType)VT, Expand);
     setLoadExtAction(ISD::EXTLOAD, (MVT::SimpleValueType)VT, Expand);
+
+    setOperationAction(ISD::MULHS, (MVT::SimpleValueType)VT, Expand);
+    setOperationAction(ISD::SMUL_LOHI, (MVT::SimpleValueType)VT, Expand);
+    setOperationAction(ISD::MULHU, (MVT::SimpleValueType)VT, Expand);
+    setOperationAction(ISD::UMUL_LOHI, (MVT::SimpleValueType)VT, Expand);
+
+    setOperationAction(ISD::BSWAP, (MVT::SimpleValueType)VT, Expand);
   }
 
   setOperationAction(ISD::ConstantFP, MVT::f32, Custom);
@@ -617,8 +580,14 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
 
   computeRegisterProperties();
 
-  // ARM does not have f32 extending load.
+  // ARM does not have floating-point extending loads.
   setLoadExtAction(ISD::EXTLOAD, MVT::f32, Expand);
+  setLoadExtAction(ISD::EXTLOAD, MVT::f16, Expand);
+
+  // ... or truncating stores
+  setTruncStoreAction(MVT::f64, MVT::f32, Expand);
+  setTruncStoreAction(MVT::f32, MVT::f16, Expand);
+  setTruncStoreAction(MVT::f64, MVT::f16, Expand);
 
   // ARM does not have i1 sign extending load.
   setLoadExtAction(ISD::SEXTLOAD, MVT::i1, Promote);
@@ -638,6 +607,11 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
     }
   }
 
+  setOperationAction(ISD::SADDO, MVT::i32, Custom);
+  setOperationAction(ISD::UADDO, MVT::i32, Custom);
+  setOperationAction(ISD::SSUBO, MVT::i32, Custom);
+  setOperationAction(ISD::USUBO, MVT::i32, Custom);
+
   // i64 operation support.
   setOperationAction(ISD::MUL,     MVT::i64, Expand);
   setOperationAction(ISD::MULHU,   MVT::i32, Expand);
@@ -733,39 +707,31 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
   setOperationAction(ISD::STACKSAVE,          MVT::Other, Expand);
   setOperationAction(ISD::STACKRESTORE,       MVT::Other, Expand);
 
-  if (!Subtarget->isTargetDarwin()) {
-    // Non-Darwin platforms may return values in these registers via the
+  if (!Subtarget->isTargetMachO()) {
+    // Non-MachO platforms may return values in these registers via the
     // personality function.
     setExceptionPointerRegister(ARM::R0);
     setExceptionSelectorRegister(ARM::R1);
   }
 
-  setOperationAction(ISD::DYNAMIC_STACKALLOC, MVT::i32, Expand);
+  if (Subtarget->getTargetTriple().isWindowsItaniumEnvironment())
+    setOperationAction(ISD::DYNAMIC_STACKALLOC, MVT::i32, Custom);
+  else
+    setOperationAction(ISD::DYNAMIC_STACKALLOC, MVT::i32, Expand);
+
   // ARMv6 Thumb1 (except for CPUs that support dmb / dsb) and earlier use
   // the default expansion.
   if (Subtarget->hasAnyDataBarrier() && !Subtarget->isThumb1Only()) {
-    // ATOMIC_FENCE needs custom lowering; the other 32-bit ones are legal and
-    // handled normally.
+    // ATOMIC_FENCE needs custom lowering; the others should have been expanded
+    // to ldrex/strex loops already.
     setOperationAction(ISD::ATOMIC_FENCE,     MVT::Other, Custom);
-    // Custom lowering for 64-bit ops
-    setOperationAction(ISD::ATOMIC_LOAD_ADD,  MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_SUB,  MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_AND,  MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_OR,   MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_XOR,  MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_SWAP,      MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_MIN,  MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_MAX,  MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_UMIN, MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_LOAD_UMAX, MVT::i64, Custom);
-    setOperationAction(ISD::ATOMIC_CMP_SWAP,  MVT::i64, Custom);
+
     // On v8, we have particularly efficient implementations of atomic fences
     // if they can be combined with nearby atomic loads and stores.
     if (!Subtarget->hasV8Ops()) {
       // Automatically insert fences (dmb ist) around ATOMIC_SWAP etc.
       setInsertFencesForAtomic(true);
     }
-    setOperationAction(ISD::ATOMIC_LOAD, MVT::i64, Custom);
   } else {
     // If there's anything we can use as a barrier, go through custom lowering
     // for ATOMIC_FENCE.
@@ -863,13 +829,20 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
       setOperationAction(ISD::FP_TO_UINT, MVT::i32, Custom);
       setOperationAction(ISD::FP_TO_SINT, MVT::i32, Custom);
     }
-    // Special handling for half-precision FP.
+
+    // v8 adds f64 <-> f16 conversion. Before that it should be expanded.
+    if (!Subtarget->hasV8Ops()) {
+      setOperationAction(ISD::FP16_TO_FP, MVT::f64, Expand);
+      setOperationAction(ISD::FP_TO_FP16, MVT::f64, Expand);
+    }
+
+    // fp16 is a special v7 extension that adds f16 <-> f32 conversions.
     if (!Subtarget->hasFP16()) {
-      setOperationAction(ISD::FP16_TO_FP32, MVT::f32, Expand);
-      setOperationAction(ISD::FP32_TO_FP16, MVT::i32, Expand);
+      setOperationAction(ISD::FP16_TO_FP, MVT::f32, Expand);
+      setOperationAction(ISD::FP_TO_FP16, MVT::f32, Expand);
     }
   }
-      
+
   // Combine sin / cos into one node or libcall if possible.
   if (Subtarget->hasSinCos()) {
     setLibcallName(RTLIB::SINCOS_F32, "sincosf");
@@ -920,44 +893,6 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
   setMinFunctionAlignment(Subtarget->isThumb() ? 1 : 2);
 }
 
-static void getExclusiveOperation(unsigned Size, AtomicOrdering Ord,
-                                  bool isThumb2, unsigned &LdrOpc,
-                                  unsigned &StrOpc) {
-  static const unsigned LoadBares[4][2] =  {{ARM::LDREXB, ARM::t2LDREXB},
-                                            {ARM::LDREXH, ARM::t2LDREXH},
-                                            {ARM::LDREX,  ARM::t2LDREX},
-                                            {ARM::LDREXD, ARM::t2LDREXD}};
-  static const unsigned LoadAcqs[4][2] =   {{ARM::LDAEXB, ARM::t2LDAEXB},
-                                            {ARM::LDAEXH, ARM::t2LDAEXH},
-                                            {ARM::LDAEX,  ARM::t2LDAEX},
-                                            {ARM::LDAEXD, ARM::t2LDAEXD}};
-  static const unsigned StoreBares[4][2] = {{ARM::STREXB, ARM::t2STREXB},
-                                            {ARM::STREXH, ARM::t2STREXH},
-                                            {ARM::STREX,  ARM::t2STREX},
-                                            {ARM::STREXD, ARM::t2STREXD}};
-  static const unsigned StoreRels[4][2] =  {{ARM::STLEXB, ARM::t2STLEXB},
-                                            {ARM::STLEXH, ARM::t2STLEXH},
-                                            {ARM::STLEX,  ARM::t2STLEX},
-                                            {ARM::STLEXD, ARM::t2STLEXD}};
-
-  const unsigned (*LoadOps)[2], (*StoreOps)[2];
-  if (Ord == Acquire || Ord == AcquireRelease || Ord == SequentiallyConsistent)
-    LoadOps = LoadAcqs;
-  else
-    LoadOps = LoadBares;
-
-  if (Ord == Release || Ord == AcquireRelease || Ord == SequentiallyConsistent)
-    StoreOps = StoreRels;
-  else
-    StoreOps = StoreBares;
-
-  assert(isPowerOf2_32(Size) && Size <= 8 &&
-         "unsupported size for atomic binary op!");
-
-  LdrOpc = LoadOps[Log2_32(Size)][isThumb2];
-  StrOpc = StoreOps[Log2_32(Size)][isThumb2];
-}
-
 // FIXME: It might make sense to define the representative register class as the
 // nearest super-register that has a non-null superset. For example, DPR_VFP2 is
 // a super-register of SPR, and DPR is a superset if DPR_VFP2. Consequently,
@@ -970,7 +905,7 @@ static void getExclusiveOperation(unsigned Size, AtomicOrdering Ord,
 // and extractions.
 std::pair<const TargetRegisterClass*, uint8_t>
 ARMTargetLowering::findRepresentativeClass(MVT VT) const{
-  const TargetRegisterClass *RRC = 0;
+  const TargetRegisterClass *RRC = nullptr;
   uint8_t Cost = 1;
   switch (VT.SimpleTy) {
   default:
@@ -1007,9 +942,8 @@ ARMTargetLowering::findRepresentativeClass(MVT VT) const{
 
 const char *ARMTargetLowering::getTargetNodeName(unsigned Opcode) const {
   switch (Opcode) {
-  default: return 0;
+  default: return nullptr;
   case ARMISD::Wrapper:       return "ARMISD::Wrapper";
-  case ARMISD::WrapperDYN:    return "ARMISD::WrapperDYN";
   case ARMISD::WrapperPIC:    return "ARMISD::WrapperPIC";
   case ARMISD::WrapperJT:     return "ARMISD::WrapperJT";
   case ARMISD::CALL:          return "ARMISD::CALL";
@@ -1064,6 +998,8 @@ const char *ARMTargetLowering::getTargetNodeName(unsigned Opcode) const {
 
   case ARMISD::PRELOAD:       return "ARMISD::PRELOAD";
 
+  case ARMISD::WIN__CHKSTK:   return "ARMISD:::WIN__CHKSTK";
+
   case ARMISD::VCEQ:          return "ARMISD::VCEQ";
   case ARMISD::VCEQZ:         return "ARMISD::VCEQZ";
   case ARMISD::VCGE:          return "ARMISD::VCGE";
@@ -1079,10 +1015,6 @@ const char *ARMTargetLowering::getTargetNodeName(unsigned Opcode) const {
   case ARMISD::VSHL:          return "ARMISD::VSHL";
   case ARMISD::VSHRs:         return "ARMISD::VSHRs";
   case ARMISD::VSHRu:         return "ARMISD::VSHRu";
-  case ARMISD::VSHLLs:        return "ARMISD::VSHLLs";
-  case ARMISD::VSHLLu:        return "ARMISD::VSHLLu";
-  case ARMISD::VSHLLi:        return "ARMISD::VSHLLi";
-  case ARMISD::VSHRN:         return "ARMISD::VSHRN";
   case ARMISD::VRSHRs:        return "ARMISD::VRSHRs";
   case ARMISD::VRSHRu:        return "ARMISD::VRSHRu";
   case ARMISD::VRSHRN:        return "ARMISD::VRSHRN";
@@ -1266,40 +1198,58 @@ static void FPCCToARMCC(ISD::CondCode CC, ARMCC::CondCodes &CondCode,
 
 #include "ARMGenCallingConv.inc"
 
-/// CCAssignFnForNode - Selects the correct CCAssignFn for a the
-/// given CallingConvention value.
-CCAssignFn *ARMTargetLowering::CCAssignFnForNode(CallingConv::ID CC,
-                                                 bool Return,
-                                                 bool isVarArg) const {
+/// getEffectiveCallingConv - Get the effective calling convention, taking into
+/// account presence of floating point hardware and calling convention
+/// limitations, such as support for variadic functions.
+CallingConv::ID
+ARMTargetLowering::getEffectiveCallingConv(CallingConv::ID CC,
+                                           bool isVarArg) const {
   switch (CC) {
   default:
     llvm_unreachable("Unsupported calling convention");
-  case CallingConv::Fast:
-    if (Subtarget->hasVFP2() && !isVarArg) {
-      if (!Subtarget->isAAPCS_ABI())
-        return (Return ? RetFastCC_ARM_APCS : FastCC_ARM_APCS);
-      // For AAPCS ABI targets, just use VFP variant of the calling convention.
-      return (Return ? RetCC_ARM_AAPCS_VFP : CC_ARM_AAPCS_VFP);
-    }
-    // Fallthrough
-  case CallingConv::C: {
-    // Use target triple & subtarget features to do actual dispatch.
+  case CallingConv::ARM_AAPCS:
+  case CallingConv::ARM_APCS:
+  case CallingConv::GHC:
+    return CC;
+  case CallingConv::ARM_AAPCS_VFP:
+    return isVarArg ? CallingConv::ARM_AAPCS : CallingConv::ARM_AAPCS_VFP;
+  case CallingConv::C:
     if (!Subtarget->isAAPCS_ABI())
-      return (Return ? RetCC_ARM_APCS : CC_ARM_APCS);
-    else if (Subtarget->hasVFP2() &&
+      return CallingConv::ARM_APCS;
+    else if (Subtarget->hasVFP2() && !Subtarget->isThumb1Only() &&
              getTargetMachine().Options.FloatABIType == FloatABI::Hard &&
              !isVarArg)
-      return (Return ? RetCC_ARM_AAPCS_VFP : CC_ARM_AAPCS_VFP);
-    return (Return ? RetCC_ARM_AAPCS : CC_ARM_AAPCS);
+      return CallingConv::ARM_AAPCS_VFP;
+    else
+      return CallingConv::ARM_AAPCS;
+  case CallingConv::Fast:
+    if (!Subtarget->isAAPCS_ABI()) {
+      if (Subtarget->hasVFP2() && !Subtarget->isThumb1Only() && !isVarArg)
+        return CallingConv::Fast;
+      return CallingConv::ARM_APCS;
+    } else if (Subtarget->hasVFP2() && !Subtarget->isThumb1Only() && !isVarArg)
+      return CallingConv::ARM_AAPCS_VFP;
+    else
+      return CallingConv::ARM_AAPCS;
   }
-  case CallingConv::ARM_AAPCS_VFP:
-    if (!isVarArg)
-      return (Return ? RetCC_ARM_AAPCS_VFP : CC_ARM_AAPCS_VFP);
-    // Fallthrough
-  case CallingConv::ARM_AAPCS:
-    return (Return ? RetCC_ARM_AAPCS : CC_ARM_AAPCS);
+}
+
+/// CCAssignFnForNode - Selects the correct CCAssignFn for the given
+/// CallingConvention.
+CCAssignFn *ARMTargetLowering::CCAssignFnForNode(CallingConv::ID CC,
+                                                 bool Return,
+                                                 bool isVarArg) const {
+  switch (getEffectiveCallingConv(CC, isVarArg)) {
+  default:
+    llvm_unreachable("Unsupported calling convention");
   case CallingConv::ARM_APCS:
     return (Return ? RetCC_ARM_APCS : CC_ARM_APCS);
+  case CallingConv::ARM_AAPCS:
+    return (Return ? RetCC_ARM_AAPCS : CC_ARM_AAPCS);
+  case CallingConv::ARM_AAPCS_VFP:
+    return (Return ? RetCC_ARM_AAPCS_VFP : CC_ARM_AAPCS_VFP);
+  case CallingConv::Fast:
+    return (Return ? RetFastCC_ARM_APCS : FastCC_ARM_APCS);
   case CallingConv::GHC:
     return (Return ? RetCC_ARM_APCS : CC_ARM_APCS_GHC);
   }
@@ -1348,6 +1298,8 @@ ARMTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag,
                                       InFlag);
       Chain = Hi.getValue(1);
       InFlag = Hi.getValue(2);
+      if (!Subtarget->isLittle())
+        std::swap (Lo, Hi);
       Val = DAG.getNode(ARMISD::VMOVDRR, dl, MVT::f64, Lo, Hi);
 
       if (VA.getLocVT() == MVT::v2f64) {
@@ -1363,6 +1315,8 @@ ARMTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag,
         Hi = DAG.getCopyFromReg(Chain, dl, VA.getLocReg(), MVT::i32, InFlag);
         Chain = Hi.getValue(1);
         InFlag = Hi.getValue(2);
+        if (!Subtarget->isLittle())
+          std::swap (Lo, Hi);
         Val = DAG.getNode(ARMISD::VMOVDRR, dl, MVT::f64, Lo, Hi);
         Val = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, MVT::v2f64, Vec, Val,
                           DAG.getConstant(1, MVT::i32));
@@ -1413,16 +1367,17 @@ void ARMTargetLowering::PassF64ArgInRegs(SDLoc dl, SelectionDAG &DAG,
 
   SDValue fmrrd = DAG.getNode(ARMISD::VMOVRRD, dl,
                               DAG.getVTList(MVT::i32, MVT::i32), Arg);
-  RegsToPass.push_back(std::make_pair(VA.getLocReg(), fmrrd));
+  unsigned id = Subtarget->isLittle() ? 0 : 1;
+  RegsToPass.push_back(std::make_pair(VA.getLocReg(), fmrrd.getValue(id)));
 
   if (NextVA.isRegLoc())
-    RegsToPass.push_back(std::make_pair(NextVA.getLocReg(), fmrrd.getValue(1)));
+    RegsToPass.push_back(std::make_pair(NextVA.getLocReg(), fmrrd.getValue(1-id)));
   else {
     assert(NextVA.isMemLoc());
-    if (StackPtr.getNode() == 0)
+    if (!StackPtr.getNode())
       StackPtr = DAG.getCopyFromReg(Chain, dl, ARM::SP, getPointerTy());
 
-    MemOpChains.push_back(LowerMemOpCallTo(Chain, StackPtr, fmrrd.getValue(1),
+    MemOpChains.push_back(LowerMemOpCallTo(Chain, StackPtr, fmrrd.getValue(1-id),
                                            dl, DAG, NextVA,
                                            Flags));
   }
@@ -1450,14 +1405,19 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   bool isStructRet    = (Outs.empty()) ? false : Outs[0].Flags.isSRet();
   bool isThisReturn   = false;
   bool isSibCall      = false;
+
   // Disable tail calls if they're not supported.
-  if (!EnableARMTailCalls && !Subtarget->supportsTailCall())
+  if (!Subtarget->supportsTailCall() || MF.getTarget().Options.DisableTailCalls)
     isTailCall = false;
+
   if (isTailCall) {
     // Check if it's really possible to do a tail call.
     isTailCall = IsEligibleForTailCallOptimization(Callee, CallConv,
                     isVarArg, isStructRet, MF.getFunction()->hasStructRetAttr(),
                                                    Outs, OutVals, Ins, DAG);
+    if (!isTailCall && CLI.CS && CLI.CS->isMustTailCall())
+      report_fatal_error("failed to perform tail call elimination on a call "
+                         "site marked musttail");
     // We don't support GuaranteedTailCallOpt for ARM, only automatically
     // detected sibcalls.
     if (isTailCall) {
@@ -1602,7 +1562,7 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
         SDVTList VTs = DAG.getVTList(MVT::Other, MVT::Glue);
         SDValue Ops[] = { Chain, Dst, Src, SizeNode, AlignNode};
         MemOpChains.push_back(DAG.getNode(ARMISD::COPY_STRUCT_BYVAL, dl, VTs,
-                                          Ops, array_lengthof(Ops)));
+                                          Ops));
       }
     } else if (!isSibCall) {
       assert(VA.isMemLoc());
@@ -1613,8 +1573,7 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   }
 
   if (!MemOpChains.empty())
-    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
-                        &MemOpChains[0], MemOpChains.size());
+    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, MemOpChains);
 
   // Build a sequence of copy-to-reg nodes chained together with token chain
   // and flag operands which copy the outgoing args into the appropriate regs.
@@ -1656,8 +1615,9 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
 
   if (EnableARMLongCalls) {
-    assert (getTargetMachine().getRelocationModel() == Reloc::Static
-            && "long-calls with non-static relocation model!");
+    assert((Subtarget->isTargetWindows() ||
+            getTargetMachine().getRelocationModel() == Reloc::Static) &&
+           "long-calls with non-static relocation model!");
     // Handle a global address or an external symbol. If it's not one of
     // those, the target's already in a register, so we don't need to do
     // anything extra.
@@ -1695,25 +1655,29 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
     const GlobalValue *GV = G->getGlobal();
     isDirect = true;
     bool isExt = GV->isDeclaration() || GV->isWeakForLinker();
-    bool isStub = (isExt && Subtarget->isTargetDarwin()) &&
+    bool isStub = (isExt && Subtarget->isTargetMachO()) &&
                    getTargetMachine().getRelocationModel() != Reloc::Static;
     isARMFunc = !Subtarget->isThumb() || isStub;
     // ARM call to a local ARM function is predicable.
     isLocalARMFunc = !Subtarget->isThumb() && (!isExt || !ARMInterworking);
     // tBX takes a register source operand.
-    if (isARMFunc && Subtarget->isThumb1Only() && !Subtarget->hasV5TOps()) {
-      unsigned ARMPCLabelIndex = AFI->createPICLabelUId();
-      ARMConstantPoolValue *CPV =
-        ARMConstantPoolConstant::Create(GV, ARMPCLabelIndex, ARMCP::CPValue, 4);
-      SDValue CPAddr = DAG.getTargetConstantPool(CPV, getPointerTy(), 4);
-      CPAddr = DAG.getNode(ARMISD::Wrapper, dl, MVT::i32, CPAddr);
-      Callee = DAG.getLoad(getPointerTy(), dl,
-                           DAG.getEntryNode(), CPAddr,
-                           MachinePointerInfo::getConstantPool(),
-                           false, false, false, 0);
-      SDValue PICLabel = DAG.getConstant(ARMPCLabelIndex, MVT::i32);
-      Callee = DAG.getNode(ARMISD::PIC_ADD, dl,
-                           getPointerTy(), Callee, PICLabel);
+    if (isStub && Subtarget->isThumb1Only() && !Subtarget->hasV5TOps()) {
+      assert(Subtarget->isTargetMachO() && "WrapperPIC use on non-MachO?");
+      Callee = DAG.getNode(ARMISD::WrapperPIC, dl, getPointerTy(),
+                           DAG.getTargetGlobalAddress(GV, dl, getPointerTy()));
+    } else if (Subtarget->isTargetCOFF()) {
+      assert(Subtarget->isTargetWindows() &&
+             "Windows is the only supported COFF target");
+      unsigned TargetFlags = GV->hasDLLImportStorageClass()
+                                 ? ARMII::MO_DLLIMPORT
+                                 : ARMII::MO_NO_FLAG;
+      Callee = DAG.getTargetGlobalAddress(GV, dl, getPointerTy(), /*Offset=*/0,
+                                          TargetFlags);
+      if (GV->hasDLLImportStorageClass())
+        Callee = DAG.getLoad(getPointerTy(), dl, DAG.getEntryNode(),
+                             DAG.getNode(ARMISD::Wrapper, dl, getPointerTy(),
+                                         Callee), MachinePointerInfo::getGOT(),
+                             false, false, false, 0);
     } else {
       // On ELF targets for PIC code, direct calls should go through the PLT
       unsigned OpFlags = 0;
@@ -1724,7 +1688,7 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
     }
   } else if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(Callee)) {
     isDirect = true;
-    bool isStub = Subtarget->isTargetDarwin() &&
+    bool isStub = Subtarget->isTargetMachO() &&
                   getTargetMachine().getRelocationModel() != Reloc::Static;
     isARMFunc = !Subtarget->isThumb() || isStub;
     // tBX takes a register source operand.
@@ -1755,8 +1719,8 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
 
   // FIXME: handle tail calls differently.
   unsigned CallOpc;
-  bool HasMinSizeAttr = MF.getFunction()->getAttributes().
-    hasAttribute(AttributeSet::FunctionIndex, Attribute::MinSize);
+  bool HasMinSizeAttr = MF.getFunction()->getAttributes().hasAttribute(
+      AttributeSet::FunctionIndex, Attribute::MinSize);
   if (Subtarget->isThumb()) {
     if ((!isDirect || isARMFunc) && !Subtarget->hasV5TOps())
       CallOpc = ARMISD::CALL_NOLINK;
@@ -1811,10 +1775,10 @@ ARMTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
 
   SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
   if (isTailCall)
-    return DAG.getNode(ARMISD::TC_RETURN, dl, NodeTys, &Ops[0], Ops.size());
+    return DAG.getNode(ARMISD::TC_RETURN, dl, NodeTys, Ops);
 
   // Returns a chain and a flag for retval copy to use.
-  Chain = DAG.getNode(CallOpc, dl, NodeTys, &Ops[0], Ops.size());
+  Chain = DAG.getNode(CallOpc, dl, NodeTys, Ops);
   InFlag = Chain.getValue(1);
 
   Chain = DAG.getCALLSEQ_END(Chain, DAG.getIntPtrConstant(NumBytes, true),
@@ -1841,22 +1805,6 @@ ARMTargetLowering::HandleByVal(
           State->getCallOrPrologue() == Call) &&
          "unhandled ParmContext");
 
-  // For in-prologue parameters handling, we also introduce stack offset
-  // for byval registers: see CallingConvLower.cpp, CCState::HandleByVal.
-  // This behaviour outsides AAPCS rules (5.5 Parameters Passing) of how
-  // NSAA should be evaluted (NSAA means "next stacked argument address").
-  // So: NextStackOffset = NSAAOffset + SizeOfByValParamsStoredInRegs.
-  // Then: NSAAOffset = NextStackOffset - SizeOfByValParamsStoredInRegs.
-  unsigned NSAAOffset = State->getNextStackOffset();
-  if (State->getCallOrPrologue() != Call) {
-    for (unsigned i = 0, e = State->getInRegsParamsCount(); i != e; ++i) {
-      unsigned RB, RE;
-      State->getInRegsParamInfo(i, RB, RE);
-      assert(NSAAOffset >= (RE-RB)*4 &&
-             "Stack offset for byval regs doesn't introduced anymore?");
-      NSAAOffset -= (RE-RB)*4;
-    }
-  }
   if ((ARM::R0 <= reg) && (reg <= ARM::R3)) {
     if (Subtarget->isAAPCS_ABI() && Align > 4) {
       unsigned AlignInRegs = Align / 4;
@@ -1871,6 +1819,7 @@ ARMTargetLowering::HandleByVal(
       // all remained GPR regs. In that case we can't split parameter, we must
       // send it to stack. We also must set NCRN to R4, so waste all
       // remained registers.
+      const unsigned NSAAOffset = State->getNextStackOffset();
       if (Subtarget->isAAPCS_ABI() && NSAAOffset != 0 && size > excess) {
         while (State->AllocateReg(GPRArgRegs, 4))
           ;
@@ -1890,18 +1839,14 @@ ARMTargetLowering::HandleByVal(
       // allocate remained amount of registers we need.
       for (unsigned i = reg+1; i != ByValRegEnd; ++i)
         State->AllocateReg(GPRArgRegs, 4);
-      // At a call site, a byval parameter that is split between
-      // registers and memory needs its size truncated here.  In a
-      // function prologue, such byval parameters are reassembled in
-      // memory, and are not truncated.
-      if (State->getCallOrPrologue() == Call) {
-        // Make remained size equal to 0 in case, when
-        // the whole structure may be stored into registers.
-        if (size < excess)
-          size = 0;
-        else
-          size -= excess;
-      }
+      // A byval parameter that is split between registers and memory needs its
+      // size truncated here.
+      // In the case where the entire structure fits in registers, we set the
+      // size in memory to zero.
+      if (size < excess)
+        size = 0;
+      else
+        size -= excess;
     }
   }
 }
@@ -2138,8 +2083,7 @@ static SDValue LowerInterruptReturn(SmallVectorImpl<SDValue> &RetOps,
 
   RetOps.insert(RetOps.begin() + 1, DAG.getConstant(LROffset, MVT::i32, false));
 
-  return DAG.getNode(ARMISD::INTRET_FLAG, DL, MVT::Other,
-                     RetOps.data(), RetOps.size());
+  return DAG.getNode(ARMISD::INTRET_FLAG, DL, MVT::Other, RetOps);
 }
 
 SDValue
@@ -2163,6 +2107,7 @@ ARMTargetLowering::LowerReturn(SDValue Chain,
   SDValue Flag;
   SmallVector<SDValue, 4> RetOps;
   RetOps.push_back(Chain); // Operand #0 = Chain (updated below)
+  bool isLittleEndian = Subtarget->isLittle();
 
   // Copy the result values into the output registers.
   for (unsigned i = 0, realRVLocIdx = 0;
@@ -2189,12 +2134,15 @@ ARMTargetLowering::LowerReturn(SDValue Chain,
         SDValue HalfGPRs = DAG.getNode(ARMISD::VMOVRRD, dl,
                                        DAG.getVTList(MVT::i32, MVT::i32), Half);
 
-        Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(), HalfGPRs, Flag);
+        Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(),
+                                 HalfGPRs.getValue(isLittleEndian ? 0 : 1),
+                                 Flag);
         Flag = Chain.getValue(1);
         RetOps.push_back(DAG.getRegister(VA.getLocReg(), VA.getLocVT()));
         VA = RVLocs[++i]; // skip ahead to next loc
         Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(),
-                                 HalfGPRs.getValue(1), Flag);
+                                 HalfGPRs.getValue(isLittleEndian ? 1 : 0),
+                                 Flag);
         Flag = Chain.getValue(1);
         RetOps.push_back(DAG.getRegister(VA.getLocReg(), VA.getLocVT()));
         VA = RVLocs[++i]; // skip ahead to next loc
@@ -2206,12 +2154,15 @@ ARMTargetLowering::LowerReturn(SDValue Chain,
       // Legalize ret f64 -> ret 2 x i32.  We always have fmrrd if f64 is
       // available.
       SDValue fmrrd = DAG.getNode(ARMISD::VMOVRRD, dl,
-                                  DAG.getVTList(MVT::i32, MVT::i32), &Arg, 1);
-      Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(), fmrrd, Flag);
+                                  DAG.getVTList(MVT::i32, MVT::i32), Arg);
+      Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(),
+                               fmrrd.getValue(isLittleEndian ? 0 : 1),
+                               Flag);
       Flag = Chain.getValue(1);
       RetOps.push_back(DAG.getRegister(VA.getLocReg(), VA.getLocVT()));
       VA = RVLocs[++i]; // skip ahead to next loc
-      Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(), fmrrd.getValue(1),
+      Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(),
+                               fmrrd.getValue(isLittleEndian ? 1 : 0),
                                Flag);
     } else
       Chain = DAG.getCopyToReg(Chain, dl, VA.getLocReg(), Arg, Flag);
@@ -2240,8 +2191,7 @@ ARMTargetLowering::LowerReturn(SDValue Chain,
     return LowerInterruptReturn(RetOps, dl, DAG);
   }
 
-  return DAG.getNode(ARMISD::RET_FLAG, dl, MVT::Other,
-                     RetOps.data(), RetOps.size());
+  return DAG.getNode(ARMISD::RET_FLAG, dl, MVT::Other, RetOps);
 }
 
 bool ARMTargetLowering::isUsedByReturnOnly(SDNode *N, SDValue &Chain) const {
@@ -2310,10 +2260,10 @@ bool ARMTargetLowering::isUsedByReturnOnly(SDNode *N, SDValue &Chain) const {
 }
 
 bool ARMTargetLowering::mayBeEmittedAsTailCall(CallInst *CI) const {
-  if (!EnableARMTailCalls && !Subtarget->supportsTailCall())
+  if (!Subtarget->supportsTailCall())
     return false;
 
-  if (!CI->isTailCall())
+  if (!CI->isTailCall() || getTargetMachine().Options.DisableTailCalls)
     return false;
 
   return !Subtarget->isThumb1Only();
@@ -2403,13 +2353,14 @@ ARMTargetLowering::LowerToTLSGeneralDynamicModel(GlobalAddressSDNode *GA,
   Entry.Node = Argument;
   Entry.Ty = (Type *) Type::getInt32Ty(*DAG.getContext());
   Args.push_back(Entry);
+
   // FIXME: is there useful debug info available here?
-  TargetLowering::CallLoweringInfo CLI(Chain,
-                (Type *) Type::getInt32Ty(*DAG.getContext()),
-                false, false, false, false,
-                0, CallingConv::C, /*isTailCall=*/false,
-                /*doesNotRet=*/false, /*isReturnValueUsed=*/true,
-                DAG.getExternalSymbol("__tls_get_addr", PtrVT), Args, DAG, dl);
+  TargetLowering::CallLoweringInfo CLI(DAG);
+  CLI.setDebugLoc(dl).setChain(Chain)
+    .setCallee(CallingConv::C, Type::getInt32Ty(*DAG.getContext()),
+               DAG.getExternalSymbol("__tls_get_addr", PtrVT), std::move(Args),
+               0);
+
   std::pair<SDValue, SDValue> CallResult = LowerCallTo(CLI);
   return CallResult.first;
 }
@@ -2516,7 +2467,7 @@ SDValue ARMTargetLowering::LowerGlobalAddressELF(SDValue Op,
 
   // If we have T2 ops, we can materialize the address directly via movt/movw
   // pair. This is always cheaper.
-  if (Subtarget->useMovt()) {
+  if (Subtarget->useMovt(DAG.getMachineFunction())) {
     ++NumMovwMovt;
     // FIXME: Once remat is capable of dealing with instructions with register
     // operands, expand this into two nodes.
@@ -2538,56 +2489,46 @@ SDValue ARMTargetLowering::LowerGlobalAddressDarwin(SDValue Op,
   const GlobalValue *GV = cast<GlobalAddressSDNode>(Op)->getGlobal();
   Reloc::Model RelocM = getTargetMachine().getRelocationModel();
 
-  // FIXME: Enable this for static codegen when tool issues are fixed.  Also
-  // update ARMFastISel::ARMMaterializeGV.
-  if (Subtarget->useMovt() && RelocM != Reloc::Static) {
+  if (Subtarget->useMovt(DAG.getMachineFunction()))
     ++NumMovwMovt;
-    // FIXME: Once remat is capable of dealing with instructions with register
-    // operands, expand this into two nodes.
-    if (RelocM == Reloc::Static)
-      return DAG.getNode(ARMISD::Wrapper, dl, PtrVT,
-                                 DAG.getTargetGlobalAddress(GV, dl, PtrVT));
-
-    unsigned Wrapper = (RelocM == Reloc::PIC_)
-      ? ARMISD::WrapperPIC : ARMISD::WrapperDYN;
-    SDValue Result = DAG.getNode(Wrapper, dl, PtrVT,
-                                 DAG.getTargetGlobalAddress(GV, dl, PtrVT));
-    if (Subtarget->GVIsIndirectSymbol(GV, RelocM))
-      Result = DAG.getLoad(PtrVT, dl, DAG.getEntryNode(), Result,
-                           MachinePointerInfo::getGOT(),
-                           false, false, false, 0);
-    return Result;
-  }
 
-  unsigned ARMPCLabelIndex = 0;
-  SDValue CPAddr;
-  if (RelocM == Reloc::Static) {
-    CPAddr = DAG.getTargetConstantPool(GV, PtrVT, 4);
-  } else {
-    ARMFunctionInfo *AFI = DAG.getMachineFunction().getInfo<ARMFunctionInfo>();
-    ARMPCLabelIndex = AFI->createPICLabelUId();
-    unsigned PCAdj = (RelocM != Reloc::PIC_) ? 0 : (Subtarget->isThumb()?4:8);
-    ARMConstantPoolValue *CPV =
-      ARMConstantPoolConstant::Create(GV, ARMPCLabelIndex, ARMCP::CPValue,
-                                      PCAdj);
-    CPAddr = DAG.getTargetConstantPool(CPV, PtrVT, 4);
-  }
-  CPAddr = DAG.getNode(ARMISD::Wrapper, dl, MVT::i32, CPAddr);
+  // FIXME: Once remat is capable of dealing with instructions with register
+  // operands, expand this into multiple nodes
+  unsigned Wrapper =
+      RelocM == Reloc::PIC_ ? ARMISD::WrapperPIC : ARMISD::Wrapper;
 
-  SDValue Result = DAG.getLoad(PtrVT, dl, DAG.getEntryNode(), CPAddr,
-                               MachinePointerInfo::getConstantPool(),
-                               false, false, false, 0);
-  SDValue Chain = Result.getValue(1);
-
-  if (RelocM == Reloc::PIC_) {
-    SDValue PICLabel = DAG.getConstant(ARMPCLabelIndex, MVT::i32);
-    Result = DAG.getNode(ARMISD::PIC_ADD, dl, PtrVT, Result, PICLabel);
-  }
+  SDValue G = DAG.getTargetGlobalAddress(GV, dl, PtrVT, 0, ARMII::MO_NONLAZY);
+  SDValue Result = DAG.getNode(Wrapper, dl, PtrVT, G);
 
   if (Subtarget->GVIsIndirectSymbol(GV, RelocM))
-    Result = DAG.getLoad(PtrVT, dl, Chain, Result, MachinePointerInfo::getGOT(),
-                         false, false, false, 0);
+    Result = DAG.getLoad(PtrVT, dl, DAG.getEntryNode(), Result,
+                         MachinePointerInfo::getGOT(), false, false, false, 0);
+  return Result;
+}
+
+SDValue ARMTargetLowering::LowerGlobalAddressWindows(SDValue Op,
+                                                     SelectionDAG &DAG) const {
+  assert(Subtarget->isTargetWindows() && "non-Windows COFF is not supported");
+  assert(Subtarget->useMovt(DAG.getMachineFunction()) &&
+         "Windows on ARM expects to use movw/movt");
+
+  const GlobalValue *GV = cast<GlobalAddressSDNode>(Op)->getGlobal();
+  const ARMII::TOF TargetFlags =
+    (GV->hasDLLImportStorageClass() ? ARMII::MO_DLLIMPORT : ARMII::MO_NO_FLAG);
+  EVT PtrVT = getPointerTy();
+  SDValue Result;
+  SDLoc DL(Op);
 
+  ++NumMovwMovt;
+
+  // FIXME: Once remat is capable of dealing with instructions with register
+  // operands, expand this into two nodes.
+  Result = DAG.getNode(ARMISD::Wrapper, DL, PtrVT,
+                       DAG.getTargetGlobalAddress(GV, DL, PtrVT, /*Offset=*/0,
+                                                  TargetFlags));
+  if (GV->hasDLLImportStorageClass())
+    Result = DAG.getLoad(PtrVT, DL, DAG.getEntryNode(), Result,
+                         MachinePointerInfo::getGOT(), false, false, false, 0);
   return Result;
 }
 
@@ -2636,6 +2577,11 @@ ARMTargetLowering::LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG,
   SDLoc dl(Op);
   switch (IntNo) {
   default: return SDValue();    // Don't custom lower most intrinsics.
+  case Intrinsic::arm_rbit: {
+    assert(Op.getOperand(1).getValueType() == MVT::i32 &&
+           "RBIT intrinsic must have i32 type!");
+    return DAG.getNode(ARMISD::RBIT, dl, MVT::i32, Op.getOperand(1));
+  }
   case Intrinsic::arm_thread_pointer: {
     EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy();
     return DAG.getNode(ARMISD::THREAD_POINTER, dl, PtrVT);
@@ -2779,7 +2725,8 @@ ARMTargetLowering::GetF64FormalArgument(CCValAssign &VA, CCValAssign &NextVA,
     Reg = MF.addLiveIn(NextVA.getLocReg(), RC);
     ArgValue2 = DAG.getCopyFromReg(Root, dl, Reg, MVT::i32);
   }
-
+  if (!Subtarget->isLittle())
+    std::swap (ArgValue, ArgValue2);
   return DAG.getNode(ARMISD::VMOVDRR, dl, MVT::f64, ArgValue, ArgValue2);
 }
 
@@ -2807,11 +2754,11 @@ ARMTargetLowering::computeRegArea(CCState &CCInfo, MachineFunction &MF,
   ArgRegsSize = NumGPRs * 4;
 
   // If parameter is split between stack and GPRs...
-  if (NumGPRs && Align == 8 &&
+  if (NumGPRs && Align > 4 &&
       (ArgRegsSize < ArgSize ||
         InRegsParamRecordIdx >= CCInfo.getInRegsParamsCount())) {
-    // Add padding for part of param recovered from GPRs, so
-    // its last byte must be at address K*8 - 1.
+    // Add padding for part of param recovered from GPRs.  For example,
+    // if Align == 8, its last byte must be at address K*8 - 1.
     // We need to do it, since remained (stack) part of parameter has
     // stack alignment, and we need to "attach" "GPRs head" without gaps
     // to it:
@@ -2821,8 +2768,7 @@ ARMTargetLowering::computeRegArea(CCState &CCInfo, MachineFunction &MF,
     //
     ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
     unsigned Padding =
-        ((ArgRegsSize + AFI->getArgRegsSaveSize() + Align - 1) & ~(Align-1)) -
-        (ArgRegsSize + AFI->getArgRegsSaveSize());
+        OffsetToAlignment(ArgRegsSize + AFI->getArgRegsSaveSize(), Align);
     ArgRegsSaveSize = ArgRegsSize + Padding;
   } else
     // We don't need to extend regs save size for byval parameters if they
@@ -2846,10 +2792,12 @@ ARMTargetLowering::StoreByValRegs(CCState &CCInfo, SelectionDAG &DAG,
                                   unsigned OffsetFromOrigArg,
                                   unsigned ArgOffset,
                                   unsigned ArgSize,
-                                  bool ForceMutable) const {
+                                  bool ForceMutable,
+                                  unsigned ByValStoreOffset,
+                                  unsigned TotalArgRegsSaveSize) const {
 
   // Currently, two use-cases possible:
-  // Case #1. Non var-args function, and we meet first byval parameter.
+  // Case #1. Non-var-args function, and we meet first byval parameter.
   //          Setup first unallocated register as first byval register;
   //          eat all remained registers
   //          (these two actions are performed by HandleByVal method).
@@ -2883,7 +2831,6 @@ ARMTargetLowering::StoreByValRegs(CCState &CCInfo, SelectionDAG &DAG,
   // Note: once stack area for byval/varargs registers
   // was initialized, it can't be initialized again.
   if (ArgRegsSaveSize) {
-
     unsigned Padding = ArgRegsSaveSize - ArgRegsSize;
 
     if (Padding) {
@@ -2892,11 +2839,18 @@ ARMTargetLowering::StoreByValRegs(CCState &CCInfo, SelectionDAG &DAG,
       AFI->setStoredByValParamsPadding(Padding);
     }
 
-    int FrameIndex = MFI->CreateFixedObject(
-                      ArgRegsSaveSize,
-                      Padding + ArgOffset,
-                      false);
+    int FrameIndex = MFI->CreateFixedObject(ArgRegsSaveSize,
+                                            Padding +
+                                              ByValStoreOffset -
+                                              (int64_t)TotalArgRegsSaveSize,
+                                            false);
     SDValue FIN = DAG.getFrameIndex(FrameIndex, getPointerTy());
+    if (Padding) {
+       MFI->CreateFixedObject(Padding,
+                              ArgOffset + ByValStoreOffset -
+                                (int64_t)ArgRegsSaveSize,
+                              false);
+    }
 
     SmallVector<SDValue, 4> MemOps;
     for (unsigned i = 0; firstRegToSaveIndex < lastRegToSaveIndex;
@@ -2921,13 +2875,18 @@ ARMTargetLowering::StoreByValRegs(CCState &CCInfo, SelectionDAG &DAG,
     AFI->setArgRegsSaveSize(ArgRegsSaveSize + AFI->getArgRegsSaveSize());
 
     if (!MemOps.empty())
-      Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
-                          &MemOps[0], MemOps.size());
+      Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, MemOps);
     return FrameIndex;
-  } else
+  } else {
+    if (ArgSize == 0) {
+      // We cannot allocate a zero-byte object for the first variadic argument,
+      // so just make up a size.
+      ArgSize = 4;
+    }
     // This will point to the next argument passed via stack.
     return MFI->CreateFixedObject(
-        4, AFI->getStoredByValParamsPadding() + ArgOffset, !ForceMutable);
+      ArgSize, ArgOffset, !ForceMutable);
+  }
 }
 
 // Setup stack frame, the va_list pointer will start from.
@@ -2935,6 +2894,7 @@ void
 ARMTargetLowering::VarArgStyleRegisters(CCState &CCInfo, SelectionDAG &DAG,
                                         SDLoc dl, SDValue &Chain,
                                         unsigned ArgOffset,
+                                        unsigned TotalArgRegsSaveSize,
                                         bool ForceMutable) const {
   MachineFunction &MF = DAG.getMachineFunction();
   ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
@@ -2945,8 +2905,9 @@ ARMTargetLowering::VarArgStyleRegisters(CCState &CCInfo, SelectionDAG &DAG,
   // If there is no regs to be stored, just point address after last
   // argument passed via stack.
   int FrameIndex =
-    StoreByValRegs(CCInfo, DAG, dl, Chain, 0, CCInfo.getInRegsParamsCount(),
-                   0, ArgOffset, 0, ForceMutable);
+    StoreByValRegs(CCInfo, DAG, dl, Chain, nullptr,
+                   CCInfo.getInRegsParamsCount(), 0, ArgOffset, 0, ForceMutable,
+                   0, TotalArgRegsSaveSize);
 
   AFI->setVarArgsFrameIndex(FrameIndex);
 }
@@ -2983,6 +2944,51 @@ ARMTargetLowering::LowerFormalArguments(SDValue Chain,
   // We also increase this value in case of varargs function.
   AFI->setArgRegsSaveSize(0);
 
+  unsigned ByValStoreOffset = 0;
+  unsigned TotalArgRegsSaveSize = 0;
+  unsigned ArgRegsSaveSizeMaxAlign = 4;
+
+  // Calculate the amount of stack space that we need to allocate to store
+  // byval and variadic arguments that are passed in registers.
+  // We need to know this before we allocate the first byval or variadic
+  // argument, as they will be allocated a stack slot below the CFA (Canonical
+  // Frame Address, the stack pointer at entry to the function).
+  for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
+    CCValAssign &VA = ArgLocs[i];
+    if (VA.isMemLoc()) {
+      int index = VA.getValNo();
+      if (index != lastInsIndex) {
+        ISD::ArgFlagsTy Flags = Ins[index].Flags;
+        if (Flags.isByVal()) {
+          unsigned ExtraArgRegsSize;
+          unsigned ExtraArgRegsSaveSize;
+          computeRegArea(CCInfo, MF, CCInfo.getInRegsParamsProceed(),
+                         Flags.getByValSize(),
+                         ExtraArgRegsSize, ExtraArgRegsSaveSize);
+
+          TotalArgRegsSaveSize += ExtraArgRegsSaveSize;
+          if (Flags.getByValAlign() > ArgRegsSaveSizeMaxAlign)
+              ArgRegsSaveSizeMaxAlign = Flags.getByValAlign();
+          CCInfo.nextInRegsParam();
+        }
+        lastInsIndex = index;
+      }
+    }
+  }
+  CCInfo.rewindByValRegsInfo();
+  lastInsIndex = -1;
+  if (isVarArg) {
+    unsigned ExtraArgRegsSize;
+    unsigned ExtraArgRegsSaveSize;
+    computeRegArea(CCInfo, MF, CCInfo.getInRegsParamsCount(), 0,
+                   ExtraArgRegsSize, ExtraArgRegsSaveSize);
+    TotalArgRegsSaveSize += ExtraArgRegsSaveSize;
+  }
+  // If the arg regs save area contains N-byte aligned values, the
+  // bottom of it must be at least N-byte aligned.
+  TotalArgRegsSaveSize = RoundUpToAlignment(TotalArgRegsSaveSize, ArgRegsSaveSizeMaxAlign);
+  TotalArgRegsSaveSize = std::min(TotalArgRegsSaveSize, 16U);
+
   for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
     CCValAssign &VA = ArgLocs[i];
     std::advance(CurOrigArg, Ins[VA.getValNo()].OrigArgIndex - CurArgIdx);
@@ -3081,18 +3087,23 @@ ARMTargetLowering::LowerFormalArguments(SDValue Chain,
           // a tail call.
           if (Flags.isByVal()) {
             unsigned CurByValIndex = CCInfo.getInRegsParamsProceed();
+
+            ByValStoreOffset = RoundUpToAlignment(ByValStoreOffset, Flags.getByValAlign());
             int FrameIndex = StoreByValRegs(
                 CCInfo, DAG, dl, Chain, CurOrigArg,
                 CurByValIndex,
                 Ins[VA.getValNo()].PartOffset,
                 VA.getLocMemOffset(),
                 Flags.getByValSize(),
-                true /*force mutable frames*/);
+                true /*force mutable frames*/,
+                ByValStoreOffset,
+                TotalArgRegsSaveSize);
+            ByValStoreOffset += Flags.getByValSize();
+            ByValStoreOffset = std::min(ByValStoreOffset, 16U);
             InVals.push_back(DAG.getFrameIndex(FrameIndex, getPointerTy()));
             CCInfo.nextInRegsParam();
           } else {
-            unsigned FIOffset = VA.getLocMemOffset() +
-                                AFI->getStoredByValParamsPadding();
+            unsigned FIOffset = VA.getLocMemOffset();
             int FI = MFI->CreateFixedObject(VA.getLocVT().getSizeInBits()/8,
                                             FIOffset, true);
 
@@ -3110,7 +3121,10 @@ ARMTargetLowering::LowerFormalArguments(SDValue Chain,
   // varargs
   if (isVarArg)
     VarArgStyleRegisters(CCInfo, DAG, dl, Chain,
-                         CCInfo.getNextStackOffset());
+                         CCInfo.getNextStackOffset(),
+                         TotalArgRegsSaveSize);
+
+  AFI->setArgumentStackSize(CCInfo.getNextStackOffset());
 
   return Chain;
 }
@@ -3224,11 +3238,96 @@ ARMTargetLowering::duplicateCmp(SDValue Cmp, SelectionDAG &DAG) const {
   return DAG.getNode(ARMISD::FMSTAT, DL, MVT::Glue, Cmp);
 }
 
+std::pair<SDValue, SDValue>
+ARMTargetLowering::getARMXALUOOp(SDValue Op, SelectionDAG &DAG,
+                                 SDValue &ARMcc) const {
+  assert(Op.getValueType() == MVT::i32 &&  "Unsupported value type");
+
+  SDValue Value, OverflowCmp;
+  SDValue LHS = Op.getOperand(0);
+  SDValue RHS = Op.getOperand(1);
+
+
+  // FIXME: We are currently always generating CMPs because we don't support
+  // generating CMN through the backend. This is not as good as the natural
+  // CMP case because it causes a register dependency and cannot be folded
+  // later.
+
+  switch (Op.getOpcode()) {
+  default:
+    llvm_unreachable("Unknown overflow instruction!");
+  case ISD::SADDO:
+    ARMcc = DAG.getConstant(ARMCC::VC, MVT::i32);
+    Value = DAG.getNode(ISD::ADD, SDLoc(Op), Op.getValueType(), LHS, RHS);
+    OverflowCmp = DAG.getNode(ARMISD::CMP, SDLoc(Op), MVT::Glue, Value, LHS);
+    break;
+  case ISD::UADDO:
+    ARMcc = DAG.getConstant(ARMCC::HS, MVT::i32);
+    Value = DAG.getNode(ISD::ADD, SDLoc(Op), Op.getValueType(), LHS, RHS);
+    OverflowCmp = DAG.getNode(ARMISD::CMP, SDLoc(Op), MVT::Glue, Value, LHS);
+    break;
+  case ISD::SSUBO:
+    ARMcc = DAG.getConstant(ARMCC::VC, MVT::i32);
+    Value = DAG.getNode(ISD::SUB, SDLoc(Op), Op.getValueType(), LHS, RHS);
+    OverflowCmp = DAG.getNode(ARMISD::CMP, SDLoc(Op), MVT::Glue, LHS, RHS);
+    break;
+  case ISD::USUBO:
+    ARMcc = DAG.getConstant(ARMCC::HS, MVT::i32);
+    Value = DAG.getNode(ISD::SUB, SDLoc(Op), Op.getValueType(), LHS, RHS);
+    OverflowCmp = DAG.getNode(ARMISD::CMP, SDLoc(Op), MVT::Glue, LHS, RHS);
+    break;
+  } // switch (...)
+
+  return std::make_pair(Value, OverflowCmp);
+}
+
+
+SDValue
+ARMTargetLowering::LowerXALUO(SDValue Op, SelectionDAG &DAG) const {
+  // Let legalize expand this if it isn't a legal type yet.
+  if (!DAG.getTargetLoweringInfo().isTypeLegal(Op.getValueType()))
+    return SDValue();
+
+  SDValue Value, OverflowCmp;
+  SDValue ARMcc;
+  std::tie(Value, OverflowCmp) = getARMXALUOOp(Op, DAG, ARMcc);
+  SDValue CCR = DAG.getRegister(ARM::CPSR, MVT::i32);
+  // We use 0 and 1 as false and true values.
+  SDValue TVal = DAG.getConstant(1, MVT::i32);
+  SDValue FVal = DAG.getConstant(0, MVT::i32);
+  EVT VT = Op.getValueType();
+
+  SDValue Overflow = DAG.getNode(ARMISD::CMOV, SDLoc(Op), VT, TVal, FVal,
+                                 ARMcc, CCR, OverflowCmp);
+
+  SDVTList VTs = DAG.getVTList(Op.getValueType(), MVT::i32);
+  return DAG.getNode(ISD::MERGE_VALUES, SDLoc(Op), VTs, Value, Overflow);
+}
+
+
 SDValue ARMTargetLowering::LowerSELECT(SDValue Op, SelectionDAG &DAG) const {
   SDValue Cond = Op.getOperand(0);
   SDValue SelectTrue = Op.getOperand(1);
   SDValue SelectFalse = Op.getOperand(2);
   SDLoc dl(Op);
+  unsigned Opc = Cond.getOpcode();
+
+  if (Cond.getResNo() == 1 &&
+      (Opc == ISD::SADDO || Opc == ISD::UADDO || Opc == ISD::SSUBO ||
+       Opc == ISD::USUBO)) {
+    if (!DAG.getTargetLoweringInfo().isTypeLegal(Cond->getValueType(0)))
+      return SDValue();
+
+    SDValue Value, OverflowCmp;
+    SDValue ARMcc;
+    std::tie(Value, OverflowCmp) = getARMXALUOOp(Cond, DAG, ARMcc);
+    SDValue CCR = DAG.getRegister(ARM::CPSR, MVT::i32);
+    EVT VT = Op.getValueType();
+
+    return DAG.getNode(ARMISD::CMOV, SDLoc(Op), VT, SelectTrue, SelectFalse,
+                       ARMcc, CCR, OverflowCmp);
+
+  }
 
   // Convert:
   //
@@ -3279,7 +3378,7 @@ SDValue ARMTargetLowering::LowerSELECT(SDValue Op, SelectionDAG &DAG) const {
 static ISD::CondCode getInverseCCForVSEL(ISD::CondCode CC) {
   if (CC == ISD::SETNE)
     return ISD::SETEQ;
-  return ISD::getSetCCSwappedOperands(CC);
+  return ISD::getSetCCInverse(CC, true);
 }
 
 static void checkVSELConstraints(ISD::CondCode CC, ARMCC::CondCodes &CondCode,
@@ -3530,7 +3629,7 @@ ARMTargetLowering::OptimizeVFPBrcond(SDValue Op, SelectionDAG &DAG) const {
     ARMcc = DAG.getConstant(CondCode, MVT::i32);
     SDVTList VTList = DAG.getVTList(MVT::Other, MVT::Glue);
     SDValue Ops[] = { Chain, ARMcc, LHS1, LHS2, RHS1, RHS2, Dest };
-    return DAG.getNode(ARMISD::BCC_i64, dl, VTList, Ops, 7);
+    return DAG.getNode(ARMISD::BCC_i64, dl, VTList, Ops);
   }
 
   return SDValue();
@@ -3570,11 +3669,11 @@ SDValue ARMTargetLowering::LowerBR_CC(SDValue Op, SelectionDAG &DAG) const {
   SDValue CCR = DAG.getRegister(ARM::CPSR, MVT::i32);
   SDVTList VTList = DAG.getVTList(MVT::Other, MVT::Glue);
   SDValue Ops[] = { Chain, Dest, ARMcc, CCR, Cmp };
-  SDValue Res = DAG.getNode(ARMISD::BRCOND, dl, VTList, Ops, 5);
+  SDValue Res = DAG.getNode(ARMISD::BRCOND, dl, VTList, Ops);
   if (CondCode2 != ARMCC::AL) {
     ARMcc = DAG.getConstant(CondCode2, MVT::i32);
     SDValue Ops[] = { Res, Dest, ARMcc, CCR, Res.getValue(1) };
-    Res = DAG.getNode(ARMISD::BRCOND, dl, VTList, Ops, 5);
+    Res = DAG.getNode(ARMISD::BRCOND, dl, VTList, Ops);
   }
   return Res;
 }
@@ -3771,7 +3870,7 @@ SDValue ARMTargetLowering::LowerFCOPYSIGN(SDValue Op, SelectionDAG &DAG) const {
   // Bitcast operand 1 to i32.
   if (SrcVT == MVT::f64)
     Tmp1 = DAG.getNode(ARMISD::VMOVRRD, dl, DAG.getVTList(MVT::i32, MVT::i32),
-                       &Tmp1, 1).getValue(1);
+                       Tmp1).getValue(1);
   Tmp1 = DAG.getNode(ISD::BITCAST, dl, MVT::i32, Tmp1);
 
   // Or in the signbit with integer operations.
@@ -3787,7 +3886,7 @@ SDValue ARMTargetLowering::LowerFCOPYSIGN(SDValue Op, SelectionDAG &DAG) const {
 
   // f64: Or the high part with signbit and then combine two parts.
   Tmp0 = DAG.getNode(ARMISD::VMOVRRD, dl, DAG.getVTList(MVT::i32, MVT::i32),
-                     &Tmp0, 1);
+                     Tmp0);
   SDValue Lo = Tmp0.getValue(0);
   SDValue Hi = DAG.getNode(ISD::AND, dl, MVT::i32, Tmp0.getValue(1), Mask2);
   Hi = DAG.getNode(ISD::OR, dl, MVT::i32, Hi, Tmp1);
@@ -3799,6 +3898,9 @@ SDValue ARMTargetLowering::LowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const{
   MachineFrameInfo *MFI = MF.getFrameInfo();
   MFI->setReturnAddressIsTaken(true);
 
+  if (verifyReturnAddressArgumentIsConstant(Op, DAG))
+    return SDValue();
+
   EVT VT = Op.getValueType();
   SDLoc dl(Op);
   unsigned Depth = cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue();
@@ -3816,14 +3918,16 @@ SDValue ARMTargetLowering::LowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const{
 }
 
 SDValue ARMTargetLowering::LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const {
-  MachineFrameInfo *MFI = DAG.getMachineFunction().getFrameInfo();
+  const ARMBaseRegisterInfo &ARI =
+    *static_cast<const ARMBaseRegisterInfo*>(RegInfo);
+  MachineFunction &MF = DAG.getMachineFunction();
+  MachineFrameInfo *MFI = MF.getFrameInfo();
   MFI->setFrameAddressIsTaken(true);
 
   EVT VT = Op.getValueType();
   SDLoc dl(Op);  // FIXME probably not meaningful
   unsigned Depth = cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue();
-  unsigned FrameReg = (Subtarget->isThumb() || Subtarget->isTargetDarwin())
-    ? ARM::R7 : ARM::R11;
+  unsigned FrameReg = ARI.getFrameRegister(MF);
   SDValue FrameAddr = DAG.getCopyFromReg(DAG.getEntryNode(), dl, FrameReg, VT);
   while (Depth--)
     FrameAddr = DAG.getLoad(VT, dl, DAG.getEntryNode(), FrameAddr,
@@ -3832,6 +3936,18 @@ SDValue ARMTargetLowering::LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const {
   return FrameAddr;
 }
 
+// FIXME? Maybe this could be a TableGen attribute on some registers and
+// this table could be generated automatically from RegInfo.
+unsigned ARMTargetLowering::getRegisterByName(const char* RegName,
+                                              EVT VT) const {
+  unsigned Reg = StringSwitch<unsigned>(RegName)
+                       .Case("sp", ARM::SP)
+                       .Default(0);
+  if (Reg)
+    return Reg;
+  report_fatal_error("Invalid register name global variable");
+}
+
 /// ExpandBITCAST - If the target supports VFP, this function is called to
 /// expand a bit convert where either the source or destination type is i64 to
 /// use a VMOVDRR or VMOVRRD node.  This should not be done when the non-i64
@@ -3861,8 +3977,15 @@ static SDValue ExpandBITCAST(SDNode *N, SelectionDAG &DAG) {
 
   // Turn f64->i64 into VMOVRRD.
   if (DstVT == MVT::i64 && TLI.isTypeLegal(SrcVT)) {
-    SDValue Cvt = DAG.getNode(ARMISD::VMOVRRD, dl,
-                              DAG.getVTList(MVT::i32, MVT::i32), &Op, 1);
+    SDValue Cvt;
+    if (TLI.isBigEndian() && SrcVT.isVector() &&
+        SrcVT.getVectorNumElements() > 1)
+      Cvt = DAG.getNode(ARMISD::VMOVRRD, dl,
+                        DAG.getVTList(MVT::i32, MVT::i32),
+                        DAG.getNode(ARMISD::VREV64, dl, SrcVT, Op));
+    else
+      Cvt = DAG.getNode(ARMISD::VMOVRRD, dl,
+                        DAG.getVTList(MVT::i32, MVT::i32), Op);
     // Merge the pieces into a single i64 value.
     return DAG.getNode(ISD::BUILD_PAIR, dl, MVT::i64, Cvt, Cvt.getValue(1));
   }
@@ -3918,7 +4041,7 @@ SDValue ARMTargetLowering::LowerShiftRightParts(SDValue Op,
                            CCR, Cmp);
 
   SDValue Ops[2] = { Lo, Hi };
-  return DAG.getMergeValues(Ops, 2, dl);
+  return DAG.getMergeValues(Ops, dl);
 }
 
 /// LowerShiftLeftParts - Lower SHL_PARTS, which returns two
@@ -3952,7 +4075,7 @@ SDValue ARMTargetLowering::LowerShiftLeftParts(SDValue Op,
                            CCR, Cmp);
 
   SDValue Ops[2] = { Lo, Hi };
-  return DAG.getMergeValues(Ops, 2, dl);
+  return DAG.getMergeValues(Ops, dl);
 }
 
 SDValue ARMTargetLowering::LowerFLT_ROUNDS_(SDValue Op,
@@ -4157,7 +4280,7 @@ static SDValue Expand64BitShift(SDNode *N, SelectionDAG &DAG,
   // First, build a SRA_FLAG/SRL_FLAG op, which shifts the top part by one and
   // captures the result into a carry flag.
   unsigned Opc = N->getOpcode() == ISD::SRL ? ARMISD::SRL_FLAG:ARMISD::SRA_FLAG;
-  Hi = DAG.getNode(Opc, dl, DAG.getVTList(MVT::i32, MVT::Glue), &Hi, 1);
+  Hi = DAG.getNode(Opc, dl, DAG.getVTList(MVT::i32, MVT::Glue), Hi);
 
   // The low part is an ARMISD::RRX operand, which shifts the carry in.
   Lo = DAG.getNode(ARMISD::RRX, dl, MVT::i32, Lo, Hi.getValue(1));
@@ -4380,7 +4503,6 @@ static SDValue isNEONModifiedImm(uint64_t SplatBits, uint64_t SplatUndef,
       // Value = 0x0000nnff: Op=x, Cmode=1100.
       OpCmode = 0xc;
       Imm = SplatBits >> 8;
-      SplatBits |= 0xff;
       break;
     }
 
@@ -4389,7 +4511,6 @@ static SDValue isNEONModifiedImm(uint64_t SplatBits, uint64_t SplatUndef,
       // Value = 0x00nnffff: Op=x, Cmode=1101.
       OpCmode = 0xd;
       Imm = SplatBits >> 16;
-      SplatBits |= 0xffff;
       break;
     }
 
@@ -4418,9 +4539,13 @@ static SDValue isNEONModifiedImm(uint64_t SplatBits, uint64_t SplatUndef,
       BitMask <<= 8;
       ImmMask <<= 1;
     }
+
+    if (DAG.getTargetLoweringInfo().isBigEndian())
+      // swap higher and lower 32 bit word
+      Imm = ((Imm & 0xf) << 4) | ((Imm & 0xf0) >> 4);
+
     // Op=1, Cmode=1110.
     OpCmode = 0x1e;
-    SplatBits = Val;
     VT = is128Bits ? MVT::v2i64 : MVT::v1i64;
     break;
   }
@@ -4917,7 +5042,7 @@ SDValue ARMTargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
           Ops.push_back(N);
           Ops.push_back(Op.getOperand(I));
           Ops.push_back(DAG.getConstant(I, MVT::i32));
-          N = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, VT, &Ops[0], 3);
+          N = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, VT, Ops);
         }
       }
       return N;
@@ -4928,7 +5053,7 @@ SDValue ARMTargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
         Ops.push_back(DAG.getNode(ISD::BITCAST, dl, MVT::i32,
                                   Op.getOperand(i)));
       EVT VecVT = EVT::getVectorVT(*DAG.getContext(), MVT::i32, NumElts);
-      SDValue Val = DAG.getNode(ISD::BUILD_VECTOR, dl, VecVT, &Ops[0], NumElts);
+      SDValue Val = DAG.getNode(ISD::BUILD_VECTOR, dl, VecVT, Ops);
       Val = LowerBUILD_VECTOR(Val, DAG, ST);
       if (Val.getNode())
         return DAG.getNode(ISD::BITCAST, dl, VT, Val);
@@ -4964,7 +5089,7 @@ SDValue ARMTargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
     SmallVector<SDValue, 8> Ops;
     for (unsigned i = 0; i < NumElts; ++i)
       Ops.push_back(DAG.getNode(ISD::BITCAST, dl, EltVT, Op.getOperand(i)));
-    SDValue Val = DAG.getNode(ARMISD::BUILD_VECTOR, dl, VecVT, &Ops[0],NumElts);
+    SDValue Val = DAG.getNode(ARMISD::BUILD_VECTOR, dl, VecVT, Ops);
     return DAG.getNode(ISD::BITCAST, dl, VT, Val);
   }
 
@@ -5271,12 +5396,10 @@ static SDValue LowerVECTOR_SHUFFLEv8i8(SDValue Op,
 
   if (V2.getNode()->getOpcode() == ISD::UNDEF)
     return DAG.getNode(ARMISD::VTBL1, DL, MVT::v8i8, V1,
-                       DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i8,
-                                   &VTBLMask[0], 8));
+                       DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i8, VTBLMask));
 
   return DAG.getNode(ARMISD::VTBL2, DL, MVT::v8i8, V1, V2,
-                     DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i8,
-                                 &VTBLMask[0], 8));
+                     DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i8, VTBLMask));
 }
 
 static SDValue LowerReverse_VECTOR_SHUFFLEv16i8_v8i16(SDValue Op,
@@ -5429,7 +5552,7 @@ static SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) {
                                   DAG.getConstant(ShuffleMask[i] & (NumElts-1),
                                                   MVT::i32)));
     }
-    SDValue Val = DAG.getNode(ARMISD::BUILD_VECTOR, dl, VecVT, &Ops[0],NumElts);
+    SDValue Val = DAG.getNode(ARMISD::BUILD_VECTOR, dl, VecVT, Ops);
     return DAG.getNode(ISD::BITCAST, dl, VT, Val);
   }
 
@@ -5666,7 +5789,7 @@ static SDValue SkipExtensionForVMULL(SDNode *N, SelectionDAG &DAG) {
     Ops.push_back(DAG.getConstant(CInt.zextOrTrunc(32), MVT::i32));
   }
   return DAG.getNode(ISD::BUILD_VECTOR, SDLoc(N),
-                     MVT::getVectorVT(TruncVT, NumElts), Ops.data(), NumElts);
+                     MVT::getVectorVT(TruncVT, NumElts), Ops);
 }
 
 static bool isAddSubSExt(SDNode *N, SelectionDAG &DAG) {
@@ -6004,12 +6127,12 @@ SDValue ARMTargetLowering::LowerFSINCOS(SDValue Op, SelectionDAG &DAG) const {
   ? "__sincos_stret" : "__sincosf_stret";
   SDValue Callee = DAG.getExternalSymbol(LibcallName, getPointerTy());
 
-  TargetLowering::
-  CallLoweringInfo CLI(DAG.getEntryNode(), Type::getVoidTy(*DAG.getContext()),
-                       false, false, false, false, 0,
-                       CallingConv::C, /*isTaillCall=*/false,
-                       /*doesNotRet=*/false, /*isReturnValueUsed*/false,
-                       Callee, Args, DAG, dl);
+  TargetLowering::CallLoweringInfo CLI(DAG);
+  CLI.setDebugLoc(dl).setChain(DAG.getEntryNode())
+    .setCallee(CallingConv::C, Type::getVoidTy(*DAG.getContext()), Callee,
+               std::move(Args), 0)
+    .setDiscardResult();
+
   std::pair<SDValue, SDValue> CallResult = LowerCallTo(CLI);
 
   SDValue LoadSin = DAG.getLoad(ArgVT, dl, CallResult.second, SRet,
@@ -6031,40 +6154,11 @@ static SDValue LowerAtomicLoadStore(SDValue Op, SelectionDAG &DAG) {
   if (cast<AtomicSDNode>(Op)->getOrdering() <= Monotonic)
     return Op;
 
-  // Aquire/Release load/store is not legal for targets without a
+  // Acquire/Release load/store is not legal for targets without a
   // dmb or equivalent available.
   return SDValue();
 }
 
-static void
-ReplaceATOMIC_OP_64(SDNode *Node, SmallVectorImpl<SDValue>& Results,
-                    SelectionDAG &DAG) {
-  SDLoc dl(Node);
-  assert (Node->getValueType(0) == MVT::i64 &&
-          "Only know how to expand i64 atomics");
-  AtomicSDNode *AN = cast<AtomicSDNode>(Node);
-
-  SmallVector<SDValue, 6> Ops;
-  Ops.push_back(Node->getOperand(0)); // Chain
-  Ops.push_back(Node->getOperand(1)); // Ptr
-  for(unsigned i=2; i<Node->getNumOperands(); i++) {
-    // Low part
-    Ops.push_back(DAG.getNode(ISD::EXTRACT_ELEMENT, dl, MVT::i32,
-                              Node->getOperand(i), DAG.getIntPtrConstant(0)));
-    // High part
-    Ops.push_back(DAG.getNode(ISD::EXTRACT_ELEMENT, dl, MVT::i32,
-                              Node->getOperand(i), DAG.getIntPtrConstant(1)));
-  }
-  SDVTList Tys = DAG.getVTList(MVT::i32, MVT::i32, MVT::Other);
-  SDValue Result =
-    DAG.getAtomic(Node->getOpcode(), dl, MVT::i64, Tys, Ops.data(), Ops.size(),
-                  cast<MemSDNode>(Node)->getMemOperand(), AN->getOrdering(),
-                  AN->getSynchScope());
-  SDValue OpsF[] = { Result.getValue(0), Result.getValue(1) };
-  Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, MVT::i64, OpsF, 2));
-  Results.push_back(Result.getValue(2));
-}
-
 static void ReplaceREADCYCLECOUNTER(SDNode *N,
                                     SmallVectorImpl<SDValue> &Results,
                                     SelectionDAG &DAG,
@@ -6085,8 +6179,7 @@ static void ReplaceREADCYCLECOUNTER(SDNode *N,
     };
 
     Cycles32 = DAG.getNode(ISD::INTRINSIC_W_CHAIN, DL,
-                           DAG.getVTList(MVT::i32, MVT::Other), &Ops[0],
-                           array_lengthof(Ops));
+                           DAG.getVTList(MVT::i32, MVT::Other), Ops);
     OutChain = Cycles32.getValue(1);
   } else {
     // Intrinsic is defined to return 0 on unsupported platforms. Technically
@@ -6109,8 +6202,15 @@ SDValue ARMTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::ConstantPool:  return LowerConstantPool(Op, DAG);
   case ISD::BlockAddress:  return LowerBlockAddress(Op, DAG);
   case ISD::GlobalAddress:
-    return Subtarget->isTargetDarwin() ? LowerGlobalAddressDarwin(Op, DAG) :
-      LowerGlobalAddressELF(Op, DAG);
+    switch (Subtarget->getTargetTriple().getObjectFormat()) {
+    default: llvm_unreachable("unknown object format");
+    case Triple::COFF:
+      return LowerGlobalAddressWindows(Op, DAG);
+    case Triple::ELF:
+      return LowerGlobalAddressELF(Op, DAG);
+    case Triple::MachO:
+      return LowerGlobalAddressDarwin(Op, DAG);
+    }
   case ISD::GlobalTLSAddress: return LowerGlobalTLSAddress(Op, DAG);
   case ISD::SELECT:        return LowerSELECT(Op, DAG);
   case ISD::SELECT_CC:     return LowerSELECT_CC(Op, DAG);
@@ -6155,11 +6255,20 @@ SDValue ARMTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::ADDE:
   case ISD::SUBC:
   case ISD::SUBE:          return LowerADDC_ADDE_SUBC_SUBE(Op, DAG);
+  case ISD::SADDO:
+  case ISD::UADDO:
+  case ISD::SSUBO:
+  case ISD::USUBO:
+    return LowerXALUO(Op, DAG);
   case ISD::ATOMIC_LOAD:
   case ISD::ATOMIC_STORE:  return LowerAtomicLoadStore(Op, DAG);
   case ISD::FSINCOS:       return LowerFSINCOS(Op, DAG);
   case ISD::SDIVREM:
   case ISD::UDIVREM:       return LowerDivRem(Op, DAG);
+  case ISD::DYNAMIC_STACKALLOC:
+    if (Subtarget->getTargetTriple().isWindowsItaniumEnvironment())
+      return LowerDYNAMIC_STACKALLOC(Op, DAG);
+    llvm_unreachable("Don't know how to custom lower this!");
   }
 }
 
@@ -6182,22 +6291,6 @@ void ARMTargetLowering::ReplaceNodeResults(SDNode *N,
   case ISD::READCYCLECOUNTER:
     ReplaceREADCYCLECOUNTER(N, Results, DAG, Subtarget);
     return;
-  case ISD::ATOMIC_STORE:
-  case ISD::ATOMIC_LOAD:
-  case ISD::ATOMIC_LOAD_ADD:
-  case ISD::ATOMIC_LOAD_AND:
-  case ISD::ATOMIC_LOAD_NAND:
-  case ISD::ATOMIC_LOAD_OR:
-  case ISD::ATOMIC_LOAD_SUB:
-  case ISD::ATOMIC_LOAD_XOR:
-  case ISD::ATOMIC_SWAP:
-  case ISD::ATOMIC_CMP_SWAP:
-  case ISD::ATOMIC_LOAD_MIN:
-  case ISD::ATOMIC_LOAD_UMIN:
-  case ISD::ATOMIC_LOAD_MAX:
-  case ISD::ATOMIC_LOAD_UMAX:
-    ReplaceATOMIC_OP_64(N, Results, DAG);
-    return;
   }
   if (Res.getNode())
     Results.push_back(Res);
@@ -6207,538 +6300,6 @@ void ARMTargetLowering::ReplaceNodeResults(SDNode *N,
 //                           ARM Scheduler Hooks
 //===----------------------------------------------------------------------===//
 
-MachineBasicBlock *
-ARMTargetLowering::EmitAtomicCmpSwap(MachineInstr *MI,
-                                     MachineBasicBlock *BB,
-                                     unsigned Size) const {
-  unsigned dest    = MI->getOperand(0).getReg();
-  unsigned ptr     = MI->getOperand(1).getReg();
-  unsigned oldval  = MI->getOperand(2).getReg();
-  unsigned newval  = MI->getOperand(3).getReg();
-  const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
-  AtomicOrdering Ord = static_cast<AtomicOrdering>(MI->getOperand(4).getImm());
-  DebugLoc dl = MI->getDebugLoc();
-  bool isThumb2 = Subtarget->isThumb2();
-
-  MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
-  unsigned scratch = MRI.createVirtualRegister(isThumb2 ?
-    (const TargetRegisterClass*)&ARM::rGPRRegClass :
-    (const TargetRegisterClass*)&ARM::GPRRegClass);
-
-  if (isThumb2) {
-    MRI.constrainRegClass(dest, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(oldval, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(newval, &ARM::rGPRRegClass);
-  }
-
-  unsigned ldrOpc, strOpc;
-  getExclusiveOperation(Size, Ord, isThumb2, ldrOpc, strOpc);
-
-  MachineFunction *MF = BB->getParent();
-  const BasicBlock *LLVM_BB = BB->getBasicBlock();
-  MachineFunction::iterator It = BB;
-  ++It; // insert the new blocks after the current block
-
-  MachineBasicBlock *loop1MBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MachineBasicBlock *loop2MBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MachineBasicBlock *exitMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MF->insert(It, loop1MBB);
-  MF->insert(It, loop2MBB);
-  MF->insert(It, exitMBB);
-
-  // Transfer the remainder of BB and its successor edges to exitMBB.
-  exitMBB->splice(exitMBB->begin(), BB,
-                  llvm::next(MachineBasicBlock::iterator(MI)),
-                  BB->end());
-  exitMBB->transferSuccessorsAndUpdatePHIs(BB);
-
-  //  thisMBB:
-  //   ...
-  //   fallthrough --> loop1MBB
-  BB->addSuccessor(loop1MBB);
-
-  // loop1MBB:
-  //   ldrex dest, [ptr]
-  //   cmp dest, oldval
-  //   bne exitMBB
-  BB = loop1MBB;
-  MachineInstrBuilder MIB = BuildMI(BB, dl, TII->get(ldrOpc), dest).addReg(ptr);
-  if (ldrOpc == ARM::t2LDREX)
-    MIB.addImm(0);
-  AddDefaultPred(MIB);
-  AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPrr : ARM::CMPrr))
-                 .addReg(dest).addReg(oldval));
-  BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-    .addMBB(exitMBB).addImm(ARMCC::NE).addReg(ARM::CPSR);
-  BB->addSuccessor(loop2MBB);
-  BB->addSuccessor(exitMBB);
-
-  // loop2MBB:
-  //   strex scratch, newval, [ptr]
-  //   cmp scratch, #0
-  //   bne loop1MBB
-  BB = loop2MBB;
-  MIB = BuildMI(BB, dl, TII->get(strOpc), scratch).addReg(newval).addReg(ptr);
-  if (strOpc == ARM::t2STREX)
-    MIB.addImm(0);
-  AddDefaultPred(MIB);
-  AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPri : ARM::CMPri))
-                 .addReg(scratch).addImm(0));
-  BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-    .addMBB(loop1MBB).addImm(ARMCC::NE).addReg(ARM::CPSR);
-  BB->addSuccessor(loop1MBB);
-  BB->addSuccessor(exitMBB);
-
-  //  exitMBB:
-  //   ...
-  BB = exitMBB;
-
-  MI->eraseFromParent();   // The instruction is gone now.
-
-  return BB;
-}
-
-MachineBasicBlock *
-ARMTargetLowering::EmitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
-                                    unsigned Size, unsigned BinOpcode) const {
-  // This also handles ATOMIC_SWAP, indicated by BinOpcode==0.
-  const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
-
-  const BasicBlock *LLVM_BB = BB->getBasicBlock();
-  MachineFunction *MF = BB->getParent();
-  MachineFunction::iterator It = BB;
-  ++It;
-
-  unsigned dest = MI->getOperand(0).getReg();
-  unsigned ptr = MI->getOperand(1).getReg();
-  unsigned incr = MI->getOperand(2).getReg();
-  AtomicOrdering Ord = static_cast<AtomicOrdering>(MI->getOperand(3).getImm());
-  DebugLoc dl = MI->getDebugLoc();
-  bool isThumb2 = Subtarget->isThumb2();
-
-  MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
-  if (isThumb2) {
-    MRI.constrainRegClass(dest, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(ptr, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(incr, &ARM::rGPRRegClass);
-  }
-
-  unsigned ldrOpc, strOpc;
-  getExclusiveOperation(Size, Ord, isThumb2, ldrOpc, strOpc);
-
-  MachineBasicBlock *loopMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MachineBasicBlock *exitMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MF->insert(It, loopMBB);
-  MF->insert(It, exitMBB);
-
-  // Transfer the remainder of BB and its successor edges to exitMBB.
-  exitMBB->splice(exitMBB->begin(), BB,
-                  llvm::next(MachineBasicBlock::iterator(MI)),
-                  BB->end());
-  exitMBB->transferSuccessorsAndUpdatePHIs(BB);
-
-  const TargetRegisterClass *TRC = isThumb2 ?
-    (const TargetRegisterClass*)&ARM::rGPRRegClass :
-    (const TargetRegisterClass*)&ARM::GPRRegClass;
-  unsigned scratch = MRI.createVirtualRegister(TRC);
-  unsigned scratch2 = (!BinOpcode) ? incr : MRI.createVirtualRegister(TRC);
-
-  //  thisMBB:
-  //   ...
-  //   fallthrough --> loopMBB
-  BB->addSuccessor(loopMBB);
-
-  //  loopMBB:
-  //   ldrex dest, ptr
-  //   <binop> scratch2, dest, incr
-  //   strex scratch, scratch2, ptr
-  //   cmp scratch, #0
-  //   bne- loopMBB
-  //   fallthrough --> exitMBB
-  BB = loopMBB;
-  MachineInstrBuilder MIB = BuildMI(BB, dl, TII->get(ldrOpc), dest).addReg(ptr);
-  if (ldrOpc == ARM::t2LDREX)
-    MIB.addImm(0);
-  AddDefaultPred(MIB);
-  if (BinOpcode) {
-    // operand order needs to go the other way for NAND
-    if (BinOpcode == ARM::BICrr || BinOpcode == ARM::t2BICrr)
-      AddDefaultPred(BuildMI(BB, dl, TII->get(BinOpcode), scratch2).
-                     addReg(incr).addReg(dest)).addReg(0);
-    else
-      AddDefaultPred(BuildMI(BB, dl, TII->get(BinOpcode), scratch2).
-                     addReg(dest).addReg(incr)).addReg(0);
-  }
-
-  MIB = BuildMI(BB, dl, TII->get(strOpc), scratch).addReg(scratch2).addReg(ptr);
-  if (strOpc == ARM::t2STREX)
-    MIB.addImm(0);
-  AddDefaultPred(MIB);
-  AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPri : ARM::CMPri))
-                 .addReg(scratch).addImm(0));
-  BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-    .addMBB(loopMBB).addImm(ARMCC::NE).addReg(ARM::CPSR);
-
-  BB->addSuccessor(loopMBB);
-  BB->addSuccessor(exitMBB);
-
-  //  exitMBB:
-  //   ...
-  BB = exitMBB;
-
-  MI->eraseFromParent();   // The instruction is gone now.
-
-  return BB;
-}
-
-MachineBasicBlock *
-ARMTargetLowering::EmitAtomicBinaryMinMax(MachineInstr *MI,
-                                          MachineBasicBlock *BB,
-                                          unsigned Size,
-                                          bool signExtend,
-                                          ARMCC::CondCodes Cond) const {
-  const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
-
-  const BasicBlock *LLVM_BB = BB->getBasicBlock();
-  MachineFunction *MF = BB->getParent();
-  MachineFunction::iterator It = BB;
-  ++It;
-
-  unsigned dest = MI->getOperand(0).getReg();
-  unsigned ptr = MI->getOperand(1).getReg();
-  unsigned incr = MI->getOperand(2).getReg();
-  unsigned oldval = dest;
-  AtomicOrdering Ord = static_cast<AtomicOrdering>(MI->getOperand(3).getImm());
-  DebugLoc dl = MI->getDebugLoc();
-  bool isThumb2 = Subtarget->isThumb2();
-
-  MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
-  if (isThumb2) {
-    MRI.constrainRegClass(dest, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(ptr, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(incr, &ARM::rGPRRegClass);
-  }
-
-  unsigned ldrOpc, strOpc, extendOpc;
-  getExclusiveOperation(Size, Ord, isThumb2, ldrOpc, strOpc);
-  switch (Size) {
-  default: llvm_unreachable("unsupported size for AtomicBinaryMinMax!");
-  case 1:
-    extendOpc = isThumb2 ? ARM::t2SXTB : ARM::SXTB;
-    break;
-  case 2:
-    extendOpc = isThumb2 ? ARM::t2SXTH : ARM::SXTH;
-    break;
-  case 4:
-    extendOpc = 0;
-    break;
-  }
-
-  MachineBasicBlock *loopMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MachineBasicBlock *exitMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MF->insert(It, loopMBB);
-  MF->insert(It, exitMBB);
-
-  // Transfer the remainder of BB and its successor edges to exitMBB.
-  exitMBB->splice(exitMBB->begin(), BB,
-                  llvm::next(MachineBasicBlock::iterator(MI)),
-                  BB->end());
-  exitMBB->transferSuccessorsAndUpdatePHIs(BB);
-
-  const TargetRegisterClass *TRC = isThumb2 ?
-    (const TargetRegisterClass*)&ARM::rGPRRegClass :
-    (const TargetRegisterClass*)&ARM::GPRRegClass;
-  unsigned scratch = MRI.createVirtualRegister(TRC);
-  unsigned scratch2 = MRI.createVirtualRegister(TRC);
-
-  //  thisMBB:
-  //   ...
-  //   fallthrough --> loopMBB
-  BB->addSuccessor(loopMBB);
-
-  //  loopMBB:
-  //   ldrex dest, ptr
-  //   (sign extend dest, if required)
-  //   cmp dest, incr
-  //   cmov.cond scratch2, incr, dest
-  //   strex scratch, scratch2, ptr
-  //   cmp scratch, #0
-  //   bne- loopMBB
-  //   fallthrough --> exitMBB
-  BB = loopMBB;
-  MachineInstrBuilder MIB = BuildMI(BB, dl, TII->get(ldrOpc), dest).addReg(ptr);
-  if (ldrOpc == ARM::t2LDREX)
-    MIB.addImm(0);
-  AddDefaultPred(MIB);
-
-  // Sign extend the value, if necessary.
-  if (signExtend && extendOpc) {
-    oldval = MRI.createVirtualRegister(isThumb2 ? &ARM::rGPRRegClass
-                                                : &ARM::GPRnopcRegClass);
-    if (!isThumb2)
-      MRI.constrainRegClass(dest, &ARM::GPRnopcRegClass);
-    AddDefaultPred(BuildMI(BB, dl, TII->get(extendOpc), oldval)
-                     .addReg(dest)
-                     .addImm(0));
-  }
-
-  // Build compare and cmov instructions.
-  AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPrr : ARM::CMPrr))
-                 .addReg(oldval).addReg(incr));
-  BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2MOVCCr : ARM::MOVCCr), scratch2)
-         .addReg(incr).addReg(oldval).addImm(Cond).addReg(ARM::CPSR);
-
-  MIB = BuildMI(BB, dl, TII->get(strOpc), scratch).addReg(scratch2).addReg(ptr);
-  if (strOpc == ARM::t2STREX)
-    MIB.addImm(0);
-  AddDefaultPred(MIB);
-  AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPri : ARM::CMPri))
-                 .addReg(scratch).addImm(0));
-  BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-    .addMBB(loopMBB).addImm(ARMCC::NE).addReg(ARM::CPSR);
-
-  BB->addSuccessor(loopMBB);
-  BB->addSuccessor(exitMBB);
-
-  //  exitMBB:
-  //   ...
-  BB = exitMBB;
-
-  MI->eraseFromParent();   // The instruction is gone now.
-
-  return BB;
-}
-
-MachineBasicBlock *
-ARMTargetLowering::EmitAtomicBinary64(MachineInstr *MI, MachineBasicBlock *BB,
-                                      unsigned Op1, unsigned Op2,
-                                      bool NeedsCarry, bool IsCmpxchg,
-                                      bool IsMinMax, ARMCC::CondCodes CC) const {
-  // This also handles ATOMIC_SWAP and ATOMIC_STORE, indicated by Op1==0.
-  const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
-
-  const BasicBlock *LLVM_BB = BB->getBasicBlock();
-  MachineFunction *MF = BB->getParent();
-  MachineFunction::iterator It = BB;
-  ++It;
-
-  bool isStore = (MI->getOpcode() == ARM::ATOMIC_STORE_I64);
-  unsigned offset = (isStore ? -2 : 0);
-  unsigned destlo = MI->getOperand(0).getReg();
-  unsigned desthi = MI->getOperand(1).getReg();
-  unsigned ptr = MI->getOperand(offset+2).getReg();
-  unsigned vallo = MI->getOperand(offset+3).getReg();
-  unsigned valhi = MI->getOperand(offset+4).getReg();
-  unsigned OrdIdx = offset + (IsCmpxchg ? 7 : 5);
-  AtomicOrdering Ord = static_cast<AtomicOrdering>(MI->getOperand(OrdIdx).getImm());
-  DebugLoc dl = MI->getDebugLoc();
-  bool isThumb2 = Subtarget->isThumb2();
-
-  MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
-  if (isThumb2) {
-    MRI.constrainRegClass(destlo, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(desthi, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(ptr, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(vallo, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(valhi, &ARM::rGPRRegClass);
-  }
-
-  unsigned ldrOpc, strOpc;
-  getExclusiveOperation(8, Ord, isThumb2, ldrOpc, strOpc);
-
-  MachineBasicBlock *loopMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MachineBasicBlock *contBB = 0, *cont2BB = 0;
-  if (IsCmpxchg || IsMinMax)
-    contBB = MF->CreateMachineBasicBlock(LLVM_BB);
-  if (IsCmpxchg)
-    cont2BB = MF->CreateMachineBasicBlock(LLVM_BB);
-  MachineBasicBlock *exitMBB = MF->CreateMachineBasicBlock(LLVM_BB);
-
-  MF->insert(It, loopMBB);
-  if (IsCmpxchg || IsMinMax) MF->insert(It, contBB);
-  if (IsCmpxchg) MF->insert(It, cont2BB);
-  MF->insert(It, exitMBB);
-
-  // Transfer the remainder of BB and its successor edges to exitMBB.
-  exitMBB->splice(exitMBB->begin(), BB,
-                  llvm::next(MachineBasicBlock::iterator(MI)),
-                  BB->end());
-  exitMBB->transferSuccessorsAndUpdatePHIs(BB);
-
-  const TargetRegisterClass *TRC = isThumb2 ?
-    (const TargetRegisterClass*)&ARM::tGPRRegClass :
-    (const TargetRegisterClass*)&ARM::GPRRegClass;
-  unsigned storesuccess = MRI.createVirtualRegister(TRC);
-
-  //  thisMBB:
-  //   ...
-  //   fallthrough --> loopMBB
-  BB->addSuccessor(loopMBB);
-
-  //  loopMBB:
-  //   ldrexd r2, r3, ptr
-  //   <binopa> r0, r2, incr
-  //   <binopb> r1, r3, incr
-  //   strexd storesuccess, r0, r1, ptr
-  //   cmp storesuccess, #0
-  //   bne- loopMBB
-  //   fallthrough --> exitMBB
-  BB = loopMBB;
-
-  if (!isStore) {
-    // Load
-    if (isThumb2) {
-      AddDefaultPred(BuildMI(BB, dl, TII->get(ldrOpc))
-                     .addReg(destlo, RegState::Define)
-                     .addReg(desthi, RegState::Define)
-                     .addReg(ptr));
-    } else {
-      unsigned GPRPair0 = MRI.createVirtualRegister(&ARM::GPRPairRegClass);
-      AddDefaultPred(BuildMI(BB, dl, TII->get(ldrOpc))
-                     .addReg(GPRPair0, RegState::Define).addReg(ptr));
-      // Copy r2/r3 into dest.  (This copy will normally be coalesced.)
-      BuildMI(BB, dl, TII->get(TargetOpcode::COPY), destlo)
-        .addReg(GPRPair0, 0, ARM::gsub_0);
-      BuildMI(BB, dl, TII->get(TargetOpcode::COPY), desthi)
-        .addReg(GPRPair0, 0, ARM::gsub_1);
-    }
-  }
-
-  unsigned StoreLo, StoreHi;
-  if (IsCmpxchg) {
-    // Add early exit
-    for (unsigned i = 0; i < 2; i++) {
-      AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPrr :
-                                                         ARM::CMPrr))
-                     .addReg(i == 0 ? destlo : desthi)
-                     .addReg(i == 0 ? vallo : valhi));
-      BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-        .addMBB(exitMBB).addImm(ARMCC::NE).addReg(ARM::CPSR);
-      BB->addSuccessor(exitMBB);
-      BB->addSuccessor(i == 0 ? contBB : cont2BB);
-      BB = (i == 0 ? contBB : cont2BB);
-    }
-
-    // Copy to physregs for strexd
-    StoreLo = MI->getOperand(5).getReg();
-    StoreHi = MI->getOperand(6).getReg();
-  } else if (Op1) {
-    // Perform binary operation
-    unsigned tmpRegLo = MRI.createVirtualRegister(TRC);
-    AddDefaultPred(BuildMI(BB, dl, TII->get(Op1), tmpRegLo)
-                   .addReg(destlo).addReg(vallo))
-        .addReg(NeedsCarry ? ARM::CPSR : 0, getDefRegState(NeedsCarry));
-    unsigned tmpRegHi = MRI.createVirtualRegister(TRC);
-    AddDefaultPred(BuildMI(BB, dl, TII->get(Op2), tmpRegHi)
-                   .addReg(desthi).addReg(valhi))
-        .addReg(IsMinMax ? ARM::CPSR : 0, getDefRegState(IsMinMax));
-
-    StoreLo = tmpRegLo;
-    StoreHi = tmpRegHi;
-  } else {
-    // Copy to physregs for strexd
-    StoreLo = vallo;
-    StoreHi = valhi;
-  }
-  if (IsMinMax) {
-    // Compare and branch to exit block.
-    BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-      .addMBB(exitMBB).addImm(CC).addReg(ARM::CPSR);
-    BB->addSuccessor(exitMBB);
-    BB->addSuccessor(contBB);
-    BB = contBB;
-    StoreLo = vallo;
-    StoreHi = valhi;
-  }
-
-  // Store
-  if (isThumb2) {
-    MRI.constrainRegClass(StoreLo, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(StoreHi, &ARM::rGPRRegClass);
-    AddDefaultPred(BuildMI(BB, dl, TII->get(strOpc), storesuccess)
-                   .addReg(StoreLo).addReg(StoreHi).addReg(ptr));
-  } else {
-    // Marshal a pair...
-    unsigned StorePair = MRI.createVirtualRegister(&ARM::GPRPairRegClass);
-    unsigned UndefPair = MRI.createVirtualRegister(&ARM::GPRPairRegClass);
-    unsigned r1 = MRI.createVirtualRegister(&ARM::GPRPairRegClass);
-    BuildMI(BB, dl, TII->get(TargetOpcode::IMPLICIT_DEF), UndefPair);
-    BuildMI(BB, dl, TII->get(TargetOpcode::INSERT_SUBREG), r1)
-      .addReg(UndefPair)
-      .addReg(StoreLo)
-      .addImm(ARM::gsub_0);
-    BuildMI(BB, dl, TII->get(TargetOpcode::INSERT_SUBREG), StorePair)
-      .addReg(r1)
-      .addReg(StoreHi)
-      .addImm(ARM::gsub_1);
-
-    // ...and store it
-    AddDefaultPred(BuildMI(BB, dl, TII->get(strOpc), storesuccess)
-                   .addReg(StorePair).addReg(ptr));
-  }
-  // Cmp+jump
-  AddDefaultPred(BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2CMPri : ARM::CMPri))
-                 .addReg(storesuccess).addImm(0));
-  BuildMI(BB, dl, TII->get(isThumb2 ? ARM::t2Bcc : ARM::Bcc))
-    .addMBB(loopMBB).addImm(ARMCC::NE).addReg(ARM::CPSR);
-
-  BB->addSuccessor(loopMBB);
-  BB->addSuccessor(exitMBB);
-
-  //  exitMBB:
-  //   ...
-  BB = exitMBB;
-
-  MI->eraseFromParent();   // The instruction is gone now.
-
-  return BB;
-}
-
-MachineBasicBlock *
-ARMTargetLowering::EmitAtomicLoad64(MachineInstr *MI, MachineBasicBlock *BB) const {
-
-  const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
-
-  unsigned destlo = MI->getOperand(0).getReg();
-  unsigned desthi = MI->getOperand(1).getReg();
-  unsigned ptr = MI->getOperand(2).getReg();
-  AtomicOrdering Ord = static_cast<AtomicOrdering>(MI->getOperand(3).getImm());
-  DebugLoc dl = MI->getDebugLoc();
-  bool isThumb2 = Subtarget->isThumb2();
-
-  MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
-  if (isThumb2) {
-    MRI.constrainRegClass(destlo, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(desthi, &ARM::rGPRRegClass);
-    MRI.constrainRegClass(ptr, &ARM::rGPRRegClass);
-  }
-  unsigned ldrOpc, strOpc;
-  getExclusiveOperation(8, Ord, isThumb2, ldrOpc, strOpc);
-
-  MachineInstrBuilder MIB = BuildMI(*BB, MI, dl, TII->get(ldrOpc));
-
-  if (isThumb2) {
-    MIB.addReg(destlo, RegState::Define)
-       .addReg(desthi, RegState::Define)
-       .addReg(ptr);
-
-  } else {
-    unsigned GPRPair0 = MRI.createVirtualRegister(&ARM::GPRPairRegClass);
-    MIB.addReg(GPRPair0, RegState::Define).addReg(ptr);
-
-    // Copy GPRPair0 into dest.  (This copy will normally be coalesced.)
-    BuildMI(*BB, MI, dl, TII->get(TargetOpcode::COPY), destlo)
-      .addReg(GPRPair0, 0, ARM::gsub_0);
-    BuildMI(*BB, MI, dl, TII->get(TargetOpcode::COPY), desthi)
-      .addReg(GPRPair0, 0, ARM::gsub_1);
-  }
-  AddDefaultPred(MIB);
-
-  MI->eraseFromParent();   // The instruction is gone now.
-
-  return BB;
-}
-
 /// SetupEntryBlockForSjLj - Insert code into the entry block that creates and
 /// registers the function context.
 void ARMTargetLowering::
@@ -7193,7 +6754,7 @@ EmitSjLjDispatchBlock(MachineInstr *MI, MachineBasicBlock *MBB) const {
   }
 
   // N.B. the order the invoke BBs are processed in doesn't matter here.
-  const uint16_t *SavedRegs = RI.getCalleeSavedRegs(MF);
+  const MCPhysReg *SavedRegs = RI.getCalleeSavedRegs(MF);
   SmallVector<MachineBasicBlock*, 64> MBBLPads;
   for (SmallPtrSet<MachineBasicBlock*, 64>::iterator
          I = InvokeBBs.begin(), E = InvokeBBs.end(); I != E; ++I) {
@@ -7390,8 +6951,8 @@ ARMTargetLowering::EmitStructByval(MachineInstr *MI,
   MachineFunction *MF = BB->getParent();
   MachineRegisterInfo &MRI = MF->getRegInfo();
   unsigned UnitSize = 0;
-  const TargetRegisterClass *TRC = 0;
-  const TargetRegisterClass *VecTRC = 0;
+  const TargetRegisterClass *TRC = nullptr;
+  const TargetRegisterClass *VecTRC = nullptr;
 
   bool IsThumb1 = Subtarget->isThumb1Only();
   bool IsThumb2 = Subtarget->isThumb2();
@@ -7425,7 +6986,7 @@ ARMTargetLowering::EmitStructByval(MachineInstr *MI,
                  ? (const TargetRegisterClass *)&ARM::DPairRegClass
                  : UnitSize == 8
                        ? (const TargetRegisterClass *)&ARM::DPRRegClass
-                       : 0;
+                       : nullptr;
 
   unsigned BytesLeft = SizeVal % UnitSize;
   unsigned LoopSize = SizeVal - BytesLeft;
@@ -7493,8 +7054,7 @@ ARMTargetLowering::EmitStructByval(MachineInstr *MI,
 
   // Transfer the remainder of BB and its successor edges to exitMBB.
   exitMBB->splice(exitMBB->begin(), BB,
-                  llvm::next(MachineBasicBlock::iterator(MI)),
-                  BB->end());
+                  std::next(MachineBasicBlock::iterator(MI)), BB->end());
   exitMBB->transferSuccessorsAndUpdatePHIs(BB);
 
   // Load an immediate to varEnd.
@@ -7608,6 +7168,72 @@ ARMTargetLowering::EmitStructByval(MachineInstr *MI,
 }
 
 MachineBasicBlock *
+ARMTargetLowering::EmitLowered__chkstk(MachineInstr *MI,
+                                       MachineBasicBlock *MBB) const {
+  const TargetMachine &TM = getTargetMachine();
+  const TargetInstrInfo &TII = *TM.getInstrInfo();
+  DebugLoc DL = MI->getDebugLoc();
+
+  assert(Subtarget->isTargetWindows() &&
+         "__chkstk is only supported on Windows");
+  assert(Subtarget->isThumb2() && "Windows on ARM requires Thumb-2 mode");
+
+  // __chkstk takes the number of words to allocate on the stack in R4, and
+  // returns the stack adjustment in number of bytes in R4.  This will not
+  // clober any other registers (other than the obvious lr).
+  //
+  // Although, technically, IP should be considered a register which may be
+  // clobbered, the call itself will not touch it.  Windows on ARM is a pure
+  // thumb-2 environment, so there is no interworking required.  As a result, we
+  // do not expect a veneer to be emitted by the linker, clobbering IP.
+  //
+  // Each module receives its own copy of __chkstk, so no import thunk is
+  // required, again, ensuring that IP is not clobbered.
+  //
+  // Finally, although some linkers may theoretically provide a trampoline for
+  // out of range calls (which is quite common due to a 32M range limitation of
+  // branches for Thumb), we can generate the long-call version via
+  // -mcmodel=large, alleviating the need for the trampoline which may clobber
+  // IP.
+
+  switch (TM.getCodeModel()) {
+  case CodeModel::Small:
+  case CodeModel::Medium:
+  case CodeModel::Default:
+  case CodeModel::Kernel:
+    BuildMI(*MBB, MI, DL, TII.get(ARM::tBL))
+      .addImm((unsigned)ARMCC::AL).addReg(0)
+      .addExternalSymbol("__chkstk")
+      .addReg(ARM::R4, RegState::Implicit | RegState::Kill)
+      .addReg(ARM::R4, RegState::Implicit | RegState::Define)
+      .addReg(ARM::R12, RegState::Implicit | RegState::Define | RegState::Dead);
+    break;
+  case CodeModel::Large:
+  case CodeModel::JITDefault: {
+    MachineRegisterInfo &MRI = MBB->getParent()->getRegInfo();
+    unsigned Reg = MRI.createVirtualRegister(&ARM::rGPRRegClass);
+
+    BuildMI(*MBB, MI, DL, TII.get(ARM::t2MOVi32imm), Reg)
+      .addExternalSymbol("__chkstk");
+    BuildMI(*MBB, MI, DL, TII.get(ARM::tBLXr))
+      .addImm((unsigned)ARMCC::AL).addReg(0)
+      .addReg(Reg, RegState::Kill)
+      .addReg(ARM::R4, RegState::Implicit | RegState::Kill)
+      .addReg(ARM::R4, RegState::Implicit | RegState::Define)
+      .addReg(ARM::R12, RegState::Implicit | RegState::Define | RegState::Dead);
+    break;
+  }
+  }
+
+  AddDefaultCC(AddDefaultPred(BuildMI(*MBB, MI, DL, TII.get(ARM::t2SUBrr),
+                                      ARM::SP)
+                              .addReg(ARM::SP).addReg(ARM::R4)));
+
+  MI->eraseFromParent();
+  return MBB;
+}
+
+MachineBasicBlock *
 ARMTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
                                                MachineBasicBlock *BB) const {
   const TargetInstrInfo *TII = getTargetMachine().getInstrInfo();
@@ -7670,131 +7296,6 @@ ARMTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     MI->eraseFromParent();
     return BB;
   }
-  case ARM::ATOMIC_LOAD_ADD_I8:
-     return EmitAtomicBinary(MI, BB, 1, isThumb2 ? ARM::t2ADDrr : ARM::ADDrr);
-  case ARM::ATOMIC_LOAD_ADD_I16:
-     return EmitAtomicBinary(MI, BB, 2, isThumb2 ? ARM::t2ADDrr : ARM::ADDrr);
-  case ARM::ATOMIC_LOAD_ADD_I32:
-     return EmitAtomicBinary(MI, BB, 4, isThumb2 ? ARM::t2ADDrr : ARM::ADDrr);
-
-  case ARM::ATOMIC_LOAD_AND_I8:
-     return EmitAtomicBinary(MI, BB, 1, isThumb2 ? ARM::t2ANDrr : ARM::ANDrr);
-  case ARM::ATOMIC_LOAD_AND_I16:
-     return EmitAtomicBinary(MI, BB, 2, isThumb2 ? ARM::t2ANDrr : ARM::ANDrr);
-  case ARM::ATOMIC_LOAD_AND_I32:
-     return EmitAtomicBinary(MI, BB, 4, isThumb2 ? ARM::t2ANDrr : ARM::ANDrr);
-
-  case ARM::ATOMIC_LOAD_OR_I8:
-     return EmitAtomicBinary(MI, BB, 1, isThumb2 ? ARM::t2ORRrr : ARM::ORRrr);
-  case ARM::ATOMIC_LOAD_OR_I16:
-     return EmitAtomicBinary(MI, BB, 2, isThumb2 ? ARM::t2ORRrr : ARM::ORRrr);
-  case ARM::ATOMIC_LOAD_OR_I32:
-     return EmitAtomicBinary(MI, BB, 4, isThumb2 ? ARM::t2ORRrr : ARM::ORRrr);
-
-  case ARM::ATOMIC_LOAD_XOR_I8:
-     return EmitAtomicBinary(MI, BB, 1, isThumb2 ? ARM::t2EORrr : ARM::EORrr);
-  case ARM::ATOMIC_LOAD_XOR_I16:
-     return EmitAtomicBinary(MI, BB, 2, isThumb2 ? ARM::t2EORrr : ARM::EORrr);
-  case ARM::ATOMIC_LOAD_XOR_I32:
-     return EmitAtomicBinary(MI, BB, 4, isThumb2 ? ARM::t2EORrr : ARM::EORrr);
-
-  case ARM::ATOMIC_LOAD_NAND_I8:
-     return EmitAtomicBinary(MI, BB, 1, isThumb2 ? ARM::t2BICrr : ARM::BICrr);
-  case ARM::ATOMIC_LOAD_NAND_I16:
-     return EmitAtomicBinary(MI, BB, 2, isThumb2 ? ARM::t2BICrr : ARM::BICrr);
-  case ARM::ATOMIC_LOAD_NAND_I32:
-     return EmitAtomicBinary(MI, BB, 4, isThumb2 ? ARM::t2BICrr : ARM::BICrr);
-
-  case ARM::ATOMIC_LOAD_SUB_I8:
-     return EmitAtomicBinary(MI, BB, 1, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr);
-  case ARM::ATOMIC_LOAD_SUB_I16:
-     return EmitAtomicBinary(MI, BB, 2, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr);
-  case ARM::ATOMIC_LOAD_SUB_I32:
-     return EmitAtomicBinary(MI, BB, 4, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr);
-
-  case ARM::ATOMIC_LOAD_MIN_I8:
-     return EmitAtomicBinaryMinMax(MI, BB, 1, true, ARMCC::LT);
-  case ARM::ATOMIC_LOAD_MIN_I16:
-     return EmitAtomicBinaryMinMax(MI, BB, 2, true, ARMCC::LT);
-  case ARM::ATOMIC_LOAD_MIN_I32:
-     return EmitAtomicBinaryMinMax(MI, BB, 4, true, ARMCC::LT);
-
-  case ARM::ATOMIC_LOAD_MAX_I8:
-     return EmitAtomicBinaryMinMax(MI, BB, 1, true, ARMCC::GT);
-  case ARM::ATOMIC_LOAD_MAX_I16:
-     return EmitAtomicBinaryMinMax(MI, BB, 2, true, ARMCC::GT);
-  case ARM::ATOMIC_LOAD_MAX_I32:
-     return EmitAtomicBinaryMinMax(MI, BB, 4, true, ARMCC::GT);
-
-  case ARM::ATOMIC_LOAD_UMIN_I8:
-     return EmitAtomicBinaryMinMax(MI, BB, 1, false, ARMCC::LO);
-  case ARM::ATOMIC_LOAD_UMIN_I16:
-     return EmitAtomicBinaryMinMax(MI, BB, 2, false, ARMCC::LO);
-  case ARM::ATOMIC_LOAD_UMIN_I32:
-     return EmitAtomicBinaryMinMax(MI, BB, 4, false, ARMCC::LO);
-
-  case ARM::ATOMIC_LOAD_UMAX_I8:
-     return EmitAtomicBinaryMinMax(MI, BB, 1, false, ARMCC::HI);
-  case ARM::ATOMIC_LOAD_UMAX_I16:
-     return EmitAtomicBinaryMinMax(MI, BB, 2, false, ARMCC::HI);
-  case ARM::ATOMIC_LOAD_UMAX_I32:
-     return EmitAtomicBinaryMinMax(MI, BB, 4, false, ARMCC::HI);
-
-  case ARM::ATOMIC_SWAP_I8:  return EmitAtomicBinary(MI, BB, 1, 0);
-  case ARM::ATOMIC_SWAP_I16: return EmitAtomicBinary(MI, BB, 2, 0);
-  case ARM::ATOMIC_SWAP_I32: return EmitAtomicBinary(MI, BB, 4, 0);
-
-  case ARM::ATOMIC_CMP_SWAP_I8:  return EmitAtomicCmpSwap(MI, BB, 1);
-  case ARM::ATOMIC_CMP_SWAP_I16: return EmitAtomicCmpSwap(MI, BB, 2);
-  case ARM::ATOMIC_CMP_SWAP_I32: return EmitAtomicCmpSwap(MI, BB, 4);
-
-  case ARM::ATOMIC_LOAD_I64:
-    return EmitAtomicLoad64(MI, BB);
-
-  case ARM::ATOMIC_LOAD_ADD_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2ADDrr : ARM::ADDrr,
-                              isThumb2 ? ARM::t2ADCrr : ARM::ADCrr,
-                              /*NeedsCarry*/ true);
-  case ARM::ATOMIC_LOAD_SUB_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr,
-                              isThumb2 ? ARM::t2SBCrr : ARM::SBCrr,
-                              /*NeedsCarry*/ true);
-  case ARM::ATOMIC_LOAD_OR_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2ORRrr : ARM::ORRrr,
-                              isThumb2 ? ARM::t2ORRrr : ARM::ORRrr);
-  case ARM::ATOMIC_LOAD_XOR_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2EORrr : ARM::EORrr,
-                              isThumb2 ? ARM::t2EORrr : ARM::EORrr);
-  case ARM::ATOMIC_LOAD_AND_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2ANDrr : ARM::ANDrr,
-                              isThumb2 ? ARM::t2ANDrr : ARM::ANDrr);
-  case ARM::ATOMIC_STORE_I64:
-  case ARM::ATOMIC_SWAP_I64:
-    return EmitAtomicBinary64(MI, BB, 0, 0, false);
-  case ARM::ATOMIC_CMP_SWAP_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr,
-                              isThumb2 ? ARM::t2SBCrr : ARM::SBCrr,
-                              /*NeedsCarry*/ false, /*IsCmpxchg*/true);
-  case ARM::ATOMIC_LOAD_MIN_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr,
-                              isThumb2 ? ARM::t2SBCrr : ARM::SBCrr,
-                              /*NeedsCarry*/ true, /*IsCmpxchg*/false,
-                              /*IsMinMax*/ true, ARMCC::LT);
-  case ARM::ATOMIC_LOAD_MAX_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr,
-                              isThumb2 ? ARM::t2SBCrr : ARM::SBCrr,
-                              /*NeedsCarry*/ true, /*IsCmpxchg*/false,
-                              /*IsMinMax*/ true, ARMCC::GE);
-  case ARM::ATOMIC_LOAD_UMIN_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr,
-                              isThumb2 ? ARM::t2SBCrr : ARM::SBCrr,
-                              /*NeedsCarry*/ true, /*IsCmpxchg*/false,
-                              /*IsMinMax*/ true, ARMCC::LO);
-  case ARM::ATOMIC_LOAD_UMAX_I64:
-    return EmitAtomicBinary64(MI, BB, isThumb2 ? ARM::t2SUBrr : ARM::SUBrr,
-                              isThumb2 ? ARM::t2SBCrr : ARM::SBCrr,
-                              /*NeedsCarry*/ true, /*IsCmpxchg*/false,
-                              /*IsMinMax*/ true, ARMCC::HS);
 
   case ARM::tMOVCCr_pseudo: {
     // To "insert" a SELECT_CC instruction, we actually have to insert the
@@ -7820,8 +7321,7 @@ ARMTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     // Transfer the remainder of BB and its successor edges to sinkMBB.
     sinkMBB->splice(sinkMBB->begin(), BB,
-                    llvm::next(MachineBasicBlock::iterator(MI)),
-                    BB->end());
+                    std::next(MachineBasicBlock::iterator(MI)), BB->end());
     sinkMBB->transferSuccessorsAndUpdatePHIs(BB);
 
     BB->addSuccessor(copy0MBB);
@@ -7854,7 +7354,7 @@ ARMTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
   case ARM::BCCi64:
   case ARM::BCCZi64: {
     // If there is an unconditional branch to the other successor, remove it.
-    BB->erase(llvm::next(MachineBasicBlock::iterator(MI)), BB->end());
+    BB->erase(std::next(MachineBasicBlock::iterator(MI)), BB->end());
 
     // Compare both parts that make up the double comparison separately for
     // equality.
@@ -7939,8 +7439,7 @@ ARMTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     // Transfer the remainder of BB and its successor edges to sinkMBB.
     SinkBB->splice(SinkBB->begin(), BB,
-      llvm::next(MachineBasicBlock::iterator(MI)),
-      BB->end());
+                   std::next(MachineBasicBlock::iterator(MI)), BB->end());
     SinkBB->transferSuccessorsAndUpdatePHIs(BB);
 
     BB->addSuccessor(RSBBB);
@@ -7983,6 +7482,8 @@ ARMTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
   case ARM::COPY_STRUCT_BYVAL_I32:
     ++NumLoopByVals;
     return EmitStructByval(MI, BB);
+  case ARM::WIN__CHKSTK:
+    return EmitLowered__chkstk(MI, BB);
   }
 }
 
@@ -8273,7 +7774,9 @@ static SDValue AddCombineToVPADDL(SDNode *N, SDValue N0, SDValue N1,
   // Get widened type and narrowed type.
   MVT widenType;
   unsigned numElem = VT.getVectorNumElements();
-  switch (VT.getVectorElementType().getSimpleVT().SimpleTy) {
+  
+  EVT inputLaneType = Vec.getValueType().getVectorElementType();
+  switch (inputLaneType.getSimpleVT().SimpleTy) {
     case MVT::i8: widenType = MVT::getVectorVT(MVT::i16, numElem); break;
     case MVT::i16: widenType = MVT::getVectorVT(MVT::i32, numElem); break;
     case MVT::i32: widenType = MVT::getVectorVT(MVT::i64, numElem); break;
@@ -8281,9 +7784,9 @@ static SDValue AddCombineToVPADDL(SDNode *N, SDValue N0, SDValue N1,
       llvm_unreachable("Invalid vector element type for padd optimization.");
   }
 
-  SDValue tmp = DAG.getNode(ISD::INTRINSIC_WO_CHAIN, SDLoc(N),
-                            widenType, &Ops[0], Ops.size());
-  return DAG.getNode(ISD::TRUNCATE, SDLoc(N), VT, tmp);
+  SDValue tmp = DAG.getNode(ISD::INTRINSIC_WO_CHAIN, SDLoc(N), widenType, Ops);
+  unsigned ExtOp = VT.bitsGT(tmp.getValueType()) ? ISD::ANY_EXTEND : ISD::TRUNCATE;
+  return DAG.getNode(ExtOp, SDLoc(N), VT, tmp);
 }
 
 static SDValue findMUL_LOHI(SDValue V) {
@@ -8341,7 +7844,7 @@ static SDValue AddCombineTo64bitMLAL(SDNode *AddcNode,
 
   // Look for the glued ADDE.
   SDNode* AddeNode = AddcNode->getGluedUser();
-  if (AddeNode == NULL)
+  if (!AddeNode)
     return SDValue();
 
   // Make sure it is really an ADDE.
@@ -8376,9 +7879,9 @@ static SDValue AddCombineTo64bitMLAL(SDNode *AddcNode,
 
   // Figure out the high and low input values to the MLAL node.
   SDValue* HiMul = &MULOp;
-  SDValue* HiAdd = NULL;
-  SDValue* LoMul = NULL;
-  SDValue* LowAdd = NULL;
+  SDValue* HiAdd = nullptr;
+  SDValue* LoMul = nullptr;
+  SDValue* LowAdd = nullptr;
 
   if (IsLeftOperandMUL)
     HiAdd = &AddeOp1;
@@ -8395,7 +7898,7 @@ static SDValue AddCombineTo64bitMLAL(SDNode *AddcNode,
     LowAdd = &AddcOp0;
   }
 
-  if (LoMul == NULL)
+  if (!LoMul)
     return SDValue();
 
   if (LoMul->getNode() != HiMul->getNode())
@@ -8412,8 +7915,7 @@ static SDValue AddCombineTo64bitMLAL(SDNode *AddcNode,
   Ops.push_back(*HiAdd);
 
   SDValue MLALNode =  DAG.getNode(FinalOpc, SDLoc(AddcNode),
-                                 DAG.getVTList(MVT::i32, MVT::i32),
-                                 &Ops[0], Ops.size());
+                                 DAG.getVTList(MVT::i32, MVT::i32), Ops);
 
   // Replace the ADDs' nodes uses by the MLA node's values.
   SDValue HiMLALResult(MLALNode.getNode(), 1);
@@ -8937,6 +8439,8 @@ static SDValue PerformVMOVRRDCombine(SDNode *N,
                                  std::min(4U, LD->getAlignment() / 2));
 
     DAG.ReplaceAllUsesOfValueWith(SDValue(LD, 1), NewLD2.getValue(1));
+    if (DCI.DAG.getTargetLoweringInfo().isBigEndian())
+      std::swap (NewLD1, NewLD2);
     SDValue Result = DCI.CombineTo(N, NewLD1, NewLD2);
     DCI.RemoveFromWorklist(LD);
     DAG.DeleteNode(LD);
@@ -9004,7 +8508,8 @@ static SDValue PerformSTORECombine(SDNode *N,
     SDLoc DL(St);
     SDValue WideVec = DAG.getNode(ISD::BITCAST, DL, WideVecVT, StVal);
     SmallVector<int, 8> ShuffleVec(NumElems * SizeRatio, -1);
-    for (unsigned i = 0; i < NumElems; ++i) ShuffleVec[i] = i * SizeRatio;
+    for (unsigned i = 0; i < NumElems; ++i)
+      ShuffleVec[i] = TLI.isBigEndian() ? (i+1) * SizeRatio - 1 : i * SizeRatio;
 
     // Can't shuffle using an illegal type.
     if (!TLI.isTypeLegal(WideVecVT)) return SDValue();
@@ -9050,8 +8555,7 @@ static SDValue PerformSTORECombine(SDNode *N,
                             Increment);
       Chains.push_back(Ch);
     }
-    return DAG.getNode(ISD::TokenFactor, DL, MVT::Other, &Chains[0],
-                       Chains.size());
+    return DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Chains);
   }
 
   if (!ISD::isNormalStore(St))
@@ -9062,16 +8566,18 @@ static SDValue PerformSTORECombine(SDNode *N,
   if (StVal.getNode()->getOpcode() == ARMISD::VMOVDRR &&
       StVal.getNode()->hasOneUse()) {
     SelectionDAG  &DAG = DCI.DAG;
+    bool isBigEndian = DAG.getTargetLoweringInfo().isBigEndian();
     SDLoc DL(St);
     SDValue BasePtr = St->getBasePtr();
     SDValue NewST1 = DAG.getStore(St->getChain(), DL,
-                                  StVal.getNode()->getOperand(0), BasePtr,
-                                  St->getPointerInfo(), St->isVolatile(),
+                                  StVal.getNode()->getOperand(isBigEndian ? 1 : 0 ),
+                                  BasePtr, St->getPointerInfo(), St->isVolatile(),
                                   St->isNonTemporal(), St->getAlignment());
 
     SDValue OffsetPtr = DAG.getNode(ISD::ADD, DL, MVT::i32, BasePtr,
                                     DAG.getConstant(4, MVT::i32));
-    return DAG.getStore(NewST1.getValue(0), DL, StVal.getNode()->getOperand(1),
+    return DAG.getStore(NewST1.getValue(0), DL,
+                        StVal.getNode()->getOperand(isBigEndian ? 0 : 1),
                         OffsetPtr, St->getPointerInfo(), St->isVolatile(),
                         St->isNonTemporal(),
                         std::min(4U, St->getAlignment() / 2));
@@ -9147,7 +8653,7 @@ static SDValue PerformBUILD_VECTORCombine(SDNode *N,
     DCI.AddToWorklist(V.getNode());
   }
   EVT FloatVT = EVT::getVectorVT(*DAG.getContext(), MVT::f64, NumElts);
-  SDValue BV = DAG.getNode(ISD::BUILD_VECTOR, dl, FloatVT, Ops.data(), NumElts);
+  SDValue BV = DAG.getNode(ISD::BUILD_VECTOR, dl, FloatVT, Ops);
   return DAG.getNode(ISD::BITCAST, dl, VT, BV);
 }
 
@@ -9230,7 +8736,7 @@ PerformARMBUILD_VECTORCombine(SDNode *N, TargetLowering::DAGCombinerInfo &DCI) {
       // Fold obvious case.
       V = V.getOperand(0);
     else {
-      V = DAG.getNode(ISD::BITCAST, SDLoc(V), MVT::i32, V); 
+      V = DAG.getNode(ISD::BITCAST, SDLoc(V), MVT::i32, V);
       // Make the DAGCombiner fold the bitcasts.
       DCI.AddToWorklist(V.getNode());
     }
@@ -9426,7 +8932,7 @@ static SDValue CombineBaseUpdate(SDNode *N,
       Tys[n] = VecTy;
     Tys[n++] = MVT::i32;
     Tys[n] = MVT::Other;
-    SDVTList SDTys = DAG.getVTList(Tys, NumResultVecs+2);
+    SDVTList SDTys = DAG.getVTList(ArrayRef<EVT>(Tys, NumResultVecs+2));
     SmallVector<SDValue, 8> Ops;
     Ops.push_back(N->getOperand(0)); // incoming chain
     Ops.push_back(N->getOperand(AddrOpIdx));
@@ -9436,8 +8942,7 @@ static SDValue CombineBaseUpdate(SDNode *N,
     }
     MemIntrinsicSDNode *MemInt = cast<MemIntrinsicSDNode>(N);
     SDValue UpdN = DAG.getMemIntrinsicNode(NewOpc, SDLoc(N), SDTys,
-                                           Ops.data(), Ops.size(),
-                                           MemInt->getMemoryVT(),
+                                           Ops, MemInt->getMemoryVT(),
                                            MemInt->getMemOperand());
 
     // Update the uses.
@@ -9506,11 +9011,11 @@ static bool CombineVLDDUP(SDNode *N, TargetLowering::DAGCombinerInfo &DCI) {
   for (n = 0; n < NumVecs; ++n)
     Tys[n] = VT;
   Tys[n] = MVT::Other;
-  SDVTList SDTys = DAG.getVTList(Tys, NumVecs+1);
+  SDVTList SDTys = DAG.getVTList(ArrayRef<EVT>(Tys, NumVecs+1));
   SDValue Ops[] = { VLD->getOperand(0), VLD->getOperand(2) };
   MemIntrinsicSDNode *VLDMemInt = cast<MemIntrinsicSDNode>(VLD);
   SDValue VLDDup = DAG.getMemIntrinsicNode(NewOpc, SDLoc(VLD), SDTys,
-                                           Ops, 2, VLDMemInt->getMemoryVT(),
+                                           Ops, VLDMemInt->getMemoryVT(),
                                            VLDMemInt->getMemOperand());
 
   // Update the uses.
@@ -9759,9 +9264,6 @@ static SDValue PerformIntrinsicCombine(SDNode *N, SelectionDAG &DAG) {
   // loads from a constant pool.
   case Intrinsic::arm_neon_vshifts:
   case Intrinsic::arm_neon_vshiftu:
-  case Intrinsic::arm_neon_vshiftls:
-  case Intrinsic::arm_neon_vshiftlu:
-  case Intrinsic::arm_neon_vshiftn:
   case Intrinsic::arm_neon_vrshifts:
   case Intrinsic::arm_neon_vrshiftu:
   case Intrinsic::arm_neon_vrshiftn:
@@ -9792,12 +9294,6 @@ static SDValue PerformIntrinsicCombine(SDNode *N, SelectionDAG &DAG) {
       }
       return SDValue();
 
-    case Intrinsic::arm_neon_vshiftls:
-    case Intrinsic::arm_neon_vshiftlu:
-      if (isVShiftLImm(N->getOperand(2), VT, true, Cnt))
-        break;
-      llvm_unreachable("invalid shift count for vshll intrinsic");
-
     case Intrinsic::arm_neon_vrshifts:
     case Intrinsic::arm_neon_vrshiftu:
       if (isVShiftRImm(N->getOperand(2), VT, false, true, Cnt))
@@ -9815,7 +9311,6 @@ static SDValue PerformIntrinsicCombine(SDNode *N, SelectionDAG &DAG) {
         break;
       llvm_unreachable("invalid shift count for vqshlu intrinsic");
 
-    case Intrinsic::arm_neon_vshiftn:
     case Intrinsic::arm_neon_vrshiftn:
     case Intrinsic::arm_neon_vqshiftns:
     case Intrinsic::arm_neon_vqshiftnu:
@@ -9838,16 +9333,6 @@ static SDValue PerformIntrinsicCombine(SDNode *N, SelectionDAG &DAG) {
     case Intrinsic::arm_neon_vshiftu:
       // Opcode already set above.
       break;
-    case Intrinsic::arm_neon_vshiftls:
-    case Intrinsic::arm_neon_vshiftlu:
-      if (Cnt == VT.getVectorElementType().getSizeInBits())
-        VShiftOpc = ARMISD::VSHLLi;
-      else
-        VShiftOpc = (IntNo == Intrinsic::arm_neon_vshiftls ?
-                     ARMISD::VSHLLs : ARMISD::VSHLLu);
-      break;
-    case Intrinsic::arm_neon_vshiftn:
-      VShiftOpc = ARMISD::VSHRN; break;
     case Intrinsic::arm_neon_vrshifts:
       VShiftOpc = ARMISD::VRSHRs; break;
     case Intrinsic::arm_neon_vrshiftu:
@@ -10128,7 +9613,7 @@ ARMTargetLowering::PerformCMOVCombine(SDNode *N, SelectionDAG &DAG) const {
 
   if (Res.getNode()) {
     APInt KnownZero, KnownOne;
-    DAG.ComputeMaskedBits(SDValue(N,0), KnownZero, KnownOne);
+    DAG.computeKnownBits(SDValue(N,0), KnownZero, KnownOne);
     // Capture demanded bits information that would be otherwise lost.
     if (KnownZero == 0xfffffffe)
       Res = DAG.getNode(ISD::AssertZext, dl, MVT::i32, Res,
@@ -10211,7 +9696,8 @@ bool ARMTargetLowering::isDesirableToTransformToIntegerOp(unsigned Opc,
   return (VT == MVT::f32) && (Opc == ISD::LOAD || Opc == ISD::STORE);
 }
 
-bool ARMTargetLowering::allowsUnalignedMemoryAccesses(EVT VT, bool *Fast) const {
+bool ARMTargetLowering::allowsUnalignedMemoryAccesses(EVT VT, unsigned,
+                                                      bool *Fast) const {
   // The AllowsUnaliged flag models the SCTLR.A setting in ARM cpus
   bool AllowsUnaligned = Subtarget->allowsUnalignedMem();
 
@@ -10233,7 +9719,7 @@ bool ARMTargetLowering::allowsUnalignedMemoryAccesses(EVT VT, bool *Fast) const
   case MVT::v2f64: {
     // For any little-endian targets with neon, we can support unaligned ld/st
     // of D and Q (e.g. {D0,D1}) registers by using vld1.i8/vst1.i8.
-    // A big-endian target may also explictly support unaligned accesses
+    // A big-endian target may also explicitly support unaligned accesses
     if (Subtarget->hasNEON() && (AllowsUnaligned || isLittleEndian())) {
       if (Fast)
         *Fast = true;
@@ -10265,11 +9751,11 @@ EVT ARMTargetLowering::getOptimalMemOpType(uint64_t Size,
     bool Fast;
     if (Size >= 16 &&
         (memOpAlign(SrcAlign, DstAlign, 16) ||
-         (allowsUnalignedMemoryAccesses(MVT::v2f64, &Fast) && Fast))) {
+         (allowsUnalignedMemoryAccesses(MVT::v2f64, 0, &Fast) && Fast))) {
       return MVT::v2f64;
     } else if (Size >= 8 &&
                (memOpAlign(SrcAlign, DstAlign, 8) ||
-                (allowsUnalignedMemoryAccesses(MVT::f64, &Fast) && Fast))) {
+                (allowsUnalignedMemoryAccesses(MVT::f64, 0, &Fast) && Fast))) {
       return MVT::f64;
     }
   }
@@ -10714,11 +10200,11 @@ bool ARMTargetLowering::getPostIndexedAddressParts(SDNode *N, SDNode *Op,
   return true;
 }
 
-void ARMTargetLowering::computeMaskedBitsForTargetNode(const SDValue Op,
-                                                       APInt &KnownZero,
-                                                       APInt &KnownOne,
-                                                       const SelectionDAG &DAG,
-                                                       unsigned Depth) const {
+void ARMTargetLowering::computeKnownBitsForTargetNode(const SDValue Op,
+                                                      APInt &KnownZero,
+                                                      APInt &KnownOne,
+                                                      const SelectionDAG &DAG,
+                                                      unsigned Depth) const {
   unsigned BitWidth = KnownOne.getBitWidth();
   KnownZero = KnownOne = APInt(BitWidth, 0);
   switch (Op.getOpcode()) {
@@ -10734,15 +10220,29 @@ void ARMTargetLowering::computeMaskedBitsForTargetNode(const SDValue Op,
     break;
   case ARMISD::CMOV: {
     // Bits are known zero/one if known on the LHS and RHS.
-    DAG.ComputeMaskedBits(Op.getOperand(0), KnownZero, KnownOne, Depth+1);
+    DAG.computeKnownBits(Op.getOperand(0), KnownZero, KnownOne, Depth+1);
     if (KnownZero == 0 && KnownOne == 0) return;
 
     APInt KnownZeroRHS, KnownOneRHS;
-    DAG.ComputeMaskedBits(Op.getOperand(1), KnownZeroRHS, KnownOneRHS, Depth+1);
+    DAG.computeKnownBits(Op.getOperand(1), KnownZeroRHS, KnownOneRHS, Depth+1);
     KnownZero &= KnownZeroRHS;
     KnownOne  &= KnownOneRHS;
     return;
   }
+  case ISD::INTRINSIC_W_CHAIN: {
+    ConstantSDNode *CN = cast<ConstantSDNode>(Op->getOperand(1));
+    Intrinsic::ID IntID = static_cast<Intrinsic::ID>(CN->getZExtValue());
+    switch (IntID) {
+    default: return;
+    case Intrinsic::arm_ldaex:
+    case Intrinsic::arm_ldrex: {
+      EVT VT = cast<MemIntrinsicSDNode>(Op)->getMemoryVT();
+      unsigned MemBits = VT.getScalarType().getSizeInBits();
+      KnownZero |= APInt::getHighBitsSet(BitWidth, BitWidth - MemBits);
+      return;
+    }
+    }
+  }
   }
 }
 
@@ -10818,7 +10318,7 @@ ARMTargetLowering::getSingleConstraintMatchWeight(
   Value *CallOperandVal = info.CallOperandVal;
     // If we don't have a value, we can't do a match,
     // but allow it at the lowest weight.
-  if (CallOperandVal == NULL)
+  if (!CallOperandVal)
     return CW_Default;
   Type *type = CallOperandVal->getType();
   // Look at the constraint type.
@@ -10897,7 +10397,7 @@ void ARMTargetLowering::LowerAsmOperandForConstraint(SDValue Op,
                                                      std::string &Constraint,
                                                      std::vector<SDValue>&Ops,
                                                      SelectionDAG &DAG) const {
-  SDValue Result(0, 0);
+  SDValue Result;
 
   // Currently only support length 1 constraints.
   if (Constraint.length() != 1) return;
@@ -11096,16 +10596,41 @@ SDValue ARMTargetLowering::LowerDivRem(SDValue Op, SelectionDAG &DAG) const {
   Type *RetTy = (Type*)StructType::get(Ty, Ty, NULL);
 
   SDLoc dl(Op);
-  TargetLowering::
-  CallLoweringInfo CLI(InChain, RetTy, isSigned, !isSigned, false, true,
-                    0, getLibcallCallingConv(LC), /*isTailCall=*/false,
-                    /*doesNotReturn=*/false, /*isReturnValueUsed=*/true,
-                    Callee, Args, DAG, dl);
-  std::pair<SDValue, SDValue> CallInfo = LowerCallTo(CLI);
+  TargetLowering::CallLoweringInfo CLI(DAG);
+  CLI.setDebugLoc(dl).setChain(InChain)
+    .setCallee(getLibcallCallingConv(LC), RetTy, Callee, std::move(Args), 0)
+    .setInRegister().setSExtResult(isSigned).setZExtResult(!isSigned);
 
+  std::pair<SDValue, SDValue> CallInfo = LowerCallTo(CLI);
   return CallInfo.first;
 }
 
+SDValue
+ARMTargetLowering::LowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG) const {
+  assert(Subtarget->isTargetWindows() && "unsupported target platform");
+  SDLoc DL(Op);
+
+  // Get the inputs.
+  SDValue Chain = Op.getOperand(0);
+  SDValue Size  = Op.getOperand(1);
+
+  SDValue Words = DAG.getNode(ISD::SRL, DL, MVT::i32, Size,
+                              DAG.getConstant(2, MVT::i32));
+
+  SDValue Flag;
+  Chain = DAG.getCopyToReg(Chain, DL, ARM::R4, Words, Flag);
+  Flag = Chain.getValue(1);
+
+  SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
+  Chain = DAG.getNode(ARMISD::WIN__CHKSTK, DL, NodeTys, Chain, Flag);
+
+  SDValue NewSP = DAG.getCopyFromReg(Chain, DL, ARM::SP, MVT::i32);
+  Chain = NewSP.getValue(1);
+
+  SDValue Ops[2] = { NewSP, Chain };
+  return DAG.getMergeValues(Ops, DL);
+}
+
 bool
 ARMTargetLowering::isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const {
   // The ARM target isn't yet aware of offsets.
@@ -11191,6 +10716,7 @@ bool ARMTargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
     Info.writeMem = true;
     return true;
   }
+  case Intrinsic::arm_ldaex:
   case Intrinsic::arm_ldrex: {
     PointerType *PtrTy = cast<PointerType>(I.getArgOperand(0)->getType());
     Info.opc = ISD::INTRINSIC_W_CHAIN;
@@ -11203,6 +10729,7 @@ bool ARMTargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
     Info.writeMem = false;
     return true;
   }
+  case Intrinsic::arm_stlex:
   case Intrinsic::arm_strex: {
     PointerType *PtrTy = cast<PointerType>(I.getArgOperand(1)->getType());
     Info.opc = ISD::INTRINSIC_W_CHAIN;
@@ -11215,6 +10742,7 @@ bool ARMTargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
     Info.writeMem = true;
     return true;
   }
+  case Intrinsic::arm_stlexd:
   case Intrinsic::arm_strexd: {
     Info.opc = ISD::INTRINSIC_W_CHAIN;
     Info.memVT = MVT::i64;
@@ -11226,6 +10754,7 @@ bool ARMTargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
     Info.writeMem = true;
     return true;
   }
+  case Intrinsic::arm_ldaexd:
   case Intrinsic::arm_ldrexd: {
     Info.opc = ISD::INTRINSIC_W_CHAIN;
     Info.memVT = MVT::i64;
@@ -11243,3 +10772,178 @@ bool ARMTargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
 
   return false;
 }
+
+/// \brief Returns true if it is beneficial to convert a load of a constant
+/// to just the constant itself.
+bool ARMTargetLowering::shouldConvertConstantLoadToIntImm(const APInt &Imm,
+                                                          Type *Ty) const {
+  assert(Ty->isIntegerTy());
+
+  unsigned Bits = Ty->getPrimitiveSizeInBits();
+  if (Bits == 0 || Bits > 32)
+    return false;
+  return true;
+}
+
+bool ARMTargetLowering::shouldExpandAtomicInIR(Instruction *Inst) const {
+  // Loads and stores less than 64-bits are already atomic; ones above that
+  // are doomed anyway, so defer to the default libcall and blame the OS when
+  // things go wrong. Cortex M doesn't have ldrexd/strexd though, so don't emit
+  // anything for those.
+  bool IsMClass = Subtarget->isMClass();
+  if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
+    unsigned Size = SI->getValueOperand()->getType()->getPrimitiveSizeInBits();
+    return Size == 64 && !IsMClass;
+  } else if (LoadInst *LI = dyn_cast<LoadInst>(Inst)) {
+    return LI->getType()->getPrimitiveSizeInBits() == 64 && !IsMClass;
+  }
+
+  // For the real atomic operations, we have ldrex/strex up to 32 bits,
+  // and up to 64 bits on the non-M profiles
+  unsigned AtomicLimit = IsMClass ? 32 : 64;
+  return Inst->getType()->getPrimitiveSizeInBits() <= AtomicLimit;
+}
+
+Value *ARMTargetLowering::emitLoadLinked(IRBuilder<> &Builder, Value *Addr,
+                                         AtomicOrdering Ord) const {
+  Module *M = Builder.GetInsertBlock()->getParent()->getParent();
+  Type *ValTy = cast<PointerType>(Addr->getType())->getElementType();
+  bool IsAcquire =
+      Ord == Acquire || Ord == AcquireRelease || Ord == SequentiallyConsistent;
+
+  // Since i64 isn't legal and intrinsics don't get type-lowered, the ldrexd
+  // intrinsic must return {i32, i32} and we have to recombine them into a
+  // single i64 here.
+  if (ValTy->getPrimitiveSizeInBits() == 64) {
+    Intrinsic::ID Int =
+        IsAcquire ? Intrinsic::arm_ldaexd : Intrinsic::arm_ldrexd;
+    Function *Ldrex = llvm::Intrinsic::getDeclaration(M, Int);
+
+    Addr = Builder.CreateBitCast(Addr, Type::getInt8PtrTy(M->getContext()));
+    Value *LoHi = Builder.CreateCall(Ldrex, Addr, "lohi");
+
+    Value *Lo = Builder.CreateExtractValue(LoHi, 0, "lo");
+    Value *Hi = Builder.CreateExtractValue(LoHi, 1, "hi");
+    if (!Subtarget->isLittle())
+      std::swap (Lo, Hi);
+    Lo = Builder.CreateZExt(Lo, ValTy, "lo64");
+    Hi = Builder.CreateZExt(Hi, ValTy, "hi64");
+    return Builder.CreateOr(
+        Lo, Builder.CreateShl(Hi, ConstantInt::get(ValTy, 32)), "val64");
+  }
+
+  Type *Tys[] = { Addr->getType() };
+  Intrinsic::ID Int = IsAcquire ? Intrinsic::arm_ldaex : Intrinsic::arm_ldrex;
+  Function *Ldrex = llvm::Intrinsic::getDeclaration(M, Int, Tys);
+
+  return Builder.CreateTruncOrBitCast(
+      Builder.CreateCall(Ldrex, Addr),
+      cast<PointerType>(Addr->getType())->getElementType());
+}
+
+Value *ARMTargetLowering::emitStoreConditional(IRBuilder<> &Builder, Value *Val,
+                                               Value *Addr,
+                                               AtomicOrdering Ord) const {
+  Module *M = Builder.GetInsertBlock()->getParent()->getParent();
+  bool IsRelease =
+      Ord == Release || Ord == AcquireRelease || Ord == SequentiallyConsistent;
+
+  // Since the intrinsics must have legal type, the i64 intrinsics take two
+  // parameters: "i32, i32". We must marshal Val into the appropriate form
+  // before the call.
+  if (Val->getType()->getPrimitiveSizeInBits() == 64) {
+    Intrinsic::ID Int =
+        IsRelease ? Intrinsic::arm_stlexd : Intrinsic::arm_strexd;
+    Function *Strex = Intrinsic::getDeclaration(M, Int);
+    Type *Int32Ty = Type::getInt32Ty(M->getContext());
+
+    Value *Lo = Builder.CreateTrunc(Val, Int32Ty, "lo");
+    Value *Hi = Builder.CreateTrunc(Builder.CreateLShr(Val, 32), Int32Ty, "hi");
+    if (!Subtarget->isLittle())
+      std::swap (Lo, Hi);
+    Addr = Builder.CreateBitCast(Addr, Type::getInt8PtrTy(M->getContext()));
+    return Builder.CreateCall3(Strex, Lo, Hi, Addr);
+  }
+
+  Intrinsic::ID Int = IsRelease ? Intrinsic::arm_stlex : Intrinsic::arm_strex;
+  Type *Tys[] = { Addr->getType() };
+  Function *Strex = Intrinsic::getDeclaration(M, Int, Tys);
+
+  return Builder.CreateCall2(
+      Strex, Builder.CreateZExtOrBitCast(
+                 Val, Strex->getFunctionType()->getParamType(0)),
+      Addr);
+}
+
+enum HABaseType {
+  HA_UNKNOWN = 0,
+  HA_FLOAT,
+  HA_DOUBLE,
+  HA_VECT64,
+  HA_VECT128
+};
+
+static bool isHomogeneousAggregate(Type *Ty, HABaseType &Base,
+                                   uint64_t &Members) {
+  if (const StructType *ST = dyn_cast<StructType>(Ty)) {
+    for (unsigned i = 0; i < ST->getNumElements(); ++i) {
+      uint64_t SubMembers = 0;
+      if (!isHomogeneousAggregate(ST->getElementType(i), Base, SubMembers))
+        return false;
+      Members += SubMembers;
+    }
+  } else if (const ArrayType *AT = dyn_cast<ArrayType>(Ty)) {
+    uint64_t SubMembers = 0;
+    if (!isHomogeneousAggregate(AT->getElementType(), Base, SubMembers))
+      return false;
+    Members += SubMembers * AT->getNumElements();
+  } else if (Ty->isFloatTy()) {
+    if (Base != HA_UNKNOWN && Base != HA_FLOAT)
+      return false;
+    Members = 1;
+    Base = HA_FLOAT;
+  } else if (Ty->isDoubleTy()) {
+    if (Base != HA_UNKNOWN && Base != HA_DOUBLE)
+      return false;
+    Members = 1;
+    Base = HA_DOUBLE;
+  } else if (const VectorType *VT = dyn_cast<VectorType>(Ty)) {
+    Members = 1;
+    switch (Base) {
+    case HA_FLOAT:
+    case HA_DOUBLE:
+      return false;
+    case HA_VECT64:
+      return VT->getBitWidth() == 64;
+    case HA_VECT128:
+      return VT->getBitWidth() == 128;
+    case HA_UNKNOWN:
+      switch (VT->getBitWidth()) {
+      case 64:
+        Base = HA_VECT64;
+        return true;
+      case 128:
+        Base = HA_VECT128;
+        return true;
+      default:
+        return false;
+      }
+    }
+  }
+
+  return (Members > 0 && Members <= 4);
+}
+
+/// \brief Return true if a type is an AAPCS-VFP homogeneous aggregate.
+bool ARMTargetLowering::functionArgumentNeedsConsecutiveRegisters(
+    Type *Ty, CallingConv::ID CallConv, bool isVarArg) const {
+  if (getEffectiveCallingConv(CallConv, isVarArg) !=
+      CallingConv::ARM_AAPCS_VFP)
+    return false;
+
+  HABaseType Base = HA_UNKNOWN;
+  uint64_t Members = 0;
+  bool result = isHomogeneousAggregate(Ty, Base, Members);
+  DEBUG(dbgs() << "isHA: " << result << " "; Ty->dump(); dbgs() << "\n");
+  return result;
+}
diff --git a/contrib/llvm/lib/Target/ARM/ARMISelLowering.h b/contrib/llvm/lib/Target/ARM/ARMISelLowering.h
index 90facddeb02b..1ace0f330f1a 100644
--- a/contrib/llvm/lib/Target/ARM/ARMISelLowering.h
+++ b/contrib/llvm/lib/Target/ARM/ARMISelLowering.h
@@ -15,17 +15,15 @@
 #ifndef ARMISELLOWERING_H
 #define ARMISELLOWERING_H
 
-#include "ARM.h"
-#include "ARMSubtarget.h"
+#include "MCTargetDesc/ARMBaseInfo.h"
 #include "llvm/CodeGen/CallingConvLower.h"
-#include "llvm/CodeGen/FastISel.h"
 #include "llvm/CodeGen/SelectionDAG.h"
 #include "llvm/Target/TargetLowering.h"
-#include "llvm/Target/TargetRegisterInfo.h"
 #include <vector>
 
 namespace llvm {
   class ARMConstantPoolValue;
+  class ARMSubtarget;
 
   namespace ARMISD {
     // ARM Specific DAG Nodes
@@ -35,8 +33,6 @@ namespace llvm {
 
       Wrapper,      // Wrapper - A wrapper node for TargetConstantPool,
                     // TargetExternalSymbol, and TargetGlobalAddress.
-      WrapperDYN,   // WrapperDYN - A wrapper node for TargetGlobalAddress in
-                    // DYN mode.
       WrapperPIC,   // WrapperPIC - A wrapper node for TargetGlobalAddress in
                     // PIC mode.
       WrapperJT,    // WrapperJT - A wrapper node for TargetJumpTable
@@ -99,6 +95,8 @@ namespace llvm {
 
       PRELOAD,      // Preload
 
+      WIN__CHKSTK,  // Windows' __chkstk call to do stack probing.
+
       VCEQ,         // Vector compare equal.
       VCEQZ,        // Vector compare equal to zero.
       VCGE,         // Vector compare greater than or equal.
@@ -115,10 +113,6 @@ namespace llvm {
       VSHL,         // ...left
       VSHRs,        // ...right (signed)
       VSHRu,        // ...right (unsigned)
-      VSHLLs,       // ...left long (signed)
-      VSHLLu,       // ...left long (unsigned)
-      VSHLLi,       // ...left long (with maximum shift count)
-      VSHRN,        // ...right narrow
 
       // Vector rounding shift by immediate:
       VRSHRs,       // ...right (signed)
@@ -240,116 +234,114 @@ namespace llvm {
   public:
     explicit ARMTargetLowering(TargetMachine &TM);
 
-    virtual unsigned getJumpTableEncoding() const;
+    unsigned getJumpTableEncoding() const override;
 
-    virtual SDValue LowerOperation(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerOperation(SDValue Op, SelectionDAG &DAG) const override;
 
     /// ReplaceNodeResults - Replace the results of node with an illegal result
     /// type with new values built out of custom code.
     ///
-    virtual void ReplaceNodeResults(SDNode *N, SmallVectorImpl<SDValue>&Results,
-                                    SelectionDAG &DAG) const;
+    void ReplaceNodeResults(SDNode *N, SmallVectorImpl<SDValue>&Results,
+                            SelectionDAG &DAG) const override;
 
-    virtual const char *getTargetNodeName(unsigned Opcode) const;
+    const char *getTargetNodeName(unsigned Opcode) const override;
 
-    virtual bool isSelectSupported(SelectSupportKind Kind) const {
+    bool isSelectSupported(SelectSupportKind Kind) const override {
       // ARM does not support scalar condition selects on vectors.
       return (Kind != ScalarCondVectorVal);
     }
 
     /// getSetCCResultType - Return the value type to use for ISD::SETCC.
-    virtual EVT getSetCCResultType(LLVMContext &Context, EVT VT) const;
+    EVT getSetCCResultType(LLVMContext &Context, EVT VT) const override;
 
-    virtual MachineBasicBlock *
+    MachineBasicBlock *
       EmitInstrWithCustomInserter(MachineInstr *MI,
-                                  MachineBasicBlock *MBB) const;
+                                  MachineBasicBlock *MBB) const override;
 
-    virtual void
-    AdjustInstrPostInstrSelection(MachineInstr *MI, SDNode *Node) const;
+    void AdjustInstrPostInstrSelection(MachineInstr *MI,
+                                       SDNode *Node) const override;
 
     SDValue PerformCMOVCombine(SDNode *N, SelectionDAG &DAG) const;
-    virtual SDValue PerformDAGCombine(SDNode *N, DAGCombinerInfo &DCI) const;
+    SDValue PerformDAGCombine(SDNode *N, DAGCombinerInfo &DCI) const override;
 
-    bool isDesirableToTransformToIntegerOp(unsigned Opc, EVT VT) const;
+    bool isDesirableToTransformToIntegerOp(unsigned Opc, EVT VT) const override;
 
     /// allowsUnalignedMemoryAccesses - Returns true if the target allows
     /// unaligned memory accesses of the specified type. Returns whether it
     /// is "fast" by reference in the second argument.
-    virtual bool allowsUnalignedMemoryAccesses(EVT VT, bool *Fast) const;
+    bool allowsUnalignedMemoryAccesses(EVT VT, unsigned AddrSpace,
+                                       bool *Fast) const override;
 
-    virtual EVT getOptimalMemOpType(uint64_t Size,
-                                    unsigned DstAlign, unsigned SrcAlign,
-                                    bool IsMemset, bool ZeroMemset,
-                                    bool MemcpyStrSrc,
-                                    MachineFunction &MF) const;
+    EVT getOptimalMemOpType(uint64_t Size,
+                            unsigned DstAlign, unsigned SrcAlign,
+                            bool IsMemset, bool ZeroMemset,
+                            bool MemcpyStrSrc,
+                            MachineFunction &MF) const override;
 
     using TargetLowering::isZExtFree;
-    virtual bool isZExtFree(SDValue Val, EVT VT2) const;
+    bool isZExtFree(SDValue Val, EVT VT2) const override;
 
-    virtual bool allowTruncateForTailCall(Type *Ty1, Type *Ty2) const;
+    bool allowTruncateForTailCall(Type *Ty1, Type *Ty2) const override;
 
 
     /// isLegalAddressingMode - Return true if the addressing mode represented
     /// by AM is legal for this target, for a load/store of the specified type.
-    virtual bool isLegalAddressingMode(const AddrMode &AM, Type *Ty)const;
+    bool isLegalAddressingMode(const AddrMode &AM, Type *Ty) const override;
     bool isLegalT2ScaledAddressingMode(const AddrMode &AM, EVT VT) const;
 
     /// isLegalICmpImmediate - Return true if the specified immediate is legal
     /// icmp immediate, that is the target has icmp instructions which can
     /// compare a register against the immediate without having to materialize
     /// the immediate into a register.
-    virtual bool isLegalICmpImmediate(int64_t Imm) const;
+    bool isLegalICmpImmediate(int64_t Imm) const override;
 
     /// isLegalAddImmediate - Return true if the specified immediate is legal
     /// add immediate, that is the target has add instructions which can
     /// add a register and the immediate without having to materialize
     /// the immediate into a register.
-    virtual bool isLegalAddImmediate(int64_t Imm) const;
+    bool isLegalAddImmediate(int64_t Imm) const override;
 
     /// getPreIndexedAddressParts - returns true by value, base pointer and
     /// offset pointer and addressing mode by reference if the node's address
     /// can be legally represented as pre-indexed load / store address.
-    virtual bool getPreIndexedAddressParts(SDNode *N, SDValue &Base,
-                                           SDValue &Offset,
-                                           ISD::MemIndexedMode &AM,
-                                           SelectionDAG &DAG) const;
+    bool getPreIndexedAddressParts(SDNode *N, SDValue &Base, SDValue &Offset,
+                                   ISD::MemIndexedMode &AM,
+                                   SelectionDAG &DAG) const override;
 
     /// getPostIndexedAddressParts - returns true by value, base pointer and
     /// offset pointer and addressing mode by reference if this node can be
     /// combined with a load / store to form a post-indexed load / store.
-    virtual bool getPostIndexedAddressParts(SDNode *N, SDNode *Op,
-                                            SDValue &Base, SDValue &Offset,
-                                            ISD::MemIndexedMode &AM,
-                                            SelectionDAG &DAG) const;
+    bool getPostIndexedAddressParts(SDNode *N, SDNode *Op, SDValue &Base,
+                                    SDValue &Offset, ISD::MemIndexedMode &AM,
+                                    SelectionDAG &DAG) const override;
 
-    virtual void computeMaskedBitsForTargetNode(const SDValue Op,
-                                                APInt &KnownZero,
-                                                APInt &KnownOne,
-                                                const SelectionDAG &DAG,
-                                                unsigned Depth) const;
+    void computeKnownBitsForTargetNode(const SDValue Op, APInt &KnownZero,
+                                       APInt &KnownOne,
+                                       const SelectionDAG &DAG,
+                                       unsigned Depth) const override;
 
 
-    virtual bool ExpandInlineAsm(CallInst *CI) const;
+    bool ExpandInlineAsm(CallInst *CI) const override;
 
-    ConstraintType getConstraintType(const std::string &Constraint) const;
+    ConstraintType
+      getConstraintType(const std::string &Constraint) const override;
 
     /// Examine constraint string and operand type and determine a weight value.
     /// The operand object must already have been set up with the operand type.
     ConstraintWeight getSingleConstraintMatchWeight(
-      AsmOperandInfo &info, const char *constraint) const;
+      AsmOperandInfo &info, const char *constraint) const override;
 
     std::pair<unsigned, const TargetRegisterClass*>
       getRegForInlineAsmConstraint(const std::string &Constraint,
-                                   MVT VT) const;
+                                   MVT VT) const override;
 
     /// LowerAsmOperandForConstraint - Lower the specified operand into the Ops
     /// vector.  If it is invalid, don't add anything to Ops. If hasMemory is
     /// true it means one of the asm constraint of the inline asm instruction
     /// being processed is 'm'.
-    virtual void LowerAsmOperandForConstraint(SDValue Op,
-                                              std::string &Constraint,
-                                              std::vector<SDValue> &Ops,
-                                              SelectionDAG &DAG) const;
+    void LowerAsmOperandForConstraint(SDValue Op, std::string &Constraint,
+                                      std::vector<SDValue> &Ops,
+                                      SelectionDAG &DAG) const override;
 
     const ARMSubtarget* getSubtarget() const {
       return Subtarget;
@@ -357,39 +349,58 @@ namespace llvm {
 
     /// getRegClassFor - Return the register class that should be used for the
     /// specified value type.
-    virtual const TargetRegisterClass *getRegClassFor(MVT VT) const;
+    const TargetRegisterClass *getRegClassFor(MVT VT) const override;
 
     /// getMaximalGlobalOffset - Returns the maximal possible offset which can
     /// be used for loads / stores from the global.
-    virtual unsigned getMaximalGlobalOffset() const;
+    unsigned getMaximalGlobalOffset() const override;
 
     /// Returns true if a cast between SrcAS and DestAS is a noop.
-    virtual bool isNoopAddrSpaceCast(unsigned SrcAS, unsigned DestAS) const {
+    bool isNoopAddrSpaceCast(unsigned SrcAS, unsigned DestAS) const override {
       // Addrspacecasts are always noops.
       return true;
     }
 
     /// createFastISel - This method returns a target specific FastISel object,
     /// or null if the target does not support "fast" ISel.
-    virtual FastISel *createFastISel(FunctionLoweringInfo &funcInfo,
-                                     const TargetLibraryInfo *libInfo) const;
+    FastISel *createFastISel(FunctionLoweringInfo &funcInfo,
+                             const TargetLibraryInfo *libInfo) const override;
 
-    Sched::Preference getSchedulingPreference(SDNode *N) const;
+    Sched::Preference getSchedulingPreference(SDNode *N) const override;
 
-    bool isShuffleMaskLegal(const SmallVectorImpl<int> &M, EVT VT) const;
-    bool isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const;
+    bool
+    isShuffleMaskLegal(const SmallVectorImpl<int> &M, EVT VT) const override;
+    bool isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const override;
 
     /// isFPImmLegal - Returns true if the target can instruction select the
     /// specified FP immediate natively. If false, the legalizer will
     /// materialize the FP immediate as a load from a constant pool.
-    virtual bool isFPImmLegal(const APFloat &Imm, EVT VT) const;
+    bool isFPImmLegal(const APFloat &Imm, EVT VT) const override;
+
+    bool getTgtMemIntrinsic(IntrinsicInfo &Info,
+                            const CallInst &I,
+                            unsigned Intrinsic) const override;
+
+    /// \brief Returns true if it is beneficial to convert a load of a constant
+    /// to just the constant itself.
+    bool shouldConvertConstantLoadToIntImm(const APInt &Imm,
+                                           Type *Ty) const override;
+
+    /// \brief Returns true if an argument of type Ty needs to be passed in a
+    /// contiguous block of registers in calling convention CallConv.
+    bool functionArgumentNeedsConsecutiveRegisters(
+        Type *Ty, CallingConv::ID CallConv, bool isVarArg) const override;
+
+    Value *emitLoadLinked(IRBuilder<> &Builder, Value *Addr,
+                          AtomicOrdering Ord) const override;
+    Value *emitStoreConditional(IRBuilder<> &Builder, Value *Val,
+                                Value *Addr, AtomicOrdering Ord) const override;
+
+    bool shouldExpandAtomicInIR(Instruction *Inst) const override;
 
-    virtual bool getTgtMemIntrinsic(IntrinsicInfo &Info,
-                                    const CallInst &I,
-                                    unsigned Intrinsic) const;
   protected:
     std::pair<const TargetRegisterClass*, uint8_t>
-    findRepresentativeClass(MVT VT) const;
+    findRepresentativeClass(MVT VT) const override;
 
   private:
     /// Subtarget - Keep a pointer to the ARMSubtarget around so that we can
@@ -407,6 +418,7 @@ namespace llvm {
     void addTypeForNEON(MVT VT, MVT PromotedLdStVT, MVT PromotedBitwiseVT);
     void addDRTypeForNEON(MVT VT);
     void addQRTypeForNEON(MVT VT);
+    std::pair<SDValue, SDValue> getARMXALUOOp(SDValue Op, SelectionDAG &DAG, SDValue &ARMcc) const;
 
     typedef SmallVector<std::pair<unsigned, SDValue>, 8> RegsToPassVector;
     void PassF64ArgInRegs(SDLoc dl, SelectionDAG &DAG,
@@ -420,6 +432,8 @@ namespace llvm {
                                  SDValue &Root, SelectionDAG &DAG,
                                  SDLoc dl) const;
 
+    CallingConv::ID getEffectiveCallingConv(CallingConv::ID CC,
+                                            bool isVarArg) const;
     CCAssignFn *CCAssignFnForNode(CallingConv::ID CC, bool Return,
                                   bool isVarArg) const;
     SDValue LowerMemOpCallTo(SDValue Chain, SDValue StackPtr, SDValue Arg,
@@ -433,6 +447,7 @@ namespace llvm {
     SDValue LowerBlockAddress(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerGlobalAddressDarwin(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerGlobalAddressELF(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerGlobalAddressWindows(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerToTLSGeneralDynamicModel(GlobalAddressSDNode *GA,
                                             SelectionDAG &DAG) const;
@@ -441,6 +456,7 @@ namespace llvm {
                                  TLSModel::Model model) const;
     SDValue LowerGLOBAL_OFFSET_TABLE(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerBR_JT(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerXALUO(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerSELECT(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerSELECT_CC(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerBR_CC(SDValue Op, SelectionDAG &DAG) const;
@@ -456,6 +472,9 @@ namespace llvm {
                               const ARMSubtarget *ST) const;
     SDValue LowerFSINCOS(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerDivRem(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG) const;
+
+    unsigned getRegisterByName(const char* RegName, EVT VT) const override;
 
     /// isFMAFasterThanFMulAndFAdd - Return true if an FMA operation is faster
     /// than a pair of fmul and fadd instructions. fmuladd intrinsics will be
@@ -466,7 +485,7 @@ namespace llvm {
     /// lower a pair of fmul and fadd to the latter so it's not clear that there
     /// would be a gain or that the gain would be worthwhile enough to risk
     /// correctness bugs.
-    virtual bool isFMAFasterThanFMulAndFAdd(EVT VT) const { return false; }
+    bool isFMAFasterThanFMulAndFAdd(EVT VT) const override { return false; }
 
     SDValue ReconstructShuffle(SDValue Op, SelectionDAG &DAG) const;
 
@@ -477,12 +496,12 @@ namespace llvm {
                             SmallVectorImpl<SDValue> &InVals,
                             bool isThisReturn, SDValue ThisVal) const;
 
-    virtual SDValue
+    SDValue
       LowerFormalArguments(SDValue Chain,
                            CallingConv::ID CallConv, bool isVarArg,
                            const SmallVectorImpl<ISD::InputArg> &Ins,
                            SDLoc dl, SelectionDAG &DAG,
-                           SmallVectorImpl<SDValue> &InVals) const;
+                           SmallVectorImpl<SDValue> &InVals) const override;
 
     int StoreByValRegs(CCState &CCInfo, SelectionDAG &DAG,
                        SDLoc dl, SDValue &Chain,
@@ -491,11 +510,14 @@ namespace llvm {
                        unsigned OffsetFromOrigArg,
                        unsigned ArgOffset,
                        unsigned ArgSize,
-                       bool ForceMutable) const;
+                       bool ForceMutable,
+                       unsigned ByValStoreOffset,
+                       unsigned TotalArgRegsSaveSize) const;
 
     void VarArgStyleRegisters(CCState &CCInfo, SelectionDAG &DAG,
                               SDLoc dl, SDValue &Chain,
                               unsigned ArgOffset,
+                              unsigned TotalArgRegsSaveSize,
                               bool ForceMutable = false) const;
 
     void computeRegArea(CCState &CCInfo, MachineFunction &MF,
@@ -504,12 +526,12 @@ namespace llvm {
                         unsigned &ArgRegsSize,
                         unsigned &ArgRegsSaveSize) const;
 
-    virtual SDValue
+    SDValue
       LowerCall(TargetLowering::CallLoweringInfo &CLI,
-                SmallVectorImpl<SDValue> &InVals) const;
+                SmallVectorImpl<SDValue> &InVals) const override;
 
     /// HandleByVal - Target-specific cleanup for ByVal support.
-    virtual void HandleByVal(CCState *, unsigned &, unsigned) const;
+    void HandleByVal(CCState *, unsigned &, unsigned) const override;
 
     /// IsEligibleForTailCallOptimization - Check whether the call is eligible
     /// for tail call optimization. Targets which want to do tail call
@@ -524,21 +546,21 @@ namespace llvm {
                                     const SmallVectorImpl<ISD::InputArg> &Ins,
                                            SelectionDAG& DAG) const;
 
-    virtual bool CanLowerReturn(CallingConv::ID CallConv,
-                                MachineFunction &MF, bool isVarArg,
-                                const SmallVectorImpl<ISD::OutputArg> &Outs,
-                                LLVMContext &Context) const;
+    bool CanLowerReturn(CallingConv::ID CallConv,
+                        MachineFunction &MF, bool isVarArg,
+                        const SmallVectorImpl<ISD::OutputArg> &Outs,
+                        LLVMContext &Context) const override;
 
-    virtual SDValue
+    SDValue
       LowerReturn(SDValue Chain,
                   CallingConv::ID CallConv, bool isVarArg,
                   const SmallVectorImpl<ISD::OutputArg> &Outs,
                   const SmallVectorImpl<SDValue> &OutVals,
-                  SDLoc dl, SelectionDAG &DAG) const;
+                  SDLoc dl, SelectionDAG &DAG) const override;
 
-    virtual bool isUsedByReturnOnly(SDNode *N, SDValue &Chain) const;
+    bool isUsedByReturnOnly(SDNode *N, SDValue &Chain) const override;
 
-    virtual bool mayBeEmittedAsTailCall(CallInst *CI) const;
+    bool mayBeEmittedAsTailCall(CallInst *CI) const override;
 
     SDValue getARMCmp(SDValue LHS, SDValue RHS, ISD::CondCode CC,
                       SDValue &ARMcc, SelectionDAG &DAG, SDLoc dl) const;
@@ -548,29 +570,6 @@ namespace llvm {
 
     SDValue OptimizeVFPBrcond(SDValue Op, SelectionDAG &DAG) const;
 
-    MachineBasicBlock *EmitAtomicCmpSwap(MachineInstr *MI,
-                                         MachineBasicBlock *BB,
-                                         unsigned Size) const;
-    MachineBasicBlock *EmitAtomicBinary(MachineInstr *MI,
-                                        MachineBasicBlock *BB,
-                                        unsigned Size,
-                                        unsigned BinOpcode) const;
-    MachineBasicBlock *EmitAtomicBinary64(MachineInstr *MI,
-                                          MachineBasicBlock *BB,
-                                          unsigned Op1,
-                                          unsigned Op2,
-                                          bool NeedsCarry = false,
-                                          bool IsCmpxchg = false,
-                                          bool IsMinMax = false,
-                                          ARMCC::CondCodes CC = ARMCC::AL) const;
-    MachineBasicBlock * EmitAtomicBinaryMinMax(MachineInstr *MI,
-                                               MachineBasicBlock *BB,
-                                               unsigned Size,
-                                               bool signExtend,
-                                               ARMCC::CondCodes Cond) const;
-    MachineBasicBlock *EmitAtomicLoad64(MachineInstr *MI,
-                                        MachineBasicBlock *BB) const;
-
     void SetupEntryBlockForSjLj(MachineInstr *MI,
                                 MachineBasicBlock *MBB,
                                 MachineBasicBlock *DispatchBB, int FI) const;
@@ -582,6 +581,9 @@ namespace llvm {
 
     MachineBasicBlock *EmitStructByval(MachineInstr *MI,
                                        MachineBasicBlock *MBB) const;
+
+    MachineBasicBlock *EmitLowered__chkstk(MachineInstr *MI,
+                                           MachineBasicBlock *MBB) const;
   };
 
   enum NEONModImmType {
@@ -590,7 +592,6 @@ namespace llvm {
     OtherModImm
   };
 
-
   namespace ARM {
     FastISel *createFastISel(FunctionLoweringInfo &funcInfo,
                              const TargetLibraryInfo *libInfo);
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrFormats.td b/contrib/llvm/lib/Target/ARM/ARMInstrFormats.td
index f93504fddb8e..59e9260b81ea 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrFormats.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrFormats.td
@@ -212,25 +212,25 @@ def msr_mask : Operand<i32> {
 //     32       imm6<5> = '1', 32 - <imm> is encoded in imm6<4:0>
 //     64       64 - <imm> is encoded in imm6<5:0>
 def shr_imm8_asm_operand : ImmAsmOperand { let Name = "ShrImm8"; }
-def shr_imm8  : Operand<i32> {
+def shr_imm8  : Operand<i32>, ImmLeaf<i32, [{ return Imm > 0 && Imm <= 8; }]> {
   let EncoderMethod = "getShiftRight8Imm";
   let DecoderMethod = "DecodeShiftRight8Imm";
   let ParserMatchClass = shr_imm8_asm_operand;
 }
 def shr_imm16_asm_operand : ImmAsmOperand { let Name = "ShrImm16"; }
-def shr_imm16 : Operand<i32> {
+def shr_imm16 : Operand<i32>, ImmLeaf<i32, [{ return Imm > 0 && Imm <= 16; }]> {
   let EncoderMethod = "getShiftRight16Imm";
   let DecoderMethod = "DecodeShiftRight16Imm";
   let ParserMatchClass = shr_imm16_asm_operand;
 }
 def shr_imm32_asm_operand : ImmAsmOperand { let Name = "ShrImm32"; }
-def shr_imm32 : Operand<i32> {
+def shr_imm32 : Operand<i32>, ImmLeaf<i32, [{ return Imm > 0 && Imm <= 32; }]> {
   let EncoderMethod = "getShiftRight32Imm";
   let DecoderMethod = "DecodeShiftRight32Imm";
   let ParserMatchClass = shr_imm32_asm_operand;
 }
 def shr_imm64_asm_operand : ImmAsmOperand { let Name = "ShrImm64"; }
-def shr_imm64 : Operand<i32> {
+def shr_imm64 : Operand<i32>, ImmLeaf<i32, [{ return Imm > 0 && Imm <= 64; }]> {
   let EncoderMethod = "getShiftRight64Imm";
   let DecoderMethod = "DecodeShiftRight64Imm";
   let ParserMatchClass = shr_imm64_asm_operand;
@@ -329,10 +329,10 @@ class InstThumb<AddrMode am, int sz, IndexMode im,
 // Pseudo-instructions for alternate assembly syntax (never used by codegen).
 // These are aliases that require C++ handling to convert to the target
 // instruction, while InstAliases can be handled directly by tblgen.
-class AsmPseudoInst<string asm, dag iops>
+class AsmPseudoInst<string asm, dag iops, dag oops = (outs)>
   : InstTemplate<AddrModeNone, 0, IndexModeNone, Pseudo, GenericDomain,
                  "", NoItinerary> {
-  let OutOperandList = (outs);
+  let OutOperandList = oops;
   let InOperandList = iops;
   let Pattern = [];
   let isCodeGenOnly = 0; // So we get asm matcher for it.
@@ -340,16 +340,16 @@ class AsmPseudoInst<string asm, dag iops>
   let isPseudo = 1;
 }
 
-class ARMAsmPseudo<string asm, dag iops> : AsmPseudoInst<asm, iops>,
-        Requires<[IsARM]>;
-class tAsmPseudo<string asm, dag iops> : AsmPseudoInst<asm, iops>,
-        Requires<[IsThumb]>;
-class t2AsmPseudo<string asm, dag iops> : AsmPseudoInst<asm, iops>,
-        Requires<[IsThumb2]>;
-class VFP2AsmPseudo<string asm, dag iops> : AsmPseudoInst<asm, iops>,
-        Requires<[HasVFP2]>;
-class NEONAsmPseudo<string asm, dag iops> : AsmPseudoInst<asm, iops>,
-        Requires<[HasNEON]>;
+class ARMAsmPseudo<string asm, dag iops, dag oops = (outs)>
+  : AsmPseudoInst<asm, iops, oops>, Requires<[IsARM]>;
+class tAsmPseudo<string asm, dag iops, dag oops = (outs)>
+  : AsmPseudoInst<asm, iops, oops>, Requires<[IsThumb]>;
+class t2AsmPseudo<string asm, dag iops, dag oops = (outs)>
+  : AsmPseudoInst<asm, iops, oops>, Requires<[IsThumb2]>;
+class VFP2AsmPseudo<string asm, dag iops, dag oops = (outs)>
+  : AsmPseudoInst<asm, iops, oops>, Requires<[HasVFP2]>;
+class NEONAsmPseudo<string asm, dag iops, dag oops = (outs)>
+  : AsmPseudoInst<asm, iops, oops>, Requires<[HasNEON]>;
 
 // Pseudo instructions for the code generator.
 class PseudoInst<dag oops, dag iops, InstrItinClass itin, list<dag> pattern>
@@ -477,6 +477,10 @@ class AXI<dag oops, dag iops, Format f, InstrItinClass itin,
           string asm, list<dag> pattern>
   : XI<oops, iops, AddrModeNone, 4, IndexModeNone, f, itin,
        asm, "", pattern>;
+class AXIM<dag oops, dag iops, AddrMode am, Format f, InstrItinClass itin,
+          string asm, list<dag> pattern>
+  : XI<oops, iops, am, 4, IndexModeNone, f, itin,
+       asm, "", pattern>;
 class AInoP<dag oops, dag iops, Format f, InstrItinClass itin,
             string opc, string asm, list<dag> pattern>
   : InoP<oops, iops, AddrModeNone, 4, IndexModeNone, f, itin,
@@ -2025,7 +2029,7 @@ class N2V<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18, bits<2> op17_16,
 // Same as N2V but not predicated.
 class N2Vnp<bits<2> op19_18, bits<2> op17_16, bits<3> op10_8, bit op7, bit op6,
             dag oops, dag iops, InstrItinClass itin, string OpcodeStr,
-            string Dt, ValueType ResTy, ValueType OpTy, list<dag> pattern>
+            string Dt, list<dag> pattern>
    : NeonInp<oops, iops, AddrModeNone, IndexModeNone, N2RegFrm, itin,
              OpcodeStr, Dt, "$Vd, $Vm", "", pattern> {
   bits<5> Vd;
@@ -2134,8 +2138,7 @@ class N3V<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op6, bit op4,
 
 class N3Vnp<bits<5> op27_23, bits<2> op21_20, bits<4> op11_8, bit op6,
                 bit op4, dag oops, dag iops,Format f, InstrItinClass itin,
-                string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy,
-                SDPatternOperator IntOp, bit Commutable, list<dag> pattern>
+                string OpcodeStr, string Dt, list<dag> pattern>
   : NeonInp<oops, iops, AddrModeNone, IndexModeNone, f, itin, OpcodeStr,
             Dt, "$Vd, $Vn, $Vm", "", pattern> {
   bits<5> Vd;
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.cpp
index df867b49ab57..f235ac225848 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.cpp
@@ -97,7 +97,7 @@ namespace {
     static char ID;
     ARMCGBR() : MachineFunctionPass(ID) {}
 
-    virtual bool runOnMachineFunction(MachineFunction &MF) {
+    bool runOnMachineFunction(MachineFunction &MF) override {
       ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
       if (AFI->getGlobalBaseReg() == 0)
         return false;
@@ -146,11 +146,11 @@ namespace {
       return true;
     }
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM PIC Global Base Reg Initialization";
     }
 
-    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+    void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.setPreservesCFG();
       MachineFunctionPass::getAnalysisUsage(AU);
     }
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.h b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.h
index 5d3e059b7038..b09958aaa950 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.h
@@ -14,10 +14,8 @@
 #ifndef ARMINSTRUCTIONINFO_H
 #define ARMINSTRUCTIONINFO_H
 
-#include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMRegisterInfo.h"
-#include "ARMSubtarget.h"
 
 namespace llvm {
   class ARMSubtarget;
@@ -28,17 +26,17 @@ public:
   explicit ARMInstrInfo(const ARMSubtarget &STI);
 
   /// getNoopForMachoTarget - Return the noop instruction to use for a noop.
-  void getNoopForMachoTarget(MCInst &NopInst) const;
+  void getNoopForMachoTarget(MCInst &NopInst) const override;
 
   // Return the non-pre/post incrementing version of 'Opc'. Return 0
   // if there is not such an opcode.
-  unsigned getUnindexedOpcode(unsigned Opc) const;
+  unsigned getUnindexedOpcode(unsigned Opc) const override;
 
   /// getRegisterInfo - TargetInstrInfo is a superset of MRegister info.  As
   /// such, whenever a client has an instance of instruction info, it should
   /// always be able to get register info as well (through this method).
   ///
-  const ARMRegisterInfo &getRegisterInfo() const { return RI; }
+  const ARMRegisterInfo &getRegisterInfo() const override { return RI; }
 };
 
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td
index 7a14b8ebf11d..a02d997f3c4e 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td
@@ -95,7 +95,6 @@ def ARMSmlal         : SDNode<"ARMISD::SMLAL", SDT_ARM64bitmlal>;
 
 // Node definitions.
 def ARMWrapper       : SDNode<"ARMISD::Wrapper",     SDTIntUnaryOp>;
-def ARMWrapperDYN    : SDNode<"ARMISD::WrapperDYN",  SDTIntUnaryOp>;
 def ARMWrapperPIC    : SDNode<"ARMISD::WrapperPIC",  SDTIntUnaryOp>;
 def ARMWrapperJT     : SDNode<"ARMISD::WrapperJT",   SDTIntBinOp>;
 
@@ -187,7 +186,8 @@ def ARMvminnm        : SDNode<"ARMISD::VMINNM", SDT_ARMVMINNM, []>;
 def HasV4T           : Predicate<"Subtarget->hasV4TOps()">,
                                  AssemblerPredicate<"HasV4TOps", "armv4t">;
 def NoV4T            : Predicate<"!Subtarget->hasV4TOps()">;
-def HasV5T           : Predicate<"Subtarget->hasV5TOps()">;
+def HasV5T           : Predicate<"Subtarget->hasV5TOps()">,
+                                 AssemblerPredicate<"HasV5TOps", "armv5t">;
 def HasV5TE          : Predicate<"Subtarget->hasV5TEOps()">,
                                  AssemblerPredicate<"HasV5TEOps", "armv5te">;
 def HasV6            : Predicate<"Subtarget->hasV6Ops()">,
@@ -244,6 +244,7 @@ def HasMP            : Predicate<"Subtarget->hasMPExtension()">,
 def HasTrustZone     : Predicate<"Subtarget->hasTrustZone()">,
                                  AssemblerPredicate<"FeatureTrustZone",
                                                     "TrustZone">;
+def HasZCZ           : Predicate<"Subtarget->hasZeroCycleZeroing()">;
 def UseNEONForFP     : Predicate<"Subtarget->useNEONForSinglePrecisionFP()">;
 def DontUseNEONForFP : Predicate<"!Subtarget->useNEONForSinglePrecisionFP()">;
 def IsThumb          : Predicate<"Subtarget->isThumb()">,
@@ -261,14 +262,16 @@ def IsARM            : Predicate<"!Subtarget->isThumb()">,
                                  AssemblerPredicate<"!ModeThumb", "arm-mode">;
 def IsIOS            : Predicate<"Subtarget->isTargetIOS()">;
 def IsNotIOS         : Predicate<"!Subtarget->isTargetIOS()">;
+def IsMachO          : Predicate<"Subtarget->isTargetMachO()">;
+def IsNotMachO       : Predicate<"!Subtarget->isTargetMachO()">;
 def IsNaCl           : Predicate<"Subtarget->isTargetNaCl()">;
 def UseNaClTrap      : Predicate<"Subtarget->useNaClTrap()">,
                                  AssemblerPredicate<"FeatureNaClTrap", "NaCl">;
 def DontUseNaClTrap  : Predicate<"!Subtarget->useNaClTrap()">;
 
 // FIXME: Eventually this will be just "hasV6T2Ops".
-def UseMovt          : Predicate<"Subtarget->useMovt()">;
-def DontUseMovt      : Predicate<"!Subtarget->useMovt()">;
+def UseMovt          : Predicate<"Subtarget->useMovt(*MF)">;
+def DontUseMovt      : Predicate<"!Subtarget->useMovt(*MF)">;
 def UseFPVMLx        : Predicate<"Subtarget->useFPVMLx()">;
 def UseMulOps        : Predicate<"Subtarget->useMulOps()">;
 
@@ -276,7 +279,8 @@ def UseMulOps        : Predicate<"Subtarget->useMulOps()">;
 // But only select them if more precision in FP computation is allowed.
 // Do not use them for Darwin platforms.
 def UseFusedMAC      : Predicate<"(TM.Options.AllowFPOpFusion =="
-                                 " FPOpFusion::Fast) && "
+                                 " FPOpFusion::Fast && "
+                                 " Subtarget->hasVFP4()) && "
                                  "!Subtarget->isTargetDarwin()">;
 def DontUseFusedMAC  : Predicate<"!(TM.Options.AllowFPOpFusion =="
                                  " FPOpFusion::Fast &&"
@@ -489,7 +493,7 @@ def neon_vcvt_imm32 : Operand<i32> {
 // rot_imm: An integer that encodes a rotate amount. Must be 8, 16, or 24.
 def rot_imm_XFORM: SDNodeXForm<imm, [{
   switch (N->getZExtValue()){
-  default: assert(0);
+  default: llvm_unreachable(nullptr);
   case 0:  return CurDAG->getTargetConstant(0, MVT::i32);
   case 8:  return CurDAG->getTargetConstant(1, MVT::i32);
   case 16: return CurDAG->getTargetConstant(2, MVT::i32);
@@ -590,7 +594,7 @@ def so_imm2part : PatLeaf<(imm), [{
 /// arm_i32imm - True for +V6T2, or true only if so_imm2part is true.
 ///
 def arm_i32imm : PatLeaf<(imm), [{
-  if (Subtarget->hasV6T2Ops())
+  if (Subtarget->useMovt(*MF))
     return true;
   return ARM_AM::isSOImmTwoPartVal((unsigned)N->getZExtValue());
 }]>;
@@ -987,6 +991,81 @@ def addrmode6oneL32 : Operand<i32>,
   let EncoderMethod = "getAddrMode6OneLane32AddressOpValue";
 }
 
+// Base class for addrmode6 with specific alignment restrictions.
+class AddrMode6Align : Operand<i32>,
+                ComplexPattern<i32, 2, "SelectAddrMode6", [], [SDNPWantParent]>{
+  let PrintMethod = "printAddrMode6Operand";
+  let MIOperandInfo = (ops GPR:$addr, i32imm:$align);
+  let EncoderMethod = "getAddrMode6AddressOpValue";
+  let DecoderMethod = "DecodeAddrMode6Operand";
+}
+
+// Special version of addrmode6 to handle no allowed alignment encoding for
+// VLD/VST instructions and checking the alignment is not specified.
+def AddrMode6AlignNoneAsmOperand : AsmOperandClass {
+  let Name = "AlignedMemoryNone";
+  let DiagnosticType = "AlignedMemoryRequiresNone";
+}
+def addrmode6alignNone : AddrMode6Align {
+  // The alignment specifier can only be omitted.
+  let ParserMatchClass = AddrMode6AlignNoneAsmOperand;
+}
+
+// Special version of addrmode6 to handle 16-bit alignment encoding for
+// VLD/VST instructions and checking the alignment value.
+def AddrMode6Align16AsmOperand : AsmOperandClass {
+  let Name = "AlignedMemory16";
+  let DiagnosticType = "AlignedMemoryRequires16";
+}
+def addrmode6align16 : AddrMode6Align {
+  // The alignment specifier can only be 16 or omitted.
+  let ParserMatchClass = AddrMode6Align16AsmOperand;
+}
+
+// Special version of addrmode6 to handle 32-bit alignment encoding for
+// VLD/VST instructions and checking the alignment value.
+def AddrMode6Align32AsmOperand : AsmOperandClass {
+  let Name = "AlignedMemory32";
+  let DiagnosticType = "AlignedMemoryRequires32";
+}
+def addrmode6align32 : AddrMode6Align {
+  // The alignment specifier can only be 32 or omitted.
+  let ParserMatchClass = AddrMode6Align32AsmOperand;
+}
+
+// Special version of addrmode6 to handle 64-bit alignment encoding for
+// VLD/VST instructions and checking the alignment value.
+def AddrMode6Align64AsmOperand : AsmOperandClass {
+  let Name = "AlignedMemory64";
+  let DiagnosticType = "AlignedMemoryRequires64";
+}
+def addrmode6align64 : AddrMode6Align {
+  // The alignment specifier can only be 64 or omitted.
+  let ParserMatchClass = AddrMode6Align64AsmOperand;
+}
+
+// Special version of addrmode6 to handle 64-bit or 128-bit alignment encoding
+// for VLD/VST instructions and checking the alignment value.
+def AddrMode6Align64or128AsmOperand : AsmOperandClass {
+  let Name = "AlignedMemory64or128";
+  let DiagnosticType = "AlignedMemoryRequires64or128";
+}
+def addrmode6align64or128 : AddrMode6Align {
+  // The alignment specifier can only be 64, 128 or omitted.
+  let ParserMatchClass = AddrMode6Align64or128AsmOperand;
+}
+
+// Special version of addrmode6 to handle 64-bit, 128-bit or 256-bit alignment
+// encoding for VLD/VST instructions and checking the alignment value.
+def AddrMode6Align64or128or256AsmOperand : AsmOperandClass {
+  let Name = "AlignedMemory64or128or256";
+  let DiagnosticType = "AlignedMemoryRequires64or128or256";
+}
+def addrmode6align64or128or256 : AddrMode6Align {
+  // The alignment specifier can only be 64, 128, 256 or omitted.
+  let ParserMatchClass = AddrMode6Align64or128or256AsmOperand;
+}
+
 // Special version of addrmode6 to handle alignment encoding for VLD-dup
 // instructions, specifically VLD4-dup.
 def addrmode6dup : Operand<i32>,
@@ -999,6 +1078,69 @@ def addrmode6dup : Operand<i32>,
   let ParserMatchClass = AddrMode6AsmOperand;
 }
 
+// Base class for addrmode6dup with specific alignment restrictions.
+class AddrMode6DupAlign : Operand<i32>,
+                ComplexPattern<i32, 2, "SelectAddrMode6", [], [SDNPWantParent]>{
+  let PrintMethod = "printAddrMode6Operand";
+  let MIOperandInfo = (ops GPR:$addr, i32imm);
+  let EncoderMethod = "getAddrMode6DupAddressOpValue";
+}
+
+// Special version of addrmode6 to handle no allowed alignment encoding for
+// VLD-dup instruction and checking the alignment is not specified.
+def AddrMode6dupAlignNoneAsmOperand : AsmOperandClass {
+  let Name = "DupAlignedMemoryNone";
+  let DiagnosticType = "DupAlignedMemoryRequiresNone";
+}
+def addrmode6dupalignNone : AddrMode6DupAlign {
+  // The alignment specifier can only be omitted.
+  let ParserMatchClass = AddrMode6dupAlignNoneAsmOperand;
+}
+
+// Special version of addrmode6 to handle 16-bit alignment encoding for VLD-dup
+// instruction and checking the alignment value.
+def AddrMode6dupAlign16AsmOperand : AsmOperandClass {
+  let Name = "DupAlignedMemory16";
+  let DiagnosticType = "DupAlignedMemoryRequires16";
+}
+def addrmode6dupalign16 : AddrMode6DupAlign {
+  // The alignment specifier can only be 16 or omitted.
+  let ParserMatchClass = AddrMode6dupAlign16AsmOperand;
+}
+
+// Special version of addrmode6 to handle 32-bit alignment encoding for VLD-dup
+// instruction and checking the alignment value.
+def AddrMode6dupAlign32AsmOperand : AsmOperandClass {
+  let Name = "DupAlignedMemory32";
+  let DiagnosticType = "DupAlignedMemoryRequires32";
+}
+def addrmode6dupalign32 : AddrMode6DupAlign {
+  // The alignment specifier can only be 32 or omitted.
+  let ParserMatchClass = AddrMode6dupAlign32AsmOperand;
+}
+
+// Special version of addrmode6 to handle 64-bit alignment encoding for VLD
+// instructions and checking the alignment value.
+def AddrMode6dupAlign64AsmOperand : AsmOperandClass {
+  let Name = "DupAlignedMemory64";
+  let DiagnosticType = "DupAlignedMemoryRequires64";
+}
+def addrmode6dupalign64 : AddrMode6DupAlign {
+  // The alignment specifier can only be 64 or omitted.
+  let ParserMatchClass = AddrMode6dupAlign64AsmOperand;
+}
+
+// Special version of addrmode6 to handle 64-bit or 128-bit alignment encoding
+// for VLD instructions and checking the alignment value.
+def AddrMode6dupAlign64or128AsmOperand : AsmOperandClass {
+  let Name = "DupAlignedMemory64or128";
+  let DiagnosticType = "DupAlignedMemoryRequires64or128";
+}
+def addrmode6dupalign64or128 : AddrMode6DupAlign {
+  // The alignment specifier can only be 64, 128 or omitted.
+  let ParserMatchClass = AddrMode6dupAlign64or128AsmOperand;
+}
+
 // addrmodepc := pc + reg
 //
 def addrmodepc : Operand<i32>,
@@ -1685,7 +1827,8 @@ PseudoInst<(outs), (ins i32imm:$amt, pred:$p), NoItinerary,
 }
 
 def HINT : AI<(outs), (ins imm0_239:$imm), MiscFrm, NoItinerary,
-              "hint", "\t$imm", []>, Requires<[IsARM, HasV6]> {
+              "hint", "\t$imm", [(int_arm_hint imm0_239:$imm)]>,
+           Requires<[IsARM, HasV6]> {
   bits<8> imm;
   let Inst{27-8} = 0b00110010000011110000;
   let Inst{7-0} = imm;
@@ -1698,8 +1841,6 @@ def : InstAlias<"wfi$p", (HINT 3, pred:$p)>, Requires<[IsARM, HasV6T2]>;
 def : InstAlias<"sev$p", (HINT 4, pred:$p)>, Requires<[IsARM, HasV6T2]>;
 def : InstAlias<"sevl$p", (HINT 5, pred:$p)>, Requires<[IsARM, HasV8]>;
 
-def : Pat<(int_arm_sevl), (HINT 5)>;
-
 def SEL : AI<(outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm), DPFrm, NoItinerary, "sel",
              "\t$Rd, $Rn, $Rm", []>, Requires<[IsARM, HasV6]> {
   bits<4> Rd;
@@ -1725,6 +1866,8 @@ def BKPT : AInoP<(outs), (ins imm0_65535:$val), MiscFrm, NoItinerary,
   let Inst{31-28} = 0xe; // AL
   let Inst{7-4} = 0b0111;
 }
+// default immediate for breakpoint mnemonic
+def : InstAlias<"bkpt", (BKPT 0)>, Requires<[IsARM]>;
 
 def HLT : AInoP<(outs), (ins imm0_65535:$val), MiscFrm, NoItinerary,
                  "hlt", "\t$val", []>, Requires<[IsARM, HasV8]> {
@@ -1770,8 +1913,8 @@ let imod = 0, iflags = 0, M = 1 in
 // Preload signals the memory system of possible future data/instruction access.
 multiclass APreLoad<bits<1> read, bits<1> data, string opc> {
 
-  def i12 : AXI<(outs), (ins addrmode_imm12:$addr), MiscFrm, IIC_Preload,
-                !strconcat(opc, "\t$addr"),
+  def i12 : AXIM<(outs), (ins addrmode_imm12:$addr), AddrMode_i12, MiscFrm,
+                IIC_Preload, !strconcat(opc, "\t$addr"),
                 [(ARMPreload addrmode_imm12:$addr, (i32 read), (i32 data))]>,
                 Sched<[WritePreLd]> {
     bits<4> Rt;
@@ -1824,6 +1967,18 @@ def DBG : AI<(outs), (ins imm0_15:$opt), MiscFrm, NoItinerary, "dbg", "\t$opt",
   let Inst{3-0} = opt;
 }
 
+// A8.8.247  UDF - Undefined (Encoding A1)
+def UDF : AInoP<(outs), (ins imm0_65535:$imm16), MiscFrm, NoItinerary,
+                "udf", "\t$imm16", [(int_arm_undefined imm0_65535:$imm16)]> {
+  bits<16> imm16;
+  let Inst{31-28} = 0b1110; // AL
+  let Inst{27-25} = 0b011;
+  let Inst{24-20} = 0b11111;
+  let Inst{19-8} = imm16{15-4};
+  let Inst{7-4} = 0b1111;
+  let Inst{3-0} = imm16{3-0};
+}
+
 /*
  * A5.4 Permanently UNDEFINED instructions.
  *
@@ -2272,11 +2427,10 @@ def LDRSB : AI3ld<0b1101, 1, (outs GPR:$Rt), (ins addrmode3:$addr), LdMiscFrm,
                    [(set GPR:$Rt, (sextloadi8 addrmode3:$addr))]>;
 
 let mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1 in {
-// Load doubleword
-def LDRD : AI3ld<0b1101, 0, (outs GPR:$Rd, GPR:$dst2),
-                 (ins addrmode3:$addr), LdMiscFrm,
-                 IIC_iLoad_d_r, "ldrd", "\t$Rd, $dst2, $addr",
-                 []>, Requires<[IsARM, HasV5TE]>;
+  // Load doubleword
+  def LDRD : AI3ld<0b1101, 0, (outs GPR:$Rt, GPR:$Rt2), (ins addrmode3:$addr),
+                   LdMiscFrm, IIC_iLoad_d_r, "ldrd", "\t$Rt, $Rt2, $addr", []>,
+             Requires<[IsARM, HasV5TE]>;
 }
 
 def LDA : AIldracq<0b00, (outs GPR:$Rt), (ins addr_offset_none:$addr),
@@ -2441,11 +2595,11 @@ def LDRT_POST_REG : AI2ldstidx<1, 0, 0, (outs GPR:$Rt, GPR:$Rn_wb),
   let DecoderMethod = "DecodeAddrMode2IdxInstruction";
 }
 
-def LDRT_POST_IMM : AI2ldstidx<1, 0, 0, (outs GPR:$Rt, GPR:$Rn_wb),
-                    (ins addr_offset_none:$addr, am2offset_imm:$offset),
-                   IndexModePost, LdFrm, IIC_iLoad_ru,
-                   "ldrt", "\t$Rt, $addr, $offset",
-                   "$addr.base = $Rn_wb", []> {
+def LDRT_POST_IMM
+  : AI2ldstidx<1, 0, 0, (outs GPR:$Rt, GPR:$Rn_wb),
+               (ins addr_offset_none:$addr, am2offset_imm:$offset),
+               IndexModePost, LdFrm, IIC_iLoad_ru,
+               "ldrt", "\t$Rt, $addr, $offset", "$addr.base = $Rn_wb", []> {
   // {12}     isAdd
   // {11-0}   imm12/Rm
   bits<14> offset;
@@ -2477,11 +2631,11 @@ def LDRBT_POST_REG : AI2ldstidx<1, 1, 0, (outs GPR:$Rt, GPR:$Rn_wb),
   let DecoderMethod = "DecodeAddrMode2IdxInstruction";
 }
 
-def LDRBT_POST_IMM : AI2ldstidx<1, 1, 0, (outs GPR:$Rt, GPR:$Rn_wb),
-                     (ins addr_offset_none:$addr, am2offset_imm:$offset),
-                    IndexModePost, LdFrm, IIC_iLoad_bh_ru,
-                    "ldrbt", "\t$Rt, $addr, $offset",
-                    "$addr.base = $Rn_wb", []> {
+def LDRBT_POST_IMM
+  : AI2ldstidx<1, 1, 0, (outs GPR:$Rt, GPR:$Rn_wb),
+               (ins addr_offset_none:$addr, am2offset_imm:$offset),
+               IndexModePost, LdFrm, IIC_iLoad_bh_ru,
+               "ldrbt", "\t$Rt, $addr, $offset", "$addr.base = $Rn_wb", []> {
   // {12}     isAdd
   // {11-0}   imm12/Rm
   bits<14> offset;
@@ -2524,6 +2678,14 @@ defm LDRHT  : AI3ldrT<0b1011, "ldrht">;
 defm LDRSHT : AI3ldrT<0b1111, "ldrsht">;
 }
 
+def LDRT_POST
+  : ARMAsmPseudo<"ldrt${q} $Rt, $addr", (ins addr_offset_none:$addr, pred:$q),
+                 (outs GPR:$Rt)>;
+
+def LDRBT_POST
+  : ARMAsmPseudo<"ldrbt${q} $Rt, $addr", (ins addr_offset_none:$addr, pred:$q),
+                 (outs GPR:$Rt)>;
+
 // Store
 
 // Stores with truncate
@@ -2532,12 +2694,12 @@ def STRH : AI3str<0b1011, (outs), (ins GPR:$Rt, addrmode3:$addr), StMiscFrm,
                [(truncstorei16 GPR:$Rt, addrmode3:$addr)]>;
 
 // Store doubleword
-let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 in
-def STRD : AI3str<0b1111, (outs), (ins GPR:$Rt, GPR:$src2, addrmode3:$addr),
-               StMiscFrm, IIC_iStore_d_r,
-               "strd", "\t$Rt, $src2, $addr", []>,
-           Requires<[IsARM, HasV5TE]> {
-  let Inst{21} = 0;
+let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 in {
+  def STRD : AI3str<0b1111, (outs), (ins GPR:$Rt, GPR:$Rt2, addrmode3:$addr),
+                    StMiscFrm, IIC_iStore_d_r, "strd", "\t$Rt, $Rt2, $addr", []>,
+             Requires<[IsARM, HasV5TE]> {
+    let Inst{21} = 0;
+  }
 }
 
 // Indexed stores
@@ -2546,7 +2708,8 @@ multiclass AI2_stridx<bit isByte, string opc,
   def _PRE_IMM : AI2ldstidx<0, isByte, 1, (outs GPR:$Rn_wb),
                             (ins GPR:$Rt, addrmode_imm12_pre:$addr), IndexModePre,
                             StFrm, iii,
-                            opc, "\t$Rt, $addr!", "$addr.base = $Rn_wb", []> {
+                            opc, "\t$Rt, $addr!",
+                            "$addr.base = $Rn_wb,@earlyclobber $Rn_wb", []> {
     bits<17> addr;
     let Inst{25} = 0;
     let Inst{23}    = addr{12};     // U (add = ('U' == 1))
@@ -2558,7 +2721,8 @@ multiclass AI2_stridx<bit isByte, string opc,
   def _PRE_REG  : AI2ldstidx<0, isByte, 1, (outs GPR:$Rn_wb),
                       (ins GPR:$Rt, ldst_so_reg:$addr),
                       IndexModePre, StFrm, iir,
-                      opc, "\t$Rt, $addr!", "$addr.base = $Rn_wb", []> {
+                      opc, "\t$Rt, $addr!",
+                      "$addr.base = $Rn_wb,@earlyclobber $Rn_wb", []> {
     bits<17> addr;
     let Inst{25} = 1;
     let Inst{23}    = addr{12};    // U (add = ('U' == 1))
@@ -2571,7 +2735,7 @@ multiclass AI2_stridx<bit isByte, string opc,
                 (ins GPR:$Rt, addr_offset_none:$addr, am2offset_reg:$offset),
                 IndexModePost, StFrm, iir,
                 opc, "\t$Rt, $addr, $offset",
-                "$addr.base = $Rn_wb", []> {
+                "$addr.base = $Rn_wb,@earlyclobber $Rn_wb", []> {
      // {12}     isAdd
      // {11-0}   imm12/Rm
      bits<14> offset;
@@ -2589,7 +2753,7 @@ multiclass AI2_stridx<bit isByte, string opc,
                 (ins GPR:$Rt, addr_offset_none:$addr, am2offset_imm:$offset),
                 IndexModePost, StFrm, iii,
                 opc, "\t$Rt, $addr, $offset",
-                "$addr.base = $Rn_wb", []> {
+                "$addr.base = $Rn_wb,@earlyclobber $Rn_wb", []> {
     // {12}     isAdd
     // {11-0}   imm12/Rm
     bits<14> offset;
@@ -2746,11 +2910,11 @@ def STRBT_POST_REG : AI2ldstidx<0, 1, 0, (outs GPR:$Rn_wb),
   let DecoderMethod = "DecodeAddrMode2IdxInstruction";
 }
 
-def STRBT_POST_IMM : AI2ldstidx<0, 1, 0, (outs GPR:$Rn_wb),
-                   (ins GPR:$Rt, addr_offset_none:$addr, am2offset_imm:$offset),
-                   IndexModePost, StFrm, IIC_iStore_bh_ru,
-                   "strbt", "\t$Rt, $addr, $offset",
-                   "$addr.base = $Rn_wb", []> {
+def STRBT_POST_IMM
+  : AI2ldstidx<0, 1, 0, (outs GPR:$Rn_wb),
+               (ins GPR:$Rt, addr_offset_none:$addr, am2offset_imm:$offset),
+               IndexModePost, StFrm, IIC_iStore_bh_ru,
+               "strbt", "\t$Rt, $addr, $offset", "$addr.base = $Rn_wb", []> {
   // {12}     isAdd
   // {11-0}   imm12/Rm
   bits<14> offset;
@@ -2763,6 +2927,10 @@ def STRBT_POST_IMM : AI2ldstidx<0, 1, 0, (outs GPR:$Rn_wb),
   let DecoderMethod = "DecodeAddrMode2IdxInstruction";
 }
 
+def STRBT_POST
+  : ARMAsmPseudo<"strbt${q} $Rt, $addr",
+                 (ins GPR:$Rt, addr_offset_none:$addr, pred:$q)>;
+
 let mayStore = 1, neverHasSideEffects = 1 in {
 def STRT_POST_REG : AI2ldstidx<0, 0, 0, (outs GPR:$Rn_wb),
                    (ins GPR:$Rt, addr_offset_none:$addr, am2offset_reg:$offset),
@@ -2783,11 +2951,11 @@ def STRT_POST_REG : AI2ldstidx<0, 0, 0, (outs GPR:$Rn_wb),
   let DecoderMethod = "DecodeAddrMode2IdxInstruction";
 }
 
-def STRT_POST_IMM : AI2ldstidx<0, 0, 0, (outs GPR:$Rn_wb),
-                   (ins GPR:$Rt, addr_offset_none:$addr, am2offset_imm:$offset),
-                   IndexModePost, StFrm, IIC_iStore_ru,
-                   "strt", "\t$Rt, $addr, $offset",
-                   "$addr.base = $Rn_wb", []> {
+def STRT_POST_IMM
+  : AI2ldstidx<0, 0, 0, (outs GPR:$Rn_wb),
+               (ins GPR:$Rt, addr_offset_none:$addr, am2offset_imm:$offset),
+               IndexModePost, StFrm, IIC_iStore_ru,
+               "strt", "\t$Rt, $addr, $offset", "$addr.base = $Rn_wb", []> {
   // {12}     isAdd
   // {11-0}   imm12/Rm
   bits<14> offset;
@@ -2801,6 +2969,9 @@ def STRT_POST_IMM : AI2ldstidx<0, 0, 0, (outs GPR:$Rn_wb),
 }
 }
 
+def STRT_POST
+  : ARMAsmPseudo<"strt${q} $Rt, $addr",
+                 (ins GPR:$Rt, addr_offset_none:$addr, pred:$q)>;
 
 multiclass AI3strT<bits<4> op, string opc> {
   def i : AI3ldstidxT<op, 0, (outs GPR:$base_wb),
@@ -3165,8 +3336,8 @@ def SBFX  : I<(outs GPRnopc:$Rd),
   let Inst{3-0}   = Rn;
 }
 
-def UBFX  : I<(outs GPR:$Rd),
-              (ins GPR:$Rn, imm0_31:$lsb, imm1_32:$width),
+def UBFX  : I<(outs GPRnopc:$Rd),
+              (ins GPRnopc:$Rn, imm0_31:$lsb, imm1_32:$width),
                AddrMode1, 4, IndexModeNone, DPFrm, IIC_iUNAsi,
                "ubfx", "\t$Rd, $Rn, $lsb, $width", "", []>,
                Requires<[IsARM, HasV6T2]> {
@@ -3601,21 +3772,22 @@ def MULv5: ARMPseudoExpand<(outs GPRnopc:$Rd), (ins GPRnopc:$Rn, GPRnopc:$Rm,
                Requires<[IsARM, NoV6, UseMulOps]>;
 }
 
-def MLA  : AsMul1I32<0b0000001, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
+def MLA  : AsMul1I32<0b0000001, (outs GPRnopc:$Rd),
+                     (ins GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra),
                      IIC_iMAC32, "mla", "\t$Rd, $Rn, $Rm, $Ra",
-                   [(set GPR:$Rd, (add (mul GPR:$Rn, GPR:$Rm), GPR:$Ra))]>,
-                   Requires<[IsARM, HasV6, UseMulOps]> {
+        [(set GPRnopc:$Rd, (add (mul GPRnopc:$Rn, GPRnopc:$Rm), GPRnopc:$Ra))]>,
+                     Requires<[IsARM, HasV6, UseMulOps]> {
   bits<4> Ra;
   let Inst{15-12} = Ra;
 }
 
 let Constraints = "@earlyclobber $Rd" in
-def MLAv5: ARMPseudoExpand<(outs GPR:$Rd),
-                           (ins GPR:$Rn, GPR:$Rm, GPR:$Ra, pred:$p, cc_out:$s),
-                           4, IIC_iMAC32,
-                        [(set GPR:$Rd, (add (mul GPR:$Rn, GPR:$Rm), GPR:$Ra))],
-                  (MLA GPR:$Rd, GPR:$Rn, GPR:$Rm, GPR:$Ra, pred:$p, cc_out:$s)>,
-                        Requires<[IsARM, NoV6]>;
+def MLAv5: ARMPseudoExpand<(outs GPRnopc:$Rd),
+                           (ins GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra,
+                            pred:$p, cc_out:$s), 4, IIC_iMAC32,
+         [(set GPRnopc:$Rd, (add (mul GPRnopc:$Rn, GPRnopc:$Rm), GPRnopc:$Ra))],
+  (MLA GPRnopc:$Rd, GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra, pred:$p, cc_out:$s)>,
+                           Requires<[IsARM, NoV6]>;
 
 def MLS  : AMul1I<0b0000011, (outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                    IIC_iMAC32, "mls", "\t$Rd, $Rn, $Rm, $Ra",
@@ -3683,7 +3855,8 @@ def UMAAL : AMul1I <0b0000010, (outs GPR:$RdLo, GPR:$RdHi),
   let Inst{3-0}   = Rn;
 }
 
-let Constraints = "$RLo = $RdLo,$RHi = $RdHi" in {
+let Constraints =
+    "@earlyclobber $RdLo,@earlyclobber $RdHi,$RLo = $RdLo,$RHi = $RdHi" in {
 def SMLALv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
                 (ins GPR:$Rn, GPR:$Rm, GPR:$RLo, GPR:$RHi, pred:$p, cc_out:$s),
                               4, IIC_iMAC64, [],
@@ -3698,14 +3871,6 @@ def UMLALv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
                            Requires<[IsARM, NoV6]>;
 }
 
-let Constraints = "@earlyclobber $RdLo,@earlyclobber $RdHi" in {
-def UMAALv5 : ARMPseudoExpand<(outs GPR:$RdLo, GPR:$RdHi),
-                              (ins GPR:$Rn, GPR:$Rm, pred:$p),
-                              4, IIC_iMAC64, [],
-          (UMAAL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, pred:$p)>,
-                           Requires<[IsARM, NoV6]>;
-}
-
 } // neverHasSideEffects
 
 // Most significant word multiply
@@ -3972,6 +4137,11 @@ def REV16 : AMiscA1I<0b01101011, 0b1011, (outs GPR:$Rd), (ins GPR:$Rm),
                Requires<[IsARM, HasV6]>,
            Sched<[WriteALU]>;
 
+def : ARMV6Pat<(srl (bswap (extloadi16 addrmode3:$addr)), (i32 16)),
+              (REV16 (LDRH addrmode3:$addr))>;
+def : ARMV6Pat<(truncstorei16 (srl (bswap GPR:$Rn), (i32 16)), addrmode3:$addr),
+               (STRH (REV16 GPR:$Rn), addrmode3:$addr)>;
+
 let AddedComplexity = 5 in
 def REVSH : AMiscA1I<0b01101111, 0b1011, (outs GPR:$Rd), (ins GPR:$Rm),
                IIC_iUNAr, "revsh", "\t$Rd, $Rm",
@@ -4276,7 +4446,7 @@ def instsyncb_opt : Operand<i32> {
   let DecoderMethod = "DecodeInstSyncBarrierOption";
 }
 
-// memory barriers protect the atomic sequences
+// Memory barriers protect the atomic sequences
 let hasSideEffects = 1 in {
 def DMB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
                 "dmb", "\t$opt", [(int_arm_dmb (i32 imm0_15:$opt))]>,
@@ -4285,7 +4455,6 @@ def DMB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
   let Inst{31-4} = 0xf57ff05;
   let Inst{3-0} = opt;
 }
-}
 
 def DSB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
                 "dsb", "\t$opt", [(int_arm_dsb (i32 imm0_15:$opt))]>,
@@ -4297,226 +4466,19 @@ def DSB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
 
 // ISB has only full system option
 def ISB : AInoP<(outs), (ins instsyncb_opt:$opt), MiscFrm, NoItinerary,
-                "isb", "\t$opt", []>,
+                "isb", "\t$opt", [(int_arm_isb (i32 imm0_15:$opt))]>,
                 Requires<[IsARM, HasDB]> {
   bits<4> opt;
   let Inst{31-4} = 0xf57ff06;
   let Inst{3-0} = opt;
 }
+}
 
 let usesCustomInserter = 1, Defs = [CPSR] in {
 
 // Pseudo instruction that combines movs + predicated rsbmi
 // to implement integer ABS
   def ABS : ARMPseudoInst<(outs GPR:$dst), (ins GPR:$src), 8, NoItinerary, []>;
-
-// Atomic pseudo-insts which will be lowered to ldrex/strex loops.
-// (64-bit pseudos use a hand-written selection code).
-  let mayLoad = 1, mayStore = 1 in {
-    def ATOMIC_LOAD_ADD_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_SUB_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_AND_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_OR_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_XOR_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_NAND_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MIN_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MAX_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMIN_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMAX_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_SWAP_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$new, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_CMP_SWAP_I8 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$old, GPR:$new, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_ADD_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_SUB_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_AND_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_OR_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_XOR_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_NAND_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MIN_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MAX_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMIN_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMAX_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_SWAP_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$new, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_CMP_SWAP_I16 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$old, GPR:$new, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_ADD_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_SUB_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_AND_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_OR_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_XOR_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_NAND_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MIN_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MAX_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMIN_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMAX_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_SWAP_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$new, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_CMP_SWAP_I32 : PseudoInst<
-      (outs GPR:$dst),
-      (ins GPR:$ptr, GPR:$old, GPR:$new, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_ADD_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_SUB_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_AND_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_OR_I64 :  PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_XOR_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_NAND_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MIN_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_MAX_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMIN_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_LOAD_UMAX_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_SWAP_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
-    def ATOMIC_CMP_SWAP_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$cmp1, GPR:$cmp2,
-           GPR:$set1, GPR:$set2, i32imm:$ordering),
-      NoItinerary, []>;
-  }
-  let mayLoad = 1 in
-    def ATOMIC_LOAD_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, i32imm:$ordering),
-      NoItinerary, []>;
-  let mayStore = 1 in
-    def ATOMIC_STORE_I64 : PseudoInst<
-      (outs GPR:$dst1, GPR:$dst2),
-      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
-      NoItinerary, []>;
 }
 
 let usesCustomInserter = 1 in {
@@ -4553,6 +4515,33 @@ def strex_4 : PatFrag<(ops node:$val, node:$ptr),
   return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i32;
 }]>;
 
+def ldaex_1 : PatFrag<(ops node:$ptr), (int_arm_ldaex node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i8;
+}]>;
+
+def ldaex_2 : PatFrag<(ops node:$ptr), (int_arm_ldaex node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i16;
+}]>;
+
+def ldaex_4 : PatFrag<(ops node:$ptr), (int_arm_ldaex node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i32;
+}]>;
+
+def stlex_1 : PatFrag<(ops node:$val, node:$ptr),
+                      (int_arm_stlex node:$val, node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i8;
+}]>;
+
+def stlex_2 : PatFrag<(ops node:$val, node:$ptr),
+                      (int_arm_stlex node:$val, node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i16;
+}]>;
+
+def stlex_4 : PatFrag<(ops node:$val, node:$ptr),
+                      (int_arm_stlex node:$val, node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i32;
+}]>;
+
 let mayLoad = 1 in {
 def LDREXB : AIldrex<0b10, (outs GPR:$Rt), (ins addr_offset_none:$addr),
                      NoItinerary, "ldrexb", "\t$Rt, $addr",
@@ -4570,11 +4559,14 @@ def LDREXD : AIldrex<0b01, (outs GPRPairOp:$Rt),(ins addr_offset_none:$addr),
 }
 
 def LDAEXB : AIldaex<0b10, (outs GPR:$Rt), (ins addr_offset_none:$addr),
-                     NoItinerary, "ldaexb", "\t$Rt, $addr", []>;
+                     NoItinerary, "ldaexb", "\t$Rt, $addr",
+                     [(set GPR:$Rt, (ldaex_1 addr_offset_none:$addr))]>;
 def LDAEXH : AIldaex<0b11, (outs GPR:$Rt), (ins addr_offset_none:$addr),
-                     NoItinerary, "ldaexh", "\t$Rt, $addr", []>;
+                     NoItinerary, "ldaexh", "\t$Rt, $addr",
+                    [(set GPR:$Rt, (ldaex_2 addr_offset_none:$addr))]>;
 def LDAEX  : AIldaex<0b00, (outs GPR:$Rt), (ins addr_offset_none:$addr),
-                     NoItinerary, "ldaex", "\t$Rt, $addr", []>;
+                     NoItinerary, "ldaex", "\t$Rt, $addr",
+                    [(set GPR:$Rt, (ldaex_4 addr_offset_none:$addr))]>;
 let hasExtraDefRegAllocReq = 1 in
 def LDAEXD : AIldaex<0b01, (outs GPRPairOp:$Rt),(ins addr_offset_none:$addr),
                       NoItinerary, "ldaexd", "\t$Rt, $addr", []> {
@@ -4585,13 +4577,16 @@ def LDAEXD : AIldaex<0b01, (outs GPRPairOp:$Rt),(ins addr_offset_none:$addr),
 let mayStore = 1, Constraints = "@earlyclobber $Rd" in {
 def STREXB: AIstrex<0b10, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
                     NoItinerary, "strexb", "\t$Rd, $Rt, $addr",
-                    [(set GPR:$Rd, (strex_1 GPR:$Rt, addr_offset_none:$addr))]>;
+                    [(set GPR:$Rd, (strex_1 GPR:$Rt,
+                                            addr_offset_none:$addr))]>;
 def STREXH: AIstrex<0b11, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
                     NoItinerary, "strexh", "\t$Rd, $Rt, $addr",
-                    [(set GPR:$Rd, (strex_2 GPR:$Rt, addr_offset_none:$addr))]>;
+                    [(set GPR:$Rd, (strex_2 GPR:$Rt,
+                                            addr_offset_none:$addr))]>;
 def STREX : AIstrex<0b00, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
                     NoItinerary, "strex", "\t$Rd, $Rt, $addr",
-                    [(set GPR:$Rd, (strex_4 GPR:$Rt, addr_offset_none:$addr))]>;
+                    [(set GPR:$Rd, (strex_4 GPR:$Rt,
+                                            addr_offset_none:$addr))]>;
 let hasExtraSrcRegAllocReq = 1 in
 def STREXD : AIstrex<0b01, (outs GPR:$Rd),
                     (ins GPRPairOp:$Rt, addr_offset_none:$addr),
@@ -4600,13 +4595,16 @@ def STREXD : AIstrex<0b01, (outs GPR:$Rd),
 }
 def STLEXB: AIstlex<0b10, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
                     NoItinerary, "stlexb", "\t$Rd, $Rt, $addr",
-                    []>;
+                    [(set GPR:$Rd,
+                          (stlex_1 GPR:$Rt, addr_offset_none:$addr))]>;
 def STLEXH: AIstlex<0b11, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
                     NoItinerary, "stlexh", "\t$Rd, $Rt, $addr",
-                    []>;
+                    [(set GPR:$Rd,
+                          (stlex_2 GPR:$Rt, addr_offset_none:$addr))]>;
 def STLEX : AIstlex<0b00, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
                     NoItinerary, "stlex", "\t$Rd, $Rt, $addr",
-                    []>;
+                    [(set GPR:$Rd,
+                          (stlex_4 GPR:$Rt, addr_offset_none:$addr))]>;
 let hasExtraSrcRegAllocReq = 1 in
 def STLEXD : AIstlex<0b01, (outs GPR:$Rd),
                     (ins GPRPairOp:$Rt, addr_offset_none:$addr),
@@ -4621,15 +4619,16 @@ def CLREX : AXI<(outs), (ins), MiscFrm, NoItinerary, "clrex",
   let Inst{31-0} = 0b11110101011111111111000000011111;
 }
 
-def : ARMPat<(and (ldrex_1 addr_offset_none:$addr), 0xff),
-             (LDREXB addr_offset_none:$addr)>;
-def : ARMPat<(and (ldrex_2 addr_offset_none:$addr), 0xffff),
-             (LDREXH addr_offset_none:$addr)>;
 def : ARMPat<(strex_1 (and GPR:$Rt, 0xff), addr_offset_none:$addr),
              (STREXB GPR:$Rt, addr_offset_none:$addr)>;
 def : ARMPat<(strex_2 (and GPR:$Rt, 0xffff), addr_offset_none:$addr),
              (STREXH GPR:$Rt, addr_offset_none:$addr)>;
 
+def : ARMPat<(stlex_1 (and GPR:$Rt, 0xff), addr_offset_none:$addr),
+             (STLEXB GPR:$Rt, addr_offset_none:$addr)>;
+def : ARMPat<(stlex_2 (and GPR:$Rt, 0xffff), addr_offset_none:$addr),
+             (STLEXH GPR:$Rt, addr_offset_none:$addr)>;
+
 class acquiring_load<PatFrag base>
   : PatFrag<(ops node:$ptr), (base node:$ptr), [{
   AtomicOrdering Ordering = cast<AtomicSDNode>(N)->getOrdering();
@@ -4960,7 +4959,7 @@ def MCR2 : MovRCopro2<"mcr2", 0 /* from ARM core register to coprocessor */,
                       [(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
                                      imm:$CRm, imm:$opc2)]>,
                       Requires<[PreV8]>;
-def : ARMInstAlias<"mcr2$ $cop, $opc1, $Rt, $CRn, $CRm",
+def : ARMInstAlias<"mcr2 $cop, $opc1, $Rt, $CRn, $CRm",
                    (MCR2 p_imm:$cop, imm0_7:$opc1, GPR:$Rt, c_imm:$CRn,
                          c_imm:$CRm, 0)>;
 def MRC2 : MovRCopro2<"mrc2", 1 /* from coprocessor to ARM core register */,
@@ -4968,7 +4967,7 @@ def MRC2 : MovRCopro2<"mrc2", 1 /* from coprocessor to ARM core register */,
                       (ins p_imm:$cop, imm0_7:$opc1, c_imm:$CRn, c_imm:$CRm,
                            imm0_7:$opc2), []>,
                       Requires<[PreV8]>;
-def : ARMInstAlias<"mrc2$ $cop, $opc1, $Rt, $CRn, $CRm",
+def : ARMInstAlias<"mrc2 $cop, $opc1, $Rt, $CRn, $CRm",
                    (MRC2 GPRwithAPSR:$Rt, p_imm:$cop, imm0_7:$opc1, c_imm:$CRn,
                          c_imm:$CRm, 0)>;
 
@@ -5097,6 +5096,19 @@ def MSRi : ABI<0b0011, (outs), (ins msr_mask:$mask,  so_imm:$a), NoItinerary,
   let Inst{11-0} = a;
 }
 
+// Dynamic stack allocation yields a _chkstk for Windows targets.  These calls
+// are needed to probe the stack when allocating more than
+// 4k bytes in one go. Touching the stack at 4K increments is necessary to
+// ensure that the guard pages used by the OS virtual memory manager are
+// allocated in correct sequence.
+// The main point of having separate instruction are extra unmodelled effects
+// (compared to ordinary calls) like stack pointer change.
+
+def win__chkstk : SDNode<"ARMISD::WIN__CHKSTK", SDTNone,
+                      [SDNPHasChain, SDNPSideEffect]>;
+let usesCustomInserter = 1, Uses = [R4], Defs = [R4, SP] in
+  def WIN__CHKSTK : PseudoInst<(outs), (ins), NoItinerary, [(win__chkstk)]>;
+
 //===----------------------------------------------------------------------===//
 // TLS Instructions
 //
@@ -5104,9 +5116,11 @@ def MSRi : ABI<0b0011, (outs), (ins msr_mask:$mask,  so_imm:$a), NoItinerary,
 // __aeabi_read_tp preserves the registers r1-r3.
 // This is a pseudo inst so that we can get the encoding right,
 // complete with fixup for the aeabi_read_tp function.
+// TPsoft is valid for ARM mode only, in case of Thumb mode a tTPsoft pattern
+// is defined in "ARMInstrThumb.td".
 let isCall = 1,
   Defs = [R0, R12, LR, CPSR], Uses = [SP] in {
-  def TPsoft : PseudoInst<(outs), (ins), IIC_Br,
+  def TPsoft : ARMPseudoInst<(outs), (ins), 4, IIC_Br,
                [(set R0, ARMthread_pointer)]>, Sched<[WriteBr]>;
 }
 
@@ -5184,6 +5198,10 @@ def MOVi32imm : PseudoInst<(outs GPR:$dst), (ins i32imm:$src), IIC_iMOVix2,
                            [(set GPR:$dst, (arm_i32imm:$src))]>,
                            Requires<[IsARM]>;
 
+def LDRLIT_ga_abs : PseudoInst<(outs GPR:$dst), (ins i32imm:$src), IIC_iLoad_i,
+                               [(set GPR:$dst, (ARMWrapper tglobaladdr:$src))]>,
+                    Requires<[IsARM, DontUseMovt]>;
+
 // Pseudo instruction that combines movw + movt + add pc (if PIC).
 // It also makes it possible to rematerialize the instructions.
 // FIXME: Remove this when we can do generalized remat and when machine licm
@@ -5194,10 +5212,17 @@ def MOV_ga_pcrel : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr),
                         [(set GPR:$dst, (ARMWrapperPIC tglobaladdr:$addr))]>,
                         Requires<[IsARM, UseMovt]>;
 
-def MOV_ga_dyn : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr),
-                             IIC_iMOVix2,
-                        [(set GPR:$dst, (ARMWrapperDYN tglobaladdr:$addr))]>,
-                        Requires<[IsARM, UseMovt]>;
+def LDRLIT_ga_pcrel : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr),
+                                 IIC_iLoadiALU,
+                                 [(set GPR:$dst,
+                                       (ARMWrapperPIC tglobaladdr:$addr))]>,
+                      Requires<[IsARM, DontUseMovt]>;
+
+def LDRLIT_ga_pcrel_ldr : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr),
+                              NoItinerary,
+                              [(set GPR:$dst,
+                                    (load (ARMWrapperPIC tglobaladdr:$addr)))]>,
+                          Requires<[IsARM, DontUseMovt]>;
 
 let AddedComplexity = 10 in
 def MOV_ga_pcrel_ldr : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr),
@@ -5207,8 +5232,6 @@ def MOV_ga_pcrel_ldr : PseudoInst<(outs GPR:$dst), (ins i32imm:$addr),
 } // isReMaterializable
 
 // ConstantPool, GlobalAddress, and JumpTable
-def : ARMPat<(ARMWrapper  tglobaladdr :$dst), (LEApcrel tglobaladdr :$dst)>,
-            Requires<[IsARM, DontUseMovt]>;
 def : ARMPat<(ARMWrapper  tconstpool  :$dst), (LEApcrel tconstpool  :$dst)>;
 def : ARMPat<(ARMWrapper  tglobaladdr :$dst), (MOVi32imm tglobaladdr :$dst)>,
             Requires<[IsARM, UseMovt]>;
@@ -5544,9 +5567,22 @@ def : ARMInstAlias<"neg${s}${p} $Rd, $Rm",
 def : InstAlias<"nop${p}", (MOVr R0, R0, pred:$p, zero_reg)>,
          Requires<[IsARM, NoV6]>;
 
-// UMULL/SMULL are available on all arches, but the instruction definitions
-// need difference constraints pre-v6. Use these aliases for the assembly
-// parsing on pre-v6.
+// MUL/UMLAL/SMLAL/UMULL/SMULL are available on all arches, but
+// the instruction definitions need difference constraints pre-v6.
+// Use these aliases for the assembly parsing on pre-v6.
+def : InstAlias<"mul${s}${p} $Rd, $Rn, $Rm",
+            (MUL GPRnopc:$Rd, GPRnopc:$Rn, GPRnopc:$Rm, pred:$p, cc_out:$s)>,
+         Requires<[IsARM, NoV6]>;
+def : InstAlias<"mla${s}${p} $Rd, $Rn, $Rm, $Ra",
+            (MLA GPRnopc:$Rd, GPRnopc:$Rn, GPRnopc:$Rm, GPRnopc:$Ra,
+             pred:$p, cc_out:$s)>,
+         Requires<[IsARM, NoV6]>;
+def : InstAlias<"smlal${s}${p} $RdLo, $RdHi, $Rn, $Rm",
+            (SMLAL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s)>,
+         Requires<[IsARM, NoV6]>;
+def : InstAlias<"umlal${s}${p} $RdLo, $RdHi, $Rn, $Rm",
+            (UMLAL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s)>,
+         Requires<[IsARM, NoV6]>;
 def : InstAlias<"smull${s}${p} $RdLo, $RdHi, $Rn, $Rm",
             (SMULL GPR:$RdLo, GPR:$RdHi, GPR:$Rn, GPR:$Rm, pred:$p, cc_out:$s)>,
          Requires<[IsARM, NoV6]>;
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrNEON.td b/contrib/llvm/lib/Target/ARM/ARMInstrNEON.td
index 0b05c08ed948..c02bb3b55f5b 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrNEON.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrNEON.td
@@ -39,6 +39,49 @@ def nImmVMOVI32 : Operand<i32> {
   let PrintMethod = "printNEONModImmOperand";
   let ParserMatchClass = nImmVMOVI32AsmOperand;
 }
+
+def nImmVMOVI16AsmOperandByteReplicate :
+  AsmOperandClass {
+  let Name = "NEONi16vmovByteReplicate";
+  let PredicateMethod = "isNEONi16ByteReplicate";
+  let RenderMethod = "addNEONvmovByteReplicateOperands";
+}
+def nImmVMOVI32AsmOperandByteReplicate :
+  AsmOperandClass {
+  let Name = "NEONi32vmovByteReplicate";
+  let PredicateMethod = "isNEONi32ByteReplicate";
+  let RenderMethod = "addNEONvmovByteReplicateOperands";
+}
+def nImmVMVNI16AsmOperandByteReplicate :
+  AsmOperandClass {
+  let Name = "NEONi16invByteReplicate";
+  let PredicateMethod = "isNEONi16ByteReplicate";
+  let RenderMethod = "addNEONinvByteReplicateOperands";
+}
+def nImmVMVNI32AsmOperandByteReplicate :
+  AsmOperandClass {
+  let Name = "NEONi32invByteReplicate";
+  let PredicateMethod = "isNEONi32ByteReplicate";
+  let RenderMethod = "addNEONinvByteReplicateOperands";
+}
+
+def nImmVMOVI16ByteReplicate : Operand<i32> {
+  let PrintMethod = "printNEONModImmOperand";
+  let ParserMatchClass = nImmVMOVI16AsmOperandByteReplicate;
+}
+def nImmVMOVI32ByteReplicate : Operand<i32> {
+  let PrintMethod = "printNEONModImmOperand";
+  let ParserMatchClass = nImmVMOVI32AsmOperandByteReplicate;
+}
+def nImmVMVNI16ByteReplicate : Operand<i32> {
+  let PrintMethod = "printNEONModImmOperand";
+  let ParserMatchClass = nImmVMVNI16AsmOperandByteReplicate;
+}
+def nImmVMVNI32ByteReplicate : Operand<i32> {
+  let PrintMethod = "printNEONModImmOperand";
+  let ParserMatchClass = nImmVMVNI32AsmOperandByteReplicate;
+}
+
 def nImmVMOVI32NegAsmOperand : AsmOperandClass { let Name = "NEONi32vmovNeg"; }
 def nImmVMOVI32Neg : Operand<i32> {
   let PrintMethod = "printNEONModImmOperand";
@@ -466,9 +509,6 @@ def SDTARMVSHINS  : SDTypeProfile<1, 3, [SDTCisInt<0>, SDTCisSameAs<0, 1>,
 def NEONvshl      : SDNode<"ARMISD::VSHL", SDTARMVSH>;
 def NEONvshrs     : SDNode<"ARMISD::VSHRs", SDTARMVSH>;
 def NEONvshru     : SDNode<"ARMISD::VSHRu", SDTARMVSH>;
-def NEONvshlls    : SDNode<"ARMISD::VSHLLs", SDTARMVSHX>;
-def NEONvshllu    : SDNode<"ARMISD::VSHLLu", SDTARMVSHX>;
-def NEONvshlli    : SDNode<"ARMISD::VSHLLi", SDTARMVSHX>;
 def NEONvshrn     : SDNode<"ARMISD::VSHRN", SDTARMVSHX>;
 
 def NEONvrshrs    : SDNode<"ARMISD::VRSHRs", SDTARMVSH>;
@@ -620,37 +660,37 @@ class VLDQQQQWBPseudo<InstrItinClass itin>
 let mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1 in {
 
 //   VLD1     : Vector Load (multiple single elements)
-class VLD1D<bits<4> op7_4, string Dt>
+class VLD1D<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<0,0b10,0b0111,op7_4, (outs VecListOneD:$Vd),
-          (ins addrmode6:$Rn), IIC_VLD1,
+          (ins AddrMode:$Rn), IIC_VLD1,
           "vld1", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
-class VLD1Q<bits<4> op7_4, string Dt>
+class VLD1Q<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<0,0b10,0b1010,op7_4, (outs VecListDPair:$Vd),
-          (ins addrmode6:$Rn), IIC_VLD1x2,
+          (ins AddrMode:$Rn), IIC_VLD1x2,
           "vld1", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{5-4} = Rn{5-4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
 
-def  VLD1d8   : VLD1D<{0,0,0,?}, "8">;
-def  VLD1d16  : VLD1D<{0,1,0,?}, "16">;
-def  VLD1d32  : VLD1D<{1,0,0,?}, "32">;
-def  VLD1d64  : VLD1D<{1,1,0,?}, "64">;
+def  VLD1d8   : VLD1D<{0,0,0,?}, "8",  addrmode6align64>;
+def  VLD1d16  : VLD1D<{0,1,0,?}, "16", addrmode6align64>;
+def  VLD1d32  : VLD1D<{1,0,0,?}, "32", addrmode6align64>;
+def  VLD1d64  : VLD1D<{1,1,0,?}, "64", addrmode6align64>;
 
-def  VLD1q8   : VLD1Q<{0,0,?,?}, "8">;
-def  VLD1q16  : VLD1Q<{0,1,?,?}, "16">;
-def  VLD1q32  : VLD1Q<{1,0,?,?}, "32">;
-def  VLD1q64  : VLD1Q<{1,1,?,?}, "64">;
+def  VLD1q8   : VLD1Q<{0,0,?,?}, "8",  addrmode6align64or128>;
+def  VLD1q16  : VLD1Q<{0,1,?,?}, "16", addrmode6align64or128>;
+def  VLD1q32  : VLD1Q<{1,0,?,?}, "32", addrmode6align64or128>;
+def  VLD1q64  : VLD1Q<{1,1,?,?}, "64", addrmode6align64or128>;
 
 // ...with address register writeback:
-multiclass VLD1DWB<bits<4> op7_4, string Dt> {
+multiclass VLD1DWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b10, 0b0111,op7_4, (outs VecListOneD:$Vd, GPR:$wb),
-                     (ins addrmode6:$Rn), IIC_VLD1u,
+                     (ins AddrMode:$Rn), IIC_VLD1u,
                      "vld1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -658,16 +698,16 @@ multiclass VLD1DWB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b10,0b0111,op7_4, (outs VecListOneD:$Vd, GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm), IIC_VLD1u,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD1u,
                         "vld1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{4} = Rn{4};
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
 }
-multiclass VLD1QWB<bits<4> op7_4, string Dt> {
+multiclass VLD1QWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b10,0b1010,op7_4, (outs VecListDPair:$Vd, GPR:$wb),
-                    (ins addrmode6:$Rn), IIC_VLD1x2u,
+                    (ins AddrMode:$Rn), IIC_VLD1x2u,
                      "vld1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -675,7 +715,7 @@ multiclass VLD1QWB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b10,0b1010,op7_4, (outs VecListDPair:$Vd, GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm), IIC_VLD1x2u,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD1x2u,
                         "vld1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{5-4} = Rn{5-4};
@@ -683,27 +723,27 @@ multiclass VLD1QWB<bits<4> op7_4, string Dt> {
   }
 }
 
-defm VLD1d8wb  : VLD1DWB<{0,0,0,?}, "8">;
-defm VLD1d16wb : VLD1DWB<{0,1,0,?}, "16">;
-defm VLD1d32wb : VLD1DWB<{1,0,0,?}, "32">;
-defm VLD1d64wb : VLD1DWB<{1,1,0,?}, "64">;
-defm VLD1q8wb  : VLD1QWB<{0,0,?,?}, "8">;
-defm VLD1q16wb : VLD1QWB<{0,1,?,?}, "16">;
-defm VLD1q32wb : VLD1QWB<{1,0,?,?}, "32">;
-defm VLD1q64wb : VLD1QWB<{1,1,?,?}, "64">;
+defm VLD1d8wb  : VLD1DWB<{0,0,0,?}, "8",  addrmode6align64>;
+defm VLD1d16wb : VLD1DWB<{0,1,0,?}, "16", addrmode6align64>;
+defm VLD1d32wb : VLD1DWB<{1,0,0,?}, "32", addrmode6align64>;
+defm VLD1d64wb : VLD1DWB<{1,1,0,?}, "64", addrmode6align64>;
+defm VLD1q8wb  : VLD1QWB<{0,0,?,?}, "8",  addrmode6align64or128>;
+defm VLD1q16wb : VLD1QWB<{0,1,?,?}, "16", addrmode6align64or128>;
+defm VLD1q32wb : VLD1QWB<{1,0,?,?}, "32", addrmode6align64or128>;
+defm VLD1q64wb : VLD1QWB<{1,1,?,?}, "64", addrmode6align64or128>;
 
 // ...with 3 registers
-class VLD1D3<bits<4> op7_4, string Dt>
+class VLD1D3<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<0,0b10,0b0110,op7_4, (outs VecListThreeD:$Vd),
-          (ins addrmode6:$Rn), IIC_VLD1x3, "vld1", Dt,
+          (ins AddrMode:$Rn), IIC_VLD1x3, "vld1", Dt,
           "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
-multiclass VLD1D3WB<bits<4> op7_4, string Dt> {
+multiclass VLD1D3WB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b10,0b0110, op7_4, (outs VecListThreeD:$Vd, GPR:$wb),
-                    (ins addrmode6:$Rn), IIC_VLD1x2u,
+                    (ins AddrMode:$Rn), IIC_VLD1x2u,
                      "vld1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -711,7 +751,7 @@ multiclass VLD1D3WB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b10,0b0110,op7_4, (outs VecListThreeD:$Vd, GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm), IIC_VLD1x2u,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD1x2u,
                         "vld1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{4} = Rn{4};
@@ -719,32 +759,32 @@ multiclass VLD1D3WB<bits<4> op7_4, string Dt> {
   }
 }
 
-def VLD1d8T      : VLD1D3<{0,0,0,?}, "8">;
-def VLD1d16T     : VLD1D3<{0,1,0,?}, "16">;
-def VLD1d32T     : VLD1D3<{1,0,0,?}, "32">;
-def VLD1d64T     : VLD1D3<{1,1,0,?}, "64">;
+def VLD1d8T      : VLD1D3<{0,0,0,?}, "8",  addrmode6align64>;
+def VLD1d16T     : VLD1D3<{0,1,0,?}, "16", addrmode6align64>;
+def VLD1d32T     : VLD1D3<{1,0,0,?}, "32", addrmode6align64>;
+def VLD1d64T     : VLD1D3<{1,1,0,?}, "64", addrmode6align64>;
 
-defm VLD1d8Twb  : VLD1D3WB<{0,0,0,?}, "8">;
-defm VLD1d16Twb : VLD1D3WB<{0,1,0,?}, "16">;
-defm VLD1d32Twb : VLD1D3WB<{1,0,0,?}, "32">;
-defm VLD1d64Twb : VLD1D3WB<{1,1,0,?}, "64">;
+defm VLD1d8Twb  : VLD1D3WB<{0,0,0,?}, "8",  addrmode6align64>;
+defm VLD1d16Twb : VLD1D3WB<{0,1,0,?}, "16", addrmode6align64>;
+defm VLD1d32Twb : VLD1D3WB<{1,0,0,?}, "32", addrmode6align64>;
+defm VLD1d64Twb : VLD1D3WB<{1,1,0,?}, "64", addrmode6align64>;
 
 def VLD1d64TPseudo : VLDQQPseudo<IIC_VLD1x3>;
 def VLD1d64TPseudoWB_fixed : VLDQQWBfixedPseudo<IIC_VLD1x3>;
 def VLD1d64TPseudoWB_register : VLDQQWBregisterPseudo<IIC_VLD1x3>;
 
 // ...with 4 registers
-class VLD1D4<bits<4> op7_4, string Dt>
+class VLD1D4<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<0, 0b10, 0b0010, op7_4, (outs VecListFourD:$Vd),
-          (ins addrmode6:$Rn), IIC_VLD1x4, "vld1", Dt,
+          (ins AddrMode:$Rn), IIC_VLD1x4, "vld1", Dt,
           "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{5-4} = Rn{5-4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
-multiclass VLD1D4WB<bits<4> op7_4, string Dt> {
+multiclass VLD1D4WB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b10,0b0010, op7_4, (outs VecListFourD:$Vd, GPR:$wb),
-                    (ins addrmode6:$Rn), IIC_VLD1x2u,
+                    (ins AddrMode:$Rn), IIC_VLD1x2u,
                      "vld1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -752,7 +792,7 @@ multiclass VLD1D4WB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b10,0b0010,op7_4, (outs VecListFourD:$Vd, GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm), IIC_VLD1x2u,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD1x2u,
                         "vld1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{5-4} = Rn{5-4};
@@ -760,15 +800,15 @@ multiclass VLD1D4WB<bits<4> op7_4, string Dt> {
   }
 }
 
-def VLD1d8Q      : VLD1D4<{0,0,?,?}, "8">;
-def VLD1d16Q     : VLD1D4<{0,1,?,?}, "16">;
-def VLD1d32Q     : VLD1D4<{1,0,?,?}, "32">;
-def VLD1d64Q     : VLD1D4<{1,1,?,?}, "64">;
+def VLD1d8Q      : VLD1D4<{0,0,?,?}, "8",  addrmode6align64or128or256>;
+def VLD1d16Q     : VLD1D4<{0,1,?,?}, "16", addrmode6align64or128or256>;
+def VLD1d32Q     : VLD1D4<{1,0,?,?}, "32", addrmode6align64or128or256>;
+def VLD1d64Q     : VLD1D4<{1,1,?,?}, "64", addrmode6align64or128or256>;
 
-defm VLD1d8Qwb   : VLD1D4WB<{0,0,?,?}, "8">;
-defm VLD1d16Qwb  : VLD1D4WB<{0,1,?,?}, "16">;
-defm VLD1d32Qwb  : VLD1D4WB<{1,0,?,?}, "32">;
-defm VLD1d64Qwb  : VLD1D4WB<{1,1,?,?}, "64">;
+defm VLD1d8Qwb   : VLD1D4WB<{0,0,?,?}, "8",  addrmode6align64or128or256>;
+defm VLD1d16Qwb  : VLD1D4WB<{0,1,?,?}, "16", addrmode6align64or128or256>;
+defm VLD1d32Qwb  : VLD1D4WB<{1,0,?,?}, "32", addrmode6align64or128or256>;
+defm VLD1d64Qwb  : VLD1D4WB<{1,1,?,?}, "64", addrmode6align64or128or256>;
 
 def VLD1d64QPseudo : VLDQQPseudo<IIC_VLD1x4>;
 def VLD1d64QPseudoWB_fixed : VLDQQWBfixedPseudo<IIC_VLD1x4>;
@@ -776,22 +816,28 @@ def VLD1d64QPseudoWB_register : VLDQQWBregisterPseudo<IIC_VLD1x4>;
 
 //   VLD2     : Vector Load (multiple 2-element structures)
 class VLD2<bits<4> op11_8, bits<4> op7_4, string Dt, RegisterOperand VdTy,
-           InstrItinClass itin>
+           InstrItinClass itin, Operand AddrMode>
   : NLdSt<0, 0b10, op11_8, op7_4, (outs VdTy:$Vd),
-          (ins addrmode6:$Rn), itin,
+          (ins AddrMode:$Rn), itin,
           "vld2", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{5-4} = Rn{5-4};
   let DecoderMethod = "DecodeVLDST2Instruction";
 }
 
-def  VLD2d8   : VLD2<0b1000, {0,0,?,?}, "8", VecListDPair, IIC_VLD2>;
-def  VLD2d16  : VLD2<0b1000, {0,1,?,?}, "16", VecListDPair, IIC_VLD2>;
-def  VLD2d32  : VLD2<0b1000, {1,0,?,?}, "32", VecListDPair, IIC_VLD2>;
+def  VLD2d8   : VLD2<0b1000, {0,0,?,?}, "8", VecListDPair, IIC_VLD2,
+                     addrmode6align64or128>;
+def  VLD2d16  : VLD2<0b1000, {0,1,?,?}, "16", VecListDPair, IIC_VLD2,
+                     addrmode6align64or128>;
+def  VLD2d32  : VLD2<0b1000, {1,0,?,?}, "32", VecListDPair, IIC_VLD2,
+                     addrmode6align64or128>;
 
-def  VLD2q8   : VLD2<0b0011, {0,0,?,?}, "8", VecListFourD, IIC_VLD2x2>;
-def  VLD2q16  : VLD2<0b0011, {0,1,?,?}, "16", VecListFourD, IIC_VLD2x2>;
-def  VLD2q32  : VLD2<0b0011, {1,0,?,?}, "32", VecListFourD, IIC_VLD2x2>;
+def  VLD2q8   : VLD2<0b0011, {0,0,?,?}, "8", VecListFourD, IIC_VLD2x2,
+                     addrmode6align64or128or256>;
+def  VLD2q16  : VLD2<0b0011, {0,1,?,?}, "16", VecListFourD, IIC_VLD2x2,
+                     addrmode6align64or128or256>;
+def  VLD2q32  : VLD2<0b0011, {1,0,?,?}, "32", VecListFourD, IIC_VLD2x2,
+                     addrmode6align64or128or256>;
 
 def  VLD2q8Pseudo  : VLDQQPseudo<IIC_VLD2x2>;
 def  VLD2q16Pseudo : VLDQQPseudo<IIC_VLD2x2>;
@@ -799,9 +845,9 @@ def  VLD2q32Pseudo : VLDQQPseudo<IIC_VLD2x2>;
 
 // ...with address register writeback:
 multiclass VLD2WB<bits<4> op11_8, bits<4> op7_4, string Dt,
-                  RegisterOperand VdTy, InstrItinClass itin> {
+                  RegisterOperand VdTy, InstrItinClass itin, Operand AddrMode> {
   def _fixed : NLdSt<0, 0b10, op11_8, op7_4, (outs VdTy:$Vd, GPR:$wb),
-                     (ins addrmode6:$Rn), itin,
+                     (ins AddrMode:$Rn), itin,
                      "vld2", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -809,7 +855,7 @@ multiclass VLD2WB<bits<4> op11_8, bits<4> op7_4, string Dt,
     let DecoderMethod = "DecodeVLDST2Instruction";
   }
   def _register : NLdSt<0, 0b10, op11_8, op7_4, (outs VdTy:$Vd, GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm), itin,
+                        (ins AddrMode:$Rn, rGPR:$Rm), itin,
                         "vld2", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{5-4} = Rn{5-4};
@@ -817,13 +863,19 @@ multiclass VLD2WB<bits<4> op11_8, bits<4> op7_4, string Dt,
   }
 }
 
-defm VLD2d8wb  : VLD2WB<0b1000, {0,0,?,?}, "8", VecListDPair, IIC_VLD2u>;
-defm VLD2d16wb : VLD2WB<0b1000, {0,1,?,?}, "16", VecListDPair, IIC_VLD2u>;
-defm VLD2d32wb : VLD2WB<0b1000, {1,0,?,?}, "32", VecListDPair, IIC_VLD2u>;
+defm VLD2d8wb  : VLD2WB<0b1000, {0,0,?,?}, "8", VecListDPair, IIC_VLD2u,
+                        addrmode6align64or128>;
+defm VLD2d16wb : VLD2WB<0b1000, {0,1,?,?}, "16", VecListDPair, IIC_VLD2u,
+                        addrmode6align64or128>;
+defm VLD2d32wb : VLD2WB<0b1000, {1,0,?,?}, "32", VecListDPair, IIC_VLD2u,
+                        addrmode6align64or128>;
 
-defm VLD2q8wb  : VLD2WB<0b0011, {0,0,?,?}, "8", VecListFourD, IIC_VLD2x2u>;
-defm VLD2q16wb : VLD2WB<0b0011, {0,1,?,?}, "16", VecListFourD, IIC_VLD2x2u>;
-defm VLD2q32wb : VLD2WB<0b0011, {1,0,?,?}, "32", VecListFourD, IIC_VLD2x2u>;
+defm VLD2q8wb  : VLD2WB<0b0011, {0,0,?,?}, "8", VecListFourD, IIC_VLD2x2u,
+                        addrmode6align64or128or256>;
+defm VLD2q16wb : VLD2WB<0b0011, {0,1,?,?}, "16", VecListFourD, IIC_VLD2x2u,
+                        addrmode6align64or128or256>;
+defm VLD2q32wb : VLD2WB<0b0011, {1,0,?,?}, "32", VecListFourD, IIC_VLD2x2u,
+                        addrmode6align64or128or256>;
 
 def VLD2q8PseudoWB_fixed     : VLDQQWBfixedPseudo<IIC_VLD2x2u>;
 def VLD2q16PseudoWB_fixed    : VLDQQWBfixedPseudo<IIC_VLD2x2u>;
@@ -833,12 +885,18 @@ def VLD2q16PseudoWB_register : VLDQQWBregisterPseudo<IIC_VLD2x2u>;
 def VLD2q32PseudoWB_register : VLDQQWBregisterPseudo<IIC_VLD2x2u>;
 
 // ...with double-spaced registers
-def  VLD2b8    : VLD2<0b1001, {0,0,?,?}, "8", VecListDPairSpaced, IIC_VLD2>;
-def  VLD2b16   : VLD2<0b1001, {0,1,?,?}, "16", VecListDPairSpaced, IIC_VLD2>;
-def  VLD2b32   : VLD2<0b1001, {1,0,?,?}, "32", VecListDPairSpaced, IIC_VLD2>;
-defm VLD2b8wb  : VLD2WB<0b1001, {0,0,?,?}, "8", VecListDPairSpaced, IIC_VLD2u>;
-defm VLD2b16wb : VLD2WB<0b1001, {0,1,?,?}, "16", VecListDPairSpaced, IIC_VLD2u>;
-defm VLD2b32wb : VLD2WB<0b1001, {1,0,?,?}, "32", VecListDPairSpaced, IIC_VLD2u>;
+def  VLD2b8    : VLD2<0b1001, {0,0,?,?}, "8", VecListDPairSpaced, IIC_VLD2,
+                      addrmode6align64or128>;
+def  VLD2b16   : VLD2<0b1001, {0,1,?,?}, "16", VecListDPairSpaced, IIC_VLD2,
+                      addrmode6align64or128>;
+def  VLD2b32   : VLD2<0b1001, {1,0,?,?}, "32", VecListDPairSpaced, IIC_VLD2,
+                      addrmode6align64or128>;
+defm VLD2b8wb  : VLD2WB<0b1001, {0,0,?,?}, "8", VecListDPairSpaced, IIC_VLD2u,
+                        addrmode6align64or128>;
+defm VLD2b16wb : VLD2WB<0b1001, {0,1,?,?}, "16", VecListDPairSpaced, IIC_VLD2u,
+                        addrmode6align64or128>;
+defm VLD2b32wb : VLD2WB<0b1001, {1,0,?,?}, "32", VecListDPairSpaced, IIC_VLD2u,
+                        addrmode6align64or128>;
 
 //   VLD3     : Vector Load (multiple 3-element structures)
 class VLD3D<bits<4> op11_8, bits<4> op7_4, string Dt>
@@ -1296,47 +1354,55 @@ def VLD4LNq32Pseudo_UPD : VLDQQQQLNWBPseudo<IIC_VLD4lnu>;
 } // mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1
 
 //   VLD1DUP  : Vector Load (single element to all lanes)
-class VLD1DUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp>
+class VLD1DUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp,
+              Operand AddrMode>
   : NLdSt<1, 0b10, 0b1100, op7_4, (outs VecListOneDAllLanes:$Vd),
-          (ins addrmode6dup:$Rn),
+          (ins AddrMode:$Rn),
           IIC_VLD1dup, "vld1", Dt, "$Vd, $Rn", "",
           [(set VecListOneDAllLanes:$Vd,
-                (Ty (NEONvdup (i32 (LoadOp addrmode6dup:$Rn)))))]> {
+                (Ty (NEONvdup (i32 (LoadOp AddrMode:$Rn)))))]> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLD1DupInstruction";
 }
-def VLD1DUPd8  : VLD1DUP<{0,0,0,?}, "8", v8i8, extloadi8>;
-def VLD1DUPd16 : VLD1DUP<{0,1,0,?}, "16", v4i16, extloadi16>;
-def VLD1DUPd32 : VLD1DUP<{1,0,0,?}, "32", v2i32, load>;
+def VLD1DUPd8  : VLD1DUP<{0,0,0,?}, "8", v8i8, extloadi8,
+                         addrmode6dupalignNone>;
+def VLD1DUPd16 : VLD1DUP<{0,1,0,?}, "16", v4i16, extloadi16,
+                         addrmode6dupalign16>;
+def VLD1DUPd32 : VLD1DUP<{1,0,0,?}, "32", v2i32, load,
+                         addrmode6dupalign32>;
 
 def : Pat<(v2f32 (NEONvdup (f32 (load addrmode6dup:$addr)))),
           (VLD1DUPd32 addrmode6:$addr)>;
 
-class VLD1QDUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp>
+class VLD1QDUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp,
+               Operand AddrMode>
   : NLdSt<1, 0b10, 0b1100, op7_4, (outs VecListDPairAllLanes:$Vd),
-          (ins addrmode6dup:$Rn), IIC_VLD1dup,
+          (ins AddrMode:$Rn), IIC_VLD1dup,
           "vld1", Dt, "$Vd, $Rn", "",
           [(set VecListDPairAllLanes:$Vd,
-                (Ty (NEONvdup (i32 (LoadOp addrmode6dup:$Rn)))))]> {
+                (Ty (NEONvdup (i32 (LoadOp AddrMode:$Rn)))))]> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLD1DupInstruction";
 }
 
-def VLD1DUPq8  : VLD1QDUP<{0,0,1,0}, "8", v16i8, extloadi8>;
-def VLD1DUPq16 : VLD1QDUP<{0,1,1,?}, "16", v8i16, extloadi16>;
-def VLD1DUPq32 : VLD1QDUP<{1,0,1,?}, "32", v4i32, load>;
+def VLD1DUPq8  : VLD1QDUP<{0,0,1,0}, "8", v16i8, extloadi8,
+                          addrmode6dupalignNone>;
+def VLD1DUPq16 : VLD1QDUP<{0,1,1,?}, "16", v8i16, extloadi16,
+                          addrmode6dupalign16>;
+def VLD1DUPq32 : VLD1QDUP<{1,0,1,?}, "32", v4i32, load,
+                          addrmode6dupalign32>;
 
 def : Pat<(v4f32 (NEONvdup (f32 (load addrmode6dup:$addr)))),
           (VLD1DUPq32 addrmode6:$addr)>;
 
 let mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1 in {
 // ...with address register writeback:
-multiclass VLD1DUPWB<bits<4> op7_4, string Dt> {
+multiclass VLD1DUPWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<1, 0b10, 0b1100, op7_4,
                      (outs VecListOneDAllLanes:$Vd, GPR:$wb),
-                     (ins addrmode6dup:$Rn), IIC_VLD1dupu,
+                     (ins AddrMode:$Rn), IIC_VLD1dupu,
                      "vld1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1345,17 +1411,17 @@ multiclass VLD1DUPWB<bits<4> op7_4, string Dt> {
   }
   def _register : NLdSt<1, 0b10, 0b1100, op7_4,
                         (outs VecListOneDAllLanes:$Vd, GPR:$wb),
-                        (ins addrmode6dup:$Rn, rGPR:$Rm), IIC_VLD1dupu,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD1dupu,
                         "vld1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{4} = Rn{4};
     let DecoderMethod = "DecodeVLD1DupInstruction";
   }
 }
-multiclass VLD1QDUPWB<bits<4> op7_4, string Dt> {
+multiclass VLD1QDUPWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<1, 0b10, 0b1100, op7_4,
                      (outs VecListDPairAllLanes:$Vd, GPR:$wb),
-                     (ins addrmode6dup:$Rn), IIC_VLD1dupu,
+                     (ins AddrMode:$Rn), IIC_VLD1dupu,
                      "vld1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1364,7 +1430,7 @@ multiclass VLD1QDUPWB<bits<4> op7_4, string Dt> {
   }
   def _register : NLdSt<1, 0b10, 0b1100, op7_4,
                         (outs VecListDPairAllLanes:$Vd, GPR:$wb),
-                        (ins addrmode6dup:$Rn, rGPR:$Rm), IIC_VLD1dupu,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD1dupu,
                         "vld1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{4} = Rn{4};
@@ -1372,38 +1438,47 @@ multiclass VLD1QDUPWB<bits<4> op7_4, string Dt> {
   }
 }
 
-defm VLD1DUPd8wb  : VLD1DUPWB<{0,0,0,0}, "8">;
-defm VLD1DUPd16wb : VLD1DUPWB<{0,1,0,?}, "16">;
-defm VLD1DUPd32wb : VLD1DUPWB<{1,0,0,?}, "32">;
+defm VLD1DUPd8wb  : VLD1DUPWB<{0,0,0,0}, "8", addrmode6dupalignNone>;
+defm VLD1DUPd16wb : VLD1DUPWB<{0,1,0,?}, "16", addrmode6dupalign16>;
+defm VLD1DUPd32wb : VLD1DUPWB<{1,0,0,?}, "32", addrmode6dupalign32>;
 
-defm VLD1DUPq8wb  : VLD1QDUPWB<{0,0,1,0}, "8">;
-defm VLD1DUPq16wb : VLD1QDUPWB<{0,1,1,?}, "16">;
-defm VLD1DUPq32wb : VLD1QDUPWB<{1,0,1,?}, "32">;
+defm VLD1DUPq8wb  : VLD1QDUPWB<{0,0,1,0}, "8", addrmode6dupalignNone>;
+defm VLD1DUPq16wb : VLD1QDUPWB<{0,1,1,?}, "16", addrmode6dupalign16>;
+defm VLD1DUPq32wb : VLD1QDUPWB<{1,0,1,?}, "32", addrmode6dupalign32>;
 
 //   VLD2DUP  : Vector Load (single 2-element structure to all lanes)
-class VLD2DUP<bits<4> op7_4, string Dt, RegisterOperand VdTy>
+class VLD2DUP<bits<4> op7_4, string Dt, RegisterOperand VdTy, Operand AddrMode>
   : NLdSt<1, 0b10, 0b1101, op7_4, (outs VdTy:$Vd),
-          (ins addrmode6dup:$Rn), IIC_VLD2dup,
+          (ins AddrMode:$Rn), IIC_VLD2dup,
           "vld2", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLD2DupInstruction";
 }
 
-def VLD2DUPd8  : VLD2DUP<{0,0,0,?}, "8",  VecListDPairAllLanes>;
-def VLD2DUPd16 : VLD2DUP<{0,1,0,?}, "16", VecListDPairAllLanes>;
-def VLD2DUPd32 : VLD2DUP<{1,0,0,?}, "32", VecListDPairAllLanes>;
+def VLD2DUPd8  : VLD2DUP<{0,0,0,?}, "8",  VecListDPairAllLanes,
+                         addrmode6dupalign16>;
+def VLD2DUPd16 : VLD2DUP<{0,1,0,?}, "16", VecListDPairAllLanes,
+                         addrmode6dupalign32>;
+def VLD2DUPd32 : VLD2DUP<{1,0,0,?}, "32", VecListDPairAllLanes,
+                         addrmode6dupalign64>;
 
+// HACK this one, VLD2DUPd8x2 must be changed at the same time with VLD2b8 or
+// "vld2.8 {d0[], d2[]}, [r4:32]" will become "vld2.8 {d0, d2}, [r4:32]".
 // ...with double-spaced registers
-def VLD2DUPd8x2  : VLD2DUP<{0,0,1,?}, "8",  VecListDPairSpacedAllLanes>;
-def VLD2DUPd16x2 : VLD2DUP<{0,1,1,?}, "16", VecListDPairSpacedAllLanes>;
-def VLD2DUPd32x2 : VLD2DUP<{1,0,1,?}, "32", VecListDPairSpacedAllLanes>;
+def VLD2DUPd8x2  : VLD2DUP<{0,0,1,?}, "8",  VecListDPairSpacedAllLanes,
+                           addrmode6dupalign16>;
+def VLD2DUPd16x2 : VLD2DUP<{0,1,1,?}, "16", VecListDPairSpacedAllLanes,
+                           addrmode6dupalign32>;
+def VLD2DUPd32x2 : VLD2DUP<{1,0,1,?}, "32", VecListDPairSpacedAllLanes,
+                           addrmode6dupalign64>;
 
 // ...with address register writeback:
-multiclass VLD2DUPWB<bits<4> op7_4, string Dt, RegisterOperand VdTy> {
+multiclass VLD2DUPWB<bits<4> op7_4, string Dt, RegisterOperand VdTy,
+                     Operand AddrMode> {
   def _fixed : NLdSt<1, 0b10, 0b1101, op7_4,
                      (outs VdTy:$Vd, GPR:$wb),
-                     (ins addrmode6dup:$Rn), IIC_VLD2dupu,
+                     (ins AddrMode:$Rn), IIC_VLD2dupu,
                      "vld2", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1412,7 +1487,7 @@ multiclass VLD2DUPWB<bits<4> op7_4, string Dt, RegisterOperand VdTy> {
   }
   def _register : NLdSt<1, 0b10, 0b1101, op7_4,
                         (outs VdTy:$Vd, GPR:$wb),
-                        (ins addrmode6dup:$Rn, rGPR:$Rm), IIC_VLD2dupu,
+                        (ins AddrMode:$Rn, rGPR:$Rm), IIC_VLD2dupu,
                         "vld2", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{4} = Rn{4};
@@ -1420,13 +1495,19 @@ multiclass VLD2DUPWB<bits<4> op7_4, string Dt, RegisterOperand VdTy> {
   }
 }
 
-defm VLD2DUPd8wb    : VLD2DUPWB<{0,0,0,0}, "8",  VecListDPairAllLanes>;
-defm VLD2DUPd16wb   : VLD2DUPWB<{0,1,0,?}, "16", VecListDPairAllLanes>;
-defm VLD2DUPd32wb   : VLD2DUPWB<{1,0,0,?}, "32", VecListDPairAllLanes>;
+defm VLD2DUPd8wb    : VLD2DUPWB<{0,0,0,0}, "8",  VecListDPairAllLanes,
+                                addrmode6dupalign16>;
+defm VLD2DUPd16wb   : VLD2DUPWB<{0,1,0,?}, "16", VecListDPairAllLanes,
+                                addrmode6dupalign32>;
+defm VLD2DUPd32wb   : VLD2DUPWB<{1,0,0,?}, "32", VecListDPairAllLanes,
+                                addrmode6dupalign64>;
 
-defm VLD2DUPd8x2wb  : VLD2DUPWB<{0,0,1,0}, "8",  VecListDPairSpacedAllLanes>;
-defm VLD2DUPd16x2wb : VLD2DUPWB<{0,1,1,?}, "16", VecListDPairSpacedAllLanes>;
-defm VLD2DUPd32x2wb : VLD2DUPWB<{1,0,1,?}, "32", VecListDPairSpacedAllLanes>;
+defm VLD2DUPd8x2wb  : VLD2DUPWB<{0,0,1,0}, "8",  VecListDPairSpacedAllLanes,
+                                addrmode6dupalign16>;
+defm VLD2DUPd16x2wb : VLD2DUPWB<{0,1,1,?}, "16", VecListDPairSpacedAllLanes,
+                                addrmode6dupalign32>;
+defm VLD2DUPd32x2wb : VLD2DUPWB<{1,0,1,?}, "32", VecListDPairSpacedAllLanes,
+                                addrmode6dupalign64>;
 
 //   VLD3DUP  : Vector Load (single 3-element structure to all lanes)
 class VLD3DUP<bits<4> op7_4, string Dt>
@@ -1452,22 +1533,22 @@ def VLD3DUPq16 : VLD3DUP<{0,1,1,?}, "16">;
 def VLD3DUPq32 : VLD3DUP<{1,0,1,?}, "32">;
 
 // ...with address register writeback:
-class VLD3DUPWB<bits<4> op7_4, string Dt>
+class VLD3DUPWB<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<1, 0b10, 0b1110, op7_4, (outs DPR:$Vd, DPR:$dst2, DPR:$dst3, GPR:$wb),
-          (ins addrmode6dup:$Rn, am6offset:$Rm), IIC_VLD3dupu,
+          (ins AddrMode:$Rn, am6offset:$Rm), IIC_VLD3dupu,
           "vld3", Dt, "\\{$Vd[], $dst2[], $dst3[]\\}, $Rn$Rm",
           "$Rn.addr = $wb", []> {
   let Inst{4} = 0;
   let DecoderMethod = "DecodeVLD3DupInstruction";
 }
 
-def VLD3DUPd8_UPD  : VLD3DUPWB<{0,0,0,0}, "8">;
-def VLD3DUPd16_UPD : VLD3DUPWB<{0,1,0,?}, "16">;
-def VLD3DUPd32_UPD : VLD3DUPWB<{1,0,0,?}, "32">;
+def VLD3DUPd8_UPD  : VLD3DUPWB<{0,0,0,0}, "8",  addrmode6dupalign64>;
+def VLD3DUPd16_UPD : VLD3DUPWB<{0,1,0,?}, "16", addrmode6dupalign64>;
+def VLD3DUPd32_UPD : VLD3DUPWB<{1,0,0,?}, "32", addrmode6dupalign64>;
 
-def VLD3DUPq8_UPD  : VLD3DUPWB<{0,0,1,0}, "8">;
-def VLD3DUPq16_UPD : VLD3DUPWB<{0,1,1,?}, "16">;
-def VLD3DUPq32_UPD : VLD3DUPWB<{1,0,1,?}, "32">;
+def VLD3DUPq8_UPD  : VLD3DUPWB<{0,0,1,0}, "8",  addrmode6dupalign64>;
+def VLD3DUPq16_UPD : VLD3DUPWB<{0,1,1,?}, "16", addrmode6dupalign64>;
+def VLD3DUPq32_UPD : VLD3DUPWB<{1,0,1,?}, "32", addrmode6dupalign64>;
 
 def VLD3DUPd8Pseudo_UPD  : VLDQQWBPseudo<IIC_VLD3dupu>;
 def VLD3DUPd16Pseudo_UPD : VLDQQWBPseudo<IIC_VLD3dupu>;
@@ -1563,35 +1644,35 @@ class VSTQQQQWBPseudo<InstrItinClass itin>
                 "$addr.addr = $wb">;
 
 //   VST1     : Vector Store (multiple single elements)
-class VST1D<bits<4> op7_4, string Dt>
-  : NLdSt<0,0b00,0b0111,op7_4, (outs), (ins addrmode6:$Rn, VecListOneD:$Vd),
+class VST1D<bits<4> op7_4, string Dt, Operand AddrMode>
+  : NLdSt<0,0b00,0b0111,op7_4, (outs), (ins AddrMode:$Rn, VecListOneD:$Vd),
           IIC_VST1, "vst1", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
-class VST1Q<bits<4> op7_4, string Dt>
-  : NLdSt<0,0b00,0b1010,op7_4, (outs), (ins addrmode6:$Rn, VecListDPair:$Vd),
+class VST1Q<bits<4> op7_4, string Dt, Operand AddrMode>
+  : NLdSt<0,0b00,0b1010,op7_4, (outs), (ins AddrMode:$Rn, VecListDPair:$Vd),
           IIC_VST1x2, "vst1", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{5-4} = Rn{5-4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
 
-def  VST1d8   : VST1D<{0,0,0,?}, "8">;
-def  VST1d16  : VST1D<{0,1,0,?}, "16">;
-def  VST1d32  : VST1D<{1,0,0,?}, "32">;
-def  VST1d64  : VST1D<{1,1,0,?}, "64">;
+def  VST1d8   : VST1D<{0,0,0,?}, "8",  addrmode6align64>;
+def  VST1d16  : VST1D<{0,1,0,?}, "16", addrmode6align64>;
+def  VST1d32  : VST1D<{1,0,0,?}, "32", addrmode6align64>;
+def  VST1d64  : VST1D<{1,1,0,?}, "64", addrmode6align64>;
 
-def  VST1q8   : VST1Q<{0,0,?,?}, "8">;
-def  VST1q16  : VST1Q<{0,1,?,?}, "16">;
-def  VST1q32  : VST1Q<{1,0,?,?}, "32">;
-def  VST1q64  : VST1Q<{1,1,?,?}, "64">;
+def  VST1q8   : VST1Q<{0,0,?,?}, "8",  addrmode6align64or128>;
+def  VST1q16  : VST1Q<{0,1,?,?}, "16", addrmode6align64or128>;
+def  VST1q32  : VST1Q<{1,0,?,?}, "32", addrmode6align64or128>;
+def  VST1q64  : VST1Q<{1,1,?,?}, "64", addrmode6align64or128>;
 
 // ...with address register writeback:
-multiclass VST1DWB<bits<4> op7_4, string Dt> {
+multiclass VST1DWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b00, 0b0111,op7_4, (outs GPR:$wb),
-                     (ins addrmode6:$Rn, VecListOneD:$Vd), IIC_VLD1u,
+                     (ins AddrMode:$Rn, VecListOneD:$Vd), IIC_VLD1u,
                      "vst1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1599,7 +1680,7 @@ multiclass VST1DWB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b00,0b0111,op7_4, (outs GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm, VecListOneD:$Vd),
+                        (ins AddrMode:$Rn, rGPR:$Rm, VecListOneD:$Vd),
                         IIC_VLD1u,
                         "vst1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
@@ -1607,9 +1688,9 @@ multiclass VST1DWB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
 }
-multiclass VST1QWB<bits<4> op7_4, string Dt> {
+multiclass VST1QWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b00,0b1010,op7_4, (outs GPR:$wb),
-                    (ins addrmode6:$Rn, VecListDPair:$Vd), IIC_VLD1x2u,
+                    (ins AddrMode:$Rn, VecListDPair:$Vd), IIC_VLD1x2u,
                      "vst1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1617,7 +1698,7 @@ multiclass VST1QWB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b00,0b1010,op7_4, (outs GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm, VecListDPair:$Vd),
+                        (ins AddrMode:$Rn, rGPR:$Rm, VecListDPair:$Vd),
                         IIC_VLD1x2u,
                         "vst1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
@@ -1626,28 +1707,28 @@ multiclass VST1QWB<bits<4> op7_4, string Dt> {
   }
 }
 
-defm VST1d8wb  : VST1DWB<{0,0,0,?}, "8">;
-defm VST1d16wb : VST1DWB<{0,1,0,?}, "16">;
-defm VST1d32wb : VST1DWB<{1,0,0,?}, "32">;
-defm VST1d64wb : VST1DWB<{1,1,0,?}, "64">;
+defm VST1d8wb  : VST1DWB<{0,0,0,?}, "8",  addrmode6align64>;
+defm VST1d16wb : VST1DWB<{0,1,0,?}, "16", addrmode6align64>;
+defm VST1d32wb : VST1DWB<{1,0,0,?}, "32", addrmode6align64>;
+defm VST1d64wb : VST1DWB<{1,1,0,?}, "64", addrmode6align64>;
 
-defm VST1q8wb  : VST1QWB<{0,0,?,?}, "8">;
-defm VST1q16wb : VST1QWB<{0,1,?,?}, "16">;
-defm VST1q32wb : VST1QWB<{1,0,?,?}, "32">;
-defm VST1q64wb : VST1QWB<{1,1,?,?}, "64">;
+defm VST1q8wb  : VST1QWB<{0,0,?,?}, "8",  addrmode6align64or128>;
+defm VST1q16wb : VST1QWB<{0,1,?,?}, "16", addrmode6align64or128>;
+defm VST1q32wb : VST1QWB<{1,0,?,?}, "32", addrmode6align64or128>;
+defm VST1q64wb : VST1QWB<{1,1,?,?}, "64", addrmode6align64or128>;
 
 // ...with 3 registers
-class VST1D3<bits<4> op7_4, string Dt>
+class VST1D3<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<0, 0b00, 0b0110, op7_4, (outs),
-          (ins addrmode6:$Rn, VecListThreeD:$Vd),
+          (ins AddrMode:$Rn, VecListThreeD:$Vd),
           IIC_VST1x3, "vst1", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
-multiclass VST1D3WB<bits<4> op7_4, string Dt> {
+multiclass VST1D3WB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b00,0b0110,op7_4, (outs GPR:$wb),
-                    (ins addrmode6:$Rn, VecListThreeD:$Vd), IIC_VLD1x3u,
+                    (ins AddrMode:$Rn, VecListThreeD:$Vd), IIC_VLD1x3u,
                      "vst1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1655,7 +1736,7 @@ multiclass VST1D3WB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b00,0b0110,op7_4, (outs GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm, VecListThreeD:$Vd),
+                        (ins AddrMode:$Rn, rGPR:$Rm, VecListThreeD:$Vd),
                         IIC_VLD1x3u,
                         "vst1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
@@ -1664,33 +1745,33 @@ multiclass VST1D3WB<bits<4> op7_4, string Dt> {
   }
 }
 
-def VST1d8T     : VST1D3<{0,0,0,?}, "8">;
-def VST1d16T    : VST1D3<{0,1,0,?}, "16">;
-def VST1d32T    : VST1D3<{1,0,0,?}, "32">;
-def VST1d64T    : VST1D3<{1,1,0,?}, "64">;
+def VST1d8T     : VST1D3<{0,0,0,?}, "8",  addrmode6align64>;
+def VST1d16T    : VST1D3<{0,1,0,?}, "16", addrmode6align64>;
+def VST1d32T    : VST1D3<{1,0,0,?}, "32", addrmode6align64>;
+def VST1d64T    : VST1D3<{1,1,0,?}, "64", addrmode6align64>;
 
-defm VST1d8Twb  : VST1D3WB<{0,0,0,?}, "8">;
-defm VST1d16Twb : VST1D3WB<{0,1,0,?}, "16">;
-defm VST1d32Twb : VST1D3WB<{1,0,0,?}, "32">;
-defm VST1d64Twb : VST1D3WB<{1,1,0,?}, "64">;
+defm VST1d8Twb  : VST1D3WB<{0,0,0,?}, "8",  addrmode6align64>;
+defm VST1d16Twb : VST1D3WB<{0,1,0,?}, "16", addrmode6align64>;
+defm VST1d32Twb : VST1D3WB<{1,0,0,?}, "32", addrmode6align64>;
+defm VST1d64Twb : VST1D3WB<{1,1,0,?}, "64", addrmode6align64>;
 
 def VST1d64TPseudo            : VSTQQPseudo<IIC_VST1x3>;
 def VST1d64TPseudoWB_fixed    : VSTQQWBfixedPseudo<IIC_VST1x3u>;
 def VST1d64TPseudoWB_register : VSTQQWBPseudo<IIC_VST1x3u>;
 
 // ...with 4 registers
-class VST1D4<bits<4> op7_4, string Dt>
+class VST1D4<bits<4> op7_4, string Dt, Operand AddrMode>
   : NLdSt<0, 0b00, 0b0010, op7_4, (outs),
-          (ins addrmode6:$Rn, VecListFourD:$Vd),
+          (ins AddrMode:$Rn, VecListFourD:$Vd),
           IIC_VST1x4, "vst1", Dt, "$Vd, $Rn", "",
           []> {
   let Rm = 0b1111;
   let Inst{5-4} = Rn{5-4};
   let DecoderMethod = "DecodeVLDST1Instruction";
 }
-multiclass VST1D4WB<bits<4> op7_4, string Dt> {
+multiclass VST1D4WB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0,0b00,0b0010,op7_4, (outs GPR:$wb),
-                    (ins addrmode6:$Rn, VecListFourD:$Vd), IIC_VLD1x4u,
+                    (ins AddrMode:$Rn, VecListFourD:$Vd), IIC_VLD1x4u,
                      "vst1", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1698,7 +1779,7 @@ multiclass VST1D4WB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST1Instruction";
   }
   def _register : NLdSt<0,0b00,0b0010,op7_4, (outs GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm, VecListFourD:$Vd),
+                        (ins AddrMode:$Rn, rGPR:$Rm, VecListFourD:$Vd),
                         IIC_VLD1x4u,
                         "vst1", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
@@ -1707,15 +1788,15 @@ multiclass VST1D4WB<bits<4> op7_4, string Dt> {
   }
 }
 
-def VST1d8Q     : VST1D4<{0,0,?,?}, "8">;
-def VST1d16Q    : VST1D4<{0,1,?,?}, "16">;
-def VST1d32Q    : VST1D4<{1,0,?,?}, "32">;
-def VST1d64Q    : VST1D4<{1,1,?,?}, "64">;
+def VST1d8Q     : VST1D4<{0,0,?,?}, "8",  addrmode6align64or128or256>;
+def VST1d16Q    : VST1D4<{0,1,?,?}, "16", addrmode6align64or128or256>;
+def VST1d32Q    : VST1D4<{1,0,?,?}, "32", addrmode6align64or128or256>;
+def VST1d64Q    : VST1D4<{1,1,?,?}, "64", addrmode6align64or128or256>;
 
-defm VST1d8Qwb  : VST1D4WB<{0,0,?,?}, "8">;
-defm VST1d16Qwb : VST1D4WB<{0,1,?,?}, "16">;
-defm VST1d32Qwb : VST1D4WB<{1,0,?,?}, "32">;
-defm VST1d64Qwb : VST1D4WB<{1,1,?,?}, "64">;
+defm VST1d8Qwb  : VST1D4WB<{0,0,?,?}, "8",  addrmode6align64or128or256>;
+defm VST1d16Qwb : VST1D4WB<{0,1,?,?}, "16", addrmode6align64or128or256>;
+defm VST1d32Qwb : VST1D4WB<{1,0,?,?}, "32", addrmode6align64or128or256>;
+defm VST1d64Qwb : VST1D4WB<{1,1,?,?}, "64", addrmode6align64or128or256>;
 
 def VST1d64QPseudo            : VSTQQPseudo<IIC_VST1x4>;
 def VST1d64QPseudoWB_fixed    : VSTQQWBfixedPseudo<IIC_VST1x4u>;
@@ -1723,21 +1804,27 @@ def VST1d64QPseudoWB_register : VSTQQWBPseudo<IIC_VST1x4u>;
 
 //   VST2     : Vector Store (multiple 2-element structures)
 class VST2<bits<4> op11_8, bits<4> op7_4, string Dt, RegisterOperand VdTy,
-            InstrItinClass itin>
-  : NLdSt<0, 0b00, op11_8, op7_4, (outs), (ins addrmode6:$Rn, VdTy:$Vd),
+            InstrItinClass itin, Operand AddrMode>
+  : NLdSt<0, 0b00, op11_8, op7_4, (outs), (ins AddrMode:$Rn, VdTy:$Vd),
           itin, "vst2", Dt, "$Vd, $Rn", "", []> {
   let Rm = 0b1111;
   let Inst{5-4} = Rn{5-4};
   let DecoderMethod = "DecodeVLDST2Instruction";
 }
 
-def  VST2d8   : VST2<0b1000, {0,0,?,?}, "8",  VecListDPair, IIC_VST2>;
-def  VST2d16  : VST2<0b1000, {0,1,?,?}, "16", VecListDPair, IIC_VST2>;
-def  VST2d32  : VST2<0b1000, {1,0,?,?}, "32", VecListDPair, IIC_VST2>;
+def  VST2d8   : VST2<0b1000, {0,0,?,?}, "8",  VecListDPair, IIC_VST2,
+                     addrmode6align64or128>;
+def  VST2d16  : VST2<0b1000, {0,1,?,?}, "16", VecListDPair, IIC_VST2,
+                     addrmode6align64or128>;
+def  VST2d32  : VST2<0b1000, {1,0,?,?}, "32", VecListDPair, IIC_VST2,
+                     addrmode6align64or128>;
 
-def  VST2q8   : VST2<0b0011, {0,0,?,?}, "8",  VecListFourD, IIC_VST2x2>;
-def  VST2q16  : VST2<0b0011, {0,1,?,?}, "16", VecListFourD, IIC_VST2x2>;
-def  VST2q32  : VST2<0b0011, {1,0,?,?}, "32", VecListFourD, IIC_VST2x2>;
+def  VST2q8   : VST2<0b0011, {0,0,?,?}, "8",  VecListFourD, IIC_VST2x2,
+                     addrmode6align64or128or256>;
+def  VST2q16  : VST2<0b0011, {0,1,?,?}, "16", VecListFourD, IIC_VST2x2,
+                     addrmode6align64or128or256>;
+def  VST2q32  : VST2<0b0011, {1,0,?,?}, "32", VecListFourD, IIC_VST2x2,
+                     addrmode6align64or128or256>;
 
 def  VST2q8Pseudo  : VSTQQPseudo<IIC_VST2x2>;
 def  VST2q16Pseudo : VSTQQPseudo<IIC_VST2x2>;
@@ -1745,9 +1832,9 @@ def  VST2q32Pseudo : VSTQQPseudo<IIC_VST2x2>;
 
 // ...with address register writeback:
 multiclass VST2DWB<bits<4> op11_8, bits<4> op7_4, string Dt,
-                   RegisterOperand VdTy> {
+                   RegisterOperand VdTy, Operand AddrMode> {
   def _fixed : NLdSt<0, 0b00, op11_8, op7_4, (outs GPR:$wb),
-                     (ins addrmode6:$Rn, VdTy:$Vd), IIC_VLD1u,
+                     (ins AddrMode:$Rn, VdTy:$Vd), IIC_VLD1u,
                      "vst2", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1755,16 +1842,16 @@ multiclass VST2DWB<bits<4> op11_8, bits<4> op7_4, string Dt,
     let DecoderMethod = "DecodeVLDST2Instruction";
   }
   def _register : NLdSt<0, 0b00, op11_8, op7_4, (outs GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm, VdTy:$Vd), IIC_VLD1u,
+                        (ins AddrMode:$Rn, rGPR:$Rm, VdTy:$Vd), IIC_VLD1u,
                         "vst2", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
     let Inst{5-4} = Rn{5-4};
     let DecoderMethod = "DecodeVLDST2Instruction";
   }
 }
-multiclass VST2QWB<bits<4> op7_4, string Dt> {
+multiclass VST2QWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<0, 0b00, 0b0011, op7_4, (outs GPR:$wb),
-                     (ins addrmode6:$Rn, VecListFourD:$Vd), IIC_VLD1u,
+                     (ins AddrMode:$Rn, VecListFourD:$Vd), IIC_VLD1u,
                      "vst2", Dt, "$Vd, $Rn!",
                      "$Rn.addr = $wb", []> {
     let Rm = 0b1101; // NLdSt will assign to the right encoding bits.
@@ -1772,7 +1859,7 @@ multiclass VST2QWB<bits<4> op7_4, string Dt> {
     let DecoderMethod = "DecodeVLDST2Instruction";
   }
   def _register : NLdSt<0, 0b00, 0b0011, op7_4, (outs GPR:$wb),
-                        (ins addrmode6:$Rn, rGPR:$Rm, VecListFourD:$Vd),
+                        (ins AddrMode:$Rn, rGPR:$Rm, VecListFourD:$Vd),
                         IIC_VLD1u,
                         "vst2", Dt, "$Vd, $Rn, $Rm",
                         "$Rn.addr = $wb", []> {
@@ -1781,13 +1868,16 @@ multiclass VST2QWB<bits<4> op7_4, string Dt> {
   }
 }
 
-defm VST2d8wb    : VST2DWB<0b1000, {0,0,?,?}, "8",  VecListDPair>;
-defm VST2d16wb   : VST2DWB<0b1000, {0,1,?,?}, "16", VecListDPair>;
-defm VST2d32wb   : VST2DWB<0b1000, {1,0,?,?}, "32", VecListDPair>;
+defm VST2d8wb    : VST2DWB<0b1000, {0,0,?,?}, "8",  VecListDPair,
+                           addrmode6align64or128>;
+defm VST2d16wb   : VST2DWB<0b1000, {0,1,?,?}, "16", VecListDPair,
+                           addrmode6align64or128>;
+defm VST2d32wb   : VST2DWB<0b1000, {1,0,?,?}, "32", VecListDPair,
+                           addrmode6align64or128>;
 
-defm VST2q8wb    : VST2QWB<{0,0,?,?}, "8">;
-defm VST2q16wb   : VST2QWB<{0,1,?,?}, "16">;
-defm VST2q32wb   : VST2QWB<{1,0,?,?}, "32">;
+defm VST2q8wb    : VST2QWB<{0,0,?,?}, "8", addrmode6align64or128or256>;
+defm VST2q16wb   : VST2QWB<{0,1,?,?}, "16", addrmode6align64or128or256>;
+defm VST2q32wb   : VST2QWB<{1,0,?,?}, "32", addrmode6align64or128or256>;
 
 def VST2q8PseudoWB_fixed     : VSTQQWBfixedPseudo<IIC_VST2x2u>;
 def VST2q16PseudoWB_fixed    : VSTQQWBfixedPseudo<IIC_VST2x2u>;
@@ -1797,12 +1887,18 @@ def VST2q16PseudoWB_register : VSTQQWBregisterPseudo<IIC_VST2x2u>;
 def VST2q32PseudoWB_register : VSTQQWBregisterPseudo<IIC_VST2x2u>;
 
 // ...with double-spaced registers
-def VST2b8      : VST2<0b1001, {0,0,?,?}, "8",  VecListDPairSpaced, IIC_VST2>;
-def VST2b16     : VST2<0b1001, {0,1,?,?}, "16", VecListDPairSpaced, IIC_VST2>;
-def VST2b32     : VST2<0b1001, {1,0,?,?}, "32", VecListDPairSpaced, IIC_VST2>;
-defm VST2b8wb   : VST2DWB<0b1001, {0,0,?,?}, "8",  VecListDPairSpaced>;
-defm VST2b16wb  : VST2DWB<0b1001, {0,1,?,?}, "16", VecListDPairSpaced>;
-defm VST2b32wb  : VST2DWB<0b1001, {1,0,?,?}, "32", VecListDPairSpaced>;
+def VST2b8      : VST2<0b1001, {0,0,?,?}, "8",  VecListDPairSpaced, IIC_VST2,
+                      addrmode6align64or128>;
+def VST2b16     : VST2<0b1001, {0,1,?,?}, "16", VecListDPairSpaced, IIC_VST2,
+                      addrmode6align64or128>;
+def VST2b32     : VST2<0b1001, {1,0,?,?}, "32", VecListDPairSpaced, IIC_VST2,
+                      addrmode6align64or128>;
+defm VST2b8wb   : VST2DWB<0b1001, {0,0,?,?}, "8",  VecListDPairSpaced,
+                          addrmode6align64or128>;
+defm VST2b16wb  : VST2DWB<0b1001, {0,1,?,?}, "16", VecListDPairSpaced,
+                          addrmode6align64or128>;
+defm VST2b32wb  : VST2DWB<0b1001, {1,0,?,?}, "32", VecListDPairSpaced,
+                          addrmode6align64or128>;
 
 //   VST3     : Vector Store (multiple 3-element structures)
 class VST3D<bits<4> op11_8, bits<4> op7_4, string Dt>
@@ -2270,9 +2366,9 @@ def : Pat<(v2f64 (dword_alignedload addrmode6:$addr)),
 def : Pat<(dword_alignedstore (v2f64 QPR:$value), addrmode6:$addr),
           (VST1q64 addrmode6:$addr, QPR:$value)>;
 def : Pat<(v2f64 (word_alignedload addrmode6:$addr)),
-          (VLD1q32 addrmode6:$addr)>;
+          (VLD1q32 addrmode6:$addr)>, Requires<[IsLE]>;
 def : Pat<(word_alignedstore (v2f64 QPR:$value), addrmode6:$addr),
-          (VST1q32 addrmode6:$addr, QPR:$value)>;
+          (VST1q32 addrmode6:$addr, QPR:$value)>, Requires<[IsLE]>;
 def : Pat<(v2f64 (hword_alignedload addrmode6:$addr)),
           (VLD1q16 addrmode6:$addr)>, Requires<[IsLE]>;
 def : Pat<(hword_alignedstore (v2f64 QPR:$value), addrmode6:$addr),
@@ -2360,14 +2456,14 @@ class N2VDIntnp<bits<2> op17_16, bits<3> op10_8, bit op7,
               InstrItinClass itin, string OpcodeStr, string Dt,
               ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N2Vnp<0b10, op17_16, op10_8, op7, 0,  (outs DPR:$Vd), (ins DPR:$Vm),
-          itin, OpcodeStr, Dt, ResTy, OpTy,
+          itin, OpcodeStr, Dt,
           [(set DPR:$Vd, (ResTy (IntOp (OpTy DPR:$Vm))))]>;
 
 class N2VQIntnp<bits<2> op17_16, bits<3> op10_8, bit op7,
               InstrItinClass itin, string OpcodeStr, string Dt,
               ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N2Vnp<0b10, op17_16, op10_8, op7, 1,  (outs QPR:$Vd), (ins QPR:$Vm),
-          itin, OpcodeStr, Dt, ResTy, OpTy,
+          itin, OpcodeStr, Dt,
           [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$Vm))))]>;
 
 // Similar to NV2VQIntnp with some more encoding bits exposed (crypto).
@@ -2375,7 +2471,7 @@ class N2VQIntXnp<bits<2> op19_18, bits<2> op17_16, bits<3> op10_8, bit op6,
               bit op7, InstrItinClass itin, string OpcodeStr, string Dt,
               ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N2Vnp<op19_18, op17_16, op10_8, op7, op6,  (outs QPR:$Vd), (ins QPR:$Vm),
-          itin, OpcodeStr, Dt, ResTy, OpTy,
+          itin, OpcodeStr, Dt,
           [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$Vm))))]>;
 
 // Same as N2VQIntXnp but with Vd as a src register.
@@ -2384,7 +2480,7 @@ class N2VQIntX2np<bits<2> op19_18, bits<2> op17_16, bits<3> op10_8, bit op6,
               ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N2Vnp<op19_18, op17_16, op10_8, op7, op6,
           (outs QPR:$Vd), (ins QPR:$src, QPR:$Vm),
-          itin, OpcodeStr, Dt, ResTy, OpTy,
+          itin, OpcodeStr, Dt,
           [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$src), (OpTy QPR:$Vm))))]> {
   let Constraints = "$src = $Vd";
 }
@@ -2558,7 +2654,6 @@ class N3VDIntnp<bits<5> op27_23, bits<2> op21_20, bits<4> op11_8, bit op6,
                 SDPatternOperator IntOp, bit Commutable>
   : N3Vnp<op27_23, op21_20, op11_8, op6, op4,
           (outs DPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin, OpcodeStr, Dt,
-          ResTy, OpTy, IntOp, Commutable,
           [(set DPR:$Vd, (ResTy (IntOp (OpTy DPR:$Vn), (OpTy DPR:$Vm))))]>;
 
 class N3VDIntSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
@@ -2612,7 +2707,6 @@ class N3VQIntnp<bits<5> op27_23, bits<2> op21_20, bits<4> op11_8, bit op6,
                 SDPatternOperator IntOp, bit Commutable>
   : N3Vnp<op27_23, op21_20, op11_8, op6, op4,
           (outs QPR:$Vd), (ins QPR:$Vn, QPR:$Vm), f, itin, OpcodeStr, Dt,
-          ResTy, OpTy, IntOp, Commutable,
           [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$Vn), (OpTy QPR:$Vm))))]>;
 
 // Same as N3VQIntnp but with Vd as a src register.
@@ -2621,8 +2715,8 @@ class N3VQInt3np<bits<5> op27_23, bits<2> op21_20, bits<4> op11_8, bit op6,
                 string Dt, ValueType ResTy, ValueType OpTy,
                 SDPatternOperator IntOp, bit Commutable>
   : N3Vnp<op27_23, op21_20, op11_8, op6, op4,
-          (outs QPR:$Vd), (ins QPR:$src, QPR:$Vn, QPR:$Vm), f, itin, OpcodeStr,
-          Dt, ResTy, OpTy, IntOp, Commutable,
+          (outs QPR:$Vd), (ins QPR:$src, QPR:$Vn, QPR:$Vm),
+          f, itin, OpcodeStr, Dt,
           [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$src), (OpTy QPR:$Vn),
                                        (OpTy QPR:$Vm))))]> {
   let Constraints = "$src = $Vd";
@@ -2942,7 +3036,6 @@ class N3VLIntnp<bits<5> op27_23, bits<2> op21_20, bits<4> op11_8, bit op6,
                 SDPatternOperator IntOp, bit Commutable>
   : N3Vnp<op27_23, op21_20, op11_8, op6, op4,
           (outs QPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin, OpcodeStr, Dt,
-          ResTy, OpTy, IntOp, Commutable,
           [(set QPR:$Vd, (ResTy (IntOp (OpTy DPR:$Vn), (OpTy DPR:$Vm))))]>;
 
 class N3VLIntSL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
@@ -3038,22 +3131,23 @@ class N2VQSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
 // Long shift by immediate.
 class N2VLSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
              string OpcodeStr, string Dt,
-             ValueType ResTy, ValueType OpTy, Operand ImmTy, SDNode OpNode>
+             ValueType ResTy, ValueType OpTy, Operand ImmTy,
+             SDPatternOperator OpNode>
   : N2VImm<op24, op23, op11_8, op7, op6, op4,
            (outs QPR:$Vd), (ins DPR:$Vm, ImmTy:$SIMM), N2RegVShLFrm,
            IIC_VSHLiD, OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "",
-           [(set QPR:$Vd, (ResTy (OpNode (OpTy DPR:$Vm),
-                                          (i32 imm:$SIMM))))]>;
+           [(set QPR:$Vd, (ResTy (OpNode (OpTy DPR:$Vm), ImmTy:$SIMM)))]>;
 
 // Narrow shift by immediate.
 class N2VNSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
              InstrItinClass itin, string OpcodeStr, string Dt,
-             ValueType ResTy, ValueType OpTy, Operand ImmTy, SDNode OpNode>
+             ValueType ResTy, ValueType OpTy, Operand ImmTy,
+             SDPatternOperator OpNode>
   : N2VImm<op24, op23, op11_8, op7, op6, op4,
            (outs DPR:$Vd), (ins QPR:$Vm, ImmTy:$SIMM), N2RegVShRFrm, itin,
            OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "",
            [(set DPR:$Vd, (ResTy (OpNode (OpTy QPR:$Vm),
-                                          (i32 imm:$SIMM))))]>;
+                                          (i32 ImmTy:$SIMM))))]>;
 
 // Shift right by immediate and accumulate,
 // both double- and quad-register.
@@ -3941,7 +4035,8 @@ multiclass N2VShInsR_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
 // Neon Shift Long operations,
 //   element sizes of 8, 16, 32 bits:
 multiclass N2VLSh_QHS<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6,
-                      bit op4, string OpcodeStr, string Dt, SDNode OpNode> {
+                      bit op4, string OpcodeStr, string Dt,
+                      SDPatternOperator OpNode> {
   def v8i16 : N2VLSh<op24, op23, op11_8, op7, op6, op4,
               OpcodeStr, !strconcat(Dt, "8"), v8i16, v8i8, imm1_7, OpNode> {
     let Inst{21-19} = 0b001; // imm6 = 001xxx
@@ -3960,7 +4055,7 @@ multiclass N2VLSh_QHS<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6,
 //   element sizes of 16, 32, 64 bits:
 multiclass N2VNSh_HSD<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6,
                       bit op4, InstrItinClass itin, string OpcodeStr, string Dt,
-                      SDNode OpNode> {
+                      SDPatternOperator OpNode> {
   def v8i8 : N2VNSh<op24, op23, op11_8, op7, op6, op4, itin,
                     OpcodeStr, !strconcat(Dt, "16"),
                     v8i8, v8i16, shr_imm8, OpNode> {
@@ -4427,14 +4522,14 @@ defm VCLTz    : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00100, 0, "vclt", "s",
 
 //   VACGE    : Vector Absolute Compare Greater Than or Equal (aka VCAGE)
 def  VACGEd   : N3VDInt<1, 0, 0b00, 0b1110, 1, N3RegFrm, IIC_VBIND, "vacge",
-                        "f32", v2i32, v2f32, int_arm_neon_vacged, 0>;
+                        "f32", v2i32, v2f32, int_arm_neon_vacge, 0>;
 def  VACGEq   : N3VQInt<1, 0, 0b00, 0b1110, 1, N3RegFrm, IIC_VBINQ, "vacge",
-                        "f32", v4i32, v4f32, int_arm_neon_vacgeq, 0>;
+                        "f32", v4i32, v4f32, int_arm_neon_vacge, 0>;
 //   VACGT    : Vector Absolute Compare Greater Than (aka VCAGT)
 def  VACGTd   : N3VDInt<1, 0, 0b10, 0b1110, 1, N3RegFrm, IIC_VBIND, "vacgt",
-                        "f32", v2i32, v2f32, int_arm_neon_vacgtd, 0>;
+                        "f32", v2i32, v2f32, int_arm_neon_vacgt, 0>;
 def  VACGTq   : N3VQInt<1, 0, 0b10, 0b1110, 1, N3RegFrm, IIC_VBINQ, "vacgt",
-                        "f32", v4i32, v4f32, int_arm_neon_vacgtq, 0>;
+                        "f32", v4i32, v4f32, int_arm_neon_vacgt, 0>;
 //   VTST     : Vector Test Bits
 defm VTST     : N3V_QHS<0, 0, 0b1000, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q,
                         IIC_VBINi4Q, "vtst", "", NEONvtst, 1>;
@@ -4946,28 +5041,51 @@ defm VSHRu    : N2VShR_QHSD<1, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "u", "VSHRu",
                             NEONvshru>;
 
 //   VSHLL    : Vector Shift Left Long
-defm VSHLLs   : N2VLSh_QHS<0, 1, 0b1010, 0, 0, 1, "vshll", "s", NEONvshlls>;
-defm VSHLLu   : N2VLSh_QHS<1, 1, 0b1010, 0, 0, 1, "vshll", "u", NEONvshllu>;
+defm VSHLLs   : N2VLSh_QHS<0, 1, 0b1010, 0, 0, 1, "vshll", "s",
+  PatFrag<(ops node:$LHS, node:$RHS), (NEONvshl (sext node:$LHS), node:$RHS)>>;
+defm VSHLLu   : N2VLSh_QHS<1, 1, 0b1010, 0, 0, 1, "vshll", "u",
+  PatFrag<(ops node:$LHS, node:$RHS), (NEONvshl (zext node:$LHS), node:$RHS)>>;
 
 //   VSHLL    : Vector Shift Left Long (with maximum shift count)
 class N2VLShMax<bit op24, bit op23, bits<6> op21_16, bits<4> op11_8, bit op7,
                 bit op6, bit op4, string OpcodeStr, string Dt, ValueType ResTy,
-                ValueType OpTy, Operand ImmTy, SDNode OpNode>
+                ValueType OpTy, Operand ImmTy>
   : N2VLSh<op24, op23, op11_8, op7, op6, op4, OpcodeStr, Dt,
-           ResTy, OpTy, ImmTy, OpNode> {
+           ResTy, OpTy, ImmTy, null_frag> {
   let Inst{21-16} = op21_16;
   let DecoderMethod = "DecodeVSHLMaxInstruction";
 }
 def  VSHLLi8  : N2VLShMax<1, 1, 0b110010, 0b0011, 0, 0, 0, "vshll", "i8",
-                          v8i16, v8i8, imm8, NEONvshlli>;
+                          v8i16, v8i8, imm8>;
 def  VSHLLi16 : N2VLShMax<1, 1, 0b110110, 0b0011, 0, 0, 0, "vshll", "i16",
-                          v4i32, v4i16, imm16, NEONvshlli>;
+                          v4i32, v4i16, imm16>;
 def  VSHLLi32 : N2VLShMax<1, 1, 0b111010, 0b0011, 0, 0, 0, "vshll", "i32",
-                          v2i64, v2i32, imm32, NEONvshlli>;
+                          v2i64, v2i32, imm32>;
+
+def : Pat<(v8i16 (NEONvshl (zext (v8i8 DPR:$Rn)), (i32 8))),
+          (VSHLLi8 DPR:$Rn, 8)>;
+def : Pat<(v4i32 (NEONvshl (zext (v4i16 DPR:$Rn)), (i32 16))),
+          (VSHLLi16 DPR:$Rn, 16)>;
+def : Pat<(v2i64 (NEONvshl (zext (v2i32 DPR:$Rn)), (i32 32))),
+          (VSHLLi32 DPR:$Rn, 32)>;
+def : Pat<(v8i16 (NEONvshl (sext (v8i8 DPR:$Rn)), (i32 8))),
+          (VSHLLi8 DPR:$Rn, 8)>;
+def : Pat<(v4i32 (NEONvshl (sext (v4i16 DPR:$Rn)), (i32 16))),
+          (VSHLLi16 DPR:$Rn, 16)>;
+def : Pat<(v2i64 (NEONvshl (sext (v2i32 DPR:$Rn)), (i32 32))),
+          (VSHLLi32 DPR:$Rn, 32)>;
 
 //   VSHRN    : Vector Shift Right and Narrow
 defm VSHRN    : N2VNSh_HSD<0,1,0b1000,0,0,1, IIC_VSHLiD, "vshrn", "i",
-                           NEONvshrn>;
+                           PatFrag<(ops node:$Rn, node:$amt),
+                                   (trunc (NEONvshrs node:$Rn, node:$amt))>>;
+
+def : Pat<(v8i8 (trunc (NEONvshru (v8i16 QPR:$Vn), shr_imm8:$amt))),
+          (VSHRNv8i8 QPR:$Vn, shr_imm8:$amt)>;
+def : Pat<(v4i16 (trunc (NEONvshru (v4i32 QPR:$Vn), shr_imm16:$amt))),
+          (VSHRNv4i16 QPR:$Vn, shr_imm16:$amt)>;
+def : Pat<(v2i32 (trunc (NEONvshru (v2i64 QPR:$Vn), shr_imm32:$amt))),
+          (VSHRNv2i32 QPR:$Vn, shr_imm32:$amt)>;
 
 //   VRSHL    : Vector Rounding Shift
 defm VRSHLs   : N3VInt_QHSDSh<0, 0, 0b0101, 0, N3RegVShFrm,
@@ -5077,9 +5195,6 @@ def : Pat<(xor (v4i32 (NEONvshrs QPR:$src, (i32 31))),
                (v4i32 (add QPR:$src, (NEONvshrs QPR:$src, (i32 31))))),
           (VABSv4i32 QPR:$src)>;
 
-def : Pat<(v2f32 (int_arm_neon_vabs (v2f32 DPR:$src))), (VABSfd DPR:$src)>;
-def : Pat<(v4f32 (int_arm_neon_vabs (v4f32 QPR:$src))), (VABSfq QPR:$src)>;
-
 //   VQABS    : Vector Saturating Absolute Value
 defm VQABS    : N2VInt_QHS<0b11, 0b11, 0b00, 0b01110, 0,
                            IIC_VQUNAiD, IIC_VQUNAiQ, "vqabs", "s",
@@ -5226,6 +5341,55 @@ def VMOVv4f32 : N1ModImm<1, 0b000, 0b1111, 0, 1, 0, 1, (outs QPR:$Vd),
                          [(set QPR:$Vd, (v4f32 (NEONvmovFPImm timm:$SIMM)))]>;
 } // isReMaterializable
 
+// Add support for bytes replication feature, so it could be GAS compatible.
+// E.g. instructions below:
+// "vmov.i32 d0, 0xffffffff"
+// "vmov.i32 d0, 0xabababab"
+// "vmov.i16 d0, 0xabab"
+// are incorrect, but we could deal with such cases.
+// For last two instructions, for example, it should emit:
+// "vmov.i8 d0, 0xab"
+def : NEONInstAlias<"vmov${p}.i16 $Vd, $Vm",
+                    (VMOVv8i8 DPR:$Vd, nImmVMOVI16ByteReplicate:$Vm, pred:$p)>;
+def : NEONInstAlias<"vmov${p}.i32 $Vd, $Vm",
+                    (VMOVv8i8 DPR:$Vd, nImmVMOVI32ByteReplicate:$Vm, pred:$p)>;
+def : NEONInstAlias<"vmov${p}.i16 $Vd, $Vm",
+                    (VMOVv16i8 QPR:$Vd, nImmVMOVI16ByteReplicate:$Vm, pred:$p)>;
+def : NEONInstAlias<"vmov${p}.i32 $Vd, $Vm",
+                    (VMOVv16i8 QPR:$Vd, nImmVMOVI32ByteReplicate:$Vm, pred:$p)>;
+
+// Also add same support for VMVN instructions. So instruction:
+// "vmvn.i32 d0, 0xabababab"
+// actually means:
+// "vmov.i8 d0, 0x54"
+def : NEONInstAlias<"vmvn${p}.i16 $Vd, $Vm",
+                    (VMOVv8i8 DPR:$Vd, nImmVMVNI16ByteReplicate:$Vm, pred:$p)>;
+def : NEONInstAlias<"vmvn${p}.i32 $Vd, $Vm",
+                    (VMOVv8i8 DPR:$Vd, nImmVMVNI32ByteReplicate:$Vm, pred:$p)>;
+def : NEONInstAlias<"vmvn${p}.i16 $Vd, $Vm",
+                    (VMOVv16i8 QPR:$Vd, nImmVMVNI16ByteReplicate:$Vm, pred:$p)>;
+def : NEONInstAlias<"vmvn${p}.i32 $Vd, $Vm",
+                    (VMOVv16i8 QPR:$Vd, nImmVMVNI32ByteReplicate:$Vm, pred:$p)>;
+
+// On some CPUs the two instructions "vmov.i32 dD, #0" and "vmov.i32 qD, #0"
+// require zero cycles to execute so they should be used wherever possible for
+// setting a register to zero.
+
+// Even without these pseudo-insts we would probably end up with the correct
+// instruction, but we could not mark the general ones with "isAsCheapAsAMove"
+// since they are sometimes rather expensive (in general).
+
+let AddedComplexity = 50, isAsCheapAsAMove = 1, isReMaterializable = 1 in {
+  def VMOVD0 : ARMPseudoExpand<(outs DPR:$Vd), (ins), 4, IIC_VMOVImm,
+                               [(set DPR:$Vd, (v2i32 NEONimmAllZerosV))],
+                               (VMOVv2i32 DPR:$Vd, 0, (ops 14, zero_reg))>,
+               Requires<[HasZCZ]>;
+  def VMOVQ0 : ARMPseudoExpand<(outs QPR:$Vd), (ins), 4, IIC_VMOVImm,
+                               [(set QPR:$Vd, (v4i32 NEONimmAllZerosV))],
+                               (VMOVv4i32 QPR:$Vd, 0, (ops 14, zero_reg))>,
+               Requires<[HasZCZ]>;
+}
+
 //   VMOV     : Vector Get Lane (move scalar to ARM core register)
 
 def VGETLNs8  : NVGetLane<{1,1,1,0,0,1,?,1}, 0b1011, {?,?},
@@ -5490,10 +5654,12 @@ def : Pat<(v4f32 (NEONvduplane (v4f32 QPR:$src), imm:$lane)),
                                    (DSubReg_i32_reg imm:$lane))),
                            (SubReg_i32_lane imm:$lane)))>;
 
-def  VDUPfdf : PseudoNeonI<(outs DPR:$dst), (ins SPR:$src), IIC_VMOVD, "",
-                    [(set DPR:$dst, (v2f32 (NEONvdup (f32 SPR:$src))))]>;
-def  VDUPfqf : PseudoNeonI<(outs QPR:$dst), (ins SPR:$src), IIC_VMOVD, "",
-                    [(set QPR:$dst, (v4f32 (NEONvdup (f32 SPR:$src))))]>;
+def : Pat<(v2f32 (NEONvdup (f32 SPR:$src))),
+          (v2f32 (VDUPLN32d (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
+                             SPR:$src, ssub_0), (i32 0)))>;
+def : Pat<(v4f32 (NEONvdup (f32 SPR:$src))),
+          (v4f32 (VDUPLN32q (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
+                             SPR:$src, ssub_0), (i32 0)))>;
 
 //   VMOVN    : Vector Narrowing Move
 defm VMOVN    : N2VN_HSD<0b11,0b11,0b10,0b00100,0,0, IIC_VMOVN,
@@ -5576,22 +5742,22 @@ def VCVTxu2fq : N2VCvtQ<1, 1, 0b1110, 0, 1, "vcvt", "f32.u32",
                         v4f32, v4i32, int_arm_neon_vcvtfxu2fp>;
 }
 
-def : NEONInstAlias<"vcvt${p}.s32.f32 $Dd, $Dm, #0", 
+def : NEONInstAlias<"vcvt${p}.s32.f32 $Dd, $Dm, #0",
                     (VCVTf2sd DPR:$Dd, DPR:$Dm, pred:$p)>;
-def : NEONInstAlias<"vcvt${p}.u32.f32 $Dd, $Dm, #0", 
+def : NEONInstAlias<"vcvt${p}.u32.f32 $Dd, $Dm, #0",
                     (VCVTf2ud DPR:$Dd, DPR:$Dm, pred:$p)>;
-def : NEONInstAlias<"vcvt${p}.f32.s32 $Dd, $Dm, #0", 
+def : NEONInstAlias<"vcvt${p}.f32.s32 $Dd, $Dm, #0",
                     (VCVTs2fd DPR:$Dd, DPR:$Dm, pred:$p)>;
-def : NEONInstAlias<"vcvt${p}.f32.u32 $Dd, $Dm, #0", 
+def : NEONInstAlias<"vcvt${p}.f32.u32 $Dd, $Dm, #0",
                     (VCVTu2fd DPR:$Dd, DPR:$Dm, pred:$p)>;
 
-def : NEONInstAlias<"vcvt${p}.s32.f32 $Qd, $Qm, #0", 
+def : NEONInstAlias<"vcvt${p}.s32.f32 $Qd, $Qm, #0",
                     (VCVTf2sq QPR:$Qd, QPR:$Qm, pred:$p)>;
-def : NEONInstAlias<"vcvt${p}.u32.f32 $Qd, $Qm, #0", 
+def : NEONInstAlias<"vcvt${p}.u32.f32 $Qd, $Qm, #0",
                     (VCVTf2uq QPR:$Qd, QPR:$Qm, pred:$p)>;
-def : NEONInstAlias<"vcvt${p}.f32.s32 $Qd, $Qm, #0", 
+def : NEONInstAlias<"vcvt${p}.f32.s32 $Qd, $Qm, #0",
                     (VCVTs2fq QPR:$Qd, QPR:$Qm, pred:$p)>;
-def : NEONInstAlias<"vcvt${p}.f32.u32 $Qd, $Qm, #0", 
+def : NEONInstAlias<"vcvt${p}.f32.u32 $Qd, $Qm, #0",
                     (VCVTu2fq QPR:$Qd, QPR:$Qm, pred:$p)>;
 
 
@@ -5874,7 +6040,7 @@ defm VRINTPN : VRINT_FPI<"p", 0b111, int_arm_neon_vrintp>;
 
 // Cryptography instructions
 let PostEncoderMethod = "NEONThumb2DataIPostEncoder",
-    DecoderNamespace = "v8Crypto" in {
+    DecoderNamespace = "v8Crypto", hasSideEffects = 0 in {
   class AES<string op, bit op7, bit op6, SDPatternOperator Int>
     : N2VQIntXnp<0b00, 0b00, 0b011, op6, op7, NoItinerary,
                  !strconcat("aes", op), "8", v16i8, v16i8, Int>,
@@ -5904,17 +6070,45 @@ def AESE : AES2Op<"e", 0, 0, int_arm_neon_aese>;
 def AESIMC : AES<"imc", 1, 1, int_arm_neon_aesimc>;
 def AESMC : AES<"mc", 1, 0, int_arm_neon_aesmc>;
 
-def SHA1H : N2SHA<"1h", 0b01, 0b010, 1, 1, int_arm_neon_sha1h>;
+def SHA1H : N2SHA<"1h", 0b01, 0b010, 1, 1, null_frag>;
 def SHA1SU1 : N2SHA2Op<"1su1", 0b10, 0b011, 1, 0, int_arm_neon_sha1su1>;
 def SHA256SU0 : N2SHA2Op<"256su0", 0b10, 0b011, 1, 1, int_arm_neon_sha256su0>;
-def SHA1C : N3SHA3Op<"1c", 0b00100, 0b00, int_arm_neon_sha1c>;
-def SHA1M : N3SHA3Op<"1m", 0b00100, 0b10, int_arm_neon_sha1m>;
-def SHA1P : N3SHA3Op<"1p", 0b00100, 0b01, int_arm_neon_sha1p>;
+def SHA1C : N3SHA3Op<"1c", 0b00100, 0b00, null_frag>;
+def SHA1M : N3SHA3Op<"1m", 0b00100, 0b10, null_frag>;
+def SHA1P : N3SHA3Op<"1p", 0b00100, 0b01, null_frag>;
 def SHA1SU0 : N3SHA3Op<"1su0", 0b00100, 0b11, int_arm_neon_sha1su0>;
 def SHA256H : N3SHA3Op<"256h", 0b00110, 0b00, int_arm_neon_sha256h>;
 def SHA256H2 : N3SHA3Op<"256h2", 0b00110, 0b01, int_arm_neon_sha256h2>;
 def SHA256SU1 : N3SHA3Op<"256su1", 0b00110, 0b10, int_arm_neon_sha256su1>;
 
+def : Pat<(i32 (int_arm_neon_sha1h i32:$Rn)),
+          (COPY_TO_REGCLASS (f32 (EXTRACT_SUBREG
+              (SHA1H (SUBREG_TO_REG (i64 0),
+                                    (f32 (COPY_TO_REGCLASS i32:$Rn, SPR)),
+                                    ssub_0)),
+              ssub_0)), GPR)>;
+
+def : Pat<(v4i32 (int_arm_neon_sha1c v4i32:$hash_abcd, i32:$hash_e, v4i32:$wk)),
+          (SHA1C v4i32:$hash_abcd,
+                 (SUBREG_TO_REG (i64 0),
+                                (f32 (COPY_TO_REGCLASS i32:$hash_e, SPR)),
+                                ssub_0),
+                 v4i32:$wk)>;
+
+def : Pat<(v4i32 (int_arm_neon_sha1m v4i32:$hash_abcd, i32:$hash_e, v4i32:$wk)),
+          (SHA1M v4i32:$hash_abcd,
+                 (SUBREG_TO_REG (i64 0),
+                                (f32 (COPY_TO_REGCLASS i32:$hash_e, SPR)),
+                                ssub_0),
+                 v4i32:$wk)>;
+
+def : Pat<(v4i32 (int_arm_neon_sha1p v4i32:$hash_abcd, i32:$hash_e, v4i32:$wk)),
+          (SHA1P v4i32:$hash_abcd,
+                 (SUBREG_TO_REG (i64 0),
+                                (f32 (COPY_TO_REGCLASS i32:$hash_e, SPR)),
+                                ssub_0),
+                 v4i32:$wk)>;
+
 //===----------------------------------------------------------------------===//
 // NEON instructions for single-precision FP math
 //===----------------------------------------------------------------------===//
@@ -5982,67 +6176,145 @@ def : Pat<(f32 (bitconvert GPR:$a)),
 //===----------------------------------------------------------------------===//
 
 // bit_convert
-def : Pat<(v1i64 (bitconvert (v2i32 DPR:$src))), (v1i64 DPR:$src)>;
-def : Pat<(v1i64 (bitconvert (v4i16 DPR:$src))), (v1i64 DPR:$src)>;
-def : Pat<(v1i64 (bitconvert (v8i8  DPR:$src))), (v1i64 DPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v1i64 (bitconvert (v2i32 DPR:$src))), (v1i64 DPR:$src)>;
+  def : Pat<(v1i64 (bitconvert (v4i16 DPR:$src))), (v1i64 DPR:$src)>;
+  def : Pat<(v1i64 (bitconvert (v8i8  DPR:$src))), (v1i64 DPR:$src)>;
+}
 def : Pat<(v1i64 (bitconvert (f64   DPR:$src))), (v1i64 DPR:$src)>;
-def : Pat<(v1i64 (bitconvert (v2f32 DPR:$src))), (v1i64 DPR:$src)>;
-def : Pat<(v2i32 (bitconvert (v1i64 DPR:$src))), (v2i32 DPR:$src)>;
-def : Pat<(v2i32 (bitconvert (v4i16 DPR:$src))), (v2i32 DPR:$src)>;
-def : Pat<(v2i32 (bitconvert (v8i8  DPR:$src))), (v2i32 DPR:$src)>;
-def : Pat<(v2i32 (bitconvert (f64   DPR:$src))), (v2i32 DPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v1i64 (bitconvert (v2f32 DPR:$src))), (v1i64 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (v1i64 DPR:$src))), (v2i32 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (v4i16 DPR:$src))), (v2i32 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (v8i8  DPR:$src))), (v2i32 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (f64   DPR:$src))), (v2i32 DPR:$src)>;
+}
 def : Pat<(v2i32 (bitconvert (v2f32 DPR:$src))), (v2i32 DPR:$src)>;
-def : Pat<(v4i16 (bitconvert (v1i64 DPR:$src))), (v4i16 DPR:$src)>;
-def : Pat<(v4i16 (bitconvert (v2i32 DPR:$src))), (v4i16 DPR:$src)>;
-def : Pat<(v4i16 (bitconvert (v8i8  DPR:$src))), (v4i16 DPR:$src)>;
-def : Pat<(v4i16 (bitconvert (f64   DPR:$src))), (v4i16 DPR:$src)>;
-def : Pat<(v4i16 (bitconvert (v2f32 DPR:$src))), (v4i16 DPR:$src)>;
-def : Pat<(v8i8  (bitconvert (v1i64 DPR:$src))), (v8i8  DPR:$src)>;
-def : Pat<(v8i8  (bitconvert (v2i32 DPR:$src))), (v8i8  DPR:$src)>;
-def : Pat<(v8i8  (bitconvert (v4i16 DPR:$src))), (v8i8  DPR:$src)>;
-def : Pat<(v8i8  (bitconvert (f64   DPR:$src))), (v8i8  DPR:$src)>;
-def : Pat<(v8i8  (bitconvert (v2f32 DPR:$src))), (v8i8  DPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v4i16 (bitconvert (v1i64 DPR:$src))), (v4i16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v2i32 DPR:$src))), (v4i16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v8i8  DPR:$src))), (v4i16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (f64   DPR:$src))), (v4i16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v2f32 DPR:$src))), (v4i16 DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v1i64 DPR:$src))), (v8i8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v2i32 DPR:$src))), (v8i8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v4i16 DPR:$src))), (v8i8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (f64   DPR:$src))), (v8i8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v2f32 DPR:$src))), (v8i8  DPR:$src)>;
+}
 def : Pat<(f64   (bitconvert (v1i64 DPR:$src))), (f64   DPR:$src)>;
-def : Pat<(f64   (bitconvert (v2i32 DPR:$src))), (f64   DPR:$src)>;
-def : Pat<(f64   (bitconvert (v4i16 DPR:$src))), (f64   DPR:$src)>;
-def : Pat<(f64   (bitconvert (v8i8  DPR:$src))), (f64   DPR:$src)>;
-def : Pat<(f64   (bitconvert (v2f32 DPR:$src))), (f64   DPR:$src)>;
-def : Pat<(v2f32 (bitconvert (f64   DPR:$src))), (v2f32 DPR:$src)>;
-def : Pat<(v2f32 (bitconvert (v1i64 DPR:$src))), (v2f32 DPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(f64   (bitconvert (v2i32 DPR:$src))), (f64   DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v4i16 DPR:$src))), (f64   DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v8i8  DPR:$src))), (f64   DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v2f32 DPR:$src))), (f64   DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (f64   DPR:$src))), (v2f32 DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (v1i64 DPR:$src))), (v2f32 DPR:$src)>;
+}
 def : Pat<(v2f32 (bitconvert (v2i32 DPR:$src))), (v2f32 DPR:$src)>;
-def : Pat<(v2f32 (bitconvert (v4i16 DPR:$src))), (v2f32 DPR:$src)>;
-def : Pat<(v2f32 (bitconvert (v8i8  DPR:$src))), (v2f32 DPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v2f32 (bitconvert (v4i16 DPR:$src))), (v2f32 DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (v8i8  DPR:$src))), (v2f32 DPR:$src)>;
+}
 
-def : Pat<(v2i64 (bitconvert (v4i32 QPR:$src))), (v2i64 QPR:$src)>;
-def : Pat<(v2i64 (bitconvert (v8i16 QPR:$src))), (v2i64 QPR:$src)>;
-def : Pat<(v2i64 (bitconvert (v16i8 QPR:$src))), (v2i64 QPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v2i64 (bitconvert (v4i32 QPR:$src))), (v2i64 QPR:$src)>;
+  def : Pat<(v2i64 (bitconvert (v8i16 QPR:$src))), (v2i64 QPR:$src)>;
+  def : Pat<(v2i64 (bitconvert (v16i8 QPR:$src))), (v2i64 QPR:$src)>;
+}
 def : Pat<(v2i64 (bitconvert (v2f64 QPR:$src))), (v2i64 QPR:$src)>;
-def : Pat<(v2i64 (bitconvert (v4f32 QPR:$src))), (v2i64 QPR:$src)>;
-def : Pat<(v4i32 (bitconvert (v2i64 QPR:$src))), (v4i32 QPR:$src)>;
-def : Pat<(v4i32 (bitconvert (v8i16 QPR:$src))), (v4i32 QPR:$src)>;
-def : Pat<(v4i32 (bitconvert (v16i8 QPR:$src))), (v4i32 QPR:$src)>;
-def : Pat<(v4i32 (bitconvert (v2f64 QPR:$src))), (v4i32 QPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v2i64 (bitconvert (v4f32 QPR:$src))), (v2i64 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v2i64 QPR:$src))), (v4i32 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v8i16 QPR:$src))), (v4i32 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v16i8 QPR:$src))), (v4i32 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v2f64 QPR:$src))), (v4i32 QPR:$src)>;
+}
 def : Pat<(v4i32 (bitconvert (v4f32 QPR:$src))), (v4i32 QPR:$src)>;
-def : Pat<(v8i16 (bitconvert (v2i64 QPR:$src))), (v8i16 QPR:$src)>;
-def : Pat<(v8i16 (bitconvert (v4i32 QPR:$src))), (v8i16 QPR:$src)>;
-def : Pat<(v8i16 (bitconvert (v16i8 QPR:$src))), (v8i16 QPR:$src)>;
-def : Pat<(v8i16 (bitconvert (v2f64 QPR:$src))), (v8i16 QPR:$src)>;
-def : Pat<(v8i16 (bitconvert (v4f32 QPR:$src))), (v8i16 QPR:$src)>;
-def : Pat<(v16i8 (bitconvert (v2i64 QPR:$src))), (v16i8 QPR:$src)>;
-def : Pat<(v16i8 (bitconvert (v4i32 QPR:$src))), (v16i8 QPR:$src)>;
-def : Pat<(v16i8 (bitconvert (v8i16 QPR:$src))), (v16i8 QPR:$src)>;
-def : Pat<(v16i8 (bitconvert (v2f64 QPR:$src))), (v16i8 QPR:$src)>;
-def : Pat<(v16i8 (bitconvert (v4f32 QPR:$src))), (v16i8 QPR:$src)>;
-def : Pat<(v4f32 (bitconvert (v2i64 QPR:$src))), (v4f32 QPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v8i16 (bitconvert (v2i64 QPR:$src))), (v8i16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v4i32 QPR:$src))), (v8i16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v16i8 QPR:$src))), (v8i16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v2f64 QPR:$src))), (v8i16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v4f32 QPR:$src))), (v8i16 QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v2i64 QPR:$src))), (v16i8 QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v4i32 QPR:$src))), (v16i8 QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v8i16 QPR:$src))), (v16i8 QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v2f64 QPR:$src))), (v16i8 QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v4f32 QPR:$src))), (v16i8 QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v2i64 QPR:$src))), (v4f32 QPR:$src)>;
+}
 def : Pat<(v4f32 (bitconvert (v4i32 QPR:$src))), (v4f32 QPR:$src)>;
-def : Pat<(v4f32 (bitconvert (v8i16 QPR:$src))), (v4f32 QPR:$src)>;
-def : Pat<(v4f32 (bitconvert (v16i8 QPR:$src))), (v4f32 QPR:$src)>;
-def : Pat<(v4f32 (bitconvert (v2f64 QPR:$src))), (v4f32 QPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v4f32 (bitconvert (v8i16 QPR:$src))), (v4f32 QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v16i8 QPR:$src))), (v4f32 QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v2f64 QPR:$src))), (v4f32 QPR:$src)>;
+}
 def : Pat<(v2f64 (bitconvert (v2i64 QPR:$src))), (v2f64 QPR:$src)>;
-def : Pat<(v2f64 (bitconvert (v4i32 QPR:$src))), (v2f64 QPR:$src)>;
-def : Pat<(v2f64 (bitconvert (v8i16 QPR:$src))), (v2f64 QPR:$src)>;
-def : Pat<(v2f64 (bitconvert (v16i8 QPR:$src))), (v2f64 QPR:$src)>;
-def : Pat<(v2f64 (bitconvert (v4f32 QPR:$src))), (v2f64 QPR:$src)>;
+let Predicates = [IsLE] in {
+  def : Pat<(v2f64 (bitconvert (v4i32 QPR:$src))), (v2f64 QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v8i16 QPR:$src))), (v2f64 QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v16i8 QPR:$src))), (v2f64 QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v4f32 QPR:$src))), (v2f64 QPR:$src)>;
+}
+
+let Predicates = [IsBE] in {
+  // 64 bit conversions
+  def : Pat<(v1i64 (bitconvert (v2i32 DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v1i64 (bitconvert (v4i16 DPR:$src))), (VREV64d16 DPR:$src)>;
+  def : Pat<(v1i64 (bitconvert (v8i8  DPR:$src))), (VREV64d8  DPR:$src)>;
+  def : Pat<(v1i64 (bitconvert (v2f32 DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (v1i64 DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (v4i16 DPR:$src))), (VREV32d16 DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (v8i8  DPR:$src))), (VREV32d8  DPR:$src)>;
+  def : Pat<(v2i32 (bitconvert (f64   DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v1i64 DPR:$src))), (VREV64d16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v2i32 DPR:$src))), (VREV32d16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v8i8  DPR:$src))), (VREV16d8  DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (f64   DPR:$src))), (VREV64d16 DPR:$src)>;
+  def : Pat<(v4i16 (bitconvert (v2f32 DPR:$src))), (VREV32d16 DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v1i64 DPR:$src))), (VREV64d8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v2i32 DPR:$src))), (VREV32d8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v4i16 DPR:$src))), (VREV16d8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (f64   DPR:$src))), (VREV64d8  DPR:$src)>;
+  def : Pat<(v8i8  (bitconvert (v2f32 DPR:$src))), (VREV32d8  DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v2i32 DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v4i16 DPR:$src))), (VREV64d16 DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v8i8  DPR:$src))), (VREV64d8  DPR:$src)>;
+  def : Pat<(f64   (bitconvert (v2f32 DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (f64   DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (v1i64 DPR:$src))), (VREV64d32 DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (v4i16 DPR:$src))), (VREV32d16 DPR:$src)>;
+  def : Pat<(v2f32 (bitconvert (v8i8  DPR:$src))), (VREV32d8  DPR:$src)>;
+
+  // 128 bit conversions
+  def : Pat<(v2i64 (bitconvert (v4i32 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v2i64 (bitconvert (v8i16 QPR:$src))), (VREV64q16 QPR:$src)>;
+  def : Pat<(v2i64 (bitconvert (v16i8 QPR:$src))), (VREV64q8  QPR:$src)>;
+  def : Pat<(v2i64 (bitconvert (v4f32 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v2i64 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v8i16 QPR:$src))), (VREV32q16 QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v16i8 QPR:$src))), (VREV32q8  QPR:$src)>;
+  def : Pat<(v4i32 (bitconvert (v2f64 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v2i64 QPR:$src))), (VREV64q16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v4i32 QPR:$src))), (VREV32q16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v16i8 QPR:$src))), (VREV16q8  QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v2f64 QPR:$src))), (VREV64q16 QPR:$src)>;
+  def : Pat<(v8i16 (bitconvert (v4f32 QPR:$src))), (VREV32q16 QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v2i64 QPR:$src))), (VREV64q8  QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v4i32 QPR:$src))), (VREV32q8  QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v8i16 QPR:$src))), (VREV16q8  QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v2f64 QPR:$src))), (VREV64q8  QPR:$src)>;
+  def : Pat<(v16i8 (bitconvert (v4f32 QPR:$src))), (VREV32q8  QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v2i64 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v8i16 QPR:$src))), (VREV32q16 QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v16i8 QPR:$src))), (VREV32q8  QPR:$src)>;
+  def : Pat<(v4f32 (bitconvert (v2f64 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v4i32 QPR:$src))), (VREV64q32 QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v8i16 QPR:$src))), (VREV64q16 QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v16i8 QPR:$src))), (VREV64q8  QPR:$src)>;
+  def : Pat<(v2f64 (bitconvert (v4f32 QPR:$src))), (VREV64q32 QPR:$src)>;
+}
 
 // Fold extracting an element out of a v2i32 into a vfp register.
 def : Pat<(f32 (bitconvert (i32 (extractelt (v2i32 DPR:$src), imm:$lane)))),
@@ -6051,7 +6323,7 @@ def : Pat<(f32 (bitconvert (i32 (extractelt (v2i32 DPR:$src), imm:$lane)))),
 // Vector lengthening move with load, matching extending loads.
 
 // extload, zextload and sextload for a standard lengthening load. Example:
-// Lengthen_Single<"8", "i16", "8"> = 
+// Lengthen_Single<"8", "i16", "8"> =
 //     Pat<(v8i16 (extloadvi8 addrmode6:$addr))
 //         (VMOVLuv8i16 (VLD1d8 addrmode6:$addr,
 //                              (f64 (IMPLICIT_DEF)), (i32 0)))>;
@@ -6078,7 +6350,7 @@ multiclass Lengthen_Single<string DestLanes, string DestTy, string SrcTy> {
 // half the lanes available. Example:
 // Lengthen_HalfSingle<"4", "i16", "8", "i16", "i8"> =
 //     Pat<(v4i16 (extloadvi8 addrmode6oneL32:$addr)),
-//         (EXTRACT_SUBREG (VMOVLuv8i16 (VLD1LNd32 addrmode6oneL32:$addr, 
+//         (EXTRACT_SUBREG (VMOVLuv8i16 (VLD1LNd32 addrmode6oneL32:$addr,
 //                                      (f64 (IMPLICIT_DEF)), (i32 0))),
 //                         dsub_0)>;
 multiclass Lengthen_HalfSingle<string DestLanes, string DestTy, string SrcTy,
@@ -6100,6 +6372,32 @@ multiclass Lengthen_HalfSingle<string DestLanes, string DestTy, string SrcTy,
          dsub_0)>;
 }
 
+// The following class definition is basically a copy of the
+// Lengthen_HalfSingle definition above, however with an additional parameter
+// "RevLanes" to select the correct VREV32dXX instruction. This is to convert
+// data loaded by VLD1LN into proper vector format in big endian mode.
+multiclass Lengthen_HalfSingle_Big_Endian<string DestLanes, string DestTy, string SrcTy,
+                               string InsnLanes, string InsnTy, string RevLanes> {
+  def _Any : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("extloadv" # SrcTy) addrmode6oneL32:$addr)),
+       (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # InsnLanes # InsnTy)
+         (!cast<Instruction>("VREV32d" # RevLanes)
+           (VLD1LNd32 addrmode6oneL32:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+         dsub_0)>;
+  def _Z   : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("zextloadv" # SrcTy) addrmode6oneL32:$addr)),
+       (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # InsnLanes # InsnTy)
+         (!cast<Instruction>("VREV32d" # RevLanes)
+           (VLD1LNd32 addrmode6oneL32:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+         dsub_0)>;
+  def _S   : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("sextloadv" # SrcTy) addrmode6oneL32:$addr)),
+       (EXTRACT_SUBREG (!cast<Instruction>("VMOVLsv" # InsnLanes # InsnTy)
+         (!cast<Instruction>("VREV32d" # RevLanes)
+           (VLD1LNd32 addrmode6oneL32:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+         dsub_0)>;
+}
+
 // extload, zextload and sextload for a lengthening load followed by another
 // lengthening load, to quadruple the initial length.
 //
@@ -6134,6 +6432,36 @@ multiclass Lengthen_Double<string DestLanes, string DestTy, string SrcTy,
              dsub_0))>;
 }
 
+// The following class definition is basically a copy of the
+// Lengthen_Double definition above, however with an additional parameter
+// "RevLanes" to select the correct VREV32dXX instruction. This is to convert
+// data loaded by VLD1LN into proper vector format in big endian mode.
+multiclass Lengthen_Double_Big_Endian<string DestLanes, string DestTy, string SrcTy,
+                           string Insn1Lanes, string Insn1Ty, string Insn2Lanes,
+                           string Insn2Ty, string RevLanes> {
+  def _Any : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("extloadv" # SrcTy) addrmode6oneL32:$addr)),
+         (!cast<Instruction>("VMOVLuv" # Insn2Lanes # Insn2Ty)
+           (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # Insn1Lanes # Insn1Ty)
+            (!cast<Instruction>("VREV32d" # RevLanes)
+             (VLD1LNd32 addrmode6oneL32:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+             dsub_0))>;
+  def _Z   : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("zextloadv" # SrcTy) addrmode6oneL32:$addr)),
+         (!cast<Instruction>("VMOVLuv" # Insn2Lanes # Insn2Ty)
+           (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # Insn1Lanes # Insn1Ty)
+            (!cast<Instruction>("VREV32d" # RevLanes)
+             (VLD1LNd32 addrmode6oneL32:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+             dsub_0))>;
+  def _S   : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("sextloadv" # SrcTy) addrmode6oneL32:$addr)),
+         (!cast<Instruction>("VMOVLsv" # Insn2Lanes # Insn2Ty)
+           (EXTRACT_SUBREG (!cast<Instruction>("VMOVLsv" # Insn1Lanes # Insn1Ty)
+            (!cast<Instruction>("VREV32d" # RevLanes)
+             (VLD1LNd32 addrmode6oneL32:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+             dsub_0))>;
+}
+
 // extload, zextload and sextload for a lengthening load followed by another
 // lengthening load, to quadruple the initial length, but which ends up only
 // requiring half the available lanes (a 64-bit outcome instead of a 128-bit).
@@ -6171,33 +6499,102 @@ multiclass Lengthen_HalfDouble<string DestLanes, string DestTy, string SrcTy,
           dsub_0)>;
 }
 
+// The following class definition is basically a copy of the
+// Lengthen_HalfDouble definition above, however with an additional VREV16d8
+// instruction to convert data loaded by VLD1LN into proper vector format
+// in big endian mode.
+multiclass Lengthen_HalfDouble_Big_Endian<string DestLanes, string DestTy, string SrcTy,
+                           string Insn1Lanes, string Insn1Ty, string Insn2Lanes,
+                           string Insn2Ty> {
+  def _Any : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("extloadv" # SrcTy) addrmode6:$addr)),
+         (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # Insn2Lanes # Insn2Ty)
+           (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # Insn1Lanes # Insn1Ty)
+            (!cast<Instruction>("VREV16d8")
+             (VLD1LNd16 addrmode6:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+             dsub_0)),
+          dsub_0)>;
+  def _Z   : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("zextloadv" # SrcTy) addrmode6:$addr)),
+         (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # Insn2Lanes # Insn2Ty)
+           (EXTRACT_SUBREG (!cast<Instruction>("VMOVLuv" # Insn1Lanes # Insn1Ty)
+            (!cast<Instruction>("VREV16d8")
+             (VLD1LNd16 addrmode6:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+             dsub_0)),
+          dsub_0)>;
+  def _S   : Pat<(!cast<ValueType>("v" # DestLanes # DestTy)
+                   (!cast<PatFrag>("sextloadv" # SrcTy) addrmode6:$addr)),
+         (EXTRACT_SUBREG (!cast<Instruction>("VMOVLsv" # Insn2Lanes # Insn2Ty)
+           (EXTRACT_SUBREG (!cast<Instruction>("VMOVLsv" # Insn1Lanes # Insn1Ty)
+            (!cast<Instruction>("VREV16d8")
+             (VLD1LNd16 addrmode6:$addr, (f64 (IMPLICIT_DEF)), (i32 0)))),
+             dsub_0)),
+          dsub_0)>;
+}
+
 defm : Lengthen_Single<"8", "i16", "8">; // v8i8 -> v8i16
 defm : Lengthen_Single<"4", "i32", "16">; // v4i16 -> v4i32
 defm : Lengthen_Single<"2", "i64", "32">; // v2i32 -> v2i64
 
-defm : Lengthen_HalfSingle<"4", "i16", "i8", "8", "i16">; // v4i8 -> v4i16
-defm : Lengthen_HalfSingle<"2", "i32", "i16", "4", "i32">; // v2i16 -> v2i32
+let Predicates = [IsLE] in {
+  defm : Lengthen_HalfSingle<"4", "i16", "i8", "8", "i16">; // v4i8 -> v4i16
+  defm : Lengthen_HalfSingle<"2", "i32", "i16", "4", "i32">; // v2i16 -> v2i32
+
+  // Double lengthening - v4i8 -> v4i16 -> v4i32
+  defm : Lengthen_Double<"4", "i32", "i8", "8", "i16", "4", "i32">;
+  // v2i8 -> v2i16 -> v2i32
+  defm : Lengthen_HalfDouble<"2", "i32", "i8", "8", "i16", "4", "i32">;
+  // v2i16 -> v2i32 -> v2i64
+  defm : Lengthen_Double<"2", "i64", "i16", "4", "i32", "2", "i64">;
+}
+
+let Predicates = [IsBE] in {
+  defm : Lengthen_HalfSingle_Big_Endian<"4", "i16", "i8", "8", "i16", "8">; // v4i8 -> v4i16
+  defm : Lengthen_HalfSingle_Big_Endian<"2", "i32", "i16", "4", "i32", "16">; // v2i16 -> v2i32
 
-// Double lengthening - v4i8 -> v4i16 -> v4i32
-defm : Lengthen_Double<"4", "i32", "i8", "8", "i16", "4", "i32">;
-// v2i8 -> v2i16 -> v2i32
-defm : Lengthen_HalfDouble<"2", "i32", "i8", "8", "i16", "4", "i32">;
-// v2i16 -> v2i32 -> v2i64
-defm : Lengthen_Double<"2", "i64", "i16", "4", "i32", "2", "i64">;
+  // Double lengthening - v4i8 -> v4i16 -> v4i32
+  defm : Lengthen_Double_Big_Endian<"4", "i32", "i8", "8", "i16", "4", "i32", "8">;
+  // v2i8 -> v2i16 -> v2i32
+  defm : Lengthen_HalfDouble_Big_Endian<"2", "i32", "i8", "8", "i16", "4", "i32">;
+  // v2i16 -> v2i32 -> v2i64
+  defm : Lengthen_Double_Big_Endian<"2", "i64", "i16", "4", "i32", "2", "i64", "16">;
+}
 
 // Triple lengthening - v2i8 -> v2i16 -> v2i32 -> v2i64
-def : Pat<(v2i64 (extloadvi8 addrmode6:$addr)),
-      (VMOVLuv2i64 (EXTRACT_SUBREG (VMOVLuv4i32 (EXTRACT_SUBREG (VMOVLuv8i16
-         (VLD1LNd16 addrmode6:$addr, 
-                    (f64 (IMPLICIT_DEF)), (i32 0))), dsub_0)), dsub_0))>;
-def : Pat<(v2i64 (zextloadvi8 addrmode6:$addr)),
-      (VMOVLuv2i64 (EXTRACT_SUBREG (VMOVLuv4i32 (EXTRACT_SUBREG (VMOVLuv8i16
-         (VLD1LNd16 addrmode6:$addr,
-                    (f64 (IMPLICIT_DEF)), (i32 0))), dsub_0)), dsub_0))>;
-def : Pat<(v2i64 (sextloadvi8 addrmode6:$addr)),
-      (VMOVLsv2i64 (EXTRACT_SUBREG (VMOVLsv4i32 (EXTRACT_SUBREG (VMOVLsv8i16
-         (VLD1LNd16 addrmode6:$addr,
-                    (f64 (IMPLICIT_DEF)), (i32 0))), dsub_0)), dsub_0))>;
+let Predicates = [IsLE] in {
+  def : Pat<(v2i64 (extloadvi8 addrmode6:$addr)),
+        (VMOVLuv2i64 (EXTRACT_SUBREG (VMOVLuv4i32 (EXTRACT_SUBREG (VMOVLuv8i16
+           (VLD1LNd16 addrmode6:$addr,
+                      (f64 (IMPLICIT_DEF)), (i32 0))), dsub_0)), dsub_0))>;
+  def : Pat<(v2i64 (zextloadvi8 addrmode6:$addr)),
+        (VMOVLuv2i64 (EXTRACT_SUBREG (VMOVLuv4i32 (EXTRACT_SUBREG (VMOVLuv8i16
+           (VLD1LNd16 addrmode6:$addr,
+                      (f64 (IMPLICIT_DEF)), (i32 0))), dsub_0)), dsub_0))>;
+  def : Pat<(v2i64 (sextloadvi8 addrmode6:$addr)),
+        (VMOVLsv2i64 (EXTRACT_SUBREG (VMOVLsv4i32 (EXTRACT_SUBREG (VMOVLsv8i16
+           (VLD1LNd16 addrmode6:$addr,
+                      (f64 (IMPLICIT_DEF)), (i32 0))), dsub_0)), dsub_0))>;
+}
+// The following patterns are basically a copy of the patterns above, 
+// however with an additional VREV16d instruction to convert data
+// loaded by VLD1LN into proper vector format in big endian mode.
+let Predicates = [IsBE] in {
+  def : Pat<(v2i64 (extloadvi8 addrmode6:$addr)),
+        (VMOVLuv2i64 (EXTRACT_SUBREG (VMOVLuv4i32 (EXTRACT_SUBREG (VMOVLuv8i16
+           (!cast<Instruction>("VREV16d8")
+             (VLD1LNd16 addrmode6:$addr,
+                        (f64 (IMPLICIT_DEF)), (i32 0)))), dsub_0)), dsub_0))>;
+  def : Pat<(v2i64 (zextloadvi8 addrmode6:$addr)),
+        (VMOVLuv2i64 (EXTRACT_SUBREG (VMOVLuv4i32 (EXTRACT_SUBREG (VMOVLuv8i16
+           (!cast<Instruction>("VREV16d8")
+             (VLD1LNd16 addrmode6:$addr,
+                        (f64 (IMPLICIT_DEF)), (i32 0)))), dsub_0)), dsub_0))>;
+  def : Pat<(v2i64 (sextloadvi8 addrmode6:$addr)),
+        (VMOVLsv2i64 (EXTRACT_SUBREG (VMOVLsv4i32 (EXTRACT_SUBREG (VMOVLsv8i16
+           (!cast<Instruction>("VREV16d8")
+             (VLD1LNd16 addrmode6:$addr,
+                        (f64 (IMPLICIT_DEF)), (i32 0)))), dsub_0)), dsub_0))>;
+}
 
 //===----------------------------------------------------------------------===//
 // Assembler aliases
@@ -6242,379 +6639,442 @@ defm : NEONDTAnyInstAlias<"vorr${p}", "$Vdn, $Vm",
 // VLD1 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VLD1LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vld1${p}", ".8", "$list, $addr",
-                 (ins VecListOneDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDByteIndexed:$list, addrmode6alignNone:$addr,
+                      pred:$p)>;
 def VLD1LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vld1${p}", ".16", "$list, $addr",
-                 (ins VecListOneDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDHWordIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VLD1LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vld1${p}", ".32", "$list, $addr",
-                 (ins VecListOneDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 
 def VLD1LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld1${p}", ".8", "$list, $addr!",
-                 (ins VecListOneDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDByteIndexed:$list, addrmode6alignNone:$addr,
+                      pred:$p)>;
 def VLD1LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld1${p}", ".16", "$list, $addr!",
-                 (ins VecListOneDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDHWordIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VLD1LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld1${p}", ".32", "$list, $addr!",
-                 (ins VecListOneDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VLD1LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld1${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListOneDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListOneDByteIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD1LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld1${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListOneDHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListOneDHWordIndexed:$list, addrmode6align16:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD1LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld1${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListOneDWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListOneDWordIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 
 // VST1 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VST1LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vst1${p}", ".8", "$list, $addr",
-                 (ins VecListOneDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDByteIndexed:$list, addrmode6alignNone:$addr,
+                      pred:$p)>;
 def VST1LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vst1${p}", ".16", "$list, $addr",
-                 (ins VecListOneDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDHWordIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VST1LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vst1${p}", ".32", "$list, $addr",
-                 (ins VecListOneDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 
 def VST1LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst1${p}", ".8", "$list, $addr!",
-                 (ins VecListOneDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDByteIndexed:$list, addrmode6alignNone:$addr,
+                      pred:$p)>;
 def VST1LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst1${p}", ".16", "$list, $addr!",
-                 (ins VecListOneDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDHWordIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VST1LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst1${p}", ".32", "$list, $addr!",
-                 (ins VecListOneDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListOneDWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VST1LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst1${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListOneDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListOneDByteIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST1LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst1${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListOneDHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListOneDHWordIndexed:$list, addrmode6align16:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST1LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst1${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListOneDWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListOneDWordIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VLD2 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VLD2LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vld2${p}", ".8", "$list, $addr",
-                 (ins VecListTwoDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDByteIndexed:$list, addrmode6align16:$addr,
+                  pred:$p)>;
 def VLD2LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vld2${p}", ".16", "$list, $addr",
-                 (ins VecListTwoDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VLD2LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vld2${p}", ".32", "$list, $addr",
-                 (ins VecListTwoDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDWordIndexed:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD2LNqAsm_16 : NEONDataTypeAsmPseudoInst<"vld2${p}", ".16", "$list, $addr",
-                 (ins VecListTwoQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VLD2LNqAsm_32 : NEONDataTypeAsmPseudoInst<"vld2${p}", ".32", "$list, $addr",
-                 (ins VecListTwoQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 
 def VLD2LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".8", "$list, $addr!",
-                 (ins VecListTwoDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDByteIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VLD2LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".16", "$list, $addr!",
-                 (ins VecListTwoDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VLD2LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".32", "$list, $addr!",
-                 (ins VecListTwoDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 def VLD2LNqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".16", "$list, $addr!",
-                 (ins VecListTwoQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VLD2LNqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".32", "$list, $addr!",
-                 (ins VecListTwoQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 def VLD2LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListTwoDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoDByteIndexed:$list, addrmode6align16:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD2LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListTwoDHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoDHWordIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD2LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListTwoDWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoDWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD2LNqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListTwoQHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoQHWordIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD2LNqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld2${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListTwoQWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoQWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 
 // VST2 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VST2LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vst2${p}", ".8", "$list, $addr",
-                 (ins VecListTwoDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDByteIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VST2LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vst2${p}", ".16", "$list, $addr",
-                 (ins VecListTwoDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VST2LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vst2${p}", ".32", "$list, $addr",
-                 (ins VecListTwoDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 def VST2LNqAsm_16 : NEONDataTypeAsmPseudoInst<"vst2${p}", ".16", "$list, $addr",
-                 (ins VecListTwoQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VST2LNqAsm_32 : NEONDataTypeAsmPseudoInst<"vst2${p}", ".32", "$list, $addr",
-                 (ins VecListTwoQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 
 def VST2LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".8", "$list, $addr!",
-                 (ins VecListTwoDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDByteIndexed:$list, addrmode6align16:$addr,
+                      pred:$p)>;
 def VST2LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".16", "$list, $addr!",
-                 (ins VecListTwoDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VST2LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".32", "$list, $addr!",
-                 (ins VecListTwoDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoDWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 def VST2LNqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".16", "$list, $addr!",
-                 (ins VecListTwoQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQHWordIndexed:$list, addrmode6align32:$addr,
+                      pred:$p)>;
 def VST2LNqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".32", "$list, $addr!",
-                 (ins VecListTwoQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+                 (ins VecListTwoQWordIndexed:$list, addrmode6align64:$addr,
+                      pred:$p)>;
 def VST2LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListTwoDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoDByteIndexed:$list, addrmode6align16:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST2LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".16","$list, $addr, $Rm",
-                  (ins VecListTwoDHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoDHWordIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST2LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListTwoDWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoDWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST2LNqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".16","$list, $addr, $Rm",
-                  (ins VecListTwoQHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoQHWordIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST2LNqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst2${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListTwoQWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListTwoQWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VLD3 all-lanes pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VLD3DUPdAsm_8 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr",
-               (ins VecListThreeDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPdAsm_16: NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr",
-               (ins VecListThreeDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPdAsm_32: NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr",
-               (ins VecListThreeDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPqAsm_8 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr",
-               (ins VecListThreeQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPqAsm_16: NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr",
-               (ins VecListThreeQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPqAsm_32: NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr",
-               (ins VecListThreeQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 
 def VLD3DUPdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPqWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
+                    pred:$p)>;
 def VLD3DUPdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeDAllLanes:$list, addrmode6:$addr,
+                  (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3DUPdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeDAllLanes:$list, addrmode6:$addr,
+                  (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3DUPdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeDAllLanes:$list, addrmode6:$addr,
+                  (ins VecListThreeDAllLanes:$list, addrmode6dupalignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3DUPqWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeQAllLanes:$list, addrmode6:$addr,
+                  (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3DUPqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeQAllLanes:$list, addrmode6:$addr,
+                  (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3DUPqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeQAllLanes:$list, addrmode6:$addr,
+                  (ins VecListThreeQAllLanes:$list, addrmode6dupalignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 
 // VLD3 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VLD3LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr",
-               (ins VecListThreeDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDByteIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr",
-               (ins VecListThreeDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr",
-               (ins VecListThreeDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNqAsm_16 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr",
-               (ins VecListThreeQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNqAsm_32 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr",
-               (ins VecListThreeQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 
 def VLD3LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDByteIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VLD3LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListThreeDByteIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeDHWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListThreeDHWordIndexed:$list,
+                       addrmode6alignNone:$addr, rGPR:$Rm, pred:$p)>;
 def VLD3LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeDWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListThreeDWordIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3LNqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeQHWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListThreeQHWordIndexed:$list,
+                       addrmode6alignNone:$addr, rGPR:$Rm, pred:$p)>;
 def VLD3LNqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeQWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListThreeQWordIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VLD3 multiple structure pseudo-instructions. These need special handling for
 // the vector operands that the normal instructions don't yet model.
 // FIXME: Remove these when the register classes and instructions are updated.
 def VLD3dAsm_8 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3dAsm_16 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3dAsm_32 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3qAsm_8 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3qAsm_16 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3qAsm_32 : NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 
 def VLD3dWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3dWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3dWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3qWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3qWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3qWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VLD3dWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeD:$list, addrmode6:$addr,
+                  (ins VecListThreeD:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3dWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeD:$list, addrmode6:$addr,
+                  (ins VecListThreeD:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3dWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeD:$list, addrmode6:$addr,
+                  (ins VecListThreeD:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3qWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeQ:$list, addrmode6:$addr,
+                  (ins VecListThreeQ:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3qWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeQ:$list, addrmode6:$addr,
+                  (ins VecListThreeQ:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD3qWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeQ:$list, addrmode6:$addr,
+                  (ins VecListThreeQ:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VST3 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VST3LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr",
-               (ins VecListThreeDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDByteIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr",
-               (ins VecListThreeDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr",
-               (ins VecListThreeDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNqAsm_16 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr",
-               (ins VecListThreeQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNqAsm_32 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr",
-               (ins VecListThreeQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 
 def VST3LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDByteIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeDWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQHWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQWordIndexed:$list, addrmode6alignNone:$addr,
+                    pred:$p)>;
 def VST3LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListThreeDByteIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeDHWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListThreeDHWordIndexed:$list,
+                       addrmode6alignNone:$addr, rGPR:$Rm, pred:$p)>;
 def VST3LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeDWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListThreeDWordIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3LNqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeQHWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListThreeQHWordIndexed:$list,
+                       addrmode6alignNone:$addr, rGPR:$Rm, pred:$p)>;
 def VST3LNqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeQWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListThreeQWordIndexed:$list, addrmode6alignNone:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 
@@ -6622,168 +7082,190 @@ def VST3LNqWB_register_Asm_32 :
 // the vector operands that the normal instructions don't yet model.
 // FIXME: Remove these when the register classes and instructions are updated.
 def VST3dAsm_8 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3dAsm_16 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3dAsm_32 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3qAsm_8 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3qAsm_16 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3qAsm_32 : NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 
 def VST3dWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3dWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3dWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeD:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3qWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr!",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3qWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr!",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3qWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr!",
-               (ins VecListThreeQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListThreeQ:$list, addrmode6align64:$addr, pred:$p)>;
 def VST3dWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeD:$list, addrmode6:$addr,
+                  (ins VecListThreeD:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3dWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeD:$list, addrmode6:$addr,
+                  (ins VecListThreeD:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3dWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeD:$list, addrmode6:$addr,
+                  (ins VecListThreeD:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3qWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListThreeQ:$list, addrmode6:$addr,
+                  (ins VecListThreeQ:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3qWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListThreeQ:$list, addrmode6:$addr,
+                  (ins VecListThreeQ:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST3qWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst3${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListThreeQ:$list, addrmode6:$addr,
+                  (ins VecListThreeQ:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VLD4 all-lanes pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VLD4DUPdAsm_8 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr",
-               (ins VecListFourDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDAllLanes:$list, addrmode6dupalign32:$addr,
+                    pred:$p)>;
 def VLD4DUPdAsm_16: NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr",
-               (ins VecListFourDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDAllLanes:$list, addrmode6dupalign64:$addr,
+                    pred:$p)>;
 def VLD4DUPdAsm_32: NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr",
-               (ins VecListFourDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDAllLanes:$list, addrmode6dupalign64or128:$addr,
+                    pred:$p)>;
 def VLD4DUPqAsm_8 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr",
-               (ins VecListFourQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQAllLanes:$list, addrmode6dupalign32:$addr,
+                    pred:$p)>;
 def VLD4DUPqAsm_16: NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr",
-               (ins VecListFourQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQAllLanes:$list, addrmode6dupalign64:$addr,
+                    pred:$p)>;
 def VLD4DUPqAsm_32: NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr",
-               (ins VecListFourQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQAllLanes:$list, addrmode6dupalign64or128:$addr,
+                    pred:$p)>;
 
 def VLD4DUPdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr!",
-               (ins VecListFourDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDAllLanes:$list, addrmode6dupalign32:$addr,
+                    pred:$p)>;
 def VLD4DUPdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr!",
-               (ins VecListFourDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDAllLanes:$list, addrmode6dupalign64:$addr,
+                    pred:$p)>;
 def VLD4DUPdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr!",
-               (ins VecListFourDAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDAllLanes:$list, addrmode6dupalign64or128:$addr,
+                    pred:$p)>;
 def VLD4DUPqWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr!",
-               (ins VecListFourQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQAllLanes:$list, addrmode6dupalign32:$addr,
+                    pred:$p)>;
 def VLD4DUPqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr!",
-               (ins VecListFourQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQAllLanes:$list, addrmode6dupalign64:$addr,
+                    pred:$p)>;
 def VLD4DUPqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr!",
-               (ins VecListFourQAllLanes:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQAllLanes:$list, addrmode6dupalign64or128:$addr,
+                    pred:$p)>;
 def VLD4DUPdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourDAllLanes:$list, addrmode6:$addr,
+                  (ins VecListFourDAllLanes:$list, addrmode6dupalign32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4DUPdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourDAllLanes:$list, addrmode6:$addr,
+                  (ins VecListFourDAllLanes:$list, addrmode6dupalign64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4DUPdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourDAllLanes:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListFourDAllLanes:$list,
+                       addrmode6dupalign64or128:$addr, rGPR:$Rm, pred:$p)>;
 def VLD4DUPqWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourQAllLanes:$list, addrmode6:$addr,
+                  (ins VecListFourQAllLanes:$list, addrmode6dupalign32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4DUPqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourQAllLanes:$list, addrmode6:$addr,
+                  (ins VecListFourQAllLanes:$list, addrmode6dupalign64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4DUPqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourQAllLanes:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListFourQAllLanes:$list,
+                       addrmode6dupalign64or128:$addr, rGPR:$Rm, pred:$p)>;
 
 
 // VLD4 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VLD4LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr",
-               (ins VecListFourDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDByteIndexed:$list, addrmode6align32:$addr,
+                    pred:$p)>;
 def VLD4LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr",
-               (ins VecListFourDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VLD4LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr",
-               (ins VecListFourDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 def VLD4LNqAsm_16 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr",
-               (ins VecListFourQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VLD4LNqAsm_32 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr",
-               (ins VecListFourQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 
 def VLD4LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr!",
-               (ins VecListFourDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDByteIndexed:$list, addrmode6align32:$addr,
+                    pred:$p)>;
 def VLD4LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr!",
-               (ins VecListFourDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VLD4LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr!",
-               (ins VecListFourDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 def VLD4LNqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr!",
-               (ins VecListFourQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VLD4LNqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr!",
-               (ins VecListFourQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 def VLD4LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListFourDByteIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourDHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListFourDHWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourDWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListFourDWordIndexed:$list,
+                       addrmode6align64or128:$addr, rGPR:$Rm, pred:$p)>;
 def VLD4LNqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourQHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListFourQHWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4LNqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourQWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListFourQWordIndexed:$list,
+                       addrmode6align64or128:$addr, rGPR:$Rm, pred:$p)>;
 
 
 
@@ -6791,168 +7273,202 @@ def VLD4LNqWB_register_Asm_32 :
 // the vector operands that the normal instructions don't yet model.
 // FIXME: Remove these when the register classes and instructions are updated.
 def VLD4dAsm_8 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4dAsm_16 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4dAsm_32 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4qAsm_8 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4qAsm_16 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4qAsm_32 : NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 
 def VLD4dWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr!",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4dWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr!",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4dWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr!",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4qWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr!",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4qWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr!",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4qWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr!",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                pred:$p)>;
 def VLD4dWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourD:$list, addrmode6:$addr,
+                  (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4dWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourD:$list, addrmode6:$addr,
+                  (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4dWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourD:$list, addrmode6:$addr,
+                  (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4qWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourQ:$list, addrmode6:$addr,
+                  (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4qWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourQ:$list, addrmode6:$addr,
+                  (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VLD4qWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vld4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourQ:$list, addrmode6:$addr,
+                  (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VST4 single-lane pseudo-instructions. These need special handling for
 // the lane index that an InstAlias can't handle, so we use these instead.
 def VST4LNdAsm_8 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr",
-               (ins VecListFourDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDByteIndexed:$list, addrmode6align32:$addr,
+                    pred:$p)>;
 def VST4LNdAsm_16 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr",
-               (ins VecListFourDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VST4LNdAsm_32 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr",
-               (ins VecListFourDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 def VST4LNqAsm_16 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr",
-               (ins VecListFourQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VST4LNqAsm_32 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr",
-               (ins VecListFourQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 
 def VST4LNdWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr!",
-               (ins VecListFourDByteIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDByteIndexed:$list, addrmode6align32:$addr,
+                    pred:$p)>;
 def VST4LNdWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr!",
-               (ins VecListFourDHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VST4LNdWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr!",
-               (ins VecListFourDWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourDWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 def VST4LNqWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr!",
-               (ins VecListFourQHWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQHWordIndexed:$list, addrmode6align64:$addr,
+                    pred:$p)>;
 def VST4LNqWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr!",
-               (ins VecListFourQWordIndexed:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQWordIndexed:$list, addrmode6align64or128:$addr,
+                    pred:$p)>;
 def VST4LNdWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourDByteIndexed:$list, addrmode6:$addr,
+                  (ins VecListFourDByteIndexed:$list, addrmode6align32:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4LNdWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourDHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListFourDHWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4LNdWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourDWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListFourDWordIndexed:$list,
+                       addrmode6align64or128:$addr, rGPR:$Rm, pred:$p)>;
 def VST4LNqWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourQHWordIndexed:$list, addrmode6:$addr,
+                  (ins VecListFourQHWordIndexed:$list, addrmode6align64:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4LNqWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourQWordIndexed:$list, addrmode6:$addr,
-                       rGPR:$Rm, pred:$p)>;
+                  (ins VecListFourQWordIndexed:$list,
+                       addrmode6align64or128:$addr, rGPR:$Rm, pred:$p)>;
 
 
 // VST4 multiple structure pseudo-instructions. These need special handling for
 // the vector operands that the normal instructions don't yet model.
 // FIXME: Remove these when the register classes and instructions are updated.
 def VST4dAsm_8 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4dAsm_16 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4dAsm_32 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4qAsm_8 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4qAsm_16 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4qAsm_32 : NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 
 def VST4dWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr!",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4dWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr!",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4dWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr!",
-               (ins VecListFourD:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4qWB_fixed_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr!",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4qWB_fixed_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr!",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4qWB_fixed_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr!",
-               (ins VecListFourQ:$list, addrmode6:$addr, pred:$p)>;
+               (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
+                    pred:$p)>;
 def VST4dWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourD:$list, addrmode6:$addr,
+                  (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4dWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourD:$list, addrmode6:$addr,
+                  (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4dWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourD:$list, addrmode6:$addr,
+                  (ins VecListFourD:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4qWB_register_Asm_8 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".8", "$list, $addr, $Rm",
-                  (ins VecListFourQ:$list, addrmode6:$addr,
+                  (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4qWB_register_Asm_16 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".16", "$list, $addr, $Rm",
-                  (ins VecListFourQ:$list, addrmode6:$addr,
+                  (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 def VST4qWB_register_Asm_32 :
         NEONDataTypeAsmPseudoInst<"vst4${p}", ".32", "$list, $addr, $Rm",
-                  (ins VecListFourQ:$list, addrmode6:$addr,
+                  (ins VecListFourQ:$list, addrmode6align64or128or256:$addr,
                        rGPR:$Rm, pred:$p)>;
 
 // VMOV/VMVN takes an optional datatype suffix
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrThumb.td b/contrib/llvm/lib/Target/ARM/ARMInstrThumb.td
index af5ef537b536..e17f73af03ec 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrThumb.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrThumb.td
@@ -269,7 +269,8 @@ class T1SystemEncoding<bits<8> opc>
   let Inst{7-0} = opc;
 }
 
-def tHINT : T1pI<(outs), (ins imm0_15:$imm), NoItinerary, "hint", "\t$imm", []>,
+def tHINT : T1pI<(outs), (ins imm0_15:$imm), NoItinerary, "hint", "\t$imm",
+                 [(int_arm_hint imm0_15:$imm)]>,
             T1SystemEncoding<0x00>,
             Requires<[IsThumb, HasV6M]> {
   bits<4> imm;
@@ -288,7 +289,6 @@ def : tHintAlias<"sev$p", (tHINT 4, pred:$p)>; // A8.6.157
 def : tInstAlias<"sevl$p", (tHINT 5, pred:$p)> {
   let Predicates = [IsThumb2, HasV8];
 }
-def : T2Pat<(int_arm_sevl), (tHINT 5)>;
 
 // The imm operand $val can be used by a debugger to store more information
 // about the breakpoint.
@@ -300,6 +300,8 @@ def tBKPT : T1I<(outs), (ins imm0_255:$val), NoItinerary, "bkpt\t$val",
   bits<8> val;
   let Inst{7-0} = val;
 }
+// default immediate for breakpoint mnemonic
+def : InstAlias<"bkpt", (tBKPT 0)>, Requires<[IsThumb]>;
 
 def tHLT : T1I<(outs), (ins imm0_63:$val), NoItinerary, "hlt\t$val",
                 []>, T1Encoding<0b101110>, Requires<[IsThumb, HasV8]> {
@@ -543,15 +545,15 @@ let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1 in {
                      (tBX GPR:$dst, (ops 14, zero_reg))>,
                      Requires<[IsThumb]>, Sched<[WriteBr]>;
   }
-  // tTAILJMPd: IOS version uses a Thumb2 branch (no Thumb1 tail calls
-  // on IOS), so it's in ARMInstrThumb2.td.
-  // Non-IOS version:
+  // tTAILJMPd: MachO version uses a Thumb2 branch (no Thumb1 tail calls
+  // on MachO), so it's in ARMInstrThumb2.td.
+  // Non-MachO version:
   let Uses = [SP] in {
     def tTAILJMPdND : tPseudoExpand<(outs),
                    (ins t_brtarget:$dst, pred:$p),
                    4, IIC_Br, [],
                    (tB t_brtarget:$dst, pred:$p)>,
-                 Requires<[IsThumb, IsNotIOS]>, Sched<[WriteBr]>;
+                 Requires<[IsThumb, IsNotMachO]>, Sched<[WriteBr]>;
   }
 }
 
@@ -1191,6 +1193,15 @@ def tTST :                      // A8.6.230
                [(ARMcmpZ (and_su tGPR:$Rn, tGPR:$Rm), 0)]>,
                Sched<[WriteALU]>;
 
+// A8.8.247  UDF - Undefined (Encoding T1)
+def tUDF : TI<(outs), (ins imm0_255:$imm8), IIC_Br, "udf\t$imm8",
+              [(int_arm_undefined imm0_255:$imm8)]>, Encoding16 {
+  bits<8> imm8;
+  let Inst{15-12} = 0b1101;
+  let Inst{11-8} = 0b1110;
+  let Inst{7-0} = imm8;
+}
+
 // Zero-extend byte
 def tUXTB :                     // A8.6.262
   T1pIMiscEncode<{0,0,1,0,1,1,?}, (outs tGPR:$Rd), (ins tGPR:$Rm),
@@ -1306,10 +1317,35 @@ def : T1Pat<(addc   tGPR:$lhs, imm8_255_neg:$rhs),
 def : T1Pat<(subc   tGPR:$lhs, tGPR:$rhs),
             (tSUBrr tGPR:$lhs, tGPR:$rhs)>;
 
-// ConstantPool, GlobalAddress
-def : T1Pat<(ARMWrapper  tglobaladdr :$dst), (tLEApcrel tglobaladdr :$dst)>;
+// Bswap 16 with load/store
+def : T1Pat<(srl (bswap (extloadi16 t_addrmode_rrs2:$addr)), (i32 16)),
+            (tREV16 (tLDRHr t_addrmode_rrs2:$addr))>;
+def : T1Pat<(srl (bswap (extloadi16 t_addrmode_is2:$addr)), (i32 16)),
+            (tREV16 (tLDRHi t_addrmode_is2:$addr))>;
+def : T1Pat<(truncstorei16 (srl (bswap tGPR:$Rn), (i32 16)),
+                           t_addrmode_rrs2:$addr),
+            (tSTRHr (tREV16 tGPR:$Rn), t_addrmode_rrs2:$addr)>;
+def : T1Pat<(truncstorei16 (srl (bswap tGPR:$Rn), (i32 16)),
+                           t_addrmode_is2:$addr),
+            (tSTRHi(tREV16 tGPR:$Rn), t_addrmode_is2:$addr)>;
+
+// ConstantPool
 def : T1Pat<(ARMWrapper  tconstpool  :$dst), (tLEApcrel tconstpool  :$dst)>;
 
+// GlobalAddress
+def tLDRLIT_ga_pcrel : PseudoInst<(outs tGPR:$dst), (ins i32imm:$addr),
+                                  IIC_iLoadiALU,
+                                  [(set tGPR:$dst,
+                                        (ARMWrapperPIC tglobaladdr:$addr))]>,
+                       Requires<[IsThumb, DontUseMovt]>;
+
+def tLDRLIT_ga_abs : PseudoInst<(outs tGPR:$dst), (ins i32imm:$src),
+                                IIC_iLoad_i,
+                                [(set tGPR:$dst,
+                                      (ARMWrapper tglobaladdr:$src))]>,
+                     Requires<[IsThumb, DontUseMovt]>;
+
+
 // JumpTable
 def : T1Pat<(ARMWrapperJT tjumptable:$dst, imm:$id),
             (tLEApcrelJT tjumptable:$dst, imm:$id)>;
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrThumb2.td b/contrib/llvm/lib/Target/ARM/ARMInstrThumb2.td
index 48acffd3a64e..85e93516807b 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrThumb2.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrThumb2.td
@@ -1445,7 +1445,7 @@ defm t2STRH:T2I_st<0b01,"strh", IIC_iStore_bh_i, IIC_iStore_bh_si,
 // Store doubleword
 let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 in
 def t2STRDi8 : T2Ii8s4<1, 0, 0, (outs),
-                       (ins GPR:$Rt, GPR:$Rt2, t2addrmode_imm8s4:$addr),
+                       (ins rGPR:$Rt, rGPR:$Rt2, t2addrmode_imm8s4:$addr),
                IIC_iStore_d_r, "strd", "\t$Rt, $Rt2, $addr", "", []>;
 
 // Indexed stores
@@ -1676,7 +1676,7 @@ defm t2PLI    : T2Ipl<0, 1, "pli">,  Requires<[IsThumb2,HasV7]>;
 // pci variant is very similar to i12, but supports negative offsets
 // from the PC. Only PLD and PLI have pci variants (not PLDW)
 class T2Iplpci<bits<1> inst, string opc> : T2Iso<(outs), (ins t2ldrlabel:$addr),
-               IIC_Preload, opc, "\t$addr", 
+               IIC_Preload, opc, "\t$addr",
                [(ARMPreload (ARMWrapper tconstpool:$addr),
                 (i32 0), (i32 inst))]>, Sched<[WritePreLd]> {
   let Inst{31-25} = 0b1111100;
@@ -1918,7 +1918,7 @@ def t2MOVi16 : T2I<(outs rGPR:$Rd), (ins imm0_65535_expr:$imm), IIC_iMOVi,
   let DecoderMethod = "DecodeT2MOVTWInstruction";
 }
 
-def : t2InstAlias<"mov${p} $Rd, $imm", 
+def : t2InstAlias<"mov${p} $Rd, $imm",
                   (t2MOVi16 rGPR:$Rd, imm256_65535_expr:$imm, pred:$p)>;
 
 def t2MOVi16_ga_pcrel : PseudoInst<(outs rGPR:$Rd),
@@ -2407,6 +2407,19 @@ def t2UBFX: T2TwoRegBitFI<
   let Inst{15} = 0;
 }
 
+// A8.8.247  UDF - Undefined (Encoding T2)
+def t2UDF : T2XI<(outs), (ins imm0_65535:$imm16), IIC_Br, "udf.w\t$imm16",
+                 [(int_arm_undefined imm0_65535:$imm16)]> {
+  bits<16> imm16;
+  let Inst{31-29} = 0b111;
+  let Inst{28-27} = 0b10;
+  let Inst{26-20} = 0b1111111;
+  let Inst{19-16} = imm16{15-12};
+  let Inst{15} = 0b1;
+  let Inst{14-12} = 0b010;
+  let Inst{11-0} = imm16{11-0};
+}
+
 // A8.6.18  BFI - Bitfield insert (Encoding T1)
 let Constraints = "$src = $Rd" in {
   def t2BFI : T2TwoRegBitFI<(outs rGPR:$Rd),
@@ -3196,27 +3209,28 @@ def t2MOVCCi32imm
 let hasSideEffects = 1 in {
 def t2DMB : T2I<(outs), (ins memb_opt:$opt), NoItinerary,
                 "dmb", "\t$opt", [(int_arm_dmb (i32 imm0_15:$opt))]>,
-                Requires<[HasDB]> {
+                Requires<[IsThumb, HasDB]> {
   bits<4> opt;
   let Inst{31-4} = 0xf3bf8f5;
   let Inst{3-0} = opt;
 }
-}
 
 def t2DSB : T2I<(outs), (ins memb_opt:$opt), NoItinerary,
                 "dsb", "\t$opt", [(int_arm_dsb (i32 imm0_15:$opt))]>,
-                Requires<[HasDB]> {
+                Requires<[IsThumb, HasDB]> {
   bits<4> opt;
   let Inst{31-4} = 0xf3bf8f4;
   let Inst{3-0} = opt;
 }
 
 def t2ISB : T2I<(outs), (ins instsyncb_opt:$opt), NoItinerary,
-                "isb", "\t$opt", []>, Requires<[HasDB]> {
+                "isb", "\t$opt", [(int_arm_isb (i32 imm0_15:$opt))]>,
+                Requires<[IsThumb, HasDB]> {
   bits<4> opt;
   let Inst{31-4} = 0xf3bf8f6;
   let Inst{3-0} = opt;
 }
+}
 
 class T2I_ldrex<bits<4> opcod, dag oops, dag iops, AddrMode am, int sz,
                 InstrItinClass itin, string opc, string asm, string cstr,
@@ -3284,15 +3298,18 @@ def t2LDREXD : T2I_ldrex<0b0111, (outs rGPR:$Rt, rGPR:$Rt2),
 def t2LDAEXB : T2I_ldrex<0b1100, (outs rGPR:$Rt), (ins addr_offset_none:$addr),
                          AddrModeNone, 4, NoItinerary,
                          "ldaexb", "\t$Rt, $addr", "",
-                         []>, Requires<[IsThumb, HasV8]>;
+                         [(set rGPR:$Rt, (ldaex_1 addr_offset_none:$addr))]>,
+               Requires<[IsThumb, HasV8]>;
 def t2LDAEXH : T2I_ldrex<0b1101, (outs rGPR:$Rt), (ins addr_offset_none:$addr),
                          AddrModeNone, 4, NoItinerary,
                          "ldaexh", "\t$Rt, $addr", "",
-                         []>, Requires<[IsThumb, HasV8]>;
+                         [(set rGPR:$Rt, (ldaex_2 addr_offset_none:$addr))]>,
+               Requires<[IsThumb, HasV8]>;
 def t2LDAEX  : Thumb2I<(outs rGPR:$Rt), (ins addr_offset_none:$addr),
                        AddrModeNone, 4, NoItinerary,
                        "ldaex", "\t$Rt, $addr", "",
-                     []>, Requires<[IsThumb, HasV8]> {
+                         [(set rGPR:$Rt, (ldaex_4 addr_offset_none:$addr))]>,
+               Requires<[IsThumb, HasV8]> {
   bits<4> Rt;
   bits<4> addr;
   let Inst{31-27} = 0b11101;
@@ -3320,21 +3337,21 @@ def t2STREXB : T2I_strex<0b0100, (outs rGPR:$Rd),
                          (ins rGPR:$Rt, addr_offset_none:$addr),
                          AddrModeNone, 4, NoItinerary,
                          "strexb", "\t$Rd, $Rt, $addr", "",
-                         [(set rGPR:$Rd, (strex_1 rGPR:$Rt,
-                                                  addr_offset_none:$addr))]>;
+                         [(set rGPR:$Rd,
+                               (strex_1 rGPR:$Rt, addr_offset_none:$addr))]>;
 def t2STREXH : T2I_strex<0b0101, (outs rGPR:$Rd),
                          (ins rGPR:$Rt, addr_offset_none:$addr),
                          AddrModeNone, 4, NoItinerary,
                          "strexh", "\t$Rd, $Rt, $addr", "",
-                         [(set rGPR:$Rd, (strex_2 rGPR:$Rt,
-                                                  addr_offset_none:$addr))]>;
+                         [(set rGPR:$Rd,
+                               (strex_2 rGPR:$Rt, addr_offset_none:$addr))]>;
 
 def t2STREX  : Thumb2I<(outs rGPR:$Rd), (ins rGPR:$Rt,
                              t2addrmode_imm0_1020s4:$addr),
                   AddrModeNone, 4, NoItinerary,
                   "strex", "\t$Rd, $Rt, $addr", "",
-                  [(set rGPR:$Rd, (strex_4 rGPR:$Rt,
-                                           t2addrmode_imm0_1020s4:$addr))]> {
+                  [(set rGPR:$Rd,
+                        (strex_4 rGPR:$Rt, t2addrmode_imm0_1020s4:$addr))]> {
   bits<4> Rd;
   bits<4> Rt;
   bits<12> addr;
@@ -3358,19 +3375,25 @@ def t2STLEXB : T2I_strex<0b1100, (outs rGPR:$Rd),
                          (ins rGPR:$Rt, addr_offset_none:$addr),
                          AddrModeNone, 4, NoItinerary,
                          "stlexb", "\t$Rd, $Rt, $addr", "",
-                         []>, Requires<[IsThumb, HasV8]>;
+                         [(set rGPR:$Rd,
+                               (stlex_1 rGPR:$Rt, addr_offset_none:$addr))]>,
+                         Requires<[IsThumb, HasV8]>;
 
 def t2STLEXH : T2I_strex<0b1101, (outs rGPR:$Rd),
                          (ins rGPR:$Rt, addr_offset_none:$addr),
                          AddrModeNone, 4, NoItinerary,
                          "stlexh", "\t$Rd, $Rt, $addr", "",
-                         []>, Requires<[IsThumb, HasV8]>;
+                         [(set rGPR:$Rd,
+                               (stlex_2 rGPR:$Rt, addr_offset_none:$addr))]>,
+                         Requires<[IsThumb, HasV8]>;
 
 def t2STLEX  : Thumb2I<(outs rGPR:$Rd), (ins rGPR:$Rt,
                              addr_offset_none:$addr),
                   AddrModeNone, 4, NoItinerary,
                   "stlex", "\t$Rd, $Rt, $addr", "",
-                  []>, Requires<[IsThumb, HasV8]> {
+                  [(set rGPR:$Rd,
+                        (stlex_4 rGPR:$Rt, addr_offset_none:$addr))]>,
+                  Requires<[IsThumb, HasV8]> {
   bits<4> Rd;
   bits<4> Rt;
   bits<4> addr;
@@ -3412,6 +3435,15 @@ def : T2Pat<(strex_1 (and GPR:$Rt, 0xff), addr_offset_none:$addr),
 def : T2Pat<(strex_2 (and GPR:$Rt, 0xffff), addr_offset_none:$addr),
             (t2STREXH GPR:$Rt, addr_offset_none:$addr)>;
 
+def : T2Pat<(and (ldaex_1 addr_offset_none:$addr), 0xff),
+            (t2LDAEXB addr_offset_none:$addr)>;
+def : T2Pat<(and (ldaex_2 addr_offset_none:$addr), 0xffff),
+            (t2LDAEXH addr_offset_none:$addr)>;
+def : T2Pat<(stlex_1 (and GPR:$Rt, 0xff), addr_offset_none:$addr),
+            (t2STLEXB GPR:$Rt, addr_offset_none:$addr)>;
+def : T2Pat<(stlex_2 (and GPR:$Rt, 0xffff), addr_offset_none:$addr),
+            (t2STLEXH GPR:$Rt, addr_offset_none:$addr)>;
+
 //===----------------------------------------------------------------------===//
 // SJLJ Exception handling intrinsics
 //   eh_sjlj_setjmp() is an instruction sequence to store the return
@@ -3477,8 +3509,8 @@ def t2B   : T2I<(outs), (ins uncondbrtarget:$target), IIC_Br,
   let Inst{25-16} = target{20-11};
   let Inst{10-0} = target{10-0};
   let DecoderMethod = "DecodeT2BInstruction";
-  let AsmMatchConverter = "cvtThumbBranches"; 
-} 
+  let AsmMatchConverter = "cvtThumbBranches";
+}
 
 let isNotDuplicable = 1, isIndirectBranch = 1 in {
 def t2BR_JT : t2PseudoInst<(outs),
@@ -3549,7 +3581,7 @@ def t2Bcc : T2I<(outs), (ins brtarget:$target), IIC_Br,
   let AsmMatchConverter = "cvtThumbBranches";
 }
 
-// Tail calls. The IOS version of thumb tail calls uses a t2 branch, so
+// Tail calls. The MachO version of thumb tail calls uses a t2 branch, so
 // it goes here.
 let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1 in {
   // IOS version.
@@ -3558,7 +3590,7 @@ let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1 in {
                    (ins uncondbrtarget:$dst, pred:$p),
                    4, IIC_Br, [],
                    (t2B uncondbrtarget:$dst, pred:$p)>,
-                 Requires<[IsThumb2, IsIOS]>, Sched<[WriteBr]>;
+                 Requires<[IsThumb2, IsMachO]>, Sched<[WriteBr]>;
 }
 
 // IT block
@@ -3653,7 +3685,8 @@ def : t2InstAlias<"cps.w $mode", (t2CPS1p imm0_31:$mode), 0>;
 
 // A6.3.4 Branches and miscellaneous control
 // Table A6-14 Change Processor State, and hint instructions
-def t2HINT : T2I<(outs), (ins imm0_239:$imm), NoItinerary, "hint", ".w\t$imm",[]> {
+def t2HINT : T2I<(outs), (ins imm0_239:$imm), NoItinerary, "hint", ".w\t$imm",
+                  [(int_arm_hint imm0_239:$imm)]> {
   bits<8> imm;
   let Inst{31-3} = 0b11110011101011111000000000000;
   let Inst{7-0} = imm;
@@ -3680,7 +3713,7 @@ def t2DBG : T2I<(outs), (ins imm0_15:$opt), NoItinerary, "dbg", "\t$opt", []> {
 
 // Secure Monitor Call is a system instruction.
 // Option = Inst{19-16}
-def t2SMC : T2I<(outs), (ins imm0_15:$opt), NoItinerary, "smc", "\t$opt", 
+def t2SMC : T2I<(outs), (ins imm0_15:$opt), NoItinerary, "smc", "\t$opt",
                 []>, Requires<[IsThumb2, HasTrustZone]> {
   let Inst{31-27} = 0b11110;
   let Inst{26-20} = 0b1111111;
@@ -3781,7 +3814,7 @@ def t2SUBS_PC_LR : T2I <(outs), (ins imm0_255:$imm), NoItinerary,
 let isReMaterializable = 1, isMoveImm = 1 in
 def t2MOVi32imm : PseudoInst<(outs rGPR:$dst), (ins i32imm:$src), IIC_iMOVix2,
                             [(set rGPR:$dst, (i32 imm:$src))]>,
-                            Requires<[IsThumb, HasV6T2]>;
+                            Requires<[IsThumb, UseMovt]>;
 
 // Pseudo instruction that combines movw + movt + add pc (if pic).
 // It also makes it possible to rematerialize the instructions.
@@ -3793,15 +3826,9 @@ def t2MOV_ga_pcrel : PseudoInst<(outs rGPR:$dst), (ins i32imm:$addr),
                           [(set rGPR:$dst, (ARMWrapperPIC tglobaladdr:$addr))]>,
                           Requires<[IsThumb2, UseMovt]>;
 
-def t2MOV_ga_dyn : PseudoInst<(outs rGPR:$dst), (ins i32imm:$addr),
-                              IIC_iMOVix2,
-                          [(set rGPR:$dst, (ARMWrapperDYN tglobaladdr:$addr))]>,
-                          Requires<[IsThumb2, UseMovt]>;
 }
 
 // ConstantPool, GlobalAddress, and JumpTable
-def : T2Pat<(ARMWrapper  tglobaladdr :$dst), (t2LEApcrel tglobaladdr :$dst)>,
-           Requires<[IsThumb2, DontUseMovt]>;
 def : T2Pat<(ARMWrapper  tconstpool  :$dst), (t2LEApcrel tconstpool  :$dst)>;
 def : T2Pat<(ARMWrapper  tglobaladdr :$dst), (t2MOVi32imm tglobaladdr :$dst)>,
            Requires<[IsThumb2, UseMovt]>;
@@ -4266,7 +4293,7 @@ def : t2InstAlias<"sbc${s}${p} $Rd, $Rn, $ShiftedRm",
 
 // Aliases for ADD without the ".w" optional width specifier.
 def : t2InstAlias<"add${s}${p} $Rd, $Rn, $imm",
-        (t2ADDri GPRnopc:$Rd, GPRnopc:$Rn, t2_so_imm:$imm, pred:$p, 
+        (t2ADDri GPRnopc:$Rd, GPRnopc:$Rn, t2_so_imm:$imm, pred:$p,
          cc_out:$s)>;
 def : t2InstAlias<"add${p} $Rd, $Rn, $imm",
            (t2ADDri12 GPRnopc:$Rd, GPR:$Rn, imm0_4095:$imm, pred:$p)>;
@@ -4371,7 +4398,7 @@ def : t2InstAlias<"ldrsh${p} $Rt, $addr",
                   (t2LDRSHs rGPR:$Rt, t2addrmode_so_reg:$addr, pred:$p)>;
 
 def : t2InstAlias<"ldr${p} $Rt, $addr",
-                  (t2LDRpci rGPR:$Rt, t2ldrlabel:$addr, pred:$p)>;
+                  (t2LDRpci GPRnopc:$Rt, t2ldrlabel:$addr, pred:$p)>;
 def : t2InstAlias<"ldrb${p} $Rt, $addr",
                   (t2LDRBpci rGPR:$Rt, t2ldrlabel:$addr, pred:$p)>;
 def : t2InstAlias<"ldrh${p} $Rt, $addr",
diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrVFP.td b/contrib/llvm/lib/Target/ARM/ARMInstrVFP.td
index a8cdc5ca0637..55a6efcb4c04 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrVFP.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrVFP.td
@@ -200,13 +200,34 @@ let mayLoad = 1, hasExtraDefRegAllocReq = 1 in
 defm VLDM : vfp_ldst_mult<"vldm", 1, IIC_fpLoad_m, IIC_fpLoad_mu>;
 
 let mayStore = 1, hasExtraSrcRegAllocReq = 1 in
-defm VSTM : vfp_ldst_mult<"vstm", 0, IIC_fpLoad_m, IIC_fpLoad_mu>;
+defm VSTM : vfp_ldst_mult<"vstm", 0, IIC_fpStore_m, IIC_fpStore_mu>;
 
 } // neverHasSideEffects
 
 def : MnemonicAlias<"vldm", "vldmia">;
 def : MnemonicAlias<"vstm", "vstmia">;
 
+// FLDM/FSTM - Load / Store multiple single / double precision registers for
+// pre-ARMv6 cores.
+// These instructions are deprecated!
+def : VFP2MnemonicAlias<"fldmias", "vldmia">;
+def : VFP2MnemonicAlias<"fldmdbs", "vldmdb">;
+def : VFP2MnemonicAlias<"fldmeas", "vldmdb">;
+def : VFP2MnemonicAlias<"fldmfds", "vldmia">;
+def : VFP2MnemonicAlias<"fldmiad", "vldmia">;
+def : VFP2MnemonicAlias<"fldmdbd", "vldmdb">;
+def : VFP2MnemonicAlias<"fldmead", "vldmdb">;
+def : VFP2MnemonicAlias<"fldmfdd", "vldmia">;
+
+def : VFP2MnemonicAlias<"fstmias", "vstmia">;
+def : VFP2MnemonicAlias<"fstmdbs", "vstmdb">;
+def : VFP2MnemonicAlias<"fstmeas", "vstmia">;
+def : VFP2MnemonicAlias<"fstmfds", "vstmdb">;
+def : VFP2MnemonicAlias<"fstmiad", "vstmia">;
+def : VFP2MnemonicAlias<"fstmdbd", "vstmdb">;
+def : VFP2MnemonicAlias<"fstmead", "vstmia">;
+def : VFP2MnemonicAlias<"fstmfdd", "vstmdb">;
+
 def : InstAlias<"vpush${p} $r", (VSTMDDB_UPD SP, pred:$p, dpr_reglist:$r)>,
                 Requires<[HasVFP2]>;
 def : InstAlias<"vpush${p} $r", (VSTMSDB_UPD SP, pred:$p, spr_reglist:$r)>,
@@ -247,7 +268,7 @@ multiclass vfp_ldstx_mult<string asm, bit L_bit> {
     AXXI4<(outs GPR:$wb), (ins GPR:$Rn, pred:$p, dpr_reglist:$regs, variable_ops),
           IndexModeUpd, !strconcat(asm, "dbx${p}\t$Rn!, $regs"), "$Rn = $wb", []> {
     let Inst{24-23} = 0b10;         // Decrement Before
-    let Inst{21}    = 1;
+    let Inst{21}    = 1;            // Writeback
     let Inst{20}    = L_bit;
   }
 }
@@ -255,6 +276,12 @@ multiclass vfp_ldstx_mult<string asm, bit L_bit> {
 defm FLDM : vfp_ldstx_mult<"fldm", 1>;
 defm FSTM : vfp_ldstx_mult<"fstm", 0>;
 
+def : VFP2MnemonicAlias<"fldmeax", "fldmdbx">;
+def : VFP2MnemonicAlias<"fldmfdx", "fldmiax">;
+
+def : VFP2MnemonicAlias<"fstmeax", "fstmiax">;
+def : VFP2MnemonicAlias<"fstmfdx", "fstmdbx">;
+
 //===----------------------------------------------------------------------===//
 // FP Binary Operations.
 //
@@ -524,12 +551,6 @@ def VCVTBSH: ASuI<0b11101, 0b11, 0b0011, 0b01, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTHS, "vcvtb", ".f16.f32\t$Sd, $Sm",
                  [/* For disassembly only; pattern left blank */]>;
 
-def : Pat<(f32_to_f16 SPR:$a),
-          (i32 (COPY_TO_REGCLASS (VCVTBSH SPR:$a), GPR))>;
-
-def : Pat<(f16_to_f32 GPR:$a),
-          (VCVTBHS (COPY_TO_REGCLASS GPR:$a, SPR))>;
-
 def VCVTTHS: ASuI<0b11101, 0b11, 0b0010, 0b11, 0, (outs SPR:$Sd), (ins SPR:$Sm),
                  /* FIXME */ IIC_fpCVTSH, "vcvtt", ".f32.f16\t$Sd, $Sm",
                  [/* For disassembly only; pattern left blank */]>;
@@ -592,6 +613,19 @@ def VCVTTDH : ADuI<0b11101, 0b11, 0b0011, 0b11, 0,
   let Inst{5}     = Dm{4};
 }
 
+def : Pat<(fp_to_f16 SPR:$a),
+          (i32 (COPY_TO_REGCLASS (VCVTBSH SPR:$a), GPR))>;
+
+def : Pat<(fp_to_f16 (f64 DPR:$a)),
+          (i32 (COPY_TO_REGCLASS (VCVTBDH DPR:$a), GPR))>;
+
+def : Pat<(f16_to_fp GPR:$a),
+          (VCVTBHS (COPY_TO_REGCLASS GPR:$a, SPR))>;
+
+def : Pat<(f64 (f16_to_fp GPR:$a)),
+          (VCVTBHD (COPY_TO_REGCLASS GPR:$a, SPR))>;
+
+
 multiclass vcvt_inst<string opc, bits<2> rm> {
   let PostEncoderMethod = "", DecoderNamespace = "VFPV8" in {
     def SS : ASuInp<0b11101, 0b11, 0b1100, 0b11, 0,
@@ -1639,7 +1673,7 @@ def FCONSTS : VFPAI<(outs SPR:$Sd), (ins vfp_f32imm:$imm),
 //===----------------------------------------------------------------------===//
 // Assembler aliases.
 //
-// A few mnemnoic aliases for pre-unifixed syntax. We don't guarantee to
+// A few mnemonic aliases for pre-unifixed syntax. We don't guarantee to
 // support them all, but supporting at least some of the basics is
 // good to be friendly.
 def : VFP2MnemonicAlias<"flds", "vldr">;
@@ -1735,3 +1769,14 @@ def : VFP2InstAlias<"vmov${p}.f64 $Dn, $Rt, $Rt2",
 // VMOVD does.
 def : VFP2InstAlias<"vmov${p} $Sd, $Sm",
                     (VMOVS SPR:$Sd, SPR:$Sm, pred:$p)>;
+
+// FCONSTD/FCONSTS alias for vmov.f64/vmov.f32
+// These aliases provide added functionality over vmov.f instructions by
+// allowing users to write assembly containing encoded floating point constants
+// (e.g. #0x70 vs #1.0).  Without these alises there is no way for the
+// assembler to accept encoded fp constants (but the equivalent fp-literal is
+// accepted directly by vmovf).
+def : VFP3InstAlias<"fconstd${p} $Dd, $val",
+                    (FCONSTD DPR:$Dd, vfp_f64imm:$val, pred:$p)>;
+def : VFP3InstAlias<"fconsts${p} $Sd, $val",
+                    (FCONSTS SPR:$Sd, vfp_f32imm:$val, pred:$p)>;
diff --git a/contrib/llvm/lib/Target/ARM/ARMJITInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMJITInfo.cpp
index 351a290e2aa0..6d1114d51aab 100644
--- a/contrib/llvm/lib/Target/ARM/ARMJITInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMJITInfo.cpp
@@ -11,12 +11,11 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "jit"
 #include "ARMJITInfo.h"
-#include "ARM.h"
 #include "ARMConstantPoolValue.h"
+#include "ARMMachineFunctionInfo.h"
 #include "ARMRelocations.h"
-#include "ARMSubtarget.h"
+#include "MCTargetDesc/ARMBaseInfo.h"
 #include "llvm/CodeGen/JITCodeEmitter.h"
 #include "llvm/IR/Function.h"
 #include "llvm/Support/Debug.h"
@@ -26,6 +25,8 @@
 #include <cstdlib>
 using namespace llvm;
 
+#define DEBUG_TYPE "jit"
+
 void ARMJITInfo::replaceMachineCodeForFunction(void *Old, void *New) {
   report_fatal_error("ARMJITInfo::replaceMachineCodeForFunction");
 }
@@ -320,13 +321,13 @@ void ARMJITInfo::relocate(void *Function, MachineRelocation *MR,
       break;
     }
     case ARM::reloc_arm_movw: {
-      ResultPtr = ResultPtr & 0xFFFF; 
+      ResultPtr = ResultPtr & 0xFFFF;
       *((intptr_t*)RelocPos) |= ResultPtr & 0xFFF;
       *((intptr_t*)RelocPos) |= ((ResultPtr >> 12) & 0xF) << 16;
       break;
     }
     case ARM::reloc_arm_movt: {
-      ResultPtr = (ResultPtr >> 16) & 0xFFFF; 
+      ResultPtr = (ResultPtr >> 16) & 0xFFFF;
       *((intptr_t*)RelocPos) |= ResultPtr & 0xFFF;
       *((intptr_t*)RelocPos) |= ((ResultPtr >> 12) & 0xF) << 16;
       break;
@@ -334,3 +335,10 @@ void ARMJITInfo::relocate(void *Function, MachineRelocation *MR,
     }
   }
 }
+
+void ARMJITInfo::Initialize(const MachineFunction &MF, bool isPIC) {
+  const ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
+  ConstPoolId2AddrMap.resize(AFI->getNumPICLabels());
+  JumpTableId2AddrMap.resize(AFI->getNumJumpTables());
+  IsPIC = isPIC;
+}
diff --git a/contrib/llvm/lib/Target/ARM/ARMJITInfo.h b/contrib/llvm/lib/Target/ARM/ARMJITInfo.h
index 23a6a9b512f4..27e2a2013404 100644
--- a/contrib/llvm/lib/Target/ARM/ARMJITInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMJITInfo.h
@@ -14,7 +14,6 @@
 #ifndef ARMJITINFO_H
 #define ARMJITINFO_H
 
-#include "ARMMachineFunctionInfo.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/CodeGen/MachineConstantPool.h"
@@ -53,45 +52,45 @@ namespace llvm {
     /// overwriting OLD with a branch to NEW.  This is used for self-modifying
     /// code.
     ///
-    virtual void replaceMachineCodeForFunction(void *Old, void *New);
+    void replaceMachineCodeForFunction(void *Old, void *New) override;
 
     /// emitGlobalValueIndirectSym - Use the specified JITCodeEmitter object
     /// to emit an indirect symbol which contains the address of the specified
     /// ptr.
-    virtual void *emitGlobalValueIndirectSym(const GlobalValue* GV, void *ptr,
-                                            JITCodeEmitter &JCE);
+    void *emitGlobalValueIndirectSym(const GlobalValue* GV, void *ptr,
+                                    JITCodeEmitter &JCE) override;
 
     // getStubLayout - Returns the size and alignment of the largest call stub
     // on ARM.
-    virtual StubLayout getStubLayout();
+    StubLayout getStubLayout() override;
 
     /// emitFunctionStub - Use the specified JITCodeEmitter object to emit a
     /// small native function that simply calls the function at the specified
     /// address.
-    virtual void *emitFunctionStub(const Function* F, void *Fn,
-                                   JITCodeEmitter &JCE);
+    void *emitFunctionStub(const Function* F, void *Fn,
+                           JITCodeEmitter &JCE) override;
 
     /// getLazyResolverFunction - Expose the lazy resolver to the JIT.
-    virtual LazyResolverFn getLazyResolverFunction(JITCompilerFn);
+    LazyResolverFn getLazyResolverFunction(JITCompilerFn) override;
 
     /// relocate - Before the JIT can run a block of code that has been emitted,
     /// it must rewrite the code to contain the actual addresses of any
     /// referenced global symbols.
-    virtual void relocate(void *Function, MachineRelocation *MR,
-                          unsigned NumRelocs, unsigned char* GOTBase);
+    void relocate(void *Function, MachineRelocation *MR,
+                  unsigned NumRelocs, unsigned char* GOTBase) override;
 
     /// hasCustomConstantPool - Allows a target to specify that constant
     /// pool address resolution is handled by the target.
-    virtual bool hasCustomConstantPool() const { return true; }
+    bool hasCustomConstantPool() const override { return true; }
 
     /// hasCustomJumpTables - Allows a target to specify that jumptables
     /// are emitted by the target.
-    virtual bool hasCustomJumpTables() const { return true; }
+    bool hasCustomJumpTables() const override { return true; }
 
     /// allocateSeparateGVMemory - If true, globals should be placed in
     /// separately allocated heap memory rather than in the same
     /// code memory allocated by JITCodeEmitter.
-    virtual bool allocateSeparateGVMemory() const {
+    bool allocateSeparateGVMemory() const override {
 #ifdef __APPLE__
       return true;
 #else
@@ -103,12 +102,7 @@ namespace llvm {
     /// Resize constant pool ids to CONSTPOOL_ENTRY addresses map; resize
     /// jump table ids to jump table bases map; remember if codegen relocation
     /// model is PIC.
-    void Initialize(const MachineFunction &MF, bool isPIC) {
-      const ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
-      ConstPoolId2AddrMap.resize(AFI->getNumPICLabels());
-      JumpTableId2AddrMap.resize(AFI->getNumJumpTables());
-      IsPIC = isPIC;
-    }
+    void Initialize(const MachineFunction &MF, bool isPIC);
 
     /// getConstantPoolEntryAddr - The ARM target puts all constant
     /// pool entries into constant islands. This returns the address of the
diff --git a/contrib/llvm/lib/Target/ARM/ARMLoadStoreOptimizer.cpp b/contrib/llvm/lib/Target/ARM/ARMLoadStoreOptimizer.cpp
index 61596d54b692..a03bcdbddd79 100644
--- a/contrib/llvm/lib/Target/ARM/ARMLoadStoreOptimizer.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMLoadStoreOptimizer.cpp
@@ -12,12 +12,14 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-ldst-opt"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMBaseRegisterInfo.h"
+#include "ARMISelLowering.h"
 #include "ARMMachineFunctionInfo.h"
+#include "ARMSubtarget.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
+#include "Thumb1RegisterInfo.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallPtrSet.h"
@@ -36,12 +38,13 @@
 #include "llvm/IR/Function.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
-#include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "arm-ldst-opt"
+
 STATISTIC(NumLDMGened , "Number of ldm instructions generated");
 STATISTIC(NumSTMGened , "Number of stm instructions generated");
 STATISTIC(NumVLDMGened, "Number of vldm instructions generated");
@@ -65,13 +68,14 @@ namespace {
     const TargetInstrInfo *TII;
     const TargetRegisterInfo *TRI;
     const ARMSubtarget *STI;
+    const TargetLowering *TL;
     ARMFunctionInfo *AFI;
     RegScavenger *RS;
-    bool isThumb2;
+    bool isThumb1, isThumb2;
 
-    virtual bool runOnMachineFunction(MachineFunction &Fn);
+    bool runOnMachineFunction(MachineFunction &Fn) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM load / store optimization pass";
     }
 
@@ -93,7 +97,10 @@ namespace {
     void findUsesOfImpDef(SmallVectorImpl<MachineOperand *> &UsesOfImpDefs,
                           const MemOpQueue &MemOps, unsigned DefReg,
                           unsigned RangeBegin, unsigned RangeEnd);
-
+    void UpdateBaseRegUses(MachineBasicBlock &MBB,
+                           MachineBasicBlock::iterator MBBI,
+                           DebugLoc dl, unsigned Base, unsigned WordOffset,
+                           ARMCC::CondCodes Pred, unsigned PredReg);
     bool MergeOps(MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI,
                   int Offset, unsigned Base, bool BaseKill, int Opcode,
                   ARMCC::CondCodes Pred, unsigned PredReg, unsigned Scratch,
@@ -119,7 +126,6 @@ namespace {
                       ARMCC::CondCodes Pred, unsigned PredReg,
                       unsigned Scratch, MemOpQueue &MemOps,
                       SmallVectorImpl<MachineBasicBlock::iterator> &Merges);
-
     void AdvanceRS(MachineBasicBlock &MBB, MemOpQueue &MemOps);
     bool FixInvalidRegPairOp(MachineBasicBlock &MBB,
                              MachineBasicBlock::iterator &MBBI);
@@ -159,6 +165,21 @@ static int getLoadStoreMultipleOpcode(int Opcode, ARM_AM::AMSubMode Mode) {
     case ARM_AM::db: return ARM::STMDB;
     case ARM_AM::ib: return ARM::STMIB;
     }
+  case ARM::tLDRi:
+    // tLDMIA is writeback-only - unless the base register is in the input
+    // reglist.
+    ++NumLDMGened;
+    switch (Mode) {
+    default: llvm_unreachable("Unhandled submode!");
+    case ARM_AM::ia: return ARM::tLDMIA;
+    }
+  case ARM::tSTRi:
+    // There is no non-writeback tSTMIA either.
+    ++NumSTMGened;
+    switch (Mode) {
+    default: llvm_unreachable("Unhandled submode!");
+    case ARM_AM::ia: return ARM::tSTMIA_UPD;
+    }
   case ARM::t2LDRi8:
   case ARM::t2LDRi12:
     ++NumLDMGened;
@@ -217,6 +238,9 @@ AMSubMode getLoadStoreMultipleSubMode(int Opcode) {
   case ARM::LDMIA_UPD:
   case ARM::STMIA:
   case ARM::STMIA_UPD:
+  case ARM::tLDMIA:
+  case ARM::tLDMIA_UPD:
+  case ARM::tSTMIA_UPD:
   case ARM::t2LDMIA_RET:
   case ARM::t2LDMIA:
   case ARM::t2LDMIA_UPD:
@@ -263,12 +287,20 @@ AMSubMode getLoadStoreMultipleSubMode(int Opcode) {
   } // end namespace ARM_AM
 } // end namespace llvm
 
+static bool isT1i32Load(unsigned Opc) {
+  return Opc == ARM::tLDRi;
+}
+
 static bool isT2i32Load(unsigned Opc) {
   return Opc == ARM::t2LDRi12 || Opc == ARM::t2LDRi8;
 }
 
 static bool isi32Load(unsigned Opc) {
-  return Opc == ARM::LDRi12 || isT2i32Load(Opc);
+  return Opc == ARM::LDRi12 || isT1i32Load(Opc) || isT2i32Load(Opc) ;
+}
+
+static bool isT1i32Store(unsigned Opc) {
+  return Opc == ARM::tSTRi;
 }
 
 static bool isT2i32Store(unsigned Opc) {
@@ -276,7 +308,102 @@ static bool isT2i32Store(unsigned Opc) {
 }
 
 static bool isi32Store(unsigned Opc) {
-  return Opc == ARM::STRi12 || isT2i32Store(Opc);
+  return Opc == ARM::STRi12 || isT1i32Store(Opc) || isT2i32Store(Opc);
+}
+
+static unsigned getImmScale(unsigned Opc) {
+  switch (Opc) {
+  default: llvm_unreachable("Unhandled opcode!");
+  case ARM::tLDRi:
+  case ARM::tSTRi:
+    return 1;
+  case ARM::tLDRHi:
+  case ARM::tSTRHi:
+    return 2;
+  case ARM::tLDRBi:
+  case ARM::tSTRBi:
+    return 4;
+  }
+}
+
+/// Update future uses of the base register with the offset introduced
+/// due to writeback. This function only works on Thumb1.
+void
+ARMLoadStoreOpt::UpdateBaseRegUses(MachineBasicBlock &MBB,
+                                   MachineBasicBlock::iterator MBBI,
+                                   DebugLoc dl, unsigned Base,
+                                   unsigned WordOffset,
+                                   ARMCC::CondCodes Pred, unsigned PredReg) {
+  assert(isThumb1 && "Can only update base register uses for Thumb1!");
+
+  // Start updating any instructions with immediate offsets. Insert a sub before
+  // the first non-updateable instruction (if any).
+  for (; MBBI != MBB.end(); ++MBBI) {
+    if (MBBI->readsRegister(Base)) {
+      unsigned Opc = MBBI->getOpcode();
+      int Offset;
+      bool InsertSub = false;
+
+      if (Opc == ARM::tLDRi  || Opc == ARM::tSTRi  ||
+          Opc == ARM::tLDRHi || Opc == ARM::tSTRHi ||
+          Opc == ARM::tLDRBi || Opc == ARM::tSTRBi) {
+        // Loads and stores with immediate offsets can be updated, but only if
+        // the new offset isn't negative.
+        // The MachineOperand containing the offset immediate is the last one
+        // before predicates.
+        MachineOperand &MO =
+          MBBI->getOperand(MBBI->getDesc().getNumOperands() - 3);
+        // The offsets are scaled by 1, 2 or 4 depending on the Opcode
+        Offset = MO.getImm() - WordOffset * getImmScale(Opc);
+        if (Offset >= 0)
+          MO.setImm(Offset);
+        else
+          InsertSub = true;
+
+      } else if (Opc == ARM::tSUBi8 || Opc == ARM::tADDi8) {
+        // SUB/ADD using this register. Merge it with the update.
+        // If the merged offset is too large, insert a new sub instead.
+        MachineOperand &MO =
+          MBBI->getOperand(MBBI->getDesc().getNumOperands() - 3);
+        Offset = (Opc == ARM::tSUBi8) ?
+          MO.getImm() + WordOffset * 4 :
+          MO.getImm() - WordOffset * 4 ;
+        if (TL->isLegalAddImmediate(Offset)) {
+          MO.setImm(Offset);
+          // The base register has now been reset, so exit early.
+          return;
+        } else {
+          InsertSub = true;
+        }
+
+      } else {
+        // Can't update the instruction.
+        InsertSub = true;
+      }
+
+      if (InsertSub) {
+        // An instruction above couldn't be updated, so insert a sub.
+        AddDefaultT1CC(BuildMI(MBB, MBBI, dl, TII->get(ARM::tSUBi8), Base))
+          .addReg(Base, getKillRegState(true)).addImm(WordOffset * 4)
+          .addImm(Pred).addReg(PredReg);
+        return;
+      }
+    }
+
+    if (MBBI->killsRegister(Base))
+      // Register got killed. Stop updating.
+      return;
+  }
+
+  // The end of the block was reached. This means register liveness escapes the
+  // block, and it's necessary to insert a sub before the last instruction.
+  if (MBB.succ_size() > 0)
+    // But only insert the SUB if there is actually a successor block.
+    // FIXME: Check more carefully if register is live at this point, e.g. by
+    // also examining the successor block's register liveness information.
+    AddDefaultT1CC(BuildMI(MBB, --MBBI, dl, TII->get(ARM::tSUBi8), Base))
+      .addReg(Base, getKillRegState(true)).addImm(WordOffset * 4)
+      .addImm(Pred).addReg(PredReg);
 }
 
 /// MergeOps - Create and insert a LDM or STM with Base as base register and
@@ -296,18 +423,19 @@ ARMLoadStoreOpt::MergeOps(MachineBasicBlock &MBB,
     return false;
 
   ARM_AM::AMSubMode Mode = ARM_AM::ia;
-  // VFP and Thumb2 do not support IB or DA modes.
+  // VFP and Thumb2 do not support IB or DA modes. Thumb1 only supports IA.
   bool isNotVFP = isi32Load(Opcode) || isi32Store(Opcode);
-  bool haveIBAndDA = isNotVFP && !isThumb2;
-  if (Offset == 4 && haveIBAndDA)
+  bool haveIBAndDA = isNotVFP && !isThumb2 && !isThumb1;
+
+  if (Offset == 4 && haveIBAndDA) {
     Mode = ARM_AM::ib;
-  else if (Offset == -4 * (int)NumRegs + 4 && haveIBAndDA)
+  } else if (Offset == -4 * (int)NumRegs + 4 && haveIBAndDA) {
     Mode = ARM_AM::da;
-  else if (Offset == -4 * (int)NumRegs && isNotVFP)
+  } else if (Offset == -4 * (int)NumRegs && isNotVFP && !isThumb1) {
     // VLDM/VSTM do not support DB mode without also updating the base reg.
     Mode = ARM_AM::db;
-  else if (Offset != 0) {
-    // Check if this is a supported opcode before we insert instructions to
+  } else if (Offset != 0) {
+    // Check if this is a supported opcode before inserting instructions to
     // calculate a new base register.
     if (!getLoadStoreMultipleOpcode(Opcode, Mode)) return false;
 
@@ -318,41 +446,98 @@ ARMLoadStoreOpt::MergeOps(MachineBasicBlock &MBB,
       return false;
 
     unsigned NewBase;
-    if (isi32Load(Opcode))
+    if (isi32Load(Opcode)) {
       // If it is a load, then just use one of the destination register to
       // use as the new base.
       NewBase = Regs[NumRegs-1].first;
-    else {
+    } else {
       // Use the scratch register to use as a new base.
       NewBase = Scratch;
       if (NewBase == 0)
         return false;
     }
-    int BaseOpc = !isThumb2 ? ARM::ADDri : ARM::t2ADDri;
+
+    int BaseOpc =
+      isThumb2 ? ARM::t2ADDri :
+      isThumb1 ? ARM::tADDi8  : ARM::ADDri;
+
     if (Offset < 0) {
-      BaseOpc = !isThumb2 ? ARM::SUBri : ARM::t2SUBri;
+      BaseOpc =
+        isThumb2 ? ARM::t2SUBri :
+        isThumb1 ? ARM::tSUBi8  : ARM::SUBri;
       Offset = - Offset;
     }
-    int ImmedOffset = isThumb2
-      ? ARM_AM::getT2SOImmVal(Offset) : ARM_AM::getSOImmVal(Offset);
-    if (ImmedOffset == -1)
-      // FIXME: Try t2ADDri12 or t2SUBri12?
-      return false;  // Probably not worth it then.
-
-    BuildMI(MBB, MBBI, dl, TII->get(BaseOpc), NewBase)
-      .addReg(Base, getKillRegState(BaseKill)).addImm(Offset)
-      .addImm(Pred).addReg(PredReg).addReg(0);
+
+    if (!TL->isLegalAddImmediate(Offset))
+      // FIXME: Try add with register operand?
+      return false; // Probably not worth it then.
+
+    if (isThumb1) {
+      if (Base != NewBase) {
+        // Need to insert a MOV to the new base first.
+        // FIXME: If the immediate fits in 3 bits, use ADD instead.
+        BuildMI(MBB, MBBI, dl, TII->get(ARM::tMOVr), NewBase)
+          .addReg(Base, getKillRegState(BaseKill))
+          .addImm(Pred).addReg(PredReg);
+      }
+      AddDefaultT1CC(BuildMI(MBB, MBBI, dl, TII->get(BaseOpc), NewBase))
+        .addReg(NewBase, getKillRegState(true)).addImm(Offset)
+        .addImm(Pred).addReg(PredReg);
+    } else {
+      BuildMI(MBB, MBBI, dl, TII->get(BaseOpc), NewBase)
+        .addReg(Base, getKillRegState(BaseKill)).addImm(Offset)
+        .addImm(Pred).addReg(PredReg).addReg(0);
+    }
+
     Base = NewBase;
-    BaseKill = true;  // New base is always killed right its use.
+    BaseKill = true; // New base is always killed straight away.
   }
 
   bool isDef = (isi32Load(Opcode) || Opcode == ARM::VLDRS ||
                 Opcode == ARM::VLDRD);
+
+  // Get LS multiple opcode. Note that for Thumb1 this might be an opcode with
+  // base register writeback.
   Opcode = getLoadStoreMultipleOpcode(Opcode, Mode);
   if (!Opcode) return false;
-  MachineInstrBuilder MIB = BuildMI(MBB, MBBI, dl, TII->get(Opcode))
-    .addReg(Base, getKillRegState(BaseKill))
-    .addImm(Pred).addReg(PredReg);
+
+  bool Writeback = isThumb1; // Thumb1 LDM/STM have base reg writeback.
+
+  // Exception: If the base register is in the input reglist, Thumb1 LDM is
+  // non-writeback. Check for this.
+  if (Opcode == ARM::tLDMIA && isThumb1)
+    for (unsigned I = 0; I < NumRegs; ++I)
+      if (Base == Regs[I].first) {
+        Writeback = false;
+        break;
+      }
+
+  MachineInstrBuilder MIB;
+
+  if (Writeback) {
+    if (Opcode == ARM::tLDMIA)
+      // Update tLDMIA with writeback if necessary.
+      Opcode = ARM::tLDMIA_UPD;
+
+    MIB = BuildMI(MBB, MBBI, dl, TII->get(Opcode));
+
+    // Thumb1: we might need to set base writeback when building the MI.
+    MIB.addReg(Base, getDefRegState(true))
+       .addReg(Base, getKillRegState(BaseKill));
+
+    // The base isn't dead after a merged instruction with writeback. Update
+    // future uses of the base with the added offset (if possible), or reset
+    // the base register as necessary.
+    if (!BaseKill)
+      UpdateBaseRegUses(MBB, MBBI, dl, Base, NumRegs, Pred, PredReg);
+  } else {
+    // No writeback, simply build the MachineInstr.
+    MIB = BuildMI(MBB, MBBI, dl, TII->get(Opcode));
+    MIB.addReg(Base, getKillRegState(BaseKill));
+  }
+
+  MIB.addImm(Pred).addReg(PredReg);
+
   for (unsigned i = 0; i != NumRegs; ++i)
     MIB = MIB.addReg(Regs[i].first, getDefRegState(isDef)
                      | getKillRegState(Regs[i].second));
@@ -484,7 +669,7 @@ void ARMLoadStoreOpt::MergeOpsUpdate(MachineBasicBlock &MBB,
     return;
 
   // Merge succeeded, update records.
-  Merges.push_back(prior(Loc));
+  Merges.push_back(std::prev(Loc));
 
   // In gathering loads together, we may have moved the imp-def of a register
   // past one of its uses. This is OK, since we know better than the rest of
@@ -492,7 +677,7 @@ void ARMLoadStoreOpt::MergeOpsUpdate(MachineBasicBlock &MBB,
   // affected uses.
   for (SmallVectorImpl<MachineOperand *>::iterator I = UsesOfImpDefs.begin(),
                                                    E = UsesOfImpDefs.end();
-       I != E; ++I)
+                                                   I != E; ++I)
     (*I)->setIsUndef();
 
   for (unsigned i = memOpsBegin; i < memOpsEnd; ++i) {
@@ -589,7 +774,6 @@ ARMLoadStoreOpt::MergeLDR_STR(MachineBasicBlock &MBB, unsigned SIndex,
   bool BaseKill = Loc->findRegisterUseOperandIdx(Base, true) != -1;
   MergeOpsUpdate(MBB, MemOps, SIndex, MemOps.size(), insertAfter, SOffset,
                  Base, BaseKill, Opcode, Pred, PredReg, Scratch, dl, Merges);
-  return;
 }
 
 static bool definesCPSR(MachineInstr *MI) {
@@ -616,6 +800,7 @@ static bool isMatchingDecrement(MachineInstr *MI, unsigned Base,
   bool CheckCPSRDef = false;
   switch (MI->getOpcode()) {
   default: return false;
+  case ARM::tSUBi8:
   case ARM::t2SUBri:
   case ARM::SUBri:
     CheckCPSRDef = true;
@@ -628,10 +813,11 @@ static bool isMatchingDecrement(MachineInstr *MI, unsigned Base,
   if (Bytes == 0 || (Limit && Bytes >= Limit))
     return false;
 
-  unsigned Scale = (MI->getOpcode() == ARM::tSUBspi) ? 4 : 1; // FIXME
+  unsigned Scale = (MI->getOpcode() == ARM::tSUBspi ||
+                    MI->getOpcode() == ARM::tSUBi8) ? 4 : 1; // FIXME
   if (!(MI->getOperand(0).getReg() == Base &&
         MI->getOperand(1).getReg() == Base &&
-        (MI->getOperand(2).getImm()*Scale) == Bytes &&
+        (MI->getOperand(2).getImm() * Scale) == Bytes &&
         getInstrPredicate(MI, MyPredReg) == Pred &&
         MyPredReg == PredReg))
     return false;
@@ -649,6 +835,7 @@ static bool isMatchingIncrement(MachineInstr *MI, unsigned Base,
   bool CheckCPSRDef = false;
   switch (MI->getOpcode()) {
   default: return false;
+  case ARM::tADDi8:
   case ARM::t2ADDri:
   case ARM::ADDri:
     CheckCPSRDef = true;
@@ -661,10 +848,11 @@ static bool isMatchingIncrement(MachineInstr *MI, unsigned Base,
     // Make sure the offset fits in 8 bits.
     return false;
 
-  unsigned Scale = (MI->getOpcode() == ARM::tADDspi) ? 4 : 1; // FIXME
+  unsigned Scale = (MI->getOpcode() == ARM::tADDspi ||
+                    MI->getOpcode() == ARM::tADDi8) ? 4 : 1; // FIXME
   if (!(MI->getOperand(0).getReg() == Base &&
         MI->getOperand(1).getReg() == Base &&
-        (MI->getOperand(2).getImm()*Scale) == Bytes &&
+        (MI->getOperand(2).getImm() * Scale) == Bytes &&
         getInstrPredicate(MI, MyPredReg) == Pred &&
         MyPredReg == PredReg))
     return false;
@@ -677,6 +865,8 @@ static inline unsigned getLSMultipleTransferSize(MachineInstr *MI) {
   default: return 0;
   case ARM::LDRi12:
   case ARM::STRi12:
+  case ARM::tLDRi:
+  case ARM::tSTRi:
   case ARM::t2LDRi8:
   case ARM::t2LDRi12:
   case ARM::t2STRi8:
@@ -695,6 +885,9 @@ static inline unsigned getLSMultipleTransferSize(MachineInstr *MI) {
   case ARM::STMDA:
   case ARM::STMDB:
   case ARM::STMIB:
+  case ARM::tLDMIA:
+  case ARM::tLDMIA_UPD:
+  case ARM::tSTMIA_UPD:
   case ARM::t2LDMIA:
   case ARM::t2LDMDB:
   case ARM::t2STMIA:
@@ -791,6 +984,9 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLSMultiple(MachineBasicBlock &MBB,
                                                MachineBasicBlock::iterator MBBI,
                                                bool &Advance,
                                                MachineBasicBlock::iterator &I) {
+  // Thumb1 is already using updating loads/stores.
+  if (isThumb1) return false;
+
   MachineInstr *MI = MBBI;
   unsigned Base = MI->getOperand(0).getReg();
   bool BaseKill = MI->getOperand(0).isKill();
@@ -812,7 +1008,7 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLSMultiple(MachineBasicBlock &MBB,
   // Try merging with the previous instruction.
   MachineBasicBlock::iterator BeginMBBI = MBB.begin();
   if (MBBI != BeginMBBI) {
-    MachineBasicBlock::iterator PrevMBBI = prior(MBBI);
+    MachineBasicBlock::iterator PrevMBBI = std::prev(MBBI);
     while (PrevMBBI != BeginMBBI && PrevMBBI->isDebugValue())
       --PrevMBBI;
     if (Mode == ARM_AM::ia &&
@@ -831,7 +1027,7 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLSMultiple(MachineBasicBlock &MBB,
   // Try merging with the next instruction.
   MachineBasicBlock::iterator EndMBBI = MBB.end();
   if (!DoMerge && MBBI != EndMBBI) {
-    MachineBasicBlock::iterator NextMBBI = llvm::next(MBBI);
+    MachineBasicBlock::iterator NextMBBI = std::next(MBBI);
     while (NextMBBI != EndMBBI && NextMBBI->isDebugValue())
       ++NextMBBI;
     if ((Mode == ARM_AM::ia || Mode == ARM_AM::ib) &&
@@ -927,6 +1123,10 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLoadStore(MachineBasicBlock &MBB,
                                                const TargetInstrInfo *TII,
                                                bool &Advance,
                                                MachineBasicBlock::iterator &I) {
+  // Thumb1 doesn't have updating LDR/STR.
+  // FIXME: Use LDM/STM with single register instead.
+  if (isThumb1) return false;
+
   MachineInstr *MI = MBBI;
   unsigned Base = MI->getOperand(1).getReg();
   bool BaseKill = MI->getOperand(1).isKill();
@@ -959,7 +1159,7 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLoadStore(MachineBasicBlock &MBB,
   // Try merging with the previous instruction.
   MachineBasicBlock::iterator BeginMBBI = MBB.begin();
   if (MBBI != BeginMBBI) {
-    MachineBasicBlock::iterator PrevMBBI = prior(MBBI);
+    MachineBasicBlock::iterator PrevMBBI = std::prev(MBBI);
     while (PrevMBBI != BeginMBBI && PrevMBBI->isDebugValue())
       --PrevMBBI;
     if (isMatchingDecrement(PrevMBBI, Base, Bytes, Limit, Pred, PredReg)) {
@@ -978,7 +1178,7 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLoadStore(MachineBasicBlock &MBB,
   // Try merging with the next instruction.
   MachineBasicBlock::iterator EndMBBI = MBB.end();
   if (!DoMerge && MBBI != EndMBBI) {
-    MachineBasicBlock::iterator NextMBBI = llvm::next(MBBI);
+    MachineBasicBlock::iterator NextMBBI = std::next(MBBI);
     while (NextMBBI != EndMBBI && NextMBBI->isDebugValue())
       ++NextMBBI;
     if (!isAM5 &&
@@ -1002,7 +1202,7 @@ bool ARMLoadStoreOpt::MergeBaseUpdateLoadStore(MachineBasicBlock &MBB,
     return false;
 
   if (isAM5) {
-    // VLDM[SD}_UPD, VSTM[SD]_UPD
+    // VLDM[SD]_UPD, VSTM[SD]_UPD
     // (There are no base-updating versions of VLDR/VSTR instructions, but the
     // updating load/store-multiple instructions can be used with only one
     // register.)
@@ -1100,6 +1300,8 @@ static bool isMemoryOp(const MachineInstr *MI) {
     return MI->getOperand(1).isReg();
   case ARM::LDRi12:
   case ARM::STRi12:
+  case ARM::tLDRi:
+  case ARM::tSTRi:
   case ARM::t2LDRi8:
   case ARM::t2LDRi12:
   case ARM::t2STRi8:
@@ -1122,7 +1324,7 @@ void ARMLoadStoreOpt::AdvanceRS(MachineBasicBlock &MBB, MemOpQueue &MemOps) {
   }
 
   if (Loc != MBB.begin())
-    RS->forward(prior(Loc));
+    RS->forward(std::prev(Loc));
 }
 
 static int getMemoryOpOffset(const MachineInstr *MI) {
@@ -1137,6 +1339,10 @@ static int getMemoryOpOffset(const MachineInstr *MI) {
       Opcode == ARM::LDRi12   || Opcode == ARM::STRi12)
     return OffField;
 
+  // Thumb1 immediate offsets are scaled by 4
+  if (Opcode == ARM::tLDRi || Opcode == ARM::tSTRi)
+    return OffField * 4;
+
   int Offset = isAM3 ? ARM_AM::getAM3Offset(OffField)
     : ARM_AM::getAM5Offset(OffField) * 4;
   if (isAM3) {
@@ -1232,7 +1438,7 @@ bool ARMLoadStoreOpt::FixInvalidRegPairOp(MachineBasicBlock &MBB,
                   getKillRegState(OddDeadKill)  | getUndefRegState(OddUndef));
         ++NumSTRD2STM;
       }
-      NewBBI = llvm::prior(MBBI);
+      NewBBI = std::prev(MBBI);
     } else {
       // Split into two instructions.
       unsigned NewOpc = (isLd)
@@ -1254,7 +1460,7 @@ bool ARMLoadStoreOpt::FixInvalidRegPairOp(MachineBasicBlock &MBB,
                       OddReg, OddDeadKill, false,
                       BaseReg, false, BaseUndef, false, OffUndef,
                       Pred, PredReg, TII, isT2);
-        NewBBI = llvm::prior(MBBI);
+        NewBBI = std::prev(MBBI);
         InsertLDR_STR(MBB, MBBI, OffImm, isLd, dl, NewOpc,
                       EvenReg, EvenDeadKill, false,
                       BaseReg, BaseKill, BaseUndef, OffKill, OffUndef,
@@ -1274,7 +1480,7 @@ bool ARMLoadStoreOpt::FixInvalidRegPairOp(MachineBasicBlock &MBB,
                       EvenReg, EvenDeadKill, EvenUndef,
                       BaseReg, false, BaseUndef, false, OffUndef,
                       Pred, PredReg, TII, isT2);
-        NewBBI = llvm::prior(MBBI);
+        NewBBI = std::prev(MBBI);
         InsertLDR_STR(MBB, MBBI, OffImm+4, isLd, dl, NewOpc2,
                       OddReg, OddDeadKill, OddUndef,
                       BaseReg, BaseKill, BaseUndef, OffKill, OffUndef,
@@ -1408,18 +1614,22 @@ bool ARMLoadStoreOpt::LoadStoreMultipleOpti(MachineBasicBlock &MBB) {
       if (MBBI == E)
         // Reach the end of the block, try merging the memory instructions.
         TryMerge = true;
-    } else
+    } else {
       TryMerge = true;
+    }
 
     if (TryMerge) {
       if (NumMemOps > 1) {
         // Try to find a free register to use as a new base in case it's needed.
         // First advance to the instruction just before the start of the chain.
         AdvanceRS(MBB, MemOps);
+
         // Find a scratch register.
-        unsigned Scratch = RS->FindUnusedReg(&ARM::GPRRegClass);
+        unsigned Scratch =
+          RS->FindUnusedReg(isThumb1 ? &ARM::tGPRRegClass : &ARM::GPRRegClass);
+
         // Process the load / store instructions.
-        RS->forward(prior(MBBI));
+        RS->forward(std::prev(MBBI));
 
         // Merge ops.
         Merges.clear();
@@ -1441,13 +1651,13 @@ bool ARMLoadStoreOpt::LoadStoreMultipleOpti(MachineBasicBlock &MBB) {
               ++NumMerges;
 
         // RS may be pointing to an instruction that's deleted.
-        RS->skipTo(prior(MBBI));
+        RS->skipTo(std::prev(MBBI));
       } else if (NumMemOps == 1) {
         // Try folding preceding/trailing base inc/dec into the single
         // load/store.
         if (MergeBaseUpdateLoadStore(MBB, MemOps[0].MBBI, TII, Advance, MBBI)) {
           ++NumMerges;
-          RS->forward(prior(MBBI));
+          RS->forward(std::prev(MBBI));
         }
       }
 
@@ -1483,6 +1693,8 @@ bool ARMLoadStoreOpt::LoadStoreMultipleOpti(MachineBasicBlock &MBB) {
 /// =>
 ///   ldmfd sp!, {..., pc}
 bool ARMLoadStoreOpt::MergeReturnIntoLDM(MachineBasicBlock &MBB) {
+  // Thumb1 LDM doesn't allow high registers.
+  if (isThumb1) return false;
   if (MBB.empty()) return false;
 
   MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
@@ -1490,7 +1702,7 @@ bool ARMLoadStoreOpt::MergeReturnIntoLDM(MachineBasicBlock &MBB) {
       (MBBI->getOpcode() == ARM::BX_RET ||
        MBBI->getOpcode() == ARM::tBX_RET ||
        MBBI->getOpcode() == ARM::MOVPCLR)) {
-    MachineInstr *PrevMI = prior(MBBI);
+    MachineInstr *PrevMI = std::prev(MBBI);
     unsigned Opcode = PrevMI->getOpcode();
     if (Opcode == ARM::LDMIA_UPD || Opcode == ARM::LDMDA_UPD ||
         Opcode == ARM::LDMDB_UPD || Opcode == ARM::LDMIB_UPD ||
@@ -1513,12 +1725,20 @@ bool ARMLoadStoreOpt::MergeReturnIntoLDM(MachineBasicBlock &MBB) {
 
 bool ARMLoadStoreOpt::runOnMachineFunction(MachineFunction &Fn) {
   const TargetMachine &TM = Fn.getTarget();
+  TL = TM.getTargetLowering();
   AFI = Fn.getInfo<ARMFunctionInfo>();
   TII = TM.getInstrInfo();
   TRI = TM.getRegisterInfo();
   STI = &TM.getSubtarget<ARMSubtarget>();
   RS = new RegScavenger();
   isThumb2 = AFI->isThumb2Function();
+  isThumb1 = AFI->isThumbFunction() && !isThumb2;
+
+  // FIXME: Temporarily disabling for Thumb-1 due to miscompiles
+  if (isThumb1) {
+    delete RS;
+    return false;
+  }
 
   bool Modified = false;
   for (MachineFunction::iterator MFI = Fn.begin(), E = Fn.end(); MFI != E;
@@ -1550,9 +1770,9 @@ namespace {
     MachineRegisterInfo *MRI;
     MachineFunction *MF;
 
-    virtual bool runOnMachineFunction(MachineFunction &Fn);
+    bool runOnMachineFunction(MachineFunction &Fn) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM pre- register allocation load / store optimization pass";
     }
 
@@ -1666,11 +1886,11 @@ ARMPreAllocLoadStoreOpt::CanFormLdStDWord(MachineInstr *Op0, MachineInstr *Op1,
   // FIXME: VLDRS / VSTRS -> VLDRD / VSTRD
   unsigned Scale = 1;
   unsigned Opcode = Op0->getOpcode();
-  if (Opcode == ARM::LDRi12)
+  if (Opcode == ARM::LDRi12) {
     NewOpc = ARM::LDRD;
-  else if (Opcode == ARM::STRi12)
+  } else if (Opcode == ARM::STRi12) {
     NewOpc = ARM::STRD;
-  else if (Opcode == ARM::t2LDRi8 || Opcode == ARM::t2LDRi12) {
+  } else if (Opcode == ARM::t2LDRi8 || Opcode == ARM::t2LDRi12) {
     NewOpc = ARM::t2LDRDi8;
     Scale = 4;
     isT2 = true;
@@ -1678,8 +1898,9 @@ ARMPreAllocLoadStoreOpt::CanFormLdStDWord(MachineInstr *Op0, MachineInstr *Op1,
     NewOpc = ARM::t2STRDi8;
     Scale = 4;
     isT2 = true;
-  } else
+  } else {
     return false;
+  }
 
   // Make sure the base address satisfies i64 ld / st alignment requirement.
   // At the moment, we ignore the memoryoperand's value.
@@ -1724,17 +1945,6 @@ ARMPreAllocLoadStoreOpt::CanFormLdStDWord(MachineInstr *Op0, MachineInstr *Op1,
   return true;
 }
 
-namespace {
-  struct OffsetCompare {
-    bool operator()(const MachineInstr *LHS, const MachineInstr *RHS) const {
-      int LOffset = getMemoryOpOffset(LHS);
-      int ROffset = getMemoryOpOffset(RHS);
-      assert(LHS == RHS || LOffset != ROffset);
-      return LOffset > ROffset;
-    }
-  };
-}
-
 bool ARMPreAllocLoadStoreOpt::RescheduleOps(MachineBasicBlock *MBB,
                                  SmallVectorImpl<MachineInstr *> &Ops,
                                  unsigned Base, bool isLd,
@@ -1742,7 +1952,13 @@ bool ARMPreAllocLoadStoreOpt::RescheduleOps(MachineBasicBlock *MBB,
   bool RetVal = false;
 
   // Sort by offset (in reverse order).
-  std::sort(Ops.begin(), Ops.end(), OffsetCompare());
+  std::sort(Ops.begin(), Ops.end(),
+            [](const MachineInstr *LHS, const MachineInstr *RHS) {
+    int LOffset = getMemoryOpOffset(LHS);
+    int ROffset = getMemoryOpOffset(RHS);
+    assert(LHS == RHS || LOffset != ROffset);
+    return LOffset > ROffset;
+  });
 
   // The loads / stores of the same base are in order. Scan them from first to
   // last and check for the following:
@@ -1751,8 +1967,8 @@ bool ARMPreAllocLoadStoreOpt::RescheduleOps(MachineBasicBlock *MBB,
   while (Ops.size() > 1) {
     unsigned FirstLoc = ~0U;
     unsigned LastLoc = 0;
-    MachineInstr *FirstOp = 0;
-    MachineInstr *LastOp = 0;
+    MachineInstr *FirstOp = nullptr;
+    MachineInstr *LastOp = nullptr;
     int LastOffset = 0;
     unsigned LastOpcode = 0;
     unsigned LastBytes = 0;
diff --git a/contrib/llvm/lib/Target/ARM/ARMMCInstLower.cpp b/contrib/llvm/lib/Target/ARM/ARMMCInstLower.cpp
index e12c9c61ab14..023f5f8e37a0 100644
--- a/contrib/llvm/lib/Target/ARM/ARMMCInstLower.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMMCInstLower.cpp
@@ -14,25 +14,27 @@
 
 #include "ARM.h"
 #include "ARMAsmPrinter.h"
+#include "MCTargetDesc/ARMBaseInfo.h"
 #include "MCTargetDesc/ARMMCExpr.h"
 #include "llvm/CodeGen/MachineBasicBlock.h"
 #include "llvm/IR/Constants.h"
+#include "llvm/IR/Mangler.h"
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCInst.h"
-#include "llvm/Target/Mangler.h"
 using namespace llvm;
 
 
 MCOperand ARMAsmPrinter::GetSymbolRef(const MachineOperand &MO,
                                       const MCSymbol *Symbol) {
   const MCExpr *Expr;
-  switch (MO.getTargetFlags()) {
+  unsigned Option = MO.getTargetFlags() & ARMII::MO_OPTION_MASK;
+  switch (Option) {
   default: {
     Expr = MCSymbolRefExpr::Create(Symbol, MCSymbolRefExpr::VK_None,
                                    OutContext);
-    switch (MO.getTargetFlags()) {
+    switch (Option) {
     default: llvm_unreachable("Unknown target flag on symbol operand");
-    case 0:
+    case ARMII::MO_NO_FLAG:
       break;
     case ARMII::MO_LO16:
       Expr = MCSymbolRefExpr::Create(Symbol, MCSymbolRefExpr::VK_None,
@@ -49,7 +51,7 @@ MCOperand ARMAsmPrinter::GetSymbolRef(const MachineOperand &MO,
   }
 
   case ARMII::MO_PLT:
-    Expr = MCSymbolRefExpr::Create(Symbol, MCSymbolRefExpr::VK_ARM_PLT,
+    Expr = MCSymbolRefExpr::Create(Symbol, MCSymbolRefExpr::VK_PLT,
                                    OutContext);
     break;
   }
@@ -81,9 +83,11 @@ bool ARMAsmPrinter::lowerOperand(const MachineOperand &MO,
     MCOp = MCOperand::CreateExpr(MCSymbolRefExpr::Create(
         MO.getMBB()->getSymbol(), OutContext));
     break;
-  case MachineOperand::MO_GlobalAddress:
-    MCOp = GetSymbolRef(MO, getSymbol(MO.getGlobal()));
+  case MachineOperand::MO_GlobalAddress: {
+    MCOp = GetSymbolRef(MO,
+                        GetARMGVSymbol(MO.getGlobal(), MO.getTargetFlags()));
     break;
+  }
   case MachineOperand::MO_ExternalSymbol:
    MCOp = GetSymbolRef(MO,
                         GetExternalSymbolSymbol(MO.getSymbolName()));
diff --git a/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.cpp
index af445e2f35aa..892b269fc181 100644
--- a/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.cpp
@@ -12,3 +12,13 @@
 using namespace llvm;
 
 void ARMFunctionInfo::anchor() { }
+
+ARMFunctionInfo::ARMFunctionInfo(MachineFunction &MF)
+    : isThumb(MF.getTarget().getSubtarget<ARMSubtarget>().isThumb()),
+      hasThumb2(MF.getTarget().getSubtarget<ARMSubtarget>().hasThumb2()),
+      StByValParamsPadding(0), ArgRegsSaveSize(0), HasStackFrame(false),
+      RestoreSPFromFP(false), LRSpilledForFarJump(false),
+      FramePtrSpillOffset(0), GPRCS1Offset(0), GPRCS2Offset(0), DPRCSOffset(0),
+      GPRCS1Size(0), GPRCS2Size(0), DPRCSSize(0), JumpTableUId(0),
+      PICLabelUId(0), VarArgsFrameIndex(0), HasITBlocks(false),
+      GlobalBaseReg(0) {}
diff --git a/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.h b/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.h
index 010edf33cea4..d3fabc3ebb04 100644
--- a/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMMachineFunctionInfo.h
@@ -19,6 +19,7 @@
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetRegisterInfo.h"
+#include "llvm/ADT/DenseMap.h"
 
 namespace llvm {
 
@@ -38,7 +39,7 @@ class ARMFunctionInfo : public MachineFunctionInfo {
 
   /// StByValParamsPadding - For parameter that is split between
   /// GPRs and memory; while recovering GPRs part, when
-  /// StackAlignment == 8, and GPRs-part-size mod 8 != 0,
+  /// StackAlignment > 4, and GPRs-part-size mod StackAlignment != 0,
   /// we need to insert gap before parameter start address. It allows to
   /// "attach" GPR-part to the part that was passed via stack.
   unsigned StByValParamsPadding;
@@ -114,6 +115,14 @@ class ARMFunctionInfo : public MachineFunctionInfo {
   /// relocation models.
   unsigned GlobalBaseReg;
 
+  /// ArgumentStackSize - amount of bytes on stack consumed by the arguments
+  /// being passed on the stack
+  unsigned ArgumentStackSize;
+
+  /// CoalescedWeights - mapping of basic blocks to the rolling counter of
+  /// coalesced weights.
+  DenseMap<const MachineBasicBlock*, unsigned> CoalescedWeights;
+
 public:
   ARMFunctionInfo() :
     isThumb(false),
@@ -126,16 +135,7 @@ public:
     JumpTableUId(0), PICLabelUId(0),
     VarArgsFrameIndex(0), HasITBlocks(false), GlobalBaseReg(0) {}
 
-  explicit ARMFunctionInfo(MachineFunction &MF) :
-    isThumb(MF.getTarget().getSubtarget<ARMSubtarget>().isThumb()),
-    hasThumb2(MF.getTarget().getSubtarget<ARMSubtarget>().hasThumb2()),
-    StByValParamsPadding(0),
-    ArgRegsSaveSize(0), HasStackFrame(false), RestoreSPFromFP(false),
-    LRSpilledForFarJump(false),
-    FramePtrSpillOffset(0), GPRCS1Offset(0), GPRCS2Offset(0), DPRCSOffset(0),
-    GPRCS1Size(0), GPRCS2Size(0), DPRCSSize(0),
-    JumpTableUId(0), PICLabelUId(0),
-    VarArgsFrameIndex(0), HasITBlocks(false), GlobalBaseReg(0) {}
+  explicit ARMFunctionInfo(MachineFunction &MF);
 
   bool isThumbFunction() const { return isThumb; }
   bool isThumb1OnlyFunction() const { return isThumb && !hasThumb2; }
@@ -182,6 +182,9 @@ public:
   void setGPRCalleeSavedArea2Size(unsigned s) { GPRCS2Size = s; }
   void setDPRCalleeSavedAreaSize(unsigned s)  { DPRCSSize = s; }
 
+  unsigned getArgumentStackSize() const { return ArgumentStackSize; }
+  void setArgumentStackSize(unsigned size) { ArgumentStackSize = size; }
+
   unsigned createJumpTableUId() {
     return JumpTableUId++;
   }
@@ -213,7 +216,7 @@ public:
 
   void recordCPEClone(unsigned CPIdx, unsigned CPCloneIdx) {
     if (!CPEClones.insert(std::make_pair(CPCloneIdx, CPIdx)).second)
-      assert(0 && "Duplicate entries!");
+      llvm_unreachable("Duplicate entries!");
   }
 
   unsigned getOriginalCPIdx(unsigned CloneIdx) const {
@@ -223,6 +226,15 @@ public:
     else
       return -1U;
   }
+
+  DenseMap<const MachineBasicBlock*, unsigned>::iterator getCoalescedWeight(
+                                                  MachineBasicBlock* MBB) {
+    auto It = CoalescedWeights.find(MBB);
+    if (It == CoalescedWeights.end()) {
+      It = CoalescedWeights.insert(std::make_pair(MBB, 0)).first;
+    }
+    return It;
+  }
 };
 } // End llvm namespace
 
diff --git a/contrib/llvm/lib/Target/ARM/ARMOptimizeBarriersPass.cpp b/contrib/llvm/lib/Target/ARM/ARMOptimizeBarriersPass.cpp
new file mode 100644
index 000000000000..2a4925572ae8
--- /dev/null
+++ b/contrib/llvm/lib/Target/ARM/ARMOptimizeBarriersPass.cpp
@@ -0,0 +1,101 @@
+//===-- ARMOptimizeBarriersPass - two DMBs without a memory access in between,
+//removed one -===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===------------------------------------------------------------------------------------------===//
+
+#include "ARM.h"
+#include "ARMMachineFunctionInfo.h"
+#include "ARMInstrInfo.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
+using namespace llvm;
+
+#define DEBUG_TYPE "double barriers"
+
+STATISTIC(NumDMBsRemoved, "Number of DMBs removed");
+
+namespace {
+class ARMOptimizeBarriersPass : public MachineFunctionPass {
+public:
+  static char ID;
+  ARMOptimizeBarriersPass() : MachineFunctionPass(ID) {}
+
+  bool runOnMachineFunction(MachineFunction &Fn) override;
+
+  const char *getPassName() const override {
+    return "optimise barriers pass";
+  }
+
+private:
+};
+char ARMOptimizeBarriersPass::ID = 0;
+}
+
+// Returns whether the instruction can safely move past a DMB instruction
+// The current implementation allows this iif MI does not have any possible
+// memory access
+static bool CanMovePastDMB(const MachineInstr *MI) {
+  return !(MI->mayLoad() ||
+          MI->mayStore() ||
+          MI->hasUnmodeledSideEffects() ||
+          MI->isCall() ||
+          MI->isReturn());
+}
+
+bool ARMOptimizeBarriersPass::runOnMachineFunction(MachineFunction &MF) {
+  // Vector to store the DMBs we will remove after the first iteration
+  std::vector<MachineInstr *> ToRemove;
+  // DMBType is the Imm value of the first operand. It determines whether it's a
+  // DMB ish, dmb sy, dmb osh, etc
+  int64_t DMBType = -1;
+
+  // Find a dmb. If we can move it until the next dmb, tag the second one for
+  // removal
+  for (auto &MBB : MF) {
+    // Will be true when we have seen a DMB, and not seen any instruction since
+    // that cannot move past a DMB
+    bool IsRemovableNextDMB = false;
+    for (auto &MI : MBB) {
+      if (MI.getOpcode() == ARM::DMB) {
+        if (IsRemovableNextDMB) {
+          // If the Imm of this DMB is the same as that of the last DMB, we can
+          // tag this second DMB for removal
+          if (MI.getOperand(0).getImm() == DMBType) {
+            ToRemove.push_back(&MI);
+          } else {
+            // If it has a different DMBType, we cannot remove it, but will scan
+            // for the next DMB, recording this DMB's type as last seen DMB type
+            DMBType = MI.getOperand(0).getImm();
+          }
+        } else {
+          // After we see a DMB, a next one is removable
+          IsRemovableNextDMB = true;
+          DMBType = MI.getOperand(0).getImm();
+        }
+      } else if (!CanMovePastDMB(&MI)) {
+        // If we find an instruction unable to pass past a DMB, a next DMB is
+        // not removable
+        IsRemovableNextDMB = false;
+      }
+    }
+  }
+  // Remove the tagged DMB
+  for (auto MI : ToRemove) {
+    MI->eraseFromParent();
+    ++NumDMBsRemoved;
+  }
+
+  return NumDMBsRemoved > 0;
+}
+
+/// createARMOptimizeBarriersPass - Returns an instance of the remove double
+/// barriers
+/// pass.
+FunctionPass *llvm::createARMOptimizeBarriersPass() {
+  return new ARMOptimizeBarriersPass();
+}
diff --git a/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.cpp
index a7880364d89a..80b4b4822ce8 100644
--- a/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.cpp
@@ -12,8 +12,6 @@
 //===----------------------------------------------------------------------===//
 
 #include "ARMRegisterInfo.h"
-#include "ARM.h"
-#include "ARMBaseInstrInfo.h"
 using namespace llvm;
 
 void ARMRegisterInfo::anchor() { }
diff --git a/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.h b/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.h
index fb1537cf9425..3e6af3f04698 100644
--- a/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.h
@@ -14,9 +14,7 @@
 #ifndef ARMREGISTERINFO_H
 #define ARMREGISTERINFO_H
 
-#include "ARM.h"
 #include "ARMBaseRegisterInfo.h"
-#include "llvm/Target/TargetRegisterInfo.h"
 
 namespace llvm {
 
diff --git a/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.td b/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.td
index d0457618ef6f..b290e7f6679c 100644
--- a/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.td
+++ b/contrib/llvm/lib/Target/ARM/ARMRegisterInfo.td
@@ -116,13 +116,13 @@ def D15 : ARMReg<15, "d15", [S30, S31]>, DwarfRegNum<[271]>;
 }
 
 // VFP3 defines 16 additional double registers
-def D16 : ARMFReg<16, "d16">, DwarfRegNum<[272]>; 
+def D16 : ARMFReg<16, "d16">, DwarfRegNum<[272]>;
 def D17 : ARMFReg<17, "d17">, DwarfRegNum<[273]>;
 def D18 : ARMFReg<18, "d18">, DwarfRegNum<[274]>;
 def D19 : ARMFReg<19, "d19">, DwarfRegNum<[275]>;
 def D20 : ARMFReg<20, "d20">, DwarfRegNum<[276]>;
 def D21 : ARMFReg<21, "d21">, DwarfRegNum<[277]>;
-def D22 : ARMFReg<22, "d22">, DwarfRegNum<[278]>; 
+def D22 : ARMFReg<22, "d22">, DwarfRegNum<[278]>;
 def D23 : ARMFReg<23, "d23">, DwarfRegNum<[279]>;
 def D24 : ARMFReg<24, "d24">, DwarfRegNum<[280]>;
 def D25 : ARMFReg<25, "d25">, DwarfRegNum<[281]>;
@@ -158,11 +158,11 @@ def Q15 : ARMReg<15, "q15", [D30, D31]>;
 // Current Program Status Register.
 // We model fpscr with two registers: FPSCR models the control bits and will be
 // reserved. FPSCR_NZCV models the flag bits and will be unreserved. APSR_NZCV
-// models the APSR when it's accessed by some special instructions. In such cases 
+// models the APSR when it's accessed by some special instructions. In such cases
 // it has the same encoding as PC.
 def CPSR       : ARMReg<0,  "cpsr">;
 def APSR       : ARMReg<1,  "apsr">;
-def APSR_NZCV  : ARMReg<15, "apsr_nzcv">; 
+def APSR_NZCV  : ARMReg<15, "apsr_nzcv">;
 def SPSR       : ARMReg<2,  "spsr">;
 def FPSCR      : ARMReg<3,  "fpscr">;
 def FPSCR_NZCV : ARMReg<3,  "fpscr_nzcv"> {
@@ -214,7 +214,7 @@ def GPRnopc : RegisterClass<"ARM", [i32], 32, (sub GPR, PC)> {
 }
 
 // GPRs without the PC but with APSR. Some instructions allow accessing the
-// APSR, while actually encoding PC in the register field. This is usefull
+// APSR, while actually encoding PC in the register field. This is useful
 // for assembly and disassembly only.
 def GPRwithAPSR : RegisterClass<"ARM", [i32], 32, (add (sub GPR, PC), APSR_NZCV)> {
   let AltOrders = [(add LR, GPRnopc), (trunc GPRnopc, 8)];
diff --git a/contrib/llvm/lib/Target/ARM/ARMScheduleA9.td b/contrib/llvm/lib/Target/ARM/ARMScheduleA9.td
index 603e775d351d..9a1d22275646 100644
--- a/contrib/llvm/lib/Target/ARM/ARMScheduleA9.td
+++ b/contrib/llvm/lib/Target/ARM/ARMScheduleA9.td
@@ -1894,16 +1894,26 @@ def CortexA9Model : SchedMachineModel {
   let MispredictPenalty = 8; // Based on estimate of pipeline depth.
 
   let Itineraries = CortexA9Itineraries;
+
+  // FIXME: Many vector operations were never given an itinerary. We
+  // haven't mapped these to the new model either.
+  let CompleteModel = 0;
 }
 
 //===----------------------------------------------------------------------===//
 // Define each kind of processor resource and number available.
+//
+// The AGU unit has BufferSize=1 so that the latency between operations
+// that use it are considered to stall other operations.
+//
+// The FP unit has BufferSize=0 so that it is a hard dispatch
+// hazard. No instruction may be dispatched while the unit is reserved.
 
 let SchedModel = CortexA9Model in {
 
 def A9UnitALU : ProcResource<2>;
 def A9UnitMul : ProcResource<1> { let Super = A9UnitALU; }
-def A9UnitAGU : ProcResource<1>;
+def A9UnitAGU : ProcResource<1> { let BufferSize = 1; }
 def A9UnitLS  : ProcResource<1>;
 def A9UnitFP  : ProcResource<1> { let BufferSize = 0; }
 def A9UnitB   : ProcResource<1>;
@@ -2217,7 +2227,7 @@ def A9WriteLMfp : SchedWriteVariant<[
   SchedVar<A9PostRA, [A9WriteLMfpPostRA]>]>;
 
 //===----------------------------------------------------------------------===//
-// Resources for other (non LDM/VLDM) Variants.
+// Resources for other (non-LDM/VLDM) Variants.
 
 // These mov immediate writers are unconditionally expanded with
 // additive latency.
@@ -2397,6 +2407,7 @@ def :ItinRW<[A9WriteV3, A9Read2], [IIC_VSUBiD, IIC_VSUBiQ, IIC_VCNTiD]>;
 // ...
 // VHADD/VRHADD/VQADD/VTST/VADH/VRADH
 def :ItinRW<[A9WriteV4, A9Read2, A9Read2], [IIC_VBINi4D, IIC_VBINi4Q]>;
+
 // VSBH/VRSBH/VHSUB/VQSUB/VABD/VCEQ/VCGE/VCGT/VMAX/VMIN/VPMAX/VPMIN/VABDL
 def :ItinRW<[A9WriteV4, A9Read2], [IIC_VSUBi4D, IIC_VSUBi4Q]>;
 // VQNEG/VQABS
@@ -2431,7 +2442,7 @@ def :ItinRW<[A9WriteV3], [IIC_VSHLiD, IIC_VSHLiQ]>;
 def :ItinRW<[A9WriteV4], [IIC_VSHLi4D, IIC_VSHLi4Q]>;
 
 // NEON permute
-def :ItinRW<[A9WriteV2], [IIC_VPERMD, IIC_VPERMQ, IIC_VEXTD]>;
+def :ItinRW<[A9WriteV2, A9WriteV2], [IIC_VPERMD, IIC_VPERMQ, IIC_VEXTD]>;
 def :ItinRW<[A9WriteV3, A9WriteV4, ReadDefault, A9Read2],
             [IIC_VPERMQ3, IIC_VEXTQ]>;
 def :ItinRW<[A9WriteV3, A9Read2], [IIC_VTB1]>;
diff --git a/contrib/llvm/lib/Target/ARM/ARMScheduleSwift.td b/contrib/llvm/lib/Target/ARM/ARMScheduleSwift.td
index 8d7dbc246095..b03d5ff44c6e 100644
--- a/contrib/llvm/lib/Target/ARM/ARMScheduleSwift.td
+++ b/contrib/llvm/lib/Target/ARM/ARMScheduleSwift.td
@@ -1721,7 +1721,7 @@ let SchedModel = SwiftModel in {
     SchedVar<SwiftLMAddr3Pred, [SwiftWriteLM9Cy, SwiftWriteLM10Cy,
                                 SwiftWriteLM13CyNo, SwiftWriteP01OneCycle,
                                 SwiftVLDMPerm3]>,
-    // Load of a Q register (not neccessarily true). We should not be mapping to
+    // Load of a Q register (not necessarily true). We should not be mapping to
     // 4 S registers, either.
     SchedVar<SwiftLMAddr4Pred, [SwiftWriteLM4Cy, SwiftWriteLM4CyNo,
                                 SwiftWriteLM4CyNo, SwiftWriteLM4CyNo]>,
@@ -1858,7 +1858,7 @@ let SchedModel = SwiftModel in {
     // Assume 5 D registers.
     SchedVar<SwiftLMAddr10Pred, [SwiftWriteSTM6]>,
     SchedVar<SwiftLMAddr11Pred, [SwiftWriteSTM12]>,
-    // Asume three Q registers.
+    // Assume three Q registers.
     SchedVar<SwiftLMAddr12Pred, [SwiftWriteSTM4]>,
     SchedVar<SwiftLMAddr13Pred, [SwiftWriteSTM14]>,
     // Assume 7 D registers.
diff --git a/contrib/llvm/lib/Target/ARM/ARMScheduleV6.td b/contrib/llvm/lib/Target/ARM/ARMScheduleV6.td
index 0ace9bc1796d..57d0bfb65049 100644
--- a/contrib/llvm/lib/Target/ARM/ARMScheduleV6.td
+++ b/contrib/llvm/lib/Target/ARM/ARMScheduleV6.td
@@ -93,7 +93,7 @@ def ARMV6Itineraries : ProcessorItineraries<
   InstrItinData<IIC_iMAC32   , [InstrStage<2, [V6_Pipe]>], [5, 1, 1, 2]>,
   InstrItinData<IIC_iMUL64   , [InstrStage<3, [V6_Pipe]>], [6, 1, 1]>,
   InstrItinData<IIC_iMAC64   , [InstrStage<3, [V6_Pipe]>], [6, 1, 1, 2]>,
-  
+
   // Integer load pipeline
   //
   // Immediate offset
@@ -181,7 +181,7 @@ def ARMV6Itineraries : ProcessorItineraries<
   //
   // Store multiple + update
   InstrItinData<IIC_iStore_mu , [InstrStage<3, [V6_Pipe]>], [2]>,
-  
+
   // Branch
   //
   // no delay slots, so the latency of a branch is unimportant
diff --git a/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.cpp
index 93add6ee33cf..3dcc0df349d2 100644
--- a/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.cpp
@@ -11,16 +11,15 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-selectiondag-info"
 #include "ARMTargetMachine.h"
 #include "llvm/CodeGen/SelectionDAG.h"
 #include "llvm/IR/DerivedTypes.h"
 using namespace llvm;
 
-ARMSelectionDAGInfo::ARMSelectionDAGInfo(const TargetMachine &TM)
-  : TargetSelectionDAGInfo(TM),
-    Subtarget(&TM.getSubtarget<ARMSubtarget>()) {
-}
+#define DEBUG_TYPE "arm-selectiondag-info"
+
+ARMSelectionDAGInfo::ARMSelectionDAGInfo(const DataLayout &DL)
+    : TargetSelectionDAGInfo(&DL) {}
 
 ARMSelectionDAGInfo::~ARMSelectionDAGInfo() {
 }
@@ -33,6 +32,7 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
                                              bool isVolatile, bool AlwaysInline,
                                              MachinePointerInfo DstPtrInfo,
                                           MachinePointerInfo SrcPtrInfo) const {
+  const ARMSubtarget &Subtarget = DAG.getTarget().getSubtarget<ARMSubtarget>();
   // Do repeated 4-byte loads and stores. To be improved.
   // This requires 4-byte alignment.
   if ((Align & 3) != 0)
@@ -43,7 +43,7 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
   if (!ConstantSize)
     return SDValue();
   uint64_t SizeVal = ConstantSize->getZExtValue();
-  if (!AlwaysInline && SizeVal > Subtarget->getMaxInlineSizeThreshold())
+  if (!AlwaysInline && SizeVal > Subtarget.getMaxInlineSizeThreshold())
     return SDValue();
 
   unsigned BytesLeft = SizeVal & 3;
@@ -52,9 +52,10 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
   EVT VT = MVT::i32;
   unsigned VTSize = 4;
   unsigned i = 0;
-  const unsigned MAX_LOADS_IN_LDM = 6;
-  SDValue TFOps[MAX_LOADS_IN_LDM];
-  SDValue Loads[MAX_LOADS_IN_LDM];
+  // Emit a maximum of 4 loads in Thumb1 since we have fewer registers
+  const unsigned MAX_LOADS_IN_LDM = Subtarget.isThumb1Only() ? 4 : 6;
+  SDValue TFOps[6];
+  SDValue Loads[6];
   uint64_t SrcOff = 0, DstOff = 0;
 
   // Emit up to MAX_LOADS_IN_LDM loads, then a TokenFactor barrier, then the
@@ -71,7 +72,8 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
       TFOps[i] = Loads[i].getValue(1);
       SrcOff += VTSize;
     }
-    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &TFOps[0], i);
+    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
+                        makeArrayRef(TFOps, i));
 
     for (i = 0;
          i < MAX_LOADS_IN_LDM && EmittedNumMemOps + i < NumMemOps; ++i) {
@@ -82,7 +84,8 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
                               isVolatile, false, 0);
       DstOff += VTSize;
     }
-    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &TFOps[0], i);
+    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
+                        makeArrayRef(TFOps, i));
 
     EmittedNumMemOps += i;
   }
@@ -112,7 +115,8 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
     SrcOff += VTSize;
     BytesLeft -= VTSize;
   }
-  Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &TFOps[0], i);
+  Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
+                      makeArrayRef(TFOps, i));
 
   i = 0;
   BytesLeft = BytesLeftSave;
@@ -133,7 +137,8 @@ ARMSelectionDAGInfo::EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
     DstOff += VTSize;
     BytesLeft -= VTSize;
   }
-  return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &TFOps[0], i);
+  return DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
+                     makeArrayRef(TFOps, i));
 }
 
 // Adjust parameters for memset, EABI uses format (ptr, size, value),
@@ -145,8 +150,10 @@ EmitTargetCodeForMemset(SelectionDAG &DAG, SDLoc dl,
                         SDValue Src, SDValue Size,
                         unsigned Align, bool isVolatile,
                         MachinePointerInfo DstPtrInfo) const {
-  // Use default for non AAPCS (or Darwin) subtargets
-  if (!Subtarget->isAAPCS_ABI() || Subtarget->isTargetDarwin())
+  const ARMSubtarget &Subtarget = DAG.getTarget().getSubtarget<ARMSubtarget>();
+  // Use default for non-AAPCS (or MachO) subtargets
+  if (!Subtarget.isAAPCS_ABI() || Subtarget.isTargetMachO() ||
+      Subtarget.isTargetWindows())
     return SDValue();
 
   const ARMTargetLowering &TLI =
@@ -179,22 +186,14 @@ EmitTargetCodeForMemset(SelectionDAG &DAG, SDLoc dl,
   Args.push_back(Entry);
 
   // Emit __eabi_memset call
-  TargetLowering::CallLoweringInfo CLI(Chain,
-                    Type::getVoidTy(*DAG.getContext()), // return type
-                    false, // return sign ext
-                    false, // return zero ext
-                    false, // is var arg
-                    false, // is in regs
-                    0,     // number of fixed arguments
-                    TLI.getLibcallCallingConv(RTLIB::MEMSET), // call conv
-                    false, // is tail call
-                    false, // does not return
-                    false, // is return val used
-                    DAG.getExternalSymbol(TLI.getLibcallName(RTLIB::MEMSET),
-                                          TLI.getPointerTy()), // callee
-                    Args, DAG, dl);
-  std::pair<SDValue,SDValue> CallResult =
-    TLI.LowerCallTo(CLI);
-
+  TargetLowering::CallLoweringInfo CLI(DAG);
+  CLI.setDebugLoc(dl).setChain(Chain)
+    .setCallee(TLI.getLibcallCallingConv(RTLIB::MEMSET),
+               Type::getVoidTy(*DAG.getContext()),
+               DAG.getExternalSymbol(TLI.getLibcallName(RTLIB::MEMSET),
+                                     TLI.getPointerTy()), std::move(Args), 0)
+    .setDiscardResult();
+
+  std::pair<SDValue,SDValue> CallResult = TLI.LowerCallTo(CLI);
   return CallResult.second;
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.h b/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.h
index 56c9375855dd..13769dc8efe0 100644
--- a/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.h
+++ b/contrib/llvm/lib/Target/ARM/ARMSelectionDAGInfo.h
@@ -36,31 +36,25 @@ namespace ARM_AM {
 }  // end namespace ARM_AM
 
 class ARMSelectionDAGInfo : public TargetSelectionDAGInfo {
-  /// Subtarget - Keep a pointer to the ARMSubtarget around so that we can
-  /// make the right decision when generating code for different targets.
-  const ARMSubtarget *Subtarget;
-
 public:
-  explicit ARMSelectionDAGInfo(const TargetMachine &TM);
+  explicit ARMSelectionDAGInfo(const DataLayout &DL);
   ~ARMSelectionDAGInfo();
 
-  virtual
   SDValue EmitTargetCodeForMemcpy(SelectionDAG &DAG, SDLoc dl,
                                   SDValue Chain,
                                   SDValue Dst, SDValue Src,
                                   SDValue Size, unsigned Align,
                                   bool isVolatile, bool AlwaysInline,
                                   MachinePointerInfo DstPtrInfo,
-                                  MachinePointerInfo SrcPtrInfo) const;
+                                  MachinePointerInfo SrcPtrInfo) const override;
 
   // Adjust parameters for memset, see RTABI section 4.3.4
-  virtual
   SDValue EmitTargetCodeForMemset(SelectionDAG &DAG, SDLoc dl,
                                   SDValue Chain,
                                   SDValue Op1, SDValue Op2,
                                   SDValue Op3, unsigned Align,
                                   bool isVolatile,
-                                  MachinePointerInfo DstPtrInfo) const;
+                                  MachinePointerInfo DstPtrInfo) const override;
 };
 
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMSubtarget.cpp b/contrib/llvm/lib/Target/ARM/ARMSubtarget.cpp
index a11629852ad7..c1b4562f4118 100644
--- a/contrib/llvm/lib/Target/ARM/ARMSubtarget.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMSubtarget.cpp
@@ -12,21 +12,33 @@
 //===----------------------------------------------------------------------===//
 
 #include "ARMSubtarget.h"
-#include "ARMBaseInstrInfo.h"
-#include "ARMBaseRegisterInfo.h"
+#include "ARMFrameLowering.h"
+#include "ARMISelLowering.h"
+#include "ARMInstrInfo.h"
+#include "ARMJITInfo.h"
+#include "ARMSelectionDAGInfo.h"
+#include "ARMSubtarget.h"
+#include "ARMMachineFunctionInfo.h"
+#include "Thumb1FrameLowering.h"
+#include "Thumb1InstrInfo.h"
+#include "Thumb2InstrInfo.h"
 #include "llvm/IR/Attributes.h"
-#include "llvm/IR/GlobalValue.h"
 #include "llvm/IR/Function.h"
+#include "llvm/IR/GlobalValue.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetOptions.h"
+#include "llvm/Target/TargetRegisterInfo.h"
+#include "llvm/CodeGen/MachineRegisterInfo.h"
+
+using namespace llvm;
+
+#define DEBUG_TYPE "arm-subtarget"
 
 #define GET_SUBTARGETINFO_TARGET_DESC
 #define GET_SUBTARGETINFO_CTOR
 #include "ARMGenSubtargetInfo.inc"
 
-using namespace llvm;
-
 static cl::opt<bool>
 ReserveR9("arm-reserve-r9", cl::Hidden,
           cl::desc("Reserve R9, making it unavailable as GPR"));
@@ -74,20 +86,89 @@ IT(cl::desc("IT block support"), cl::Hidden, cl::init(DefaultIT),
                          "Allow IT blocks based on ARMv7"),
               clEnumValEnd));
 
-ARMSubtarget::ARMSubtarget(const std::string &TT, const std::string &CPU,
-                           const std::string &FS, const TargetOptions &Options)
-  : ARMGenSubtargetInfo(TT, CPU, FS)
-  , ARMProcFamily(Others)
-  , ARMProcClass(None)
-  , stackAlignment(4)
-  , CPUString(CPU)
-  , TargetTriple(TT)
-  , Options(Options)
-  , TargetABI(ARM_ABI_APCS) {
+static std::string computeDataLayout(ARMSubtarget &ST) {
+  std::string Ret = "";
+
+  if (ST.isLittle())
+    // Little endian.
+    Ret += "e";
+  else
+    // Big endian.
+    Ret += "E";
+
+  Ret += DataLayout::getManglingComponent(ST.getTargetTriple());
+
+  // Pointers are 32 bits and aligned to 32 bits.
+  Ret += "-p:32:32";
+
+  // On thumb, i16,i18 and i1 have natural aligment requirements, but we try to
+  // align to 32.
+  if (ST.isThumb())
+    Ret += "-i1:8:32-i8:8:32-i16:16:32";
+
+  // ABIs other than APCS have 64 bit integers with natural alignment.
+  if (!ST.isAPCS_ABI())
+    Ret += "-i64:64";
+
+  // We have 64 bits floats. The APCS ABI requires them to be aligned to 32
+  // bits, others to 64 bits. We always try to align to 64 bits.
+  if (ST.isAPCS_ABI())
+    Ret += "-f64:32:64";
+
+  // We have 128 and 64 bit vectors. The APCS ABI aligns them to 32 bits, others
+  // to 64. We always ty to give them natural alignment.
+  if (ST.isAPCS_ABI())
+    Ret += "-v64:32:64-v128:32:128";
+  else
+    Ret += "-v128:64:128";
+
+  // On thumb and APCS, only try to align aggregates to 32 bits (the default is
+  // 64 bits).
+  if (ST.isThumb() || ST.isAPCS_ABI())
+    Ret += "-a:0:32";
+
+  // Integer registers are 32 bits.
+  Ret += "-n32";
+
+  // The stack is 128 bit aligned on NaCl, 64 bit aligned on AAPCS and 32 bit
+  // aligned everywhere else.
+  if (ST.isTargetNaCl())
+    Ret += "-S128";
+  else if (ST.isAAPCS_ABI())
+    Ret += "-S64";
+  else
+    Ret += "-S32";
+
+  return Ret;
+}
+
+/// initializeSubtargetDependencies - Initializes using a CPU and feature string
+/// so that we can use initializer lists for subtarget initialization.
+ARMSubtarget &ARMSubtarget::initializeSubtargetDependencies(StringRef CPU,
+                                                            StringRef FS) {
   initializeEnvironment();
   resetSubtargetFeatures(CPU, FS);
+  return *this;
 }
 
+ARMSubtarget::ARMSubtarget(const std::string &TT, const std::string &CPU,
+                           const std::string &FS, TargetMachine &TM,
+                           bool IsLittle, const TargetOptions &Options)
+    : ARMGenSubtargetInfo(TT, CPU, FS), ARMProcFamily(Others),
+      ARMProcClass(None), stackAlignment(4), CPUString(CPU), IsLittle(IsLittle),
+      TargetTriple(TT), Options(Options), TargetABI(ARM_ABI_UNKNOWN),
+      DL(computeDataLayout(initializeSubtargetDependencies(CPU, FS))),
+      TSInfo(DL), JITInfo(),
+      InstrInfo(isThumb1Only()
+                    ? (ARMBaseInstrInfo *)new Thumb1InstrInfo(*this)
+                    : !isThumb()
+                          ? (ARMBaseInstrInfo *)new ARMInstrInfo(*this)
+                          : (ARMBaseInstrInfo *)new Thumb2InstrInfo(*this)),
+      TLInfo(TM),
+      FrameLowering(!isThumb1Only()
+                        ? new ARMFrameLowering(*this)
+                        : (ARMFrameLowering *)new Thumb1FrameLowering(*this)) {}
+
 void ARMSubtarget::initializeEnvironment() {
   HasV4TOps = false;
   HasV5TOps = false;
@@ -110,7 +191,6 @@ void ARMSubtarget::initializeEnvironment() {
   InThumbMode = false;
   HasThumb2 = false;
   NoARM = false;
-  PostRAScheduler = false;
   IsR9Reserved = ReserveR9;
   UseMovt = false;
   SupportsTailCall = false;
@@ -131,6 +211,7 @@ void ARMSubtarget::initializeEnvironment() {
   HasTrustZone = false;
   HasCrypto = false;
   HasCRC = false;
+  HasZeroCycleZeroing = false;
   AllowsUnalignedMem = false;
   Thumb2DSP = false;
   UseNaClTrap = false;
@@ -175,10 +256,9 @@ void ARMSubtarget::resetSubtargetFeatures(StringRef CPU, StringRef FS) {
   }
   ParseSubtargetFeatures(CPUString, ArchFS);
 
-  // Thumb2 implies at least V6T2. FIXME: Fix tests to explicitly specify a
-  // ARM version or CPU and then remove this.
-  if (!HasV6T2Ops && hasThumb2())
-    HasV4TOps = HasV5TOps = HasV5TEOps = HasV6Ops = HasV6MOps = HasV6T2Ops = true;
+  // FIXME: This used enable V6T2 support implicitly for Thumb2 mode.
+  // Assert this for now to make the change obvious.
+  assert(hasV6T2Ops() || !hasThumb2());
 
   // Keep a pointer to static instruction cost data for the specified CPU.
   SchedModel = getSchedModelForCPU(CPUString);
@@ -186,34 +266,54 @@ void ARMSubtarget::resetSubtargetFeatures(StringRef CPU, StringRef FS) {
   // Initialize scheduling itinerary for the specified CPU.
   InstrItins = getInstrItineraryForCPU(CPUString);
 
-  if ((TargetTriple.getTriple().find("eabi") != std::string::npos) ||
-      (isTargetIOS() && isMClass()))
-    // FIXME: We might want to separate AAPCS and EABI. Some systems, e.g.
-    // Darwin-EABI conforms to AACPS but not the rest of EABI.
+  if (TargetABI == ARM_ABI_UNKNOWN) {
+    switch (TargetTriple.getEnvironment()) {
+    case Triple::Android:
+    case Triple::EABI:
+    case Triple::EABIHF:
+    case Triple::GNUEABI:
+    case Triple::GNUEABIHF:
+      TargetABI = ARM_ABI_AAPCS;
+      break;
+    default:
+      if ((isTargetIOS() && isMClass()) ||
+          (TargetTriple.isOSBinFormatMachO() &&
+           TargetTriple.getOS() == Triple::UnknownOS))
+        TargetABI = ARM_ABI_AAPCS;
+      else
+        TargetABI = ARM_ABI_APCS;
+      break;
+    }
+  }
+
+  // FIXME: this is invalid for WindowsCE
+  if (isTargetWindows()) {
     TargetABI = ARM_ABI_AAPCS;
+    NoARM = true;
+  }
 
   if (isAAPCS_ABI())
     stackAlignment = 8;
+  if (isTargetNaCl())
+    stackAlignment = 16;
 
   UseMovt = hasV6T2Ops() && ArmUseMOVT;
 
-  if (!isTargetIOS()) {
-    IsR9Reserved = ReserveR9;
-  } else {
+  if (isTargetMachO()) {
     IsR9Reserved = ReserveR9 | !HasV6Ops;
-    SupportsTailCall = !getTargetTriple().isOSVersionLT(5, 0);
+    SupportsTailCall = !isTargetIOS() || !getTargetTriple().isOSVersionLT(5, 0);
+  } else {
+    IsR9Reserved = ReserveR9;
+    SupportsTailCall = !isThumb1Only();
   }
 
-  if (!isThumb() || hasThumb2())
-    PostRAScheduler = true;
-
   switch (Align) {
     case DefaultAlign:
       // Assume pre-ARMv6 doesn't support unaligned accesses.
       //
       // ARMv6 may or may not support unaligned accesses depending on the
       // SCTLR.U bit, which is architecture-specific. We assume ARMv6
-      // Darwin targets support unaligned accesses, and others don't.
+      // Darwin and NetBSD targets support unaligned accesses, and others don't.
       //
       // ARMv7 always has SCTLR.U set to 1, but it has a new SCTLR.A bit
       // which raises an alignment fault on unaligned accesses. Linux
@@ -222,9 +322,15 @@ void ARMSubtarget::resetSubtargetFeatures(StringRef CPU, StringRef FS) {
       // Linux targets support unaligned accesses. The same goes for NaCl.
       //
       // The above behavior is consistent with GCC.
-      AllowsUnalignedMem = (
-          (hasV7Ops() && (isTargetLinux() || isTargetNaCl())) ||
-          (hasV6Ops() && isTargetDarwin()));
+      AllowsUnalignedMem =
+          (hasV7Ops() && (isTargetLinux() || isTargetNaCl() ||
+                          isTargetNetBSD())) ||
+          (hasV6Ops() && (isTargetMachO() || isTargetNetBSD()));
+      // The one exception is cortex-m0, which despite being v6, does not
+      // support unaligned accesses. Rather than make the above boolean
+      // expression even more obtuse, just override the value here.
+      if (isThumb1Only() && isMClass())
+        AllowsUnalignedMem = false;
       break;
     case StrictAlign:
       AllowsUnalignedMem = false;
@@ -266,7 +372,7 @@ ARMSubtarget::GVIsIndirectSymbol(const GlobalValue *GV,
   if (GV->isDeclaration() && !GV->isMaterializable())
     isDecl = true;
 
-  if (!isTargetDarwin()) {
+  if (!isTargetMachO()) {
     // Extra load is needed for all externally visible.
     if (GV->hasLocalLinkage() || GV->hasHiddenVisibility())
       return false;
@@ -315,10 +421,20 @@ bool ARMSubtarget::hasSinCos() const {
     !getTargetTriple().isOSVersionLT(7, 0);
 }
 
-bool ARMSubtarget::enablePostRAScheduler(
-           CodeGenOpt::Level OptLevel,
-           TargetSubtargetInfo::AntiDepBreakMode& Mode,
-           RegClassVector& CriticalPathRCs) const {
-  Mode = TargetSubtargetInfo::ANTIDEP_NONE;
-  return PostRAScheduler && OptLevel >= CodeGenOpt::Default;
+// This overrides the PostRAScheduler bit in the SchedModel for any CPU.
+bool ARMSubtarget::enablePostMachineScheduler() const {
+  return (!isThumb() || hasThumb2());
+}
+
+bool ARMSubtarget::enableAtomicExpandLoadLinked() const {
+  return hasAnyDataBarrier() && !isThumb1Only();
+}
+
+bool ARMSubtarget::useMovt(const MachineFunction &MF) const {
+  // NOTE Windows on ARM needs to use mov.w/mov.t pairs to materialise 32-bit
+  // immediates as it is inherently position independent, and may be out of
+  // range otherwise.
+  return UseMovt && (isTargetWindows() ||
+                     !MF.getFunction()->getAttributes().hasAttribute(
+                         AttributeSet::FunctionIndex, Attribute::MinSize));
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMSubtarget.h b/contrib/llvm/lib/Target/ARM/ARMSubtarget.h
index 5276901bbb92..f8283b08d485 100644
--- a/contrib/llvm/lib/Target/ARM/ARMSubtarget.h
+++ b/contrib/llvm/lib/Target/ARM/ARMSubtarget.h
@@ -14,8 +14,20 @@
 #ifndef ARMSUBTARGET_H
 #define ARMSUBTARGET_H
 
+
+#include "ARMFrameLowering.h"
+#include "ARMISelLowering.h"
+#include "ARMInstrInfo.h"
+#include "ARMJITInfo.h"
+#include "ARMSelectionDAGInfo.h"
+#include "ARMSubtarget.h"
+#include "Thumb1FrameLowering.h"
+#include "Thumb1InstrInfo.h"
+#include "Thumb2InstrInfo.h"
+#include "ARMJITInfo.h"
 #include "MCTargetDesc/ARMMCTargetDesc.h"
 #include "llvm/ADT/Triple.h"
+#include "llvm/IR/DataLayout.h"
 #include "llvm/MC/MCInstrItineraries.h"
 #include "llvm/Target/TargetSubtargetInfo.h"
 #include <string>
@@ -31,7 +43,8 @@ class TargetOptions;
 class ARMSubtarget : public ARMGenSubtargetInfo {
 protected:
   enum ARMProcFamilyEnum {
-    Others, CortexA5, CortexA8, CortexA9, CortexA15, CortexR5, Swift, CortexA53, CortexA57
+    Others, CortexA5, CortexA7, CortexA8, CortexA9, CortexA12, CortexA15,
+    CortexR5, Swift, CortexA53, CortexA57, Krait
   };
   enum ARMProcClassEnum {
     None, AClass, RClass, MClass
@@ -92,9 +105,6 @@ protected:
   /// NoARM - True if subtarget does not support ARM mode execution.
   bool NoARM;
 
-  /// PostRAScheduler - True if using post-register-allocation scheduler.
-  bool PostRAScheduler;
-
   /// IsR9Reserved - True if R9 is a not available as general purpose register.
   bool IsR9Reserved;
 
@@ -172,6 +182,10 @@ protected:
   /// HasCRC - if true, processor supports CRC instructions
   bool HasCRC;
 
+  /// If true, the instructions "vmov.i32 d0, #0" and "vmov.i32 q0, #0" are
+  /// particularly effective at zeroing a VFP register.
+  bool HasZeroCycleZeroing;
+
   /// AllowsUnalignedMem - If true, the subtarget allows unaligned memory
   /// accesses for some types.  For details, see
   /// ARMTargetLowering::allowsUnalignedMemoryAccesses().
@@ -198,6 +212,9 @@ protected:
   /// CPUString - String name of used CPU.
   std::string CPUString;
 
+  /// IsLittle - The target is Little Endian
+  bool IsLittle;
+
   /// TargetTriple - What processor and OS we're targeting.
   Triple TargetTriple;
 
@@ -212,6 +229,7 @@ protected:
 
  public:
   enum {
+    ARM_ABI_UNKNOWN,
     ARM_ABI_APCS,
     ARM_ABI_AAPCS // ARM EABI
   } TargetABI;
@@ -220,22 +238,45 @@ protected:
   /// of the specified triple.
   ///
   ARMSubtarget(const std::string &TT, const std::string &CPU,
-               const std::string &FS, const TargetOptions &Options);
+               const std::string &FS, TargetMachine &TM, bool IsLittle,
+               const TargetOptions &Options);
 
   /// getMaxInlineSizeThreshold - Returns the maximum memset / memcpy size
   /// that still makes it profitable to inline the call.
   unsigned getMaxInlineSizeThreshold() const {
-    // FIXME: For now, we don't lower memcpy's to loads / stores for Thumb1.
-    // Change this once Thumb1 ldmia / stmia support is added.
-    return isThumb1Only() ? 0 : 64;
+    return 64;
   }
   /// ParseSubtargetFeatures - Parses features string setting specified
   /// subtarget options.  Definition of function is auto generated by tblgen.
   void ParseSubtargetFeatures(StringRef CPU, StringRef FS);
 
   /// \brief Reset the features for the ARM target.
-  virtual void resetSubtargetFeatures(const MachineFunction *MF);
+  void resetSubtargetFeatures(const MachineFunction *MF) override;
+
+  /// initializeSubtargetDependencies - Initializes using a CPU and feature string
+  /// so that we can use initializer lists for subtarget initialization.
+  ARMSubtarget &initializeSubtargetDependencies(StringRef CPU, StringRef FS);
+
+  const DataLayout *getDataLayout() const { return &DL; }
+  const ARMSelectionDAGInfo *getSelectionDAGInfo() const { return &TSInfo; }
+  ARMJITInfo *getJITInfo() { return &JITInfo; }
+  const ARMBaseInstrInfo *getInstrInfo() const { return InstrInfo.get(); }
+  const ARMTargetLowering *getTargetLowering() const { return &TLInfo; }
+  const ARMFrameLowering *getFrameLowering() const { return FrameLowering.get(); }
+  const ARMBaseRegisterInfo *getRegisterInfo() const {
+    return &InstrInfo->getRegisterInfo();
+  }
+
 private:
+  const DataLayout DL;
+  ARMSelectionDAGInfo TSInfo;
+  ARMJITInfo JITInfo;
+  // Either Thumb1InstrInfo or Thumb2InstrInfo.
+  std::unique_ptr<ARMBaseInstrInfo> InstrInfo;
+  ARMTargetLowering   TLInfo;
+  // Either Thumb1FrameLowering or ARMFrameLowering.
+  std::unique_ptr<ARMFrameLowering> FrameLowering;
+
   void initializeEnvironment();
   void resetSubtargetFeatures(StringRef CPU, StringRef FS);
 public:
@@ -251,13 +292,15 @@ public:
   bool hasV8Ops()   const { return HasV8Ops;  }
 
   bool isCortexA5() const { return ARMProcFamily == CortexA5; }
+  bool isCortexA7() const { return ARMProcFamily == CortexA7; }
   bool isCortexA8() const { return ARMProcFamily == CortexA8; }
   bool isCortexA9() const { return ARMProcFamily == CortexA9; }
   bool isCortexA15() const { return ARMProcFamily == CortexA15; }
   bool isSwift()    const { return ARMProcFamily == Swift; }
   bool isCortexM3() const { return CPUString == "cortex-m3"; }
-  bool isLikeA9() const { return isCortexA9() || isCortexA15(); }
+  bool isLikeA9() const { return isCortexA9() || isCortexA15() || isKrait(); }
   bool isCortexR5() const { return ARMProcFamily == CortexR5; }
+  bool isKrait() const { return ARMProcFamily == Krait; }
 
   bool hasARMOps() const { return !NoARM; }
 
@@ -286,6 +329,7 @@ public:
   bool isFPOnlySP() const { return FPOnlySP; }
   bool hasPerfMon() const { return HasPerfMon; }
   bool hasTrustZone() const { return HasTrustZone; }
+  bool hasZeroCycleZeroing() const { return HasZeroCycleZeroing; }
   bool prefers32BitThumb() const { return Pref32BitThumb; }
   bool avoidCPSRPartialUpdate() const { return AvoidCPSRPartialUpdate; }
   bool avoidMOVsShifterOperand() const { return AvoidMOVsShifterOperand; }
@@ -299,22 +343,59 @@ public:
 
   const Triple &getTargetTriple() const { return TargetTriple; }
 
-  bool isTargetIOS() const { return TargetTriple.isiOS(); }
   bool isTargetDarwin() const { return TargetTriple.isOSDarwin(); }
-  bool isTargetNaCl() const { return TargetTriple.isOSNaCl(); }
+  bool isTargetIOS() const { return TargetTriple.isiOS(); }
   bool isTargetLinux() const { return TargetTriple.isOSLinux(); }
-  bool isTargetELF() const { return !isTargetDarwin(); }
+  bool isTargetNaCl() const { return TargetTriple.isOSNaCl(); }
+  bool isTargetNetBSD() const { return TargetTriple.getOS() == Triple::NetBSD; }
+  bool isTargetWindows() const { return TargetTriple.isOSWindows(); }
+
+  bool isTargetCOFF() const { return TargetTriple.isOSBinFormatCOFF(); }
+  bool isTargetELF() const { return TargetTriple.isOSBinFormatELF(); }
+  bool isTargetMachO() const { return TargetTriple.isOSBinFormatMachO(); }
+
   // ARM EABI is the bare-metal EABI described in ARM ABI documents and
   // can be accessed via -target arm-none-eabi. This is NOT GNUEABI.
   // FIXME: Add a flag for bare-metal for that target and set Triple::EABI
   // even for GNUEABI, so we can make a distinction here and still conform to
   // the EABI on GNU (and Android) mode. This requires change in Clang, too.
+  // FIXME: The Darwin exception is temporary, while we move users to
+  // "*-*-*-macho" triples as quickly as possible.
   bool isTargetAEABI() const {
-    return TargetTriple.getEnvironment() == Triple::EABI;
+    return (TargetTriple.getEnvironment() == Triple::EABI ||
+            TargetTriple.getEnvironment() == Triple::EABIHF) &&
+           !isTargetDarwin() && !isTargetWindows();
   }
 
-  bool isAPCS_ABI() const { return TargetABI == ARM_ABI_APCS; }
-  bool isAAPCS_ABI() const { return TargetABI == ARM_ABI_AAPCS; }
+  // ARM Targets that support EHABI exception handling standard
+  // Darwin uses SjLj. Other targets might need more checks.
+  bool isTargetEHABICompatible() const {
+    return (TargetTriple.getEnvironment() == Triple::EABI ||
+            TargetTriple.getEnvironment() == Triple::GNUEABI ||
+            TargetTriple.getEnvironment() == Triple::EABIHF ||
+            TargetTriple.getEnvironment() == Triple::GNUEABIHF ||
+            TargetTriple.getEnvironment() == Triple::Android) &&
+           !isTargetDarwin() && !isTargetWindows();
+  }
+
+  bool isTargetHardFloat() const {
+    // FIXME: this is invalid for WindowsCE
+    return TargetTriple.getEnvironment() == Triple::GNUEABIHF ||
+           TargetTriple.getEnvironment() == Triple::EABIHF ||
+           isTargetWindows();
+  }
+  bool isTargetAndroid() const {
+    return TargetTriple.getEnvironment() == Triple::Android;
+  }
+
+  bool isAPCS_ABI() const {
+    assert(TargetABI != ARM_ABI_UNKNOWN);
+    return TargetABI == ARM_ABI_APCS;
+  }
+  bool isAAPCS_ABI() const {
+    assert(TargetABI != ARM_ABI_UNKNOWN);
+    return TargetABI == ARM_ABI_AAPCS;
+  }
 
   bool isThumb() const { return InThumbMode; }
   bool isThumb1Only() const { return InThumbMode && !HasThumb2; }
@@ -326,7 +407,8 @@ public:
 
   bool isR9Reserved() const { return IsR9Reserved; }
 
-  bool useMovt() const { return UseMovt && hasV6T2Ops(); }
+  bool useMovt(const MachineFunction &MF) const;
+
   bool supportsTailCall() const { return SupportsTailCall; }
 
   bool allowsUnalignedMem() const { return AllowsUnalignedMem; }
@@ -335,16 +417,19 @@ public:
 
   const std::string & getCPUString() const { return CPUString; }
 
+  bool isLittle() const { return IsLittle; }
+
   unsigned getMispredictionPenalty() const;
-  
+
   /// This function returns true if the target has sincos() routine in its
   /// compiler runtime or math libraries.
   bool hasSinCos() const;
 
-  /// enablePostRAScheduler - True at 'More' optimization.
-  bool enablePostRAScheduler(CodeGenOpt::Level OptLevel,
-                             TargetSubtargetInfo::AntiDepBreakMode& Mode,
-                             RegClassVector& CriticalPathRCs) const;
+  /// True for some subtargets at > -O0.
+  bool enablePostMachineScheduler() const override;
+
+  // enableAtomicExpandLoadLinked - True if we need to expand our atomics.
+  bool enableAtomicExpandLoadLinked() const override;
 
   /// getInstrItins - Return the instruction itineraies based on subtarget
   /// selection.
@@ -358,6 +443,7 @@ public:
   /// GVIsIndirectSymbol - true if the GV will be accessed via an indirect
   /// symbol.
   bool GVIsIndirectSymbol(const GlobalValue *GV, Reloc::Model RelocM) const;
+
 };
 } // End llvm namespace
 
diff --git a/contrib/llvm/lib/Target/ARM/ARMTargetMachine.cpp b/contrib/llvm/lib/Target/ARM/ARMTargetMachine.cpp
index c2bf78877875..d85194b75ecb 100644
--- a/contrib/llvm/lib/Target/ARM/ARMTargetMachine.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMTargetMachine.cpp
@@ -10,8 +10,8 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "ARMTargetMachine.h"
 #include "ARM.h"
+#include "ARMTargetMachine.h"
 #include "ARMFrameLowering.h"
 #include "llvm/CodeGen/Passes.h"
 #include "llvm/MC/MCAsmInfo.h"
@@ -24,19 +24,22 @@
 using namespace llvm;
 
 static cl::opt<bool>
-EnableGlobalMerge("global-merge", cl::Hidden,
-                  cl::desc("Enable global merge pass"),
-                  cl::init(true));
-
-static cl::opt<bool>
 DisableA15SDOptimization("disable-a15-sd-optimization", cl::Hidden,
                    cl::desc("Inhibit optimization of S->D register accesses on A15"),
                    cl::init(false));
 
+static cl::opt<bool>
+EnableAtomicTidy("arm-atomic-cfg-tidy", cl::Hidden,
+                 cl::desc("Run SimplifyCFG after expanding atomic operations"
+                          " to make use of cmpxchg flow-based information"),
+                 cl::init(true));
+
 extern "C" void LLVMInitializeARMTarget() {
   // Register the target.
-  RegisterTargetMachine<ARMTargetMachine> X(TheARMTarget);
-  RegisterTargetMachine<ThumbTargetMachine> Y(TheThumbTarget);
+  RegisterTargetMachine<ARMLETargetMachine> X(TheARMLETarget);
+  RegisterTargetMachine<ARMBETargetMachine> Y(TheARMBETarget);
+  RegisterTargetMachine<ThumbLETargetMachine> A(TheThumbLETarget);
+  RegisterTargetMachine<ThumbBETargetMachine> B(TheThumbBETarget);
 }
 
 
@@ -46,14 +49,14 @@ ARMBaseTargetMachine::ARMBaseTargetMachine(const Target &T, StringRef TT,
                                            StringRef CPU, StringRef FS,
                                            const TargetOptions &Options,
                                            Reloc::Model RM, CodeModel::Model CM,
-                                           CodeGenOpt::Level OL)
-  : LLVMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL),
-    Subtarget(TT, CPU, FS, Options),
-    JITInfo(),
-    InstrItins(Subtarget.getInstrItineraryData()) {
-  // Default to soft float ABI
+                                           CodeGenOpt::Level OL, bool isLittle)
+    : LLVMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL),
+      Subtarget(TT, CPU, FS, *this, isLittle, Options) {
+
+  // Default to triple-appropriate float ABI
   if (Options.FloatABIType == FloatABI::Default)
-    this->Options.FloatABIType = FloatABI::Soft;
+    this->Options.FloatABIType =
+        Subtarget.isTargetHardFloat() ? FloatABI::Hard : FloatABI::Soft;
 }
 
 void ARMBaseTargetMachine::addAnalysisPasses(PassManagerBase &PM) {
@@ -67,60 +70,65 @@ void ARMBaseTargetMachine::addAnalysisPasses(PassManagerBase &PM) {
 
 void ARMTargetMachine::anchor() { }
 
-ARMTargetMachine::ARMTargetMachine(const Target &T, StringRef TT,
-                                   StringRef CPU, StringRef FS,
-                                   const TargetOptions &Options,
+ARMTargetMachine::ARMTargetMachine(const Target &T, StringRef TT, StringRef CPU,
+                                   StringRef FS, const TargetOptions &Options,
                                    Reloc::Model RM, CodeModel::Model CM,
-                                   CodeGenOpt::Level OL)
-  : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL),
-    InstrInfo(Subtarget),
-    DL(Subtarget.isAPCS_ABI() ?
-               std::string("e-p:32:32-f64:32:64-i64:32:64-"
-                           "v128:32:128-v64:32:64-n32-S32") :
-               Subtarget.isAAPCS_ABI() ?
-               std::string("e-p:32:32-f64:64:64-i64:64:64-"
-                           "v128:64:128-v64:64:64-n32-S64") :
-               std::string("e-p:32:32-f64:64:64-i64:64:64-"
-                           "v128:64:128-v64:64:64-n32-S32")),
-    TLInfo(*this),
-    TSInfo(*this),
-    FrameLowering(Subtarget) {
+                                   CodeGenOpt::Level OL, bool isLittle)
+    : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, isLittle) {
   initAsmInfo();
   if (!Subtarget.hasARMOps())
     report_fatal_error("CPU: '" + Subtarget.getCPUString() + "' does not "
                        "support ARM mode execution!");
 }
 
+void ARMLETargetMachine::anchor() { }
+
+ARMLETargetMachine::ARMLETargetMachine(const Target &T, StringRef TT,
+                                       StringRef CPU, StringRef FS,
+                                       const TargetOptions &Options,
+                                       Reloc::Model RM, CodeModel::Model CM,
+                                       CodeGenOpt::Level OL)
+    : ARMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, true) {}
+
+void ARMBETargetMachine::anchor() { }
+
+ARMBETargetMachine::ARMBETargetMachine(const Target &T, StringRef TT,
+                                       StringRef CPU, StringRef FS,
+                                       const TargetOptions &Options,
+                                       Reloc::Model RM, CodeModel::Model CM,
+                                       CodeGenOpt::Level OL)
+    : ARMTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, false) {}
+
 void ThumbTargetMachine::anchor() { }
 
 ThumbTargetMachine::ThumbTargetMachine(const Target &T, StringRef TT,
                                        StringRef CPU, StringRef FS,
                                        const TargetOptions &Options,
                                        Reloc::Model RM, CodeModel::Model CM,
-                                       CodeGenOpt::Level OL)
-  : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL),
-    InstrInfo(Subtarget.hasThumb2()
-              ? ((ARMBaseInstrInfo*)new Thumb2InstrInfo(Subtarget))
-              : ((ARMBaseInstrInfo*)new Thumb1InstrInfo(Subtarget))),
-    DL(Subtarget.isAPCS_ABI() ?
-               std::string("e-p:32:32-f64:32:64-i64:32:64-"
-                           "i16:16:32-i8:8:32-i1:8:32-"
-                           "v128:32:128-v64:32:64-a:0:32-n32-S32") :
-               Subtarget.isAAPCS_ABI() ?
-               std::string("e-p:32:32-f64:64:64-i64:64:64-"
-                           "i16:16:32-i8:8:32-i1:8:32-"
-                           "v128:64:128-v64:64:64-a:0:32-n32-S64") :
-               std::string("e-p:32:32-f64:64:64-i64:64:64-"
-                           "i16:16:32-i8:8:32-i1:8:32-"
-                           "v128:64:128-v64:64:64-a:0:32-n32-S32")),
-    TLInfo(*this),
-    TSInfo(*this),
-    FrameLowering(Subtarget.hasThumb2()
-              ? new ARMFrameLowering(Subtarget)
-              : (ARMFrameLowering*)new Thumb1FrameLowering(Subtarget)) {
+                                       CodeGenOpt::Level OL, bool isLittle)
+    : ARMBaseTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL,
+                           isLittle) {
   initAsmInfo();
 }
 
+void ThumbLETargetMachine::anchor() { }
+
+ThumbLETargetMachine::ThumbLETargetMachine(const Target &T, StringRef TT,
+                                           StringRef CPU, StringRef FS,
+                                           const TargetOptions &Options,
+                                           Reloc::Model RM, CodeModel::Model CM,
+                                           CodeGenOpt::Level OL)
+    : ThumbTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, true) {}
+
+void ThumbBETargetMachine::anchor() { }
+
+ThumbBETargetMachine::ThumbBETargetMachine(const Target &T, StringRef TT,
+                                           StringRef CPU, StringRef FS,
+                                           const TargetOptions &Options,
+                                           Reloc::Model RM, CodeModel::Model CM,
+                                           CodeGenOpt::Level OL)
+    : ThumbTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL, false) {}
+
 namespace {
 /// ARM Code Generator Pass Configuration Options.
 class ARMPassConfig : public TargetPassConfig {
@@ -136,11 +144,12 @@ public:
     return *getARMTargetMachine().getSubtargetImpl();
   }
 
-  virtual bool addPreISel();
-  virtual bool addInstSelector();
-  virtual bool addPreRegAlloc();
-  virtual bool addPreSched2();
-  virtual bool addPreEmitPass();
+  void addIRPasses() override;
+  bool addPreISel() override;
+  bool addInstSelector() override;
+  bool addPreRegAlloc() override;
+  bool addPreSched2() override;
+  bool addPreEmitPass() override;
 };
 } // namespace
 
@@ -148,8 +157,22 @@ TargetPassConfig *ARMBaseTargetMachine::createPassConfig(PassManagerBase &PM) {
   return new ARMPassConfig(this, PM);
 }
 
+void ARMPassConfig::addIRPasses() {
+  addPass(createAtomicExpandLoadLinkedPass(TM));
+
+  // Cmpxchg instructions are often used with a subsequent comparison to
+  // determine whether it succeeded. We can exploit existing control-flow in
+  // ldrex/strex loops to simplify this, but it needs tidying up.
+  const ARMSubtarget *Subtarget = &getARMSubtarget();
+  if (Subtarget->hasAnyDataBarrier() && !Subtarget->isThumb1Only())
+    if (TM->getOptLevel() != CodeGenOpt::None && EnableAtomicTidy)
+      addPass(createCFGSimplificationPass());
+
+  TargetPassConfig::addIRPasses();
+}
+
 bool ARMPassConfig::addPreISel() {
-  if (TM->getOptLevel() != CodeGenOpt::None && EnableGlobalMerge)
+  if (TM->getOptLevel() != CodeGenOpt::None)
     addPass(createGlobalMergePass(TM));
 
   return false;
@@ -166,8 +189,7 @@ bool ARMPassConfig::addInstSelector() {
 }
 
 bool ARMPassConfig::addPreRegAlloc() {
-  // FIXME: temporarily disabling load / store optimization pass for Thumb1.
-  if (getOptLevel() != CodeGenOpt::None && !getARMSubtarget().isThumb1Only())
+  if (getOptLevel() != CodeGenOpt::None)
     addPass(createARMLoadStoreOptimizationPass(true));
   if (getOptLevel() != CodeGenOpt::None && getARMSubtarget().isCortexA9())
     addPass(createMLxExpansionPass());
@@ -181,12 +203,10 @@ bool ARMPassConfig::addPreRegAlloc() {
 }
 
 bool ARMPassConfig::addPreSched2() {
-  // FIXME: temporarily disabling load / store optimization pass for Thumb1.
   if (getOptLevel() != CodeGenOpt::None) {
-    if (!getARMSubtarget().isThumb1Only()) {
-      addPass(createARMLoadStoreOptimizationPass());
-      printAndVerify("After ARM load / store optimizer");
-    }
+    addPass(createARMLoadStoreOptimizationPass());
+    printAndVerify("After ARM load / store optimizer");
+
     if (getARMSubtarget().hasNEON())
       addPass(createExecutionDependencyFixPass(&ARM::DPRRegClass));
   }
@@ -219,6 +239,7 @@ bool ARMPassConfig::addPreEmitPass() {
     addPass(&UnpackMachineBundlesID);
   }
 
+  addPass(createARMOptimizeBarriersPass());
   addPass(createARMConstantIslandPass());
 
   return true;
diff --git a/contrib/llvm/lib/Target/ARM/ARMTargetMachine.h b/contrib/llvm/lib/Target/ARM/ARMTargetMachine.h
index d4caf5ca6e19..b72b1df4af83 100644
--- a/contrib/llvm/lib/Target/ARM/ARMTargetMachine.h
+++ b/contrib/llvm/lib/Target/ARM/ARMTargetMachine.h
@@ -14,18 +14,9 @@
 #ifndef ARMTARGETMACHINE_H
 #define ARMTARGETMACHINE_H
 
-#include "ARMFrameLowering.h"
-#include "ARMISelLowering.h"
 #include "ARMInstrInfo.h"
-#include "ARMJITInfo.h"
-#include "ARMSelectionDAGInfo.h"
 #include "ARMSubtarget.h"
-#include "Thumb1FrameLowering.h"
-#include "Thumb1InstrInfo.h"
-#include "Thumb2InstrInfo.h"
-#include "llvm/ADT/OwningPtr.h"
 #include "llvm/IR/DataLayout.h"
-#include "llvm/MC/MCStreamer.h"
 #include "llvm/Target/TargetMachine.h"
 
 namespace llvm {
@@ -33,68 +24,76 @@ namespace llvm {
 class ARMBaseTargetMachine : public LLVMTargetMachine {
 protected:
   ARMSubtarget        Subtarget;
-private:
-  ARMJITInfo          JITInfo;
-  InstrItineraryData  InstrItins;
-
 public:
   ARMBaseTargetMachine(const Target &T, StringRef TT,
                        StringRef CPU, StringRef FS,
                        const TargetOptions &Options,
                        Reloc::Model RM, CodeModel::Model CM,
-                       CodeGenOpt::Level OL);
+                       CodeGenOpt::Level OL,
+                       bool isLittle);
 
-  virtual       ARMJITInfo       *getJITInfo()         { return &JITInfo; }
-  virtual const ARMSubtarget  *getSubtargetImpl() const { return &Subtarget; }
-  virtual const ARMTargetLowering *getTargetLowering() const {
-    // Implemented by derived classes
-    llvm_unreachable("getTargetLowering not implemented");
+  const ARMSubtarget *getSubtargetImpl() const override { return &Subtarget; }
+  const ARMBaseRegisterInfo *getRegisterInfo() const override {
+    return getSubtargetImpl()->getRegisterInfo();
+  }
+  const ARMTargetLowering *getTargetLowering() const override {
+    return getSubtargetImpl()->getTargetLowering();
+  }
+  const ARMSelectionDAGInfo *getSelectionDAGInfo() const override {
+    return getSubtargetImpl()->getSelectionDAGInfo();
+  }
+  const ARMBaseInstrInfo *getInstrInfo() const override {
+    return getSubtargetImpl()->getInstrInfo();
   }
-  virtual const InstrItineraryData *getInstrItineraryData() const {
-    return &InstrItins;
+  const ARMFrameLowering *getFrameLowering() const override {
+    return getSubtargetImpl()->getFrameLowering();
   }
+  const InstrItineraryData *getInstrItineraryData() const override {
+    return &getSubtargetImpl()->getInstrItineraryData();
+  }
+  const DataLayout *getDataLayout() const override {
+    return getSubtargetImpl()->getDataLayout();
+  }
+  ARMJITInfo *getJITInfo() override { return Subtarget.getJITInfo(); }
 
   /// \brief Register ARM analysis passes with a pass manager.
-  virtual void addAnalysisPasses(PassManagerBase &PM);
+  void addAnalysisPasses(PassManagerBase &PM) override;
 
   // Pass Pipeline Configuration
-  virtual TargetPassConfig *createPassConfig(PassManagerBase &PM);
+  TargetPassConfig *createPassConfig(PassManagerBase &PM) override;
 
-  virtual bool addCodeEmitter(PassManagerBase &PM, JITCodeEmitter &MCE);
+  bool addCodeEmitter(PassManagerBase &PM, JITCodeEmitter &MCE) override;
 };
 
 /// ARMTargetMachine - ARM target machine.
 ///
 class ARMTargetMachine : public ARMBaseTargetMachine {
   virtual void anchor();
-  ARMInstrInfo        InstrInfo;
-  const DataLayout    DL;       // Calculates type size & alignment
-  ARMTargetLowering   TLInfo;
-  ARMSelectionDAGInfo TSInfo;
-  ARMFrameLowering    FrameLowering;
  public:
-  ARMTargetMachine(const Target &T, StringRef TT,
-                   StringRef CPU, StringRef FS,
-                   const TargetOptions &Options,
-                   Reloc::Model RM, CodeModel::Model CM,
-                   CodeGenOpt::Level OL);
-
-  virtual const ARMRegisterInfo  *getRegisterInfo() const {
-    return &InstrInfo.getRegisterInfo();
-  }
+   ARMTargetMachine(const Target &T, StringRef TT, StringRef CPU, StringRef FS,
+                    const TargetOptions &Options, Reloc::Model RM,
+                    CodeModel::Model CM, CodeGenOpt::Level OL, bool isLittle);
+};
 
-  virtual const ARMTargetLowering *getTargetLowering() const {
-    return &TLInfo;
-  }
+/// ARMLETargetMachine - ARM little endian target machine.
+///
+class ARMLETargetMachine : public ARMTargetMachine {
+  void anchor() override;
+public:
+  ARMLETargetMachine(const Target &T, StringRef TT,
+                     StringRef CPU, StringRef FS, const TargetOptions &Options,
+                     Reloc::Model RM, CodeModel::Model CM,
+                     CodeGenOpt::Level OL);
+};
 
-  virtual const ARMSelectionDAGInfo* getSelectionDAGInfo() const {
-    return &TSInfo;
-  }
-  virtual const ARMFrameLowering *getFrameLowering() const {
-    return &FrameLowering;
-  }
-  virtual const ARMInstrInfo     *getInstrInfo() const { return &InstrInfo; }
-  virtual const DataLayout       *getDataLayout() const { return &DL; }
+/// ARMBETargetMachine - ARM big endian target machine.
+///
+class ARMBETargetMachine : public ARMTargetMachine {
+  void anchor() override;
+public:
+  ARMBETargetMachine(const Target &T, StringRef TT, StringRef CPU, StringRef FS,
+                     const TargetOptions &Options, Reloc::Model RM,
+                     CodeModel::Model CM, CodeGenOpt::Level OL);
 };
 
 /// ThumbTargetMachine - Thumb target machine.
@@ -103,42 +102,32 @@ class ARMTargetMachine : public ARMBaseTargetMachine {
 ///
 class ThumbTargetMachine : public ARMBaseTargetMachine {
   virtual void anchor();
-  // Either Thumb1InstrInfo or Thumb2InstrInfo.
-  OwningPtr<ARMBaseInstrInfo> InstrInfo;
-  const DataLayout    DL;   // Calculates type size & alignment
-  ARMTargetLowering   TLInfo;
-  ARMSelectionDAGInfo TSInfo;
-  // Either Thumb1FrameLowering or ARMFrameLowering.
-  OwningPtr<ARMFrameLowering> FrameLowering;
 public:
-  ThumbTargetMachine(const Target &T, StringRef TT,
-                     StringRef CPU, StringRef FS,
-                     const TargetOptions &Options,
-                     Reloc::Model RM, CodeModel::Model CM,
-                     CodeGenOpt::Level OL);
-
-  /// returns either Thumb1RegisterInfo or Thumb2RegisterInfo
-  virtual const ARMBaseRegisterInfo *getRegisterInfo() const {
-    return &InstrInfo->getRegisterInfo();
-  }
-
-  virtual const ARMTargetLowering *getTargetLowering() const {
-    return &TLInfo;
-  }
+  ThumbTargetMachine(const Target &T, StringRef TT, StringRef CPU, StringRef FS,
+                     const TargetOptions &Options, Reloc::Model RM,
+                     CodeModel::Model CM, CodeGenOpt::Level OL, bool isLittle);
+};
 
-  virtual const ARMSelectionDAGInfo *getSelectionDAGInfo() const {
-    return &TSInfo;
-  }
+/// ThumbLETargetMachine - Thumb little endian target machine.
+///
+class ThumbLETargetMachine : public ThumbTargetMachine {
+  void anchor() override;
+public:
+  ThumbLETargetMachine(const Target &T, StringRef TT, StringRef CPU,
+                       StringRef FS, const TargetOptions &Options,
+                       Reloc::Model RM, CodeModel::Model CM,
+                       CodeGenOpt::Level OL);
+};
 
-  /// returns either Thumb1InstrInfo or Thumb2InstrInfo
-  virtual const ARMBaseInstrInfo *getInstrInfo() const {
-    return InstrInfo.get();
-  }
-  /// returns either Thumb1FrameLowering or ARMFrameLowering
-  virtual const ARMFrameLowering *getFrameLowering() const {
-    return FrameLowering.get();
-  }
-  virtual const DataLayout       *getDataLayout() const { return &DL; }
+/// ThumbBETargetMachine - Thumb big endian target machine.
+///
+class ThumbBETargetMachine : public ThumbTargetMachine {
+  void anchor() override;
+public:
+  ThumbBETargetMachine(const Target &T, StringRef TT, StringRef CPU,
+                       StringRef FS, const TargetOptions &Options,
+                       Reloc::Model RM, CodeModel::Model CM,
+                       CodeGenOpt::Level OL);
 };
 
 } // end namespace llvm
diff --git a/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.cpp b/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.cpp
index 7ec71b20c64d..48238bfcf696 100644
--- a/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.cpp
@@ -10,13 +10,14 @@
 #include "ARMTargetObjectFile.h"
 #include "ARMSubtarget.h"
 #include "llvm/ADT/StringExtras.h"
+#include "llvm/IR/Mangler.h"
+#include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/MCContext.h"
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCSectionELF.h"
 #include "llvm/Support/Dwarf.h"
 #include "llvm/Support/ELF.h"
-#include "llvm/Target/Mangler.h"
-#include "llvm/Target/TargetMachine.h"
+#include "llvm/Target/TargetLowering.h"
 using namespace llvm;
 using namespace dwarf;
 
@@ -31,7 +32,7 @@ void ARMElfTargetObjectFile::Initialize(MCContext &Ctx,
   InitializeELF(isAAPCS_ABI);
 
   if (isAAPCS_ABI) {
-    LSDASection = NULL;
+    LSDASection = nullptr;
   }
 
   AttributesSection =
@@ -41,13 +42,22 @@ void ARMElfTargetObjectFile::Initialize(MCContext &Ctx,
                                SectionKind::getMetadata());
 }
 
-const MCExpr *ARMElfTargetObjectFile::
-getTTypeGlobalReference(const GlobalValue *GV, Mangler *Mang,
-                        MachineModuleInfo *MMI, unsigned Encoding,
-                        MCStreamer &Streamer) const {
+const MCExpr *ARMElfTargetObjectFile::getTTypeGlobalReference(
+    const GlobalValue *GV, unsigned Encoding, Mangler &Mang,
+    const TargetMachine &TM, MachineModuleInfo *MMI,
+    MCStreamer &Streamer) const {
+  if (TM.getMCAsmInfo()->getExceptionHandlingType() != ExceptionHandling::ARM)
+    return TargetLoweringObjectFileELF::getTTypeGlobalReference(
+        GV, Encoding, Mang, TM, MMI, Streamer);
+
   assert(Encoding == DW_EH_PE_absptr && "Can handle absptr encoding only");
 
-  return MCSymbolRefExpr::Create(getSymbol(*Mang, GV),
-                                 MCSymbolRefExpr::VK_ARM_TARGET2,
+  return MCSymbolRefExpr::Create(TM.getSymbol(GV, Mang),
+                                 MCSymbolRefExpr::VK_ARM_TARGET2, getContext());
+}
+
+const MCExpr *ARMElfTargetObjectFile::
+getDebugThreadLocalSymbol(const MCSymbol *Sym) const {
+  return MCSymbolRefExpr::Create(Sym, MCSymbolRefExpr::VK_ARM_TLSLDO,
                                  getContext());
 }
diff --git a/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.h b/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.h
index 7f60727e5305..c926421848e5 100644
--- a/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.h
+++ b/contrib/llvm/lib/Target/ARM/ARMTargetObjectFile.h
@@ -23,19 +23,19 @@ protected:
 public:
   ARMElfTargetObjectFile() :
     TargetLoweringObjectFileELF(),
-    AttributesSection(NULL)
+    AttributesSection(nullptr)
   {}
 
-  virtual void Initialize(MCContext &Ctx, const TargetMachine &TM);
+  void Initialize(MCContext &Ctx, const TargetMachine &TM) override;
 
   const MCExpr *
-  getTTypeGlobalReference(const GlobalValue *GV, Mangler *Mang,
-                          MachineModuleInfo *MMI, unsigned Encoding,
-                          MCStreamer &Streamer) const;
-  
-  virtual const MCSection *getAttributesSection() const {
-    return AttributesSection;
-  }
+  getTTypeGlobalReference(const GlobalValue *GV, unsigned Encoding,
+                          Mangler &Mang, const TargetMachine &TM,
+                          MachineModuleInfo *MMI,
+                          MCStreamer &Streamer) const override;
+
+  /// \brief Describe a TLS variable address within debug info.
+  const MCExpr *getDebugThreadLocalSymbol(const MCSymbol *Sym) const override;
 };
 
 } // end namespace llvm
diff --git a/contrib/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp b/contrib/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
index 6bbb38facc24..a2ace629baa1 100644
--- a/contrib/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
@@ -14,17 +14,18 @@
 ///
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "armtti"
 #include "ARM.h"
 #include "ARMTargetMachine.h"
 #include "llvm/Analysis/TargetTransformInfo.h"
 #include "llvm/Support/Debug.h"
-#include "llvm/Target/TargetLowering.h"
 #include "llvm/Target/CostTable.h"
+#include "llvm/Target/TargetLowering.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "armtti"
+
 // Declare the pass initialization routine locally as target-specific passes
-// don't havve a target-wide initialization entry point, and so we rely on the
+// don't have a target-wide initialization entry point, and so we rely on the
 // pass constructor initialization.
 namespace llvm {
 void initializeARMTTIPass(PassRegistry &);
@@ -32,7 +33,7 @@ void initializeARMTTIPass(PassRegistry &);
 
 namespace {
 
-class ARMTTI : public ImmutablePass, public TargetTransformInfo {
+class ARMTTI final : public ImmutablePass, public TargetTransformInfo {
   const ARMBaseTargetMachine *TM;
   const ARMSubtarget *ST;
   const ARMTargetLowering *TLI;
@@ -42,7 +43,7 @@ class ARMTTI : public ImmutablePass, public TargetTransformInfo {
   unsigned getScalarizationOverhead(Type *Ty, bool Insert, bool Extract) const;
 
 public:
-  ARMTTI() : ImmutablePass(ID), TM(0), ST(0), TLI(0) {
+  ARMTTI() : ImmutablePass(ID), TM(nullptr), ST(nullptr), TLI(nullptr) {
     llvm_unreachable("This pass cannot be directly constructed");
   }
 
@@ -52,15 +53,11 @@ public:
     initializeARMTTIPass(*PassRegistry::getPassRegistry());
   }
 
-  virtual void initializePass() {
+  void initializePass() override {
     pushTTIStack(this);
   }
 
-  virtual void finalizePass() {
-    popTTIStack();
-  }
-
-  virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+  void getAnalysisUsage(AnalysisUsage &AU) const override {
     TargetTransformInfo::getAnalysisUsage(AU);
   }
 
@@ -68,7 +65,7 @@ public:
   static char ID;
 
   /// Provide necessary pointer adjustments for the two base classes.
-  virtual void *getAdjustedAnalysisPointer(const void *ID) {
+  void *getAdjustedAnalysisPointer(const void *ID) override {
     if (ID == &TargetTransformInfo::ID)
       return (TargetTransformInfo*)this;
     return this;
@@ -76,8 +73,8 @@ public:
 
   /// \name Scalar TTI Implementations
   /// @{
-
-  virtual unsigned getIntImmCost(const APInt &Imm, Type *Ty) const;
+  using TargetTransformInfo::getIntImmCost;
+  unsigned getIntImmCost(const APInt &Imm, Type *Ty) const override;
 
   /// @}
 
@@ -85,7 +82,7 @@ public:
   /// \name Vector TTI Implementations
   /// @{
 
-  unsigned getNumberOfRegisters(bool Vector) const {
+  unsigned getNumberOfRegisters(bool Vector) const override {
     if (Vector) {
       if (ST->hasNEON())
         return 16;
@@ -94,10 +91,10 @@ public:
 
     if (ST->isThumb1Only())
       return 8;
-    return 16;
+    return 13;
   }
 
-  unsigned getRegisterBitWidth(bool Vector) const {
+  unsigned getRegisterBitWidth(bool Vector) const override {
     if (Vector) {
       if (ST->hasNEON())
         return 128;
@@ -107,7 +104,7 @@ public:
     return 32;
   }
 
-  unsigned getMaximumUnrollFactor() const {
+  unsigned getMaximumUnrollFactor() const override {
     // These are out of order CPUs:
     if (ST->isCortexA15() || ST->isSwift())
       return 2;
@@ -115,23 +112,27 @@ public:
   }
 
   unsigned getShuffleCost(ShuffleKind Kind, Type *Tp,
-                          int Index, Type *SubTp) const;
+                          int Index, Type *SubTp) const override;
 
   unsigned getCastInstrCost(unsigned Opcode, Type *Dst,
-                                      Type *Src) const;
+                            Type *Src) const override;
 
-  unsigned getCmpSelInstrCost(unsigned Opcode, Type *ValTy, Type *CondTy) const;
+  unsigned getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
+                              Type *CondTy) const override;
 
-  unsigned getVectorInstrCost(unsigned Opcode, Type *Val, unsigned Index) const;
+  unsigned getVectorInstrCost(unsigned Opcode, Type *Val,
+                              unsigned Index) const override;
 
-  unsigned getAddressComputationCost(Type *Val, bool IsComplex) const;
+  unsigned getAddressComputationCost(Type *Val,
+                                     bool IsComplex) const override;
 
-  unsigned getArithmeticInstrCost(unsigned Opcode, Type *Ty,
-                                  OperandValueKind Op1Info = OK_AnyValue,
-                                  OperandValueKind Op2Info = OK_AnyValue) const;
+  unsigned
+  getArithmeticInstrCost(unsigned Opcode, Type *Ty,
+                         OperandValueKind Op1Info = OK_AnyValue,
+                         OperandValueKind Op2Info = OK_AnyValue) const override;
 
   unsigned getMemoryOpCost(unsigned Opcode, Type *Src, unsigned Alignment,
-                           unsigned AddressSpace) const;
+                           unsigned AddressSpace) const override;
   /// @}
 };
 
@@ -162,25 +163,25 @@ unsigned ARMTTI::getIntImmCost(const APInt &Imm, Type *Ty) const {
         (ARM_AM::getSOImmVal(~ZImmVal) != -1))
       return 1;
     return ST->hasV6T2Ops() ? 2 : 3;
-  } else if (ST->isThumb2()) {
+  }
+  if (ST->isThumb2()) {
     if ((SImmVal >= 0 && SImmVal < 65536) ||
         (ARM_AM::getT2SOImmVal(ZImmVal) != -1) ||
         (ARM_AM::getT2SOImmVal(~ZImmVal) != -1))
       return 1;
     return ST->hasV6T2Ops() ? 2 : 3;
-  } else /*Thumb1*/ {
-    if (SImmVal >= 0 && SImmVal < 256)
-      return 1;
-    if ((~ZImmVal < 256) || ARM_AM::isThumbImmShiftedVal(ZImmVal))
-      return 2;
-    // Load from constantpool.
-    return 3;
   }
-  return 2;
+  // Thumb1.
+  if (SImmVal >= 0 && SImmVal < 256)
+    return 1;
+  if ((~ZImmVal < 256) || ARM_AM::isThumbImmShiftedVal(ZImmVal))
+    return 2;
+  // Load from constantpool.
+  return 3;
 }
 
 unsigned ARMTTI::getCastInstrCost(unsigned Opcode, Type *Dst,
-                                    Type *Src) const {
+                                  Type *Src) const {
   int ISD = TLI->InstructionOpcodeToISD(Opcode);
   assert(ISD && "Invalid opcode");
 
@@ -442,34 +443,62 @@ unsigned ARMTTI::getAddressComputationCost(Type *Ty, bool IsComplex) const {
 
 unsigned ARMTTI::getShuffleCost(ShuffleKind Kind, Type *Tp, int Index,
                                 Type *SubTp) const {
-  // We only handle costs of reverse shuffles for now.
-  if (Kind != SK_Reverse)
+  // We only handle costs of reverse and alternate shuffles for now.
+  if (Kind != SK_Reverse && Kind != SK_Alternate)
     return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
 
-  static const CostTblEntry<MVT::SimpleValueType> NEONShuffleTbl[] = {
-    // Reverse shuffle cost one instruction if we are shuffling within a double
-    // word (vrev) or two if we shuffle a quad word (vrev, vext).
-    { ISD::VECTOR_SHUFFLE, MVT::v2i32, 1 },
-    { ISD::VECTOR_SHUFFLE, MVT::v2f32, 1 },
-    { ISD::VECTOR_SHUFFLE, MVT::v2i64, 1 },
-    { ISD::VECTOR_SHUFFLE, MVT::v2f64, 1 },
-
-    { ISD::VECTOR_SHUFFLE, MVT::v4i32, 2 },
-    { ISD::VECTOR_SHUFFLE, MVT::v4f32, 2 },
-    { ISD::VECTOR_SHUFFLE, MVT::v8i16, 2 },
-    { ISD::VECTOR_SHUFFLE, MVT::v16i8, 2 }
-  };
+  if (Kind == SK_Reverse) {
+    static const CostTblEntry<MVT::SimpleValueType> NEONShuffleTbl[] = {
+        // Reverse shuffle cost one instruction if we are shuffling within a
+        // double word (vrev) or two if we shuffle a quad word (vrev, vext).
+        {ISD::VECTOR_SHUFFLE, MVT::v2i32, 1},
+        {ISD::VECTOR_SHUFFLE, MVT::v2f32, 1},
+        {ISD::VECTOR_SHUFFLE, MVT::v2i64, 1},
+        {ISD::VECTOR_SHUFFLE, MVT::v2f64, 1},
 
-  std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Tp);
+        {ISD::VECTOR_SHUFFLE, MVT::v4i32, 2},
+        {ISD::VECTOR_SHUFFLE, MVT::v4f32, 2},
+        {ISD::VECTOR_SHUFFLE, MVT::v8i16, 2},
+        {ISD::VECTOR_SHUFFLE, MVT::v16i8, 2}};
 
-  int Idx = CostTableLookup(NEONShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
-  if (Idx == -1)
-    return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
+    std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Tp);
+
+    int Idx = CostTableLookup(NEONShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
+    if (Idx == -1)
+      return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
 
-  return LT.first * NEONShuffleTbl[Idx].Cost;
+    return LT.first * NEONShuffleTbl[Idx].Cost;
+  }
+  if (Kind == SK_Alternate) {
+    static const CostTblEntry<MVT::SimpleValueType> NEONAltShuffleTbl[] = {
+        // Alt shuffle cost table for ARM. Cost is the number of instructions
+        // required to create the shuffled vector.
+
+        {ISD::VECTOR_SHUFFLE, MVT::v2f32, 1},
+        {ISD::VECTOR_SHUFFLE, MVT::v2i64, 1},
+        {ISD::VECTOR_SHUFFLE, MVT::v2f64, 1},
+        {ISD::VECTOR_SHUFFLE, MVT::v2i32, 1},
+
+        {ISD::VECTOR_SHUFFLE, MVT::v4i32, 2},
+        {ISD::VECTOR_SHUFFLE, MVT::v4f32, 2},
+        {ISD::VECTOR_SHUFFLE, MVT::v4i16, 2},
+
+        {ISD::VECTOR_SHUFFLE, MVT::v8i16, 16},
+
+        {ISD::VECTOR_SHUFFLE, MVT::v16i8, 32}};
+
+    std::pair<unsigned, MVT> LT = TLI->getTypeLegalizationCost(Tp);
+    int Idx =
+        CostTableLookup(NEONAltShuffleTbl, ISD::VECTOR_SHUFFLE, LT.second);
+    if (Idx == -1)
+      return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
+    return LT.first * NEONAltShuffleTbl[Idx].Cost;
+  }
+  return TargetTransformInfo::getShuffleCost(Kind, Tp, Index, SubTp);
 }
 
-unsigned ARMTTI::getArithmeticInstrCost(unsigned Opcode, Type *Ty, OperandValueKind Op1Info,
+unsigned ARMTTI::getArithmeticInstrCost(unsigned Opcode, Type *Ty,
+                                        OperandValueKind Op1Info,
                                         OperandValueKind Op2Info) const {
 
   int ISDOpcode = TLI->InstructionOpcodeToISD(Opcode);
@@ -533,7 +562,7 @@ unsigned ARMTTI::getArithmeticInstrCost(unsigned Opcode, Type *Ty, OperandValueK
   // creates a sequence of shift, and, or instructions to construct values.
   // These sequences are recognized by the ISel and have zero-cost. Not so for
   // the vectorized code. Because we have support for v2i64 but not i64 those
-  // sequences look particularily beneficial to vectorize.
+  // sequences look particularly beneficial to vectorize.
   // To work around this we increase the cost of v2i64 operations to make them
   // seem less beneficial.
   if (LT.second == MVT::v2i64 &&
diff --git a/contrib/llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp b/contrib/llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp
index e3f9e0dc609a..b62706c45fbf 100644
--- a/contrib/llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp
+++ b/contrib/llvm/lib/Target/ARM/AsmParser/ARMAsmParser.cpp
@@ -7,33 +7,40 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "ARMBuildAttrs.h"
 #include "ARMFPUName.h"
 #include "ARMFeatures.h"
-#include "llvm/MC/MCTargetAsmParser.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
+#include "MCTargetDesc/ARMArchName.h"
 #include "MCTargetDesc/ARMBaseInfo.h"
 #include "MCTargetDesc/ARMMCExpr.h"
-#include "llvm/ADT/BitVector.h"
-#include "llvm/ADT/OwningPtr.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallVector.h"
+#include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCContext.h"
+#include "llvm/MC/MCDisassembler.h"
 #include "llvm/MC/MCELFStreamer.h"
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCInst.h"
 #include "llvm/MC/MCInstrDesc.h"
 #include "llvm/MC/MCInstrInfo.h"
+#include "llvm/MC/MCObjectFileInfo.h"
 #include "llvm/MC/MCParser/MCAsmLexer.h"
 #include "llvm/MC/MCParser/MCAsmParser.h"
 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
 #include "llvm/MC/MCRegisterInfo.h"
+#include "llvm/MC/MCSection.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSubtargetInfo.h"
+#include "llvm/MC/MCSymbol.h"
+#include "llvm/MC/MCTargetAsmParser.h"
+#include "llvm/Support/ARMBuildAttributes.h"
+#include "llvm/Support/ARMEHABI.h"
+#include "llvm/Support/COFF.h"
+#include "llvm/Support/Debug.h"
 #include "llvm/Support/ELF.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/SourceMgr.h"
@@ -48,31 +55,90 @@ class ARMOperand;
 
 enum VectorLaneTy { NoLanes, AllLanes, IndexedLane };
 
+class UnwindContext {
+  MCAsmParser &Parser;
+
+  typedef SmallVector<SMLoc, 4> Locs;
+
+  Locs FnStartLocs;
+  Locs CantUnwindLocs;
+  Locs PersonalityLocs;
+  Locs PersonalityIndexLocs;
+  Locs HandlerDataLocs;
+  int FPReg;
+
+public:
+  UnwindContext(MCAsmParser &P) : Parser(P), FPReg(ARM::SP) {}
+
+  bool hasFnStart() const { return !FnStartLocs.empty(); }
+  bool cantUnwind() const { return !CantUnwindLocs.empty(); }
+  bool hasHandlerData() const { return !HandlerDataLocs.empty(); }
+  bool hasPersonality() const {
+    return !(PersonalityLocs.empty() && PersonalityIndexLocs.empty());
+  }
+
+  void recordFnStart(SMLoc L) { FnStartLocs.push_back(L); }
+  void recordCantUnwind(SMLoc L) { CantUnwindLocs.push_back(L); }
+  void recordPersonality(SMLoc L) { PersonalityLocs.push_back(L); }
+  void recordHandlerData(SMLoc L) { HandlerDataLocs.push_back(L); }
+  void recordPersonalityIndex(SMLoc L) { PersonalityIndexLocs.push_back(L); }
+
+  void saveFPReg(int Reg) { FPReg = Reg; }
+  int getFPReg() const { return FPReg; }
+
+  void emitFnStartLocNotes() const {
+    for (Locs::const_iterator FI = FnStartLocs.begin(), FE = FnStartLocs.end();
+         FI != FE; ++FI)
+      Parser.Note(*FI, ".fnstart was specified here");
+  }
+  void emitCantUnwindLocNotes() const {
+    for (Locs::const_iterator UI = CantUnwindLocs.begin(),
+                              UE = CantUnwindLocs.end(); UI != UE; ++UI)
+      Parser.Note(*UI, ".cantunwind was specified here");
+  }
+  void emitHandlerDataLocNotes() const {
+    for (Locs::const_iterator HI = HandlerDataLocs.begin(),
+                              HE = HandlerDataLocs.end(); HI != HE; ++HI)
+      Parser.Note(*HI, ".handlerdata was specified here");
+  }
+  void emitPersonalityLocNotes() const {
+    for (Locs::const_iterator PI = PersonalityLocs.begin(),
+                              PE = PersonalityLocs.end(),
+                              PII = PersonalityIndexLocs.begin(),
+                              PIE = PersonalityIndexLocs.end();
+         PI != PE || PII != PIE;) {
+      if (PI != PE && (PII == PIE || PI->getPointer() < PII->getPointer()))
+        Parser.Note(*PI++, ".personality was specified here");
+      else if (PII != PIE && (PI == PE || PII->getPointer() < PI->getPointer()))
+        Parser.Note(*PII++, ".personalityindex was specified here");
+      else
+        llvm_unreachable(".personality and .personalityindex cannot be "
+                         "at the same location");
+    }
+  }
+
+  void reset() {
+    FnStartLocs = Locs();
+    CantUnwindLocs = Locs();
+    PersonalityLocs = Locs();
+    HandlerDataLocs = Locs();
+    PersonalityIndexLocs = Locs();
+    FPReg = ARM::SP;
+  }
+};
+
 class ARMAsmParser : public MCTargetAsmParser {
   MCSubtargetInfo &STI;
   MCAsmParser &Parser;
   const MCInstrInfo &MII;
   const MCRegisterInfo *MRI;
+  UnwindContext UC;
 
   ARMTargetStreamer &getTargetStreamer() {
-    MCTargetStreamer &TS = getParser().getStreamer().getTargetStreamer();
+    MCTargetStreamer &TS = *getParser().getStreamer().getTargetStreamer();
     return static_cast<ARMTargetStreamer &>(TS);
   }
 
-  // Unwind directives state
-  SMLoc FnStartLoc;
-  SMLoc CantUnwindLoc;
-  SMLoc PersonalityLoc;
-  SMLoc HandlerDataLoc;
-  int FPReg;
-  void resetUnwindDirectiveParserState() {
-    FnStartLoc = SMLoc();
-    CantUnwindLoc = SMLoc();
-    PersonalityLoc = SMLoc();
-    HandlerDataLoc = SMLoc();
-    FPReg = -1;
-  }
-
   // Map of register aliases registers via the .req directive.
   StringMap<unsigned> RegisterReqs;
 
@@ -111,6 +177,9 @@ class ARMAsmParser : public MCTargetAsmParser {
   MCAsmParser &getParser() const { return Parser; }
   MCAsmLexer &getLexer() const { return Parser.getLexer(); }
 
+  void Note(SMLoc L, const Twine &Msg, ArrayRef<SMRange> Ranges = None) {
+    return Parser.Note(L, Msg, Ranges);
+  }
   bool Warning(SMLoc L, const Twine &Msg,
                ArrayRef<SMRange> Ranges = None) {
     return Parser.Warning(L, Msg, Ranges);
@@ -121,15 +190,15 @@ class ARMAsmParser : public MCTargetAsmParser {
   }
 
   int tryParseRegister();
-  bool tryParseRegisterWithWriteBack(SmallVectorImpl<MCParsedAsmOperand*> &);
-  int tryParseShiftRegister(SmallVectorImpl<MCParsedAsmOperand*> &);
-  bool parseRegisterList(SmallVectorImpl<MCParsedAsmOperand*> &);
-  bool parseMemory(SmallVectorImpl<MCParsedAsmOperand*> &);
-  bool parseOperand(SmallVectorImpl<MCParsedAsmOperand*> &, StringRef Mnemonic);
+  bool tryParseRegisterWithWriteBack(OperandVector &);
+  int tryParseShiftRegister(OperandVector &);
+  bool parseRegisterList(OperandVector &);
+  bool parseMemory(OperandVector &);
+  bool parseOperand(OperandVector &, StringRef Mnemonic);
   bool parsePrefix(ARMMCExpr::VariantKind &RefKind);
   bool parseMemRegOffsetShift(ARM_AM::ShiftOpc &ShiftType,
                               unsigned &ShiftAmount);
-  bool parseDirectiveWord(unsigned Size, SMLoc L);
+  bool parseLiteralValues(unsigned Size, SMLoc L);
   bool parseDirectiveThumb(SMLoc L);
   bool parseDirectiveARM(SMLoc L);
   bool parseDirectiveThumbFunc(SMLoc L);
@@ -149,6 +218,17 @@ class ARMAsmParser : public MCTargetAsmParser {
   bool parseDirectiveSetFP(SMLoc L);
   bool parseDirectivePad(SMLoc L);
   bool parseDirectiveRegSave(SMLoc L, bool IsVector);
+  bool parseDirectiveInst(SMLoc L, char Suffix = '\0');
+  bool parseDirectiveLtorg(SMLoc L);
+  bool parseDirectiveEven(SMLoc L);
+  bool parseDirectivePersonalityIndex(SMLoc L);
+  bool parseDirectiveUnwindRaw(SMLoc L);
+  bool parseDirectiveTLSDescSeq(SMLoc L);
+  bool parseDirectiveMovSP(SMLoc L);
+  bool parseDirectiveObjectArch(SMLoc L);
+  bool parseDirectiveArchExtension(SMLoc L);
+  bool parseDirectiveAlign(SMLoc L);
+  bool parseDirectiveThumbSet(SMLoc L);
 
   StringRef splitMnemonic(StringRef Mnemonic, unsigned &PredicationCode,
                           bool &CarrySetting, unsigned &ProcessorIMod,
@@ -202,54 +282,42 @@ class ARMAsmParser : public MCTargetAsmParser {
 
   /// }
 
-  OperandMatchResultTy parseITCondCode(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseCoprocNumOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseCoprocRegOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseCoprocOptionOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseMemBarrierOptOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseInstSyncBarrierOptOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseProcIFlagsOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseMSRMaskOperand(
-    SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parsePKHImm(SmallVectorImpl<MCParsedAsmOperand*> &O,
-                                   StringRef Op, int Low, int High);
-  OperandMatchResultTy parsePKHLSLImm(SmallVectorImpl<MCParsedAsmOperand*> &O) {
+  OperandMatchResultTy parseITCondCode(OperandVector &);
+  OperandMatchResultTy parseCoprocNumOperand(OperandVector &);
+  OperandMatchResultTy parseCoprocRegOperand(OperandVector &);
+  OperandMatchResultTy parseCoprocOptionOperand(OperandVector &);
+  OperandMatchResultTy parseMemBarrierOptOperand(OperandVector &);
+  OperandMatchResultTy parseInstSyncBarrierOptOperand(OperandVector &);
+  OperandMatchResultTy parseProcIFlagsOperand(OperandVector &);
+  OperandMatchResultTy parseMSRMaskOperand(OperandVector &);
+  OperandMatchResultTy parsePKHImm(OperandVector &O, StringRef Op, int Low,
+                                   int High);
+  OperandMatchResultTy parsePKHLSLImm(OperandVector &O) {
     return parsePKHImm(O, "lsl", 0, 31);
   }
-  OperandMatchResultTy parsePKHASRImm(SmallVectorImpl<MCParsedAsmOperand*> &O) {
+  OperandMatchResultTy parsePKHASRImm(OperandVector &O) {
     return parsePKHImm(O, "asr", 1, 32);
   }
-  OperandMatchResultTy parseSetEndImm(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseShifterImm(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseRotImm(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseBitfield(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parsePostIdxReg(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseAM3Offset(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseFPImm(SmallVectorImpl<MCParsedAsmOperand*>&);
-  OperandMatchResultTy parseVectorList(SmallVectorImpl<MCParsedAsmOperand*>&);
+  OperandMatchResultTy parseSetEndImm(OperandVector &);
+  OperandMatchResultTy parseShifterImm(OperandVector &);
+  OperandMatchResultTy parseRotImm(OperandVector &);
+  OperandMatchResultTy parseBitfield(OperandVector &);
+  OperandMatchResultTy parsePostIdxReg(OperandVector &);
+  OperandMatchResultTy parseAM3Offset(OperandVector &);
+  OperandMatchResultTy parseFPImm(OperandVector &);
+  OperandMatchResultTy parseVectorList(OperandVector &);
   OperandMatchResultTy parseVectorLane(VectorLaneTy &LaneKind, unsigned &Index,
                                        SMLoc &EndLoc);
 
   // Asm Match Converter Methods
-  void cvtThumbMultiply(MCInst &Inst,
-                        const SmallVectorImpl<MCParsedAsmOperand*> &);
-  void cvtThumbBranches(MCInst &Inst,
-                        const SmallVectorImpl<MCParsedAsmOperand*> &);
-                        
-  bool validateInstruction(MCInst &Inst,
-                           const SmallVectorImpl<MCParsedAsmOperand*> &Ops);
-  bool processInstruction(MCInst &Inst,
-                          const SmallVectorImpl<MCParsedAsmOperand*> &Ops);
-  bool shouldOmitCCOutOperand(StringRef Mnemonic,
-                              SmallVectorImpl<MCParsedAsmOperand*> &Operands);
-  bool shouldOmitPredicateOperand(StringRef Mnemonic,
-                              SmallVectorImpl<MCParsedAsmOperand*> &Operands);
+  void cvtThumbMultiply(MCInst &Inst, const OperandVector &);
+  void cvtThumbBranches(MCInst &Inst, const OperandVector &);
+
+  bool validateInstruction(MCInst &Inst, const OperandVector &Ops);
+  bool processInstruction(MCInst &Inst, const OperandVector &Ops);
+  bool shouldOmitCCOutOperand(StringRef Mnemonic, OperandVector &Operands);
+  bool shouldOmitPredicateOperand(StringRef Mnemonic, OperandVector &Operands);
+
 public:
   enum ARMMatchResultTy {
     Match_RequiresITBlock = FIRST_TARGET_MATCH_RESULT_TY,
@@ -262,8 +330,9 @@ public:
   };
 
   ARMAsmParser(MCSubtargetInfo &_STI, MCAsmParser &_Parser,
-               const MCInstrInfo &MII)
-      : MCTargetAsmParser(), STI(_STI), Parser(_Parser), MII(MII), FPReg(-1) {
+               const MCInstrInfo &MII,
+               const MCTargetOptions &Options)
+      : MCTargetAsmParser(), STI(_STI), Parser(_Parser), MII(MII), UC(_Parser) {
     MCAsmParserExtension::Initialize(_Parser);
 
     // Cache the MCRegisterInfo.
@@ -279,21 +348,20 @@ public:
   }
 
   // Implementation of the MCTargetAsmParser interface:
-  bool ParseRegister(unsigned &RegNo, SMLoc &StartLoc, SMLoc &EndLoc);
+  bool ParseRegister(unsigned &RegNo, SMLoc &StartLoc, SMLoc &EndLoc) override;
   bool ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
-                        SMLoc NameLoc,
-                        SmallVectorImpl<MCParsedAsmOperand*> &Operands);
-  bool ParseDirective(AsmToken DirectiveID);
+                        SMLoc NameLoc, OperandVector &Operands) override;
+  bool ParseDirective(AsmToken DirectiveID) override;
 
-  unsigned validateTargetOperandClass(MCParsedAsmOperand *Op, unsigned Kind);
-  unsigned checkTargetMatchPredicate(MCInst &Inst);
+  unsigned validateTargetOperandClass(MCParsedAsmOperand &Op,
+                                      unsigned Kind) override;
+  unsigned checkTargetMatchPredicate(MCInst &Inst) override;
 
   bool MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
-                               SmallVectorImpl<MCParsedAsmOperand*> &Operands,
-                               MCStreamer &Out, unsigned &ErrorInfo,
-                               bool MatchingInlineAsm);
-  void onLabelParsed(MCSymbol *Symbol);
-
+                               OperandVector &Operands, MCStreamer &Out,
+                               unsigned &ErrorInfo,
+                               bool MatchingInlineAsm) override;
+  void onLabelParsed(MCSymbol *Symbol) override;
 };
 } // end anonymous namespace
 
@@ -332,7 +400,7 @@ class ARMOperand : public MCParsedAsmOperand {
     k_Token
   } Kind;
 
-  SMLoc StartLoc, EndLoc;
+  SMLoc StartLoc, EndLoc, AlignmentLoc;
   SmallVector<unsigned, 8> Registers;
 
   struct CCOp {
@@ -463,8 +531,8 @@ class ARMOperand : public MCParsedAsmOperand {
     struct BitfieldOp Bitfield;
   };
 
-  ARMOperand(KindTy K) : MCParsedAsmOperand(), Kind(K) {}
 public:
+  ARMOperand(KindTy K) : MCParsedAsmOperand(), Kind(K) {}
   ARMOperand(const ARMOperand &o) : MCParsedAsmOperand() {
     Kind = o.Kind;
     StartLoc = o.StartLoc;
@@ -542,13 +610,19 @@ public:
   }
 
   /// getStartLoc - Get the location of the first token of this operand.
-  SMLoc getStartLoc() const { return StartLoc; }
+  SMLoc getStartLoc() const override { return StartLoc; }
   /// getEndLoc - Get the location of the last token of this operand.
-  SMLoc getEndLoc() const { return EndLoc; }
+  SMLoc getEndLoc() const override { return EndLoc; }
   /// getLocRange - Get the range between the first and last token of this
   /// operand.
   SMRange getLocRange() const { return SMRange(StartLoc, EndLoc); }
 
+  /// getAlignmentLoc - Get the location of the Alignment token of this operand.
+  SMLoc getAlignmentLoc() const {
+    assert(Kind == k_Memory && "Invalid access!");
+    return AlignmentLoc;
+  }
+
   ARMCC::CondCodes getCondCode() const {
     assert(Kind == k_CondCode && "Invalid access!");
     return CC.Val;
@@ -564,7 +638,7 @@ public:
     return StringRef(Tok.Data, Tok.Length);
   }
 
-  unsigned getReg() const {
+  unsigned getReg() const override {
     assert((Kind == k_Register || Kind == k_CCOut) && "Invalid access!");
     return Reg.RegNum;
   }
@@ -612,7 +686,7 @@ public:
   bool isCCOut() const { return Kind == k_CCOut; }
   bool isITMask() const { return Kind == k_ITCondMask; }
   bool isITCondCode() const { return Kind == k_CondCode; }
-  bool isImm() const { return Kind == k_Immediate; }
+  bool isImm() const override { return Kind == k_Immediate; }
   // checks whether this operand is an unsigned offset which fits is a field
   // of specified width and scaled by a specific number of bits
   template<unsigned width, unsigned scale>
@@ -988,14 +1062,14 @@ public:
     int64_t Value = CE->getValue();
     return Value == 1 || Value == 0;
   }
-  bool isReg() const { return Kind == k_Register; }
+  bool isReg() const override { return Kind == k_Register; }
   bool isRegList() const { return Kind == k_RegisterList; }
   bool isDPRRegList() const { return Kind == k_DPRRegisterList; }
   bool isSPRRegList() const { return Kind == k_SPRRegisterList; }
-  bool isToken() const { return Kind == k_Token; }
+  bool isToken() const override { return Kind == k_Token; }
   bool isMemBarrierOpt() const { return Kind == k_MemBarrierOpt; }
   bool isInstSyncBarrierOpt() const { return Kind == k_InstSyncBarrierOpt; }
-  bool isMem() const { return Kind == k_Memory; }
+  bool isMem() const override { return Kind == k_Memory; }
   bool isShifterImm() const { return Kind == k_ShifterImmediate; }
   bool isRegShiftedReg() const { return Kind == k_ShiftedRegister; }
   bool isRegShiftedImm() const { return Kind == k_ShiftedImmediate; }
@@ -1005,12 +1079,12 @@ public:
   bool isPostIdxReg() const {
     return Kind == k_PostIndexRegister && PostIdxReg.ShiftTy ==ARM_AM::no_shift;
   }
-  bool isMemNoOffset(bool alignOK = false) const {
+  bool isMemNoOffset(bool alignOK = false, unsigned Alignment = 0) const {
     if (!isMem())
       return false;
     // No offset of any kind.
-    return Memory.OffsetRegNum == 0 && Memory.OffsetImm == 0 &&
-     (alignOK || Memory.Alignment == 0);
+    return Memory.OffsetRegNum == 0 && Memory.OffsetImm == nullptr &&
+     (alignOK || Memory.Alignment == Alignment);
   }
   bool isMemPCRelImm12() const {
     if (!isMem() || Memory.OffsetRegNum != 0 || Memory.Alignment != 0)
@@ -1026,6 +1100,65 @@ public:
   bool isAlignedMemory() const {
     return isMemNoOffset(true);
   }
+  bool isAlignedMemoryNone() const {
+    return isMemNoOffset(false, 0);
+  }
+  bool isDupAlignedMemoryNone() const {
+    return isMemNoOffset(false, 0);
+  }
+  bool isAlignedMemory16() const {
+    if (isMemNoOffset(false, 2)) // alignment in bytes for 16-bits is 2.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isDupAlignedMemory16() const {
+    if (isMemNoOffset(false, 2)) // alignment in bytes for 16-bits is 2.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isAlignedMemory32() const {
+    if (isMemNoOffset(false, 4)) // alignment in bytes for 32-bits is 4.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isDupAlignedMemory32() const {
+    if (isMemNoOffset(false, 4)) // alignment in bytes for 32-bits is 4.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isAlignedMemory64() const {
+    if (isMemNoOffset(false, 8)) // alignment in bytes for 64-bits is 8.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isDupAlignedMemory64() const {
+    if (isMemNoOffset(false, 8)) // alignment in bytes for 64-bits is 8.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isAlignedMemory64or128() const {
+    if (isMemNoOffset(false, 8)) // alignment in bytes for 64-bits is 8.
+      return true;
+    if (isMemNoOffset(false, 16)) // alignment in bytes for 128-bits is 16.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isDupAlignedMemory64or128() const {
+    if (isMemNoOffset(false, 8)) // alignment in bytes for 64-bits is 8.
+      return true;
+    if (isMemNoOffset(false, 16)) // alignment in bytes for 128-bits is 16.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
+  bool isAlignedMemory64or128or256() const {
+    if (isMemNoOffset(false, 8)) // alignment in bytes for 64-bits is 8.
+      return true;
+    if (isMemNoOffset(false, 16)) // alignment in bytes for 128-bits is 16.
+      return true;
+    if (isMemNoOffset(false, 32)) // alignment in bytes for 256-bits is 32.
+      return true;
+    return isMemNoOffset(false, 0);
+  }
   bool isAddrMode2() const {
     if (!isMem() || Memory.Alignment != 0) return false;
     // Check for register offset.
@@ -1282,6 +1415,7 @@ public:
   }
 
   bool isVecListDPairSpaced() const {
+    if (Kind != k_VectorList) return false;
     if (isSingleSpacedVectorList()) return false;
     return (ARMMCRegisterClasses[ARM::DPairSpcRegClassID]
               .contains(VectorList.RegNum));
@@ -1460,7 +1594,10 @@ public:
   }
 
   bool isNEONi16splat() const {
-    if (!isImm()) return false;
+    if (isNEONByteReplicate(2))
+      return false; // Leave that for bytes replication and forbid by default.
+    if (!isImm())
+      return false;
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     // Must be a constant.
     if (!CE) return false;
@@ -1470,7 +1607,10 @@ public:
   }
 
   bool isNEONi32splat() const {
-    if (!isImm()) return false;
+    if (isNEONByteReplicate(4))
+      return false; // Leave that for bytes replication and forbid by default.
+    if (!isImm())
+      return false;
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     // Must be a constant.
     if (!CE) return false;
@@ -1482,11 +1622,36 @@ public:
       (Value >= 0x01000000 && Value <= 0xff000000);
   }
 
+  bool isNEONByteReplicate(unsigned NumBytes) const {
+    if (!isImm())
+      return false;
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    // Must be a constant.
+    if (!CE)
+      return false;
+    int64_t Value = CE->getValue();
+    if (!Value)
+      return false; // Don't bother with zero.
+
+    unsigned char B = Value & 0xff;
+    for (unsigned i = 1; i < NumBytes; ++i) {
+      Value >>= 8;
+      if ((Value & 0xff) != B)
+        return false;
+    }
+    return true;
+  }
+  bool isNEONi16ByteReplicate() const { return isNEONByteReplicate(2); }
+  bool isNEONi32ByteReplicate() const { return isNEONByteReplicate(4); }
   bool isNEONi32vmov() const {
-    if (!isImm()) return false;
+    if (isNEONByteReplicate(4))
+      return false; // Let it to be classified as byte-replicate case.
+    if (!isImm())
+      return false;
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
     // Must be a constant.
-    if (!CE) return false;
+    if (!CE)
+      return false;
     int64_t Value = CE->getValue();
     // i32 value with set bits only in one byte X000, 0X00, 00X0, or 000X,
     // for VMOV/VMVN only, 00Xf or 0Xff are also accepted.
@@ -1527,7 +1692,7 @@ public:
 
   void addExpr(MCInst &Inst, const MCExpr *Expr) const {
     // Add as immediates when possible.  Null MCExpr = 0.
-    if (Expr == 0)
+    if (!Expr)
       Inst.addOperand(MCOperand::CreateImm(0));
     else if (const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Expr))
       Inst.addOperand(MCOperand::CreateImm(CE->getValue()));
@@ -1580,7 +1745,7 @@ public:
   void addRegShiftedRegOperands(MCInst &Inst, unsigned N) const {
     assert(N == 3 && "Invalid number of operands!");
     assert(isRegShiftedReg() &&
-           "addRegShiftedRegOperands() on non RegShiftedReg!");
+           "addRegShiftedRegOperands() on non-RegShiftedReg!");
     Inst.addOperand(MCOperand::CreateReg(RegShiftedReg.SrcReg));
     Inst.addOperand(MCOperand::CreateReg(RegShiftedReg.ShiftReg));
     Inst.addOperand(MCOperand::CreateImm(
@@ -1590,7 +1755,7 @@ public:
   void addRegShiftedImmOperands(MCInst &Inst, unsigned N) const {
     assert(N == 2 && "Invalid number of operands!");
     assert(isRegShiftedImm() &&
-           "addRegShiftedImmOperands() on non RegShiftedImm!");
+           "addRegShiftedImmOperands() on non-RegShiftedImm!");
     Inst.addOperand(MCOperand::CreateReg(RegShiftedImm.SrcReg));
     // Shift of #32 is encoded as 0 where permitted
     unsigned Imm = (RegShiftedImm.ShiftImm == 32 ? 0 : RegShiftedImm.ShiftImm);
@@ -1841,6 +2006,50 @@ public:
     Inst.addOperand(MCOperand::CreateImm(Memory.Alignment));
   }
 
+  void addDupAlignedMemoryNoneOperands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addAlignedMemoryNoneOperands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addAlignedMemory16Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addDupAlignedMemory16Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addAlignedMemory32Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addDupAlignedMemory32Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addAlignedMemory64Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addDupAlignedMemory64Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addAlignedMemory64or128Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addDupAlignedMemory64or128Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
+  void addAlignedMemory64or128or256Operands(MCInst &Inst, unsigned N) const {
+    addAlignedMemoryOperands(Inst, N);
+  }
+
   void addAddrMode2Operands(MCInst &Inst, unsigned N) const {
     assert(N == 3 && "Invalid number of operands!");
     int32_t Val = Memory.OffsetImm ? Memory.OffsetImm->getValue() : 0;
@@ -2190,6 +2399,19 @@ public:
     Inst.addOperand(MCOperand::CreateImm(Value));
   }
 
+  void addNEONinvByteReplicateOperands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    // The immediate encodes the type of constant as well as the value.
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    unsigned Value = CE->getValue();
+    assert((Inst.getOpcode() == ARM::VMOVv8i8 ||
+            Inst.getOpcode() == ARM::VMOVv16i8) &&
+           "All vmvn instructions that wants to replicate non-zero byte "
+           "always must be replaced with VMOVv8i8 or VMOVv16i8.");
+    unsigned B = ((~Value) & 0xff);
+    B |= 0xe00; // cmode = 0b1110
+    Inst.addOperand(MCOperand::CreateImm(B));
+  }
   void addNEONi32vmovOperands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     // The immediate encodes the type of constant as well as the value.
@@ -2204,6 +2426,19 @@ public:
     Inst.addOperand(MCOperand::CreateImm(Value));
   }
 
+  void addNEONvmovByteReplicateOperands(MCInst &Inst, unsigned N) const {
+    assert(N == 1 && "Invalid number of operands!");
+    // The immediate encodes the type of constant as well as the value.
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(getImm());
+    unsigned Value = CE->getValue();
+    assert((Inst.getOpcode() == ARM::VMOVv8i8 ||
+            Inst.getOpcode() == ARM::VMOVv16i8) &&
+           "All instructions that wants to replicate non-zero byte "
+           "always must be replaced with VMOVv8i8 or VMOVv16i8.");
+    unsigned B = Value & 0xff;
+    B |= 0xe00; // cmode = 0b1110
+    Inst.addOperand(MCOperand::CreateImm(B));
+  }
   void addNEONi32vmovNegOperands(MCInst &Inst, unsigned N) const {
     assert(N == 1 && "Invalid number of operands!");
     // The immediate encodes the type of constant as well as the value.
@@ -2230,58 +2465,60 @@ public:
     Inst.addOperand(MCOperand::CreateImm(Imm | 0x1e00));
   }
 
-  virtual void print(raw_ostream &OS) const;
+  void print(raw_ostream &OS) const override;
 
-  static ARMOperand *CreateITMask(unsigned Mask, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_ITCondMask);
+  static std::unique_ptr<ARMOperand> CreateITMask(unsigned Mask, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_ITCondMask);
     Op->ITMask.Mask = Mask;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateCondCode(ARMCC::CondCodes CC, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_CondCode);
+  static std::unique_ptr<ARMOperand> CreateCondCode(ARMCC::CondCodes CC,
+                                                    SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_CondCode);
     Op->CC.Val = CC;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateCoprocNum(unsigned CopVal, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_CoprocNum);
+  static std::unique_ptr<ARMOperand> CreateCoprocNum(unsigned CopVal, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_CoprocNum);
     Op->Cop.Val = CopVal;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateCoprocReg(unsigned CopVal, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_CoprocReg);
+  static std::unique_ptr<ARMOperand> CreateCoprocReg(unsigned CopVal, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_CoprocReg);
     Op->Cop.Val = CopVal;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateCoprocOption(unsigned Val, SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_CoprocOption);
+  static std::unique_ptr<ARMOperand> CreateCoprocOption(unsigned Val, SMLoc S,
+                                                        SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_CoprocOption);
     Op->Cop.Val = Val;
     Op->StartLoc = S;
     Op->EndLoc = E;
     return Op;
   }
 
-  static ARMOperand *CreateCCOut(unsigned RegNum, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_CCOut);
+  static std::unique_ptr<ARMOperand> CreateCCOut(unsigned RegNum, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_CCOut);
     Op->Reg.RegNum = RegNum;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateToken(StringRef Str, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_Token);
+  static std::unique_ptr<ARMOperand> CreateToken(StringRef Str, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_Token);
     Op->Tok.Data = Str.data();
     Op->Tok.Length = Str.size();
     Op->StartLoc = S;
@@ -2289,20 +2526,20 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateReg(unsigned RegNum, SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_Register);
+  static std::unique_ptr<ARMOperand> CreateReg(unsigned RegNum, SMLoc S,
+                                               SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_Register);
     Op->Reg.RegNum = RegNum;
     Op->StartLoc = S;
     Op->EndLoc = E;
     return Op;
   }
 
-  static ARMOperand *CreateShiftedRegister(ARM_AM::ShiftOpc ShTy,
-                                           unsigned SrcReg,
-                                           unsigned ShiftReg,
-                                           unsigned ShiftImm,
-                                           SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_ShiftedRegister);
+  static std::unique_ptr<ARMOperand>
+  CreateShiftedRegister(ARM_AM::ShiftOpc ShTy, unsigned SrcReg,
+                        unsigned ShiftReg, unsigned ShiftImm, SMLoc S,
+                        SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_ShiftedRegister);
     Op->RegShiftedReg.ShiftTy = ShTy;
     Op->RegShiftedReg.SrcReg = SrcReg;
     Op->RegShiftedReg.ShiftReg = ShiftReg;
@@ -2312,11 +2549,10 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateShiftedImmediate(ARM_AM::ShiftOpc ShTy,
-                                            unsigned SrcReg,
-                                            unsigned ShiftImm,
-                                            SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_ShiftedImmediate);
+  static std::unique_ptr<ARMOperand>
+  CreateShiftedImmediate(ARM_AM::ShiftOpc ShTy, unsigned SrcReg,
+                         unsigned ShiftImm, SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_ShiftedImmediate);
     Op->RegShiftedImm.ShiftTy = ShTy;
     Op->RegShiftedImm.SrcReg = SrcReg;
     Op->RegShiftedImm.ShiftImm = ShiftImm;
@@ -2325,9 +2561,9 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateShifterImm(bool isASR, unsigned Imm,
-                                   SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_ShifterImmediate);
+  static std::unique_ptr<ARMOperand> CreateShifterImm(bool isASR, unsigned Imm,
+                                                      SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_ShifterImmediate);
     Op->ShifterImm.isASR = isASR;
     Op->ShifterImm.Imm = Imm;
     Op->StartLoc = S;
@@ -2335,17 +2571,18 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateRotImm(unsigned Imm, SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_RotateImmediate);
+  static std::unique_ptr<ARMOperand> CreateRotImm(unsigned Imm, SMLoc S,
+                                                  SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_RotateImmediate);
     Op->RotImm.Imm = Imm;
     Op->StartLoc = S;
     Op->EndLoc = E;
     return Op;
   }
 
-  static ARMOperand *CreateBitfield(unsigned LSB, unsigned Width,
-                                    SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_BitfieldDescriptor);
+  static std::unique_ptr<ARMOperand>
+  CreateBitfield(unsigned LSB, unsigned Width, SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_BitfieldDescriptor);
     Op->Bitfield.LSB = LSB;
     Op->Bitfield.Width = Width;
     Op->StartLoc = S;
@@ -2353,8 +2590,8 @@ public:
     return Op;
   }
 
-  static ARMOperand *
-  CreateRegList(SmallVectorImpl<std::pair<unsigned, unsigned> > &Regs,
+  static std::unique_ptr<ARMOperand>
+  CreateRegList(SmallVectorImpl<std::pair<unsigned, unsigned>> &Regs,
                 SMLoc StartLoc, SMLoc EndLoc) {
     assert (Regs.size() > 0 && "RegList contains no registers?");
     KindTy Kind = k_RegisterList;
@@ -2368,7 +2605,7 @@ public:
     // Sort based on the register encoding values.
     array_pod_sort(Regs.begin(), Regs.end());
 
-    ARMOperand *Op = new ARMOperand(Kind);
+    auto Op = make_unique<ARMOperand>(Kind);
     for (SmallVectorImpl<std::pair<unsigned, unsigned> >::const_iterator
            I = Regs.begin(), E = Regs.end(); I != E; ++I)
       Op->Registers.push_back(I->second);
@@ -2377,9 +2614,11 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateVectorList(unsigned RegNum, unsigned Count,
-                                      bool isDoubleSpaced, SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_VectorList);
+  static std::unique_ptr<ARMOperand> CreateVectorList(unsigned RegNum,
+                                                      unsigned Count,
+                                                      bool isDoubleSpaced,
+                                                      SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_VectorList);
     Op->VectorList.RegNum = RegNum;
     Op->VectorList.Count = Count;
     Op->VectorList.isDoubleSpaced = isDoubleSpaced;
@@ -2388,10 +2627,10 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateVectorListAllLanes(unsigned RegNum, unsigned Count,
-                                              bool isDoubleSpaced,
-                                              SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_VectorListAllLanes);
+  static std::unique_ptr<ARMOperand>
+  CreateVectorListAllLanes(unsigned RegNum, unsigned Count, bool isDoubleSpaced,
+                           SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_VectorListAllLanes);
     Op->VectorList.RegNum = RegNum;
     Op->VectorList.Count = Count;
     Op->VectorList.isDoubleSpaced = isDoubleSpaced;
@@ -2400,11 +2639,10 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateVectorListIndexed(unsigned RegNum, unsigned Count,
-                                             unsigned Index,
-                                             bool isDoubleSpaced,
-                                             SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_VectorListIndexed);
+  static std::unique_ptr<ARMOperand>
+  CreateVectorListIndexed(unsigned RegNum, unsigned Count, unsigned Index,
+                          bool isDoubleSpaced, SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_VectorListIndexed);
     Op->VectorList.RegNum = RegNum;
     Op->VectorList.Count = Count;
     Op->VectorList.LaneIndex = Index;
@@ -2414,32 +2652,30 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateVectorIndex(unsigned Idx, SMLoc S, SMLoc E,
-                                       MCContext &Ctx) {
-    ARMOperand *Op = new ARMOperand(k_VectorIndex);
+  static std::unique_ptr<ARMOperand>
+  CreateVectorIndex(unsigned Idx, SMLoc S, SMLoc E, MCContext &Ctx) {
+    auto Op = make_unique<ARMOperand>(k_VectorIndex);
     Op->VectorIndex.Val = Idx;
     Op->StartLoc = S;
     Op->EndLoc = E;
     return Op;
   }
 
-  static ARMOperand *CreateImm(const MCExpr *Val, SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_Immediate);
+  static std::unique_ptr<ARMOperand> CreateImm(const MCExpr *Val, SMLoc S,
+                                               SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_Immediate);
     Op->Imm.Val = Val;
     Op->StartLoc = S;
     Op->EndLoc = E;
     return Op;
   }
 
-  static ARMOperand *CreateMem(unsigned BaseRegNum,
-                               const MCConstantExpr *OffsetImm,
-                               unsigned OffsetRegNum,
-                               ARM_AM::ShiftOpc ShiftType,
-                               unsigned ShiftImm,
-                               unsigned Alignment,
-                               bool isNegative,
-                               SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_Memory);
+  static std::unique_ptr<ARMOperand>
+  CreateMem(unsigned BaseRegNum, const MCConstantExpr *OffsetImm,
+            unsigned OffsetRegNum, ARM_AM::ShiftOpc ShiftType,
+            unsigned ShiftImm, unsigned Alignment, bool isNegative, SMLoc S,
+            SMLoc E, SMLoc AlignmentLoc = SMLoc()) {
+    auto Op = make_unique<ARMOperand>(k_Memory);
     Op->Memory.BaseRegNum = BaseRegNum;
     Op->Memory.OffsetImm = OffsetImm;
     Op->Memory.OffsetRegNum = OffsetRegNum;
@@ -2449,14 +2685,14 @@ public:
     Op->Memory.isNegative = isNegative;
     Op->StartLoc = S;
     Op->EndLoc = E;
+    Op->AlignmentLoc = AlignmentLoc;
     return Op;
   }
 
-  static ARMOperand *CreatePostIdxReg(unsigned RegNum, bool isAdd,
-                                      ARM_AM::ShiftOpc ShiftTy,
-                                      unsigned ShiftImm,
-                                      SMLoc S, SMLoc E) {
-    ARMOperand *Op = new ARMOperand(k_PostIndexRegister);
+  static std::unique_ptr<ARMOperand>
+  CreatePostIdxReg(unsigned RegNum, bool isAdd, ARM_AM::ShiftOpc ShiftTy,
+                   unsigned ShiftImm, SMLoc S, SMLoc E) {
+    auto Op = make_unique<ARMOperand>(k_PostIndexRegister);
     Op->PostIdxReg.RegNum = RegNum;
     Op->PostIdxReg.isAdd = isAdd;
     Op->PostIdxReg.ShiftTy = ShiftTy;
@@ -2466,33 +2702,35 @@ public:
     return Op;
   }
 
-  static ARMOperand *CreateMemBarrierOpt(ARM_MB::MemBOpt Opt, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_MemBarrierOpt);
+  static std::unique_ptr<ARMOperand> CreateMemBarrierOpt(ARM_MB::MemBOpt Opt,
+                                                         SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_MemBarrierOpt);
     Op->MBOpt.Val = Opt;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateInstSyncBarrierOpt(ARM_ISB::InstSyncBOpt Opt,
-                                              SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_InstSyncBarrierOpt);
+  static std::unique_ptr<ARMOperand>
+  CreateInstSyncBarrierOpt(ARM_ISB::InstSyncBOpt Opt, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_InstSyncBarrierOpt);
     Op->ISBOpt.Val = Opt;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateProcIFlags(ARM_PROC::IFlags IFlags, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_ProcIFlags);
+  static std::unique_ptr<ARMOperand> CreateProcIFlags(ARM_PROC::IFlags IFlags,
+                                                      SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_ProcIFlags);
     Op->IFlags.Val = IFlags;
     Op->StartLoc = S;
     Op->EndLoc = S;
     return Op;
   }
 
-  static ARMOperand *CreateMSRMask(unsigned MMask, SMLoc S) {
-    ARMOperand *Op = new ARMOperand(k_MSRMask);
+  static std::unique_ptr<ARMOperand> CreateMSRMask(unsigned MMask, SMLoc S) {
+    auto Op = make_unique<ARMOperand>(k_MSRMask);
     Op->MMask.Val = MMask;
     Op->StartLoc = S;
     Op->EndLoc = S;
@@ -2696,11 +2934,11 @@ int ARMAsmParser::tryParseRegister() {
 // occurs, return -1. An irrecoverable error is one where tokens have been
 // consumed in the process of trying to parse the shifter (i.e., when it is
 // indeed a shifter operand, but malformed).
-int ARMAsmParser::tryParseShiftRegister(
-                               SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+int ARMAsmParser::tryParseShiftRegister(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
-  assert(Tok.is(AsmToken::Identifier) && "Token is not an Identifier");
+  if (Tok.isNot(AsmToken::Identifier))
+    return -1; 
 
   std::string lowerCase = Tok.getString().lower();
   ARM_AM::ShiftOpc ShiftTy = StringSwitch<ARM_AM::ShiftOpc>(lowerCase)
@@ -2720,7 +2958,8 @@ int ARMAsmParser::tryParseShiftRegister(
   // The source register for the shift has already been added to the
   // operand list, so we need to pop it off and combine it into the shifted
   // register operand instead.
-  OwningPtr<ARMOperand> PrevOp((ARMOperand*)Operands.pop_back_val());
+  std::unique_ptr<ARMOperand> PrevOp(
+      (ARMOperand *)Operands.pop_back_val().release());
   if (!PrevOp->isReg())
     return Error(PrevOp->getStartLoc(), "shift must be of a register");
   int SrcReg = PrevOp->getReg();
@@ -2739,7 +2978,7 @@ int ARMAsmParser::tryParseShiftRegister(
         Parser.getTok().is(AsmToken::Dollar)) {
       Parser.Lex(); // Eat hash.
       SMLoc ImmLoc = Parser.getTok().getLoc();
-      const MCExpr *ShiftExpr = 0;
+      const MCExpr *ShiftExpr = nullptr;
       if (getParser().parseExpression(ShiftExpr, EndLoc)) {
         Error(ImmLoc, "invalid immediate shift value");
         return -1;
@@ -2769,12 +3008,12 @@ int ARMAsmParser::tryParseShiftRegister(
       EndLoc = Parser.getTok().getEndLoc();
       ShiftReg = tryParseRegister();
       if (ShiftReg == -1) {
-        Error (L, "expected immediate or register in shift operand");
+        Error(L, "expected immediate or register in shift operand");
         return -1;
       }
     } else {
-      Error (Parser.getTok().getLoc(),
-                    "expected immediate or register in shift operand");
+      Error(Parser.getTok().getLoc(),
+            "expected immediate or register in shift operand");
       return -1;
     }
   }
@@ -2797,8 +3036,7 @@ int ARMAsmParser::tryParseShiftRegister(
 ///
 /// TODO this is likely to change to allow different register types and or to
 /// parse for a specific register type.
-bool ARMAsmParser::
-tryParseRegisterWithWriteBack(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+bool ARMAsmParser::tryParseRegisterWithWriteBack(OperandVector &Operands) {
   const AsmToken &RegTok = Parser.getTok();
   int RegNo = tryParseRegister();
   if (RegNo == -1)
@@ -2844,17 +3082,25 @@ tryParseRegisterWithWriteBack(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 }
 
 /// MatchCoprocessorOperandName - Try to parse an coprocessor related
-/// instruction with a symbolic operand name. Example: "p1", "p7", "c3",
-/// "c5", ...
+/// instruction with a symbolic operand name.
+/// We accept "crN" syntax for GAS compatibility.
+/// <operand-name> ::= <prefix><number>
+/// If CoprocOp is 'c', then:
+///   <prefix> ::= c | cr
+/// If CoprocOp is 'p', then :
+///   <prefix> ::= p
+/// <number> ::= integer in range [0, 15]
 static int MatchCoprocessorOperandName(StringRef Name, char CoprocOp) {
   // Use the same layout as the tablegen'erated register name matcher. Ugly,
   // but efficient.
+  if (Name.size() < 2 || Name[0] != CoprocOp)
+    return -1;
+  Name = (Name[1] == 'r') ? Name.drop_front(2) : Name.drop_front();
+
   switch (Name.size()) {
   default: return -1;
-  case 2:
-    if (Name[0] != CoprocOp)
-      return -1;
-    switch (Name[1]) {
+  case 1:
+    switch (Name[0]) {
     default:  return -1;
     case '0': return 0;
     case '1': return 1;
@@ -2867,10 +3113,10 @@ static int MatchCoprocessorOperandName(StringRef Name, char CoprocOp) {
     case '8': return 8;
     case '9': return 9;
     }
-  case 3:
-    if (Name[0] != CoprocOp || Name[1] != '1')
+  case 2:
+    if (Name[0] != '1')
       return -1;
-    switch (Name[2]) {
+    switch (Name[1]) {
     default:  return -1;
     // p10 and p11 are invalid for coproc instructions (reserved for FP/NEON)
     case '0': return CoprocOp == 'p'? -1: 10;
@@ -2884,8 +3130,8 @@ static int MatchCoprocessorOperandName(StringRef Name, char CoprocOp) {
 }
 
 /// parseITCondCode - Try to parse a condition code for an IT instruction.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseITCondCode(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseITCondCode(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   if (!Tok.is(AsmToken::Identifier))
@@ -2921,8 +3167,8 @@ parseITCondCode(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 /// parseCoprocNumOperand - Try to parse an coprocessor number operand. The
 /// token must be an Identifier when called, and if it is a coprocessor
 /// number, the token is eaten and the operand is added to the operand list.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseCoprocNumOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseCoprocNumOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   if (Tok.isNot(AsmToken::Identifier))
@@ -2940,8 +3186,8 @@ parseCoprocNumOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 /// parseCoprocRegOperand - Try to parse an coprocessor register operand. The
 /// token must be an Identifier when called, and if it is a coprocessor
 /// number, the token is eaten and the operand is added to the operand list.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseCoprocRegOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseCoprocRegOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   if (Tok.isNot(AsmToken::Identifier))
@@ -2958,8 +3204,8 @@ parseCoprocRegOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 
 /// parseCoprocOptionOperand - Try to parse an coprocessor option operand.
 /// coproc_option : '{' imm0_255 '}'
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseCoprocOptionOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseCoprocOptionOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
 
   // If this isn't a '{', this isn't a coprocessor immediate operand.
@@ -3036,8 +3282,7 @@ static unsigned getDRegFromQReg(unsigned QReg) {
 }
 
 /// Parse a register list.
-bool ARMAsmParser::
-parseRegisterList(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+bool ARMAsmParser::parseRegisterList(OperandVector &Operands) {
   assert(Parser.getTok().is(AsmToken::LCurly) &&
          "Token is not a Left Curly Brace");
   SMLoc S = Parser.getTok().getLoc();
@@ -3218,8 +3463,8 @@ parseVectorLane(VectorLaneTy &LaneKind, unsigned &Index, SMLoc &EndLoc) {
 }
 
 // parse a vector register list
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseVectorList(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseVectorList(OperandVector &Operands) {
   VectorLaneTy LaneKind;
   unsigned LaneIndex;
   SMLoc S = Parser.getTok().getLoc();
@@ -3469,8 +3714,8 @@ parseVectorList(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 }
 
 /// parseMemBarrierOptOperand - Try to parse DSB/DMB data barrier options.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseMemBarrierOptOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseMemBarrierOptOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   unsigned Opt;
@@ -3518,7 +3763,7 @@ parseMemBarrierOptOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
       Error(Loc, "illegal expression");
       return MatchOperand_ParseFail;
     }
-    
+
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(MemBarrierID);
     if (!CE) {
       Error(Loc, "constant expression expected");
@@ -3540,8 +3785,8 @@ parseMemBarrierOptOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 }
 
 /// parseInstSyncBarrierOptOperand - Try to parse ISB inst sync barrier options.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseInstSyncBarrierOptOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseInstSyncBarrierOptOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   unsigned Opt;
@@ -3591,8 +3836,8 @@ parseInstSyncBarrierOptOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 
 
 /// parseProcIFlagsOperand - Try to parse iflags from CPS instruction.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseProcIFlagsOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseProcIFlagsOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   if (!Tok.is(AsmToken::Identifier)) 
@@ -3625,8 +3870,8 @@ parseProcIFlagsOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 }
 
 /// parseMSRMaskOperand - Try to parse mask flags from MSR instruction.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseMSRMaskOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseMSRMaskOperand(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   const AsmToken &Tok = Parser.getTok();
   if (!Tok.is(AsmToken::Identifier))
@@ -3753,9 +3998,9 @@ parseMSRMaskOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   return MatchOperand_Success;
 }
 
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parsePKHImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands, StringRef Op,
-            int Low, int High) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parsePKHImm(OperandVector &Operands, StringRef Op, int Low,
+                          int High) {
   const AsmToken &Tok = Parser.getTok();
   if (Tok.isNot(AsmToken::Identifier)) {
     Error(Parser.getTok().getLoc(), Op + " operand expected.");
@@ -3801,8 +4046,8 @@ parsePKHImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands, StringRef Op,
   return MatchOperand_Success;
 }
 
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseSetEndImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseSetEndImm(OperandVector &Operands) {
   const AsmToken &Tok = Parser.getTok();
   SMLoc S = Tok.getLoc();
   if (Tok.isNot(AsmToken::Identifier)) {
@@ -3830,8 +4075,8 @@ parseSetEndImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 ///     lsl #n  'n' in [0,31]
 ///     asr #n  'n' in [1,32]
 ///             n == 32 encoded as n == 0.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseShifterImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseShifterImm(OperandVector &Operands) {
   const AsmToken &Tok = Parser.getTok();
   SMLoc S = Tok.getLoc();
   if (Tok.isNot(AsmToken::Identifier)) {
@@ -3900,8 +4145,8 @@ parseShifterImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 /// parseRotImm - Parse the shifter immediate operand for SXTB/UXTB family
 /// of instructions. Legal values are:
 ///     ror #n  'n' in {0, 8, 16, 24}
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseRotImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseRotImm(OperandVector &Operands) {
   const AsmToken &Tok = Parser.getTok();
   SMLoc S = Tok.getLoc();
   if (Tok.isNot(AsmToken::Identifier))
@@ -3946,8 +4191,8 @@ parseRotImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   return MatchOperand_Success;
 }
 
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseBitfield(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseBitfield(OperandVector &Operands) {
   SMLoc S = Parser.getTok().getLoc();
   // The bitfield descriptor is really two operands, the LSB and the width.
   if (Parser.getTok().isNot(AsmToken::Hash) &&
@@ -4014,8 +4259,8 @@ parseBitfield(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   return MatchOperand_Success;
 }
 
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parsePostIdxReg(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parsePostIdxReg(OperandVector &Operands) {
   // Check for a post-index addressing register operand. Specifically:
   // postidx_reg := '+' register {, shift}
   //              | '-' register {, shift}
@@ -4063,8 +4308,8 @@ parsePostIdxReg(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   return MatchOperand_Success;
 }
 
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseAM3Offset(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseAM3Offset(OperandVector &Operands) {
   // Check for a post-index addressing register operand. Specifically:
   // am3offset := '+' register
   //              | '-' register
@@ -4117,7 +4362,7 @@ parseAM3Offset(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
     isAdd = false;
     haveEaten = true;
   }
-  
+
   Tok = Parser.getTok();
   int Reg = tryParseRegister();
   if (Reg == -1) {
@@ -4136,26 +4381,24 @@ parseAM3Offset(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 /// Convert parsed operands to MCInst.  Needed here because this instruction
 /// only has two register operands, but multiplication is commutative so
 /// assemblers should accept both "mul rD, rN, rD" and "mul rD, rD, rN".
-void ARMAsmParser::
-cvtThumbMultiply(MCInst &Inst,
-           const SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
-  ((ARMOperand*)Operands[3])->addRegOperands(Inst, 1);
-  ((ARMOperand*)Operands[1])->addCCOutOperands(Inst, 1);
+void ARMAsmParser::cvtThumbMultiply(MCInst &Inst,
+                                    const OperandVector &Operands) {
+  ((ARMOperand &)*Operands[3]).addRegOperands(Inst, 1);
+  ((ARMOperand &)*Operands[1]).addCCOutOperands(Inst, 1);
   // If we have a three-operand form, make sure to set Rn to be the operand
   // that isn't the same as Rd.
   unsigned RegOp = 4;
   if (Operands.size() == 6 &&
-      ((ARMOperand*)Operands[4])->getReg() ==
-        ((ARMOperand*)Operands[3])->getReg())
+      ((ARMOperand &)*Operands[4]).getReg() ==
+          ((ARMOperand &)*Operands[3]).getReg())
     RegOp = 5;
-  ((ARMOperand*)Operands[RegOp])->addRegOperands(Inst, 1);
+  ((ARMOperand &)*Operands[RegOp]).addRegOperands(Inst, 1);
   Inst.addOperand(Inst.getOperand(0));
-  ((ARMOperand*)Operands[2])->addCondCodeOperands(Inst, 2);
+  ((ARMOperand &)*Operands[2]).addCondCodeOperands(Inst, 2);
 }
 
-void ARMAsmParser::
-cvtThumbBranches(MCInst &Inst,
-           const SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+void ARMAsmParser::cvtThumbBranches(MCInst &Inst,
+                                    const OperandVector &Operands) {
   int CondOp = -1, ImmOp = -1;
   switch(Inst.getOpcode()) {
     case ARM::tB:
@@ -4178,7 +4421,7 @@ cvtThumbBranches(MCInst &Inst,
   } else {
     // outside IT blocks we can only have unconditional branches with AL
     // condition code or conditional branches with non-AL condition code
-    unsigned Cond = static_cast<ARMOperand*>(Operands[CondOp])->getCondCode();
+    unsigned Cond = static_cast<ARMOperand &>(*Operands[CondOp]).getCondCode();
     switch(Inst.getOpcode()) {
       case ARM::tB:
       case ARM::tBcc: 
@@ -4190,32 +4433,31 @@ cvtThumbBranches(MCInst &Inst,
         break;
     }
   }
-  
+
   // now decide on encoding size based on branch target range
   switch(Inst.getOpcode()) {
     // classify tB as either t2B or t1B based on range of immediate operand
     case ARM::tB: {
-      ARMOperand* op = static_cast<ARMOperand*>(Operands[ImmOp]);
-      if(!op->isSignedOffset<11, 1>() && isThumbTwo()) 
+      ARMOperand &op = static_cast<ARMOperand &>(*Operands[ImmOp]);
+      if (!op.isSignedOffset<11, 1>() && isThumbTwo())
         Inst.setOpcode(ARM::t2B);
       break;
     }
     // classify tBcc as either t2Bcc or t1Bcc based on range of immediate operand
     case ARM::tBcc: {
-      ARMOperand* op = static_cast<ARMOperand*>(Operands[ImmOp]);
-      if(!op->isSignedOffset<8, 1>() && isThumbTwo())
+      ARMOperand &op = static_cast<ARMOperand &>(*Operands[ImmOp]);
+      if (!op.isSignedOffset<8, 1>() && isThumbTwo())
         Inst.setOpcode(ARM::t2Bcc);
       break;
     }
   }
-  ((ARMOperand*)Operands[ImmOp])->addImmOperands(Inst, 1);
-  ((ARMOperand*)Operands[CondOp])->addCondCodeOperands(Inst, 2);
+  ((ARMOperand &)*Operands[ImmOp]).addImmOperands(Inst, 1);
+  ((ARMOperand &)*Operands[CondOp]).addCondCodeOperands(Inst, 2);
 }
 
 /// Parse an ARM memory expression, return false if successful else return true
 /// or an error.  The first token must be a '[' when called.
-bool ARMAsmParser::
-parseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+bool ARMAsmParser::parseMemory(OperandVector &Operands) {
   SMLoc S, E;
   assert(Parser.getTok().is(AsmToken::LBrac) &&
          "Token is not a Left Bracket");
@@ -4237,8 +4479,9 @@ parseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
     E = Tok.getEndLoc();
     Parser.Lex(); // Eat right bracket token.
 
-    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, 0, 0, ARM_AM::no_shift,
-                                             0, 0, false, S, E));
+    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, nullptr, 0,
+                                             ARM_AM::no_shift, 0, 0, false,
+                                             S, E));
 
     // If there's a pre-indexing writeback marker, '!', just add it as a token
     // operand. It's rather odd, but syntactically valid.
@@ -4260,6 +4503,7 @@ parseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   if (Parser.getTok().is(AsmToken::Colon)) {
     Parser.Lex(); // Eat the ':'.
     E = Parser.getTok().getLoc();
+    SMLoc AlignmentLoc = Tok.getLoc();
 
     const MCExpr *Expr;
     if (getParser().parseExpression(Expr))
@@ -4292,9 +4536,9 @@ parseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 
     // Don't worry about range checking the value here. That's handled by
     // the is*() predicates.
-    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, 0, 0,
+    Operands.push_back(ARMOperand::CreateMem(BaseRegNum, nullptr, 0,
                                              ARM_AM::no_shift, 0, Align,
-                                             false, S, E));
+                                             false, S, E, AlignmentLoc));
 
     // If there's a pre-indexing writeback marker, '!', just add it as a token
     // operand.
@@ -4385,7 +4629,7 @@ parseMemory(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   E = Parser.getTok().getEndLoc();
   Parser.Lex(); // Eat right bracket token.
 
-  Operands.push_back(ARMOperand::CreateMem(BaseRegNum, 0, OffsetRegNum,
+  Operands.push_back(ARMOperand::CreateMem(BaseRegNum, nullptr, OffsetRegNum,
                                            ShiftType, ShiftImm, 0, isNegative,
                                            S, E));
 
@@ -4463,8 +4707,8 @@ bool ARMAsmParser::parseMemRegOffsetShift(ARM_AM::ShiftOpc &St,
 }
 
 /// parseFPImm - A floating point immediate expression operand.
-ARMAsmParser::OperandMatchResultTy ARMAsmParser::
-parseFPImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+ARMAsmParser::OperandMatchResultTy
+ARMAsmParser::parseFPImm(OperandVector &Operands) {
   // Anything that can accept a floating point constant as an operand
   // needs to go through here, as the regular parseExpression is
   // integer only.
@@ -4490,9 +4734,13 @@ parseFPImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   // integer constant. Make sure we don't try to parse an FPImm
   // for these:
   // vmov.i{8|16|32|64} <dreg|qreg>, #imm
-  ARMOperand *TyOp = static_cast<ARMOperand*>(Operands[2]);
-  if (!TyOp->isToken() || (TyOp->getToken() != ".f32" &&
-                           TyOp->getToken() != ".f64"))
+  ARMOperand &TyOp = static_cast<ARMOperand &>(*Operands[2]);
+  bool isVmovf = TyOp.isToken() &&
+                 (TyOp.getToken() == ".f32" || TyOp.getToken() == ".f64");
+  ARMOperand &Mnemonic = static_cast<ARMOperand &>(*Operands[0]);
+  bool isFconst = Mnemonic.isToken() && (Mnemonic.getToken() == "fconstd" ||
+                                         Mnemonic.getToken() == "fconsts");
+  if (!(isVmovf || isFconst))
     return MatchOperand_NoMatch;
 
   Parser.Lex(); // Eat '#' or '$'.
@@ -4505,7 +4753,7 @@ parseFPImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   }
   const AsmToken &Tok = Parser.getTok();
   SMLoc Loc = Tok.getLoc();
-  if (Tok.is(AsmToken::Real)) {
+  if (Tok.is(AsmToken::Real) && isVmovf) {
     APFloat RealVal(APFloat::IEEEsingle, Tok.getString());
     uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue();
     // If we had a '-' in front, toggle the sign bit.
@@ -4518,15 +4766,16 @@ parseFPImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
   }
   // Also handle plain integers. Instructions which allow floating point
   // immediates also allow a raw encoded 8-bit value.
-  if (Tok.is(AsmToken::Integer)) {
+  if (Tok.is(AsmToken::Integer) && isFconst) {
     int64_t Val = Tok.getIntVal();
     Parser.Lex(); // Eat the token.
     if (Val > 255 || Val < 0) {
       Error(Loc, "encoded floating point value out of range");
       return MatchOperand_ParseFail;
     }
-    double RealVal = ARM_AM::getFPImmFloat(Val);
-    Val = APFloat(APFloat::IEEEdouble, RealVal).bitcastToAPInt().getZExtValue();
+    float RealVal = ARM_AM::getFPImmFloat(Val);
+    Val = APFloat(RealVal).bitcastToAPInt().getZExtValue();
+
     Operands.push_back(ARMOperand::CreateImm(
         MCConstantExpr::Create(Val, getContext()), S,
         Parser.getTok().getLoc()));
@@ -4539,8 +4788,7 @@ parseFPImm(SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
 
 /// Parse a arm instruction operand.  For now this parses the operand regardless
 /// of the mnemonic.
-bool ARMAsmParser::parseOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands,
-                                StringRef Mnemonic) {
+bool ARMAsmParser::parseOperand(OperandVector &Operands, StringRef Mnemonic) {
   SMLoc S, E;
 
   // Check if the current operand has a custom associated parser, if so, try to
@@ -4623,7 +4871,7 @@ bool ARMAsmParser::parseOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands,
       Operands.push_back(ARMOperand::CreateImm(ImmVal, S, E));
 
       // There can be a trailing '!' on operands that we want as a separate
-      // '!' Token operand. Handle that here. For example, the compatibilty
+      // '!' Token operand. Handle that here. For example, the compatibility
       // alias for 'srsdb sp!, #imm' is 'srsdb #imm!'.
       if (Parser.getTok().is(AsmToken::Exclaim)) {
         Operands.push_back(ARMOperand::CreateToken(Parser.getTok().getString(),
@@ -4653,6 +4901,20 @@ bool ARMAsmParser::parseOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands,
     Operands.push_back(ARMOperand::CreateImm(ExprVal, S, E));
     return false;
   }
+  case AsmToken::Equal: {
+    if (Mnemonic != "ldr") // only parse for ldr pseudo (e.g. ldr r0, =val)
+      return Error(Parser.getTok().getLoc(), "unexpected token in operand");
+
+    Parser.Lex(); // Eat '='
+    const MCExpr *SubExprVal;
+    if (getParser().parseExpression(SubExprVal))
+      return true;
+    E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
+
+    const MCExpr *CPLoc = getTargetStreamer().addConstantPoolEntry(SubExprVal);
+    Operands.push_back(ARMOperand::CreateImm(CPLoc, S, E));
+    return false;
+  }
   }
 }
 
@@ -4661,6 +4923,10 @@ bool ARMAsmParser::parseOperand(SmallVectorImpl<MCParsedAsmOperand*> &Operands,
 bool ARMAsmParser::parsePrefix(ARMMCExpr::VariantKind &RefKind) {
   RefKind = ARMMCExpr::VK_ARM_None;
 
+  // consume an optional '#' (GNU compatibility)
+  if (getLexer().is(AsmToken::Hash))
+    Parser.Lex();
+
   // :lower16: and :upper16: modifiers
   assert(getLexer().is(AsmToken::Colon) && "expected a :");
   Parser.Lex(); // Eat ':'
@@ -4763,7 +5029,7 @@ StringRef ARMAsmParser::splitMnemonic(StringRef Mnemonic,
         Mnemonic == "fmrs" || Mnemonic == "fsqrts" || Mnemonic == "fsubs" ||
         Mnemonic == "fsts" || Mnemonic == "fcpys" || Mnemonic == "fdivs" ||
         Mnemonic == "fmuls" || Mnemonic == "fcmps" || Mnemonic == "fcmpzs" ||
-        Mnemonic == "vfms" || Mnemonic == "vfnms" ||
+        Mnemonic == "vfms" || Mnemonic == "vfnms" || Mnemonic == "fconsts" ||
         (Mnemonic == "movs" && isThumb()))) {
     Mnemonic = Mnemonic.slice(0, Mnemonic.size() - 1);
     CarrySetting = true;
@@ -4817,8 +5083,9 @@ getMnemonicAcceptInfo(StringRef Mnemonic, StringRef FullInst,
 
   if (Mnemonic == "bkpt" || Mnemonic == "cbnz" || Mnemonic == "setend" ||
       Mnemonic == "cps" ||  Mnemonic == "it" ||  Mnemonic == "cbz" ||
-      Mnemonic == "trap" || Mnemonic == "hlt" || Mnemonic.startswith("crc32") ||
-      Mnemonic.startswith("cps") || Mnemonic.startswith("vsel") ||
+      Mnemonic == "trap" || Mnemonic == "hlt" || Mnemonic == "udf" ||
+      Mnemonic.startswith("crc32") || Mnemonic.startswith("cps") ||
+      Mnemonic.startswith("vsel") ||
       Mnemonic == "vmaxnm" || Mnemonic == "vminnm" || Mnemonic == "vcvta" ||
       Mnemonic == "vcvtn" || Mnemonic == "vcvtp" || Mnemonic == "vcvtm" ||
       Mnemonic == "vrinta" || Mnemonic == "vrintn" || Mnemonic == "vrintp" ||
@@ -4847,7 +5114,7 @@ getMnemonicAcceptInfo(StringRef Mnemonic, StringRef FullInst,
 }
 
 bool ARMAsmParser::shouldOmitCCOutOperand(StringRef Mnemonic,
-                               SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+                                          OperandVector &Operands) {
   // FIXME: This is all horribly hacky. We really need a better way to deal
   // with optional operands like this in the matcher table.
 
@@ -4860,17 +5127,17 @@ bool ARMAsmParser::shouldOmitCCOutOperand(StringRef Mnemonic,
   // conditionally adding the cc_out in the first place because we need
   // to check the type of the parsed immediate operand.
   if (Mnemonic == "mov" && Operands.size() > 4 && !isThumb() &&
-      !static_cast<ARMOperand*>(Operands[4])->isARMSOImm() &&
-      static_cast<ARMOperand*>(Operands[4])->isImm0_65535Expr() &&
-      static_cast<ARMOperand*>(Operands[1])->getReg() == 0)
+      !static_cast<ARMOperand &>(*Operands[4]).isARMSOImm() &&
+      static_cast<ARMOperand &>(*Operands[4]).isImm0_65535Expr() &&
+      static_cast<ARMOperand &>(*Operands[1]).getReg() == 0)
     return true;
 
   // Register-register 'add' for thumb does not have a cc_out operand
   // when there are only two register operands.
   if (isThumb() && Mnemonic == "add" && Operands.size() == 5 &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->isReg() &&
-      static_cast<ARMOperand*>(Operands[1])->getReg() == 0)
+      static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[1]).getReg() == 0)
     return true;
   // Register-register 'add' for thumb does not have a cc_out operand
   // when it's an ADD Rdm, SP, {Rdm|#imm0_255} instruction. We do
@@ -4878,13 +5145,12 @@ bool ARMAsmParser::shouldOmitCCOutOperand(StringRef Mnemonic,
   // that can handle a different range and has a cc_out operand.
   if (((isThumb() && Mnemonic == "add") ||
        (isThumbTwo() && Mnemonic == "sub")) &&
-      Operands.size() == 6 &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->getReg() == ARM::SP &&
-      static_cast<ARMOperand*>(Operands[1])->getReg() == 0 &&
-      ((Mnemonic == "add" &&static_cast<ARMOperand*>(Operands[5])->isReg()) ||
-       static_cast<ARMOperand*>(Operands[5])->isImm0_1020s4()))
+      Operands.size() == 6 && static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).getReg() == ARM::SP &&
+      static_cast<ARMOperand &>(*Operands[1]).getReg() == 0 &&
+      ((Mnemonic == "add" && static_cast<ARMOperand &>(*Operands[5]).isReg()) ||
+       static_cast<ARMOperand &>(*Operands[5]).isImm0_1020s4()))
     return true;
   // For Thumb2, add/sub immediate does not have a cc_out operand for the
   // imm0_4095 variant. That's the least-preferred variant when
@@ -4892,23 +5158,22 @@ bool ARMAsmParser::shouldOmitCCOutOperand(StringRef Mnemonic,
   // should remove the cc_out operand, we have to explicitly check that
   // it's not one of the other variants. Ugh.
   if (isThumbTwo() && (Mnemonic == "add" || Mnemonic == "sub") &&
-      Operands.size() == 6 &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->isReg() &&
-      static_cast<ARMOperand*>(Operands[5])->isImm()) {
+      Operands.size() == 6 && static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[5]).isImm()) {
     // Nest conditions rather than one big 'if' statement for readability.
     //
     // If both registers are low, we're in an IT block, and the immediate is
     // in range, we should use encoding T1 instead, which has a cc_out.
     if (inITBlock() &&
-        isARMLowRegister(static_cast<ARMOperand*>(Operands[3])->getReg()) &&
-        isARMLowRegister(static_cast<ARMOperand*>(Operands[4])->getReg()) &&
-        static_cast<ARMOperand*>(Operands[5])->isImm0_7())
+        isARMLowRegister(static_cast<ARMOperand &>(*Operands[3]).getReg()) &&
+        isARMLowRegister(static_cast<ARMOperand &>(*Operands[4]).getReg()) &&
+        static_cast<ARMOperand &>(*Operands[5]).isImm0_7())
       return false;
     // Check against T3. If the second register is the PC, this is an
     // alternate form of ADR, which uses encoding T4, so check for that too.
-    if (static_cast<ARMOperand*>(Operands[4])->getReg() != ARM::PC &&
-        static_cast<ARMOperand*>(Operands[5])->isT2SOImm())
+    if (static_cast<ARMOperand &>(*Operands[4]).getReg() != ARM::PC &&
+        static_cast<ARMOperand &>(*Operands[5]).isT2SOImm())
       return false;
 
     // Otherwise, we use encoding T4, which does not have a cc_out
@@ -4920,35 +5185,34 @@ bool ARMAsmParser::shouldOmitCCOutOperand(StringRef Mnemonic,
   // if we have a "mul" mnemonic in Thumb mode, check if we'll be able to
   // use the 16-bit encoding or not.
   if (isThumbTwo() && Mnemonic == "mul" && Operands.size() == 6 &&
-      static_cast<ARMOperand*>(Operands[1])->getReg() == 0 &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->isReg() &&
-      static_cast<ARMOperand*>(Operands[5])->isReg() &&
+      static_cast<ARMOperand &>(*Operands[1]).getReg() == 0 &&
+      static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[5]).isReg() &&
       // If the registers aren't low regs, the destination reg isn't the
       // same as one of the source regs, or the cc_out operand is zero
       // outside of an IT block, we have to use the 32-bit encoding, so
       // remove the cc_out operand.
-      (!isARMLowRegister(static_cast<ARMOperand*>(Operands[3])->getReg()) ||
-       !isARMLowRegister(static_cast<ARMOperand*>(Operands[4])->getReg()) ||
-       !isARMLowRegister(static_cast<ARMOperand*>(Operands[5])->getReg()) ||
-       !inITBlock() ||
-       (static_cast<ARMOperand*>(Operands[3])->getReg() !=
-        static_cast<ARMOperand*>(Operands[5])->getReg() &&
-        static_cast<ARMOperand*>(Operands[3])->getReg() !=
-        static_cast<ARMOperand*>(Operands[4])->getReg())))
+      (!isARMLowRegister(static_cast<ARMOperand &>(*Operands[3]).getReg()) ||
+       !isARMLowRegister(static_cast<ARMOperand &>(*Operands[4]).getReg()) ||
+       !isARMLowRegister(static_cast<ARMOperand &>(*Operands[5]).getReg()) ||
+       !inITBlock() || (static_cast<ARMOperand &>(*Operands[3]).getReg() !=
+                            static_cast<ARMOperand &>(*Operands[5]).getReg() &&
+                        static_cast<ARMOperand &>(*Operands[3]).getReg() !=
+                            static_cast<ARMOperand &>(*Operands[4]).getReg())))
     return true;
 
   // Also check the 'mul' syntax variant that doesn't specify an explicit
   // destination register.
   if (isThumbTwo() && Mnemonic == "mul" && Operands.size() == 5 &&
-      static_cast<ARMOperand*>(Operands[1])->getReg() == 0 &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->isReg() &&
+      static_cast<ARMOperand &>(*Operands[1]).getReg() == 0 &&
+      static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).isReg() &&
       // If the registers aren't low regs  or the cc_out operand is zero
       // outside of an IT block, we have to use the 32-bit encoding, so
       // remove the cc_out operand.
-      (!isARMLowRegister(static_cast<ARMOperand*>(Operands[3])->getReg()) ||
-       !isARMLowRegister(static_cast<ARMOperand*>(Operands[4])->getReg()) ||
+      (!isARMLowRegister(static_cast<ARMOperand &>(*Operands[3]).getReg()) ||
+       !isARMLowRegister(static_cast<ARMOperand &>(*Operands[4]).getReg()) ||
        !inITBlock()))
     return true;
 
@@ -4961,32 +5225,32 @@ bool ARMAsmParser::shouldOmitCCOutOperand(StringRef Mnemonic,
   // anyway.
   if (isThumb() && (Mnemonic == "add" || Mnemonic == "sub") &&
       (Operands.size() == 5 || Operands.size() == 6) &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[3])->getReg() == ARM::SP &&
-      static_cast<ARMOperand*>(Operands[1])->getReg() == 0 &&
-      (static_cast<ARMOperand*>(Operands[4])->isImm() ||
+      static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[3]).getReg() == ARM::SP &&
+      static_cast<ARMOperand &>(*Operands[1]).getReg() == 0 &&
+      (static_cast<ARMOperand &>(*Operands[4]).isImm() ||
        (Operands.size() == 6 &&
-        static_cast<ARMOperand*>(Operands[5])->isImm())))
+        static_cast<ARMOperand &>(*Operands[5]).isImm())))
     return true;
 
   return false;
 }
 
-bool ARMAsmParser::shouldOmitPredicateOperand(
-    StringRef Mnemonic, SmallVectorImpl<MCParsedAsmOperand *> &Operands) {
+bool ARMAsmParser::shouldOmitPredicateOperand(StringRef Mnemonic,
+                                              OperandVector &Operands) {
   // VRINT{Z, R, X} have a predicate operand in VFP, but not in NEON
   unsigned RegIdx = 3;
   if ((Mnemonic == "vrintz" || Mnemonic == "vrintx" || Mnemonic == "vrintr") &&
-      static_cast<ARMOperand *>(Operands[2])->getToken() == ".f32") {
-    if (static_cast<ARMOperand *>(Operands[3])->isToken() &&
-        static_cast<ARMOperand *>(Operands[3])->getToken() == ".f32")
+      static_cast<ARMOperand &>(*Operands[2]).getToken() == ".f32") {
+    if (static_cast<ARMOperand &>(*Operands[3]).isToken() &&
+        static_cast<ARMOperand &>(*Operands[3]).getToken() == ".f32")
       RegIdx = 4;
 
-    if (static_cast<ARMOperand *>(Operands[RegIdx])->isReg() &&
-        (ARMMCRegisterClasses[ARM::DPRRegClassID]
-             .contains(static_cast<ARMOperand *>(Operands[RegIdx])->getReg()) ||
-         ARMMCRegisterClasses[ARM::QPRRegClassID]
-             .contains(static_cast<ARMOperand *>(Operands[RegIdx])->getReg())))
+    if (static_cast<ARMOperand &>(*Operands[RegIdx]).isReg() &&
+        (ARMMCRegisterClasses[ARM::DPRRegClassID].contains(
+             static_cast<ARMOperand &>(*Operands[RegIdx]).getReg()) ||
+         ARMMCRegisterClasses[ARM::QPRRegClassID].contains(
+             static_cast<ARMOperand &>(*Operands[RegIdx]).getReg())))
       return true;
   }
   return false;
@@ -5009,12 +5273,39 @@ static bool doesIgnoreDataTypeSuffix(StringRef Mnemonic, StringRef DT) {
 }
 static void applyMnemonicAliases(StringRef &Mnemonic, unsigned Features,
                                  unsigned VariantID);
+
+static bool RequiresVFPRegListValidation(StringRef Inst,
+                                         bool &AcceptSinglePrecisionOnly,
+                                         bool &AcceptDoublePrecisionOnly) {
+  if (Inst.size() < 7)
+    return false;
+
+  if (Inst.startswith("fldm") || Inst.startswith("fstm")) {
+    StringRef AddressingMode = Inst.substr(4, 2);
+    if (AddressingMode == "ia" || AddressingMode == "db" ||
+        AddressingMode == "ea" || AddressingMode == "fd") {
+      AcceptSinglePrecisionOnly = Inst[6] == 's';
+      AcceptDoublePrecisionOnly = Inst[6] == 'd' || Inst[6] == 'x';
+      return true;
+    }
+  }
+
+  return false;
+}
+
 /// Parse an arm instruction mnemonic followed by its operands.
 bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
-                                    SMLoc NameLoc,
-                               SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+                                    SMLoc NameLoc, OperandVector &Operands) {
+  // FIXME: Can this be done via tablegen in some fashion?
+  bool RequireVFPRegisterListCheck;
+  bool AcceptSinglePrecisionOnly;
+  bool AcceptDoublePrecisionOnly;
+  RequireVFPRegisterListCheck =
+    RequiresVFPRegListValidation(Name, AcceptSinglePrecisionOnly,
+                                 AcceptDoublePrecisionOnly);
+
   // Apply mnemonic aliases before doing anything else, as the destination
-  // mnemnonic may include suffices and we want to handle them normally.
+  // mnemonic may include suffices and we want to handle them normally.
   // The generic tblgen'erated code does this later, at the start of
   // MatchInstructionImpl(), but that's too late for aliases that include
   // any sort of suffix.
@@ -5141,6 +5432,7 @@ bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
     // For for ARM mode generate an error if the .n qualifier is used.
     if (ExtraToken == ".n" && !isThumb()) {
       SMLoc Loc = SMLoc::getFromPointer(NameLoc.getPointer() + Start);
+      Parser.eatToEndOfStatement();
       return Error(Loc, "instruction with .n (narrow) qualifier not allowed in "
                    "arm mode");
     }
@@ -5181,6 +5473,16 @@ bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
 
   Parser.Lex(); // Consume the EndOfStatement
 
+  if (RequireVFPRegisterListCheck) {
+    ARMOperand &Op = static_cast<ARMOperand &>(*Operands.back());
+    if (AcceptSinglePrecisionOnly && !Op.isSPRRegList())
+      return Error(Op.getStartLoc(),
+                   "VFP/Neon single precision register expected");
+    if (AcceptDoublePrecisionOnly && !Op.isDPRRegList())
+      return Error(Op.getStartLoc(),
+                   "VFP/Neon double precision register expected");
+  }
+
   // Some instructions, mostly Thumb, have forms for the same mnemonic that
   // do and don't have a cc_out optional-def operand. With some spot-checks
   // of the operand list, we can figure out which variant we're trying to
@@ -5188,20 +5490,14 @@ bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
   // try to remove a cc_out operand that was explicitly set on the the
   // mnemonic, of course (CarrySetting == true). Reason number #317 the
   // table driven matcher doesn't fit well with the ARM instruction set.
-  if (!CarrySetting && shouldOmitCCOutOperand(Mnemonic, Operands)) {
-    ARMOperand *Op = static_cast<ARMOperand*>(Operands[1]);
+  if (!CarrySetting && shouldOmitCCOutOperand(Mnemonic, Operands))
     Operands.erase(Operands.begin() + 1);
-    delete Op;
-  }
 
   // Some instructions have the same mnemonic, but don't always
   // have a predicate. Distinguish them here and delete the
   // predicate if needed.
-  if (shouldOmitPredicateOperand(Mnemonic, Operands)) {
-    ARMOperand *Op = static_cast<ARMOperand*>(Operands[1]);
+  if (shouldOmitPredicateOperand(Mnemonic, Operands))
     Operands.erase(Operands.begin() + 1);
-    delete Op;
-  }
 
   // ARM mode 'blx' need special handling, as the register operand version
   // is predicable, but the label operand version is not. So, we can't rely
@@ -5209,11 +5505,8 @@ bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
   // a k_CondCode operand in the list. If we're trying to match the label
   // version, remove the k_CondCode operand here.
   if (!isThumb() && Mnemonic == "blx" && Operands.size() == 3 &&
-      static_cast<ARMOperand*>(Operands[2])->isImm()) {
-    ARMOperand *Op = static_cast<ARMOperand*>(Operands[1]);
+      static_cast<ARMOperand &>(*Operands[2]).isImm())
     Operands.erase(Operands.begin() + 1);
-    delete Op;
-  }
 
   // Adjust operands of ldrexd/strexd to MCK_GPRPair.
   // ldrexd/strexd require even/odd GPR pair. To enforce this constraint,
@@ -5226,32 +5519,50 @@ bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
        Mnemonic == "stlexd")) {
     bool isLoad = (Mnemonic == "ldrexd" || Mnemonic == "ldaexd");
     unsigned Idx = isLoad ? 2 : 3;
-    ARMOperand* Op1 = static_cast<ARMOperand*>(Operands[Idx]);
-    ARMOperand* Op2 = static_cast<ARMOperand*>(Operands[Idx+1]);
+    ARMOperand &Op1 = static_cast<ARMOperand &>(*Operands[Idx]);
+    ARMOperand &Op2 = static_cast<ARMOperand &>(*Operands[Idx + 1]);
 
     const MCRegisterClass& MRC = MRI->getRegClass(ARM::GPRRegClassID);
     // Adjust only if Op1 and Op2 are GPRs.
-    if (Op1->isReg() && Op2->isReg() && MRC.contains(Op1->getReg()) &&
-        MRC.contains(Op2->getReg())) {
-      unsigned Reg1 = Op1->getReg();
-      unsigned Reg2 = Op2->getReg();
+    if (Op1.isReg() && Op2.isReg() && MRC.contains(Op1.getReg()) &&
+        MRC.contains(Op2.getReg())) {
+      unsigned Reg1 = Op1.getReg();
+      unsigned Reg2 = Op2.getReg();
       unsigned Rt = MRI->getEncodingValue(Reg1);
       unsigned Rt2 = MRI->getEncodingValue(Reg2);
 
       // Rt2 must be Rt + 1 and Rt must be even.
       if (Rt + 1 != Rt2 || (Rt & 1)) {
-        Error(Op2->getStartLoc(), isLoad ?
-            "destination operands must be sequential" :
-            "source operands must be sequential");
+        Error(Op2.getStartLoc(), isLoad
+                                     ? "destination operands must be sequential"
+                                     : "source operands must be sequential");
         return true;
       }
       unsigned NewReg = MRI->getMatchingSuperReg(Reg1, ARM::gsub_0,
           &(MRI->getRegClass(ARM::GPRPairRegClassID)));
-      Operands.erase(Operands.begin() + Idx, Operands.begin() + Idx + 2);
-      Operands.insert(Operands.begin() + Idx, ARMOperand::CreateReg(
-            NewReg, Op1->getStartLoc(), Op2->getEndLoc()));
-      delete Op1;
-      delete Op2;
+      Operands[Idx] =
+          ARMOperand::CreateReg(NewReg, Op1.getStartLoc(), Op2.getEndLoc());
+      Operands.erase(Operands.begin() + Idx + 1);
+    }
+  }
+
+  // GNU Assembler extension (compatibility)
+  if ((Mnemonic == "ldrd" || Mnemonic == "strd")) {
+    ARMOperand &Op2 = static_cast<ARMOperand &>(*Operands[2]);
+    ARMOperand &Op3 = static_cast<ARMOperand &>(*Operands[3]);
+    if (Op3.isMem()) {
+      assert(Op2.isReg() && "expected register argument");
+
+      unsigned SuperReg = MRI->getMatchingSuperReg(
+          Op2.getReg(), ARM::gsub_0, &MRI->getRegClass(ARM::GPRPairRegClassID));
+
+      assert(SuperReg && "expected register pair");
+
+      unsigned PairedReg = MRI->getSubReg(SuperReg, ARM::gsub_1);
+
+      Operands.insert(
+          Operands.begin() + 3,
+          ARMOperand::CreateReg(PairedReg, Op2.getStartLoc(), Op2.getEndLoc()));
     }
   }
 
@@ -5261,19 +5572,13 @@ bool ARMAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
   // so the Mnemonic is the original name "subs" and delete the predicate
   // operand so it will match the table entry.
   if (isThumbTwo() && Mnemonic == "sub" && Operands.size() == 6 &&
-      static_cast<ARMOperand*>(Operands[3])->isReg() &&
-      static_cast<ARMOperand*>(Operands[3])->getReg() == ARM::PC &&
-      static_cast<ARMOperand*>(Operands[4])->isReg() &&
-      static_cast<ARMOperand*>(Operands[4])->getReg() == ARM::LR &&
-      static_cast<ARMOperand*>(Operands[5])->isImm()) {
-    ARMOperand *Op0 = static_cast<ARMOperand*>(Operands[0]);
-    Operands.erase(Operands.begin());
-    delete Op0;
-    Operands.insert(Operands.begin(), ARMOperand::CreateToken(Name, NameLoc));
-
-    ARMOperand *Op1 = static_cast<ARMOperand*>(Operands[1]);
+      static_cast<ARMOperand &>(*Operands[3]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[3]).getReg() == ARM::PC &&
+      static_cast<ARMOperand &>(*Operands[4]).isReg() &&
+      static_cast<ARMOperand &>(*Operands[4]).getReg() == ARM::LR &&
+      static_cast<ARMOperand &>(*Operands[5]).isImm()) {
+    Operands.front() = ARMOperand::CreateToken(Name, NameLoc);
     Operands.erase(Operands.begin() + 1);
-    delete Op1;
   }
   return false;
 }
@@ -5319,9 +5624,8 @@ static bool instIsBreakpoint(const MCInst &Inst) {
 }
 
 // FIXME: We would really like to be able to tablegen'erate this.
-bool ARMAsmParser::
-validateInstruction(MCInst &Inst,
-                    const SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+bool ARMAsmParser::validateInstruction(MCInst &Inst,
+                                       const OperandVector &Operands) {
   const MCInstrDesc &MCID = MII.get(Inst.getOpcode());
   SMLoc Loc = Operands[0]->getStartLoc();
 
@@ -5344,7 +5648,7 @@ validateInstruction(MCInst &Inst,
       // Find the condition code Operand to get its SMLoc information.
       SMLoc CondLoc;
       for (unsigned I = 1; I < Operands.size(); ++I)
-        if (static_cast<ARMOperand*>(Operands[I])->isCondCode())
+        if (static_cast<ARMOperand &>(*Operands[I]).isCondCode())
           CondLoc = Operands[I]->getStartLoc();
       return Error(CondLoc, "incorrect condition in IT block; got '" +
                    StringRef(ARMCondCodeToString(ARMCC::CondCodes(Cond))) +
@@ -5444,8 +5748,8 @@ validateInstruction(MCInst &Inst,
     // in the register list.
     unsigned Rn = Inst.getOperand(0).getReg();
     bool HasWritebackToken =
-      (static_cast<ARMOperand*>(Operands[3])->isToken() &&
-       static_cast<ARMOperand*>(Operands[3])->getToken() == "!");
+        (static_cast<ARMOperand &>(*Operands[3]).isToken() &&
+         static_cast<ARMOperand &>(*Operands[3]).getToken() == "!");
     bool ListContainsBase;
     if (checkLowRegisterList(Inst, 3, Rn, 0, ListContainsBase) && !isThumbTwo())
       return Error(Operands[3 + HasWritebackToken]->getStartLoc(),
@@ -5496,7 +5800,6 @@ validateInstruction(MCInst &Inst,
   case ARM::sysSTMIB_UPD:
     return Error(Operands[2]->getStartLoc(),
                  "system STM cannot have writeback register");
-    break;
   case ARM::tMUL: {
     // The second source operand must be the same register as the destination
     // operand.
@@ -5506,11 +5809,10 @@ validateInstruction(MCInst &Inst,
     // this first statement is always true for the new Inst.  Essentially, the
     // destination is unconditionally copied into the second source operand
     // without checking to see if it matches what we actually parsed.
-    if (Operands.size() == 6 &&
-        (((ARMOperand*)Operands[3])->getReg() !=
-         ((ARMOperand*)Operands[5])->getReg()) &&
-        (((ARMOperand*)Operands[3])->getReg() !=
-         ((ARMOperand*)Operands[4])->getReg())) {
+    if (Operands.size() == 6 && (((ARMOperand &)*Operands[3]).getReg() !=
+                                 ((ARMOperand &)*Operands[5]).getReg()) &&
+        (((ARMOperand &)*Operands[3]).getReg() !=
+         ((ARMOperand &)*Operands[4]).getReg())) {
       return Error(Operands[3]->getStartLoc(),
                    "destination register must match source register");
     }
@@ -5563,26 +5865,50 @@ validateInstruction(MCInst &Inst,
   }
   // Final range checking for Thumb unconditional branch instructions.
   case ARM::tB:
-    if (!(static_cast<ARMOperand*>(Operands[2]))->isSignedOffset<11, 1>())
+    if (!(static_cast<ARMOperand &>(*Operands[2])).isSignedOffset<11, 1>())
       return Error(Operands[2]->getStartLoc(), "branch target out of range");
     break;
   case ARM::t2B: {
     int op = (Operands[2]->isImm()) ? 2 : 3;
-    if (!(static_cast<ARMOperand*>(Operands[op]))->isSignedOffset<24, 1>())
+    if (!static_cast<ARMOperand &>(*Operands[op]).isSignedOffset<24, 1>())
       return Error(Operands[op]->getStartLoc(), "branch target out of range");
     break;
   }
   // Final range checking for Thumb conditional branch instructions.
   case ARM::tBcc:
-    if (!(static_cast<ARMOperand*>(Operands[2]))->isSignedOffset<8, 1>())
+    if (!static_cast<ARMOperand &>(*Operands[2]).isSignedOffset<8, 1>())
       return Error(Operands[2]->getStartLoc(), "branch target out of range");
     break;
   case ARM::t2Bcc: {
     int Op = (Operands[2]->isImm()) ? 2 : 3;
-    if (!(static_cast<ARMOperand*>(Operands[Op]))->isSignedOffset<20, 1>())
+    if (!static_cast<ARMOperand &>(*Operands[Op]).isSignedOffset<20, 1>())
       return Error(Operands[Op]->getStartLoc(), "branch target out of range");
     break;
   }
+  case ARM::MOVi16:
+  case ARM::t2MOVi16:
+  case ARM::t2MOVTi16:
+    {
+    // We want to avoid misleadingly allowing something like "mov r0, <symbol>"
+    // especially when we turn it into a movw and the expression <symbol> does
+    // not have a :lower16: or :upper16 as part of the expression.  We don't
+    // want the behavior of silently truncating, which can be unexpected and
+    // lead to bugs that are difficult to find since this is an easy mistake
+    // to make.
+    int i = (Operands[3]->isImm()) ? 3 : 4;
+    ARMOperand &Op = static_cast<ARMOperand &>(*Operands[i]);
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Op.getImm());
+    if (CE) break;
+    const MCExpr *E = dyn_cast<MCExpr>(Op.getImm());
+    if (!E) break;
+    const ARMMCExpr *ARM16Expr = dyn_cast<ARMMCExpr>(E);
+    if (!ARM16Expr || (ARM16Expr->getKind() != ARMMCExpr::VK_ARM_HI16 &&
+                       ARM16Expr->getKind() != ARMMCExpr::VK_ARM_LO16))
+      return Error(
+          Op.getStartLoc(),
+          "immediate expression for mov requires :lower16: or :upper16");
+    break;
+  }
   }
 
   return false;
@@ -5733,7 +6059,7 @@ static unsigned getRealVLDOpcode(unsigned Opc, unsigned &Spacing) {
   case ARM::VLD3DUPdWB_fixed_Asm_16: Spacing = 1; return ARM::VLD3DUPd16_UPD;
   case ARM::VLD3DUPdWB_fixed_Asm_32: Spacing = 1; return ARM::VLD3DUPd32_UPD;
   case ARM::VLD3DUPqWB_fixed_Asm_8: Spacing = 1; return ARM::VLD3DUPq8_UPD;
-  case ARM::VLD3DUPqWB_fixed_Asm_16: Spacing = 1; return ARM::VLD3DUPq16_UPD;
+  case ARM::VLD3DUPqWB_fixed_Asm_16: Spacing = 2; return ARM::VLD3DUPq16_UPD;
   case ARM::VLD3DUPqWB_fixed_Asm_32: Spacing = 2; return ARM::VLD3DUPq32_UPD;
   case ARM::VLD3DUPdWB_register_Asm_8:  Spacing = 1; return ARM::VLD3DUPd8_UPD;
   case ARM::VLD3DUPdWB_register_Asm_16: Spacing = 1; return ARM::VLD3DUPd16_UPD;
@@ -5789,7 +6115,7 @@ static unsigned getRealVLDOpcode(unsigned Opc, unsigned &Spacing) {
   case ARM::VLD4LNdWB_fixed_Asm_8:  Spacing = 1; return ARM::VLD4LNd8_UPD;
   case ARM::VLD4LNdWB_fixed_Asm_16: Spacing = 1; return ARM::VLD4LNd16_UPD;
   case ARM::VLD4LNdWB_fixed_Asm_32: Spacing = 1; return ARM::VLD4LNd32_UPD;
-  case ARM::VLD4LNqWB_fixed_Asm_16: Spacing = 1; return ARM::VLD4LNq16_UPD;
+  case ARM::VLD4LNqWB_fixed_Asm_16: Spacing = 2; return ARM::VLD4LNq16_UPD;
   case ARM::VLD4LNqWB_fixed_Asm_32: Spacing = 2; return ARM::VLD4LNq32_UPD;
   case ARM::VLD4LNdWB_register_Asm_8:  Spacing = 1; return ARM::VLD4LNd8_UPD;
   case ARM::VLD4LNdWB_register_Asm_16: Spacing = 1; return ARM::VLD4LNd16_UPD;
@@ -5844,10 +6170,45 @@ static unsigned getRealVLDOpcode(unsigned Opc, unsigned &Spacing) {
   }
 }
 
-bool ARMAsmParser::
-processInstruction(MCInst &Inst,
-                   const SmallVectorImpl<MCParsedAsmOperand*> &Operands) {
+bool ARMAsmParser::processInstruction(MCInst &Inst,
+                                      const OperandVector &Operands) {
   switch (Inst.getOpcode()) {
+  // Alias for alternate form of 'ldr{,b}t Rt, [Rn], #imm' instruction.
+  case ARM::LDRT_POST:
+  case ARM::LDRBT_POST: {
+    const unsigned Opcode =
+      (Inst.getOpcode() == ARM::LDRT_POST) ? ARM::LDRT_POST_IMM
+                                           : ARM::LDRBT_POST_IMM;
+    MCInst TmpInst;
+    TmpInst.setOpcode(Opcode);
+    TmpInst.addOperand(Inst.getOperand(0));
+    TmpInst.addOperand(Inst.getOperand(1));
+    TmpInst.addOperand(Inst.getOperand(1));
+    TmpInst.addOperand(MCOperand::CreateReg(0));
+    TmpInst.addOperand(MCOperand::CreateImm(0));
+    TmpInst.addOperand(Inst.getOperand(2));
+    TmpInst.addOperand(Inst.getOperand(3));
+    Inst = TmpInst;
+    return true;
+  }
+  // Alias for alternate form of 'str{,b}t Rt, [Rn], #imm' instruction.
+  case ARM::STRT_POST:
+  case ARM::STRBT_POST: {
+    const unsigned Opcode =
+      (Inst.getOpcode() == ARM::STRT_POST) ? ARM::STRT_POST_IMM
+                                           : ARM::STRBT_POST_IMM;
+    MCInst TmpInst;
+    TmpInst.setOpcode(Opcode);
+    TmpInst.addOperand(Inst.getOperand(1));
+    TmpInst.addOperand(Inst.getOperand(0));
+    TmpInst.addOperand(Inst.getOperand(1));
+    TmpInst.addOperand(MCOperand::CreateReg(0));
+    TmpInst.addOperand(MCOperand::CreateImm(0));
+    TmpInst.addOperand(Inst.getOperand(2));
+    TmpInst.addOperand(Inst.getOperand(3));
+    Inst = TmpInst;
+    return true;
+  }
   // Alias for alternate form of 'ADR Rd, #imm' instruction.
   case ARM::ADDri: {
     if (Inst.getOperand(1).getReg() != ARM::PC ||
@@ -5867,8 +6228,8 @@ processInstruction(MCInst &Inst,
     // Select the narrow version if the immediate will fit.
     if (Inst.getOperand(1).getImm() > 0 &&
         Inst.getOperand(1).getImm() <= 0xff &&
-        !(static_cast<ARMOperand*>(Operands[2])->isToken() &&
-         static_cast<ARMOperand*>(Operands[2])->getToken() == ".w"))
+        !(static_cast<ARMOperand &>(*Operands[2]).isToken() &&
+          static_cast<ARMOperand &>(*Operands[2]).getToken() == ".w"))
       Inst.setOpcode(ARM::tLDRpci);
     else
       Inst.setOpcode(ARM::t2LDRpci);
@@ -6958,8 +7319,8 @@ processInstruction(MCInst &Inst,
     if (isARMLowRegister(Inst.getOperand(0).getReg()) &&
         Inst.getOperand(0).getReg() == Inst.getOperand(1).getReg() &&
         Inst.getOperand(5).getReg() == (inITBlock() ? 0 : ARM::CPSR) &&
-        !(static_cast<ARMOperand*>(Operands[3])->isToken() &&
-         static_cast<ARMOperand*>(Operands[3])->getToken() == ".w")) {
+        !(static_cast<ARMOperand &>(*Operands[3]).isToken() &&
+          static_cast<ARMOperand &>(*Operands[3]).getToken() == ".w")) {
       unsigned NewOpc;
       switch (Inst.getOpcode()) {
       default: llvm_unreachable("unexpected opcode");
@@ -7162,7 +7523,7 @@ processInstruction(MCInst &Inst,
   case ARM::LDMIA_UPD:
     // If this is a load of a single register via a 'pop', then we should use
     // a post-indexed LDR instruction instead, per the ARM ARM.
-    if (static_cast<ARMOperand*>(Operands[0])->getToken() == "pop" &&
+    if (static_cast<ARMOperand &>(*Operands[0]).getToken() == "pop" &&
         Inst.getNumOperands() == 5) {
       MCInst TmpInst;
       TmpInst.setOpcode(ARM::LDR_POST_IMM);
@@ -7180,7 +7541,7 @@ processInstruction(MCInst &Inst,
   case ARM::STMDB_UPD:
     // If this is a store of a single register via a 'push', then we should use
     // a pre-indexed STR instruction instead, per the ARM ARM.
-    if (static_cast<ARMOperand*>(Operands[0])->getToken() == "push" &&
+    if (static_cast<ARMOperand &>(*Operands[0]).getToken() == "push" &&
         Inst.getNumOperands() == 5) {
       MCInst TmpInst;
       TmpInst.setOpcode(ARM::STR_PRE_IMM);
@@ -7196,7 +7557,7 @@ processInstruction(MCInst &Inst,
   case ARM::t2ADDri12:
     // If the immediate fits for encoding T3 (t2ADDri) and the generic "add"
     // mnemonic was used (not "addw"), encoding T3 is preferred.
-    if (static_cast<ARMOperand*>(Operands[0])->getToken() != "add" ||
+    if (static_cast<ARMOperand &>(*Operands[0]).getToken() != "add" ||
         ARM_AM::getT2SOImmVal(Inst.getOperand(2).getImm()) == -1)
       break;
     Inst.setOpcode(ARM::t2ADDri);
@@ -7205,7 +7566,7 @@ processInstruction(MCInst &Inst,
   case ARM::t2SUBri12:
     // If the immediate fits for encoding T3 (t2SUBri) and the generic "sub"
     // mnemonic was used (not "subw"), encoding T3 is preferred.
-    if (static_cast<ARMOperand*>(Operands[0])->getToken() != "sub" ||
+    if (static_cast<ARMOperand &>(*Operands[0]).getToken() != "sub" ||
         ARM_AM::getT2SOImmVal(Inst.getOperand(2).getImm()) == -1)
       break;
     Inst.setOpcode(ARM::t2SUBri);
@@ -7241,9 +7602,9 @@ processInstruction(MCInst &Inst,
         !isARMLowRegister(Inst.getOperand(0).getReg()) ||
         (unsigned)Inst.getOperand(2).getImm() > 255 ||
         ((!inITBlock() && Inst.getOperand(5).getReg() != ARM::CPSR) ||
-        (inITBlock() && Inst.getOperand(5).getReg() != 0)) ||
-        (static_cast<ARMOperand*>(Operands[3])->isToken() &&
-         static_cast<ARMOperand*>(Operands[3])->getToken() == ".w"))
+         (inITBlock() && Inst.getOperand(5).getReg() != 0)) ||
+        (static_cast<ARMOperand &>(*Operands[3]).isToken() &&
+         static_cast<ARMOperand &>(*Operands[3]).getToken() == ".w"))
       break;
     MCInst TmpInst;
     TmpInst.setOpcode(Inst.getOpcode() == ARM::t2ADDri ?
@@ -7264,8 +7625,8 @@ processInstruction(MCInst &Inst,
     // 'as' behaviour. Make sure the wide encoding wasn't explicit.
     if (Inst.getOperand(0).getReg() != Inst.getOperand(1).getReg() ||
         Inst.getOperand(5).getReg() != 0 ||
-        (static_cast<ARMOperand*>(Operands[3])->isToken() &&
-         static_cast<ARMOperand*>(Operands[3])->getToken() == ".w"))
+        (static_cast<ARMOperand &>(*Operands[3]).isToken() &&
+         static_cast<ARMOperand &>(*Operands[3]).getToken() == ".w"))
       break;
     MCInst TmpInst;
     TmpInst.setOpcode(ARM::tADDhirr);
@@ -7322,8 +7683,8 @@ processInstruction(MCInst &Inst,
     // an error in validateInstruction().
     unsigned Rn = Inst.getOperand(0).getReg();
     bool hasWritebackToken =
-      (static_cast<ARMOperand*>(Operands[3])->isToken() &&
-       static_cast<ARMOperand*>(Operands[3])->getToken() == "!");
+        (static_cast<ARMOperand &>(*Operands[3]).isToken() &&
+         static_cast<ARMOperand &>(*Operands[3]).getToken() == "!");
     bool listContainsBase;
     if (checkLowRegisterList(Inst, 3, Rn, 0, listContainsBase) ||
         (!listContainsBase && !hasWritebackToken) ||
@@ -7385,10 +7746,10 @@ processInstruction(MCInst &Inst,
     if (isARMLowRegister(Inst.getOperand(0).getReg()) &&
         (unsigned)Inst.getOperand(1).getImm() <= 255 &&
         ((!inITBlock() && Inst.getOperand(2).getImm() == ARMCC::AL &&
-         Inst.getOperand(4).getReg() == ARM::CPSR) ||
-        (inITBlock() && Inst.getOperand(4).getReg() == 0)) &&
-        (!static_cast<ARMOperand*>(Operands[2])->isToken() ||
-         static_cast<ARMOperand*>(Operands[2])->getToken() != ".w")) {
+          Inst.getOperand(4).getReg() == ARM::CPSR) ||
+         (inITBlock() && Inst.getOperand(4).getReg() == 0)) &&
+        (!static_cast<ARMOperand &>(*Operands[2]).isToken() ||
+         static_cast<ARMOperand &>(*Operands[2]).getToken() != ".w")) {
       // The operands aren't in the same order for tMOVi8...
       MCInst TmpInst;
       TmpInst.setOpcode(ARM::tMOVi8);
@@ -7409,8 +7770,8 @@ processInstruction(MCInst &Inst,
         isARMLowRegister(Inst.getOperand(1).getReg()) &&
         Inst.getOperand(2).getImm() == ARMCC::AL &&
         Inst.getOperand(4).getReg() == ARM::CPSR &&
-        (!static_cast<ARMOperand*>(Operands[2])->isToken() ||
-         static_cast<ARMOperand*>(Operands[2])->getToken() != ".w")) {
+        (!static_cast<ARMOperand &>(*Operands[2]).isToken() ||
+         static_cast<ARMOperand &>(*Operands[2]).getToken() != ".w")) {
       // The operands aren't the same for tMOV[S]r... (no cc_out)
       MCInst TmpInst;
       TmpInst.setOpcode(Inst.getOperand(4).getReg() ? ARM::tMOVSr : ARM::tMOVr);
@@ -7432,8 +7793,8 @@ processInstruction(MCInst &Inst,
     if (isARMLowRegister(Inst.getOperand(0).getReg()) &&
         isARMLowRegister(Inst.getOperand(1).getReg()) &&
         Inst.getOperand(2).getImm() == 0 &&
-        (!static_cast<ARMOperand*>(Operands[2])->isToken() ||
-         static_cast<ARMOperand*>(Operands[2])->getToken() != ".w")) {
+        (!static_cast<ARMOperand &>(*Operands[2]).isToken() ||
+         static_cast<ARMOperand &>(*Operands[2]).getToken() != ".w")) {
       unsigned NewOpc;
       switch (Inst.getOpcode()) {
       default: llvm_unreachable("Illegal opcode!");
@@ -7545,9 +7906,10 @@ processInstruction(MCInst &Inst,
          isARMLowRegister(Inst.getOperand(2).getReg())) &&
         Inst.getOperand(0).getReg() == Inst.getOperand(1).getReg() &&
         ((!inITBlock() && Inst.getOperand(5).getReg() == ARM::CPSR) ||
-         (inITBlock() && Inst.getOperand(5).getReg() != ARM::CPSR)) && 
-        (!static_cast<ARMOperand*>(Operands[3])->isToken() ||
-         !static_cast<ARMOperand*>(Operands[3])->getToken().equals_lower(".w"))) {
+         (inITBlock() && Inst.getOperand(5).getReg() != ARM::CPSR)) &&
+        (!static_cast<ARMOperand &>(*Operands[3]).isToken() ||
+         !static_cast<ARMOperand &>(*Operands[3]).getToken().equals_lower(
+             ".w"))) {
       unsigned NewOpc;
       switch (Inst.getOpcode()) {
         default: llvm_unreachable("unexpected opcode");
@@ -7584,9 +7946,10 @@ processInstruction(MCInst &Inst,
         (Inst.getOperand(0).getReg() == Inst.getOperand(1).getReg() ||
          Inst.getOperand(0).getReg() == Inst.getOperand(2).getReg()) &&
         ((!inITBlock() && Inst.getOperand(5).getReg() == ARM::CPSR) ||
-         (inITBlock() && Inst.getOperand(5).getReg() != ARM::CPSR)) && 
-        (!static_cast<ARMOperand*>(Operands[3])->isToken() ||
-         !static_cast<ARMOperand*>(Operands[3])->getToken().equals_lower(".w"))) {
+         (inITBlock() && Inst.getOperand(5).getReg() != ARM::CPSR)) &&
+        (!static_cast<ARMOperand &>(*Operands[3]).isToken() ||
+         !static_cast<ARMOperand &>(*Operands[3]).getToken().equals_lower(
+             ".w"))) {
       unsigned NewOpc;
       switch (Inst.getOpcode()) {
         default: llvm_unreachable("unexpected opcode");
@@ -7659,12 +8022,17 @@ unsigned ARMAsmParser::checkTargetMatchPredicate(MCInst &Inst) {
   return Match_Success;
 }
 
+namespace llvm {
+template <> inline bool IsCPSRDead<MCInst>(MCInst *Instr) {
+  return true; // In an assembly source, no need to second-guess
+}
+}
+
 static const char *getSubtargetFeatureName(unsigned Val);
-bool ARMAsmParser::
-MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
-                        SmallVectorImpl<MCParsedAsmOperand*> &Operands,
-                        MCStreamer &Out, unsigned &ErrorInfo,
-                        bool MatchingInlineAsm) {
+bool ARMAsmParser::MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
+                                           OperandVector &Operands,
+                                           MCStreamer &Out, unsigned &ErrorInfo,
+                                           bool MatchingInlineAsm) {
   MCInst Inst;
   unsigned MatchResult;
 
@@ -7694,7 +8062,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
 
       // Only after the instruction is fully processed, we can validate it
       if (wasInITBlock && hasV8Ops() && isThumb() &&
-          !isV8EligibleForIT(&Inst, 2)) {
+          !isV8EligibleForIT(&Inst)) {
         Warning(IDLoc, "deprecated instruction in IT block");
       }
     }
@@ -7710,7 +8078,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
       return false;
 
     Inst.setLoc(IDLoc);
-    Out.EmitInstruction(Inst);
+    Out.EmitInstruction(Inst, STI);
     return false;
   case Match_MissingFeature: {
     assert(ErrorInfo && "Unknown missing feature!");
@@ -7733,7 +8101,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
       if (ErrorInfo >= Operands.size())
         return Error(IDLoc, "too few operands for instruction");
 
-      ErrorLoc = ((ARMOperand*)Operands[ErrorInfo])->getStartLoc();
+      ErrorLoc = ((ARMOperand &)*Operands[ErrorInfo]).getStartLoc();
       if (ErrorLoc == SMLoc()) ErrorLoc = IDLoc;
     }
 
@@ -7741,7 +8109,7 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
   }
   case Match_MnemonicFail:
     return Error(IDLoc, "invalid instruction",
-                 ((ARMOperand*)Operands[0])->getLocRange());
+                 ((ARMOperand &)*Operands[0]).getLocRange());
   case Match_RequiresNotITBlock:
     return Error(IDLoc, "flag setting instruction only valid outside IT block");
   case Match_RequiresITBlock:
@@ -7751,15 +8119,51 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
   case Match_RequiresThumb2:
     return Error(IDLoc, "instruction variant requires Thumb2");
   case Match_ImmRange0_15: {
-    SMLoc ErrorLoc = ((ARMOperand*)Operands[ErrorInfo])->getStartLoc();
+    SMLoc ErrorLoc = ((ARMOperand &)*Operands[ErrorInfo]).getStartLoc();
     if (ErrorLoc == SMLoc()) ErrorLoc = IDLoc;
     return Error(ErrorLoc, "immediate operand must be in the range [0,15]");
   }
   case Match_ImmRange0_239: {
-    SMLoc ErrorLoc = ((ARMOperand*)Operands[ErrorInfo])->getStartLoc();
+    SMLoc ErrorLoc = ((ARMOperand &)*Operands[ErrorInfo]).getStartLoc();
     if (ErrorLoc == SMLoc()) ErrorLoc = IDLoc;
     return Error(ErrorLoc, "immediate operand must be in the range [0,239]");
   }
+  case Match_AlignedMemoryRequiresNone:
+  case Match_DupAlignedMemoryRequiresNone:
+  case Match_AlignedMemoryRequires16:
+  case Match_DupAlignedMemoryRequires16:
+  case Match_AlignedMemoryRequires32:
+  case Match_DupAlignedMemoryRequires32:
+  case Match_AlignedMemoryRequires64:
+  case Match_DupAlignedMemoryRequires64:
+  case Match_AlignedMemoryRequires64or128:
+  case Match_DupAlignedMemoryRequires64or128:
+  case Match_AlignedMemoryRequires64or128or256:
+  {
+    SMLoc ErrorLoc = ((ARMOperand &)*Operands[ErrorInfo]).getAlignmentLoc();
+    if (ErrorLoc == SMLoc()) ErrorLoc = IDLoc;
+    switch (MatchResult) {
+      default:
+        llvm_unreachable("Missing Match_Aligned type");
+      case Match_AlignedMemoryRequiresNone:
+      case Match_DupAlignedMemoryRequiresNone:
+        return Error(ErrorLoc, "alignment must be omitted");
+      case Match_AlignedMemoryRequires16:
+      case Match_DupAlignedMemoryRequires16:
+        return Error(ErrorLoc, "alignment must be 16 or omitted");
+      case Match_AlignedMemoryRequires32:
+      case Match_DupAlignedMemoryRequires32:
+        return Error(ErrorLoc, "alignment must be 32 or omitted");
+      case Match_AlignedMemoryRequires64:
+      case Match_DupAlignedMemoryRequires64:
+        return Error(ErrorLoc, "alignment must be 64 or omitted");
+      case Match_AlignedMemoryRequires64or128:
+      case Match_DupAlignedMemoryRequires64or128:
+        return Error(ErrorLoc, "alignment must be 64, 128 or omitted");
+      case Match_AlignedMemoryRequires64or128or256:
+        return Error(ErrorLoc, "alignment must be 64, 128, 256 or omitted");
+    }
+  }
   }
 
   llvm_unreachable("Implement any new match types added!");
@@ -7767,9 +8171,15 @@ MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
 
 /// parseDirective parses the arm specific directives
 bool ARMAsmParser::ParseDirective(AsmToken DirectiveID) {
+  const MCObjectFileInfo::Environment Format =
+    getContext().getObjectFileInfo()->getObjectFileType();
+  bool IsMachO = Format == MCObjectFileInfo::IsMachO;
+
   StringRef IDVal = DirectiveID.getIdentifier();
   if (IDVal == ".word")
-    return parseDirectiveWord(4, DirectiveID.getLoc());
+    return parseLiteralValues(4, DirectiveID.getLoc());
+  else if (IDVal == ".short" || IDVal == ".hword")
+    return parseLiteralValues(2, DirectiveID.getLoc());
   else if (IDVal == ".thumb")
     return parseDirectiveThumb(DirectiveID.getLoc());
   else if (IDVal == ".arm")
@@ -7782,16 +8192,6 @@ bool ARMAsmParser::ParseDirective(AsmToken DirectiveID) {
     return parseDirectiveSyntax(DirectiveID.getLoc());
   else if (IDVal == ".unreq")
     return parseDirectiveUnreq(DirectiveID.getLoc());
-  else if (IDVal == ".arch")
-    return parseDirectiveArch(DirectiveID.getLoc());
-  else if (IDVal == ".eabi_attribute")
-    return parseDirectiveEabiAttr(DirectiveID.getLoc());
-  else if (IDVal == ".cpu")
-    return parseDirectiveCPU(DirectiveID.getLoc());
-  else if (IDVal == ".fpu")
-    return parseDirectiveFPU(DirectiveID.getLoc());
-  else if (IDVal == ".fnstart")
-    return parseDirectiveFnStart(DirectiveID.getLoc());
   else if (IDVal == ".fnend")
     return parseDirectiveFnEnd(DirectiveID.getLoc());
   else if (IDVal == ".cantunwind")
@@ -7808,17 +8208,61 @@ bool ARMAsmParser::ParseDirective(AsmToken DirectiveID) {
     return parseDirectiveRegSave(DirectiveID.getLoc(), false);
   else if (IDVal == ".vsave")
     return parseDirectiveRegSave(DirectiveID.getLoc(), true);
+  else if (IDVal == ".ltorg" || IDVal == ".pool")
+    return parseDirectiveLtorg(DirectiveID.getLoc());
+  else if (IDVal == ".even")
+    return parseDirectiveEven(DirectiveID.getLoc());
+  else if (IDVal == ".personalityindex")
+    return parseDirectivePersonalityIndex(DirectiveID.getLoc());
+  else if (IDVal == ".unwind_raw")
+    return parseDirectiveUnwindRaw(DirectiveID.getLoc());
+  else if (IDVal == ".movsp")
+    return parseDirectiveMovSP(DirectiveID.getLoc());
+  else if (IDVal == ".arch_extension")
+    return parseDirectiveArchExtension(DirectiveID.getLoc());
+  else if (IDVal == ".align")
+    return parseDirectiveAlign(DirectiveID.getLoc());
+  else if (IDVal == ".thumb_set")
+    return parseDirectiveThumbSet(DirectiveID.getLoc());
+
+  if (!IsMachO) {
+    if (IDVal == ".arch")
+      return parseDirectiveArch(DirectiveID.getLoc());
+    else if (IDVal == ".cpu")
+      return parseDirectiveCPU(DirectiveID.getLoc());
+    else if (IDVal == ".eabi_attribute")
+      return parseDirectiveEabiAttr(DirectiveID.getLoc());
+    else if (IDVal == ".fpu")
+      return parseDirectiveFPU(DirectiveID.getLoc());
+    else if (IDVal == ".fnstart")
+      return parseDirectiveFnStart(DirectiveID.getLoc());
+    else if (IDVal == ".inst")
+      return parseDirectiveInst(DirectiveID.getLoc());
+    else if (IDVal == ".inst.n")
+      return parseDirectiveInst(DirectiveID.getLoc(), 'n');
+    else if (IDVal == ".inst.w")
+      return parseDirectiveInst(DirectiveID.getLoc(), 'w');
+    else if (IDVal == ".object_arch")
+      return parseDirectiveObjectArch(DirectiveID.getLoc());
+    else if (IDVal == ".tlsdescseq")
+      return parseDirectiveTLSDescSeq(DirectiveID.getLoc());
+  }
+
   return true;
 }
 
-/// parseDirectiveWord
-///  ::= .word [ expression (, expression)* ]
-bool ARMAsmParser::parseDirectiveWord(unsigned Size, SMLoc L) {
+/// parseLiteralValues
+///  ::= .hword expression [, expression]*
+///  ::= .short expression [, expression]*
+///  ::= .word expression [, expression]*
+bool ARMAsmParser::parseLiteralValues(unsigned Size, SMLoc L) {
   if (getLexer().isNot(AsmToken::EndOfStatement)) {
     for (;;) {
       const MCExpr *Value;
-      if (getParser().parseExpression(Value))
-        return true;
+      if (getParser().parseExpression(Value)) {
+        Parser.eatToEndOfStatement();
+        return false;
+      }
 
       getParser().getStreamer().EmitValue(Value, Size);
 
@@ -7826,8 +8270,10 @@ bool ARMAsmParser::parseDirectiveWord(unsigned Size, SMLoc L) {
         break;
 
       // FIXME: Improve diagnostic.
-      if (getLexer().isNot(AsmToken::Comma))
-        return Error(L, "unexpected token in directive");
+      if (getLexer().isNot(AsmToken::Comma)) {
+        Error(L, "unexpected token in directive");
+        return false;
+      }
       Parser.Lex();
     }
   }
@@ -7839,15 +8285,20 @@ bool ARMAsmParser::parseDirectiveWord(unsigned Size, SMLoc L) {
 /// parseDirectiveThumb
 ///  ::= .thumb
 bool ARMAsmParser::parseDirectiveThumb(SMLoc L) {
-  if (getLexer().isNot(AsmToken::EndOfStatement))
-    return Error(L, "unexpected token in directive");
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(L, "unexpected token in directive");
+    return false;
+  }
   Parser.Lex();
 
-  if (!hasThumb())
-    return Error(L, "target does not support Thumb mode");
+  if (!hasThumb()) {
+    Error(L, "target does not support Thumb mode");
+    return false;
+  }
 
   if (!isThumb())
     SwitchMode();
+
   getParser().getStreamer().EmitAssemblerFlag(MCAF_Code16);
   return false;
 }
@@ -7855,15 +8306,20 @@ bool ARMAsmParser::parseDirectiveThumb(SMLoc L) {
 /// parseDirectiveARM
 ///  ::= .arm
 bool ARMAsmParser::parseDirectiveARM(SMLoc L) {
-  if (getLexer().isNot(AsmToken::EndOfStatement))
-    return Error(L, "unexpected token in directive");
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(L, "unexpected token in directive");
+    return false;
+  }
   Parser.Lex();
 
-  if (!hasARM())
-    return Error(L, "target does not support ARM mode");
+  if (!hasARM()) {
+    Error(L, "target does not support ARM mode");
+    return false;
+  }
 
   if (isThumb())
     SwitchMode();
+
   getParser().getStreamer().EmitAssemblerFlag(MCAF_Code32);
   return false;
 }
@@ -7886,8 +8342,11 @@ bool ARMAsmParser::parseDirectiveThumbFunc(SMLoc L) {
   if (isMachO) {
     const AsmToken &Tok = Parser.getTok();
     if (Tok.isNot(AsmToken::EndOfStatement)) {
-      if (Tok.isNot(AsmToken::Identifier) && Tok.isNot(AsmToken::String))
-        return Error(L, "unexpected token in .thumb_func directive");
+      if (Tok.isNot(AsmToken::Identifier) && Tok.isNot(AsmToken::String)) {
+        Error(L, "unexpected token in .thumb_func directive");
+        return false;
+      }
+
       MCSymbol *Func =
           getParser().getContext().GetOrCreateSymbol(Tok.getIdentifier());
       getParser().getStreamer().EmitThumbFunc(Func);
@@ -7896,11 +8355,12 @@ bool ARMAsmParser::parseDirectiveThumbFunc(SMLoc L) {
     }
   }
 
-  if (getLexer().isNot(AsmToken::EndOfStatement))
-    return Error(L, "unexpected token in directive");
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(L, "unexpected token in directive");
+    return false;
+  }
 
   NextSymbolIsThumb = true;
-
   return false;
 }
 
@@ -7908,18 +8368,26 @@ bool ARMAsmParser::parseDirectiveThumbFunc(SMLoc L) {
 ///  ::= .syntax unified | divided
 bool ARMAsmParser::parseDirectiveSyntax(SMLoc L) {
   const AsmToken &Tok = Parser.getTok();
-  if (Tok.isNot(AsmToken::Identifier))
-    return Error(L, "unexpected token in .syntax directive");
+  if (Tok.isNot(AsmToken::Identifier)) {
+    Error(L, "unexpected token in .syntax directive");
+    return false;
+  }
+
   StringRef Mode = Tok.getString();
-  if (Mode == "unified" || Mode == "UNIFIED")
+  if (Mode == "unified" || Mode == "UNIFIED") {
     Parser.Lex();
-  else if (Mode == "divided" || Mode == "DIVIDED")
-    return Error(L, "'.syntax divided' arm asssembly not supported");
-  else
-    return Error(L, "unrecognized syntax mode in .syntax directive");
+  } else if (Mode == "divided" || Mode == "DIVIDED") {
+    Error(L, "'.syntax divided' arm asssembly not supported");
+    return false;
+  } else {
+    Error(L, "unrecognized syntax mode in .syntax directive");
+    return false;
+  }
 
-  if (getLexer().isNot(AsmToken::EndOfStatement))
-    return Error(Parser.getTok().getLoc(), "unexpected token in directive");
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(Parser.getTok().getLoc(), "unexpected token in directive");
+    return false;
+  }
   Parser.Lex();
 
   // TODO tell the MC streamer the mode
@@ -7931,30 +8399,37 @@ bool ARMAsmParser::parseDirectiveSyntax(SMLoc L) {
 ///  ::= .code 16 | 32
 bool ARMAsmParser::parseDirectiveCode(SMLoc L) {
   const AsmToken &Tok = Parser.getTok();
-  if (Tok.isNot(AsmToken::Integer))
-    return Error(L, "unexpected token in .code directive");
+  if (Tok.isNot(AsmToken::Integer)) {
+    Error(L, "unexpected token in .code directive");
+    return false;
+  }
   int64_t Val = Parser.getTok().getIntVal();
-  if (Val == 16)
-    Parser.Lex();
-  else if (Val == 32)
-    Parser.Lex();
-  else
-    return Error(L, "invalid operand to .code directive");
+  if (Val != 16 && Val != 32) {
+    Error(L, "invalid operand to .code directive");
+    return false;
+  }
+  Parser.Lex();
 
-  if (getLexer().isNot(AsmToken::EndOfStatement))
-    return Error(Parser.getTok().getLoc(), "unexpected token in directive");
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(Parser.getTok().getLoc(), "unexpected token in directive");
+    return false;
+  }
   Parser.Lex();
 
   if (Val == 16) {
-    if (!hasThumb())
-      return Error(L, "target does not support Thumb mode");
+    if (!hasThumb()) {
+      Error(L, "target does not support Thumb mode");
+      return false;
+    }
 
     if (!isThumb())
       SwitchMode();
     getParser().getStreamer().EmitAssemblerFlag(MCAF_Code16);
   } else {
-    if (!hasARM())
-      return Error(L, "target does not support ARM mode");
+    if (!hasARM()) {
+      Error(L, "target does not support ARM mode");
+      return false;
+    }
 
     if (isThumb())
       SwitchMode();
@@ -7972,21 +8447,23 @@ bool ARMAsmParser::parseDirectiveReq(StringRef Name, SMLoc L) {
   SMLoc SRegLoc, ERegLoc;
   if (ParseRegister(Reg, SRegLoc, ERegLoc)) {
     Parser.eatToEndOfStatement();
-    return Error(SRegLoc, "register name expected");
+    Error(SRegLoc, "register name expected");
+    return false;
   }
 
   // Shouldn't be anything else.
   if (Parser.getTok().isNot(AsmToken::EndOfStatement)) {
     Parser.eatToEndOfStatement();
-    return Error(Parser.getTok().getLoc(),
-                 "unexpected input in .req directive.");
+    Error(Parser.getTok().getLoc(), "unexpected input in .req directive.");
+    return false;
   }
 
   Parser.Lex(); // Consume the EndOfStatement
 
-  if (RegisterReqs.GetOrCreateValue(Name, Reg).getValue() != Reg)
-    return Error(SRegLoc, "redefinition of '" + Name +
-                          "' does not match original.");
+  if (RegisterReqs.GetOrCreateValue(Name, Reg).getValue() != Reg) {
+    Error(SRegLoc, "redefinition of '" + Name + "' does not match original.");
+    return false;
+  }
 
   return false;
 }
@@ -7996,9 +8473,10 @@ bool ARMAsmParser::parseDirectiveReq(StringRef Name, SMLoc L) {
 bool ARMAsmParser::parseDirectiveUnreq(SMLoc L) {
   if (Parser.getTok().isNot(AsmToken::Identifier)) {
     Parser.eatToEndOfStatement();
-    return Error(L, "unexpected input in .unreq directive.");
+    Error(L, "unexpected input in .unreq directive.");
+    return false;
   }
-  RegisterReqs.erase(Parser.getTok().getIdentifier());
+  RegisterReqs.erase(Parser.getTok().getIdentifier().lower());
   Parser.Lex(); // Eat the identifier.
   return false;
 }
@@ -8006,28 +8484,128 @@ bool ARMAsmParser::parseDirectiveUnreq(SMLoc L) {
 /// parseDirectiveArch
 ///  ::= .arch token
 bool ARMAsmParser::parseDirectiveArch(SMLoc L) {
-  return true;
+  StringRef Arch = getParser().parseStringToEndOfStatement().trim();
+
+  unsigned ID = StringSwitch<unsigned>(Arch)
+#define ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH) \
+    .Case(NAME, ARM::ID)
+#define ARM_ARCH_ALIAS(NAME, ID) \
+    .Case(NAME, ARM::ID)
+#include "MCTargetDesc/ARMArchName.def"
+    .Default(ARM::INVALID_ARCH);
+
+  if (ID == ARM::INVALID_ARCH) {
+    Error(L, "Unknown arch name");
+    return false;
+  }
+
+  getTargetStreamer().emitArch(ID);
+  return false;
 }
 
 /// parseDirectiveEabiAttr
-///  ::= .eabi_attribute int, int
+///  ::= .eabi_attribute int, int [, "str"]
+///  ::= .eabi_attribute Tag_name, int [, "str"]
 bool ARMAsmParser::parseDirectiveEabiAttr(SMLoc L) {
-  if (Parser.getTok().isNot(AsmToken::Integer))
-    return Error(L, "integer expected");
-  int64_t Tag = Parser.getTok().getIntVal();
-  Parser.Lex(); // eat tag integer
+  int64_t Tag;
+  SMLoc TagLoc;
+  TagLoc = Parser.getTok().getLoc();
+  if (Parser.getTok().is(AsmToken::Identifier)) {
+    StringRef Name = Parser.getTok().getIdentifier();
+    Tag = ARMBuildAttrs::AttrTypeFromString(Name);
+    if (Tag == -1) {
+      Error(TagLoc, "attribute name not recognised: " + Name);
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+    Parser.Lex();
+  } else {
+    const MCExpr *AttrExpr;
+
+    TagLoc = Parser.getTok().getLoc();
+    if (Parser.parseExpression(AttrExpr)) {
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(AttrExpr);
+    if (!CE) {
+      Error(TagLoc, "expected numeric constant");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
 
-  if (Parser.getTok().isNot(AsmToken::Comma))
-    return Error(L, "comma expected");
+    Tag = CE->getValue();
+  }
+
+  if (Parser.getTok().isNot(AsmToken::Comma)) {
+    Error(Parser.getTok().getLoc(), "comma expected");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
   Parser.Lex(); // skip comma
 
-  L = Parser.getTok().getLoc();
-  if (Parser.getTok().isNot(AsmToken::Integer))
-    return Error(L, "integer expected");
-  int64_t Value = Parser.getTok().getIntVal();
-  Parser.Lex(); // eat value integer
+  StringRef StringValue = "";
+  bool IsStringValue = false;
+
+  int64_t IntegerValue = 0;
+  bool IsIntegerValue = false;
+
+  if (Tag == ARMBuildAttrs::CPU_raw_name || Tag == ARMBuildAttrs::CPU_name)
+    IsStringValue = true;
+  else if (Tag == ARMBuildAttrs::compatibility) {
+    IsStringValue = true;
+    IsIntegerValue = true;
+  } else if (Tag < 32 || Tag % 2 == 0)
+    IsIntegerValue = true;
+  else if (Tag % 2 == 1)
+    IsStringValue = true;
+  else
+    llvm_unreachable("invalid tag type");
+
+  if (IsIntegerValue) {
+    const MCExpr *ValueExpr;
+    SMLoc ValueExprLoc = Parser.getTok().getLoc();
+    if (Parser.parseExpression(ValueExpr)) {
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(ValueExpr);
+    if (!CE) {
+      Error(ValueExprLoc, "expected numeric constant");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
 
-  getTargetStreamer().emitAttribute(Tag, Value);
+    IntegerValue = CE->getValue();
+  }
+
+  if (Tag == ARMBuildAttrs::compatibility) {
+    if (Parser.getTok().isNot(AsmToken::Comma))
+      IsStringValue = false;
+    else
+      Parser.Lex();
+  }
+
+  if (IsStringValue) {
+    if (Parser.getTok().isNot(AsmToken::String)) {
+      Error(Parser.getTok().getLoc(), "bad string constant");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    StringValue = Parser.getTok().getStringContents();
+    Parser.Lex();
+  }
+
+  if (IsIntegerValue && IsStringValue) {
+    assert(Tag == ARMBuildAttrs::compatibility);
+    getTargetStreamer().emitIntTextAttribute(Tag, IntegerValue, StringValue);
+  } else if (IsIntegerValue)
+    getTargetStreamer().emitAttribute(Tag, IntegerValue);
+  else if (IsStringValue)
+    getTargetStreamer().emitTextAttribute(Tag, StringValue);
   return false;
 }
 
@@ -8049,8 +8627,10 @@ bool ARMAsmParser::parseDirectiveFPU(SMLoc L) {
 #include "ARMFPUName.def"
     .Default(ARM::INVALID_FPU);
 
-  if (ID == ARM::INVALID_FPU)
-    return Error(L, "Unknown FPU name");
+  if (ID == ARM::INVALID_FPU) {
+    Error(L, "Unknown FPU name");
+    return false;
+  }
 
   getTargetStreamer().emitFPU(ID);
   return false;
@@ -8059,14 +8639,18 @@ bool ARMAsmParser::parseDirectiveFPU(SMLoc L) {
 /// parseDirectiveFnStart
 ///  ::= .fnstart
 bool ARMAsmParser::parseDirectiveFnStart(SMLoc L) {
-  if (FnStartLoc.isValid()) {
+  if (UC.hasFnStart()) {
     Error(L, ".fnstart starts before the end of previous one");
-    Error(FnStartLoc, "previous .fnstart starts here");
-    return true;
+    UC.emitFnStartLocNotes();
+    return false;
   }
 
-  FnStartLoc = L;
+  // Reset the unwind directives parser state
+  UC.reset();
+
   getTargetStreamer().emitFnStart();
+
+  UC.recordFnStart(L);
   return false;
 }
 
@@ -8074,31 +8658,37 @@ bool ARMAsmParser::parseDirectiveFnStart(SMLoc L) {
 ///  ::= .fnend
 bool ARMAsmParser::parseDirectiveFnEnd(SMLoc L) {
   // Check the ordering of unwind directives
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .fnend directive");
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .fnend directive");
+    return false;
+  }
 
   // Reset the unwind directives parser state
-  resetUnwindDirectiveParserState();
   getTargetStreamer().emitFnEnd();
+
+  UC.reset();
   return false;
 }
 
 /// parseDirectiveCantUnwind
 ///  ::= .cantunwind
 bool ARMAsmParser::parseDirectiveCantUnwind(SMLoc L) {
+  UC.recordCantUnwind(L);
+
   // Check the ordering of unwind directives
-  CantUnwindLoc = L;
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .cantunwind directive");
-  if (HandlerDataLoc.isValid()) {
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .cantunwind directive");
+    return false;
+  }
+  if (UC.hasHandlerData()) {
     Error(L, ".cantunwind can't be used with .handlerdata directive");
-    Error(HandlerDataLoc, ".handlerdata was specified here");
-    return true;
+    UC.emitHandlerDataLocNotes();
+    return false;
   }
-  if (PersonalityLoc.isValid()) {
+  if (UC.hasPersonality()) {
     Error(L, ".cantunwind can't be used with .personality directive");
-    Error(PersonalityLoc, ".personality was specified here");
-    return true;
+    UC.emitPersonalityLocNotes();
+    return false;
   }
 
   getTargetStreamer().emitCantUnwind();
@@ -8108,25 +8698,37 @@ bool ARMAsmParser::parseDirectiveCantUnwind(SMLoc L) {
 /// parseDirectivePersonality
 ///  ::= .personality name
 bool ARMAsmParser::parseDirectivePersonality(SMLoc L) {
+  bool HasExistingPersonality = UC.hasPersonality();
+
+  UC.recordPersonality(L);
+
   // Check the ordering of unwind directives
-  PersonalityLoc = L;
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .personality directive");
-  if (CantUnwindLoc.isValid()) {
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .personality directive");
+    return false;
+  }
+  if (UC.cantUnwind()) {
     Error(L, ".personality can't be used with .cantunwind directive");
-    Error(CantUnwindLoc, ".cantunwind was specified here");
-    return true;
+    UC.emitCantUnwindLocNotes();
+    return false;
   }
-  if (HandlerDataLoc.isValid()) {
+  if (UC.hasHandlerData()) {
     Error(L, ".personality must precede .handlerdata directive");
-    Error(HandlerDataLoc, ".handlerdata was specified here");
-    return true;
+    UC.emitHandlerDataLocNotes();
+    return false;
+  }
+  if (HasExistingPersonality) {
+    Parser.eatToEndOfStatement();
+    Error(L, "multiple personality directives");
+    UC.emitPersonalityLocNotes();
+    return false;
   }
 
   // Parse the name of the personality routine
   if (Parser.getTok().isNot(AsmToken::Identifier)) {
     Parser.eatToEndOfStatement();
-    return Error(L, "unexpected input in .personality directive.");
+    Error(L, "unexpected input in .personality directive.");
+    return false;
   }
   StringRef Name(Parser.getTok().getIdentifier());
   Parser.Lex();
@@ -8139,14 +8741,17 @@ bool ARMAsmParser::parseDirectivePersonality(SMLoc L) {
 /// parseDirectiveHandlerData
 ///  ::= .handlerdata
 bool ARMAsmParser::parseDirectiveHandlerData(SMLoc L) {
+  UC.recordHandlerData(L);
+
   // Check the ordering of unwind directives
-  HandlerDataLoc = L;
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .personality directive");
-  if (CantUnwindLoc.isValid()) {
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .personality directive");
+    return false;
+  }
+  if (UC.cantUnwind()) {
     Error(L, ".handlerdata can't be used with .cantunwind directive");
-    Error(CantUnwindLoc, ".cantunwind was specified here");
-    return true;
+    UC.emitCantUnwindLocNotes();
+    return false;
   }
 
   getTargetStreamer().emitHandlerData();
@@ -8157,34 +8762,45 @@ bool ARMAsmParser::parseDirectiveHandlerData(SMLoc L) {
 ///  ::= .setfp fpreg, spreg [, offset]
 bool ARMAsmParser::parseDirectiveSetFP(SMLoc L) {
   // Check the ordering of unwind directives
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .setfp directive");
-  if (HandlerDataLoc.isValid())
-    return Error(L, ".setfp must precede .handlerdata directive");
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .setfp directive");
+    return false;
+  }
+  if (UC.hasHandlerData()) {
+    Error(L, ".setfp must precede .handlerdata directive");
+    return false;
+  }
 
   // Parse fpreg
-  SMLoc NewFPRegLoc = Parser.getTok().getLoc();
-  int NewFPReg = tryParseRegister();
-  if (NewFPReg == -1)
-    return Error(NewFPRegLoc, "frame pointer register expected");
+  SMLoc FPRegLoc = Parser.getTok().getLoc();
+  int FPReg = tryParseRegister();
+  if (FPReg == -1) {
+    Error(FPRegLoc, "frame pointer register expected");
+    return false;
+  }
 
   // Consume comma
-  if (!Parser.getTok().is(AsmToken::Comma))
-    return Error(Parser.getTok().getLoc(), "comma expected");
+  if (Parser.getTok().isNot(AsmToken::Comma)) {
+    Error(Parser.getTok().getLoc(), "comma expected");
+    return false;
+  }
   Parser.Lex(); // skip comma
 
   // Parse spreg
-  SMLoc NewSPRegLoc = Parser.getTok().getLoc();
-  int NewSPReg = tryParseRegister();
-  if (NewSPReg == -1)
-    return Error(NewSPRegLoc, "stack pointer register expected");
+  SMLoc SPRegLoc = Parser.getTok().getLoc();
+  int SPReg = tryParseRegister();
+  if (SPReg == -1) {
+    Error(SPRegLoc, "stack pointer register expected");
+    return false;
+  }
 
-  if (NewSPReg != ARM::SP && NewSPReg != FPReg)
-    return Error(NewSPRegLoc,
-                 "register should be either $sp or the latest fp register");
+  if (SPReg != ARM::SP && SPReg != UC.getFPReg()) {
+    Error(SPRegLoc, "register should be either $sp or the latest fp register");
+    return false;
+  }
 
   // Update the frame pointer register
-  FPReg = NewFPReg;
+  UC.saveFPReg(FPReg);
 
   // Parse offset
   int64_t Offset = 0;
@@ -8193,24 +8809,29 @@ bool ARMAsmParser::parseDirectiveSetFP(SMLoc L) {
 
     if (Parser.getTok().isNot(AsmToken::Hash) &&
         Parser.getTok().isNot(AsmToken::Dollar)) {
-      return Error(Parser.getTok().getLoc(), "'#' expected");
+      Error(Parser.getTok().getLoc(), "'#' expected");
+      return false;
     }
     Parser.Lex(); // skip hash token.
 
     const MCExpr *OffsetExpr;
     SMLoc ExLoc = Parser.getTok().getLoc();
     SMLoc EndLoc;
-    if (getParser().parseExpression(OffsetExpr, EndLoc))
-      return Error(ExLoc, "malformed setfp offset");
+    if (getParser().parseExpression(OffsetExpr, EndLoc)) {
+      Error(ExLoc, "malformed setfp offset");
+      return false;
+    }
     const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(OffsetExpr);
-    if (!CE)
-      return Error(ExLoc, "setfp offset must be an immediate");
+    if (!CE) {
+      Error(ExLoc, "setfp offset must be an immediate");
+      return false;
+    }
 
     Offset = CE->getValue();
   }
 
-  getTargetStreamer().emitSetFP(static_cast<unsigned>(NewFPReg),
-                                static_cast<unsigned>(NewSPReg), Offset);
+  getTargetStreamer().emitSetFP(static_cast<unsigned>(FPReg),
+                                static_cast<unsigned>(SPReg), Offset);
   return false;
 }
 
@@ -8218,26 +8839,35 @@ bool ARMAsmParser::parseDirectiveSetFP(SMLoc L) {
 ///  ::= .pad offset
 bool ARMAsmParser::parseDirectivePad(SMLoc L) {
   // Check the ordering of unwind directives
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .pad directive");
-  if (HandlerDataLoc.isValid())
-    return Error(L, ".pad must precede .handlerdata directive");
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .pad directive");
+    return false;
+  }
+  if (UC.hasHandlerData()) {
+    Error(L, ".pad must precede .handlerdata directive");
+    return false;
+  }
 
   // Parse the offset
   if (Parser.getTok().isNot(AsmToken::Hash) &&
       Parser.getTok().isNot(AsmToken::Dollar)) {
-    return Error(Parser.getTok().getLoc(), "'#' expected");
+    Error(Parser.getTok().getLoc(), "'#' expected");
+    return false;
   }
   Parser.Lex(); // skip hash token.
 
   const MCExpr *OffsetExpr;
   SMLoc ExLoc = Parser.getTok().getLoc();
   SMLoc EndLoc;
-  if (getParser().parseExpression(OffsetExpr, EndLoc))
-    return Error(ExLoc, "malformed pad offset");
+  if (getParser().parseExpression(OffsetExpr, EndLoc)) {
+    Error(ExLoc, "malformed pad offset");
+    return false;
+  }
   const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(OffsetExpr);
-  if (!CE)
-    return Error(ExLoc, "pad offset must be an immediate");
+  if (!CE) {
+    Error(ExLoc, "pad offset must be an immediate");
+    return false;
+  }
 
   getTargetStreamer().emitPad(CE->getValue());
   return false;
@@ -8248,37 +8878,470 @@ bool ARMAsmParser::parseDirectivePad(SMLoc L) {
 ///  ::= .vsave { registers }
 bool ARMAsmParser::parseDirectiveRegSave(SMLoc L, bool IsVector) {
   // Check the ordering of unwind directives
-  if (!FnStartLoc.isValid())
-    return Error(L, ".fnstart must precede .save or .vsave directives");
-  if (HandlerDataLoc.isValid())
-    return Error(L, ".save or .vsave must precede .handlerdata directive");
+  if (!UC.hasFnStart()) {
+    Error(L, ".fnstart must precede .save or .vsave directives");
+    return false;
+  }
+  if (UC.hasHandlerData()) {
+    Error(L, ".save or .vsave must precede .handlerdata directive");
+    return false;
+  }
 
   // RAII object to make sure parsed operands are deleted.
-  struct CleanupObject {
-    SmallVector<MCParsedAsmOperand *, 1> Operands;
-    ~CleanupObject() {
-      for (unsigned I = 0, E = Operands.size(); I != E; ++I)
-        delete Operands[I];
-    }
-  } CO;
+  SmallVector<std::unique_ptr<MCParsedAsmOperand>, 1> Operands;
 
   // Parse the register list
-  if (parseRegisterList(CO.Operands))
+  if (parseRegisterList(Operands))
+    return false;
+  ARMOperand &Op = (ARMOperand &)*Operands[0];
+  if (!IsVector && !Op.isRegList()) {
+    Error(L, ".save expects GPR registers");
+    return false;
+  }
+  if (IsVector && !Op.isDPRRegList()) {
+    Error(L, ".vsave expects DPR registers");
+    return false;
+  }
+
+  getTargetStreamer().emitRegSave(Op.getRegList(), IsVector);
+  return false;
+}
+
+/// parseDirectiveInst
+///  ::= .inst opcode [, ...]
+///  ::= .inst.n opcode [, ...]
+///  ::= .inst.w opcode [, ...]
+bool ARMAsmParser::parseDirectiveInst(SMLoc Loc, char Suffix) {
+  int Width;
+
+  if (isThumb()) {
+    switch (Suffix) {
+    case 'n':
+      Width = 2;
+      break;
+    case 'w':
+      Width = 4;
+      break;
+    default:
+      Parser.eatToEndOfStatement();
+      Error(Loc, "cannot determine Thumb instruction size, "
+                 "use inst.n/inst.w instead");
+      return false;
+    }
+  } else {
+    if (Suffix) {
+      Parser.eatToEndOfStatement();
+      Error(Loc, "width suffixes are invalid in ARM mode");
+      return false;
+    }
+    Width = 4;
+  }
+
+  if (getLexer().is(AsmToken::EndOfStatement)) {
+    Parser.eatToEndOfStatement();
+    Error(Loc, "expected expression following directive");
+    return false;
+  }
+
+  for (;;) {
+    const MCExpr *Expr;
+
+    if (getParser().parseExpression(Expr)) {
+      Error(Loc, "expected expression");
+      return false;
+    }
+
+    const MCConstantExpr *Value = dyn_cast_or_null<MCConstantExpr>(Expr);
+    if (!Value) {
+      Error(Loc, "expected constant expression");
+      return false;
+    }
+
+    switch (Width) {
+    case 2:
+      if (Value->getValue() > 0xffff) {
+        Error(Loc, "inst.n operand is too big, use inst.w instead");
+        return false;
+      }
+      break;
+    case 4:
+      if (Value->getValue() > 0xffffffff) {
+        Error(Loc,
+              StringRef(Suffix ? "inst.w" : "inst") + " operand is too big");
+        return false;
+      }
+      break;
+    default:
+      llvm_unreachable("only supported widths are 2 and 4");
+    }
+
+    getTargetStreamer().emitInst(Value->getValue(), Suffix);
+
+    if (getLexer().is(AsmToken::EndOfStatement))
+      break;
+
+    if (getLexer().isNot(AsmToken::Comma)) {
+      Error(Loc, "unexpected token in directive");
+      return false;
+    }
+
+    Parser.Lex();
+  }
+
+  Parser.Lex();
+  return false;
+}
+
+/// parseDirectiveLtorg
+///  ::= .ltorg | .pool
+bool ARMAsmParser::parseDirectiveLtorg(SMLoc L) {
+  getTargetStreamer().emitCurrentConstantPool();
+  return false;
+}
+
+bool ARMAsmParser::parseDirectiveEven(SMLoc L) {
+  const MCSection *Section = getStreamer().getCurrentSection().first;
+
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    TokError("unexpected token in directive");
+    return false;
+  }
+
+  if (!Section) {
+    getStreamer().InitSections();
+    Section = getStreamer().getCurrentSection().first;
+  }
+
+  assert(Section && "must have section to emit alignment");
+  if (Section->UseCodeAlign())
+    getStreamer().EmitCodeAlignment(2);
+  else
+    getStreamer().EmitValueToAlignment(2);
+
+  return false;
+}
+
+/// parseDirectivePersonalityIndex
+///   ::= .personalityindex index
+bool ARMAsmParser::parseDirectivePersonalityIndex(SMLoc L) {
+  bool HasExistingPersonality = UC.hasPersonality();
+
+  UC.recordPersonalityIndex(L);
+
+  if (!UC.hasFnStart()) {
+    Parser.eatToEndOfStatement();
+    Error(L, ".fnstart must precede .personalityindex directive");
+    return false;
+  }
+  if (UC.cantUnwind()) {
+    Parser.eatToEndOfStatement();
+    Error(L, ".personalityindex cannot be used with .cantunwind");
+    UC.emitCantUnwindLocNotes();
+    return false;
+  }
+  if (UC.hasHandlerData()) {
+    Parser.eatToEndOfStatement();
+    Error(L, ".personalityindex must precede .handlerdata directive");
+    UC.emitHandlerDataLocNotes();
+    return false;
+  }
+  if (HasExistingPersonality) {
+    Parser.eatToEndOfStatement();
+    Error(L, "multiple personality directives");
+    UC.emitPersonalityLocNotes();
+    return false;
+  }
+
+  const MCExpr *IndexExpression;
+  SMLoc IndexLoc = Parser.getTok().getLoc();
+  if (Parser.parseExpression(IndexExpression)) {
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(IndexExpression);
+  if (!CE) {
+    Parser.eatToEndOfStatement();
+    Error(IndexLoc, "index must be a constant number");
+    return false;
+  }
+  if (CE->getValue() < 0 ||
+      CE->getValue() >= ARM::EHABI::NUM_PERSONALITY_INDEX) {
+    Parser.eatToEndOfStatement();
+    Error(IndexLoc, "personality routine index should be in range [0-3]");
+    return false;
+  }
+
+  getTargetStreamer().emitPersonalityIndex(CE->getValue());
+  return false;
+}
+
+/// parseDirectiveUnwindRaw
+///   ::= .unwind_raw offset, opcode [, opcode...]
+bool ARMAsmParser::parseDirectiveUnwindRaw(SMLoc L) {
+  if (!UC.hasFnStart()) {
+    Parser.eatToEndOfStatement();
+    Error(L, ".fnstart must precede .unwind_raw directives");
+    return false;
+  }
+
+  int64_t StackOffset;
+
+  const MCExpr *OffsetExpr;
+  SMLoc OffsetLoc = getLexer().getLoc();
+  if (getLexer().is(AsmToken::EndOfStatement) ||
+      getParser().parseExpression(OffsetExpr)) {
+    Error(OffsetLoc, "expected expression");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(OffsetExpr);
+  if (!CE) {
+    Error(OffsetLoc, "offset must be a constant");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  StackOffset = CE->getValue();
+
+  if (getLexer().isNot(AsmToken::Comma)) {
+    Error(getLexer().getLoc(), "expected comma");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+  Parser.Lex();
+
+  SmallVector<uint8_t, 16> Opcodes;
+  for (;;) {
+    const MCExpr *OE;
+
+    SMLoc OpcodeLoc = getLexer().getLoc();
+    if (getLexer().is(AsmToken::EndOfStatement) || Parser.parseExpression(OE)) {
+      Error(OpcodeLoc, "expected opcode expression");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    const MCConstantExpr *OC = dyn_cast<MCConstantExpr>(OE);
+    if (!OC) {
+      Error(OpcodeLoc, "opcode value must be a constant");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    const int64_t Opcode = OC->getValue();
+    if (Opcode & ~0xff) {
+      Error(OpcodeLoc, "invalid opcode");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    Opcodes.push_back(uint8_t(Opcode));
+
+    if (getLexer().is(AsmToken::EndOfStatement))
+      break;
+
+    if (getLexer().isNot(AsmToken::Comma)) {
+      Error(getLexer().getLoc(), "unexpected token in directive");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+
+    Parser.Lex();
+  }
+
+  getTargetStreamer().emitUnwindRaw(StackOffset, Opcodes);
+
+  Parser.Lex();
+  return false;
+}
+
+/// parseDirectiveTLSDescSeq
+///   ::= .tlsdescseq tls-variable
+bool ARMAsmParser::parseDirectiveTLSDescSeq(SMLoc L) {
+  if (getLexer().isNot(AsmToken::Identifier)) {
+    TokError("expected variable after '.tlsdescseq' directive");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  const MCSymbolRefExpr *SRE =
+    MCSymbolRefExpr::Create(Parser.getTok().getIdentifier(),
+                            MCSymbolRefExpr::VK_ARM_TLSDESCSEQ, getContext());
+  Lex();
+
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(Parser.getTok().getLoc(), "unexpected token");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  getTargetStreamer().AnnotateTLSDescriptorSequence(SRE);
+  return false;
+}
+
+/// parseDirectiveMovSP
+///  ::= .movsp reg [, #offset]
+bool ARMAsmParser::parseDirectiveMovSP(SMLoc L) {
+  if (!UC.hasFnStart()) {
+    Parser.eatToEndOfStatement();
+    Error(L, ".fnstart must precede .movsp directives");
+    return false;
+  }
+  if (UC.getFPReg() != ARM::SP) {
+    Parser.eatToEndOfStatement();
+    Error(L, "unexpected .movsp directive");
+    return false;
+  }
+
+  SMLoc SPRegLoc = Parser.getTok().getLoc();
+  int SPReg = tryParseRegister();
+  if (SPReg == -1) {
+    Parser.eatToEndOfStatement();
+    Error(SPRegLoc, "register expected");
+    return false;
+  }
+
+  if (SPReg == ARM::SP || SPReg == ARM::PC) {
+    Parser.eatToEndOfStatement();
+    Error(SPRegLoc, "sp and pc are not permitted in .movsp directive");
+    return false;
+  }
+
+  int64_t Offset = 0;
+  if (Parser.getTok().is(AsmToken::Comma)) {
+    Parser.Lex();
+
+    if (Parser.getTok().isNot(AsmToken::Hash)) {
+      Error(Parser.getTok().getLoc(), "expected #constant");
+      Parser.eatToEndOfStatement();
+      return false;
+    }
+    Parser.Lex();
+
+    const MCExpr *OffsetExpr;
+    SMLoc OffsetLoc = Parser.getTok().getLoc();
+    if (Parser.parseExpression(OffsetExpr)) {
+      Parser.eatToEndOfStatement();
+      Error(OffsetLoc, "malformed offset expression");
+      return false;
+    }
+
+    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(OffsetExpr);
+    if (!CE) {
+      Parser.eatToEndOfStatement();
+      Error(OffsetLoc, "offset must be an immediate constant");
+      return false;
+    }
+
+    Offset = CE->getValue();
+  }
+
+  getTargetStreamer().emitMovSP(SPReg, Offset);
+  UC.saveFPReg(SPReg);
+
+  return false;
+}
+
+/// parseDirectiveObjectArch
+///   ::= .object_arch name
+bool ARMAsmParser::parseDirectiveObjectArch(SMLoc L) {
+  if (getLexer().isNot(AsmToken::Identifier)) {
+    Error(getLexer().getLoc(), "unexpected token");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  StringRef Arch = Parser.getTok().getString();
+  SMLoc ArchLoc = Parser.getTok().getLoc();
+  getLexer().Lex();
+
+  unsigned ID = StringSwitch<unsigned>(Arch)
+#define ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH) \
+    .Case(NAME, ARM::ID)
+#define ARM_ARCH_ALIAS(NAME, ID) \
+    .Case(NAME, ARM::ID)
+#include "MCTargetDesc/ARMArchName.def"
+#undef ARM_ARCH_NAME
+#undef ARM_ARCH_ALIAS
+    .Default(ARM::INVALID_ARCH);
+
+  if (ID == ARM::INVALID_ARCH) {
+    Error(ArchLoc, "unknown architecture '" + Arch + "'");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  getTargetStreamer().emitObjectArch(ID);
+
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    Error(getLexer().getLoc(), "unexpected token");
+    Parser.eatToEndOfStatement();
+  }
+
+  return false;
+}
+
+/// parseDirectiveAlign
+///   ::= .align
+bool ARMAsmParser::parseDirectiveAlign(SMLoc L) {
+  // NOTE: if this is not the end of the statement, fall back to the target
+  // agnostic handling for this directive which will correctly handle this.
+  if (getLexer().isNot(AsmToken::EndOfStatement))
     return true;
-  ARMOperand *Op = (ARMOperand*)CO.Operands[0];
-  if (!IsVector && !Op->isRegList())
-    return Error(L, ".save expects GPR registers");
-  if (IsVector && !Op->isDPRRegList())
-    return Error(L, ".vsave expects DPR registers");
 
-  getTargetStreamer().emitRegSave(Op->getRegList(), IsVector);
+  // '.align' is target specifically handled to mean 2**2 byte alignment.
+  if (getStreamer().getCurrentSection().first->UseCodeAlign())
+    getStreamer().EmitCodeAlignment(4, 0);
+  else
+    getStreamer().EmitValueToAlignment(4, 0, 1, 0);
+
+  return false;
+}
+
+/// parseDirectiveThumbSet
+///  ::= .thumb_set name, value
+bool ARMAsmParser::parseDirectiveThumbSet(SMLoc L) {
+  StringRef Name;
+  if (Parser.parseIdentifier(Name)) {
+    TokError("expected identifier after '.thumb_set'");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  if (getLexer().isNot(AsmToken::Comma)) {
+    TokError("expected comma after name '" + Name + "'");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+  Lex();
+
+  const MCExpr *Value;
+  if (Parser.parseExpression(Value)) {
+    TokError("missing expression");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  if (getLexer().isNot(AsmToken::EndOfStatement)) {
+    TokError("unexpected token");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+  Lex();
+
+  MCSymbol *Alias = getContext().GetOrCreateSymbol(Name);
+  getTargetStreamer().emitThumbSet(Alias, Value);
   return false;
 }
 
 /// Force static initialization.
 extern "C" void LLVMInitializeARMAsmParser() {
-  RegisterMCAsmParser<ARMAsmParser> X(TheARMTarget);
-  RegisterMCAsmParser<ARMAsmParser> Y(TheThumbTarget);
+  RegisterMCAsmParser<ARMAsmParser> X(TheARMLETarget);
+  RegisterMCAsmParser<ARMAsmParser> Y(TheARMBETarget);
+  RegisterMCAsmParser<ARMAsmParser> A(TheThumbLETarget);
+  RegisterMCAsmParser<ARMAsmParser> B(TheThumbBETarget);
 }
 
 #define GET_REGISTER_MATCHER
@@ -8286,20 +9349,113 @@ extern "C" void LLVMInitializeARMAsmParser() {
 #define GET_MATCHER_IMPLEMENTATION
 #include "ARMGenAsmMatcher.inc"
 
+static const struct ExtMapEntry {
+  const char *Extension;
+  const unsigned ArchCheck;
+  const uint64_t Features;
+} Extensions[] = {
+  { "crc", Feature_HasV8, ARM::FeatureCRC },
+  { "crypto",  Feature_HasV8,
+    ARM::FeatureCrypto | ARM::FeatureNEON | ARM::FeatureFPARMv8 },
+  { "fp", Feature_HasV8, ARM::FeatureFPARMv8 },
+  { "idiv", Feature_HasV7 | Feature_IsNotMClass,
+    ARM::FeatureHWDiv | ARM::FeatureHWDivARM },
+  // FIXME: iWMMXT not supported
+  { "iwmmxt", Feature_None, 0 },
+  // FIXME: iWMMXT2 not supported
+  { "iwmmxt2", Feature_None, 0 },
+  // FIXME: Maverick not supported
+  { "maverick", Feature_None, 0 },
+  { "mp", Feature_HasV7 | Feature_IsNotMClass, ARM::FeatureMP },
+  // FIXME: ARMv6-m OS Extensions feature not checked
+  { "os", Feature_None, 0 },
+  // FIXME: Also available in ARMv6-K
+  { "sec", Feature_HasV7, ARM::FeatureTrustZone },
+  { "simd", Feature_HasV8, ARM::FeatureNEON | ARM::FeatureFPARMv8 },
+  // FIXME: Only available in A-class, isel not predicated
+  { "virt", Feature_HasV7, ARM::FeatureVirtualization },
+  // FIXME: xscale not supported
+  { "xscale", Feature_None, 0 },
+};
+
+/// parseDirectiveArchExtension
+///   ::= .arch_extension [no]feature
+bool ARMAsmParser::parseDirectiveArchExtension(SMLoc L) {
+  if (getLexer().isNot(AsmToken::Identifier)) {
+    Error(getLexer().getLoc(), "unexpected token");
+    Parser.eatToEndOfStatement();
+    return false;
+  }
+
+  StringRef Extension = Parser.getTok().getString();
+  SMLoc ExtLoc = Parser.getTok().getLoc();
+  getLexer().Lex();
+
+  bool EnableFeature = true;
+  if (Extension.startswith_lower("no")) {
+    EnableFeature = false;
+    Extension = Extension.substr(2);
+  }
+
+  for (unsigned EI = 0, EE = array_lengthof(Extensions); EI != EE; ++EI) {
+    if (Extensions[EI].Extension != Extension)
+      continue;
+
+    unsigned FB = getAvailableFeatures();
+    if ((FB & Extensions[EI].ArchCheck) != Extensions[EI].ArchCheck) {
+      Error(ExtLoc, "architectural extension '" + Extension + "' is not "
+            "allowed for the current base architecture");
+      return false;
+    }
+
+    if (!Extensions[EI].Features)
+      report_fatal_error("unsupported architectural extension: " + Extension);
+
+    if (EnableFeature)
+      FB |= ComputeAvailableFeatures(Extensions[EI].Features);
+    else
+      FB &= ~ComputeAvailableFeatures(Extensions[EI].Features);
+
+    setAvailableFeatures(FB);
+    return false;
+  }
+
+  Error(ExtLoc, "unknown architectural extension: " + Extension);
+  Parser.eatToEndOfStatement();
+  return false;
+}
+
 // Define this matcher function after the auto-generated include so we
 // have the match class enum definitions.
-unsigned ARMAsmParser::validateTargetOperandClass(MCParsedAsmOperand *AsmOp,
+unsigned ARMAsmParser::validateTargetOperandClass(MCParsedAsmOperand &AsmOp,
                                                   unsigned Kind) {
-  ARMOperand *Op = static_cast<ARMOperand*>(AsmOp);
+  ARMOperand &Op = static_cast<ARMOperand &>(AsmOp);
   // If the kind is a token for a literal immediate, check if our asm
   // operand matches. This is for InstAliases which have a fixed-value
   // immediate in the syntax.
-  if (Kind == MCK__35_0 && Op->isImm()) {
-    const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Op->getImm());
-    if (!CE)
-      return Match_InvalidOperand;
-    if (CE->getValue() == 0)
+  switch (Kind) {
+  default: break;
+  case MCK__35_0:
+    if (Op.isImm())
+      if (const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Op.getImm()))
+        if (CE->getValue() == 0)
+          return Match_Success;
+    break;
+  case MCK_ARMSOImm:
+    if (Op.isImm()) {
+      const MCExpr *SOExpr = Op.getImm();
+      int64_t Value;
+      if (!SOExpr->EvaluateAsAbsolute(Value))
+        return Match_Success;
+      assert((Value >= INT32_MIN && Value <= UINT32_MAX) &&
+             "expression value must be representable in 32 bits");
+    }
+    break;
+  case MCK_GPRPair:
+    if (Op.isReg() &&
+        MRI->getRegClass(ARM::GPRRegClassID).contains(Op.getReg()))
       return Match_Success;
+    break;
   }
   return Match_InvalidOperand;
 }
diff --git a/contrib/llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp b/contrib/llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp
index 9c7988f8a857..4d4038deac7c 100644
--- a/contrib/llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp
+++ b/contrib/llvm/lib/Target/ARM/Disassembler/ARMDisassembler.cpp
@@ -7,8 +7,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "arm-disassembler"
-
 #include "llvm/MC/MCDisassembler.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "MCTargetDesc/ARMBaseInfo.h"
@@ -29,6 +27,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "arm-disassembler"
+
 typedef MCDisassembler::DecodeStatus DecodeStatus;
 
 namespace {
@@ -90,20 +90,18 @@ class ARMDisassembler : public MCDisassembler {
 public:
   /// Constructor     - Initializes the disassembler.
   ///
-  ARMDisassembler(const MCSubtargetInfo &STI) :
-    MCDisassembler(STI) {
+  ARMDisassembler(const MCSubtargetInfo &STI, MCContext &Ctx) :
+    MCDisassembler(STI, Ctx) {
   }
 
   ~ARMDisassembler() {
   }
 
   /// getInstruction - See MCDisassembler.
-  DecodeStatus getInstruction(MCInst &instr,
-                              uint64_t &size,
-                              const MemoryObject &region,
-                              uint64_t address,
+  DecodeStatus getInstruction(MCInst &instr, uint64_t &size,
+                              const MemoryObject &region, uint64_t address,
                               raw_ostream &vStream,
-                              raw_ostream &cStream) const;
+                              raw_ostream &cStream) const override;
 };
 
 /// ThumbDisassembler - Thumb disassembler for all Thumb platforms.
@@ -111,20 +109,18 @@ class ThumbDisassembler : public MCDisassembler {
 public:
   /// Constructor     - Initializes the disassembler.
   ///
-  ThumbDisassembler(const MCSubtargetInfo &STI) :
-    MCDisassembler(STI) {
+  ThumbDisassembler(const MCSubtargetInfo &STI, MCContext &Ctx) :
+    MCDisassembler(STI, Ctx) {
   }
 
   ~ThumbDisassembler() {
   }
 
   /// getInstruction - See MCDisassembler.
-  DecodeStatus getInstruction(MCInst &instr,
-                              uint64_t &size,
-                              const MemoryObject &region,
-                              uint64_t address,
+  DecodeStatus getInstruction(MCInst &instr, uint64_t &size,
+                              const MemoryObject &region, uint64_t address,
                               raw_ostream &vStream,
-                              raw_ostream &cStream) const;
+                              raw_ostream &cStream) const override;
 
 private:
   mutable ITStatus ITBlock;
@@ -404,12 +400,16 @@ static DecodeStatus DecodeMRRC2(llvm::MCInst &Inst, unsigned Val,
                                 uint64_t Address, const void *Decoder);
 #include "ARMGenDisassemblerTables.inc"
 
-static MCDisassembler *createARMDisassembler(const Target &T, const MCSubtargetInfo &STI) {
-  return new ARMDisassembler(STI);
+static MCDisassembler *createARMDisassembler(const Target &T,
+                                             const MCSubtargetInfo &STI,
+                                             MCContext &Ctx) {
+  return new ARMDisassembler(STI, Ctx);
 }
 
-static MCDisassembler *createThumbDisassembler(const Target &T, const MCSubtargetInfo &STI) {
-  return new ThumbDisassembler(STI);
+static MCDisassembler *createThumbDisassembler(const Target &T,
+                                               const MCSubtargetInfo &STI,
+                                               MCContext &Ctx) {
+  return new ThumbDisassembler(STI, Ctx);
 }
 
 DecodeStatus ARMDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
@@ -860,9 +860,13 @@ DecodeStatus ThumbDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
 
 
 extern "C" void LLVMInitializeARMDisassembler() {
-  TargetRegistry::RegisterMCDisassembler(TheARMTarget,
+  TargetRegistry::RegisterMCDisassembler(TheARMLETarget,
+                                         createARMDisassembler);
+  TargetRegistry::RegisterMCDisassembler(TheARMBETarget,
                                          createARMDisassembler);
-  TargetRegistry::RegisterMCDisassembler(TheThumbTarget,
+  TargetRegistry::RegisterMCDisassembler(TheThumbLETarget,
+                                         createThumbDisassembler);
+  TargetRegistry::RegisterMCDisassembler(TheThumbBETarget,
                                          createThumbDisassembler);
 }
 
diff --git a/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.cpp b/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.cpp
index f89702853d41..228fb5756ca8 100644
--- a/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.cpp
+++ b/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "asm-printer"
 #include "ARMInstPrinter.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "MCTargetDesc/ARMBaseInfo.h"
@@ -23,6 +22,8 @@
 #include "llvm/Support/raw_ostream.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "asm-printer"
+
 #include "ARMGenAsmWriter.inc"
 
 /// translateShiftImm - Convert shift immediate from 0-31 to 1-32 for printing.
@@ -307,17 +308,30 @@ void ARMInstPrinter::printOperand(const MCInst *MI, unsigned OpNo,
       << markup(">");
   } else {
     assert(Op.isExpr() && "unknown operand kind in printOperand");
-    // If a symbolic branch target was added as a constant expression then print
-    // that address in hex. And only print 32 unsigned bits for the address.
-    const MCConstantExpr *BranchTarget = dyn_cast<MCConstantExpr>(Op.getExpr());
-    int64_t Address;
-    if (BranchTarget && BranchTarget->EvaluateAsAbsolute(Address)) {
-      O << "0x";
-      O.write_hex((uint32_t)Address);
+    const MCExpr *Expr = Op.getExpr();
+    switch (Expr->getKind()) {
+    case MCExpr::Binary:
+      O << '#' << *Expr;
+      break;
+    case MCExpr::Constant: {
+      // If a symbolic branch target was added as a constant expression then
+      // print that address in hex. And only print 32 unsigned bits for the
+      // address.
+      const MCConstantExpr *Constant = cast<MCConstantExpr>(Expr);
+      int64_t TargetAddress;
+      if (!Constant->EvaluateAsAbsolute(TargetAddress)) {
+        O << '#' << *Expr;
+      } else {
+        O << "0x";
+        O.write_hex(static_cast<uint32_t>(TargetAddress));
+      }
+      break;
     }
-    else {
-      // Otherwise, just print the expression.
-      O << *Op.getExpr();
+    default:
+      // FIXME: Should we always treat this as if it is a constant literal and
+      // prefix it with '#'?
+      O << *Expr;
+      break;
     }
   }
 }
@@ -1078,13 +1092,13 @@ void ARMInstPrinter::printAddrModeImm12Operand(const MCInst *MI, unsigned OpNum,
   if (isSub) {
     O << ", "
       << markup("<imm:")
-      << "#-" << -OffImm
+      << "#-" << formatImm(-OffImm)
       << markup(">");
   }
   else if (AlwaysPrintImm0 || OffImm > 0) {
     O << ", "
       << markup("<imm:")
-      << "#" << OffImm
+      << "#" << formatImm(OffImm)
       << markup(">");
   }
   O << "]" << markup(">");
diff --git a/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.h b/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.h
index 15ae8d1267f7..f671fe49bb66 100644
--- a/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.h
+++ b/contrib/llvm/lib/Target/ARM/InstPrinter/ARMInstPrinter.h
@@ -26,8 +26,8 @@ public:
   ARMInstPrinter(const MCAsmInfo &MAI, const MCInstrInfo &MII,
                  const MCRegisterInfo &MRI, const MCSubtargetInfo &STI);
 
-  virtual void printInst(const MCInst *MI, raw_ostream &O, StringRef Annot);
-  virtual void printRegName(raw_ostream &OS, unsigned RegNo) const;
+  void printInst(const MCInst *MI, raw_ostream &O, StringRef Annot) override;
+  void printRegName(raw_ostream &OS, unsigned RegNo) const override;
 
   // Autogenerated by tblgen.
   void printInstruction(const MCInst *MI, raw_ostream &O);
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMArchName.def b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMArchName.def
new file mode 100644
index 000000000000..9f007a035a88
--- /dev/null
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMArchName.def
@@ -0,0 +1,50 @@
+//===-- ARMArchName.def - List of the ARM arch names ------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file contains the list of the supported ARM architecture names,
+// i.e. the supported value for -march= option.
+//
+//===----------------------------------------------------------------------===//
+
+// NOTE: NO INCLUDE GUARD DESIRED!
+
+#ifndef ARM_ARCH_NAME
+#error "You must define ARM_ARCH_NAME before including ARMArchName.def"
+#endif
+
+// ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH)
+ARM_ARCH_NAME("armv2",   ARMV2,   "2",       v4)
+ARM_ARCH_NAME("armv2a",  ARMV2A,  "2A",      v4)
+ARM_ARCH_NAME("armv3",   ARMV3,   "3",       v4)
+ARM_ARCH_NAME("armv3m",  ARMV3M,  "3M",      v4)
+ARM_ARCH_NAME("armv4",   ARMV4,   "4",       v4)
+ARM_ARCH_NAME("armv4t",  ARMV4T,  "4T",      v4T)
+ARM_ARCH_NAME("armv5",   ARMV5,   "5",       v5T)
+ARM_ARCH_NAME("armv5t",  ARMV5T,  "5T",      v5T)
+ARM_ARCH_NAME("armv5te", ARMV5TE, "5TE",     v5TE)
+ARM_ARCH_NAME("armv6",   ARMV6,   "6",       v6)
+ARM_ARCH_NAME("armv6j",  ARMV6J,  "6J",      v6)
+ARM_ARCH_NAME("armv6t2", ARMV6T2, "6T2",     v6T2)
+ARM_ARCH_NAME("armv6z",  ARMV6Z,  "6Z",      v6KZ)
+ARM_ARCH_NAME("armv6zk", ARMV6ZK, "6ZK",     v6KZ)
+ARM_ARCH_NAME("armv6-m", ARMV6M,  "6-M",     v6_M)
+ARM_ARCH_NAME("armv7",   ARMV7,   "7",       v7)
+ARM_ARCH_NAME("armv7-a", ARMV7A,  "7-A",     v7)
+ARM_ARCH_ALIAS("armv7a", ARMV7A)
+ARM_ARCH_NAME("armv7-r", ARMV7R,  "7-R",     v7)
+ARM_ARCH_ALIAS("armv7r", ARMV7R)
+ARM_ARCH_NAME("armv7-m", ARMV7M,  "7-M",     v7)
+ARM_ARCH_ALIAS("armv7m", ARMV7M)
+ARM_ARCH_NAME("armv8-a", ARMV8A,  "8-A",     v8)
+ARM_ARCH_ALIAS("armv8a", ARMV8A)
+ARM_ARCH_NAME("iwmmxt",  IWMMXT,  "iwmmxt",  v5TE)
+ARM_ARCH_NAME("iwmmxt2", IWMMXT2, "iwmmxt2", v5TE)
+
+#undef ARM_ARCH_NAME
+#undef ARM_ARCH_ALIAS
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMArchName.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMArchName.h
new file mode 100644
index 000000000000..34b9fc126ff1
--- /dev/null
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMArchName.h
@@ -0,0 +1,27 @@
+//===-- ARMArchName.h - List of the ARM arch names --------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef ARMARCHNAME_H
+#define ARMARCHNAME_H
+
+namespace llvm {
+namespace ARM {
+
+enum ArchKind {
+  INVALID_ARCH = 0
+
+#define ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH) , ID
+#define ARM_ARCH_ALIAS(NAME, ID) /* empty */
+#include "ARMArchName.def"
+};
+
+} // namespace ARM
+} // namespace llvm
+
+#endif // ARMARCHNAME_H
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMAsmBackend.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMAsmBackend.cpp
index 5615b808fc11..7acd9cc4d302 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMAsmBackend.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMAsmBackend.cpp
@@ -41,24 +41,27 @@ public:
 
 class ARMAsmBackend : public MCAsmBackend {
   const MCSubtargetInfo* STI;
-  bool isThumbMode;  // Currently emitting Thumb code.
+  bool isThumbMode;     // Currently emitting Thumb code.
+  bool IsLittleEndian;  // Big or little endian.
 public:
-  ARMAsmBackend(const Target &T, const StringRef TT)
+  ARMAsmBackend(const Target &T, const StringRef TT, bool IsLittle)
     : MCAsmBackend(), STI(ARM_MC::createARMMCSubtargetInfo(TT, "", "")),
-      isThumbMode(TT.startswith("thumb")) {}
+      isThumbMode(TT.startswith("thumb")), IsLittleEndian(IsLittle) {}
 
   ~ARMAsmBackend() {
     delete STI;
   }
 
-  unsigned getNumFixupKinds() const { return ARM::NumTargetFixupKinds; }
+  unsigned getNumFixupKinds() const override {
+    return ARM::NumTargetFixupKinds;
+  }
 
   bool hasNOP() const {
     return (STI->getFeatureBits() & ARM::HasV6T2Ops) != 0;
   }
 
-  const MCFixupKindInfo &getFixupKindInfo(MCFixupKind Kind) const {
-    const static MCFixupKindInfo Infos[ARM::NumTargetFixupKinds] = {
+  const MCFixupKindInfo &getFixupKindInfo(MCFixupKind Kind) const override {
+    const static MCFixupKindInfo InfosLE[ARM::NumTargetFixupKinds] = {
 // This table *must* be in the order that the fixup_* kinds are defined in
 // ARMFixupKinds.h.
 //
@@ -94,10 +97,43 @@ public:
 { "fixup_arm_movw_lo16",     0,            20,  0 },
 { "fixup_t2_movt_hi16",      0,            20,  0 },
 { "fixup_t2_movw_lo16",      0,            20,  0 },
-{ "fixup_arm_movt_hi16_pcrel", 0,          20,  MCFixupKindInfo::FKF_IsPCRel },
-{ "fixup_arm_movw_lo16_pcrel", 0,          20,  MCFixupKindInfo::FKF_IsPCRel },
-{ "fixup_t2_movt_hi16_pcrel", 0,           20,  MCFixupKindInfo::FKF_IsPCRel },
-{ "fixup_t2_movw_lo16_pcrel", 0,           20,  MCFixupKindInfo::FKF_IsPCRel },
+    };
+    const static MCFixupKindInfo InfosBE[ARM::NumTargetFixupKinds] = {
+// This table *must* be in the order that the fixup_* kinds are defined in
+// ARMFixupKinds.h.
+//
+// Name                      Offset (bits) Size (bits)     Flags
+{ "fixup_arm_ldst_pcrel_12", 0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_t2_ldst_pcrel_12",  0,            32,  MCFixupKindInfo::FKF_IsPCRel |
+                                   MCFixupKindInfo::FKF_IsAlignedDownTo32Bits},
+{ "fixup_arm_pcrel_10_unscaled", 0,        32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_pcrel_10",      0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_t2_pcrel_10",       0,            32,  MCFixupKindInfo::FKF_IsPCRel |
+                                   MCFixupKindInfo::FKF_IsAlignedDownTo32Bits},
+{ "fixup_thumb_adr_pcrel_10",8,            8,   MCFixupKindInfo::FKF_IsPCRel |
+                                   MCFixupKindInfo::FKF_IsAlignedDownTo32Bits},
+{ "fixup_arm_adr_pcrel_12",  0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_t2_adr_pcrel_12",   0,            32,  MCFixupKindInfo::FKF_IsPCRel |
+                                   MCFixupKindInfo::FKF_IsAlignedDownTo32Bits},
+{ "fixup_arm_condbranch",    8,            24,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_uncondbranch",  8,            24,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_t2_condbranch",     0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_t2_uncondbranch",   0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_thumb_br",      0,            16,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_uncondbl",      8,            24,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_condbl",        8,            24,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_blx",           8,            24,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_thumb_bl",      0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_thumb_blx",     0,            32,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_thumb_cb",      0,            16,  MCFixupKindInfo::FKF_IsPCRel },
+{ "fixup_arm_thumb_cp",      8,             8,  MCFixupKindInfo::FKF_IsPCRel |
+                                   MCFixupKindInfo::FKF_IsAlignedDownTo32Bits},
+{ "fixup_arm_thumb_bcc",     8,             8,  MCFixupKindInfo::FKF_IsPCRel },
+// movw / movt: 16-bits immediate but scattered into two chunks 0 - 12, 16 - 19.
+{ "fixup_arm_movt_hi16",     12,           20,  0 },
+{ "fixup_arm_movw_lo16",     12,           20,  0 },
+{ "fixup_t2_movt_hi16",      12,           20,  0 },
+{ "fixup_t2_movw_lo16",      12,           20,  0 },
     };
 
     if (Kind < FirstTargetFixupKind)
@@ -105,32 +141,31 @@ public:
 
     assert(unsigned(Kind - FirstTargetFixupKind) < getNumFixupKinds() &&
            "Invalid kind!");
-    return Infos[Kind - FirstTargetFixupKind];
+    return (IsLittleEndian ? InfosLE : InfosBE)[Kind - FirstTargetFixupKind];
   }
 
   /// processFixupValue - Target hook to process the literal value of a fixup
   /// if necessary.
   void processFixupValue(const MCAssembler &Asm, const MCAsmLayout &Layout,
                          const MCFixup &Fixup, const MCFragment *DF,
-                         MCValue &Target, uint64_t &Value,
-                         bool &IsResolved);
+                         const MCValue &Target, uint64_t &Value,
+                         bool &IsResolved) override;
 
 
   void applyFixup(const MCFixup &Fixup, char *Data, unsigned DataSize,
-                  uint64_t Value) const;
+                  uint64_t Value, bool IsPCRel) const override;
 
-  bool mayNeedRelaxation(const MCInst &Inst) const;
+  bool mayNeedRelaxation(const MCInst &Inst) const override;
 
-  bool fixupNeedsRelaxation(const MCFixup &Fixup,
-                            uint64_t Value,
+  bool fixupNeedsRelaxation(const MCFixup &Fixup, uint64_t Value,
                             const MCRelaxableFragment *DF,
-                            const MCAsmLayout &Layout) const;
+                            const MCAsmLayout &Layout) const override;
 
-  void relaxInstruction(const MCInst &Inst, MCInst &Res) const;
+  void relaxInstruction(const MCInst &Inst, MCInst &Res) const override;
 
-  bool writeNopData(uint64_t Count, MCObjectWriter *OW) const;
+  bool writeNopData(uint64_t Count, MCObjectWriter *OW) const override;
 
-  void handleAssemblerFlag(MCAssemblerFlag Flag) {
+  void handleAssemblerFlag(MCAssemblerFlag Flag) override {
     switch (Flag) {
     default: break;
     case MCAF_Code16:
@@ -145,6 +180,7 @@ public:
   unsigned getPointerSize() const { return 4; }
   bool isThumb() const { return isThumbMode; }
   void setIsThumb(bool it) { isThumbMode = it; }
+  bool isLittle() const { return IsLittleEndian; }
 };
 } // end anonymous namespace
 
@@ -155,6 +191,8 @@ static unsigned getRelaxedOpcode(unsigned Op) {
   case ARM::tLDRpci:    return ARM::t2LDRpci;
   case ARM::tADR:       return ARM::t2ADR;
   case ARM::tB:         return ARM::t2B;
+  case ARM::tCBZ:       return ARM::tHINT;
+  case ARM::tCBNZ:      return ARM::tHINT;
   }
 }
 
@@ -196,6 +234,12 @@ bool ARMAsmBackend::fixupNeedsRelaxation(const MCFixup &Fixup,
     int64_t Offset = int64_t(Value) - 4;
     return Offset > 1020 || Offset < 0 || Offset & 3;
   }
+  case ARM::fixup_arm_thumb_cb:
+    // If we have a Thumb CBZ or CBNZ instruction and its target is the next
+    // instruction it is is actually out of range for the instruction.
+    // It will be changed to a NOP.
+    int64_t Offset = (Value & ~1);
+    return Offset == 2;
   }
   llvm_unreachable("Unexpected fixup kind in fixupNeedsRelaxation()!");
 }
@@ -212,7 +256,18 @@ void ARMAsmBackend::relaxInstruction(const MCInst &Inst, MCInst &Res) const {
     report_fatal_error("unexpected instruction to relax: " + OS.str());
   }
 
-  // The instructions we're relaxing have (so far) the same operands.
+  // If we are changing Thumb CBZ or CBNZ instruction to a NOP, aka tHINT, we
+  // have to change the operands too.
+  if ((Inst.getOpcode() == ARM::tCBZ || Inst.getOpcode() == ARM::tCBNZ) &&
+      RelaxedOp == ARM::tHINT) {
+    Res.setOpcode(RelaxedOp);
+    Res.addOperand(MCOperand::CreateImm(0));
+    Res.addOperand(MCOperand::CreateImm(14));
+    Res.addOperand(MCOperand::CreateReg(0));
+    return;
+  } 
+
+  // The rest of instructions we're relaxing have the same operands.
   // We just need to update to the proper opcode.
   Res = Inst;
   Res.setOpcode(RelaxedOp);
@@ -251,8 +306,36 @@ bool ARMAsmBackend::writeNopData(uint64_t Count, MCObjectWriter *OW) const {
   return true;
 }
 
+static uint32_t swapHalfWords(uint32_t Value, bool IsLittleEndian) {
+  if (IsLittleEndian) {
+    // Note that the halfwords are stored high first and low second in thumb;
+    // so we need to swap the fixup value here to map properly.
+    uint32_t Swapped = (Value & 0xFFFF0000) >> 16;
+    Swapped |= (Value & 0x0000FFFF) << 16;
+    return Swapped;
+  }
+  else
+    return Value;
+}
+
+static uint32_t joinHalfWords(uint32_t FirstHalf, uint32_t SecondHalf,
+                              bool IsLittleEndian) {
+  uint32_t Value;
+
+  if (IsLittleEndian) {
+    Value = (SecondHalf & 0xFFFF) << 16;
+    Value |= (FirstHalf & 0xFFFF);
+  } else {
+    Value = (SecondHalf & 0xFFFF);
+    Value |= (FirstHalf & 0xFFFF) << 16;
+  }
+
+  return Value;
+}
+
 static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
-                                 MCContext *Ctx = NULL) {
+                                 bool IsPCRel, MCContext *Ctx,
+                                 bool IsLittleEndian) {
   unsigned Kind = Fixup.getKind();
   switch (Kind) {
   default:
@@ -261,12 +344,15 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
   case FK_Data_2:
   case FK_Data_4:
     return Value;
+  case FK_SecRel_2:
+    return Value;
+  case FK_SecRel_4:
+    return Value;
   case ARM::fixup_arm_movt_hi16:
-    Value >>= 16;
+    if (!IsPCRel)
+      Value >>= 16;
     // Fallthrough
-  case ARM::fixup_arm_movw_lo16:
-  case ARM::fixup_arm_movt_hi16_pcrel:
-  case ARM::fixup_arm_movw_lo16_pcrel: {
+  case ARM::fixup_arm_movw_lo16: {
     unsigned Hi4 = (Value & 0xF000) >> 12;
     unsigned Lo12 = Value & 0x0FFF;
     // inst{19-16} = Hi4;
@@ -275,12 +361,10 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
     return Value;
   }
   case ARM::fixup_t2_movt_hi16:
-    Value >>= 16;
+    if (!IsPCRel)
+      Value >>= 16;
     // Fallthrough
-  case ARM::fixup_t2_movw_lo16:
-  case ARM::fixup_t2_movt_hi16_pcrel:  //FIXME: Shouldn't this be shifted like
-                                       // the other hi16 fixup?
-  case ARM::fixup_t2_movw_lo16_pcrel: {
+  case ARM::fixup_t2_movw_lo16: {
     unsigned Hi4 = (Value & 0xF000) >> 12;
     unsigned i = (Value & 0x800) >> 11;
     unsigned Mid3 = (Value & 0x700) >> 8;
@@ -290,9 +374,7 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
     // inst{14-12} = Mid3;
     // inst{7-0} = Lo8;
     Value = (Hi4 << 16) | (i << 26) | (Mid3 << 12) | (Lo8);
-    uint64_t swapped = (Value & 0xFFFF0000) >> 16;
-    swapped |= (Value & 0x0000FFFF) << 16;
-    return swapped;
+    return swapHalfWords(Value, IsLittleEndian);
   }
   case ARM::fixup_arm_ldst_pcrel_12:
     // ARM PC-relative values are offset by 8.
@@ -312,11 +394,8 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
 
     // Same addressing mode as fixup_arm_pcrel_10,
     // but with 16-bit halfwords swapped.
-    if (Kind == ARM::fixup_t2_ldst_pcrel_12) {
-      uint64_t swapped = (Value & 0xFFFF0000) >> 16;
-      swapped |= (Value & 0x0000FFFF) << 16;
-      return swapped;
-    }
+    if (Kind == ARM::fixup_t2_ldst_pcrel_12)
+      return swapHalfWords(Value, IsLittleEndian);
 
     return Value;
   }
@@ -349,9 +428,7 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
     out |= (Value & 0x700) << 4;
     out |= (Value & 0x0FF);
 
-    uint64_t swapped = (out & 0xFFFF0000) >> 16;
-    swapped |= (out & 0x0000FFFF) << 16;
-    return swapped;
+    return swapHalfWords(out, IsLittleEndian);
   }
 
   case ARM::fixup_arm_condbranch:
@@ -361,6 +438,9 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
   case ARM::fixup_arm_blx:
     // These values don't encode the low two bits since they're always zero.
     // Offset by 8 just as above.
+    if (const MCSymbolRefExpr *SRE = dyn_cast<MCSymbolRefExpr>(Fixup.getValue()))
+      if (SRE->getKind() == MCSymbolRefExpr::VK_ARM_TLSCALL)
+        return 0;
     return 0xffffff & ((Value - 8) >> 2);
   case ARM::fixup_t2_uncondbranch: {
     Value = Value - 4;
@@ -379,9 +459,7 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
     out |= (Value & 0x1FF800)  << 5; // imm6 field
     out |= (Value & 0x0007FF);        // imm11 field
 
-    uint64_t swapped = (out & 0xFFFF0000) >> 16;
-    swapped |= (out & 0x0000FFFF) << 16;
-    return swapped;
+    return swapHalfWords(out, IsLittleEndian);
   }
   case ARM::fixup_t2_condbranch: {
     Value = Value - 4;
@@ -394,70 +472,64 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
     out |= (Value & 0x1F800) << 5; // imm6 field
     out |= (Value & 0x007FF);      // imm11 field
 
-    uint32_t swapped = (out & 0xFFFF0000) >> 16;
-    swapped |= (out & 0x0000FFFF) << 16;
-    return swapped;
+    return swapHalfWords(out, IsLittleEndian);
   }
   case ARM::fixup_arm_thumb_bl: {
-     // The value doesn't encode the low bit (always zero) and is offset by
-     // four. The 32-bit immediate value is encoded as
-     //   imm32 = SignExtend(S:I1:I2:imm10:imm11:0)
-     // where I1 = NOT(J1 ^ S) and I2 = NOT(J2 ^ S).
-     // The value is encoded into disjoint bit positions in the destination
-     // opcode. x = unchanged, I = immediate value bit, S = sign extension bit,
-     // J = either J1 or J2 bit
-     //
-     //   BL:  xxxxxSIIIIIIIIII xxJxJIIIIIIIIIII
-     //
-     // Note that the halfwords are stored high first, low second; so we need
-     // to transpose the fixup value here to map properly.
-     uint32_t offset = (Value - 4) >> 1;
-     uint32_t signBit = (offset & 0x800000) >> 23;
-     uint32_t I1Bit = (offset & 0x400000) >> 22;
-     uint32_t J1Bit = (I1Bit ^ 0x1) ^ signBit;
-     uint32_t I2Bit = (offset & 0x200000) >> 21;
-     uint32_t J2Bit = (I2Bit ^ 0x1) ^ signBit;
-     uint32_t imm10Bits = (offset & 0x1FF800) >> 11;
-     uint32_t imm11Bits = (offset & 0x000007FF);
-
-     uint32_t Binary = 0;
-     uint32_t firstHalf = (((uint16_t)signBit << 10) | (uint16_t)imm10Bits);
-     uint32_t secondHalf = (((uint16_t)J1Bit << 13) | ((uint16_t)J2Bit << 11) |
-                           (uint16_t)imm11Bits);
-     Binary |= secondHalf << 16;
-     Binary |= firstHalf;
-     return Binary;
-
+    // The value doesn't encode the low bit (always zero) and is offset by
+    // four. The 32-bit immediate value is encoded as
+    //   imm32 = SignExtend(S:I1:I2:imm10:imm11:0)
+    // where I1 = NOT(J1 ^ S) and I2 = NOT(J2 ^ S).
+    // The value is encoded into disjoint bit positions in the destination
+    // opcode. x = unchanged, I = immediate value bit, S = sign extension bit,
+    // J = either J1 or J2 bit
+    //
+    //   BL:  xxxxxSIIIIIIIIII xxJxJIIIIIIIIIII
+    //
+    // Note that the halfwords are stored high first, low second; so we need
+    // to transpose the fixup value here to map properly.
+    uint32_t offset = (Value - 4) >> 1;
+    uint32_t signBit = (offset & 0x800000) >> 23;
+    uint32_t I1Bit = (offset & 0x400000) >> 22;
+    uint32_t J1Bit = (I1Bit ^ 0x1) ^ signBit;
+    uint32_t I2Bit = (offset & 0x200000) >> 21;
+    uint32_t J2Bit = (I2Bit ^ 0x1) ^ signBit;
+    uint32_t imm10Bits = (offset & 0x1FF800) >> 11;
+    uint32_t imm11Bits = (offset & 0x000007FF);
+
+    uint32_t FirstHalf = (((uint16_t)signBit << 10) | (uint16_t)imm10Bits);
+    uint32_t SecondHalf = (((uint16_t)J1Bit << 13) | ((uint16_t)J2Bit << 11) |
+                          (uint16_t)imm11Bits);
+    return joinHalfWords(FirstHalf, SecondHalf, IsLittleEndian);
   }
   case ARM::fixup_arm_thumb_blx: {
-     // The value doesn't encode the low two bits (always zero) and is offset by
-     // four (see fixup_arm_thumb_cp). The 32-bit immediate value is encoded as
-     //   imm32 = SignExtend(S:I1:I2:imm10H:imm10L:00)
-     // where I1 = NOT(J1 ^ S) and I2 = NOT(J2 ^ S).
-     // The value is encoded into disjoint bit positions in the destination
-     // opcode. x = unchanged, I = immediate value bit, S = sign extension bit,
-     // J = either J1 or J2 bit, 0 = zero.
-     //
-     //   BLX: xxxxxSIIIIIIIIII xxJxJIIIIIIIIII0
-     //
-     // Note that the halfwords are stored high first, low second; so we need
-     // to transpose the fixup value here to map properly.
-     uint32_t offset = (Value - 2) >> 2;
-     uint32_t signBit = (offset & 0x400000) >> 22;
-     uint32_t I1Bit = (offset & 0x200000) >> 21;
-     uint32_t J1Bit = (I1Bit ^ 0x1) ^ signBit;
-     uint32_t I2Bit = (offset & 0x100000) >> 20;
-     uint32_t J2Bit = (I2Bit ^ 0x1) ^ signBit;
-     uint32_t imm10HBits = (offset & 0xFFC00) >> 10;
-     uint32_t imm10LBits = (offset & 0x3FF);
-
-     uint32_t Binary = 0;
-     uint32_t firstHalf = (((uint16_t)signBit << 10) | (uint16_t)imm10HBits);
-     uint32_t secondHalf = (((uint16_t)J1Bit << 13) | ((uint16_t)J2Bit << 11) |
-                           ((uint16_t)imm10LBits) << 1);
-     Binary |= secondHalf << 16;
-     Binary |= firstHalf;
-     return Binary;
+    // The value doesn't encode the low two bits (always zero) and is offset by
+    // four (see fixup_arm_thumb_cp). The 32-bit immediate value is encoded as
+    //   imm32 = SignExtend(S:I1:I2:imm10H:imm10L:00)
+    // where I1 = NOT(J1 ^ S) and I2 = NOT(J2 ^ S).
+    // The value is encoded into disjoint bit positions in the destination
+    // opcode. x = unchanged, I = immediate value bit, S = sign extension bit,
+    // J = either J1 or J2 bit, 0 = zero.
+    //
+    //   BLX: xxxxxSIIIIIIIIII xxJxJIIIIIIIIII0
+    //
+    // Note that the halfwords are stored high first, low second; so we need
+    // to transpose the fixup value here to map properly.
+    uint32_t offset = (Value - 2) >> 2;
+    if (const MCSymbolRefExpr *SRE = dyn_cast<MCSymbolRefExpr>(Fixup.getValue()))
+      if (SRE->getKind() == MCSymbolRefExpr::VK_ARM_TLSCALL)
+        offset = 0;
+    uint32_t signBit = (offset & 0x400000) >> 22;
+    uint32_t I1Bit = (offset & 0x200000) >> 21;
+    uint32_t J1Bit = (I1Bit ^ 0x1) ^ signBit;
+    uint32_t I2Bit = (offset & 0x100000) >> 20;
+    uint32_t J2Bit = (I2Bit ^ 0x1) ^ signBit;
+    uint32_t imm10HBits = (offset & 0xFFC00) >> 10;
+    uint32_t imm10LBits = (offset & 0x3FF);
+
+    uint32_t FirstHalf = (((uint16_t)signBit << 10) | (uint16_t)imm10HBits);
+    uint32_t SecondHalf = (((uint16_t)J1Bit << 13) | ((uint16_t)J2Bit << 11) |
+                          ((uint16_t)imm10LBits) << 1);
+    return joinHalfWords(FirstHalf, SecondHalf, IsLittleEndian);
   }
   case ARM::fixup_arm_thumb_cp:
     // Offset by 4, and don't encode the low two bits. Two bytes of that
@@ -509,11 +581,8 @@ static unsigned adjustFixupValue(const MCFixup &Fixup, uint64_t Value,
 
     // Same addressing mode as fixup_arm_pcrel_10, but with 16-bit halfwords
     // swapped.
-    if (Kind == ARM::fixup_t2_pcrel_10) {
-      uint32_t swapped = (Value & 0xFFFF0000) >> 16;
-      swapped |= (Value & 0x0000FFFF) << 16;
-      return swapped;
-    }
+    if (Kind == ARM::fixup_t2_pcrel_10)
+      return swapHalfWords(Value, IsLittleEndian);
 
     return Value;
   }
@@ -524,7 +593,7 @@ void ARMAsmBackend::processFixupValue(const MCAssembler &Asm,
                                       const MCAsmLayout &Layout,
                                       const MCFixup &Fixup,
                                       const MCFragment *DF,
-                                      MCValue &Target, uint64_t &Value,
+                                      const MCValue &Target, uint64_t &Value,
                                       bool &IsResolved) {
   const MCSymbolRefExpr *A = Target.getSymA();
   // Some fixups to thumb function symbols need the low bit (thumb bit)
@@ -541,11 +610,18 @@ void ARMAsmBackend::processFixupValue(const MCAssembler &Asm,
         Value |= 1;
     }
   }
+  // For Thumb1 BL instruction, it is possible to be a long jump between
+  // the basic blocks of the same function.  Thus, we would like to resolve
+  // the offset when the destination has the same MCFragment.
+  if (A && (unsigned)Fixup.getKind() == ARM::fixup_arm_thumb_bl) {
+    const MCSymbol &Sym = A->getSymbol().AliasedSymbol();
+    const MCSymbolData &SymData = Asm.getSymbolData(Sym);
+    IsResolved = (SymData.getFragment() == DF);
+  }
   // We must always generate a relocation for BL/BLX instructions if we have
   // a symbol to reference, as the linker relies on knowing the destination
   // symbol's thumb-ness to get interworking right.
   if (A && ((unsigned)Fixup.getKind() == ARM::fixup_arm_thumb_blx ||
-            (unsigned)Fixup.getKind() == ARM::fixup_arm_thumb_bl ||
             (unsigned)Fixup.getKind() == ARM::fixup_arm_blx ||
             (unsigned)Fixup.getKind() == ARM::fixup_arm_uncondbl ||
             (unsigned)Fixup.getKind() == ARM::fixup_arm_condbl))
@@ -554,7 +630,8 @@ void ARMAsmBackend::processFixupValue(const MCAssembler &Asm,
   // Try to get the encoded value for the fixup as-if we're mapping it into
   // the instruction. This allows adjustFixupValue() to issue a diagnostic
   // if the value aren't invalid.
-  (void)adjustFixupValue(Fixup, Value, &Asm.getContext());
+  (void)adjustFixupValue(Fixup, Value, false, &Asm.getContext(),
+                         IsLittleEndian);
 }
 
 /// getFixupKindNumBytes - The number of bytes the fixup may change.
@@ -595,33 +672,101 @@ static unsigned getFixupKindNumBytes(unsigned Kind) {
   case ARM::fixup_arm_thumb_blx:
   case ARM::fixup_arm_movt_hi16:
   case ARM::fixup_arm_movw_lo16:
-  case ARM::fixup_arm_movt_hi16_pcrel:
-  case ARM::fixup_arm_movw_lo16_pcrel:
   case ARM::fixup_t2_movt_hi16:
   case ARM::fixup_t2_movw_lo16:
-  case ARM::fixup_t2_movt_hi16_pcrel:
-  case ARM::fixup_t2_movw_lo16_pcrel:
+    return 4;
+
+  case FK_SecRel_2:
+    return 2;
+  case FK_SecRel_4:
+    return 4;
+  }
+}
+
+/// getFixupKindContainerSizeBytes - The number of bytes of the
+/// container involved in big endian.
+static unsigned getFixupKindContainerSizeBytes(unsigned Kind) {
+  switch (Kind) {
+  default:
+    llvm_unreachable("Unknown fixup kind!");
+
+  case FK_Data_1:
+    return 1;
+  case FK_Data_2:
+    return 2;
+  case FK_Data_4:
+    return 4;
+
+  case ARM::fixup_arm_thumb_bcc:
+  case ARM::fixup_arm_thumb_cp:
+  case ARM::fixup_thumb_adr_pcrel_10:
+  case ARM::fixup_arm_thumb_br:
+  case ARM::fixup_arm_thumb_cb:
+    // Instruction size is 2 bytes.
+    return 2;
+
+  case ARM::fixup_arm_pcrel_10_unscaled:
+  case ARM::fixup_arm_ldst_pcrel_12:
+  case ARM::fixup_arm_pcrel_10:
+  case ARM::fixup_arm_adr_pcrel_12:
+  case ARM::fixup_arm_uncondbl:
+  case ARM::fixup_arm_condbl:
+  case ARM::fixup_arm_blx:
+  case ARM::fixup_arm_condbranch:
+  case ARM::fixup_arm_uncondbranch:
+  case ARM::fixup_t2_ldst_pcrel_12:
+  case ARM::fixup_t2_condbranch:
+  case ARM::fixup_t2_uncondbranch:
+  case ARM::fixup_t2_pcrel_10:
+  case ARM::fixup_t2_adr_pcrel_12:
+  case ARM::fixup_arm_thumb_bl:
+  case ARM::fixup_arm_thumb_blx:
+  case ARM::fixup_arm_movt_hi16:
+  case ARM::fixup_arm_movw_lo16:
+  case ARM::fixup_t2_movt_hi16:
+  case ARM::fixup_t2_movw_lo16:
+    // Instruction size is 4 bytes.
     return 4;
   }
 }
 
 void ARMAsmBackend::applyFixup(const MCFixup &Fixup, char *Data,
-                               unsigned DataSize, uint64_t Value) const {
+                               unsigned DataSize, uint64_t Value,
+                               bool IsPCRel) const {
   unsigned NumBytes = getFixupKindNumBytes(Fixup.getKind());
-  Value = adjustFixupValue(Fixup, Value);
+  Value = adjustFixupValue(Fixup, Value, IsPCRel, nullptr, IsLittleEndian);
   if (!Value) return;           // Doesn't change encoding.
 
   unsigned Offset = Fixup.getOffset();
   assert(Offset + NumBytes <= DataSize && "Invalid fixup offset!");
 
+  // Used to point to big endian bytes.
+  unsigned FullSizeBytes;
+  if (!IsLittleEndian) {
+    FullSizeBytes = getFixupKindContainerSizeBytes(Fixup.getKind());
+    assert((Offset + FullSizeBytes) <= DataSize && "Invalid fixup size!");
+    assert(NumBytes <= FullSizeBytes && "Invalid fixup size!");
+  }
+
   // For each byte of the fragment that the fixup touches, mask in the bits from
   // the fixup value. The Value has been "split up" into the appropriate
   // bitfields above.
-  for (unsigned i = 0; i != NumBytes; ++i)
-    Data[Offset + i] |= uint8_t((Value >> (i * 8)) & 0xff);
+  for (unsigned i = 0; i != NumBytes; ++i) {
+    unsigned Idx = IsLittleEndian ? i : (FullSizeBytes - 1 - i);
+    Data[Offset + Idx] |= uint8_t((Value >> (i * 8)) & 0xff);
+  }
 }
 
 namespace {
+// FIXME: This should be in a separate file.
+class ARMWinCOFFAsmBackend : public ARMAsmBackend {
+public:
+  ARMWinCOFFAsmBackend(const Target &T, const StringRef &Triple)
+    : ARMAsmBackend(T, Triple, true) { }
+  MCObjectWriter *createObjectWriter(raw_ostream &OS) const override {
+    return createARMWinCOFFObjectWriter(OS, /*Is64Bit=*/false);
+  }
+};
 
 // FIXME: This should be in a separate file.
 // ELF is an ELF of course...
@@ -629,11 +774,11 @@ class ELFARMAsmBackend : public ARMAsmBackend {
 public:
   uint8_t OSABI;
   ELFARMAsmBackend(const Target &T, const StringRef TT,
-                   uint8_t _OSABI)
-    : ARMAsmBackend(T, TT), OSABI(_OSABI) { }
+                   uint8_t OSABI, bool IsLittle)
+    : ARMAsmBackend(T, TT, IsLittle), OSABI(OSABI) { }
 
-  MCObjectWriter *createObjectWriter(raw_ostream &OS) const {
-    return createARMELFObjectWriter(OS, OSABI);
+  MCObjectWriter *createObjectWriter(raw_ostream &OS) const override {
+    return createARMELFObjectWriter(OS, OSABI, isLittle());
   }
 };
 
@@ -643,29 +788,28 @@ public:
   const MachO::CPUSubTypeARM Subtype;
   DarwinARMAsmBackend(const Target &T, const StringRef TT,
                       MachO::CPUSubTypeARM st)
-    : ARMAsmBackend(T, TT), Subtype(st) {
+    : ARMAsmBackend(T, TT, /* IsLittleEndian */ true), Subtype(st) {
       HasDataInCodeSupport = true;
     }
 
-  MCObjectWriter *createObjectWriter(raw_ostream &OS) const {
+  MCObjectWriter *createObjectWriter(raw_ostream &OS) const override {
     return createARMMachObjectWriter(OS, /*Is64Bit=*/false,
                                      MachO::CPU_TYPE_ARM,
                                      Subtype);
   }
-
-  virtual bool doesSectionRequireSymbols(const MCSection &Section) const {
-    return false;
-  }
 };
 
 } // end anonymous namespace
 
 MCAsmBackend *llvm::createARMAsmBackend(const Target &T,
                                         const MCRegisterInfo &MRI,
-                                        StringRef TT, StringRef CPU) {
+                                        StringRef TT, StringRef CPU,
+                                        bool isLittle) {
   Triple TheTriple(TT);
 
-  if (TheTriple.isOSDarwin()) {
+  switch (TheTriple.getObjectFormat()) {
+  default: llvm_unreachable("unsupported object format");
+  case Triple::MachO: {
     MachO::CPUSubTypeARM CS =
       StringSwitch<MachO::CPUSubTypeARM>(TheTriple.getArchName())
       .Cases("armv4t", "thumbv4t", MachO::CPU_SUBTYPE_ARM_V4T)
@@ -673,7 +817,6 @@ MCAsmBackend *llvm::createARMAsmBackend(const Target &T,
       .Cases("armv6", "thumbv6", MachO::CPU_SUBTYPE_ARM_V6)
       .Cases("armv6m", "thumbv6m", MachO::CPU_SUBTYPE_ARM_V6M)
       .Cases("armv7em", "thumbv7em", MachO::CPU_SUBTYPE_ARM_V7EM)
-      .Cases("armv7f", "thumbv7f", MachO::CPU_SUBTYPE_ARM_V7F)
       .Cases("armv7k", "thumbv7k", MachO::CPU_SUBTYPE_ARM_V7K)
       .Cases("armv7m", "thumbv7m", MachO::CPU_SUBTYPE_ARM_V7M)
       .Cases("armv7s", "thumbv7s", MachO::CPU_SUBTYPE_ARM_V7S)
@@ -681,13 +824,37 @@ MCAsmBackend *llvm::createARMAsmBackend(const Target &T,
 
     return new DarwinARMAsmBackend(T, TT, CS);
   }
+  case Triple::COFF:
+    assert(TheTriple.isOSWindows() && "non-Windows ARM COFF is not supported");
+    return new ARMWinCOFFAsmBackend(T, TT);
+  case Triple::ELF:
+    assert(TheTriple.isOSBinFormatELF() && "using ELF for non-ELF target");
+    uint8_t OSABI = MCELFObjectTargetWriter::getOSABI(Triple(TT).getOS());
+    return new ELFARMAsmBackend(T, TT, OSABI, isLittle);
+  }
+}
 
-#if 0
-  // FIXME: Introduce yet another checker but assert(0).
-  if (TheTriple.isOSBinFormatCOFF())
-    assert(0 && "Windows not supported on ARM");
-#endif
+MCAsmBackend *llvm::createARMLEAsmBackend(const Target &T,
+                                          const MCRegisterInfo &MRI,
+                                          StringRef TT, StringRef CPU) {
+  return createARMAsmBackend(T, MRI, TT, CPU, true);
+}
+
+MCAsmBackend *llvm::createARMBEAsmBackend(const Target &T,
+                                          const MCRegisterInfo &MRI,
+                                          StringRef TT, StringRef CPU) {
+  return createARMAsmBackend(T, MRI, TT, CPU, false);
+}
 
-  uint8_t OSABI = MCELFObjectTargetWriter::getOSABI(Triple(TT).getOS());
-  return new ELFARMAsmBackend(T, TT, OSABI);
+MCAsmBackend *llvm::createThumbLEAsmBackend(const Target &T,
+                                          const MCRegisterInfo &MRI,
+                                          StringRef TT, StringRef CPU) {
+  return createARMAsmBackend(T, MRI, TT, CPU, true);
 }
+
+MCAsmBackend *llvm::createThumbBEAsmBackend(const Target &T,
+                                          const MCRegisterInfo &MRI,
+                                          StringRef TT, StringRef CPU) {
+  return createARMAsmBackend(T, MRI, TT, CPU, false);
+}
+
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h
index af939fc19129..1686d76b8e1b 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h
@@ -183,7 +183,8 @@ namespace ARM_ISB {
 
   inline static const char *InstSyncBOptToString(unsigned val) {
     switch (val) {
-      default: llvm_unreachable("Unkown memory operation");
+    default:
+      llvm_unreachable("Unknown memory operation");
       case RESERVED_0:  return "#0x0";
       case RESERVED_1:  return "#0x1";
       case RESERVED_2:  return "#0x2";
@@ -278,42 +279,41 @@ namespace ARMII {
     //===------------------------------------------------------------------===//
     // ARM Specific MachineOperand flags.
 
-    MO_NO_FLAG,
+    MO_NO_FLAG = 0,
 
     /// MO_LO16 - On a symbol operand, this represents a relocation containing
     /// lower 16 bit of the address. Used only via movw instruction.
-    MO_LO16,
+    MO_LO16 = 0x1,
 
     /// MO_HI16 - On a symbol operand, this represents a relocation containing
     /// higher 16 bit of the address. Used only via movt instruction.
-    MO_HI16,
-
-    /// MO_LO16_NONLAZY - On a symbol operand "FOO", this represents a
-    /// relocation containing lower 16 bit of the non-lazy-ptr indirect symbol,
-    /// i.e. "FOO$non_lazy_ptr".
-    /// Used only via movw instruction.
-    MO_LO16_NONLAZY,
-
-    /// MO_HI16_NONLAZY - On a symbol operand "FOO", this represents a
-    /// relocation containing lower 16 bit of the non-lazy-ptr indirect symbol,
-    /// i.e. "FOO$non_lazy_ptr". Used only via movt instruction.
-    MO_HI16_NONLAZY,
-
-    /// MO_LO16_NONLAZY_PIC - On a symbol operand "FOO", this represents a
-    /// relocation containing lower 16 bit of the PC relative address of the
-    /// non-lazy-ptr indirect symbol, i.e. "FOO$non_lazy_ptr - LABEL".
-    /// Used only via movw instruction.
-    MO_LO16_NONLAZY_PIC,
-
-    /// MO_HI16_NONLAZY_PIC - On a symbol operand "FOO", this represents a
-    /// relocation containing lower 16 bit of the PC relative address of the
-    /// non-lazy-ptr indirect symbol, i.e. "FOO$non_lazy_ptr - LABEL".
-    /// Used only via movt instruction.
-    MO_HI16_NONLAZY_PIC,
+    MO_HI16 = 0x2,
 
     /// MO_PLT - On a symbol operand, this represents an ELF PLT reference on a
     /// call operand.
-    MO_PLT
+    MO_PLT = 0x3,
+
+    /// MO_OPTION_MASK - Most flags are mutually exclusive; this mask selects
+    /// just that part of the flag set.
+    MO_OPTION_MASK = 0x3f,
+
+    /// MO_DLLIMPORT - On a symbol operand, this represents that the reference
+    /// to the symbol is for an import stub.  This is used for DLL import
+    /// storage class indication on Windows.
+    MO_DLLIMPORT = 0x40,
+
+    /// MO_NONLAZY - This is an independent flag, on a symbol operand "FOO" it
+    /// represents a symbol which, if indirect, will get special Darwin mangling
+    /// as a non-lazy-ptr indirect symbol (i.e. "L_FOO$non_lazy_ptr"). Can be
+    /// combined with MO_LO16, MO_HI16 or MO_NO_FLAG (in a constant-pool, for
+    /// example).
+    MO_NONLAZY = 0x80,
+
+    // It's undefined behaviour if an enum overflows the range between its
+    // smallest and largest values, but since these are |ed together, it can
+    // happen. Put a sentinel in (values of this enum are stored as "unsigned
+    // char").
+    MO_UNUSED_MAXIMUM = 0xff
   };
 
   enum {
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFObjectWriter.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFObjectWriter.cpp
index f98bbd204c7a..a86601b6bb5b 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFObjectWriter.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFObjectWriter.cpp
@@ -34,14 +34,11 @@ namespace {
 
     virtual ~ARMELFObjectWriter();
 
-    virtual unsigned GetRelocType(const MCValue &Target, const MCFixup &Fixup,
-                                  bool IsPCRel, bool IsRelocWithSymbol,
-                                  int64_t Addend) const;
-    virtual const MCSymbol *ExplicitRelSym(const MCAssembler &Asm,
-                                   const MCValue &Target,
-                                   const MCFragment &F,
-                                   const MCFixup &Fixup,
-                                   bool IsPCRel) const;
+    unsigned GetRelocType(const MCValue &Target, const MCFixup &Fixup,
+                          bool IsPCRel) const override;
+
+    bool needsRelocateWithSymbol(const MCSymbolData &SD,
+                                 unsigned Type) const override;
   };
 }
 
@@ -52,91 +49,19 @@ ARMELFObjectWriter::ARMELFObjectWriter(uint8_t OSABI)
 
 ARMELFObjectWriter::~ARMELFObjectWriter() {}
 
-// In ARM, _MergedGlobals and other most symbols get emitted directly.
-// I.e. not as an offset to a section symbol.
-// This code is an approximation of what ARM/gcc does.
-
-STATISTIC(PCRelCount, "Total number of PIC Relocations");
-STATISTIC(NonPCRelCount, "Total number of non-PIC relocations");
-
-const MCSymbol *ARMELFObjectWriter::ExplicitRelSym(const MCAssembler &Asm,
-                                                   const MCValue &Target,
-                                                   const MCFragment &F,
-                                                   const MCFixup &Fixup,
-                                                   bool IsPCRel) const {
-  const MCSymbol &Symbol = Target.getSymA()->getSymbol().AliasedSymbol();
-  bool EmitThisSym = false;
-
-  const MCSectionELF &Section =
-    static_cast<const MCSectionELF&>(Symbol.getSection());
-  bool InNormalSection = true;
-  unsigned RelocType = 0;
-  RelocType = GetRelocTypeInner(Target, Fixup, IsPCRel);
-
-  DEBUG(
-      const MCSymbolRefExpr::VariantKind Kind = Target.getSymA()->getKind();
-      MCSymbolRefExpr::VariantKind Kind2;
-      Kind2 = Target.getSymB() ?  Target.getSymB()->getKind() :
-        MCSymbolRefExpr::VK_None;
-      dbgs() << "considering symbol "
-        << Section.getSectionName() << "/"
-        << Symbol.getName() << "/"
-        << " Rel:" << (unsigned)RelocType
-        << " Kind: " << (int)Kind << "/" << (int)Kind2
-        << " Tmp:"
-        << Symbol.isAbsolute() << "/" << Symbol.isDefined() << "/"
-        << Symbol.isVariable() << "/" << Symbol.isTemporary()
-        << " Counts:" << PCRelCount << "/" << NonPCRelCount << "\n");
-
-  if (IsPCRel) { ++PCRelCount;
-    switch (RelocType) {
-    default:
-      // Most relocation types are emitted as explicit symbols
-      InNormalSection =
-        StringSwitch<bool>(Section.getSectionName())
-        .Case(".data.rel.ro.local", false)
-        .Case(".data.rel", false)
-        .Case(".bss", false)
-        .Default(true);
-      EmitThisSym = true;
-      break;
-    case ELF::R_ARM_ABS32:
-      // But things get strange with R_ARM_ABS32
-      // In this case, most things that go in .rodata show up
-      // as section relative relocations
-      InNormalSection =
-        StringSwitch<bool>(Section.getSectionName())
-        .Case(".data.rel.ro.local", false)
-        .Case(".data.rel", false)
-        .Case(".rodata", false)
-        .Case(".bss", false)
-        .Default(true);
-      EmitThisSym = false;
-      break;
-    }
-  } else {
-    NonPCRelCount++;
-    InNormalSection =
-      StringSwitch<bool>(Section.getSectionName())
-      .Case(".data.rel.ro.local", false)
-      .Case(".rodata", false)
-      .Case(".data.rel", false)
-      .Case(".bss", false)
-      .Default(true);
-
-    switch (RelocType) {
-    default: EmitThisSym = true; break;
-    case ELF::R_ARM_ABS32: EmitThisSym = false; break;
-    case ELF::R_ARM_PREL31: EmitThisSym = false; break;
-    }
-  }
-
-  if (EmitThisSym)
-    return &Symbol;
-  if (! Symbol.isTemporary() && InNormalSection) {
-    return &Symbol;
+bool ARMELFObjectWriter::needsRelocateWithSymbol(const MCSymbolData &SD,
+                                                 unsigned Type) const {
+  // FIXME: This is extremelly conservative. This really needs to use a
+  // whitelist with a clear explanation for why each realocation needs to
+  // point to the symbol, not to the section.
+  switch (Type) {
+  default:
+    return true;
+
+  case ELF::R_ARM_PREL31:
+  case ELF::R_ARM_ABS32:
+    return false;
   }
-  return NULL;
 }
 
 // Need to examine the Fixup when determining whether to 
@@ -144,17 +69,14 @@ const MCSymbol *ARMELFObjectWriter::ExplicitRelSym(const MCAssembler &Asm,
 // offset
 unsigned ARMELFObjectWriter::GetRelocType(const MCValue &Target,
                                           const MCFixup &Fixup,
-                                          bool IsPCRel,
-                                          bool IsRelocWithSymbol,
-                                          int64_t Addend) const {
+                                          bool IsPCRel) const {
   return GetRelocTypeInner(Target, Fixup, IsPCRel);
 }
 
 unsigned ARMELFObjectWriter::GetRelocTypeInner(const MCValue &Target,
                                                const MCFixup &Fixup,
                                                bool IsPCRel) const  {
-  MCSymbolRefExpr::VariantKind Modifier = Target.isAbsolute() ?
-    MCSymbolRefExpr::VK_None : Target.getSymA()->getKind();
+  MCSymbolRefExpr::VariantKind Modifier = Target.getAccessVariant();
 
   unsigned Type = 0;
   if (IsPCRel) {
@@ -166,19 +88,25 @@ unsigned ARMELFObjectWriter::GetRelocTypeInner(const MCValue &Target,
       case MCSymbolRefExpr::VK_None:
         Type = ELF::R_ARM_REL32;
         break;
-      case MCSymbolRefExpr::VK_ARM_TLSGD:
+      case MCSymbolRefExpr::VK_TLSGD:
         llvm_unreachable("unimplemented");
-      case MCSymbolRefExpr::VK_ARM_GOTTPOFF:
+      case MCSymbolRefExpr::VK_GOTTPOFF:
         Type = ELF::R_ARM_TLS_IE32;
         break;
+      case MCSymbolRefExpr::VK_GOTPCREL:
+        Type = ELF::R_ARM_GOT_PREL;
+        break;
       }
       break;
     case ARM::fixup_arm_blx:
     case ARM::fixup_arm_uncondbl:
       switch (Modifier) {
-      case MCSymbolRefExpr::VK_ARM_PLT:
+      case MCSymbolRefExpr::VK_PLT:
         Type = ELF::R_ARM_PLT32;
         break;
+      case MCSymbolRefExpr::VK_ARM_TLSCALL:
+        Type = ELF::R_ARM_TLS_CALL;
+        break;
       default:
         Type = ELF::R_ARM_CALL;
         break;
@@ -194,24 +122,27 @@ unsigned ARMELFObjectWriter::GetRelocTypeInner(const MCValue &Target,
       Type = ELF::R_ARM_THM_JUMP24;
       break;
     case ARM::fixup_arm_movt_hi16:
-    case ARM::fixup_arm_movt_hi16_pcrel:
       Type = ELF::R_ARM_MOVT_PREL;
       break;
     case ARM::fixup_arm_movw_lo16:
-    case ARM::fixup_arm_movw_lo16_pcrel:
       Type = ELF::R_ARM_MOVW_PREL_NC;
       break;
     case ARM::fixup_t2_movt_hi16:
-    case ARM::fixup_t2_movt_hi16_pcrel:
       Type = ELF::R_ARM_THM_MOVT_PREL;
       break;
     case ARM::fixup_t2_movw_lo16:
-    case ARM::fixup_t2_movw_lo16_pcrel:
       Type = ELF::R_ARM_THM_MOVW_PREL_NC;
       break;
     case ARM::fixup_arm_thumb_bl:
     case ARM::fixup_arm_thumb_blx:
-      Type = ELF::R_ARM_THM_CALL;
+      switch (Modifier) {
+      case MCSymbolRefExpr::VK_ARM_TLSCALL:
+        Type = ELF::R_ARM_THM_TLS_CALL;
+        break;
+      default:
+        Type = ELF::R_ARM_THM_CALL;
+        break;
+      }
       break;
     }
   } else {
@@ -223,24 +154,27 @@ unsigned ARMELFObjectWriter::GetRelocTypeInner(const MCValue &Target,
       case MCSymbolRefExpr::VK_ARM_NONE:
         Type = ELF::R_ARM_NONE;
         break;
-      case MCSymbolRefExpr::VK_ARM_GOT:
+      case MCSymbolRefExpr::VK_GOT:
         Type = ELF::R_ARM_GOT_BREL;
         break;
-      case MCSymbolRefExpr::VK_ARM_TLSGD:
+      case MCSymbolRefExpr::VK_TLSGD:
         Type = ELF::R_ARM_TLS_GD32;
         break;
-      case MCSymbolRefExpr::VK_ARM_TPOFF:
+      case MCSymbolRefExpr::VK_TPOFF:
         Type = ELF::R_ARM_TLS_LE32;
         break;
-      case MCSymbolRefExpr::VK_ARM_GOTTPOFF:
+      case MCSymbolRefExpr::VK_GOTTPOFF:
         Type = ELF::R_ARM_TLS_IE32;
         break;
       case MCSymbolRefExpr::VK_None:
         Type = ELF::R_ARM_ABS32;
         break;
-      case MCSymbolRefExpr::VK_ARM_GOTOFF:
+      case MCSymbolRefExpr::VK_GOTOFF:
         Type = ELF::R_ARM_GOTOFF32;
         break;
+      case MCSymbolRefExpr::VK_GOTPCREL:
+        Type = ELF::R_ARM_GOT_PREL;
+        break;
       case MCSymbolRefExpr::VK_ARM_TARGET1:
         Type = ELF::R_ARM_TARGET1;
         break;
@@ -250,6 +184,18 @@ unsigned ARMELFObjectWriter::GetRelocTypeInner(const MCValue &Target,
       case MCSymbolRefExpr::VK_ARM_PREL31:
         Type = ELF::R_ARM_PREL31;
         break;
+      case MCSymbolRefExpr::VK_ARM_TLSLDO:
+        Type = ELF::R_ARM_TLS_LDO32;
+        break;
+      case MCSymbolRefExpr::VK_ARM_TLSCALL:
+        Type = ELF::R_ARM_TLS_CALL;
+        break;
+      case MCSymbolRefExpr::VK_ARM_TLSDESC:
+        Type = ELF::R_ARM_TLS_GOTDESC;
+        break;
+      case MCSymbolRefExpr::VK_ARM_TLSDESCSEQ:
+        Type = ELF::R_ARM_TLS_DESCSEQ;
+        break;
       }
       break;
     case ARM::fixup_arm_ldst_pcrel_12:
@@ -283,7 +229,8 @@ unsigned ARMELFObjectWriter::GetRelocTypeInner(const MCValue &Target,
 }
 
 MCObjectWriter *llvm::createARMELFObjectWriter(raw_ostream &OS,
-                                               uint8_t OSABI) {
+                                               uint8_t OSABI,
+                                               bool IsLittleEndian) {
   MCELFObjectTargetWriter *MOTW = new ARMELFObjectWriter(OSABI);
-  return createELFObjectWriter(MOTW, OS,  /*IsLittleEndian=*/true);
+  return createELFObjectWriter(MOTW, OS, IsLittleEndian);
 }
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFStreamer.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFStreamer.cpp
index 471897de5c1c..7b5d8b01dfe6 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFStreamer.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMELFStreamer.cpp
@@ -13,14 +13,14 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "ARMBuildAttrs.h"
+#include "ARMArchName.h"
 #include "ARMFPUName.h"
 #include "ARMRegisterInfo.h"
-#include "ARMUnwindOp.h"
 #include "ARMUnwindOpAsm.h"
-#include "llvm/ADT/SmallPtrSet.h"
+#include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/MC/MCAsmBackend.h"
+#include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCCodeEmitter.h"
 #include "llvm/MC/MCContext.h"
@@ -30,6 +30,7 @@
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCInst.h"
 #include "llvm/MC/MCInstPrinter.h"
+#include "llvm/MC/MCObjectFileInfo.h"
 #include "llvm/MC/MCObjectStreamer.h"
 #include "llvm/MC/MCRegisterInfo.h"
 #include "llvm/MC/MCSection.h"
@@ -37,16 +38,20 @@
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSymbol.h"
 #include "llvm/MC/MCValue.h"
+#include "llvm/Support/ARMBuildAttributes.h"
+#include "llvm/Support/ARMEHABI.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ELF.h"
 #include "llvm/Support/FormattedStream.h"
+#include "llvm/Support/LEB128.h"
 #include "llvm/Support/raw_ostream.h"
 #include <algorithm>
 
 using namespace llvm;
 
 static std::string GetAEABIUnwindPersonalityName(unsigned Index) {
-  assert(Index < NUM_PERSONALITY_INDEX && "Invalid personality index");
+  assert(Index < ARM::EHABI::NUM_PERSONALITY_INDEX &&
+         "Invalid personality index");
   return (Twine("__aeabi_unwind_cpp_pr") + Twine(Index)).str();
 }
 
@@ -58,7 +63,46 @@ static const char *GetFPUName(unsigned ID) {
 #define ARM_FPU_NAME(NAME, ID) case ARM::ID: return NAME;
 #include "ARMFPUName.def"
   }
-  return NULL;
+  return nullptr;
+}
+
+static const char *GetArchName(unsigned ID) {
+  switch (ID) {
+  default:
+    llvm_unreachable("Unknown ARCH kind");
+    break;
+#define ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH) \
+  case ARM::ID: return NAME;
+#define ARM_ARCH_ALIAS(NAME, ID) /* empty */
+#include "ARMArchName.def"
+  }
+  return nullptr;
+}
+
+static const char *GetArchDefaultCPUName(unsigned ID) {
+  switch (ID) {
+  default:
+    llvm_unreachable("Unknown ARCH kind");
+    break;
+#define ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH) \
+  case ARM::ID: return DEFAULT_CPU_NAME;
+#define ARM_ARCH_ALIAS(NAME, ID) /* empty */
+#include "ARMArchName.def"
+  }
+  return nullptr;
+}
+
+static unsigned GetArchDefaultCPUArch(unsigned ID) {
+  switch (ID) {
+  default:
+    llvm_unreachable("Unknown ARCH kind");
+    break;
+#define ARM_ARCH_NAME(NAME, ID, DEFAULT_CPU_NAME, DEFAULT_CPU_ARCH) \
+  case ARM::ID: return ARMBuildAttrs::DEFAULT_CPU_ARCH;
+#define ARM_ARCH_ALIAS(NAME, ID) /* empty */
+#include "ARMArchName.def"
+  }
+  return 0;
 }
 
 namespace {
@@ -68,36 +112,56 @@ class ARMELFStreamer;
 class ARMTargetAsmStreamer : public ARMTargetStreamer {
   formatted_raw_ostream &OS;
   MCInstPrinter &InstPrinter;
-
-  virtual void emitFnStart();
-  virtual void emitFnEnd();
-  virtual void emitCantUnwind();
-  virtual void emitPersonality(const MCSymbol *Personality);
-  virtual void emitHandlerData();
-  virtual void emitSetFP(unsigned FpReg, unsigned SpReg, int64_t Offset = 0);
-  virtual void emitPad(int64_t Offset);
-  virtual void emitRegSave(const SmallVectorImpl<unsigned> &RegList,
-                           bool isVector);
-
-  virtual void switchVendor(StringRef Vendor);
-  virtual void emitAttribute(unsigned Attribute, unsigned Value);
-  virtual void emitTextAttribute(unsigned Attribute, StringRef String);
-  virtual void emitFPU(unsigned FPU);
-  virtual void finishAttributeSection();
+  bool IsVerboseAsm;
+
+  void emitFnStart() override;
+  void emitFnEnd() override;
+  void emitCantUnwind() override;
+  void emitPersonality(const MCSymbol *Personality) override;
+  void emitPersonalityIndex(unsigned Index) override;
+  void emitHandlerData() override;
+  void emitSetFP(unsigned FpReg, unsigned SpReg, int64_t Offset = 0) override;
+  void emitMovSP(unsigned Reg, int64_t Offset = 0) override;
+  void emitPad(int64_t Offset) override;
+  void emitRegSave(const SmallVectorImpl<unsigned> &RegList,
+                   bool isVector) override;
+  void emitUnwindRaw(int64_t Offset,
+                     const SmallVectorImpl<uint8_t> &Opcodes) override;
+
+  void switchVendor(StringRef Vendor) override;
+  void emitAttribute(unsigned Attribute, unsigned Value) override;
+  void emitTextAttribute(unsigned Attribute, StringRef String) override;
+  void emitIntTextAttribute(unsigned Attribute, unsigned IntValue,
+                            StringRef StrinValue) override;
+  void emitArch(unsigned Arch) override;
+  void emitObjectArch(unsigned Arch) override;
+  void emitFPU(unsigned FPU) override;
+  void emitInst(uint32_t Inst, char Suffix = '\0') override;
+  void finishAttributeSection() override;
+
+  void AnnotateTLSDescriptorSequence(const MCSymbolRefExpr *SRE) override;
+  void emitThumbSet(MCSymbol *Symbol, const MCExpr *Value) override;
 
 public:
-  ARMTargetAsmStreamer(formatted_raw_ostream &OS, MCInstPrinter &InstPrinter);
+  ARMTargetAsmStreamer(MCStreamer &S, formatted_raw_ostream &OS,
+                       MCInstPrinter &InstPrinter, bool VerboseAsm);
 };
 
-ARMTargetAsmStreamer::ARMTargetAsmStreamer(formatted_raw_ostream &OS,
-                                           MCInstPrinter &InstPrinter)
-    : OS(OS), InstPrinter(InstPrinter) {}
+ARMTargetAsmStreamer::ARMTargetAsmStreamer(MCStreamer &S,
+                                           formatted_raw_ostream &OS,
+                                           MCInstPrinter &InstPrinter,
+                                           bool VerboseAsm)
+    : ARMTargetStreamer(S), OS(OS), InstPrinter(InstPrinter),
+      IsVerboseAsm(VerboseAsm) {}
 void ARMTargetAsmStreamer::emitFnStart() { OS << "\t.fnstart\n"; }
 void ARMTargetAsmStreamer::emitFnEnd() { OS << "\t.fnend\n"; }
 void ARMTargetAsmStreamer::emitCantUnwind() { OS << "\t.cantunwind\n"; }
 void ARMTargetAsmStreamer::emitPersonality(const MCSymbol *Personality) {
   OS << "\t.personality " << Personality->getName() << '\n';
 }
+void ARMTargetAsmStreamer::emitPersonalityIndex(unsigned Index) {
+  OS << "\t.personalityindex " << Index << '\n';
+}
 void ARMTargetAsmStreamer::emitHandlerData() { OS << "\t.handlerdata\n"; }
 void ARMTargetAsmStreamer::emitSetFP(unsigned FpReg, unsigned SpReg,
                                      int64_t Offset) {
@@ -109,6 +173,16 @@ void ARMTargetAsmStreamer::emitSetFP(unsigned FpReg, unsigned SpReg,
     OS << ", #" << Offset;
   OS << '\n';
 }
+void ARMTargetAsmStreamer::emitMovSP(unsigned Reg, int64_t Offset) {
+  assert((Reg != ARM::SP && Reg != ARM::PC) &&
+         "the operand of .movsp cannot be either sp or pc");
+
+  OS << "\t.movsp\t";
+  InstPrinter.printRegName(OS, Reg);
+  if (Offset)
+    OS << ", #" << Offset;
+  OS << '\n';
+}
 void ARMTargetAsmStreamer::emitPad(int64_t Offset) {
   OS << "\t.pad\t#" << Offset << '\n';
 }
@@ -132,22 +206,82 @@ void ARMTargetAsmStreamer::emitRegSave(const SmallVectorImpl<unsigned> &RegList,
 void ARMTargetAsmStreamer::switchVendor(StringRef Vendor) {
 }
 void ARMTargetAsmStreamer::emitAttribute(unsigned Attribute, unsigned Value) {
-  OS << "\t.eabi_attribute\t" << Attribute << ", " << Twine(Value) << "\n";
+  OS << "\t.eabi_attribute\t" << Attribute << ", " << Twine(Value);
+  if (IsVerboseAsm) {
+    StringRef Name = ARMBuildAttrs::AttrTypeAsString(Attribute);
+    if (!Name.empty())
+      OS << "\t@ " << Name;
+  }
+  OS << "\n";
 }
 void ARMTargetAsmStreamer::emitTextAttribute(unsigned Attribute,
                                              StringRef String) {
   switch (Attribute) {
-  default: llvm_unreachable("Unsupported Text attribute in ASM Mode");
   case ARMBuildAttrs::CPU_name:
-    OS << "\t.cpu\t" << String.lower() << "\n";
+    OS << "\t.cpu\t" << String.lower();
+    break;
+  default:
+    OS << "\t.eabi_attribute\t" << Attribute << ", \"" << String << "\"";
+    if (IsVerboseAsm) {
+      StringRef Name = ARMBuildAttrs::AttrTypeAsString(Attribute);
+      if (!Name.empty())
+        OS << "\t@ " << Name;
+    }
     break;
   }
+  OS << "\n";
+}
+void ARMTargetAsmStreamer::emitIntTextAttribute(unsigned Attribute,
+                                                unsigned IntValue,
+                                                StringRef StringValue) {
+  switch (Attribute) {
+  default: llvm_unreachable("unsupported multi-value attribute in asm mode");
+  case ARMBuildAttrs::compatibility:
+    OS << "\t.eabi_attribute\t" << Attribute << ", " << IntValue;
+    if (!StringValue.empty())
+      OS << ", \"" << StringValue << "\"";
+    if (IsVerboseAsm)
+      OS << "\t@ " << ARMBuildAttrs::AttrTypeAsString(Attribute);
+    break;
+  }
+  OS << "\n";
+}
+void ARMTargetAsmStreamer::emitArch(unsigned Arch) {
+  OS << "\t.arch\t" << GetArchName(Arch) << "\n";
+}
+void ARMTargetAsmStreamer::emitObjectArch(unsigned Arch) {
+  OS << "\t.object_arch\t" << GetArchName(Arch) << '\n';
 }
 void ARMTargetAsmStreamer::emitFPU(unsigned FPU) {
   OS << "\t.fpu\t" << GetFPUName(FPU) << "\n";
 }
 void ARMTargetAsmStreamer::finishAttributeSection() {
 }
+void
+ARMTargetAsmStreamer::AnnotateTLSDescriptorSequence(const MCSymbolRefExpr *S) {
+  OS << "\t.tlsdescseq\t" << S->getSymbol().getName();
+}
+
+void ARMTargetAsmStreamer::emitThumbSet(MCSymbol *Symbol, const MCExpr *Value) {
+  OS << "\t.thumb_set\t" << *Symbol << ", " << *Value << '\n';
+}
+
+void ARMTargetAsmStreamer::emitInst(uint32_t Inst, char Suffix) {
+  OS << "\t.inst";
+  if (Suffix)
+    OS << "." << Suffix;
+  OS << "\t0x" << utohexstr(Inst) << "\n";
+}
+
+void ARMTargetAsmStreamer::emitUnwindRaw(int64_t Offset,
+                                      const SmallVectorImpl<uint8_t> &Opcodes) {
+  OS << "\t.unwind_raw " << Offset;
+  for (SmallVectorImpl<uint8_t>::const_iterator OCI = Opcodes.begin(),
+                                                OCE = Opcodes.end();
+       OCI != OCE; ++OCI)
+    OS << ", 0x" << utohexstr(*OCI);
+  OS << '\n';
+}
 
 class ARMTargetELFStreamer : public ARMTargetStreamer {
 private:
@@ -158,7 +292,8 @@ private:
     enum {
       HiddenAttribute = 0,
       NumericAttribute,
-      TextAttribute
+      TextAttribute,
+      NumericAndTextAttributes
     } Type;
     unsigned Tag;
     unsigned IntValue;
@@ -171,26 +306,17 @@ private:
 
   StringRef CurrentVendor;
   unsigned FPU;
+  unsigned Arch;
+  unsigned EmittedArch;
   SmallVector<AttributeItem, 64> Contents;
 
   const MCSection *AttributeSection;
 
-  // FIXME: this should be in a more generic place, but
-  // getULEBSize() is in MCAsmInfo and will be moved to MCDwarf
-  static size_t getULEBSize(int Value) {
-    size_t Size = 0;
-    do {
-      Value >>= 7;
-      Size += sizeof(int8_t); // Is this really necessary?
-    } while (Value);
-    return Size;
-  }
-
   AttributeItem *getAttributeItem(unsigned Attribute) {
     for (size_t i = 0; i < Contents.size(); ++i)
       if (Contents[i].Tag == Attribute)
         return &Contents[i];
-    return 0;
+    return nullptr;
   }
 
   void setAttributeItem(unsigned Attribute, unsigned Value,
@@ -199,6 +325,7 @@ private:
     if (AttributeItem *Item = getAttributeItem(Attribute)) {
       if (!OverwriteExisting)
         return;
+      Item->Type = AttributeItem::NumericAttribute;
       Item->IntValue = Value;
       return;
     }
@@ -219,6 +346,7 @@ private:
     if (AttributeItem *Item = getAttributeItem(Attribute)) {
       if (!OverwriteExisting)
         return;
+      Item->Type = AttributeItem::TextAttribute;
       Item->StringValue = Value;
       return;
     }
@@ -233,33 +361,69 @@ private:
     Contents.push_back(Item);
   }
 
+  void setAttributeItems(unsigned Attribute, unsigned IntValue,
+                         StringRef StringValue, bool OverwriteExisting) {
+    // Look for existing attribute item
+    if (AttributeItem *Item = getAttributeItem(Attribute)) {
+      if (!OverwriteExisting)
+        return;
+      Item->Type = AttributeItem::NumericAndTextAttributes;
+      Item->IntValue = IntValue;
+      Item->StringValue = StringValue;
+      return;
+    }
+
+    // Create new attribute item
+    AttributeItem Item = {
+      AttributeItem::NumericAndTextAttributes,
+      Attribute,
+      IntValue,
+      StringValue
+    };
+    Contents.push_back(Item);
+  }
+
+  void emitArchDefaultAttributes();
   void emitFPUDefaultAttributes();
 
   ARMELFStreamer &getStreamer();
 
-  virtual void emitFnStart();
-  virtual void emitFnEnd();
-  virtual void emitCantUnwind();
-  virtual void emitPersonality(const MCSymbol *Personality);
-  virtual void emitHandlerData();
-  virtual void emitSetFP(unsigned FpReg, unsigned SpReg, int64_t Offset = 0);
-  virtual void emitPad(int64_t Offset);
-  virtual void emitRegSave(const SmallVectorImpl<unsigned> &RegList,
-                           bool isVector);
-
-  virtual void switchVendor(StringRef Vendor);
-  virtual void emitAttribute(unsigned Attribute, unsigned Value);
-  virtual void emitTextAttribute(unsigned Attribute, StringRef String);
-  virtual void emitFPU(unsigned FPU);
-  virtual void finishAttributeSection();
+  void emitFnStart() override;
+  void emitFnEnd() override;
+  void emitCantUnwind() override;
+  void emitPersonality(const MCSymbol *Personality) override;
+  void emitPersonalityIndex(unsigned Index) override;
+  void emitHandlerData() override;
+  void emitSetFP(unsigned FpReg, unsigned SpReg, int64_t Offset = 0) override;
+  void emitMovSP(unsigned Reg, int64_t Offset = 0) override;
+  void emitPad(int64_t Offset) override;
+  void emitRegSave(const SmallVectorImpl<unsigned> &RegList,
+                   bool isVector) override;
+  void emitUnwindRaw(int64_t Offset,
+                     const SmallVectorImpl<uint8_t> &Opcodes) override;
+
+  void switchVendor(StringRef Vendor) override;
+  void emitAttribute(unsigned Attribute, unsigned Value) override;
+  void emitTextAttribute(unsigned Attribute, StringRef String) override;
+  void emitIntTextAttribute(unsigned Attribute, unsigned IntValue,
+                            StringRef StringValue) override;
+  void emitArch(unsigned Arch) override;
+  void emitObjectArch(unsigned Arch) override;
+  void emitFPU(unsigned FPU) override;
+  void emitInst(uint32_t Inst, char Suffix = '\0') override;
+  void finishAttributeSection() override;
+  void emitLabel(MCSymbol *Symbol) override;
+
+  void AnnotateTLSDescriptorSequence(const MCSymbolRefExpr *SRE) override;
+  void emitThumbSet(MCSymbol *Symbol, const MCExpr *Value) override;
 
   size_t calculateContentSize() const;
 
 public:
-  ARMTargetELFStreamer()
-    : ARMTargetStreamer(), CurrentVendor("aeabi"), FPU(ARM::INVALID_FPU),
-      AttributeSection(0) {
-  }
+  ARMTargetELFStreamer(MCStreamer &S)
+    : ARMTargetStreamer(S), CurrentVendor("aeabi"), FPU(ARM::INVALID_FPU),
+      Arch(ARM::INVALID_ARCH), EmittedArch(ARM::INVALID_ARCH),
+      AttributeSection(nullptr) {}
 };
 
 /// Extend the generic ELFStreamer class so that it can emit mapping symbols at
@@ -278,30 +442,32 @@ class ARMELFStreamer : public MCELFStreamer {
 public:
   friend class ARMTargetELFStreamer;
 
-  ARMELFStreamer(MCContext &Context, MCTargetStreamer *TargetStreamer,
-                 MCAsmBackend &TAB, raw_ostream &OS, MCCodeEmitter *Emitter,
-                 bool IsThumb)
-      : MCELFStreamer(Context, TargetStreamer, TAB, OS, Emitter),
-        IsThumb(IsThumb), MappingSymbolCounter(0), LastEMS(EMS_None) {
+  ARMELFStreamer(MCContext &Context, MCAsmBackend &TAB, raw_ostream &OS,
+                 MCCodeEmitter *Emitter, bool IsThumb)
+      : MCELFStreamer(Context, TAB, OS, Emitter), IsThumb(IsThumb),
+        MappingSymbolCounter(0), LastEMS(EMS_None) {
     Reset();
   }
 
   ~ARMELFStreamer() {}
 
-  virtual void FinishImpl();
+  void FinishImpl() override;
 
   // ARM exception handling directives
   void emitFnStart();
   void emitFnEnd();
   void emitCantUnwind();
   void emitPersonality(const MCSymbol *Per);
+  void emitPersonalityIndex(unsigned index);
   void emitHandlerData();
   void emitSetFP(unsigned NewFpReg, unsigned NewSpReg, int64_t Offset = 0);
+  void emitMovSP(unsigned Reg, int64_t Offset = 0);
   void emitPad(int64_t Offset);
   void emitRegSave(const SmallVectorImpl<unsigned> &RegList, bool isVector);
+  void emitUnwindRaw(int64_t Offset, const SmallVectorImpl<uint8_t> &Opcodes);
 
-  virtual void ChangeSection(const MCSection *Section,
-                             const MCExpr *Subsection) {
+  void ChangeSection(const MCSection *Section,
+                     const MCExpr *Subsection) override {
     // We have to keep track of the mapping symbol state of any sections we
     // use. Each one should start off as EMS_None, which is provided as the
     // default constructor by DenseMap::lookup.
@@ -314,19 +480,58 @@ public:
   /// This function is the one used to emit instruction data into the ELF
   /// streamer. We override it to add the appropriate mapping symbol if
   /// necessary.
-  virtual void EmitInstruction(const MCInst& Inst) {
+  void EmitInstruction(const MCInst& Inst,
+                       const MCSubtargetInfo &STI) override {
     if (IsThumb)
       EmitThumbMappingSymbol();
     else
       EmitARMMappingSymbol();
 
-    MCELFStreamer::EmitInstruction(Inst);
+    MCELFStreamer::EmitInstruction(Inst, STI);
+  }
+
+  void emitInst(uint32_t Inst, char Suffix) {
+    unsigned Size;
+    char Buffer[4];
+    const bool LittleEndian = getContext().getAsmInfo()->isLittleEndian();
+
+    switch (Suffix) {
+    case '\0':
+      Size = 4;
+
+      assert(!IsThumb);
+      EmitARMMappingSymbol();
+      for (unsigned II = 0, IE = Size; II != IE; II++) {
+        const unsigned I = LittleEndian ? (Size - II - 1) : II;
+        Buffer[Size - II - 1] = uint8_t(Inst >> I * CHAR_BIT);
+      }
+
+      break;
+    case 'n':
+    case 'w':
+      Size = (Suffix == 'n' ? 2 : 4);
+
+      assert(IsThumb);
+      EmitThumbMappingSymbol();
+      for (unsigned II = 0, IE = Size; II != IE; II = II + 2) {
+        const unsigned I0 = LittleEndian ? II + 0 : (Size - II - 1);
+        const unsigned I1 = LittleEndian ? II + 1 : (Size - II - 2);
+        Buffer[Size - II - 2] = uint8_t(Inst >> I0 * CHAR_BIT);
+        Buffer[Size - II - 1] = uint8_t(Inst >> I1 * CHAR_BIT);
+      }
+
+      break;
+    default:
+      llvm_unreachable("Invalid Suffix");
+    }
+
+    MCELFStreamer::EmitBytes(StringRef(Buffer, Size));
   }
 
   /// This is one of the functions used to emit data into an ELF section, so the
   /// ARM streamer overrides it to add the appropriate mapping symbol ($d) if
   /// necessary.
-  virtual void EmitBytes(StringRef Data) {
+  void EmitBytes(StringRef Data) override {
     EmitDataMappingSymbol();
     MCELFStreamer::EmitBytes(Data);
   }
@@ -334,12 +539,13 @@ public:
   /// This is one of the functions used to emit data into an ELF section, so the
   /// ARM streamer overrides it to add the appropriate mapping symbol ($d) if
   /// necessary.
-  virtual void EmitValueImpl(const MCExpr *Value, unsigned Size) {
+  void EmitValueImpl(const MCExpr *Value, unsigned Size,
+                     const SMLoc &Loc) override {
     EmitDataMappingSymbol();
     MCELFStreamer::EmitValueImpl(Value, Size);
   }
 
-  virtual void EmitAssemblerFlag(MCAssemblerFlag Flag) {
+  void EmitAssemblerFlag(MCAssemblerFlag Flag) override {
     MCELFStreamer::EmitAssemblerFlag(Flag);
 
     switch (Flag) {
@@ -402,13 +608,9 @@ private:
     Symbol->setVariableValue(Value);
   }
 
-  void EmitThumbFunc(MCSymbol *Func) {
-    // FIXME: Anything needed here to flag the function as thumb?
-
+  void EmitThumbFunc(MCSymbol *Func) override {
     getAssembler().setIsThumbFunc(Func);
-
-    MCSymbolData &SD = getAssembler().getOrCreateSymbolData(*Func);
-    SD.setFlags(SD.getFlags() | ELF_Other_ThumbFunc);
+    EmitSymbolAttribute(Func, MCSA_ELF_TypeFunction);
   }
 
   // Helper functions for ARM exception handling directives
@@ -423,6 +625,8 @@ private:
   void SwitchToExTabSection(const MCSymbol &FnStart);
   void SwitchToExIdxSection(const MCSymbol &FnStart);
 
+  void EmitFixup(const MCExpr *Expr, MCFixupKind Kind);
+
   bool IsThumb;
   int64_t MappingSymbolCounter;
 
@@ -446,8 +650,7 @@ private:
 } // end anonymous namespace
 
 ARMELFStreamer &ARMTargetELFStreamer::getStreamer() {
-  ARMELFStreamer *S = static_cast<ARMELFStreamer *>(Streamer);
-  return *S;
+  return static_cast<ARMELFStreamer &>(Streamer);
 }
 
 void ARMTargetELFStreamer::emitFnStart() { getStreamer().emitFnStart(); }
@@ -456,6 +659,9 @@ void ARMTargetELFStreamer::emitCantUnwind() { getStreamer().emitCantUnwind(); }
 void ARMTargetELFStreamer::emitPersonality(const MCSymbol *Personality) {
   getStreamer().emitPersonality(Personality);
 }
+void ARMTargetELFStreamer::emitPersonalityIndex(unsigned Index) {
+  getStreamer().emitPersonalityIndex(Index);
+}
 void ARMTargetELFStreamer::emitHandlerData() {
   getStreamer().emitHandlerData();
 }
@@ -463,6 +669,9 @@ void ARMTargetELFStreamer::emitSetFP(unsigned FpReg, unsigned SpReg,
                                      int64_t Offset) {
   getStreamer().emitSetFP(FpReg, SpReg, Offset);
 }
+void ARMTargetELFStreamer::emitMovSP(unsigned Reg, int64_t Offset) {
+  getStreamer().emitMovSP(Reg, Offset);
+}
 void ARMTargetELFStreamer::emitPad(int64_t Offset) {
   getStreamer().emitPad(Offset);
 }
@@ -470,6 +679,10 @@ void ARMTargetELFStreamer::emitRegSave(const SmallVectorImpl<unsigned> &RegList,
                                        bool isVector) {
   getStreamer().emitRegSave(RegList, isVector);
 }
+void ARMTargetELFStreamer::emitUnwindRaw(int64_t Offset,
+                                      const SmallVectorImpl<uint8_t> &Opcodes) {
+  getStreamer().emitUnwindRaw(Offset, Opcodes);
+}
 void ARMTargetELFStreamer::switchVendor(StringRef Vendor) {
   assert(!Vendor.empty() && "Vendor cannot be empty.");
 
@@ -491,6 +704,108 @@ void ARMTargetELFStreamer::emitTextAttribute(unsigned Attribute,
                                              StringRef Value) {
   setAttributeItem(Attribute, Value, /* OverwriteExisting= */ true);
 }
+void ARMTargetELFStreamer::emitIntTextAttribute(unsigned Attribute,
+                                                unsigned IntValue,
+                                                StringRef StringValue) {
+  setAttributeItems(Attribute, IntValue, StringValue,
+                    /* OverwriteExisting= */ true);
+}
+void ARMTargetELFStreamer::emitArch(unsigned Value) {
+  Arch = Value;
+}
+void ARMTargetELFStreamer::emitObjectArch(unsigned Value) {
+  EmittedArch = Value;
+}
+void ARMTargetELFStreamer::emitArchDefaultAttributes() {
+  using namespace ARMBuildAttrs;
+
+  setAttributeItem(CPU_name, GetArchDefaultCPUName(Arch), false);
+  if (EmittedArch == ARM::INVALID_ARCH)
+    setAttributeItem(CPU_arch, GetArchDefaultCPUArch(Arch), false);
+  else
+    setAttributeItem(CPU_arch, GetArchDefaultCPUArch(EmittedArch), false);
+
+  switch (Arch) {
+  case ARM::ARMV2:
+  case ARM::ARMV2A:
+  case ARM::ARMV3:
+  case ARM::ARMV3M:
+  case ARM::ARMV4:
+  case ARM::ARMV5:
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    break;
+
+  case ARM::ARMV4T:
+  case ARM::ARMV5T:
+  case ARM::ARMV5TE:
+  case ARM::ARMV6:
+  case ARM::ARMV6J:
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, Allowed, false);
+    break;
+
+  case ARM::ARMV6T2:
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, AllowThumb32, false);
+    break;
+
+  case ARM::ARMV6Z:
+  case ARM::ARMV6ZK:
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, Allowed, false);
+    setAttributeItem(Virtualization_use, AllowTZ, false);
+    break;
+
+  case ARM::ARMV6M:
+    setAttributeItem(THUMB_ISA_use, Allowed, false);
+    break;
+
+  case ARM::ARMV7:
+    setAttributeItem(THUMB_ISA_use, AllowThumb32, false);
+    break;
+
+  case ARM::ARMV7A:
+    setAttributeItem(CPU_arch_profile, ApplicationProfile, false);
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, AllowThumb32, false);
+    break;
+
+  case ARM::ARMV7R:
+    setAttributeItem(CPU_arch_profile, RealTimeProfile, false);
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, AllowThumb32, false);
+    break;
+
+  case ARM::ARMV7M:
+    setAttributeItem(CPU_arch_profile, MicroControllerProfile, false);
+    setAttributeItem(THUMB_ISA_use, AllowThumb32, false);
+    break;
+
+  case ARM::ARMV8A:
+    setAttributeItem(CPU_arch_profile, ApplicationProfile, false);
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, AllowThumb32, false);
+    setAttributeItem(MPextension_use, Allowed, false);
+    setAttributeItem(Virtualization_use, AllowTZVirtualization, false);
+    break;
+
+  case ARM::IWMMXT:
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, Allowed, false);
+    setAttributeItem(WMMX_arch, AllowWMMXv1, false);
+    break;
+
+  case ARM::IWMMXT2:
+    setAttributeItem(ARM_ISA_use, Allowed, false);
+    setAttributeItem(THUMB_ISA_use, Allowed, false);
+    setAttributeItem(WMMX_arch, AllowWMMXv2, false);
+    break;
+
+  default:
+    report_fatal_error("Unknown Arch: " + Twine(Arch));
+    break;
+  }
+}
 void ARMTargetELFStreamer::emitFPU(unsigned Value) {
   FPU = Value;
 }
@@ -498,43 +813,43 @@ void ARMTargetELFStreamer::emitFPUDefaultAttributes() {
   switch (FPU) {
   case ARM::VFP:
   case ARM::VFPV2:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv2,
                      /* OverwriteExisting= */ false);
     break;
 
   case ARM::VFPV3:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv3A,
                      /* OverwriteExisting= */ false);
     break;
 
   case ARM::VFPV3_D16:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv3B,
                      /* OverwriteExisting= */ false);
     break;
 
   case ARM::VFPV4:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv4A,
                      /* OverwriteExisting= */ false);
     break;
 
   case ARM::VFPV4_D16:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv4B,
                      /* OverwriteExisting= */ false);
     break;
 
   case ARM::FP_ARMV8:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPARMv8A,
                      /* OverwriteExisting= */ false);
     break;
 
   case ARM::NEON:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv3A,
                      /* OverwriteExisting= */ false);
     setAttributeItem(ARMBuildAttrs::Advanced_SIMD_arch,
@@ -543,7 +858,7 @@ void ARMTargetELFStreamer::emitFPUDefaultAttributes() {
     break;
 
   case ARM::NEON_VFPV4:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPv4A,
                      /* OverwriteExisting= */ false);
     setAttributeItem(ARMBuildAttrs::Advanced_SIMD_arch,
@@ -553,7 +868,7 @@ void ARMTargetELFStreamer::emitFPUDefaultAttributes() {
 
   case ARM::NEON_FP_ARMV8:
   case ARM::CRYPTO_NEON_FP_ARMV8:
-    setAttributeItem(ARMBuildAttrs::VFP_arch,
+    setAttributeItem(ARMBuildAttrs::FP_arch,
                      ARMBuildAttrs::AllowFPARMv8A,
                      /* OverwriteExisting= */ false);
     setAttributeItem(ARMBuildAttrs::Advanced_SIMD_arch,
@@ -561,6 +876,9 @@ void ARMTargetELFStreamer::emitFPUDefaultAttributes() {
                      /* OverwriteExisting= */ false);
     break;
 
+  case ARM::SOFTVFP:
+    break;
+
   default:
     report_fatal_error("Unknown FPU: " + Twine(FPU));
     break;
@@ -574,13 +892,18 @@ size_t ARMTargetELFStreamer::calculateContentSize() const {
     case AttributeItem::HiddenAttribute:
       break;
     case AttributeItem::NumericAttribute:
-      Result += getULEBSize(item.Tag);
-      Result += getULEBSize(item.IntValue);
+      Result += getULEB128Size(item.Tag);
+      Result += getULEB128Size(item.IntValue);
       break;
     case AttributeItem::TextAttribute:
-      Result += getULEBSize(item.Tag);
+      Result += getULEB128Size(item.Tag);
       Result += item.StringValue.size() + 1; // string + '\0'
       break;
+    case AttributeItem::NumericAndTextAttributes:
+      Result += getULEB128Size(item.Tag);
+      Result += getULEB128Size(item.IntValue);
+      Result += item.StringValue.size() + 1; // string + '\0';
+      break;
     }
   }
   return Result;
@@ -597,6 +920,9 @@ void ARMTargetELFStreamer::finishAttributeSection() {
   if (FPU != ARM::INVALID_FPU)
     emitFPUDefaultAttributes();
 
+  if (Arch != ARM::INVALID_ARCH)
+    emitArchDefaultAttributes();
+
   if (Contents.empty())
     return;
 
@@ -648,6 +974,11 @@ void ARMTargetELFStreamer::finishAttributeSection() {
       Streamer.EmitBytes(item.StringValue.upper());
       Streamer.EmitIntValue(0, 1); // '\0'
       break;
+    case AttributeItem::NumericAndTextAttributes:
+      Streamer.EmitULEB128IntValue(item.IntValue);
+      Streamer.EmitBytes(item.StringValue.upper());
+      Streamer.EmitIntValue(0, 1); // '\0'
+      break;
     }
   }
 
@@ -655,8 +986,41 @@ void ARMTargetELFStreamer::finishAttributeSection() {
   FPU = ARM::INVALID_FPU;
 }
 
+void ARMTargetELFStreamer::emitLabel(MCSymbol *Symbol) {
+  ARMELFStreamer &Streamer = getStreamer();
+  if (!Streamer.IsThumb)
+    return;
+
+  const MCSymbolData &SD = Streamer.getOrCreateSymbolData(Symbol);
+  unsigned Type = MCELF::GetType(SD);
+  if (Type == ELF_STT_Func || Type == ELF_STT_GnuIFunc)
+    Streamer.EmitThumbFunc(Symbol);
+}
+
+void
+ARMTargetELFStreamer::AnnotateTLSDescriptorSequence(const MCSymbolRefExpr *S) {
+  getStreamer().EmitFixup(S, FK_Data_4);
+}
+
+void ARMTargetELFStreamer::emitThumbSet(MCSymbol *Symbol, const MCExpr *Value) {
+  if (const MCSymbolRefExpr *SRE = dyn_cast<MCSymbolRefExpr>(Value)) {
+    const MCSymbol &Sym = SRE->getSymbol();
+    if (!Sym.isDefined()) {
+      getStreamer().EmitAssignment(Symbol, Value);
+      return;
+    }
+  }
+
+  getStreamer().EmitThumbFunc(Symbol);
+  getStreamer().EmitAssignment(Symbol, Value);
+}
+
+void ARMTargetELFStreamer::emitInst(uint32_t Inst, char Suffix) {
+  getStreamer().emitInst(Inst, Suffix);
+}
+
 void ARMELFStreamer::FinishImpl() {
-  MCTargetStreamer &TS = getTargetStreamer();
+  MCTargetStreamer &TS = *getTargetStreamer();
   ARMTargetStreamer &ATS = static_cast<ARMTargetStreamer &>(TS);
   ATS.finishAttributeSection();
 
@@ -679,7 +1043,7 @@ inline void ARMELFStreamer::SwitchToEHSection(const char *Prefix,
   }
 
   // Get .ARM.extab or .ARM.exidx section
-  const MCSectionELF *EHSection = NULL;
+  const MCSectionELF *EHSection = nullptr;
   if (const MCSymbol *Group = FnSection.getGroup()) {
     EHSection = getContext().getELFSection(
       EHSecName, Type, Flags | ELF::SHF_GROUP, Kind,
@@ -691,7 +1055,7 @@ inline void ARMELFStreamer::SwitchToEHSection(const char *Prefix,
 
   // Switch to .ARM.extab or .ARM.exidx section
   SwitchSection(EHSection);
-  EmitCodeAlignment(4, 0);
+  EmitCodeAlignment(4);
 }
 
 inline void ARMELFStreamer::SwitchToExTabSection(const MCSymbol &FnStart) {
@@ -709,12 +1073,17 @@ inline void ARMELFStreamer::SwitchToExIdxSection(const MCSymbol &FnStart) {
                     SectionKind::getDataRel(),
                     FnStart);
 }
+void ARMELFStreamer::EmitFixup(const MCExpr *Expr, MCFixupKind Kind) {
+  MCDataFragment *Frag = getOrCreateDataFragment();
+  Frag->getFixups().push_back(MCFixup::Create(Frag->getContents().size(), Expr,
+                                              Kind));
+}
 
 void ARMELFStreamer::Reset() {
-  ExTab = NULL;
-  FnStart = NULL;
-  Personality = NULL;
-  PersonalityIndex = NUM_PERSONALITY_INDEX;
+  ExTab = nullptr;
+  FnStart = nullptr;
+  Personality = nullptr;
+  PersonalityIndex = ARM::EHABI::NUM_PERSONALITY_INDEX;
   FPReg = ARM::SP;
   FPOffset = 0;
   SPOffset = 0;
@@ -727,13 +1096,13 @@ void ARMELFStreamer::Reset() {
 }
 
 void ARMELFStreamer::emitFnStart() {
-  assert(FnStart == 0);
+  assert(FnStart == nullptr);
   FnStart = getContext().CreateTempSymbol();
   EmitLabel(FnStart);
 }
 
 void ARMELFStreamer::emitFnEnd() {
-  assert(FnStart && ".fnstart must preceeds .fnend");
+  assert(FnStart && ".fnstart must precedes .fnend");
 
   // Emit unwind opcodes if there is no .handlerdata directive
   if (!ExTab && !CantUnwind)
@@ -742,7 +1111,7 @@ void ARMELFStreamer::emitFnEnd() {
   // Emit the exception index table entry
   SwitchToExIdxSection(*FnStart);
 
-  if (PersonalityIndex < NUM_PERSONALITY_INDEX)
+  if (PersonalityIndex < ARM::EHABI::NUM_PERSONALITY_INDEX)
     EmitPersonalityFixup(GetAEABIUnwindPersonalityName(PersonalityIndex));
 
   const MCSymbolRefExpr *FnStartRef =
@@ -753,7 +1122,7 @@ void ARMELFStreamer::emitFnEnd() {
   EmitValue(FnStartRef, 4);
 
   if (CantUnwind) {
-    EmitIntValue(EXIDX_CANTUNWIND, 4);
+    EmitIntValue(ARM::EHABI::EXIDX_CANTUNWIND, 4);
   } else if (ExTab) {
     // Emit a reference to the unwind opcodes in the ".ARM.extab" section.
     const MCSymbolRefExpr *ExTabEntryRef =
@@ -765,12 +1134,15 @@ void ARMELFStreamer::emitFnEnd() {
     // For the __aeabi_unwind_cpp_pr0, we have to emit the unwind opcodes in
     // the second word of exception index table entry.  The size of the unwind
     // opcodes should always be 4 bytes.
-    assert(PersonalityIndex == AEABI_UNWIND_CPP_PR0 &&
-           "Compact model must use __aeabi_cpp_unwind_pr0 as personality");
+    assert(PersonalityIndex == ARM::EHABI::AEABI_UNWIND_CPP_PR0 &&
+           "Compact model must use __aeabi_unwind_cpp_pr0 as personality");
     assert(Opcodes.size() == 4u &&
-           "Unwind opcode size for __aeabi_cpp_unwind_pr0 must be equal to 4");
-    EmitBytes(StringRef(reinterpret_cast<const char*>(Opcodes.data()),
-                        Opcodes.size()));
+           "Unwind opcode size for __aeabi_unwind_cpp_pr0 must be equal to 4");
+    uint64_t Intval = Opcodes[0] |
+                      Opcodes[1] << 8 |
+                      Opcodes[2] << 16 |
+                      Opcodes[3] << 24;
+    EmitIntValue(Intval, Opcodes.size());
   }
 
   // Switch to the section containing FnStart
@@ -789,7 +1161,7 @@ void ARMELFStreamer::EmitPersonalityFixup(StringRef Name) {
   const MCSymbolRefExpr *PersonalityRef = MCSymbolRefExpr::Create(
       PersonalitySym, MCSymbolRefExpr::VK_ARM_NONE, getContext());
 
-  AddValueSymbols(PersonalityRef);
+  visitUsedExpr(*PersonalityRef);
   MCDataFragment *DF = getOrCreateDataFragment();
   DF->getFixups().push_back(MCFixup::Create(DF->getContents().size(),
                                             PersonalityRef,
@@ -820,7 +1192,7 @@ void ARMELFStreamer::FlushUnwindOpcodes(bool NoHandlerData) {
   // For compact model 0, we have to emit the unwind opcodes in the .ARM.exidx
   // section.  Thus, we don't have to create an entry in the .ARM.extab
   // section.
-  if (NoHandlerData && PersonalityIndex == AEABI_UNWIND_CPP_PR0)
+  if (NoHandlerData && PersonalityIndex == ARM::EHABI::AEABI_UNWIND_CPP_PR0)
     return;
 
   // Switch to .ARM.extab section.
@@ -842,8 +1214,15 @@ void ARMELFStreamer::FlushUnwindOpcodes(bool NoHandlerData) {
   }
 
   // Emit unwind opcodes
-  EmitBytes(StringRef(reinterpret_cast<const char *>(Opcodes.data()),
-                      Opcodes.size()));
+  assert((Opcodes.size() % 4) == 0 &&
+         "Unwind opcode size for __aeabi_cpp_unwind_pr0 must be multiple of 4");
+  for (unsigned I = 0; I != Opcodes.size(); I += 4) {
+    uint64_t Intval = Opcodes[I] |
+                      Opcodes[I + 1] << 8 |
+                      Opcodes[I + 2] << 16 |
+                      Opcodes[I + 3] << 24;
+    EmitIntValue(Intval, 4);
+  }
 
   // According to ARM EHABI section 9.2, if the __aeabi_unwind_cpp_pr1() or
   // __aeabi_unwind_cpp_pr2() is used, then the handler data must be emitted
@@ -863,6 +1242,11 @@ void ARMELFStreamer::emitPersonality(const MCSymbol *Per) {
   UnwindOpAsm.setPersonality(Per);
 }
 
+void ARMELFStreamer::emitPersonalityIndex(unsigned Index) {
+  assert(Index < ARM::EHABI::NUM_PERSONALITY_INDEX && "invalid index");
+  PersonalityIndex = Index;
+}
+
 void ARMELFStreamer::emitSetFP(unsigned NewFPReg, unsigned NewSPReg,
                                int64_t Offset) {
   assert((NewSPReg == ARM::SP || NewSPReg == FPReg) &&
@@ -877,6 +1261,20 @@ void ARMELFStreamer::emitSetFP(unsigned NewFPReg, unsigned NewSPReg,
     FPOffset += Offset;
 }
 
+void ARMELFStreamer::emitMovSP(unsigned Reg, int64_t Offset) {
+  assert((Reg != ARM::SP && Reg != ARM::PC) &&
+         "the operand of .movsp cannot be either sp or pc");
+  assert(FPReg == ARM::SP && "current FP must be SP");
+
+  FlushPendingOffset();
+
+  FPReg = Reg;
+  FPOffset = SPOffset + Offset;
+
+  const MCRegisterInfo *MRI = getContext().getRegisterInfo();
+  UnwindOpAsm.EmitSetSP(MRI->getEncodingValue(FPReg));
+}
+
 void ARMELFStreamer::emitPad(int64_t Offset) {
   // Track the change of the $sp offset
   SPOffset -= Offset;
@@ -916,27 +1314,37 @@ void ARMELFStreamer::emitRegSave(const SmallVectorImpl<unsigned> &RegList,
     UnwindOpAsm.EmitRegSave(Mask);
 }
 
+void ARMELFStreamer::emitUnwindRaw(int64_t Offset,
+                                   const SmallVectorImpl<uint8_t> &Opcodes) {
+  FlushPendingOffset();
+  SPOffset = SPOffset - Offset;
+  UnwindOpAsm.EmitRaw(Opcodes);
+}
+
 namespace llvm {
 
 MCStreamer *createMCAsmStreamer(MCContext &Ctx, formatted_raw_ostream &OS,
-                                bool isVerboseAsm, bool useLoc, bool useCFI,
-                                bool useDwarfDirectory,
+                                bool isVerboseAsm, bool useDwarfDirectory,
                                 MCInstPrinter *InstPrint, MCCodeEmitter *CE,
                                 MCAsmBackend *TAB, bool ShowInst) {
-  ARMTargetAsmStreamer *S = new ARMTargetAsmStreamer(OS, *InstPrint);
+  MCStreamer *S = llvm::createAsmStreamer(
+      Ctx, OS, isVerboseAsm, useDwarfDirectory, InstPrint, CE, TAB, ShowInst);
+  new ARMTargetAsmStreamer(*S, OS, *InstPrint, isVerboseAsm);
+  return S;
+}
 
-  return llvm::createAsmStreamer(Ctx, S, OS, isVerboseAsm, useLoc, useCFI,
-                                 useDwarfDirectory, InstPrint, CE, TAB,
-                                 ShowInst);
+MCStreamer *createARMNullStreamer(MCContext &Ctx) {
+  MCStreamer *S = llvm::createNullStreamer(Ctx);
+  new ARMTargetStreamer(*S);
+  return S;
 }
 
   MCELFStreamer* createARMELFStreamer(MCContext &Context, MCAsmBackend &TAB,
                                       raw_ostream &OS, MCCodeEmitter *Emitter,
                                       bool RelaxAll, bool NoExecStack,
                                       bool IsThumb) {
-    ARMTargetELFStreamer *TS = new ARMTargetELFStreamer();
-    ARMELFStreamer *S =
-        new ARMELFStreamer(Context, TS, TAB, OS, Emitter, IsThumb);
+    ARMELFStreamer *S = new ARMELFStreamer(Context, TAB, OS, Emitter, IsThumb);
+    new ARMTargetELFStreamer(*S);
     // FIXME: This should eventually end up somewhere else where more
     // intelligent flag decisions can be made. For now we are just maintaining
     // the status quo for ARM and setting EF_ARM_EABI_VER5 as the default.
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMFixupKinds.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMFixupKinds.h
index 0085feb82069..bfd9e3364825 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMFixupKinds.h
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMFixupKinds.h
@@ -100,15 +100,6 @@ enum Fixups {
   fixup_t2_movt_hi16, // :upper16:
   fixup_t2_movw_lo16, // :lower16:
 
-  // It is possible to create an "immediate" that happens to be pcrel.
-  // movw r0, :lower16:Foo-(Bar+8) and movt  r0, :upper16:Foo-(Bar+8)
-  // result in different reloc tags than the above two.
-  // Needed to support ELF::R_ARM_MOVT_PREL and ELF::R_ARM_MOVW_PREL_NC
-  fixup_arm_movt_hi16_pcrel, // :upper16:
-  fixup_arm_movw_lo16_pcrel, // :lower16:
-  fixup_t2_movt_hi16_pcrel, // :upper16:
-  fixup_t2_movw_lo16_pcrel, // :lower16:
-
   // Marker
   LastTargetFixupKind,
   NumTargetFixupKinds = LastTargetFixupKind - FirstTargetFixupKind
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.cpp
index ad796e660e96..7a19208cffc7 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.cpp
@@ -13,19 +13,19 @@
 
 #include "ARMMCAsmInfo.h"
 #include "llvm/Support/CommandLine.h"
+#include "llvm/ADT/Triple.h"
 
 using namespace llvm;
 
-cl::opt<bool>
-EnableARMEHABI("arm-enable-ehabi", cl::Hidden,
-  cl::desc("Generate ARM EHABI tables"),
-  cl::init(false));
-
-
 void ARMMCAsmInfoDarwin::anchor() { }
 
-ARMMCAsmInfoDarwin::ARMMCAsmInfoDarwin() {
-  Data64bitsDirective = 0;
+ARMMCAsmInfoDarwin::ARMMCAsmInfoDarwin(StringRef TT) {
+  Triple TheTriple(TT);
+  if ((TheTriple.getArch() == Triple::armeb) ||
+      (TheTriple.getArch() == Triple::thumbeb))
+    IsLittleEndian = false;
+
+  Data64bitsDirective = nullptr;
   CommentString = "@";
   Code16Directive = ".code\t16";
   Code32Directive = ".code\t32";
@@ -35,17 +35,23 @@ ARMMCAsmInfoDarwin::ARMMCAsmInfoDarwin() {
 
   // Exceptions handling
   ExceptionsType = ExceptionHandling::SjLj;
+
+  UseIntegratedAssembler = true;
 }
 
 void ARMELFMCAsmInfo::anchor() { }
 
-ARMELFMCAsmInfo::ARMELFMCAsmInfo() {
+ARMELFMCAsmInfo::ARMELFMCAsmInfo(StringRef TT) {
+  Triple TheTriple(TT);
+  if ((TheTriple.getArch() == Triple::armeb) ||
+      (TheTriple.getArch() == Triple::thumbeb))
+    IsLittleEndian = false;
+
   // ".comm align is in bytes but .align is pow-2."
   AlignmentIsInBytes = false;
 
-  Data64bitsDirective = 0;
+  Data64bitsDirective = nullptr;
   CommentString = "@";
-  PrivateGlobalPrefix = ".L";
   Code16Directive = ".code\t16";
   Code32Directive = ".code\t32";
 
@@ -53,6 +59,56 @@ ARMELFMCAsmInfo::ARMELFMCAsmInfo() {
   SupportsDebugInformation = true;
 
   // Exceptions handling
-  if (EnableARMEHABI)
+  switch (TheTriple.getOS()) {
+  case Triple::NetBSD:
+    ExceptionsType = ExceptionHandling::DwarfCFI;
+    break;
+  default:
     ExceptionsType = ExceptionHandling::ARM;
+    break;
+  }
+
+  // foo(plt) instead of foo@plt
+  UseParensForSymbolVariant = true;
+
+  UseIntegratedAssembler = true;
+}
+
+void ARMELFMCAsmInfo::setUseIntegratedAssembler(bool Value) {
+  UseIntegratedAssembler = Value;
+  if (!UseIntegratedAssembler) {
+    // gas doesn't handle VFP register names in cfi directives,
+    // so don't use register names with external assembler.
+    // See https://sourceware.org/bugzilla/show_bug.cgi?id=16694
+    DwarfRegNumForCFI = true;
+  }
+}
+
+void ARMCOFFMCAsmInfoMicrosoft::anchor() { }
+
+ARMCOFFMCAsmInfoMicrosoft::ARMCOFFMCAsmInfoMicrosoft() {
+  AlignmentIsInBytes = false;
+
+  PrivateGlobalPrefix = "$M";
 }
+
+void ARMCOFFMCAsmInfoGNU::anchor() { }
+
+ARMCOFFMCAsmInfoGNU::ARMCOFFMCAsmInfoGNU() {
+  AlignmentIsInBytes = false;
+  HasSingleParameterDotFile = true;
+
+  CommentString = "@";
+  Code16Directive = ".code\t16";
+  Code32Directive = ".code\t32";
+  PrivateGlobalPrefix = ".L";
+
+  HasLEB128 = true;
+  SupportsDebugInformation = true;
+  ExceptionsType = ExceptionHandling::None;
+  UseParensForSymbolVariant = true;
+
+  UseIntegratedAssembler = false;
+  DwarfRegNumForCFI = true;
+}
+
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.h
index e1f716d936ad..51cfa0adc1a9 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.h
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.h
@@ -14,21 +14,36 @@
 #ifndef LLVM_ARMTARGETASMINFO_H
 #define LLVM_ARMTARGETASMINFO_H
 
+#include "llvm/MC/MCAsmInfoCOFF.h"
 #include "llvm/MC/MCAsmInfoDarwin.h"
 #include "llvm/MC/MCAsmInfoELF.h"
 
 namespace llvm {
 
   class ARMMCAsmInfoDarwin : public MCAsmInfoDarwin {
-    virtual void anchor();
+    void anchor() override;
   public:
-    explicit ARMMCAsmInfoDarwin();
+    explicit ARMMCAsmInfoDarwin(StringRef TT);
   };
 
   class ARMELFMCAsmInfo : public MCAsmInfoELF {
-    virtual void anchor();
+    void anchor() override;
   public:
-    explicit ARMELFMCAsmInfo();
+    explicit ARMELFMCAsmInfo(StringRef TT);
+
+    void setUseIntegratedAssembler(bool Value) override;
+  };
+
+  class ARMCOFFMCAsmInfoMicrosoft : public MCAsmInfoMicrosoft {
+    void anchor() override;
+  public:
+    explicit ARMCOFFMCAsmInfoMicrosoft();
+  };
+
+  class ARMCOFFMCAsmInfoGNU : public MCAsmInfoGNUCOFF {
+    void anchor() override;
+  public:
+    explicit ARMCOFFMCAsmInfoGNU();
   };
 
 } // namespace llvm
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCCodeEmitter.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCCodeEmitter.cpp
index 4382d0d97144..b8ee55574972 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCCodeEmitter.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCCodeEmitter.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "mccodeemitter"
 #include "MCTargetDesc/ARMMCTargetDesc.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "MCTargetDesc/ARMBaseInfo.h"
@@ -26,10 +25,13 @@
 #include "llvm/MC/MCInstrInfo.h"
 #include "llvm/MC/MCRegisterInfo.h"
 #include "llvm/MC/MCSubtargetInfo.h"
+#include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/raw_ostream.h"
 
 using namespace llvm;
 
+#define DEBUG_TYPE "mccodeemitter"
+
 STATISTIC(MCNumEmitted, "Number of MC instructions emitted.");
 STATISTIC(MCNumCPRelocations, "Number of constant pool relocations created.");
 
@@ -38,27 +40,25 @@ class ARMMCCodeEmitter : public MCCodeEmitter {
   ARMMCCodeEmitter(const ARMMCCodeEmitter &) LLVM_DELETED_FUNCTION;
   void operator=(const ARMMCCodeEmitter &) LLVM_DELETED_FUNCTION;
   const MCInstrInfo &MCII;
-  const MCSubtargetInfo &STI;
   const MCContext &CTX;
+  bool IsLittleEndian;
 
 public:
-  ARMMCCodeEmitter(const MCInstrInfo &mcii, const MCSubtargetInfo &sti,
-                   MCContext &ctx)
-    : MCII(mcii), STI(sti), CTX(ctx) {
+  ARMMCCodeEmitter(const MCInstrInfo &mcii, MCContext &ctx, bool IsLittle)
+    : MCII(mcii), CTX(ctx), IsLittleEndian(IsLittle) {
   }
 
   ~ARMMCCodeEmitter() {}
 
-  bool isThumb() const {
-    // FIXME: Can tablegen auto-generate this?
+  bool isThumb(const MCSubtargetInfo &STI) const {
     return (STI.getFeatureBits() & ARM::ModeThumb) != 0;
   }
-  bool isThumb2() const {
-    return isThumb() && (STI.getFeatureBits() & ARM::FeatureThumb2) != 0;
+  bool isThumb2(const MCSubtargetInfo &STI) const {
+    return isThumb(STI) && (STI.getFeatureBits() & ARM::FeatureThumb2) != 0;
   }
-  bool isTargetDarwin() const {
+  bool isTargetMachO(const MCSubtargetInfo &STI) const {
     Triple TT(STI.getTargetTriple());
-  	return TT.isOSDarwin();
+    return TT.isOSBinFormatMachO();
   }
 
   unsigned getMachineSoImmOpValue(unsigned SoImm) const;
@@ -66,107 +66,131 @@ public:
   // getBinaryCodeForInstr - TableGen'erated function for getting the
   // binary encoding for an instruction.
   uint64_t getBinaryCodeForInstr(const MCInst &MI,
-                                 SmallVectorImpl<MCFixup> &Fixups) const;
+                                 SmallVectorImpl<MCFixup> &Fixups,
+                                 const MCSubtargetInfo &STI) const;
 
   /// getMachineOpValue - Return binary encoding of operand. If the machine
   /// operand requires relocation, record the relocation and return zero.
   unsigned getMachineOpValue(const MCInst &MI,const MCOperand &MO,
-                             SmallVectorImpl<MCFixup> &Fixups) const;
+                             SmallVectorImpl<MCFixup> &Fixups,
+                             const MCSubtargetInfo &STI) const;
 
   /// getHiLo16ImmOpValue - Return the encoding for the hi / low 16-bit of
   /// the specified operand. This is used for operands with :lower16: and
   /// :upper16: prefixes.
   uint32_t getHiLo16ImmOpValue(const MCInst &MI, unsigned OpIdx,
-                               SmallVectorImpl<MCFixup> &Fixups) const;
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const;
 
   bool EncodeAddrModeOpValues(const MCInst &MI, unsigned OpIdx,
                               unsigned &Reg, unsigned &Imm,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
 
   /// getThumbBLTargetOpValue - Return encoding info for Thumb immediate
   /// BL branch target.
   uint32_t getThumbBLTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                   SmallVectorImpl<MCFixup> &Fixups) const;
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
 
   /// getThumbBLXTargetOpValue - Return encoding info for Thumb immediate
   /// BLX branch target.
   uint32_t getThumbBLXTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                    SmallVectorImpl<MCFixup> &Fixups) const;
+                                    SmallVectorImpl<MCFixup> &Fixups,
+                                    const MCSubtargetInfo &STI) const;
 
   /// getThumbBRTargetOpValue - Return encoding info for Thumb branch target.
   uint32_t getThumbBRTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                   SmallVectorImpl<MCFixup> &Fixups) const;
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
 
   /// getThumbBCCTargetOpValue - Return encoding info for Thumb branch target.
   uint32_t getThumbBCCTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                    SmallVectorImpl<MCFixup> &Fixups) const;
+                                    SmallVectorImpl<MCFixup> &Fixups,
+                                    const MCSubtargetInfo &STI) const;
 
   /// getThumbCBTargetOpValue - Return encoding info for Thumb branch target.
   uint32_t getThumbCBTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                   SmallVectorImpl<MCFixup> &Fixups) const;
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
 
   /// getBranchTargetOpValue - Return encoding info for 24-bit immediate
   /// branch target.
   uint32_t getBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                  SmallVectorImpl<MCFixup> &Fixups) const;
+                                  SmallVectorImpl<MCFixup> &Fixups,
+                                  const MCSubtargetInfo &STI) const;
 
   /// getUnconditionalBranchTargetOpValue - Return encoding info for 24-bit
   /// immediate Thumb2 direct branch target.
   uint32_t getUnconditionalBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                  SmallVectorImpl<MCFixup> &Fixups) const;
+                                  SmallVectorImpl<MCFixup> &Fixups,
+                                  const MCSubtargetInfo &STI) const;
 
   /// getARMBranchTargetOpValue - Return encoding info for 24-bit immediate
   /// branch target.
   uint32_t getARMBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                     SmallVectorImpl<MCFixup> &Fixups) const;
+                                     SmallVectorImpl<MCFixup> &Fixups,
+                                     const MCSubtargetInfo &STI) const;
   uint32_t getARMBLTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                 SmallVectorImpl<MCFixup> &Fixups) const;
+                                 SmallVectorImpl<MCFixup> &Fixups,
+                                 const MCSubtargetInfo &STI) const;
   uint32_t getARMBLXTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                                  SmallVectorImpl<MCFixup> &Fixups) const;
+                                  SmallVectorImpl<MCFixup> &Fixups,
+                                  const MCSubtargetInfo &STI) const;
 
   /// getAdrLabelOpValue - Return encoding info for 12-bit immediate
   /// ADR label target.
   uint32_t getAdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
   uint32_t getThumbAdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
   uint32_t getT2AdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
 
 
   /// getAddrModeImm12OpValue - Return encoding info for 'reg +/- imm12'
   /// operand.
   uint32_t getAddrModeImm12OpValue(const MCInst &MI, unsigned OpIdx,
-                                   SmallVectorImpl<MCFixup> &Fixups) const;
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
 
   /// getThumbAddrModeRegRegOpValue - Return encoding for 'reg + reg' operand.
   uint32_t getThumbAddrModeRegRegOpValue(const MCInst &MI, unsigned OpIdx,
-                                         SmallVectorImpl<MCFixup> &Fixups)const;
+                                         SmallVectorImpl<MCFixup> &Fixups,
+                                         const MCSubtargetInfo &STI) const;
 
   /// getT2AddrModeImm8s4OpValue - Return encoding info for 'reg +/- imm8<<2'
   /// operand.
   uint32_t getT2AddrModeImm8s4OpValue(const MCInst &MI, unsigned OpIdx,
-                                   SmallVectorImpl<MCFixup> &Fixups) const;
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
 
   /// getT2AddrModeImm0_1020s4OpValue - Return encoding info for 'reg + imm8<<2'
   /// operand.
   uint32_t getT2AddrModeImm0_1020s4OpValue(const MCInst &MI, unsigned OpIdx,
-                                   SmallVectorImpl<MCFixup> &Fixups) const;
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const;
 
   /// getT2Imm8s4OpValue - Return encoding info for '+/- imm8<<2'
   /// operand.
   uint32_t getT2Imm8s4OpValue(const MCInst &MI, unsigned OpIdx,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
 
 
   /// getLdStSORegOpValue - Return encoding info for 'reg +/- reg shop imm'
   /// operand as needed by load/store instructions.
   uint32_t getLdStSORegOpValue(const MCInst &MI, unsigned OpIdx,
-                               SmallVectorImpl<MCFixup> &Fixups) const;
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const;
 
   /// getLdStmModeOpValue - Return encoding for load/store multiple mode.
   uint32_t getLdStmModeOpValue(const MCInst &MI, unsigned OpIdx,
-                               SmallVectorImpl<MCFixup> &Fixups) const {
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const {
     ARM_AM::AMSubMode Mode = (ARM_AM::AMSubMode)MI.getOperand(OpIdx).getImm();
     switch (Mode) {
     default: llvm_unreachable("Unknown addressing sub-mode!");
@@ -192,44 +216,54 @@ public:
 
   /// getAddrMode2OpValue - Return encoding for addrmode2 operands.
   uint32_t getAddrMode2OpValue(const MCInst &MI, unsigned OpIdx,
-                               SmallVectorImpl<MCFixup> &Fixups) const;
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const;
 
   /// getAddrMode2OffsetOpValue - Return encoding for am2offset operands.
   uint32_t getAddrMode2OffsetOpValue(const MCInst &MI, unsigned OpIdx,
-                                     SmallVectorImpl<MCFixup> &Fixups) const;
+                                     SmallVectorImpl<MCFixup> &Fixups,
+                                     const MCSubtargetInfo &STI) const;
 
   /// getPostIdxRegOpValue - Return encoding for postidx_reg operands.
   uint32_t getPostIdxRegOpValue(const MCInst &MI, unsigned OpIdx,
-                                SmallVectorImpl<MCFixup> &Fixups) const;
+                                SmallVectorImpl<MCFixup> &Fixups,
+                                const MCSubtargetInfo &STI) const;
 
   /// getAddrMode3OffsetOpValue - Return encoding for am3offset operands.
   uint32_t getAddrMode3OffsetOpValue(const MCInst &MI, unsigned OpIdx,
-                                     SmallVectorImpl<MCFixup> &Fixups) const;
+                                     SmallVectorImpl<MCFixup> &Fixups,
+                                     const MCSubtargetInfo &STI) const;
 
   /// getAddrMode3OpValue - Return encoding for addrmode3 operands.
   uint32_t getAddrMode3OpValue(const MCInst &MI, unsigned OpIdx,
-                               SmallVectorImpl<MCFixup> &Fixups) const;
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const;
 
   /// getAddrModeThumbSPOpValue - Return encoding info for 'reg +/- imm12'
   /// operand.
   uint32_t getAddrModeThumbSPOpValue(const MCInst &MI, unsigned OpIdx,
-                                     SmallVectorImpl<MCFixup> &Fixups) const;
+                                     SmallVectorImpl<MCFixup> &Fixups,
+                                     const MCSubtargetInfo &STI) const;
 
   /// getAddrModeISOpValue - Encode the t_addrmode_is# operands.
   uint32_t getAddrModeISOpValue(const MCInst &MI, unsigned OpIdx,
-                                SmallVectorImpl<MCFixup> &Fixups) const;
+                                SmallVectorImpl<MCFixup> &Fixups,
+                                const MCSubtargetInfo &STI) const;
 
   /// getAddrModePCOpValue - Return encoding for t_addrmode_pc operands.
   uint32_t getAddrModePCOpValue(const MCInst &MI, unsigned OpIdx,
-                                SmallVectorImpl<MCFixup> &Fixups) const;
+                                SmallVectorImpl<MCFixup> &Fixups,
+                                const MCSubtargetInfo &STI) const;
 
   /// getAddrMode5OpValue - Return encoding info for 'reg +/- imm8' operand.
   uint32_t getAddrMode5OpValue(const MCInst &MI, unsigned OpIdx,
-                               SmallVectorImpl<MCFixup> &Fixups) const;
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const;
 
   /// getCCOutOpValue - Return encoding of the 's' bit.
   unsigned getCCOutOpValue(const MCInst &MI, unsigned Op,
-                           SmallVectorImpl<MCFixup> &Fixups) const {
+                           SmallVectorImpl<MCFixup> &Fixups,
+                           const MCSubtargetInfo &STI) const {
     // The operand is either reg0 or CPSR. The 's' bit is encoded as '0' or
     // '1' respectively.
     return MI.getOperand(Op).getReg() == ARM::CPSR;
@@ -237,8 +271,27 @@ public:
 
   /// getSOImmOpValue - Return an encoded 12-bit shifted-immediate value.
   unsigned getSOImmOpValue(const MCInst &MI, unsigned Op,
-                           SmallVectorImpl<MCFixup> &Fixups) const {
-    unsigned SoImm = MI.getOperand(Op).getImm();
+                           SmallVectorImpl<MCFixup> &Fixups,
+                           const MCSubtargetInfo &STI) const {
+
+    const MCOperand &MO = MI.getOperand(Op);
+
+    // We expect MO to be an immediate or an expression,
+    // if it is an immediate - that's fine, just encode the value.
+    // Otherwise - create a Fixup.
+    if (MO.isExpr()) {
+      const MCExpr *Expr = MO.getExpr();
+      // In instruction code this value always encoded as lowest 12 bits,
+      // so we don't have to perform any specific adjustments.
+      // Due to requirements of relocatable records we have to use FK_Data_4.
+      // See ARMELFObjectWriter::ExplicitRelSym and
+      //     ARMELFObjectWriter::GetRelocTypeInner for more details.
+      MCFixupKind Kind = MCFixupKind(FK_Data_4);
+      Fixups.push_back(MCFixup::Create(0, Expr, Kind, MI.getLoc()));
+      return 0;
+    }
+
+    unsigned SoImm = MO.getImm();
     int SoImmVal = ARM_AM::getSOImmVal(SoImm);
     assert(SoImmVal != -1 && "Not a valid so_imm value!");
 
@@ -253,7 +306,8 @@ public:
 
   /// getT2SOImmOpValue - Return an encoded 12-bit shifted-immediate value.
   unsigned getT2SOImmOpValue(const MCInst &MI, unsigned Op,
-                           SmallVectorImpl<MCFixup> &Fixups) const {
+                           SmallVectorImpl<MCFixup> &Fixups,
+                           const MCSubtargetInfo &STI) const {
     unsigned SoImm = MI.getOperand(Op).getImm();
     unsigned Encoded =  ARM_AM::getT2SOImmVal(SoImm);
     assert(Encoded != ~0U && "Not a Thumb2 so_imm value?");
@@ -261,64 +315,88 @@ public:
   }
 
   unsigned getT2AddrModeSORegOpValue(const MCInst &MI, unsigned OpNum,
-    SmallVectorImpl<MCFixup> &Fixups) const;
+    SmallVectorImpl<MCFixup> &Fixups,
+    const MCSubtargetInfo &STI) const;
   unsigned getT2AddrModeImm8OpValue(const MCInst &MI, unsigned OpNum,
-    SmallVectorImpl<MCFixup> &Fixups) const;
+    SmallVectorImpl<MCFixup> &Fixups,
+    const MCSubtargetInfo &STI) const;
   unsigned getT2AddrModeImm8OffsetOpValue(const MCInst &MI, unsigned OpNum,
-    SmallVectorImpl<MCFixup> &Fixups) const;
+    SmallVectorImpl<MCFixup> &Fixups,
+    const MCSubtargetInfo &STI) const;
   unsigned getT2AddrModeImm12OffsetOpValue(const MCInst &MI, unsigned OpNum,
-    SmallVectorImpl<MCFixup> &Fixups) const;
+    SmallVectorImpl<MCFixup> &Fixups,
+    const MCSubtargetInfo &STI) const;
 
   /// getSORegOpValue - Return an encoded so_reg shifted register value.
   unsigned getSORegRegOpValue(const MCInst &MI, unsigned Op,
-                           SmallVectorImpl<MCFixup> &Fixups) const;
+                           SmallVectorImpl<MCFixup> &Fixups,
+                           const MCSubtargetInfo &STI) const;
   unsigned getSORegImmOpValue(const MCInst &MI, unsigned Op,
-                           SmallVectorImpl<MCFixup> &Fixups) const;
+                           SmallVectorImpl<MCFixup> &Fixups,
+                           const MCSubtargetInfo &STI) const;
   unsigned getT2SORegOpValue(const MCInst &MI, unsigned Op,
-                             SmallVectorImpl<MCFixup> &Fixups) const;
+                             SmallVectorImpl<MCFixup> &Fixups,
+                             const MCSubtargetInfo &STI) const;
 
   unsigned getNEONVcvtImm32OpValue(const MCInst &MI, unsigned Op,
-                                   SmallVectorImpl<MCFixup> &Fixups) const {
+                                   SmallVectorImpl<MCFixup> &Fixups,
+                                   const MCSubtargetInfo &STI) const {
     return 64 - MI.getOperand(Op).getImm();
   }
 
   unsigned getBitfieldInvertedMaskOpValue(const MCInst &MI, unsigned Op,
-                                      SmallVectorImpl<MCFixup> &Fixups) const;
+                                      SmallVectorImpl<MCFixup> &Fixups,
+                                      const MCSubtargetInfo &STI) const;
 
   unsigned getRegisterListOpValue(const MCInst &MI, unsigned Op,
-                                  SmallVectorImpl<MCFixup> &Fixups) const;
+                                  SmallVectorImpl<MCFixup> &Fixups,
+                                  const MCSubtargetInfo &STI) const;
   unsigned getAddrMode6AddressOpValue(const MCInst &MI, unsigned Op,
-                                      SmallVectorImpl<MCFixup> &Fixups) const;
+                                      SmallVectorImpl<MCFixup> &Fixups,
+                                      const MCSubtargetInfo &STI) const;
   unsigned getAddrMode6OneLane32AddressOpValue(const MCInst &MI, unsigned Op,
-                                        SmallVectorImpl<MCFixup> &Fixups) const;
+                                        SmallVectorImpl<MCFixup> &Fixups,
+                                        const MCSubtargetInfo &STI) const;
   unsigned getAddrMode6DupAddressOpValue(const MCInst &MI, unsigned Op,
-                                        SmallVectorImpl<MCFixup> &Fixups) const;
+                                        SmallVectorImpl<MCFixup> &Fixups,
+                                        const MCSubtargetInfo &STI) const;
   unsigned getAddrMode6OffsetOpValue(const MCInst &MI, unsigned Op,
-                                     SmallVectorImpl<MCFixup> &Fixups) const;
+                                     SmallVectorImpl<MCFixup> &Fixups,
+                                     const MCSubtargetInfo &STI) const;
 
   unsigned getShiftRight8Imm(const MCInst &MI, unsigned Op,
-                             SmallVectorImpl<MCFixup> &Fixups) const;
+                             SmallVectorImpl<MCFixup> &Fixups,
+                             const MCSubtargetInfo &STI) const;
   unsigned getShiftRight16Imm(const MCInst &MI, unsigned Op,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
   unsigned getShiftRight32Imm(const MCInst &MI, unsigned Op,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
   unsigned getShiftRight64Imm(const MCInst &MI, unsigned Op,
-                              SmallVectorImpl<MCFixup> &Fixups) const;
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const;
 
   unsigned getThumbSRImmOpValue(const MCInst &MI, unsigned Op,
-                                 SmallVectorImpl<MCFixup> &Fixups) const;
+                                 SmallVectorImpl<MCFixup> &Fixups,
+                                 const MCSubtargetInfo &STI) const;
 
   unsigned NEONThumb2DataIPostEncoder(const MCInst &MI,
-                                      unsigned EncodedValue) const;
+                                      unsigned EncodedValue,
+                                      const MCSubtargetInfo &STI) const;
   unsigned NEONThumb2LoadStorePostEncoder(const MCInst &MI,
-                                          unsigned EncodedValue) const;
+                                          unsigned EncodedValue,
+                                          const MCSubtargetInfo &STI) const;
   unsigned NEONThumb2DupPostEncoder(const MCInst &MI,
-                                    unsigned EncodedValue) const;
+                                    unsigned EncodedValue,
+                                    const MCSubtargetInfo &STI) const;
   unsigned NEONThumb2V8PostEncoder(const MCInst &MI,
-                                   unsigned EncodedValue) const;
+                                   unsigned EncodedValue,
+                                   const MCSubtargetInfo &STI) const;
 
   unsigned VFPThumb2PostEncoder(const MCInst &MI,
-                                unsigned EncodedValue) const;
+                                unsigned EncodedValue,
+                                const MCSubtargetInfo &STI) const;
 
   void EmitByte(unsigned char C, raw_ostream &OS) const {
     OS << (char)C;
@@ -327,30 +405,39 @@ public:
   void EmitConstant(uint64_t Val, unsigned Size, raw_ostream &OS) const {
     // Output the constant in little endian byte order.
     for (unsigned i = 0; i != Size; ++i) {
-      EmitByte(Val & 255, OS);
-      Val >>= 8;
+      unsigned Shift = IsLittleEndian ? i * 8 : (Size - 1 - i) * 8;
+      EmitByte((Val >> Shift) & 0xff, OS);
     }
   }
 
   void EncodeInstruction(const MCInst &MI, raw_ostream &OS,
-                         SmallVectorImpl<MCFixup> &Fixups) const;
+                         SmallVectorImpl<MCFixup> &Fixups,
+                         const MCSubtargetInfo &STI) const override;
 };
 
 } // end anonymous namespace
 
-MCCodeEmitter *llvm::createARMMCCodeEmitter(const MCInstrInfo &MCII,
-                                            const MCRegisterInfo &MRI,
-                                            const MCSubtargetInfo &STI,
-                                            MCContext &Ctx) {
-  return new ARMMCCodeEmitter(MCII, STI, Ctx);
+MCCodeEmitter *llvm::createARMLEMCCodeEmitter(const MCInstrInfo &MCII,
+                                              const MCRegisterInfo &MRI,
+                                              const MCSubtargetInfo &STI,
+                                              MCContext &Ctx) {
+  return new ARMMCCodeEmitter(MCII, Ctx, true);
+}
+
+MCCodeEmitter *llvm::createARMBEMCCodeEmitter(const MCInstrInfo &MCII,
+                                              const MCRegisterInfo &MRI,
+                                              const MCSubtargetInfo &STI,
+                                              MCContext &Ctx) {
+  return new ARMMCCodeEmitter(MCII, Ctx, false);
 }
 
 /// NEONThumb2DataIPostEncoder - Post-process encoded NEON data-processing
 /// instructions, and rewrite them to their Thumb2 form if we are currently in
 /// Thumb2 mode.
 unsigned ARMMCCodeEmitter::NEONThumb2DataIPostEncoder(const MCInst &MI,
-                                                 unsigned EncodedValue) const {
-  if (isThumb2()) {
+                                                 unsigned EncodedValue,
+                                                 const MCSubtargetInfo &STI) const {
+  if (isThumb2(STI)) {
     // NEON Thumb2 data-processsing encodings are very simple: bit 24 is moved
     // to bit 12 of the high half-word (i.e. bit 28), and bits 27-24 are
     // set to 1111.
@@ -368,8 +455,9 @@ unsigned ARMMCCodeEmitter::NEONThumb2DataIPostEncoder(const MCInst &MI,
 /// instructions, and rewrite them to their Thumb2 form if we are currently in
 /// Thumb2 mode.
 unsigned ARMMCCodeEmitter::NEONThumb2LoadStorePostEncoder(const MCInst &MI,
-                                                 unsigned EncodedValue) const {
-  if (isThumb2()) {
+                                                 unsigned EncodedValue,
+                                                 const MCSubtargetInfo &STI) const {
+  if (isThumb2(STI)) {
     EncodedValue &= 0xF0FFFFFF;
     EncodedValue |= 0x09000000;
   }
@@ -381,8 +469,9 @@ unsigned ARMMCCodeEmitter::NEONThumb2LoadStorePostEncoder(const MCInst &MI,
 /// instructions, and rewrite them to their Thumb2 form if we are currently in
 /// Thumb2 mode.
 unsigned ARMMCCodeEmitter::NEONThumb2DupPostEncoder(const MCInst &MI,
-                                                 unsigned EncodedValue) const {
-  if (isThumb2()) {
+                                                 unsigned EncodedValue,
+                                                 const MCSubtargetInfo &STI) const {
+  if (isThumb2(STI)) {
     EncodedValue &= 0x00FFFFFF;
     EncodedValue |= 0xEE000000;
   }
@@ -393,8 +482,9 @@ unsigned ARMMCCodeEmitter::NEONThumb2DupPostEncoder(const MCInst &MI,
 /// Post-process encoded NEON v8 instructions, and rewrite them to Thumb2 form
 /// if we are in Thumb2.
 unsigned ARMMCCodeEmitter::NEONThumb2V8PostEncoder(const MCInst &MI,
-                                                 unsigned EncodedValue) const {
-  if (isThumb2()) {
+                                                 unsigned EncodedValue,
+                                                 const MCSubtargetInfo &STI) const {
+  if (isThumb2(STI)) {
     EncodedValue |= 0xC000000; // Set bits 27-26
   }
 
@@ -404,8 +494,9 @@ unsigned ARMMCCodeEmitter::NEONThumb2V8PostEncoder(const MCInst &MI,
 /// VFPThumb2PostEncoder - Post-process encoded VFP instructions and rewrite
 /// them to their Thumb2 form if we are currently in Thumb2 mode.
 unsigned ARMMCCodeEmitter::
-VFPThumb2PostEncoder(const MCInst &MI, unsigned EncodedValue) const {
-  if (isThumb2()) {
+VFPThumb2PostEncoder(const MCInst &MI, unsigned EncodedValue,
+                     const MCSubtargetInfo &STI) const {
+  if (isThumb2(STI)) {
     EncodedValue &= 0x0FFFFFFF;
     EncodedValue |= 0xE0000000;
   }
@@ -416,7 +507,8 @@ VFPThumb2PostEncoder(const MCInst &MI, unsigned EncodedValue) const {
 /// operand requires relocation, record the relocation and return zero.
 unsigned ARMMCCodeEmitter::
 getMachineOpValue(const MCInst &MI, const MCOperand &MO,
-                  SmallVectorImpl<MCFixup> &Fixups) const {
+                  SmallVectorImpl<MCFixup> &Fixups,
+                  const MCSubtargetInfo &STI) const {
   if (MO.isReg()) {
     unsigned Reg = MO.getReg();
     unsigned RegNo = CTX.getRegisterInfo()->getEncodingValue(Reg);
@@ -444,7 +536,8 @@ getMachineOpValue(const MCInst &MI, const MCOperand &MO,
 /// getAddrModeImmOpValue - Return encoding info for 'reg +/- imm' operand.
 bool ARMMCCodeEmitter::
 EncodeAddrModeOpValues(const MCInst &MI, unsigned OpIdx, unsigned &Reg,
-                       unsigned &Imm, SmallVectorImpl<MCFixup> &Fixups) const {
+                       unsigned &Imm, SmallVectorImpl<MCFixup> &Fixups,
+ const MCSubtargetInfo &STI) const {
   const MCOperand &MO  = MI.getOperand(OpIdx);
   const MCOperand &MO1 = MI.getOperand(OpIdx + 1);
 
@@ -473,7 +566,8 @@ EncodeAddrModeOpValues(const MCInst &MI, unsigned OpIdx, unsigned &Reg,
 /// which is either an immediate or requires a fixup.
 static uint32_t getBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
                                        unsigned FixupKind,
-                                       SmallVectorImpl<MCFixup> &Fixups) {
+                                       SmallVectorImpl<MCFixup> &Fixups,
+                                       const MCSubtargetInfo &STI) {
   const MCOperand &MO = MI.getOperand(OpIdx);
 
   // If the destination is an immediate, we have nothing to do.
@@ -509,11 +603,12 @@ static int32_t encodeThumbBLOffset(int32_t offset) {
 /// getThumbBLTargetOpValue - Return encoding info for immediate branch target.
 uint32_t ARMMCCodeEmitter::
 getThumbBLTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                        SmallVectorImpl<MCFixup> &Fixups) const {
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_bl,
-                                    Fixups);
+                                    Fixups, STI);
   return encodeThumbBLOffset(MO.getImm());
 }
 
@@ -521,43 +616,47 @@ getThumbBLTargetOpValue(const MCInst &MI, unsigned OpIdx,
 /// BLX branch target.
 uint32_t ARMMCCodeEmitter::
 getThumbBLXTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                         SmallVectorImpl<MCFixup> &Fixups) const {
+                         SmallVectorImpl<MCFixup> &Fixups,
+                         const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_blx,
-                                    Fixups);
+                                    Fixups, STI);
   return encodeThumbBLOffset(MO.getImm());
 }
 
 /// getThumbBRTargetOpValue - Return encoding info for Thumb branch target.
 uint32_t ARMMCCodeEmitter::
 getThumbBRTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                        SmallVectorImpl<MCFixup> &Fixups) const {
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_br,
-                                    Fixups);
+                                    Fixups, STI);
   return (MO.getImm() >> 1);
 }
 
 /// getThumbBCCTargetOpValue - Return encoding info for Thumb branch target.
 uint32_t ARMMCCodeEmitter::
 getThumbBCCTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                         SmallVectorImpl<MCFixup> &Fixups) const {
+                         SmallVectorImpl<MCFixup> &Fixups,
+                         const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_bcc,
-                                    Fixups);
+                                    Fixups, STI);
   return (MO.getImm() >> 1);
 }
 
 /// getThumbCBTargetOpValue - Return encoding info for Thumb branch target.
 uint32_t ARMMCCodeEmitter::
 getThumbCBTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                        SmallVectorImpl<MCFixup> &Fixups) const {
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
-    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_cb, Fixups);
+    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_cb, Fixups, STI);
   return (MO.getImm() >> 1);
 }
 
@@ -582,27 +681,29 @@ static bool HasConditionalBranch(const MCInst &MI) {
 /// target.
 uint32_t ARMMCCodeEmitter::
 getBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                       SmallVectorImpl<MCFixup> &Fixups) const {
+                       SmallVectorImpl<MCFixup> &Fixups,
+                       const MCSubtargetInfo &STI) const {
   // FIXME: This really, really shouldn't use TargetMachine. We don't want
   // coupling between MC and TM anywhere we can help it.
-  if (isThumb2())
+  if (isThumb2(STI))
     return
-      ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_t2_condbranch, Fixups);
-  return getARMBranchTargetOpValue(MI, OpIdx, Fixups);
+      ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_t2_condbranch, Fixups, STI);
+  return getARMBranchTargetOpValue(MI, OpIdx, Fixups, STI);
 }
 
 /// getBranchTargetOpValue - Return encoding info for 24-bit immediate branch
 /// target.
 uint32_t ARMMCCodeEmitter::
 getARMBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr()) {
     if (HasConditionalBranch(MI))
       return ::getBranchTargetOpValue(MI, OpIdx,
-                                      ARM::fixup_arm_condbranch, Fixups);
+                                      ARM::fixup_arm_condbranch, Fixups, STI);
     return ::getBranchTargetOpValue(MI, OpIdx,
-                                    ARM::fixup_arm_uncondbranch, Fixups);
+                                    ARM::fixup_arm_uncondbranch, Fixups, STI);
   }
 
   return MO.getImm() >> 2;
@@ -610,13 +711,14 @@ getARMBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
 
 uint32_t ARMMCCodeEmitter::
 getARMBLTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr()) {
     if (HasConditionalBranch(MI))
       return ::getBranchTargetOpValue(MI, OpIdx, 
-                                      ARM::fixup_arm_condbl, Fixups);
-    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_uncondbl, Fixups);
+                                      ARM::fixup_arm_condbl, Fixups, STI);
+    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_uncondbl, Fixups, STI);
   }
 
   return MO.getImm() >> 2;
@@ -624,10 +726,11 @@ getARMBLTargetOpValue(const MCInst &MI, unsigned OpIdx,
 
 uint32_t ARMMCCodeEmitter::
 getARMBLXTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
-    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_blx, Fixups);
+    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_blx, Fixups, STI);
 
   return MO.getImm() >> 1;
 }
@@ -636,12 +739,13 @@ getARMBLXTargetOpValue(const MCInst &MI, unsigned OpIdx,
 /// immediate branch target.
 uint32_t ARMMCCodeEmitter::
 getUnconditionalBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
-                       SmallVectorImpl<MCFixup> &Fixups) const {
+                       SmallVectorImpl<MCFixup> &Fixups,
+                       const MCSubtargetInfo &STI) const {
   unsigned Val = 0;
   const MCOperand MO = MI.getOperand(OpIdx);
     
   if(MO.isExpr())
-    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_t2_uncondbranch, Fixups);
+    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_t2_uncondbranch, Fixups, STI);
   else 
     Val = MO.getImm() >> 1;
 
@@ -665,11 +769,12 @@ getUnconditionalBranchTargetOpValue(const MCInst &MI, unsigned OpIdx,
 /// ADR label target.
 uint32_t ARMMCCodeEmitter::
 getAdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_adr_pcrel_12,
-                                    Fixups);
+                                    Fixups, STI);
   int64_t offset = MO.getImm();
   uint32_t Val = 0x2000;
 
@@ -705,11 +810,12 @@ getAdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
 /// target.
 uint32_t ARMMCCodeEmitter::
 getT2AdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_t2_adr_pcrel_12,
-                                    Fixups);
+                                    Fixups, STI);
   int32_t Val = MO.getImm();
   if (Val == INT32_MIN)
     Val = 0x1000;
@@ -724,11 +830,12 @@ getT2AdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
 /// target.
 uint32_t ARMMCCodeEmitter::
 getThumbAdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
     return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_thumb_adr_pcrel_10,
-                                    Fixups);
+                                    Fixups, STI);
   return MO.getImm();
 }
 
@@ -736,7 +843,8 @@ getThumbAdrLabelOpValue(const MCInst &MI, unsigned OpIdx,
 /// operand.
 uint32_t ARMMCCodeEmitter::
 getThumbAddrModeRegRegOpValue(const MCInst &MI, unsigned OpIdx,
-                              SmallVectorImpl<MCFixup> &) const {
+                              SmallVectorImpl<MCFixup> &,
+                              const MCSubtargetInfo &STI) const {
   // [Rn, Rm]
   //   {5-3} = Rm
   //   {2-0} = Rn
@@ -750,7 +858,8 @@ getThumbAddrModeRegRegOpValue(const MCInst &MI, unsigned OpIdx,
 /// getAddrModeImm12OpValue - Return encoding info for 'reg +/- imm12' operand.
 uint32_t ARMMCCodeEmitter::
 getAddrModeImm12OpValue(const MCInst &MI, unsigned OpIdx,
-                        SmallVectorImpl<MCFixup> &Fixups) const {
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
   // {17-13} = reg
   // {12}    = (U)nsigned (add == '1', sub == '0')
   // {11-0}  = imm12
@@ -767,7 +876,7 @@ getAddrModeImm12OpValue(const MCInst &MI, unsigned OpIdx,
       isAdd = false ; // 'U' bit is set as part of the fixup.
 
       MCFixupKind Kind;
-      if (isThumb2())
+      if (isThumb2(STI))
         Kind = MCFixupKind(ARM::fixup_t2_ldst_pcrel_12);
       else
         Kind = MCFixupKind(ARM::fixup_arm_ldst_pcrel_12);
@@ -787,7 +896,7 @@ getAddrModeImm12OpValue(const MCInst &MI, unsigned OpIdx,
       Imm12 = Offset;
     }
   } else
-    isAdd = EncodeAddrModeOpValues(MI, OpIdx, Reg, Imm12, Fixups);
+    isAdd = EncodeAddrModeOpValues(MI, OpIdx, Reg, Imm12, Fixups, STI);
 
   uint32_t Binary = Imm12 & 0xfff;
   // Immediate is always encoded as positive. The 'U' bit controls add vs sub.
@@ -801,7 +910,8 @@ getAddrModeImm12OpValue(const MCInst &MI, unsigned OpIdx,
 /// '+/- imm8<<2' operand.
 uint32_t ARMMCCodeEmitter::
 getT2Imm8s4OpValue(const MCInst &MI, unsigned OpIdx,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   // FIXME: The immediate operand should have already been encoded like this
   // before ever getting here. The encoder method should just need to combine
   // the MI operands for the register and the offset into a single
@@ -832,7 +942,8 @@ getT2Imm8s4OpValue(const MCInst &MI, unsigned OpIdx,
 /// 'reg +/- imm8<<2' operand.
 uint32_t ARMMCCodeEmitter::
 getT2AddrModeImm8s4OpValue(const MCInst &MI, unsigned OpIdx,
-                        SmallVectorImpl<MCFixup> &Fixups) const {
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
   // {12-9} = reg
   // {8}    = (U)nsigned (add == '1', sub == '0')
   // {7-0}  = imm8
@@ -852,7 +963,7 @@ getT2AddrModeImm8s4OpValue(const MCInst &MI, unsigned OpIdx,
 
     ++MCNumCPRelocations;
   } else
-    isAdd = EncodeAddrModeOpValues(MI, OpIdx, Reg, Imm8, Fixups);
+    isAdd = EncodeAddrModeOpValues(MI, OpIdx, Reg, Imm8, Fixups, STI);
 
   // FIXME: The immediate operand should have already been encoded like this
   // before ever getting here. The encoder method should just need to combine
@@ -872,7 +983,8 @@ getT2AddrModeImm8s4OpValue(const MCInst &MI, unsigned OpIdx,
 /// 'reg + imm8<<2' operand.
 uint32_t ARMMCCodeEmitter::
 getT2AddrModeImm0_1020s4OpValue(const MCInst &MI, unsigned OpIdx,
-                        SmallVectorImpl<MCFixup> &Fixups) const {
+                        SmallVectorImpl<MCFixup> &Fixups,
+                        const MCSubtargetInfo &STI) const {
   // {11-8} = reg
   // {7-0}  = imm8
   const MCOperand &MO = MI.getOperand(OpIdx);
@@ -882,22 +994,10 @@ getT2AddrModeImm0_1020s4OpValue(const MCInst &MI, unsigned OpIdx,
   return (Reg << 8) | Imm8;
 }
 
-// FIXME: This routine assumes that a binary
-// expression will always result in a PCRel expression
-// In reality, its only true if one or more subexpressions
-// is itself a PCRel (i.e. "." in asm or some other pcrel construct)
-// but this is good enough for now.
-static bool EvaluateAsPCRel(const MCExpr *Expr) {
-  switch (Expr->getKind()) {
-  default: llvm_unreachable("Unexpected expression type");
-  case MCExpr::SymbolRef: return false;
-  case MCExpr::Binary: return true;
-  }
-}
-
 uint32_t
 ARMMCCodeEmitter::getHiLo16ImmOpValue(const MCInst &MI, unsigned OpIdx,
-                                      SmallVectorImpl<MCFixup> &Fixups) const {
+                                      SmallVectorImpl<MCFixup> &Fixups,
+                                      const MCSubtargetInfo &STI) const {
   // {20-16} = imm{15-12}
   // {11-0}  = imm{11-0}
   const MCOperand &MO = MI.getOperand(OpIdx);
@@ -912,51 +1012,48 @@ ARMMCCodeEmitter::getHiLo16ImmOpValue(const MCInst &MI, unsigned OpIdx,
     const ARMMCExpr *ARM16Expr = cast<ARMMCExpr>(E);
     E = ARM16Expr->getSubExpr();
 
+    if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(E)) {
+      const int64_t Value = MCE->getValue();
+      if (Value > UINT32_MAX)
+        report_fatal_error("constant value truncated (limited to 32-bit)");
+
+      switch (ARM16Expr->getKind()) {
+      case ARMMCExpr::VK_ARM_HI16:
+        return (int32_t(Value) & 0xffff0000) >> 16;
+      case ARMMCExpr::VK_ARM_LO16:
+        return (int32_t(Value) & 0x0000ffff);
+      default: llvm_unreachable("Unsupported ARMFixup");
+      }
+    }
+
     switch (ARM16Expr->getKind()) {
     default: llvm_unreachable("Unsupported ARMFixup");
     case ARMMCExpr::VK_ARM_HI16:
-      if (!isTargetDarwin() && EvaluateAsPCRel(E))
-        Kind = MCFixupKind(isThumb2()
-                           ? ARM::fixup_t2_movt_hi16_pcrel
-                           : ARM::fixup_arm_movt_hi16_pcrel);
-      else
-        Kind = MCFixupKind(isThumb2()
-                           ? ARM::fixup_t2_movt_hi16
-                           : ARM::fixup_arm_movt_hi16);
+      Kind = MCFixupKind(isThumb2(STI) ? ARM::fixup_t2_movt_hi16
+                                       : ARM::fixup_arm_movt_hi16);
       break;
     case ARMMCExpr::VK_ARM_LO16:
-      if (!isTargetDarwin() && EvaluateAsPCRel(E))
-        Kind = MCFixupKind(isThumb2()
-                           ? ARM::fixup_t2_movw_lo16_pcrel
-                           : ARM::fixup_arm_movw_lo16_pcrel);
-      else
-        Kind = MCFixupKind(isThumb2()
-                           ? ARM::fixup_t2_movw_lo16
-                           : ARM::fixup_arm_movw_lo16);
+      Kind = MCFixupKind(isThumb2(STI) ? ARM::fixup_t2_movw_lo16
+                                       : ARM::fixup_arm_movw_lo16);
       break;
     }
+
     Fixups.push_back(MCFixup::Create(0, E, Kind, MI.getLoc()));
     return 0;
   }
   // If the expression doesn't have :upper16: or :lower16: on it,
-  // it's just a plain immediate expression, and those evaluate to
+  // it's just a plain immediate expression, previously those evaluated to
   // the lower 16 bits of the expression regardless of whether
-  // we have a movt or a movw.
-  if (!isTargetDarwin() && EvaluateAsPCRel(E))
-    Kind = MCFixupKind(isThumb2()
-                       ? ARM::fixup_t2_movw_lo16_pcrel
-                       : ARM::fixup_arm_movw_lo16_pcrel);
-  else
-    Kind = MCFixupKind(isThumb2()
-                       ? ARM::fixup_t2_movw_lo16
-                       : ARM::fixup_arm_movw_lo16);
-  Fixups.push_back(MCFixup::Create(0, E, Kind, MI.getLoc()));
-  return 0;
+  // we have a movt or a movw, but that led to misleadingly results.
+  // This is now disallowed in the the AsmParser in validateInstruction()
+  // so this should never happen.
+  llvm_unreachable("expression without :upper16: or :lower16:");
 }
 
 uint32_t ARMMCCodeEmitter::
 getLdStSORegOpValue(const MCInst &MI, unsigned OpIdx,
-                    SmallVectorImpl<MCFixup> &Fixups) const {
+                    SmallVectorImpl<MCFixup> &Fixups,
+                    const MCSubtargetInfo &STI) const {
   const MCOperand &MO = MI.getOperand(OpIdx);
   const MCOperand &MO1 = MI.getOperand(OpIdx+1);
   const MCOperand &MO2 = MI.getOperand(OpIdx+2);
@@ -989,21 +1086,23 @@ getLdStSORegOpValue(const MCInst &MI, unsigned OpIdx,
 
 uint32_t ARMMCCodeEmitter::
 getAddrMode2OpValue(const MCInst &MI, unsigned OpIdx,
-                    SmallVectorImpl<MCFixup> &Fixups) const {
+                    SmallVectorImpl<MCFixup> &Fixups,
+                    const MCSubtargetInfo &STI) const {
   // {17-14}  Rn
   // {13}     1 == imm12, 0 == Rm
   // {12}     isAdd
   // {11-0}   imm12/Rm
   const MCOperand &MO = MI.getOperand(OpIdx);
   unsigned Rn = CTX.getRegisterInfo()->getEncodingValue(MO.getReg());
-  uint32_t Binary = getAddrMode2OffsetOpValue(MI, OpIdx + 1, Fixups);
+  uint32_t Binary = getAddrMode2OffsetOpValue(MI, OpIdx + 1, Fixups, STI);
   Binary |= Rn << 14;
   return Binary;
 }
 
 uint32_t ARMMCCodeEmitter::
 getAddrMode2OffsetOpValue(const MCInst &MI, unsigned OpIdx,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   // {13}     1 == imm12, 0 == Rm
   // {12}     isAdd
   // {11-0}   imm12/Rm
@@ -1025,7 +1124,8 @@ getAddrMode2OffsetOpValue(const MCInst &MI, unsigned OpIdx,
 
 uint32_t ARMMCCodeEmitter::
 getPostIdxRegOpValue(const MCInst &MI, unsigned OpIdx,
-                     SmallVectorImpl<MCFixup> &Fixups) const {
+                     SmallVectorImpl<MCFixup> &Fixups,
+                     const MCSubtargetInfo &STI) const {
   // {4}      isAdd
   // {3-0}    Rm
   const MCOperand &MO = MI.getOperand(OpIdx);
@@ -1036,7 +1136,8 @@ getPostIdxRegOpValue(const MCInst &MI, unsigned OpIdx,
 
 uint32_t ARMMCCodeEmitter::
 getAddrMode3OffsetOpValue(const MCInst &MI, unsigned OpIdx,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   // {9}      1 == imm8, 0 == Rm
   // {8}      isAdd
   // {7-4}    imm7_4/zero
@@ -1055,7 +1156,8 @@ getAddrMode3OffsetOpValue(const MCInst &MI, unsigned OpIdx,
 
 uint32_t ARMMCCodeEmitter::
 getAddrMode3OpValue(const MCInst &MI, unsigned OpIdx,
-                    SmallVectorImpl<MCFixup> &Fixups) const {
+                    SmallVectorImpl<MCFixup> &Fixups,
+                    const MCSubtargetInfo &STI) const {
   // {13}     1 == imm8, 0 == Rm
   // {12-9}   Rn
   // {8}      isAdd
@@ -1091,7 +1193,8 @@ getAddrMode3OpValue(const MCInst &MI, unsigned OpIdx,
 /// getAddrModeThumbSPOpValue - Encode the t_addrmode_sp operands.
 uint32_t ARMMCCodeEmitter::
 getAddrModeThumbSPOpValue(const MCInst &MI, unsigned OpIdx,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   // [SP, #imm]
   //   {7-0} = imm8
   const MCOperand &MO1 = MI.getOperand(OpIdx + 1);
@@ -1106,7 +1209,8 @@ getAddrModeThumbSPOpValue(const MCInst &MI, unsigned OpIdx,
 /// getAddrModeISOpValue - Encode the t_addrmode_is# operands.
 uint32_t ARMMCCodeEmitter::
 getAddrModeISOpValue(const MCInst &MI, unsigned OpIdx,
-                     SmallVectorImpl<MCFixup> &Fixups) const {
+                     SmallVectorImpl<MCFixup> &Fixups,
+                     const MCSubtargetInfo &STI) const {
   // [Rn, #imm]
   //   {7-3} = imm5
   //   {2-0} = Rn
@@ -1120,17 +1224,19 @@ getAddrModeISOpValue(const MCInst &MI, unsigned OpIdx,
 /// getAddrModePCOpValue - Return encoding for t_addrmode_pc operands.
 uint32_t ARMMCCodeEmitter::
 getAddrModePCOpValue(const MCInst &MI, unsigned OpIdx,
-                     SmallVectorImpl<MCFixup> &Fixups) const {
+                     SmallVectorImpl<MCFixup> &Fixups,
+                     const MCSubtargetInfo &STI) const {
   const MCOperand MO = MI.getOperand(OpIdx);
   if (MO.isExpr())
-    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_cp, Fixups);
+    return ::getBranchTargetOpValue(MI, OpIdx, ARM::fixup_arm_thumb_cp, Fixups, STI);
   return (MO.getImm() >> 2);
 }
 
 /// getAddrMode5OpValue - Return encoding info for 'reg +/- imm10' operand.
 uint32_t ARMMCCodeEmitter::
 getAddrMode5OpValue(const MCInst &MI, unsigned OpIdx,
-                    SmallVectorImpl<MCFixup> &Fixups) const {
+                    SmallVectorImpl<MCFixup> &Fixups,
+                    const MCSubtargetInfo &STI) const {
   // {12-9} = reg
   // {8}    = (U)nsigned (add == '1', sub == '0')
   // {7-0}  = imm8
@@ -1146,7 +1252,7 @@ getAddrMode5OpValue(const MCInst &MI, unsigned OpIdx,
     assert(MO.isExpr() && "Unexpected machine operand type!");
     const MCExpr *Expr = MO.getExpr();
     MCFixupKind Kind;
-    if (isThumb2())
+    if (isThumb2(STI))
       Kind = MCFixupKind(ARM::fixup_t2_pcrel_10);
     else
       Kind = MCFixupKind(ARM::fixup_arm_pcrel_10);
@@ -1154,7 +1260,7 @@ getAddrMode5OpValue(const MCInst &MI, unsigned OpIdx,
 
     ++MCNumCPRelocations;
   } else {
-    EncodeAddrModeOpValues(MI, OpIdx, Reg, Imm8, Fixups);
+    EncodeAddrModeOpValues(MI, OpIdx, Reg, Imm8, Fixups, STI);
     isAdd = ARM_AM::getAM5Op(Imm8) == ARM_AM::add;
   }
 
@@ -1168,7 +1274,8 @@ getAddrMode5OpValue(const MCInst &MI, unsigned OpIdx,
 
 unsigned ARMMCCodeEmitter::
 getSORegRegOpValue(const MCInst &MI, unsigned OpIdx,
-                SmallVectorImpl<MCFixup> &Fixups) const {
+                SmallVectorImpl<MCFixup> &Fixups,
+                const MCSubtargetInfo &STI) const {
   // Sub-operands are [reg, reg, imm]. The first register is Rm, the reg to be
   // shifted. The second is Rs, the amount to shift by, and the third specifies
   // the type of the shift.
@@ -1215,7 +1322,8 @@ getSORegRegOpValue(const MCInst &MI, unsigned OpIdx,
 
 unsigned ARMMCCodeEmitter::
 getSORegImmOpValue(const MCInst &MI, unsigned OpIdx,
-                SmallVectorImpl<MCFixup> &Fixups) const {
+                SmallVectorImpl<MCFixup> &Fixups,
+                const MCSubtargetInfo &STI) const {
   // Sub-operands are [reg, imm]. The first register is Rm, the reg to be
   // shifted. The second is the amount to shift by.
   //
@@ -1261,7 +1369,8 @@ getSORegImmOpValue(const MCInst &MI, unsigned OpIdx,
 
 unsigned ARMMCCodeEmitter::
 getT2AddrModeSORegOpValue(const MCInst &MI, unsigned OpNum,
-                SmallVectorImpl<MCFixup> &Fixups) const {
+                SmallVectorImpl<MCFixup> &Fixups,
+                const MCSubtargetInfo &STI) const {
   const MCOperand &MO1 = MI.getOperand(OpNum);
   const MCOperand &MO2 = MI.getOperand(OpNum+1);
   const MCOperand &MO3 = MI.getOperand(OpNum+2);
@@ -1279,7 +1388,8 @@ getT2AddrModeSORegOpValue(const MCInst &MI, unsigned OpNum,
 
 unsigned ARMMCCodeEmitter::
 getT2AddrModeImm8OpValue(const MCInst &MI, unsigned OpNum,
-                         SmallVectorImpl<MCFixup> &Fixups) const {
+                         SmallVectorImpl<MCFixup> &Fixups,
+                         const MCSubtargetInfo &STI) const {
   const MCOperand &MO1 = MI.getOperand(OpNum);
   const MCOperand &MO2 = MI.getOperand(OpNum+1);
 
@@ -1300,7 +1410,8 @@ getT2AddrModeImm8OpValue(const MCInst &MI, unsigned OpNum,
 
 unsigned ARMMCCodeEmitter::
 getT2AddrModeImm8OffsetOpValue(const MCInst &MI, unsigned OpNum,
-                         SmallVectorImpl<MCFixup> &Fixups) const {
+                         SmallVectorImpl<MCFixup> &Fixups,
+                         const MCSubtargetInfo &STI) const {
   const MCOperand &MO1 = MI.getOperand(OpNum);
 
   // FIXME: Needs fixup support.
@@ -1316,7 +1427,8 @@ getT2AddrModeImm8OffsetOpValue(const MCInst &MI, unsigned OpNum,
 
 unsigned ARMMCCodeEmitter::
 getT2AddrModeImm12OffsetOpValue(const MCInst &MI, unsigned OpNum,
-                         SmallVectorImpl<MCFixup> &Fixups) const {
+                         SmallVectorImpl<MCFixup> &Fixups,
+                         const MCSubtargetInfo &STI) const {
   const MCOperand &MO1 = MI.getOperand(OpNum);
 
   // FIXME: Needs fixup support.
@@ -1332,7 +1444,8 @@ getT2AddrModeImm12OffsetOpValue(const MCInst &MI, unsigned OpNum,
 
 unsigned ARMMCCodeEmitter::
 getT2SORegOpValue(const MCInst &MI, unsigned OpIdx,
-                SmallVectorImpl<MCFixup> &Fixups) const {
+                SmallVectorImpl<MCFixup> &Fixups,
+                const MCSubtargetInfo &STI) const {
   // Sub-operands are [reg, imm]. The first register is Rm, the reg to be
   // shifted. The second is the amount to shift by.
   //
@@ -1374,7 +1487,8 @@ getT2SORegOpValue(const MCInst &MI, unsigned OpIdx,
 
 unsigned ARMMCCodeEmitter::
 getBitfieldInvertedMaskOpValue(const MCInst &MI, unsigned Op,
-                               SmallVectorImpl<MCFixup> &Fixups) const {
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const {
   // 10 bits. lower 5 bits are are the lsb of the mask, high five bits are the
   // msb of the mask.
   const MCOperand &MO = MI.getOperand(Op);
@@ -1387,7 +1501,8 @@ getBitfieldInvertedMaskOpValue(const MCInst &MI, unsigned Op,
 
 unsigned ARMMCCodeEmitter::
 getRegisterListOpValue(const MCInst &MI, unsigned Op,
-                       SmallVectorImpl<MCFixup> &Fixups) const {
+                       SmallVectorImpl<MCFixup> &Fixups,
+                       const MCSubtargetInfo &STI) const {
   // VLDM/VSTM:
   //   {12-8} = Vd
   //   {7-0}  = Number of registers
@@ -1423,7 +1538,8 @@ getRegisterListOpValue(const MCInst &MI, unsigned Op,
 /// with the alignment operand.
 unsigned ARMMCCodeEmitter::
 getAddrMode6AddressOpValue(const MCInst &MI, unsigned Op,
-                           SmallVectorImpl<MCFixup> &Fixups) const {
+                           SmallVectorImpl<MCFixup> &Fixups,
+                           const MCSubtargetInfo &STI) const {
   const MCOperand &Reg = MI.getOperand(Op);
   const MCOperand &Imm = MI.getOperand(Op + 1);
 
@@ -1446,7 +1562,8 @@ getAddrMode6AddressOpValue(const MCInst &MI, unsigned Op,
 /// along  with the alignment operand for use in VST1 and VLD1 with size 32.
 unsigned ARMMCCodeEmitter::
 getAddrMode6OneLane32AddressOpValue(const MCInst &MI, unsigned Op,
-                                    SmallVectorImpl<MCFixup> &Fixups) const {
+                                    SmallVectorImpl<MCFixup> &Fixups,
+                                    const MCSubtargetInfo &STI) const {
   const MCOperand &Reg = MI.getOperand(Op);
   const MCOperand &Imm = MI.getOperand(Op + 1);
 
@@ -1472,7 +1589,8 @@ getAddrMode6OneLane32AddressOpValue(const MCInst &MI, unsigned Op,
 /// different for VLD4-dup.
 unsigned ARMMCCodeEmitter::
 getAddrMode6DupAddressOpValue(const MCInst &MI, unsigned Op,
-                              SmallVectorImpl<MCFixup> &Fixups) const {
+                              SmallVectorImpl<MCFixup> &Fixups,
+                              const MCSubtargetInfo &STI) const {
   const MCOperand &Reg = MI.getOperand(Op);
   const MCOperand &Imm = MI.getOperand(Op + 1);
 
@@ -1492,7 +1610,8 @@ getAddrMode6DupAddressOpValue(const MCInst &MI, unsigned Op,
 
 unsigned ARMMCCodeEmitter::
 getAddrMode6OffsetOpValue(const MCInst &MI, unsigned Op,
-                          SmallVectorImpl<MCFixup> &Fixups) const {
+                          SmallVectorImpl<MCFixup> &Fixups,
+                          const MCSubtargetInfo &STI) const {
   const MCOperand &MO = MI.getOperand(Op);
   if (MO.getReg() == 0) return 0x0D;
   return CTX.getRegisterInfo()->getEncodingValue(MO.getReg());
@@ -1500,31 +1619,36 @@ getAddrMode6OffsetOpValue(const MCInst &MI, unsigned Op,
 
 unsigned ARMMCCodeEmitter::
 getShiftRight8Imm(const MCInst &MI, unsigned Op,
-                  SmallVectorImpl<MCFixup> &Fixups) const {
+                  SmallVectorImpl<MCFixup> &Fixups,
+                  const MCSubtargetInfo &STI) const {
   return 8 - MI.getOperand(Op).getImm();
 }
 
 unsigned ARMMCCodeEmitter::
 getShiftRight16Imm(const MCInst &MI, unsigned Op,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   return 16 - MI.getOperand(Op).getImm();
 }
 
 unsigned ARMMCCodeEmitter::
 getShiftRight32Imm(const MCInst &MI, unsigned Op,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   return 32 - MI.getOperand(Op).getImm();
 }
 
 unsigned ARMMCCodeEmitter::
 getShiftRight64Imm(const MCInst &MI, unsigned Op,
-                   SmallVectorImpl<MCFixup> &Fixups) const {
+                   SmallVectorImpl<MCFixup> &Fixups,
+                   const MCSubtargetInfo &STI) const {
   return 64 - MI.getOperand(Op).getImm();
 }
 
 void ARMMCCodeEmitter::
 EncodeInstruction(const MCInst &MI, raw_ostream &OS,
-                  SmallVectorImpl<MCFixup> &Fixups) const {
+                  SmallVectorImpl<MCFixup> &Fixups,
+                  const MCSubtargetInfo &STI) const {
   // Pseudo instructions don't get encoded.
   const MCInstrDesc &Desc = MCII.get(MI.getOpcode());
   uint64_t TSFlags = Desc.TSFlags;
@@ -1537,10 +1661,10 @@ EncodeInstruction(const MCInst &MI, raw_ostream &OS,
   else
     llvm_unreachable("Unexpected instruction size!");
 
-  uint32_t Binary = getBinaryCodeForInstr(MI, Fixups);
+  uint32_t Binary = getBinaryCodeForInstr(MI, Fixups, STI);
   // Thumb 32-bit wide instructions need to emit the high order halfword
   // first.
-  if (isThumb() && Size == 4) {
+  if (isThumb(STI) && Size == 4) {
     EmitConstant(Binary >> 16, 2, OS);
     EmitConstant(Binary & 0xffff, 2, OS);
   } else
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.cpp
index fc8505b052bd..e545e3c2f301 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.cpp
@@ -7,12 +7,13 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "armmcexpr"
 #include "ARMMCExpr.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCContext.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "armmcexpr"
+
 const ARMMCExpr*
 ARMMCExpr::Create(VariantKind Kind, const MCExpr *Expr,
                        MCContext &Ctx) {
@@ -40,33 +41,6 @@ ARMMCExpr::EvaluateAsRelocatableImpl(MCValue &Res,
   return false;
 }
 
-// FIXME: This basically copies MCObjectStreamer::AddValueSymbols. Perhaps
-// that method should be made public?
-static void AddValueSymbols_(const MCExpr *Value, MCAssembler *Asm) {
-  switch (Value->getKind()) {
-  case MCExpr::Target:
-    llvm_unreachable("Can't handle nested target expr!");
-
-  case MCExpr::Constant:
-    break;
-
-  case MCExpr::Binary: {
-    const MCBinaryExpr *BE = cast<MCBinaryExpr>(Value);
-    AddValueSymbols_(BE->getLHS(), Asm);
-    AddValueSymbols_(BE->getRHS(), Asm);
-    break;
-  }
-
-  case MCExpr::SymbolRef:
-    Asm->getOrCreateSymbolData(cast<MCSymbolRefExpr>(Value)->getSymbol());
-    break;
-
-  case MCExpr::Unary:
-    AddValueSymbols_(cast<MCUnaryExpr>(Value)->getSubExpr(), Asm);
-    break;
-  }
-}
-
-void ARMMCExpr::AddValueSymbols(MCAssembler *Asm) const {
-  AddValueSymbols_(getSubExpr(), Asm);
+void ARMMCExpr::visitUsedExpr(MCStreamer &Streamer) const {
+  Streamer.visitUsedExpr(*getSubExpr());
 }
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.h
index cd4067a52955..c5c0b10f8ad9 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.h
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCExpr.h
@@ -56,16 +56,16 @@ public:
 
   /// @}
 
-  void PrintImpl(raw_ostream &OS) const;
+  void PrintImpl(raw_ostream &OS) const override;
   bool EvaluateAsRelocatableImpl(MCValue &Res,
-                                 const MCAsmLayout *Layout) const;
-  void AddValueSymbols(MCAssembler *) const;
-  const MCSection *FindAssociatedSection() const {
+                                 const MCAsmLayout *Layout) const override;
+  void visitUsedExpr(MCStreamer &Streamer) const override;
+  const MCSection *FindAssociatedSection() const override {
     return getSubExpr()->FindAssociatedSection();
   }
 
   // There are no TLS ARMMCExprs at the moment.
-  void fixELFSymbolsInTLSFixups(MCAssembler &Asm) const {}
+  void fixELFSymbolsInTLSFixups(MCAssembler &Asm) const override {}
 
   static bool classof(const MCExpr *E) {
     return E->getKind() == MCExpr::Target;
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp
index a99de0e78230..6a3ec8fcc486 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp
@@ -21,6 +21,7 @@
 #include "llvm/MC/MCInstrAnalysis.h"
 #include "llvm/MC/MCInstrInfo.h"
 #include "llvm/MC/MCRegisterInfo.h"
+#include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSubtargetInfo.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/TargetRegistry.h"
@@ -83,89 +84,87 @@ static bool getITDeprecationInfo(MCInst &MI, MCSubtargetInfo &STI,
 std::string ARM_MC::ParseARMTriple(StringRef TT, StringRef CPU) {
   Triple triple(TT);
 
-  // Set the boolean corresponding to the current target triple, or the default
-  // if one cannot be determined, to true.
-  unsigned Len = TT.size();
-  unsigned Idx = 0;
-
-  // FIXME: Enhance Triple helper class to extract ARM version.
-  bool isThumb = false;
-  if (Len >= 5 && TT.substr(0, 4) == "armv")
-    Idx = 4;
-  else if (Len >= 6 && TT.substr(0, 5) == "thumb") {
-    isThumb = true;
-    if (Len >= 7 && TT[5] == 'v')
-      Idx = 6;
-  }
+  bool isThumb = triple.getArch() == Triple::thumb ||
+                 triple.getArch() == Triple::thumbeb;
 
   bool NoCPU = CPU == "generic" || CPU.empty();
   std::string ARMArchFeature;
-  if (Idx) {
-    unsigned SubVer = TT[Idx];
-    if (SubVer == '8') {
-      if (NoCPU)
-        // v8a: FeatureDB, FeatureFPARMv8, FeatureNEON, FeatureDSPThumb2, FeatureMP,
-        //      FeatureHWDiv, FeatureHWDivARM, FeatureTrustZone, FeatureT2XtPk, FeatureCrypto, FeatureCRC
-        ARMArchFeature = "+v8,+db,+fp-armv8,+neon,+t2dsp,+mp,+hwdiv,+hwdiv-arm,+trustzone,+t2xtpk,+crypto,+crc";
-      else
-        // Use CPU to figure out the exact features
-        ARMArchFeature = "+v8";
-    } else if (SubVer == '7') {
-      if (Len >= Idx+2 && TT[Idx+1] == 'm') {
-        isThumb = true;
-        if (NoCPU)
-          // v7m: FeatureNoARM, FeatureDB, FeatureHWDiv, FeatureMClass
-          ARMArchFeature = "+v7,+noarm,+db,+hwdiv,+mclass";
-        else
-          // Use CPU to figure out the exact features.
-          ARMArchFeature = "+v7";
-      } else if (Len >= Idx+3 && TT[Idx+1] == 'e'&& TT[Idx+2] == 'm') {
-        if (NoCPU)
-          // v7em: FeatureNoARM, FeatureDB, FeatureHWDiv, FeatureDSPThumb2,
-          //       FeatureT2XtPk, FeatureMClass
-          ARMArchFeature = "+v7,+noarm,+db,+hwdiv,+t2dsp,t2xtpk,+mclass";
-        else
-          // Use CPU to figure out the exact features.
-          ARMArchFeature = "+v7";
-      } else if (Len >= Idx+2 && TT[Idx+1] == 's') {
-        if (NoCPU)
-          // v7s: FeatureNEON, FeatureDB, FeatureDSPThumb2, FeatureT2XtPk
-          //      Swift
-          ARMArchFeature = "+v7,+swift,+neon,+db,+t2dsp,+t2xtpk";
-        else
-          // Use CPU to figure out the exact features.
-          ARMArchFeature = "+v7";
-      } else {
-        // v7 CPUs have lots of different feature sets. If no CPU is specified,
-        // then assume v7a (e.g. cortex-a8) feature set. Otherwise, return
-        // the "minimum" feature set and use CPU string to figure out the exact
-        // features.
-        if (NoCPU)
-          // v7a: FeatureNEON, FeatureDB, FeatureDSPThumb2, FeatureT2XtPk
-          ARMArchFeature = "+v7,+neon,+db,+t2dsp,+t2xtpk";
-        else
-          // Use CPU to figure out the exact features.
-          ARMArchFeature = "+v7";
-      }
-    } else if (SubVer == '6') {
-      if (Len >= Idx+3 && TT[Idx+1] == 't' && TT[Idx+2] == '2')
-        ARMArchFeature = "+v6t2";
-      else if (Len >= Idx+2 && TT[Idx+1] == 'm') {
-        isThumb = true;
-        if (NoCPU)
-          // v6m: FeatureNoARM, FeatureMClass
-          ARMArchFeature = "+v6m,+noarm,+mclass";
-        else
-          ARMArchFeature = "+v6";
-      } else
-        ARMArchFeature = "+v6";
-    } else if (SubVer == '5') {
-      if (Len >= Idx+3 && TT[Idx+1] == 't' && TT[Idx+2] == 'e')
-        ARMArchFeature = "+v5te";
-      else
-        ARMArchFeature = "+v5t";
-    } else if (SubVer == '4' && Len >= Idx+2 && TT[Idx+1] == 't')
-      ARMArchFeature = "+v4t";
+  switch (triple.getSubArch()) {
+  case Triple::ARMSubArch_v8:
+    if (NoCPU)
+      // v8a: FeatureDB, FeatureFPARMv8, FeatureNEON, FeatureDSPThumb2,
+      //      FeatureMP, FeatureHWDiv, FeatureHWDivARM, FeatureTrustZone,
+      //      FeatureT2XtPk, FeatureCrypto, FeatureCRC
+      ARMArchFeature = "+v8,+db,+fp-armv8,+neon,+t2dsp,+mp,+hwdiv,+hwdiv-arm,"
+                       "+trustzone,+t2xtpk,+crypto,+crc";
+    else
+      // Use CPU to figure out the exact features
+      ARMArchFeature = "+v8";
+    break;
+  case Triple::ARMSubArch_v7m:
+    isThumb = true;
+    if (NoCPU)
+      // v7m: FeatureNoARM, FeatureDB, FeatureHWDiv, FeatureMClass
+      ARMArchFeature = "+v7,+noarm,+db,+hwdiv,+mclass";
+    else
+      // Use CPU to figure out the exact features.
+      ARMArchFeature = "+v7";
+    break;
+  case Triple::ARMSubArch_v7em:
+    if (NoCPU)
+      // v7em: FeatureNoARM, FeatureDB, FeatureHWDiv, FeatureDSPThumb2,
+      //       FeatureT2XtPk, FeatureMClass
+      ARMArchFeature = "+v7,+noarm,+db,+hwdiv,+t2dsp,t2xtpk,+mclass";
+    else
+      // Use CPU to figure out the exact features.
+      ARMArchFeature = "+v7";
+    break;
+  case Triple::ARMSubArch_v7s:
+    if (NoCPU)
+      // v7s: FeatureNEON, FeatureDB, FeatureDSPThumb2, FeatureHasRAS
+      //      Swift
+      ARMArchFeature = "+v7,+swift,+neon,+db,+t2dsp,+ras";
+    else
+      // Use CPU to figure out the exact features.
+      ARMArchFeature = "+v7";
+    break;
+  case Triple::ARMSubArch_v7:
+    // v7 CPUs have lots of different feature sets. If no CPU is specified,
+    // then assume v7a (e.g. cortex-a8) feature set. Otherwise, return
+    // the "minimum" feature set and use CPU string to figure out the exact
+    // features.
+    if (NoCPU)
+      // v7a: FeatureNEON, FeatureDB, FeatureDSPThumb2, FeatureT2XtPk
+      ARMArchFeature = "+v7,+neon,+db,+t2dsp,+t2xtpk";
+    else
+      // Use CPU to figure out the exact features.
+      ARMArchFeature = "+v7";
+    break;
+  case Triple::ARMSubArch_v6t2:
+    ARMArchFeature = "+v6t2";
+    break;
+  case Triple::ARMSubArch_v6m:
+    isThumb = true;
+    if (NoCPU)
+      // v6m: FeatureNoARM, FeatureMClass
+      ARMArchFeature = "+v6m,+noarm,+mclass";
+    else
+      ARMArchFeature = "+v6";
+    break;
+  case Triple::ARMSubArch_v6:
+    ARMArchFeature = "+v6";
+    break;
+  case Triple::ARMSubArch_v5te:
+    ARMArchFeature = "+v5te";
+    break;
+  case Triple::ARMSubArch_v5:
+    ARMArchFeature = "+v5t";
+    break;
+  case Triple::ARMSubArch_v4t:
+    ARMArchFeature = "+v4t";
+    break;
+  case Triple::NoSubArch:
+    break;
   }
 
   if (isThumb) {
@@ -215,10 +214,37 @@ static MCRegisterInfo *createARMMCRegisterInfo(StringRef Triple) {
 static MCAsmInfo *createARMMCAsmInfo(const MCRegisterInfo &MRI, StringRef TT) {
   Triple TheTriple(TT);
 
-  if (TheTriple.isOSDarwin())
-    return new ARMMCAsmInfoDarwin();
+  MCAsmInfo *MAI;
+  switch (TheTriple.getOS()) {
+  case llvm::Triple::Darwin:
+  case llvm::Triple::IOS:
+  case llvm::Triple::MacOSX:
+    MAI = new ARMMCAsmInfoDarwin(TT);
+    break;
+  case llvm::Triple::Win32:
+    switch (TheTriple.getEnvironment()) {
+    case llvm::Triple::Itanium:
+      MAI = new ARMCOFFMCAsmInfoGNU();
+      break;
+    case llvm::Triple::MSVC:
+      MAI = new ARMCOFFMCAsmInfoMicrosoft();
+      break;
+    default:
+      llvm_unreachable("invalid environment");
+    }
+    break;
+  default:
+    if (TheTriple.isOSBinFormatMachO())
+      MAI = new ARMMCAsmInfoDarwin(TT);
+    else
+      MAI = new ARMELFMCAsmInfo(TT);
+    break;
+  }
+
+  unsigned Reg = MRI.getDwarfRegNum(ARM::SP, true);
+  MAI->addInitialFrameState(MCCFIInstruction::createDefCfa(nullptr, Reg, 0));
 
-  return new ARMELFMCAsmInfo();
+  return MAI;
 }
 
 static MCCodeGenInfo *createARMMCCodeGenInfo(StringRef TT, Reloc::Model RM,
@@ -239,19 +265,25 @@ static MCStreamer *createMCStreamer(const Target &T, StringRef TT,
                                     MCContext &Ctx, MCAsmBackend &MAB,
                                     raw_ostream &OS,
                                     MCCodeEmitter *Emitter,
+                                    const MCSubtargetInfo &STI,
                                     bool RelaxAll,
                                     bool NoExecStack) {
   Triple TheTriple(TT);
 
-  if (TheTriple.isOSDarwin())
-    return createMachOStreamer(Ctx, MAB, OS, Emitter, false);
-
-  if (TheTriple.isOSWindows()) {
-    llvm_unreachable("ARM does not support Windows COFF format");
+  switch (TheTriple.getObjectFormat()) {
+  default: llvm_unreachable("unsupported object format");
+  case Triple::MachO: {
+    MCStreamer *S = createMachOStreamer(Ctx, MAB, OS, Emitter, false);
+    new ARMTargetStreamer(*S);
+    return S;
+  }
+  case Triple::COFF:
+    assert(TheTriple.isOSWindows() && "non-Windows ARM COFF is not supported");
+    return createARMWinCOFFStreamer(Ctx, MAB, *Emitter, OS);
+  case Triple::ELF:
+    return createARMELFStreamer(Ctx, MAB, OS, Emitter, false, NoExecStack,
+                                TheTriple.getArch() == Triple::thumb);
   }
-
-  return createARMELFStreamer(Ctx, MAB, OS, Emitter, false, NoExecStack,
-                              TheTriple.getArch() == Triple::thumb);
 }
 
 static MCInstPrinter *createARMMCInstPrinter(const Target &T,
@@ -262,13 +294,13 @@ static MCInstPrinter *createARMMCInstPrinter(const Target &T,
                                              const MCSubtargetInfo &STI) {
   if (SyntaxVariant == 0)
     return new ARMInstPrinter(MAI, MII, MRI, STI);
-  return 0;
+  return nullptr;
 }
 
 static MCRelocationInfo *createARMMCRelocationInfo(StringRef TT,
                                                    MCContext &Ctx) {
   Triple TheTriple(TT);
-  if (TheTriple.isEnvironmentMachO())
+  if (TheTriple.isOSBinFormatMachO())
     return createARMMachORelocationInfo(Ctx);
   // Default to the stock relocation info.
   return llvm::createMCRelocationInfo(TT, Ctx);
@@ -280,14 +312,14 @@ class ARMMCInstrAnalysis : public MCInstrAnalysis {
 public:
   ARMMCInstrAnalysis(const MCInstrInfo *Info) : MCInstrAnalysis(Info) {}
 
-  virtual bool isUnconditionalBranch(const MCInst &Inst) const {
+  bool isUnconditionalBranch(const MCInst &Inst) const override {
     // BCCs with the "always" predicate are unconditional branches.
     if (Inst.getOpcode() == ARM::Bcc && Inst.getOperand(1).getImm()==ARMCC::AL)
       return true;
     return MCInstrAnalysis::isUnconditionalBranch(Inst);
   }
 
-  virtual bool isConditionalBranch(const MCInst &Inst) const {
+  bool isConditionalBranch(const MCInst &Inst) const override {
     // BCCs with the "always" predicate are unconditional branches.
     if (Inst.getOpcode() == ARM::Bcc && Inst.getOperand(1).getImm()==ARMCC::AL)
       return false;
@@ -295,7 +327,7 @@ public:
   }
 
   bool evaluateBranch(const MCInst &Inst, uint64_t Addr,
-                      uint64_t Size, uint64_t &Target) const {
+                      uint64_t Size, uint64_t &Target) const override {
     // We only handle PCRel branches for now.
     if (Info->get(Inst.getOpcode()).OpInfo[0].OperandType!=MCOI::OPERAND_PCREL)
       return false;
@@ -316,56 +348,100 @@ static MCInstrAnalysis *createARMMCInstrAnalysis(const MCInstrInfo *Info) {
 // Force static initialization.
 extern "C" void LLVMInitializeARMTargetMC() {
   // Register the MC asm info.
-  RegisterMCAsmInfoFn A(TheARMTarget, createARMMCAsmInfo);
-  RegisterMCAsmInfoFn B(TheThumbTarget, createARMMCAsmInfo);
+  RegisterMCAsmInfoFn X(TheARMLETarget, createARMMCAsmInfo);
+  RegisterMCAsmInfoFn Y(TheARMBETarget, createARMMCAsmInfo);
+  RegisterMCAsmInfoFn A(TheThumbLETarget, createARMMCAsmInfo);
+  RegisterMCAsmInfoFn B(TheThumbBETarget, createARMMCAsmInfo);
 
   // Register the MC codegen info.
-  TargetRegistry::RegisterMCCodeGenInfo(TheARMTarget, createARMMCCodeGenInfo);
-  TargetRegistry::RegisterMCCodeGenInfo(TheThumbTarget, createARMMCCodeGenInfo);
+  TargetRegistry::RegisterMCCodeGenInfo(TheARMLETarget, createARMMCCodeGenInfo);
+  TargetRegistry::RegisterMCCodeGenInfo(TheARMBETarget, createARMMCCodeGenInfo);
+  TargetRegistry::RegisterMCCodeGenInfo(TheThumbLETarget, createARMMCCodeGenInfo);
+  TargetRegistry::RegisterMCCodeGenInfo(TheThumbBETarget, createARMMCCodeGenInfo);
 
   // Register the MC instruction info.
-  TargetRegistry::RegisterMCInstrInfo(TheARMTarget, createARMMCInstrInfo);
-  TargetRegistry::RegisterMCInstrInfo(TheThumbTarget, createARMMCInstrInfo);
+  TargetRegistry::RegisterMCInstrInfo(TheARMLETarget, createARMMCInstrInfo);
+  TargetRegistry::RegisterMCInstrInfo(TheARMBETarget, createARMMCInstrInfo);
+  TargetRegistry::RegisterMCInstrInfo(TheThumbLETarget, createARMMCInstrInfo);
+  TargetRegistry::RegisterMCInstrInfo(TheThumbBETarget, createARMMCInstrInfo);
 
   // Register the MC register info.
-  TargetRegistry::RegisterMCRegInfo(TheARMTarget, createARMMCRegisterInfo);
-  TargetRegistry::RegisterMCRegInfo(TheThumbTarget, createARMMCRegisterInfo);
+  TargetRegistry::RegisterMCRegInfo(TheARMLETarget, createARMMCRegisterInfo);
+  TargetRegistry::RegisterMCRegInfo(TheARMBETarget, createARMMCRegisterInfo);
+  TargetRegistry::RegisterMCRegInfo(TheThumbLETarget, createARMMCRegisterInfo);
+  TargetRegistry::RegisterMCRegInfo(TheThumbBETarget, createARMMCRegisterInfo);
 
   // Register the MC subtarget info.
-  TargetRegistry::RegisterMCSubtargetInfo(TheARMTarget,
+  TargetRegistry::RegisterMCSubtargetInfo(TheARMLETarget,
+                                          ARM_MC::createARMMCSubtargetInfo);
+  TargetRegistry::RegisterMCSubtargetInfo(TheARMBETarget,
                                           ARM_MC::createARMMCSubtargetInfo);
-  TargetRegistry::RegisterMCSubtargetInfo(TheThumbTarget,
+  TargetRegistry::RegisterMCSubtargetInfo(TheThumbLETarget,
+                                          ARM_MC::createARMMCSubtargetInfo);
+  TargetRegistry::RegisterMCSubtargetInfo(TheThumbBETarget,
                                           ARM_MC::createARMMCSubtargetInfo);
 
   // Register the MC instruction analyzer.
-  TargetRegistry::RegisterMCInstrAnalysis(TheARMTarget,
+  TargetRegistry::RegisterMCInstrAnalysis(TheARMLETarget,
+                                          createARMMCInstrAnalysis);
+  TargetRegistry::RegisterMCInstrAnalysis(TheARMBETarget,
+                                          createARMMCInstrAnalysis);
+  TargetRegistry::RegisterMCInstrAnalysis(TheThumbLETarget,
                                           createARMMCInstrAnalysis);
-  TargetRegistry::RegisterMCInstrAnalysis(TheThumbTarget,
+  TargetRegistry::RegisterMCInstrAnalysis(TheThumbBETarget,
                                           createARMMCInstrAnalysis);
 
   // Register the MC Code Emitter
-  TargetRegistry::RegisterMCCodeEmitter(TheARMTarget, createARMMCCodeEmitter);
-  TargetRegistry::RegisterMCCodeEmitter(TheThumbTarget, createARMMCCodeEmitter);
+  TargetRegistry::RegisterMCCodeEmitter(TheARMLETarget,
+                                        createARMLEMCCodeEmitter);
+  TargetRegistry::RegisterMCCodeEmitter(TheARMBETarget,
+                                        createARMBEMCCodeEmitter);
+  TargetRegistry::RegisterMCCodeEmitter(TheThumbLETarget,
+                                        createARMLEMCCodeEmitter);
+  TargetRegistry::RegisterMCCodeEmitter(TheThumbBETarget,
+                                        createARMBEMCCodeEmitter);
 
   // Register the asm backend.
-  TargetRegistry::RegisterMCAsmBackend(TheARMTarget, createARMAsmBackend);
-  TargetRegistry::RegisterMCAsmBackend(TheThumbTarget, createARMAsmBackend);
+  TargetRegistry::RegisterMCAsmBackend(TheARMLETarget, createARMLEAsmBackend);
+  TargetRegistry::RegisterMCAsmBackend(TheARMBETarget, createARMBEAsmBackend);
+  TargetRegistry::RegisterMCAsmBackend(TheThumbLETarget,
+                                       createThumbLEAsmBackend);
+  TargetRegistry::RegisterMCAsmBackend(TheThumbBETarget,
+                                       createThumbBEAsmBackend);
 
   // Register the object streamer.
-  TargetRegistry::RegisterMCObjectStreamer(TheARMTarget, createMCStreamer);
-  TargetRegistry::RegisterMCObjectStreamer(TheThumbTarget, createMCStreamer);
+  TargetRegistry::RegisterMCObjectStreamer(TheARMLETarget, createMCStreamer);
+  TargetRegistry::RegisterMCObjectStreamer(TheARMBETarget, createMCStreamer);
+  TargetRegistry::RegisterMCObjectStreamer(TheThumbLETarget, createMCStreamer);
+  TargetRegistry::RegisterMCObjectStreamer(TheThumbBETarget, createMCStreamer);
 
   // Register the asm streamer.
-  TargetRegistry::RegisterAsmStreamer(TheARMTarget, createMCAsmStreamer);
-  TargetRegistry::RegisterAsmStreamer(TheThumbTarget, createMCAsmStreamer);
+  TargetRegistry::RegisterAsmStreamer(TheARMLETarget, createMCAsmStreamer);
+  TargetRegistry::RegisterAsmStreamer(TheARMBETarget, createMCAsmStreamer);
+  TargetRegistry::RegisterAsmStreamer(TheThumbLETarget, createMCAsmStreamer);
+  TargetRegistry::RegisterAsmStreamer(TheThumbBETarget, createMCAsmStreamer);
+
+  // Register the null streamer.
+  TargetRegistry::RegisterNullStreamer(TheARMLETarget, createARMNullStreamer);
+  TargetRegistry::RegisterNullStreamer(TheARMBETarget, createARMNullStreamer);
+  TargetRegistry::RegisterNullStreamer(TheThumbLETarget, createARMNullStreamer);
+  TargetRegistry::RegisterNullStreamer(TheThumbBETarget, createARMNullStreamer);
 
   // Register the MCInstPrinter.
-  TargetRegistry::RegisterMCInstPrinter(TheARMTarget, createARMMCInstPrinter);
-  TargetRegistry::RegisterMCInstPrinter(TheThumbTarget, createARMMCInstPrinter);
+  TargetRegistry::RegisterMCInstPrinter(TheARMLETarget, createARMMCInstPrinter);
+  TargetRegistry::RegisterMCInstPrinter(TheARMBETarget, createARMMCInstPrinter);
+  TargetRegistry::RegisterMCInstPrinter(TheThumbLETarget,
+                                        createARMMCInstPrinter);
+  TargetRegistry::RegisterMCInstPrinter(TheThumbBETarget,
+                                        createARMMCInstPrinter);
 
   // Register the MC relocation info.
-  TargetRegistry::RegisterMCRelocationInfo(TheARMTarget,
+  TargetRegistry::RegisterMCRelocationInfo(TheARMLETarget,
+                                           createARMMCRelocationInfo);
+  TargetRegistry::RegisterMCRelocationInfo(TheARMBETarget,
+                                           createARMMCRelocationInfo);
+  TargetRegistry::RegisterMCRelocationInfo(TheThumbLETarget,
                                            createARMMCRelocationInfo);
-  TargetRegistry::RegisterMCRelocationInfo(TheThumbTarget,
+  TargetRegistry::RegisterMCRelocationInfo(TheThumbBETarget,
                                            createARMMCRelocationInfo);
 }
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h
index 959be8b55c3a..5326e564f363 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h
@@ -33,7 +33,8 @@ class StringRef;
 class Target;
 class raw_ostream;
 
-extern Target TheARMTarget, TheThumbTarget;
+extern Target TheARMLETarget, TheThumbLETarget;
+extern Target TheARMBETarget, TheThumbBETarget;
 
 namespace ARM_MC {
   std::string ParseARMTriple(StringRef TT, StringRef CPU);
@@ -46,22 +47,47 @@ namespace ARM_MC {
 }
 
 MCStreamer *createMCAsmStreamer(MCContext &Ctx, formatted_raw_ostream &OS,
-                                bool isVerboseAsm, bool useLoc, bool useCFI,
-                                bool useDwarfDirectory,
+                                bool isVerboseAsm, bool useDwarfDirectory,
                                 MCInstPrinter *InstPrint, MCCodeEmitter *CE,
                                 MCAsmBackend *TAB, bool ShowInst);
 
-MCCodeEmitter *createARMMCCodeEmitter(const MCInstrInfo &MCII,
-                                      const MCRegisterInfo &MRI,
-                                      const MCSubtargetInfo &STI,
-                                      MCContext &Ctx);
+MCStreamer *createARMNullStreamer(MCContext &Ctx);
+
+MCCodeEmitter *createARMLEMCCodeEmitter(const MCInstrInfo &MCII,
+                                        const MCRegisterInfo &MRI,
+                                        const MCSubtargetInfo &STI,
+                                        MCContext &Ctx);
+
+MCCodeEmitter *createARMBEMCCodeEmitter(const MCInstrInfo &MCII,
+                                        const MCRegisterInfo &MRI,
+                                        const MCSubtargetInfo &STI,
+                                        MCContext &Ctx);
 
 MCAsmBackend *createARMAsmBackend(const Target &T, const MCRegisterInfo &MRI,
+                                  StringRef TT, StringRef CPU,
+                                  bool IsLittleEndian);
+
+MCAsmBackend *createARMLEAsmBackend(const Target &T, const MCRegisterInfo &MRI,
+                                  StringRef TT, StringRef CPU);
+
+MCAsmBackend *createARMBEAsmBackend(const Target &T, const MCRegisterInfo &MRI,
                                   StringRef TT, StringRef CPU);
 
+MCAsmBackend *createThumbLEAsmBackend(const Target &T, const MCRegisterInfo &MRI,
+                                      StringRef TT, StringRef CPU);
+
+MCAsmBackend *createThumbBEAsmBackend(const Target &T, const MCRegisterInfo &MRI,
+                                      StringRef TT, StringRef CPU);
+
+/// createARMWinCOFFStreamer - Construct a PE/COFF machine code streamer which
+/// will generate a PE/COFF object file.
+MCStreamer *createARMWinCOFFStreamer(MCContext &Context, MCAsmBackend &MAB,
+                                     MCCodeEmitter &Emitter, raw_ostream &OS);
+
 /// createARMELFObjectWriter - Construct an ELF Mach-O object writer.
 MCObjectWriter *createARMELFObjectWriter(raw_ostream &OS,
-                                         uint8_t OSABI);
+                                         uint8_t OSABI,
+                                         bool IsLittleEndian);
 
 /// createARMMachObjectWriter - Construct an ARM Mach-O object writer.
 MCObjectWriter *createARMMachObjectWriter(raw_ostream &OS,
@@ -69,6 +95,8 @@ MCObjectWriter *createARMMachObjectWriter(raw_ostream &OS,
                                           uint32_t CPUType,
                                           uint32_t CPUSubtype);
 
+/// createARMWinCOFFObjectWriter - Construct an ARM PE/COFF object writer.
+MCObjectWriter *createARMWinCOFFObjectWriter(raw_ostream &OS, bool Is64Bit);
 
 /// createARMMachORelocationInfo - Construct ARM Mach-O relocation info.
 MCRelocationInfo *createARMMachORelocationInfo(MCContext &Ctx);
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachORelocationInfo.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachORelocationInfo.cpp
index 807c9483bc38..d4b00e6e4fb5 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachORelocationInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachORelocationInfo.cpp
@@ -9,10 +9,10 @@
 
 #include "MCTargetDesc/ARMMCTargetDesc.h"
 #include "ARMMCExpr.h"
+#include "llvm-c/Disassembler.h"
 #include "llvm/MC/MCContext.h"
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCRelocationInfo.h"
-#include "llvm-c/Disassembler.h"
 
 using namespace llvm;
 using namespace object;
@@ -23,7 +23,7 @@ public:
   ARMMachORelocationInfo(MCContext &Ctx) : MCRelocationInfo(Ctx) {}
 
   const MCExpr *createExprForCAPIVariantKind(const MCExpr *SubExpr,
-                                             unsigned VariantKind) {
+                                             unsigned VariantKind) override {
     switch(VariantKind) {
     case LLVMDisassembler_VariantKind_ARM_HI16:
       return ARMMCExpr::CreateUpper16(SubExpr, Ctx);
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachObjectWriter.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachObjectWriter.cpp
index 1f681bac2242..186776a1944f 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachObjectWriter.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMMachObjectWriter.cpp
@@ -32,6 +32,7 @@ class ARMMachObjectWriter : public MCMachObjectTargetWriter {
                                     const MCFragment *Fragment,
                                     const MCFixup &Fixup,
                                     MCValue Target,
+                                    unsigned Type,
                                     unsigned Log2Size,
                                     uint64_t &FixedValue);
   void RecordARMScatteredHalfRelocation(MachObjectWriter *Writer,
@@ -56,7 +57,7 @@ public:
   void RecordRelocation(MachObjectWriter *Writer,
                         const MCAssembler &Asm, const MCAsmLayout &Layout,
                         const MCFragment *Fragment, const MCFixup &Fixup,
-                        MCValue Target, uint64_t &FixedValue);
+                        MCValue Target, uint64_t &FixedValue) override;
 };
 }
 
@@ -82,10 +83,14 @@ static bool getARMFixupKindMachOInfo(unsigned Kind, unsigned &RelocType,
     Log2Size = llvm::Log2_32(8);
     return true;
 
-    // Handle 24-bit branch kinds.
+    // These fixups are expected to always be resolvable at assembly time and
+    // have no relocations supported.
   case ARM::fixup_arm_ldst_pcrel_12:
   case ARM::fixup_arm_pcrel_10:
   case ARM::fixup_arm_adr_pcrel_12:
+    return false;
+
+    // Handle 24-bit branch kinds.
   case ARM::fixup_arm_condbranch:
   case ARM::fixup_arm_uncondbranch:
   case ARM::fixup_arm_uncondbl:
@@ -119,23 +124,19 @@ static bool getARMFixupKindMachOInfo(unsigned Kind, unsigned &RelocType,
   //      0 - arm instructions
   //      1 - thumb instructions
   case ARM::fixup_arm_movt_hi16:
-  case ARM::fixup_arm_movt_hi16_pcrel:
     RelocType = unsigned(MachO::ARM_RELOC_HALF);
     Log2Size = 1;
     return true;
   case ARM::fixup_t2_movt_hi16:
-  case ARM::fixup_t2_movt_hi16_pcrel:
     RelocType = unsigned(MachO::ARM_RELOC_HALF);
     Log2Size = 3;
     return true;
 
   case ARM::fixup_arm_movw_lo16:
-  case ARM::fixup_arm_movw_lo16_pcrel:
     RelocType = unsigned(MachO::ARM_RELOC_HALF);
     Log2Size = 0;
     return true;
   case ARM::fixup_t2_movw_lo16:
-  case ARM::fixup_t2_movw_lo16_pcrel:
     RelocType = unsigned(MachO::ARM_RELOC_HALF);
     Log2Size = 2;
     return true;
@@ -156,7 +157,7 @@ RecordARMScatteredHalfRelocation(MachObjectWriter *Writer,
 
   // See <reloc.h>.
   const MCSymbol *A = &Target.getSymA()->getSymbol();
-  MCSymbolData *A_SD = &Asm.getSymbolData(*A);
+  const MCSymbolData *A_SD = &Asm.getSymbolData(*A);
 
   if (!A_SD->getFragment())
     Asm.getContext().FatalError(Fixup.getLoc(),
@@ -170,7 +171,7 @@ RecordARMScatteredHalfRelocation(MachObjectWriter *Writer,
   FixedValue += SecAddr;
 
   if (const MCSymbolRefExpr *B = Target.getSymB()) {
-    MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
+    const MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
 
     if (!B_SD->getFragment())
       Asm.getContext().FatalError(Fixup.getLoc(),
@@ -202,22 +203,19 @@ RecordARMScatteredHalfRelocation(MachObjectWriter *Writer,
   switch ((unsigned)Fixup.getKind()) {
   default: break;
   case ARM::fixup_arm_movt_hi16:
-  case ARM::fixup_arm_movt_hi16_pcrel:
     MovtBit = 1;
     // The thumb bit shouldn't be set in the 'other-half' bit of the
     // relocation, but it will be set in FixedValue if the base symbol
     // is a thumb function. Clear it out here.
-    if (A_SD->getFlags() & SF_ThumbFunc)
+    if (Asm.isThumbFunc(A))
       FixedValue &= 0xfffffffe;
     break;
   case ARM::fixup_t2_movt_hi16:
-  case ARM::fixup_t2_movt_hi16_pcrel:
-    if (A_SD->getFlags() & SF_ThumbFunc)
+    if (Asm.isThumbFunc(A))
       FixedValue &= 0xfffffffe;
     MovtBit = 1;
     // Fallthrough
   case ARM::fixup_t2_movw_lo16:
-  case ARM::fixup_t2_movw_lo16_pcrel:
     ThumbBit = 1;
     break;
   }
@@ -254,15 +252,15 @@ void ARMMachObjectWriter::RecordARMScatteredRelocation(MachObjectWriter *Writer,
                                                     const MCFragment *Fragment,
                                                     const MCFixup &Fixup,
                                                     MCValue Target,
+                                                    unsigned Type,
                                                     unsigned Log2Size,
                                                     uint64_t &FixedValue) {
   uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
   unsigned IsPCRel = Writer->isFixupKindPCRel(Asm, Fixup.getKind());
-  unsigned Type = MachO::ARM_RELOC_VANILLA;
 
   // See <reloc.h>.
   const MCSymbol *A = &Target.getSymA()->getSymbol();
-  MCSymbolData *A_SD = &Asm.getSymbolData(*A);
+  const MCSymbolData *A_SD = &Asm.getSymbolData(*A);
 
   if (!A_SD->getFragment())
     Asm.getContext().FatalError(Fixup.getLoc(),
@@ -275,7 +273,8 @@ void ARMMachObjectWriter::RecordARMScatteredRelocation(MachObjectWriter *Writer,
   uint32_t Value2 = 0;
 
   if (const MCSymbolRefExpr *B = Target.getSymB()) {
-    MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
+    assert(Type == MachO::ARM_RELOC_VANILLA && "invalid reloc for 2 symbols");
+    const MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
 
     if (!B_SD->getFragment())
       Asm.getContext().FatalError(Fixup.getLoc(),
@@ -377,11 +376,12 @@ void ARMMachObjectWriter::RecordRelocation(MachObjectWriter *Writer,
       return RecordARMScatteredHalfRelocation(Writer, Asm, Layout, Fragment,
                                               Fixup, Target, FixedValue);
     return RecordARMScatteredRelocation(Writer, Asm, Layout, Fragment, Fixup,
-                                        Target, Log2Size, FixedValue);
+                                        Target, RelocType, Log2Size,
+                                        FixedValue);
   }
 
   // Get the symbol data, if any.
-  MCSymbolData *SD = 0;
+  const MCSymbolData *SD = nullptr;
   if (Target.getSymA())
     SD = &Asm.getSymbolData(Target.getSymA()->getSymbol());
 
@@ -395,7 +395,8 @@ void ARMMachObjectWriter::RecordRelocation(MachObjectWriter *Writer,
     Offset += 1 << Log2Size;
   if (Offset && SD && !Writer->doesSymbolRequireExternRelocation(SD))
     return RecordARMScatteredRelocation(Writer, Asm, Layout, Fragment, Fixup,
-                                        Target, Log2Size, FixedValue);
+                                        Target, RelocType, Log2Size,
+                                        FixedValue);
 
   // See <reloc.h>.
   uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
@@ -461,15 +462,11 @@ void ARMMachObjectWriter::RecordRelocation(MachObjectWriter *Writer,
     switch ((unsigned)Fixup.getKind()) {
     default: break;
     case ARM::fixup_arm_movw_lo16:
-    case ARM::fixup_arm_movw_lo16_pcrel:
     case ARM::fixup_t2_movw_lo16:
-    case ARM::fixup_t2_movw_lo16_pcrel:
       Value = (FixedValue >> 16) & 0xffff;
       break;
     case ARM::fixup_arm_movt_hi16:
-    case ARM::fixup_arm_movt_hi16_pcrel:
     case ARM::fixup_t2_movt_hi16:
-    case ARM::fixup_t2_movt_hi16_pcrel:
       Value = FixedValue & 0xffff;
       break;
     }
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMTargetStreamer.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMTargetStreamer.cpp
new file mode 100644
index 000000000000..8acd7aff6bca
--- /dev/null
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMTargetStreamer.cpp
@@ -0,0 +1,73 @@
+//===- ARMTargetStreamer.cpp - ARMTargetStreamer class --*- C++ -*---------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file implements the ARMTargetStreamer class.
+//
+//===----------------------------------------------------------------------===//
+#include "llvm/ADT/MapVector.h"
+#include "llvm/MC/ConstantPools.h"
+#include "llvm/MC/MCContext.h"
+#include "llvm/MC/MCExpr.h"
+#include "llvm/MC/MCStreamer.h"
+
+using namespace llvm;
+//
+// ARMTargetStreamer Implemenation
+//
+ARMTargetStreamer::ARMTargetStreamer(MCStreamer &S)
+    : MCTargetStreamer(S), ConstantPools(new AssemblerConstantPools()) {}
+
+ARMTargetStreamer::~ARMTargetStreamer() {}
+
+// The constant pool handling is shared by all ARMTargetStreamer
+// implementations.
+const MCExpr *ARMTargetStreamer::addConstantPoolEntry(const MCExpr *Expr) {
+  return ConstantPools->addEntry(Streamer, Expr, 4);
+}
+
+void ARMTargetStreamer::emitCurrentConstantPool() {
+  ConstantPools->emitForCurrentSection(Streamer);
+}
+
+// finish() - write out any non-empty assembler constant pools.
+void ARMTargetStreamer::finish() { ConstantPools->emitAll(Streamer); }
+
+// The remaining callbacks should be handled separately by each
+// streamer.
+void ARMTargetStreamer::emitFnStart() {}
+void ARMTargetStreamer::emitFnEnd() {}
+void ARMTargetStreamer::emitCantUnwind() {}
+void ARMTargetStreamer::emitPersonality(const MCSymbol *Personality) {}
+void ARMTargetStreamer::emitPersonalityIndex(unsigned Index) {}
+void ARMTargetStreamer::emitHandlerData() {}
+void ARMTargetStreamer::emitSetFP(unsigned FpReg, unsigned SpReg,
+                                  int64_t Offset) {}
+void ARMTargetStreamer::emitMovSP(unsigned Reg, int64_t Offset) {}
+void ARMTargetStreamer::emitPad(int64_t Offset) {}
+void ARMTargetStreamer::emitRegSave(const SmallVectorImpl<unsigned> &RegList,
+                                    bool isVector) {}
+void ARMTargetStreamer::emitUnwindRaw(int64_t StackOffset,
+                                      const SmallVectorImpl<uint8_t> &Opcodes) {
+}
+void ARMTargetStreamer::switchVendor(StringRef Vendor) {}
+void ARMTargetStreamer::emitAttribute(unsigned Attribute, unsigned Value) {}
+void ARMTargetStreamer::emitTextAttribute(unsigned Attribute,
+                                          StringRef String) {}
+void ARMTargetStreamer::emitIntTextAttribute(unsigned Attribute,
+                                             unsigned IntValue,
+                                             StringRef StringValue) {}
+void ARMTargetStreamer::emitArch(unsigned Arch) {}
+void ARMTargetStreamer::emitObjectArch(unsigned Arch) {}
+void ARMTargetStreamer::emitFPU(unsigned FPU) {}
+void ARMTargetStreamer::finishAttributeSection() {}
+void ARMTargetStreamer::emitInst(uint32_t Inst, char Suffix) {}
+void
+ARMTargetStreamer::AnnotateTLSDescriptorSequence(const MCSymbolRefExpr *SRE) {}
+
+void ARMTargetStreamer::emitThumbSet(MCSymbol *Symbol, const MCExpr *Value) {}
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOp.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOp.h
deleted file mode 100644
index fa4add65a8df..000000000000
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOp.h
+++ /dev/null
@@ -1,125 +0,0 @@
-//===-- ARMUnwindOp.h - ARM Unwind Opcodes ----------------------*- C++ -*-===//
-//
-//                     The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-//
-// This file defines the constants for the ARM unwind opcodes and exception
-// handling table entry kinds.
-//
-//===----------------------------------------------------------------------===//
-
-#ifndef ARM_UNWIND_OP_H
-#define ARM_UNWIND_OP_H
-
-namespace llvm {
-
-  /// ARM exception handling table entry kinds
-  enum ARMEHTEntryKind {
-    EHT_GENERIC = 0x00,
-    EHT_COMPACT = 0x80
-  };
-
-  enum {
-    /// Special entry for the function never unwind
-    EXIDX_CANTUNWIND = 0x1
-  };
-
-  /// ARM-defined frame unwinding opcodes
-  enum ARMUnwindOpcodes {
-    // Format: 00xxxxxx
-    // Purpose: vsp = vsp + ((x << 2) + 4)
-    UNWIND_OPCODE_INC_VSP = 0x00,
-
-    // Format: 01xxxxxx
-    // Purpose: vsp = vsp - ((x << 2) + 4)
-    UNWIND_OPCODE_DEC_VSP = 0x40,
-
-    // Format: 10000000 00000000
-    // Purpose: refuse to unwind
-    UNWIND_OPCODE_REFUSE = 0x8000,
-
-    // Format: 1000xxxx xxxxxxxx
-    // Purpose: pop r[15:12], r[11:4]
-    // Constraint: x != 0
-    UNWIND_OPCODE_POP_REG_MASK_R4 = 0x8000,
-
-    // Format: 1001xxxx
-    // Purpose: vsp = r[x]
-    // Constraint: x != 13 && x != 15
-    UNWIND_OPCODE_SET_VSP = 0x90,
-
-    // Format: 10100xxx
-    // Purpose: pop r[(4+x):4]
-    UNWIND_OPCODE_POP_REG_RANGE_R4 = 0xa0,
-
-    // Format: 10101xxx
-    // Purpose: pop r14, r[(4+x):4]
-    UNWIND_OPCODE_POP_REG_RANGE_R4_R14 = 0xa8,
-
-    // Format: 10110000
-    // Purpose: finish
-    UNWIND_OPCODE_FINISH = 0xb0,
-
-    // Format: 10110001 0000xxxx
-    // Purpose: pop r[3:0]
-    // Constraint: x != 0
-    UNWIND_OPCODE_POP_REG_MASK = 0xb100,
-
-    // Format: 10110010 x(uleb128)
-    // Purpose: vsp = vsp + ((x << 2) + 0x204)
-    UNWIND_OPCODE_INC_VSP_ULEB128 = 0xb2,
-
-    // Format: 10110011 xxxxyyyy
-    // Purpose: pop d[(x+y):x]
-    UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDX = 0xb300,
-
-    // Format: 10111xxx
-    // Purpose: pop d[(8+x):8]
-    UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDX_D8 = 0xb8,
-
-    // Format: 11000xxx
-    // Purpose: pop wR[(10+x):10]
-    UNWIND_OPCODE_POP_WIRELESS_MMX_REG_RANGE_WR10 = 0xc0,
-
-    // Format: 11000110 xxxxyyyy
-    // Purpose: pop wR[(x+y):x]
-    UNWIND_OPCODE_POP_WIRELESS_MMX_REG_RANGE = 0xc600,
-
-    // Format: 11000111 0000xxxx
-    // Purpose: pop wCGR[3:0]
-    // Constraint: x != 0
-    UNWIND_OPCODE_POP_WIRELESS_MMX_REG_MASK = 0xc700,
-
-    // Format: 11001000 xxxxyyyy
-    // Purpose: pop d[(16+x+y):(16+x)]
-    UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD_D16 = 0xc800,
-
-    // Format: 11001001 xxxxyyyy
-    // Purpose: pop d[(x+y):x]
-    UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD = 0xc900,
-
-    // Format: 11010xxx
-    // Purpose: pop d[(8+x):8]
-    UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD_D8 = 0xd0
-  };
-
-  /// ARM-defined Personality Routine Index
-  enum ARMPersonalityRoutineIndex {
-    // To make the exception handling table become more compact, ARM defined
-    // several personality routines in EHABI.  There are 3 different
-    // personality routines in ARM EHABI currently.  It is possible to have 16
-    // pre-defined personality routines at most.
-    AEABI_UNWIND_CPP_PR0 = 0,
-    AEABI_UNWIND_CPP_PR1 = 1,
-    AEABI_UNWIND_CPP_PR2 = 2,
-
-    NUM_PERSONALITY_INDEX
-  };
-
-}
-
-#endif // ARM_UNWIND_OP_H
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.cpp
index c94337081884..593fe349b1d4 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.cpp
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.cpp
@@ -13,8 +13,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "ARMUnwindOpAsm.h"
-
-#include "ARMUnwindOp.h"
+#include "llvm/Support/ARMEHABI.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/LEB128.h"
 
@@ -50,14 +49,15 @@ namespace {
 
     /// Emit the personality index prefix.
     inline void EmitPersonalityIndex(unsigned PI) {
-      assert(PI < NUM_PERSONALITY_INDEX && "Invalid personality prefix");
-      EmitByte(EHT_COMPACT | PI);
+      assert(PI < ARM::EHABI::NUM_PERSONALITY_INDEX &&
+             "Invalid personality prefix");
+      EmitByte(ARM::EHABI::EHT_COMPACT | PI);
     }
 
     /// Fill the rest of bytes with FINISH opcode.
     inline void FillFinishOpcode() {
       while (Pos < Vec.size())
-        EmitByte(UNWIND_OPCODE_FINISH);
+        EmitByte(ARM::EHABI::UNWIND_OPCODE_FINISH);
     }
   };
 }
@@ -85,22 +85,22 @@ void UnwindOpcodeAssembler::EmitRegSave(uint32_t RegSave) {
     uint32_t UnmaskedReg = RegSave & 0xfff0u & (~Mask);
     if (UnmaskedReg == 0u) {
       // Pop r[4 : (4 + n)]
-      EmitInt8(UNWIND_OPCODE_POP_REG_RANGE_R4 | Range);
+      EmitInt8(ARM::EHABI::UNWIND_OPCODE_POP_REG_RANGE_R4 | Range);
       RegSave &= 0x000fu;
     } else if (UnmaskedReg == (1u << 14)) {
       // Pop r[14] + r[4 : (4 + n)]
-      EmitInt8(UNWIND_OPCODE_POP_REG_RANGE_R4_R14 | Range);
+      EmitInt8(ARM::EHABI::UNWIND_OPCODE_POP_REG_RANGE_R4_R14 | Range);
       RegSave &= 0x000fu;
     }
   }
 
   // Two bytes opcode to save register r15-r4
   if ((RegSave & 0xfff0u) != 0)
-    EmitInt16(UNWIND_OPCODE_POP_REG_MASK_R4 | (RegSave >> 4));
+    EmitInt16(ARM::EHABI::UNWIND_OPCODE_POP_REG_MASK_R4 | (RegSave >> 4));
 
   // Opcode to save register r3-r0
   if ((RegSave & 0x000fu) != 0)
-    EmitInt16(UNWIND_OPCODE_POP_REG_MASK | (RegSave & 0x000fu));
+    EmitInt16(ARM::EHABI::UNWIND_OPCODE_POP_REG_MASK | (RegSave & 0x000fu));
 }
 
 /// Emit unwind opcodes for .vsave directives
@@ -125,7 +125,7 @@ void UnwindOpcodeAssembler::EmitVFPRegSave(uint32_t VFPRegSave) {
       Bit >>= 1;
     }
 
-    EmitInt16(UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD_D16 |
+    EmitInt16(ARM::EHABI::UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD_D16 |
               ((i - 16) << 4) | Range);
   }
 
@@ -147,34 +147,36 @@ void UnwindOpcodeAssembler::EmitVFPRegSave(uint32_t VFPRegSave) {
       Bit >>= 1;
     }
 
-    EmitInt16(UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD | (i << 4) | Range);
+    EmitInt16(ARM::EHABI::UNWIND_OPCODE_POP_VFP_REG_RANGE_FSTMFDD | (i << 4) |
+              Range);
   }
 }
 
 /// Emit unwind opcodes to copy address from source register to $sp.
 void UnwindOpcodeAssembler::EmitSetSP(uint16_t Reg) {
-  EmitInt8(UNWIND_OPCODE_SET_VSP | Reg);
+  EmitInt8(ARM::EHABI::UNWIND_OPCODE_SET_VSP | Reg);
 }
 
 /// Emit unwind opcodes to add $sp with an offset.
 void UnwindOpcodeAssembler::EmitSPOffset(int64_t Offset) {
   if (Offset > 0x200) {
     uint8_t Buff[16];
-    Buff[0] = UNWIND_OPCODE_INC_VSP_ULEB128;
+    Buff[0] = ARM::EHABI::UNWIND_OPCODE_INC_VSP_ULEB128;
     size_t ULEBSize = encodeULEB128((Offset - 0x204) >> 2, Buff + 1);
     EmitBytes(Buff, ULEBSize + 1);
   } else if (Offset > 0) {
     if (Offset > 0x100) {
-      EmitInt8(UNWIND_OPCODE_INC_VSP | 0x3fu);
+      EmitInt8(ARM::EHABI::UNWIND_OPCODE_INC_VSP | 0x3fu);
       Offset -= 0x100;
     }
-    EmitInt8(UNWIND_OPCODE_INC_VSP | static_cast<uint8_t>((Offset - 4) >> 2));
+    EmitInt8(ARM::EHABI::UNWIND_OPCODE_INC_VSP |
+             static_cast<uint8_t>((Offset - 4) >> 2));
   } else if (Offset < 0) {
     while (Offset < -0x100) {
-      EmitInt8(UNWIND_OPCODE_DEC_VSP | 0x3fu);
+      EmitInt8(ARM::EHABI::UNWIND_OPCODE_DEC_VSP | 0x3fu);
       Offset += 0x100;
     }
-    EmitInt8(UNWIND_OPCODE_DEC_VSP |
+    EmitInt8(ARM::EHABI::UNWIND_OPCODE_DEC_VSP |
              static_cast<uint8_t>(((-Offset) - 4) >> 2));
   }
 }
@@ -186,20 +188,23 @@ void UnwindOpcodeAssembler::Finalize(unsigned &PersonalityIndex,
 
   if (HasPersonality) {
     // User-specifed personality routine: [ SIZE , OP1 , OP2 , ... ]
-    PersonalityIndex = NUM_PERSONALITY_INDEX;
+    PersonalityIndex = ARM::EHABI::NUM_PERSONALITY_INDEX;
     size_t TotalSize = Ops.size() + 1;
     size_t RoundUpSize = (TotalSize + 3) / 4 * 4;
     Result.resize(RoundUpSize);
     OpStreamer.EmitSize(RoundUpSize);
   } else {
-    if (Ops.size() <= 3) {
+    // If no personalityindex is specified, select ane
+    if (PersonalityIndex == ARM::EHABI::NUM_PERSONALITY_INDEX)
+      PersonalityIndex = (Ops.size() <= 3) ? ARM::EHABI::AEABI_UNWIND_CPP_PR0
+                                           : ARM::EHABI::AEABI_UNWIND_CPP_PR1;
+    if (PersonalityIndex == ARM::EHABI::AEABI_UNWIND_CPP_PR0) {
       // __aeabi_unwind_cpp_pr0: [ 0x80 , OP1 , OP2 , OP3 ]
-      PersonalityIndex = AEABI_UNWIND_CPP_PR0;
+      assert(Ops.size() <= 3 && "too many opcodes for __aeabi_unwind_cpp_pr0");
       Result.resize(4);
       OpStreamer.EmitPersonalityIndex(PersonalityIndex);
     } else {
-      // __aeabi_unwind_cpp_pr1: [ 0x81 , SIZE , OP1 , OP2 , ... ]
-      PersonalityIndex = AEABI_UNWIND_CPP_PR1;
+      // __aeabi_unwind_cpp_pr{1,2}: [ {0x81,0x82} , SIZE , OP1 , OP2 , ... ]
       size_t TotalSize = Ops.size() + 2;
       size_t RoundUpSize = (TotalSize + 3) / 4 * 4;
       Result.resize(RoundUpSize);
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.h b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.h
index ac67c6efabb7..cd5875924300 100644
--- a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.h
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMUnwindOpAsm.h
@@ -15,10 +15,8 @@
 #ifndef ARM_UNWIND_OP_ASM_H
 #define ARM_UNWIND_OP_ASM_H
 
-#include "ARMUnwindOp.h"
-
 #include "llvm/ADT/SmallVector.h"
-#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/ARMEHABI.h"
 #include "llvm/Support/DataTypes.h"
 
 namespace llvm {
@@ -45,7 +43,7 @@ public:
     HasPersonality = 0;
   }
 
-  /// Set the personality index
+  /// Set the personality
   void setPersonality(const MCSymbol *Per) {
     HasPersonality = 1;
   }
@@ -62,6 +60,12 @@ public:
   /// Emit unwind opcodes to add $sp with an offset.
   void EmitSPOffset(int64_t Offset);
 
+  /// Emit unwind raw opcodes
+  void EmitRaw(const SmallVectorImpl<uint8_t> &Opcodes) {
+    Ops.insert(Ops.end(), Opcodes.begin(), Opcodes.end());
+    OpBegins.push_back(OpBegins.back() + Opcodes.size());
+  }
+
   /// Finalize the unwind opcode sequence for EmitBytes()
   void Finalize(unsigned &PersonalityIndex,
                 SmallVectorImpl<uint8_t> &Result);
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMWinCOFFObjectWriter.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMWinCOFFObjectWriter.cpp
new file mode 100644
index 000000000000..d31f1f41c697
--- /dev/null
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMWinCOFFObjectWriter.cpp
@@ -0,0 +1,82 @@
+//===-- ARMWinCOFFObjectWriter.cpp - ARM Windows COFF Object Writer -- C++ -==//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "MCTargetDesc/ARMFixupKinds.h"
+#include "llvm/MC/MCFixup.h"
+#include "llvm/MC/MCValue.h"
+#include "llvm/MC/MCWinCOFFObjectWriter.h"
+#include "llvm/Support/COFF.h"
+#include "llvm/Support/Debug.h"
+
+using namespace llvm;
+
+namespace {
+class ARMWinCOFFObjectWriter : public MCWinCOFFObjectTargetWriter {
+public:
+  ARMWinCOFFObjectWriter(bool Is64Bit)
+    : MCWinCOFFObjectTargetWriter(COFF::IMAGE_FILE_MACHINE_ARMNT) {
+    assert(!Is64Bit && "AArch64 support not yet implemented");
+  }
+  virtual ~ARMWinCOFFObjectWriter() { }
+
+  unsigned getRelocType(const MCValue &Target, const MCFixup &Fixup,
+                        bool IsCrossSection) const override;
+
+  bool recordRelocation(const MCFixup &) const override;
+};
+
+unsigned ARMWinCOFFObjectWriter::getRelocType(const MCValue &Target,
+                                              const MCFixup &Fixup,
+                                              bool IsCrossSection) const {
+  assert(getMachine() == COFF::IMAGE_FILE_MACHINE_ARMNT &&
+         "AArch64 support not yet implemented");
+
+  MCSymbolRefExpr::VariantKind Modifier =
+    Target.isAbsolute() ? MCSymbolRefExpr::VK_None : Target.getSymA()->getKind();
+
+  switch (static_cast<unsigned>(Fixup.getKind())) {
+  default: llvm_unreachable("unsupported relocation type");
+  case FK_Data_4:
+    switch (Modifier) {
+    case MCSymbolRefExpr::VK_COFF_IMGREL32:
+      return COFF::IMAGE_REL_ARM_ADDR32NB;
+    case MCSymbolRefExpr::VK_SECREL:
+      return COFF::IMAGE_REL_ARM_SECREL;
+    default:
+      return COFF::IMAGE_REL_ARM_ADDR32;
+    }
+  case FK_SecRel_2:
+    return COFF::IMAGE_REL_ARM_SECTION;
+  case FK_SecRel_4:
+    return COFF::IMAGE_REL_ARM_SECREL;
+  case ARM::fixup_t2_condbranch:
+    return COFF::IMAGE_REL_ARM_BRANCH20T;
+  case ARM::fixup_t2_uncondbranch:
+    return COFF::IMAGE_REL_ARM_BRANCH24T;
+  case ARM::fixup_arm_thumb_bl:
+  case ARM::fixup_arm_thumb_blx:
+    return COFF::IMAGE_REL_ARM_BLX23T;
+  case ARM::fixup_t2_movw_lo16:
+  case ARM::fixup_t2_movt_hi16:
+    return COFF::IMAGE_REL_ARM_MOV32T;
+  }
+}
+
+bool ARMWinCOFFObjectWriter::recordRelocation(const MCFixup &Fixup) const {
+  return static_cast<unsigned>(Fixup.getKind()) != ARM::fixup_t2_movt_hi16;
+}
+}
+
+namespace llvm {
+MCObjectWriter *createARMWinCOFFObjectWriter(raw_ostream &OS, bool Is64Bit) {
+  MCWinCOFFObjectTargetWriter *MOTW = new ARMWinCOFFObjectWriter(Is64Bit);
+  return createWinCOFFObjectWriter(MOTW, OS);
+}
+}
+
diff --git a/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMWinCOFFStreamer.cpp b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMWinCOFFStreamer.cpp
new file mode 100644
index 000000000000..b344ced2f67a
--- /dev/null
+++ b/contrib/llvm/lib/Target/ARM/MCTargetDesc/ARMWinCOFFStreamer.cpp
@@ -0,0 +1,46 @@
+//===-- ARMWinCOFFStreamer.cpp - ARM Target WinCOFF Streamer ----*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "ARMMCTargetDesc.h"
+#include "llvm/MC/MCWinCOFFStreamer.h"
+
+using namespace llvm;
+
+namespace {
+class ARMWinCOFFStreamer : public MCWinCOFFStreamer {
+public:
+  ARMWinCOFFStreamer(MCContext &C, MCAsmBackend &AB, MCCodeEmitter &CE,
+                     raw_ostream &OS)
+    : MCWinCOFFStreamer(C, AB, CE, OS) { }
+
+  void EmitAssemblerFlag(MCAssemblerFlag Flag) override;
+  void EmitThumbFunc(MCSymbol *Symbol) override;
+};
+
+void ARMWinCOFFStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) {
+  switch (Flag) {
+  default: llvm_unreachable("not implemented");
+  case MCAF_SyntaxUnified:
+  case MCAF_Code16:
+    break;
+  }
+}
+
+void ARMWinCOFFStreamer::EmitThumbFunc(MCSymbol *Symbol) {
+  getAssembler().setIsThumbFunc(Symbol);
+}
+}
+
+namespace llvm {
+MCStreamer *createARMWinCOFFStreamer(MCContext &Context, MCAsmBackend &MAB,
+                                     MCCodeEmitter &Emitter, raw_ostream &OS) {
+  return new ARMWinCOFFStreamer(Context, MAB, Emitter, OS);
+}
+}
+
diff --git a/contrib/llvm/lib/Target/ARM/MLxExpansionPass.cpp b/contrib/llvm/lib/Target/ARM/MLxExpansionPass.cpp
index 2e266c2e9624..f6d24e9e4f93 100644
--- a/contrib/llvm/lib/Target/ARM/MLxExpansionPass.cpp
+++ b/contrib/llvm/lib/Target/ARM/MLxExpansionPass.cpp
@@ -12,7 +12,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "mlx-expansion"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMSubtarget.h"
@@ -28,6 +27,8 @@
 #include "llvm/Target/TargetRegisterInfo.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "mlx-expansion"
+
 static cl::opt<bool>
 ForceExapnd("expand-all-fp-mlx", cl::init(false), cl::Hidden);
 static cl::opt<unsigned>
@@ -40,9 +41,9 @@ namespace {
     static char ID;
     MLxExpansion() : MachineFunctionPass(ID) {}
 
-    virtual bool runOnMachineFunction(MachineFunction &Fn);
+    bool runOnMachineFunction(MachineFunction &Fn) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "ARM MLA / MLS expansion pass";
     }
 
@@ -73,7 +74,7 @@ namespace {
 }
 
 void MLxExpansion::clearStack() {
-  std::fill(LastMIs, LastMIs + 4, (MachineInstr*)0);
+  std::fill(LastMIs, LastMIs + 4, nullptr);
   MIIdx = 0;
 }
 
@@ -88,7 +89,7 @@ MachineInstr *MLxExpansion::getAccDefMI(MachineInstr *MI) const {
   // real definition MI. This is important for _sfp instructions.
   unsigned Reg = MI->getOperand(1).getReg();
   if (TargetRegisterInfo::isPhysicalRegister(Reg))
-    return 0;
+    return nullptr;
 
   MachineBasicBlock *MBB = MI->getParent();
   MachineInstr *DefMI = MRI->getVRegDef(Reg);
@@ -120,7 +121,7 @@ unsigned MLxExpansion::getDefReg(MachineInstr *MI) const {
     return Reg;
 
   MachineBasicBlock *MBB = MI->getParent();
-  MachineInstr *UseMI = &*MRI->use_nodbg_begin(Reg);
+  MachineInstr *UseMI = &*MRI->use_instr_nodbg_begin(Reg);
   if (UseMI->getParent() != MBB)
     return Reg;
 
@@ -129,7 +130,7 @@ unsigned MLxExpansion::getDefReg(MachineInstr *MI) const {
     if (TargetRegisterInfo::isPhysicalRegister(Reg) ||
         !MRI->hasOneNonDBGUse(Reg))
       return Reg;
-    UseMI = &*MRI->use_nodbg_begin(Reg);
+    UseMI = &*MRI->use_instr_nodbg_begin(Reg);
     if (UseMI->getParent() != MBB)
       return Reg;
   }
@@ -312,9 +313,9 @@ MLxExpansion::ExpandFPMLxInstruction(MachineBasicBlock &MBB, MachineInstr *MI,
       dbgs() << "Expanding: " << *MI;
       dbgs() << "  to:\n";
       MachineBasicBlock::iterator MII = MI;
-      MII = llvm::prior(MII);
+      MII = std::prev(MII);
       MachineInstr &MI2 = *MII;
-      MII = llvm::prior(MII);
+      MII = std::prev(MII);
       MachineInstr &MI1 = *MII;
       dbgs() << "    " << MI1;
       dbgs() << "    " << MI2;
@@ -335,7 +336,7 @@ bool MLxExpansion::ExpandFPMLxInstructions(MachineBasicBlock &MBB) {
   while (MII != E) {
     MachineInstr *MI = &*MII;
 
-    if (MI->isLabel() || MI->isImplicitDef() || MI->isCopy()) {
+    if (MI->isPosition() || MI->isImplicitDef() || MI->isCopy()) {
       ++MII;
       continue;
     }
@@ -352,7 +353,7 @@ bool MLxExpansion::ExpandFPMLxInstructions(MachineBasicBlock &MBB) {
     if (Domain == ARMII::DomainGeneral) {
       if (++Skip == 2)
         // Assume dual issues of non-VFP / NEON instructions.
-        pushStack(0);
+        pushStack(nullptr);
     } else {
       Skip = 0;
 
@@ -385,11 +386,8 @@ bool MLxExpansion::runOnMachineFunction(MachineFunction &Fn) {
   isSwift = STI->isSwift();
 
   bool Modified = false;
-  for (MachineFunction::iterator MFI = Fn.begin(), E = Fn.end(); MFI != E;
-       ++MFI) {
-    MachineBasicBlock &MBB = *MFI;
+  for (MachineBasicBlock &MBB : Fn)
     Modified |= ExpandFPMLxInstructions(MBB);
-  }
 
   return Modified;
 }
diff --git a/contrib/llvm/lib/Target/ARM/TargetInfo/ARMTargetInfo.cpp b/contrib/llvm/lib/Target/ARM/TargetInfo/ARMTargetInfo.cpp
index fa5681fb12bf..e4646713d4d7 100644
--- a/contrib/llvm/lib/Target/ARM/TargetInfo/ARMTargetInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/TargetInfo/ARMTargetInfo.cpp
@@ -7,17 +7,22 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "ARM.h"
+#include "MCTargetDesc/ARMMCTargetDesc.h"
 #include "llvm/IR/Module.h"
 #include "llvm/Support/TargetRegistry.h"
 using namespace llvm;
 
-Target llvm::TheARMTarget, llvm::TheThumbTarget;
+Target llvm::TheARMLETarget,   llvm::TheARMBETarget;
+Target llvm::TheThumbLETarget, llvm::TheThumbBETarget;
 
 extern "C" void LLVMInitializeARMTargetInfo() { 
   RegisterTarget<Triple::arm, /*HasJIT=*/true>
-    X(TheARMTarget, "arm", "ARM");
+    X(TheARMLETarget, "arm", "ARM");
+  RegisterTarget<Triple::armeb, /*HasJIT=*/true>
+    Y(TheARMBETarget, "armeb", "ARM (big endian)");
 
   RegisterTarget<Triple::thumb, /*HasJIT=*/true>
-    Y(TheThumbTarget, "thumb", "Thumb");
+    A(TheThumbLETarget, "thumb", "Thumb");
+  RegisterTarget<Triple::thumbeb, /*HasJIT=*/true>
+    B(TheThumbBETarget, "thumbeb", "Thumb (big endian)");
 }
diff --git a/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.cpp b/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.cpp
index cfb33f5b8212..baa97a7c479a 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.cpp
@@ -16,10 +16,14 @@
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/CodeGen/MachineModuleInfo.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 
 using namespace llvm;
 
+Thumb1FrameLowering::Thumb1FrameLowering(const ARMSubtarget &sti)
+    : ARMFrameLowering(sti) {}
+
 bool Thumb1FrameLowering::hasReservedCallFrame(const MachineFunction &MF) const{
   const MachineFrameInfo *FFI = MF.getFrameInfo();
   unsigned CFSize = FFI->getMaxCallFrameSize();
@@ -83,6 +87,8 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
   MachineBasicBlock::iterator MBBI = MBB.begin();
   MachineFrameInfo  *MFI = MF.getFrameInfo();
   ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
+  MachineModuleInfo &MMI = MF.getMMI();
+  const MCRegisterInfo *MRI = MMI.getContext().getRegisterInfo();
   const Thumb1RegisterInfo *RegInfo =
     static_cast<const Thumb1RegisterInfo*>(MF.getTarget().getRegisterInfo());
   const Thumb1InstrInfo &TII =
@@ -91,10 +97,13 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
   unsigned Align = MF.getTarget().getFrameLowering()->getStackAlignment();
   unsigned ArgRegsSaveSize = AFI->getArgRegsSaveSize(Align);
   unsigned NumBytes = MFI->getStackSize();
+  assert(NumBytes >= ArgRegsSaveSize &&
+         "ArgRegsSaveSize is included in NumBytes");
   const std::vector<CalleeSavedInfo> &CSI = MFI->getCalleeSavedInfo();
   DebugLoc dl = MBBI != MBB.end() ? MBBI->getDebugLoc() : DebugLoc();
   unsigned FramePtr = RegInfo->getFrameRegister(MF);
   unsigned BasePtr = RegInfo->getBaseRegister();
+  int CFAOffset = 0;
 
   // Thumb add/sub sp, imm8 instructions implicitly multiply the offset by 4.
   NumBytes = (NumBytes + 3) & ~3;
@@ -105,14 +114,26 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
   unsigned GPRCS1Size = 0, GPRCS2Size = 0, DPRCSSize = 0;
   int FramePtrSpillFI = 0;
 
-  if (ArgRegsSaveSize)
+  if (ArgRegsSaveSize) {
     emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, -ArgRegsSaveSize,
                  MachineInstr::FrameSetup);
+    CFAOffset -= ArgRegsSaveSize;
+    unsigned CFIIndex = MMI.addFrameInst(
+        MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+    BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+        .addCFIIndex(CFIIndex);
+  }
 
   if (!AFI->hasStackFrame()) {
-    if (NumBytes != 0)
-      emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, -NumBytes,
+    if (NumBytes - ArgRegsSaveSize != 0) {
+      emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, -(NumBytes - ArgRegsSaveSize),
                    MachineInstr::FrameSetup);
+      CFAOffset -= NumBytes - ArgRegsSaveSize;
+      unsigned CFIIndex = MMI.addFrameInst(
+          MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
     return;
   }
 
@@ -120,6 +141,15 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
     unsigned Reg = CSI[i].getReg();
     int FI = CSI[i].getFrameIdx();
     switch (Reg) {
+    case ARM::R8:
+    case ARM::R9:
+    case ARM::R10:
+    case ARM::R11:
+      if (STI.isTargetMachO()) {
+        GPRCS2Size += 4;
+        break;
+      }
+      // fallthrough
     case ARM::R4:
     case ARM::R5:
     case ARM::R6:
@@ -129,17 +159,6 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
         FramePtrSpillFI = FI;
       GPRCS1Size += 4;
       break;
-    case ARM::R8:
-    case ARM::R9:
-    case ARM::R10:
-    case ARM::R11:
-      if (Reg == FramePtr)
-        FramePtrSpillFI = FI;
-      if (STI.isTargetIOS())
-        GPRCS2Size += 4;
-      else
-        GPRCS1Size += 4;
-      break;
     default:
       DPRCSSize += 8;
     }
@@ -152,7 +171,7 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
   }
 
   // Determine starting offsets of spill areas.
-  unsigned DPRCSOffset  = NumBytes - (GPRCS1Size + GPRCS2Size + DPRCSSize);
+  unsigned DPRCSOffset  = NumBytes - ArgRegsSaveSize - (GPRCS1Size + GPRCS2Size + DPRCSSize);
   unsigned GPRCS2Offset = DPRCSOffset + DPRCSSize;
   unsigned GPRCS1Offset = GPRCS2Offset + GPRCS2Size;
   bool HasFP = hasFP(MF);
@@ -165,27 +184,89 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
   NumBytes = DPRCSOffset;
 
   int FramePtrOffsetInBlock = 0;
-  if (tryFoldSPUpdateIntoPushPop(MF, prior(MBBI), NumBytes)) {
+  unsigned adjustedGPRCS1Size = GPRCS1Size;
+  if (tryFoldSPUpdateIntoPushPop(STI, MF, std::prev(MBBI), NumBytes)) {
     FramePtrOffsetInBlock = NumBytes;
+    adjustedGPRCS1Size += NumBytes;
     NumBytes = 0;
   }
 
+  if (adjustedGPRCS1Size) {
+    CFAOffset -= adjustedGPRCS1Size;
+    unsigned CFIIndex = MMI.addFrameInst(
+        MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+    BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+        .addCFIIndex(CFIIndex);
+  }
+  for (std::vector<CalleeSavedInfo>::const_iterator I = CSI.begin(),
+         E = CSI.end(); I != E; ++I) {
+    unsigned Reg = I->getReg();
+    int FI = I->getFrameIdx();
+    switch (Reg) {
+    case ARM::R8:
+    case ARM::R9:
+    case ARM::R10:
+    case ARM::R11:
+    case ARM::R12:
+      if (STI.isTargetMachO())
+        break;
+      // fallthough
+    case ARM::R0:
+    case ARM::R1:
+    case ARM::R2:
+    case ARM::R3:
+    case ARM::R4:
+    case ARM::R5:
+    case ARM::R6:
+    case ARM::R7:
+    case ARM::LR:
+      unsigned CFIIndex = MMI.addFrameInst(MCCFIInstruction::createOffset(
+          nullptr, MRI->getDwarfRegNum(Reg, true), MFI->getObjectOffset(FI)));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+      break;
+    }
+  }
+
+
   // Adjust FP so it point to the stack slot that contains the previous FP.
   if (HasFP) {
-    FramePtrOffsetInBlock += MFI->getObjectOffset(FramePtrSpillFI) + GPRCS1Size;
+    FramePtrOffsetInBlock += MFI->getObjectOffset(FramePtrSpillFI)
+		                     + GPRCS1Size + ArgRegsSaveSize;
     AddDefaultPred(BuildMI(MBB, MBBI, dl, TII.get(ARM::tADDrSPi), FramePtr)
       .addReg(ARM::SP).addImm(FramePtrOffsetInBlock / 4)
       .setMIFlags(MachineInstr::FrameSetup));
+    if(FramePtrOffsetInBlock) {
+      CFAOffset += FramePtrOffsetInBlock;
+      unsigned CFIIndex = MMI.addFrameInst(MCCFIInstruction::createDefCfa(
+          nullptr, MRI->getDwarfRegNum(FramePtr, true), CFAOffset));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    } else {
+      unsigned CFIIndex =
+          MMI.addFrameInst(MCCFIInstruction::createDefCfaRegister(
+              nullptr, MRI->getDwarfRegNum(FramePtr, true)));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
     if (NumBytes > 508)
       // If offset is > 508 then sp cannot be adjusted in a single instruction,
       // try restoring from fp instead.
       AFI->setShouldRestoreSPFromFP(true);
   }
 
-  if (NumBytes)
+  if (NumBytes) {
     // Insert it after all the callee-save spills.
     emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, -NumBytes,
                  MachineInstr::FrameSetup);
+    if (!HasFP) {
+      CFAOffset -= NumBytes;
+      unsigned CFIIndex = MMI.addFrameInst(
+          MCCFIInstruction::createDefCfaOffset(nullptr, CFAOffset));
+      BuildMI(MBB, MBBI, dl, TII.get(TargetOpcode::CFI_INSTRUCTION))
+          .addCFIIndex(CFIIndex);
+    }
+  }
 
   if (STI.isTargetELF() && HasFP)
     MFI->setOffsetAdjustment(MFI->getOffsetAdjustment() -
@@ -215,7 +296,7 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
     AFI->setShouldRestoreSPFromFP(true);
 }
 
-static bool isCSRestore(MachineInstr *MI, const uint16_t *CSRegs) {
+static bool isCSRestore(MachineInstr *MI, const MCPhysReg *CSRegs) {
   if (MI->getOpcode() == ARM::tLDRspi &&
       MI->getOperand(1).isFI() &&
       isCalleeSavedRegister(MI->getOperand(0).getReg(), CSRegs))
@@ -248,12 +329,14 @@ void Thumb1FrameLowering::emitEpilogue(MachineFunction &MF,
   unsigned Align = MF.getTarget().getFrameLowering()->getStackAlignment();
   unsigned ArgRegsSaveSize = AFI->getArgRegsSaveSize(Align);
   int NumBytes = (int)MFI->getStackSize();
-  const uint16_t *CSRegs = RegInfo->getCalleeSavedRegs();
+  assert((unsigned)NumBytes >= ArgRegsSaveSize &&
+         "ArgRegsSaveSize is included in NumBytes");
+  const MCPhysReg *CSRegs = RegInfo->getCalleeSavedRegs();
   unsigned FramePtr = RegInfo->getFrameRegister(MF);
 
   if (!AFI->hasStackFrame()) {
-    if (NumBytes != 0)
-      emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, NumBytes);
+    if (NumBytes - ArgRegsSaveSize != 0)
+      emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, NumBytes - ArgRegsSaveSize);
   } else {
     // Unwind MBBI to point to first LDR / VLDRD.
     if (MBBI != MBB.begin()) {
@@ -267,7 +350,8 @@ void Thumb1FrameLowering::emitEpilogue(MachineFunction &MF,
     // Move SP to start of FP callee save spill area.
     NumBytes -= (AFI->getGPRCalleeSavedArea1Size() +
                  AFI->getGPRCalleeSavedArea2Size() +
-                 AFI->getDPRCalleeSavedAreaSize());
+                 AFI->getDPRCalleeSavedAreaSize() +
+                 ArgRegsSaveSize);
 
     if (AFI->shouldRestoreSPFromFP()) {
       NumBytes = AFI->getFramePtrSpillOffset() - NumBytes;
@@ -289,11 +373,11 @@ void Thumb1FrameLowering::emitEpilogue(MachineFunction &MF,
     } else {
       if (MBBI->getOpcode() == ARM::tBX_RET &&
           &MBB.front() != MBBI &&
-          prior(MBBI)->getOpcode() == ARM::tPOP) {
-        MachineBasicBlock::iterator PMBBI = prior(MBBI);
-        if (!tryFoldSPUpdateIntoPushPop(MF, PMBBI, NumBytes))
+          std::prev(MBBI)->getOpcode() == ARM::tPOP) {
+        MachineBasicBlock::iterator PMBBI = std::prev(MBBI);
+        if (!tryFoldSPUpdateIntoPushPop(STI, MF, PMBBI, NumBytes))
           emitSPUpdate(MBB, PMBBI, TII, dl, *RegInfo, NumBytes);
-      } else if (!tryFoldSPUpdateIntoPushPop(MF, MBBI, NumBytes))
+      } else if (!tryFoldSPUpdateIntoPushPop(STI, MF, MBBI, NumBytes))
         emitSPUpdate(MBB, MBBI, TII, dl, *RegInfo, NumBytes);
     }
   }
@@ -304,9 +388,9 @@ void Thumb1FrameLowering::emitEpilogue(MachineFunction &MF,
     // we need to update the SP after popping the value. Therefore, we
     // pop the old LR into R3 as a temporary.
 
-    // Move back past the callee-saved register restoration
-    while (MBBI != MBB.end() && isCSRestore(MBBI, CSRegs))
-      ++MBBI;
+    // Get the last instruction, tBX_RET
+    MBBI = MBB.getLastNonDebugInstr();
+    assert (MBBI->getOpcode() == ARM::tBX_RET);
     // Epilogue for vararg functions: pop LR to R3 and branch off it.
     AddDefaultPred(BuildMI(MBB, MBBI, dl, TII.get(ARM::tPOP)))
       .addReg(ARM::R3, RegState::Define);
diff --git a/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.h b/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.h
index 5a300afd5d36..a227f8ece73f 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.h
+++ b/contrib/llvm/lib/Target/ARM/Thumb1FrameLowering.h
@@ -11,44 +11,40 @@
 //
 //===----------------------------------------------------------------------===//
 
-#ifndef __THUMB_FRAMEINFO_H_
-#define __THUMB_FRAMEINFO_H_
+#ifndef LLVM_ARM_THUMB1FRAMELOWERING_H
+#define LLVM_ARM_THUMB1FRAMELOWERING_H
 
-#include "ARM.h"
 #include "ARMFrameLowering.h"
-#include "ARMSubtarget.h"
 #include "Thumb1InstrInfo.h"
 #include "Thumb1RegisterInfo.h"
 #include "llvm/Target/TargetFrameLowering.h"
 
 namespace llvm {
-  class ARMSubtarget;
 
 class Thumb1FrameLowering : public ARMFrameLowering {
 public:
-  explicit Thumb1FrameLowering(const ARMSubtarget &sti)
-    : ARMFrameLowering(sti) {
-  }
+  explicit Thumb1FrameLowering(const ARMSubtarget &sti);
 
   /// emitProlog/emitEpilog - These methods insert prolog and epilog code into
   /// the function.
-  void emitPrologue(MachineFunction &MF) const;
-  void emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const;
+  void emitPrologue(MachineFunction &MF) const override;
+  void emitEpilogue(MachineFunction &MF, MachineBasicBlock &MBB) const override;
 
   bool spillCalleeSavedRegisters(MachineBasicBlock &MBB,
                                  MachineBasicBlock::iterator MI,
                                  const std::vector<CalleeSavedInfo> &CSI,
-                                 const TargetRegisterInfo *TRI) const;
+                                 const TargetRegisterInfo *TRI) const override;
   bool restoreCalleeSavedRegisters(MachineBasicBlock &MBB,
-                                   MachineBasicBlock::iterator MI,
-                                   const std::vector<CalleeSavedInfo> &CSI,
-                                   const TargetRegisterInfo *TRI) const;
+                                  MachineBasicBlock::iterator MI,
+                                  const std::vector<CalleeSavedInfo> &CSI,
+                                  const TargetRegisterInfo *TRI) const override;
 
-  bool hasReservedCallFrame(const MachineFunction &MF) const;
+  bool hasReservedCallFrame(const MachineFunction &MF) const override;
 
-  void eliminateCallFramePseudoInstr(MachineFunction &MF,
-                                     MachineBasicBlock &MBB,
-                                     MachineBasicBlock::iterator MI) const;
+  void
+  eliminateCallFramePseudoInstr(MachineFunction &MF,
+                                MachineBasicBlock &MBB,
+                                MachineBasicBlock::iterator MI) const override;
 };
 
 } // End llvm namespace
diff --git a/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp b/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp
index 22a925e0ffbc..68cbb5c580df 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.cpp
@@ -12,7 +12,6 @@
 //===----------------------------------------------------------------------===//
 
 #include "Thumb1InstrInfo.h"
-#include "ARM.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineMemOperand.h"
diff --git a/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.h b/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.h
index 36af20492d4e..c5845b75e587 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.h
+++ b/contrib/llvm/lib/Target/ARM/Thumb1InstrInfo.h
@@ -14,7 +14,6 @@
 #ifndef THUMB1INSTRUCTIONINFO_H
 #define THUMB1INSTRUCTIONINFO_H
 
-#include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "Thumb1RegisterInfo.h"
 
@@ -27,33 +26,33 @@ public:
   explicit Thumb1InstrInfo(const ARMSubtarget &STI);
 
   /// getNoopForMachoTarget - Return the noop instruction to use for a noop.
-  void getNoopForMachoTarget(MCInst &NopInst) const;
+  void getNoopForMachoTarget(MCInst &NopInst) const override;
 
   // Return the non-pre/post incrementing version of 'Opc'. Return 0
   // if there is not such an opcode.
-  unsigned getUnindexedOpcode(unsigned Opc) const;
+  unsigned getUnindexedOpcode(unsigned Opc) const override;
 
   /// getRegisterInfo - TargetInstrInfo is a superset of MRegister info.  As
   /// such, whenever a client has an instance of instruction info, it should
   /// always be able to get register info as well (through this method).
   ///
-  const Thumb1RegisterInfo &getRegisterInfo() const { return RI; }
+  const Thumb1RegisterInfo &getRegisterInfo() const override { return RI; }
 
   void copyPhysReg(MachineBasicBlock &MBB,
                    MachineBasicBlock::iterator I, DebugLoc DL,
                    unsigned DestReg, unsigned SrcReg,
-                   bool KillSrc) const;
+                   bool KillSrc) const override;
   void storeRegToStackSlot(MachineBasicBlock &MBB,
                            MachineBasicBlock::iterator MBBI,
                            unsigned SrcReg, bool isKill, int FrameIndex,
                            const TargetRegisterClass *RC,
-                           const TargetRegisterInfo *TRI) const;
+                           const TargetRegisterInfo *TRI) const override;
 
   void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI,
                             unsigned DestReg, int FrameIndex,
                             const TargetRegisterClass *RC,
-                            const TargetRegisterInfo *TRI) const;
+                            const TargetRegisterInfo *TRI) const override;
 
 };
 }
diff --git a/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.cpp b/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.cpp
index 65a7221d5db7..f907b143ef1b 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.cpp
@@ -13,7 +13,6 @@
 //===----------------------------------------------------------------------===//
 
 #include "Thumb1RegisterInfo.h"
-#include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "ARMMachineFunctionInfo.h"
 #include "ARMSubtarget.h"
@@ -30,7 +29,6 @@
 #include "llvm/IR/LLVMContext.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/ErrorHandling.h"
-#include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetFrameLowering.h"
 #include "llvm/Target/TargetMachine.h"
 
@@ -421,7 +419,7 @@ rewriteFrameIndex(MachineBasicBlock::iterator II, unsigned FrameRegIdx,
         MI.getOperand(FrameRegIdx+1).ChangeToImmediate(Mask);
       }
       Offset = (Offset - Mask * Scale);
-      MachineBasicBlock::iterator NII = llvm::next(II);
+      MachineBasicBlock::iterator NII = std::next(II);
       emitThumbRegPlusImmediate(MBB, NII, dl, DestReg, DestReg, Offset, TII,
                                 *this);
     } else {
@@ -484,10 +482,8 @@ rewriteFrameIndex(MachineBasicBlock::iterator II, unsigned FrameRegIdx,
   return Offset == 0;
 }
 
-void
-Thumb1RegisterInfo::resolveFrameIndex(MachineBasicBlock::iterator I,
-                                      unsigned BaseReg, int64_t Offset) const {
-  MachineInstr &MI = *I;
+void Thumb1RegisterInfo::resolveFrameIndex(MachineInstr &MI, unsigned BaseReg,
+                                           int64_t Offset) const {
   const ARMBaseInstrInfo &TII =
     *static_cast<const ARMBaseInstrInfo*>(
       MI.getParent()->getParent()->getTarget().getInstrInfo());
diff --git a/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.h b/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.h
index 9689b23146a0..0c0abbe99042 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.h
+++ b/contrib/llvm/lib/Target/ARM/Thumb1RegisterInfo.h
@@ -15,7 +15,6 @@
 #ifndef THUMB1REGISTERINFO_H
 #define THUMB1REGISTERINFO_H
 
-#include "ARM.h"
 #include "ARMBaseRegisterInfo.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 
@@ -27,21 +26,20 @@ struct Thumb1RegisterInfo : public ARMBaseRegisterInfo {
 public:
   Thumb1RegisterInfo(const ARMSubtarget &STI);
 
-  const TargetRegisterClass*
-  getLargestLegalSuperClass(const TargetRegisterClass *RC) const;
+  const TargetRegisterClass *
+  getLargestLegalSuperClass(const TargetRegisterClass *RC) const override;
 
-  const TargetRegisterClass*
-  getPointerRegClass(const MachineFunction &MF, unsigned Kind = 0) const;
+  const TargetRegisterClass *
+  getPointerRegClass(const MachineFunction &MF,
+                     unsigned Kind = 0) const override;
 
   /// emitLoadConstPool - Emits a load from constpool to materialize the
   /// specified immediate.
- void emitLoadConstPool(MachineBasicBlock &MBB,
-                        MachineBasicBlock::iterator &MBBI,
-                        DebugLoc dl,
-                        unsigned DestReg, unsigned SubIdx, int Val,
-                        ARMCC::CondCodes Pred = ARMCC::AL,
-                        unsigned PredReg = 0,
-                        unsigned MIFlags = MachineInstr::NoFlags) const;
+  void
+  emitLoadConstPool(MachineBasicBlock &MBB, MachineBasicBlock::iterator &MBBI,
+                    DebugLoc dl, unsigned DestReg, unsigned SubIdx, int Val,
+                    ARMCC::CondCodes Pred = ARMCC::AL, unsigned PredReg = 0,
+                    unsigned MIFlags = MachineInstr::NoFlags) const override;
 
   // rewrite MI to access 'Offset' bytes from the FP. Update Offset to be
   // however much remains to be handled. Return 'true' if no further
@@ -49,16 +47,16 @@ public:
   bool rewriteFrameIndex(MachineBasicBlock::iterator II, unsigned FrameRegIdx,
                          unsigned FrameReg, int &Offset,
                          const ARMBaseInstrInfo &TII) const;
-  void resolveFrameIndex(MachineBasicBlock::iterator I,
-                         unsigned BaseReg, int64_t Offset) const;
+  void resolveFrameIndex(MachineInstr &MI, unsigned BaseReg,
+                         int64_t Offset) const override;
   bool saveScavengerRegister(MachineBasicBlock &MBB,
                              MachineBasicBlock::iterator I,
                              MachineBasicBlock::iterator &UseMI,
                              const TargetRegisterClass *RC,
-                             unsigned Reg) const;
+                             unsigned Reg) const override;
   void eliminateFrameIndex(MachineBasicBlock::iterator II,
                            int SPAdj, unsigned FIOperandNum,
-                           RegScavenger *RS = NULL) const;
+                           RegScavenger *RS = nullptr) const override;
 };
 }
 
diff --git a/contrib/llvm/lib/Target/ARM/Thumb2ITBlockPass.cpp b/contrib/llvm/lib/Target/ARM/Thumb2ITBlockPass.cpp
index 0b7d3bb7754f..edb9ff3a24f2 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb2ITBlockPass.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb2ITBlockPass.cpp
@@ -7,7 +7,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "thumb2-it"
 #include "ARM.h"
 #include "ARMMachineFunctionInfo.h"
 #include "Thumb2InstrInfo.h"
@@ -19,6 +18,8 @@
 #include "llvm/CodeGen/MachineInstrBundle.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "thumb2-it"
+
 STATISTIC(NumITs,        "Number of IT blocks inserted");
 STATISTIC(NumMovedInsts, "Number of predicated instructions moved");
 
@@ -33,9 +34,9 @@ namespace {
     const TargetRegisterInfo *TRI;
     ARMFunctionInfo *AFI;
 
-    virtual bool runOnMachineFunction(MachineFunction &Fn);
+    bool runOnMachineFunction(MachineFunction &Fn) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "Thumb IT blocks insertion pass";
     }
 
@@ -242,7 +243,7 @@ bool Thumb2ITBlockPass::InsertITInstructions(MachineBasicBlock &MBB) {
 
     // Finalize the bundle.
     MachineBasicBlock::instr_iterator LI = LastITMI;
-    finalizeBundle(MBB, InsertPos.getInstrIterator(), llvm::next(LI));
+    finalizeBundle(MBB, InsertPos.getInstrIterator(), std::next(LI));
 
     Modified = true;
     ++NumITs;
diff --git a/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.cpp b/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.cpp
index 91788ac4f893..a9df006fb8c1 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.cpp
@@ -12,7 +12,6 @@
 //===----------------------------------------------------------------------===//
 
 #include "Thumb2InstrInfo.h"
-#include "ARM.h"
 #include "ARMConstantPoolValue.h"
 #include "ARMMachineFunctionInfo.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
diff --git a/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.h b/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.h
index 2cdcd06e9350..34d45d307ff4 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.h
+++ b/contrib/llvm/lib/Target/ARM/Thumb2InstrInfo.h
@@ -14,7 +14,6 @@
 #ifndef THUMB2INSTRUCTIONINFO_H
 #define THUMB2INSTRUCTIONINFO_H
 
-#include "ARM.h"
 #include "ARMBaseInstrInfo.h"
 #include "Thumb2RegisterInfo.h"
 
@@ -28,40 +27,40 @@ public:
   explicit Thumb2InstrInfo(const ARMSubtarget &STI);
 
   /// getNoopForMachoTarget - Return the noop instruction to use for a noop.
-  void getNoopForMachoTarget(MCInst &NopInst) const;
+  void getNoopForMachoTarget(MCInst &NopInst) const override;
 
   // Return the non-pre/post incrementing version of 'Opc'. Return 0
   // if there is not such an opcode.
-  unsigned getUnindexedOpcode(unsigned Opc) const;
+  unsigned getUnindexedOpcode(unsigned Opc) const override;
 
   void ReplaceTailWithBranchTo(MachineBasicBlock::iterator Tail,
-                               MachineBasicBlock *NewDest) const;
+                               MachineBasicBlock *NewDest) const override;
 
   bool isLegalToSplitMBBAt(MachineBasicBlock &MBB,
-                           MachineBasicBlock::iterator MBBI) const;
+                           MachineBasicBlock::iterator MBBI) const override;
 
   void copyPhysReg(MachineBasicBlock &MBB,
                    MachineBasicBlock::iterator I, DebugLoc DL,
                    unsigned DestReg, unsigned SrcReg,
-                   bool KillSrc) const;
+                   bool KillSrc) const override;
 
   void storeRegToStackSlot(MachineBasicBlock &MBB,
                            MachineBasicBlock::iterator MBBI,
                            unsigned SrcReg, bool isKill, int FrameIndex,
                            const TargetRegisterClass *RC,
-                           const TargetRegisterInfo *TRI) const;
+                           const TargetRegisterInfo *TRI) const override;
 
   void loadRegFromStackSlot(MachineBasicBlock &MBB,
                             MachineBasicBlock::iterator MBBI,
                             unsigned DestReg, int FrameIndex,
                             const TargetRegisterClass *RC,
-                            const TargetRegisterInfo *TRI) const;
+                            const TargetRegisterInfo *TRI) const override;
 
   /// getRegisterInfo - TargetInstrInfo is a superset of MRegister info.  As
   /// such, whenever a client has an instance of instruction info, it should
   /// always be able to get register info as well (through this method).
   ///
-  const Thumb2RegisterInfo &getRegisterInfo() const { return RI; }
+  const Thumb2RegisterInfo &getRegisterInfo() const override { return RI; }
 };
 
 /// getITInstrPredicate - Valid only in Thumb2 mode. This function is identical
diff --git a/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.cpp b/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.cpp
index 4cb827f308ef..782d81fd424d 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.cpp
@@ -14,7 +14,6 @@
 
 #include "Thumb2RegisterInfo.h"
 #include "ARM.h"
-#include "ARMBaseInstrInfo.h"
 #include "ARMSubtarget.h"
 #include "llvm/CodeGen/MachineConstantPool.h"
 #include "llvm/CodeGen/MachineFunction.h"
@@ -22,6 +21,8 @@
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DerivedTypes.h"
 #include "llvm/IR/Function.h"
+#include "llvm/Target/TargetInstrInfo.h"
+#include "llvm/Target/TargetMachine.h"
 using namespace llvm;
 
 Thumb2RegisterInfo::Thumb2RegisterInfo(const ARMSubtarget &sti)
diff --git a/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.h b/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.h
index b1d63fa86d01..8a33e6cea919 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.h
+++ b/contrib/llvm/lib/Target/ARM/Thumb2RegisterInfo.h
@@ -15,9 +15,7 @@
 #ifndef THUMB2REGISTERINFO_H
 #define THUMB2REGISTERINFO_H
 
-#include "ARM.h"
 #include "ARMBaseRegisterInfo.h"
-#include "llvm/Target/TargetRegisterInfo.h"
 
 namespace llvm {
 
@@ -29,13 +27,11 @@ public:
 
   /// emitLoadConstPool - Emits a load from constpool to materialize the
   /// specified immediate.
-  void emitLoadConstPool(MachineBasicBlock &MBB,
-                         MachineBasicBlock::iterator &MBBI,
-                         DebugLoc dl,
-                         unsigned DestReg, unsigned SubIdx, int Val,
-                         ARMCC::CondCodes Pred = ARMCC::AL,
-                         unsigned PredReg = 0,
-                         unsigned MIFlags = MachineInstr::NoFlags) const;
+  void
+  emitLoadConstPool(MachineBasicBlock &MBB, MachineBasicBlock::iterator &MBBI,
+                    DebugLoc dl, unsigned DestReg, unsigned SubIdx, int Val,
+                    ARMCC::CondCodes Pred = ARMCC::AL, unsigned PredReg = 0,
+                    unsigned MIFlags = MachineInstr::NoFlags) const override;
 };
 }
 
diff --git a/contrib/llvm/lib/Target/ARM/Thumb2SizeReduction.cpp b/contrib/llvm/lib/Target/ARM/Thumb2SizeReduction.cpp
index 4795aae12fd6..09debe76f1b6 100644
--- a/contrib/llvm/lib/Target/ARM/Thumb2SizeReduction.cpp
+++ b/contrib/llvm/lib/Target/ARM/Thumb2SizeReduction.cpp
@@ -7,10 +7,8 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "t2-reduce-size"
 #include "ARM.h"
 #include "ARMBaseInstrInfo.h"
-#include "ARMBaseRegisterInfo.h"
 #include "ARMSubtarget.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "Thumb2InstrInfo.h"
@@ -23,9 +21,11 @@
 #include "llvm/IR/Function.h"        // To access Function attributes
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
-#include "llvm/Support/raw_ostream.h"
+#include "llvm/Target/TargetMachine.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "t2-reduce-size"
+
 STATISTIC(NumNarrows,  "Number of 32-bit instrs reduced to 16-bit ones");
 STATISTIC(Num2Addrs,   "Number of 32-bit instrs reduced to 2addr 16-bit ones");
 STATISTIC(NumLdSts,    "Number of 32-bit load / store reduced to 16-bit ones");
@@ -137,9 +137,9 @@ namespace {
     const Thumb2InstrInfo *TII;
     const ARMSubtarget *STI;
 
-    virtual bool runOnMachineFunction(MachineFunction &MF);
+    bool runOnMachineFunction(MachineFunction &MF) override;
 
-    virtual const char *getPassName() const {
+    const char *getPassName() const override {
       return "Thumb2 instruction size reduction pass";
     }
 
@@ -256,8 +256,7 @@ Thumb2SizeReduce::canAddPseudoFlagDep(MachineInstr *Use, bool FirstInSelfLoop) {
     return HighLatencyCPSR || FirstInSelfLoop;
 
   SmallSet<unsigned, 2> Defs;
-  for (unsigned i = 0, e = CPSRDef->getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = CPSRDef->getOperand(i);
+  for (const MachineOperand &MO : CPSRDef->operands()) {
     if (!MO.isReg() || MO.isUndef() || MO.isUse())
       continue;
     unsigned Reg = MO.getReg();
@@ -266,8 +265,7 @@ Thumb2SizeReduce::canAddPseudoFlagDep(MachineInstr *Use, bool FirstInSelfLoop) {
     Defs.insert(Reg);
   }
 
-  for (unsigned i = 0, e = Use->getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = Use->getOperand(i);
+  for (const MachineOperand &MO : Use->operands()) {
     if (!MO.isReg() || MO.isUndef() || MO.isDef())
       continue;
     unsigned Reg = MO.getReg();
@@ -858,8 +856,7 @@ Thumb2SizeReduce::ReduceToNarrow(MachineBasicBlock &MBB, MachineInstr *MI,
 
 static bool UpdateCPSRDef(MachineInstr &MI, bool LiveCPSR, bool &DefCPSR) {
   bool HasDef = false;
-  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = MI.getOperand(i);
+  for (const MachineOperand &MO : MI.operands()) {
     if (!MO.isReg() || MO.isUndef() || MO.isUse())
       continue;
     if (MO.getReg() != ARM::CPSR)
@@ -874,8 +871,7 @@ static bool UpdateCPSRDef(MachineInstr &MI, bool LiveCPSR, bool &DefCPSR) {
 }
 
 static bool UpdateCPSRUse(MachineInstr &MI, bool LiveCPSR) {
-  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = MI.getOperand(i);
+  for (const MachineOperand &MO : MI.operands()) {
     if (!MO.isReg() || MO.isUndef() || MO.isDef())
       continue;
     if (MO.getReg() != ARM::CPSR)
@@ -920,15 +916,14 @@ bool Thumb2SizeReduce::ReduceMBB(MachineBasicBlock &MBB) {
 
   // Yes, CPSR could be livein.
   bool LiveCPSR = MBB.isLiveIn(ARM::CPSR);
-  MachineInstr *BundleMI = 0;
+  MachineInstr *BundleMI = nullptr;
 
-  CPSRDef = 0;
+  CPSRDef = nullptr;
   HighLatencyCPSR = false;
 
   // Check predecessors for the latest CPSRDef.
-  for (MachineBasicBlock::pred_iterator
-       I = MBB.pred_begin(), E = MBB.pred_end(); I != E; ++I) {
-    const MBBInfo &PInfo = BlockInfo[(*I)->getNumber()];
+  for (auto *Pred : MBB.predecessors()) {
+    const MBBInfo &PInfo = BlockInfo[Pred->getNumber()];
     if (!PInfo.Visited) {
       // Since blocks are visited in RPO, this must be a back-edge.
       continue;
@@ -945,7 +940,7 @@ bool Thumb2SizeReduce::ReduceMBB(MachineBasicBlock &MBB) {
   MachineBasicBlock::instr_iterator MII = MBB.instr_begin(),E = MBB.instr_end();
   MachineBasicBlock::instr_iterator NextMII;
   for (; MII != E; MII = NextMII) {
-    NextMII = llvm::next(MII);
+    NextMII = std::next(MII);
 
     MachineInstr *MI = &*MII;
     if (MI->isBundle()) {
@@ -962,7 +957,7 @@ bool Thumb2SizeReduce::ReduceMBB(MachineBasicBlock &MBB) {
 
     if (ReduceMI(MBB, MI, LiveCPSR, IsSelfLoop)) {
       Modified = true;
-      MachineBasicBlock::instr_iterator I = prior(NextMII);
+      MachineBasicBlock::instr_iterator I = std::prev(NextMII);
       MI = &*I;
       // Removing and reinserting the first instruction in a bundle will break
       // up the bundle. Fix the bundling if it was broken.
@@ -980,13 +975,16 @@ bool Thumb2SizeReduce::ReduceMBB(MachineBasicBlock &MBB) {
       MachineOperand *MO = BundleMI->findRegisterDefOperand(ARM::CPSR);
       if (MO && !MO->isDead())
         LiveCPSR = true;
+      MO = BundleMI->findRegisterUseOperand(ARM::CPSR);
+      if (MO && !MO->isKill())
+        LiveCPSR = true;
     }
 
     bool DefCPSR = false;
     LiveCPSR = UpdateCPSRDef(*MI, LiveCPSR, DefCPSR);
     if (MI->isCall()) {
       // Calls don't really set CPSR.
-      CPSRDef = 0;
+      CPSRDef = nullptr;
       HighLatencyCPSR = false;
       IsSelfLoop = false;
     } else if (DefCPSR) {
@@ -1012,8 +1010,8 @@ bool Thumb2SizeReduce::runOnMachineFunction(MachineFunction &MF) {
   AttributeSet FnAttrs = MF.getFunction()->getAttributes();
   OptimizeSize = FnAttrs.hasAttribute(AttributeSet::FunctionIndex,
                                       Attribute::OptimizeForSize);
-  MinimizeSize = FnAttrs.hasAttribute(AttributeSet::FunctionIndex,
-                                      Attribute::MinSize);
+  MinimizeSize =
+      FnAttrs.hasAttribute(AttributeSet::FunctionIndex, Attribute::MinSize);
 
   BlockInfo.clear();
   BlockInfo.resize(MF.getNumBlockIDs());