vendor/llvm/llvm-release_39-r276489

71 files changed, 5140 insertions, 2263 deletions

diff --git a/lib/Target/PowerPC/AsmParser/Makefile b/lib/Target/PowerPC/AsmParser/Makefile
deleted file mode 100644
index c8a8915685ea5..0000000000000
--- a/lib/Target/PowerPC/AsmParser/Makefile
+++ /dev/null
@@ -1,15 +0,0 @@
-##===- lib/Target/PowerPC/AsmParser/Makefile ----------------*- Makefile -*-===##
-#
-#                     The LLVM Compiler Infrastructure
-#
-# This file is distributed under the University of Illinois Open Source
-# License. See LICENSE.TXT for details.
-#
-##===----------------------------------------------------------------------===##
-LEVEL = ../../../..
-LIBRARYNAME = LLVMPowerPCAsmParser
-
-# Hack: we need to include 'main' PowerPC target directory to grab private headers
-CPP.Flags += -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/..
-
-include $(LEVEL)/Makefile.common
diff --git a/lib/Target/PowerPC/AsmParser/PPCAsmParser.cpp b/lib/Target/PowerPC/AsmParser/PPCAsmParser.cpp
index 220c70a485427..4181775fc6dab 100644
--- a/lib/Target/PowerPC/AsmParser/PPCAsmParser.cpp
+++ b/lib/Target/PowerPC/AsmParser/PPCAsmParser.cpp
@@ -1,4 +1,4 @@
-//===-- PPCAsmParser.cpp - Parse PowerPC asm to MCInst instructions ---------===//
+//===-- PPCAsmParser.cpp - Parse PowerPC asm to MCInst instructions -------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -7,12 +7,10 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "MCTargetDesc/PPCMCTargetDesc.h"
 #include "MCTargetDesc/PPCMCExpr.h"
+#include "MCTargetDesc/PPCMCTargetDesc.h"
 #include "PPCTargetStreamer.h"
 #include "llvm/ADT/STLExtras.h"
-#include "llvm/ADT/SmallString.h"
-#include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/MC/MCContext.h"
@@ -22,11 +20,11 @@
 #include "llvm/MC/MCParser/MCAsmLexer.h"
 #include "llvm/MC/MCParser/MCAsmParser.h"
 #include "llvm/MC/MCParser/MCParsedAsmOperand.h"
+#include "llvm/MC/MCParser/MCTargetAsmParser.h"
 #include "llvm/MC/MCRegisterInfo.h"
 #include "llvm/MC/MCStreamer.h"
-#include "llvm/MC/MCSymbolELF.h"
 #include "llvm/MC/MCSubtargetInfo.h"
-#include "llvm/MC/MCTargetAsmParser.h"
+#include "llvm/MC/MCSymbolELF.h"
 #include "llvm/Support/SourceMgr.h"
 #include "llvm/Support/TargetRegistry.h"
 #include "llvm/Support/raw_ostream.h"
@@ -294,7 +292,7 @@ public:
                const MCInstrInfo &MII, const MCTargetOptions &Options)
     : MCTargetAsmParser(Options, STI), MII(MII) {
     // Check for 64-bit vs. 32-bit pointer mode.
-    Triple TheTriple(STI.getTargetTriple());
+    const Triple &TheTriple = STI.getTargetTriple();
     IsPPC64 = (TheTriple.getArch() == Triple::ppc64 ||
                TheTriple.getArch() == Triple::ppc64le);
     IsDarwin = TheTriple.isMacOSX();
@@ -378,6 +376,10 @@ public:
     }
   }
 
+  // Disable use of sized deallocation due to overallocation of PPCOperand
+  // objects in CreateTokenWithStringCopy.
+  void operator delete(void *p) { ::operator delete(p); }
+
   /// getStartLoc - Get the location of the first token of this operand.
   SMLoc getStartLoc() const override { return StartLoc; }
 
@@ -392,13 +394,15 @@ public:
     return Imm.Val;
   }
   int64_t getImmS16Context() const {
-    assert((Kind == Immediate || Kind == ContextImmediate) && "Invalid access!");
+    assert((Kind == Immediate || Kind == ContextImmediate) &&
+           "Invalid access!");
     if (Kind == Immediate)
       return Imm.Val;
     return static_cast<int16_t>(Imm.Val);
   }
   int64_t getImmU16Context() const {
-    assert((Kind == Immediate || Kind == ContextImmediate) && "Invalid access!");
+    assert((Kind == Immediate || Kind == ContextImmediate) &&
+           "Invalid access!");
     return Imm.Val;
   }
 
@@ -443,7 +447,9 @@ public:
   }
 
   bool isToken() const override { return Kind == Token; }
-  bool isImm() const override { return Kind == Immediate || Kind == Expression; }
+  bool isImm() const override {
+    return Kind == Immediate || Kind == Expression;
+  }
   bool isU1Imm() const { return Kind == Immediate && isUInt<1>(getImm()); }
   bool isU2Imm() const { return Kind == Immediate && isUInt<2>(getImm()); }
   bool isU3Imm() const { return Kind == Immediate && isUInt<3>(getImm()); }
@@ -454,13 +460,15 @@ public:
   bool isU6ImmX2() const { return Kind == Immediate &&
                                   isUInt<6>(getImm()) &&
                                   (getImm() & 1) == 0; }
+  bool isU7Imm() const { return Kind == Immediate && isUInt<7>(getImm()); }
   bool isU7ImmX4() const { return Kind == Immediate &&
                                   isUInt<7>(getImm()) &&
                                   (getImm() & 3) == 0; }
+  bool isU8Imm() const { return Kind == Immediate && isUInt<8>(getImm()); }
   bool isU8ImmX8() const { return Kind == Immediate &&
                                   isUInt<8>(getImm()) &&
                                   (getImm() & 7) == 0; }
-  
+
   bool isU10Imm() const { return Kind == Immediate && isUInt<10>(getImm()); }
   bool isU12Imm() const { return Kind == Immediate && isUInt<12>(getImm()); }
   bool isU16Imm() const {
@@ -488,6 +496,9 @@ public:
   bool isS16ImmX4() const { return Kind == Expression ||
                                    (Kind == Immediate && isInt<16>(getImm()) &&
                                     (getImm() & 3) == 0); }
+  bool isS16ImmX16() const { return Kind == Expression ||
+                                    (Kind == Immediate && isInt<16>(getImm()) &&
+                                     (getImm() & 15) == 0); }
   bool isS17Imm() const {
     switch (Kind) {
       case Expression:
@@ -521,7 +532,9 @@ public:
                                  (Kind == Immediate && isInt<16>(getImm()) &&
                                   (getImm() & 3) == 0); }
   bool isRegNumber() const { return Kind == Immediate && isUInt<5>(getImm()); }
-  bool isVSRegNumber() const { return Kind == Immediate && isUInt<6>(getImm()); }
+  bool isVSRegNumber() const {
+    return Kind == Immediate && isUInt<6>(getImm());
+  }
   bool isCCRegNumber() const { return (Kind == Expression
                                        && isUInt<3>(getExprCRVal())) ||
                                       (Kind == Immediate
@@ -1190,6 +1203,29 @@ void PPCAsmParser::ProcessInstruction(MCInst &Inst,
     }
     break;
   }
+  case PPC::CP_COPYx:
+  case PPC::CP_COPY_FIRST: {
+    MCInst TmpInst;
+    TmpInst.setOpcode(PPC::CP_COPY);
+    TmpInst.addOperand(Inst.getOperand(0));
+    TmpInst.addOperand(Inst.getOperand(1));
+    TmpInst.addOperand(MCOperand::createImm(Opcode == PPC::CP_COPYx ? 0 : 1));
+
+    Inst = TmpInst;
+    break;
+  }
+  case PPC::CP_PASTEx :
+  case PPC::CP_PASTE_LAST: {
+    MCInst TmpInst;
+    TmpInst.setOpcode(Opcode == PPC::CP_PASTEx ?
+                      PPC::CP_PASTE : PPC::CP_PASTEo);
+    TmpInst.addOperand(Inst.getOperand(0));
+    TmpInst.addOperand(Inst.getOperand(1));
+    TmpInst.addOperand(MCOperand::createImm(Opcode == PPC::CP_PASTEx ? 0 : 1));
+
+    Inst = TmpInst;
+    break;
+  }
   }
 }
 
@@ -1454,8 +1490,8 @@ ParseExpression(const MCExpr *&EVal) {
 /// This differs from the default "parseExpression" in that it handles detection
 /// of the \code hi16(), ha16() and lo16() \endcode modifiers.  At present,
 /// parseExpression() doesn't recognise the modifiers when in the Darwin/MachO
-/// syntax form so it is done here.  TODO: Determine if there is merit in arranging
-/// for this to be done at a higher level.
+/// syntax form so it is done here.  TODO: Determine if there is merit in
+/// arranging for this to be done at a higher level.
 bool PPCAsmParser::
 ParseDarwinExpression(const MCExpr *&EVal) {
   MCAsmParser &Parser = getParser();
@@ -1674,7 +1710,7 @@ bool PPCAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
   while (getLexer().isNot(AsmToken::EndOfStatement) &&
          getLexer().is(AsmToken::Comma)) {
     // Consume the comma token
-    getLexer().Lex();
+    Lex();
 
     // Parse the next operand
     if (ParseOperand(Operands))
diff --git a/lib/Target/PowerPC/CMakeLists.txt b/lib/Target/PowerPC/CMakeLists.txt
index c31ababafbe7c..4842c3b7a656a 100644
--- a/lib/Target/PowerPC/CMakeLists.txt
+++ b/lib/Target/PowerPC/CMakeLists.txt
@@ -16,6 +16,7 @@ add_llvm_target(PowerPCCodeGen
   PPCBoolRetToInt.cpp
   PPCAsmPrinter.cpp
   PPCBranchSelector.cpp
+  PPCCCState.cpp
   PPCCTRLoops.cpp
   PPCHazardRecognizers.cpp
   PPCInstrInfo.cpp
@@ -24,12 +25,12 @@ add_llvm_target(PowerPCCodeGen
   PPCEarlyReturn.cpp
   PPCFastISel.cpp
   PPCFrameLowering.cpp
-  PPCLoopDataPrefetch.cpp
   PPCLoopPreIncPrep.cpp
   PPCMCInstLower.cpp
   PPCMachineFunctionInfo.cpp
   PPCMIPeephole.cpp
   PPCRegisterInfo.cpp
+  PPCQPXLoadSplat.cpp
   PPCSubtarget.cpp
   PPCTargetMachine.cpp
   PPCTargetObjectFile.cpp
diff --git a/lib/Target/PowerPC/Disassembler/Makefile b/lib/Target/PowerPC/Disassembler/Makefile
deleted file mode 100644
index 86e3b4752207f..0000000000000
--- a/lib/Target/PowerPC/Disassembler/Makefile
+++ /dev/null
@@ -1,16 +0,0 @@
-##===-- lib/Target/PowerPC/Disassembler/Makefile -----------*- Makefile -*-===##
-#
-#                     The LLVM Compiler Infrastructure
-#
-# This file is distributed under the University of Illinois Open Source
-# License. See LICENSE.TXT for details.
-#
-##===----------------------------------------------------------------------===##
-
-LEVEL = ../../../..
-LIBRARYNAME = LLVMPowerPCDisassembler
-
-# Hack: we need to include 'main' PPC target directory to grab private headers
-CPP.Flags += -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/..
-
-include $(LEVEL)/Makefile.common
diff --git a/lib/Target/PowerPC/Disassembler/PPCDisassembler.cpp b/lib/Target/PowerPC/Disassembler/PPCDisassembler.cpp
index 1fc84fb765512..6ea4fb1bfbc33 100644
--- a/lib/Target/PowerPC/Disassembler/PPCDisassembler.cpp
+++ b/lib/Target/PowerPC/Disassembler/PPCDisassembler.cpp
@@ -8,7 +8,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "PPC.h"
-#include "llvm/MC/MCDisassembler.h"
+#include "llvm/MC/MCDisassembler/MCDisassembler.h"
 #include "llvm/MC/MCFixedLenDisassembler.h"
 #include "llvm/MC/MCInst.h"
 #include "llvm/MC/MCSubtargetInfo.h"
@@ -368,6 +368,21 @@ static DecodeStatus decodeMemRIXOperands(MCInst &Inst, uint64_t Imm,
   return MCDisassembler::Success;
 }
 
+static DecodeStatus decodeMemRIX16Operands(MCInst &Inst, uint64_t Imm,
+                                         int64_t Address, const void *Decoder) {
+  // Decode the memrix16 field (imm, reg), which has the low 12-bits as the
+  // displacement with 16-byte aligned, and the next 5 bits as the register #.
+
+  uint64_t Base = Imm >> 12;
+  uint64_t Disp = Imm & 0xFFF;
+
+  assert(Base < 32 && "Invalid base register");
+
+  Inst.addOperand(MCOperand::createImm(SignExtend64<16>(Disp << 4)));
+  Inst.addOperand(MCOperand::createReg(GP0Regs[Base]));
+  return MCDisassembler::Success;
+}
+
 static DecodeStatus decodeCRBitMOperand(MCInst &Inst, uint64_t Imm,
                                         int64_t Address, const void *Decoder) {
   // The cr bit encoding is 0x80 >> cr_reg_num.
diff --git a/lib/Target/PowerPC/InstPrinter/Makefile b/lib/Target/PowerPC/InstPrinter/Makefile
deleted file mode 100644
index f097e84248ff3..0000000000000
--- a/lib/Target/PowerPC/InstPrinter/Makefile
+++ /dev/null
@@ -1,16 +0,0 @@
-##===- lib/Target/PowerPC/AsmPrinter/Makefile --------------*- Makefile -*-===##
-#
-#                     The LLVM Compiler Infrastructure
-#
-# This file is distributed under the University of Illinois Open Source
-# License. See LICENSE.TXT for details.
-#
-##===----------------------------------------------------------------------===##
-
-LEVEL = ../../../..
-LIBRARYNAME = LLVMPowerPCAsmPrinter
-
-# Hack: we need to include 'main' powerpc target directory to grab private headers
-CPP.Flags += -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/..
-
-include $(LEVEL)/Makefile.common
diff --git a/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp b/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp
index 5e1d22789056a..d9d9b4f180f71 100644
--- a/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp
+++ b/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.cpp
@@ -136,17 +136,6 @@ void PPCInstPrinter::printInst(const MCInst *MI, raw_ostream &O,
     return;
   }
   
-  // For fast-isel, a COPY_TO_REGCLASS may survive this long.  This is
-  // used when converting a 32-bit float to a 64-bit float as part of
-  // conversion to an integer (see PPCFastISel.cpp:SelectFPToI()),
-  // as otherwise we have problems with incorrect register classes
-  // in machine instruction verification.  For now, just avoid trying
-  // to print it as such an instruction has no effect (a 32-bit float
-  // in a register is already in 64-bit form, just with lower
-  // precision).  FIXME: Is there a better solution?
-  if (MI->getOpcode() == TargetOpcode::COPY_TO_REGCLASS)
-    return;
-
   if (!printAliasInstr(MI, O))
     printInstruction(MI, O);
   printAnnotation(O, Annot);
@@ -299,6 +288,20 @@ void PPCInstPrinter::printU6ImmOperand(const MCInst *MI, unsigned OpNo,
   O << (unsigned int)Value;
 }
 
+void PPCInstPrinter::printU7ImmOperand(const MCInst *MI, unsigned OpNo,
+                                       raw_ostream &O) {
+  unsigned int Value = MI->getOperand(OpNo).getImm();
+  assert(Value <= 127 && "Invalid u7imm argument!");
+  O << (unsigned int)Value;
+}
+
+void PPCInstPrinter::printU8ImmOperand(const MCInst *MI, unsigned OpNo,
+                                       raw_ostream &O) {
+  unsigned int Value = MI->getOperand(OpNo).getImm();
+  assert(Value <= 255 && "Invalid u8imm argument!");
+  O << (unsigned int)Value;
+}
+
 void PPCInstPrinter::printU10ImmOperand(const MCInst *MI, unsigned OpNo,
                                         raw_ostream &O) {
   unsigned short Value = MI->getOperand(OpNo).getImm();
diff --git a/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.h b/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.h
index 53eb727d0b07f..d0ffeff0247cf 100644
--- a/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.h
+++ b/lib/Target/PowerPC/InstPrinter/PPCInstPrinter.h
@@ -53,6 +53,8 @@ public:
   void printS5ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printU5ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printU6ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
+  void printU7ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
+  void printU8ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printU10ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printU12ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
   void printS16ImmOperand(const MCInst *MI, unsigned OpNo, raw_ostream &O);
diff --git a/lib/Target/PowerPC/MCTargetDesc/Makefile b/lib/Target/PowerPC/MCTargetDesc/Makefile
deleted file mode 100644
index 9db66622cced3..0000000000000
--- a/lib/Target/PowerPC/MCTargetDesc/Makefile
+++ /dev/null
@@ -1,16 +0,0 @@
-##===- lib/Target/PowerPC/TargetDesc/Makefile --------------*- Makefile -*-===##
-#
-#                     The LLVM Compiler Infrastructure
-#
-# This file is distributed under the University of Illinois Open Source
-# License. See LICENSE.TXT for details.
-#
-##===----------------------------------------------------------------------===##
-
-LEVEL = ../../../..
-LIBRARYNAME = LLVMPowerPCDesc
-
-# Hack: we need to include 'main' target directory to grab private headers
-CPP.Flags += -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/..
-
-include $(LEVEL)/Makefile.common
diff --git a/lib/Target/PowerPC/MCTargetDesc/PPCAsmBackend.cpp b/lib/Target/PowerPC/MCTargetDesc/PPCAsmBackend.cpp
index b6dd595ffb0e0..9100ecb4aa371 100644
--- a/lib/Target/PowerPC/MCTargetDesc/PPCAsmBackend.cpp
+++ b/lib/Target/PowerPC/MCTargetDesc/PPCAsmBackend.cpp
@@ -168,8 +168,8 @@ public:
     llvm_unreachable("relaxInstruction() unimplemented");
   }
 
-
-  void relaxInstruction(const MCInst &Inst, MCInst &Res) const override {
+  void relaxInstruction(const MCInst &Inst, const MCSubtargetInfo &STI,
+                        MCInst &Res) const override {
     // FIXME.
     llvm_unreachable("relaxInstruction() unimplemented");
   }
diff --git a/lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp b/lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp
index dd994956870fd..fd279c60f3f59 100644
--- a/lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp
+++ b/lib/Target/PowerPC/MCTargetDesc/PPCELFObjectWriter.cpp
@@ -25,8 +25,8 @@ namespace {
     PPCELFObjectWriter(bool Is64Bit, uint8_t OSABI);
 
   protected:
-    unsigned GetRelocType(const MCValue &Target, const MCFixup &Fixup,
-                          bool IsPCRel) const override;
+    unsigned getRelocType(MCContext &Ctx, const MCValue &Target,
+                          const MCFixup &Fixup, bool IsPCRel) const override;
 
     bool needsRelocateWithSymbol(const MCSymbol &Sym,
                                  unsigned Type) const override;
@@ -66,7 +66,7 @@ static MCSymbolRefExpr::VariantKind getAccessVariant(const MCValue &Target,
   llvm_unreachable("unknown PPCMCExpr kind");
 }
 
-unsigned PPCELFObjectWriter::GetRelocType(const MCValue &Target,
+unsigned PPCELFObjectWriter::getRelocType(MCContext &Ctx, const MCValue &Target,
                                           const MCFixup &Fixup,
                                           bool IsPCRel) const {
   MCSymbolRefExpr::VariantKind Modifier = getAccessVariant(Target, Fixup);
@@ -186,7 +186,7 @@ unsigned PPCELFObjectWriter::GetRelocType(const MCValue &Target,
       case MCSymbolRefExpr::VK_PPC_TOC_HA:
         Type = ELF::R_PPC64_TOC16_HA;
         break;
-      case MCSymbolRefExpr::VK_PPC_TPREL:
+      case MCSymbolRefExpr::VK_TPREL:
         Type = ELF::R_PPC_TPREL16;
         break;
       case MCSymbolRefExpr::VK_PPC_TPREL_LO:
@@ -210,7 +210,7 @@ unsigned PPCELFObjectWriter::GetRelocType(const MCValue &Target,
       case MCSymbolRefExpr::VK_PPC_TPREL_HIGHESTA:
         Type = ELF::R_PPC64_TPREL16_HIGHESTA;
         break;
-      case MCSymbolRefExpr::VK_PPC_DTPREL:
+      case MCSymbolRefExpr::VK_DTPREL:
         Type = ELF::R_PPC64_DTPREL16;
         break;
       case MCSymbolRefExpr::VK_PPC_DTPREL_LO:
@@ -319,13 +319,13 @@ unsigned PPCELFObjectWriter::GetRelocType(const MCValue &Target,
       case MCSymbolRefExpr::VK_PPC_TOC_LO:
         Type = ELF::R_PPC64_TOC16_LO_DS;
         break;
-      case MCSymbolRefExpr::VK_PPC_TPREL:
+      case MCSymbolRefExpr::VK_TPREL:
         Type = ELF::R_PPC64_TPREL16_DS;
         break;
       case MCSymbolRefExpr::VK_PPC_TPREL_LO:
         Type = ELF::R_PPC64_TPREL16_LO_DS;
         break;
-      case MCSymbolRefExpr::VK_PPC_DTPREL:
+      case MCSymbolRefExpr::VK_DTPREL:
         Type = ELF::R_PPC64_DTPREL16_DS;
         break;
       case MCSymbolRefExpr::VK_PPC_DTPREL_LO:
@@ -380,10 +380,10 @@ unsigned PPCELFObjectWriter::GetRelocType(const MCValue &Target,
       case MCSymbolRefExpr::VK_PPC_DTPMOD:
         Type = ELF::R_PPC64_DTPMOD64;
         break;
-      case MCSymbolRefExpr::VK_PPC_TPREL:
+      case MCSymbolRefExpr::VK_TPREL:
         Type = ELF::R_PPC64_TPREL64;
         break;
-      case MCSymbolRefExpr::VK_PPC_DTPREL:
+      case MCSymbolRefExpr::VK_DTPREL:
         Type = ELF::R_PPC64_DTPREL64;
         break;
       }
diff --git a/lib/Target/PowerPC/MCTargetDesc/PPCMCCodeEmitter.cpp b/lib/Target/PowerPC/MCTargetDesc/PPCMCCodeEmitter.cpp
index b7291561c75db..e7b2d8369f2f8 100644
--- a/lib/Target/PowerPC/MCTargetDesc/PPCMCCodeEmitter.cpp
+++ b/lib/Target/PowerPC/MCTargetDesc/PPCMCCodeEmitter.cpp
@@ -69,6 +69,9 @@ public:
   unsigned getMemRIXEncoding(const MCInst &MI, unsigned OpNo,
                              SmallVectorImpl<MCFixup> &Fixups,
                              const MCSubtargetInfo &STI) const;
+  unsigned getMemRIX16Encoding(const MCInst &MI, unsigned OpNo,
+                               SmallVectorImpl<MCFixup> &Fixups,
+                               const MCSubtargetInfo &STI) const;
   unsigned getSPE8DisEncoding(const MCInst &MI, unsigned OpNo,
                               SmallVectorImpl<MCFixup> &Fixups,
                               const MCSubtargetInfo &STI) const;
@@ -102,19 +105,16 @@ public:
   void encodeInstruction(const MCInst &MI, raw_ostream &OS,
                          SmallVectorImpl<MCFixup> &Fixups,
                          const MCSubtargetInfo &STI) const override {
-    // For fast-isel, a float COPY_TO_REGCLASS can survive this long.
-    // It's just a nop to keep the register classes happy, so don't
-    // generate anything.
     unsigned Opcode = MI.getOpcode();
     const MCInstrDesc &Desc = MCII.get(Opcode);
-    if (Opcode == TargetOpcode::COPY_TO_REGCLASS)
-      return;
 
     uint64_t Bits = getBinaryCodeForInstr(MI, Fixups, STI);
 
     // Output the constant in big/little endian byte order.
     unsigned Size = Desc.getSize();
     switch (Size) {
+    case 0:
+      break;
     case 4:
       if (IsLittleEndian) {
         support::endian::Writer<support::little>(OS).write<uint32_t>(Bits);
@@ -249,6 +249,19 @@ unsigned PPCMCCodeEmitter::getMemRIXEncoding(const MCInst &MI, unsigned OpNo,
   return RegBits;
 }
 
+unsigned PPCMCCodeEmitter::getMemRIX16Encoding(const MCInst &MI, unsigned OpNo,
+                                       SmallVectorImpl<MCFixup> &Fixups,
+                                       const MCSubtargetInfo &STI) const {
+  // Encode (imm, reg) as a memrix16, which has the low 12-bits as the
+  // displacement and the next 5 bits as the register #.
+  assert(MI.getOperand(OpNo+1).isReg());
+  unsigned RegBits = getMachineOpValue(MI, MI.getOperand(OpNo+1), Fixups, STI) << 12;
+
+  const MCOperand &MO = MI.getOperand(OpNo);
+  assert(MO.isImm());
+
+  return ((getMachineOpValue(MI, MO, Fixups, STI) >> 4) & 0xFFF) | RegBits;
+}
 
 unsigned PPCMCCodeEmitter::getSPE8DisEncoding(const MCInst &MI, unsigned OpNo,
                                               SmallVectorImpl<MCFixup> &Fixups,
diff --git a/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp b/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp
index 30f232a9a91e0..c9074448fe45e 100644
--- a/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp
+++ b/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp
@@ -15,7 +15,6 @@
 #include "InstPrinter/PPCInstPrinter.h"
 #include "PPCMCAsmInfo.h"
 #include "PPCTargetStreamer.h"
-#include "llvm/MC/MCCodeGenInfo.h"
 #include "llvm/MC/MCContext.h"
 #include "llvm/MC/MCELFStreamer.h"
 #include "llvm/MC/MCExpr.h"
@@ -87,24 +86,13 @@ static MCAsmInfo *createPPCMCAsmInfo(const MCRegisterInfo &MRI,
   return MAI;
 }
 
-static MCCodeGenInfo *createPPCMCCodeGenInfo(const Triple &TT, Reloc::Model RM,
-                                             CodeModel::Model CM,
-                                             CodeGenOpt::Level OL) {
-  MCCodeGenInfo *X = new MCCodeGenInfo();
-
-  if (RM == Reloc::Default) {
-    if (TT.isOSDarwin())
-      RM = Reloc::DynamicNoPIC;
-    else
-      RM = Reloc::Static;
-  }
+static void adjustCodeGenOpts(const Triple &TT, Reloc::Model RM,
+                              CodeModel::Model &CM) {
   if (CM == CodeModel::Default) {
     if (!TT.isOSDarwin() &&
         (TT.getArch() == Triple::ppc64 || TT.getArch() == Triple::ppc64le))
       CM = CodeModel::Medium;
   }
-  X->initMCCodeGenInfo(RM, CM, OL);
-  return X;
 }
 
 namespace {
@@ -245,7 +233,7 @@ extern "C" void LLVMInitializePowerPCTargetMC() {
     RegisterMCAsmInfoFn C(*T, createPPCMCAsmInfo);
 
     // Register the MC codegen info.
-    TargetRegistry::RegisterMCCodeGenInfo(*T, createPPCMCCodeGenInfo);
+    TargetRegistry::registerMCAdjustCodeGenOpts(*T, adjustCodeGenOpts);
 
     // Register the MC instruction info.
     TargetRegistry::RegisterMCInstrInfo(*T, createPPCMCInstrInfo);
diff --git a/lib/Target/PowerPC/MCTargetDesc/PPCMachObjectWriter.cpp b/lib/Target/PowerPC/MCTargetDesc/PPCMachObjectWriter.cpp
index b54a0e1b86b14..1f38a8c947e73 100644
--- a/lib/Target/PowerPC/MCTargetDesc/PPCMachObjectWriter.cpp
+++ b/lib/Target/PowerPC/MCTargetDesc/PPCMachObjectWriter.cpp
@@ -79,7 +79,7 @@ static unsigned getFixupKindLog2Size(unsigned Kind) {
 }
 
 /// Translates generic PPC fixup kind to Mach-O/PPC relocation type enum.
-/// Outline based on PPCELFObjectWriter::GetRelocType().
+/// Outline based on PPCELFObjectWriter::getRelocType().
 static unsigned getRelocType(const MCValue &Target,
                              const MCFixupKind FixupKind, // from
                                                           // Fixup.getKind()
diff --git a/lib/Target/PowerPC/Makefile b/lib/Target/PowerPC/Makefile
deleted file mode 100644
index cf516f4e5ec93..0000000000000
--- a/lib/Target/PowerPC/Makefile
+++ /dev/null
@@ -1,24 +0,0 @@
-##===- lib/Target/PowerPC/Makefile -------------------------*- Makefile -*-===##
-#
-#                     The LLVM Compiler Infrastructure
-#
-# This file is distributed under the University of Illinois Open Source
-# License. See LICENSE.TXT for details.
-#
-##===----------------------------------------------------------------------===##
-
-LEVEL = ../../..
-LIBRARYNAME = LLVMPowerPCCodeGen
-TARGET = PPC
-
-# Make sure that tblgen is run, first thing.
-BUILT_SOURCES = PPCGenRegisterInfo.inc PPCGenAsmMatcher.inc \
-                PPCGenAsmWriter.inc  \
-                PPCGenInstrInfo.inc PPCGenDAGISel.inc \
-                PPCGenSubtargetInfo.inc PPCGenCallingConv.inc \
-                PPCGenMCCodeEmitter.inc PPCGenFastISel.inc \
-                PPCGenDisassemblerTables.inc
-
-DIRS = AsmParser Disassembler InstPrinter TargetInfo MCTargetDesc
-
-include $(LEVEL)/Makefile.common
diff --git a/lib/Target/PowerPC/PPC.h b/lib/Target/PowerPC/PPC.h
index a259ed3fd3276..e01f49dce81e7 100644
--- a/lib/Target/PowerPC/PPC.h
+++ b/lib/Target/PowerPC/PPC.h
@@ -16,7 +16,6 @@
 #define LLVM_LIB_TARGET_POWERPC_PPC_H
 
 #include "MCTargetDesc/PPCMCTargetDesc.h"
-#include <string>
 
 // GCC #defines PPC on Linux but we use it as our namespace name
 #undef PPC
@@ -34,7 +33,6 @@ namespace llvm {
 #ifndef NDEBUG
   FunctionPass *createPPCCTRLoopsVerify();
 #endif
-  FunctionPass *createPPCLoopDataPrefetchPass();
   FunctionPass *createPPCLoopPreIncPrepPass(PPCTargetMachine &TM);
   FunctionPass *createPPCTOCRegDepsPass();
   FunctionPass *createPPCEarlyReturnPass();
@@ -43,6 +41,7 @@ namespace llvm {
   FunctionPass *createPPCVSXSwapRemovalPass();
   FunctionPass *createPPCMIPeepholePass();
   FunctionPass *createPPCBranchSelectionPass();
+  FunctionPass *createPPCQPXLoadSplatPass();
   FunctionPass *createPPCISelDag(PPCTargetMachine &TM);
   FunctionPass *createPPCTLSDynamicCallPass();
   FunctionPass *createPPCBoolRetToIntPass();
@@ -60,13 +59,12 @@ namespace llvm {
     //===------------------------------------------------------------------===//
     // PPC Specific MachineOperand flags.
     MO_NO_FLAG,
-    
-    /// MO_PLT_OR_STUB - On a symbol operand "FOO", this indicates that the
-    /// reference is actually to the "FOO$stub" or "FOO@plt" symbol.  This is
-    /// used for calls and jumps to external functions on Tiger and earlier, and
+
+    /// On a symbol operand "FOO", this indicates that the reference is actually
+    /// to "FOO@plt".  This is used for calls and jumps to external functions on
     /// for PIC calls on Linux and ELF systems.
-    MO_PLT_OR_STUB = 1,
-    
+    MO_PLT = 1,
+
     /// MO_PIC_FLAG - If this bit is set, the symbol reference is relative to
     /// the function's picbase, e.g. lo16(symbol-picbase).
     MO_PIC_FLAG = 2,
@@ -74,7 +72,7 @@ namespace llvm {
     /// MO_NLP_FLAG - If this bit is set, the symbol reference is actually to
     /// the non_lazy_ptr for the global, e.g. lo16(symbol$non_lazy_ptr-picbase).
     MO_NLP_FLAG = 4,
-    
+
     /// MO_NLP_HIDDEN_FLAG - If this bit is set, the symbol reference is to a
     /// symbol with hidden visibility.  This causes a different kind of
     /// non-lazy-pointer to be generated.
@@ -93,11 +91,11 @@ namespace llvm {
     /// These values identify relocations on immediates folded
     /// into memory operations.
     MO_DTPREL_LO = 5 << 4,
-    MO_TLSLD_LO  = 6 << 4,
-    MO_TOC_LO    = 7 << 4,
+    MO_TLSLD_LO = 6 << 4,
+    MO_TOC_LO = 7 << 4,
 
     // Symbol for VK_PPC_TLS fixup attached to an ADD instruction
-    MO_TLS       = 8 << 4
+    MO_TLS = 8 << 4
   };
   } // end namespace PPCII
   
diff --git a/lib/Target/PowerPC/PPC.td b/lib/Target/PowerPC/PPC.td
index b03be12cfd97f..b40b530f4c5d2 100644
--- a/lib/Target/PowerPC/PPC.td
+++ b/lib/Target/PowerPC/PPC.td
@@ -37,16 +37,19 @@ def Directive64  : SubtargetFeature<"", "DarwinDirective", "PPC::DIR_64", "">;
 def DirectiveA2  : SubtargetFeature<"", "DarwinDirective", "PPC::DIR_A2", "">;
 def DirectiveE500mc : SubtargetFeature<"", "DarwinDirective",
                                        "PPC::DIR_E500mc", "">;
-def DirectiveE5500  : SubtargetFeature<"", "DarwinDirective", 
+def DirectiveE5500  : SubtargetFeature<"", "DarwinDirective",
                                        "PPC::DIR_E5500", "">;
 def DirectivePwr3: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR3", "">;
 def DirectivePwr4: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR4", "">;
 def DirectivePwr5: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR5", "">;
-def DirectivePwr5x: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR5X", "">;
+def DirectivePwr5x
+    : SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR5X", "">;
 def DirectivePwr6: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR6", "">;
-def DirectivePwr6x: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR6X", "">;
+def DirectivePwr6x
+    : SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR6X", "">;
 def DirectivePwr7: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR7", "">;
 def DirectivePwr8: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR8", "">;
+def DirectivePwr9: SubtargetFeature<"", "DarwinDirective", "PPC::DIR_PWR9", "">;
 
 def Feature64Bit     : SubtargetFeature<"64bit","Has64BitSupport", "true",
                                         "Enable 64-bit instructions">;
@@ -86,8 +89,6 @@ def FeatureFPCVT     : SubtargetFeature<"fpcvt", "HasFPCVT", "true",
   "Enable fc[ft]* (unsigned and single-precision) and lfiwzx instructions">;
 def FeatureISEL      : SubtargetFeature<"isel","HasISEL", "true",
                                         "Enable the isel instruction">;
-def FeaturePOPCNTD   : SubtargetFeature<"popcntd","HasPOPCNTD", "true",
-                                        "Enable the popcnt[dw] instructions">;
 def FeatureBPERMD    : SubtargetFeature<"bpermd", "HasBPERMD", "true",
                                         "Enable the bpermd instruction">;
 def FeatureExtDiv    : SubtargetFeature<"extdiv", "HasExtDiv", "true",
@@ -145,24 +146,43 @@ def FeatureFloat128 :
   SubtargetFeature<"float128", "HasFloat128", "true",
                    "Enable the __float128 data type for IEEE-754R Binary128.",
                    [FeatureVSX]>;
+def FeaturePOPCNTD   : SubtargetFeature<"popcntd","HasPOPCNTD",
+                                        "POPCNTD_Fast",
+                                        "Enable the popcnt[dw] instructions">;
+// Note that for the a2/a2q processor models we should not use popcnt[dw] by
+// default. These processors do support the instructions, but they're
+// microcoded, and the software emulation is about twice as fast.
+def FeatureSlowPOPCNTD : SubtargetFeature<"slow-popcntd","HasPOPCNTD",
+                                          "POPCNTD_Slow",
+                                          "Has slow popcnt[dw] instructions">;
 
 def DeprecatedDST    : SubtargetFeature<"", "DeprecatedDST", "true",
   "Treat vector data stream cache control instructions as deprecated">;
 
-/*  Since new processors generally contain a superset of features of those that
-    came before them, the idea is to make implementations of new processors
-    less error prone and easier to read.
-    Namely:
-        list<SubtargetFeature> Power8FeatureList = ...
-        list<SubtargetFeature> FutureProcessorSpecificFeatureList =
-            [ features that Power8 does not support ]
-        list<SubtargetFeature> FutureProcessorFeatureList =
-            !listconcat(Power8FeatureList, FutureProcessorSpecificFeatureList)
+def FeatureISA3_0 : SubtargetFeature<"isa-v30-instructions", "IsISA3_0",
+                                     "true",
+                                     "Enable instructions added in ISA 3.0.">;
+def FeatureP9Altivec : SubtargetFeature<"power9-altivec", "HasP9Altivec", "true",
+                                        "Enable POWER9 Altivec instructions",
+                                        [FeatureISA3_0, FeatureP8Altivec]>;
+def FeatureP9Vector  : SubtargetFeature<"power9-vector", "HasP9Vector", "true",
+                                        "Enable POWER9 vector instructions",
+                                        [FeatureISA3_0, FeatureP8Vector,
+                                         FeatureP9Altivec]>;
 
-    Makes it explicit and obvious what is new in FutureProcesor vs. Power8 as
-    well as providing a single point of definition if the feature set will be
-    used elsewhere.
-*/
+// Since new processors generally contain a superset of features of those that
+// came before them, the idea is to make implementations of new processors
+// less error prone and easier to read.
+// Namely:
+//     list<SubtargetFeature> Power8FeatureList = ...
+//     list<SubtargetFeature> FutureProcessorSpecificFeatureList =
+//         [ features that Power8 does not support ]
+//     list<SubtargetFeature> FutureProcessorFeatureList =
+//         !listconcat(Power8FeatureList, FutureProcessorSpecificFeatureList)
+
+// Makes it explicit and obvious what is new in FutureProcesor vs. Power8 as
+// well as providing a single point of definition if the feature set will be
+// used elsewhere.
 def ProcessorFeatures {
   list<SubtargetFeature> Power7FeatureList =
       [DirectivePwr7, FeatureAltivec, FeatureVSX,
@@ -180,6 +200,10 @@ def ProcessorFeatures {
        FeatureFusion];
   list<SubtargetFeature> Power8FeatureList =
       !listconcat(Power7FeatureList, Power8SpecificFeatures);
+  list<SubtargetFeature> Power9SpecificFeatures =
+      [FeatureP9Altivec, FeatureP9Vector, FeatureISA3_0];
+  list<SubtargetFeature> Power9FeatureList =
+      !listconcat(Power8FeatureList, Power9SpecificFeatures);
 }
 
 // Note: Future features to add when support is extended to more
@@ -331,16 +355,17 @@ def : ProcessorModel<"a2", PPCA2Model,
                    FeatureFRSQRTE, FeatureFRSQRTES, FeatureRecipPrec,
                    FeatureSTFIWX, FeatureLFIWAX,
                    FeatureFPRND, FeatureFPCVT, FeatureISEL,
-                   FeaturePOPCNTD, FeatureCMPB, FeatureLDBRX, Feature64Bit
-               /*, Feature64BitRegs */, FeatureMFTB]>;
+                   FeatureSlowPOPCNTD, FeatureCMPB, FeatureLDBRX,
+                   Feature64Bit /*, Feature64BitRegs */, FeatureMFTB]>;
 def : ProcessorModel<"a2q", PPCA2Model,
                   [DirectiveA2, FeatureICBT, FeatureBookE, FeatureMFOCRF,
                    FeatureFCPSGN, FeatureFSqrt, FeatureFRE, FeatureFRES,
                    FeatureFRSQRTE, FeatureFRSQRTES, FeatureRecipPrec,
                    FeatureSTFIWX, FeatureLFIWAX,
                    FeatureFPRND, FeatureFPCVT, FeatureISEL,
-                   FeaturePOPCNTD, FeatureCMPB, FeatureLDBRX, Feature64Bit
-               /*, Feature64BitRegs */, FeatureQPX, FeatureMFTB]>;
+                   FeatureSlowPOPCNTD, FeatureCMPB, FeatureLDBRX,
+                   Feature64Bit /*, Feature64BitRegs */, FeatureQPX,
+                   FeatureMFTB]>;
 def : ProcessorModel<"pwr3", G5Model,
                   [DirectivePwr3, FeatureAltivec,
                    FeatureFRES, FeatureFRSQRTE, FeatureMFOCRF,
@@ -377,6 +402,8 @@ def : ProcessorModel<"pwr6x", G5Model,
                    FeatureMFTB, DeprecatedDST]>;
 def : ProcessorModel<"pwr7", P7Model, ProcessorFeatures.Power7FeatureList>;
 def : ProcessorModel<"pwr8", P8Model, ProcessorFeatures.Power8FeatureList>;
+// FIXME: Same as P8 until the POWER9 scheduling info is available
+def : ProcessorModel<"pwr9", P8Model, ProcessorFeatures.Power9FeatureList>;
 def : Processor<"ppc", G3Itineraries, [Directive32, FeatureMFTB]>;
 def : ProcessorModel<"ppc64", G5Model,
                   [Directive64, FeatureAltivec,
diff --git a/lib/Target/PowerPC/PPCAsmPrinter.cpp b/lib/Target/PowerPC/PPCAsmPrinter.cpp
index ec354c209ca04..76c52ab6cf1ef 100644
--- a/lib/Target/PowerPC/PPCAsmPrinter.cpp
+++ b/lib/Target/PowerPC/PPCAsmPrinter.cpp
@@ -25,7 +25,6 @@
 #include "PPCTargetMachine.h"
 #include "PPCTargetStreamer.h"
 #include "llvm/ADT/MapVector.h"
-#include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/CodeGen/AsmPrinter.h"
 #include "llvm/CodeGen/MachineConstantPool.h"
@@ -50,7 +49,6 @@
 #include "llvm/MC/MCSectionMachO.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSymbolELF.h"
-#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ELF.h"
 #include "llvm/Support/ErrorHandling.h"
@@ -82,6 +80,12 @@ public:
 
     MCSymbol *lookUpOrCreateTOCEntry(MCSymbol *Sym);
 
+    virtual bool doInitialization(Module &M) override {
+      if (!TOC.empty())
+        TOC.clear();
+      return AsmPrinter::doInitialization(M);
+    }
+
     void EmitInstruction(const MachineInstr *MI) override;
 
     void printOperand(const MachineInstr *MI, unsigned OpNo, raw_ostream &O);
@@ -138,8 +142,6 @@ public:
 
     bool doFinalization(Module &M) override;
     void EmitStartOfAsmFile(Module &M) override;
-
-    void EmitFunctionStubs(const MachineModuleInfoMachO::SymbolListTy &Stubs);
   };
 } // end of anonymous namespace
 
@@ -195,29 +197,14 @@ void PPCAsmPrinter::printOperand(const MachineInstr *MI, unsigned OpNo,
     MCSymbol *SymToPrint;
 
     // External or weakly linked global variables need non-lazily-resolved stubs
-    if (TM.getRelocationModel() != Reloc::Static &&
-        !GV->isStrongDefinitionForLinker()) {
-      if (!GV->hasHiddenVisibility()) {
-        SymToPrint = getSymbolWithGlobalValueBase(GV, "$non_lazy_ptr");
-        MachineModuleInfoImpl::StubValueTy &StubSym =
-            MMI->getObjFileInfo<MachineModuleInfoMachO>().getGVStubEntry(
-                SymToPrint);
-        if (!StubSym.getPointer())
-          StubSym = MachineModuleInfoImpl::
-            StubValueTy(getSymbol(GV), !GV->hasInternalLinkage());
-      } else if (GV->isDeclaration() || GV->hasCommonLinkage() ||
-                 GV->hasAvailableExternallyLinkage()) {
-        SymToPrint = getSymbolWithGlobalValueBase(GV, "$non_lazy_ptr");
-
-        MachineModuleInfoImpl::StubValueTy &StubSym =
-            MMI->getObjFileInfo<MachineModuleInfoMachO>().getHiddenGVStubEntry(
-                SymToPrint);
-        if (!StubSym.getPointer())
-          StubSym = MachineModuleInfoImpl::
-            StubValueTy(getSymbol(GV), !GV->hasInternalLinkage());
-      } else {
-        SymToPrint = getSymbol(GV);
-      }
+    if (Subtarget->hasLazyResolverStub(GV)) {
+      SymToPrint = getSymbolWithGlobalValueBase(GV, "$non_lazy_ptr");
+      MachineModuleInfoImpl::StubValueTy &StubSym =
+          MMI->getObjFileInfo<MachineModuleInfoMachO>().getGVStubEntry(
+              SymToPrint);
+      if (!StubSym.getPointer())
+        StubSym = MachineModuleInfoImpl::StubValueTy(getSymbol(GV),
+                                                     !GV->hasInternalLinkage());
     } else {
       SymToPrint = getSymbol(GV);
     }
@@ -470,7 +457,7 @@ void PPCAsmPrinter::EmitTlsCall(const MachineInstr *MI,
          "GETtls[ld]ADDR[32] must read GPR3");
 
   if (!Subtarget->isPPC64() && !Subtarget->isDarwin() &&
-      TM.getRelocationModel() == Reloc::PIC_)
+      isPositionIndependent())
     Kind = MCSymbolRefExpr::VK_PLT;
   const MCSymbolRefExpr *TlsRef =
     MCSymbolRefExpr::create(TlsGetAddr, Kind, OutContext);
@@ -597,7 +584,7 @@ void PPCAsmPrinter::EmitInstruction(const MachineInstr *MI) {
     else if (MO.isBlockAddress())
       MOSymbol = GetBlockAddressSymbol(MO.getBlockAddress());
 
-    if (PL == PICLevel::Small) {
+    if (PL == PICLevel::SmallPIC) {
       const MCExpr *Exp =
         MCSymbolRefExpr::create(MOSymbol, MCSymbolRefExpr::VK_GOT,
                                 OutContext);
@@ -1038,10 +1025,10 @@ void PPCLinuxAsmPrinter::EmitStartOfAsmFile(Module &M) {
   }
 
   if (static_cast<const PPCTargetMachine &>(TM).isPPC64() ||
-      TM.getRelocationModel() != Reloc::PIC_)
+      !isPositionIndependent())
     return AsmPrinter::EmitStartOfAsmFile(M);
 
-  if (M.getPICLevel() == PICLevel::Small)
+  if (M.getPICLevel() == PICLevel::SmallPIC)
     return AsmPrinter::EmitStartOfAsmFile(M);
 
   OutStreamer->SwitchSection(OutContext.getELFSection(
@@ -1067,8 +1054,8 @@ void PPCLinuxAsmPrinter::EmitStartOfAsmFile(Module &M) {
 void PPCLinuxAsmPrinter::EmitFunctionEntryLabel() {
   // linux/ppc32 - Normal entry label.
   if (!Subtarget->isPPC64() &&
-      (TM.getRelocationModel() != Reloc::PIC_ ||
-       MF->getFunction()->getParent()->getPICLevel() == PICLevel::Small))
+      (!isPositionIndependent() ||
+       MF->getFunction()->getParent()->getPICLevel() == PICLevel::SmallPIC))
     return AsmPrinter::EmitFunctionEntryLabel();
 
   if (!Subtarget->isPPC64()) {
@@ -1302,8 +1289,10 @@ void PPCDarwinAsmPrinter::EmitStartOfAsmFile(Module &M) {
     "power6",
     "power6x",
     "power7",
+    // FIXME: why is power8 missing here?
     "ppc64",
-    "ppc64le"
+    "ppc64le",
+    "power9"
   };
 
   // Get the numerically largest directive.
@@ -1350,161 +1339,6 @@ void PPCDarwinAsmPrinter::EmitStartOfAsmFile(Module &M) {
   OutStreamer->SwitchSection(getObjFileLowering().getTextSection());
 }
 
-static MCSymbol *GetLazyPtr(MCSymbol *Sym, MCContext &Ctx) {
-  // Remove $stub suffix, add $lazy_ptr.
-  StringRef NoStub = Sym->getName().substr(0, Sym->getName().size()-5);
-  return Ctx.getOrCreateSymbol(NoStub + "$lazy_ptr");
-}
-
-static MCSymbol *GetAnonSym(MCSymbol *Sym, MCContext &Ctx) {
-  // Add $tmp suffix to $stub, yielding $stub$tmp.
-  return Ctx.getOrCreateSymbol(Sym->getName() + "$tmp");
-}
-
-void PPCDarwinAsmPrinter::
-EmitFunctionStubs(const MachineModuleInfoMachO::SymbolListTy &Stubs) {
-  bool isPPC64 = getDataLayout().getPointerSizeInBits() == 64;
-
-  // Construct a local MCSubtargetInfo and shadow EmitToStreamer here.
-  // This is because the MachineFunction won't exist (but have not yet been
-  // freed) and since we're at the global level we can use the default
-  // constructed subtarget.
-  std::unique_ptr<MCSubtargetInfo> STI(TM.getTarget().createMCSubtargetInfo(
-      TM.getTargetTriple().str(), TM.getTargetCPU(),
-      TM.getTargetFeatureString()));
-  auto EmitToStreamer = [&STI] (MCStreamer &S, const MCInst &Inst) {
-    S.EmitInstruction(Inst, *STI);
-  };
-
-  const TargetLoweringObjectFileMachO &TLOFMacho =
-      static_cast<const TargetLoweringObjectFileMachO &>(getObjFileLowering());
-
-  // .lazy_symbol_pointer
-  MCSection *LSPSection = TLOFMacho.getLazySymbolPointerSection();
-
-  // Output stubs for dynamically-linked functions
-  if (TM.getRelocationModel() == Reloc::PIC_) {
-    MCSection *StubSection = OutContext.getMachOSection(
-        "__TEXT", "__picsymbolstub1",
-        MachO::S_SYMBOL_STUBS | MachO::S_ATTR_PURE_INSTRUCTIONS, 32,
-        SectionKind::getText());
-    for (unsigned i = 0, e = Stubs.size(); i != e; ++i) {
-      OutStreamer->SwitchSection(StubSection);
-      EmitAlignment(4);
-
-      MCSymbol *Stub = Stubs[i].first;
-      MCSymbol *RawSym = Stubs[i].second.getPointer();
-      MCSymbol *LazyPtr = GetLazyPtr(Stub, OutContext);
-      MCSymbol *AnonSymbol = GetAnonSym(Stub, OutContext);
-
-      OutStreamer->EmitLabel(Stub);
-      OutStreamer->EmitSymbolAttribute(RawSym, MCSA_IndirectSymbol);
-
-      const MCExpr *Anon = MCSymbolRefExpr::create(AnonSymbol, OutContext);
-      const MCExpr *LazyPtrExpr = MCSymbolRefExpr::create(LazyPtr, OutContext);
-      const MCExpr *Sub =
-        MCBinaryExpr::createSub(LazyPtrExpr, Anon, OutContext);
-
-      // mflr r0
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::MFLR).addReg(PPC::R0));
-      // bcl 20, 31, AnonSymbol
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::BCLalways).addExpr(Anon));
-      OutStreamer->EmitLabel(AnonSymbol);
-      // mflr r11
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::MFLR).addReg(PPC::R11));
-      // addis r11, r11, ha16(LazyPtr - AnonSymbol)
-      const MCExpr *SubHa16 = PPCMCExpr::createHa(Sub, true, OutContext);
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::ADDIS)
-        .addReg(PPC::R11)
-        .addReg(PPC::R11)
-        .addExpr(SubHa16));
-      // mtlr r0
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::MTLR).addReg(PPC::R0));
-
-      // ldu r12, lo16(LazyPtr - AnonSymbol)(r11)
-      // lwzu r12, lo16(LazyPtr - AnonSymbol)(r11)
-      const MCExpr *SubLo16 = PPCMCExpr::createLo(Sub, true, OutContext);
-      EmitToStreamer(*OutStreamer, MCInstBuilder(isPPC64 ? PPC::LDU : PPC::LWZU)
-        .addReg(PPC::R12)
-        .addExpr(SubLo16).addExpr(SubLo16)
-        .addReg(PPC::R11));
-      // mtctr r12
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::MTCTR).addReg(PPC::R12));
-      // bctr
-      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::BCTR));
-
-      OutStreamer->SwitchSection(LSPSection);
-      OutStreamer->EmitLabel(LazyPtr);
-      OutStreamer->EmitSymbolAttribute(RawSym, MCSA_IndirectSymbol);
-
-      MCSymbol *DyldStubBindingHelper =
-        OutContext.getOrCreateSymbol(StringRef("dyld_stub_binding_helper"));
-      if (isPPC64) {
-        // .quad dyld_stub_binding_helper
-        OutStreamer->EmitSymbolValue(DyldStubBindingHelper, 8);
-      } else {
-        // .long dyld_stub_binding_helper
-        OutStreamer->EmitSymbolValue(DyldStubBindingHelper, 4);
-      }
-    }
-    OutStreamer->AddBlankLine();
-    return;
-  }
-
-  MCSection *StubSection = OutContext.getMachOSection(
-      "__TEXT", "__symbol_stub1",
-      MachO::S_SYMBOL_STUBS | MachO::S_ATTR_PURE_INSTRUCTIONS, 16,
-      SectionKind::getText());
-  for (unsigned i = 0, e = Stubs.size(); i != e; ++i) {
-    MCSymbol *Stub = Stubs[i].first;
-    MCSymbol *RawSym = Stubs[i].second.getPointer();
-    MCSymbol *LazyPtr = GetLazyPtr(Stub, OutContext);
-    const MCExpr *LazyPtrExpr = MCSymbolRefExpr::create(LazyPtr, OutContext);
-
-    OutStreamer->SwitchSection(StubSection);
-    EmitAlignment(4);
-    OutStreamer->EmitLabel(Stub);
-    OutStreamer->EmitSymbolAttribute(RawSym, MCSA_IndirectSymbol);
-
-    // lis r11, ha16(LazyPtr)
-    const MCExpr *LazyPtrHa16 =
-      PPCMCExpr::createHa(LazyPtrExpr, true, OutContext);
-    EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::LIS)
-      .addReg(PPC::R11)
-      .addExpr(LazyPtrHa16));
-
-    // ldu r12, lo16(LazyPtr)(r11)
-    // lwzu r12, lo16(LazyPtr)(r11)
-    const MCExpr *LazyPtrLo16 =
-      PPCMCExpr::createLo(LazyPtrExpr, true, OutContext);
-    EmitToStreamer(*OutStreamer, MCInstBuilder(isPPC64 ? PPC::LDU : PPC::LWZU)
-      .addReg(PPC::R12)
-      .addExpr(LazyPtrLo16).addExpr(LazyPtrLo16)
-      .addReg(PPC::R11));
-
-    // mtctr r12
-    EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::MTCTR).addReg(PPC::R12));
-    // bctr
-    EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::BCTR));
-
-    OutStreamer->SwitchSection(LSPSection);
-    OutStreamer->EmitLabel(LazyPtr);
-    OutStreamer->EmitSymbolAttribute(RawSym, MCSA_IndirectSymbol);
-
-    MCSymbol *DyldStubBindingHelper =
-      OutContext.getOrCreateSymbol(StringRef("dyld_stub_binding_helper"));
-    if (isPPC64) {
-      // .quad dyld_stub_binding_helper
-      OutStreamer->EmitSymbolValue(DyldStubBindingHelper, 8);
-    } else {
-      // .long dyld_stub_binding_helper
-      OutStreamer->EmitSymbolValue(DyldStubBindingHelper, 4);
-    }
-  }
-
-  OutStreamer->AddBlankLine();
-}
-
 bool PPCDarwinAsmPrinter::doFinalization(Module &M) {
   bool isPPC64 = getDataLayout().getPointerSizeInBits() == 64;
 
@@ -1514,10 +1348,6 @@ bool PPCDarwinAsmPrinter::doFinalization(Module &M) {
   MachineModuleInfoMachO &MMIMacho =
       MMI->getObjFileInfo<MachineModuleInfoMachO>();
 
-  MachineModuleInfoMachO::SymbolListTy Stubs = MMIMacho.GetFnStubList();
-  if (!Stubs.empty())
-    EmitFunctionStubs(Stubs);
-
   if (MAI->doesSupportExceptionHandling() && MMI) {
     // Add the (possibly multiple) personalities to the set of global values.
     // Only referenced functions get into the Personalities list.
@@ -1534,7 +1364,7 @@ bool PPCDarwinAsmPrinter::doFinalization(Module &M) {
   }
 
   // Output stubs for dynamically-linked functions.
-  Stubs = MMIMacho.GetGVStubList();
+  MachineModuleInfoMachO::SymbolListTy Stubs = MMIMacho.GetGVStubList();
 
   // Output macho stubs for external and common global variables.
   if (!Stubs.empty()) {
@@ -1568,25 +1398,6 @@ bool PPCDarwinAsmPrinter::doFinalization(Module &M) {
     OutStreamer->AddBlankLine();
   }
 
-  Stubs = MMIMacho.GetHiddenGVStubList();
-  if (!Stubs.empty()) {
-    OutStreamer->SwitchSection(getObjFileLowering().getDataSection());
-    EmitAlignment(isPPC64 ? 3 : 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),
-                             isPPC64 ? 8 : 4/*size*/);
-    }
-
-    Stubs.clear();
-    OutStreamer->AddBlankLine();
-  }
-
   // Funny Darwin hack: This flag tells the linker that no global symbols
   // contain code that falls through to other global symbols (e.g. the obvious
   // implementation of multiple entry points).  If this doesn't occur, the
diff --git a/lib/Target/PowerPC/PPCBoolRetToInt.cpp b/lib/Target/PowerPC/PPCBoolRetToInt.cpp
index 7920240bc2b9e..bfb4d8756901b 100644
--- a/lib/Target/PowerPC/PPCBoolRetToInt.cpp
+++ b/lib/Target/PowerPC/PPCBoolRetToInt.cpp
@@ -119,7 +119,7 @@ class PPCBoolRetToInt : public FunctionPass {
             Promotable.insert(P);
 
     SmallVector<const PHINode *, 8> ToRemove;
-    for (const auto &P : Promotable) {
+    for (const PHINode *P : Promotable) {
       // Condition 2 and 3
       auto IsValidUser = [] (const Value *V) -> bool {
         return isa<ReturnInst>(V) || isa<CallInst>(V) || isa<PHINode>(V) ||
@@ -146,7 +146,7 @@ class PPCBoolRetToInt : public FunctionPass {
         Promotable.erase(User);
       ToRemove.clear();
 
-      for (const auto &P : Promotable) {
+      for (const PHINode *P : Promotable) {
         // Condition 4 and 5
         const auto &Users = P->users();
         const auto &Operands = P->operands();
@@ -168,6 +168,9 @@ class PPCBoolRetToInt : public FunctionPass {
   }
 
   bool runOnFunction(Function &F) {
+    if (skipFunction(F))
+      return false;
+
     PHINodeSet PromotablePHINodes = getPromotablePHINodes(F);
     B2IMap Bool2IntMap;
     bool Changed = false;
@@ -199,11 +202,11 @@ class PPCBoolRetToInt : public FunctionPass {
     // Presently, we only know how to handle PHINode, Constant, and Arguments.
     // Potentially, bitwise operations (AND, OR, XOR, NOT) and sign extension
     // could also be handled in the future.
-    for (const auto &V : Defs)
+    for (Value *V : Defs)
       if (!isa<PHINode>(V) && !isa<Constant>(V) && !isa<Argument>(V))
         return false;
 
-    for (const auto &V : Defs)
+    for (Value *V : Defs)
       if (const PHINode *P = dyn_cast<PHINode>(V))
         if (!PromotablePHINodes.count(P))
           return false;
@@ -214,7 +217,7 @@ class PPCBoolRetToInt : public FunctionPass {
       ++NumBoolCallPromotion;
     ++NumBoolToIntPromotion;
 
-    for (const auto &V : Defs)
+    for (Value *V : Defs)
       if (!BoolToIntMap.count(V))
         BoolToIntMap[V] = translate(V);
 
diff --git a/lib/Target/PowerPC/PPCBranchSelector.cpp b/lib/Target/PowerPC/PPCBranchSelector.cpp
index 73a5305197ad7..4d63c5b5703c4 100644
--- a/lib/Target/PowerPC/PPCBranchSelector.cpp
+++ b/lib/Target/PowerPC/PPCBranchSelector.cpp
@@ -46,6 +46,11 @@ namespace {
 
     bool runOnMachineFunction(MachineFunction &Fn) override;
 
+    MachineFunctionProperties getRequiredProperties() const override {
+      return MachineFunctionProperties().set(
+          MachineFunctionProperties::Property::AllVRegsAllocated);
+    }
+
     const char *getPassName() const override {
       return "PowerPC Branch Selector";
     }
@@ -102,10 +107,9 @@ bool PPCBSel::runOnMachineFunction(MachineFunction &Fn) {
     }
 
     unsigned BlockSize = 0;
-    for (MachineBasicBlock::iterator MBBI = MBB->begin(), EE = MBB->end();
-         MBBI != EE; ++MBBI)
-      BlockSize += TII->GetInstSizeInBytes(MBBI);
-    
+    for (MachineInstr &MI : *MBB)
+      BlockSize += TII->GetInstSizeInBytes(MI);
+
     BlockSizes[MBB->getNumber()] = BlockSize;
     FuncSize += BlockSize;
   }
@@ -151,7 +155,7 @@ bool PPCBSel::runOnMachineFunction(MachineFunction &Fn) {
           Dest = I->getOperand(0).getMBB();
 
         if (!Dest) {
-          MBBStartOffset += TII->GetInstSizeInBytes(I);
+          MBBStartOffset += TII->GetInstSizeInBytes(*I);
           continue;
         }
         
@@ -234,4 +238,3 @@ bool PPCBSel::runOnMachineFunction(MachineFunction &Fn) {
   BlockSizes.clear();
   return true;
 }
-
diff --git a/lib/Target/PowerPC/PPCCCState.cpp b/lib/Target/PowerPC/PPCCCState.cpp
new file mode 100644
index 0000000000000..5510a95430f51
--- /dev/null
+++ b/lib/Target/PowerPC/PPCCCState.cpp
@@ -0,0 +1,36 @@
+//===---- PPCCCState.cpp - CCState with PowerPC specific extensions ---------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "PPCCCState.h"
+#include "PPCSubtarget.h"
+#include "llvm/IR/Module.h"
+using namespace llvm;
+
+// Identify lowered values that originated from ppcf128 arguments and record
+// this.
+void PPCCCState::PreAnalyzeCallOperands(
+    const SmallVectorImpl<ISD::OutputArg> &Outs) {
+  for (const auto &I : Outs) {
+    if (I.ArgVT == llvm::MVT::ppcf128)
+      OriginalArgWasPPCF128.push_back(true);
+    else
+      OriginalArgWasPPCF128.push_back(false);
+  }
+}
+
+void PPCCCState::PreAnalyzeFormalArguments(
+    const SmallVectorImpl<ISD::InputArg> &Ins) {
+  for (const auto &I : Ins) {
+    if (I.ArgVT == llvm::MVT::ppcf128) {
+      OriginalArgWasPPCF128.push_back(true);
+    } else {
+      OriginalArgWasPPCF128.push_back(false);
+    }
+  }
+}
\ No newline at end of file
diff --git a/lib/Target/PowerPC/PPCCCState.h b/lib/Target/PowerPC/PPCCCState.h
new file mode 100644
index 0000000000000..9be9f11dbea38
--- /dev/null
+++ b/lib/Target/PowerPC/PPCCCState.h
@@ -0,0 +1,42 @@
+//===---- PPCCCState.h - CCState with PowerPC specific extensions -----------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef PPCCCSTATE_H
+#define PPCCCSTATE_H
+
+#include "PPCISelLowering.h"
+#include "llvm/ADT/SmallVector.h"
+#include "llvm/CodeGen/CallingConvLower.h"
+
+namespace llvm {
+
+class PPCCCState : public CCState {
+public:
+
+  void
+  PreAnalyzeCallOperands(const SmallVectorImpl<ISD::OutputArg> &Outs);
+  void
+  PreAnalyzeFormalArguments(const SmallVectorImpl<ISD::InputArg> &Ins);
+
+private:
+
+  // Records whether the value has been lowered from an ppcf128.
+  SmallVector<bool, 4> OriginalArgWasPPCF128;
+
+public:
+  PPCCCState(CallingConv::ID CC, bool isVarArg, MachineFunction &MF,
+             SmallVectorImpl<CCValAssign> &locs, LLVMContext &C)
+        : CCState(CC, isVarArg, MF, locs, C) {}
+
+  bool WasOriginalArgPPCF128(unsigned ValNo) { return OriginalArgWasPPCF128[ValNo]; }
+  void clearWasPPCF128() { OriginalArgWasPPCF128.clear(); }
+};
+}
+
+#endif
diff --git a/lib/Target/PowerPC/PPCCTRLoops.cpp b/lib/Target/PowerPC/PPCCTRLoops.cpp
index b6ac4d54d4c77..8752266359175 100644
--- a/lib/Target/PowerPC/PPCCTRLoops.cpp
+++ b/lib/Target/PowerPC/PPCCTRLoops.cpp
@@ -54,9 +54,6 @@
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #endif
 
-#include <algorithm>
-#include <vector>
-
 using namespace llvm;
 
 #define DEBUG_TYPE "ctrloops"
@@ -169,6 +166,9 @@ FunctionPass *llvm::createPPCCTRLoopsVerify() {
 #endif // NDEBUG
 
 bool PPCCTRLoops::runOnFunction(Function &F) {
+  if (skipFunction(F))
+    return false;
+
   LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
   SE = &getAnalysis<ScalarEvolutionWrapperPass>().getSE();
   DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
@@ -245,7 +245,7 @@ bool PPCCTRLoops::mightUseCTR(const Triple &TT, BasicBlock *BB) {
       if (Function *F = CI->getCalledFunction()) {
         // Most intrinsics don't become function calls, but some might.
         // sin, cos, exp and log are always calls.
-        unsigned Opcode;
+        unsigned Opcode = 0;
         if (F->getIntrinsicID() != Intrinsic::not_intrinsic) {
           switch (F->getIntrinsicID()) {
           default: continue;
@@ -305,6 +305,8 @@ bool PPCCTRLoops::mightUseCTR(const Triple &TT, BasicBlock *BB) {
           case Intrinsic::rint:      Opcode = ISD::FRINT;      break;
           case Intrinsic::nearbyint: Opcode = ISD::FNEARBYINT; break;
           case Intrinsic::round:     Opcode = ISD::FROUND;     break;
+          case Intrinsic::minnum:    Opcode = ISD::FMINNUM;    break;
+          case Intrinsic::maxnum:    Opcode = ISD::FMAXNUM;    break;
           }
         }
 
@@ -364,8 +366,18 @@ bool PPCCTRLoops::mightUseCTR(const Triple &TT, BasicBlock *BB) {
           case LibFunc::truncf:
           case LibFunc::truncl:
             Opcode = ISD::FTRUNC; break;
+          case LibFunc::fmin:
+          case LibFunc::fminf:
+          case LibFunc::fminl:
+            Opcode = ISD::FMINNUM; break;
+          case LibFunc::fmax:
+          case LibFunc::fmaxf:
+          case LibFunc::fmaxl:
+            Opcode = ISD::FMAXNUM; break;
           }
+        }
 
+        if (Opcode) {
           auto &DL = CI->getModule()->getDataLayout();
           MVT VTy = TLI->getSimpleValueType(DL, CI->getArgOperand(0)->getType(),
                                             true);
@@ -422,6 +434,25 @@ bool PPCCTRLoops::mightUseCTR(const Triple &TT, BasicBlock *BB) {
       if (SI->getNumCases() + 1 >= (unsigned)TLI->getMinimumJumpTableEntries())
         return true;
     }
+
+    if (TM->getSubtargetImpl(*BB->getParent())->getTargetLowering()->useSoftFloat()) {
+      switch(J->getOpcode()) {
+      case Instruction::FAdd:
+      case Instruction::FSub:
+      case Instruction::FMul:
+      case Instruction::FDiv:
+      case Instruction::FRem:
+      case Instruction::FPTrunc:
+      case Instruction::FPExt:
+      case Instruction::FPToUI:
+      case Instruction::FPToSI:
+      case Instruction::UIToFP:
+      case Instruction::SIToFP:
+      case Instruction::FCmp:
+        return true;
+      }
+    }
+
     for (Value *Operand : J->operands())
       if (memAddrUsesCTR(TM, Operand))
         return true;
diff --git a/lib/Target/PowerPC/PPCCallingConv.td b/lib/Target/PowerPC/PPCCallingConv.td
index 5bc9124f80858..53d2f77ff9188 100644
--- a/lib/Target/PowerPC/PPCCallingConv.td
+++ b/lib/Target/PowerPC/PPCCallingConv.td
@@ -23,6 +23,9 @@ class CCIfNotSubtarget<string F, CCAction A>
                        "(State.getMachineFunction().getSubtarget()).",
                      F),
           A>;
+class CCIfOrigArgWasNotPPCF128<CCAction A>
+    : CCIf<"!static_cast<PPCCCState *>(&State)->WasOriginalArgPPCF128(ValNo)",
+           A>;
 
 //===----------------------------------------------------------------------===//
 // Return Value Calling Convention
@@ -109,7 +112,7 @@ def RetCC_PPC64_ELF_FIS : CallingConv<[
   CCIfType<[i8],   CCPromoteToType<i64>>,
   CCIfType<[i16],  CCPromoteToType<i64>>,
   CCIfType<[i32],  CCPromoteToType<i64>>,
-  CCIfType<[i64],  CCAssignToReg<[X3, X4]>>,
+  CCIfType<[i64],  CCAssignToReg<[X3, X4, X5, X6]>>,
   CCIfType<[i128], CCAssignToReg<[X3, X4, X5, X6]>>,
   CCIfType<[f32],  CCAssignToReg<[F1, F2, F3, F4, F5, F6, F7, F8]>>,
   CCIfType<[f64],  CCAssignToReg<[F1, F2, F3, F4, F5, F6, F7, F8]>>,
@@ -131,7 +134,14 @@ def CC_PPC32_SVR4_Common : CallingConv<[
 
   // The ABI requires i64 to be passed in two adjacent registers with the first
   // register having an odd register number.
-  CCIfType<[i32], CCIfSplit<CCCustom<"CC_PPC32_SVR4_Custom_AlignArgRegs">>>,
+  CCIfType<[i32],
+  CCIfSplit<CCIfSubtarget<"useSoftFloat()", 
+            CCIfOrigArgWasNotPPCF128<
+            CCCustom<"CC_PPC32_SVR4_Custom_AlignArgRegs">>>>>,
+  
+  CCIfType<[i32],
+  CCIfSplit<CCIfNotSubtarget<"useSoftFloat()", 
+                            CCCustom<"CC_PPC32_SVR4_Custom_AlignArgRegs">>>>,
 
   // The 'nest' parameter, if any, is passed in R11.
   CCIfNest<CCAssignToReg<[R11]>>,
@@ -243,12 +253,23 @@ def CSR_SVR464   : CalleeSavedRegs<(add X14, X15, X16, X17, X18, X19, X20,
                                         F27, F28, F29, F30, F31, CR2, CR3, CR4
                                    )>;
 
+// CSRs that are handled by prologue, epilogue.
+def CSR_SRV464_TLS_PE : CalleeSavedRegs<(add)>;
+
+def CSR_SVR464_ViaCopy : CalleeSavedRegs<(add CSR_SVR464)>;
+
 def CSR_SVR464_Altivec : CalleeSavedRegs<(add CSR_SVR464, CSR_Altivec)>;
 
+def CSR_SVR464_Altivec_ViaCopy : CalleeSavedRegs<(add CSR_SVR464_Altivec)>;
+
 def CSR_SVR464_R2 : CalleeSavedRegs<(add CSR_SVR464, X2)>;
 
+def CSR_SVR464_R2_ViaCopy : CalleeSavedRegs<(add CSR_SVR464_R2)>;
+
 def CSR_SVR464_R2_Altivec : CalleeSavedRegs<(add CSR_SVR464_Altivec, X2)>;
 
+def CSR_SVR464_R2_Altivec_ViaCopy : CalleeSavedRegs<(add CSR_SVR464_R2_Altivec)>;
+
 def CSR_NoRegs : CalleeSavedRegs<(add)>;
 
 def CSR_64_AllRegs: CalleeSavedRegs<(add X0, (sequence "X%u", 3, 10),
diff --git a/lib/Target/PowerPC/PPCEarlyReturn.cpp b/lib/Target/PowerPC/PPCEarlyReturn.cpp
index 7cb1bb54c7251..fcd2f50e1e3dc 100644
--- a/lib/Target/PowerPC/PPCEarlyReturn.cpp
+++ b/lib/Target/PowerPC/PPCEarlyReturn.cpp
@@ -12,10 +12,10 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "PPCInstrInfo.h"
-#include "MCTargetDesc/PPCPredicates.h"
 #include "PPC.h"
+#include "MCTargetDesc/PPCPredicates.h"
 #include "PPCInstrBuilder.h"
+#include "PPCInstrInfo.h"
 #include "PPCMachineFunctionInfo.h"
 #include "PPCTargetMachine.h"
 #include "llvm/ADT/STLExtras.h"
@@ -26,7 +26,6 @@
 #include "llvm/CodeGen/MachineMemOperand.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/MC/MCAsmInfo.h"
-#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/TargetRegistry.h"
@@ -84,7 +83,7 @@ protected:
               // This is an unconditional branch to the return. Replace the
               // branch with a blr.
               BuildMI(**PI, J, J->getDebugLoc(), TII->get(I->getOpcode()))
-                  .copyImplicitOps(I);
+                  .copyImplicitOps(*I);
               MachineBasicBlock::iterator K = J--;
               K->eraseFromParent();
               BlockChanged = true;
@@ -98,7 +97,7 @@ protected:
               BuildMI(**PI, J, J->getDebugLoc(), TII->get(PPC::BCCLR))
                   .addImm(J->getOperand(0).getImm())
                   .addReg(J->getOperand(1).getReg())
-                  .copyImplicitOps(I);
+                  .copyImplicitOps(*I);
               MachineBasicBlock::iterator K = J--;
               K->eraseFromParent();
               BlockChanged = true;
@@ -113,7 +112,7 @@ protected:
                   **PI, J, J->getDebugLoc(),
                   TII->get(J->getOpcode() == PPC::BC ? PPC::BCLR : PPC::BCLRn))
                   .addReg(J->getOperand(0).getReg())
-                  .copyImplicitOps(I);
+                  .copyImplicitOps(*I);
               MachineBasicBlock::iterator K = J--;
               K->eraseFromParent();
               BlockChanged = true;
@@ -174,6 +173,9 @@ protected:
 
 public:
     bool runOnMachineFunction(MachineFunction &MF) override {
+      if (skipFunction(*MF.getFunction()))
+        return false;
+
       TII = MF.getSubtarget().getInstrInfo();
 
       bool Changed = false;
@@ -192,6 +194,11 @@ public:
       return Changed;
     }
 
+    MachineFunctionProperties getRequiredProperties() const override {
+      return MachineFunctionProperties().set(
+          MachineFunctionProperties::Property::AllVRegsAllocated);
+    }
+
     void getAnalysisUsage(AnalysisUsage &AU) const override {
       MachineFunctionPass::getAnalysisUsage(AU);
     }
@@ -204,4 +211,3 @@ INITIALIZE_PASS(PPCEarlyReturn, DEBUG_TYPE,
 char PPCEarlyReturn::ID = 0;
 FunctionPass*
 llvm::createPPCEarlyReturnPass() { return new PPCEarlyReturn(); }
-
diff --git a/lib/Target/PowerPC/PPCFastISel.cpp b/lib/Target/PowerPC/PPCFastISel.cpp
index 16dcd468c91d6..7e92042d2f969 100644
--- a/lib/Target/PowerPC/PPCFastISel.cpp
+++ b/lib/Target/PowerPC/PPCFastISel.cpp
@@ -16,6 +16,7 @@
 #include "PPC.h"
 #include "MCTargetDesc/PPCPredicates.h"
 #include "PPCCallingConv.h"
+#include "PPCCCState.h"
 #include "PPCISelLowering.h"
 #include "PPCMachineFunctionInfo.h"
 #include "PPCSubtarget.h"
@@ -158,7 +159,7 @@ class PPCFastISel final : public FastISel {
                      unsigned FP64LoadOpc = PPC::LFD);
     bool PPCEmitStore(MVT VT, unsigned SrcReg, Address &Addr);
     bool PPCComputeAddress(const Value *Obj, Address &Addr);
-    void PPCSimplifyAddress(Address &Addr, MVT VT, bool &UseOffset,
+    void PPCSimplifyAddress(Address &Addr, bool &UseOffset,
                             unsigned &IndexReg);
     bool PPCEmitIntExt(MVT SrcVT, unsigned SrcReg, MVT DestVT,
                            unsigned DestReg, bool IsZExt);
@@ -185,7 +186,7 @@ class PPCFastISel final : public FastISel {
                          unsigned &NumBytes,
                          bool IsVarArg);
     bool finishCall(MVT RetVT, CallLoweringInfo &CLI, unsigned &NumBytes);
-    CCAssignFn *usePPC32CCs(unsigned Flag);
+    LLVM_ATTRIBUTE_UNUSED CCAssignFn *usePPC32CCs(unsigned Flag);
 
   private:
   #include "PPCGenFastISel.inc"
@@ -196,7 +197,7 @@ class PPCFastISel final : public FastISel {
 
 #include "PPCGenCallingConv.inc"
 
-// Function whose sole purpose is to kill compiler warnings 
+// Function whose sole purpose is to kill compiler warnings
 // stemming from unused functions included from PPCGenCallingConv.inc.
 CCAssignFn *PPCFastISel::usePPC32CCs(unsigned Flag) {
   if (Flag == 1)
@@ -213,13 +214,29 @@ static Optional<PPC::Predicate> getComparePred(CmpInst::Predicate Pred) {
   switch (Pred) {
     // These are not representable with any single compare.
     case CmpInst::FCMP_FALSE:
+    case CmpInst::FCMP_TRUE:
+    // Major concern about the following 6 cases is NaN result. The comparison
+    // result consists of 4 bits, indicating lt, eq, gt and un (unordered),
+    // only one of which will be set. The result is generated by fcmpu
+    // instruction. However, bc instruction only inspects one of the first 3
+    // bits, so when un is set, bc instruction may jump to to an undesired
+    // place.
+    //
+    // More specifically, if we expect an unordered comparison and un is set, we
+    // expect to always go to true branch; in such case UEQ, UGT and ULT still
+    // give false, which are undesired; but UNE, UGE, ULE happen to give true,
+    // since they are tested by inspecting !eq, !lt, !gt, respectively.
+    //
+    // Similarly, for ordered comparison, when un is set, we always expect the
+    // result to be false. In such case OGT, OLT and OEQ is good, since they are
+    // actually testing GT, LT, and EQ respectively, which are false. OGE, OLE
+    // and ONE are tested through !lt, !gt and !eq, and these are true.
     case CmpInst::FCMP_UEQ:
     case CmpInst::FCMP_UGT:
-    case CmpInst::FCMP_UGE:
     case CmpInst::FCMP_ULT:
-    case CmpInst::FCMP_ULE:
-    case CmpInst::FCMP_UNE:
-    case CmpInst::FCMP_TRUE:
+    case CmpInst::FCMP_OGE:
+    case CmpInst::FCMP_OLE:
+    case CmpInst::FCMP_ONE:
     default:
       return Optional<PPC::Predicate>();
 
@@ -232,7 +249,7 @@ static Optional<PPC::Predicate> getComparePred(CmpInst::Predicate Pred) {
     case CmpInst::ICMP_SGT:
       return PPC::PRED_GT;
 
-    case CmpInst::FCMP_OGE:
+    case CmpInst::FCMP_UGE:
     case CmpInst::ICMP_UGE:
     case CmpInst::ICMP_SGE:
       return PPC::PRED_GE;
@@ -242,12 +259,12 @@ static Optional<PPC::Predicate> getComparePred(CmpInst::Predicate Pred) {
     case CmpInst::ICMP_SLT:
       return PPC::PRED_LT;
 
-    case CmpInst::FCMP_OLE:
+    case CmpInst::FCMP_ULE:
     case CmpInst::ICMP_ULE:
     case CmpInst::ICMP_SLE:
       return PPC::PRED_LE;
 
-    case CmpInst::FCMP_ONE:
+    case CmpInst::FCMP_UNE:
     case CmpInst::ICMP_NE:
       return PPC::PRED_NE;
 
@@ -412,7 +429,7 @@ bool PPCFastISel::PPCComputeAddress(const Value *Obj, Address &Addr) {
 // Fix up some addresses that can't be used directly.  For example, if
 // an offset won't fit in an instruction field, we may need to move it
 // into an index register.
-void PPCFastISel::PPCSimplifyAddress(Address &Addr, MVT VT, bool &UseOffset,
+void PPCFastISel::PPCSimplifyAddress(Address &Addr, bool &UseOffset,
                                      unsigned &IndexReg) {
 
   // Check whether the offset fits in the instruction field.
@@ -431,8 +448,7 @@ void PPCFastISel::PPCSimplifyAddress(Address &Addr, MVT VT, bool &UseOffset,
   }
 
   if (!UseOffset) {
-    IntegerType *OffsetTy = ((VT == MVT::i32) ? Type::getInt32Ty(*Context)
-                             : Type::getInt64Ty(*Context));
+    IntegerType *OffsetTy = Type::getInt64Ty(*Context);
     const ConstantInt *Offset =
       ConstantInt::getSigned(OffsetTy, (int64_t)(Addr.Offset));
     IndexReg = PPCMaterializeInt(Offset, MVT::i64);
@@ -501,7 +517,7 @@ bool PPCFastISel::PPCEmitLoad(MVT VT, unsigned &ResultReg, Address &Addr,
   // If necessary, materialize the offset into a register and use
   // the indexed form.  Also handle stack pointers with special needs.
   unsigned IndexReg = 0;
-  PPCSimplifyAddress(Addr, VT, UseOffset, IndexReg);
+  PPCSimplifyAddress(Addr, UseOffset, IndexReg);
 
   // If this is a potential VSX load with an offset of 0, a VSX indexed load can
   // be used.
@@ -637,7 +653,7 @@ bool PPCFastISel::PPCEmitStore(MVT VT, unsigned SrcReg, Address &Addr) {
   // If necessary, materialize the offset into a register and use
   // the indexed form.  Also handle stack pointers with special needs.
   unsigned IndexReg = 0;
-  PPCSimplifyAddress(Addr, VT, UseOffset, IndexReg);
+  PPCSimplifyAddress(Addr, UseOffset, IndexReg);
 
   // If this is a potential VSX store with an offset of 0, a VSX indexed store
   // can be used.
@@ -1068,10 +1084,10 @@ unsigned PPCFastISel::PPCMoveToIntReg(const Instruction *I, MVT VT,
   if (!PPCEmitStore(MVT::f64, SrcReg, Addr))
     return 0;
 
-  // Reload it into a GPR.  If we want an i32, modify the address
-  // to have a 4-byte offset so we load from the right place.
+  // Reload it into a GPR.  If we want an i32 on big endian, modify the
+  // address to have a 4-byte offset so we load from the right place.
   if (VT == MVT::i32)
-    Addr.Offset = 4;
+    Addr.Offset = (PPCSubTarget->isLittleEndian()) ? 0 : 4;
 
   // Look at the currently assigned register for this instruction
   // to determine the required register class.
@@ -1115,14 +1131,13 @@ bool PPCFastISel::SelectFPToI(const Instruction *I, bool IsSigned) {
     return false;
 
   // Convert f32 to f64 if necessary.  This is just a meaningless copy
-  // to get the register class right.  COPY_TO_REGCLASS is needed since
-  // a COPY from F4RC to F8RC is converted to a F4RC-F4RC copy downstream.
+  // to get the register class right.
   const TargetRegisterClass *InRC = MRI.getRegClass(SrcReg);
   if (InRC == &PPC::F4RCRegClass) {
     unsigned TmpReg = createResultReg(&PPC::F8RCRegClass);
     BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
-            TII.get(TargetOpcode::COPY_TO_REGCLASS), TmpReg)
-      .addReg(SrcReg).addImm(PPC::F8RCRegClassID);
+            TII.get(TargetOpcode::COPY), TmpReg)
+      .addReg(SrcReg);
     SrcReg = TmpReg;
   }
 
@@ -1583,6 +1598,9 @@ bool PPCFastISel::SelectRet(const Instruction *I) {
   if (!FuncInfo.CanLowerReturn)
     return false;
 
+  if (TLI.supportSplitCSR(FuncInfo.MF))
+    return false;
+
   const ReturnInst *Ret = cast<ReturnInst>(I);
   const Function &F = *I->getParent()->getParent();
 
@@ -2071,7 +2089,6 @@ unsigned PPCFastISel::PPCMaterialize64BitInt(int64_t Imm,
   return TmpReg3;
 }
 
-
 // Materialize an integer constant into a register, and return
 // the register number (or zero if we failed to handle it).
 unsigned PPCFastISel::PPCMaterializeInt(const ConstantInt *CI, MVT VT,
@@ -2085,12 +2102,12 @@ unsigned PPCFastISel::PPCMaterializeInt(const ConstantInt *CI, MVT VT,
     return ImmReg;
   }
 
-  if (VT != MVT::i64 && VT != MVT::i32 && VT != MVT::i16 &&
-      VT != MVT::i8 && VT != MVT::i1) 
+  if (VT != MVT::i64 && VT != MVT::i32 && VT != MVT::i16 && VT != MVT::i8 &&
+      VT != MVT::i1)
     return 0;
 
-  const TargetRegisterClass *RC = ((VT == MVT::i64) ? &PPC::G8RCRegClass :
-                                   &PPC::GPRCRegClass);
+  const TargetRegisterClass *RC =
+      ((VT == MVT::i64) ? &PPC::G8RCRegClass : &PPC::GPRCRegClass);
   int64_t Imm = UseSExt ? CI->getSExtValue() : CI->getZExtValue();
 
   // If the constant is in range, use a load-immediate.
diff --git a/lib/Target/PowerPC/PPCFrameLowering.cpp b/lib/Target/PowerPC/PPCFrameLowering.cpp
index 3fd509ae27f48..c480430dd29a9 100644
--- a/lib/Target/PowerPC/PPCFrameLowering.cpp
+++ b/lib/Target/PowerPC/PPCFrameLowering.cpp
@@ -76,9 +76,7 @@ static unsigned computeBasePointerSaveOffset(const PPCSubtarget &STI) {
   // SVR4 ABI: First slot in the general register save area.
   return STI.isPPC64()
              ? -16U
-             : (STI.getTargetMachine().getRelocationModel() == Reloc::PIC_)
-                   ? -12U
-                   : -8U;
+             : STI.getTargetMachine().isPositionIndependent() ? -12U : -8U;
 }
 
 PPCFrameLowering::PPCFrameLowering(const PPCSubtarget &STI)
@@ -596,7 +594,7 @@ PPCFrameLowering::findScratchRegister(MachineBasicBlock *MBB,
       (!UseAtEnd && (&MBB->getParent()->front() == MBB)))
     return true;
 
-  RS.enterBasicBlock(MBB);
+  RS.enterBasicBlock(*MBB);
 
   if (UseAtEnd && !MBB->empty()) {
     // The scratch register will be used at the end of the block, so must
@@ -653,7 +651,7 @@ PPCFrameLowering::findScratchRegister(MachineBasicBlock *MBB,
 
   // Now that we've done our best to provide both registers, double check
   // whether we were unable to provide enough.
-  if (BV.count() < (TwoUniqueRegsRequired ? 2 : 1))
+  if (BV.count() < (TwoUniqueRegsRequired ? 2U : 1U))
     return false;
 
   return true;
@@ -838,13 +836,20 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF,
   // If we need to spill the CR and the LR but we don't have two separate
   // registers available, we must spill them one at a time
   if (MustSaveCR && SingleScratchReg && MustSaveLR) {
-    // FIXME: In the ELFv2 ABI, we are not required to save all CR fields.
-    // If only one or two CR fields are clobbered, it could be more
-    // efficient to use mfocrf to selectively save just those fields.
+    // In the ELFv2 ABI, we are not required to save all CR fields.
+    // If only one or two CR fields are clobbered, it is more efficient to use
+    // mfocrf to selectively save just those fields, because mfocrf has short
+    // latency compares to mfcr.
+    unsigned MfcrOpcode = PPC::MFCR8;
+    unsigned CrState = RegState::ImplicitKill;
+    if (isELFv2ABI && MustSaveCRs.size() == 1) {
+      MfcrOpcode = PPC::MFOCRF8;
+      CrState = RegState::Kill;
+    }
     MachineInstrBuilder MIB =
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::MFCR8), TempReg);
+      BuildMI(MBB, MBBI, dl, TII.get(MfcrOpcode), TempReg);
     for (unsigned i = 0, e = MustSaveCRs.size(); i != e; ++i)
-      MIB.addReg(MustSaveCRs[i], RegState::ImplicitKill);
+      MIB.addReg(MustSaveCRs[i], CrState);
     BuildMI(MBB, MBBI, dl, TII.get(PPC::STW8))
       .addReg(TempReg, getKillRegState(true))
       .addImm(8)
@@ -856,13 +861,20 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF,
 
   if (MustSaveCR &&
       !(SingleScratchReg && MustSaveLR)) { // will only occur for PPC64
-    // FIXME: In the ELFv2 ABI, we are not required to save all CR fields.
-    // If only one or two CR fields are clobbered, it could be more
-    // efficient to use mfocrf to selectively save just those fields.
+    // In the ELFv2 ABI, we are not required to save all CR fields.
+    // If only one or two CR fields are clobbered, it is more efficient to use
+    // mfocrf to selectively save just those fields, because mfocrf has short
+    // latency compares to mfcr.
+    unsigned MfcrOpcode = PPC::MFCR8;
+    unsigned CrState = RegState::ImplicitKill;
+    if (isELFv2ABI && MustSaveCRs.size() == 1) {
+      MfcrOpcode = PPC::MFOCRF8;
+      CrState = RegState::Kill;
+    }
     MachineInstrBuilder MIB =
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::MFCR8), TempReg);
+      BuildMI(MBB, MBBI, dl, TII.get(MfcrOpcode), TempReg);
     for (unsigned i = 0, e = MustSaveCRs.size(); i != e; ++i)
-      MIB.addReg(MustSaveCRs[i], RegState::ImplicitKill);
+      MIB.addReg(MustSaveCRs[i], CrState);
   }
 
   if (HasFP)
@@ -889,7 +901,7 @@ void PPCFrameLowering::emitPrologue(MachineFunction &MF,
   if (MustSaveLR)
     // FIXME: On PPC32 SVR4, we must not spill before claiming the stackframe.
     BuildMI(MBB, MBBI, dl, StoreInst)
-      .addReg(ScratchReg)
+      .addReg(ScratchReg, getKillRegState(true))
       .addImm(LROffset)
       .addReg(SPReg);
 
@@ -1315,36 +1327,53 @@ void PPCFrameLowering::emitEpilogue(MachineFunction &MF,
           .addReg(FPReg)
           .addReg(ScratchReg);
       }
-    } else if (RetOpcode == PPC::TCRETURNdi) {
-      MBBI = MBB.getLastNonDebugInstr();
-      MachineOperand &JumpTarget = MBBI->getOperand(0);
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILB)).
-        addGlobalAddress(JumpTarget.getGlobal(), JumpTarget.getOffset());
-    } else if (RetOpcode == PPC::TCRETURNri) {
-      MBBI = MBB.getLastNonDebugInstr();
-      assert(MBBI->getOperand(0).isReg() && "Expecting register operand.");
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBCTR));
-    } else if (RetOpcode == PPC::TCRETURNai) {
-      MBBI = MBB.getLastNonDebugInstr();
-      MachineOperand &JumpTarget = MBBI->getOperand(0);
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBA)).addImm(JumpTarget.getImm());
-    } else if (RetOpcode == PPC::TCRETURNdi8) {
-      MBBI = MBB.getLastNonDebugInstr();
-      MachineOperand &JumpTarget = MBBI->getOperand(0);
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILB8)).
-        addGlobalAddress(JumpTarget.getGlobal(), JumpTarget.getOffset());
-    } else if (RetOpcode == PPC::TCRETURNri8) {
-      MBBI = MBB.getLastNonDebugInstr();
-      assert(MBBI->getOperand(0).isReg() && "Expecting register operand.");
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBCTR8));
-    } else if (RetOpcode == PPC::TCRETURNai8) {
-      MBBI = MBB.getLastNonDebugInstr();
-      MachineOperand &JumpTarget = MBBI->getOperand(0);
-      BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBA8)).addImm(JumpTarget.getImm());
+    } else {
+      createTailCallBranchInstr(MBB);
     }
   }
 }
 
+void PPCFrameLowering::createTailCallBranchInstr(MachineBasicBlock &MBB) const {
+  MachineBasicBlock::iterator MBBI = MBB.getFirstTerminator();
+  DebugLoc dl;
+
+  if (MBBI != MBB.end())
+    dl = MBBI->getDebugLoc();
+
+  const PPCInstrInfo &TII =
+      *static_cast<const PPCInstrInfo *>(Subtarget.getInstrInfo());
+
+  // Create branch instruction for pseudo tail call return instruction
+  unsigned RetOpcode = MBBI->getOpcode();
+  if (RetOpcode == PPC::TCRETURNdi) {
+    MBBI = MBB.getLastNonDebugInstr();
+    MachineOperand &JumpTarget = MBBI->getOperand(0);
+    BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILB)).
+      addGlobalAddress(JumpTarget.getGlobal(), JumpTarget.getOffset());
+  } else if (RetOpcode == PPC::TCRETURNri) {
+    MBBI = MBB.getLastNonDebugInstr();
+    assert(MBBI->getOperand(0).isReg() && "Expecting register operand.");
+    BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBCTR));
+  } else if (RetOpcode == PPC::TCRETURNai) {
+    MBBI = MBB.getLastNonDebugInstr();
+    MachineOperand &JumpTarget = MBBI->getOperand(0);
+    BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBA)).addImm(JumpTarget.getImm());
+  } else if (RetOpcode == PPC::TCRETURNdi8) {
+    MBBI = MBB.getLastNonDebugInstr();
+    MachineOperand &JumpTarget = MBBI->getOperand(0);
+    BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILB8)).
+      addGlobalAddress(JumpTarget.getGlobal(), JumpTarget.getOffset());
+  } else if (RetOpcode == PPC::TCRETURNri8) {
+    MBBI = MBB.getLastNonDebugInstr();
+    assert(MBBI->getOperand(0).isReg() && "Expecting register operand.");
+    BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBCTR8));
+  } else if (RetOpcode == PPC::TCRETURNai8) {
+    MBBI = MBB.getLastNonDebugInstr();
+    MachineOperand &JumpTarget = MBBI->getOperand(0);
+    BuildMI(MBB, MBBI, dl, TII.get(PPC::TAILBA8)).addImm(JumpTarget.getImm());
+  }
+}
+
 void PPCFrameLowering::determineCalleeSaves(MachineFunction &MF,
                                             BitVector &SavedRegs,
                                             RegScavenger *RS) const {
@@ -1421,6 +1450,18 @@ void PPCFrameLowering::processFunctionBeforeFrameFinalized(MachineFunction &MF,
   MachineFrameInfo *FFI = MF.getFrameInfo();
   const std::vector<CalleeSavedInfo> &CSI = FFI->getCalleeSavedInfo();
 
+  // If the function is shrink-wrapped, and if the function has a tail call, the
+  // tail call might not be in the new RestoreBlock, so real branch instruction
+  // won't be generated by emitEpilogue(), because shrink-wrap has chosen new
+  // RestoreBlock. So we handle this case here.
+  if (FFI->getSavePoint() && FFI->hasTailCall()) {
+    MachineBasicBlock *RestoreBlock = FFI->getRestorePoint();
+    for (MachineBasicBlock &MBB : MF) {
+      if (MBB.isReturnBlock() && (&MBB) != RestoreBlock)
+        createTailCallBranchInstr(MBB);
+    }
+  }
+
   // Early exit if no callee saved registers are modified!
   if (CSI.empty() && !needsFP(MF)) {
     addScavengingSpillSlot(MF, RS);
@@ -1770,7 +1811,7 @@ restoreCRs(bool isPPC64, bool is31,
                .addReg(MoveReg, getKillRegState(true)));
 }
 
-void PPCFrameLowering::
+MachineBasicBlock::iterator PPCFrameLowering::
 eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
                               MachineBasicBlock::iterator I) const {
   const TargetInstrInfo &TII = *Subtarget.getInstrInfo();
@@ -1787,7 +1828,7 @@ eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
       unsigned LISInstr = is64Bit ? PPC::LIS8 : PPC::LIS;
       unsigned ORIInstr = is64Bit ? PPC::ORI8 : PPC::ORI;
       MachineInstr *MI = I;
-      DebugLoc dl = MI->getDebugLoc();
+      const DebugLoc &dl = MI->getDebugLoc();
 
       if (isInt<16>(CalleeAmt)) {
         BuildMI(MBB, I, dl, TII.get(ADDIInstr), StackReg)
@@ -1807,7 +1848,7 @@ eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
     }
   }
   // Simply discard ADJCALLSTACKDOWN, ADJCALLSTACKUP instructions.
-  MBB.erase(I);
+  return MBB.erase(I);
 }
 
 bool
diff --git a/lib/Target/PowerPC/PPCFrameLowering.h b/lib/Target/PowerPC/PPCFrameLowering.h
index f1f3f0b831a70..28b0c57f0ffb5 100644
--- a/lib/Target/PowerPC/PPCFrameLowering.h
+++ b/lib/Target/PowerPC/PPCFrameLowering.h
@@ -66,6 +66,13 @@ class PPCFrameLowering: public TargetFrameLowering {
                            unsigned *SR2 = nullptr) const;
   bool twoUniqueScratchRegsRequired(MachineBasicBlock *MBB) const;
 
+  /**
+   * \brief Create branch instruction for PPC::TCRETURN* (tail call return)
+   *
+   * \param[in] MBB that is terminated by PPC::TCRETURN*
+   */
+  void createTailCallBranchInstr(MachineBasicBlock &MBB) const;
+
 public:
   PPCFrameLowering(const PPCSubtarget &STI);
 
@@ -93,9 +100,9 @@ public:
                                  const std::vector<CalleeSavedInfo> &CSI,
                                  const TargetRegisterInfo *TRI) const override;
 
-  void eliminateCallFramePseudoInstr(MachineFunction &MF,
-                                  MachineBasicBlock &MBB,
-                                  MachineBasicBlock::iterator I) const override;
+  MachineBasicBlock::iterator
+  eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
+                                MachineBasicBlock::iterator I) const override;
 
   bool restoreCalleeSavedRegisters(MachineBasicBlock &MBB,
                                   MachineBasicBlock::iterator MI,
diff --git a/lib/Target/PowerPC/PPCHazardRecognizers.cpp b/lib/Target/PowerPC/PPCHazardRecognizers.cpp
index 7234e30fa73e6..caab67d68b171 100644
--- a/lib/Target/PowerPC/PPCHazardRecognizers.cpp
+++ b/lib/Target/PowerPC/PPCHazardRecognizers.cpp
@@ -162,8 +162,9 @@ unsigned PPCDispatchGroupSBHazardRecognizer::PreEmitNoops(SUnit *SU) {
     unsigned Directive =
         DAG->MF.getSubtarget<PPCSubtarget>().getDarwinDirective();
     // If we're using a special group-terminating nop, then we need only one.
+    // FIXME: the same for P9 as previous gen until POWER9 scheduling is ready
     if (Directive == PPC::DIR_PWR6 || Directive == PPC::DIR_PWR7 ||
-        Directive == PPC::DIR_PWR8 )
+        Directive == PPC::DIR_PWR8 || Directive == PPC::DIR_PWR9)
       return 1;
 
     return 5 - CurSlots;
@@ -223,8 +224,10 @@ void PPCDispatchGroupSBHazardRecognizer::EmitNoop() {
       DAG->MF.getSubtarget<PPCSubtarget>().getDarwinDirective();
   // If the group has now filled all of its slots, or if we're using a special
   // group-terminating nop, the group is complete.
+  // FIXME: the same for P9 as previous gen until POWER9 scheduling is ready
   if (Directive == PPC::DIR_PWR6 || Directive == PPC::DIR_PWR7 ||
-      Directive == PPC::DIR_PWR8 || CurSlots == 6)  {
+      Directive == PPC::DIR_PWR8 || Directive == PPC::DIR_PWR8 ||
+      CurSlots == 6) {
     CurGroup.clear();
     CurSlots = CurBranches = 0;
   } else {
diff --git a/lib/Target/PowerPC/PPCISelDAGToDAG.cpp b/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
index 1eaa8118ba0a7..0e9b2daa0cb5b 100644
--- a/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
+++ b/lib/Target/PowerPC/PPCISelDAGToDAG.cpp
@@ -59,10 +59,6 @@ static cl::opt<bool> EnableBranchHint(
     cl::desc("Enable static hinting of branches on ppc"),
     cl::Hidden);
 
-namespace llvm {
-  void initializePPCDAGToDAGISelPass(PassRegistry&);
-}
-
 namespace {
   //===--------------------------------------------------------------------===//
   /// PPCDAGToDAGISel - PPC specific code to select PPC machine
@@ -75,9 +71,7 @@ namespace {
     unsigned GlobalBaseReg;
   public:
     explicit PPCDAGToDAGISel(PPCTargetMachine &tm)
-        : SelectionDAGISel(tm), TM(tm) {
-      initializePPCDAGToDAGISelPass(*PassRegistry::getPassRegistry());
-    }
+        : SelectionDAGISel(tm), TM(tm) {}
 
     bool runOnMachineFunction(MachineFunction &MF) override {
       // Make sure we re-emit a set of the global base reg if necessary
@@ -97,18 +91,18 @@ namespace {
 
     /// getI32Imm - Return a target constant with the specified value, of type
     /// i32.
-    inline SDValue getI32Imm(unsigned Imm, SDLoc dl) {
+    inline SDValue getI32Imm(unsigned Imm, const SDLoc &dl) {
       return CurDAG->getTargetConstant(Imm, dl, MVT::i32);
     }
 
     /// getI64Imm - Return a target constant with the specified value, of type
     /// i64.
-    inline SDValue getI64Imm(uint64_t Imm, SDLoc dl) {
+    inline SDValue getI64Imm(uint64_t Imm, const SDLoc &dl) {
       return CurDAG->getTargetConstant(Imm, dl, MVT::i64);
     }
 
     /// getSmallIPtrImm - Return a target constant of pointer type.
-    inline SDValue getSmallIPtrImm(unsigned Imm, SDLoc dl) {
+    inline SDValue getSmallIPtrImm(unsigned Imm, const SDLoc &dl) {
       return CurDAG->getTargetConstant(
           Imm, dl, PPCLowering->getPointerTy(CurDAG->getDataLayout()));
     }
@@ -122,18 +116,19 @@ namespace {
     /// base register.  Return the virtual register that holds this value.
     SDNode *getGlobalBaseReg();
 
-    SDNode *getFrameIndex(SDNode *SN, SDNode *N, unsigned Offset = 0);
+    void selectFrameIndex(SDNode *SN, SDNode *N, unsigned Offset = 0);
 
     // Select - Convert the specified operand from a target-independent to a
     // target-specific node if it hasn't already been changed.
-    SDNode *Select(SDNode *N) override;
+    void Select(SDNode *N) override;
 
-    SDNode *SelectBitfieldInsert(SDNode *N);
-    SDNode *SelectBitPermutation(SDNode *N);
+    bool tryBitfieldInsert(SDNode *N);
+    bool tryBitPermutation(SDNode *N);
 
     /// SelectCC - Select a comparison of the specified values with the
     /// specified condition code, returning the CR# of the expression.
-    SDValue SelectCC(SDValue LHS, SDValue RHS, ISD::CondCode CC, SDLoc dl);
+    SDValue SelectCC(SDValue LHS, SDValue RHS, ISD::CondCode CC,
+                     const SDLoc &dl);
 
     /// SelectAddrImm - Returns true if the address N can be represented by
     /// a base register plus a signed 16-bit displacement [r+imm].
@@ -228,7 +223,7 @@ namespace {
 #include "PPCGenDAGISel.inc"
 
 private:
-    SDNode *SelectSETCC(SDNode *N);
+    bool trySETCC(SDNode *N);
 
     void PeepholePPC64();
     void PeepholePPC64ZExt();
@@ -240,7 +235,7 @@ private:
     bool AllUsersSelectZero(SDNode *N);
     void SwapAllSelectUsers(SDNode *N);
 
-    SDNode *transferMemOperands(SDNode *N, SDNode *Result);
+    void transferMemOperands(SDNode *N, SDNode *Result);
   };
 }
 
@@ -324,7 +319,7 @@ SDNode *PPCDAGToDAGISel::getGlobalBaseReg() {
     if (PPCLowering->getPointerTy(CurDAG->getDataLayout()) == MVT::i32) {
       if (PPCSubTarget->isTargetELF()) {
         GlobalBaseReg = PPC::R30;
-        if (M->getPICLevel() == PICLevel::Small) {
+        if (M->getPICLevel() == PICLevel::SmallPIC) {
           BuildMI(FirstMBB, MBBI, dl, TII.get(PPC::MoveGOTtoLR));
           BuildMI(FirstMBB, MBBI, dl, TII.get(PPC::MFLR), GlobalBaseReg);
           MF->getInfo<PPCFunctionInfo>()->setUsesPICBase(true);
@@ -458,16 +453,17 @@ static bool isOpcWithIntImmediate(SDNode *N, unsigned Opc, unsigned& Imm) {
          && isInt32Immediate(N->getOperand(1).getNode(), Imm);
 }
 
-SDNode *PPCDAGToDAGISel::getFrameIndex(SDNode *SN, SDNode *N, unsigned Offset) {
+void PPCDAGToDAGISel::selectFrameIndex(SDNode *SN, SDNode *N, unsigned Offset) {
   SDLoc dl(SN);
   int FI = cast<FrameIndexSDNode>(N)->getIndex();
   SDValue TFI = CurDAG->getTargetFrameIndex(FI, N->getValueType(0));
   unsigned Opc = N->getValueType(0) == MVT::i32 ? PPC::ADDI : PPC::ADDI8;
   if (SN->hasOneUse())
-    return CurDAG->SelectNodeTo(SN, Opc, N->getValueType(0), TFI,
-                                getSmallIPtrImm(Offset, dl));
-  return CurDAG->getMachineNode(Opc, dl, N->getValueType(0), TFI,
-                                getSmallIPtrImm(Offset, dl));
+    CurDAG->SelectNodeTo(SN, Opc, N->getValueType(0), TFI,
+                         getSmallIPtrImm(Offset, dl));
+  else
+    ReplaceNode(SN, CurDAG->getMachineNode(Opc, dl, N->getValueType(0), TFI,
+                                           getSmallIPtrImm(Offset, dl)));
 }
 
 bool PPCDAGToDAGISel::isRotateAndMask(SDNode *N, unsigned Mask,
@@ -512,9 +508,9 @@ bool PPCDAGToDAGISel::isRotateAndMask(SDNode *N, unsigned Mask,
   return false;
 }
 
-/// SelectBitfieldInsert - turn an or of two masked values into
-/// the rotate left word immediate then mask insert (rlwimi) instruction.
-SDNode *PPCDAGToDAGISel::SelectBitfieldInsert(SDNode *N) {
+/// Turn an or of two masked values into the rotate left word immediate then
+/// mask insert (rlwimi) instruction.
+bool PPCDAGToDAGISel::tryBitfieldInsert(SDNode *N) {
   SDValue Op0 = N->getOperand(0);
   SDValue Op1 = N->getOperand(1);
   SDLoc dl(N);
@@ -584,15 +580,16 @@ SDNode *PPCDAGToDAGISel::SelectBitfieldInsert(SDNode *N) {
       SH &= 31;
       SDValue Ops[] = { Op0, Op1, getI32Imm(SH, dl), getI32Imm(MB, dl),
                           getI32Imm(ME, dl) };
-      return CurDAG->getMachineNode(PPC::RLWIMI, dl, MVT::i32, Ops);
+      ReplaceNode(N, CurDAG->getMachineNode(PPC::RLWIMI, dl, MVT::i32, Ops));
+      return true;
     }
   }
-  return nullptr;
+  return false;
 }
 
 // Predict the number of instructions that would be generated by calling
-// SelectInt64(N).
-static unsigned SelectInt64CountDirect(int64_t Imm) {
+// getInt64(N).
+static unsigned getInt64CountDirect(int64_t Imm) {
   // Assume no remaining bits.
   unsigned Remainder = 0;
   // Assume no shift required.
@@ -653,17 +650,17 @@ static uint64_t Rot64(uint64_t Imm, unsigned R) {
   return (Imm << R) | (Imm >> (64 - R));
 }
 
-static unsigned SelectInt64Count(int64_t Imm) {
-  unsigned Count = SelectInt64CountDirect(Imm);
+static unsigned getInt64Count(int64_t Imm) {
+  unsigned Count = getInt64CountDirect(Imm);
   if (Count == 1)
     return Count;
 
   for (unsigned r = 1; r < 63; ++r) {
     uint64_t RImm = Rot64(Imm, r);
-    unsigned RCount = SelectInt64CountDirect(RImm) + 1;
+    unsigned RCount = getInt64CountDirect(RImm) + 1;
     Count = std::min(Count, RCount);
 
-    // See comments in SelectInt64 for an explanation of the logic below.
+    // See comments in getInt64 for an explanation of the logic below.
     unsigned LS = findLastSet(RImm);
     if (LS != r-1)
       continue;
@@ -671,16 +668,17 @@ static unsigned SelectInt64Count(int64_t Imm) {
     uint64_t OnesMask = -(int64_t) (UINT64_C(1) << (LS+1));
     uint64_t RImmWithOnes = RImm | OnesMask;
 
-    RCount = SelectInt64CountDirect(RImmWithOnes) + 1;
+    RCount = getInt64CountDirect(RImmWithOnes) + 1;
     Count = std::min(Count, RCount);
   }
 
   return Count;
 }
 
-// Select a 64-bit constant. For cost-modeling purposes, SelectInt64Count
+// Select a 64-bit constant. For cost-modeling purposes, getInt64Count
 // (above) needs to be kept in sync with this function.
-static SDNode *SelectInt64Direct(SelectionDAG *CurDAG, SDLoc dl, int64_t Imm) {
+static SDNode *getInt64Direct(SelectionDAG *CurDAG, const SDLoc &dl,
+                              int64_t Imm) {
   // Assume no remaining bits.
   unsigned Remainder = 0;
   // Assume no shift required.
@@ -754,10 +752,10 @@ static SDNode *SelectInt64Direct(SelectionDAG *CurDAG, SDLoc dl, int64_t Imm) {
   return Result;
 }
 
-static SDNode *SelectInt64(SelectionDAG *CurDAG, SDLoc dl, int64_t Imm) {
-  unsigned Count = SelectInt64CountDirect(Imm);
+static SDNode *getInt64(SelectionDAG *CurDAG, const SDLoc &dl, int64_t Imm) {
+  unsigned Count = getInt64CountDirect(Imm);
   if (Count == 1)
-    return SelectInt64Direct(CurDAG, dl, Imm);
+    return getInt64Direct(CurDAG, dl, Imm);
 
   unsigned RMin = 0;
 
@@ -766,7 +764,7 @@ static SDNode *SelectInt64(SelectionDAG *CurDAG, SDLoc dl, int64_t Imm) {
 
   for (unsigned r = 1; r < 63; ++r) {
     uint64_t RImm = Rot64(Imm, r);
-    unsigned RCount = SelectInt64CountDirect(RImm) + 1;
+    unsigned RCount = getInt64CountDirect(RImm) + 1;
     if (RCount < Count) {
       Count = RCount;
       RMin = r;
@@ -789,7 +787,7 @@ static SDNode *SelectInt64(SelectionDAG *CurDAG, SDLoc dl, int64_t Imm) {
     uint64_t OnesMask = -(int64_t) (UINT64_C(1) << (LS+1));
     uint64_t RImmWithOnes = RImm | OnesMask;
 
-    RCount = SelectInt64CountDirect(RImmWithOnes) + 1;
+    RCount = getInt64CountDirect(RImmWithOnes) + 1;
     if (RCount < Count) {
       Count = RCount;
       RMin = r;
@@ -799,24 +797,24 @@ static SDNode *SelectInt64(SelectionDAG *CurDAG, SDLoc dl, int64_t Imm) {
   }
 
   if (!RMin)
-    return SelectInt64Direct(CurDAG, dl, Imm);
+    return getInt64Direct(CurDAG, dl, Imm);
 
   auto getI32Imm = [CurDAG, dl](unsigned Imm) {
       return CurDAG->getTargetConstant(Imm, dl, MVT::i32);
   };
 
-  SDValue Val = SDValue(SelectInt64Direct(CurDAG, dl, MatImm), 0);
+  SDValue Val = SDValue(getInt64Direct(CurDAG, dl, MatImm), 0);
   return CurDAG->getMachineNode(PPC::RLDICR, dl, MVT::i64, Val,
                                 getI32Imm(64 - RMin), getI32Imm(MaskEnd));
 }
 
 // Select a 64-bit constant.
-static SDNode *SelectInt64(SelectionDAG *CurDAG, SDNode *N) {
+static SDNode *getInt64(SelectionDAG *CurDAG, SDNode *N) {
   SDLoc dl(N);
 
   // Get 64 bit value.
   int64_t Imm = cast<ConstantSDNode>(N)->getZExtValue();
-  return SelectInt64(CurDAG, dl, Imm);
+  return getInt64(CurDAG, dl, Imm);
 }
 
 namespace {
@@ -1209,7 +1207,7 @@ class BitPermutationSelector {
                  "bit group ends at index 63 but there is another?");
           auto IN = BitGroups.begin();
 
-          if (IP->Repl32 && IN->Repl32 && I->V == IP->V && I->V == IN->V && 
+          if (IP->Repl32 && IN->Repl32 && I->V == IP->V && I->V == IN->V &&
               (I->RLAmt % 32) == IP->RLAmt && (I->RLAmt % 32) == IN->RLAmt &&
               IP->EndIdx == 31 && IN->StartIdx == 0 && I != IP &&
               IsAllLow32(*I)) {
@@ -1252,7 +1250,7 @@ class BitPermutationSelector {
     }
   }
 
-  SDValue getI32Imm(unsigned Imm, SDLoc dl) {
+  SDValue getI32Imm(unsigned Imm, const SDLoc &dl) {
     return CurDAG->getTargetConstant(Imm, dl, MVT::i32);
   }
 
@@ -1270,7 +1268,7 @@ class BitPermutationSelector {
   // Depending on the number of groups for a particular value, it might be
   // better to rotate, mask explicitly (using andi/andis), and then or the
   // result. Select this part of the result first.
-  void SelectAndParts32(SDLoc dl, SDValue &Res, unsigned *InstCnt) {
+  void SelectAndParts32(const SDLoc &dl, SDValue &Res, unsigned *InstCnt) {
     if (BPermRewriterNoMasking)
       return;
 
@@ -1466,8 +1464,8 @@ class BitPermutationSelector {
 
   // For 64-bit values, not all combinations of rotates and masks are
   // available. Produce one if it is available.
-  SDValue SelectRotMask64(SDValue V, SDLoc dl, unsigned RLAmt, bool Repl32,
-                          unsigned MaskStart, unsigned MaskEnd,
+  SDValue SelectRotMask64(SDValue V, const SDLoc &dl, unsigned RLAmt,
+                          bool Repl32, unsigned MaskStart, unsigned MaskEnd,
                           unsigned *InstCnt = nullptr) {
     // In the notation used by the instructions, 'start' and 'end' are reversed
     // because bits are counted from high to low order.
@@ -1527,8 +1525,8 @@ class BitPermutationSelector {
 
   // For 64-bit values, not all combinations of rotates and masks are
   // available. Produce a rotate-mask-and-insert if one is available.
-  SDValue SelectRotMaskIns64(SDValue Base, SDValue V, SDLoc dl, unsigned RLAmt,
-                             bool Repl32, unsigned MaskStart,
+  SDValue SelectRotMaskIns64(SDValue Base, SDValue V, const SDLoc &dl,
+                             unsigned RLAmt, bool Repl32, unsigned MaskStart,
                              unsigned MaskEnd, unsigned *InstCnt = nullptr) {
     // In the notation used by the instructions, 'start' and 'end' are reversed
     // because bits are counted from high to low order.
@@ -1574,7 +1572,7 @@ class BitPermutationSelector {
     return SelectRotMaskIns64(Base, V, dl, RLAmt2, false, MaskStart, MaskEnd);
   }
 
-  void SelectAndParts64(SDLoc dl, SDValue &Res, unsigned *InstCnt) {
+  void SelectAndParts64(const SDLoc &dl, SDValue &Res, unsigned *InstCnt) {
     if (BPermRewriterNoMasking)
       return;
 
@@ -1646,7 +1644,7 @@ class BitPermutationSelector {
         NumAndInsts += (unsigned) (ANDIMask != 0) + (unsigned) (ANDISMask != 0) +
                        (unsigned) (ANDIMask != 0 && ANDISMask != 0);
       else
-        NumAndInsts += SelectInt64Count(Mask) + /* and */ 1;
+        NumAndInsts += getInt64Count(Mask) + /* and */ 1;
 
       unsigned NumRLInsts = 0;
       bool FirstBG = true;
@@ -1709,7 +1707,7 @@ class BitPermutationSelector {
           TotalVal = SDValue(CurDAG->getMachineNode(PPC::OR8, dl, MVT::i64,
                                ANDIVal, ANDISVal), 0);
       } else {
-        TotalVal = SDValue(SelectInt64(CurDAG, dl, Mask), 0);
+        TotalVal = SDValue(getInt64(CurDAG, dl, Mask), 0);
         TotalVal =
           SDValue(CurDAG->getMachineNode(PPC::AND8, dl, MVT::i64,
                                          VRot, TotalVal), 0);
@@ -1852,9 +1850,9 @@ class BitPermutationSelector {
           Res = SDValue(CurDAG->getMachineNode(PPC::OR8, dl, MVT::i64,
                           ANDIVal, ANDISVal), 0);
       } else {
-        if (InstCnt) *InstCnt += SelectInt64Count(Mask) + /* and */ 1;
+        if (InstCnt) *InstCnt += getInt64Count(Mask) + /* and */ 1;
 
-        SDValue MaskVal = SDValue(SelectInt64(CurDAG, dl, Mask), 0);
+        SDValue MaskVal = SDValue(getInt64(CurDAG, dl, Mask), 0);
         Res =
           SDValue(CurDAG->getMachineNode(PPC::AND8, dl, MVT::i64,
                                          Res, MaskVal), 0);
@@ -1955,13 +1953,13 @@ public:
 };
 } // anonymous namespace
 
-SDNode *PPCDAGToDAGISel::SelectBitPermutation(SDNode *N) {
+bool PPCDAGToDAGISel::tryBitPermutation(SDNode *N) {
   if (N->getValueType(0) != MVT::i32 &&
       N->getValueType(0) != MVT::i64)
-    return nullptr;
+    return false;
 
   if (!UseBitPermRewriter)
-    return nullptr;
+    return false;
 
   switch (N->getOpcode()) {
   default: break;
@@ -1971,17 +1969,21 @@ SDNode *PPCDAGToDAGISel::SelectBitPermutation(SDNode *N) {
   case ISD::AND:
   case ISD::OR: {
     BitPermutationSelector BPS(CurDAG);
-    return BPS.Select(N);
+    if (SDNode *New = BPS.Select(N)) {
+      ReplaceNode(N, New);
+      return true;
+    }
+    return false;
   }
   }
 
-  return nullptr;
+  return false;
 }
 
 /// SelectCC - Select a comparison of the specified values with the specified
 /// condition code, returning the CR# of the expression.
-SDValue PPCDAGToDAGISel::SelectCC(SDValue LHS, SDValue RHS,
-                                    ISD::CondCode CC, SDLoc dl) {
+SDValue PPCDAGToDAGISel::SelectCC(SDValue LHS, SDValue RHS, ISD::CondCode CC,
+                                  const SDLoc &dl) {
   // Always select the LHS.
   unsigned Opc;
 
@@ -2255,7 +2257,7 @@ static unsigned int getVCmpInst(MVT VecVT, ISD::CondCode CC,
   }
 }
 
-SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
+bool PPCDAGToDAGISel::trySETCC(SDNode *N) {
   SDLoc dl(N);
   unsigned Imm;
   ISD::CondCode CC = cast<CondCodeSDNode>(N->getOperand(2))->get();
@@ -2276,20 +2278,22 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
         Op = SDValue(CurDAG->getMachineNode(PPC::CNTLZW, dl, MVT::i32, Op), 0);
         SDValue Ops[] = { Op, getI32Imm(27, dl), getI32Imm(5, dl),
                           getI32Imm(31, dl) };
-        return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        return true;
       }
       case ISD::SETNE: {
         if (isPPC64) break;
         SDValue AD =
           SDValue(CurDAG->getMachineNode(PPC::ADDIC, dl, MVT::i32, MVT::Glue,
                                          Op, getI32Imm(~0U, dl)), 0);
-        return CurDAG->SelectNodeTo(N, PPC::SUBFE, MVT::i32, AD, Op,
-                                    AD.getValue(1));
+        CurDAG->SelectNodeTo(N, PPC::SUBFE, MVT::i32, AD, Op, AD.getValue(1));
+        return true;
       }
       case ISD::SETLT: {
         SDValue Ops[] = { Op, getI32Imm(1, dl), getI32Imm(31, dl),
                           getI32Imm(31, dl) };
-        return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        return true;
       }
       case ISD::SETGT: {
         SDValue T =
@@ -2297,7 +2301,8 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
         T = SDValue(CurDAG->getMachineNode(PPC::ANDC, dl, MVT::i32, T, Op), 0);
         SDValue Ops[] = { T, getI32Imm(1, dl), getI32Imm(31, dl),
                           getI32Imm(31, dl) };
-        return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        return true;
       }
       }
     } else if (Imm == ~0U) {        // setcc op, -1
@@ -2308,18 +2313,20 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
         if (isPPC64) break;
         Op = SDValue(CurDAG->getMachineNode(PPC::ADDIC, dl, MVT::i32, MVT::Glue,
                                             Op, getI32Imm(1, dl)), 0);
-        return CurDAG->SelectNodeTo(N, PPC::ADDZE, MVT::i32,
-                              SDValue(CurDAG->getMachineNode(PPC::LI, dl,
-                                                             MVT::i32,
-                                                             getI32Imm(0, dl)),
-                                      0), Op.getValue(1));
+        CurDAG->SelectNodeTo(N, PPC::ADDZE, MVT::i32,
+                             SDValue(CurDAG->getMachineNode(PPC::LI, dl,
+                                                            MVT::i32,
+                                                            getI32Imm(0, dl)),
+                                     0), Op.getValue(1));
+        return true;
       case ISD::SETNE: {
         if (isPPC64) break;
         Op = SDValue(CurDAG->getMachineNode(PPC::NOR, dl, MVT::i32, Op, Op), 0);
         SDNode *AD = CurDAG->getMachineNode(PPC::ADDIC, dl, MVT::i32, MVT::Glue,
                                             Op, getI32Imm(~0U, dl));
-        return CurDAG->SelectNodeTo(N, PPC::SUBFE, MVT::i32, SDValue(AD, 0),
-                                    Op, SDValue(AD, 1));
+        CurDAG->SelectNodeTo(N, PPC::SUBFE, MVT::i32, SDValue(AD, 0), Op,
+                             SDValue(AD, 1));
+        return true;
       }
       case ISD::SETLT: {
         SDValue AD = SDValue(CurDAG->getMachineNode(PPC::ADDI, dl, MVT::i32, Op,
@@ -2328,14 +2335,15 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
                                                     Op), 0);
         SDValue Ops[] = { AN, getI32Imm(1, dl), getI32Imm(31, dl),
                           getI32Imm(31, dl) };
-        return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+        return true;
       }
       case ISD::SETGT: {
         SDValue Ops[] = { Op, getI32Imm(1, dl), getI32Imm(31, dl),
                           getI32Imm(31, dl) };
         Op = SDValue(CurDAG->getMachineNode(PPC::RLWINM, dl, MVT::i32, Ops), 0);
-        return CurDAG->SelectNodeTo(N, PPC::XORI, MVT::i32, Op,
-                                    getI32Imm(1, dl));
+        CurDAG->SelectNodeTo(N, PPC::XORI, MVT::i32, Op, getI32Imm(1, dl));
+        return true;
       }
       }
     }
@@ -2348,7 +2356,7 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
   // vector compare operations return the same type as the operands.
   if (LHS.getValueType().isVector()) {
     if (PPCSubTarget->hasQPX())
-      return nullptr;
+      return false;
 
     EVT VecVT = LHS.getValueType();
     bool Swap, Negate;
@@ -2360,16 +2368,17 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
     EVT ResVT = VecVT.changeVectorElementTypeToInteger();
     if (Negate) {
       SDValue VCmp(CurDAG->getMachineNode(VCmpInst, dl, ResVT, LHS, RHS), 0);
-      return CurDAG->SelectNodeTo(N, PPCSubTarget->hasVSX() ? PPC::XXLNOR :
-                                                              PPC::VNOR,
-                                  ResVT, VCmp, VCmp);
+      CurDAG->SelectNodeTo(N, PPCSubTarget->hasVSX() ? PPC::XXLNOR : PPC::VNOR,
+                           ResVT, VCmp, VCmp);
+      return true;
     }
 
-    return CurDAG->SelectNodeTo(N, VCmpInst, ResVT, LHS, RHS);
+    CurDAG->SelectNodeTo(N, VCmpInst, ResVT, LHS, RHS);
+    return true;
   }
 
   if (PPCSubTarget->useCRBits())
-    return nullptr;
+    return false;
 
   bool Inv;
   unsigned Idx = getCRIdxForSetCC(CC, Inv);
@@ -2388,31 +2397,33 @@ SDNode *PPCDAGToDAGISel::SelectSETCC(SDNode *N) {
 
   SDValue Ops[] = { IntCR, getI32Imm((32 - (3 - Idx)) & 31, dl),
                       getI32Imm(31, dl), getI32Imm(31, dl) };
-  if (!Inv)
-    return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+  if (!Inv) {
+    CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+    return true;
+  }
 
   // Get the specified bit.
   SDValue Tmp =
     SDValue(CurDAG->getMachineNode(PPC::RLWINM, dl, MVT::i32, Ops), 0);
-  return CurDAG->SelectNodeTo(N, PPC::XORI, MVT::i32, Tmp, getI32Imm(1, dl));
+  CurDAG->SelectNodeTo(N, PPC::XORI, MVT::i32, Tmp, getI32Imm(1, dl));
+  return true;
 }
 
-SDNode *PPCDAGToDAGISel::transferMemOperands(SDNode *N, SDNode *Result) {
+void PPCDAGToDAGISel::transferMemOperands(SDNode *N, SDNode *Result) {
   // Transfer memoperands.
   MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
   MemOp[0] = cast<MemSDNode>(N)->getMemOperand();
   cast<MachineSDNode>(Result)->setMemRefs(MemOp, MemOp + 1);
-  return Result;
 }
 
 
 // Select - Convert the specified operand from a target-independent to a
 // target-specific node if it hasn't already been changed.
-SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
+void PPCDAGToDAGISel::Select(SDNode *N) {
   SDLoc dl(N);
   if (N->isMachineOpcode()) {
     N->setNodeId(-1);
-    return nullptr;   // Already selected.
+    return;   // Already selected.
   }
 
   // In case any misguided DAG-level optimizations form an ADD with a
@@ -2423,39 +2434,44 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
     llvm_unreachable("Invalid ADD with TargetConstant operand");
 
   // Try matching complex bit permutations before doing anything else.
-  if (SDNode *NN = SelectBitPermutation(N))
-    return NN;
+  if (tryBitPermutation(N))
+    return;
 
   switch (N->getOpcode()) {
   default: break;
 
   case ISD::Constant: {
-    if (N->getValueType(0) == MVT::i64)
-      return SelectInt64(CurDAG, N);
+    if (N->getValueType(0) == MVT::i64) {
+      ReplaceNode(N, getInt64(CurDAG, N));
+      return;
+    }
     break;
   }
 
   case ISD::SETCC: {
-    SDNode *SN = SelectSETCC(N);
-    if (SN)
-      return SN;
+    if (trySETCC(N))
+      return;
     break;
   }
   case PPCISD::GlobalBaseReg:
-    return getGlobalBaseReg();
+    ReplaceNode(N, getGlobalBaseReg());
+    return;
 
   case ISD::FrameIndex:
-    return getFrameIndex(N, N);
+    selectFrameIndex(N, N);
+    return;
 
   case PPCISD::MFOCRF: {
     SDValue InFlag = N->getOperand(1);
-    return CurDAG->getMachineNode(PPC::MFOCRF, dl, MVT::i32,
-                                  N->getOperand(0), InFlag);
+    ReplaceNode(N, CurDAG->getMachineNode(PPC::MFOCRF, dl, MVT::i32,
+                                          N->getOperand(0), InFlag));
+    return;
   }
 
   case PPCISD::READ_TIME_BASE: {
-    return CurDAG->getMachineNode(PPC::ReadTB, dl, MVT::i32, MVT::i32,
-                                  MVT::Other, N->getOperand(0));
+    ReplaceNode(N, CurDAG->getMachineNode(PPC::ReadTB, dl, MVT::i32, MVT::i32,
+                                          MVT::Other, N->getOperand(0)));
+    return;
   }
 
   case PPCISD::SRA_ADDZE: {
@@ -2468,16 +2484,18 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDNode *Op =
         CurDAG->getMachineNode(PPC::SRADI, dl, MVT::i64, MVT::Glue,
                                N0, ShiftAmt);
-      return CurDAG->SelectNodeTo(N, PPC::ADDZE8, MVT::i64,
-                                  SDValue(Op, 0), SDValue(Op, 1));
+      CurDAG->SelectNodeTo(N, PPC::ADDZE8, MVT::i64, SDValue(Op, 0),
+                           SDValue(Op, 1));
+      return;
     } else {
       assert(N->getValueType(0) == MVT::i32 &&
              "Expecting i64 or i32 in PPCISD::SRA_ADDZE");
       SDNode *Op =
         CurDAG->getMachineNode(PPC::SRAWI, dl, MVT::i32, MVT::Glue,
                                N0, ShiftAmt);
-      return CurDAG->SelectNodeTo(N, PPC::ADDZE, MVT::i32,
-                                  SDValue(Op, 0), SDValue(Op, 1));
+      CurDAG->SelectNodeTo(N, PPC::ADDZE, MVT::i32, SDValue(Op, 0),
+                           SDValue(Op, 1));
+      return;
     }
   }
 
@@ -2524,11 +2542,12 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue Chain = LD->getChain();
       SDValue Base = LD->getBasePtr();
       SDValue Ops[] = { Offset, Base, Chain };
-      return transferMemOperands(
-          N, CurDAG->getMachineNode(
-                 Opcode, dl, LD->getValueType(0),
-                 PPCLowering->getPointerTy(CurDAG->getDataLayout()), MVT::Other,
-                 Ops));
+      SDNode *MN = CurDAG->getMachineNode(
+          Opcode, dl, LD->getValueType(0),
+          PPCLowering->getPointerTy(CurDAG->getDataLayout()), MVT::Other, Ops);
+      transferMemOperands(N, MN);
+      ReplaceNode(N, MN);
+      return;
     } else {
       unsigned Opcode;
       bool isSExt = LD->getExtensionType() == ISD::SEXTLOAD;
@@ -2563,11 +2582,12 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue Chain = LD->getChain();
       SDValue Base = LD->getBasePtr();
       SDValue Ops[] = { Base, Offset, Chain };
-      return transferMemOperands(
-          N, CurDAG->getMachineNode(
-                 Opcode, dl, LD->getValueType(0),
-                 PPCLowering->getPointerTy(CurDAG->getDataLayout()), MVT::Other,
-                 Ops));
+      SDNode *MN = CurDAG->getMachineNode(
+          Opcode, dl, LD->getValueType(0),
+          PPCLowering->getPointerTy(CurDAG->getDataLayout()), MVT::Other, Ops);
+      transferMemOperands(N, MN);
+      ReplaceNode(N, MN);
+      return;
     }
   }
 
@@ -2582,7 +2602,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue Val = N->getOperand(0).getOperand(0);
       SDValue Ops[] = { Val, getI32Imm(SH, dl), getI32Imm(MB, dl),
                         getI32Imm(ME, dl) };
-      return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      return;
     }
     // If this is just a masked value where the input is not handled above, and
     // is not a rotate-left (handled by a pattern in the .td file), emit rlwinm
@@ -2592,7 +2613,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue Val = N->getOperand(0);
       SDValue Ops[] = { Val, getI32Imm(0, dl), getI32Imm(MB, dl),
                         getI32Imm(ME, dl) };
-      return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      return;
     }
     // If this is a 64-bit zero-extension mask, emit rldicl.
     if (isInt64Immediate(N->getOperand(1).getNode(), Imm64) &&
@@ -2614,12 +2636,13 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       }
 
       SDValue Ops[] = { Val, getI32Imm(SH, dl), getI32Imm(MB, dl) };
-      return CurDAG->SelectNodeTo(N, PPC::RLDICL, MVT::i64, Ops);
+      CurDAG->SelectNodeTo(N, PPC::RLDICL, MVT::i64, Ops);
+      return;
     }
     // AND X, 0 -> 0, not "rlwinm 32".
     if (isInt32Immediate(N->getOperand(1), Imm) && (Imm == 0)) {
       ReplaceUses(SDValue(N, 0), N->getOperand(1));
-      return nullptr;
+      return;
     }
     // ISD::OR doesn't get all the bitfield insertion fun.
     // (and (or x, c1), c2) where isRunOfOnes(~(c1^c2)) might be a
@@ -2645,7 +2668,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
                             N->getOperand(0).getOperand(1),
                             getI32Imm(0, dl), getI32Imm(MB, dl),
                             getI32Imm(ME, dl) };
-        return CurDAG->getMachineNode(PPC::RLWIMI, dl, MVT::i32, Ops);
+        ReplaceNode(N, CurDAG->getMachineNode(PPC::RLWIMI, dl, MVT::i32, Ops));
+        return;
       }
     }
 
@@ -2654,8 +2678,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
   }
   case ISD::OR: {
     if (N->getValueType(0) == MVT::i32)
-      if (SDNode *I = SelectBitfieldInsert(N))
-        return I;
+      if (tryBitfieldInsert(N))
+        return;
 
     short Imm;
     if (N->getOperand(0)->getOpcode() == ISD::FrameIndex &&
@@ -2665,8 +2689,10 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
 
       // If this is equivalent to an add, then we can fold it with the
       // FrameIndex calculation.
-      if ((LHSKnownZero.getZExtValue()|~(uint64_t)Imm) == ~0ULL)
-        return getFrameIndex(N, N->getOperand(0).getNode(), (int)Imm);
+      if ((LHSKnownZero.getZExtValue()|~(uint64_t)Imm) == ~0ULL) {
+        selectFrameIndex(N, N->getOperand(0).getNode(), (int)Imm);
+        return;
+      }
     }
 
     // Other cases are autogenerated.
@@ -2675,8 +2701,10 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
   case ISD::ADD: {
     short Imm;
     if (N->getOperand(0)->getOpcode() == ISD::FrameIndex &&
-        isIntS16Immediate(N->getOperand(1), Imm))
-      return getFrameIndex(N, N->getOperand(0).getNode(), (int)Imm);
+        isIntS16Immediate(N->getOperand(1), Imm)) {
+      selectFrameIndex(N, N->getOperand(0).getNode(), (int)Imm);
+      return;
+    }
 
     break;
   }
@@ -2687,7 +2715,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue Ops[] = { N->getOperand(0).getOperand(0),
                           getI32Imm(SH, dl), getI32Imm(MB, dl),
                           getI32Imm(ME, dl) };
-      return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      return;
     }
 
     // Other cases are autogenerated.
@@ -2700,7 +2729,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue Ops[] = { N->getOperand(0).getOperand(0),
                           getI32Imm(SH, dl), getI32Imm(MB, dl),
                           getI32Imm(ME, dl) };
-      return CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      CurDAG->SelectNodeTo(N, PPC::RLWINM, MVT::i32, Ops);
+      return;
     }
 
     // Other cases are autogenerated.
@@ -2726,9 +2756,9 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       CurDAG->getTargetConstant(N->getOpcode() == PPCISD::ANDIo_1_EQ_BIT ?
                                 PPC::sub_eq : PPC::sub_gt, dl, MVT::i32);
 
-    return CurDAG->SelectNodeTo(N, TargetOpcode::EXTRACT_SUBREG, MVT::i1,
-                                CR0Reg, SRIdxVal,
-                                SDValue(AndI.getNode(), 1) /* glue */);
+    CurDAG->SelectNodeTo(N, TargetOpcode::EXTRACT_SUBREG, MVT::i1, CR0Reg,
+                         SRIdxVal, SDValue(AndI.getNode(), 1) /* glue */);
+    return;
   }
   case ISD::SELECT_CC: {
     ISD::CondCode CC = cast<CondCodeSDNode>(N->getOperand(4))->get();
@@ -2753,9 +2783,9 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
               SDNode *Tmp =
                 CurDAG->getMachineNode(PPC::ADDIC, dl, MVT::i32, MVT::Glue,
                                        N->getOperand(0), getI32Imm(~0U, dl));
-              return CurDAG->SelectNodeTo(N, PPC::SUBFE, MVT::i32,
-                                          SDValue(Tmp, 0), N->getOperand(0),
-                                          SDValue(Tmp, 1));
+              CurDAG->SelectNodeTo(N, PPC::SUBFE, MVT::i32, SDValue(Tmp, 0),
+                                   N->getOperand(0), SDValue(Tmp, 1));
+              return;
             }
 
     SDValue CCReg = SelectCC(N->getOperand(0), N->getOperand(1), CC, dl);
@@ -2786,7 +2816,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue NotCAndF(CurDAG->getMachineNode(PPC::CRAND, dl, MVT::i1,
                                               NotC, N->getOperand(3)), 0);
 
-      return CurDAG->SelectNodeTo(N, PPC::CROR, MVT::i1, CAndT, NotCAndF);
+      CurDAG->SelectNodeTo(N, PPC::CROR, MVT::i1, CAndT, NotCAndF);
+      return;
     }
 
     unsigned BROpc = getPredicateForSetCC(CC);
@@ -2820,12 +2851,14 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
 
     SDValue Ops[] = { CCReg, N->getOperand(2), N->getOperand(3),
                         getI32Imm(BROpc, dl) };
-    return CurDAG->SelectNodeTo(N, SelectCCOp, N->getValueType(0), Ops);
+    CurDAG->SelectNodeTo(N, SelectCCOp, N->getValueType(0), Ops);
+    return;
   }
   case ISD::VSELECT:
     if (PPCSubTarget->hasVSX()) {
       SDValue Ops[] = { N->getOperand(2), N->getOperand(1), N->getOperand(0) };
-      return CurDAG->SelectNodeTo(N, PPC::XXSEL, N->getValueType(0), Ops);
+      CurDAG->SelectNodeTo(N, PPC::XXSEL, N->getValueType(0), Ops);
+      return;
     }
 
     break;
@@ -2856,8 +2889,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
             SelectAddrIdxOnly(LD->getBasePtr(), Base, Offset)) {
           SDValue Chain = LD->getChain();
           SDValue Ops[] = { Base, Offset, Chain };
-          return CurDAG->SelectNodeTo(N, PPC::LXVDSX,
-                                      N->getValueType(0), Ops);
+          CurDAG->SelectNodeTo(N, PPC::LXVDSX, N->getValueType(0), Ops);
+          return;
         }
       }
 
@@ -2873,7 +2906,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue DMV = CurDAG->getTargetConstant(DM[1] | (DM[0] << 1), dl,
                                               MVT::i32);
       SDValue Ops[] = { Op1, Op2, DMV };
-      return CurDAG->SelectNodeTo(N, PPC::XXPERMDI, N->getValueType(0), Ops);
+      CurDAG->SelectNodeTo(N, PPC::XXPERMDI, N->getValueType(0), Ops);
+      return;
     }
 
     break;
@@ -2881,10 +2915,11 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
   case PPCISD::BDZ: {
     bool IsPPC64 = PPCSubTarget->isPPC64();
     SDValue Ops[] = { N->getOperand(1), N->getOperand(0) };
-    return CurDAG->SelectNodeTo(N, N->getOpcode() == PPCISD::BDNZ ?
-                                   (IsPPC64 ? PPC::BDNZ8 : PPC::BDNZ) :
-                                   (IsPPC64 ? PPC::BDZ8 : PPC::BDZ),
-                                MVT::Other, Ops);
+    CurDAG->SelectNodeTo(N, N->getOpcode() == PPCISD::BDNZ
+                                ? (IsPPC64 ? PPC::BDNZ8 : PPC::BDNZ)
+                                : (IsPPC64 ? PPC::BDZ8 : PPC::BDZ),
+                         MVT::Other, Ops);
+    return;
   }
   case PPCISD::COND_BRANCH: {
     // Op #0 is the Chain.
@@ -2900,7 +2935,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
     SDValue Pred = getI32Imm(PCC, dl);
     SDValue Ops[] = { Pred, N->getOperand(2), N->getOperand(3),
       N->getOperand(0), N->getOperand(4) };
-    return CurDAG->SelectNodeTo(N, PPC::BCC, MVT::Other, Ops);
+    CurDAG->SelectNodeTo(N, PPC::BCC, MVT::Other, Ops);
+    return;
   }
   case ISD::BR_CC: {
     ISD::CondCode CC = cast<CondCodeSDNode>(N->getOperand(1))->get();
@@ -2922,8 +2958,9 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue BitComp(CurDAG->getMachineNode(Opc, dl, MVT::i1,
                                              N->getOperand(Swap ? 3 : 2),
                                              N->getOperand(Swap ? 2 : 3)), 0);
-      return CurDAG->SelectNodeTo(N, PPC::BC, MVT::Other,
-                                  BitComp, N->getOperand(4), N->getOperand(0));
+      CurDAG->SelectNodeTo(N, PPC::BC, MVT::Other, BitComp, N->getOperand(4),
+                           N->getOperand(0));
+      return;
     }
 
     if (EnableBranchHint)
@@ -2932,7 +2969,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
     SDValue CondCode = SelectCC(N->getOperand(2), N->getOperand(3), CC, dl);
     SDValue Ops[] = { getI32Imm(PCC, dl), CondCode,
                         N->getOperand(4), N->getOperand(0) };
-    return CurDAG->SelectNodeTo(N, PPC::BCC, MVT::Other, Ops);
+    CurDAG->SelectNodeTo(N, PPC::BCC, MVT::Other, Ops);
+    return;
   }
   case ISD::BRIND: {
     // FIXME: Should custom lower this.
@@ -2942,15 +2980,19 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
     unsigned Reg = Target.getValueType() == MVT::i32 ? PPC::BCTR : PPC::BCTR8;
     Chain = SDValue(CurDAG->getMachineNode(Opc, dl, MVT::Glue, Target,
                                            Chain), 0);
-    return CurDAG->SelectNodeTo(N, Reg, MVT::Other, Chain);
+    CurDAG->SelectNodeTo(N, Reg, MVT::Other, Chain);
+    return;
   }
   case PPCISD::TOC_ENTRY: {
     assert ((PPCSubTarget->isPPC64() || PPCSubTarget->isSVR4ABI()) &&
             "Only supported for 64-bit ABI and 32-bit SVR4");
     if (PPCSubTarget->isSVR4ABI() && !PPCSubTarget->isPPC64()) {
       SDValue GA = N->getOperand(0);
-      return transferMemOperands(N, CurDAG->getMachineNode(PPC::LWZtoc, dl,
-                                      MVT::i32, GA, N->getOperand(1)));
+      SDNode *MN = CurDAG->getMachineNode(PPC::LWZtoc, dl, MVT::i32, GA,
+                                          N->getOperand(1));
+      transferMemOperands(N, MN);
+      ReplaceNode(N, MN);
+      return;
     }
 
     // For medium and large code model, we generate two instructions as
@@ -2971,29 +3013,38 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
                                          TOCbase, GA);
 
     if (isa<JumpTableSDNode>(GA) || isa<BlockAddressSDNode>(GA) ||
-        CModel == CodeModel::Large)
-      return transferMemOperands(N, CurDAG->getMachineNode(PPC::LDtocL, dl,
-                                      MVT::i64, GA, SDValue(Tmp, 0)));
+        CModel == CodeModel::Large) {
+      SDNode *MN = CurDAG->getMachineNode(PPC::LDtocL, dl, MVT::i64, GA,
+                                          SDValue(Tmp, 0));
+      transferMemOperands(N, MN);
+      ReplaceNode(N, MN);
+      return;
+    }
 
     if (GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(GA)) {
       const GlobalValue *GV = G->getGlobal();
       unsigned char GVFlags = PPCSubTarget->classifyGlobalReference(GV);
       if (GVFlags & PPCII::MO_NLP_FLAG) {
-        return transferMemOperands(N, CurDAG->getMachineNode(PPC::LDtocL, dl,
-                                        MVT::i64, GA, SDValue(Tmp, 0)));
+        SDNode *MN = CurDAG->getMachineNode(PPC::LDtocL, dl, MVT::i64, GA,
+                                            SDValue(Tmp, 0));
+        transferMemOperands(N, MN);
+        ReplaceNode(N, MN);
+        return;
       }
     }
 
-    return CurDAG->getMachineNode(PPC::ADDItocL, dl, MVT::i64,
-                                  SDValue(Tmp, 0), GA);
+    ReplaceNode(N, CurDAG->getMachineNode(PPC::ADDItocL, dl, MVT::i64,
+                                          SDValue(Tmp, 0), GA));
+    return;
   }
   case PPCISD::PPC32_PICGOT: {
     // Generate a PIC-safe GOT reference.
     assert(!PPCSubTarget->isPPC64() && PPCSubTarget->isSVR4ABI() &&
       "PPCISD::PPC32_PICGOT is only supported for 32-bit SVR4");
-    return CurDAG->SelectNodeTo(
-        N, PPC::PPC32PICGOT, PPCLowering->getPointerTy(CurDAG->getDataLayout()),
-        MVT::i32);
+    CurDAG->SelectNodeTo(N, PPC::PPC32PICGOT,
+                         PPCLowering->getPointerTy(CurDAG->getDataLayout()),
+                         MVT::i32);
+    return;
   }
   case PPCISD::VADD_SPLAT: {
     // This expands into one of three sequences, depending on whether
@@ -3035,7 +3086,8 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDValue EltVal = getI32Imm(Elt >> 1, dl);
       SDNode *Tmp = CurDAG->getMachineNode(Opc1, dl, VT, EltVal);
       SDValue TmpVal = SDValue(Tmp, 0);
-      return CurDAG->getMachineNode(Opc2, dl, VT, TmpVal, TmpVal);
+      ReplaceNode(N, CurDAG->getMachineNode(Opc2, dl, VT, TmpVal, TmpVal));
+      return;
 
     } else if (Elt > 0) {
       // Elt is odd and positive, in the range [17,31].
@@ -3048,8 +3100,9 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDNode *Tmp1 = CurDAG->getMachineNode(Opc1, dl, VT, EltVal);
       EltVal = getI32Imm(-16, dl);
       SDNode *Tmp2 = CurDAG->getMachineNode(Opc1, dl, VT, EltVal);
-      return CurDAG->getMachineNode(Opc3, dl, VT, SDValue(Tmp1, 0),
-                                    SDValue(Tmp2, 0));
+      ReplaceNode(N, CurDAG->getMachineNode(Opc3, dl, VT, SDValue(Tmp1, 0),
+                                            SDValue(Tmp2, 0)));
+      return;
 
     } else {
       // Elt is odd and negative, in the range [-31,-17].
@@ -3062,13 +3115,14 @@ SDNode *PPCDAGToDAGISel::Select(SDNode *N) {
       SDNode *Tmp1 = CurDAG->getMachineNode(Opc1, dl, VT, EltVal);
       EltVal = getI32Imm(-16, dl);
       SDNode *Tmp2 = CurDAG->getMachineNode(Opc1, dl, VT, EltVal);
-      return CurDAG->getMachineNode(Opc2, dl, VT, SDValue(Tmp1, 0),
-                                    SDValue(Tmp2, 0));
+      ReplaceNode(N, CurDAG->getMachineNode(Opc2, dl, VT, SDValue(Tmp1, 0),
+                                            SDValue(Tmp2, 0)));
+      return;
     }
   }
   }
 
-  return SelectCode(N);
+  SelectCode(N);
 }
 
 // If the target supports the cmpb instruction, do the idiom recognition here.
@@ -3565,11 +3619,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(0),
                                            MachineNode->getOperand(1).
                                              getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CRNAND, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(0),
-                                           MachineNode->getOperand(1)),
+                                           MachineNode->getOperand(1));
           SelectSwap = true;
+        }
         break;
       case PPC::CRNAND:
         if (MachineNode->getOperand(0) == MachineNode->getOperand(1))
@@ -3603,11 +3658,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(1).
                                                       getOperand(0),
                                            MachineNode->getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CRAND, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(0),
-                                           MachineNode->getOperand(1)),
+                                           MachineNode->getOperand(1));
           SelectSwap = true;
+        }
         break;
       case PPC::CROR:
         if (MachineNode->getOperand(0) == MachineNode->getOperand(1))
@@ -3635,11 +3691,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(0),
                                            MachineNode->getOperand(1).
                                              getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CRNOR, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(0),
-                                           MachineNode->getOperand(1)),
+                                           MachineNode->getOperand(1));
           SelectSwap = true;
+        }
         break;
       case PPC::CRXOR:
         if (MachineNode->getOperand(0) == MachineNode->getOperand(1))
@@ -3674,11 +3731,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(0),
                                            MachineNode->getOperand(1).
                                              getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CREQV, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(0),
-                                           MachineNode->getOperand(1)),
+                                           MachineNode->getOperand(1));
           SelectSwap = true;
+        }
         break;
       case PPC::CRNOR:
         if (Op1Set || Op2Set)
@@ -3707,11 +3765,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(1).
                                                       getOperand(0),
                                            MachineNode->getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CROR, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(0),
-                                           MachineNode->getOperand(1)),
+                                           MachineNode->getOperand(1));
           SelectSwap = true;
+        }
         break;
       case PPC::CREQV:
         if (MachineNode->getOperand(0) == MachineNode->getOperand(1))
@@ -3746,11 +3805,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(0),
                                            MachineNode->getOperand(1).
                                              getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CRXOR, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(0),
-                                           MachineNode->getOperand(1)),
+                                           MachineNode->getOperand(1));
           SelectSwap = true;
+        }
         break;
       case PPC::CRANDC:
         if (MachineNode->getOperand(0) == MachineNode->getOperand(1))
@@ -3781,11 +3841,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(0),
                                            MachineNode->getOperand(1).
                                              getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CRORC, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(1),
-                                           MachineNode->getOperand(0)),
+                                           MachineNode->getOperand(0));
           SelectSwap = true;
+        }
         break;
       case PPC::CRORC:
         if (MachineNode->getOperand(0) == MachineNode->getOperand(1))
@@ -3816,11 +3877,12 @@ void PPCDAGToDAGISel::PeepholeCROps() {
                                            MVT::i1, MachineNode->getOperand(0),
                                            MachineNode->getOperand(1).
                                              getOperand(0));
-        else if (AllUsersSelectZero(MachineNode))
+        else if (AllUsersSelectZero(MachineNode)) {
           ResNode = CurDAG->getMachineNode(PPC::CRANDC, SDLoc(MachineNode),
                                            MVT::i1, MachineNode->getOperand(1),
-                                           MachineNode->getOperand(0)),
+                                           MachineNode->getOperand(0));
           SelectSwap = true;
+        }
         break;
       case PPC::SELECT_I4:
       case PPC::SELECT_I8:
@@ -4365,15 +4427,3 @@ void PPCDAGToDAGISel::PeepholePPC64() {
 FunctionPass *llvm::createPPCISelDag(PPCTargetMachine &TM) {
   return new PPCDAGToDAGISel(TM);
 }
-
-static void initializePassOnce(PassRegistry &Registry) {
-  const char *Name = "PowerPC DAG->DAG Pattern Instruction Selection";
-  PassInfo *PI = new PassInfo(Name, "ppc-codegen", &SelectionDAGISel::ID,
-                              nullptr, false, false);
-  Registry.registerPass(*PI, true);
-}
-
-void llvm::initializePPCDAGToDAGISelPass(PassRegistry &Registry) {
-  CALL_ONCE_INITIALIZATION(initializePassOnce);
-}
-
diff --git a/lib/Target/PowerPC/PPCISelLowering.cpp b/lib/Target/PowerPC/PPCISelLowering.cpp
index af9ad077a7ce1..6e3c830a82431 100644
--- a/lib/Target/PowerPC/PPCISelLowering.cpp
+++ b/lib/Target/PowerPC/PPCISelLowering.cpp
@@ -14,11 +14,13 @@
 #include "PPCISelLowering.h"
 #include "MCTargetDesc/PPCPredicates.h"
 #include "PPCCallingConv.h"
+#include "PPCCCState.h"
 #include "PPCMachineFunctionInfo.h"
 #include "PPCPerfectShuffle.h"
 #include "PPCTargetMachine.h"
 #include "PPCTargetObjectFile.h"
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/ADT/Triple.h"
 #include "llvm/CodeGen/CallingConvLower.h"
@@ -36,12 +38,16 @@
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/Format.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetOptions.h"
+#include <list>
 
 using namespace llvm;
 
+#define DEBUG_TYPE "ppc-lowering"
+
 static cl::opt<bool> DisablePPCPreinc("disable-ppc-preinc",
 cl::desc("disable preincrement load/store generation on PPC"), cl::Hidden);
 
@@ -51,6 +57,12 @@ cl::desc("disable setting the node scheduling preference to ILP on PPC"), cl::Hi
 static cl::opt<bool> DisablePPCUnaligned("disable-ppc-unaligned",
 cl::desc("disable unaligned load/store generation on PPC"), cl::Hidden);
 
+static cl::opt<bool> DisableSCO("disable-ppc-sco",
+cl::desc("disable sibling call optimization on ppc"), cl::Hidden);
+
+STATISTIC(NumTailCalls, "Number of tail calls");
+STATISTIC(NumSiblingCalls, "Number of sibling calls");
+
 // FIXME: Remove this once the bug has been fixed!
 extern cl::opt<bool> ANDIGlueBug;
 
@@ -68,7 +80,7 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
 
   // Set up the register classes.
   addRegisterClass(MVT::i32, &PPC::GPRCRegClass);
-  if (!Subtarget.useSoftFloat()) {
+  if (!useSoftFloat()) {
     addRegisterClass(MVT::f32, &PPC::F4RCRegClass);
     addRegisterClass(MVT::f64, &PPC::F8RCRegClass);
   }
@@ -207,14 +219,10 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
   // PowerPC does not have BSWAP, CTPOP or CTTZ
   setOperationAction(ISD::BSWAP, MVT::i32  , Expand);
   setOperationAction(ISD::CTTZ , MVT::i32  , Expand);
-  setOperationAction(ISD::CTTZ_ZERO_UNDEF, MVT::i32, Expand);
-  setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i32, Expand);
   setOperationAction(ISD::BSWAP, MVT::i64  , Expand);
   setOperationAction(ISD::CTTZ , MVT::i64  , Expand);
-  setOperationAction(ISD::CTTZ_ZERO_UNDEF, MVT::i64, Expand);
-  setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i64, Expand);
 
-  if (Subtarget.hasPOPCNTD()) {
+  if (Subtarget.hasPOPCNTD() == PPCSubtarget::POPCNTD_Fast) {
     setOperationAction(ISD::CTPOP, MVT::i32  , Legal);
     setOperationAction(ISD::CTPOP, MVT::i64  , Legal);
   } else {
@@ -255,7 +263,7 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
   setOperationAction(ISD::SINT_TO_FP, MVT::i32, Expand);
   setOperationAction(ISD::UINT_TO_FP, MVT::i32, Expand);
 
-  if (Subtarget.hasDirectMove()) {
+  if (Subtarget.hasDirectMove() && isPPC64) {
     setOperationAction(ISD::BITCAST, MVT::f32, Legal);
     setOperationAction(ISD::BITCAST, MVT::i32, Legal);
     setOperationAction(ISD::BITCAST, MVT::i64, Legal);
@@ -479,9 +487,7 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
       setOperationAction(ISD::SCALAR_TO_VECTOR, VT, Expand);
       setOperationAction(ISD::FPOW, VT, Expand);
       setOperationAction(ISD::BSWAP, VT, Expand);
-      setOperationAction(ISD::CTLZ_ZERO_UNDEF, VT, Expand);
       setOperationAction(ISD::CTTZ, VT, Expand);
-      setOperationAction(ISD::CTTZ_ZERO_UNDEF, VT, Expand);
       setOperationAction(ISD::VSELECT, VT, Expand);
       setOperationAction(ISD::SIGN_EXTEND_INREG, VT, Expand);
       setOperationAction(ISD::ROTL, VT, Expand);
@@ -557,7 +563,7 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
         setOperationAction(ISD::SCALAR_TO_VECTOR, MVT::v4f32, Legal);
         setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::v4f32, Legal);
       }
-      if (Subtarget.hasDirectMove()) {
+      if (Subtarget.hasDirectMove() && isPPC64) {
         setOperationAction(ISD::SCALAR_TO_VECTOR, MVT::v16i8, Legal);
         setOperationAction(ISD::SCALAR_TO_VECTOR, MVT::v8i16, Legal);
         setOperationAction(ISD::SCALAR_TO_VECTOR, MVT::v4i32, Legal);
@@ -647,6 +653,11 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
       setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::v2i16, Custom);
       setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::v2i8, Custom);
 
+      setOperationAction(ISD::FNEG, MVT::v4f32, Legal);
+      setOperationAction(ISD::FNEG, MVT::v2f64, Legal);
+      setOperationAction(ISD::FABS, MVT::v4f32, Legal);
+      setOperationAction(ISD::FABS, MVT::v2f64, Legal);
+
       addRegisterClass(MVT::v2i64, &PPC::VSRCRegClass);
     }
 
@@ -654,6 +665,10 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
       addRegisterClass(MVT::v2i64, &PPC::VRRCRegClass);
       addRegisterClass(MVT::v1i128, &PPC::VRRCRegClass);
     }
+    if (Subtarget.hasP9Vector()) {
+      setOperationAction(ISD::INSERT_VECTOR_ELT, MVT::v4i32, Legal);
+      setOperationAction(ISD::INSERT_VECTOR_ELT, MVT::v4f32, Legal);
+    }
   }
 
   if (Subtarget.hasQPX()) {
@@ -840,6 +855,7 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
 
   // We have target-specific dag combine patterns for the following nodes:
   setTargetDAGCombine(ISD::SINT_TO_FP);
+  setTargetDAGCombine(ISD::BUILD_VECTOR);
   if (Subtarget.hasFPCVT())
     setTargetDAGCombine(ISD::UINT_TO_FP);
   setTargetDAGCombine(ISD::LOAD);
@@ -906,13 +922,12 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
   case PPC::DIR_PWR6X:
   case PPC::DIR_PWR7:
   case PPC::DIR_PWR8:
+  case PPC::DIR_PWR9:
     setPrefFunctionAlignment(4);
     setPrefLoopAlignment(4);
     break;
   }
 
-  setInsertFencesForAtomic(true);
-
   if (Subtarget.enableMachineScheduler())
     setSchedulingPreference(Sched::Source);
   else
@@ -1006,6 +1021,9 @@ const char *PPCTargetLowering::getTargetNodeName(unsigned Opcode) const {
   case PPCISD::VMADDFP:         return "PPCISD::VMADDFP";
   case PPCISD::VNMSUBFP:        return "PPCISD::VNMSUBFP";
   case PPCISD::VPERM:           return "PPCISD::VPERM";
+  case PPCISD::XXSPLT:          return "PPCISD::XXSPLT";
+  case PPCISD::XXINSERT:        return "PPCISD::XXINSERT";
+  case PPCISD::VECSHL:          return "PPCISD::VECSHL";
   case PPCISD::CMPB:            return "PPCISD::CMPB";
   case PPCISD::Hi:              return "PPCISD::Hi";
   case PPCISD::Lo:              return "PPCISD::Lo";
@@ -1030,6 +1048,8 @@ const char *PPCTargetLowering::getTargetNodeName(unsigned Opcode) const {
   case PPCISD::MFVSR:           return "PPCISD::MFVSR";
   case PPCISD::MTVSRA:          return "PPCISD::MTVSRA";
   case PPCISD::MTVSRZ:          return "PPCISD::MTVSRZ";
+  case PPCISD::SINT_VEC_TO_FP:  return "PPCISD::SINT_VEC_TO_FP";
+  case PPCISD::UINT_VEC_TO_FP:  return "PPCISD::UINT_VEC_TO_FP";
   case PPCISD::ANDIo_1_EQ_BIT:  return "PPCISD::ANDIo_1_EQ_BIT";
   case PPCISD::ANDIo_1_GT_BIT:  return "PPCISD::ANDIo_1_GT_BIT";
   case PPCISD::VCMP:            return "PPCISD::VCMP";
@@ -1069,6 +1089,7 @@ const char *PPCTargetLowering::getTargetNodeName(unsigned Opcode) const {
   case PPCISD::MFBHRBE:         return "PPCISD::MFBHRBE";
   case PPCISD::RFEBB:           return "PPCISD::RFEBB";
   case PPCISD::XXSWAPD:         return "PPCISD::XXSWAPD";
+  case PPCISD::SWAP_NO_CHAIN:   return "PPCISD::SWAP_NO_CHAIN";
   case PPCISD::QVFPERM:         return "PPCISD::QVFPERM";
   case PPCISD::QVGPCI:          return "PPCISD::QVGPCI";
   case PPCISD::QVALIGNI:        return "PPCISD::QVALIGNI";
@@ -1480,6 +1501,91 @@ bool PPC::isSplatShuffleMask(ShuffleVectorSDNode *N, unsigned EltSize) {
   return true;
 }
 
+bool PPC::isXXINSERTWMask(ShuffleVectorSDNode *N, unsigned &ShiftElts,
+                          unsigned &InsertAtByte, bool &Swap, bool IsLE) {
+
+  // Check that the mask is shuffling words
+  for (unsigned i = 0; i < 4; ++i) {
+    unsigned B0 = N->getMaskElt(i*4);
+    unsigned B1 = N->getMaskElt(i*4+1);
+    unsigned B2 = N->getMaskElt(i*4+2);
+    unsigned B3 = N->getMaskElt(i*4+3);
+    if (B0 % 4)
+      return false;
+    if (B1 != B0+1 || B2 != B1+1 || B3 != B2+1)
+      return false;
+  }
+
+  // Now we look at mask elements 0,4,8,12
+  unsigned M0 = N->getMaskElt(0) / 4;
+  unsigned M1 = N->getMaskElt(4) / 4;
+  unsigned M2 = N->getMaskElt(8) / 4;
+  unsigned M3 = N->getMaskElt(12) / 4;
+  unsigned LittleEndianShifts[] = { 2, 1, 0, 3 };
+  unsigned BigEndianShifts[] = { 3, 0, 1, 2 };
+
+  // Below, let H and L be arbitrary elements of the shuffle mask
+  // where H is in the range [4,7] and L is in the range [0,3].
+  // H, 1, 2, 3 or L, 5, 6, 7
+  if ((M0 > 3 && M1 == 1 && M2 == 2 && M3 == 3) ||
+      (M0 < 4 && M1 == 5 && M2 == 6 && M3 == 7)) {
+    ShiftElts = IsLE ? LittleEndianShifts[M0 & 0x3] : BigEndianShifts[M0 & 0x3];
+    InsertAtByte = IsLE ? 12 : 0;
+    Swap = M0 < 4;
+    return true;
+  }
+  // 0, H, 2, 3 or 4, L, 6, 7
+  if ((M1 > 3 && M0 == 0 && M2 == 2 && M3 == 3) ||
+      (M1 < 4 && M0 == 4 && M2 == 6 && M3 == 7)) {
+    ShiftElts = IsLE ? LittleEndianShifts[M1 & 0x3] : BigEndianShifts[M1 & 0x3];
+    InsertAtByte = IsLE ? 8 : 4;
+    Swap = M1 < 4;
+    return true;
+  }
+  // 0, 1, H, 3 or 4, 5, L, 7
+  if ((M2 > 3 && M0 == 0 && M1 == 1 && M3 == 3) ||
+      (M2 < 4 && M0 == 4 && M1 == 5 && M3 == 7)) {
+    ShiftElts = IsLE ? LittleEndianShifts[M2 & 0x3] : BigEndianShifts[M2 & 0x3];
+    InsertAtByte = IsLE ? 4 : 8;
+    Swap = M2 < 4;
+    return true;
+  }
+  // 0, 1, 2, H or 4, 5, 6, L
+  if ((M3 > 3 && M0 == 0 && M1 == 1 && M2 == 2) ||
+      (M3 < 4 && M0 == 4 && M1 == 5 && M2 == 6)) {
+    ShiftElts = IsLE ? LittleEndianShifts[M3 & 0x3] : BigEndianShifts[M3 & 0x3];
+    InsertAtByte = IsLE ? 0 : 12;
+    Swap = M3 < 4;
+    return true;
+  }
+
+  // If both vector operands for the shuffle are the same vector, the mask will
+  // contain only elements from the first one and the second one will be undef.
+  if (N->getOperand(1).isUndef()) {
+    ShiftElts = 0;
+    Swap = true;
+    unsigned XXINSERTWSrcElem = IsLE ? 2 : 1;
+    if (M0 == XXINSERTWSrcElem && M1 == 1 && M2 == 2 && M3 == 3) {
+      InsertAtByte = IsLE ? 12 : 0;
+      return true;
+    }
+    if (M0 == 0 && M1 == XXINSERTWSrcElem && M2 == 2 && M3 == 3) {
+      InsertAtByte = IsLE ? 8 : 4;
+      return true;
+    }
+    if (M0 == 0 && M1 == 1 && M2 == XXINSERTWSrcElem && M3 == 3) {
+      InsertAtByte = IsLE ? 4 : 8;
+      return true;
+    }
+    if (M0 == 0 && M1 == 1 && M2 == 2 && M3 == XXINSERTWSrcElem) {
+      InsertAtByte = IsLE ? 0 : 12;
+      return true;
+    }
+  }
+
+  return false;
+}
+
 /// getVSPLTImmediate - Return the appropriate VSPLT* immediate to splat the
 /// specified isSplatShuffleMask VECTOR_SHUFFLE mask.
 unsigned PPC::getVSPLTImmediate(SDNode *N, unsigned EltSize,
@@ -1511,7 +1617,7 @@ SDValue PPC::get_VSPLTI_elt(SDNode *N, unsigned ByteSize, SelectionDAG &DAG) {
 
     // See if all of the elements in the buildvector agree across.
     for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) {
-      if (N->getOperand(i).getOpcode() == ISD::UNDEF) continue;
+      if (N->getOperand(i).isUndef()) continue;
       // If the element isn't a constant, bail fully out.
       if (!isa<ConstantSDNode>(N->getOperand(i))) return SDValue();
 
@@ -1557,7 +1663,7 @@ SDValue PPC::get_VSPLTI_elt(SDNode *N, unsigned ByteSize, SelectionDAG &DAG) {
 
   // Check to see if this buildvec has a single non-undef value in its elements.
   for (unsigned i = 0, e = N->getNumOperands(); i != e; ++i) {
-    if (N->getOperand(i).getOpcode() == ISD::UNDEF) continue;
+    if (N->getOperand(i).isUndef()) continue;
     if (!OpVal.getNode())
       OpVal = N->getOperand(i);
     else if (OpVal != N->getOperand(i))
@@ -1950,19 +2056,16 @@ bool PPCTargetLowering::getPreIndexedAddressParts(SDNode *N, SDValue &Base,
 //  LowerOperation implementation
 //===----------------------------------------------------------------------===//
 
-/// GetLabelAccessInfo - Return true if we should reference labels using a
-/// PICBase, set the HiOpFlags and LoOpFlags to the target MO flags.
-static bool GetLabelAccessInfo(const TargetMachine &TM,
-                               const PPCSubtarget &Subtarget,
+/// Return true if we should reference labels using a PICBase, set the HiOpFlags
+/// and LoOpFlags to the target MO flags.
+static void getLabelAccessInfo(bool IsPIC, const PPCSubtarget &Subtarget,
                                unsigned &HiOpFlags, unsigned &LoOpFlags,
                                const GlobalValue *GV = nullptr) {
   HiOpFlags = PPCII::MO_HA;
   LoOpFlags = PPCII::MO_LO;
 
   // Don't use the pic base if not in PIC relocation model.
-  bool isPIC = TM.getRelocationModel() == Reloc::PIC_;
-
-  if (isPIC) {
+  if (IsPIC) {
     HiOpFlags |= PPCII::MO_PIC_FLAG;
     LoOpFlags |= PPCII::MO_PIC_FLAG;
   }
@@ -1978,8 +2081,6 @@ static bool GetLabelAccessInfo(const TargetMachine &TM,
       LoOpFlags |= PPCII::MO_NLP_HIDDEN_FLAG;
     }
   }
-
-  return isPIC;
 }
 
 static SDValue LowerLabelRef(SDValue HiPart, SDValue LoPart, bool isPIC,
@@ -2010,7 +2111,7 @@ static void setUsesTOCBasePtr(SelectionDAG &DAG) {
   setUsesTOCBasePtr(DAG.getMachineFunction());
 }
 
-static SDValue getTOCEntry(SelectionDAG &DAG, SDLoc dl, bool Is64Bit,
+static SDValue getTOCEntry(SelectionDAG &DAG, const SDLoc &dl, bool Is64Bit,
                            SDValue GA) {
   EVT VT = Is64Bit ? MVT::i64 : MVT::i32;
   SDValue Reg = Is64Bit ? DAG.getRegister(PPC::X2, VT) :
@@ -2038,10 +2139,10 @@ SDValue PPCTargetLowering::LowerConstantPool(SDValue Op,
   }
 
   unsigned MOHiFlag, MOLoFlag;
-  bool isPIC =
-      GetLabelAccessInfo(DAG.getTarget(), Subtarget, MOHiFlag, MOLoFlag);
+  bool IsPIC = isPositionIndependent();
+  getLabelAccessInfo(IsPIC, Subtarget, MOHiFlag, MOLoFlag);
 
-  if (isPIC && Subtarget.isSVR4ABI()) {
+  if (IsPIC && Subtarget.isSVR4ABI()) {
     SDValue GA = DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment(),
                                            PPCII::MO_PIC_FLAG);
     return getTOCEntry(DAG, SDLoc(CP), false, GA);
@@ -2051,7 +2152,7 @@ SDValue PPCTargetLowering::LowerConstantPool(SDValue Op,
     DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment(), 0, MOHiFlag);
   SDValue CPILo =
     DAG.getTargetConstantPool(C, PtrVT, CP->getAlignment(), 0, MOLoFlag);
-  return LowerLabelRef(CPIHi, CPILo, isPIC, DAG);
+  return LowerLabelRef(CPIHi, CPILo, IsPIC, DAG);
 }
 
 SDValue PPCTargetLowering::LowerJumpTable(SDValue Op, SelectionDAG &DAG) const {
@@ -2067,10 +2168,10 @@ SDValue PPCTargetLowering::LowerJumpTable(SDValue Op, SelectionDAG &DAG) const {
   }
 
   unsigned MOHiFlag, MOLoFlag;
-  bool isPIC =
-      GetLabelAccessInfo(DAG.getTarget(), Subtarget, MOHiFlag, MOLoFlag);
+  bool IsPIC = isPositionIndependent();
+  getLabelAccessInfo(IsPIC, Subtarget, MOHiFlag, MOLoFlag);
 
-  if (isPIC && Subtarget.isSVR4ABI()) {
+  if (IsPIC && Subtarget.isSVR4ABI()) {
     SDValue GA = DAG.getTargetJumpTable(JT->getIndex(), PtrVT,
                                         PPCII::MO_PIC_FLAG);
     return getTOCEntry(DAG, SDLoc(GA), false, GA);
@@ -2078,7 +2179,7 @@ SDValue PPCTargetLowering::LowerJumpTable(SDValue Op, SelectionDAG &DAG) const {
 
   SDValue JTIHi = DAG.getTargetJumpTable(JT->getIndex(), PtrVT, MOHiFlag);
   SDValue JTILo = DAG.getTargetJumpTable(JT->getIndex(), PtrVT, MOLoFlag);
-  return LowerLabelRef(JTIHi, JTILo, isPIC, DAG);
+  return LowerLabelRef(JTIHi, JTILo, IsPIC, DAG);
 }
 
 SDValue PPCTargetLowering::LowerBlockAddress(SDValue Op,
@@ -2096,11 +2197,11 @@ SDValue PPCTargetLowering::LowerBlockAddress(SDValue Op,
   }
 
   unsigned MOHiFlag, MOLoFlag;
-  bool isPIC =
-      GetLabelAccessInfo(DAG.getTarget(), Subtarget, MOHiFlag, MOLoFlag);
+  bool IsPIC = isPositionIndependent();
+  getLabelAccessInfo(IsPIC, Subtarget, MOHiFlag, MOLoFlag);
   SDValue TgtBAHi = DAG.getTargetBlockAddress(BA, PtrVT, 0, MOHiFlag);
   SDValue TgtBALo = DAG.getTargetBlockAddress(BA, PtrVT, 0, MOLoFlag);
-  return LowerLabelRef(TgtBAHi, TgtBALo, isPIC, DAG);
+  return LowerLabelRef(TgtBAHi, TgtBALo, IsPIC, DAG);
 }
 
 SDValue PPCTargetLowering::LowerGlobalTLSAddress(SDValue Op,
@@ -2160,7 +2261,7 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(SDValue Op,
       GOTPtr = DAG.getNode(PPCISD::ADDIS_TLSGD_HA, dl, PtrVT,
                                    GOTReg, TGA);
     } else {
-      if (picLevel == PICLevel::Small)
+      if (picLevel == PICLevel::SmallPIC)
         GOTPtr = DAG.getNode(PPCISD::GlobalBaseReg, dl, PtrVT);
       else
         GOTPtr = DAG.getNode(PPCISD::PPC32_PICGOT, dl, PtrVT);
@@ -2178,7 +2279,7 @@ SDValue PPCTargetLowering::LowerGlobalTLSAddress(SDValue Op,
       GOTPtr = DAG.getNode(PPCISD::ADDIS_TLSLD_HA, dl, PtrVT,
                            GOTReg, TGA);
     } else {
-      if (picLevel == PICLevel::Small)
+      if (picLevel == PICLevel::SmallPIC)
         GOTPtr = DAG.getNode(PPCISD::GlobalBaseReg, dl, PtrVT);
       else
         GOTPtr = DAG.getNode(PPCISD::PPC32_PICGOT, dl, PtrVT);
@@ -2209,10 +2310,10 @@ SDValue PPCTargetLowering::LowerGlobalAddress(SDValue Op,
   }
 
   unsigned MOHiFlag, MOLoFlag;
-  bool isPIC =
-      GetLabelAccessInfo(DAG.getTarget(), Subtarget, MOHiFlag, MOLoFlag, GV);
+  bool IsPIC = isPositionIndependent();
+  getLabelAccessInfo(IsPIC, Subtarget, MOHiFlag, MOLoFlag, GV);
 
-  if (isPIC && Subtarget.isSVR4ABI()) {
+  if (IsPIC && Subtarget.isSVR4ABI()) {
     SDValue GA = DAG.getTargetGlobalAddress(GV, DL, PtrVT,
                                             GSDN->getOffset(),
                                             PPCII::MO_PIC_FLAG);
@@ -2224,13 +2325,12 @@ SDValue PPCTargetLowering::LowerGlobalAddress(SDValue Op,
   SDValue GALo =
     DAG.getTargetGlobalAddress(GV, DL, PtrVT, GSDN->getOffset(), MOLoFlag);
 
-  SDValue Ptr = LowerLabelRef(GAHi, GALo, isPIC, DAG);
+  SDValue Ptr = LowerLabelRef(GAHi, GALo, IsPIC, DAG);
 
   // If the global reference is actually to a non-lazy-pointer, we have to do an
   // extra load to get the address of the global.
   if (MOHiFlag & PPCII::MO_NLP_FLAG)
-    Ptr = DAG.getLoad(PtrVT, DL, DAG.getEntryNode(), Ptr, MachinePointerInfo(),
-                      false, false, false, 0);
+    Ptr = DAG.getLoad(PtrVT, DL, DAG.getEntryNode(), Ptr, MachinePointerInfo());
   return Ptr;
 }
 
@@ -2260,7 +2360,7 @@ SDValue PPCTargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) const {
   }
 
   // If we're comparing for equality to zero, expose the fact that this is
-  // implented as a ctlz/srl pair on ppc, so that the dag combiner can
+  // implemented as a ctlz/srl pair on ppc, so that the dag combiner can
   // fold the new nodes.
   if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Op.getOperand(1))) {
     if (C->isNullValue() && CC == ISD::SETEQ) {
@@ -2298,11 +2398,10 @@ SDValue PPCTargetLowering::LowerSETCC(SDValue Op, SelectionDAG &DAG) const {
   return SDValue();
 }
 
-SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
-                                      const PPCSubtarget &Subtarget) const {
+SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG) const {
   SDNode *Node = Op.getNode();
   EVT VT = Node->getValueType(0);
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
   SDValue InChain = Node->getOperand(0);
   SDValue VAListPtr = Node->getOperand(1);
   const Value *SV = cast<SrcValueSDNode>(Node->getOperand(2))->getValue();
@@ -2312,8 +2411,7 @@ SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
 
   // gpr_index
   SDValue GprIndex = DAG.getExtLoad(ISD::ZEXTLOAD, dl, MVT::i32, InChain,
-                                    VAListPtr, MachinePointerInfo(SV), MVT::i8,
-                                    false, false, false, 0);
+                                    VAListPtr, MachinePointerInfo(SV), MVT::i8);
   InChain = GprIndex.getValue(1);
 
   if (VT == MVT::i64) {
@@ -2335,8 +2433,7 @@ SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
 
   // fpr
   SDValue FprIndex = DAG.getExtLoad(ISD::ZEXTLOAD, dl, MVT::i32, InChain,
-                                    FprPtr, MachinePointerInfo(SV), MVT::i8,
-                                    false, false, false, 0);
+                                    FprPtr, MachinePointerInfo(SV), MVT::i8);
   InChain = FprIndex.getValue(1);
 
   SDValue RegSaveAreaPtr = DAG.getNode(ISD::ADD, dl, PtrVT, VAListPtr,
@@ -2346,14 +2443,12 @@ SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
                                         DAG.getConstant(4, dl, MVT::i32));
 
   // areas
-  SDValue OverflowArea = DAG.getLoad(MVT::i32, dl, InChain, OverflowAreaPtr,
-                                     MachinePointerInfo(), false, false,
-                                     false, 0);
+  SDValue OverflowArea =
+      DAG.getLoad(MVT::i32, dl, InChain, OverflowAreaPtr, MachinePointerInfo());
   InChain = OverflowArea.getValue(1);
 
-  SDValue RegSaveArea = DAG.getLoad(MVT::i32, dl, InChain, RegSaveAreaPtr,
-                                    MachinePointerInfo(), false, false,
-                                    false, 0);
+  SDValue RegSaveArea =
+      DAG.getLoad(MVT::i32, dl, InChain, RegSaveAreaPtr, MachinePointerInfo());
   InChain = RegSaveArea.getValue(1);
 
   // select overflow_area if index > 8
@@ -2383,8 +2478,7 @@ SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
 
   InChain = DAG.getTruncStore(InChain, dl, IndexPlus1,
                               VT.isInteger() ? VAListPtr : FprPtr,
-                              MachinePointerInfo(SV),
-                              MVT::i8, false, false, 0);
+                              MachinePointerInfo(SV), MVT::i8);
 
   // determine if we should load from reg_save_area or overflow_area
   SDValue Result = DAG.getNode(ISD::SELECT, dl, PtrVT, CC, OurReg, OverflowArea);
@@ -2397,17 +2491,13 @@ SDValue PPCTargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG,
   OverflowArea = DAG.getNode(ISD::SELECT, dl, MVT::i32, CC, OverflowArea,
                              OverflowAreaPlusN);
 
-  InChain = DAG.getTruncStore(InChain, dl, OverflowArea,
-                              OverflowAreaPtr,
-                              MachinePointerInfo(),
-                              MVT::i32, false, false, 0);
+  InChain = DAG.getTruncStore(InChain, dl, OverflowArea, OverflowAreaPtr,
+                              MachinePointerInfo(), MVT::i32);
 
-  return DAG.getLoad(VT, dl, InChain, Result, MachinePointerInfo(),
-                     false, false, false, 0);
+  return DAG.getLoad(VT, dl, InChain, Result, MachinePointerInfo());
 }
 
-SDValue PPCTargetLowering::LowerVACOPY(SDValue Op, SelectionDAG &DAG,
-                                       const PPCSubtarget &Subtarget) const {
+SDValue PPCTargetLowering::LowerVACOPY(SDValue Op, SelectionDAG &DAG) const {
   assert(!Subtarget.isPPC64() && "LowerVACOPY is PPC32 only");
 
   // We have to copy the entire va_list struct:
@@ -2431,7 +2521,7 @@ SDValue PPCTargetLowering::LowerINIT_TRAMPOLINE(SDValue Op,
   SDValue Nest = Op.getOperand(3); // 'nest' parameter value
   SDLoc dl(Op);
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
   bool isPPC64 = (PtrVT == MVT::i64);
   Type *IntPtrTy = DAG.getDataLayout().getIntPtrType(*DAG.getContext());
 
@@ -2454,28 +2544,26 @@ SDValue PPCTargetLowering::LowerINIT_TRAMPOLINE(SDValue Op,
   CLI.setDebugLoc(dl).setChain(Chain)
     .setCallee(CallingConv::C, Type::getVoidTy(*DAG.getContext()),
                DAG.getExternalSymbol("__trampoline_setup", PtrVT),
-               std::move(Args), 0);
+               std::move(Args));
 
   std::pair<SDValue, SDValue> CallResult = LowerCallTo(CLI);
   return CallResult.second;
 }
 
-SDValue PPCTargetLowering::LowerVASTART(SDValue Op, SelectionDAG &DAG,
-                                        const PPCSubtarget &Subtarget) const {
+SDValue PPCTargetLowering::LowerVASTART(SDValue Op, SelectionDAG &DAG) const {
   MachineFunction &MF = DAG.getMachineFunction();
   PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
 
   SDLoc dl(Op);
 
   if (Subtarget.isDarwinABI() || Subtarget.isPPC64()) {
     // vastart just stores the address of the VarArgsFrameIndex slot into the
     // memory location argument.
-    EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
     SDValue FR = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
     const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
     return DAG.getStore(Op.getOperand(0), dl, FR, Op.getOperand(1),
-                        MachinePointerInfo(SV),
-                        false, false, 0);
+                        MachinePointerInfo(SV));
   }
 
   // For the 32-bit SVR4 ABI we follow the layout of the va_list struct.
@@ -2504,9 +2592,6 @@ SDValue PPCTargetLowering::LowerVASTART(SDValue Op, SelectionDAG &DAG,
 
   SDValue ArgGPR = DAG.getConstant(FuncInfo->getVarArgsNumGPR(), dl, MVT::i32);
   SDValue ArgFPR = DAG.getConstant(FuncInfo->getVarArgsNumFPR(), dl, MVT::i32);
-
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
-
   SDValue StackOffsetFI = DAG.getFrameIndex(FuncInfo->getVarArgsStackOffset(),
                                             PtrVT);
   SDValue FR = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(),
@@ -2524,35 +2609,29 @@ SDValue PPCTargetLowering::LowerVASTART(SDValue Op, SelectionDAG &DAG,
   const Value *SV = cast<SrcValueSDNode>(Op.getOperand(2))->getValue();
 
   // Store first byte : number of int regs
-  SDValue firstStore = DAG.getTruncStore(Op.getOperand(0), dl, ArgGPR,
-                                         Op.getOperand(1),
-                                         MachinePointerInfo(SV),
-                                         MVT::i8, false, false, 0);
+  SDValue firstStore =
+      DAG.getTruncStore(Op.getOperand(0), dl, ArgGPR, Op.getOperand(1),
+                        MachinePointerInfo(SV), MVT::i8);
   uint64_t nextOffset = FPROffset;
   SDValue nextPtr = DAG.getNode(ISD::ADD, dl, PtrVT, Op.getOperand(1),
                                   ConstFPROffset);
 
   // Store second byte : number of float regs
   SDValue secondStore =
-    DAG.getTruncStore(firstStore, dl, ArgFPR, nextPtr,
-                      MachinePointerInfo(SV, nextOffset), MVT::i8,
-                      false, false, 0);
+      DAG.getTruncStore(firstStore, dl, ArgFPR, nextPtr,
+                        MachinePointerInfo(SV, nextOffset), MVT::i8);
   nextOffset += StackOffset;
   nextPtr = DAG.getNode(ISD::ADD, dl, PtrVT, nextPtr, ConstStackOffset);
 
   // Store second word : arguments given on stack
-  SDValue thirdStore =
-    DAG.getStore(secondStore, dl, StackOffsetFI, nextPtr,
-                 MachinePointerInfo(SV, nextOffset),
-                 false, false, 0);
+  SDValue thirdStore = DAG.getStore(secondStore, dl, StackOffsetFI, nextPtr,
+                                    MachinePointerInfo(SV, nextOffset));
   nextOffset += FrameOffset;
   nextPtr = DAG.getNode(ISD::ADD, dl, PtrVT, nextPtr, ConstFrameOffset);
 
   // Store third word : arguments given in registers
   return DAG.getStore(thirdStore, dl, FR, nextPtr,
-                      MachinePointerInfo(SV, nextOffset),
-                      false, false, 0);
-
+                      MachinePointerInfo(SV, nextOffset));
 }
 
 #include "PPCGenCallingConv.inc"
@@ -2762,14 +2841,10 @@ static unsigned EnsureStackAlignment(const PPCFrameLowering *Lowering,
   return NumBytes;
 }
 
-SDValue
-PPCTargetLowering::LowerFormalArguments(SDValue Chain,
-                                        CallingConv::ID CallConv, bool isVarArg,
-                                        const SmallVectorImpl<ISD::InputArg>
-                                          &Ins,
-                                        SDLoc dl, SelectionDAG &DAG,
-                                        SmallVectorImpl<SDValue> &InVals)
-                                          const {
+SDValue PPCTargetLowering::LowerFormalArguments(
+    SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const {
   if (Subtarget.isSVR4ABI()) {
     if (Subtarget.isPPC64())
       return LowerFormalArguments_64SVR4(Chain, CallConv, isVarArg, Ins,
@@ -2783,14 +2858,10 @@ PPCTargetLowering::LowerFormalArguments(SDValue Chain,
   }
 }
 
-SDValue
-PPCTargetLowering::LowerFormalArguments_32SVR4(
-                                      SDValue Chain,
-                                      CallingConv::ID CallConv, bool isVarArg,
-                                      const SmallVectorImpl<ISD::InputArg>
-                                        &Ins,
-                                      SDLoc dl, SelectionDAG &DAG,
-                                      SmallVectorImpl<SDValue> &InVals) const {
+SDValue PPCTargetLowering::LowerFormalArguments_32SVR4(
+    SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const {
 
   // 32-bit SVR4 ABI Stack Frame Layout:
   //              +-----------------------------------+
@@ -2825,7 +2896,7 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
   MachineFrameInfo *MFI = MF.getFrameInfo();
   PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
   // Potential tail calls could cause overwriting of argument stack slots.
   bool isImmutable = !(getTargetMachine().Options.GuaranteedTailCallOpt &&
                        (CallConv == CallingConv::Fast));
@@ -2833,14 +2904,17 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
 
   // Assign locations to all of the incoming arguments.
   SmallVector<CCValAssign, 16> ArgLocs;
-  CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), ArgLocs,
+  PPCCCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), ArgLocs,
                  *DAG.getContext());
 
   // Reserve space for the linkage area on the stack.
   unsigned LinkageSize = Subtarget.getFrameLowering()->getLinkageSize();
   CCInfo.AllocateStack(LinkageSize, PtrByteSize);
+  if (useSoftFloat())
+    CCInfo.PreAnalyzeFormalArguments(Ins);
 
   CCInfo.AnalyzeFormalArguments(Ins, CC_PPC32_SVR4);
+  CCInfo.clearWasPPCF128();
 
   for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
     CCValAssign &VA = ArgLocs[i];
@@ -2908,9 +2982,8 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
 
       // Create load nodes to retrieve arguments from the stack.
       SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
-      InVals.push_back(DAG.getLoad(VA.getValVT(), dl, Chain, FIN,
-                                   MachinePointerInfo(),
-                                   false, false, false, 0));
+      InVals.push_back(
+          DAG.getLoad(VA.getValVT(), dl, Chain, FIN, MachinePointerInfo()));
     }
   }
 
@@ -2955,7 +3028,7 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
     };
     unsigned NumFPArgRegs = array_lengthof(FPArgRegs);
 
-    if (Subtarget.useSoftFloat())
+    if (useSoftFloat())
        NumFPArgRegs = 0;
 
     FuncInfo->setVarArgsNumGPR(CCInfo.getFirstUnallocated(GPArgRegs));
@@ -2973,8 +3046,8 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
     SDValue FIN = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
 
     // The fixed integer arguments of a variadic function are stored to the
-    // VarArgsFrameIndex on the stack so that they may be loaded by deferencing
-    // the result of va_next.
+    // VarArgsFrameIndex on the stack so that they may be loaded by
+    // dereferencing the result of va_next.
     for (unsigned GPRIndex = 0; GPRIndex != NumGPArgRegs; ++GPRIndex) {
       // Get an existing live-in vreg, or add a new one.
       unsigned VReg = MF.getRegInfo().getLiveInVirtReg(GPArgRegs[GPRIndex]);
@@ -2982,8 +3055,8 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
         VReg = MF.addLiveIn(GPArgRegs[GPRIndex], &PPC::GPRCRegClass);
 
       SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
-      SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
-                                   MachinePointerInfo(), false, false, 0);
+      SDValue Store =
+          DAG.getStore(Val.getValue(1), dl, Val, FIN, MachinePointerInfo());
       MemOps.push_back(Store);
       // Increment the address by four for the next argument to store
       SDValue PtrOff = DAG.getConstant(PtrVT.getSizeInBits()/8, dl, PtrVT);
@@ -3001,8 +3074,8 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
         VReg = MF.addLiveIn(FPArgRegs[FPRIndex], &PPC::F8RCRegClass);
 
       SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, MVT::f64);
-      SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
-                                   MachinePointerInfo(), false, false, 0);
+      SDValue Store =
+          DAG.getStore(Val.getValue(1), dl, Val, FIN, MachinePointerInfo());
       MemOps.push_back(Store);
       // Increment the address by eight for the next argument to store
       SDValue PtrOff = DAG.getConstant(MVT(MVT::f64).getSizeInBits()/8, dl,
@@ -3019,10 +3092,10 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
 
 // PPC64 passes i8, i16, and i32 values in i64 registers. Promote
 // value to MVT::i64 and then truncate to the correct register size.
-SDValue
-PPCTargetLowering::extendArgForPPC64(ISD::ArgFlagsTy Flags, EVT ObjectVT,
-                                     SelectionDAG &DAG, SDValue ArgVal,
-                                     SDLoc dl) const {
+SDValue PPCTargetLowering::extendArgForPPC64(ISD::ArgFlagsTy Flags,
+                                             EVT ObjectVT, SelectionDAG &DAG,
+                                             SDValue ArgVal,
+                                             const SDLoc &dl) const {
   if (Flags.isSExt())
     ArgVal = DAG.getNode(ISD::AssertSext, dl, MVT::i64, ArgVal,
                          DAG.getValueType(ObjectVT));
@@ -3033,14 +3106,10 @@ PPCTargetLowering::extendArgForPPC64(ISD::ArgFlagsTy Flags, EVT ObjectVT,
   return DAG.getNode(ISD::TRUNCATE, dl, ObjectVT, ArgVal);
 }
 
-SDValue
-PPCTargetLowering::LowerFormalArguments_64SVR4(
-                                      SDValue Chain,
-                                      CallingConv::ID CallConv, bool isVarArg,
-                                      const SmallVectorImpl<ISD::InputArg>
-                                        &Ins,
-                                      SDLoc dl, SelectionDAG &DAG,
-                                      SmallVectorImpl<SDValue> &InVals) const {
+SDValue PPCTargetLowering::LowerFormalArguments_64SVR4(
+    SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const {
   // TODO: add description of PPC stack frame format, or at least some docs.
   //
   bool isELFv2ABI = Subtarget.isELFv2ABI();
@@ -3052,7 +3121,7 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
   assert(!(CallConv == CallingConv::Fast && isVarArg) &&
          "fastcc not supported on varargs functions");
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
   // Potential tail calls could cause overwriting of argument stack slots.
   bool isImmutable = !(getTargetMachine().Options.GuaranteedTailCallOpt &&
                        (CallConv == CallingConv::Fast));
@@ -3199,15 +3268,13 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
             EVT ObjType = (ObjSize == 1 ? MVT::i8 :
                            (ObjSize == 2 ? MVT::i16 : MVT::i32));
             Store = DAG.getTruncStore(Val.getValue(1), dl, Val, Arg,
-                                      MachinePointerInfo(&*FuncArg), ObjType,
-                                      false, false, 0);
+                                      MachinePointerInfo(&*FuncArg), ObjType);
           } else {
             // For sizes that don't fit a truncating store (3, 5, 6, 7),
             // store the whole register as-is to the parameter save area
             // slot.
-            Store =
-                DAG.getStore(Val.getValue(1), dl, Val, FIN,
-                             MachinePointerInfo(&*FuncArg), false, false, 0);
+            Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
+                                 MachinePointerInfo(&*FuncArg));
           }
 
           MemOps.push_back(Store);
@@ -3234,9 +3301,8 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
           SDValue Off = DAG.getConstant(j, dl, PtrVT);
           Addr = DAG.getNode(ISD::ADD, dl, Off.getValueType(), Addr, Off);
         }
-        SDValue Store =
-            DAG.getStore(Val.getValue(1), dl, Val, Addr,
-                         MachinePointerInfo(&*FuncArg, j), false, false, 0);
+        SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, Addr,
+                                     MachinePointerInfo(&*FuncArg, j));
         MemOps.push_back(Store);
         ++GPR_idx;
       }
@@ -3402,8 +3468,7 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
         CurArgOffset += ArgSize - ObjSize;
       int FI = MFI->CreateFixedObject(ObjSize, CurArgOffset, isImmutable);
       SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
-      ArgVal = DAG.getLoad(ObjectVT, dl, Chain, FIN, MachinePointerInfo(),
-                           false, false, false, 0);
+      ArgVal = DAG.getLoad(ObjectVT, dl, Chain, FIN, MachinePointerInfo());
     }
 
     InVals.push_back(ArgVal);
@@ -3434,14 +3499,14 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
     SDValue FIN = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
 
     // If this function is vararg, store any remaining integer argument regs
-    // to their spots on the stack so that they may be loaded by deferencing the
-    // result of va_next.
+    // to their spots on the stack so that they may be loaded by dereferencing
+    // the result of va_next.
     for (GPR_idx = (ArgOffset - LinkageSize) / PtrByteSize;
          GPR_idx < Num_GPR_Regs; ++GPR_idx) {
       unsigned VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::G8RCRegClass);
       SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
-      SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
-                                   MachinePointerInfo(), false, false, 0);
+      SDValue Store =
+          DAG.getStore(Val.getValue(1), dl, Val, FIN, MachinePointerInfo());
       MemOps.push_back(Store);
       // Increment the address by four for the next argument to store
       SDValue PtrOff = DAG.getConstant(PtrByteSize, dl, PtrVT);
@@ -3455,21 +3520,17 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
   return Chain;
 }
 
-SDValue
-PPCTargetLowering::LowerFormalArguments_Darwin(
-                                      SDValue Chain,
-                                      CallingConv::ID CallConv, bool isVarArg,
-                                      const SmallVectorImpl<ISD::InputArg>
-                                        &Ins,
-                                      SDLoc dl, SelectionDAG &DAG,
-                                      SmallVectorImpl<SDValue> &InVals) const {
+SDValue PPCTargetLowering::LowerFormalArguments_Darwin(
+    SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const {
   // TODO: add description of PPC stack frame format, or at least some docs.
   //
   MachineFunction &MF = DAG.getMachineFunction();
   MachineFrameInfo *MFI = MF.getFrameInfo();
   PPCFunctionInfo *FuncInfo = MF.getInfo<PPCFunctionInfo>();
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
   bool isPPC64 = PtrVT == MVT::i64;
   // Potential tail calls could cause overwriting of argument stack slots.
   bool isImmutable = !(getTargetMachine().Options.GuaranteedTailCallOpt &&
@@ -3613,9 +3674,9 @@ PPCTargetLowering::LowerFormalArguments_Darwin(
             VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::GPRCRegClass);
           SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
           EVT ObjType = ObjSize == 1 ? MVT::i8 : MVT::i16;
-          SDValue Store = DAG.getTruncStore(Val.getValue(1), dl, Val, FIN,
-                                            MachinePointerInfo(&*FuncArg),
-                                            ObjType, false, false, 0);
+          SDValue Store =
+              DAG.getTruncStore(Val.getValue(1), dl, Val, FIN,
+                                MachinePointerInfo(&*FuncArg), ObjType);
           MemOps.push_back(Store);
           ++GPR_idx;
         }
@@ -3637,9 +3698,8 @@ PPCTargetLowering::LowerFormalArguments_Darwin(
           int FI = MFI->CreateFixedObject(PtrByteSize, ArgOffset, true);
           SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
           SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
-          SDValue Store =
-              DAG.getStore(Val.getValue(1), dl, Val, FIN,
-                           MachinePointerInfo(&*FuncArg, j), false, false, 0);
+          SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
+                                       MachinePointerInfo(&*FuncArg, j));
           MemOps.push_back(Store);
           ++GPR_idx;
           ArgOffset += PtrByteSize;
@@ -3760,8 +3820,7 @@ PPCTargetLowering::LowerFormalArguments_Darwin(
                                       CurArgOffset + (ArgSize - ObjSize),
                                       isImmutable);
       SDValue FIN = DAG.getFrameIndex(FI, PtrVT);
-      ArgVal = DAG.getLoad(ObjectVT, dl, Chain, FIN, MachinePointerInfo(),
-                           false, false, false, 0);
+      ArgVal = DAG.getLoad(ObjectVT, dl, Chain, FIN, MachinePointerInfo());
     }
 
     InVals.push_back(ArgVal);
@@ -3795,8 +3854,8 @@ PPCTargetLowering::LowerFormalArguments_Darwin(
     SDValue FIN = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), PtrVT);
 
     // If this function is vararg, store any remaining integer argument regs
-    // to their spots on the stack so that they may be loaded by deferencing the
-    // result of va_next.
+    // to their spots on the stack so that they may be loaded by dereferencing
+    // the result of va_next.
     for (; GPR_idx != Num_GPR_Regs; ++GPR_idx) {
       unsigned VReg;
 
@@ -3806,8 +3865,8 @@ PPCTargetLowering::LowerFormalArguments_Darwin(
         VReg = MF.addLiveIn(GPR[GPR_idx], &PPC::GPRCRegClass);
 
       SDValue Val = DAG.getCopyFromReg(Chain, dl, VReg, PtrVT);
-      SDValue Store = DAG.getStore(Val.getValue(1), dl, Val, FIN,
-                                   MachinePointerInfo(), false, false, 0);
+      SDValue Store =
+          DAG.getStore(Val.getValue(1), dl, Val, FIN, MachinePointerInfo());
       MemOps.push_back(Store);
       // Increment the address by four for the next argument to store
       SDValue PtrOff = DAG.getConstant(PtrVT.getSizeInBits()/8, dl, PtrVT);
@@ -3838,6 +3897,176 @@ static int CalculateTailCallSPDiff(SelectionDAG& DAG, bool isTailCall,
   return SPDiff;
 }
 
+static bool isFunctionGlobalAddress(SDValue Callee);
+
+static bool
+resideInSameModule(SDValue Callee, Reloc::Model RelMod) {
+  // If !G, Callee can be an external symbol.
+  GlobalAddressSDNode *G = dyn_cast<GlobalAddressSDNode>(Callee);
+  if (!G) return false;
+
+  const GlobalValue *GV = G->getGlobal();
+
+  if (GV->isDeclaration()) return false;
+
+  switch(GV->getLinkage()) {
+  default: llvm_unreachable("unknow linkage type");
+  case GlobalValue::AvailableExternallyLinkage:
+  case GlobalValue::ExternalWeakLinkage:
+    return false;
+
+  // Callee with weak linkage is allowed if it has hidden or protected
+  // visibility
+  case GlobalValue::LinkOnceAnyLinkage:
+  case GlobalValue::LinkOnceODRLinkage: // e.g. c++ inline functions
+  case GlobalValue::WeakAnyLinkage:
+  case GlobalValue::WeakODRLinkage:     // e.g. c++ template instantiation
+    if (GV->hasDefaultVisibility())
+      return false;
+
+  case GlobalValue::ExternalLinkage:
+  case GlobalValue::InternalLinkage:
+  case GlobalValue::PrivateLinkage:
+    break;
+  }
+
+  // With '-fPIC', calling default visiblity function need insert 'nop' after
+  // function call, no matter that function resides in same module or not, so
+  // we treat it as in different module.
+  if (RelMod == Reloc::PIC_ && GV->hasDefaultVisibility())
+    return false;
+
+  return true;
+}
+
+static bool
+needStackSlotPassParameters(const PPCSubtarget &Subtarget,
+                            const SmallVectorImpl<ISD::OutputArg> &Outs) {
+  assert(Subtarget.isSVR4ABI() && Subtarget.isPPC64());
+
+  const unsigned PtrByteSize = 8;
+  const unsigned LinkageSize = Subtarget.getFrameLowering()->getLinkageSize();
+
+  static const MCPhysReg GPR[] = {
+    PPC::X3, PPC::X4, PPC::X5, PPC::X6,
+    PPC::X7, PPC::X8, PPC::X9, PPC::X10,
+  };
+  static const MCPhysReg VR[] = {
+    PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
+    PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
+  };
+
+  const unsigned NumGPRs = array_lengthof(GPR);
+  const unsigned NumFPRs = 13;
+  const unsigned NumVRs = array_lengthof(VR);
+  const unsigned ParamAreaSize = NumGPRs * PtrByteSize;
+
+  unsigned NumBytes = LinkageSize;
+  unsigned AvailableFPRs = NumFPRs;
+  unsigned AvailableVRs = NumVRs;
+
+  for (const ISD::OutputArg& Param : Outs) {
+    if (Param.Flags.isNest()) continue;
+
+    if (CalculateStackSlotUsed(Param.VT, Param.ArgVT, Param.Flags,
+                               PtrByteSize, LinkageSize, ParamAreaSize,
+                               NumBytes, AvailableFPRs, AvailableVRs,
+                               Subtarget.hasQPX()))
+      return true;
+  }
+  return false;
+}
+
+static bool
+hasSameArgumentList(const Function *CallerFn, ImmutableCallSite *CS) {
+  if (CS->arg_size() != CallerFn->getArgumentList().size())
+    return false;
+
+  ImmutableCallSite::arg_iterator CalleeArgIter = CS->arg_begin();
+  ImmutableCallSite::arg_iterator CalleeArgEnd = CS->arg_end();
+  Function::const_arg_iterator CallerArgIter = CallerFn->arg_begin();
+
+  for (; CalleeArgIter != CalleeArgEnd; ++CalleeArgIter, ++CallerArgIter) {
+    const Value* CalleeArg = *CalleeArgIter;
+    const Value* CallerArg = &(*CallerArgIter);
+    if (CalleeArg == CallerArg)
+      continue;
+
+    // e.g. @caller([4 x i64] %a, [4 x i64] %b) {
+    //        tail call @callee([4 x i64] undef, [4 x i64] %b)
+    //      }
+    // 1st argument of callee is undef and has the same type as caller.
+    if (CalleeArg->getType() == CallerArg->getType() &&
+        isa<UndefValue>(CalleeArg))
+      continue;
+
+    return false;
+  }
+
+  return true;
+}
+
+bool
+PPCTargetLowering::IsEligibleForTailCallOptimization_64SVR4(
+                                    SDValue Callee,
+                                    CallingConv::ID CalleeCC,
+                                    ImmutableCallSite *CS,
+                                    bool isVarArg,
+                                    const SmallVectorImpl<ISD::OutputArg> &Outs,
+                                    const SmallVectorImpl<ISD::InputArg> &Ins,
+                                    SelectionDAG& DAG) const {
+  bool TailCallOpt = getTargetMachine().Options.GuaranteedTailCallOpt;
+
+  if (DisableSCO && !TailCallOpt) return false;
+
+  // Variadic argument functions are not supported.
+  if (isVarArg) return false;
+
+  MachineFunction &MF = DAG.getMachineFunction();
+  CallingConv::ID CallerCC = MF.getFunction()->getCallingConv();
+
+  // Tail or Sibling call optimization (TCO/SCO) needs callee and caller has
+  // the same calling convention
+  if (CallerCC != CalleeCC) return false;
+
+  // SCO support C calling convention
+  if (CalleeCC != CallingConv::Fast && CalleeCC != CallingConv::C)
+    return false;
+
+  // Functions containing by val parameters are not supported.
+  if (std::any_of(Ins.begin(), Ins.end(),
+                  [](const ISD::InputArg& IA) { return IA.Flags.isByVal(); }))
+    return false;
+
+  // No TCO/SCO on indirect call because Caller have to restore its TOC
+  if (!isFunctionGlobalAddress(Callee) &&
+      !isa<ExternalSymbolSDNode>(Callee))
+    return false;
+
+  // Check if Callee resides in the same module, because for now, PPC64 SVR4 ABI
+  // (ELFv1/ELFv2) doesn't allow tail calls to a symbol resides in another
+  // module.
+  // ref: https://bugzilla.mozilla.org/show_bug.cgi?id=973977
+  if (!resideInSameModule(Callee, getTargetMachine().getRelocationModel()))
+    return false;
+
+  // TCO allows altering callee ABI, so we don't have to check further.
+  if (CalleeCC == CallingConv::Fast && TailCallOpt)
+    return true;
+
+  if (DisableSCO) return false;
+
+  // If callee use the same argument list that caller is using, then we can
+  // apply SCO on this case. If it is not, then we need to check if callee needs
+  // stack for passing arguments.
+  if (!hasSameArgumentList(MF.getFunction(), CS) &&
+      needStackSlotPassParameters(Subtarget, Outs)) {
+    return false;
+  }
+
+  return true;
+}
+
 /// IsEligibleForTailCallOptimization - Check whether the call is eligible
 /// for tail call optimization. Targets which want to do tail call
 /// optimization should implement this function.
@@ -3888,9 +4117,11 @@ static SDNode *isBLACompatibleAddress(SDValue Op, SelectionDAG &DAG) {
       SignExtend32<26>(Addr) != Addr)
     return nullptr;  // Top 6 bits have to be sext of immediate.
 
-  return DAG.getConstant((int)C->getZExtValue() >> 2, SDLoc(Op),
-                         DAG.getTargetLoweringInfo().getPointerTy(
-                             DAG.getDataLayout())).getNode();
+  return DAG
+      .getConstant(
+          (int)C->getZExtValue() >> 2, SDLoc(Op),
+          DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout()))
+      .getNode();
 }
 
 namespace {
@@ -3905,12 +4136,10 @@ struct TailCallArgumentInfo {
 }
 
 /// StoreTailCallArgumentsToStackSlot - Stores arguments to their stack slot.
-static void
-StoreTailCallArgumentsToStackSlot(SelectionDAG &DAG,
-                                           SDValue Chain,
-                   const SmallVectorImpl<TailCallArgumentInfo> &TailCallArgs,
-                   SmallVectorImpl<SDValue> &MemOpChains,
-                   SDLoc dl) {
+static void StoreTailCallArgumentsToStackSlot(
+    SelectionDAG &DAG, SDValue Chain,
+    const SmallVectorImpl<TailCallArgumentInfo> &TailCallArgs,
+    SmallVectorImpl<SDValue> &MemOpChains, const SDLoc &dl) {
   for (unsigned i = 0, e = TailCallArgs.size(); i != e; ++i) {
     SDValue Arg = TailCallArgs[i].Arg;
     SDValue FIN = TailCallArgs[i].FrameIdxOp;
@@ -3918,48 +4147,40 @@ StoreTailCallArgumentsToStackSlot(SelectionDAG &DAG,
     // Store relative to framepointer.
     MemOpChains.push_back(DAG.getStore(
         Chain, dl, Arg, FIN,
-        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FI), false,
-        false, 0));
+        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FI)));
   }
 }
 
 /// EmitTailCallStoreFPAndRetAddr - Move the frame pointer and return address to
 /// the appropriate stack slot for the tail call optimized function call.
-static SDValue EmitTailCallStoreFPAndRetAddr(SelectionDAG &DAG,
-                                               MachineFunction &MF,
-                                               SDValue Chain,
-                                               SDValue OldRetAddr,
-                                               SDValue OldFP,
-                                               int SPDiff,
-                                               bool isPPC64,
-                                               bool isDarwinABI,
-                                               SDLoc dl) {
+static SDValue EmitTailCallStoreFPAndRetAddr(SelectionDAG &DAG, SDValue Chain,
+                                             SDValue OldRetAddr, SDValue OldFP,
+                                             int SPDiff, const SDLoc &dl) {
   if (SPDiff) {
     // Calculate the new stack slot for the return address.
+    MachineFunction &MF = DAG.getMachineFunction();
+    const PPCSubtarget &Subtarget = MF.getSubtarget<PPCSubtarget>();
+    const PPCFrameLowering *FL = Subtarget.getFrameLowering();
+    bool isPPC64 = Subtarget.isPPC64();
     int SlotSize = isPPC64 ? 8 : 4;
-    const PPCFrameLowering *FL =
-        MF.getSubtarget<PPCSubtarget>().getFrameLowering();
     int NewRetAddrLoc = SPDiff + FL->getReturnSaveOffset();
     int NewRetAddr = MF.getFrameInfo()->CreateFixedObject(SlotSize,
                                                           NewRetAddrLoc, true);
     EVT VT = isPPC64 ? MVT::i64 : MVT::i32;
     SDValue NewRetAddrFrIdx = DAG.getFrameIndex(NewRetAddr, VT);
-    Chain = DAG.getStore(
-        Chain, dl, OldRetAddr, NewRetAddrFrIdx,
-        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), NewRetAddr),
-        false, false, 0);
+    Chain = DAG.getStore(Chain, dl, OldRetAddr, NewRetAddrFrIdx,
+                         MachinePointerInfo::getFixedStack(MF, NewRetAddr));
 
     // When using the 32/64-bit SVR4 ABI there is no need to move the FP stack
     // slot as the FP is never overwritten.
-    if (isDarwinABI) {
+    if (Subtarget.isDarwinABI()) {
       int NewFPLoc = SPDiff + FL->getFramePointerSaveOffset();
       int NewFPIdx = MF.getFrameInfo()->CreateFixedObject(SlotSize, NewFPLoc,
                                                           true);
       SDValue NewFramePtrIdx = DAG.getFrameIndex(NewFPIdx, VT);
-      Chain = DAG.getStore(
-          Chain, dl, OldFP, NewFramePtrIdx,
-          MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), NewFPIdx),
-          false, false, 0);
+      Chain = DAG.getStore(Chain, dl, OldFP, NewFramePtrIdx,
+                           MachinePointerInfo::getFixedStack(
+                               DAG.getMachineFunction(), NewFPIdx));
     }
   }
   return Chain;
@@ -3986,27 +4207,21 @@ CalculateTailCallArgDest(SelectionDAG &DAG, MachineFunction &MF, bool isPPC64,
 /// EmitTCFPAndRetAddrLoad - Emit load from frame pointer and return address
 /// stack slot. Returns the chain as result and the loaded frame pointers in
 /// LROpOut/FPOpout. Used when tail calling.
-SDValue PPCTargetLowering::EmitTailCallLoadFPAndRetAddr(SelectionDAG & DAG,
-                                                        int SPDiff,
-                                                        SDValue Chain,
-                                                        SDValue &LROpOut,
-                                                        SDValue &FPOpOut,
-                                                        bool isDarwinABI,
-                                                        SDLoc dl) const {
+SDValue PPCTargetLowering::EmitTailCallLoadFPAndRetAddr(
+    SelectionDAG &DAG, int SPDiff, SDValue Chain, SDValue &LROpOut,
+    SDValue &FPOpOut, const SDLoc &dl) const {
   if (SPDiff) {
     // Load the LR and FP stack slot for later adjusting.
     EVT VT = Subtarget.isPPC64() ? MVT::i64 : MVT::i32;
     LROpOut = getReturnAddrFrameIndex(DAG);
-    LROpOut = DAG.getLoad(VT, dl, Chain, LROpOut, MachinePointerInfo(),
-                          false, false, false, 0);
+    LROpOut = DAG.getLoad(VT, dl, Chain, LROpOut, MachinePointerInfo());
     Chain = SDValue(LROpOut.getNode(), 1);
 
     // When using the 32/64-bit SVR4 ABI there is no need to load the FP stack
     // slot as the FP is never overwritten.
-    if (isDarwinABI) {
+    if (Subtarget.isDarwinABI()) {
       FPOpOut = getFramePointerFrameIndex(DAG);
-      FPOpOut = DAG.getLoad(VT, dl, Chain, FPOpOut, MachinePointerInfo(),
-                            false, false, false, 0);
+      FPOpOut = DAG.getLoad(VT, dl, Chain, FPOpOut, MachinePointerInfo());
       Chain = SDValue(FPOpOut.getNode(), 1);
     }
   }
@@ -4019,10 +4234,9 @@ SDValue PPCTargetLowering::EmitTailCallLoadFPAndRetAddr(SelectionDAG & DAG,
 /// a byval function parameter.
 /// Sometimes what we are copying is the end of a larger object, the part that
 /// does not fit in registers.
-static SDValue
-CreateCopyOfByValArgument(SDValue Src, SDValue Dst, SDValue Chain,
-                          ISD::ArgFlagsTy Flags, SelectionDAG &DAG,
-                          SDLoc dl) {
+static SDValue CreateCopyOfByValArgument(SDValue Src, SDValue Dst,
+                                         SDValue Chain, ISD::ArgFlagsTy Flags,
+                                         SelectionDAG &DAG, const SDLoc &dl) {
   SDValue SizeNode = DAG.getConstant(Flags.getByValSize(), dl, MVT::i32);
   return DAG.getMemcpy(Chain, dl, Dst, Src, SizeNode, Flags.getByValAlign(),
                        false, false, false, MachinePointerInfo(),
@@ -4031,13 +4245,11 @@ CreateCopyOfByValArgument(SDValue Src, SDValue Dst, SDValue Chain,
 
 /// LowerMemOpCallTo - Store the argument to the stack or remember it in case of
 /// tail calls.
-static void
-LowerMemOpCallTo(SelectionDAG &DAG, MachineFunction &MF, SDValue Chain,
-                 SDValue Arg, SDValue PtrOff, int SPDiff,
-                 unsigned ArgOffset, bool isPPC64, bool isTailCall,
-                 bool isVector, SmallVectorImpl<SDValue> &MemOpChains,
-                 SmallVectorImpl<TailCallArgumentInfo> &TailCallArguments,
-                 SDLoc dl) {
+static void LowerMemOpCallTo(
+    SelectionDAG &DAG, MachineFunction &MF, SDValue Chain, SDValue Arg,
+    SDValue PtrOff, int SPDiff, unsigned ArgOffset, bool isPPC64,
+    bool isTailCall, bool isVector, SmallVectorImpl<SDValue> &MemOpChains,
+    SmallVectorImpl<TailCallArgumentInfo> &TailCallArguments, const SDLoc &dl) {
   EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
   if (!isTailCall) {
     if (isVector) {
@@ -4049,20 +4261,18 @@ LowerMemOpCallTo(SelectionDAG &DAG, MachineFunction &MF, SDValue Chain,
       PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr,
                            DAG.getConstant(ArgOffset, dl, PtrVT));
     }
-    MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, PtrOff,
-                                       MachinePointerInfo(), false, false, 0));
-  // Calculate and remember argument location.
+    MemOpChains.push_back(
+        DAG.getStore(Chain, dl, Arg, PtrOff, MachinePointerInfo()));
+    // Calculate and remember argument location.
   } else CalculateTailCallArgDest(DAG, MF, isPPC64, Arg, SPDiff, ArgOffset,
                                   TailCallArguments);
 }
 
-static
-void PrepareTailCall(SelectionDAG &DAG, SDValue &InFlag, SDValue &Chain,
-                     SDLoc dl, bool isPPC64, int SPDiff, unsigned NumBytes,
-                     SDValue LROp, SDValue FPOp, bool isDarwinABI,
-                     SmallVectorImpl<TailCallArgumentInfo> &TailCallArguments) {
-  MachineFunction &MF = DAG.getMachineFunction();
-
+static void
+PrepareTailCall(SelectionDAG &DAG, SDValue &InFlag, SDValue &Chain,
+                const SDLoc &dl, int SPDiff, unsigned NumBytes, SDValue LROp,
+                SDValue FPOp,
+                SmallVectorImpl<TailCallArgumentInfo> &TailCallArguments) {
   // Emit a sequence of copyto/copyfrom virtual registers for arguments that
   // might overwrite each other in case of tail call optimization.
   SmallVector<SDValue, 8> MemOpChains2;
@@ -4074,8 +4284,7 @@ void PrepareTailCall(SelectionDAG &DAG, SDValue &InFlag, SDValue &Chain,
     Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, MemOpChains2);
 
   // Store the return address to the appropriate stack slot.
-  Chain = EmitTailCallStoreFPAndRetAddr(DAG, MF, Chain, LROp, FPOp, SPDiff,
-                                        isPPC64, isDarwinABI, dl);
+  Chain = EmitTailCallStoreFPAndRetAddr(DAG, Chain, LROp, FPOp, SPDiff, dl);
 
   // Emit callseq_end just before tailcall node.
   Chain = DAG.getCALLSEQ_END(Chain, DAG.getIntPtrConstant(NumBytes, dl, true),
@@ -4091,19 +4300,19 @@ static bool isFunctionGlobalAddress(SDValue Callee) {
         Callee.getOpcode() == ISD::TargetGlobalTLSAddress)
       return false;
 
-    return G->getGlobal()->getType()->getElementType()->isFunctionTy();
+    return G->getGlobal()->getValueType()->isFunctionTy();
   }
 
   return false;
 }
 
-static
-unsigned PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag,
-                     SDValue &Chain, SDValue CallSeqStart, SDLoc dl, int SPDiff,
-                     bool isTailCall, bool IsPatchPoint, bool hasNest,
-                     SmallVectorImpl<std::pair<unsigned, SDValue> > &RegsToPass,
-                     SmallVectorImpl<SDValue> &Ops, std::vector<EVT> &NodeTys,
-                     ImmutableCallSite *CS, const PPCSubtarget &Subtarget) {
+static unsigned
+PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag, SDValue &Chain,
+            SDValue CallSeqStart, const SDLoc &dl, int SPDiff, bool isTailCall,
+            bool isPatchPoint, bool hasNest,
+            SmallVectorImpl<std::pair<unsigned, SDValue>> &RegsToPass,
+            SmallVectorImpl<SDValue> &Ops, std::vector<EVT> &NodeTys,
+            ImmutableCallSite *CS, const PPCSubtarget &Subtarget) {
 
   bool isPPC64 = Subtarget.isPPC64();
   bool isSVR4ABI = Subtarget.isSVR4ABI();
@@ -4123,23 +4332,24 @@ unsigned PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag,
       needIndirectCall = false;
     }
 
+  // PC-relative references to external symbols should go through $stub, unless
+  // we're building with the leopard linker or later, which automatically
+  // synthesizes these stubs.
+  const TargetMachine &TM = DAG.getTarget();
+  const Module *Mod = DAG.getMachineFunction().getFunction()->getParent();
+  const GlobalValue *GV = nullptr;
+  if (auto *G = dyn_cast<GlobalAddressSDNode>(Callee))
+    GV = G->getGlobal();
+  bool Local = TM.shouldAssumeDSOLocal(*Mod, GV);
+  bool UsePlt = !Local && Subtarget.isTargetELF() && !isPPC64;
+
   if (isFunctionGlobalAddress(Callee)) {
     GlobalAddressSDNode *G = cast<GlobalAddressSDNode>(Callee);
     // A call to a TLS address is actually an indirect call to a
     // thread-specific pointer.
     unsigned OpFlags = 0;
-    if ((DAG.getTarget().getRelocationModel() != Reloc::Static &&
-         (Subtarget.getTargetTriple().isMacOSX() &&
-          Subtarget.getTargetTriple().isMacOSXVersionLT(10, 5)) &&
-         !G->getGlobal()->isStrongDefinitionForLinker()) ||
-        (Subtarget.isTargetELF() && !isPPC64 &&
-         !G->getGlobal()->hasLocalLinkage() &&
-         DAG.getTarget().getRelocationModel() == Reloc::PIC_)) {
-      // PC-relative references to external symbols should go through $stub,
-      // unless we're building with the leopard linker or later, which
-      // automatically synthesizes these stubs.
-      OpFlags = PPCII::MO_PLT_OR_STUB;
-    }
+    if (UsePlt)
+      OpFlags = PPCII::MO_PLT;
 
     // If the callee is a GlobalAddress/ExternalSymbol node (quite common,
     // every direct call is) turn it into a TargetGlobalAddress /
@@ -4152,23 +4362,15 @@ unsigned PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag,
   if (ExternalSymbolSDNode *S = dyn_cast<ExternalSymbolSDNode>(Callee)) {
     unsigned char OpFlags = 0;
 
-    if ((DAG.getTarget().getRelocationModel() != Reloc::Static &&
-         (Subtarget.getTargetTriple().isMacOSX() &&
-          Subtarget.getTargetTriple().isMacOSXVersionLT(10, 5))) ||
-        (Subtarget.isTargetELF() && !isPPC64 &&
-         DAG.getTarget().getRelocationModel() == Reloc::PIC_)) {
-      // PC-relative references to external symbols should go through $stub,
-      // unless we're building with the leopard linker or later, which
-      // automatically synthesizes these stubs.
-      OpFlags = PPCII::MO_PLT_OR_STUB;
-    }
+    if (UsePlt)
+      OpFlags = PPCII::MO_PLT;
 
     Callee = DAG.getTargetExternalSymbol(S->getSymbol(), Callee.getValueType(),
                                          OpFlags);
     needIndirectCall = false;
   }
 
-  if (IsPatchPoint) {
+  if (isPatchPoint) {
     // We'll form an invalid direct call when lowering a patchpoint; the full
     // sequence for an indirect call is complicated, and many of the
     // instructions introduced might have side effects (and, thus, can't be
@@ -4217,24 +4419,26 @@ unsigned PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag,
       if (LDChain.getValueType() == MVT::Glue)
         LDChain = CallSeqStart.getValue(CallSeqStart->getNumValues()-2);
 
-      bool LoadsInv = Subtarget.hasInvariantFunctionDescriptors();
+      auto MMOFlags = Subtarget.hasInvariantFunctionDescriptors()
+                          ? MachineMemOperand::MOInvariant
+                          : MachineMemOperand::MONone;
 
       MachinePointerInfo MPI(CS ? CS->getCalledValue() : nullptr);
       SDValue LoadFuncPtr = DAG.getLoad(MVT::i64, dl, LDChain, Callee, MPI,
-                                        false, false, LoadsInv, 8);
+                                        /* Alignment = */ 8, MMOFlags);
 
       // Load environment pointer into r11.
       SDValue PtrOff = DAG.getIntPtrConstant(16, dl);
       SDValue AddPtr = DAG.getNode(ISD::ADD, dl, MVT::i64, Callee, PtrOff);
-      SDValue LoadEnvPtr = DAG.getLoad(MVT::i64, dl, LDChain, AddPtr,
-                                       MPI.getWithOffset(16), false, false,
-                                       LoadsInv, 8);
+      SDValue LoadEnvPtr =
+          DAG.getLoad(MVT::i64, dl, LDChain, AddPtr, MPI.getWithOffset(16),
+                      /* Alignment = */ 8, MMOFlags);
 
       SDValue TOCOff = DAG.getIntPtrConstant(8, dl);
       SDValue AddTOC = DAG.getNode(ISD::ADD, dl, MVT::i64, Callee, TOCOff);
-      SDValue TOCPtr = DAG.getLoad(MVT::i64, dl, LDChain, AddTOC,
-                                   MPI.getWithOffset(8), false, false,
-                                   LoadsInv, 8);
+      SDValue TOCPtr =
+          DAG.getLoad(MVT::i64, dl, LDChain, AddTOC, MPI.getWithOffset(8),
+                      /* Alignment = */ 8, MMOFlags);
 
       setUsesTOCBasePtr(DAG);
       SDValue TOCVal = DAG.getCopyToReg(Chain, dl, PPC::X2, TOCPtr,
@@ -4292,7 +4496,7 @@ unsigned PrepareCall(SelectionDAG &DAG, SDValue &Callee, SDValue &InFlag,
 
   // All calls, in both the ELF V1 and V2 ABIs, need the TOC register live
   // into the call.
-  if (isSVR4ABI && isPPC64 && !IsPatchPoint) {
+  if (isSVR4ABI && isPPC64 && !isPatchPoint) {
     setUsesTOCBasePtr(DAG);
     Ops.push_back(DAG.getRegister(PPC::X2, PtrVT));
   }
@@ -4308,12 +4512,10 @@ bool isLocalCall(const SDValue &Callee)
   return false;
 }
 
-SDValue
-PPCTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag,
-                                   CallingConv::ID CallConv, bool isVarArg,
-                                   const SmallVectorImpl<ISD::InputArg> &Ins,
-                                   SDLoc dl, SelectionDAG &DAG,
-                                   SmallVectorImpl<SDValue> &InVals) const {
+SDValue PPCTargetLowering::LowerCallResult(
+    SDValue Chain, SDValue InFlag, CallingConv::ID CallConv, bool isVarArg,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const {
 
   SmallVector<CCValAssign, 16> RVLocs;
   CCState CCRetInfo(CallConv, isVarArg, DAG.getMachineFunction(), RVLocs,
@@ -4354,23 +4556,18 @@ PPCTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag,
   return Chain;
 }
 
-SDValue
-PPCTargetLowering::FinishCall(CallingConv::ID CallConv, SDLoc dl,
-                              bool isTailCall, bool isVarArg, bool IsPatchPoint,
-                              bool hasNest, SelectionDAG &DAG,
-                              SmallVector<std::pair<unsigned, SDValue>, 8>
-                                &RegsToPass,
-                              SDValue InFlag, SDValue Chain,
-                              SDValue CallSeqStart, SDValue &Callee,
-                              int SPDiff, unsigned NumBytes,
-                              const SmallVectorImpl<ISD::InputArg> &Ins,
-                              SmallVectorImpl<SDValue> &InVals,
-                              ImmutableCallSite *CS) const {
+SDValue PPCTargetLowering::FinishCall(
+    CallingConv::ID CallConv, const SDLoc &dl, bool isTailCall, bool isVarArg,
+    bool isPatchPoint, bool hasNest, SelectionDAG &DAG,
+    SmallVector<std::pair<unsigned, SDValue>, 8> &RegsToPass, SDValue InFlag,
+    SDValue Chain, SDValue CallSeqStart, SDValue &Callee, int SPDiff,
+    unsigned NumBytes, const SmallVectorImpl<ISD::InputArg> &Ins,
+    SmallVectorImpl<SDValue> &InVals, ImmutableCallSite *CS) const {
 
   std::vector<EVT> NodeTys;
   SmallVector<SDValue, 8> Ops;
   unsigned CallOpc = PrepareCall(DAG, Callee, InFlag, Chain, CallSeqStart, dl,
-                                 SPDiff, isTailCall, IsPatchPoint, hasNest,
+                                 SPDiff, isTailCall, isPatchPoint, hasNest,
                                  RegsToPass, Ops, NodeTys, CS, Subtarget);
 
   // Add implicit use of CR bit 6 for 32-bit SVR4 vararg calls
@@ -4417,7 +4614,7 @@ PPCTargetLowering::FinishCall(CallingConv::ID CallConv, SDLoc dl,
   // same TOC), the NOP will remain unchanged.
 
   if (!isTailCall && Subtarget.isSVR4ABI()&& Subtarget.isPPC64() &&
-      !IsPatchPoint) {
+      !isPatchPoint) {
     if (CallOpc == PPCISD::BCTRL) {
       // This is a call through a function pointer.
       // Restore the caller TOC from the save area into R2.
@@ -4430,7 +4627,7 @@ PPCTargetLowering::FinishCall(CallingConv::ID CallConv, SDLoc dl,
       // allocated and an unnecessary move instruction being generated.
       CallOpc = PPCISD::BCTRL_LOAD_TOC;
 
-      EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+      EVT PtrVT = getPointerTy(DAG.getDataLayout());
       SDValue StackPtr = DAG.getRegister(PPC::X1, PtrVT);
       unsigned TOCSaveOffset = Subtarget.getFrameLowering()->getTOCSaveOffset();
       SDValue TOCOff = DAG.getIntPtrConstant(TOCSaveOffset, dl);
@@ -4472,12 +4669,35 @@ PPCTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   bool &isTailCall                      = CLI.IsTailCall;
   CallingConv::ID CallConv              = CLI.CallConv;
   bool isVarArg                         = CLI.IsVarArg;
-  bool IsPatchPoint                     = CLI.IsPatchPoint;
+  bool isPatchPoint                     = CLI.IsPatchPoint;
   ImmutableCallSite *CS                 = CLI.CS;
 
-  if (isTailCall)
-    isTailCall = IsEligibleForTailCallOptimization(Callee, CallConv, isVarArg,
-                                                   Ins, DAG);
+  if (isTailCall) {
+    if (Subtarget.isSVR4ABI() && Subtarget.isPPC64())
+      isTailCall =
+        IsEligibleForTailCallOptimization_64SVR4(Callee, CallConv, CS,
+                                                 isVarArg, Outs, Ins, DAG);
+    else
+      isTailCall = IsEligibleForTailCallOptimization(Callee, CallConv, isVarArg,
+                                                     Ins, DAG);
+    if (isTailCall) {
+      ++NumTailCalls;
+      if (!getTargetMachine().Options.GuaranteedTailCallOpt)
+        ++NumSiblingCalls;
+
+      assert(isa<GlobalAddressSDNode>(Callee) &&
+             "Callee should be an llvm::Function object.");
+      DEBUG(
+        const GlobalValue *GV = cast<GlobalAddressSDNode>(Callee)->getGlobal();
+        const unsigned Width = 80 - strlen("TCO caller: ")
+                                  - strlen(", callee linkage: 0, 0");
+        dbgs() << "TCO caller: "
+               << left_justify(DAG.getMachineFunction().getName(), Width)
+               << ", callee linkage: "
+               << GV->getVisibility() << ", " << GV->getLinkage() << "\n"
+      );
+    }
+  }
 
   if (!isTailCall && CS && CS->isMustTailCall())
     report_fatal_error("failed to perform tail call elimination on a call "
@@ -4486,29 +4706,27 @@ PPCTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   if (Subtarget.isSVR4ABI()) {
     if (Subtarget.isPPC64())
       return LowerCall_64SVR4(Chain, Callee, CallConv, isVarArg,
-                              isTailCall, IsPatchPoint, Outs, OutVals, Ins,
+                              isTailCall, isPatchPoint, Outs, OutVals, Ins,
                               dl, DAG, InVals, CS);
     else
       return LowerCall_32SVR4(Chain, Callee, CallConv, isVarArg,
-                              isTailCall, IsPatchPoint, Outs, OutVals, Ins,
+                              isTailCall, isPatchPoint, Outs, OutVals, Ins,
                               dl, DAG, InVals, CS);
   }
 
   return LowerCall_Darwin(Chain, Callee, CallConv, isVarArg,
-                          isTailCall, IsPatchPoint, Outs, OutVals, Ins,
+                          isTailCall, isPatchPoint, Outs, OutVals, Ins,
                           dl, DAG, InVals, CS);
 }
 
-SDValue
-PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
-                                    CallingConv::ID CallConv, bool isVarArg,
-                                    bool isTailCall, bool IsPatchPoint,
-                                    const SmallVectorImpl<ISD::OutputArg> &Outs,
-                                    const SmallVectorImpl<SDValue> &OutVals,
-                                    const SmallVectorImpl<ISD::InputArg> &Ins,
-                                    SDLoc dl, SelectionDAG &DAG,
-                                    SmallVectorImpl<SDValue> &InVals,
-                                    ImmutableCallSite *CS) const {
+SDValue PPCTargetLowering::LowerCall_32SVR4(
+    SDValue Chain, SDValue Callee, CallingConv::ID CallConv, bool isVarArg,
+    bool isTailCall, bool isPatchPoint,
+    const SmallVectorImpl<ISD::OutputArg> &Outs,
+    const SmallVectorImpl<SDValue> &OutVals,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals,
+    ImmutableCallSite *CS) const {
   // See PPCTargetLowering::LowerFormalArguments_32SVR4() for a description
   // of the 32-bit SVR4 ABI stack frame layout.
 
@@ -4534,12 +4752,13 @@ PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
 
   // Assign locations to all of the outgoing arguments.
   SmallVector<CCValAssign, 16> ArgLocs;
-  CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), ArgLocs,
-                 *DAG.getContext());
+  PPCCCState CCInfo(CallConv, isVarArg, MF, ArgLocs, *DAG.getContext());
 
   // Reserve space for the linkage area on the stack.
   CCInfo.AllocateStack(Subtarget.getFrameLowering()->getLinkageSize(),
                        PtrByteSize);
+  if (useSoftFloat())
+    CCInfo.PreAnalyzeCallOperands(Outs);
 
   if (isVarArg) {
     // Handle fixed and variable vector arguments differently.
@@ -4572,11 +4791,11 @@ PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
     // All arguments are treated the same.
     CCInfo.AnalyzeCallOperands(Outs, CC_PPC32_SVR4);
   }
+  CCInfo.clearWasPPCF128();
 
   // Assign locations to all of the outgoing aggregate by value arguments.
   SmallVector<CCValAssign, 16> ByValArgLocs;
-  CCState CCByValInfo(CallConv, isVarArg, DAG.getMachineFunction(),
-                      ByValArgLocs, *DAG.getContext());
+  CCState CCByValInfo(CallConv, isVarArg, MF, ByValArgLocs, *DAG.getContext());
 
   // Reserve stack space for the allocations in CCInfo.
   CCByValInfo.AllocateStack(CCInfo.getNextStackOffset(), PtrByteSize);
@@ -4601,8 +4820,7 @@ PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
   // Load the return address and frame pointer so it can be moved somewhere else
   // later.
   SDValue LROp, FPOp;
-  Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, false,
-                                       dl);
+  Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, dl);
 
   // Set up a copy of the stack pointer for use loading and storing any
   // arguments that may not fit in the registers available for argument
@@ -4676,9 +4894,8 @@ PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
         PtrOff = DAG.getNode(ISD::ADD, dl, getPointerTy(MF.getDataLayout()),
                              StackPtr, PtrOff);
 
-        MemOpChains.push_back(DAG.getStore(Chain, dl, Arg, PtrOff,
-                                           MachinePointerInfo(),
-                                           false, false, 0));
+        MemOpChains.push_back(
+            DAG.getStore(Chain, dl, Arg, PtrOff, MachinePointerInfo()));
       } else {
         // Calculate and remember argument location.
         CalculateTailCallArgDest(DAG, MF, false, Arg, SPDiff, LocMemOffset,
@@ -4712,10 +4929,10 @@ PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
   }
 
   if (isTailCall)
-    PrepareTailCall(DAG, InFlag, Chain, dl, false, SPDiff, NumBytes, LROp, FPOp,
-                    false, TailCallArguments);
+    PrepareTailCall(DAG, InFlag, Chain, dl, SPDiff, NumBytes, LROp, FPOp,
+                    TailCallArguments);
 
-  return FinishCall(CallConv, dl, isTailCall, isVarArg, IsPatchPoint,
+  return FinishCall(CallConv, dl, isTailCall, isVarArg, isPatchPoint,
                     /* unused except on PPC64 ELFv1 */ false, DAG,
                     RegsToPass, InFlag, Chain, CallSeqStart, Callee, SPDiff,
                     NumBytes, Ins, InVals, CS);
@@ -4723,12 +4940,9 @@ PPCTargetLowering::LowerCall_32SVR4(SDValue Chain, SDValue Callee,
 
 // Copy an argument into memory, being careful to do this outside the
 // call sequence for the call to which the argument belongs.
-SDValue
-PPCTargetLowering::createMemcpyOutsideCallSeq(SDValue Arg, SDValue PtrOff,
-                                              SDValue CallSeqStart,
-                                              ISD::ArgFlagsTy Flags,
-                                              SelectionDAG &DAG,
-                                              SDLoc dl) const {
+SDValue PPCTargetLowering::createMemcpyOutsideCallSeq(
+    SDValue Arg, SDValue PtrOff, SDValue CallSeqStart, ISD::ArgFlagsTy Flags,
+    SelectionDAG &DAG, const SDLoc &dl) const {
   SDValue MemcpyCall = CreateCopyOfByValArgument(Arg, PtrOff,
                         CallSeqStart.getNode()->getOperand(0),
                         Flags, DAG, dl);
@@ -4741,27 +4955,29 @@ PPCTargetLowering::createMemcpyOutsideCallSeq(SDValue Arg, SDValue PtrOff,
   return NewCallSeqStart;
 }
 
-SDValue
-PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
-                                    CallingConv::ID CallConv, bool isVarArg,
-                                    bool isTailCall, bool IsPatchPoint,
-                                    const SmallVectorImpl<ISD::OutputArg> &Outs,
-                                    const SmallVectorImpl<SDValue> &OutVals,
-                                    const SmallVectorImpl<ISD::InputArg> &Ins,
-                                    SDLoc dl, SelectionDAG &DAG,
-                                    SmallVectorImpl<SDValue> &InVals,
-                                    ImmutableCallSite *CS) const {
+SDValue PPCTargetLowering::LowerCall_64SVR4(
+    SDValue Chain, SDValue Callee, CallingConv::ID CallConv, bool isVarArg,
+    bool isTailCall, bool isPatchPoint,
+    const SmallVectorImpl<ISD::OutputArg> &Outs,
+    const SmallVectorImpl<SDValue> &OutVals,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals,
+    ImmutableCallSite *CS) const {
 
   bool isELFv2ABI = Subtarget.isELFv2ABI();
   bool isLittleEndian = Subtarget.isLittleEndian();
   unsigned NumOps = Outs.size();
   bool hasNest = false;
+  bool IsSibCall = false;
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
   unsigned PtrByteSize = 8;
 
   MachineFunction &MF = DAG.getMachineFunction();
 
+  if (isTailCall && !getTargetMachine().Options.GuaranteedTailCallOpt)
+    IsSibCall = true;
+
   // Mark this function as potentially containing a function that contains a
   // tail call. As a consequence the frame pointer will be used for dynamicalloc
   // and restoring the callers stack pointer in this functions epilog. This is
@@ -4881,9 +5097,12 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
       CallConv == CallingConv::Fast)
     NumBytes = EnsureStackAlignment(Subtarget.getFrameLowering(), NumBytes);
 
+  int SPDiff = 0;
+
   // Calculate by how many bytes the stack has to be adjusted in case of tail
   // call optimization.
-  int SPDiff = CalculateTailCallSPDiff(DAG, isTailCall, NumBytes);
+  if (!IsSibCall)
+    SPDiff = CalculateTailCallSPDiff(DAG, isTailCall, NumBytes);
 
   // To protect arguments on the stack from being clobbered in a tail call,
   // force all the loads to happen before doing any other lowering.
@@ -4892,15 +5111,15 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
 
   // Adjust the stack pointer for the new arguments...
   // These operations are automatically eliminated by the prolog/epilog pass
-  Chain = DAG.getCALLSEQ_START(Chain, DAG.getIntPtrConstant(NumBytes, dl, true),
-                               dl);
+  if (!IsSibCall)
+    Chain = DAG.getCALLSEQ_START(Chain,
+                                 DAG.getIntPtrConstant(NumBytes, dl, true), dl);
   SDValue CallSeqStart = Chain;
 
   // Load the return address and frame pointer so it can be move somewhere else
   // later.
   SDValue LROp, FPOp;
-  Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, true,
-                                       dl);
+  Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, dl);
 
   // Set up a copy of the stack pointer for use loading and storing any
   // arguments that may not fit in the registers available for argument
@@ -4980,8 +5199,7 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
         EVT VT = (Size==1) ? MVT::i8 : ((Size==2) ? MVT::i16 : MVT::i32);
         if (GPR_idx != NumGPRs) {
           SDValue Load = DAG.getExtLoad(ISD::EXTLOAD, dl, PtrVT, Chain, Arg,
-                                        MachinePointerInfo(), VT,
-                                        false, false, false, 0);
+                                        MachinePointerInfo(), VT);
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
 
@@ -5041,9 +5259,8 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
                                                           Flags, DAG, dl);
 
         // Load the slot into the register.
-        SDValue Load = DAG.getLoad(PtrVT, dl, Chain, PtrOff,
-                                   MachinePointerInfo(),
-                                   false, false, false, 0);
+        SDValue Load =
+            DAG.getLoad(PtrVT, dl, Chain, PtrOff, MachinePointerInfo());
         MemOpChains.push_back(Load.getValue(1));
         RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
 
@@ -5058,9 +5275,8 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
         SDValue Const = DAG.getConstant(j, dl, PtrOff.getValueType());
         SDValue AddArg = DAG.getNode(ISD::ADD, dl, PtrVT, Arg, Const);
         if (GPR_idx != NumGPRs) {
-          SDValue Load = DAG.getLoad(PtrVT, dl, Chain, AddArg,
-                                     MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(PtrVT, dl, Chain, AddArg, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
           ArgOffset += PtrByteSize;
@@ -5214,13 +5430,12 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
       if (isVarArg) {
         // We could elide this store in the case where the object fits
         // entirely in R registers.  Maybe later.
-        SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
-                                     MachinePointerInfo(), false, false, 0);
+        SDValue Store =
+            DAG.getStore(Chain, dl, Arg, PtrOff, MachinePointerInfo());
         MemOpChains.push_back(Store);
         if (VR_idx != NumVRs) {
-          SDValue Load = DAG.getLoad(MVT::v4f32, dl, Store, PtrOff,
-                                     MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(MVT::v4f32, dl, Store, PtrOff, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
 
           unsigned VReg = (Arg.getSimpleValueType() == MVT::v2f64 ||
@@ -5236,8 +5451,8 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
             break;
           SDValue Ix = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff,
                                    DAG.getConstant(i, dl, PtrVT));
-          SDValue Load = DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
         }
@@ -5278,13 +5493,12 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
       if (isVarArg) {
         // We could elide this store in the case where the object fits
         // entirely in R registers.  Maybe later.
-        SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
-                                     MachinePointerInfo(), false, false, 0);
+        SDValue Store =
+            DAG.getStore(Chain, dl, Arg, PtrOff, MachinePointerInfo());
         MemOpChains.push_back(Store);
         if (QFPR_idx != NumQFPRs) {
-          SDValue Load = DAG.getLoad(IsF32 ? MVT::v4f32 : MVT::v4f64, dl,
-                                     Store, PtrOff, MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load = DAG.getLoad(IsF32 ? MVT::v4f32 : MVT::v4f64, dl, Store,
+                                     PtrOff, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(QFPR[QFPR_idx++], Load));
         }
@@ -5294,8 +5508,8 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
             break;
           SDValue Ix = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff,
                                    DAG.getConstant(i, dl, PtrVT));
-          SDValue Load = DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
         }
@@ -5332,7 +5546,7 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
   // Check if this is an indirect call (MTCTR/BCTRL).
   // See PrepareCall() for more information about calls through function
   // pointers in the 64-bit SVR4 ABI.
-  if (!isTailCall && !IsPatchPoint &&
+  if (!isTailCall && !isPatchPoint &&
       !isFunctionGlobalAddress(Callee) &&
       !isa<ExternalSymbolSDNode>(Callee)) {
     // Load r2 into a virtual register and store it to the TOC save area.
@@ -5344,12 +5558,11 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
     SDValue AddPtr = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr, PtrOff);
     Chain = DAG.getStore(
         Val.getValue(1), dl, Val, AddPtr,
-        MachinePointerInfo::getStack(DAG.getMachineFunction(), TOCSaveOffset),
-        false, false, 0);
+        MachinePointerInfo::getStack(DAG.getMachineFunction(), TOCSaveOffset));
     // In the ELFv2 ABI, R12 must contain the address of an indirect callee.
     // This does not mean the MTCTR instruction must use R12; it's easier
     // to model this as an extra parameter, so do that.
-    if (isELFv2ABI && !IsPatchPoint)
+    if (isELFv2ABI && !isPatchPoint)
       RegsToPass.push_back(std::make_pair((unsigned)PPC::X12, Callee));
   }
 
@@ -5362,29 +5575,27 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
     InFlag = Chain.getValue(1);
   }
 
-  if (isTailCall)
-    PrepareTailCall(DAG, InFlag, Chain, dl, true, SPDiff, NumBytes, LROp,
-                    FPOp, true, TailCallArguments);
+  if (isTailCall && !IsSibCall)
+    PrepareTailCall(DAG, InFlag, Chain, dl, SPDiff, NumBytes, LROp, FPOp,
+                    TailCallArguments);
 
-  return FinishCall(CallConv, dl, isTailCall, isVarArg, IsPatchPoint, hasNest,
+  return FinishCall(CallConv, dl, isTailCall, isVarArg, isPatchPoint, hasNest,
                     DAG, RegsToPass, InFlag, Chain, CallSeqStart, Callee,
                     SPDiff, NumBytes, Ins, InVals, CS);
 }
 
-SDValue
-PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
-                                    CallingConv::ID CallConv, bool isVarArg,
-                                    bool isTailCall, bool IsPatchPoint,
-                                    const SmallVectorImpl<ISD::OutputArg> &Outs,
-                                    const SmallVectorImpl<SDValue> &OutVals,
-                                    const SmallVectorImpl<ISD::InputArg> &Ins,
-                                    SDLoc dl, SelectionDAG &DAG,
-                                    SmallVectorImpl<SDValue> &InVals,
-                                    ImmutableCallSite *CS) const {
+SDValue PPCTargetLowering::LowerCall_Darwin(
+    SDValue Chain, SDValue Callee, CallingConv::ID CallConv, bool isVarArg,
+    bool isTailCall, bool isPatchPoint,
+    const SmallVectorImpl<ISD::OutputArg> &Outs,
+    const SmallVectorImpl<SDValue> &OutVals,
+    const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+    SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals,
+    ImmutableCallSite *CS) const {
 
   unsigned NumOps = Outs.size();
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
   bool isPPC64 = PtrVT == MVT::i64;
   unsigned PtrByteSize = isPPC64 ? 8 : 4;
 
@@ -5467,8 +5678,7 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
   // Load the return address and frame pointer so it can be move somewhere else
   // later.
   SDValue LROp, FPOp;
-  Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, true,
-                                       dl);
+  Chain = EmitTailCallLoadFPAndRetAddr(DAG, SPDiff, Chain, LROp, FPOp, dl);
 
   // Set up a copy of the stack pointer for use loading and storing any
   // arguments that may not fit in the registers available for argument
@@ -5538,8 +5748,7 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
         EVT VT = (Size==1) ? MVT::i8 : MVT::i16;
         if (GPR_idx != NumGPRs) {
           SDValue Load = DAG.getExtLoad(ISD::EXTLOAD, dl, PtrVT, Chain, Arg,
-                                        MachinePointerInfo(), VT,
-                                        false, false, false, 0);
+                                        MachinePointerInfo(), VT);
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
 
@@ -5569,9 +5778,8 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
         SDValue Const = DAG.getConstant(j, dl, PtrOff.getValueType());
         SDValue AddArg = DAG.getNode(ISD::ADD, dl, PtrVT, Arg, Const);
         if (GPR_idx != NumGPRs) {
-          SDValue Load = DAG.getLoad(PtrVT, dl, Chain, AddArg,
-                                     MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(PtrVT, dl, Chain, AddArg, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
           ArgOffset += PtrByteSize;
@@ -5606,24 +5814,22 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
         RegsToPass.push_back(std::make_pair(FPR[FPR_idx++], Arg));
 
         if (isVarArg) {
-          SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
-                                       MachinePointerInfo(), false, false, 0);
+          SDValue Store =
+              DAG.getStore(Chain, dl, Arg, PtrOff, MachinePointerInfo());
           MemOpChains.push_back(Store);
 
           // Float varargs are always shadowed in available integer registers
           if (GPR_idx != NumGPRs) {
-            SDValue Load = DAG.getLoad(PtrVT, dl, Store, PtrOff,
-                                       MachinePointerInfo(), false, false,
-                                       false, 0);
+            SDValue Load =
+                DAG.getLoad(PtrVT, dl, Store, PtrOff, MachinePointerInfo());
             MemOpChains.push_back(Load.getValue(1));
             RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
           }
           if (GPR_idx != NumGPRs && Arg.getValueType() == MVT::f64 && !isPPC64){
             SDValue ConstFour = DAG.getConstant(4, dl, PtrOff.getValueType());
             PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff, ConstFour);
-            SDValue Load = DAG.getLoad(PtrVT, dl, Store, PtrOff,
-                                       MachinePointerInfo(),
-                                       false, false, false, 0);
+            SDValue Load =
+                DAG.getLoad(PtrVT, dl, Store, PtrOff, MachinePointerInfo());
             MemOpChains.push_back(Load.getValue(1));
             RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
           }
@@ -5665,13 +5871,12 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
         // entirely in R registers.  Maybe later.
         PtrOff = DAG.getNode(ISD::ADD, dl, PtrVT, StackPtr,
                              DAG.getConstant(ArgOffset, dl, PtrVT));
-        SDValue Store = DAG.getStore(Chain, dl, Arg, PtrOff,
-                                     MachinePointerInfo(), false, false, 0);
+        SDValue Store =
+            DAG.getStore(Chain, dl, Arg, PtrOff, MachinePointerInfo());
         MemOpChains.push_back(Store);
         if (VR_idx != NumVRs) {
-          SDValue Load = DAG.getLoad(MVT::v4f32, dl, Store, PtrOff,
-                                     MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(MVT::v4f32, dl, Store, PtrOff, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(VR[VR_idx++], Load));
         }
@@ -5681,8 +5886,8 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
             break;
           SDValue Ix = DAG.getNode(ISD::ADD, dl, PtrVT, PtrOff,
                                    DAG.getConstant(i, dl, PtrVT));
-          SDValue Load = DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo(),
-                                     false, false, false, 0);
+          SDValue Load =
+              DAG.getLoad(PtrVT, dl, Store, Ix, MachinePointerInfo());
           MemOpChains.push_back(Load.getValue(1));
           RegsToPass.push_back(std::make_pair(GPR[GPR_idx++], Load));
         }
@@ -5754,10 +5959,10 @@ PPCTargetLowering::LowerCall_Darwin(SDValue Chain, SDValue Callee,
   }
 
   if (isTailCall)
-    PrepareTailCall(DAG, InFlag, Chain, dl, isPPC64, SPDiff, NumBytes, LROp,
-                    FPOp, true, TailCallArguments);
+    PrepareTailCall(DAG, InFlag, Chain, dl, SPDiff, NumBytes, LROp, FPOp,
+                    TailCallArguments);
 
-  return FinishCall(CallConv, dl, isTailCall, isVarArg, IsPatchPoint,
+  return FinishCall(CallConv, dl, isTailCall, isVarArg, isPatchPoint,
                     /* unused except on PPC64 ELFv1 */ false, DAG,
                     RegsToPass, InFlag, Chain, CallSeqStart, Callee, SPDiff,
                     NumBytes, Ins, InVals, CS);
@@ -5774,11 +5979,11 @@ PPCTargetLowering::CanLowerReturn(CallingConv::ID CallConv,
 }
 
 SDValue
-PPCTargetLowering::LowerReturn(SDValue Chain,
-                               CallingConv::ID CallConv, bool isVarArg,
+PPCTargetLowering::LowerReturn(SDValue Chain, CallingConv::ID CallConv,
+                               bool isVarArg,
                                const SmallVectorImpl<ISD::OutputArg> &Outs,
                                const SmallVectorImpl<SDValue> &OutVals,
-                               SDLoc dl, SelectionDAG &DAG) const {
+                               const SDLoc &dl, SelectionDAG &DAG) const {
 
   SmallVector<CCValAssign, 16> RVLocs;
   CCState CCInfo(CallConv, isVarArg, DAG.getMachineFunction(), RVLocs,
@@ -5814,6 +6019,25 @@ PPCTargetLowering::LowerReturn(SDValue Chain,
     RetOps.push_back(DAG.getRegister(VA.getLocReg(), VA.getLocVT()));
   }
 
+  const PPCRegisterInfo *TRI = Subtarget.getRegisterInfo();
+  const MCPhysReg *I =
+    TRI->getCalleeSavedRegsViaCopy(&DAG.getMachineFunction());
+  if (I) {
+    for (; *I; ++I) {
+
+      if (PPC::G8RCRegClass.contains(*I))
+        RetOps.push_back(DAG.getRegister(*I, MVT::i64));
+      else if (PPC::F8RCRegClass.contains(*I))
+        RetOps.push_back(DAG.getRegister(*I, MVT::getFloatingPointVT(64)));
+      else if (PPC::CRRCRegClass.contains(*I))
+        RetOps.push_back(DAG.getRegister(*I, MVT::i1));
+      else if (PPC::VRRCRegClass.contains(*I))
+        RetOps.push_back(DAG.getRegister(*I, MVT::Other));
+      else
+        llvm_unreachable("Unexpected register class in CSRsViaCopy!");
+    }
+  }
+
   RetOps[0] = Chain;  // Update chain.
 
   // Add the flag if we have it.
@@ -5823,8 +6047,9 @@ PPCTargetLowering::LowerReturn(SDValue Chain,
   return DAG.getNode(PPCISD::RET_FLAG, dl, MVT::Other, RetOps);
 }
 
-SDValue PPCTargetLowering::LowerGET_DYNAMIC_AREA_OFFSET(
-    SDValue Op, SelectionDAG &DAG, const PPCSubtarget &Subtarget) const {
+SDValue
+PPCTargetLowering::LowerGET_DYNAMIC_AREA_OFFSET(SDValue Op,
+                                                SelectionDAG &DAG) const {
   SDLoc dl(Op);
 
   // Get the corect type for integers.
@@ -5839,13 +6064,13 @@ SDValue PPCTargetLowering::LowerGET_DYNAMIC_AREA_OFFSET(
   return DAG.getNode(PPCISD::DYNAREAOFFSET, dl, VTs, Ops);
 }
 
-SDValue PPCTargetLowering::LowerSTACKRESTORE(SDValue Op, SelectionDAG &DAG,
-                                   const PPCSubtarget &Subtarget) const {
+SDValue PPCTargetLowering::LowerSTACKRESTORE(SDValue Op,
+                                             SelectionDAG &DAG) const {
   // When we pop the dynamic allocation we need to restore the SP link.
   SDLoc dl(Op);
 
   // Get the corect type for pointers.
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
 
   // Construct the stack pointer operand.
   bool isPPC64 = Subtarget.isPPC64();
@@ -5857,22 +6082,20 @@ SDValue PPCTargetLowering::LowerSTACKRESTORE(SDValue Op, SelectionDAG &DAG,
   SDValue SaveSP = Op.getOperand(1);
 
   // Load the old link SP.
-  SDValue LoadLinkSP = DAG.getLoad(PtrVT, dl, Chain, StackPtr,
-                                   MachinePointerInfo(),
-                                   false, false, false, 0);
+  SDValue LoadLinkSP =
+      DAG.getLoad(PtrVT, dl, Chain, StackPtr, MachinePointerInfo());
 
   // Restore the stack pointer.
   Chain = DAG.getCopyToReg(LoadLinkSP.getValue(1), dl, SP, SaveSP);
 
   // Store the old link SP.
-  return DAG.getStore(Chain, dl, LoadLinkSP, StackPtr, MachinePointerInfo(),
-                      false, false, 0);
+  return DAG.getStore(Chain, dl, LoadLinkSP, StackPtr, MachinePointerInfo());
 }
 
 SDValue PPCTargetLowering::getReturnAddrFrameIndex(SelectionDAG &DAG) const {
   MachineFunction &MF = DAG.getMachineFunction();
   bool isPPC64 = Subtarget.isPPC64();
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
 
   // Get current frame pointer save index.  The users of this index will be
   // primarily DYNALLOC instructions.
@@ -5895,7 +6118,7 @@ SDValue
 PPCTargetLowering::getFramePointerFrameIndex(SelectionDAG & DAG) const {
   MachineFunction &MF = DAG.getMachineFunction();
   bool isPPC64 = Subtarget.isPPC64();
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
 
   // Get current frame pointer save index.  The users of this index will be
   // primarily DYNALLOC instructions.
@@ -5915,15 +6138,14 @@ PPCTargetLowering::getFramePointerFrameIndex(SelectionDAG & DAG) const {
 }
 
 SDValue PPCTargetLowering::LowerDYNAMIC_STACKALLOC(SDValue Op,
-                                         SelectionDAG &DAG,
-                                         const PPCSubtarget &Subtarget) const {
+                                                   SelectionDAG &DAG) const {
   // Get the inputs.
   SDValue Chain = Op.getOperand(0);
   SDValue Size  = Op.getOperand(1);
   SDLoc dl(Op);
 
   // Get the corect type for pointers.
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+  EVT PtrVT = getPointerTy(DAG.getDataLayout());
   // Negate the size.
   SDValue NegSize = DAG.getNode(ISD::SUB, dl, PtrVT,
                                 DAG.getConstant(0, dl, PtrVT), Size);
@@ -6113,7 +6335,7 @@ SDValue PPCTargetLowering::LowerSELECT_CC(SDValue Op, SelectionDAG &DAG) const {
 
 void PPCTargetLowering::LowerFP_TO_INTForReuse(SDValue Op, ReuseLoadInfo &RLI,
                                                SelectionDAG &DAG,
-                                               SDLoc dl) const {
+                                               const SDLoc &dl) const {
   assert(Op.getOperand(0).getValueType().isFloatingPoint());
   SDValue Src = Op.getOperand(0);
   if (Src.getValueType() == MVT::f32)
@@ -6156,15 +6378,14 @@ void PPCTargetLowering::LowerFP_TO_INTForReuse(SDValue Op, ReuseLoadInfo &RLI,
     Chain = DAG.getMemIntrinsicNode(PPCISD::STFIWX, dl,
               DAG.getVTList(MVT::Other), Ops, MVT::i32, MMO);
   } else
-    Chain = DAG.getStore(DAG.getEntryNode(), dl, Tmp, FIPtr,
-                         MPI, false, false, 0);
+    Chain = DAG.getStore(DAG.getEntryNode(), dl, Tmp, FIPtr, MPI);
 
   // Result is a load from the stack slot.  If loading 4 bytes, make sure to
-  // add in a bias.
+  // add in a bias on big endian.
   if (Op.getValueType() == MVT::i32 && !i32Stack) {
     FIPtr = DAG.getNode(ISD::ADD, dl, FIPtr.getValueType(), FIPtr,
                         DAG.getConstant(4, dl, FIPtr.getValueType()));
-    MPI = MPI.getWithOffset(4);
+    MPI = MPI.getWithOffset(Subtarget.isLittleEndian() ? 0 : 4);
   }
 
   RLI.Chain = Chain;
@@ -6177,7 +6398,7 @@ void PPCTargetLowering::LowerFP_TO_INTForReuse(SDValue Op, ReuseLoadInfo &RLI,
 /// need for load/store combinations.
 SDValue PPCTargetLowering::LowerFP_TO_INTDirectMove(SDValue Op,
                                                     SelectionDAG &DAG,
-                                                    SDLoc dl) const {
+                                                    const SDLoc &dl) const {
   assert(Op.getOperand(0).getValueType().isFloatingPoint());
   SDValue Src = Op.getOperand(0);
 
@@ -6208,16 +6429,18 @@ SDValue PPCTargetLowering::LowerFP_TO_INTDirectMove(SDValue Op,
 }
 
 SDValue PPCTargetLowering::LowerFP_TO_INT(SDValue Op, SelectionDAG &DAG,
-                                          SDLoc dl) const {
+                                          const SDLoc &dl) const {
   if (Subtarget.hasDirectMove() && Subtarget.isPPC64())
     return LowerFP_TO_INTDirectMove(Op, DAG, dl);
 
   ReuseLoadInfo RLI;
   LowerFP_TO_INTForReuse(Op, RLI, DAG, dl);
 
-  return DAG.getLoad(Op.getValueType(), dl, RLI.Chain, RLI.Ptr, RLI.MPI, false,
-                     false, RLI.IsInvariant, RLI.Alignment, RLI.AAInfo,
-                     RLI.Ranges);
+  return DAG.getLoad(Op.getValueType(), dl, RLI.Chain, RLI.Ptr, RLI.MPI,
+                     RLI.Alignment,
+                     RLI.IsInvariant ? MachineMemOperand::MOInvariant
+                                     : MachineMemOperand::MONone,
+                     RLI.AAInfo, RLI.Ranges);
 }
 
 // We're trying to insert a regular store, S, and then a load, L. If the
@@ -6251,7 +6474,7 @@ bool PPCTargetLowering::canReuseLoadAddress(SDValue Op, EVT MemVT,
     return false;
 
   RLI.Ptr = LD->getBasePtr();
-  if (LD->isIndexed() && LD->getOffset().getOpcode() != ISD::UNDEF) {
+  if (LD->isIndexed() && !LD->getOffset().isUndef()) {
     assert(LD->getAddressingMode() == ISD::PRE_INC &&
            "Non-pre-inc AM on PPC?");
     RLI.Ptr = DAG.getNode(ISD::ADD, dl, RLI.Ptr.getValueType(), RLI.Ptr,
@@ -6289,12 +6512,36 @@ void PPCTargetLowering::spliceIntoChain(SDValue ResChain,
   DAG.UpdateNodeOperands(TF.getNode(), ResChain, NewResChain);
 }
 
+/// \brief Analyze profitability of direct move
+/// prefer float load to int load plus direct move
+/// when there is no integer use of int load
+static bool directMoveIsProfitable(const SDValue &Op) {
+  SDNode *Origin = Op.getOperand(0).getNode();
+  if (Origin->getOpcode() != ISD::LOAD)
+    return true;
+
+  for (SDNode::use_iterator UI = Origin->use_begin(),
+                            UE = Origin->use_end();
+       UI != UE; ++UI) {
+
+    // Only look at the users of the loaded value.
+    if (UI.getUse().get().getResNo() != 0)
+      continue;
+
+    if (UI->getOpcode() != ISD::SINT_TO_FP &&
+        UI->getOpcode() != ISD::UINT_TO_FP)
+      return true;
+  }
+
+  return false;
+}
+
 /// \brief Custom lowers integer to floating point conversions to use
 /// the direct move instructions available in ISA 2.07 to avoid the
 /// need for load/store combinations.
 SDValue PPCTargetLowering::LowerINT_TO_FPDirectMove(SDValue Op,
                                                     SelectionDAG &DAG,
-                                                    SDLoc dl) const {
+                                                    const SDLoc &dl) const {
   assert((Op.getValueType() == MVT::f32 ||
           Op.getValueType() == MVT::f64) &&
          "Invalid floating point type as target of conversion");
@@ -6335,9 +6582,7 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
     // This can be done with an fma and the 0.5 constant: (V+1.0)*0.5 = 0.5*V+0.5
     Value = DAG.getNode(PPCISD::QBFLT, dl, MVT::v4f64, Value);
 
-    SDValue FPHalfs = DAG.getConstantFP(0.5, dl, MVT::f64);
-    FPHalfs = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4f64, FPHalfs, FPHalfs,
-                          FPHalfs, FPHalfs);
+    SDValue FPHalfs = DAG.getConstantFP(0.5, dl, MVT::v4f64);
 
     Value = DAG.getNode(ISD::FMA, dl, MVT::v4f64, Value, FPHalfs, FPHalfs);
 
@@ -6359,7 +6604,8 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
 
   // If we have direct moves, we can do all the conversion, skip the store/load
   // however, without FPCVT we can't do most conversions.
-  if (Subtarget.hasDirectMove() && Subtarget.isPPC64() && Subtarget.hasFPCVT())
+  if (Subtarget.hasDirectMove() && directMoveIsProfitable(Op) &&
+      Subtarget.isPPC64() && Subtarget.hasFPCVT())
     return LowerINT_TO_FPDirectMove(Op, DAG, dl);
 
   assert((Op.getOpcode() == ISD::SINT_TO_FP || Subtarget.hasFPCVT()) &&
@@ -6429,9 +6675,11 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
 
     MachineFunction &MF = DAG.getMachineFunction();
     if (canReuseLoadAddress(SINT, MVT::i64, RLI, DAG)) {
-      Bits = DAG.getLoad(MVT::f64, dl, RLI.Chain, RLI.Ptr, RLI.MPI, false,
-                         false, RLI.IsInvariant, RLI.Alignment, RLI.AAInfo,
-                         RLI.Ranges);
+      Bits =
+          DAG.getLoad(MVT::f64, dl, RLI.Chain, RLI.Ptr, RLI.MPI, RLI.Alignment,
+                      RLI.IsInvariant ? MachineMemOperand::MOInvariant
+                                      : MachineMemOperand::MONone,
+                      RLI.AAInfo, RLI.Ranges);
       spliceIntoChain(RLI.ResChain, Bits.getValue(1), DAG);
     } else if (Subtarget.hasLFIWAX() &&
                canReuseLoadAddress(SINT, MVT::i32, RLI, DAG, ISD::SEXTLOAD)) {
@@ -6459,15 +6707,15 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
                  SINT.getOpcode() == ISD::ZERO_EXTEND)) &&
                SINT.getOperand(0).getValueType() == MVT::i32) {
       MachineFrameInfo *FrameInfo = MF.getFrameInfo();
-      EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(DAG.getDataLayout());
+      EVT PtrVT = getPointerTy(DAG.getDataLayout());
 
       int FrameIdx = FrameInfo->CreateStackObject(4, 4, false);
       SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
 
-      SDValue Store = DAG.getStore(
-          DAG.getEntryNode(), dl, SINT.getOperand(0), FIdx,
-          MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FrameIdx),
-          false, false, 0);
+      SDValue Store =
+          DAG.getStore(DAG.getEntryNode(), dl, SINT.getOperand(0), FIdx,
+                       MachinePointerInfo::getFixedStack(
+                           DAG.getMachineFunction(), FrameIdx));
 
       assert(cast<StoreSDNode>(Store)->getMemoryVT() == MVT::i32 &&
              "Expected an i32 store");
@@ -6505,7 +6753,7 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
   // then lfd it and fcfid it.
   MachineFunction &MF = DAG.getMachineFunction();
   MachineFrameInfo *FrameInfo = MF.getFrameInfo();
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
 
   SDValue Ld;
   if (Subtarget.hasLFIWAX() || Subtarget.hasFPCVT()) {
@@ -6516,10 +6764,10 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
       int FrameIdx = FrameInfo->CreateStackObject(4, 4, false);
       SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
 
-      SDValue Store = DAG.getStore(
-          DAG.getEntryNode(), dl, Op.getOperand(0), FIdx,
-          MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FrameIdx),
-          false, false, 0);
+      SDValue Store =
+          DAG.getStore(DAG.getEntryNode(), dl, Op.getOperand(0), FIdx,
+                       MachinePointerInfo::getFixedStack(
+                           DAG.getMachineFunction(), FrameIdx));
 
       assert(cast<StoreSDNode>(Store)->getMemoryVT() == MVT::i32 &&
              "Expected an i32 store");
@@ -6554,14 +6802,12 @@ SDValue PPCTargetLowering::LowerINT_TO_FP(SDValue Op,
     // STD the extended value into the stack slot.
     SDValue Store = DAG.getStore(
         DAG.getEntryNode(), dl, Ext64, FIdx,
-        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FrameIdx),
-        false, false, 0);
+        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FrameIdx));
 
     // Load the value as a double.
     Ld = DAG.getLoad(
         MVT::f64, dl, Store, FIdx,
-        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FrameIdx),
-        false, false, false, 0);
+        MachinePointerInfo::getFixedStack(DAG.getMachineFunction(), FrameIdx));
   }
 
   // FCFID it and return it.
@@ -6596,7 +6842,7 @@ SDValue PPCTargetLowering::LowerFLT_ROUNDS_(SDValue Op,
 
   MachineFunction &MF = DAG.getMachineFunction();
   EVT VT = Op.getValueType();
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
 
   // Save FP Control Word to register
   EVT NodeTys[] = {
@@ -6608,14 +6854,13 @@ SDValue PPCTargetLowering::LowerFLT_ROUNDS_(SDValue Op,
   // Save FP register to stack slot
   int SSFI = MF.getFrameInfo()->CreateStackObject(8, 8, false);
   SDValue StackSlot = DAG.getFrameIndex(SSFI, PtrVT);
-  SDValue Store = DAG.getStore(DAG.getEntryNode(), dl, Chain,
-                               StackSlot, MachinePointerInfo(), false, false,0);
+  SDValue Store = DAG.getStore(DAG.getEntryNode(), dl, Chain, StackSlot,
+                               MachinePointerInfo());
 
   // Load FP Control Word from low 32 bits of stack slot.
   SDValue Four = DAG.getConstant(4, dl, PtrVT);
   SDValue Addr = DAG.getNode(ISD::ADD, dl, PtrVT, StackSlot, Four);
-  SDValue CWD = DAG.getLoad(MVT::i32, dl, Store, Addr, MachinePointerInfo(),
-                            false, false, false, 0);
+  SDValue CWD = DAG.getLoad(MVT::i32, dl, Store, Addr, MachinePointerInfo());
 
   // Transform as necessary
   SDValue CWD1 =
@@ -6730,7 +6975,7 @@ SDValue PPCTargetLowering::LowerSRA_PARTS(SDValue Op, SelectionDAG &DAG) const {
 /// BuildSplatI - Build a canonical splati of Val with an element size of
 /// SplatSize.  Cast the result to VT.
 static SDValue BuildSplatI(int Val, unsigned SplatSize, EVT VT,
-                             SelectionDAG &DAG, SDLoc dl) {
+                           SelectionDAG &DAG, const SDLoc &dl) {
   assert(Val >= -16 && Val <= 15 && "vsplti is out of range!");
 
   static const MVT VTys[] = { // canonical VT to use for each size.
@@ -6746,18 +6991,13 @@ static SDValue BuildSplatI(int Val, unsigned SplatSize, EVT VT,
   EVT CanonicalVT = VTys[SplatSize-1];
 
   // Build a canonical splat for this value.
-  SDValue Elt = DAG.getConstant(Val, dl, MVT::i32);
-  SmallVector<SDValue, 8> Ops;
-  Ops.assign(CanonicalVT.getVectorNumElements(), Elt);
-  SDValue Res = DAG.getNode(ISD::BUILD_VECTOR, dl, CanonicalVT, Ops);
-  return DAG.getNode(ISD::BITCAST, dl, ReqVT, Res);
+  return DAG.getBitcast(ReqVT, DAG.getConstant(Val, dl, CanonicalVT));
 }
 
 /// BuildIntrinsicOp - Return a unary operator intrinsic node with the
 /// specified intrinsic ID.
-static SDValue BuildIntrinsicOp(unsigned IID, SDValue Op,
-                                SelectionDAG &DAG, SDLoc dl,
-                                EVT DestVT = MVT::Other) {
+static SDValue BuildIntrinsicOp(unsigned IID, SDValue Op, SelectionDAG &DAG,
+                                const SDLoc &dl, EVT DestVT = MVT::Other) {
   if (DestVT == MVT::Other) DestVT = Op.getValueType();
   return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, dl, DestVT,
                      DAG.getConstant(IID, dl, MVT::i32), Op);
@@ -6766,7 +7006,7 @@ static SDValue BuildIntrinsicOp(unsigned IID, SDValue Op,
 /// BuildIntrinsicOp - Return a binary operator intrinsic node with the
 /// specified intrinsic ID.
 static SDValue BuildIntrinsicOp(unsigned IID, SDValue LHS, SDValue RHS,
-                                SelectionDAG &DAG, SDLoc dl,
+                                SelectionDAG &DAG, const SDLoc &dl,
                                 EVT DestVT = MVT::Other) {
   if (DestVT == MVT::Other) DestVT = LHS.getValueType();
   return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, dl, DestVT,
@@ -6776,8 +7016,8 @@ static SDValue BuildIntrinsicOp(unsigned IID, SDValue LHS, SDValue RHS,
 /// BuildIntrinsicOp - Return a ternary operator intrinsic node with the
 /// specified intrinsic ID.
 static SDValue BuildIntrinsicOp(unsigned IID, SDValue Op0, SDValue Op1,
-                                SDValue Op2, SelectionDAG &DAG,
-                                SDLoc dl, EVT DestVT = MVT::Other) {
+                                SDValue Op2, SelectionDAG &DAG, const SDLoc &dl,
+                                EVT DestVT = MVT::Other) {
   if (DestVT == MVT::Other) DestVT = Op0.getValueType();
   return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, dl, DestVT,
                      DAG.getConstant(IID, dl, MVT::i32), Op0, Op1, Op2);
@@ -6785,8 +7025,8 @@ static SDValue BuildIntrinsicOp(unsigned IID, SDValue Op0, SDValue Op1,
 
 /// BuildVSLDOI - Return a VECTOR_SHUFFLE that is a vsldoi of the specified
 /// amount.  The result has the specified value type.
-static SDValue BuildVSLDOI(SDValue LHS, SDValue RHS, unsigned Amt,
-                             EVT VT, SelectionDAG &DAG, SDLoc dl) {
+static SDValue BuildVSLDOI(SDValue LHS, SDValue RHS, unsigned Amt, EVT VT,
+                           SelectionDAG &DAG, const SDLoc &dl) {
   // Force LHS/RHS to be the right type.
   LHS = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, LHS);
   RHS = DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, RHS);
@@ -6825,7 +7065,7 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
 
     bool IsConst = true;
     for (unsigned i = 0; i < 4; ++i) {
-      if (BVN->getOperand(i).getOpcode() == ISD::UNDEF) continue;
+      if (BVN->getOperand(i).isUndef()) continue;
       if (!isa<ConstantSDNode>(BVN->getOperand(i))) {
         IsConst = false;
         break;
@@ -6838,12 +7078,12 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
       Constant *NegOne =
         ConstantFP::get(Type::getFloatTy(*DAG.getContext()), -1.0);
 
-      SmallVector<Constant*, 4> CV(4, NegOne);
+      Constant *CV[4];
       for (unsigned i = 0; i < 4; ++i) {
-        if (BVN->getOperand(i).getOpcode() == ISD::UNDEF)
+        if (BVN->getOperand(i).isUndef())
           CV[i] = UndefValue::get(Type::getFloatTy(*DAG.getContext()));
         else if (isNullConstant(BVN->getOperand(i)))
-          continue;
+          CV[i] = NegOne;
         else
           CV[i] = One;
       }
@@ -6852,15 +7092,8 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
       SDValue CPIdx = DAG.getConstantPool(CP, getPointerTy(DAG.getDataLayout()),
                                           16 /* alignment */);
 
-      SmallVector<SDValue, 2> Ops;
-      Ops.push_back(DAG.getEntryNode());
-      Ops.push_back(CPIdx);
-
-      SmallVector<EVT, 2> ValueVTs;
-      ValueVTs.push_back(MVT::v4i1);
-      ValueVTs.push_back(MVT::Other); // chain
-      SDVTList VTs = DAG.getVTList(ValueVTs);
-
+      SDValue Ops[] = {DAG.getEntryNode(), CPIdx};
+      SDVTList VTs = DAG.getVTList({MVT::v4i1, /*chain*/ MVT::Other});
       return DAG.getMemIntrinsicNode(
           PPCISD::QVLFSb, dl, VTs, Ops, MVT::v4f32,
           MachinePointerInfo::getConstantPool(DAG.getMachineFunction()));
@@ -6868,7 +7101,7 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
 
     SmallVector<SDValue, 4> Stores;
     for (unsigned i = 0; i < 4; ++i) {
-      if (BVN->getOperand(i).getOpcode() == ISD::UNDEF) continue;
+      if (BVN->getOperand(i).isUndef()) continue;
 
       unsigned Offset = 4*i;
       SDValue Idx = DAG.getConstant(Offset, dl, FIdx.getValueType());
@@ -6876,19 +7109,16 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
 
       unsigned StoreSize = BVN->getOperand(i).getValueType().getStoreSize();
       if (StoreSize > 4) {
-        Stores.push_back(DAG.getTruncStore(DAG.getEntryNode(), dl,
-                                           BVN->getOperand(i), Idx,
-                                           PtrInfo.getWithOffset(Offset),
-                                           MVT::i32, false, false, 0));
+        Stores.push_back(
+            DAG.getTruncStore(DAG.getEntryNode(), dl, BVN->getOperand(i), Idx,
+                              PtrInfo.getWithOffset(Offset), MVT::i32));
       } else {
         SDValue StoreValue = BVN->getOperand(i);
         if (StoreSize < 4)
           StoreValue = DAG.getNode(ISD::ANY_EXTEND, dl, MVT::i32, StoreValue);
 
-        Stores.push_back(DAG.getStore(DAG.getEntryNode(), dl,
-                                      StoreValue, Idx,
-                                      PtrInfo.getWithOffset(Offset),
-                                      false, false, 0));
+        Stores.push_back(DAG.getStore(DAG.getEntryNode(), dl, StoreValue, Idx,
+                                      PtrInfo.getWithOffset(Offset)));
       }
     }
 
@@ -6903,15 +7133,10 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
     // is typed as v4f64 because the QPX register integer states are not
     // explicitly represented.
 
-    SmallVector<SDValue, 2> Ops;
-    Ops.push_back(StoreChain);
-    Ops.push_back(DAG.getConstant(Intrinsic::ppc_qpx_qvlfiwz, dl, MVT::i32));
-    Ops.push_back(FIdx);
-
-    SmallVector<EVT, 2> ValueVTs;
-    ValueVTs.push_back(MVT::v4f64);
-    ValueVTs.push_back(MVT::Other); // chain
-    SDVTList VTs = DAG.getVTList(ValueVTs);
+    SDValue Ops[] = {StoreChain,
+                     DAG.getConstant(Intrinsic::ppc_qpx_qvlfiwz, dl, MVT::i32),
+                     FIdx};
+    SDVTList VTs = DAG.getVTList({MVT::v4f64, /*chain*/ MVT::Other});
 
     SDValue LoadedVect = DAG.getMemIntrinsicNode(ISD::INTRINSIC_W_CHAIN,
       dl, VTs, Ops, MVT::v4i32, PtrInfo);
@@ -6919,9 +7144,7 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
       DAG.getConstant(Intrinsic::ppc_qpx_qvfcfidu, dl, MVT::i32),
       LoadedVect);
 
-    SDValue FPZeros = DAG.getConstantFP(0.0, dl, MVT::f64);
-    FPZeros = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4f64,
-                          FPZeros, FPZeros, FPZeros, FPZeros);
+    SDValue FPZeros = DAG.getConstantFP(0.0, dl, MVT::v4f64);
 
     return DAG.getSetCC(dl, MVT::v4i1, LoadedVect, FPZeros, ISD::SETEQ);
   }
@@ -6949,8 +7172,7 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
   if (SplatBits == 0) {
     // Canonicalize all zero vectors to be v4i32.
     if (Op.getValueType() != MVT::v4i32 || HasAnyUndefs) {
-      SDValue Z = DAG.getConstant(0, dl, MVT::i32);
-      Z = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Z, Z, Z, Z);
+      SDValue Z = DAG.getConstant(0, dl, MVT::v4i32);
       Op = DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), Z);
     }
     return Op;
@@ -7089,7 +7311,7 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
 /// the specified operations to build the shuffle.
 static SDValue GeneratePerfectShuffle(unsigned PFEntry, SDValue LHS,
                                       SDValue RHS, SelectionDAG &DAG,
-                                      SDLoc dl) {
+                                      const SDLoc &dl) {
   unsigned OpNum = (PFEntry >> 26) & 0x0F;
   unsigned LHSID = (PFEntry >> 13) & ((1 << 13)-1);
   unsigned RHSID = (PFEntry >>  0) & ((1 << 13)-1);
@@ -7175,11 +7397,50 @@ SDValue PPCTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
   EVT VT = Op.getValueType();
   bool isLittleEndian = Subtarget.isLittleEndian();
 
+  unsigned ShiftElts, InsertAtByte;
+  bool Swap;
+  if (Subtarget.hasP9Vector() &&
+      PPC::isXXINSERTWMask(SVOp, ShiftElts, InsertAtByte, Swap,
+                           isLittleEndian)) {
+    if (Swap)
+      std::swap(V1, V2);
+    SDValue Conv1 = DAG.getNode(ISD::BITCAST, dl, MVT::v4i32, V1);
+    SDValue Conv2 = DAG.getNode(ISD::BITCAST, dl, MVT::v4i32, V2);
+    if (ShiftElts) {
+      SDValue Shl = DAG.getNode(PPCISD::VECSHL, dl, MVT::v4i32, Conv2, Conv2,
+                                DAG.getConstant(ShiftElts, dl, MVT::i32));
+      SDValue Ins = DAG.getNode(PPCISD::XXINSERT, dl, MVT::v4i32, Conv1, Shl,
+                                DAG.getConstant(InsertAtByte, dl, MVT::i32));
+      return DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, Ins);
+    }
+    SDValue Ins = DAG.getNode(PPCISD::XXINSERT, dl, MVT::v4i32, Conv1, Conv2,
+                              DAG.getConstant(InsertAtByte, dl, MVT::i32));
+    return DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, Ins);
+  }
+
+  if (Subtarget.hasVSX()) {
+    if (V2.isUndef() && PPC::isSplatShuffleMask(SVOp, 4)) {
+      int SplatIdx = PPC::getVSPLTImmediate(SVOp, 4, DAG);
+      SDValue Conv = DAG.getNode(ISD::BITCAST, dl, MVT::v4i32, V1);
+      SDValue Splat = DAG.getNode(PPCISD::XXSPLT, dl, MVT::v4i32, Conv,
+                                  DAG.getConstant(SplatIdx, dl, MVT::i32));
+      return DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, Splat);
+    }
+
+    // Left shifts of 8 bytes are actually swaps. Convert accordingly.
+    if (V2.isUndef() && PPC::isVSLDOIShuffleMask(SVOp, 1, DAG) == 8) {
+      SDValue Conv = DAG.getNode(ISD::BITCAST, dl, MVT::v2f64, V1);
+      SDValue Swap = DAG.getNode(PPCISD::SWAP_NO_CHAIN, dl, MVT::v2f64, Conv);
+      return DAG.getNode(ISD::BITCAST, dl, MVT::v16i8, Swap);
+    }
+
+  }
+
   if (Subtarget.hasQPX()) {
     if (VT.getVectorNumElements() != 4)
       return SDValue();
 
-    if (V2.getOpcode() == ISD::UNDEF) V2 = V1;
+    if (V2.isUndef()) V2 = V1;
 
     int AlignIdx = PPC::isQVALIGNIShuffleMask(SVOp);
     if (AlignIdx != -1) {
@@ -7192,9 +7453,6 @@ SDValue PPCTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
         SplatIdx -= 4;
       }
 
-      // FIXME: If SplatIdx == 0 and the input came from a load, then there is
-      // nothing to do.
-
       return DAG.getNode(PPCISD::QVESPLATI, dl, VT, V1,
                          DAG.getConstant(SplatIdx, dl, MVT::i32));
     }
@@ -7217,7 +7475,7 @@ SDValue PPCTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
   // Cases that are handled by instructions that take permute immediates
   // (such as vsplt*) should be left as VECTOR_SHUFFLE nodes so they can be
   // selected by the instruction selector.
-  if (V2.getOpcode() == ISD::UNDEF) {
+  if (V2.isUndef()) {
     if (PPC::isSplatShuffleMask(SVOp, 1) ||
         PPC::isSplatShuffleMask(SVOp, 2) ||
         PPC::isSplatShuffleMask(SVOp, 4) ||
@@ -7315,7 +7573,7 @@ SDValue PPCTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
 
   // Lower this to a VPERM(V1, V2, V3) expression, where V3 is a constant
   // vector that will get spilled to the constant pool.
-  if (V2.getOpcode() == ISD::UNDEF) V2 = V1;
+  if (V2.isUndef()) V2 = V1;
 
   // The SHUFFLE_VECTOR mask is almost exactly what we want for vperm, except
   // that it is in input element units, not in bytes.  Convert now.
@@ -7340,8 +7598,7 @@ SDValue PPCTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
                                              MVT::i32));
   }
 
-  SDValue VPermMask = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v16i8,
-                                  ResultMask);
+  SDValue VPermMask = DAG.getBuildVector(MVT::v16i8, dl, ResultMask);
   if (isLittleEndian)
     return DAG.getNode(PPCISD::VPERM, dl, V1.getValueType(),
                        V2, V1, VPermMask);
@@ -7468,6 +7725,16 @@ static bool getVectorCompareInfo(SDValue Intrin, int &CompareOpc,
 /// lower, do it, otherwise return null.
 SDValue PPCTargetLowering::LowerINTRINSIC_WO_CHAIN(SDValue Op,
                                                    SelectionDAG &DAG) const {
+  unsigned IntrinsicID =
+    cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue();
+
+  if (IntrinsicID == Intrinsic::thread_pointer) {
+    // Reads the thread pointer register, used for __builtin_thread_pointer.
+    bool is64bit = Subtarget.isPPC64();
+    return DAG.getRegister(is64bit ? PPC::X13 : PPC::R2,
+                           is64bit ? MVT::i64 : MVT::i32);
+  }
+
   // If this is a lowered altivec predicate compare, CompareOpc is set to the
   // opcode number of the comparison.
   SDLoc dl(Op);
@@ -7566,12 +7833,10 @@ SDValue PPCTargetLowering::LowerSCALAR_TO_VECTOR(SDValue Op,
   SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
 
   // Store the input value into Value#0 of the stack slot.
-  SDValue Store = DAG.getStore(DAG.getEntryNode(), dl,
-                               Op.getOperand(0), FIdx, MachinePointerInfo(),
-                               false, false, 0);
+  SDValue Store = DAG.getStore(DAG.getEntryNode(), dl, Op.getOperand(0), FIdx,
+                               MachinePointerInfo());
   // Load it out.
-  return DAG.getLoad(Op.getValueType(), dl, Store, FIdx, MachinePointerInfo(),
-                     false, false, false, 0);
+  return DAG.getLoad(Op.getValueType(), dl, Store, FIdx, MachinePointerInfo());
 }
 
 SDValue PPCTargetLowering::LowerEXTRACT_VECTOR_ELT(SDValue Op,
@@ -7594,9 +7859,7 @@ SDValue PPCTargetLowering::LowerEXTRACT_VECTOR_ELT(SDValue Op,
 
   // FIXME: We can make this an f32 vector, but the BUILD_VECTOR code needs to
   // understand how to form the extending load.
-  SDValue FPHalfs = DAG.getConstantFP(0.5, dl, MVT::f64);
-  FPHalfs = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4f64,
-                        FPHalfs, FPHalfs, FPHalfs, FPHalfs);
+  SDValue FPHalfs = DAG.getConstantFP(0.5, dl, MVT::v4f64);
 
   Value = DAG.getNode(ISD::FMA, dl, MVT::v4f64, Value, FPHalfs, FPHalfs);
 
@@ -7613,15 +7876,10 @@ SDValue PPCTargetLowering::LowerEXTRACT_VECTOR_ELT(SDValue Op,
   SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
 
   SDValue StoreChain = DAG.getEntryNode();
-  SmallVector<SDValue, 2> Ops;
-  Ops.push_back(StoreChain);
-  Ops.push_back(DAG.getConstant(Intrinsic::ppc_qpx_qvstfiw, dl, MVT::i32));
-  Ops.push_back(Value);
-  Ops.push_back(FIdx);
-
-  SmallVector<EVT, 2> ValueVTs;
-  ValueVTs.push_back(MVT::Other); // chain
-  SDVTList VTs = DAG.getVTList(ValueVTs);
+  SDValue Ops[] = {StoreChain,
+                   DAG.getConstant(Intrinsic::ppc_qpx_qvstfiw, dl, MVT::i32),
+                   Value, FIdx};
+  SDVTList VTs = DAG.getVTList(/*chain*/ MVT::Other);
 
   StoreChain = DAG.getMemIntrinsicNode(ISD::INTRINSIC_VOID,
     dl, VTs, Ops, MVT::v4i32, PtrInfo);
@@ -7631,9 +7889,8 @@ SDValue PPCTargetLowering::LowerEXTRACT_VECTOR_ELT(SDValue Op,
   SDValue Idx = DAG.getConstant(Offset, dl, FIdx.getValueType());
   Idx = DAG.getNode(ISD::ADD, dl, FIdx.getValueType(), FIdx, Idx);
 
-  SDValue IntVal = DAG.getLoad(MVT::i32, dl, StoreChain, Idx,
-                               PtrInfo.getWithOffset(Offset),
-                               false, false, false, 0);
+  SDValue IntVal =
+      DAG.getLoad(MVT::i32, dl, StoreChain, Idx, PtrInfo.getWithOffset(Offset));
 
   if (!Subtarget.useCRBits())
     return IntVal;
@@ -7662,24 +7919,20 @@ SDValue PPCTargetLowering::LowerVectorLoad(SDValue Op,
         ScalarMemVT = MemVT.getScalarType();
     unsigned Stride = ScalarMemVT.getStoreSize();
 
-    SmallVector<SDValue, 8> Vals, LoadChains;
+    SDValue Vals[4], LoadChains[4];
     for (unsigned Idx = 0; Idx < 4; ++Idx) {
       SDValue Load;
       if (ScalarVT != ScalarMemVT)
-        Load =
-          DAG.getExtLoad(LN->getExtensionType(), dl, ScalarVT, LoadChain,
-                         BasePtr,
-                         LN->getPointerInfo().getWithOffset(Idx*Stride),
-                         ScalarMemVT, LN->isVolatile(), LN->isNonTemporal(),
-                         LN->isInvariant(), MinAlign(Alignment, Idx*Stride),
-                         LN->getAAInfo());
+        Load = DAG.getExtLoad(LN->getExtensionType(), dl, ScalarVT, LoadChain,
+                              BasePtr,
+                              LN->getPointerInfo().getWithOffset(Idx * Stride),
+                              ScalarMemVT, MinAlign(Alignment, Idx * Stride),
+                              LN->getMemOperand()->getFlags(), LN->getAAInfo());
       else
-        Load =
-          DAG.getLoad(ScalarVT, dl, LoadChain, BasePtr,
-                       LN->getPointerInfo().getWithOffset(Idx*Stride),
-                       LN->isVolatile(), LN->isNonTemporal(),
-                       LN->isInvariant(), MinAlign(Alignment, Idx*Stride),
-                       LN->getAAInfo());
+        Load = DAG.getLoad(ScalarVT, dl, LoadChain, BasePtr,
+                           LN->getPointerInfo().getWithOffset(Idx * Stride),
+                           MinAlign(Alignment, Idx * Stride),
+                           LN->getMemOperand()->getFlags(), LN->getAAInfo());
 
       if (Idx == 0 && LN->isIndexed()) {
         assert(LN->getAddressingMode() == ISD::PRE_INC &&
@@ -7688,8 +7941,8 @@ SDValue PPCTargetLowering::LowerVectorLoad(SDValue Op,
                                   LN->getAddressingMode());
       }
 
-      Vals.push_back(Load);
-      LoadChains.push_back(Load.getValue(1));
+      Vals[Idx] = Load;
+      LoadChains[Idx] = Load.getValue(1);
 
       BasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr,
                             DAG.getConstant(Stride, dl,
@@ -7697,8 +7950,7 @@ SDValue PPCTargetLowering::LowerVectorLoad(SDValue Op,
     }
 
     SDValue TF =  DAG.getNode(ISD::TokenFactor, dl, MVT::Other, LoadChains);
-    SDValue Value = DAG.getNode(ISD::BUILD_VECTOR, dl,
-                                Op.getValueType(), Vals);
+    SDValue Value = DAG.getBuildVector(Op.getValueType(), dl, Vals);
 
     if (LN->isIndexed()) {
       SDValue RetOps[] = { Value, Vals[0].getValue(1), TF };
@@ -7715,23 +7967,20 @@ SDValue PPCTargetLowering::LowerVectorLoad(SDValue Op,
   // To lower v4i1 from a byte array, we load the byte elements of the
   // vector and then reuse the BUILD_VECTOR logic.
 
-  SmallVector<SDValue, 4> VectElmts, VectElmtChains;
+  SDValue VectElmts[4], VectElmtChains[4];
   for (unsigned i = 0; i < 4; ++i) {
     SDValue Idx = DAG.getConstant(i, dl, BasePtr.getValueType());
     Idx = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr, Idx);
 
-    VectElmts.push_back(DAG.getExtLoad(ISD::EXTLOAD,
-                        dl, MVT::i32, LoadChain, Idx,
-                        LN->getPointerInfo().getWithOffset(i),
-                        MVT::i8 /* memory type */,
-                        LN->isVolatile(), LN->isNonTemporal(),
-                        LN->isInvariant(),
-                        1 /* alignment */, LN->getAAInfo()));
-    VectElmtChains.push_back(VectElmts[i].getValue(1));
+    VectElmts[i] = DAG.getExtLoad(
+        ISD::EXTLOAD, dl, MVT::i32, LoadChain, Idx,
+        LN->getPointerInfo().getWithOffset(i), MVT::i8,
+        /* Alignment = */ 1, LN->getMemOperand()->getFlags(), LN->getAAInfo());
+    VectElmtChains[i] = VectElmts[i].getValue(1);
   }
 
   LoadChain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, VectElmtChains);
-  SDValue Value = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i1, VectElmts);
+  SDValue Value = DAG.getBuildVector(MVT::v4i1, dl, VectElmts);
 
   SDValue RVals[] = { Value, LoadChain };
   return DAG.getMergeValues(RVals, dl);
@@ -7759,7 +8008,7 @@ SDValue PPCTargetLowering::LowerVectorStore(SDValue Op,
         ScalarMemVT = MemVT.getScalarType();
     unsigned Stride = ScalarMemVT.getStoreSize();
 
-    SmallVector<SDValue, 8> Stores;
+    SDValue Stores[4];
     for (unsigned Idx = 0; Idx < 4; ++Idx) {
       SDValue Ex = DAG.getNode(
           ISD::EXTRACT_VECTOR_ELT, dl, ScalarVT, Value,
@@ -7767,16 +8016,15 @@ SDValue PPCTargetLowering::LowerVectorStore(SDValue Op,
       SDValue Store;
       if (ScalarVT != ScalarMemVT)
         Store =
-          DAG.getTruncStore(StoreChain, dl, Ex, BasePtr,
-                            SN->getPointerInfo().getWithOffset(Idx*Stride),
-                            ScalarMemVT, SN->isVolatile(), SN->isNonTemporal(),
-                            MinAlign(Alignment, Idx*Stride), SN->getAAInfo());
+            DAG.getTruncStore(StoreChain, dl, Ex, BasePtr,
+                              SN->getPointerInfo().getWithOffset(Idx * Stride),
+                              ScalarMemVT, MinAlign(Alignment, Idx * Stride),
+                              SN->getMemOperand()->getFlags(), SN->getAAInfo());
       else
-        Store =
-          DAG.getStore(StoreChain, dl, Ex, BasePtr,
-                       SN->getPointerInfo().getWithOffset(Idx*Stride),
-                       SN->isVolatile(), SN->isNonTemporal(),
-                       MinAlign(Alignment, Idx*Stride), SN->getAAInfo());
+        Store = DAG.getStore(StoreChain, dl, Ex, BasePtr,
+                             SN->getPointerInfo().getWithOffset(Idx * Stride),
+                             MinAlign(Alignment, Idx * Stride),
+                             SN->getMemOperand()->getFlags(), SN->getAAInfo());
 
       if (Idx == 0 && SN->isIndexed()) {
         assert(SN->getAddressingMode() == ISD::PRE_INC &&
@@ -7788,7 +8036,7 @@ SDValue PPCTargetLowering::LowerVectorStore(SDValue Op,
       BasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr,
                             DAG.getConstant(Stride, dl,
                                             BasePtr.getValueType()));
-      Stores.push_back(Store);
+      Stores[Idx] = Store;
     }
 
     SDValue TF =  DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Stores);
@@ -7811,9 +8059,7 @@ SDValue PPCTargetLowering::LowerVectorStore(SDValue Op,
 
   // FIXME: We can make this an f32 vector, but the BUILD_VECTOR code needs to
   // understand how to form the extending load.
-  SDValue FPHalfs = DAG.getConstantFP(0.5, dl, MVT::f64);
-  FPHalfs = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4f64,
-                        FPHalfs, FPHalfs, FPHalfs, FPHalfs);
+  SDValue FPHalfs = DAG.getConstantFP(0.5, dl, MVT::v4f64);
 
   Value = DAG.getNode(ISD::FMA, dl, MVT::v4f64, Value, FPHalfs, FPHalfs);
 
@@ -7829,43 +8075,37 @@ SDValue PPCTargetLowering::LowerVectorStore(SDValue Op,
   EVT PtrVT = getPointerTy(DAG.getDataLayout());
   SDValue FIdx = DAG.getFrameIndex(FrameIdx, PtrVT);
 
-  SmallVector<SDValue, 2> Ops;
-  Ops.push_back(StoreChain);
-  Ops.push_back(DAG.getConstant(Intrinsic::ppc_qpx_qvstfiw, dl, MVT::i32));
-  Ops.push_back(Value);
-  Ops.push_back(FIdx);
-
-  SmallVector<EVT, 2> ValueVTs;
-  ValueVTs.push_back(MVT::Other); // chain
-  SDVTList VTs = DAG.getVTList(ValueVTs);
+  SDValue Ops[] = {StoreChain,
+                   DAG.getConstant(Intrinsic::ppc_qpx_qvstfiw, dl, MVT::i32),
+                   Value, FIdx};
+  SDVTList VTs = DAG.getVTList(/*chain*/ MVT::Other);
 
   StoreChain = DAG.getMemIntrinsicNode(ISD::INTRINSIC_VOID,
     dl, VTs, Ops, MVT::v4i32, PtrInfo);
 
   // Move data into the byte array.
-  SmallVector<SDValue, 4> Loads, LoadChains;
+  SDValue Loads[4], LoadChains[4];
   for (unsigned i = 0; i < 4; ++i) {
     unsigned Offset = 4*i;
     SDValue Idx = DAG.getConstant(Offset, dl, FIdx.getValueType());
     Idx = DAG.getNode(ISD::ADD, dl, FIdx.getValueType(), FIdx, Idx);
 
-    Loads.push_back(DAG.getLoad(MVT::i32, dl, StoreChain, Idx,
-                                   PtrInfo.getWithOffset(Offset),
-                                   false, false, false, 0));
-    LoadChains.push_back(Loads[i].getValue(1));
+    Loads[i] = DAG.getLoad(MVT::i32, dl, StoreChain, Idx,
+                           PtrInfo.getWithOffset(Offset));
+    LoadChains[i] = Loads[i].getValue(1);
   }
 
   StoreChain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, LoadChains);
 
-  SmallVector<SDValue, 4> Stores;
+  SDValue Stores[4];
   for (unsigned i = 0; i < 4; ++i) {
     SDValue Idx = DAG.getConstant(i, dl, BasePtr.getValueType());
     Idx = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr, Idx);
 
-    Stores.push_back(DAG.getTruncStore(
+    Stores[i] = DAG.getTruncStore(
         StoreChain, dl, Loads[i], Idx, SN->getPointerInfo().getWithOffset(i),
-        MVT::i8 /* memory type */, SN->isNonTemporal(), SN->isVolatile(),
-        1 /* alignment */, SN->getAAInfo()));
+        MVT::i8, /* Alignment = */ 1, SN->getMemOperand()->getFlags(),
+        SN->getAAInfo());
   }
 
   StoreChain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Stores);
@@ -7958,18 +8198,22 @@ SDValue PPCTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::INIT_TRAMPOLINE:    return LowerINIT_TRAMPOLINE(Op, DAG);
   case ISD::ADJUST_TRAMPOLINE:  return LowerADJUST_TRAMPOLINE(Op, DAG);
   case ISD::VASTART:
-    return LowerVASTART(Op, DAG, Subtarget);
+    return LowerVASTART(Op, DAG);
 
   case ISD::VAARG:
-    return LowerVAARG(Op, DAG, Subtarget);
+    return LowerVAARG(Op, DAG);
 
   case ISD::VACOPY:
-    return LowerVACOPY(Op, DAG, Subtarget);
+    return LowerVACOPY(Op, DAG);
+
+  case ISD::STACKRESTORE:
+    return LowerSTACKRESTORE(Op, DAG);
 
-  case ISD::STACKRESTORE:       return LowerSTACKRESTORE(Op, DAG, Subtarget);
   case ISD::DYNAMIC_STACKALLOC:
-    return LowerDYNAMIC_STACKALLOC(Op, DAG, Subtarget);
-  case ISD::GET_DYNAMIC_AREA_OFFSET: return LowerGET_DYNAMIC_AREA_OFFSET(Op, DAG, Subtarget);
+    return LowerDYNAMIC_STACKALLOC(Op, DAG);
+
+  case ISD::GET_DYNAMIC_AREA_OFFSET:
+    return LowerGET_DYNAMIC_AREA_OFFSET(Op, DAG);
 
   case ISD::EH_SJLJ_SETJMP:     return lowerEH_SJLJ_SETJMP(Op, DAG);
   case ISD::EH_SJLJ_LONGJMP:    return lowerEH_SJLJ_LONGJMP(Op, DAG);
@@ -8048,7 +8292,7 @@ void PPCTargetLowering::ReplaceNodeResults(SDNode *N,
     EVT VT = N->getValueType(0);
 
     if (VT == MVT::i64) {
-      SDValue NewNode = LowerVAARG(SDValue(N, 1), DAG, Subtarget);
+      SDValue NewNode = LowerVAARG(SDValue(N, 1), DAG);
 
       Results.push_back(NewNode);
       Results.push_back(NewNode.getValue(1));
@@ -8099,9 +8343,9 @@ static Instruction* callIntrinsic(IRBuilder<> &Builder, Intrinsic::ID Id) {
 Instruction* PPCTargetLowering::emitLeadingFence(IRBuilder<> &Builder,
                                          AtomicOrdering Ord, bool IsStore,
                                          bool IsLoad) const {
-  if (Ord == SequentiallyConsistent)
+  if (Ord == AtomicOrdering::SequentiallyConsistent)
     return callIntrinsic(Builder, Intrinsic::ppc_sync);
-  if (isAtLeastRelease(Ord))
+  if (isReleaseOrStronger(Ord))
     return callIntrinsic(Builder, Intrinsic::ppc_lwsync);
   return nullptr;
 }
@@ -8109,7 +8353,7 @@ Instruction* PPCTargetLowering::emitLeadingFence(IRBuilder<> &Builder,
 Instruction* PPCTargetLowering::emitTrailingFence(IRBuilder<> &Builder,
                                           AtomicOrdering Ord, bool IsStore,
                                           bool IsLoad) const {
-  if (IsLoad && isAtLeastAcquire(Ord))
+  if (IsLoad && isAcquireOrStronger(Ord))
     return callIntrinsic(Builder, Intrinsic::ppc_lwsync);
   // FIXME: this is too conservative, a dependent branch + isync is enough.
   // See http://www.cl.cam.ac.uk/~pes20/cpp/cpp0xmappings.html and
@@ -8119,7 +8363,7 @@ Instruction* PPCTargetLowering::emitTrailingFence(IRBuilder<> &Builder,
 }
 
 MachineBasicBlock *
-PPCTargetLowering::EmitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
+PPCTargetLowering::EmitAtomicBinary(MachineInstr &MI, MachineBasicBlock *BB,
                                     unsigned AtomicSize,
                                     unsigned BinOpcode) const {
   // This also handles ATOMIC_SWAP, indicated by BinOpcode==0.
@@ -8154,11 +8398,11 @@ PPCTargetLowering::EmitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
   MachineFunction *F = BB->getParent();
   MachineFunction::iterator It = ++BB->getIterator();
 
-  unsigned dest = MI->getOperand(0).getReg();
-  unsigned ptrA = MI->getOperand(1).getReg();
-  unsigned ptrB = MI->getOperand(2).getReg();
-  unsigned incr = MI->getOperand(3).getReg();
-  DebugLoc dl = MI->getDebugLoc();
+  unsigned dest = MI.getOperand(0).getReg();
+  unsigned ptrA = MI.getOperand(1).getReg();
+  unsigned ptrB = MI.getOperand(2).getReg();
+  unsigned incr = MI.getOperand(3).getReg();
+  DebugLoc dl = MI.getDebugLoc();
 
   MachineBasicBlock *loopMBB = F->CreateMachineBasicBlock(LLVM_BB);
   MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
@@ -8203,9 +8447,9 @@ PPCTargetLowering::EmitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
 }
 
 MachineBasicBlock *
-PPCTargetLowering::EmitPartwordAtomicBinary(MachineInstr *MI,
+PPCTargetLowering::EmitPartwordAtomicBinary(MachineInstr &MI,
                                             MachineBasicBlock *BB,
-                                            bool is8bit,    // operation
+                                            bool is8bit, // operation
                                             unsigned BinOpcode) const {
   // If we support part-word atomic mnemonics, just use them
   if (Subtarget.hasPartwordAtomics())
@@ -8224,11 +8468,11 @@ PPCTargetLowering::EmitPartwordAtomicBinary(MachineInstr *MI,
   MachineFunction *F = BB->getParent();
   MachineFunction::iterator It = ++BB->getIterator();
 
-  unsigned dest = MI->getOperand(0).getReg();
-  unsigned ptrA = MI->getOperand(1).getReg();
-  unsigned ptrB = MI->getOperand(2).getReg();
-  unsigned incr = MI->getOperand(3).getReg();
-  DebugLoc dl = MI->getDebugLoc();
+  unsigned dest = MI.getOperand(0).getReg();
+  unsigned ptrA = MI.getOperand(1).getReg();
+  unsigned ptrB = MI.getOperand(2).getReg();
+  unsigned incr = MI.getOperand(3).getReg();
+  DebugLoc dl = MI.getDebugLoc();
 
   MachineBasicBlock *loopMBB = F->CreateMachineBasicBlock(LLVM_BB);
   MachineBasicBlock *exitMBB = F->CreateMachineBasicBlock(LLVM_BB);
@@ -8334,10 +8578,10 @@ PPCTargetLowering::EmitPartwordAtomicBinary(MachineInstr *MI,
   return BB;
 }
 
-llvm::MachineBasicBlock*
-PPCTargetLowering::emitEHSjLjSetJmp(MachineInstr *MI,
+llvm::MachineBasicBlock *
+PPCTargetLowering::emitEHSjLjSetJmp(MachineInstr &MI,
                                     MachineBasicBlock *MBB) const {
-  DebugLoc DL = MI->getDebugLoc();
+  DebugLoc DL = MI.getDebugLoc();
   const TargetInstrInfo *TII = Subtarget.getInstrInfo();
 
   MachineFunction *MF = MBB->getParent();
@@ -8347,10 +8591,10 @@ PPCTargetLowering::emitEHSjLjSetJmp(MachineInstr *MI,
   MachineFunction::iterator I = ++MBB->getIterator();
 
   // Memory Reference
-  MachineInstr::mmo_iterator MMOBegin = MI->memoperands_begin();
-  MachineInstr::mmo_iterator MMOEnd = MI->memoperands_end();
+  MachineInstr::mmo_iterator MMOBegin = MI.memoperands_begin();
+  MachineInstr::mmo_iterator MMOEnd = MI.memoperands_end();
 
-  unsigned DstReg = MI->getOperand(0).getReg();
+  unsigned DstReg = MI.getOperand(0).getReg();
   const TargetRegisterClass *RC = MRI.getRegClass(DstReg);
   assert(RC->hasType(MVT::i32) && "Invalid destination!");
   unsigned mainDstReg = MRI.createVirtualRegister(RC);
@@ -8407,7 +8651,7 @@ PPCTargetLowering::emitEHSjLjSetJmp(MachineInstr *MI,
   // Prepare IP either in reg.
   const TargetRegisterClass *PtrRC = getRegClassFor(PVT);
   unsigned LabelReg = MRI.createVirtualRegister(PtrRC);
-  unsigned BufReg = MI->getOperand(1).getReg();
+  unsigned BufReg = MI.getOperand(1).getReg();
 
   if (Subtarget.isPPC64() && Subtarget.isSVR4ABI()) {
     setUsesTOCBasePtr(*MBB->getParent());
@@ -8477,22 +8721,22 @@ PPCTargetLowering::emitEHSjLjSetJmp(MachineInstr *MI,
     .addReg(mainDstReg).addMBB(mainMBB)
     .addReg(restoreDstReg).addMBB(thisMBB);
 
-  MI->eraseFromParent();
+  MI.eraseFromParent();
   return sinkMBB;
 }
 
 MachineBasicBlock *
-PPCTargetLowering::emitEHSjLjLongJmp(MachineInstr *MI,
+PPCTargetLowering::emitEHSjLjLongJmp(MachineInstr &MI,
                                      MachineBasicBlock *MBB) const {
-  DebugLoc DL = MI->getDebugLoc();
+  DebugLoc DL = MI.getDebugLoc();
   const TargetInstrInfo *TII = Subtarget.getInstrInfo();
 
   MachineFunction *MF = MBB->getParent();
   MachineRegisterInfo &MRI = MF->getRegInfo();
 
   // Memory Reference
-  MachineInstr::mmo_iterator MMOBegin = MI->memoperands_begin();
-  MachineInstr::mmo_iterator MMOEnd = MI->memoperands_end();
+  MachineInstr::mmo_iterator MMOBegin = MI.memoperands_begin();
+  MachineInstr::mmo_iterator MMOEnd = MI.memoperands_end();
 
   MVT PVT = getPointerTy(MF->getDataLayout());
   assert((PVT == MVT::i64 || PVT == MVT::i32) &&
@@ -8507,10 +8751,8 @@ PPCTargetLowering::emitEHSjLjLongJmp(MachineInstr *MI,
   unsigned BP =
       (PVT == MVT::i64)
           ? PPC::X30
-          : (Subtarget.isSVR4ABI() &&
-                     MF->getTarget().getRelocationModel() == Reloc::PIC_
-                 ? PPC::R29
-                 : PPC::R30);
+          : (Subtarget.isSVR4ABI() && isPositionIndependent() ? PPC::R29
+                                                              : PPC::R30);
 
   MachineInstrBuilder MIB;
 
@@ -8519,7 +8761,7 @@ PPCTargetLowering::emitEHSjLjLongJmp(MachineInstr *MI,
   const int64_t TOCOffset   = 3 * PVT.getStoreSize();
   const int64_t BPOffset    = 4 * PVT.getStoreSize();
 
-  unsigned BufReg = MI->getOperand(0).getReg();
+  unsigned BufReg = MI.getOperand(0).getReg();
 
   // Reload FP (the jumped-to function may not have had a
   // frame pointer, and if so, then its r31 will be restored
@@ -8586,34 +8828,34 @@ PPCTargetLowering::emitEHSjLjLongJmp(MachineInstr *MI,
           TII->get(PVT == MVT::i64 ? PPC::MTCTR8 : PPC::MTCTR)).addReg(Tmp);
   BuildMI(*MBB, MI, DL, TII->get(PVT == MVT::i64 ? PPC::BCTR8 : PPC::BCTR));
 
-  MI->eraseFromParent();
+  MI.eraseFromParent();
   return MBB;
 }
 
 MachineBasicBlock *
-PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
+PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
                                                MachineBasicBlock *BB) const {
-  if (MI->getOpcode() == TargetOpcode::STACKMAP ||
-      MI->getOpcode() == TargetOpcode::PATCHPOINT) {
+  if (MI.getOpcode() == TargetOpcode::STACKMAP ||
+      MI.getOpcode() == TargetOpcode::PATCHPOINT) {
     if (Subtarget.isPPC64() && Subtarget.isSVR4ABI() &&
-        MI->getOpcode() == TargetOpcode::PATCHPOINT) {
+        MI.getOpcode() == TargetOpcode::PATCHPOINT) {
       // Call lowering should have added an r2 operand to indicate a dependence
       // on the TOC base pointer value. It can't however, because there is no
       // way to mark the dependence as implicit there, and so the stackmap code
       // will confuse it with a regular operand. Instead, add the dependence
       // here.
       setUsesTOCBasePtr(*BB->getParent());
-      MI->addOperand(MachineOperand::CreateReg(PPC::X2, false, true));
+      MI.addOperand(MachineOperand::CreateReg(PPC::X2, false, true));
     }
 
     return emitPatchPoint(MI, BB);
   }
 
-  if (MI->getOpcode() == PPC::EH_SjLj_SetJmp32 ||
-      MI->getOpcode() == PPC::EH_SjLj_SetJmp64) {
+  if (MI.getOpcode() == PPC::EH_SjLj_SetJmp32 ||
+      MI.getOpcode() == PPC::EH_SjLj_SetJmp64) {
     return emitEHSjLjSetJmp(MI, BB);
-  } else if (MI->getOpcode() == PPC::EH_SjLj_LongJmp32 ||
-             MI->getOpcode() == PPC::EH_SjLj_LongJmp64) {
+  } else if (MI.getOpcode() == PPC::EH_SjLj_LongJmp32 ||
+             MI.getOpcode() == PPC::EH_SjLj_LongJmp64) {
     return emitEHSjLjLongJmp(MI, BB);
   }
 
@@ -8626,44 +8868,43 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
   MachineFunction *F = BB->getParent();
 
-  if (Subtarget.hasISEL() && (MI->getOpcode() == PPC::SELECT_CC_I4 ||
-                              MI->getOpcode() == PPC::SELECT_CC_I8 ||
-                              MI->getOpcode() == PPC::SELECT_I4 ||
-                              MI->getOpcode() == PPC::SELECT_I8)) {
+  if (Subtarget.hasISEL() &&
+      (MI.getOpcode() == PPC::SELECT_CC_I4 ||
+       MI.getOpcode() == PPC::SELECT_CC_I8 ||
+       MI.getOpcode() == PPC::SELECT_I4 || MI.getOpcode() == PPC::SELECT_I8)) {
     SmallVector<MachineOperand, 2> Cond;
-    if (MI->getOpcode() == PPC::SELECT_CC_I4 ||
-        MI->getOpcode() == PPC::SELECT_CC_I8)
-      Cond.push_back(MI->getOperand(4));
+    if (MI.getOpcode() == PPC::SELECT_CC_I4 ||
+        MI.getOpcode() == PPC::SELECT_CC_I8)
+      Cond.push_back(MI.getOperand(4));
     else
       Cond.push_back(MachineOperand::CreateImm(PPC::PRED_BIT_SET));
-    Cond.push_back(MI->getOperand(1));
-
-    DebugLoc dl = MI->getDebugLoc();
-    TII->insertSelect(*BB, MI, dl, MI->getOperand(0).getReg(),
-                      Cond, MI->getOperand(2).getReg(),
-                      MI->getOperand(3).getReg());
-  } else if (MI->getOpcode() == PPC::SELECT_CC_I4 ||
-             MI->getOpcode() == PPC::SELECT_CC_I8 ||
-             MI->getOpcode() == PPC::SELECT_CC_F4 ||
-             MI->getOpcode() == PPC::SELECT_CC_F8 ||
-             MI->getOpcode() == PPC::SELECT_CC_QFRC ||
-             MI->getOpcode() == PPC::SELECT_CC_QSRC ||
-             MI->getOpcode() == PPC::SELECT_CC_QBRC ||
-             MI->getOpcode() == PPC::SELECT_CC_VRRC ||
-             MI->getOpcode() == PPC::SELECT_CC_VSFRC ||
-             MI->getOpcode() == PPC::SELECT_CC_VSSRC ||
-             MI->getOpcode() == PPC::SELECT_CC_VSRC ||
-             MI->getOpcode() == PPC::SELECT_I4 ||
-             MI->getOpcode() == PPC::SELECT_I8 ||
-             MI->getOpcode() == PPC::SELECT_F4 ||
-             MI->getOpcode() == PPC::SELECT_F8 ||
-             MI->getOpcode() == PPC::SELECT_QFRC ||
-             MI->getOpcode() == PPC::SELECT_QSRC ||
-             MI->getOpcode() == PPC::SELECT_QBRC ||
-             MI->getOpcode() == PPC::SELECT_VRRC ||
-             MI->getOpcode() == PPC::SELECT_VSFRC ||
-             MI->getOpcode() == PPC::SELECT_VSSRC ||
-             MI->getOpcode() == PPC::SELECT_VSRC) {
+    Cond.push_back(MI.getOperand(1));
+
+    DebugLoc dl = MI.getDebugLoc();
+    TII->insertSelect(*BB, MI, dl, MI.getOperand(0).getReg(), Cond,
+                      MI.getOperand(2).getReg(), MI.getOperand(3).getReg());
+  } else if (MI.getOpcode() == PPC::SELECT_CC_I4 ||
+             MI.getOpcode() == PPC::SELECT_CC_I8 ||
+             MI.getOpcode() == PPC::SELECT_CC_F4 ||
+             MI.getOpcode() == PPC::SELECT_CC_F8 ||
+             MI.getOpcode() == PPC::SELECT_CC_QFRC ||
+             MI.getOpcode() == PPC::SELECT_CC_QSRC ||
+             MI.getOpcode() == PPC::SELECT_CC_QBRC ||
+             MI.getOpcode() == PPC::SELECT_CC_VRRC ||
+             MI.getOpcode() == PPC::SELECT_CC_VSFRC ||
+             MI.getOpcode() == PPC::SELECT_CC_VSSRC ||
+             MI.getOpcode() == PPC::SELECT_CC_VSRC ||
+             MI.getOpcode() == PPC::SELECT_I4 ||
+             MI.getOpcode() == PPC::SELECT_I8 ||
+             MI.getOpcode() == PPC::SELECT_F4 ||
+             MI.getOpcode() == PPC::SELECT_F8 ||
+             MI.getOpcode() == PPC::SELECT_QFRC ||
+             MI.getOpcode() == PPC::SELECT_QSRC ||
+             MI.getOpcode() == PPC::SELECT_QBRC ||
+             MI.getOpcode() == PPC::SELECT_VRRC ||
+             MI.getOpcode() == PPC::SELECT_VSFRC ||
+             MI.getOpcode() == PPC::SELECT_VSSRC ||
+             MI.getOpcode() == PPC::SELECT_VSRC) {
     // The incoming instruction knows the destination vreg to set, the
     // condition code register to branch on, the true/false values to
     // select between, and a branch opcode to use.
@@ -8677,7 +8918,7 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     MachineBasicBlock *thisMBB = BB;
     MachineBasicBlock *copy0MBB = F->CreateMachineBasicBlock(LLVM_BB);
     MachineBasicBlock *sinkMBB = F->CreateMachineBasicBlock(LLVM_BB);
-    DebugLoc dl = MI->getDebugLoc();
+    DebugLoc dl = MI.getDebugLoc();
     F->insert(It, copy0MBB);
     F->insert(It, sinkMBB);
 
@@ -8690,23 +8931,24 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     BB->addSuccessor(copy0MBB);
     BB->addSuccessor(sinkMBB);
 
-    if (MI->getOpcode() == PPC::SELECT_I4 ||
-        MI->getOpcode() == PPC::SELECT_I8 ||
-        MI->getOpcode() == PPC::SELECT_F4 ||
-        MI->getOpcode() == PPC::SELECT_F8 ||
-        MI->getOpcode() == PPC::SELECT_QFRC ||
-        MI->getOpcode() == PPC::SELECT_QSRC ||
-        MI->getOpcode() == PPC::SELECT_QBRC ||
-        MI->getOpcode() == PPC::SELECT_VRRC ||
-        MI->getOpcode() == PPC::SELECT_VSFRC ||
-        MI->getOpcode() == PPC::SELECT_VSSRC ||
-        MI->getOpcode() == PPC::SELECT_VSRC) {
+    if (MI.getOpcode() == PPC::SELECT_I4 || MI.getOpcode() == PPC::SELECT_I8 ||
+        MI.getOpcode() == PPC::SELECT_F4 || MI.getOpcode() == PPC::SELECT_F8 ||
+        MI.getOpcode() == PPC::SELECT_QFRC ||
+        MI.getOpcode() == PPC::SELECT_QSRC ||
+        MI.getOpcode() == PPC::SELECT_QBRC ||
+        MI.getOpcode() == PPC::SELECT_VRRC ||
+        MI.getOpcode() == PPC::SELECT_VSFRC ||
+        MI.getOpcode() == PPC::SELECT_VSSRC ||
+        MI.getOpcode() == PPC::SELECT_VSRC) {
       BuildMI(BB, dl, TII->get(PPC::BC))
-        .addReg(MI->getOperand(1).getReg()).addMBB(sinkMBB);
+          .addReg(MI.getOperand(1).getReg())
+          .addMBB(sinkMBB);
     } else {
-      unsigned SelectPred = MI->getOperand(4).getImm();
+      unsigned SelectPred = MI.getOperand(4).getImm();
       BuildMI(BB, dl, TII->get(PPC::BCC))
-        .addImm(SelectPred).addReg(MI->getOperand(1).getReg()).addMBB(sinkMBB);
+          .addImm(SelectPred)
+          .addReg(MI.getOperand(1).getReg())
+          .addMBB(sinkMBB);
     }
 
     //  copy0MBB:
@@ -8721,11 +8963,12 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     //   %Result = phi [ %FalseValue, copy0MBB ], [ %TrueValue, thisMBB ]
     //  ...
     BB = sinkMBB;
-    BuildMI(*BB, BB->begin(), dl,
-            TII->get(PPC::PHI), MI->getOperand(0).getReg())
-      .addReg(MI->getOperand(3).getReg()).addMBB(copy0MBB)
-      .addReg(MI->getOperand(2).getReg()).addMBB(thisMBB);
-  } else if (MI->getOpcode() == PPC::ReadTB) {
+    BuildMI(*BB, BB->begin(), dl, TII->get(PPC::PHI), MI.getOperand(0).getReg())
+        .addReg(MI.getOperand(3).getReg())
+        .addMBB(copy0MBB)
+        .addReg(MI.getOperand(2).getReg())
+        .addMBB(thisMBB);
+  } else if (MI.getOpcode() == PPC::ReadTB) {
     // To read the 64-bit time-base register on a 32-bit target, we read the
     // two halves. Should the counter have wrapped while it was being read, we
     // need to try again.
@@ -8740,7 +8983,7 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     MachineBasicBlock *readMBB = F->CreateMachineBasicBlock(LLVM_BB);
     MachineBasicBlock *sinkMBB = F->CreateMachineBasicBlock(LLVM_BB);
-    DebugLoc dl = MI->getDebugLoc();
+    DebugLoc dl = MI.getDebugLoc();
     F->insert(It, readMBB);
     F->insert(It, sinkMBB);
 
@@ -8754,8 +8997,8 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     MachineRegisterInfo &RegInfo = F->getRegInfo();
     unsigned ReadAgainReg = RegInfo.createVirtualRegister(&PPC::GPRCRegClass);
-    unsigned LoReg = MI->getOperand(0).getReg();
-    unsigned HiReg = MI->getOperand(1).getReg();
+    unsigned LoReg = MI.getOperand(0).getReg();
+    unsigned HiReg = MI.getOperand(1).getReg();
 
     BuildMI(BB, dl, TII->get(PPC::MFSPR), HiReg).addImm(269);
     BuildMI(BB, dl, TII->get(PPC::MFSPR), LoReg).addImm(268);
@@ -8770,81 +9013,80 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     BB->addSuccessor(readMBB);
     BB->addSuccessor(sinkMBB);
-  }
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I8)
+  } else if (MI.getOpcode() == PPC::ATOMIC_LOAD_ADD_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::ADD4);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_ADD_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::ADD4);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_ADD_I32)
     BB = EmitAtomicBinary(MI, BB, 4, PPC::ADD4);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_ADD_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_ADD_I64)
     BB = EmitAtomicBinary(MI, BB, 8, PPC::ADD8);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I8)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_AND_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::AND);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_AND_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::AND);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_AND_I32)
     BB = EmitAtomicBinary(MI, BB, 4, PPC::AND);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_AND_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_AND_I64)
     BB = EmitAtomicBinary(MI, BB, 8, PPC::AND8);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I8)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_OR_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::OR);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_OR_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::OR);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_OR_I32)
     BB = EmitAtomicBinary(MI, BB, 4, PPC::OR);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_OR_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_OR_I64)
     BB = EmitAtomicBinary(MI, BB, 8, PPC::OR8);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I8)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_XOR_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::XOR);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_XOR_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::XOR);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_XOR_I32)
     BB = EmitAtomicBinary(MI, BB, 4, PPC::XOR);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_XOR_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_XOR_I64)
     BB = EmitAtomicBinary(MI, BB, 8, PPC::XOR8);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I8)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_NAND_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::NAND);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_NAND_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::NAND);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_NAND_I32)
     BB = EmitAtomicBinary(MI, BB, 4, PPC::NAND);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_NAND_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_NAND_I64)
     BB = EmitAtomicBinary(MI, BB, 8, PPC::NAND8);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I8)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_SUB_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, PPC::SUBF);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_SUB_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, PPC::SUBF);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_SUB_I32)
     BB = EmitAtomicBinary(MI, BB, 4, PPC::SUBF);
-  else if (MI->getOpcode() == PPC::ATOMIC_LOAD_SUB_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_LOAD_SUB_I64)
     BB = EmitAtomicBinary(MI, BB, 8, PPC::SUBF8);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I8)
+  else if (MI.getOpcode() == PPC::ATOMIC_SWAP_I8)
     BB = EmitPartwordAtomicBinary(MI, BB, true, 0);
-  else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I16)
+  else if (MI.getOpcode() == PPC::ATOMIC_SWAP_I16)
     BB = EmitPartwordAtomicBinary(MI, BB, false, 0);
-  else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I32)
+  else if (MI.getOpcode() == PPC::ATOMIC_SWAP_I32)
     BB = EmitAtomicBinary(MI, BB, 4, 0);
-  else if (MI->getOpcode() == PPC::ATOMIC_SWAP_I64)
+  else if (MI.getOpcode() == PPC::ATOMIC_SWAP_I64)
     BB = EmitAtomicBinary(MI, BB, 8, 0);
 
-  else if (MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I32 ||
-           MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I64 ||
+  else if (MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I32 ||
+           MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I64 ||
            (Subtarget.hasPartwordAtomics() &&
-            MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I8) ||
+            MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I8) ||
            (Subtarget.hasPartwordAtomics() &&
-            MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I16)) {
-    bool is64bit = MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I64;
+            MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I16)) {
+    bool is64bit = MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I64;
 
     auto LoadMnemonic = PPC::LDARX;
     auto StoreMnemonic = PPC::STDCX;
-    switch(MI->getOpcode()) {
+    switch (MI.getOpcode()) {
     default:
       llvm_unreachable("Compare and swap of unknown size");
     case PPC::ATOMIC_CMP_SWAP_I8:
@@ -8866,12 +9108,12 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
       StoreMnemonic = PPC::STDCX;
       break;
     }
-    unsigned dest   = MI->getOperand(0).getReg();
-    unsigned ptrA   = MI->getOperand(1).getReg();
-    unsigned ptrB   = MI->getOperand(2).getReg();
-    unsigned oldval = MI->getOperand(3).getReg();
-    unsigned newval = MI->getOperand(4).getReg();
-    DebugLoc dl     = MI->getDebugLoc();
+    unsigned dest = MI.getOperand(0).getReg();
+    unsigned ptrA = MI.getOperand(1).getReg();
+    unsigned ptrB = MI.getOperand(2).getReg();
+    unsigned oldval = MI.getOperand(3).getReg();
+    unsigned newval = MI.getOperand(4).getReg();
+    DebugLoc dl = MI.getDebugLoc();
 
     MachineBasicBlock *loop1MBB = F->CreateMachineBasicBlock(LLVM_BB);
     MachineBasicBlock *loop2MBB = F->CreateMachineBasicBlock(LLVM_BB);
@@ -8928,20 +9170,20 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     //  exitMBB:
     //   ...
     BB = exitMBB;
-  } else if (MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I8 ||
-             MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I16) {
+  } else if (MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I8 ||
+             MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I16) {
     // We must use 64-bit registers for addresses when targeting 64-bit,
     // since we're actually doing arithmetic on them.  Other registers
     // can be 32-bit.
     bool is64bit = Subtarget.isPPC64();
-    bool is8bit = MI->getOpcode() == PPC::ATOMIC_CMP_SWAP_I8;
+    bool is8bit = MI.getOpcode() == PPC::ATOMIC_CMP_SWAP_I8;
 
-    unsigned dest   = MI->getOperand(0).getReg();
-    unsigned ptrA   = MI->getOperand(1).getReg();
-    unsigned ptrB   = MI->getOperand(2).getReg();
-    unsigned oldval = MI->getOperand(3).getReg();
-    unsigned newval = MI->getOperand(4).getReg();
-    DebugLoc dl     = MI->getDebugLoc();
+    unsigned dest = MI.getOperand(0).getReg();
+    unsigned ptrA = MI.getOperand(1).getReg();
+    unsigned ptrB = MI.getOperand(2).getReg();
+    unsigned oldval = MI.getOperand(3).getReg();
+    unsigned newval = MI.getOperand(4).getReg();
+    DebugLoc dl = MI.getDebugLoc();
 
     MachineBasicBlock *loop1MBB = F->CreateMachineBasicBlock(LLVM_BB);
     MachineBasicBlock *loop2MBB = F->CreateMachineBasicBlock(LLVM_BB);
@@ -9076,14 +9318,14 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     BB = exitMBB;
     BuildMI(*BB, BB->begin(), dl, TII->get(PPC::SRW),dest).addReg(TmpReg)
       .addReg(ShiftReg);
-  } else if (MI->getOpcode() == PPC::FADDrtz) {
+  } else if (MI.getOpcode() == PPC::FADDrtz) {
     // This pseudo performs an FADD with rounding mode temporarily forced
     // to round-to-zero.  We emit this via custom inserter since the FPSCR
     // is not modeled at the SelectionDAG level.
-    unsigned Dest = MI->getOperand(0).getReg();
-    unsigned Src1 = MI->getOperand(1).getReg();
-    unsigned Src2 = MI->getOperand(2).getReg();
-    DebugLoc dl   = MI->getDebugLoc();
+    unsigned Dest = MI.getOperand(0).getReg();
+    unsigned Src1 = MI.getOperand(1).getReg();
+    unsigned Src2 = MI.getOperand(2).getReg();
+    DebugLoc dl = MI.getDebugLoc();
 
     MachineRegisterInfo &RegInfo = F->getRegInfo();
     unsigned MFFSReg = RegInfo.createVirtualRegister(&PPC::F8RCRegClass);
@@ -9100,29 +9342,31 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
 
     // Restore FPSCR value.
     BuildMI(*BB, MI, dl, TII->get(PPC::MTFSFb)).addImm(1).addReg(MFFSReg);
-  } else if (MI->getOpcode() == PPC::ANDIo_1_EQ_BIT ||
-             MI->getOpcode() == PPC::ANDIo_1_GT_BIT ||
-             MI->getOpcode() == PPC::ANDIo_1_EQ_BIT8 ||
-             MI->getOpcode() == PPC::ANDIo_1_GT_BIT8) {
-    unsigned Opcode = (MI->getOpcode() == PPC::ANDIo_1_EQ_BIT8 ||
-                       MI->getOpcode() == PPC::ANDIo_1_GT_BIT8) ?
-                      PPC::ANDIo8 : PPC::ANDIo;
-    bool isEQ = (MI->getOpcode() == PPC::ANDIo_1_EQ_BIT ||
-                 MI->getOpcode() == PPC::ANDIo_1_EQ_BIT8);
+  } else if (MI.getOpcode() == PPC::ANDIo_1_EQ_BIT ||
+             MI.getOpcode() == PPC::ANDIo_1_GT_BIT ||
+             MI.getOpcode() == PPC::ANDIo_1_EQ_BIT8 ||
+             MI.getOpcode() == PPC::ANDIo_1_GT_BIT8) {
+    unsigned Opcode = (MI.getOpcode() == PPC::ANDIo_1_EQ_BIT8 ||
+                       MI.getOpcode() == PPC::ANDIo_1_GT_BIT8)
+                          ? PPC::ANDIo8
+                          : PPC::ANDIo;
+    bool isEQ = (MI.getOpcode() == PPC::ANDIo_1_EQ_BIT ||
+                 MI.getOpcode() == PPC::ANDIo_1_EQ_BIT8);
 
     MachineRegisterInfo &RegInfo = F->getRegInfo();
     unsigned Dest = RegInfo.createVirtualRegister(Opcode == PPC::ANDIo ?
                                                   &PPC::GPRCRegClass :
                                                   &PPC::G8RCRegClass);
 
-    DebugLoc dl   = MI->getDebugLoc();
+    DebugLoc dl = MI.getDebugLoc();
     BuildMI(*BB, MI, dl, TII->get(Opcode), Dest)
-      .addReg(MI->getOperand(1).getReg()).addImm(1);
+        .addReg(MI.getOperand(1).getReg())
+        .addImm(1);
     BuildMI(*BB, MI, dl, TII->get(TargetOpcode::COPY),
-            MI->getOperand(0).getReg())
-      .addReg(isEQ ? PPC::CR0EQ : PPC::CR0GT);
-  } else if (MI->getOpcode() == PPC::TCHECK_RET) {
-    DebugLoc Dl = MI->getDebugLoc();
+            MI.getOperand(0).getReg())
+        .addReg(isEQ ? PPC::CR0EQ : PPC::CR0GT);
+  } else if (MI.getOpcode() == PPC::TCHECK_RET) {
+    DebugLoc Dl = MI.getDebugLoc();
     MachineRegisterInfo &RegInfo = F->getRegInfo();
     unsigned CRReg = RegInfo.createVirtualRegister(&PPC::CRRCRegClass);
     BuildMI(*BB, MI, Dl, TII->get(PPC::TCHECK), CRReg);
@@ -9131,7 +9375,7 @@ PPCTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI,
     llvm_unreachable("Unexpected instr type to insert");
   }
 
-  MI->eraseFromParent();   // The pseudo instruction is gone now.
+  MI.eraseFromParent(); // The pseudo instruction is gone now.
   return BB;
 }
 
@@ -9650,14 +9894,18 @@ SDValue PPCTargetLowering::DAGCombineTruncBoolExt(SDNode *N,
       DAG.ReplaceAllUsesOfValueWith(Inputs[i], Inputs[i].getOperand(0));
   }
 
+  std::list<HandleSDNode> PromOpHandles;
+  for (auto &PromOp : PromOps)
+    PromOpHandles.emplace_back(PromOp);
+
   // Replace all operations (these are all the same, but have a different
   // (i1) return type). DAG.getNode will validate that the types of
   // a binary operator match, so go through the list in reverse so that
   // we've likely promoted both operands first. Any intermediate truncations or
   // extensions disappear.
-  while (!PromOps.empty()) {
-    SDValue PromOp = PromOps.back();
-    PromOps.pop_back();
+  while (!PromOpHandles.empty()) {
+    SDValue PromOp = PromOpHandles.back().getValue();
+    PromOpHandles.pop_back();
 
     if (PromOp.getOpcode() == ISD::TRUNCATE ||
         PromOp.getOpcode() == ISD::SIGN_EXTEND ||
@@ -9666,7 +9914,7 @@ SDValue PPCTargetLowering::DAGCombineTruncBoolExt(SDNode *N,
       if (!isa<ConstantSDNode>(PromOp.getOperand(0)) &&
           PromOp.getOperand(0).getValueType() != MVT::i1) {
         // The operand is not yet ready (see comment below).
-        PromOps.insert(PromOps.begin(), PromOp);
+        PromOpHandles.emplace_front(PromOp);
         continue;
       }
 
@@ -9693,7 +9941,7 @@ SDValue PPCTargetLowering::DAGCombineTruncBoolExt(SDNode *N,
       // promoted (this should be rare because we're going through the
       // list backward, but if one of the operands has several users in
       // this cluster of to-be-promoted nodes, it is possible).
-      PromOps.insert(PromOps.begin(), PromOp);
+      PromOpHandles.emplace_front(PromOp);
       continue;
     }
 
@@ -9900,13 +10148,17 @@ SDValue PPCTargetLowering::DAGCombineExtBoolTrunc(SDNode *N,
         DAG.getAnyExtOrTrunc(InSrc, dl, N->getValueType(0)));
   }
 
+  std::list<HandleSDNode> PromOpHandles;
+  for (auto &PromOp : PromOps)
+    PromOpHandles.emplace_back(PromOp);
+
   // Replace all operations (these are all the same, but have a different
   // (promoted) return type). DAG.getNode will validate that the types of
   // a binary operator match, so go through the list in reverse so that
   // we've likely promoted both operands first.
-  while (!PromOps.empty()) {
-    SDValue PromOp = PromOps.back();
-    PromOps.pop_back();
+  while (!PromOpHandles.empty()) {
+    SDValue PromOp = PromOpHandles.back().getValue();
+    PromOpHandles.pop_back();
 
     unsigned C;
     switch (PromOp.getOpcode()) {
@@ -9923,7 +10175,7 @@ SDValue PPCTargetLowering::DAGCombineExtBoolTrunc(SDNode *N,
       // promoted (this should be rare because we're going through the
       // list backward, but if one of the operands has several users in
       // this cluster of to-be-promoted nodes, it is possible).
-      PromOps.insert(PromOps.begin(), PromOp);
+      PromOpHandles.emplace_front(PromOp);
       continue;
     }
 
@@ -9935,7 +10187,7 @@ SDValue PPCTargetLowering::DAGCombineExtBoolTrunc(SDNode *N,
            PromOp.getOperand(0).getValueType() != N->getValueType(0)) ||
           (SelectTruncOp[1].count(PromOp.getNode()) &&
            PromOp.getOperand(1).getValueType() != N->getValueType(0))) {
-        PromOps.insert(PromOps.begin(), PromOp);
+        PromOpHandles.emplace_front(PromOp);
         continue;
       }
     }
@@ -9997,6 +10249,59 @@ SDValue PPCTargetLowering::DAGCombineExtBoolTrunc(SDNode *N,
       ShiftCst);
 }
 
+SDValue PPCTargetLowering::DAGCombineBuildVector(SDNode *N,
+                                                 DAGCombinerInfo &DCI) const {
+  assert(N->getOpcode() == ISD::BUILD_VECTOR &&
+         "Should be called with a BUILD_VECTOR node");
+
+  SelectionDAG &DAG = DCI.DAG;
+  SDLoc dl(N);
+  if (N->getValueType(0) != MVT::v2f64 || !Subtarget.hasVSX())
+    return SDValue();
+
+  // Looking for:
+  // (build_vector ([su]int_to_fp (extractelt 0)), [su]int_to_fp (extractelt 1))
+  if (N->getOperand(0).getOpcode() != ISD::SINT_TO_FP &&
+      N->getOperand(0).getOpcode() != ISD::UINT_TO_FP)
+    return SDValue();
+  if (N->getOperand(1).getOpcode() != ISD::SINT_TO_FP &&
+      N->getOperand(1).getOpcode() != ISD::UINT_TO_FP)
+    return SDValue();
+  if (N->getOperand(0).getOpcode() != N->getOperand(1).getOpcode())
+    return SDValue();
+
+  SDValue Ext1 = N->getOperand(0).getOperand(0);
+  SDValue Ext2 = N->getOperand(1).getOperand(0);
+  if(Ext1.getOpcode() != ISD::EXTRACT_VECTOR_ELT ||
+     Ext2.getOpcode() != ISD::EXTRACT_VECTOR_ELT)
+    return SDValue();
+
+  ConstantSDNode *Ext1Op = dyn_cast<ConstantSDNode>(Ext1.getOperand(1));
+  ConstantSDNode *Ext2Op = dyn_cast<ConstantSDNode>(Ext2.getOperand(1));
+  if (!Ext1Op || !Ext2Op)
+    return SDValue();
+  if (Ext1.getValueType() != MVT::i32 ||
+      Ext2.getValueType() != MVT::i32)
+  if (Ext1.getOperand(0) != Ext2.getOperand(0))
+    return SDValue();
+
+  int FirstElem = Ext1Op->getZExtValue();
+  int SecondElem = Ext2Op->getZExtValue();
+  int SubvecIdx;
+  if (FirstElem == 0 && SecondElem == 1)
+    SubvecIdx = Subtarget.isLittleEndian() ? 1 : 0;
+  else if (FirstElem == 2 && SecondElem == 3)
+    SubvecIdx = Subtarget.isLittleEndian() ? 0 : 1;
+  else
+    return SDValue();
+
+  SDValue SrcVec = Ext1.getOperand(0);
+  auto NodeType = (N->getOperand(1).getOpcode() == ISD::SINT_TO_FP) ?
+    PPCISD::SINT_VEC_TO_FP : PPCISD::UINT_VEC_TO_FP;
+  return DAG.getNode(NodeType, dl, MVT::v2f64,
+                     SrcVec, DAG.getIntPtrConstant(SubvecIdx, dl));
+}
+
 SDValue PPCTargetLowering::combineFPToIntToFP(SDNode *N,
                                               DAGCombinerInfo &DCI) const {
   assert((N->getOpcode() == ISD::SINT_TO_FP ||
@@ -10109,13 +10414,24 @@ SDValue PPCTargetLowering::expandVSXLoadForLE(SDNode *N,
   MVT VecTy = N->getValueType(0).getSimpleVT();
   SDValue LoadOps[] = { Chain, Base };
   SDValue Load = DAG.getMemIntrinsicNode(PPCISD::LXVD2X, dl,
-                                         DAG.getVTList(VecTy, MVT::Other),
-                                         LoadOps, VecTy, MMO);
+                                         DAG.getVTList(MVT::v2f64, MVT::Other),
+                                         LoadOps, MVT::v2f64, MMO);
+
   DCI.AddToWorklist(Load.getNode());
   Chain = Load.getValue(1);
-  SDValue Swap = DAG.getNode(PPCISD::XXSWAPD, dl,
-                             DAG.getVTList(VecTy, MVT::Other), Chain, Load);
+  SDValue Swap = DAG.getNode(
+      PPCISD::XXSWAPD, dl, DAG.getVTList(MVT::v2f64, MVT::Other), Chain, Load);
   DCI.AddToWorklist(Swap.getNode());
+
+  // Add a bitcast if the resulting load type doesn't match v2f64.
+  if (VecTy != MVT::v2f64) {
+    SDValue N = DAG.getNode(ISD::BITCAST, dl, VecTy, Swap);
+    DCI.AddToWorklist(N.getNode());
+    // Package {bitcast value, swap's chain} to match Load's shape.
+    return DAG.getNode(ISD::MERGE_VALUES, dl, DAG.getVTList(VecTy, MVT::Other),
+                       N, Swap.getValue(1));
+  }
+
   return Swap;
 }
 
@@ -10159,8 +10475,15 @@ SDValue PPCTargetLowering::expandVSXStoreForLE(SDNode *N,
 
   SDValue Src = N->getOperand(SrcOpnd);
   MVT VecTy = Src.getValueType().getSimpleVT();
+
+  // All stores are done as v2f64 and possible bit cast.
+  if (VecTy != MVT::v2f64) {
+    Src = DAG.getNode(ISD::BITCAST, dl, MVT::v2f64, Src);
+    DCI.AddToWorklist(Src.getNode());
+  }
+
   SDValue Swap = DAG.getNode(PPCISD::XXSWAPD, dl,
-                             DAG.getVTList(VecTy, MVT::Other), Chain, Src);
+                             DAG.getVTList(MVT::v2f64, MVT::Other), Chain, Src);
   DCI.AddToWorklist(Swap.getNode());
   Chain = Swap.getValue(1);
   SDValue StoreOps[] = { Chain, Swap, Base };
@@ -10277,6 +10600,111 @@ SDValue PPCTargetLowering::PerformDAGCombine(SDNode *N,
         return expandVSXLoadForLE(N, DCI);
     }
 
+    // We sometimes end up with a 64-bit integer load, from which we extract
+    // two single-precision floating-point numbers. This happens with
+    // std::complex<float>, and other similar structures, because of the way we
+    // canonicalize structure copies. However, if we lack direct moves,
+    // then the final bitcasts from the extracted integer values to the
+    // floating-point numbers turn into store/load pairs. Even with direct moves,
+    // just loading the two floating-point numbers is likely better.
+    auto ReplaceTwoFloatLoad = [&]() {
+      if (VT != MVT::i64)
+        return false;
+
+      if (LD->getExtensionType() != ISD::NON_EXTLOAD ||
+          LD->isVolatile())
+        return false;
+
+      //  We're looking for a sequence like this:
+      //  t13: i64,ch = load<LD8[%ref.tmp]> t0, t6, undef:i64
+      //      t16: i64 = srl t13, Constant:i32<32>
+      //    t17: i32 = truncate t16
+      //  t18: f32 = bitcast t17
+      //    t19: i32 = truncate t13
+      //  t20: f32 = bitcast t19
+
+      if (!LD->hasNUsesOfValue(2, 0))
+        return false;
+
+      auto UI = LD->use_begin();
+      while (UI.getUse().getResNo() != 0) ++UI;
+      SDNode *Trunc = *UI++;
+      while (UI.getUse().getResNo() != 0) ++UI;
+      SDNode *RightShift = *UI;
+      if (Trunc->getOpcode() != ISD::TRUNCATE)
+        std::swap(Trunc, RightShift);
+
+      if (Trunc->getOpcode() != ISD::TRUNCATE ||
+          Trunc->getValueType(0) != MVT::i32 ||
+          !Trunc->hasOneUse())
+        return false;
+      if (RightShift->getOpcode() != ISD::SRL ||
+          !isa<ConstantSDNode>(RightShift->getOperand(1)) ||
+          RightShift->getConstantOperandVal(1) != 32 ||
+          !RightShift->hasOneUse())
+        return false;
+
+      SDNode *Trunc2 = *RightShift->use_begin();
+      if (Trunc2->getOpcode() != ISD::TRUNCATE ||
+          Trunc2->getValueType(0) != MVT::i32 ||
+          !Trunc2->hasOneUse())
+        return false;
+
+      SDNode *Bitcast = *Trunc->use_begin();
+      SDNode *Bitcast2 = *Trunc2->use_begin();
+
+      if (Bitcast->getOpcode() != ISD::BITCAST ||
+          Bitcast->getValueType(0) != MVT::f32)
+        return false;
+      if (Bitcast2->getOpcode() != ISD::BITCAST ||
+          Bitcast2->getValueType(0) != MVT::f32)
+        return false;
+
+      if (Subtarget.isLittleEndian())
+        std::swap(Bitcast, Bitcast2);
+
+      // Bitcast has the second float (in memory-layout order) and Bitcast2
+      // has the first one.
+
+      SDValue BasePtr = LD->getBasePtr();
+      if (LD->isIndexed()) {
+        assert(LD->getAddressingMode() == ISD::PRE_INC &&
+               "Non-pre-inc AM on PPC?");
+        BasePtr =
+          DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr,
+                      LD->getOffset());
+      }
+
+      auto MMOFlags =
+          LD->getMemOperand()->getFlags() & ~MachineMemOperand::MOVolatile;
+      SDValue FloatLoad = DAG.getLoad(MVT::f32, dl, LD->getChain(), BasePtr,
+                                      LD->getPointerInfo(), LD->getAlignment(),
+                                      MMOFlags, LD->getAAInfo());
+      SDValue AddPtr =
+        DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(),
+                    BasePtr, DAG.getIntPtrConstant(4, dl));
+      SDValue FloatLoad2 = DAG.getLoad(
+          MVT::f32, dl, SDValue(FloatLoad.getNode(), 1), AddPtr,
+          LD->getPointerInfo().getWithOffset(4),
+          MinAlign(LD->getAlignment(), 4), MMOFlags, LD->getAAInfo());
+
+      if (LD->isIndexed()) {
+        // Note that DAGCombine should re-form any pre-increment load(s) from
+        // what is produced here if that makes sense.
+        DAG.ReplaceAllUsesOfValueWith(SDValue(LD, 1), BasePtr);
+      }
+
+      DCI.CombineTo(Bitcast2, FloatLoad);
+      DCI.CombineTo(Bitcast, FloatLoad2);
+
+      DAG.ReplaceAllUsesOfValueWith(SDValue(LD, LD->isIndexed() ? 2 : 1),
+                                    SDValue(FloatLoad2.getNode(), 1));
+      return true;
+    };
+
+    if (ReplaceTwoFloatLoad())
+      return SDValue(N, 0);
+
     EVT MemVT = LD->getMemoryVT();
     Type *Ty = MemVT.getTypeForEVT(*DAG.getContext());
     unsigned ABIAlignment = DAG.getDataLayout().getABITypeAlignment(Ty);
@@ -10710,6 +11138,8 @@ SDValue PPCTargetLowering::PerformDAGCombine(SDNode *N,
     }
     break;
   }
+  case ISD::BUILD_VECTOR:
+    return DAGCombineBuildVector(N, DCI);
   }
 
   return SDValue();
@@ -10801,7 +11231,8 @@ unsigned PPCTargetLowering::getPrefLoopAlignment(MachineLoop *ML) const {
   case PPC::DIR_PWR6:
   case PPC::DIR_PWR6X:
   case PPC::DIR_PWR7:
-  case PPC::DIR_PWR8: {
+  case PPC::DIR_PWR8:
+  case PPC::DIR_PWR9: {
     if (!ML)
       break;
 
@@ -10812,7 +11243,7 @@ unsigned PPCTargetLowering::getPrefLoopAlignment(MachineLoop *ML) const {
     uint64_t LoopSize = 0;
     for (auto I = ML->block_begin(), IE = ML->block_end(); I != IE; ++I)
       for (auto J = (*I)->begin(), JE = (*I)->end(); J != JE; ++J) {
-        LoopSize += TII->GetInstSizeInBytes(J);
+        LoopSize += TII->GetInstSizeInBytes(*J);
         if (LoopSize > 32)
           break;
       }
@@ -10837,6 +11268,7 @@ PPCTargetLowering::getConstraintType(StringRef Constraint) const {
     case 'b':
     case 'r':
     case 'f':
+    case 'd':
     case 'v':
     case 'y':
       return C_RegisterClass;
@@ -10928,6 +11360,10 @@ PPCTargetLowering::getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI,
       if (VT == MVT::i64 && Subtarget.isPPC64())
         return std::make_pair(0U, &PPC::G8RCRegClass);
       return std::make_pair(0U, &PPC::GPRCRegClass);
+    // 'd' and 'f' constraints are both defined to be "the floating point
+    // registers", where one is for 32-bit and the other for 64-bit. We don't
+    // really care overly much here so just give them all the same reg classes.
+    case 'd':
     case 'f':
       if (VT == MVT::f32 || VT == MVT::i32)
         return std::make_pair(0U, &PPC::F4RCRegClass);
@@ -11126,13 +11562,13 @@ SDValue PPCTargetLowering::LowerRETURNADDR(SDValue Op,
                         isPPC64 ? MVT::i64 : MVT::i32);
     return DAG.getLoad(PtrVT, dl, DAG.getEntryNode(),
                        DAG.getNode(ISD::ADD, dl, PtrVT, FrameAddr, Offset),
-                       MachinePointerInfo(), false, false, false, 0);
+                       MachinePointerInfo());
   }
 
   // Just load the return address off the stack.
   SDValue RetAddrFI = getReturnAddrFrameIndex(DAG);
   return DAG.getLoad(PtrVT, dl, DAG.getEntryNode(), RetAddrFI,
-                     MachinePointerInfo(), false, false, false, 0);
+                     MachinePointerInfo());
 }
 
 SDValue PPCTargetLowering::LowerFRAMEADDR(SDValue Op,
@@ -11144,7 +11580,7 @@ SDValue PPCTargetLowering::LowerFRAMEADDR(SDValue Op,
   MachineFrameInfo *MFI = MF.getFrameInfo();
   MFI->setFrameAddressIsTaken(true);
 
-  EVT PtrVT = DAG.getTargetLoweringInfo().getPointerTy(MF.getDataLayout());
+  EVT PtrVT = getPointerTy(MF.getDataLayout());
   bool isPPC64 = PtrVT == MVT::i64;
 
   // Naked functions never have a frame pointer, and so we use r1. For all
@@ -11159,8 +11595,7 @@ SDValue PPCTargetLowering::LowerFRAMEADDR(SDValue Op,
                                          PtrVT);
   while (Depth--)
     FrameAddr = DAG.getLoad(Op.getValueType(), dl, DAG.getEntryNode(),
-                            FrameAddr, MachinePointerInfo(), false, false,
-                            false, 0);
+                            FrameAddr, MachinePointerInfo());
   return FrameAddr;
 }
 
@@ -11567,10 +12002,8 @@ PPCTargetLowering::shouldExpandBuildVectorWithShuffles(
   if (VT == MVT::v2i64)
     return Subtarget.hasDirectMove(); // Don't need stack ops with direct moves
 
-  if (Subtarget.hasQPX()) {
-    if (VT == MVT::v4f32 || VT == MVT::v4f64 || VT == MVT::v4i1)
-      return true;
-  }
+  if (Subtarget.hasVSX() || Subtarget.hasQPX())
+    return true;
 
   return TargetLowering::shouldExpandBuildVectorWithShuffles(VT, DefinedValues);
 }
@@ -11588,3 +12021,70 @@ PPCTargetLowering::createFastISel(FunctionLoweringInfo &FuncInfo,
                                   const TargetLibraryInfo *LibInfo) const {
   return PPC::createFastISel(FuncInfo, LibInfo);
 }
+
+void PPCTargetLowering::initializeSplitCSR(MachineBasicBlock *Entry) const {
+  if (Subtarget.isDarwinABI()) return;
+  if (!Subtarget.isPPC64()) return;
+
+  // Update IsSplitCSR in PPCFunctionInfo
+  PPCFunctionInfo *PFI = Entry->getParent()->getInfo<PPCFunctionInfo>();
+  PFI->setIsSplitCSR(true);
+}
+
+void PPCTargetLowering::insertCopiesSplitCSR(
+  MachineBasicBlock *Entry,
+  const SmallVectorImpl<MachineBasicBlock *> &Exits) const {
+  const PPCRegisterInfo *TRI = Subtarget.getRegisterInfo();
+  const MCPhysReg *IStart = TRI->getCalleeSavedRegsViaCopy(Entry->getParent());
+  if (!IStart)
+    return;
+
+  const TargetInstrInfo *TII = Subtarget.getInstrInfo();
+  MachineRegisterInfo *MRI = &Entry->getParent()->getRegInfo();
+  MachineBasicBlock::iterator MBBI = Entry->begin();
+  for (const MCPhysReg *I = IStart; *I; ++I) {
+    const TargetRegisterClass *RC = nullptr;
+    if (PPC::G8RCRegClass.contains(*I))
+      RC = &PPC::G8RCRegClass;
+    else if (PPC::F8RCRegClass.contains(*I))
+      RC = &PPC::F8RCRegClass;
+    else if (PPC::CRRCRegClass.contains(*I))
+      RC = &PPC::CRRCRegClass;
+    else if (PPC::VRRCRegClass.contains(*I))
+      RC = &PPC::VRRCRegClass;
+    else
+      llvm_unreachable("Unexpected register class in CSRsViaCopy!");
+
+    unsigned NewVR = MRI->createVirtualRegister(RC);
+    // Create copy from CSR to a virtual register.
+    // FIXME: this currently does not emit CFI pseudo-instructions, it works
+    // fine for CXX_FAST_TLS since the C++-style TLS access functions should be
+    // nounwind. If we want to generalize this later, we may need to emit
+    // CFI pseudo-instructions.
+    assert(Entry->getParent()->getFunction()->hasFnAttribute(
+             Attribute::NoUnwind) &&
+           "Function should be nounwind in insertCopiesSplitCSR!");
+    Entry->addLiveIn(*I);
+    BuildMI(*Entry, MBBI, DebugLoc(), TII->get(TargetOpcode::COPY), NewVR)
+      .addReg(*I);
+
+    // Insert the copy-back instructions right before the terminator
+    for (auto *Exit : Exits)
+      BuildMI(*Exit, Exit->getFirstTerminator(), DebugLoc(),
+              TII->get(TargetOpcode::COPY), *I)
+        .addReg(NewVR);
+  }
+}
+
+// Override to enable LOAD_STACK_GUARD lowering on Linux.
+bool PPCTargetLowering::useLoadStackGuardNode() const {
+  if (!Subtarget.isTargetLinux())
+    return TargetLowering::useLoadStackGuardNode();
+  return true;
+}
+
+// Override to disable global variable loading on Linux.
+void PPCTargetLowering::insertSSPDeclarations(Module &M) const {
+  if (!Subtarget.isTargetLinux())
+    return TargetLowering::insertSSPDeclarations(M);
+}
diff --git a/lib/Target/PowerPC/PPCISelLowering.h b/lib/Target/PowerPC/PPCISelLowering.h
index 44bcb8942cfcc..e3be8074e62ec 100644
--- a/lib/Target/PowerPC/PPCISelLowering.h
+++ b/lib/Target/PowerPC/PPCISelLowering.h
@@ -61,6 +61,18 @@ namespace llvm {
       ///
       VPERM,
 
+      /// XXSPLT - The PPC VSX splat instructions
+      ///
+      XXSPLT,
+
+      /// XXINSERT - The PPC VSX insert instruction
+      ///
+      XXINSERT,
+
+      /// VECSHL - The PPC VSX shift left instruction
+      ///
+      VECSHL,
+
       /// The CMPB instruction (takes two operands of i32 or i64).
       CMPB,
 
@@ -133,6 +145,16 @@ namespace llvm {
       /// Direct move from a GPR to a VSX register (zero)
       MTVSRZ,
 
+      /// Extract a subvector from signed integer vector and convert to FP.
+      /// It is primarily used to convert a (widened) illegal integer vector
+      /// type to a legal floating point vector type.
+      /// For example v2i32 -> widened to v4i32 -> v2f64
+      SINT_VEC_TO_FP,
+
+      /// Extract a subvector from unsigned integer vector and convert to FP.
+      /// As with SINT_VEC_TO_FP, used for converting illegal types.
+      UINT_VEC_TO_FP,
+
       // FIXME: Remove these once the ANDI glue bug is fixed:
       /// i1 = ANDIo_1_[EQ|GT]_BIT(i32 or i64 x) - Represents the result of the
       /// eq or gt bit of CR0 after executing andi. x, 1. This is used to
@@ -297,6 +319,10 @@ namespace llvm {
       /// of outputs.
       XXSWAPD,
 
+      /// An SDNode for swaps that are not associated with any loads/stores
+      /// and thereby have no chain.
+      SWAP_NO_CHAIN,
+
       /// QVFPERM = This corresponds to the QPX qvfperm instruction.
       QVFPERM,
 
@@ -402,6 +428,16 @@ namespace llvm {
     /// VSPLTB/VSPLTH/VSPLTW.
     bool isSplatShuffleMask(ShuffleVectorSDNode *N, unsigned EltSize);
 
+    /// isXXINSERTWMask - Return true if this VECTOR_SHUFFLE can be handled by
+    /// the XXINSERTW instruction introduced in ISA 3.0. This is essentially any
+    /// shuffle of v4f32/v4i32 vectors that just inserts one element from one
+    /// vector into the other. This function will also set a couple of
+    /// output parameters for how much the source vector needs to be shifted and
+    /// what byte number needs to be specified for the instruction to put the
+    /// element in the desired location of the target vector.
+    bool isXXINSERTWMask(ShuffleVectorSDNode *N, unsigned &ShiftElts,
+                         unsigned &InsertAtByte, bool &Swap, bool IsLE);
+
     /// getVSPLTImmediate - Return the appropriate VSPLT* immediate to splat the
     /// specified isSplatShuffleMask VECTOR_SHUFFLE mask.
     unsigned getVSPLTImmediate(SDNode *N, unsigned EltSize, SelectionDAG &DAG);
@@ -428,6 +464,20 @@ namespace llvm {
     /// DAG node.
     const char *getTargetNodeName(unsigned Opcode) const override;
 
+    /// getPreferredVectorAction - The code we generate when vector types are
+    /// legalized by promoting the integer element type is often much worse
+    /// than code we generate if we widen the type for applicable vector types.
+    /// The issue with promoting is that the vector is scalaraized, individual
+    /// elements promoted and then the vector is rebuilt. So say we load a pair
+    /// of v4i8's and shuffle them. This will turn into a mess of 8 extending
+    /// loads, moves back into VSR's (or memory ops if we don't have moves) and
+    /// then the VPERM for the shuffle. All in all a very slow sequence.
+    TargetLoweringBase::LegalizeTypeAction getPreferredVectorAction(EVT VT)
+      const override {
+      if (VT.getVectorElementType().getSizeInBits() % 8 == 0)
+        return TypeWidenVector;
+      return TargetLoweringBase::getPreferredVectorAction(VT);
+    }
     bool useSoftFloat() const override;
 
     MVT getScalarShiftAmountTy(const DataLayout &, EVT) const override {
@@ -442,6 +492,18 @@ namespace llvm {
       return true;
     }
 
+    bool supportSplitCSR(MachineFunction *MF) const override {
+      return
+        MF->getFunction()->getCallingConv() == CallingConv::CXX_FAST_TLS &&
+        MF->getFunction()->hasFnAttribute(Attribute::NoUnwind);
+    }
+
+    void initializeSplitCSR(MachineBasicBlock *Entry) const override;
+
+    void insertCopiesSplitCSR(
+      MachineBasicBlock *Entry,
+      const SmallVectorImpl<MachineBasicBlock *> &Exits) const override;
+
     /// getSetCCResultType - Return the ISD::SETCC ValueType
     EVT getSetCCResultType(const DataLayout &DL, LLVMContext &Context,
                            EVT VT) const override;
@@ -508,26 +570,31 @@ namespace llvm {
 
     unsigned getPrefLoopAlignment(MachineLoop *ML) const override;
 
+    bool shouldInsertFencesForAtomic(const Instruction *I) const override {
+      return true;
+    }
+
     Instruction* emitLeadingFence(IRBuilder<> &Builder, AtomicOrdering Ord,
                                   bool IsStore, bool IsLoad) const override;
     Instruction* emitTrailingFence(IRBuilder<> &Builder, AtomicOrdering Ord,
                                    bool IsStore, bool IsLoad) const override;
 
     MachineBasicBlock *
-      EmitInstrWithCustomInserter(MachineInstr *MI,
-                                  MachineBasicBlock *MBB) const override;
-    MachineBasicBlock *EmitAtomicBinary(MachineInstr *MI,
+    EmitInstrWithCustomInserter(MachineInstr &MI,
+                                MachineBasicBlock *MBB) const override;
+    MachineBasicBlock *EmitAtomicBinary(MachineInstr &MI,
                                         MachineBasicBlock *MBB,
                                         unsigned AtomicSize,
                                         unsigned BinOpcode) const;
-    MachineBasicBlock *EmitPartwordAtomicBinary(MachineInstr *MI,
+    MachineBasicBlock *EmitPartwordAtomicBinary(MachineInstr &MI,
                                                 MachineBasicBlock *MBB,
-                                            bool is8bit, unsigned Opcode) const;
+                                                bool is8bit,
+                                                unsigned Opcode) const;
 
-    MachineBasicBlock *emitEHSjLjSetJmp(MachineInstr *MI,
+    MachineBasicBlock *emitEHSjLjSetJmp(MachineInstr &MI,
                                         MachineBasicBlock *MBB) const;
 
-    MachineBasicBlock *emitEHSjLjLongJmp(MachineInstr *MI,
+    MachineBasicBlock *emitEHSjLjLongJmp(MachineInstr &MI,
                                          MachineBasicBlock *MBB) const;
 
     ConstraintType getConstraintType(StringRef Constraint) const override;
@@ -672,6 +739,10 @@ namespace llvm {
     unsigned
     getExceptionSelectorRegister(const Constant *PersonalityFn) const override;
 
+    /// Override to support customized stack guard loading.
+    bool useLoadStackGuardNode() const override;
+    void insertSSPDeclarations(Module &M) const override;
+
   private:
     struct ReuseLoadInfo {
       SDValue Ptr;
@@ -693,11 +764,11 @@ namespace llvm {
                          SelectionDAG &DAG) const;
 
     void LowerFP_TO_INTForReuse(SDValue Op, ReuseLoadInfo &RLI,
-                                SelectionDAG &DAG, SDLoc dl) const;
+                                SelectionDAG &DAG, const SDLoc &dl) const;
     SDValue LowerFP_TO_INTDirectMove(SDValue Op, SelectionDAG &DAG,
-                                     SDLoc dl) const;
+                                     const SDLoc &dl) const;
     SDValue LowerINT_TO_FPDirectMove(SDValue Op, SelectionDAG &DAG,
-                                     SDLoc dl) const;
+                                     const SDLoc &dl) const;
 
     SDValue getFramePointerFrameIndex(SelectionDAG & DAG) const;
     SDValue getReturnAddrFrameIndex(SelectionDAG & DAG) const;
@@ -709,13 +780,20 @@ namespace llvm {
                                       const SmallVectorImpl<ISD::InputArg> &Ins,
                                       SelectionDAG& DAG) const;
 
-    SDValue EmitTailCallLoadFPAndRetAddr(SelectionDAG & DAG,
-                                         int SPDiff,
-                                         SDValue Chain,
-                                         SDValue &LROpOut,
+    bool
+    IsEligibleForTailCallOptimization_64SVR4(
+                                    SDValue Callee,
+                                    CallingConv::ID CalleeCC,
+                                    ImmutableCallSite *CS,
+                                    bool isVarArg,
+                                    const SmallVectorImpl<ISD::OutputArg> &Outs,
+                                    const SmallVectorImpl<ISD::InputArg> &Ins,
+                                    SelectionDAG& DAG) const;
+
+    SDValue EmitTailCallLoadFPAndRetAddr(SelectionDAG &DAG, int SPDiff,
+                                         SDValue Chain, SDValue &LROpOut,
                                          SDValue &FPOpOut,
-                                         bool isDarwinABI,
-                                         SDLoc dl) const;
+                                         const SDLoc &dl) const;
 
     SDValue LowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const;
@@ -727,23 +805,18 @@ namespace llvm {
     SDValue LowerSETCC(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerINIT_TRAMPOLINE(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerADJUST_TRAMPOLINE(SDValue Op, SelectionDAG &DAG) const;
-    SDValue LowerVASTART(SDValue Op, SelectionDAG &DAG,
-                         const PPCSubtarget &Subtarget) const;
-    SDValue LowerVAARG(SDValue Op, SelectionDAG &DAG,
-                       const PPCSubtarget &Subtarget) const;
-    SDValue LowerVACOPY(SDValue Op, SelectionDAG &DAG,
-                        const PPCSubtarget &Subtarget) const;
-    SDValue LowerSTACKRESTORE(SDValue Op, SelectionDAG &DAG,
-                                const PPCSubtarget &Subtarget) const;
-    SDValue LowerGET_DYNAMIC_AREA_OFFSET(SDValue Op, SelectionDAG &DAG,
-                                         const PPCSubtarget &Subtarget) const;
-    SDValue LowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG,
-                                      const PPCSubtarget &Subtarget) const;
+    SDValue LowerVASTART(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerVAARG(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerVACOPY(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerSTACKRESTORE(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerGET_DYNAMIC_AREA_OFFSET(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerLOAD(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerSTORE(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerTRUNCATE(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerSELECT_CC(SDValue Op, SelectionDAG &DAG) const;
-    SDValue LowerFP_TO_INT(SDValue Op, SelectionDAG &DAG, SDLoc dl) const;
+    SDValue LowerFP_TO_INT(SDValue Op, SelectionDAG &DAG,
+                           const SDLoc &dl) const;
     SDValue LowerINT_TO_FP(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerFLT_ROUNDS_(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerSHL_PARTS(SDValue Op, SelectionDAG &DAG) const;
@@ -763,26 +836,23 @@ namespace llvm {
     SDValue LowerCallResult(SDValue Chain, SDValue InFlag,
                             CallingConv::ID CallConv, bool isVarArg,
                             const SmallVectorImpl<ISD::InputArg> &Ins,
-                            SDLoc dl, SelectionDAG &DAG,
+                            const SDLoc &dl, SelectionDAG &DAG,
                             SmallVectorImpl<SDValue> &InVals) const;
-    SDValue FinishCall(CallingConv::ID CallConv, SDLoc dl, bool isTailCall,
-                       bool isVarArg, bool IsPatchPoint, bool hasNest,
-                       SelectionDAG &DAG,
-                       SmallVector<std::pair<unsigned, SDValue>, 8>
-                         &RegsToPass,
+    SDValue FinishCall(CallingConv::ID CallConv, const SDLoc &dl,
+                       bool isTailCall, bool isVarArg, bool isPatchPoint,
+                       bool hasNest, SelectionDAG &DAG,
+                       SmallVector<std::pair<unsigned, SDValue>, 8> &RegsToPass,
                        SDValue InFlag, SDValue Chain, SDValue CallSeqStart,
-                       SDValue &Callee,
-                       int SPDiff, unsigned NumBytes,
+                       SDValue &Callee, int SPDiff, unsigned NumBytes,
                        const SmallVectorImpl<ISD::InputArg> &Ins,
                        SmallVectorImpl<SDValue> &InVals,
                        ImmutableCallSite *CS) const;
 
     SDValue
-      LowerFormalArguments(SDValue Chain,
-                           CallingConv::ID CallConv, bool isVarArg,
-                           const SmallVectorImpl<ISD::InputArg> &Ins,
-                           SDLoc dl, SelectionDAG &DAG,
-                           SmallVectorImpl<SDValue> &InVals) const override;
+    LowerFormalArguments(SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+                         const SmallVectorImpl<ISD::InputArg> &Ins,
+                         const SDLoc &dl, SelectionDAG &DAG,
+                         SmallVectorImpl<SDValue> &InVals) const override;
 
     SDValue
       LowerCall(TargetLowering::CallLoweringInfo &CLI,
@@ -794,75 +864,66 @@ namespace llvm {
                    const SmallVectorImpl<ISD::OutputArg> &Outs,
                    LLVMContext &Context) const override;
 
-    SDValue
-      LowerReturn(SDValue Chain,
-                  CallingConv::ID CallConv, bool isVarArg,
-                  const SmallVectorImpl<ISD::OutputArg> &Outs,
-                  const SmallVectorImpl<SDValue> &OutVals,
-                  SDLoc dl, SelectionDAG &DAG) const override;
-
-    SDValue
-      extendArgForPPC64(ISD::ArgFlagsTy Flags, EVT ObjectVT, SelectionDAG &DAG,
-                        SDValue ArgVal, SDLoc dl) const;
-
-    SDValue
-      LowerFormalArguments_Darwin(SDValue Chain,
-                                  CallingConv::ID CallConv, bool isVarArg,
-                                  const SmallVectorImpl<ISD::InputArg> &Ins,
-                                  SDLoc dl, SelectionDAG &DAG,
-                                  SmallVectorImpl<SDValue> &InVals) const;
-    SDValue
-      LowerFormalArguments_64SVR4(SDValue Chain,
-                                  CallingConv::ID CallConv, bool isVarArg,
-                                  const SmallVectorImpl<ISD::InputArg> &Ins,
-                                  SDLoc dl, SelectionDAG &DAG,
-                                  SmallVectorImpl<SDValue> &InVals) const;
-    SDValue
-      LowerFormalArguments_32SVR4(SDValue Chain,
-                                  CallingConv::ID CallConv, bool isVarArg,
-                                  const SmallVectorImpl<ISD::InputArg> &Ins,
-                                  SDLoc dl, SelectionDAG &DAG,
-                                  SmallVectorImpl<SDValue> &InVals) const;
-
-    SDValue
-      createMemcpyOutsideCallSeq(SDValue Arg, SDValue PtrOff,
-                                 SDValue CallSeqStart, ISD::ArgFlagsTy Flags,
-                                 SelectionDAG &DAG, SDLoc dl) const;
-
-    SDValue
-      LowerCall_Darwin(SDValue Chain, SDValue Callee,
-                       CallingConv::ID CallConv,
-                       bool isVarArg, bool isTailCall, bool IsPatchPoint,
-                       const SmallVectorImpl<ISD::OutputArg> &Outs,
-                       const SmallVectorImpl<SDValue> &OutVals,
-                       const SmallVectorImpl<ISD::InputArg> &Ins,
-                       SDLoc dl, SelectionDAG &DAG,
-                       SmallVectorImpl<SDValue> &InVals,
-                       ImmutableCallSite *CS) const;
-    SDValue
-      LowerCall_64SVR4(SDValue Chain, SDValue Callee,
-                       CallingConv::ID CallConv,
-                       bool isVarArg, bool isTailCall, bool IsPatchPoint,
-                       const SmallVectorImpl<ISD::OutputArg> &Outs,
-                       const SmallVectorImpl<SDValue> &OutVals,
-                       const SmallVectorImpl<ISD::InputArg> &Ins,
-                       SDLoc dl, SelectionDAG &DAG,
-                       SmallVectorImpl<SDValue> &InVals,
-                       ImmutableCallSite *CS) const;
-    SDValue
-    LowerCall_32SVR4(SDValue Chain, SDValue Callee, CallingConv::ID CallConv,
-                     bool isVarArg, bool isTailCall, bool IsPatchPoint,
-                     const SmallVectorImpl<ISD::OutputArg> &Outs,
-                     const SmallVectorImpl<SDValue> &OutVals,
-                     const SmallVectorImpl<ISD::InputArg> &Ins,
-                     SDLoc dl, SelectionDAG &DAG,
-                     SmallVectorImpl<SDValue> &InVals,
-                     ImmutableCallSite *CS) const;
+    SDValue LowerReturn(SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+                        const SmallVectorImpl<ISD::OutputArg> &Outs,
+                        const SmallVectorImpl<SDValue> &OutVals,
+                        const SDLoc &dl, SelectionDAG &DAG) const override;
+
+    SDValue extendArgForPPC64(ISD::ArgFlagsTy Flags, EVT ObjectVT,
+                              SelectionDAG &DAG, SDValue ArgVal,
+                              const SDLoc &dl) const;
+
+    SDValue LowerFormalArguments_Darwin(
+        SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+        const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+        SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const;
+    SDValue LowerFormalArguments_64SVR4(
+        SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+        const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+        SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const;
+    SDValue LowerFormalArguments_32SVR4(
+        SDValue Chain, CallingConv::ID CallConv, bool isVarArg,
+        const SmallVectorImpl<ISD::InputArg> &Ins, const SDLoc &dl,
+        SelectionDAG &DAG, SmallVectorImpl<SDValue> &InVals) const;
+
+    SDValue createMemcpyOutsideCallSeq(SDValue Arg, SDValue PtrOff,
+                                       SDValue CallSeqStart,
+                                       ISD::ArgFlagsTy Flags, SelectionDAG &DAG,
+                                       const SDLoc &dl) const;
+
+    SDValue LowerCall_Darwin(SDValue Chain, SDValue Callee,
+                             CallingConv::ID CallConv, bool isVarArg,
+                             bool isTailCall, bool isPatchPoint,
+                             const SmallVectorImpl<ISD::OutputArg> &Outs,
+                             const SmallVectorImpl<SDValue> &OutVals,
+                             const SmallVectorImpl<ISD::InputArg> &Ins,
+                             const SDLoc &dl, SelectionDAG &DAG,
+                             SmallVectorImpl<SDValue> &InVals,
+                             ImmutableCallSite *CS) const;
+    SDValue LowerCall_64SVR4(SDValue Chain, SDValue Callee,
+                             CallingConv::ID CallConv, bool isVarArg,
+                             bool isTailCall, bool isPatchPoint,
+                             const SmallVectorImpl<ISD::OutputArg> &Outs,
+                             const SmallVectorImpl<SDValue> &OutVals,
+                             const SmallVectorImpl<ISD::InputArg> &Ins,
+                             const SDLoc &dl, SelectionDAG &DAG,
+                             SmallVectorImpl<SDValue> &InVals,
+                             ImmutableCallSite *CS) const;
+    SDValue LowerCall_32SVR4(SDValue Chain, SDValue Callee,
+                             CallingConv::ID CallConv, bool isVarArg,
+                             bool isTailCall, bool isPatchPoint,
+                             const SmallVectorImpl<ISD::OutputArg> &Outs,
+                             const SmallVectorImpl<SDValue> &OutVals,
+                             const SmallVectorImpl<ISD::InputArg> &Ins,
+                             const SDLoc &dl, SelectionDAG &DAG,
+                             SmallVectorImpl<SDValue> &InVals,
+                             ImmutableCallSite *CS) const;
 
     SDValue lowerEH_SJLJ_SETJMP(SDValue Op, SelectionDAG &DAG) const;
     SDValue lowerEH_SJLJ_LONGJMP(SDValue Op, SelectionDAG &DAG) const;
 
     SDValue DAGCombineExtBoolTrunc(SDNode *N, DAGCombinerInfo &DCI) const;
+    SDValue DAGCombineBuildVector(SDNode *N, DAGCombinerInfo &DCI) const;
     SDValue DAGCombineTruncBoolExt(SDNode *N, DAGCombinerInfo &DCI) const;
     SDValue combineFPToIntToFP(SDNode *N, DAGCombinerInfo &DCI) const;
 
diff --git a/lib/Target/PowerPC/PPCInstr64Bit.td b/lib/Target/PowerPC/PPCInstr64Bit.td
index 79e4fe379c2d9..e7eb8a16180a3 100644
--- a/lib/Target/PowerPC/PPCInstr64Bit.td
+++ b/lib/Target/PowerPC/PPCInstr64Bit.td
@@ -244,12 +244,22 @@ def LDARX : XForm_1<31,  84, (outs g8rc:$rD), (ins memrr:$ptr),
 // (EH=1 - see Power ISA 2.07 Book II 4.4.2)
 def LDARXL : XForm_1<31,  84, (outs g8rc:$rD), (ins memrr:$ptr),
                      "ldarx $rD, $ptr, 1", IIC_LdStLDARX, []>, isDOT;
+
+let hasExtraDefRegAllocReq = 1 in
+def LDAT : X_RD5_RS5_IM5<31, 614, (outs g8rc:$rD), (ins g8rc:$rA, u5imm:$FC),
+                         "ldat $rD, $rA, $FC", IIC_LdStLoad>, isPPC64,
+           Requires<[IsISA3_0]>;
 }
 
 let Defs = [CR0], mayStore = 1, hasSideEffects = 0 in
 def STDCX : XForm_1<31, 214, (outs), (ins g8rc:$rS, memrr:$dst),
                     "stdcx. $rS, $dst", IIC_LdStSTDCX, []>, isDOT;
 
+let mayStore = 1, hasSideEffects = 0 in
+def STDAT : X_RD5_RS5_IM5<31, 742, (outs), (ins g8rc:$rS, g8rc:$rA, u5imm:$FC),
+                          "stdat $rS, $rA, $FC", IIC_LdStStore>, isPPC64,
+            Requires<[IsISA3_0]>;
+
 let Interpretation64Bit = 1, isCodeGenOnly = 1 in {
 let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1, Uses = [RM] in
 def TCRETURNdi8 :Pseudo< (outs),
@@ -476,8 +486,10 @@ defm ADD8  : XOForm_1r<31, 266, 0, (outs g8rc:$rT), (ins g8rc:$rA, g8rc:$rB),
                        "add", "$rT, $rA, $rB", IIC_IntSimple,
                        [(set i64:$rT, (add i64:$rA, i64:$rB))]>;
 // ADD8 has a special form: reg = ADD8(reg, sym@tls) for use by the
-// initial-exec thread-local storage model.
-def ADD8TLS  : XOForm_1<31, 266, 0, (outs g8rc:$rT), (ins g8rc:$rA, tlsreg:$rB),
+// initial-exec thread-local storage model.  We need to forbid r0 here -
+// while it works for add just fine, the linker can relax this to local-exec
+// addi, which won't work for r0.
+def ADD8TLS  : XOForm_1<31, 266, 0, (outs g8rc:$rT), (ins g8rc_nox0:$rA, tlsreg:$rB),
                         "add $rT, $rA, $rB", IIC_IntSimple,
                         [(set i64:$rT, (add i64:$rA, tglobaltlsaddr:$rB))]>;
                      
@@ -502,11 +514,11 @@ let Defs = [CARRY] in {
 def SUBFIC8: DForm_2< 8, (outs g8rc:$rD), (ins g8rc:$rA, s16imm64:$imm),
                      "subfic $rD, $rA, $imm", IIC_IntGeneral,
                      [(set i64:$rD, (subc imm64SExt16:$imm, i64:$rA))]>;
-defm SUBFC8 : XOForm_1r<31, 8, 0, (outs g8rc:$rT), (ins g8rc:$rA, g8rc:$rB),
+}
+defm SUBFC8 : XOForm_1rc<31, 8, 0, (outs g8rc:$rT), (ins g8rc:$rA, g8rc:$rB),
                         "subfc", "$rT, $rA, $rB", IIC_IntGeneral,
                         [(set i64:$rT, (subc i64:$rB, i64:$rA))]>,
                         PPC970_DGroup_Cracked;
-}
 defm SUBF8 : XOForm_1r<31, 40, 0, (outs g8rc:$rT), (ins g8rc:$rA, g8rc:$rB),
                        "subf", "$rT, $rA, $rB", IIC_IntGeneral,
                        [(set i64:$rT, (sub i64:$rB, i64:$rA))]>;
@@ -564,6 +576,14 @@ let isCompare = 1, hasSideEffects = 0 in {
   def CMPLDI : DForm_6_ext<10, (outs crrc:$dst), (ins g8rc:$src1, u16imm64:$src2),
                            "cmpldi $dst, $src1, $src2",
                            IIC_IntCompare>, isPPC64;
+  let Interpretation64Bit = 1, isCodeGenOnly = 1 in
+  def CMPRB8 : X_BF3_L1_RS5_RS5<31, 192, (outs crbitrc:$BF),
+                                (ins u1imm:$L, g8rc:$rA, g8rc:$rB),
+                                "cmprb $BF, $L, $rA, $rB", IIC_IntCompare, []>,
+               Requires<[IsISA3_0]>;
+  def CMPEQB : X_BF3_RS5_RS5<31, 224, (outs crbitrc:$BF),
+                             (ins g8rc:$rA, g8rc:$rB), "cmpeqb $BF, $rA, $rB",
+                             IIC_IntCompare, []>, Requires<[IsISA3_0]>;
 }
 
 let hasSideEffects = 0 in {
@@ -580,6 +600,9 @@ defm SRAD : XForm_6rc<31, 794, (outs g8rc:$rA), (ins g8rc:$rS, gprc:$rB),
 let Interpretation64Bit = 1, isCodeGenOnly = 1 in {
 defm CNTLZW8 : XForm_11r<31,  26, (outs g8rc:$rA), (ins g8rc:$rS),
                         "cntlzw", "$rA, $rS", IIC_IntGeneral, []>;
+defm CNTTZW8 : XForm_11r<31, 538, (outs g8rc:$rA), (ins g8rc:$rS),
+                        "cnttzw", "$rA, $rS", IIC_IntGeneral, []>,
+               Requires<[IsISA3_0]>;
 
 defm EXTSB8 : XForm_11r<31, 954, (outs g8rc:$rA), (ins g8rc:$rS),
                         "extsb", "$rA, $rS", IIC_IntSimple,
@@ -613,9 +636,12 @@ defm EXTSW_32_64 : XForm_11r<31, 986, (outs g8rc:$rA), (ins gprc:$rS),
 defm SRADI  : XSForm_1rc<31, 413, (outs g8rc:$rA), (ins g8rc:$rS, u6imm:$SH),
                          "sradi", "$rA, $rS, $SH", IIC_IntRotateDI,
                          [(set i64:$rA, (sra i64:$rS, (i32 imm:$SH)))]>, isPPC64;
-defm CNTLZD : XForm_11r<31, 58, (outs g8rc:$rA), (ins g8rc:$rS),
+defm CNTLZD : XForm_11r<31,  58, (outs g8rc:$rA), (ins g8rc:$rS),
                         "cntlzd", "$rA, $rS", IIC_IntGeneral,
                         [(set i64:$rA, (ctlz i64:$rS))]>;
+defm CNTTZD : XForm_11r<31, 570, (outs g8rc:$rA), (ins g8rc:$rS),
+                        "cnttzd", "$rA, $rS", IIC_IntGeneral,
+                        [(set i64:$rA, (cttz i64:$rS))]>, Requires<[IsISA3_0]>;
 def POPCNTD : XForm_11<31, 506, (outs g8rc:$rA), (ins g8rc:$rS),
                        "popcntd $rA, $rS", IIC_IntGeneral,
                        [(set i64:$rA, (ctpop i64:$rS))]>;
@@ -905,6 +931,10 @@ def LDUX : XForm_1<31, 53, (outs g8rc:$rD, ptr_rc_nor0:$ea_result),
                    "ldux $rD, $addr", IIC_LdStLDUX,
                    []>, RegConstraint<"$addr.ptrreg = $ea_result">,
                    NoEncode<"$ea_result">, isPPC64;
+
+def LDMX : XForm_1<31, 309, (outs g8rc:$rD), (ins memrr:$src),
+                   "ldmx $rD, $src", IIC_LdStLD, []>, isPPC64,
+           Requires<[IsISA3_0]>;
 }
 }
 
@@ -1246,3 +1276,24 @@ def : Pat<(atomic_load_64 xaddr:$src),  (LDX memrr:$src)>;
 
 def : Pat<(atomic_store_64 ixaddr:$ptr, i64:$val), (STD  g8rc:$val, memrix:$ptr)>;
 def : Pat<(atomic_store_64 xaddr:$ptr,  i64:$val), (STDX g8rc:$val, memrr:$ptr)>;
+
+let Predicates = [IsISA3_0] in {
+
+class X_L1_RA5_RB5<bits<6> opcode, bits<10> xo, string opc, RegisterOperand ty,
+                   InstrItinClass itin, list<dag> pattern>
+  : X_L1_RS5_RS5<opcode, xo, (outs), (ins ty:$rA, ty:$rB, u1imm:$L),
+                 !strconcat(opc, " $rA, $rB, $L"), itin, pattern>;
+
+let Interpretation64Bit = 1, isCodeGenOnly = 1 in {
+def CP_COPY8   : X_L1_RA5_RB5<31, 774, "copy"  , g8rc, IIC_LdStCOPY, []>;
+def CP_PASTE8  : X_L1_RA5_RB5<31, 902, "paste" , g8rc, IIC_LdStPASTE, []>;
+def CP_PASTE8o : X_L1_RA5_RB5<31, 902, "paste.", g8rc, IIC_LdStPASTE, []>,isDOT;
+}
+
+// SLB Invalidate Entry Global
+def SLBIEG : XForm_26<31, 466, (outs), (ins gprc:$RS, gprc:$RB),
+                      "slbieg $RS, $RB", IIC_SprSLBIEG, []>;
+// SLB Synchronize
+def SLBSYNC : XForm_0<31, 338, (outs), (ins), "slbsync", IIC_SprSLBSYNC, []>;
+
+} // IsISA3_0
diff --git a/lib/Target/PowerPC/PPCInstrAltivec.td b/lib/Target/PowerPC/PPCInstrAltivec.td
index 53674681b2137..e1c4673c2d7f5 100644
--- a/lib/Target/PowerPC/PPCInstrAltivec.td
+++ b/lib/Target/PowerPC/PPCInstrAltivec.td
@@ -1213,3 +1213,187 @@ def VNCIPHERLAST : VX1_Int_Ty<1353, "vncipherlast",
                               int_ppc_altivec_crypto_vncipherlast, v2i64>;
 def VSBOX : VXBX_Int_Ty<1480, "vsbox", int_ppc_altivec_crypto_vsbox, v2i64>;
 } // HasP8Crypto
+
+// The following altivec instructions were introduced in Power ISA 3.0
+def HasP9Altivec : Predicate<"PPCSubTarget->hasP9Altivec()">;
+let Predicates = [HasP9Altivec] in {
+
+// Vector Compare Not Equal (Zero)
+class P9VCMP<bits<10> xo, string asmstr, ValueType Ty>
+  : VXRForm_1<xo, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB), asmstr,
+              IIC_VecFPCompare, []>;
+class P9VCMPo<bits<10> xo, string asmstr, ValueType Ty>
+  : VXRForm_1<xo, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB), asmstr,
+              IIC_VecFPCompare, []> {
+  let Defs = [CR6];
+  let RC = 1;
+}
+
+// i8 element comparisons.
+def VCMPNEB   : P9VCMP <  7, "vcmpneb $vD, $vA, $vB"  , v16i8>;
+def VCMPNEBo  : P9VCMPo<  7, "vcmpneb. $vD, $vA, $vB" , v16i8>;
+def VCMPNEZB  : P9VCMP <263, "vcmpnezb $vD, $vA, $vB" , v16i8>;
+def VCMPNEZBo : P9VCMPo<263, "vcmpnezb. $vD, $vA, $vB", v16i8>;
+
+// i16 element comparisons.
+def VCMPNEH   : P9VCMP < 71, "vcmpneh $vD, $vA, $vB"  , v8i16>;
+def VCMPNEHo  : P9VCMPo< 71, "vcmpneh. $vD, $vA, $vB" , v8i16>;
+def VCMPNEZH  : P9VCMP <327, "vcmpnezh $vD, $vA, $vB" , v8i16>;
+def VCMPNEZHo : P9VCMPo<327, "vcmpnezh. $vD, $vA, $vB", v8i16>;
+
+// i32 element comparisons.
+def VCMPNEW   : P9VCMP <135, "vcmpnew $vD, $vA, $vB"  , v4i32>;
+def VCMPNEWo  : P9VCMPo<135, "vcmpnew. $vD, $vA, $vB" , v4i32>;
+def VCMPNEZW  : P9VCMP <391, "vcmpnezw $vD, $vA, $vB" , v4i32>;
+def VCMPNEZWo : P9VCMPo<391, "vcmpnezw. $vD, $vA, $vB", v4i32>;
+
+// VX-Form: [PO VRT / UIM VRB XO].
+// We use VXForm_1 to implement it, that is, we use "VRA" (5 bit) to represent
+// "/ UIM" (1 + 4 bit)
+class VX1_VT5_UIM5_VB5<bits<11> xo, string opc, list<dag> pattern>
+  : VXForm_1<xo, (outs vrrc:$vD), (ins u4imm:$UIMM, vrrc:$vB),
+             !strconcat(opc, " $vD, $vB, $UIMM"), IIC_VecGeneral, pattern>;
+
+class VX1_RT5_RA5_VB5<bits<11> xo, string opc, list<dag> pattern>
+  : VXForm_1<xo, (outs g8rc:$rD), (ins g8rc:$rA, vrrc:$vB),
+             !strconcat(opc, " $rD, $rA, $vB"), IIC_VecGeneral, pattern>;
+
+// Vector Extract Unsigned
+def VEXTRACTUB : VX1_VT5_UIM5_VB5<525, "vextractub", []>;
+def VEXTRACTUH : VX1_VT5_UIM5_VB5<589, "vextractuh", []>;
+def VEXTRACTUW : VX1_VT5_UIM5_VB5<653, "vextractuw", []>;
+def VEXTRACTD  : VX1_VT5_UIM5_VB5<717, "vextractd" , []>;
+
+// Vector Extract Unsigned Byte/Halfword/Word Left/Right-Indexed
+def VEXTUBLX : VX1_RT5_RA5_VB5<1549, "vextublx", []>;
+def VEXTUBRX : VX1_RT5_RA5_VB5<1805, "vextubrx", []>;
+def VEXTUHLX : VX1_RT5_RA5_VB5<1613, "vextuhlx", []>;
+def VEXTUHRX : VX1_RT5_RA5_VB5<1869, "vextuhrx", []>;
+def VEXTUWLX : VX1_RT5_RA5_VB5<1677, "vextuwlx", []>;
+def VEXTUWRX : VX1_RT5_RA5_VB5<1933, "vextuwrx", []>;
+
+// Vector Insert Element Instructions
+def VINSERTB : VX1_VT5_UIM5_VB5<781, "vinsertb", []>;
+def VINSERTH : VX1_VT5_UIM5_VB5<845, "vinserth", []>;
+def VINSERTW : VX1_VT5_UIM5_VB5<909, "vinsertw", []>;
+def VINSERTD : VX1_VT5_UIM5_VB5<973, "vinsertd", []>;
+
+class VX_VT5_EO5_VB5<bits<11> xo, bits<5> eo, string opc, list<dag> pattern>
+  : VXForm_RD5_XO5_RS5<xo, eo, (outs vrrc:$vD), (ins vrrc:$vB),
+                       !strconcat(opc, " $vD, $vB"), IIC_VecGeneral, pattern>;
+
+// Vector Count Leading/Trailing Zero LSB. Result is placed into GPR[rD]
+def VCLZLSBB : VXForm_RD5_XO5_RS5<1538, 0, (outs g8rc:$rD), (ins vrrc:$vB),
+                                  "vclzlsbb $rD, $vB", IIC_VecGeneral, []>;
+def VCTZLSBB : VXForm_RD5_XO5_RS5<1538, 1, (outs g8rc:$rD), (ins vrrc:$vB),
+                                  "vctzlsbb $rD, $vB", IIC_VecGeneral, []>;
+// Vector Count Trailing Zeros
+def VCTZB : VX_VT5_EO5_VB5<1538, 28, "vctzb", []>;
+def VCTZH : VX_VT5_EO5_VB5<1538, 29, "vctzh", []>;
+def VCTZW : VX_VT5_EO5_VB5<1538, 30, "vctzw", []>;
+def VCTZD : VX_VT5_EO5_VB5<1538, 31, "vctzd", []>;
+
+// Vector Extend Sign
+def VEXTSB2W : VX_VT5_EO5_VB5<1538, 16, "vextsb2w", []>;
+def VEXTSH2W : VX_VT5_EO5_VB5<1538, 17, "vextsh2w", []>;
+def VEXTSB2D : VX_VT5_EO5_VB5<1538, 24, "vextsb2d", []>;
+def VEXTSH2D : VX_VT5_EO5_VB5<1538, 25, "vextsh2d", []>;
+def VEXTSW2D : VX_VT5_EO5_VB5<1538, 26, "vextsw2d", []>;
+
+// Vector Integer Negate
+def VNEGW : VX_VT5_EO5_VB5<1538, 6, "vnegw", []>;
+def VNEGD : VX_VT5_EO5_VB5<1538, 7, "vnegd", []>;
+
+// Vector Parity Byte
+def VPRTYBW : VX_VT5_EO5_VB5<1538,  8, "vprtybw", []>;
+def VPRTYBD : VX_VT5_EO5_VB5<1538,  9, "vprtybd", []>;
+def VPRTYBQ : VX_VT5_EO5_VB5<1538, 10, "vprtybq", []>;
+
+// Vector (Bit) Permute (Right-indexed)
+def VBPERMD : VXForm_1<1484, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+                       "vbpermd $vD, $vA, $vB", IIC_VecFP, []>;
+def VPERMR : VAForm_1a<59, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB, vrrc:$vC),
+                       "vpermr $vD, $vA, $vB, $vC", IIC_VecFP, []>;
+
+class VX1_VT5_VA5_VB5<bits<11> xo, string opc, list<dag> pattern>
+  : VXForm_1<xo, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+             !strconcat(opc, " $vD, $vA, $vB"), IIC_VecFP, pattern>;
+
+// Vector Rotate Left Mask/Mask-Insert
+def VRLWNM : VX1_VT5_VA5_VB5<389, "vrlwnm", []>;
+def VRLWMI : VX1_VT5_VA5_VB5<133, "vrlwmi", []>;
+def VRLDNM : VX1_VT5_VA5_VB5<453, "vrldnm", []>;
+def VRLDMI : VX1_VT5_VA5_VB5<197, "vrldmi", []>;
+
+// Vector Shift Left/Right
+def VSLV : VX1_VT5_VA5_VB5<1860, "vslv", []>;
+def VSRV : VX1_VT5_VA5_VB5<1796, "vsrv", []>;
+
+// Vector Multiply-by-10 (& Write Carry) Unsigned Quadword
+def VMUL10UQ   : VXForm_BX<513, (outs vrrc:$vD), (ins vrrc:$vA),
+                           "vmul10uq $vD, $vA", IIC_VecFP, []>;
+def VMUL10CUQ  : VXForm_BX<  1, (outs vrrc:$vD), (ins vrrc:$vA),
+                           "vmul10cuq $vD, $vA", IIC_VecFP, []>;
+
+// Vector Multiply-by-10 Extended (& Write Carry) Unsigned Quadword
+def VMUL10EUQ  : VX1_VT5_VA5_VB5<577, "vmul10euq" , []>;
+def VMUL10ECUQ : VX1_VT5_VA5_VB5< 65, "vmul10ecuq", []>;
+
+// Decimal Integer Format Conversion Instructions
+
+// [PO VRT EO VRB 1 PS XO], "_o" means CR6 is set.
+class VX_VT5_EO5_VB5_PS1_XO9_o<bits<5> eo, bits<9> xo, string opc,
+                               list<dag> pattern>
+  : VX_RD5_EO5_RS5_PS1_XO9<eo, xo, (outs vrrc:$vD), (ins vrrc:$vB, u1imm:$PS),
+                        !strconcat(opc, " $vD, $vB, $PS"), IIC_VecFP, pattern> {
+  let Defs = [CR6];
+}
+
+// [PO VRT EO VRB 1 / XO]
+class VX_VT5_EO5_VB5_XO9_o<bits<5> eo, bits<9> xo, string opc,
+                           list<dag> pattern>
+  : VX_RD5_EO5_RS5_PS1_XO9<eo, xo, (outs vrrc:$vD), (ins vrrc:$vB),
+                           !strconcat(opc, " $vD, $vB"), IIC_VecFP, pattern> {
+  let Defs = [CR6];
+  let PS = 0;
+}
+
+// Decimal Convert From/to National/Zoned/Signed-QWord
+def BCDCFNo  : VX_VT5_EO5_VB5_PS1_XO9_o<7, 385, "bcdcfn." , []>;
+def BCDCFZo  : VX_VT5_EO5_VB5_PS1_XO9_o<6, 385, "bcdcfz." , []>;
+def BCDCTNo  : VX_VT5_EO5_VB5_XO9_o    <5, 385, "bcdctn." , []>;
+def BCDCTZo  : VX_VT5_EO5_VB5_PS1_XO9_o<4, 385, "bcdctz." , []>;
+def BCDCFSQo : VX_VT5_EO5_VB5_PS1_XO9_o<2, 385, "bcdcfsq.", []>;
+def BCDCTSQo : VX_VT5_EO5_VB5_XO9_o    <0, 385, "bcdctsq.", []>;
+
+// Decimal Copy-Sign/Set-Sign
+let Defs = [CR6] in
+def BCDCPSGNo : VX1_VT5_VA5_VB5<833, "bcdcpsgn.", []>;
+
+def BCDSETSGNo : VX_VT5_EO5_VB5_PS1_XO9_o<31, 385, "bcdsetsgn.", []>;
+
+// [PO VRT VRA VRB 1 PS XO], "_o" means CR6 is set.
+class VX_VT5_VA5_VB5_PS1_XO9_o<bits<9> xo, string opc, list<dag> pattern>
+  : VX_RD5_RSp5_PS1_XO9<xo,
+                   (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB, u1imm:$PS),
+                   !strconcat(opc, " $vD, $vA, $vB, $PS"), IIC_VecFP, pattern> {
+  let Defs = [CR6];
+}
+
+// [PO VRT VRA VRB 1 / XO]
+class VX_VT5_VA5_VB5_XO9_o<bits<9> xo, string opc, list<dag> pattern>
+  : VX_RD5_RSp5_PS1_XO9<xo, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+                        !strconcat(opc, " $vD, $vA, $vB"), IIC_VecFP, pattern> {
+  let Defs = [CR6];
+  let PS = 0;
+}
+
+// Decimal Shift/Unsigned-Shift/Shift-and-Round
+def BCDSo :  VX_VT5_VA5_VB5_PS1_XO9_o<193, "bcds." , []>;
+def BCDUSo : VX_VT5_VA5_VB5_XO9_o    <129, "bcdus.", []>;
+def BCDSRo : VX_VT5_VA5_VB5_PS1_XO9_o<449, "bcdsr.", []>;
+
+// Decimal (Unsigned) Truncate
+def BCDTRUNCo :  VX_VT5_VA5_VB5_PS1_XO9_o<257, "bcdtrunc." , []>;
+def BCDUTRUNCo : VX_VT5_VA5_VB5_XO9_o    <321, "bcdutrunc.", []>;
+} // end HasP9Altivec
diff --git a/lib/Target/PowerPC/PPCInstrFormats.td b/lib/Target/PowerPC/PPCInstrFormats.td
index 4e03ed27653f5..5acff75595445 100644
--- a/lib/Target/PowerPC/PPCInstrFormats.td
+++ b/lib/Target/PowerPC/PPCInstrFormats.td
@@ -360,6 +360,21 @@ class DSForm_1<bits<6> opcode, bits<2> xo, dag OOL, dag IOL, string asmstr,
   let Inst{30-31} = xo;
 }
 
+// DQ-Form: [PO T RA DQ TX XO] or [PO S RA DQ SX XO]
+class DQ_RD6_RS5_DQ12<bits<6> opcode, bits<3> xo, dag OOL, dag IOL,
+                      string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<6>  XT;
+  bits<17> DS_RA;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = XT{4-0};
+  let Inst{11-15} = DS_RA{16-12};  // Register #
+  let Inst{16-27} = DS_RA{11-0};   // Displacement.
+  let Inst{28}    = XT{5};
+  let Inst{29-31} = xo;
+}
 
 // 1.7.6 X-Form
 class XForm_base_r3xo<bits<6> opcode, bits<10> xo, dag OOL, dag IOL, string asmstr, 
@@ -747,6 +762,107 @@ class XForm_htm3<bits<6> opcode, bits<10> xo, dag OOL, dag IOL, string asmstr,
   let Inst{31}    = RC;
 }
 
+// [PO RT RA RB XO /]
+class X_BF3_L1_RS5_RS5<bits<6> opcode, bits<10> xo, dag OOL, dag IOL,
+                       string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<3> BF;
+  bits<1> L;
+  bits<5> RA;
+  bits<5> RB;
+
+  let Pattern = pattern;
+
+  let Inst{6-8}   = BF;
+  let Inst{9}     = 0;
+  let Inst{10}    = L;
+  let Inst{11-15} = RA;
+  let Inst{16-20} = RB;
+  let Inst{21-30} = xo;
+  let Inst{31}    = 0;
+}
+
+// Same as XForm_17 but with GPR's and new naming convention
+class X_BF3_RS5_RS5<bits<6> opcode, bits<10> xo, dag OOL, dag IOL,
+                    string asmstr, InstrItinClass itin, list<dag> pattern>
+         : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<3> BF;
+  bits<5> RA;
+  bits<5> RB;
+
+  let Pattern = pattern;
+
+  let Inst{6-8}   = BF;
+  let Inst{9-10}  = 0;
+  let Inst{11-15} = RA;
+  let Inst{16-20} = RB;
+  let Inst{21-30} = xo;
+  let Inst{31}    = 0;
+}
+
+// e.g. [PO VRT XO VRB XO /] or [PO VRT XO VRB XO RO]
+class X_RD5_XO5_RS5<bits<6> opcode, bits<5> xo2, bits<10> xo, dag OOL, dag IOL,
+                    string asmstr, InstrItinClass itin, list<dag> pattern>
+  : XForm_base_r3xo<opcode, xo, OOL, IOL, asmstr, itin, pattern> {
+  let A = xo2;
+}
+
+class X_BF3_DCMX7_RS5<bits<6> opcode, bits<10> xo, dag OOL, dag IOL,
+                      string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<3> BF;
+  bits<7> DCMX;
+  bits<5> VB;
+
+  let Pattern = pattern;
+
+  let Inst{6-8}  = BF;
+  let Inst{9-15} = DCMX;
+  let Inst{16-20} = VB;
+  let Inst{21-30} = xo;
+  let Inst{31}    = 0;
+}
+
+class X_RD6_IMM8<bits<6> opcode, bits<10> xo, dag OOL, dag IOL,
+                 string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<6> XT;
+  bits<8> IMM8;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = XT{4-0};
+  let Inst{11-12} = 0;
+  let Inst{13-20} = IMM8;
+  let Inst{21-30} = xo;
+  let Inst{31}    = XT{5};
+}
+
+// XForm_base_r3xo for instructions such as P9 atomics where we don't want
+// to specify an SDAG pattern for matching.
+class X_RD5_RS5_IM5<bits<6> opcode, bits<10> xo, dag OOL, dag IOL,
+                    string asmstr, InstrItinClass itin>
+  : XForm_base_r3xo<opcode, xo, OOL, IOL, asmstr, itin, []> {
+}
+
+class X_BF3<bits<6> opcode, bits<10> xo, dag OOL, dag IOL, string asmstr,
+            InstrItinClass itin>
+  : XForm_17<opcode, xo, OOL, IOL, asmstr, itin> {
+  let FRA = 0;
+  let FRB = 0;
+}
+
+// [PO /// L RA RB XO /]
+class X_L1_RS5_RS5<bits<6> opcode, bits<10> xo, dag OOL, dag IOL,
+                   string asmstr, InstrItinClass itin, list<dag> pattern>
+  : XForm_16<opcode, xo, OOL, IOL, asmstr, itin> {
+  let BF = 0;
+  let Pattern = pattern;
+
+  bit RC = 0;
+  let Inst{31} = RC;
+}
+
 // XX*-Form (VSX)
 class XX1Form<bits<6> opcode, bits<10> xo, dag OOL, dag IOL, string asmstr, 
               InstrItinClass itin, list<dag> pattern>
@@ -820,6 +936,95 @@ class XX2Form_2<bits<6> opcode, bits<9> xo, dag OOL, dag IOL, string asmstr,
   let Inst{31}    = XT{5};
 }
 
+class XX2_RD6_UIM5_RS6<bits<6> opcode, bits<9> xo, dag OOL, dag IOL,
+                       string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<6> XT;
+  bits<6> XB;
+  bits<5> UIM5;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = XT{4-0};
+  let Inst{11-15} = UIM5;
+  let Inst{16-20} = XB{4-0};
+  let Inst{21-29} = xo;
+  let Inst{30}    = XB{5};
+  let Inst{31}    = XT{5};
+}
+
+// [PO T XO B XO BX /]
+class XX2_RD5_XO5_RS6<bits<6> opcode, bits<5> xo2, bits<9> xo, dag OOL, dag IOL,
+                       string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<5> RT;
+  bits<6> XB;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = RT;
+  let Inst{11-15} = xo2;
+  let Inst{16-20} = XB{4-0};
+  let Inst{21-29} = xo;
+  let Inst{30}    = XB{5};
+  let Inst{31}    = 0;
+}
+
+// [PO T XO B XO BX TX]
+class XX2_RD6_XO5_RS6<bits<6> opcode, bits<5> xo2, bits<9> xo, dag OOL, dag IOL,
+                      string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<6> XT;
+  bits<6> XB;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = XT{4-0};
+  let Inst{11-15} = xo2;
+  let Inst{16-20} = XB{4-0};
+  let Inst{21-29} = xo;
+  let Inst{30}    = XB{5};
+  let Inst{31}    = XT{5};
+}
+
+class XX2_BF3_DCMX7_RS6<bits<6> opcode, bits<9> xo, dag OOL, dag IOL,
+                      string asmstr, InstrItinClass itin, list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<3> BF;
+  bits<7> DCMX;
+  bits<6> XB;
+
+  let Pattern = pattern;
+
+  let Inst{6-8}  = BF;
+  let Inst{9-15} = DCMX;
+  let Inst{16-20} = XB{4-0};
+  let Inst{21-29} = xo;
+  let Inst{30}    = XB{5};
+  let Inst{31}    = 0;
+}
+
+class XX2_RD6_DCMX7_RS6<bits<6> opcode, bits<4> xo1, bits<3> xo2,
+                        dag OOL, dag IOL, string asmstr, InstrItinClass itin,
+                        list<dag> pattern>
+  : I<opcode, OOL, IOL, asmstr, itin> {
+  bits<6> XT;
+  bits<7> DCMX;
+  bits<6> XB;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = XT{4-0};
+  let Inst{11-15} = DCMX{4-0};
+  let Inst{16-20} = XB{4-0};
+  let Inst{21-24} = xo1;
+  let Inst{25}    = DCMX{5};
+  let Inst{26-28} = xo2;
+  let Inst{29}    = DCMX{6};
+  let Inst{30}    = XB{5};
+  let Inst{31}    = XT{5};
+}
+
 class XX3Form<bits<6> opcode, bits<8> xo, dag OOL, dag IOL, string asmstr, 
               InstrItinClass itin, list<dag> pattern>
   : I<opcode, OOL, IOL, asmstr, itin> {
@@ -1571,6 +1776,21 @@ class VXForm_5<bits<11> xo, dag OOL, dag IOL, string asmstr,
   let Inst{21-31} = xo;
 }
 
+// e.g. [PO VRT EO VRB XO]
+class VXForm_RD5_XO5_RS5<bits<11> xo, bits<5> eo, dag OOL, dag IOL,
+                         string asmstr, InstrItinClass itin, list<dag> pattern>
+    : I<4, OOL, IOL, asmstr, itin> {
+  bits<5> RD;
+  bits<5> VB;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = RD;
+  let Inst{11-15} = eo;
+  let Inst{16-20} = VB;
+  let Inst{21-31} = xo;
+}
+
 /// VXForm_CR - VX crypto instructions with "VRT, VRA, ST, SIX"
 class VXForm_CR<bits<11> xo, dag OOL, dag IOL, string asmstr,
                InstrItinClass itin, list<dag> pattern>
@@ -1622,6 +1842,44 @@ class VXRForm_1<bits<10> xo, dag OOL, dag IOL, string asmstr,
   let Inst{22-31} = xo;
 }
 
+// VX-Form: [PO VRT EO VRB 1 PS XO]
+class VX_RD5_EO5_RS5_PS1_XO9<bits<5> eo, bits<9> xo,
+                             dag OOL, dag IOL, string asmstr,
+                             InstrItinClass itin, list<dag> pattern>
+  : I<4, OOL, IOL, asmstr, itin> {
+  bits<5> VD;
+  bits<5> VB;
+  bit PS;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = VD;
+  let Inst{11-15} = eo;
+  let Inst{16-20} = VB;
+  let Inst{21}    = 1;
+  let Inst{22}    = PS;
+  let Inst{23-31} = xo;
+}
+
+// VX-Form: [PO VRT VRA VRB 1 PS XO] or [PO VRT VRA VRB 1 / XO]
+class VX_RD5_RSp5_PS1_XO9<bits<9> xo, dag OOL, dag IOL, string asmstr,
+                          InstrItinClass itin, list<dag> pattern>
+  : I<4, OOL, IOL, asmstr, itin> {
+  bits<5> VD;
+  bits<5> VA;
+  bits<5> VB;
+  bit PS;
+
+  let Pattern = pattern;
+
+  let Inst{6-10}  = VD;
+  let Inst{11-15} = VA;
+  let Inst{16-20} = VB;
+  let Inst{21}    = 1;
+  let Inst{22}    = PS;
+  let Inst{23-31} = xo;
+}
+
 // Z23-Form (used by QPX)
 class Z23Form_1<bits<6> opcode, bits<8> xo, dag OOL, dag IOL, string asmstr, 
               InstrItinClass itin, list<dag> pattern>
diff --git a/lib/Target/PowerPC/PPCInstrInfo.cpp b/lib/Target/PowerPC/PPCInstrInfo.cpp
index dcff6ad2486f8..b6ae70ec1a2dd 100644
--- a/lib/Target/PowerPC/PPCInstrInfo.cpp
+++ b/lib/Target/PowerPC/PPCInstrInfo.cpp
@@ -93,6 +93,7 @@ PPCInstrInfo::CreateTargetPostRAHazardRecognizer(const InstrItineraryData *II,
   unsigned Directive =
       DAG->MF.getSubtarget<PPCSubtarget>().getDarwinDirective();
 
+  // FIXME: Leaving this as-is until we have POWER9 scheduling info
   if (Directive == PPC::DIR_PWR7 || Directive == PPC::DIR_PWR8)
     return new PPCDispatchGroupSBHazardRecognizer(II, DAG);
 
@@ -108,7 +109,7 @@ PPCInstrInfo::CreateTargetPostRAHazardRecognizer(const InstrItineraryData *II,
 }
 
 unsigned PPCInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
-                                       const MachineInstr *MI,
+                                       const MachineInstr &MI,
                                        unsigned *PredCost) const {
   if (!ItinData || UseOldLatencyCalc)
     return PPCGenInstrInfo::getInstrLatency(ItinData, MI, PredCost);
@@ -121,9 +122,9 @@ unsigned PPCInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
   // is an output).
 
   unsigned Latency = 1;
-  unsigned DefClass = MI->getDesc().getSchedClass();
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = MI->getOperand(i);
+  unsigned DefClass = MI.getDesc().getSchedClass();
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI.getOperand(i);
     if (!MO.isReg() || !MO.isDef() || MO.isImplicit())
       continue;
 
@@ -138,22 +139,22 @@ unsigned PPCInstrInfo::getInstrLatency(const InstrItineraryData *ItinData,
 }
 
 int PPCInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
-                                    const MachineInstr *DefMI, unsigned DefIdx,
-                                    const MachineInstr *UseMI,
+                                    const MachineInstr &DefMI, unsigned DefIdx,
+                                    const MachineInstr &UseMI,
                                     unsigned UseIdx) const {
   int Latency = PPCGenInstrInfo::getOperandLatency(ItinData, DefMI, DefIdx,
                                                    UseMI, UseIdx);
 
-  if (!DefMI->getParent())
+  if (!DefMI.getParent())
     return Latency;
 
-  const MachineOperand &DefMO = DefMI->getOperand(DefIdx);
+  const MachineOperand &DefMO = DefMI.getOperand(DefIdx);
   unsigned Reg = DefMO.getReg();
 
   bool IsRegCR;
   if (TargetRegisterInfo::isVirtualRegister(Reg)) {
     const MachineRegisterInfo *MRI =
-      &DefMI->getParent()->getParent()->getRegInfo();
+        &DefMI.getParent()->getParent()->getRegInfo();
     IsRegCR = MRI->getRegClass(Reg)->hasSuperClassEq(&PPC::CRRCRegClass) ||
               MRI->getRegClass(Reg)->hasSuperClassEq(&PPC::CRBITRCRegClass);
   } else {
@@ -161,7 +162,7 @@ int PPCInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
               PPC::CRBITRCRegClass.contains(Reg);
   }
 
-  if (UseMI->isBranch() && IsRegCR) {
+  if (UseMI.isBranch() && IsRegCR) {
     if (Latency < 0)
       Latency = getInstrLatency(ItinData, DefMI);
 
@@ -181,6 +182,7 @@ int PPCInstrInfo::getOperandLatency(const InstrItineraryData *ItinData,
     case PPC::DIR_PWR6X:
     case PPC::DIR_PWR7:
     case PPC::DIR_PWR8:
+    // FIXME: Is this needed for POWER9?
       Latency += 2;
       break;
     }
@@ -258,10 +260,10 @@ bool PPCInstrInfo::isCoalescableExtInstr(const MachineInstr &MI,
   }
 }
 
-unsigned PPCInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
+unsigned PPCInstrInfo::isLoadFromStackSlot(const MachineInstr &MI,
                                            int &FrameIndex) const {
   // Note: This list must be kept consistent with LoadRegFromStackSlot.
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case PPC::LD:
   case PPC::LWZ:
@@ -277,20 +279,20 @@ unsigned PPCInstrInfo::isLoadFromStackSlot(const MachineInstr *MI,
   case PPC::RESTORE_VRSAVE:
     // Check for the operands added by addFrameReference (the immediate is the
     // offset which defaults to 0).
-    if (MI->getOperand(1).isImm() && !MI->getOperand(1).getImm() &&
-        MI->getOperand(2).isFI()) {
-      FrameIndex = MI->getOperand(2).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isImm() && !MI.getOperand(1).getImm() &&
+        MI.getOperand(2).isFI()) {
+      FrameIndex = MI.getOperand(2).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
   return 0;
 }
 
-unsigned PPCInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
+unsigned PPCInstrInfo::isStoreToStackSlot(const MachineInstr &MI,
                                           int &FrameIndex) const {
   // Note: This list must be kept consistent with StoreRegToStackSlot.
-  switch (MI->getOpcode()) {
+  switch (MI.getOpcode()) {
   default: break;
   case PPC::STD:
   case PPC::STW:
@@ -306,25 +308,23 @@ unsigned PPCInstrInfo::isStoreToStackSlot(const MachineInstr *MI,
   case PPC::SPILL_VRSAVE:
     // Check for the operands added by addFrameReference (the immediate is the
     // offset which defaults to 0).
-    if (MI->getOperand(1).isImm() && !MI->getOperand(1).getImm() &&
-        MI->getOperand(2).isFI()) {
-      FrameIndex = MI->getOperand(2).getIndex();
-      return MI->getOperand(0).getReg();
+    if (MI.getOperand(1).isImm() && !MI.getOperand(1).getImm() &&
+        MI.getOperand(2).isFI()) {
+      FrameIndex = MI.getOperand(2).getIndex();
+      return MI.getOperand(0).getReg();
     }
     break;
   }
   return 0;
 }
 
-MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr *MI,
-                                                   bool NewMI,
+MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                                    unsigned OpIdx1,
                                                    unsigned OpIdx2) const {
-  MachineFunction &MF = *MI->getParent()->getParent();
+  MachineFunction &MF = *MI.getParent()->getParent();
 
   // Normal instructions can be commuted the obvious way.
-  if (MI->getOpcode() != PPC::RLWIMI &&
-      MI->getOpcode() != PPC::RLWIMIo)
+  if (MI.getOpcode() != PPC::RLWIMI && MI.getOpcode() != PPC::RLWIMIo)
     return TargetInstrInfo::commuteInstructionImpl(MI, NewMI, OpIdx1, OpIdx2);
   // Note that RLWIMI can be commuted as a 32-bit instruction, but not as a
   // 64-bit instruction (so we don't handle PPC::RLWIMI8 here), because
@@ -332,7 +332,7 @@ MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr *MI,
   // to the high-bits of the mask (and, thus, the result).
 
   // Cannot commute if it has a non-zero rotate count.
-  if (MI->getOperand(3).getImm() != 0)
+  if (MI.getOperand(3).getImm() != 0)
     return nullptr;
 
   // If we have a zero rotate count, we have:
@@ -345,28 +345,28 @@ MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr *MI,
   // Swap op1/op2
   assert(((OpIdx1 == 1 && OpIdx2 == 2) || (OpIdx1 == 2 && OpIdx2 == 1)) &&
          "Only the operands 1 and 2 can be swapped in RLSIMI/RLWIMIo.");
-  unsigned Reg0 = MI->getOperand(0).getReg();
-  unsigned Reg1 = MI->getOperand(1).getReg();
-  unsigned Reg2 = MI->getOperand(2).getReg();
-  unsigned SubReg1 = MI->getOperand(1).getSubReg();
-  unsigned SubReg2 = MI->getOperand(2).getSubReg();
-  bool Reg1IsKill = MI->getOperand(1).isKill();
-  bool Reg2IsKill = MI->getOperand(2).isKill();
+  unsigned Reg0 = MI.getOperand(0).getReg();
+  unsigned Reg1 = MI.getOperand(1).getReg();
+  unsigned Reg2 = MI.getOperand(2).getReg();
+  unsigned SubReg1 = MI.getOperand(1).getSubReg();
+  unsigned SubReg2 = MI.getOperand(2).getSubReg();
+  bool Reg1IsKill = MI.getOperand(1).isKill();
+  bool Reg2IsKill = MI.getOperand(2).isKill();
   bool ChangeReg0 = false;
   // If machine instrs are no longer in two-address forms, update
   // destination register as well.
   if (Reg0 == Reg1) {
     // Must be two address instruction!
-    assert(MI->getDesc().getOperandConstraint(0, MCOI::TIED_TO) &&
+    assert(MI.getDesc().getOperandConstraint(0, MCOI::TIED_TO) &&
            "Expecting a two-address instruction!");
-    assert(MI->getOperand(0).getSubReg() == SubReg1 && "Tied subreg mismatch");
+    assert(MI.getOperand(0).getSubReg() == SubReg1 && "Tied subreg mismatch");
     Reg2IsKill = false;
     ChangeReg0 = true;
   }
 
   // Masks.
-  unsigned MB = MI->getOperand(4).getImm();
-  unsigned ME = MI->getOperand(5).getImm();
+  unsigned MB = MI.getOperand(4).getImm();
+  unsigned ME = MI.getOperand(5).getImm();
 
   // We can't commute a trivial mask (there is no way to represent an all-zero
   // mask).
@@ -375,40 +375,40 @@ MachineInstr *PPCInstrInfo::commuteInstructionImpl(MachineInstr *MI,
 
   if (NewMI) {
     // Create a new instruction.
-    unsigned Reg0 = ChangeReg0 ? Reg2 : MI->getOperand(0).getReg();
-    bool Reg0IsDead = MI->getOperand(0).isDead();
-    return BuildMI(MF, MI->getDebugLoc(), MI->getDesc())
-      .addReg(Reg0, RegState::Define | getDeadRegState(Reg0IsDead))
-      .addReg(Reg2, getKillRegState(Reg2IsKill))
-      .addReg(Reg1, getKillRegState(Reg1IsKill))
-      .addImm((ME+1) & 31)
-      .addImm((MB-1) & 31);
+    unsigned Reg0 = ChangeReg0 ? Reg2 : MI.getOperand(0).getReg();
+    bool Reg0IsDead = MI.getOperand(0).isDead();
+    return BuildMI(MF, MI.getDebugLoc(), MI.getDesc())
+        .addReg(Reg0, RegState::Define | getDeadRegState(Reg0IsDead))
+        .addReg(Reg2, getKillRegState(Reg2IsKill))
+        .addReg(Reg1, getKillRegState(Reg1IsKill))
+        .addImm((ME + 1) & 31)
+        .addImm((MB - 1) & 31);
   }
 
   if (ChangeReg0) {
-    MI->getOperand(0).setReg(Reg2);
-    MI->getOperand(0).setSubReg(SubReg2);
+    MI.getOperand(0).setReg(Reg2);
+    MI.getOperand(0).setSubReg(SubReg2);
   }
-  MI->getOperand(2).setReg(Reg1);
-  MI->getOperand(1).setReg(Reg2);
-  MI->getOperand(2).setSubReg(SubReg1);
-  MI->getOperand(1).setSubReg(SubReg2);
-  MI->getOperand(2).setIsKill(Reg1IsKill);
-  MI->getOperand(1).setIsKill(Reg2IsKill);
+  MI.getOperand(2).setReg(Reg1);
+  MI.getOperand(1).setReg(Reg2);
+  MI.getOperand(2).setSubReg(SubReg1);
+  MI.getOperand(1).setSubReg(SubReg2);
+  MI.getOperand(2).setIsKill(Reg1IsKill);
+  MI.getOperand(1).setIsKill(Reg2IsKill);
 
   // Swap the mask around.
-  MI->getOperand(4).setImm((ME+1) & 31);
-  MI->getOperand(5).setImm((MB-1) & 31);
-  return MI;
+  MI.getOperand(4).setImm((ME + 1) & 31);
+  MI.getOperand(5).setImm((MB - 1) & 31);
+  return &MI;
 }
 
-bool PPCInstrInfo::findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1,
+bool PPCInstrInfo::findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                                          unsigned &SrcOpIdx2) const {
   // For VSX A-Type FMA instructions, it is the first two operands that can be
   // commuted, however, because the non-encoded tied input operand is listed
   // first, the operands to swap are actually the second and third.
 
-  int AltOpc = PPC::getAltVSXFMAOpcode(MI->getOpcode());
+  int AltOpc = PPC::getAltVSXFMAOpcode(MI.getOpcode());
   if (AltOpc == -1)
     return TargetInstrInfo::findCommutedOpIndices(MI, SrcOpIdx1, SrcOpIdx2);
 
@@ -428,6 +428,8 @@ void PPCInstrInfo::insertNoop(MachineBasicBlock &MBB,
   case PPC::DIR_PWR6: Opcode = PPC::NOP_GT_PWR6; break;
   case PPC::DIR_PWR7: Opcode = PPC::NOP_GT_PWR7; break;
   case PPC::DIR_PWR8: Opcode = PPC::NOP_GT_PWR7; break; /* FIXME: Update when P8 InstrScheduling model is ready */
+  // FIXME: Update when POWER9 scheduling model is ready.
+  case PPC::DIR_PWR9: Opcode = PPC::NOP_GT_PWR7; break;
   }
 
   DebugLoc DL;
@@ -442,7 +444,8 @@ void PPCInstrInfo::getNoopForMachoTarget(MCInst &NopInst) const {
 // Branch analysis.
 // Note: If the condition register is set to CTR or CTR8 then this is a
 // BDNZ (imm == 1) or BDZ (imm == 0) branch.
-bool PPCInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
+bool PPCInstrInfo::analyzeBranch(MachineBasicBlock &MBB,
+                                 MachineBasicBlock *&TBB,
                                  MachineBasicBlock *&FBB,
                                  SmallVectorImpl<MachineOperand> &Cond,
                                  bool AllowModify) const {
@@ -453,14 +456,14 @@ bool PPCInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
   if (I == MBB.end())
     return false;
 
-  if (!isUnpredicatedTerminator(I))
+  if (!isUnpredicatedTerminator(*I))
     return false;
 
   // Get the last instruction in the block.
   MachineInstr *LastInst = I;
 
   // If there is only one terminator instruction, process it.
-  if (I == MBB.begin() || !isUnpredicatedTerminator(--I)) {
+  if (I == MBB.begin() || !isUnpredicatedTerminator(*--I)) {
     if (LastInst->getOpcode() == PPC::B) {
       if (!LastInst->getOperand(0).isMBB())
         return true;
@@ -522,8 +525,7 @@ bool PPCInstrInfo::AnalyzeBranch(MachineBasicBlock &MBB,MachineBasicBlock *&TBB,
   MachineInstr *SecondLastInst = I;
 
   // If there are three terminators, we don't know what sort of block this is.
-  if (SecondLastInst && I != MBB.begin() &&
-      isUnpredicatedTerminator(--I))
+  if (SecondLastInst && I != MBB.begin() && isUnpredicatedTerminator(*--I))
     return true;
 
   // If the block ends with PPC::B and PPC:BCC, handle it.
@@ -633,11 +635,11 @@ unsigned PPCInstrInfo::RemoveBranch(MachineBasicBlock &MBB) const {
   return 2;
 }
 
-unsigned
-PPCInstrInfo::InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
-                           MachineBasicBlock *FBB,
-                           ArrayRef<MachineOperand> Cond,
-                           DebugLoc DL) const {
+unsigned PPCInstrInfo::InsertBranch(MachineBasicBlock &MBB,
+                                    MachineBasicBlock *TBB,
+                                    MachineBasicBlock *FBB,
+                                    ArrayRef<MachineOperand> Cond,
+                                    const DebugLoc &DL) const {
   // Shouldn't be a fall through.
   assert(TBB && "InsertBranch must not be told to insert a fallthrough");
   assert((Cond.size() == 2 || Cond.size() == 0) &&
@@ -721,9 +723,10 @@ bool PPCInstrInfo::canInsertSelect(const MachineBasicBlock &MBB,
 }
 
 void PPCInstrInfo::insertSelect(MachineBasicBlock &MBB,
-                                MachineBasicBlock::iterator MI, DebugLoc dl,
-                                unsigned DestReg, ArrayRef<MachineOperand> Cond,
-                                unsigned TrueReg, unsigned FalseReg) const {
+                                MachineBasicBlock::iterator MI,
+                                const DebugLoc &dl, unsigned DestReg,
+                                ArrayRef<MachineOperand> Cond, unsigned TrueReg,
+                                unsigned FalseReg) const {
   assert(Cond.size() == 2 &&
          "PPC branch conditions have two components!");
 
@@ -746,8 +749,8 @@ void PPCInstrInfo::insertSelect(MachineBasicBlock &MBB,
   unsigned OpCode = Is64Bit ? PPC::ISEL8 : PPC::ISEL;
   auto SelectPred = static_cast<PPC::Predicate>(Cond[0].getImm());
 
-  unsigned SubIdx;
-  bool SwapOps;
+  unsigned SubIdx = 0;
+  bool SwapOps = false;
   switch (SelectPred) {
   case PPC::PRED_EQ:
   case PPC::PRED_EQ_MINUS:
@@ -835,9 +838,9 @@ static unsigned getCRBitValue(unsigned CRBit) {
 }
 
 void PPCInstrInfo::copyPhysReg(MachineBasicBlock &MBB,
-                               MachineBasicBlock::iterator I, DebugLoc DL,
-                               unsigned DestReg, unsigned SrcReg,
-                               bool KillSrc) const {
+                               MachineBasicBlock::iterator I,
+                               const DebugLoc &DL, unsigned DestReg,
+                               unsigned SrcReg, bool KillSrc) const {
   // We can end up with self copies and similar things as a result of VSX copy
   // legalization. Promote them here.
   const TargetRegisterInfo *TRI = &getRegisterInfo();
@@ -883,8 +886,8 @@ void PPCInstrInfo::copyPhysReg(MachineBasicBlock &MBB,
   if (PPC::CRBITRCRegClass.contains(SrcReg) &&
       PPC::GPRCRegClass.contains(DestReg)) {
     unsigned CRReg = getCRFromCRBit(SrcReg);
-    BuildMI(MBB, I, DL, get(PPC::MFOCRF), DestReg)
-       .addReg(CRReg), getKillRegState(KillSrc);
+    BuildMI(MBB, I, DL, get(PPC::MFOCRF), DestReg).addReg(CRReg);
+    getKillRegState(KillSrc);
     // Rotate the CR bit in the CR fields to be the least significant bit and
     // then mask with 0x1 (MB = ME = 31).
     BuildMI(MBB, I, DL, get(PPC::RLWINM), DestReg)
@@ -895,13 +898,13 @@ void PPCInstrInfo::copyPhysReg(MachineBasicBlock &MBB,
     return;
   } else if (PPC::CRRCRegClass.contains(SrcReg) &&
       PPC::G8RCRegClass.contains(DestReg)) {
-    BuildMI(MBB, I, DL, get(PPC::MFOCRF8), DestReg)
-       .addReg(SrcReg), getKillRegState(KillSrc);
+    BuildMI(MBB, I, DL, get(PPC::MFOCRF8), DestReg).addReg(SrcReg);
+    getKillRegState(KillSrc);
     return;
   } else if (PPC::CRRCRegClass.contains(SrcReg) &&
       PPC::GPRCRegClass.contains(DestReg)) {
-    BuildMI(MBB, I, DL, get(PPC::MFOCRF), DestReg)
-       .addReg(SrcReg), getKillRegState(KillSrc);
+    BuildMI(MBB, I, DL, get(PPC::MFOCRF), DestReg).addReg(SrcReg);
+    getKillRegState(KillSrc);
     return;
    }
 
@@ -1085,12 +1088,11 @@ PPCInstrInfo::storeRegToStackSlot(MachineBasicBlock &MBB,
   NewMIs.back()->addMemOperand(MF, MMO);
 }
 
-bool
-PPCInstrInfo::LoadRegFromStackSlot(MachineFunction &MF, DebugLoc DL,
-                                   unsigned DestReg, int FrameIdx,
-                                   const TargetRegisterClass *RC,
-                                   SmallVectorImpl<MachineInstr*> &NewMIs,
-                                   bool &NonRI, bool &SpillsVRS) const{
+bool PPCInstrInfo::LoadRegFromStackSlot(MachineFunction &MF, const DebugLoc &DL,
+                                        unsigned DestReg, int FrameIdx,
+                                        const TargetRegisterClass *RC,
+                                        SmallVectorImpl<MachineInstr *> &NewMIs,
+                                        bool &NonRI, bool &SpillsVRS) const {
   // Note: If additional load instructions are added here,
   // update isLoadFromStackSlot.
 
@@ -1208,35 +1210,35 @@ ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const {
   return false;
 }
 
-bool PPCInstrInfo::FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
-                             unsigned Reg, MachineRegisterInfo *MRI) const {
+bool PPCInstrInfo::FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI,
+                                 unsigned Reg, MachineRegisterInfo *MRI) const {
   // For some instructions, it is legal to fold ZERO into the RA register field.
   // A zero immediate should always be loaded with a single li.
-  unsigned DefOpc = DefMI->getOpcode();
+  unsigned DefOpc = DefMI.getOpcode();
   if (DefOpc != PPC::LI && DefOpc != PPC::LI8)
     return false;
-  if (!DefMI->getOperand(1).isImm())
+  if (!DefMI.getOperand(1).isImm())
     return false;
-  if (DefMI->getOperand(1).getImm() != 0)
+  if (DefMI.getOperand(1).getImm() != 0)
     return false;
 
   // Note that we cannot here invert the arguments of an isel in order to fold
   // a ZERO into what is presented as the second argument. All we have here
   // is the condition bit, and that might come from a CR-logical bit operation.
 
-  const MCInstrDesc &UseMCID = UseMI->getDesc();
+  const MCInstrDesc &UseMCID = UseMI.getDesc();
 
   // Only fold into real machine instructions.
   if (UseMCID.isPseudo())
     return false;
 
   unsigned UseIdx;
-  for (UseIdx = 0; UseIdx < UseMI->getNumOperands(); ++UseIdx)
-    if (UseMI->getOperand(UseIdx).isReg() &&
-        UseMI->getOperand(UseIdx).getReg() == Reg)
+  for (UseIdx = 0; UseIdx < UseMI.getNumOperands(); ++UseIdx)
+    if (UseMI.getOperand(UseIdx).isReg() &&
+        UseMI.getOperand(UseIdx).getReg() == Reg)
       break;
 
-  assert(UseIdx < UseMI->getNumOperands() && "Cannot find Reg in UseMI");
+  assert(UseIdx < UseMI.getNumOperands() && "Cannot find Reg in UseMI");
   assert(UseIdx < UseMCID.getNumOperands() && "No operand description for Reg");
 
   const MCOperandInfo *UseInfo = &UseMCID.OpInfo[UseIdx];
@@ -1268,10 +1270,10 @@ bool PPCInstrInfo::FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
   }
 
   bool DeleteDef = MRI->hasOneNonDBGUse(Reg);
-  UseMI->getOperand(UseIdx).setReg(ZeroReg);
+  UseMI.getOperand(UseIdx).setReg(ZeroReg);
 
   if (DeleteDef)
-    DefMI->eraseFromParent();
+    DefMI.eraseFromParent();
 
   return true;
 }
@@ -1299,7 +1301,7 @@ bool PPCInstrInfo::isProfitableToIfCvt(MachineBasicBlock &TMBB,
 }
 
 
-bool PPCInstrInfo::isPredicated(const MachineInstr *MI) const {
+bool PPCInstrInfo::isPredicated(const MachineInstr &MI) const {
   // The predicated branches are identified by their type, not really by the
   // explicit presence of a predicate. Furthermore, some of them can be
   // predicated more than once. Because if conversion won't try to predicate
@@ -1310,73 +1312,71 @@ bool PPCInstrInfo::isPredicated(const MachineInstr *MI) const {
   return false;
 }
 
-bool PPCInstrInfo::isUnpredicatedTerminator(const MachineInstr *MI) const {
-  if (!MI->isTerminator())
+bool PPCInstrInfo::isUnpredicatedTerminator(const MachineInstr &MI) const {
+  if (!MI.isTerminator())
     return false;
 
   // Conditional branch is a special case.
-  if (MI->isBranch() && !MI->isBarrier())
+  if (MI.isBranch() && !MI.isBarrier())
     return true;
 
   return !isPredicated(MI);
 }
 
-bool PPCInstrInfo::PredicateInstruction(MachineInstr *MI,
+bool PPCInstrInfo::PredicateInstruction(MachineInstr &MI,
                                         ArrayRef<MachineOperand> Pred) const {
-  unsigned OpC = MI->getOpcode();
+  unsigned OpC = MI.getOpcode();
   if (OpC == PPC::BLR || OpC == PPC::BLR8) {
     if (Pred[1].getReg() == PPC::CTR8 || Pred[1].getReg() == PPC::CTR) {
       bool isPPC64 = Subtarget.isPPC64();
-      MI->setDesc(get(Pred[0].getImm() ?
-                      (isPPC64 ? PPC::BDNZLR8 : PPC::BDNZLR) :
-                      (isPPC64 ? PPC::BDZLR8  : PPC::BDZLR)));
+      MI.setDesc(get(Pred[0].getImm() ? (isPPC64 ? PPC::BDNZLR8 : PPC::BDNZLR)
+                                      : (isPPC64 ? PPC::BDZLR8 : PPC::BDZLR)));
     } else if (Pred[0].getImm() == PPC::PRED_BIT_SET) {
-      MI->setDesc(get(PPC::BCLR));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(Pred[1].getReg());
+      MI.setDesc(get(PPC::BCLR));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addReg(Pred[1].getReg());
     } else if (Pred[0].getImm() == PPC::PRED_BIT_UNSET) {
-      MI->setDesc(get(PPC::BCLRn));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(Pred[1].getReg());
+      MI.setDesc(get(PPC::BCLRn));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addReg(Pred[1].getReg());
     } else {
-      MI->setDesc(get(PPC::BCCLR));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addImm(Pred[0].getImm())
-        .addReg(Pred[1].getReg());
+      MI.setDesc(get(PPC::BCCLR));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addImm(Pred[0].getImm())
+          .addReg(Pred[1].getReg());
     }
 
     return true;
   } else if (OpC == PPC::B) {
     if (Pred[1].getReg() == PPC::CTR8 || Pred[1].getReg() == PPC::CTR) {
       bool isPPC64 = Subtarget.isPPC64();
-      MI->setDesc(get(Pred[0].getImm() ?
-                      (isPPC64 ? PPC::BDNZ8 : PPC::BDNZ) :
-                      (isPPC64 ? PPC::BDZ8  : PPC::BDZ)));
+      MI.setDesc(get(Pred[0].getImm() ? (isPPC64 ? PPC::BDNZ8 : PPC::BDNZ)
+                                      : (isPPC64 ? PPC::BDZ8 : PPC::BDZ)));
     } else if (Pred[0].getImm() == PPC::PRED_BIT_SET) {
-      MachineBasicBlock *MBB = MI->getOperand(0).getMBB();
-      MI->RemoveOperand(0);
+      MachineBasicBlock *MBB = MI.getOperand(0).getMBB();
+      MI.RemoveOperand(0);
 
-      MI->setDesc(get(PPC::BC));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(Pred[1].getReg())
-        .addMBB(MBB);
+      MI.setDesc(get(PPC::BC));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addReg(Pred[1].getReg())
+          .addMBB(MBB);
     } else if (Pred[0].getImm() == PPC::PRED_BIT_UNSET) {
-      MachineBasicBlock *MBB = MI->getOperand(0).getMBB();
-      MI->RemoveOperand(0);
+      MachineBasicBlock *MBB = MI.getOperand(0).getMBB();
+      MI.RemoveOperand(0);
 
-      MI->setDesc(get(PPC::BCn));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(Pred[1].getReg())
-        .addMBB(MBB);
+      MI.setDesc(get(PPC::BCn));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addReg(Pred[1].getReg())
+          .addMBB(MBB);
     } else {
-      MachineBasicBlock *MBB = MI->getOperand(0).getMBB();
-      MI->RemoveOperand(0);
-
-      MI->setDesc(get(PPC::BCC));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addImm(Pred[0].getImm())
-        .addReg(Pred[1].getReg())
-        .addMBB(MBB);
+      MachineBasicBlock *MBB = MI.getOperand(0).getMBB();
+      MI.RemoveOperand(0);
+
+      MI.setDesc(get(PPC::BCC));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addImm(Pred[0].getImm())
+          .addReg(Pred[1].getReg())
+          .addMBB(MBB);
     }
 
     return true;
@@ -1389,24 +1389,24 @@ bool PPCInstrInfo::PredicateInstruction(MachineInstr *MI,
     bool isPPC64 = Subtarget.isPPC64();
 
     if (Pred[0].getImm() == PPC::PRED_BIT_SET) {
-      MI->setDesc(get(isPPC64 ? (setLR ? PPC::BCCTRL8 : PPC::BCCTR8) :
-                                (setLR ? PPC::BCCTRL  : PPC::BCCTR)));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(Pred[1].getReg());
+      MI.setDesc(get(isPPC64 ? (setLR ? PPC::BCCTRL8 : PPC::BCCTR8)
+                             : (setLR ? PPC::BCCTRL : PPC::BCCTR)));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addReg(Pred[1].getReg());
       return true;
     } else if (Pred[0].getImm() == PPC::PRED_BIT_UNSET) {
-      MI->setDesc(get(isPPC64 ? (setLR ? PPC::BCCTRL8n : PPC::BCCTR8n) :
-                                (setLR ? PPC::BCCTRLn  : PPC::BCCTRn)));
-      MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-        .addReg(Pred[1].getReg());
+      MI.setDesc(get(isPPC64 ? (setLR ? PPC::BCCTRL8n : PPC::BCCTR8n)
+                             : (setLR ? PPC::BCCTRLn : PPC::BCCTRn)));
+      MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+          .addReg(Pred[1].getReg());
       return true;
     }
 
-    MI->setDesc(get(isPPC64 ? (setLR ? PPC::BCCCTRL8 : PPC::BCCCTR8) :
-                              (setLR ? PPC::BCCCTRL  : PPC::BCCCTR)));
-    MachineInstrBuilder(*MI->getParent()->getParent(), MI)
-      .addImm(Pred[0].getImm())
-      .addReg(Pred[1].getReg());
+    MI.setDesc(get(isPPC64 ? (setLR ? PPC::BCCCTRL8 : PPC::BCCCTR8)
+                           : (setLR ? PPC::BCCCTRL : PPC::BCCCTR)));
+    MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+        .addImm(Pred[0].getImm())
+        .addReg(Pred[1].getReg());
     return true;
   }
 
@@ -1444,7 +1444,7 @@ bool PPCInstrInfo::SubsumesPredicate(ArrayRef<MachineOperand> Pred1,
   return false;
 }
 
-bool PPCInstrInfo::DefinesPredicate(MachineInstr *MI,
+bool PPCInstrInfo::DefinesPredicate(MachineInstr &MI,
                                     std::vector<MachineOperand> &Pred) const {
   // Note: At the present time, the contents of Pred from this function is
   // unused by IfConversion. This implementation follows ARM by pushing the
@@ -1457,8 +1457,8 @@ bool PPCInstrInfo::DefinesPredicate(MachineInstr *MI,
       &PPC::CTRRCRegClass, &PPC::CTRRC8RegClass };
 
   bool Found = false;
-  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
-    const MachineOperand &MO = MI->getOperand(i);
+  for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI.getOperand(i);
     for (unsigned c = 0; c < array_lengthof(RCs) && !Found; ++c) {
       const TargetRegisterClass *RC = RCs[c];
       if (MO.isReg()) {
@@ -1480,8 +1480,8 @@ bool PPCInstrInfo::DefinesPredicate(MachineInstr *MI,
   return Found;
 }
 
-bool PPCInstrInfo::isPredicable(MachineInstr *MI) const {
-  unsigned OpC = MI->getOpcode();
+bool PPCInstrInfo::isPredicable(MachineInstr &MI) const {
+  unsigned OpC = MI.getOpcode();
   switch (OpC) {
   default:
     return false;
@@ -1496,10 +1496,10 @@ bool PPCInstrInfo::isPredicable(MachineInstr *MI) const {
   }
 }
 
-bool PPCInstrInfo::analyzeCompare(const MachineInstr *MI,
-                                  unsigned &SrcReg, unsigned &SrcReg2,
-                                  int &Mask, int &Value) const {
-  unsigned Opc = MI->getOpcode();
+bool PPCInstrInfo::analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                                  unsigned &SrcReg2, int &Mask,
+                                  int &Value) const {
+  unsigned Opc = MI.getOpcode();
 
   switch (Opc) {
   default: return false;
@@ -1507,9 +1507,9 @@ bool PPCInstrInfo::analyzeCompare(const MachineInstr *MI,
   case PPC::CMPLWI:
   case PPC::CMPDI:
   case PPC::CMPLDI:
-    SrcReg = MI->getOperand(1).getReg();
+    SrcReg = MI.getOperand(1).getReg();
     SrcReg2 = 0;
-    Value = MI->getOperand(2).getImm();
+    Value = MI.getOperand(2).getImm();
     Mask = 0xFFFF;
     return true;
   case PPC::CMPW:
@@ -1518,21 +1518,20 @@ bool PPCInstrInfo::analyzeCompare(const MachineInstr *MI,
   case PPC::CMPLD:
   case PPC::FCMPUS:
   case PPC::FCMPUD:
-    SrcReg = MI->getOperand(1).getReg();
-    SrcReg2 = MI->getOperand(2).getReg();
+    SrcReg = MI.getOperand(1).getReg();
+    SrcReg2 = MI.getOperand(2).getReg();
     return true;
   }
 }
 
-bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
-                                        unsigned SrcReg, unsigned SrcReg2,
-                                        int Mask, int Value,
+bool PPCInstrInfo::optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
+                                        unsigned SrcReg2, int Mask, int Value,
                                         const MachineRegisterInfo *MRI) const {
   if (DisableCmpOpt)
     return false;
 
-  int OpC = CmpInstr->getOpcode();
-  unsigned CRReg = CmpInstr->getOperand(0).getReg();
+  int OpC = CmpInstr.getOpcode();
+  unsigned CRReg = CmpInstr.getOperand(0).getReg();
 
   // FP record forms set CR1 based on the execption status bits, not a
   // comparison with zero.
@@ -1571,11 +1570,18 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
       } else
         return false;
     } else if (is32BitUnsignedCompare) {
+      // 32-bit rotate and mask instructions are zero extending only if MB <= ME
+      bool isZeroExtendingRotate  =
+          (MIOpC == PPC::RLWINM || MIOpC == PPC::RLWINMo ||
+           MIOpC == PPC::RLWNM || MIOpC == PPC::RLWNMo)
+          && MI->getOperand(3).getImm() <= MI->getOperand(4).getImm();
+
       // We can perform this optimization, equality only, if MI is
       // zero-extending.
       if (MIOpC == PPC::CNTLZW || MIOpC == PPC::CNTLZWo ||
           MIOpC == PPC::SLW    || MIOpC == PPC::SLWo ||
-          MIOpC == PPC::SRW    || MIOpC == PPC::SRWo) {
+          MIOpC == PPC::SRW    || MIOpC == PPC::SRWo ||
+          isZeroExtendingRotate) {
         noSub = true;
         equalityOnly = true;
       } else
@@ -1608,8 +1614,8 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
   MachineBasicBlock::iterator I = CmpInstr;
 
   // Scan forward to find the first use of the compare.
-  for (MachineBasicBlock::iterator EL = CmpInstr->getParent()->end();
-       I != EL; ++I) {
+  for (MachineBasicBlock::iterator EL = CmpInstr.getParent()->end(); I != EL;
+       ++I) {
     bool FoundUse = false;
     for (MachineRegisterInfo::use_instr_iterator J =MRI->use_instr_begin(CRReg),
          JE = MRI->use_instr_end(); J != JE; ++J)
@@ -1633,7 +1639,7 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
   // same BB as the comparison. This is to allow the check below to avoid calls
   // (and other explicit clobbers); instead we should really check for these
   // more explicitly (in at least a few predecessors).
-  else if (MI->getParent() != CmpInstr->getParent() || Value != 0) {
+  else if (MI->getParent() != CmpInstr.getParent() || Value != 0) {
     // PPC does not have a record-form SUBri.
     return false;
   }
@@ -1643,16 +1649,14 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
   --I;
 
   // Get ready to iterate backward from CmpInstr.
-  MachineBasicBlock::iterator E = MI,
-                              B = CmpInstr->getParent()->begin();
+  MachineBasicBlock::iterator E = MI, B = CmpInstr.getParent()->begin();
 
   for (; I != E && !noSub; --I) {
     const MachineInstr &Instr = *I;
     unsigned IOpC = Instr.getOpcode();
 
-    if (&*I != CmpInstr && (
-        Instr.modifiesRegister(PPC::CR0, TRI) ||
-        Instr.readsRegister(PPC::CR0, TRI)))
+    if (&*I != &CmpInstr && (Instr.modifiesRegister(PPC::CR0, TRI) ||
+                             Instr.readsRegister(PPC::CR0, TRI)))
       // This instruction modifies or uses the record condition register after
       // the one we want to change. While we could do this transformation, it
       // would likely not be profitable. This transformation removes one
@@ -1752,13 +1756,17 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
   // Create a new virtual register to hold the value of the CR set by the
   // record-form instruction. If the instruction was not previously in
   // record form, then set the kill flag on the CR.
-  CmpInstr->eraseFromParent();
+  CmpInstr.eraseFromParent();
 
   MachineBasicBlock::iterator MII = MI;
   BuildMI(*MI->getParent(), std::next(MII), MI->getDebugLoc(),
           get(TargetOpcode::COPY), CRReg)
     .addReg(PPC::CR0, MIOpC != NewOpC ? RegState::Kill : 0);
 
+  // Even if CR0 register were dead before, it is alive now since the
+  // instruction we just built uses it.
+  MI->clearRegisterDeads(PPC::CR0);
+
   if (MIOpC != NewOpC) {
     // We need to be careful here: we're replacing one instruction with
     // another, and we need to make sure that we get all of the right
@@ -1783,6 +1791,8 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
           MI->addOperand(*MI->getParent()->getParent(),
                          MachineOperand::CreateReg(*ImpUses, false, true));
   }
+  assert(MI->definesRegister(PPC::CR0) &&
+         "Record-form instruction does not define cr0?");
 
   // Modify the condition code of operands in OperandsToUpdate.
   // Since we have SUB(r1, r2) and CMP(r2, r1), the condition code needs to
@@ -1799,17 +1809,17 @@ bool PPCInstrInfo::optimizeCompareInstr(MachineInstr *CmpInstr,
 /// GetInstSize - Return the number of bytes of code the specified
 /// instruction may be.  This returns the maximum number of bytes.
 ///
-unsigned PPCInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
-  unsigned Opcode = MI->getOpcode();
+unsigned PPCInstrInfo::GetInstSizeInBytes(const MachineInstr &MI) const {
+  unsigned Opcode = MI.getOpcode();
 
   if (Opcode == PPC::INLINEASM) {
-    const MachineFunction *MF = MI->getParent()->getParent();
-    const char *AsmStr = MI->getOperand(0).getSymbolName();
+    const MachineFunction *MF = MI.getParent()->getParent();
+    const char *AsmStr = MI.getOperand(0).getSymbolName();
     return getInlineAsmLength(AsmStr, *MF->getTarget().getMCAsmInfo());
   } else if (Opcode == TargetOpcode::STACKMAP) {
-    return MI->getOperand(1).getImm();
+    return MI.getOperand(1).getImm();
   } else if (Opcode == TargetOpcode::PATCHPOINT) {
-    PatchPointOpers Opers(MI);
+    PatchPointOpers Opers(&MI);
     return Opers.getMetaOper(PatchPointOpers::NBytesPos).getImm();
   } else {
     const MCInstrDesc &Desc = get(Opcode);
@@ -1842,10 +1852,26 @@ ArrayRef<std::pair<unsigned, const char *>>
 PPCInstrInfo::getSerializableBitmaskMachineOperandTargetFlags() const {
   using namespace PPCII;
   static const std::pair<unsigned, const char *> TargetFlags[] = {
-      {MO_PLT_OR_STUB, "ppc-plt-or-stub"},
+      {MO_PLT, "ppc-plt"},
       {MO_PIC_FLAG, "ppc-pic"},
       {MO_NLP_FLAG, "ppc-nlp"},
       {MO_NLP_HIDDEN_FLAG, "ppc-nlp-hidden"}};
   return makeArrayRef(TargetFlags);
 }
 
+bool PPCInstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
+  switch (MI.getOpcode()) {
+  case TargetOpcode::LOAD_STACK_GUARD: {
+    assert(Subtarget.isTargetLinux() &&
+           "Only Linux target is expected to contain LOAD_STACK_GUARD");
+    const int64_t Offset = Subtarget.isPPC64() ? -0x7010 : -0x7008;
+    const unsigned Reg = Subtarget.isPPC64() ? PPC::X13 : PPC::R2;
+    MI.setDesc(get(Subtarget.isPPC64() ? PPC::LD : PPC::LWZ));
+    MachineInstrBuilder(*MI.getParent()->getParent(), MI)
+        .addImm(Offset)
+        .addReg(Reg);
+    return true;
+  }
+  }
+  return false;
+}
diff --git a/lib/Target/PowerPC/PPCInstrInfo.h b/lib/Target/PowerPC/PPCInstrInfo.h
index c3c3a480a6aad..98baf125bdff4 100644
--- a/lib/Target/PowerPC/PPCInstrInfo.h
+++ b/lib/Target/PowerPC/PPCInstrInfo.h
@@ -73,10 +73,10 @@ class PPCInstrInfo : public PPCGenInstrInfo {
                            const TargetRegisterClass *RC,
                            SmallVectorImpl<MachineInstr*> &NewMIs,
                            bool &NonRI, bool &SpillsVRS) const;
-  bool LoadRegFromStackSlot(MachineFunction &MF, DebugLoc DL,
+  bool LoadRegFromStackSlot(MachineFunction &MF, const DebugLoc &DL,
                             unsigned DestReg, int FrameIdx,
                             const TargetRegisterClass *RC,
-                            SmallVectorImpl<MachineInstr*> &NewMIs,
+                            SmallVectorImpl<MachineInstr *> &NewMIs,
                             bool &NonRI, bool &SpillsVRS) const;
   virtual void anchor();
 
@@ -91,8 +91,7 @@ protected:
   ///
   /// For example, we can commute rlwimi instructions, but only if the
   /// rotate amt is zero.  We also have to munge the immediates a bit.
-  MachineInstr *commuteInstructionImpl(MachineInstr *MI,
-                                       bool NewMI,
+  MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
                                        unsigned OpIdx1,
                                        unsigned OpIdx2) const override;
 
@@ -113,12 +112,12 @@ public:
                                      const ScheduleDAG *DAG) const override;
 
   unsigned getInstrLatency(const InstrItineraryData *ItinData,
-                           const MachineInstr *MI,
+                           const MachineInstr &MI,
                            unsigned *PredCost = nullptr) const override;
 
   int getOperandLatency(const InstrItineraryData *ItinData,
-                        const MachineInstr *DefMI, unsigned DefIdx,
-                        const MachineInstr *UseMI,
+                        const MachineInstr &DefMI, unsigned DefIdx,
+                        const MachineInstr &UseMI,
                         unsigned UseIdx) const override;
   int getOperandLatency(const InstrItineraryData *ItinData,
                         SDNode *DefNode, unsigned DefIdx,
@@ -128,7 +127,7 @@ public:
   }
 
   bool hasLowDefLatency(const TargetSchedModel &SchedModel,
-                        const MachineInstr *DefMI,
+                        const MachineInstr &DefMI,
                         unsigned DefIdx) const override {
     // Machine LICM should hoist all instructions in low-register-pressure
     // situations; none are sufficiently free to justify leaving in a loop
@@ -152,12 +151,12 @@ public:
   bool isCoalescableExtInstr(const MachineInstr &MI,
                              unsigned &SrcReg, unsigned &DstReg,
                              unsigned &SubIdx) const override;
-  unsigned isLoadFromStackSlot(const MachineInstr *MI,
+  unsigned isLoadFromStackSlot(const MachineInstr &MI,
                                int &FrameIndex) const override;
-  unsigned isStoreToStackSlot(const MachineInstr *MI,
+  unsigned isStoreToStackSlot(const MachineInstr &MI,
                               int &FrameIndex) const override;
 
-  bool findCommutedOpIndices(MachineInstr *MI, unsigned &SrcOpIdx1,
+  bool findCommutedOpIndices(MachineInstr &MI, unsigned &SrcOpIdx1,
                              unsigned &SrcOpIdx2) const override;
 
   void insertNoop(MachineBasicBlock &MBB,
@@ -165,25 +164,25 @@ public:
 
 
   // Branch analysis.
-  bool AnalyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
+  bool analyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                      MachineBasicBlock *&FBB,
                      SmallVectorImpl<MachineOperand> &Cond,
                      bool AllowModify) const override;
   unsigned RemoveBranch(MachineBasicBlock &MBB) const override;
   unsigned InsertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
                         MachineBasicBlock *FBB, ArrayRef<MachineOperand> Cond,
-                        DebugLoc DL) const override;
+                        const DebugLoc &DL) const override;
 
   // Select analysis.
   bool canInsertSelect(const MachineBasicBlock &, ArrayRef<MachineOperand> Cond,
                        unsigned, unsigned, int &, int &, int &) const override;
   void insertSelect(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
-                    DebugLoc DL, unsigned DstReg, ArrayRef<MachineOperand> Cond,
-                    unsigned TrueReg, unsigned FalseReg) const override;
+                    const DebugLoc &DL, unsigned DstReg,
+                    ArrayRef<MachineOperand> Cond, unsigned TrueReg,
+                    unsigned FalseReg) const override;
 
-  void copyPhysReg(MachineBasicBlock &MBB,
-                   MachineBasicBlock::iterator I, DebugLoc DL,
-                   unsigned DestReg, unsigned SrcReg,
+  void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator I,
+                   const DebugLoc &DL, unsigned DestReg, unsigned SrcReg,
                    bool KillSrc) const override;
 
   void storeRegToStackSlot(MachineBasicBlock &MBB,
@@ -201,8 +200,8 @@ public:
   bool
   ReverseBranchCondition(SmallVectorImpl<MachineOperand> &Cond) const override;
 
-  bool FoldImmediate(MachineInstr *UseMI, MachineInstr *DefMI,
-                     unsigned Reg, MachineRegisterInfo *MRI) const override;
+  bool FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI, unsigned Reg,
+                     MachineRegisterInfo *MRI) const override;
 
   // If conversion by predication (only supported by some branch instructions).
   // All of the profitability checks always return true; it is always
@@ -230,37 +229,34 @@ public:
   }
 
   // Predication support.
-  bool isPredicated(const MachineInstr *MI) const override;
+  bool isPredicated(const MachineInstr &MI) const override;
 
-  bool isUnpredicatedTerminator(const MachineInstr *MI) const override;
+  bool isUnpredicatedTerminator(const MachineInstr &MI) const override;
 
-  bool PredicateInstruction(MachineInstr *MI,
+  bool PredicateInstruction(MachineInstr &MI,
                             ArrayRef<MachineOperand> Pred) const override;
 
   bool SubsumesPredicate(ArrayRef<MachineOperand> Pred1,
                          ArrayRef<MachineOperand> Pred2) const override;
 
-  bool DefinesPredicate(MachineInstr *MI,
+  bool DefinesPredicate(MachineInstr &MI,
                         std::vector<MachineOperand> &Pred) const override;
 
-  bool isPredicable(MachineInstr *MI) const override;
+  bool isPredicable(MachineInstr &MI) const override;
 
   // Comparison optimization.
 
+  bool analyzeCompare(const MachineInstr &MI, unsigned &SrcReg,
+                      unsigned &SrcReg2, int &Mask, int &Value) const override;
 
-  bool analyzeCompare(const MachineInstr *MI,
-                      unsigned &SrcReg, unsigned &SrcReg2,
-                      int &Mask, int &Value) const override;
-
-  bool optimizeCompareInstr(MachineInstr *CmpInstr,
-                            unsigned SrcReg, unsigned SrcReg2,
-                            int Mask, int Value,
+  bool optimizeCompareInstr(MachineInstr &CmpInstr, unsigned SrcReg,
+                            unsigned SrcReg2, int Mask, int Value,
                             const MachineRegisterInfo *MRI) const override;
 
   /// GetInstSize - Return the number of bytes of code the specified
   /// instruction may be.  This returns the maximum number of bytes.
   ///
-  unsigned GetInstSizeInBytes(const MachineInstr *MI) const;
+  unsigned GetInstSizeInBytes(const MachineInstr &MI) const;
 
   void getNoopForMachoTarget(MCInst &NopInst) const override;
 
@@ -272,6 +268,9 @@ public:
 
   ArrayRef<std::pair<unsigned, const char *>>
   getSerializableBitmaskMachineOperandTargetFlags() const override;
+
+  // Lower pseudo instructions after register allocation.
+  bool expandPostRAPseudo(MachineInstr &MI) const override;
 };
 
 }
diff --git a/lib/Target/PowerPC/PPCInstrInfo.td b/lib/Target/PowerPC/PPCInstrInfo.td
index ce0f9e6f52a70..4a42a947c6cbd 100644
--- a/lib/Target/PowerPC/PPCInstrInfo.td
+++ b/lib/Target/PowerPC/PPCInstrInfo.td
@@ -31,6 +31,18 @@ def SDT_PPCvperm   : SDTypeProfile<1, 3, [
   SDTCisVT<3, v16i8>, SDTCisSameAs<0, 1>, SDTCisSameAs<0, 2>
 ]>;
 
+def SDT_PPCVecSplat : SDTypeProfile<1, 2, [ SDTCisVec<0>,
+  SDTCisVec<1>, SDTCisInt<2>
+]>;
+
+def SDT_PPCVecShift : SDTypeProfile<1, 3, [ SDTCisVec<0>,
+  SDTCisVec<1>, SDTCisVec<2>, SDTCisInt<3>
+]>;
+
+def SDT_PPCVecInsert : SDTypeProfile<1, 3, [ SDTCisVec<0>,
+  SDTCisVec<1>, SDTCisVec<2>, SDTCisInt<3>
+]>;
+
 def SDT_PPCvcmp : SDTypeProfile<1, 3, [
   SDTCisSameAs<0, 1>, SDTCisSameAs<1, 2>, SDTCisVT<3, i32>
 ]>;
@@ -140,7 +152,10 @@ def PPCaddiTlsldLAddr : SDNode<"PPCISD::ADDI_TLSLD_L_ADDR",
 def PPCaddisDtprelHA : SDNode<"PPCISD::ADDIS_DTPREL_HA", SDTIntBinOp>;
 def PPCaddiDtprelL   : SDNode<"PPCISD::ADDI_DTPREL_L", SDTIntBinOp>;
 
-def PPCvperm    : SDNode<"PPCISD::VPERM", SDT_PPCvperm, []>;
+def PPCvperm     : SDNode<"PPCISD::VPERM", SDT_PPCvperm, []>;
+def PPCxxsplt    : SDNode<"PPCISD::XXSPLT", SDT_PPCVecSplat, []>;
+def PPCxxinsert  : SDNode<"PPCISD::XXINSERT", SDT_PPCVecInsert, []>;
+def PPCvecshl    : SDNode<"PPCISD::VECSHL", SDT_PPCVecShift, []>;
 
 def PPCqvfperm   : SDNode<"PPCISD::QVFPERM", SDT_PPCqvfperm, []>;
 def PPCqvgpci    : SDNode<"PPCISD::QVGPCI", SDT_PPCqvgpci, []>;
@@ -257,7 +272,7 @@ def HI16 : SDNodeXForm<imm, [{
 
 def HA16 : SDNodeXForm<imm, [{
   // Transformation function: shift the immediate value down into the low bits.
-  signed int Val = N->getZExtValue();
+  int Val = N->getZExtValue();
   return getI32Imm((Val - (signed short)Val) >> 16, SDLoc(N));
 }]>;
 def MB : SDNodeXForm<imm, [{
@@ -507,6 +522,24 @@ def u6imm   : Operand<i32> {
   let ParserMatchClass = PPCU6ImmAsmOperand;
   let DecoderMethod = "decodeUImmOperand<6>";
 }
+def PPCU7ImmAsmOperand : AsmOperandClass {
+  let Name = "U7Imm"; let PredicateMethod = "isU7Imm";
+  let RenderMethod = "addImmOperands";
+}
+def u7imm   : Operand<i32> {
+  let PrintMethod = "printU7ImmOperand";
+  let ParserMatchClass = PPCU7ImmAsmOperand;
+  let DecoderMethod = "decodeUImmOperand<7>";
+}
+def PPCU8ImmAsmOperand : AsmOperandClass {
+  let Name = "U8Imm"; let PredicateMethod = "isU8Imm";
+  let RenderMethod = "addImmOperands";
+}
+def u8imm   : Operand<i32> {
+  let PrintMethod = "printU8ImmOperand";
+  let ParserMatchClass = PPCU8ImmAsmOperand;
+  let DecoderMethod = "decodeUImmOperand<8>";
+}
 def PPCU10ImmAsmOperand : AsmOperandClass {
   let Name = "U10Imm"; let PredicateMethod = "isU10Imm";
   let RenderMethod = "addImmOperands";
@@ -635,6 +668,13 @@ def PPCDispRIXOperand : AsmOperandClass {
 def dispRIX : Operand<iPTR> {
   let ParserMatchClass = PPCDispRIXOperand;
 }
+def PPCDispRIX16Operand : AsmOperandClass {
+ let Name = "DispRIX16"; let PredicateMethod = "isS16ImmX16";
+ let RenderMethod = "addImmOperands";
+}
+def dispRIX16 : Operand<iPTR> {
+  let ParserMatchClass = PPCDispRIX16Operand;
+}
 def PPCDispSPE8Operand : AsmOperandClass {
  let Name = "DispSPE8"; let PredicateMethod = "isU8ImmX8";
  let RenderMethod = "addImmOperands";
@@ -673,6 +713,12 @@ def memrix : Operand<iPTR> {   // memri where the imm is 4-aligned.
   let EncoderMethod = "getMemRIXEncoding";
   let DecoderMethod = "decodeMemRIXOperands";
 }
+def memrix16 : Operand<iPTR> { // memri, imm is 16-aligned, 12-bit, Inst{16:27}
+  let PrintMethod = "printMemRegImm";
+  let MIOperandInfo = (ops dispRIX16:$imm, ptr_rc_nor0:$reg);
+  let EncoderMethod = "getMemRIX16Encoding";
+  let DecoderMethod = "decodeMemRIX16Operands";
+}
 def spe8dis : Operand<iPTR> {   // SPE displacement where the imm is 8-aligned.
   let PrintMethod = "printMemRegImm";
   let MIOperandInfo = (ops dispSPE8:$imm, ptr_rc_nor0:$reg);
@@ -746,6 +792,7 @@ def NoNaNsFPMath : Predicate<"TM.Options.NoNaNsFPMath">;
 def NaNsFPMath   : Predicate<"!TM.Options.NoNaNsFPMath">;
 def HasBPERMD : Predicate<"PPCSubTarget->hasBPERMD()">;
 def HasExtDiv : Predicate<"PPCSubTarget->hasExtDiv()">;
+def IsISA3_0 : Predicate<"PPCSubTarget->isISA3_0()">;
 
 //===----------------------------------------------------------------------===//
 // PowerPC Multiclass Definitions.
@@ -1365,7 +1412,10 @@ let hasSideEffects = 1, isBarrier = 1, usesCustomInserter = 1 in {
                           Requires<[In32BitMode]>;
 }
 
-let isBranch = 1, isTerminator = 1 in {
+// This pseudo is never removed from the function, as it serves as
+// a terminator.  Size is set to 0 to prevent the builtin assembler
+// from emitting it.
+let isBranch = 1, isTerminator = 1, Size = 0 in {
   def EH_SjLj_Setup : Pseudo<(outs), (ins directbrtarget:$dst),
                         "#EH_SjLj_Setup\t$dst", []>;
 }
@@ -1543,6 +1593,13 @@ def LHARXL : XForm_1<31,  116, (outs gprc:$rD), (ins memrr:$src),
 
 def LWARXL : XForm_1<31,  20, (outs gprc:$rD), (ins memrr:$src),
                      "lwarx $rD, $src, 1", IIC_LdStLWARX, []>, isDOT;
+
+// The atomic instructions use the destination register as well as the next one
+// or two registers in order (modulo 31).
+let hasExtraSrcRegAllocReq = 1 in
+def LWAT : X_RD5_RS5_IM5<31, 582, (outs gprc:$rD), (ins gprc:$rA, u5imm:$FC),
+                         "lwat $rD, $rA, $FC", IIC_LdStLoad>,
+           Requires<[IsISA3_0]>;
 }
 
 let Defs = [CR0], mayStore = 1, hasSideEffects = 0 in {
@@ -1558,6 +1615,11 @@ def STWCX : XForm_1<31, 150, (outs), (ins gprc:$rS, memrr:$dst),
                     "stwcx. $rS, $dst", IIC_LdStSTWCX, []>, isDOT;
 }
 
+let mayStore = 1, hasSideEffects = 0 in
+def STWAT : X_RD5_RS5_IM5<31, 710, (outs), (ins gprc:$rS, gprc:$rA, u5imm:$FC),
+                          "stwat $rS, $rA, $FC", IIC_LdStStore>,
+            Requires<[IsISA3_0]>;
+
 let isTerminator = 1, isBarrier = 1, hasCtrlDep = 1 in
 def TRAP  : XForm_24<31, 4, (outs), (ins), "trap", IIC_LdStLoad, [(trap)]>;
 
@@ -1947,6 +2009,10 @@ let isCompare = 1, hasSideEffects = 0 in {
                           "cmpwi $crD, $rA, $imm", IIC_IntCompare>;
   def CMPLWI : DForm_6_ext<10, (outs crrc:$dst), (ins gprc:$src1, u16imm:$src2),
                            "cmplwi $dst, $src1, $src2", IIC_IntCompare>;
+  def CMPRB  : X_BF3_L1_RS5_RS5<31, 192, (outs crbitrc:$BF),
+                                (ins u1imm:$L, g8rc:$rA, g8rc:$rB),
+                                "cmprb $BF, $L, $rA, $rB", IIC_IntCompare, []>,
+               Requires<[IsISA3_0]>;
 }
 }
 
@@ -2000,6 +2066,9 @@ defm SRAWI : XForm_10rc<31, 824, (outs gprc:$rA), (ins gprc:$rS, u5imm:$SH),
 defm CNTLZW : XForm_11r<31,  26, (outs gprc:$rA), (ins gprc:$rS),
                         "cntlzw", "$rA, $rS", IIC_IntGeneral,
                         [(set i32:$rA, (ctlz i32:$rS))]>;
+defm CNTTZW : XForm_11r<31, 538, (outs gprc:$rA), (ins gprc:$rS),
+                        "cnttzw", "$rA, $rS", IIC_IntGeneral,
+                        [(set i32:$rA, (cttz i32:$rS))]>, Requires<[IsISA3_0]>;
 defm EXTSB  : XForm_11r<31, 954, (outs gprc:$rA), (ins gprc:$rS),
                         "extsb", "$rA, $rS", IIC_IntSimple,
                         [(set i32:$rA, (sext_inreg i32:$rS, i8))]>;
@@ -2286,6 +2355,10 @@ let isCodeGenOnly = 1 in {
                   PPC970_DGroup_First, PPC970_Unit_FXU;
 }
 
+// Aliases for mtvrsave/mfvrsave to mfspr/mtspr.
+def : InstAlias<"mtvrsave $rS", (MTVRSAVE gprc:$rS)>;
+def : InstAlias<"mfvrsave $rS", (MFVRSAVE gprc:$rS)>;
+
 // SPILL_VRSAVE - Indicate that we're dumping the VRSAVE register,
 // so we'll need to scavenge a register for it.
 let mayStore = 1 in
@@ -2328,6 +2401,9 @@ def MFCR : XFXForm_3<31, 19, (outs gprc:$rT), (ins),
                      "mfcr $rT", IIC_SprMFCR>,
                      PPC970_MicroCode, PPC970_Unit_CRU;
 } // hasExtraSrcRegAllocReq = 1
+
+def MCRXRX : X_BF3<31, 576, (outs crrc:$BF), (ins),
+                   "mcrxrx $BF", IIC_BrMCRX>, Requires<[IsISA3_0]>;
 } // hasSideEffects = 0
 
 // Pseudo instruction to perform FADD in round-to-zero mode.
@@ -4138,3 +4214,33 @@ def : Pat<(atomic_store_32 iaddr:$ptr, i32:$val), (STW  gprc:$val, memri:$ptr)>;
 def : Pat<(atomic_store_8  xaddr:$ptr, i32:$val), (STBX gprc:$val, memrr:$ptr)>;
 def : Pat<(atomic_store_16 xaddr:$ptr, i32:$val), (STHX gprc:$val, memrr:$ptr)>;
 def : Pat<(atomic_store_32 xaddr:$ptr, i32:$val), (STWX gprc:$val, memrr:$ptr)>;
+
+let Predicates = [IsISA3_0] in {
+
+// Copy-Paste Facility
+// We prefix 'CP' to COPY due to name conflict in Target.td. We also prefix to
+// PASTE for naming consistency.
+let mayLoad = 1 in
+def CP_COPY   : X_L1_RA5_RB5<31, 774, "copy"  , gprc, IIC_LdStCOPY, []>;
+
+let mayStore = 1 in
+def CP_PASTE  : X_L1_RA5_RB5<31, 902, "paste" , gprc, IIC_LdStPASTE, []>;
+
+let mayStore = 1, Defs = [CR0] in
+def CP_PASTEo : X_L1_RA5_RB5<31, 902, "paste.", gprc, IIC_LdStPASTE, []>, isDOT;
+
+def CP_COPYx  : PPCAsmPseudo<"copy $rA, $rB" , (ins gprc:$rA, gprc:$rB)>;
+def CP_PASTEx : PPCAsmPseudo<"paste $rA, $rB", (ins gprc:$rA, gprc:$rB)>;
+def CP_COPY_FIRST : PPCAsmPseudo<"copy_first $rA, $rB",
+                                  (ins gprc:$rA, gprc:$rB)>;
+def CP_PASTE_LAST : PPCAsmPseudo<"paste_last $rA, $rB",
+                                  (ins gprc:$rA, gprc:$rB)>;
+def CP_ABORT : XForm_0<31, 838, (outs), (ins), "cp_abort", IIC_SprABORT, []>;
+
+// Message Synchronize
+def MSGSYNC : XForm_0<31, 886, (outs), (ins), "msgsync", IIC_SprMSGSYNC, []>;
+
+// Power-Saving Mode Instruction:
+def STOP : XForm_0<19, 370, (outs), (ins), "stop", IIC_SprSTOP, []>;
+
+} // IsISA3_0
diff --git a/lib/Target/PowerPC/PPCInstrVSX.td b/lib/Target/PowerPC/PPCInstrVSX.td
index df1142cb42f3d..a02ace00a76f6 100644
--- a/lib/Target/PowerPC/PPCInstrVSX.td
+++ b/lib/Target/PowerPC/PPCInstrVSX.td
@@ -57,6 +57,9 @@ def SDT_PPCstxvd2x : SDTypeProfile<0, 2, [
 def SDT_PPCxxswapd : SDTypeProfile<1, 1, [
   SDTCisSameAs<0, 1>
 ]>;
+def SDTVecConv : SDTypeProfile<1, 2, [
+  SDTCisVec<0>, SDTCisVec<1>, SDTCisPtrTy<2>
+]>;
 
 def PPClxvd2x  : SDNode<"PPCISD::LXVD2X", SDT_PPClxvd2x,
                         [SDNPHasChain, SDNPMayLoad]>;
@@ -66,6 +69,9 @@ def PPCxxswapd : SDNode<"PPCISD::XXSWAPD", SDT_PPCxxswapd, [SDNPHasChain]>;
 def PPCmfvsr : SDNode<"PPCISD::MFVSR", SDTUnaryOp, []>;
 def PPCmtvsra : SDNode<"PPCISD::MTVSRA", SDTUnaryOp, []>;
 def PPCmtvsrz : SDNode<"PPCISD::MTVSRZ", SDTUnaryOp, []>;
+def PPCsvec2fp : SDNode<"PPCISD::SINT_VEC_TO_FP", SDTVecConv, []>;
+def PPCuvec2fp: SDNode<"PPCISD::UINT_VEC_TO_FP", SDTVecConv, []>;
+def PPCswapNoChain : SDNode<"PPCISD::SWAP_NO_CHAIN", SDT_PPCxxswapd>;
 
 multiclass XX3Form_Rcr<bits<6> opcode, bits<7> xo, string asmbase,
                     string asmstr, InstrItinClass itin, Intrinsic Int,
@@ -608,7 +614,8 @@ let Uses = [RM] in {
                       "xvcvsxwdp $XT, $XB", IIC_VecFP, []>;
   def XVCVSXWSP : XX2Form<60, 184,
                       (outs vsrc:$XT), (ins vsrc:$XB),
-                      "xvcvsxwsp $XT, $XB", IIC_VecFP, []>;
+                      "xvcvsxwsp $XT, $XB", IIC_VecFP,
+                      [(set v4f32:$XT, (sint_to_fp v4i32:$XB))]>;
   def XVCVUXDDP : XX2Form<60, 488,
                       (outs vsrc:$XT), (ins vsrc:$XB),
                       "xvcvuxddp $XT, $XB", IIC_VecFP,
@@ -772,10 +779,14 @@ let Uses = [RM] in {
 
   def XXSLDWI : XX3Form_2<60, 2,
                        (outs vsrc:$XT), (ins vsrc:$XA, vsrc:$XB, u2imm:$SHW),
-                       "xxsldwi $XT, $XA, $XB, $SHW", IIC_VecPerm, []>;
+                       "xxsldwi $XT, $XA, $XB, $SHW", IIC_VecPerm,
+                       [(set v4i32:$XT, (PPCvecshl v4i32:$XA, v4i32:$XB,
+                                                  imm32SExt16:$SHW))]>;
   def XXSPLTW : XX2Form_2<60, 164,
                        (outs vsrc:$XT), (ins vsrc:$XB, u2imm:$UIM),
-                       "xxspltw $XT, $XB, $UIM", IIC_VecPerm, []>;
+                       "xxspltw $XT, $XB, $UIM", IIC_VecPerm,
+                       [(set v4i32:$XT,
+                             (PPCxxsplt v4i32:$XB, imm32SExt16:$UIM))]>;
 } // hasSideEffects
 
 // SELECT_CC_* - Used to implement the SELECT_CC DAG operation.  Expanded after
@@ -926,6 +937,16 @@ def : Pat<(sext_inreg v2i64:$C, v2i32),
 def : Pat<(v2f64 (sint_to_fp (sext_inreg v2i64:$C, v2i32))),
           (XVCVSXWDP (XXSLDWI $C, $C, 1))>;
 
+def : Pat<(v2f64 (PPCsvec2fp v4i32:$C, 0)),
+          (v2f64 (XVCVSXWDP (v2i64 (XXMRGHW $C, $C))))>;
+def : Pat<(v2f64 (PPCsvec2fp v4i32:$C, 1)),
+          (v2f64 (XVCVSXWDP (v2i64 (XXMRGLW $C, $C))))>;
+
+def : Pat<(v2f64 (PPCuvec2fp v4i32:$C, 0)),
+          (v2f64 (XVCVUXWDP (v2i64 (XXMRGHW $C, $C))))>;
+def : Pat<(v2f64 (PPCuvec2fp v4i32:$C, 1)),
+          (v2f64 (XVCVUXWDP (v2i64 (XXMRGLW $C, $C))))>;
+
 // Loads.
 def : Pat<(v2f64 (load xoaddr:$src)), (LXVD2X xoaddr:$src)>;
 def : Pat<(v2i64 (load xoaddr:$src)), (LXVD2X xoaddr:$src)>;
@@ -945,6 +966,7 @@ def : Pat<(v2f64 (PPCxxswapd v2f64:$src)), (XXPERMDI $src, $src, 2)>;
 def : Pat<(v2i64 (PPCxxswapd v2i64:$src)), (XXPERMDI $src, $src, 2)>;
 def : Pat<(v4f32 (PPCxxswapd v4f32:$src)), (XXPERMDI $src, $src, 2)>;
 def : Pat<(v4i32 (PPCxxswapd v4i32:$src)), (XXPERMDI $src, $src, 2)>;
+def : Pat<(v2f64 (PPCswapNoChain v2f64:$src)), (XXPERMDI $src, $src, 2)>;
 
 // Selects.
 def : Pat<(v2f64 (selectcc i1:$lhs, i1:$rhs, v2f64:$tval, v2f64:$fval, SETLT)),
@@ -1007,6 +1029,28 @@ def : Pat<(int_ppc_vsx_xvrsqrtesp v4f32:$A),
 def : Pat<(int_ppc_vsx_xvrsqrtedp v2f64:$A),
           (XVRSQRTEDP $A)>;
 
+let Predicates = [IsLittleEndian] in {
+def : Pat<(f64 (PPCfcfid (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+          (f64 (XSCVSXDDP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+def : Pat<(f64 (PPCfcfid (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+          (f64 (XSCVSXDDP (COPY_TO_REGCLASS (f64 (COPY_TO_REGCLASS $S, VSRC)), VSFRC)))>;
+def : Pat<(f64 (PPCfcfidu (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+          (f64 (XSCVUXDDP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+def : Pat<(f64 (PPCfcfidu (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+          (f64 (XSCVUXDDP (COPY_TO_REGCLASS (f64 (COPY_TO_REGCLASS $S, VSRC)), VSFRC)))>;
+} // IsLittleEndian
+
+let Predicates = [IsBigEndian] in {
+def : Pat<(f64 (PPCfcfid (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+          (f64 (XSCVSXDDP (COPY_TO_REGCLASS $S, VSFRC)))>;
+def : Pat<(f64 (PPCfcfid (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+          (f64 (XSCVSXDDP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+def : Pat<(f64 (PPCfcfidu (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+          (f64 (XSCVUXDDP (COPY_TO_REGCLASS $S, VSFRC)))>;
+def : Pat<(f64 (PPCfcfidu (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+          (f64 (XSCVUXDDP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+} // IsBigEndian
+
 } // AddedComplexity
 } // HasVSX
 
@@ -1213,10 +1257,31 @@ let AddedComplexity = 400 in { // Prefer VSX patterns over non-VSX patterns.
   def XSCVSPDPN : XX2Form<60, 331, (outs vssrc:$XT), (ins vsrc:$XB),
                           "xscvspdpn $XT, $XB", IIC_VecFP, []>;
 
+  let Predicates = [IsLittleEndian] in {
+  def : Pat<(f32 (PPCfcfids (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+            (f32 (XSCVSXDSP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+  def : Pat<(f32 (PPCfcfids (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+            (f32 (XSCVSXDSP (COPY_TO_REGCLASS (f64 (COPY_TO_REGCLASS $S, VSRC)), VSFRC)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+            (f32 (XSCVUXDSP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+            (f32 (XSCVUXDSP (COPY_TO_REGCLASS (f64 (COPY_TO_REGCLASS $S, VSRC)), VSFRC)))>;
+  }
+
+  let Predicates = [IsBigEndian] in {
+  def : Pat<(f32 (PPCfcfids (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+            (f32 (XSCVSXDSP (COPY_TO_REGCLASS $S, VSFRC)))>;
+  def : Pat<(f32 (PPCfcfids (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+            (f32 (XSCVSXDSP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsra (i64 (vector_extract v2i64:$S, 0))))),
+            (f32 (XSCVUXDSP (COPY_TO_REGCLASS $S, VSFRC)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsra (i64 (vector_extract v2i64:$S, 1))))),
+            (f32 (XSCVUXDSP (COPY_TO_REGCLASS (XXPERMDI $S, $S, 2), VSFRC)))>;
+  }
 } // AddedComplexity = 400
 } // HasP8Vector
 
-let Predicates = [HasDirectMove, HasVSX] in {
+let Predicates = [HasDirectMove] in {
   // VSX direct move instructions
   def MFVSRD : XX1_RS6_RD5_XO<31, 51, (outs g8rc:$rA), (ins vsfrc:$XT),
                               "mfvsrd $rA, $XT", IIC_VecGeneral,
@@ -1235,7 +1300,22 @@ let Predicates = [HasDirectMove, HasVSX] in {
   def MTVSRWZ : XX1_RS6_RD5_XO<31, 243, (outs vsfrc:$XT), (ins gprc:$rA),
                                "mtvsrwz $XT, $rA", IIC_VecGeneral,
                                [(set f64:$XT, (PPCmtvsrz i32:$rA))]>;
-} // HasDirectMove, HasVSX
+} // HasDirectMove
+
+let Predicates = [IsISA3_0, HasDirectMove] in {
+  def MTVSRWS: XX1_RS6_RD5_XO<31, 403, (outs vsrc:$XT), (ins gprc:$rA),
+                              "mtvsrws $XT, $rA", IIC_VecGeneral,
+                              []>;
+
+  def MTVSRDD: XX1Form<31, 435, (outs vsrc:$XT), (ins g8rc:$rA, g8rc:$rB),
+                       "mtvsrdd $XT, $rA, $rB", IIC_VecGeneral,
+                       []>, Requires<[In64BitMode]>;
+
+  def MFVSRLD: XX1_RS6_RD5_XO<31, 307, (outs g8rc:$rA), (ins vsrc:$XT),
+                              "mfvsrld $rA, $XT", IIC_VecGeneral,
+                              []>, Requires<[In64BitMode]>;
+
+} // IsISA3_0, HasDirectMove
 
 /*  Direct moves of various widths from GPR's into VSR's. Each move lines
     the value up into element 0 (both BE and LE). Namely, entities smaller than
@@ -1285,7 +1365,7 @@ def VectorExtractions {
                        (v2i64 (COPY_TO_REGCLASS $S, VSRC)), sub_64));
 
   // Word extraction
-  dag LE_WORD_0 = (MFVSRWZ (EXTRACT_SUBREG (XXSLDWI $S, $S, 2), sub_64));
+  dag LE_WORD_0 = (MFVSRWZ (EXTRACT_SUBREG (XXPERMDI $S, $S, 2), sub_64));
   dag LE_WORD_1 = (MFVSRWZ (EXTRACT_SUBREG (XXSLDWI $S, $S, 1), sub_64));
   dag LE_WORD_2 = (MFVSRWZ (EXTRACT_SUBREG
                              (v2i64 (COPY_TO_REGCLASS $S, VSRC)), sub_64));
@@ -1555,7 +1635,7 @@ let Predicates = [IsBigEndian, HasP8Vector] in {
   def : Pat<(f32 (vector_extract v4f32:$S, 1)),
             (f32 (XSCVSPDPN (XXSLDWI $S, $S, 1)))>;
   def : Pat<(f32 (vector_extract v4f32:$S, 2)),
-            (f32 (XSCVSPDPN (XXSLDWI $S, $S, 2)))>;
+            (f32 (XSCVSPDPN (XXPERMDI $S, $S, 2)))>;
   def : Pat<(f32 (vector_extract v4f32:$S, 3)),
             (f32 (XSCVSPDPN (XXSLDWI $S, $S, 3)))>;
   def : Pat<(f32 (vector_extract v4f32:$S, i64:$Idx)),
@@ -1660,7 +1740,7 @@ let Predicates = [IsLittleEndian, HasP8Vector] in {
   def : Pat<(f32 (vector_extract v4f32:$S, 0)),
             (f32 (XSCVSPDPN (XXSLDWI $S, $S, 3)))>;
   def : Pat<(f32 (vector_extract v4f32:$S, 1)),
-            (f32 (XSCVSPDPN (XXSLDWI $S, $S, 2)))>;
+            (f32 (XSCVSPDPN (XXPERMDI $S, $S, 2)))>;
   def : Pat<(f32 (vector_extract v4f32:$S, 2)),
             (f32 (XSCVSPDPN (XXSLDWI $S, $S, 1)))>;
   def : Pat<(f32 (vector_extract v4f32:$S, 3)),
@@ -1783,3 +1863,423 @@ def : Pat<(i64 (bitconvert f64:$S)),
 def : Pat<(f64 (bitconvert i64:$S)),
           (f64 (MTVSRD $S))>;
 }
+
+def AlignValues {
+  dag F32_TO_BE_WORD1 = (v4f32 (XXSLDWI (XSCVDPSPN $B), (XSCVDPSPN $B), 3));
+  dag I32_TO_BE_WORD1 = (COPY_TO_REGCLASS (MTVSRWZ $B), VSRC);
+}
+
+// The following VSX instructions were introduced in Power ISA 3.0
+def HasP9Vector : Predicate<"PPCSubTarget->hasP9Vector()">;
+let AddedComplexity = 400, Predicates = [HasP9Vector] in {
+
+  // [PO VRT XO VRB XO /]
+  class X_VT5_XO5_VB5<bits<6> opcode, bits<5> xo2, bits<10> xo, string opc,
+                      list<dag> pattern>
+    : X_RD5_XO5_RS5<opcode, xo2, xo, (outs vrrc:$vT), (ins vrrc:$vB),
+                    !strconcat(opc, " $vT, $vB"), IIC_VecFP, pattern>;
+
+  // [PO VRT XO VRB XO RO], Round to Odd version of [PO VRT XO VRB XO /]
+  class X_VT5_XO5_VB5_Ro<bits<6> opcode, bits<5> xo2, bits<10> xo, string opc,
+                         list<dag> pattern>
+    : X_VT5_XO5_VB5<opcode, xo2, xo, opc, pattern>, isDOT;
+
+  // [PO VRT XO VRB XO /], but the VRB is only used the left 64 bits (or less),
+  // So we use different operand class for VRB
+  class X_VT5_XO5_VB5_TyVB<bits<6> opcode, bits<5> xo2, bits<10> xo, string opc,
+                           RegisterOperand vbtype, list<dag> pattern>
+    : X_RD5_XO5_RS5<opcode, xo2, xo, (outs vrrc:$vT), (ins vbtype:$vB),
+                    !strconcat(opc, " $vT, $vB"), IIC_VecFP, pattern>;
+
+  // [PO T XO B XO BX /]
+  class XX2_RT5_XO5_XB6<bits<6> opcode, bits<5> xo2, bits<9> xo, string opc,
+                        list<dag> pattern>
+    : XX2_RD5_XO5_RS6<opcode, xo2, xo, (outs g8rc:$rT), (ins vsfrc:$XB),
+                      !strconcat(opc, " $rT, $XB"), IIC_VecFP, pattern>;
+
+  // [PO T XO B XO BX TX]
+  class XX2_XT6_XO5_XB6<bits<6> opcode, bits<5> xo2, bits<9> xo, string opc,
+                        RegisterOperand vtype, list<dag> pattern>
+    : XX2_RD6_XO5_RS6<opcode, xo2, xo, (outs vtype:$XT), (ins vtype:$XB),
+                      !strconcat(opc, " $XT, $XB"), IIC_VecFP, pattern>;
+
+  // [PO T A B XO AX BX TX], src and dest register use different operand class
+  class XX3_XT5_XA5_XB5<bits<6> opcode, bits<8> xo, string opc,
+                  RegisterOperand xty, RegisterOperand aty, RegisterOperand bty,
+                  InstrItinClass itin, list<dag> pattern>
+    : XX3Form<opcode, xo, (outs xty:$XT), (ins aty:$XA, bty:$XB),
+              !strconcat(opc, " $XT, $XA, $XB"), itin, pattern>;
+
+  // [PO VRT VRA VRB XO /]
+  class X_VT5_VA5_VB5<bits<6> opcode, bits<10> xo, string opc,
+                      list<dag> pattern>
+    : XForm_1<opcode, xo, (outs vrrc:$vT), (ins vrrc:$vA, vrrc:$vB),
+              !strconcat(opc, " $vT, $vA, $vB"), IIC_VecFP, pattern>;
+
+  // [PO VRT VRA VRB XO RO], Round to Odd version of [PO VRT VRA VRB XO /]
+  class X_VT5_VA5_VB5_Ro<bits<6> opcode, bits<10> xo, string opc,
+                         list<dag> pattern>
+    : X_VT5_VA5_VB5<opcode, xo, opc, pattern>, isDOT;
+
+  //===--------------------------------------------------------------------===//
+  // Quad-Precision Scalar Move Instructions:
+
+  // Copy Sign
+  def XSCPSGNQP : X_VT5_VA5_VB5<63, 100, "xscpsgnqp", []>;
+
+  // Absolute/Negative-Absolute/Negate
+  def XSABSQP   : X_VT5_XO5_VB5<63,  0, 804, "xsabsqp" , []>;
+  def XSNABSQP  : X_VT5_XO5_VB5<63,  8, 804, "xsnabsqp", []>;
+  def XSNEGQP   : X_VT5_XO5_VB5<63, 16, 804, "xsnegqp" , []>;
+
+  //===--------------------------------------------------------------------===//
+  // Quad-Precision Scalar Floating-Point Arithmetic Instructions:
+
+  // Add/Divide/Multiply/Subtract
+  def XSADDQP   : X_VT5_VA5_VB5   <63,   4, "xsaddqp" , []>;
+  def XSADDQPO  : X_VT5_VA5_VB5_Ro<63,   4, "xsaddqpo", []>;
+  def XSDIVQP   : X_VT5_VA5_VB5   <63, 548, "xsdivqp" , []>;
+  def XSDIVQPO  : X_VT5_VA5_VB5_Ro<63, 548, "xsdivqpo", []>;
+  def XSMULQP   : X_VT5_VA5_VB5   <63,  36, "xsmulqp" , []>;
+  def XSMULQPO  : X_VT5_VA5_VB5_Ro<63,  36, "xsmulqpo", []>;
+  def XSSUBQP   : X_VT5_VA5_VB5   <63, 516, "xssubqp" , []>;
+  def XSSUBQPO  : X_VT5_VA5_VB5_Ro<63, 516, "xssubqpo", []>;
+
+  // Square-Root
+  def XSSQRTQP  : X_VT5_XO5_VB5   <63, 27, 804, "xssqrtqp" , []>;
+  def XSSQRTQPO : X_VT5_XO5_VB5_Ro<63, 27, 804, "xssqrtqpo", []>;
+
+  // (Negative) Multiply-{Add/Subtract}
+  def XSMADDQP  : X_VT5_VA5_VB5   <63, 388, "xsmaddqp"  , []>;
+  def XSMADDQPO : X_VT5_VA5_VB5_Ro<63, 388, "xsmaddqpo" , []>;
+  def XSMSUBQP  : X_VT5_VA5_VB5   <63, 420, "xsmsubqp"  , []>;
+  def XSMSUBQPO : X_VT5_VA5_VB5_Ro<63, 420, "xsmsubqpo" , []>;
+  def XSNMADDQP : X_VT5_VA5_VB5   <63, 452, "xsnmaddqp" , []>;
+  def XSNMADDQPO: X_VT5_VA5_VB5_Ro<63, 452, "xsnmaddqpo", []>;
+  def XSNMSUBQP : X_VT5_VA5_VB5   <63, 484, "xsnmsubqp" , []>;
+  def XSNMSUBQPO: X_VT5_VA5_VB5_Ro<63, 484, "xsnmsubqpo", []>;
+
+  //===--------------------------------------------------------------------===//
+  // Quad/Double-Precision Compare Instructions:
+
+  // [PO BF // VRA VRB XO /]
+  class X_BF3_VA5_VB5<bits<6> opcode, bits<10> xo, string opc,
+                      list<dag> pattern>
+    : XForm_17<opcode, xo, (outs crrc:$crD), (ins vrrc:$VA, vrrc:$VB),
+               !strconcat(opc, " $crD, $VA, $VB"), IIC_FPCompare> {
+    let Pattern = pattern;
+  }
+
+  // QP Compare Ordered/Unordered
+  def XSCMPOQP : X_BF3_VA5_VB5<63, 132, "xscmpoqp", []>;
+  def XSCMPUQP : X_BF3_VA5_VB5<63, 644, "xscmpuqp", []>;
+
+  // DP/QP Compare Exponents
+  def XSCMPEXPDP : XX3Form_1<60, 59,
+                             (outs crrc:$crD), (ins vsfrc:$XA, vsfrc:$XB),
+                             "xscmpexpdp $crD, $XA, $XB", IIC_FPCompare, []>;
+  def XSCMPEXPQP : X_BF3_VA5_VB5<63, 164, "xscmpexpqp", []>;
+
+  // DP Compare ==, >=, >, !=
+  // Use vsrc for XT, because the entire register of XT is set.
+  // XT.dword[1] = 0x0000_0000_0000_0000
+  def XSCMPEQDP : XX3_XT5_XA5_XB5<60,  3, "xscmpeqdp", vsrc, vsfrc, vsfrc,
+                                  IIC_FPCompare, []>;
+  def XSCMPGEDP : XX3_XT5_XA5_XB5<60, 19, "xscmpgedp", vsrc, vsfrc, vsfrc,
+                                  IIC_FPCompare, []>;
+  def XSCMPGTDP : XX3_XT5_XA5_XB5<60, 11, "xscmpgtdp", vsrc, vsfrc, vsfrc,
+                                  IIC_FPCompare, []>;
+  def XSCMPNEDP : XX3_XT5_XA5_XB5<60, 27, "xscmpnedp", vsrc, vsfrc, vsfrc,
+                                  IIC_FPCompare, []>;
+  // Vector Compare Not Equal
+  def XVCMPNEDP  : XX3Form_Rc<60, 123,
+                              (outs vsrc:$XT), (ins vsrc:$XA, vsrc:$XB),
+                              "xvcmpnedp  $XT, $XA, $XB", IIC_VecFPCompare, []>;
+  let Defs = [CR6] in
+  def XVCMPNEDPo : XX3Form_Rc<60, 123,
+                              (outs vsrc:$XT), (ins vsrc:$XA, vsrc:$XB),
+                              "xvcmpnedp. $XT, $XA, $XB", IIC_VecFPCompare, []>,
+                              isDOT;
+  def XVCMPNESP  : XX3Form_Rc<60,  91,
+                              (outs vsrc:$XT), (ins vsrc:$XA, vsrc:$XB),
+                              "xvcmpnesp  $XT, $XA, $XB", IIC_VecFPCompare, []>;
+  let Defs = [CR6] in
+  def XVCMPNESPo : XX3Form_Rc<60,  91,
+                              (outs vsrc:$XT), (ins vsrc:$XA, vsrc:$XB),
+                              "xvcmpnesp. $XT, $XA, $XB", IIC_VecFPCompare, []>,
+                              isDOT;
+
+  //===--------------------------------------------------------------------===//
+  // Quad-Precision Floating-Point Conversion Instructions:
+
+  // Convert DP -> QP
+  def XSCVDPQP  : X_VT5_XO5_VB5_TyVB<63, 22, 836, "xscvdpqp", vsfrc, []>;
+
+  // Round & Convert QP -> DP (dword[1] is set to zero)
+  def XSCVQPDP  : X_VT5_XO5_VB5   <63, 20, 836, "xscvqpdp" , []>;
+  def XSCVQPDPO : X_VT5_XO5_VB5_Ro<63, 20, 836, "xscvqpdpo", []>;
+
+  // Truncate & Convert QP -> (Un)Signed (D)Word (dword[1] is set to zero)
+  def XSCVQPSDZ : X_VT5_XO5_VB5<63, 25, 836, "xscvqpsdz", []>;
+  def XSCVQPSWZ : X_VT5_XO5_VB5<63,  9, 836, "xscvqpswz", []>;
+  def XSCVQPUDZ : X_VT5_XO5_VB5<63, 17, 836, "xscvqpudz", []>;
+  def XSCVQPUWZ : X_VT5_XO5_VB5<63,  1, 836, "xscvqpuwz", []>;
+
+  // Convert (Un)Signed DWord -> QP
+  def XSCVSDQP  : X_VT5_XO5_VB5_TyVB<63, 10, 836, "xscvsdqp", vsfrc, []>;
+  def XSCVUDQP  : X_VT5_XO5_VB5_TyVB<63,  2, 836, "xscvudqp", vsfrc, []>;
+
+  //===--------------------------------------------------------------------===//
+  // Round to Floating-Point Integer Instructions
+
+  // (Round &) Convert DP <-> HP
+  // Note! xscvdphp's src and dest register both use the left 64 bits, so we use
+  // vsfrc for src and dest register. xscvhpdp's src only use the left 16 bits,
+  // but we still use vsfrc for it.
+  def XSCVDPHP : XX2_XT6_XO5_XB6<60, 17, 347, "xscvdphp", vsfrc, []>;
+  def XSCVHPDP : XX2_XT6_XO5_XB6<60, 16, 347, "xscvhpdp", vsfrc, []>;
+
+  // Vector HP -> SP
+  def XVCVHPSP : XX2_XT6_XO5_XB6<60, 24, 475, "xvcvhpsp", vsrc, []>;
+  def XVCVSPHP : XX2_XT6_XO5_XB6<60, 25, 475, "xvcvsphp", vsrc, []>;
+
+  class Z23_VT5_R1_VB5_RMC2_EX1<bits<6> opcode, bits<8> xo, bit ex, string opc,
+                                list<dag> pattern>
+    : Z23Form_1<opcode, xo,
+                (outs vrrc:$vT), (ins u1imm:$r, vrrc:$vB, u2imm:$rmc),
+                !strconcat(opc, " $r, $vT, $vB, $rmc"), IIC_VecFP, pattern> {
+    let RC = ex;
+  }
+
+  // Round to Quad-Precision Integer [with Inexact]
+  def XSRQPI   : Z23_VT5_R1_VB5_RMC2_EX1<63,  5, 0, "xsrqpi" , []>;
+  def XSRQPIX  : Z23_VT5_R1_VB5_RMC2_EX1<63,  5, 1, "xsrqpix", []>;
+
+  // Round Quad-Precision to Double-Extended Precision (fp80)
+  def XSRQPXP  : Z23_VT5_R1_VB5_RMC2_EX1<63, 37, 0, "xsrqpxp", []>;
+
+  //===--------------------------------------------------------------------===//
+  // Insert/Extract Instructions
+
+  // Insert Exponent DP/QP
+  // XT NOTE: XT.dword[1] = 0xUUUU_UUUU_UUUU_UUUU
+  def XSIEXPDP : XX1Form <60, 918, (outs vsrc:$XT), (ins g8rc:$rA, g8rc:$rB),
+                          "xsiexpdp $XT, $rA, $rB", IIC_VecFP, []>;
+  // vB NOTE: only vB.dword[0] is used, that's why we don't use
+  //          X_VT5_VA5_VB5 form
+  def XSIEXPQP : XForm_18<63, 868, (outs vrrc:$vT), (ins vrrc:$vA, vsfrc:$vB),
+                          "xsiexpqp $vT, $vA, $vB", IIC_VecFP, []>;
+
+  // Extract Exponent/Significand DP/QP
+  def XSXEXPDP : XX2_RT5_XO5_XB6<60,  0, 347, "xsxexpdp", []>;
+  def XSXSIGDP : XX2_RT5_XO5_XB6<60,  1, 347, "xsxsigdp", []>;
+  def XSXEXPQP : X_VT5_XO5_VB5  <63,  2, 804, "xsxexpqp", []>;
+  def XSXSIGQP : X_VT5_XO5_VB5  <63, 18, 804, "xsxsigqp", []>;
+
+  // Vector Insert Word
+  // XB NOTE: Only XB.dword[1] is used, but we use vsrc on XB.
+  def XXINSERTW   :
+    XX2_RD6_UIM5_RS6<60, 181, (outs vsrc:$XT),
+                     (ins vsrc:$XTi, vsrc:$XB, u4imm:$UIM),
+                     "xxinsertw $XT, $XB, $UIM", IIC_VecFP,
+                     [(set v4i32:$XT, (PPCxxinsert v4i32:$XTi, v4i32:$XB,
+                                                   imm32SExt16:$UIM))]>,
+                     RegConstraint<"$XTi = $XT">, NoEncode<"$XTi">;
+
+  // Vector Extract Unsigned Word
+  def XXEXTRACTUW : XX2_RD6_UIM5_RS6<60, 165,
+                                  (outs vsfrc:$XT), (ins vsrc:$XB, u4imm:$UIMM),
+                                  "xxextractuw $XT, $XB, $UIMM", IIC_VecFP, []>;
+
+  // Vector Insert Exponent DP/SP
+  def XVIEXPDP : XX3_XT5_XA5_XB5<60, 248, "xviexpdp", vsrc, vsrc, vsrc,
+                                 IIC_VecFP, []>;
+  def XVIEXPSP : XX3_XT5_XA5_XB5<60, 216, "xviexpsp", vsrc, vsrc, vsrc,
+                                 IIC_VecFP, []>;
+
+  // Vector Extract Exponent/Significand DP/SP
+  def XVXEXPDP : XX2_XT6_XO5_XB6<60,  0, 475, "xvxexpdp", vsrc, []>;
+  def XVXEXPSP : XX2_XT6_XO5_XB6<60,  8, 475, "xvxexpsp", vsrc, []>;
+  def XVXSIGDP : XX2_XT6_XO5_XB6<60,  1, 475, "xvxsigdp", vsrc, []>;
+  def XVXSIGSP : XX2_XT6_XO5_XB6<60,  9, 475, "xvxsigsp", vsrc, []>;
+
+  //===--------------------------------------------------------------------===//
+
+  // Test Data Class SP/DP/QP
+  def XSTSTDCSP : XX2_BF3_DCMX7_RS6<60, 298,
+                              (outs crrc:$BF), (ins u7imm:$DCMX, vsfrc:$XB),
+                              "xststdcsp $BF, $XB, $DCMX", IIC_VecFP, []>;
+  def XSTSTDCDP : XX2_BF3_DCMX7_RS6<60, 362,
+                              (outs crrc:$BF), (ins u7imm:$DCMX, vsfrc:$XB),
+                              "xststdcdp $BF, $XB, $DCMX", IIC_VecFP, []>;
+  def XSTSTDCQP : X_BF3_DCMX7_RS5  <63, 708,
+                              (outs crrc:$BF), (ins u7imm:$DCMX, vrrc:$vB),
+                              "xststdcqp $BF, $vB, $DCMX", IIC_VecFP, []>;
+
+  // Vector Test Data Class SP/DP
+  def XVTSTDCSP : XX2_RD6_DCMX7_RS6<60, 13, 5,
+                              (outs vsrc:$XT), (ins u7imm:$DCMX, vsrc:$XB),
+                              "xvtstdcsp $XT, $XB, $DCMX", IIC_VecFP, []>;
+  def XVTSTDCDP : XX2_RD6_DCMX7_RS6<60, 15, 5,
+                              (outs vsrc:$XT), (ins u7imm:$DCMX, vsrc:$XB),
+                              "xvtstdcdp $XT, $XB, $DCMX", IIC_VecFP, []>;
+
+  //===--------------------------------------------------------------------===//
+
+  // Maximum/Minimum Type-C/Type-J DP
+  // XT.dword[1] = 0xUUUU_UUUU_UUUU_UUUU, so we use vsrc for XT
+  def XSMAXCDP : XX3_XT5_XA5_XB5<60, 128, "xsmaxcdp", vsrc, vsfrc, vsfrc,
+                                 IIC_VecFP, []>;
+  def XSMAXJDP : XX3_XT5_XA5_XB5<60, 144, "xsmaxjdp", vsrc, vsfrc, vsfrc,
+                                 IIC_VecFP, []>;
+  def XSMINCDP : XX3_XT5_XA5_XB5<60, 136, "xsmincdp", vsrc, vsfrc, vsfrc,
+                                 IIC_VecFP, []>;
+  def XSMINJDP : XX3_XT5_XA5_XB5<60, 152, "xsminjdp", vsrc, vsfrc, vsfrc,
+                                 IIC_VecFP, []>;
+
+  //===--------------------------------------------------------------------===//
+
+  // Vector Byte-Reverse H/W/D/Q Word
+  def XXBRH : XX2_XT6_XO5_XB6<60,  7, 475, "xxbrh", vsrc, []>;
+  def XXBRW : XX2_XT6_XO5_XB6<60, 15, 475, "xxbrw", vsrc, []>;
+  def XXBRD : XX2_XT6_XO5_XB6<60, 23, 475, "xxbrd", vsrc, []>;
+  def XXBRQ : XX2_XT6_XO5_XB6<60, 31, 475, "xxbrq", vsrc, []>;
+
+  // Vector Permute
+  def XXPERM  : XX3_XT5_XA5_XB5<60, 26, "xxperm" , vsrc, vsrc, vsrc,
+                                IIC_VecPerm, []>;
+  def XXPERMR : XX3_XT5_XA5_XB5<60, 58, "xxpermr", vsrc, vsrc, vsrc,
+                                IIC_VecPerm, []>;
+
+  // Vector Splat Immediate Byte
+  def XXSPLTIB : X_RD6_IMM8<60, 360, (outs vsrc:$XT), (ins u8imm:$IMM8),
+                            "xxspltib $XT, $IMM8", IIC_VecPerm, []>;
+
+  //===--------------------------------------------------------------------===//
+  // Vector/Scalar Load/Store Instructions
+
+  let mayLoad = 1 in {
+  // Load Vector
+  def LXV : DQ_RD6_RS5_DQ12<61, 1, (outs vsrc:$XT), (ins memrix16:$src),
+                            "lxv $XT, $src", IIC_LdStLFD, []>;
+  // Load DWord
+  def LXSD  : DSForm_1<57, 2, (outs vrrc:$vD), (ins memrix:$src),
+                       "lxsd $vD, $src", IIC_LdStLFD, []>;
+  // Load SP from src, convert it to DP, and place in dword[0]
+  def LXSSP : DSForm_1<57, 3, (outs vrrc:$vD), (ins memrix:$src),
+                       "lxssp $vD, $src", IIC_LdStLFD, []>;
+
+  // [PO T RA RB XO TX] almost equal to [PO S RA RB XO SX], but has different
+  // "out" and "in" dag
+  class X_XT6_RA5_RB5<bits<6> opcode, bits<10> xo, string opc,
+                      RegisterOperand vtype, list<dag> pattern>
+    : XX1Form<opcode, xo, (outs vtype:$XT), (ins memrr:$src),
+              !strconcat(opc, " $XT, $src"), IIC_LdStLFD, pattern>;
+
+  // Load as Integer Byte/Halfword & Zero Indexed
+  def LXSIBZX : X_XT6_RA5_RB5<31, 781, "lxsibzx", vsfrc, []>;
+  def LXSIHZX : X_XT6_RA5_RB5<31, 813, "lxsihzx", vsfrc, []>;
+
+  // Load Vector Halfword*8/Byte*16 Indexed
+  def LXVH8X  : X_XT6_RA5_RB5<31, 812, "lxvh8x" , vsrc, []>;
+  def LXVB16X : X_XT6_RA5_RB5<31, 876, "lxvb16x", vsrc, []>;
+
+  // Load Vector Indexed
+  def LXVX    : X_XT6_RA5_RB5<31, 268, "lxvx"   , vsrc, []>;
+
+  // Load Vector (Left-justified) with Length
+  def LXVL    : X_XT6_RA5_RB5<31, 269, "lxvl"   , vsrc, []>;
+  def LXVLL   : X_XT6_RA5_RB5<31, 301, "lxvll"  , vsrc, []>;
+
+  // Load Vector Word & Splat Indexed
+  def LXVWSX  : X_XT6_RA5_RB5<31, 364, "lxvwsx" , vsrc, []>;
+  } // end mayLoad
+
+  let mayStore = 1 in {
+  // Store Vector
+  def STXV : DQ_RD6_RS5_DQ12<61, 5, (outs), (ins vsrc:$XT, memrix16:$dst),
+                             "stxv $XT, $dst", IIC_LdStSTFD, []>;
+  // Store DWord
+  def STXSD  : DSForm_1<61, 2, (outs), (ins vrrc:$vS, memrix:$dst),
+                        "stxsd $vS, $dst", IIC_LdStSTFD, []>;
+  // Convert DP of dword[0] to SP, and Store to dst
+  def STXSSP : DSForm_1<61, 3, (outs), (ins vrrc:$vS, memrix:$dst),
+                        "stxssp $vS, $dst", IIC_LdStSTFD, []>;
+
+  // [PO S RA RB XO SX]
+  class X_XS6_RA5_RB5<bits<6> opcode, bits<10> xo, string opc,
+                      RegisterOperand vtype, list<dag> pattern>
+    : XX1Form<opcode, xo, (outs), (ins vtype:$XT, memrr:$dst),
+              !strconcat(opc, " $XT, $dst"), IIC_LdStSTFD, pattern>;
+
+  // Store as Integer Byte/Halfword Indexed
+  def STXSIBX  : X_XS6_RA5_RB5<31,  909, "stxsibx" , vsfrc, []>;
+  def STXSIHX  : X_XS6_RA5_RB5<31,  941, "stxsihx" , vsfrc, []>;
+
+  // Store Vector Halfword*8/Byte*16 Indexed
+  def STXVH8X  : X_XS6_RA5_RB5<31,  940, "stxvh8x" , vsrc, []>;
+  def STXVB16X : X_XS6_RA5_RB5<31, 1004, "stxvb16x", vsrc, []>;
+
+  // Store Vector Indexed
+  def STXVX    : X_XS6_RA5_RB5<31,  396, "stxvx"   , vsrc, []>;
+
+  // Store Vector (Left-justified) with Length
+  def STXVL    : X_XS6_RA5_RB5<31,  397, "stxvl"   , vsrc, []>;
+  def STXVLL   : X_XS6_RA5_RB5<31,  429, "stxvll"  , vsrc, []>;
+  } // end mayStore
+
+  // Patterns for which instructions from ISA 3.0 are a better match
+  let Predicates = [IsLittleEndian, HasP9Vector] in {
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 0))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 12)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 1))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 8)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 2))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 4)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 3))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 0)))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 0)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 12))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 1)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 8))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 2)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 4))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 3)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 0))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 0)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 12))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 1)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 8))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 2)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 4))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 3)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 0))>;
+  } // IsLittleEndian, HasP9Vector
+
+  let Predicates = [IsBigEndian, HasP9Vector] in {
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 0))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 0)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 1))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 4)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 2))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 8)))>;
+  def : Pat<(f32 (PPCfcfidus (PPCmtvsrz (i32 (extractelt v4i32:$A, 3))))),
+            (f32 (XSCVUXDSP (XXEXTRACTUW $A, 12)))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 0)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 0))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 1)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 4))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 2)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 8))>;
+  def : Pat<(v4i32 (insertelt v4i32:$A, i32:$B, 3)),
+            (v4i32 (XXINSERTW v4i32:$A, AlignValues.I32_TO_BE_WORD1, 12))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 0)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 0))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 1)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 4))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 2)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 8))>;
+  def : Pat<(v4f32 (insertelt v4f32:$A, f32:$B, 3)),
+            (v4f32 (XXINSERTW v4f32:$A, AlignValues.F32_TO_BE_WORD1, 12))>;
+  } // IsLittleEndian, HasP9Vector
+} // end HasP9Vector, AddedComplexity
diff --git a/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp b/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp
deleted file mode 100644
index e3a35d5df358c..0000000000000
--- a/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp
+++ /dev/null
@@ -1,233 +0,0 @@
-//===-------- PPCLoopDataPrefetch.cpp - Loop Data Prefetching Pass --------===//
-//
-//                     The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-//
-// This file implements a Loop Data Prefetching Pass.
-//
-//===----------------------------------------------------------------------===//
-
-#define DEBUG_TYPE "ppc-loop-data-prefetch"
-#include "PPC.h"
-#include "llvm/Transforms/Scalar.h"
-#include "llvm/ADT/DepthFirstIterator.h"
-#include "llvm/ADT/Statistic.h"
-#include "llvm/Analysis/AssumptionCache.h"
-#include "llvm/Analysis/CodeMetrics.h"
-#include "llvm/Analysis/InstructionSimplify.h"
-#include "llvm/Analysis/LoopInfo.h"
-#include "llvm/Analysis/ScalarEvolution.h"
-#include "llvm/Analysis/ScalarEvolutionAliasAnalysis.h"
-#include "llvm/Analysis/ScalarEvolutionExpander.h"
-#include "llvm/Analysis/ScalarEvolutionExpressions.h"
-#include "llvm/Analysis/TargetTransformInfo.h"
-#include "llvm/Analysis/ValueTracking.h"
-#include "llvm/IR/CFG.h"
-#include "llvm/IR/Dominators.h"
-#include "llvm/IR/Function.h"
-#include "llvm/IR/IntrinsicInst.h"
-#include "llvm/IR/Module.h"
-#include "llvm/Support/CommandLine.h"
-#include "llvm/Support/Debug.h"
-#include "llvm/Transforms/Utils/BasicBlockUtils.h"
-#include "llvm/Transforms/Utils/Local.h"
-#include "llvm/Transforms/Utils/ValueMapper.h"
-using namespace llvm;
-
-// By default, we limit this to creating 16 PHIs (which is a little over half
-// of the allocatable register set).
-static cl::opt<bool>
-PrefetchWrites("ppc-loop-prefetch-writes", cl::Hidden, cl::init(false),
-               cl::desc("Prefetch write addresses"));
-
-// This seems like a reasonable default for the BG/Q (this pass is enabled, by
-// default, only on the BG/Q).
-static cl::opt<unsigned>
-PrefDist("ppc-loop-prefetch-distance", cl::Hidden, cl::init(300),
-         cl::desc("The loop prefetch distance"));
-
-static cl::opt<unsigned>
-CacheLineSize("ppc-loop-prefetch-cache-line", cl::Hidden, cl::init(64),
-              cl::desc("The loop prefetch cache line size"));
-
-namespace llvm {
-  void initializePPCLoopDataPrefetchPass(PassRegistry&);
-}
-
-namespace {
-
-  class PPCLoopDataPrefetch : public FunctionPass {
-  public:
-    static char ID; // Pass ID, replacement for typeid
-    PPCLoopDataPrefetch() : FunctionPass(ID) {
-      initializePPCLoopDataPrefetchPass(*PassRegistry::getPassRegistry());
-    }
-
-    void getAnalysisUsage(AnalysisUsage &AU) const override {
-      AU.addRequired<AssumptionCacheTracker>();
-      AU.addPreserved<DominatorTreeWrapperPass>();
-      AU.addRequired<LoopInfoWrapperPass>();
-      AU.addPreserved<LoopInfoWrapperPass>();
-      AU.addRequired<ScalarEvolutionWrapperPass>();
-      // FIXME: For some reason, preserving SE here breaks LSR (even if
-      // this pass changes nothing).
-      // AU.addPreserved<ScalarEvolutionWrapperPass>();
-      AU.addRequired<TargetTransformInfoWrapperPass>();
-    }
-
-    bool runOnFunction(Function &F) override;
-    bool runOnLoop(Loop *L);
-
-  private:
-    AssumptionCache *AC;
-    LoopInfo *LI;
-    ScalarEvolution *SE;
-    const TargetTransformInfo *TTI;
-    const DataLayout *DL;
-  };
-}
-
-char PPCLoopDataPrefetch::ID = 0;
-INITIALIZE_PASS_BEGIN(PPCLoopDataPrefetch, "ppc-loop-data-prefetch",
-                      "PPC Loop Data Prefetch", false, false)
-INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
-INITIALIZE_PASS_DEPENDENCY(TargetTransformInfoWrapperPass)
-INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass)
-INITIALIZE_PASS_DEPENDENCY(ScalarEvolutionWrapperPass)
-INITIALIZE_PASS_END(PPCLoopDataPrefetch, "ppc-loop-data-prefetch",
-                    "PPC Loop Data Prefetch", false, false)
-
-FunctionPass *llvm::createPPCLoopDataPrefetchPass() { return new PPCLoopDataPrefetch(); }
-
-bool PPCLoopDataPrefetch::runOnFunction(Function &F) {
-  LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
-  SE = &getAnalysis<ScalarEvolutionWrapperPass>().getSE();
-  DL = &F.getParent()->getDataLayout();
-  AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
-  TTI = &getAnalysis<TargetTransformInfoWrapperPass>().getTTI(F);
-
-  bool MadeChange = false;
-
-  for (auto I = LI->begin(), IE = LI->end(); I != IE; ++I)
-    for (auto L = df_begin(*I), LE = df_end(*I); L != LE; ++L)
-      MadeChange |= runOnLoop(*L);
-
-  return MadeChange;
-}
-
-bool PPCLoopDataPrefetch::runOnLoop(Loop *L) {
-  bool MadeChange = false;
-
-  // Only prefetch in the inner-most loop
-  if (!L->empty())
-    return MadeChange;
-
-  SmallPtrSet<const Value *, 32> EphValues;
-  CodeMetrics::collectEphemeralValues(L, AC, EphValues);
-
-  // Calculate the number of iterations ahead to prefetch
-  CodeMetrics Metrics;
-  for (Loop::block_iterator I = L->block_begin(), IE = L->block_end();
-       I != IE; ++I) {
-
-    // If the loop already has prefetches, then assume that the user knows
-    // what he or she is doing and don't add any more.
-    for (BasicBlock::iterator J = (*I)->begin(), JE = (*I)->end();
-         J != JE; ++J)
-      if (CallInst *CI = dyn_cast<CallInst>(J))
-        if (Function *F = CI->getCalledFunction())
-          if (F->getIntrinsicID() == Intrinsic::prefetch)
-            return MadeChange;
-
-    Metrics.analyzeBasicBlock(*I, *TTI, EphValues);
-  }
-  unsigned LoopSize = Metrics.NumInsts;
-  if (!LoopSize)
-    LoopSize = 1;
-
-  unsigned ItersAhead = PrefDist/LoopSize;
-  if (!ItersAhead)
-    ItersAhead = 1;
-
-  SmallVector<std::pair<Instruction *, const SCEVAddRecExpr *>, 16> PrefLoads;
-  for (Loop::block_iterator I = L->block_begin(), IE = L->block_end();
-       I != IE; ++I) {
-    for (BasicBlock::iterator J = (*I)->begin(), JE = (*I)->end();
-        J != JE; ++J) {
-      Value *PtrValue;
-      Instruction *MemI;
-
-      if (LoadInst *LMemI = dyn_cast<LoadInst>(J)) {
-        MemI = LMemI;
-        PtrValue = LMemI->getPointerOperand();
-      } else if (StoreInst *SMemI = dyn_cast<StoreInst>(J)) {
-        if (!PrefetchWrites) continue;
-        MemI = SMemI;
-        PtrValue = SMemI->getPointerOperand();
-      } else continue;
-
-      unsigned PtrAddrSpace = PtrValue->getType()->getPointerAddressSpace();
-      if (PtrAddrSpace)
-        continue;
-
-      if (L->isLoopInvariant(PtrValue))
-        continue;
-
-      const SCEV *LSCEV = SE->getSCEV(PtrValue);
-      const SCEVAddRecExpr *LSCEVAddRec = dyn_cast<SCEVAddRecExpr>(LSCEV);
-      if (!LSCEVAddRec)
-        continue;
-
-      // We don't want to double prefetch individual cache lines. If this load
-      // is known to be within one cache line of some other load that has
-      // already been prefetched, then don't prefetch this one as well.
-      bool DupPref = false;
-      for (SmallVector<std::pair<Instruction *, const SCEVAddRecExpr *>,
-             16>::iterator K = PrefLoads.begin(), KE = PrefLoads.end();
-           K != KE; ++K) {
-        const SCEV *PtrDiff = SE->getMinusSCEV(LSCEVAddRec, K->second);
-        if (const SCEVConstant *ConstPtrDiff =
-            dyn_cast<SCEVConstant>(PtrDiff)) {
-          int64_t PD = std::abs(ConstPtrDiff->getValue()->getSExtValue());
-          if (PD < (int64_t) CacheLineSize) {
-            DupPref = true;
-            break;
-          }
-        }
-      }
-      if (DupPref)
-        continue;
-
-      const SCEV *NextLSCEV = SE->getAddExpr(LSCEVAddRec, SE->getMulExpr(
-        SE->getConstant(LSCEVAddRec->getType(), ItersAhead),
-        LSCEVAddRec->getStepRecurrence(*SE)));
-      if (!isSafeToExpand(NextLSCEV, *SE))
-        continue;
-
-      PrefLoads.push_back(std::make_pair(MemI, LSCEVAddRec));
-
-      Type *I8Ptr = Type::getInt8PtrTy((*I)->getContext(), PtrAddrSpace);
-      SCEVExpander SCEVE(*SE, J->getModule()->getDataLayout(), "prefaddr");
-      Value *PrefPtrValue = SCEVE.expandCodeFor(NextLSCEV, I8Ptr, MemI);
-
-      IRBuilder<> Builder(MemI);
-      Module *M = (*I)->getParent()->getParent();
-      Type *I32 = Type::getInt32Ty((*I)->getContext());
-      Value *PrefetchFunc = Intrinsic::getDeclaration(M, Intrinsic::prefetch);
-      Builder.CreateCall(
-          PrefetchFunc,
-          {PrefPtrValue,
-           ConstantInt::get(I32, MemI->mayReadFromMemory() ? 0 : 1),
-           ConstantInt::get(I32, 3), ConstantInt::get(I32, 1)});
-
-      MadeChange = true;
-    }
-  }
-
-  return MadeChange;
-}
-
diff --git a/lib/Target/PowerPC/PPCLoopPreIncPrep.cpp b/lib/Target/PowerPC/PPCLoopPreIncPrep.cpp
index 5e188268fee92..48a71cfc2a6ef 100644
--- a/lib/Target/PowerPC/PPCLoopPreIncPrep.cpp
+++ b/lib/Target/PowerPC/PPCLoopPreIncPrep.cpp
@@ -144,6 +144,9 @@ static Value *GetPointerOperand(Value *MemI) {
 }
 
 bool PPCLoopPreIncPrep::runOnFunction(Function &F) {
+  if (skipFunction(F))
+    return false;
+
   LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
   SE = &getAnalysis<ScalarEvolutionWrapperPass>().getSE();
   auto *DTWP = getAnalysisIfAvailable<DominatorTreeWrapperPass>();
diff --git a/lib/Target/PowerPC/PPCMCInstLower.cpp b/lib/Target/PowerPC/PPCMCInstLower.cpp
index 44a692d4bb42d..18377a44a7f89 100644
--- a/lib/Target/PowerPC/PPCMCInstLower.cpp
+++ b/lib/Target/PowerPC/PPCMCInstLower.cpp
@@ -40,21 +40,15 @@ static MCSymbol *GetSymbolFromOperand(const MachineOperand &MO, AsmPrinter &AP){
   Mangler *Mang = AP.Mang;
   const DataLayout &DL = AP.getDataLayout();
   MCContext &Ctx = AP.OutContext;
-  bool isDarwin = TM.getTargetTriple().isOSDarwin();
 
   SmallString<128> Name;
   StringRef Suffix;
-  if (MO.getTargetFlags() == PPCII::MO_PLT_OR_STUB) {
-    if (isDarwin)
-      Suffix = "$stub";
-  } else if (MO.getTargetFlags() & PPCII::MO_NLP_FLAG)
+  if (MO.getTargetFlags() & PPCII::MO_NLP_FLAG)
     Suffix = "$non_lazy_ptr";
 
   if (!Suffix.empty())
     Name += DL.getPrivateGlobalPrefix();
 
-  unsigned PrefixLen = Name.size();
-
   if (!MO.isGlobal()) {
     assert(MO.isSymbol() && "Isn't a symbol reference");
     Mangler::getNameWithPrefix(Name, MO.getSymbolName(), DL);
@@ -63,42 +57,16 @@ static MCSymbol *GetSymbolFromOperand(const MachineOperand &MO, AsmPrinter &AP){
     TM.getNameWithPrefix(Name, GV, *Mang);
   }
 
-  unsigned OrigLen = Name.size() - PrefixLen;
-
   Name += Suffix;
   MCSymbol *Sym = Ctx.getOrCreateSymbol(Name);
-  StringRef OrigName = StringRef(Name).substr(PrefixLen, OrigLen);
-
-  // If the target flags on the operand changes the name of the symbol, do that
-  // before we return the symbol.
-  if (MO.getTargetFlags() == PPCII::MO_PLT_OR_STUB && isDarwin) {
-    MachineModuleInfoImpl::StubValueTy &StubSym =
-      getMachOMMI(AP).getFnStubEntry(Sym);
-    if (StubSym.getPointer())
-      return Sym;
-    
-    if (MO.isGlobal()) {
-      StubSym =
-      MachineModuleInfoImpl::
-      StubValueTy(AP.getSymbol(MO.getGlobal()),
-                  !MO.getGlobal()->hasInternalLinkage());
-    } else {
-      StubSym =
-      MachineModuleInfoImpl::
-      StubValueTy(Ctx.getOrCreateSymbol(OrigName), false);
-    }
-    return Sym;
-  }
 
   // If the symbol reference is actually to a non_lazy_ptr, not to the symbol,
   // then add the suffix.
   if (MO.getTargetFlags() & PPCII::MO_NLP_FLAG) {
     MachineModuleInfoMachO &MachO = getMachOMMI(AP);
-    
-    MachineModuleInfoImpl::StubValueTy &StubSym =
-      (MO.getTargetFlags() & PPCII::MO_NLP_HIDDEN_FLAG) ? 
-         MachO.getHiddenGVStubEntry(Sym) : MachO.getGVStubEntry(Sym);
-    
+
+    MachineModuleInfoImpl::StubValueTy &StubSym = MachO.getGVStubEntry(Sym);
+
     if (!StubSym.getPointer()) {
       assert(MO.isGlobal() && "Extern symbol not handled yet");
       StubSym = MachineModuleInfoImpl::
@@ -139,7 +107,7 @@ static MCOperand GetSymbolRef(const MachineOperand &MO, const MCSymbol *Symbol,
       break;
   }
 
-  if (MO.getTargetFlags() == PPCII::MO_PLT_OR_STUB && !isDarwin)
+  if (MO.getTargetFlags() == PPCII::MO_PLT)
     RefKind = MCSymbolRefExpr::VK_PLT;
 
   const MCExpr *Expr = MCSymbolRefExpr::create(Symbol, RefKind, Ctx);
diff --git a/lib/Target/PowerPC/PPCMIPeephole.cpp b/lib/Target/PowerPC/PPCMIPeephole.cpp
index fe339d70d7dec..a57a83d7aa93b 100644
--- a/lib/Target/PowerPC/PPCMIPeephole.cpp
+++ b/lib/Target/PowerPC/PPCMIPeephole.cpp
@@ -63,6 +63,8 @@ private:
 public:
   // Main entry point for this pass.
   bool runOnMachineFunction(MachineFunction &MF) override {
+    if (skipFunction(*MF.getFunction()))
+      return false;
     initialize(MF);
     return simplifyCode();
   }
diff --git a/lib/Target/PowerPC/PPCMachineFunctionInfo.h b/lib/Target/PowerPC/PPCMachineFunctionInfo.h
index 10a8ce068d40a..4c29aa06f048f 100644
--- a/lib/Target/PowerPC/PPCMachineFunctionInfo.h
+++ b/lib/Target/PowerPC/PPCMachineFunctionInfo.h
@@ -104,6 +104,10 @@ class PPCFunctionInfo : public MachineFunctionInfo {
   /// Whether this uses the PIC Base register or not.
   bool UsesPICBase;
 
+  /// True if this function has a subset of CSRs that is handled explicitly via
+  /// copies
+  bool IsSplitCSR;
+
 public:
   explicit PPCFunctionInfo(MachineFunction &MF) 
     : FramePointerSaveIndex(0),
@@ -125,7 +129,8 @@ public:
       VarArgsNumFPR(0),
       CRSpillFrameIndex(0),
       MF(MF),
-      UsesPICBase(0) {}
+      UsesPICBase(0),
+      IsSplitCSR(false) {}
 
   int getFramePointerSaveIndex() const { return FramePointerSaveIndex; }
   void setFramePointerSaveIndex(int Idx) { FramePointerSaveIndex = Idx; }
@@ -196,6 +201,9 @@ public:
   void setUsesPICBase(bool uses) { UsesPICBase = uses; }
   bool usesPICBase() const { return UsesPICBase; }
 
+  bool isSplitCSR() const { return IsSplitCSR; }
+  void setIsSplitCSR(bool s) { IsSplitCSR = s; }
+
   MCSymbol *getPICOffsetSymbol() const;
 
   MCSymbol *getGlobalEPSymbol() const;
diff --git a/lib/Target/PowerPC/PPCQPXLoadSplat.cpp b/lib/Target/PowerPC/PPCQPXLoadSplat.cpp
new file mode 100644
index 0000000000000..bfe20c12974b6
--- /dev/null
+++ b/lib/Target/PowerPC/PPCQPXLoadSplat.cpp
@@ -0,0 +1,166 @@
+//===----- PPCQPXLoadSplat.cpp - QPX Load Splat Simplification ------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// The QPX vector registers overlay the scalar floating-point registers, and
+// any scalar floating-point loads splat their value across all vector lanes.
+// Thus, if we have a scalar load followed by a splat, we can remove the splat
+// (i.e. replace the load with a load-and-splat pseudo instruction).
+//
+// This pass must run after anything that might do store-to-load forwarding.
+//
+//===----------------------------------------------------------------------===//
+
+#include "PPC.h"
+#include "PPCInstrBuilder.h"
+#include "PPCInstrInfo.h"
+#include "llvm/ADT/SmallVector.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
+#include "llvm/Support/MathExtras.h"
+#include "llvm/Target/TargetMachine.h"
+#include "llvm/Target/TargetSubtargetInfo.h"
+using namespace llvm;
+
+#define DEBUG_TYPE "ppc-qpx-load-splat"
+
+STATISTIC(NumSimplified, "Number of QPX load splats simplified");
+
+namespace llvm {
+  void initializePPCQPXLoadSplatPass(PassRegistry&);
+}
+
+namespace {
+  struct PPCQPXLoadSplat : public MachineFunctionPass {
+    static char ID;
+    PPCQPXLoadSplat() : MachineFunctionPass(ID) {
+      initializePPCQPXLoadSplatPass(*PassRegistry::getPassRegistry());
+    }
+
+    bool runOnMachineFunction(MachineFunction &Fn) override;
+
+    const char *getPassName() const override {
+      return "PowerPC QPX Load Splat Simplification";
+    }
+  };
+  char PPCQPXLoadSplat::ID = 0;
+}
+
+INITIALIZE_PASS(PPCQPXLoadSplat, "ppc-qpx-load-splat",
+                "PowerPC QPX Load Splat Simplification",
+                false, false)
+
+FunctionPass *llvm::createPPCQPXLoadSplatPass() {
+  return new PPCQPXLoadSplat();
+}
+
+bool PPCQPXLoadSplat::runOnMachineFunction(MachineFunction &MF) {
+  if (skipFunction(*MF.getFunction()))
+    return false;
+
+  bool MadeChange = false;
+  const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
+
+  for (auto MFI = MF.begin(), MFIE = MF.end(); MFI != MFIE; ++MFI) {
+    MachineBasicBlock *MBB = &*MFI;
+    SmallVector<MachineInstr *, 4> Splats;
+
+    for (auto MBBI = MBB->rbegin(); MBBI != MBB->rend(); ++MBBI) {
+      MachineInstr *MI = &*MBBI;
+
+      if (MI->hasUnmodeledSideEffects() || MI->isCall()) {
+        Splats.clear();
+        continue;
+      }
+
+      // We're looking for a sequence like this:
+      // %F0<def> = LFD 0, %X3<kill>, %QF0<imp-def>; mem:LD8[%a](tbaa=!2)
+      // %QF1<def> = QVESPLATI %QF0<kill>, 0, %RM<imp-use>
+
+      for (auto SI = Splats.begin(); SI != Splats.end();) {
+        MachineInstr *SMI = *SI;
+        unsigned SplatReg = SMI->getOperand(0).getReg();
+        unsigned SrcReg = SMI->getOperand(1).getReg();
+
+        if (MI->modifiesRegister(SrcReg, TRI)) {
+          switch (MI->getOpcode()) {
+          default:
+            SI = Splats.erase(SI);
+            continue;
+          case PPC::LFS:
+          case PPC::LFD:
+          case PPC::LFSU:
+          case PPC::LFDU:
+          case PPC::LFSUX:
+          case PPC::LFDUX:
+          case PPC::LFSX:
+          case PPC::LFDX:
+          case PPC::LFIWAX:
+          case PPC::LFIWZX:
+            if (SplatReg != SrcReg) {
+              // We need to change the load to define the scalar subregister of
+              // the QPX splat source register.
+              unsigned SubRegIndex =
+                TRI->getSubRegIndex(SrcReg, MI->getOperand(0).getReg());
+              unsigned SplatSubReg = TRI->getSubReg(SplatReg, SubRegIndex);
+
+              // Substitute both the explicit defined register, and also the
+              // implicit def of the containing QPX register.
+              MI->getOperand(0).setReg(SplatSubReg);
+              MI->substituteRegister(SrcReg, SplatReg, 0, *TRI);
+            }
+
+            SI = Splats.erase(SI);
+
+            // If SMI is directly after MI, then MBBI's base iterator is
+            // pointing at SMI.  Adjust MBBI around the call to erase SMI to
+            // avoid invalidating MBBI.
+            ++MBBI;
+            SMI->eraseFromParent();
+            --MBBI;
+
+            ++NumSimplified;
+            MadeChange = true;
+            continue;
+          }
+        }
+
+        // If this instruction defines the splat register, then we cannot move
+        // the previous definition above it. If it reads from the splat
+        // register, then it must already be alive from some previous
+        // definition, and if the splat register is different from the source
+        // register, then this definition must not be the load for which we're
+        // searching.
+        if (MI->modifiesRegister(SplatReg, TRI) ||
+            (SrcReg != SplatReg &&
+             MI->readsRegister(SplatReg, TRI))) {
+          SI = Splats.erase(SI);
+          continue;
+        }
+
+        ++SI;
+      }
+
+      if (MI->getOpcode() != PPC::QVESPLATI &&
+          MI->getOpcode() != PPC::QVESPLATIs &&
+          MI->getOpcode() != PPC::QVESPLATIb)
+        continue;
+      if (MI->getOperand(2).getImm() != 0)
+        continue;
+
+      // If there are other uses of the scalar value after this, replacing
+      // those uses might be non-trivial.
+      if (!MI->getOperand(1).isKill())
+        continue;
+
+      Splats.push_back(MI);
+    }
+  }
+
+  return MadeChange;
+}
diff --git a/lib/Target/PowerPC/PPCRegisterInfo.cpp b/lib/Target/PowerPC/PPCRegisterInfo.cpp
index 934bdf6224187..f0161a03d2d47 100644
--- a/lib/Target/PowerPC/PPCRegisterInfo.cpp
+++ b/lib/Target/PowerPC/PPCRegisterInfo.cpp
@@ -116,6 +116,9 @@ PPCRegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
                : (Subtarget.hasAltivec() ? CSR_Darwin32_Altivec_SaveList
                                          : CSR_Darwin32_SaveList);
 
+  if (TM.isPPC64() && MF->getInfo<PPCFunctionInfo>()->isSplitCSR())
+    return CSR_SRV464_TLS_PE_SaveList;
+
   // On PPC64, we might need to save r2 (but only if it is not reserved).
   bool SaveR2 = MF->getRegInfo().isAllocatable(PPC::X2);
 
@@ -128,6 +131,31 @@ PPCRegisterInfo::getCalleeSavedRegs(const MachineFunction *MF) const {
                                        : CSR_SVR432_SaveList);
 }
 
+const MCPhysReg *
+PPCRegisterInfo::getCalleeSavedRegsViaCopy(const MachineFunction *MF) const {
+  assert(MF && "Invalid MachineFunction pointer.");
+  const PPCSubtarget &Subtarget = MF->getSubtarget<PPCSubtarget>();
+  if (Subtarget.isDarwinABI())
+    return nullptr;
+  if (!TM.isPPC64())
+    return nullptr;
+  if (MF->getFunction()->getCallingConv() != CallingConv::CXX_FAST_TLS)
+    return nullptr;
+  if (!MF->getInfo<PPCFunctionInfo>()->isSplitCSR())
+    return nullptr;
+
+  // On PPC64, we might need to save r2 (but only if it is not reserved).
+  bool SaveR2 = !getReservedRegs(*MF).test(PPC::X2);
+  if (Subtarget.hasAltivec())
+    return SaveR2
+      ? CSR_SVR464_R2_Altivec_ViaCopy_SaveList
+      : CSR_SVR464_Altivec_ViaCopy_SaveList;
+  else
+    return SaveR2
+      ? CSR_SVR464_R2_ViaCopy_SaveList
+      : CSR_SVR464_ViaCopy_SaveList;
+}
+
 const uint32_t *
 PPCRegisterInfo::getCallPreservedMask(const MachineFunction &MF,
                                       CallingConv::ID CC) const {
@@ -232,16 +260,15 @@ BitVector PPCRegisterInfo::getReservedRegs(const MachineFunction &MF) const {
   if (TFI->needsFP(MF))
     Reserved.set(PPC::R31);
 
+  bool IsPositionIndependent = TM.isPositionIndependent();
   if (hasBasePointer(MF)) {
-    if (Subtarget.isSVR4ABI() && !TM.isPPC64() &&
-        TM.getRelocationModel() == Reloc::PIC_)
+    if (Subtarget.isSVR4ABI() && !TM.isPPC64() && IsPositionIndependent)
       Reserved.set(PPC::R29);
     else
       Reserved.set(PPC::R30);
   }
 
-  if (Subtarget.isSVR4ABI() && !TM.isPPC64() &&
-      TM.getRelocationModel() == Reloc::PIC_)
+  if (Subtarget.isSVR4ABI() && !TM.isPPC64() && IsPositionIndependent)
     Reserved.set(PPC::R30);
 
   // Reserve Altivec registers when Altivec is unavailable.
@@ -907,8 +934,7 @@ unsigned PPCRegisterInfo::getBaseRegister(const MachineFunction &MF) const {
   if (TM.isPPC64())
     return PPC::X30;
 
-  if (Subtarget.isSVR4ABI() &&
-      TM.getRelocationModel() == Reloc::PIC_)
+  if (Subtarget.isSVR4ABI() && TM.isPositionIndependent())
     return PPC::R29;
 
   return PPC::R30;
diff --git a/lib/Target/PowerPC/PPCRegisterInfo.h b/lib/Target/PowerPC/PPCRegisterInfo.h
index b15fde83c9f3a..459502eeb2e9e 100644
--- a/lib/Target/PowerPC/PPCRegisterInfo.h
+++ b/lib/Target/PowerPC/PPCRegisterInfo.h
@@ -75,6 +75,7 @@ public:
 
   /// Code Generation virtual methods...
   const MCPhysReg *getCalleeSavedRegs(const MachineFunction *MF) const override;
+  const MCPhysReg *getCalleeSavedRegsViaCopy(const MachineFunction *MF) const override;
   const uint32_t *getCallPreservedMask(const MachineFunction &MF,
                                        CallingConv::ID CC) const override;
   const uint32_t *getNoPreservedMask() const override;
diff --git a/lib/Target/PowerPC/PPCSchedule.td b/lib/Target/PowerPC/PPCSchedule.td
index d0954a11cd6a9..b4d72eff2b857 100644
--- a/lib/Target/PowerPC/PPCSchedule.td
+++ b/lib/Target/PowerPC/PPCSchedule.td
@@ -70,6 +70,8 @@ def IIC_LdStSTFDU    : InstrItinClass;
 def IIC_LdStSTVEBX   : InstrItinClass;
 def IIC_LdStSTWCX    : InstrItinClass;
 def IIC_LdStSync     : InstrItinClass;
+def IIC_LdStCOPY     : InstrItinClass;
+def IIC_LdStPASTE    : InstrItinClass;
 def IIC_SprISYNC     : InstrItinClass;
 def IIC_SprMFSR      : InstrItinClass;
 def IIC_SprMTMSR     : InstrItinClass;
@@ -104,12 +106,17 @@ def IIC_VecVSR       : InstrItinClass;
 def IIC_SprMTMSRD    : InstrItinClass;
 def IIC_SprSLIE      : InstrItinClass;
 def IIC_SprSLBIE     : InstrItinClass;
+def IIC_SprSLBIEG    : InstrItinClass;
 def IIC_SprSLBMTE    : InstrItinClass;
 def IIC_SprSLBMFEE   : InstrItinClass;
 def IIC_SprSLBIA     : InstrItinClass;
+def IIC_SprSLBSYNC   : InstrItinClass;
 def IIC_SprTLBIA     : InstrItinClass;
 def IIC_SprTLBIEL    : InstrItinClass;
 def IIC_SprTLBIE     : InstrItinClass;
+def IIC_SprABORT     : InstrItinClass;
+def IIC_SprMSGSYNC   : InstrItinClass;
+def IIC_SprSTOP      : InstrItinClass;
 
 //===----------------------------------------------------------------------===//
 // Processor instruction itineraries.
diff --git a/lib/Target/PowerPC/PPCSchedule440.td b/lib/Target/PowerPC/PPCSchedule440.td
index 04a43bc03251a..2455e5e52de5e 100644
--- a/lib/Target/PowerPC/PPCSchedule440.td
+++ b/lib/Target/PowerPC/PPCSchedule440.td
@@ -597,11 +597,12 @@ def PPC440Itineraries : ProcessorItineraries<
 
 def PPC440Model : SchedMachineModel {
   let IssueWidth = 2;  // 2 instructions are dispatched per cycle.
-  let MinLatency = -1; // OperandCycles are interpreted as MinLatency.
   let LoadLatency = 5; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
 
+  let CompleteModel = 0;
+
   let Itineraries = PPC440Itineraries;
 }
 
diff --git a/lib/Target/PowerPC/PPCScheduleA2.td b/lib/Target/PowerPC/PPCScheduleA2.td
index 21a357a2efcf8..54cfae5d74b7b 100644
--- a/lib/Target/PowerPC/PPCScheduleA2.td
+++ b/lib/Target/PowerPC/PPCScheduleA2.td
@@ -1,10 +1,10 @@
 //===- PPCScheduleA2.td - PPC A2 Scheduling Definitions --*- tablegen -*-===//
-// 
+//
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
-// 
+//
 //===----------------------------------------------------------------------===//
 
 // Primary reference:
@@ -160,12 +160,13 @@ def PPCA2Itineraries : ProcessorItineraries<
 
 def PPCA2Model : SchedMachineModel {
   let IssueWidth = 1;  // 1 instruction is dispatched per cycle.
-  let MinLatency = -1; // OperandCycles are interpreted as MinLatency.
   let LoadLatency = 6; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
   let MispredictPenalty = 13;
 
+  let CompleteModel = 0;
+
   let Itineraries = PPCA2Itineraries;
 }
 
diff --git a/lib/Target/PowerPC/PPCScheduleE500mc.td b/lib/Target/PowerPC/PPCScheduleE500mc.td
index 36b8517dabf1f..f687d326b52d9 100644
--- a/lib/Target/PowerPC/PPCScheduleE500mc.td
+++ b/lib/Target/PowerPC/PPCScheduleE500mc.td
@@ -7,9 +7,9 @@
 //
 //===----------------------------------------------------------------------===//
 //
-// This file defines the itinerary class data for the Freescale e500mc 32-bit 
+// This file defines the itinerary class data for the Freescale e500mc 32-bit
 // Power processor.
-// 
+//
 // All information is derived from the "e500mc Core Reference Manual",
 // Freescale Document Number E500MCRM, Rev. 1, 03/2012.
 //
@@ -25,12 +25,12 @@ def E500_DIS1 : FuncUnit; // Dispatch stage - insn 2
 //  * Execute
 //    6 pipelined execution units: SFX0, SFX1, BU, FPU, LSU, CFX.
 //    Some instructions can only execute in SFX0 but not SFX1.
-//    The CFX has a bypass path, allowing non-divide instructions to execute 
+//    The CFX has a bypass path, allowing non-divide instructions to execute
 //    while a divide instruction is executed.
 def E500_SFX0  : FuncUnit; // Simple unit 0
 def E500_SFX1  : FuncUnit; // Simple unit 1
 def E500_BU    : FuncUnit; // Branch unit
-def E500_CFX_DivBypass 
+def E500_CFX_DivBypass
                : FuncUnit; // CFX divide bypass path
 def E500_CFX_0 : FuncUnit; // CFX pipeline
 def E500_LSU_0 : FuncUnit; // LSU pipeline
@@ -271,12 +271,12 @@ def PPCE500mcItineraries : ProcessorItineraries<
                                  [NoBypass, E500_GPR_Bypass]>,
   InstrItinData<IIC_FPGeneral,   [InstrStage<1, [E500_DIS0, E500_DIS1], 0>,
                                   InstrStage<2, [E500_FPU_0]>],
-                                 [11, 1, 1], // Latency = 8, Repeat rate = 2 
+                                 [11, 1, 1], // Latency = 8, Repeat rate = 2
                                  [E500_FPR_Bypass,
                                   E500_FPR_Bypass, E500_FPR_Bypass]>,
   InstrItinData<IIC_FPAddSub,    [InstrStage<1, [E500_DIS0, E500_DIS1], 0>,
                                   InstrStage<4, [E500_FPU_0]>],
-                                 [13, 1, 1], // Latency = 10, Repeat rate = 4 
+                                 [13, 1, 1], // Latency = 10, Repeat rate = 4
                                  [E500_FPR_Bypass,
                                   E500_FPR_Bypass, E500_FPR_Bypass]>,
   InstrItinData<IIC_FPCompare,   [InstrStage<1, [E500_DIS0, E500_DIS1], 0>,
@@ -311,10 +311,11 @@ def PPCE500mcItineraries : ProcessorItineraries<
 
 def PPCE500mcModel : SchedMachineModel {
   let IssueWidth = 2;  // 2 micro-ops are dispatched per cycle.
-  let MinLatency = -1; // OperandCycles are interpreted as MinLatency.
   let LoadLatency = 5; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
 
+  let CompleteModel = 0;
+
   let Itineraries = PPCE500mcItineraries;
 }
diff --git a/lib/Target/PowerPC/PPCScheduleE5500.td b/lib/Target/PowerPC/PPCScheduleE5500.td
index 7c2693ef0d4fe..5db886cf8f940 100644
--- a/lib/Target/PowerPC/PPCScheduleE5500.td
+++ b/lib/Target/PowerPC/PPCScheduleE5500.td
@@ -7,9 +7,9 @@
 //
 //===----------------------------------------------------------------------===//
 //
-// This file defines the itinerary class data for the Freescale e5500 64-bit 
+// This file defines the itinerary class data for the Freescale e5500 64-bit
 // Power processor.
-// 
+//
 // All information is derived from the "e5500 Core Reference Manual",
 // Freescale Document Number e5500RM, Rev. 1, 03/2012.
 //
@@ -25,16 +25,16 @@ def E5500_DIS1 : FuncUnit;
 
 //  * Execute
 //    6 pipelined execution units: SFX0, SFX1, BU, FPU, LSU, CFX.
-//    The CFX has a bypass path, allowing non-divide instructions to execute 
+//    The CFX has a bypass path, allowing non-divide instructions to execute
 //    while a divide instruction is being executed.
 def E5500_SFX0  : FuncUnit; // Simple unit 0
 def E5500_SFX1  : FuncUnit; // Simple unit 1
 def E5500_BU    : FuncUnit; // Branch unit
-def E5500_CFX_DivBypass 
+def E5500_CFX_DivBypass
                 : FuncUnit; // CFX divide bypass path
 def E5500_CFX_0 : FuncUnit; // CFX pipeline stage 0
 
-def E5500_CFX_1 : FuncUnit; // CFX pipeline stage 1 
+def E5500_CFX_1 : FuncUnit; // CFX pipeline stage 1
 
 def E5500_LSU_0 : FuncUnit; // LSU pipeline
 def E5500_FPU_0 : FuncUnit; // FPU pipeline
@@ -331,12 +331,12 @@ def PPCE5500Itineraries : ProcessorItineraries<
                                  [E5500_GPR_Bypass]>,
   InstrItinData<IIC_FPGeneral,   [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
                                   InstrStage<1, [E5500_FPU_0]>],
-                                 [11, 2, 2], // Latency = 7, Repeat rate = 1 
+                                 [11, 2, 2], // Latency = 7, Repeat rate = 1
                                  [E5500_FPR_Bypass,
                                   E5500_FPR_Bypass, E5500_FPR_Bypass]>,
   InstrItinData<IIC_FPAddSub,    [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
                                   InstrStage<1, [E5500_FPU_0]>],
-                                 [11, 2, 2], // Latency = 7, Repeat rate = 1 
+                                 [11, 2, 2], // Latency = 7, Repeat rate = 1
                                  [E5500_FPR_Bypass,
                                   E5500_FPR_Bypass, E5500_FPR_Bypass]>,
   InstrItinData<IIC_FPCompare,   [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
@@ -351,7 +351,7 @@ def PPCE5500Itineraries : ProcessorItineraries<
                                   E5500_FPR_Bypass, E5500_FPR_Bypass]>,
   InstrItinData<IIC_FPDivS,      [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
                                   InstrStage<16, [E5500_FPU_0]>],
-                                 [24, 2, 2], // Latency = 20, Repeat rate = 16 
+                                 [24, 2, 2], // Latency = 20, Repeat rate = 16
                                  [E5500_FPR_Bypass,
                                   E5500_FPR_Bypass, E5500_FPR_Bypass]>,
   InstrItinData<IIC_FPFused,     [InstrStage<1, [E5500_DIS0, E5500_DIS1], 0>,
@@ -371,10 +371,11 @@ def PPCE5500Itineraries : ProcessorItineraries<
 
 def PPCE5500Model : SchedMachineModel {
   let IssueWidth = 2;  // 2 micro-ops are dispatched per cycle.
-  let MinLatency = -1; // OperandCycles are interpreted as MinLatency.
   let LoadLatency = 6; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
 
+  let CompleteModel = 0;
+
   let Itineraries = PPCE5500Itineraries;
 }
diff --git a/lib/Target/PowerPC/PPCScheduleG5.td b/lib/Target/PowerPC/PPCScheduleG5.td
index a3b73ab4454fd..b5a9f96d45aef 100644
--- a/lib/Target/PowerPC/PPCScheduleG5.td
+++ b/lib/Target/PowerPC/PPCScheduleG5.td
@@ -40,7 +40,7 @@ def G5Itineraries : ProcessorItineraries<
   InstrItinData<IIC_IntMulLI    , [InstrStage<4, [G5_IU1, G5_IU2]>]>,
   InstrItinData<IIC_IntRFID     , [InstrStage<1, [G5_IU2]>]>,
   InstrItinData<IIC_IntRotateD  , [InstrStage<2, [G5_IU1, G5_IU2]>]>,
-  InstrItinData<IIC_IntRotateDI , [InstrStage<2, [G5_IU1, G5_IU2]>]>,  
+  InstrItinData<IIC_IntRotateDI , [InstrStage<2, [G5_IU1, G5_IU2]>]>,
   InstrItinData<IIC_IntRotate   , [InstrStage<4, [G5_IU1, G5_IU2]>]>,
   InstrItinData<IIC_IntShift    , [InstrStage<2, [G5_IU1, G5_IU2]>]>,
   InstrItinData<IIC_IntTrapD    , [InstrStage<1, [G5_IU1, G5_IU2]>]>,
@@ -51,14 +51,14 @@ def G5Itineraries : ProcessorItineraries<
   InstrItinData<IIC_BrMCRX      , [InstrStage<3, [G5_BPU]>]>,
   InstrItinData<IIC_LdStDCBF    , [InstrStage<3, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLoad    , [InstrStage<3, [G5_SLU]>]>,
-  InstrItinData<IIC_LdStLoadUpd , [InstrStage<3, [G5_SLU]>]>,  
-  InstrItinData<IIC_LdStLoadUpdX, [InstrStage<3, [G5_SLU]>]>,  
+  InstrItinData<IIC_LdStLoadUpd , [InstrStage<3, [G5_SLU]>]>,
+  InstrItinData<IIC_LdStLoadUpdX, [InstrStage<3, [G5_SLU]>]>,
   InstrItinData<IIC_LdStStore   , [InstrStage<3, [G5_SLU]>]>,
-  InstrItinData<IIC_LdStStoreUpd, [InstrStage<3, [G5_SLU]>]>,  
+  InstrItinData<IIC_LdStStoreUpd, [InstrStage<3, [G5_SLU]>]>,
   InstrItinData<IIC_LdStDSS     , [InstrStage<10, [G5_SLU]>]>,
   InstrItinData<IIC_LdStICBI    , [InstrStage<40, [G5_SLU]>]>,
   InstrItinData<IIC_LdStSTFD    , [InstrStage<4, [G5_SLU]>]>,
-  InstrItinData<IIC_LdStSTFDU   , [InstrStage<4, [G5_SLU]>]>,  
+  InstrItinData<IIC_LdStSTFDU   , [InstrStage<4, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLD      , [InstrStage<3, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLDU     , [InstrStage<3, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLDUX    , [InstrStage<3, [G5_SLU]>]>,
@@ -67,8 +67,8 @@ def G5Itineraries : ProcessorItineraries<
   InstrItinData<IIC_LdStLFDU    , [InstrStage<5, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLFDUX   , [InstrStage<5, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLHA     , [InstrStage<5, [G5_SLU]>]>,
-  InstrItinData<IIC_LdStLHAU    , [InstrStage<5, [G5_SLU]>]>,  
-  InstrItinData<IIC_LdStLHAUX   , [InstrStage<5, [G5_SLU]>]>,  
+  InstrItinData<IIC_LdStLHAU    , [InstrStage<5, [G5_SLU]>]>,
+  InstrItinData<IIC_LdStLHAUX   , [InstrStage<5, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLMW     , [InstrStage<64, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLVecX   , [InstrStage<3, [G5_SLU]>]>,
   InstrItinData<IIC_LdStLWA     , [InstrStage<5, [G5_SLU]>]>,
@@ -118,12 +118,13 @@ def G5Itineraries : ProcessorItineraries<
 
 def G5Model : SchedMachineModel {
   let IssueWidth = 4;  // 4 (non-branch) instructions are dispatched per cycle.
-  let MinLatency = 0;  // Out-of-order dispatch.
   let LoadLatency = 3; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
   let MispredictPenalty = 16;
 
+  let CompleteModel = 0;
+
   let Itineraries = G5Itineraries;
 }
 
diff --git a/lib/Target/PowerPC/PPCScheduleP7.td b/lib/Target/PowerPC/PPCScheduleP7.td
index 267f56726180b..a8678f56900e4 100644
--- a/lib/Target/PowerPC/PPCScheduleP7.td
+++ b/lib/Target/PowerPC/PPCScheduleP7.td
@@ -382,7 +382,6 @@ def P7Model : SchedMachineModel {
                        // branches), but the total internal issue bandwidth per
                        // cycle (from all queues) is 8.
 
-  let MinLatency = 0;  // Out-of-order dispatch.
   let LoadLatency = 3; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
@@ -391,6 +390,8 @@ def P7Model : SchedMachineModel {
   // Try to make sure we have at least 10 dispatch groups in a loop.
   let LoopMicroOpBufferSize = 40;
 
+  let CompleteModel = 0;
+
   let Itineraries = P7Itineraries;
 }
 
diff --git a/lib/Target/PowerPC/PPCScheduleP8.td b/lib/Target/PowerPC/PPCScheduleP8.td
index 69e6d05c66049..8e52da583a0d4 100644
--- a/lib/Target/PowerPC/PPCScheduleP8.td
+++ b/lib/Target/PowerPC/PPCScheduleP8.td
@@ -391,7 +391,6 @@ def P8Model : SchedMachineModel {
                        // up to six non-branch instructions.
                        // up to two branches in a dispatch group.
 
-  let MinLatency = 0;  // Out-of-order dispatch.
   let LoadLatency = 3; // Optimistic load latency assuming bypass.
                        // This is overriden by OperandCycles if the
                        // Itineraries are queried instead.
@@ -400,6 +399,8 @@ def P8Model : SchedMachineModel {
   // Try to make sure we have at least 10 dispatch groups in a loop.
   let LoopMicroOpBufferSize = 60;
 
+  let CompleteModel = 0;
+
   let Itineraries = P8Itineraries;
 }
 
diff --git a/lib/Target/PowerPC/PPCSubtarget.cpp b/lib/Target/PowerPC/PPCSubtarget.cpp
index c0fcb6cbb9dcb..603f0fccc7c63 100644
--- a/lib/Target/PowerPC/PPCSubtarget.cpp
+++ b/lib/Target/PowerPC/PPCSubtarget.cpp
@@ -70,6 +70,8 @@ void PPCSubtarget::initializeEnvironment() {
   HasP8Vector = false;
   HasP8Altivec = false;
   HasP8Crypto = false;
+  HasP9Vector = false;
+  HasP9Altivec = false;
   HasFCPSGN = false;
   HasFSQRT = false;
   HasFRE = false;
@@ -82,7 +84,6 @@ void PPCSubtarget::initializeEnvironment() {
   HasFPRND = false;
   HasFPCVT = false;
   HasISEL = false;
-  HasPOPCNTD = false;
   HasBPERMD = false;
   HasExtDiv = false;
   HasCMPB = false;
@@ -103,12 +104,15 @@ void PPCSubtarget::initializeEnvironment() {
   HasHTM = false;
   HasFusion = false;
   HasFloat128 = false;
+  IsISA3_0 = false;
+
+  HasPOPCNTD = POPCNTD_Unavailable;
 }
 
 void PPCSubtarget::initSubtargetFeatures(StringRef CPU, StringRef FS) {
   // Determine default and user specified characteristics
   std::string CPUName = CPU;
-  if (CPUName.empty()) {
+  if (CPUName.empty() || CPU == "generic") {
     // If cross-compiling with -march=ppc64le without -mcpu
     if (TargetTriple.getArch() == Triple::ppc64le)
       CPUName = "ppc64le";
@@ -142,18 +146,20 @@ void PPCSubtarget::initSubtargetFeatures(StringRef CPU, StringRef FS) {
   IsLittleEndian = (TargetTriple.getArch() == Triple::ppc64le);
 }
 
-/// hasLazyResolverStub - Return true if accesses to the specified global have
-/// to go through a dyld lazy resolution stub.  This means that an extra load
-/// is required to get the address of the global.
+/// Return true if accesses to the specified global have to go through a dyld
+/// lazy resolution stub.  This means that an extra load is required to get the
+/// address of the global.
 bool PPCSubtarget::hasLazyResolverStub(const GlobalValue *GV) const {
-  // We never have stubs if HasLazyResolverStubs=false or if in static mode.
-  if (!HasLazyResolverStubs || TM.getRelocationModel() == Reloc::Static)
-    return false;
-  bool isDecl = GV->isDeclaration();
-  if (GV->hasHiddenVisibility() && !isDecl && !GV->hasCommonLinkage())
+  if (!HasLazyResolverStubs)
     return false;
-  return GV->hasWeakLinkage() || GV->hasLinkOnceLinkage() ||
-         GV->hasCommonLinkage() || isDecl;
+  if (!TM.shouldAssumeDSOLocal(*GV->getParent(), GV))
+    return true;
+  // 32 bit macho has no relocation for a-b if a is undefined, even if b is in
+  // the section that is being relocated. This means we have to use o load even
+  // for GVs that are known to be local to the dso.
+  if (GV->isDeclarationForLinker() || GV->hasCommonLinkage())
+    return true;
+  return false;
 }
 
 // Embedded cores need aggressive scheduling (and some others also benefit).
@@ -166,6 +172,8 @@ static bool needsAggressiveScheduling(unsigned Directive) {
   case PPC::DIR_E5500:
   case PPC::DIR_PWR7:
   case PPC::DIR_PWR8:
+  // FIXME: Same as P8 until POWER9 scheduling info is available
+  case PPC::DIR_PWR9:
     return true;
   }
 }
@@ -191,8 +199,6 @@ void PPCSubtarget::getCriticalPathRCs(RegClassVector &CriticalPathRCs) const {
 }
 
 void PPCSubtarget::overrideSchedPolicy(MachineSchedPolicy &Policy,
-                                       MachineInstr *begin,
-                                       MachineInstr *end,
                                        unsigned NumRegionInstrs) const {
   if (needsAggressiveScheduling(DarwinDirective)) {
     Policy.OnlyTopDown = false;
diff --git a/lib/Target/PowerPC/PPCSubtarget.h b/lib/Target/PowerPC/PPCSubtarget.h
index 4f5c95c1483fa..9fe286a3b7a9a 100644
--- a/lib/Target/PowerPC/PPCSubtarget.h
+++ b/lib/Target/PowerPC/PPCSubtarget.h
@@ -18,9 +18,9 @@
 #include "PPCISelLowering.h"
 #include "PPCInstrInfo.h"
 #include "llvm/ADT/Triple.h"
+#include "llvm/CodeGen/SelectionDAGTargetInfo.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/MC/MCInstrItineraries.h"
-#include "llvm/Target/TargetSelectionDAGInfo.h"
 #include "llvm/Target/TargetSubtargetInfo.h"
 #include <string>
 
@@ -56,6 +56,7 @@ namespace PPC {
     DIR_PWR6X,
     DIR_PWR7,
     DIR_PWR8,
+    DIR_PWR9,
     DIR_64
   };
 }
@@ -64,6 +65,13 @@ class GlobalValue;
 class TargetMachine;
 
 class PPCSubtarget : public PPCGenSubtargetInfo {
+public:
+  enum POPCNTDKind {
+    POPCNTD_Unavailable,
+    POPCNTD_Slow,
+    POPCNTD_Fast
+  };
+
 protected:
   /// TargetTriple - What processor and OS we're targeting.
   Triple TargetTriple;
@@ -92,6 +100,8 @@ protected:
   bool HasP8Vector;
   bool HasP8Altivec;
   bool HasP8Crypto;
+  bool HasP9Vector;
+  bool HasP9Altivec;
   bool HasFCPSGN;
   bool HasFSQRT;
   bool HasFRE, HasFRES, HasFRSQRTE, HasFRSQRTES;
@@ -101,7 +111,6 @@ protected:
   bool HasFPRND;
   bool HasFPCVT;
   bool HasISEL;
-  bool HasPOPCNTD;
   bool HasBPERMD;
   bool HasExtDiv;
   bool HasCMPB;
@@ -122,6 +131,9 @@ protected:
   bool HasHTM;
   bool HasFusion;
   bool HasFloat128;
+  bool IsISA3_0;
+
+  POPCNTDKind HasPOPCNTD;
 
   /// When targeting QPX running a stock PPC64 Linux kernel where the stack
   /// alignment has not been changed, we need to keep the 16-byte alignment
@@ -132,7 +144,7 @@ protected:
   PPCFrameLowering FrameLowering;
   PPCInstrInfo InstrInfo;
   PPCTargetLowering TLInfo;
-  TargetSelectionDAGInfo TSInfo;
+  SelectionDAGTargetInfo TSInfo;
 
 public:
   /// This constructor initializes the data members to match that
@@ -167,7 +179,7 @@ public:
   const PPCTargetLowering *getTargetLowering() const override {
     return &TLInfo;
   }
-  const TargetSelectionDAGInfo *getSelectionDAGInfo() const override {
+  const SelectionDAGTargetInfo *getSelectionDAGInfo() const override {
     return &TSInfo;
   }
   const PPCRegisterInfo *getRegisterInfo() const override {
@@ -230,9 +242,10 @@ public:
   bool hasP8Vector() const { return HasP8Vector; }
   bool hasP8Altivec() const { return HasP8Altivec; }
   bool hasP8Crypto() const { return HasP8Crypto; }
+  bool hasP9Vector() const { return HasP9Vector; }
+  bool hasP9Altivec() const { return HasP9Altivec; }
   bool hasMFOCRF() const { return HasMFOCRF; }
   bool hasISEL() const { return HasISEL; }
-  bool hasPOPCNTD() const { return HasPOPCNTD; }
   bool hasBPERMD() const { return HasBPERMD; }
   bool hasExtDiv() const { return HasExtDiv; }
   bool hasCMPB() const { return HasCMPB; }
@@ -261,6 +274,9 @@ public:
   bool hasHTM() const { return HasHTM; }
   bool hasFusion() const { return HasFusion; }
   bool hasFloat128() const { return HasFloat128; }
+  bool isISA3_0() const { return IsISA3_0; }
+
+  POPCNTDKind hasPOPCNTD() const { return HasPOPCNTD; }
 
   const Triple &getTargetTriple() const { return TargetTriple; }
 
@@ -271,6 +287,7 @@ public:
 
   bool isTargetELF() const { return TargetTriple.isOSBinFormatELF(); }
   bool isTargetMachO() const { return TargetTriple.isOSBinFormatMachO(); }
+  bool isTargetLinux() const { return TargetTriple.isOSLinux(); }
 
   bool isDarwinABI() const { return isTargetMachO() || isDarwin(); }
   bool isSVR4ABI() const { return !isDarwinABI(); }
@@ -286,8 +303,6 @@ public:
   void getCriticalPathRCs(RegClassVector &CriticalPathRCs) const override;
 
   void overrideSchedPolicy(MachineSchedPolicy &Policy,
-                           MachineInstr *begin,
-                           MachineInstr *end,
                            unsigned NumRegionInstrs) const override;
   bool useAA() const override;
 
diff --git a/lib/Target/PowerPC/PPCTLSDynamicCall.cpp b/lib/Target/PowerPC/PPCTLSDynamicCall.cpp
index a9d2e888f4b7c..61ce48ecd04ff 100644
--- a/lib/Target/PowerPC/PPCTLSDynamicCall.cpp
+++ b/lib/Target/PowerPC/PPCTLSDynamicCall.cpp
@@ -73,10 +73,7 @@ protected:
         DebugLoc DL = MI->getDebugLoc();
         unsigned GPR3 = Is64Bit ? PPC::X3 : PPC::R3;
         unsigned Opc1, Opc2;
-        SmallVector<unsigned, 4> OrigRegs;
-        OrigRegs.push_back(OutReg);
-        OrigRegs.push_back(InReg);
-        OrigRegs.push_back(GPR3);
+        const unsigned OrigRegs[] = {OutReg, InReg, GPR3};
 
         switch (MI->getOpcode()) {
         default:
diff --git a/lib/Target/PowerPC/PPCTOCRegDeps.cpp b/lib/Target/PowerPC/PPCTOCRegDeps.cpp
index bf165c9edc6e6..7c53a5601790e 100644
--- a/lib/Target/PowerPC/PPCTOCRegDeps.cpp
+++ b/lib/Target/PowerPC/PPCTOCRegDeps.cpp
@@ -61,10 +61,10 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "PPCInstrInfo.h"
-#include "MCTargetDesc/PPCPredicates.h"
 #include "PPC.h"
+#include "MCTargetDesc/PPCPredicates.h"
 #include "PPCInstrBuilder.h"
+#include "PPCInstrInfo.h"
 #include "PPCMachineFunctionInfo.h"
 #include "PPCTargetMachine.h"
 #include "llvm/ADT/STLExtras.h"
@@ -74,7 +74,6 @@
 #include "llvm/CodeGen/MachineInstr.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/MC/MCAsmInfo.h"
-#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/TargetRegistry.h"
diff --git a/lib/Target/PowerPC/PPCTargetMachine.cpp b/lib/Target/PowerPC/PPCTargetMachine.cpp
index d24b590317f59..a4ff86cb1e217 100644
--- a/lib/Target/PowerPC/PPCTargetMachine.cpp
+++ b/lib/Target/PowerPC/PPCTargetMachine.cpp
@@ -15,7 +15,9 @@
 #include "PPC.h"
 #include "PPCTargetObjectFile.h"
 #include "PPCTargetTransformInfo.h"
+#include "llvm/CodeGen/LiveVariables.h"
 #include "llvm/CodeGen/Passes.h"
+#include "llvm/CodeGen/TargetPassConfig.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/LegacyPassManager.h"
 #include "llvm/MC/MCStreamer.h"
@@ -43,6 +45,10 @@ opt<bool> DisableVSXSwapRemoval("disable-ppc-vsx-swap-removal", cl::Hidden,
                                 cl::desc("Disable VSX Swap Removal for PPC"));
 
 static cl::
+opt<bool> DisableQPXLoadSplat("disable-ppc-qpx-load-splat", cl::Hidden,
+                              cl::desc("Disable QPX load splat simplification"));
+
+static cl::
 opt<bool> DisableMIPeephole("disable-ppc-peephole", cl::Hidden,
                             cl::desc("Disable machine peepholes for PPC"));
 
@@ -172,6 +178,16 @@ static PPCTargetMachine::PPCABI computeTargetABI(const Triple &TT,
   return PPCTargetMachine::PPC_ABI_UNKNOWN;
 }
 
+static Reloc::Model getEffectiveRelocModel(const Triple &TT,
+                                           Optional<Reloc::Model> RM) {
+  if (!RM.hasValue()) {
+    if (TT.isOSDarwin())
+      return Reloc::DynamicNoPIC;
+    return Reloc::Static;
+  }
+  return *RM;
+}
+
 // The FeatureString here is a little subtle. We are modifying the feature
 // string with what are (currently) non-function specific overrides as it goes
 // into the LLVMTargetMachine constructor and then using the stored value in the
@@ -179,10 +195,11 @@ static PPCTargetMachine::PPCABI computeTargetABI(const Triple &TT,
 PPCTargetMachine::PPCTargetMachine(const Target &T, const Triple &TT,
                                    StringRef CPU, StringRef FS,
                                    const TargetOptions &Options,
-                                   Reloc::Model RM, CodeModel::Model CM,
-                                   CodeGenOpt::Level OL)
+                                   Optional<Reloc::Model> RM,
+                                   CodeModel::Model CM, CodeGenOpt::Level OL)
     : LLVMTargetMachine(T, getDataLayoutString(TT), TT, CPU,
-                        computeFSAdditions(FS, OL, TT), Options, RM, CM, OL),
+                        computeFSAdditions(FS, OL, TT), Options,
+                        getEffectiveRelocModel(TT, RM), CM, OL),
       TLOF(createTLOF(getTargetTriple())),
       TargetABI(computeTargetABI(TT, Options)),
       Subtarget(TargetTriple, CPU, computeFSAdditions(FS, OL, TT), *this) {
@@ -214,7 +231,8 @@ void PPC32TargetMachine::anchor() { }
 PPC32TargetMachine::PPC32TargetMachine(const Target &T, const Triple &TT,
                                        StringRef CPU, StringRef FS,
                                        const TargetOptions &Options,
-                                       Reloc::Model RM, CodeModel::Model CM,
+                                       Optional<Reloc::Model> RM,
+                                       CodeModel::Model CM,
                                        CodeGenOpt::Level OL)
     : PPCTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL) {}
 
@@ -223,7 +241,8 @@ void PPC64TargetMachine::anchor() { }
 PPC64TargetMachine::PPC64TargetMachine(const Target &T, const Triple &TT,
                                        StringRef CPU, StringRef FS,
                                        const TargetOptions &Options,
-                                       Reloc::Model RM, CodeModel::Model CM,
+                                       Optional<Reloc::Model> RM,
+                                       CodeModel::Model CM,
                                        CodeGenOpt::Level OL)
     : PPCTargetMachine(T, TT, CPU, FS, Options, RM, CM, OL) {}
 
@@ -245,8 +264,7 @@ PPCTargetMachine::getSubtargetImpl(const Function &F) const {
   // it as a key for the subtarget since that can be the only difference
   // between two functions.
   bool SoftFloat =
-    F.hasFnAttribute("use-soft-float") &&
-    F.getFnAttribute("use-soft-float").getValueAsString() == "true";
+      F.getFnAttribute("use-soft-float").getValueAsString() == "true";
   // If the soft float attribute is set on the function turn on the soft float
   // subtarget feature.
   if (SoftFloat)
@@ -313,9 +331,9 @@ void PPCPassConfig::addIRPasses() {
   if (EnablePrefetch.getNumOccurrences() > 0)
     UsePrefetching = EnablePrefetch;
   if (UsePrefetching)
-    addPass(createPPCLoopDataPrefetchPass());
+    addPass(createLoopDataPrefetchPass());
 
-  if (TM->getOptLevel() == CodeGenOpt::Aggressive && EnableGEPOpt) {
+  if (TM->getOptLevel() >= CodeGenOpt::Default && EnableGEPOpt) {
     // Call SeparateConstOffsetFromGEP pass to extract constants within indices
     // and lower a GEP with multiple indices to either arithmetic operations or
     // multiple GEPs with single index.
@@ -379,18 +397,35 @@ void PPCPassConfig::addMachineSSAOptimization() {
 }
 
 void PPCPassConfig::addPreRegAlloc() {
-  initializePPCVSXFMAMutatePass(*PassRegistry::getPassRegistry());
-  insertPass(VSXFMAMutateEarly ? &RegisterCoalescerID : &MachineSchedulerID,
-             &PPCVSXFMAMutateID);
-  if (getPPCTargetMachine().getRelocationModel() == Reloc::PIC_)
+  if (getOptLevel() != CodeGenOpt::None) {
+    initializePPCVSXFMAMutatePass(*PassRegistry::getPassRegistry());
+    insertPass(VSXFMAMutateEarly ? &RegisterCoalescerID : &MachineSchedulerID,
+               &PPCVSXFMAMutateID);
+  }
+
+  // FIXME: We probably don't need to run these for -fPIE.
+  if (getPPCTargetMachine().isPositionIndependent()) {
+    // FIXME: LiveVariables should not be necessary here!
+    // PPCTLSDYnamicCallPass uses LiveIntervals which previously dependet on
+    // LiveVariables. This (unnecessary) dependency has been removed now,
+    // however a stage-2 clang build fails without LiveVariables computed here.
+    addPass(&LiveVariablesID, false);
     addPass(createPPCTLSDynamicCallPass());
+  }
   if (EnableExtraTOCRegDeps)
     addPass(createPPCTOCRegDepsPass());
 }
 
 void PPCPassConfig::addPreSched2() {
-  if (getOptLevel() != CodeGenOpt::None)
+  if (getOptLevel() != CodeGenOpt::None) {
     addPass(&IfConverterID);
+
+    // This optimization must happen after anything that might do store-to-load
+    // forwarding. Here we're after RA (and, thus, when spills are inserted)
+    // but before post-RA scheduling.
+    if (!DisableQPXLoadSplat)
+      addPass(createPPCQPXLoadSplatPass());
+  }
 }
 
 void PPCPassConfig::addPreEmitPass() {
diff --git a/lib/Target/PowerPC/PPCTargetMachine.h b/lib/Target/PowerPC/PPCTargetMachine.h
index 6496339519a1d..59b4f1e30c0e5 100644
--- a/lib/Target/PowerPC/PPCTargetMachine.h
+++ b/lib/Target/PowerPC/PPCTargetMachine.h
@@ -21,7 +21,7 @@
 
 namespace llvm {
 
-/// PPCTargetMachine - Common code between 32-bit and 64-bit PowerPC targets.
+/// Common code between 32-bit and 64-bit PowerPC targets.
 ///
 class PPCTargetMachine : public LLVMTargetMachine {
 public:
@@ -35,8 +35,9 @@ private:
 
 public:
   PPCTargetMachine(const Target &T, const Triple &TT, StringRef CPU,
-                   StringRef FS, const TargetOptions &Options, Reloc::Model RM,
-                   CodeModel::Model CM, CodeGenOpt::Level OL);
+                   StringRef FS, const TargetOptions &Options,
+                   Optional<Reloc::Model> RM, CodeModel::Model CM,
+                   CodeGenOpt::Level OL);
 
   ~PPCTargetMachine() override;
 
@@ -57,25 +58,25 @@ public:
   };
 };
 
-/// PPC32TargetMachine - PowerPC 32-bit target machine.
+/// PowerPC 32-bit target machine.
 ///
 class PPC32TargetMachine : public PPCTargetMachine {
   virtual void anchor();
 public:
   PPC32TargetMachine(const Target &T, const Triple &TT, StringRef CPU,
                      StringRef FS, const TargetOptions &Options,
-                     Reloc::Model RM, CodeModel::Model CM,
+                     Optional<Reloc::Model> RM, CodeModel::Model CM,
                      CodeGenOpt::Level OL);
 };
 
-/// PPC64TargetMachine - PowerPC 64-bit target machine.
+/// PowerPC 64-bit target machine.
 ///
 class PPC64TargetMachine : public PPCTargetMachine {
   virtual void anchor();
 public:
   PPC64TargetMachine(const Target &T, const Triple &TT, StringRef CPU,
                      StringRef FS, const TargetOptions &Options,
-                     Reloc::Model RM, CodeModel::Model CM,
+                     Optional<Reloc::Model> RM, CodeModel::Model CM,
                      CodeGenOpt::Level OL);
 };
 
diff --git a/lib/Target/PowerPC/PPCTargetObjectFile.cpp b/lib/Target/PowerPC/PPCTargetObjectFile.cpp
index 798bb9d6b8920..8f660355c0acd 100644
--- a/lib/Target/PowerPC/PPCTargetObjectFile.cpp
+++ b/lib/Target/PowerPC/PPCTargetObjectFile.cpp
@@ -53,7 +53,7 @@ MCSection *PPC64LinuxTargetObjectFile::SelectSectionForGlobal(
 const MCExpr *PPC64LinuxTargetObjectFile::
 getDebugThreadLocalSymbol(const MCSymbol *Sym) const {
   const MCExpr *Expr =
-    MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_PPC_DTPREL, getContext());
+    MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_DTPREL, getContext());
   return MCBinaryExpr::createAdd(Expr,
                                  MCConstantExpr::create(0x8000, getContext()),
                                  getContext());
diff --git a/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index cd86dabd5abea..9331e41fb9c19 100644
--- a/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -21,6 +21,12 @@ using namespace llvm;
 static cl::opt<bool> DisablePPCConstHoist("disable-ppc-constant-hoisting",
 cl::desc("disable constant hoisting on PPC"), cl::init(false), cl::Hidden);
 
+// This is currently only used for the data prefetch pass which is only enabled
+// for BG/Q by default.
+static cl::opt<unsigned>
+CacheLineSize("ppc-loop-prefetch-cache-line", cl::Hidden, cl::init(64),
+              cl::desc("The loop prefetch cache line size"));
+
 //===----------------------------------------------------------------------===//
 //
 // PPC cost model.
@@ -30,8 +36,9 @@ cl::desc("disable constant hoisting on PPC"), cl::init(false), cl::Hidden);
 TargetTransformInfo::PopcntSupportKind
 PPCTTIImpl::getPopcntSupport(unsigned TyWidth) {
   assert(isPowerOf2_32(TyWidth) && "Ty width must be power of 2");
-  if (ST->hasPOPCNTD() && TyWidth <= 64)
-    return TTI::PSK_FastHardware;
+  if (ST->hasPOPCNTD() != PPCSubtarget::POPCNTD_Unavailable && TyWidth <= 64)
+    return ST->hasPOPCNTD() == PPCSubtarget::POPCNTD_Slow ?
+             TTI::PSK_SlowHardware : TTI::PSK_FastHardware;
   return TTI::PSK_Software;
 }
 
@@ -230,6 +237,18 @@ unsigned PPCTTIImpl::getRegisterBitWidth(bool Vector) {
 
 }
 
+unsigned PPCTTIImpl::getCacheLineSize() {
+  // This is currently only used for the data prefetch pass which is only
+  // enabled for BG/Q by default.
+  return CacheLineSize;
+}
+
+unsigned PPCTTIImpl::getPrefetchDistance() {
+  // This seems like a reasonable default for the BG/Q (this pass is enabled, by
+  // default, only on the BG/Q).
+  return 300;
+}
+
 unsigned PPCTTIImpl::getMaxInterleaveFactor(unsigned VF) {
   unsigned Directive = ST->getDarwinDirective();
   // The 440 has no SIMD support, but floating-point instructions
@@ -248,8 +267,9 @@ unsigned PPCTTIImpl::getMaxInterleaveFactor(unsigned VF) {
 
   // For P7 and P8, floating-point instructions have a 6-cycle latency and
   // there are two execution units, so unroll by 12x for latency hiding.
-  if (Directive == PPC::DIR_PWR7 ||
-      Directive == PPC::DIR_PWR8)
+  // FIXME: the same for P9 as previous gen until POWER9 scheduling is ready
+  if (Directive == PPC::DIR_PWR7 || Directive == PPC::DIR_PWR8 ||
+      Directive == PPC::DIR_PWR9)
     return 12;
 
   // For most things, modern systems have two execution units (and
@@ -355,7 +375,7 @@ int PPCTTIImpl::getMemoryOpCost(unsigned Opcode, Type *Src, unsigned Alignment,
   // If we can use the permutation-based load sequence, then this is also
   // relatively cheap (not counting loop-invariant instructions): one load plus
   // one permute (the last load in a series has extra cost, but we're
-  // neglecting that here). Note that on the P7, we should do unaligned loads
+  // neglecting that here). Note that on the P7, we could do unaligned loads
   // for Altivec types using the VSX instructions, but that's more expensive
   // than using the permutation-based load sequence. On the P8, that's no
   // longer true.
diff --git a/lib/Target/PowerPC/PPCTargetTransformInfo.h b/lib/Target/PowerPC/PPCTargetTransformInfo.h
index 04c1b02235f0d..5ea9a543cdb18 100644
--- a/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -70,6 +70,8 @@ public:
   bool enableInterleavedAccessVectorization();
   unsigned getNumberOfRegisters(bool Vector);
   unsigned getRegisterBitWidth(bool Vector);
+  unsigned getCacheLineSize();
+  unsigned getPrefetchDistance();
   unsigned getMaxInterleaveFactor(unsigned VF);
   int getArithmeticInstrCost(
       unsigned Opcode, Type *Ty,
diff --git a/lib/Target/PowerPC/PPCVSXCopy.cpp b/lib/Target/PowerPC/PPCVSXCopy.cpp
index 782583ce3423d..60f1ad5585ffc 100644
--- a/lib/Target/PowerPC/PPCVSXCopy.cpp
+++ b/lib/Target/PowerPC/PPCVSXCopy.cpp
@@ -13,11 +13,11 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "PPCInstrInfo.h"
-#include "MCTargetDesc/PPCPredicates.h"
 #include "PPC.h"
+#include "MCTargetDesc/PPCPredicates.h"
 #include "PPCHazardRecognizers.h"
 #include "PPCInstrBuilder.h"
+#include "PPCInstrInfo.h"
 #include "PPCMachineFunctionInfo.h"
 #include "PPCTargetMachine.h"
 #include "llvm/ADT/STLExtras.h"
@@ -28,7 +28,6 @@
 #include "llvm/CodeGen/MachineMemOperand.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/MC/MCAsmInfo.h"
-#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/TargetRegistry.h"
diff --git a/lib/Target/PowerPC/PPCVSXFMAMutate.cpp b/lib/Target/PowerPC/PPCVSXFMAMutate.cpp
index 6b19a2f7118b2..7c22cb22bfa5f 100644
--- a/lib/Target/PowerPC/PPCVSXFMAMutate.cpp
+++ b/lib/Target/PowerPC/PPCVSXFMAMutate.cpp
@@ -38,8 +38,14 @@
 
 using namespace llvm;
 
-static cl::opt<bool> DisableVSXFMAMutate("disable-ppc-vsx-fma-mutation",
-cl::desc("Disable VSX FMA instruction mutation"), cl::Hidden);
+// Temporarily disable FMA mutation by default, since it doesn't handle
+// cross-basic-block intervals well.
+// See: http://lists.llvm.org/pipermail/llvm-dev/2016-February/095669.html
+//      http://reviews.llvm.org/D17087
+static cl::opt<bool> DisableVSXFMAMutate(
+    "disable-ppc-vsx-fma-mutation",
+    cl::desc("Disable VSX FMA instruction mutation"), cl::init(true),
+    cl::Hidden);
 
 #define DEBUG_TYPE "ppc-vsx-fma-mutate"
 
@@ -99,7 +105,7 @@ protected:
         //                         %RM<imp-use>; VSLRC:%vreg16,%vreg18,%vreg9
         // and we remove: %vreg5<def> = COPY %vreg9; VSLRC:%vreg5,%vreg9
 
-        SlotIndex FMAIdx = LIS->getInstructionIndex(MI);
+        SlotIndex FMAIdx = LIS->getInstructionIndex(*MI);
 
         VNInfo *AddendValNo =
           LIS->getInterval(MI->getOperand(1).getReg()).Query(FMAIdx).valueIn();
@@ -168,21 +174,32 @@ protected:
         if (OtherUsers || KillsAddendSrc)
           continue;
 
-        // Find one of the product operands that is killed by this instruction.
 
+        // The transformation doesn't work well with things like:
+        //    %vreg5 = A-form-op %vreg5, %vreg11, %vreg5;
+        // unless vreg11 is also a kill, so skip when it is not,
+        // and check operand 3 to see it is also a kill to handle the case:
+        //   %vreg5 = A-form-op %vreg5, %vreg5, %vreg11;
+        // where vreg5 and vreg11 are both kills. This case would be skipped
+        // otherwise.
+        unsigned OldFMAReg = MI->getOperand(0).getReg();
+
+        // Find one of the product operands that is killed by this instruction.
         unsigned KilledProdOp = 0, OtherProdOp = 0;
-        if (LIS->getInterval(MI->getOperand(2).getReg())
-                     .Query(FMAIdx).isKill()) {
+        unsigned Reg2 = MI->getOperand(2).getReg();
+        unsigned Reg3 = MI->getOperand(3).getReg();
+        if (LIS->getInterval(Reg2).Query(FMAIdx).isKill()
+            && Reg2 != OldFMAReg) {
           KilledProdOp = 2;
           OtherProdOp  = 3;
-        } else if (LIS->getInterval(MI->getOperand(3).getReg())
-                     .Query(FMAIdx).isKill()) {
+        } else if (LIS->getInterval(Reg3).Query(FMAIdx).isKill()
+            && Reg3 != OldFMAReg) {
           KilledProdOp = 3;
           OtherProdOp  = 2;
         }
 
-        // If there are no killed product operands, then this transformation is
-        // likely not profitable.
+        // If there are no usable killed product operands, then this
+        // transformation is likely not profitable.
         if (!KilledProdOp)
           continue;
 
@@ -212,14 +229,6 @@ protected:
         bool KilledProdRegUndef = MI->getOperand(KilledProdOp).isUndef();
         bool OtherProdRegUndef  = MI->getOperand(OtherProdOp).isUndef();
 
-        unsigned OldFMAReg = MI->getOperand(0).getReg();
-
-        // The transformation doesn't work well with things like:
-        //    %vreg5 = A-form-op %vreg5, %vreg11, %vreg5;
-        // so leave such things alone.
-        if (OldFMAReg == KilledProdReg)
-          continue;
-
         // If there isn't a class that fits, we can't perform the transform.
         // This is needed for correctness with a mixture of VSX and Altivec
         // instructions to make sure that a low VSX register is not assigned to
@@ -236,23 +245,33 @@ protected:
         MI->getOperand(0).setReg(KilledProdReg);
         MI->getOperand(1).setReg(KilledProdReg);
         MI->getOperand(3).setReg(AddendSrcReg);
-        MI->getOperand(2).setReg(OtherProdReg);
 
         MI->getOperand(0).setSubReg(KilledProdSubReg);
         MI->getOperand(1).setSubReg(KilledProdSubReg);
         MI->getOperand(3).setSubReg(AddSubReg);
-        MI->getOperand(2).setSubReg(OtherProdSubReg);
 
         MI->getOperand(1).setIsKill(KilledProdRegKill);
         MI->getOperand(3).setIsKill(AddRegKill);
-        MI->getOperand(2).setIsKill(OtherProdRegKill);
 
         MI->getOperand(1).setIsUndef(KilledProdRegUndef);
         MI->getOperand(3).setIsUndef(AddRegUndef);
-        MI->getOperand(2).setIsUndef(OtherProdRegUndef);
 
         MI->setDesc(TII->get(AltOpc));
 
+        // If the addend is also a multiplicand, replace it with the addend
+        // source in both places.
+        if (OtherProdReg == AddendMI->getOperand(0).getReg()) {
+          MI->getOperand(2).setReg(AddendSrcReg);
+          MI->getOperand(2).setSubReg(AddSubReg);
+          MI->getOperand(2).setIsKill(AddRegKill);
+          MI->getOperand(2).setIsUndef(AddRegUndef);
+        } else {
+          MI->getOperand(2).setReg(OtherProdReg);
+          MI->getOperand(2).setSubReg(OtherProdSubReg);
+          MI->getOperand(2).setIsKill(OtherProdRegKill);
+          MI->getOperand(2).setIsUndef(OtherProdRegUndef);
+        }
+
         DEBUG(dbgs() << " -> " << *MI);
 
         // The killed product operand was killed here, so we can reuse it now
@@ -312,7 +331,7 @@ protected:
         // Remove the (now unused) copy.
 
         DEBUG(dbgs() << "  removing: " << *AddendMI << '\n');
-        LIS->RemoveMachineInstrFromMaps(AddendMI);
+        LIS->RemoveMachineInstrFromMaps(*AddendMI);
         AddendMI->eraseFromParent();
 
         Changed = true;
@@ -323,6 +342,9 @@ protected:
 
 public:
     bool runOnMachineFunction(MachineFunction &MF) override {
+      if (skipFunction(*MF.getFunction()))
+        return false;
+
       // If we don't have VSX then go ahead and return without doing
       // anything.
       const PPCSubtarget &STI = MF.getSubtarget<PPCSubtarget>();
diff --git a/lib/Target/PowerPC/PPCVSXSwapRemoval.cpp b/lib/Target/PowerPC/PPCVSXSwapRemoval.cpp
index 27c540fcf2111..d53c8e38254fc 100644
--- a/lib/Target/PowerPC/PPCVSXSwapRemoval.cpp
+++ b/lib/Target/PowerPC/PPCVSXSwapRemoval.cpp
@@ -191,6 +191,9 @@ private:
 public:
   // Main entry point for this pass.
   bool runOnMachineFunction(MachineFunction &MF) override {
+    if (skipFunction(*MF.getFunction()))
+      return false;
+
     // If we don't have VSX on the subtarget, don't do anything.
     const PPCSubtarget &STI = MF.getSubtarget<PPCSubtarget>();
     if (!STI.hasVSX())
@@ -404,9 +407,9 @@ bool PPCVSXSwapRemoval::gatherVectorInstructions() {
       case PPC::VSPLTB:
       case PPC::VSPLTH:
       case PPC::VSPLTW:
+      case PPC::XXSPLTW:
         // Splats are lane-sensitive, but we can use special handling
-        // to adjust the source lane for the splat.  This is not yet
-        // implemented.  When it is, we need to uncomment the following:
+        // to adjust the source lane for the splat.
         SwapVector[VecIdx].IsSwappable = 1;
         SwapVector[VecIdx].SpecialHandling = SHValues::SH_SPLAT;
         break;
@@ -512,7 +515,6 @@ bool PPCVSXSwapRemoval::gatherVectorInstructions() {
       // permute control vectors (for shift values 1, 2, 3).  However,
       // VPERM has a more restrictive register class.
       case PPC::XXSLDWI:
-      case PPC::XXSPLTW:
         break;
       }
     }
@@ -690,6 +692,7 @@ void PPCVSXSwapRemoval::recordUnoptimizableWebs() {
       MachineInstr *MI = SwapVector[EntryIdx].VSEMI;
       unsigned UseReg = MI->getOperand(0).getReg();
       MachineInstr *DefMI = MRI->getVRegDef(UseReg);
+      unsigned DefReg = DefMI->getOperand(0).getReg();
       int DefIdx = SwapMap[DefMI];
 
       if (!SwapVector[DefIdx].IsSwap || SwapVector[DefIdx].IsLoad ||
@@ -705,6 +708,25 @@ void PPCVSXSwapRemoval::recordUnoptimizableWebs() {
         DEBUG(MI->dump());
         DEBUG(dbgs() << "\n");
       }
+
+      // Ensure all uses of the register defined by DefMI feed store
+      // instructions
+      for (MachineInstr &UseMI : MRI->use_nodbg_instructions(DefReg)) {
+        int UseIdx = SwapMap[&UseMI];
+
+        if (SwapVector[UseIdx].VSEMI->getOpcode() != MI->getOpcode()) {
+          SwapVector[Repr].WebRejected = 1;
+
+          DEBUG(dbgs() <<
+                format("Web %d rejected for swap not feeding only stores\n",
+                       Repr));
+          DEBUG(dbgs() << "  def " << " : ");
+          DEBUG(DefMI->dump());
+          DEBUG(dbgs() << "  use " << UseIdx << ": ");
+          DEBUG(SwapVector[UseIdx].VSEMI->dump());
+          DEBUG(dbgs() << "\n");
+        }
+      }
     }
   }
 
@@ -803,12 +825,21 @@ void PPCVSXSwapRemoval::handleSpecialSwappables(int EntryIdx) {
       llvm_unreachable("Unexpected splat opcode");
     case PPC::VSPLTB: NElts = 16; break;
     case PPC::VSPLTH: NElts = 8;  break;
-    case PPC::VSPLTW: NElts = 4;  break;
+    case PPC::VSPLTW:
+    case PPC::XXSPLTW: NElts = 4;  break;
     }
 
-    unsigned EltNo = MI->getOperand(1).getImm();
+    unsigned EltNo;
+    if (MI->getOpcode() == PPC::XXSPLTW)
+      EltNo = MI->getOperand(2).getImm();
+    else
+      EltNo = MI->getOperand(1).getImm();
+
     EltNo = (EltNo + NElts / 2) % NElts;
-    MI->getOperand(1).setImm(EltNo);
+    if (MI->getOpcode() == PPC::XXSPLTW)
+      MI->getOperand(2).setImm(EltNo);
+    else
+      MI->getOperand(1).setImm(EltNo);
 
     DEBUG(dbgs() << "  Into: ");
     DEBUG(MI->dump());
diff --git a/lib/Target/PowerPC/README.txt b/lib/Target/PowerPC/README.txt
index 01233ae4f5782..f1d4ca7b7fab5 100644
--- a/lib/Target/PowerPC/README.txt
+++ b/lib/Target/PowerPC/README.txt
@@ -589,6 +589,17 @@ entry:
 	%tmp34 = zext i1 %tmp3 to i32		; <i32> [#uses=1]
 	ret i32 %tmp34
 }
+
+//===---------------------------------------------------------------------===//
+for the following code:
+
+void foo (float *__restrict__ a, int *__restrict__ b, int n) {
+      a[n] = b[n]  * 2.321;
+}
+
+we load b[n] to GPR, then move it VSX register and convert it float. We should 
+use vsx scalar integer load instructions to avoid direct moves
+
 //===----------------------------------------------------------------------===//
 ; RUN: llvm-as < %s | llc -march=ppc32 | not grep fneg
 
diff --git a/lib/Target/PowerPC/README_P9.txt b/lib/Target/PowerPC/README_P9.txt
new file mode 100644
index 0000000000000..d56f7cca7b211
--- /dev/null
+++ b/lib/Target/PowerPC/README_P9.txt
@@ -0,0 +1,605 @@
+//===- README_P9.txt - Notes for improving Power9 code gen ----------------===//
+
+TODO: Instructions Need Implement Instrinstics or Map to LLVM IR
+
+Altivec:
+- Vector Compare Not Equal (Zero):
+  vcmpneb(.) vcmpneh(.) vcmpnew(.)
+  vcmpnezb(.) vcmpnezh(.) vcmpnezw(.)
+  . Same as other VCMP*, use VCMP/VCMPo form (support intrinsic)
+
+- Vector Extract Unsigned: vextractub vextractuh vextractuw vextractd
+  . Don't use llvm extractelement because they have different semantics
+  . Use instrinstics:
+    (set v2i64:$vD, (int_ppc_altivec_vextractub v16i8:$vA, imm:$UIMM))
+    (set v2i64:$vD, (int_ppc_altivec_vextractuh v8i16:$vA, imm:$UIMM))
+    (set v2i64:$vD, (int_ppc_altivec_vextractuw v4i32:$vA, imm:$UIMM))
+    (set v2i64:$vD, (int_ppc_altivec_vextractd  v2i64:$vA, imm:$UIMM))
+
+- Vector Extract Unsigned Byte Left/Right-Indexed:
+  vextublx vextubrx vextuhlx vextuhrx vextuwlx vextuwrx
+  . Use instrinstics:
+    // Left-Indexed
+    (set i64:$rD, (int_ppc_altivec_vextublx i64:$rA, v16i8:$vB))
+    (set i64:$rD, (int_ppc_altivec_vextuhlx i64:$rA, v8i16:$vB))
+    (set i64:$rD, (int_ppc_altivec_vextuwlx i64:$rA, v4i32:$vB))
+
+    // Right-Indexed
+    (set i64:$rD, (int_ppc_altivec_vextubrx i64:$rA, v16i8:$vB))
+    (set i64:$rD, (int_ppc_altivec_vextuhrx i64:$rA, v8i16:$vB))
+    (set i64:$rD, (int_ppc_altivec_vextuwrx i64:$rA, v4i32:$vB))
+
+- Vector Insert Element Instructions: vinsertb vinsertd vinserth vinsertw
+    (set v16i8:$vD, (int_ppc_altivec_vinsertb v16i8:$vA, imm:$UIMM))
+    (set v8i16:$vD, (int_ppc_altivec_vinsertd v8i16:$vA, imm:$UIMM))
+    (set v4i32:$vD, (int_ppc_altivec_vinserth v4i32:$vA, imm:$UIMM))
+    (set v2i64:$vD, (int_ppc_altivec_vinsertw v2i64:$vA, imm:$UIMM))
+
+- Vector Count Leading/Trailing Zero LSB. Result is placed into GPR[rD]:
+  vclzlsbb vctzlsbb
+  . Use intrinsic:
+    (set i64:$rD, (int_ppc_altivec_vclzlsbb v16i8:$vB))
+    (set i64:$rD, (int_ppc_altivec_vctzlsbb v16i8:$vB))
+
+- Vector Count Trailing Zeros: vctzb vctzh vctzw vctzd
+  . Map to llvm cttz
+    (set v16i8:$vD, (cttz v16i8:$vB))     // vctzb
+    (set v8i16:$vD, (cttz v8i16:$vB))     // vctzh
+    (set v4i32:$vD, (cttz v4i32:$vB))     // vctzw
+    (set v2i64:$vD, (cttz v2i64:$vB))     // vctzd
+
+- Vector Extend Sign: vextsb2w vextsh2w vextsb2d vextsh2d vextsw2d
+  . vextsb2w:
+    (set v4i32:$vD, (sext v4i8:$vB))
+
+    // PowerISA_V3.0:
+    do i = 0 to 3
+       VR[VRT].word[i] ← EXTS32(VR[VRB].word[i].byte[3])
+    end
+
+  . vextsh2w:
+    (set v4i32:$vD, (sext v4i16:$vB))
+
+    // PowerISA_V3.0:
+    do i = 0 to 3
+       VR[VRT].word[i] ← EXTS32(VR[VRB].word[i].hword[1])
+    end
+
+  . vextsb2d
+    (set v2i64:$vD, (sext v2i8:$vB))
+
+    // PowerISA_V3.0:
+    do i = 0 to 1
+       VR[VRT].dword[i] ← EXTS64(VR[VRB].dword[i].byte[7])
+    end
+
+  . vextsh2d
+    (set v2i64:$vD, (sext v2i16:$vB))
+
+    // PowerISA_V3.0:
+    do i = 0 to 1
+       VR[VRT].dword[i] ← EXTS64(VR[VRB].dword[i].hword[3])
+    end
+
+  . vextsw2d
+    (set v2i64:$vD, (sext v2i32:$vB))
+
+    // PowerISA_V3.0:
+    do i = 0 to 1
+       VR[VRT].dword[i] ← EXTS64(VR[VRB].dword[i].word[1])
+    end
+
+- Vector Integer Negate: vnegw vnegd
+  . Map to llvm ineg
+    (set v4i32:$rT, (ineg v4i32:$rA))       // vnegw
+    (set v2i64:$rT, (ineg v2i64:$rA))       // vnegd
+
+- Vector Parity Byte: vprtybw vprtybd vprtybq
+  . Use intrinsic:
+    (set v4i32:$rD, (int_ppc_altivec_vprtybw v4i32:$vB))
+    (set v2i64:$rD, (int_ppc_altivec_vprtybd v2i64:$vB))
+    (set v1i128:$rD, (int_ppc_altivec_vprtybq v1i128:$vB))
+
+- Vector (Bit) Permute (Right-indexed):
+  . vbpermd: Same as "vbpermq", use VX1_Int_Ty2:
+    VX1_Int_Ty2<1484, "vbpermd", int_ppc_altivec_vbpermd, v2i64, v2i64>;
+
+  . vpermr: use VA1a_Int_Ty3
+    VA1a_Int_Ty3<59, "vpermr", int_ppc_altivec_vpermr, v16i8, v16i8, v16i8>;
+
+- Vector Rotate Left Mask/Mask-Insert: vrlwnm vrlwmi vrldnm vrldmi
+  . Use intrinsic:
+    VX1_Int_Ty<389, "vrlwnm", int_ppc_altivec_vrlwnm, v4i32>;
+    VX1_Int_Ty<133, "vrlwmi", int_ppc_altivec_vrlwmi, v4i32>;
+    VX1_Int_Ty<453, "vrldnm", int_ppc_altivec_vrldnm, v2i64>;
+    VX1_Int_Ty<197, "vrldmi", int_ppc_altivec_vrldmi, v2i64>;
+
+- Vector Shift Left/Right: vslv vsrv
+  . Use intrinsic, don't map to llvm shl and lshr, because they have different
+    semantics, e.g. vslv:
+
+      do i = 0 to 15
+         sh ← VR[VRB].byte[i].bit[5:7]
+         VR[VRT].byte[i] ← src.byte[i:i+1].bit[sh:sh+7]
+      end
+
+    VR[VRT].byte[i] is composed of 2 bytes from src.byte[i:i+1]
+
+  . VX1_Int_Ty<1860, "vslv", int_ppc_altivec_vslv, v16i8>;
+    VX1_Int_Ty<1796, "vsrv", int_ppc_altivec_vsrv, v16i8>;
+
+- Vector Multiply-by-10 (& Write Carry) Unsigned Quadword:
+  vmul10uq vmul10cuq
+  . Use intrinsic:
+    VX1_Int_Ty<513, "vmul10uq",   int_ppc_altivec_vmul10uq,  v1i128>;
+    VX1_Int_Ty<  1, "vmul10cuq",  int_ppc_altivec_vmul10cuq, v1i128>;
+
+- Vector Multiply-by-10 Extended (& Write Carry) Unsigned Quadword:
+  vmul10euq vmul10ecuq
+  . Use intrinsic:
+    VX1_Int_Ty<577, "vmul10euq",  int_ppc_altivec_vmul10euq, v1i128>;
+    VX1_Int_Ty< 65, "vmul10ecuq", int_ppc_altivec_vmul10ecuq, v1i128>;
+
+- Decimal Convert From/to National/Zoned/Signed-QWord:
+  bcdcfn. bcdcfz. bcdctn. bcdctz. bcdcfsq. bcdctsq.
+  . Use instrinstics:
+    (set v1i128:$vD, (int_ppc_altivec_bcdcfno  v1i128:$vB, i1:$PS))
+    (set v1i128:$vD, (int_ppc_altivec_bcdcfzo  v1i128:$vB, i1:$PS))
+    (set v1i128:$vD, (int_ppc_altivec_bcdctno  v1i128:$vB))
+    (set v1i128:$vD, (int_ppc_altivec_bcdctzo  v1i128:$vB, i1:$PS))
+    (set v1i128:$vD, (int_ppc_altivec_bcdcfsqo v1i128:$vB, i1:$PS))
+    (set v1i128:$vD, (int_ppc_altivec_bcdctsqo v1i128:$vB))
+
+- Decimal Copy-Sign/Set-Sign: bcdcpsgn. bcdsetsgn.
+  . Use instrinstics:
+    (set v1i128:$vD, (int_ppc_altivec_bcdcpsgno v1i128:$vA, v1i128:$vB))
+    (set v1i128:$vD, (int_ppc_altivec_bcdsetsgno v1i128:$vB, i1:$PS))
+
+- Decimal Shift/Unsigned-Shift/Shift-and-Round: bcds. bcdus. bcdsr.
+  . Use instrinstics:
+    (set v1i128:$vD, (int_ppc_altivec_bcdso  v1i128:$vA, v1i128:$vB, i1:$PS))
+    (set v1i128:$vD, (int_ppc_altivec_bcduso v1i128:$vA, v1i128:$vB))
+    (set v1i128:$vD, (int_ppc_altivec_bcdsro v1i128:$vA, v1i128:$vB, i1:$PS))
+
+  . Note! Their VA is accessed only 1 byte, i.e. VA.byte[7]
+
+- Decimal (Unsigned) Truncate: bcdtrunc. bcdutrunc.
+  . Use instrinstics:
+    (set v1i128:$vD, (int_ppc_altivec_bcdso  v1i128:$vA, v1i128:$vB, i1:$PS))
+    (set v1i128:$vD, (int_ppc_altivec_bcduso v1i128:$vA, v1i128:$vB))
+
+  . Note! Their VA is accessed only 2 byte, i.e. VA.hword[3] (VA.bit[48:63])
+
+VSX:
+- QP Copy Sign: xscpsgnqp
+  . Similar to xscpsgndp
+  . (set f128:$vT, (fcopysign f128:$vB, f128:$vA)
+
+- QP Absolute/Negative-Absolute/Negate: xsabsqp xsnabsqp xsnegqp
+  . Similar to xsabsdp/xsnabsdp/xsnegdp
+  . (set f128:$vT, (fabs f128:$vB))             // xsabsqp
+    (set f128:$vT, (fneg (fabs f128:$vB)))      // xsnabsqp
+    (set f128:$vT, (fneg f128:$vB))             // xsnegqp
+
+- QP Add/Divide/Multiply/Subtract/Square-Root:
+  xsaddqp xsdivqp xsmulqp xssubqp xssqrtqp
+  . Similar to xsadddp
+  . isCommutable = 1
+    (set f128:$vT, (fadd f128:$vA, f128:$vB))   // xsaddqp
+    (set f128:$vT, (fmul f128:$vA, f128:$vB))   // xsmulqp
+
+  . isCommutable = 0
+    (set f128:$vT, (fdiv f128:$vA, f128:$vB))   // xsdivqp
+    (set f128:$vT, (fsub f128:$vA, f128:$vB))   // xssubqp
+    (set f128:$vT, (fsqrt f128:$vB)))           // xssqrtqp
+
+- Round to Odd of QP Add/Divide/Multiply/Subtract/Square-Root:
+  xsaddqpo xsdivqpo xsmulqpo xssubqpo xssqrtqpo
+  . Similar to xsrsqrtedp??
+      def XSRSQRTEDP : XX2Form<60, 74,
+                               (outs vsfrc:$XT), (ins vsfrc:$XB),
+                               "xsrsqrtedp $XT, $XB", IIC_VecFP,
+                               [(set f64:$XT, (PPCfrsqrte f64:$XB))]>;
+
+  . Define DAG Node in PPCInstrInfo.td:
+    def PPCfaddrto: SDNode<"PPCISD::FADDRTO", SDTFPBinOp, []>;
+    def PPCfdivrto: SDNode<"PPCISD::FDIVRTO", SDTFPBinOp, []>;
+    def PPCfmulrto: SDNode<"PPCISD::FMULRTO", SDTFPBinOp, []>;
+    def PPCfsubrto: SDNode<"PPCISD::FSUBRTO", SDTFPBinOp, []>;
+    def PPCfsqrtrto: SDNode<"PPCISD::FSQRTRTO", SDTFPUnaryOp, []>;
+
+    DAG patterns of each instruction (PPCInstrVSX.td):
+    . isCommutable = 1
+      (set f128:$vT, (PPCfaddrto f128:$vA, f128:$vB))   // xsaddqpo
+      (set f128:$vT, (PPCfmulrto f128:$vA, f128:$vB))   // xsmulqpo
+
+    . isCommutable = 0
+      (set f128:$vT, (PPCfdivrto f128:$vA, f128:$vB))   // xsdivqpo
+      (set f128:$vT, (PPCfsubrto f128:$vA, f128:$vB))   // xssubqpo
+      (set f128:$vT, (PPCfsqrtrto f128:$vB))            // xssqrtqpo
+
+- QP (Negative) Multiply-{Add/Subtract}: xsmaddqp xsmsubqp xsnmaddqp xsnmsubqp
+  . Ref: xsmaddadp/xsmsubadp/xsnmaddadp/xsnmsubadp
+
+  . isCommutable = 1
+    // xsmaddqp
+    [(set f128:$vT, (fma f128:$vA, f128:$vB, f128:$vTi))]>,
+    RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+    AltVSXFMARel;
+
+    // xsmsubqp
+    [(set f128:$vT, (fma f128:$vA, f128:$vB, (fneg f128:$vTi)))]>,
+    RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+    AltVSXFMARel;
+
+    // xsnmaddqp
+    [(set f128:$vT, (fneg (fma f128:$vA, f128:$vB, f128:$vTi)))]>,
+    RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+    AltVSXFMARel;
+
+    // xsnmsubqp
+    [(set f128:$vT, (fneg (fma f128:$vA, f128:$vB, (fneg f128:$vTi))))]>,
+    RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+    AltVSXFMARel;
+
+- Round to Odd of QP (Negative) Multiply-{Add/Subtract}:
+  xsmaddqpo xsmsubqpo xsnmaddqpo xsnmsubqpo
+  . Similar to xsrsqrtedp??
+
+  . Define DAG Node in PPCInstrInfo.td:
+    def PPCfmarto: SDNode<"PPCISD::FMARTO", SDTFPTernaryOp, []>;
+
+    It looks like we only need to define "PPCfmarto" for these instructions,
+    because according to PowerISA_V3.0, these instructions perform RTO on
+    fma's result:
+        xsmaddqp(o)
+        v      ← bfp_MULTIPLY_ADD(src1, src3, src2)
+        rnd    ← bfp_ROUND_TO_BFP128(RO, FPSCR.RN, v)
+        result ← bfp_CONVERT_TO_BFP128(rnd)
+
+        xsmsubqp(o)
+        v      ← bfp_MULTIPLY_ADD(src1, src3, bfp_NEGATE(src2))
+        rnd    ← bfp_ROUND_TO_BFP128(RO, FPSCR.RN, v)
+        result ← bfp_CONVERT_TO_BFP128(rnd)
+
+        xsnmaddqp(o)
+        v      ← bfp_MULTIPLY_ADD(src1,src3,src2)
+        rnd    ← bfp_NEGATE(bfp_ROUND_TO_BFP128(RO, FPSCR.RN, v))
+        result ← bfp_CONVERT_TO_BFP128(rnd)
+
+        xsnmsubqp(o)
+        v      ← bfp_MULTIPLY_ADD(src1, src3, bfp_NEGATE(src2))
+        rnd    ← bfp_NEGATE(bfp_ROUND_TO_BFP128(RO, FPSCR.RN, v))
+        result ← bfp_CONVERT_TO_BFP128(rnd)
+
+    DAG patterns of each instruction (PPCInstrVSX.td):
+    . isCommutable = 1
+      // xsmaddqpo
+      [(set f128:$vT, (PPCfmarto f128:$vA, f128:$vB, f128:$vTi))]>,
+      RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+      AltVSXFMARel;
+
+      // xsmsubqpo
+      [(set f128:$vT, (PPCfmarto f128:$vA, f128:$vB, (fneg f128:$vTi)))]>,
+      RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+      AltVSXFMARel;
+
+      // xsnmaddqpo
+      [(set f128:$vT, (fneg (PPCfmarto f128:$vA, f128:$vB, f128:$vTi)))]>,
+      RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+      AltVSXFMARel;
+
+      // xsnmsubqpo
+      [(set f128:$vT, (fneg (PPCfmarto f128:$vA, f128:$vB, (fneg f128:$vTi))))]>,
+      RegConstraint<"$vTi = $vT">, NoEncode<"$vTi">,
+      AltVSXFMARel;
+
+- QP Compare Ordered/Unordered: xscmpoqp xscmpuqp
+  . ref: XSCMPUDP
+      def XSCMPUDP : XX3Form_1<60, 35,
+                               (outs crrc:$crD), (ins vsfrc:$XA, vsfrc:$XB),
+                               "xscmpudp $crD, $XA, $XB", IIC_FPCompare, []>;
+
+  . No SDAG, intrinsic, builtin are required??
+    Or llvm fcmp order/unorder compare??
+
+- DP/QP Compare Exponents: xscmpexpdp xscmpexpqp
+  . No SDAG, intrinsic, builtin are required?
+
+- DP Compare ==, >=, >, !=: xscmpeqdp xscmpgedp xscmpgtdp xscmpnedp
+  . I checked existing instruction "XSCMPUDP". They are different in target
+    register. "XSCMPUDP" write to CR field, xscmp*dp write to VSX register
+
+  . Use instrinsic:
+    (set i128:$XT, (int_ppc_vsx_xscmpeqdp f64:$XA, f64:$XB))
+    (set i128:$XT, (int_ppc_vsx_xscmpgedp f64:$XA, f64:$XB))
+    (set i128:$XT, (int_ppc_vsx_xscmpgtdp f64:$XA, f64:$XB))
+    (set i128:$XT, (int_ppc_vsx_xscmpnedp f64:$XA, f64:$XB))
+
+- Vector Compare Not Equal: xvcmpnedp xvcmpnedp. xvcmpnesp xvcmpnesp.
+  . Similar to xvcmpeqdp:
+      defm XVCMPEQDP : XX3Form_Rcr<60, 99,
+                                 "xvcmpeqdp", "$XT, $XA, $XB", IIC_VecFPCompare,
+                                 int_ppc_vsx_xvcmpeqdp, v2i64, v2f64>;
+
+  . So we should use "XX3Form_Rcr" to implement instrinsic
+
+- Convert DP -> QP: xscvdpqp
+  . Similar to XSCVDPSP:
+      def XSCVDPSP : XX2Form<60, 265,
+                          (outs vsfrc:$XT), (ins vsfrc:$XB),
+                          "xscvdpsp $XT, $XB", IIC_VecFP, []>;
+  . So, No SDAG, intrinsic, builtin are required??
+
+- Round & Convert QP -> DP (dword[1] is set to zero): xscvqpdp xscvqpdpo
+  . Similar to XSCVDPSP
+  . No SDAG, intrinsic, builtin are required??
+
+- Truncate & Convert QP -> (Un)Signed (D)Word (dword[1] is set to zero):
+  xscvqpsdz xscvqpswz xscvqpudz xscvqpuwz
+  . According to PowerISA_V3.0, these are similar to "XSCVDPSXDS", "XSCVDPSXWS",
+    "XSCVDPUXDS", "XSCVDPUXWS"
+
+  . DAG patterns:
+    (set f128:$XT, (PPCfctidz f128:$XB))    // xscvqpsdz
+    (set f128:$XT, (PPCfctiwz f128:$XB))    // xscvqpswz
+    (set f128:$XT, (PPCfctiduz f128:$XB))   // xscvqpudz
+    (set f128:$XT, (PPCfctiwuz f128:$XB))   // xscvqpuwz
+
+- Convert (Un)Signed DWord -> QP: xscvsdqp xscvudqp
+  . Similar to XSCVSXDSP
+  . (set f128:$XT, (PPCfcfids f64:$XB))     // xscvsdqp
+    (set f128:$XT, (PPCfcfidus f64:$XB))    // xscvudqp
+
+- (Round &) Convert DP <-> HP: xscvdphp xscvhpdp
+  . Similar to XSCVDPSP
+  . No SDAG, intrinsic, builtin are required??
+
+- Vector HP -> SP: xvcvhpsp xvcvsphp
+  . Similar to XVCVDPSP:
+      def XVCVDPSP : XX2Form<60, 393,
+                          (outs vsrc:$XT), (ins vsrc:$XB),
+                          "xvcvdpsp $XT, $XB", IIC_VecFP, []>;
+  . No SDAG, intrinsic, builtin are required??
+
+- Round to Quad-Precision Integer: xsrqpi xsrqpix
+  . These are combination of "XSRDPI", "XSRDPIC", "XSRDPIM", .., because you
+    need to assign rounding mode in instruction
+  . Provide builtin?
+    (set f128:$vT, (int_ppc_vsx_xsrqpi f128:$vB))
+    (set f128:$vT, (int_ppc_vsx_xsrqpix f128:$vB))
+
+- Round Quad-Precision to Double-Extended Precision (fp80): xsrqpxp
+  . Provide builtin?
+    (set f128:$vT, (int_ppc_vsx_xsrqpxp f128:$vB))
+
+Fixed Point Facility:
+
+- Exploit cmprb and cmpeqb (perhaps for something like
+  isalpha/isdigit/isupper/islower and isspace respectivelly). This can
+  perhaps be done through a builtin.
+
+- Provide testing for cnttz[dw]
+- Insert Exponent DP/QP: xsiexpdp xsiexpqp
+  . Use intrinsic?
+  . xsiexpdp:
+    // Note: rA and rB are the unsigned integer value.
+    (set f128:$XT, (int_ppc_vsx_xsiexpdp i64:$rA, i64:$rB))
+
+  . xsiexpqp:
+    (set f128:$vT, (int_ppc_vsx_xsiexpqp f128:$vA, f64:$vB))
+
+- Extract Exponent/Significand DP/QP: xsxexpdp xsxsigdp xsxexpqp xsxsigqp
+  . Use intrinsic?
+  . (set i64:$rT, (int_ppc_vsx_xsxexpdp f64$XB))    // xsxexpdp
+    (set i64:$rT, (int_ppc_vsx_xsxsigdp f64$XB))    // xsxsigdp
+    (set f128:$vT, (int_ppc_vsx_xsxexpqp f128$vB))  // xsxexpqp
+    (set f128:$vT, (int_ppc_vsx_xsxsigqp f128$vB))  // xsxsigqp
+
+- Vector Insert Word: xxinsertw
+  - Useful for inserting f32/i32 elements into vectors (the element to be
+    inserted needs to be prepared)
+  . Note: llvm has insertelem in "Vector Operations"
+    ; yields <n x <ty>>
+    <result> = insertelement <n x <ty>> <val>, <ty> <elt>, <ty2> <idx>
+
+    But how to map to it??
+    [(set v1f128:$XT, (insertelement v1f128:$XTi, f128:$XB, i4:$UIMM))]>,
+    RegConstraint<"$XTi = $XT">, NoEncode<"$XTi">,
+
+  . Or use intrinsic?
+    (set v1f128:$XT, (int_ppc_vsx_xxinsertw v1f128:$XTi, f128:$XB, i4:$UIMM))
+
+- Vector Extract Unsigned Word: xxextractuw
+  - Not useful for extraction of f32 from v4f32 (the current pattern is better -
+    shift->convert)
+  - It is useful for (uint_to_fp (vector_extract v4i32, N))
+  - Unfortunately, it can't be used for (sint_to_fp (vector_extract v4i32, N))
+  . Note: llvm has extractelement in "Vector Operations"
+    ; yields <ty>
+    <result> = extractelement <n x <ty>> <val>, <ty2> <idx>
+
+    How to map to it??
+    [(set f128:$XT, (extractelement v1f128:$XB, i4:$UIMM))]
+
+  . Or use intrinsic?
+    (set f128:$XT, (int_ppc_vsx_xxextractuw v1f128:$XB, i4:$UIMM))
+
+- Vector Insert Exponent DP/SP: xviexpdp xviexpsp
+  . Use intrinsic
+    (set v2f64:$XT, (int_ppc_vsx_xviexpdp v2f64:$XA, v2f64:$XB))
+    (set v4f32:$XT, (int_ppc_vsx_xviexpsp v4f32:$XA, v4f32:$XB))
+
+- Vector Extract Exponent/Significand DP/SP: xvxexpdp xvxexpsp xvxsigdp xvxsigsp
+  . Use intrinsic
+    (set v2f64:$XT, (int_ppc_vsx_xvxexpdp v2f64:$XB))
+    (set v4f32:$XT, (int_ppc_vsx_xvxexpsp v4f32:$XB))
+    (set v2f64:$XT, (int_ppc_vsx_xvxsigdp v2f64:$XB))
+    (set v4f32:$XT, (int_ppc_vsx_xvxsigsp v4f32:$XB))
+
+- Test Data Class SP/DP/QP: xststdcsp xststdcdp xststdcqp
+  . No SDAG, intrinsic, builtin are required?
+    Because it seems that we have no way to map BF field?
+
+    Instruction Form: [PO T XO B XO BX TX]
+    Asm: xststd* BF,XB,DCMX
+
+    BF is an index to CR register field.
+
+- Vector Test Data Class SP/DP: xvtstdcsp xvtstdcdp
+  . Use intrinsic
+    (set v4f32:$XT, (int_ppc_vsx_xvtstdcsp v4f32:$XB, i7:$DCMX))
+    (set v2f64:$XT, (int_ppc_vsx_xvtstdcdp v2f64:$XB, i7:$DCMX))
+
+- Maximum/Minimum Type-C/Type-J DP: xsmaxcdp xsmaxjdp xsmincdp xsminjdp
+  . PowerISA_V3.0:
+    "xsmaxcdp can be used to implement the C/C++/Java conditional operation
+     (x>y)?x:y for single-precision and double-precision arguments."
+
+    Note! c type and j type have different behavior when:
+    1. Either input is NaN
+    2. Both input are +-Infinity, +-Zero
+
+  . dtype map to llvm fmaxnum/fminnum
+    jtype use intrinsic
+
+  . xsmaxcdp xsmincdp
+    (set f64:$XT, (fmaxnum f64:$XA, f64:$XB))
+    (set f64:$XT, (fminnum f64:$XA, f64:$XB))
+
+  . xsmaxjdp xsminjdp
+    (set f64:$XT, (int_ppc_vsx_xsmaxjdp f64:$XA, f64:$XB))
+    (set f64:$XT, (int_ppc_vsx_xsminjdp f64:$XA, f64:$XB))
+
+- Vector Byte-Reverse H/W/D/Q Word: xxbrh xxbrw xxbrd xxbrq
+  . Use intrinsic
+    (set v8i16:$XT, (int_ppc_vsx_xxbrh v8i16:$XB))
+    (set v4i32:$XT, (int_ppc_vsx_xxbrw v4i32:$XB))
+    (set v2i64:$XT, (int_ppc_vsx_xxbrd v2i64:$XB))
+    (set v1i128:$XT, (int_ppc_vsx_xxbrq v1i128:$XB))
+
+- Vector Permute: xxperm xxpermr
+  . I have checked "PPCxxswapd" in PPCInstrVSX.td, but they are different
+  . Use intrinsic
+    (set v16i8:$XT, (int_ppc_vsx_xxperm v16i8:$XA, v16i8:$XB))
+    (set v16i8:$XT, (int_ppc_vsx_xxpermr v16i8:$XA, v16i8:$XB))
+
+- Vector Splat Immediate Byte: xxspltib
+  . Similar to XXSPLTW:
+      def XXSPLTW : XX2Form_2<60, 164,
+                           (outs vsrc:$XT), (ins vsrc:$XB, u2imm:$UIM),
+                           "xxspltw $XT, $XB, $UIM", IIC_VecPerm, []>;
+
+  . No SDAG, intrinsic, builtin are required?
+
+- Load/Store Vector: lxv stxv
+  . Has likely SDAG match:
+    (set v?:$XT, (load ix16addr:$src))
+    (set v?:$XT, (store ix16addr:$dst))
+
+  . Need define ix16addr in PPCInstrInfo.td
+    ix16addr: 16-byte aligned, see "def memrix16" in PPCInstrInfo.td
+
+- Load/Store Vector Indexed: lxvx stxvx
+  . Has likely SDAG match:
+    (set v?:$XT, (load xoaddr:$src))
+    (set v?:$XT, (store xoaddr:$dst))
+
+- Load/Store DWord: lxsd stxsd
+  . Similar to lxsdx/stxsdx:
+    def LXSDX : XX1Form<31, 588,
+                        (outs vsfrc:$XT), (ins memrr:$src),
+                        "lxsdx $XT, $src", IIC_LdStLFD,
+                        [(set f64:$XT, (load xoaddr:$src))]>;
+
+  . (set f64:$XT, (load ixaddr:$src))
+    (set f64:$XT, (store ixaddr:$dst))
+
+- Load/Store SP, with conversion from/to DP: lxssp stxssp
+  . Similar to lxsspx/stxsspx:
+    def LXSSPX : XX1Form<31, 524, (outs vssrc:$XT), (ins memrr:$src),
+                         "lxsspx $XT, $src", IIC_LdStLFD,
+                         [(set f32:$XT, (load xoaddr:$src))]>;
+
+  . (set f32:$XT, (load ixaddr:$src))
+    (set f32:$XT, (store ixaddr:$dst))
+
+- Load as Integer Byte/Halfword & Zero Indexed: lxsibzx lxsihzx
+  . Similar to lxsiwzx:
+    def LXSIWZX : XX1Form<31, 12, (outs vsfrc:$XT), (ins memrr:$src),
+                          "lxsiwzx $XT, $src", IIC_LdStLFD,
+                          [(set f64:$XT, (PPClfiwzx xoaddr:$src))]>;
+
+  . (set f64:$XT, (PPClfiwzx xoaddr:$src))
+
+- Store as Integer Byte/Halfword Indexed: stxsibx stxsihx
+  . Similar to stxsiwx:
+    def STXSIWX : XX1Form<31, 140, (outs), (ins vsfrc:$XT, memrr:$dst),
+                          "stxsiwx $XT, $dst", IIC_LdStSTFD,
+                          [(PPCstfiwx f64:$XT, xoaddr:$dst)]>;
+
+  . (PPCstfiwx f64:$XT, xoaddr:$dst)
+
+- Load Vector Halfword*8/Byte*16 Indexed: lxvh8x lxvb16x
+  . Similar to lxvd2x/lxvw4x:
+    def LXVD2X : XX1Form<31, 844,
+                         (outs vsrc:$XT), (ins memrr:$src),
+                         "lxvd2x $XT, $src", IIC_LdStLFD,
+                         [(set v2f64:$XT, (int_ppc_vsx_lxvd2x xoaddr:$src))]>;
+
+  . (set v8i16:$XT, (int_ppc_vsx_lxvh8x xoaddr:$src))
+    (set v16i8:$XT, (int_ppc_vsx_lxvb16x xoaddr:$src))
+
+- Store Vector Halfword*8/Byte*16 Indexed: stxvh8x stxvb16x
+  . Similar to stxvd2x/stxvw4x:
+    def STXVD2X : XX1Form<31, 972,
+                         (outs), (ins vsrc:$XT, memrr:$dst),
+                         "stxvd2x $XT, $dst", IIC_LdStSTFD,
+                         [(store v2f64:$XT, xoaddr:$dst)]>;
+
+  . (store v8i16:$XT, xoaddr:$dst)
+    (store v16i8:$XT, xoaddr:$dst)
+
+- Load/Store Vector (Left-justified) with Length: lxvl lxvll stxvl stxvll
+  . Likely needs an intrinsic
+  . (set v?:$XT, (int_ppc_vsx_lxvl xoaddr:$src))
+    (set v?:$XT, (int_ppc_vsx_lxvll xoaddr:$src))
+
+  . (int_ppc_vsx_stxvl xoaddr:$dst))
+    (int_ppc_vsx_stxvll xoaddr:$dst))
+
+- Load Vector Word & Splat Indexed: lxvwsx
+  . Likely needs an intrinsic
+  . (set v?:$XT, (int_ppc_vsx_lxvwsx xoaddr:$src))
+
+Atomic operations (l[dw]at, st[dw]at):
+- Provide custom lowering for common atomic operations to use these
+  instructions with the correct Function Code
+- Ensure the operands are in the correct register (i.e. RT+1, RT+2)
+- Provide builtins since not all FC's necessarily have an existing LLVM
+  atomic operation
+
+Load Doubleword Monitored (ldmx):
+- Investigate whether there are any uses for this. It seems to be related to
+  Garbage Collection so it isn't likely to be all that useful for most
+  languages we deal with.
+
+Move to CR from XER Extended (mcrxrx):
+- Is there a use for this in LLVM?
+
+Fixed Point Facility:
+
+- Copy-Paste Facility: copy copy_first cp_abort paste paste. paste_last
+  . Use instrinstics:
+    (int_ppc_copy_first i32:$rA, i32:$rB)
+    (int_ppc_copy i32:$rA, i32:$rB)
+
+    (int_ppc_paste i32:$rA, i32:$rB)
+    (int_ppc_paste_last i32:$rA, i32:$rB)
+
+    (int_cp_abort)
+
+- Message Synchronize: msgsync
+- SLB*: slbieg slbsync
+- stop
+  . No instrinstics
diff --git a/lib/Target/PowerPC/TargetInfo/Makefile b/lib/Target/PowerPC/TargetInfo/Makefile
deleted file mode 100644
index 2d0560d275f97..0000000000000
--- a/lib/Target/PowerPC/TargetInfo/Makefile
+++ /dev/null
@@ -1,15 +0,0 @@
-##===- lib/Target/PowerPC/TargetInfo/Makefile --------------*- Makefile -*-===##
-#
-#                     The LLVM Compiler Infrastructure
-#
-# This file is distributed under the University of Illinois Open Source
-# License. See LICENSE.TXT for details.
-#
-##===----------------------------------------------------------------------===##
-LEVEL = ../../../..
-LIBRARYNAME = LLVMPowerPCInfo
-
-# Hack: we need to include 'main' target directory to grab private headers
-override CPPFLAGS += -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/..
-
-include $(LEVEL)/Makefile.common
diff --git a/lib/Target/PowerPC/p9-instrs.txt b/lib/Target/PowerPC/p9-instrs.txt
new file mode 100644
index 0000000000000..a70582aca3989
--- /dev/null
+++ b/lib/Target/PowerPC/p9-instrs.txt
@@ -0,0 +1,442 @@
+Content:
+========
+. Remaining Instructions (Total 56 Instructions, include 2 unknow instructions) 
+. Done (Total 155 Instructions: 101 VSX, 54 Altivec)
+
+//------------------------------------------------------------------------------
+//. Remaining Instructions
+//------------------------------------------------------------------------------
+GCC reference: https://sourceware.org/ml/binutils/2015-11/msg00071.html
+
+// Add PC Immediate Shifted  DX-form p69
+[PO RT d1 d0 XO d2]         addpcis     RT,D
+                            subpcis Rx,value = addpcis Rx,-value
+
+// 6.17.2 Decimal Integer Format Conversion Instructions
+
+// Decimal Convert From National VX-form p352
+[PO VRT EO VRB 1 PS XO]     bcdcfn.     VRT,VRB,PS
+
+// Decimal Convert From Zoned VX-form p353
+[PO VRT EO VRB 1 PS XO]     bcdcfz.     VRT,VRB,PS
+
+// Decimal Convert To National VX-form p354
+[PO VRT EO VRB 1 / XO]      bcdctn.     VRT,VRB
+
+// Decimal Convert To Zoned VX-form p355
+[PO VRT EO VRB 1 PS XO]     bcdctz.     VRT,VRB,PS
+
+// Decimal Convert From Signed Quadword VX-form p356
+[PO VRT EO VRB 1 PS XO]     bcdcfsq.    VRT,VRB,PS
+
+// Decimal Convert To Signed Quadword VX-form p356
+[PO VRT EO VRB 1 / XO]      bcdctsq.    VRT,VRB
+
+// 6.17.3 Decimal Integer Sign Manipulation Instructions
+
+// Decimal Copy Sign VX-form p358
+[PO VRT VRA VRB XO]         bcdcpsgn.   VRT,VRA,VRB
+
+// Decimal Set Sign VX-form p358
+[PO VRT EO VRB 1 PS XO]     bcdsetsgn.  VRT,VRB,PS
+
+// Decimal Shift VX-form p359
+[PO VRT VRA VRB 1 PS XO]    bcds.       VRT,VRA,VRB,PS
+
+// Decimal Unsigned Shift VX-form p360
+[PO VRT VRA VRB 1 / XO]     bcdus.      VRT,VRA,VRB
+
+// Decimal Shift and Round VX-form p361
+[PO VRT VRA VRB 1 PS XO]    bcdsr.      VRT,VRA,VRB,PS
+
+// 6.17.5 Decimal Integer Truncate Instructions
+
+// Decimal Truncate VX-form p362
+[PO VRT VRA VRB 1 PS XO]    bcdtrunc.   VRT,VRA,VRB,PS
+
+// Decimal Unsigned Truncate VX-form p363
+[PO VRT VRA VRB 1 / XO]     bcdutrunc.  VRT,VRA,VRB
+
+// 3.3.10.1 Character-Type Compare Instructions
+
+// Compare Ranged Byte X-form p87
+[PO BF / L RA RB XO /]      cmprb       BF,L,RA,RB
+
+// Compare Equal Byte X-form p88
+[PO BF // RA RB XO /]       cmpeqb      BF,RA,RB
+
+// 3.3.13 Fixed-Point Logical Instructions
+
+// Count Trailing Zeros Word X-form p95
+[PO RS RA /// XO Rc]        cnttzw(.)   RA,RS
+
+// 3.3.13.1 64-bit Fixed-Point Logical Instructions
+
+// Count Trailing Zeros Doubleword  X-form p98
+[PO RS RA /// XO Rc]        cnttzd(.)   RA,RS
+
+// 4.4 Copy-Paste Facility
+
+// Copy X-form p858
+[PO /// L RA RB XO /]       copy        RA,RB,L
+                            copy_first = copy RA, RB, 1
+// CP_Abort p860
+[PO /// /// /// XO /]       cp_abort
+
+// Paste p859
+[PO /// L RA RB XO Rc]      paste(.)    RA,RB,L
+                            paste_last = paste RA,RB,1
+
+// 3.3.9 Fixed-Point Arithmetic Instructions
+
+// Deliver A Random Number X-form p79
+[PO RT /// L /// XO /]      darn        RT,L
+
+// Multiply-Add High Doubleword VA-form p81
+[PO RT RA RB RC XO]         maddhd      RT,RA.RB,RC
+
+// Multiply-Add High Doubleword Unsigned VA-form  p81
+[PO RT RA RB RC XO]         maddhdu     RT,RA.RB,RC
+
+// Multiply-Add Low Doubleword VA-form p81
+[PO RT RA RB RC XO]         maddld      RT,RA.RB,RC
+
+// Modulo Signed Word X-form p76
+[PO RT RA RB XO /]          modsw       RT,RA,RB
+
+// Modulo Unsigned Word X-form p76
+[PO RT RA RB XO /]          moduw       RT,RA,RB 
+
+// Modulo Signed Doubleword X-form p84
+[PO RT RA RB XO /]          modsd       RT,RA,RB
+
+// Modulo Unsigned Doubleword X-form p84
+[PO RT RA RB XO /]          modud       RT,RA,RB
+
+
+// DFP Test Significance Immediate [Quad] X-form p204
+[PO BF / UIM FRB XO /]      dtstsfi     BF,UIM,FRB
+[PO BF / UIM FRBp XO /]     dtstsfiq    BF,UIM,FRBp
+
+// 3.3.14.2.1 64-bit Fixed-Point Shift Instructions
+
+// Extend-Sign Word and Shift Left Immediate XS-form p109
+[PO RS RA sh XO sh Rc]      extswsli(.) RA,RS,SH
+
+// 4.5.1 Load Atomic
+
+// Load Word Atomic   X-form p864
+[PO RT RA FC XO /]          lwat        RT,RA,FC
+
+// Load Doubleword Atomic X-form p864
+[PO RT RA FC XO /]          ldat        RT,RA,FC
+
+// 4.5.2 Store Atomic
+
+// Store Word Atomic   X-form p866
+[PO RS RA FC XO /]          stwat       RS,RA,FC
+
+// Store Doubleword Atomic   X-form p866
+[PO RS RA FC XO /]          stdat       RS,RA,FC
+
+// 3.3.2.1 64-bit Fixed-Point Load Instructions 
+
+// Load Doubleword Monitored Indexed X-form p54
+[PO RT RA RB XO /]          ldmx        RT,RA,RB
+
+// 3.3.16 Move To/From Vector-Scalar Register Instructions
+
+// Move From VSR Lower Doubleword XX1-form p111
+[PO S RA /// XO SX]         mfvsrld     RA,XS
+
+// Move To VSR Double Doubleword XX1-form p114
+[PO T RA RB XO TX]          mtvsrdd     XT,RA,RB
+
+// Move To VSR Word & Splat XX1-form p115
+[PO T RA /// XO TX]         mtvsrws     XT,RA
+
+// Move to CR from XER Extended X-form p119
+[PO BF // /// /// XO /]     mcrxrx      BF
+
+// Set Boolean X-form p121
+[PO RT BFA // /// XO /]     setb        RT,BFA
+
+// Message Synchronize X-form p1126
+[PO /// /// /// XO /]       msgsync
+
+// SLB Invalidate Entry Global  X-form p1026
+[PO RS /// RB XO /]         slbieg      RS,RB 
+
+// SLB Synchronize  X-form p1031
+[PO /// /// /// XO /]       slbsync
+
+// 3.3.2.1 Power-Saving Mode Instruction
+
+// stop    XL-form p957
+[PO /// /// /// XO /]       stop
+
+// 4.6.4 Wait Instruction
+// Wait X-form p880
+[PO /// WC /// /// XO /]    wait
+
+// Unknow Instructions:
+urfid
+- gcc's implementation:
+    {"urfid",	XL(19,306),	0xffffffff,  POWER9,	PPCNONE,	{0}},
+    (4c 00 02 64|64 02 00 4c) 	urfid
+
+rmieg
+- gcc's implementation: 
+    {"rmieg",	X(31,882),	XRTRA_MASK,  POWER9,	PPCNONE,	{RB}},
+    (7c 00 f6 e4|e4 f6 00 7c) 	rmieg   r30
+
+//------------------------------------------------------------------------------
+//. Done:
+//------------------------------------------------------------------------------
+
+//======================================
+"vsx instructions"
+
+//--------------------------------------
+"7.6.1.2.1 VSX Scalar Move Instructions"
+// VSX Scalar Quad-Precision Move Instructions
+
+// VSX Scalar Copy Sign Quad-Precision X-form p.553
+[PO VRT VRA VRB XO /] xscpsgnqp
+
+// VSX Scalar Absolute Quad-Precision X-form 531
+// VSX Scalar Negate Quad-Precision X-form 627
+// VSX Scalar Negative Absolute Quad-Precision X-form 626
+[PO VRT XO VRB XO /] xsabsqp xsnegqp xsnabsqp
+
+//--------------------------------------
+"7.6.1.3 VSX Floating-Point Arithmetic Instructions"
+
+// VSX Scalar Quad-Precision Elementary Arithmetic
+
+// VSX Scalar Add Quad-Precision [using round to Odd] X-form 539
+// VSX Scalar Divide Quad-Precision [using round to Odd] X-form 584
+// VSX Scalar Multiply Quad-Precision [using round to Odd] X-form 622
+[PO VRT VRA VRB XO RO] xsaddqp xsaddqpo xsdivqp xsdivqpo xsmulqp xsmulqpo
+
+// VSX Scalar Square Root Quad-Precision [using round to Odd] X-form 662
+// VSX Scalar Subtract Quad-Precision [using round to Odd] X-form 667
+                       xssubqp xssubqpo
+
+[PO VRT XO VRB XO RO] xssqrtqp xssqrtqpo
+
+// VSX Scalar Quad-Precision Multiply-Add Arithmetic Instructions
+
+// VSX Scalar Multiply-Add Quad-Precision [using round to Odd] X-form 596
+// VSX Scalar Multiply-Subtract Quad-Precision [using round to Odd] X-form 617
+// VSX Scalar Negative Multiply-Add Quad-Precision [using round to Odd] X-form 636
+// VSX Scalar Negative Multiply-Subtract Quad-Precision [using round to Odd]
+// X-form 645
+[PO VRT VRA VRB XO RO] xsmaddqp xsmaddqpo xsmsubqp xsmsubqpo 
+                       xsnmaddqp xsnmaddqpo xsnmsubqp xsnmsubqpo
+
+22
+//--------------------------------------
+"7.6.1.4 VSX Floating-Point Compare Instructions"
+
+// VSX Scalar Quad-Precision Compare Instructions
+
+// VSX Scalar Compare Ordered Quad-Precision X-form 549
+// VSX Scalar Compare Unordered Quad-Precision X-form 552
+[PO BF // VRA VRB XO /] xscmpoqp xscmpuqp 
+
+"7.6.1.8 VSX Scalar Floating-Point Support Instructions"
+// VSX Scalar Compare Exponents Quad-Precision X-form p. 541 542
+[PO BF // A B XO AX BX /] xscmpexpdp 
+[PO BF // VRA VRB XO /] xscmpexpqp
+
+// VSX Scalar Compare DP, XX3-form, p.543 544 545
+// VSX Scalar Compare Equal Double-Precision, 
+[PO T A B XO AX BX TX]  xscmpeqdp xscmpgedp xscmpgtdp xscmpnedp
+
+// VSX Vector Compare Not Equal Double-Precision XX3-form 691
+[PO T A B Rc XO AX BX TX] xvcmpnedp xvcmpnedp. xvcmpnesp xvcmpnesp.
+
+//--------------------------------------
+"7.6.1.5 VSX FP-FP Conversion Instructions"
+// VSX Scalar Quad-Precision Floating-Point Conversion Instructions
+
+// VSX Scalar round & Convert Quad-Precision format to Double-Precision format
+// [using round to Odd] X-form 567
+[PO VRT XO VRB XO /] xscvqpdp xscvqpdpo (actually [PO VRT XO VRB XO RO])
+[PO VRT XO VRB XO /] xscvdpqp
+
+// VSX Scalar Quad-Precision Convert to Integer Instructions
+
+// VSX Scalar truncate & Convert Quad-Precision format to Signed Doubleword format
+// 568 570 572 574
+[PO VRT XO VRB XO /] xscvqpsdz xscvqpswz xscvqpudz xscvqpuwz
+576 = 580            xscvsdqp xscvudqp
+
+"7.6.1.7 VSX Round to Floating-Point Integer Instructions"
+// VSX Scalar round & Convert Double-Precision format to Half-Precision format
+// XX2-form 554 566
+[PO T XO B XO BX TX] xscvdphp xscvhpdp
+
+// VSX Vector Convert Half-Precision format to Single-Precision format
+// XX2-form 703 705
+[PO T XO B XO BX TX] xvcvhpsp xvcvsphp
+
+// VSX Scalar Round to Quad-Precision Integer [with Inexact] Z23-form 654
+[PO VRT /// R VRB RMC XO EX] xsrqpi xsrqpix
+
+// VSX Scalar Round Quad-Precision to Double-Extended Precision Z23-form 656
+[PO VRT /// R VRB RMC XO /] xsrqpxp 
+def XSRQPXP : Z23Form_1<63, 37,
+                        (outs vrrc:$vT), (ins u5imm:$R, vrrc:$vB, u2imm:$RMC),
+                        "xsrqpxp $vT, $R, $vB, $RMC"), IIC_VecFP, []>;
+
+27~28
+//--------------------------------------
+// VSX Scalar Insert Exponent Double-Precision X-form 588
+// VSX Scalar Insert Exponent Quad-Precision X-form 589
+[PO VT rA rB XO /]  xsiexpdp 
+[PO VRT VRA VRB XO /]  xsiexpqp
+
+// VSX Vector Insert Exponent Double-Precision XX3-form 722
+[PO T A B XO AX BX TX] xviexpdp xviexpsp
+
+// VSX Vector Extract Unsigned Word XX2-form 788
+// VSX Vector Insert Word XX2-form
+[PO T / UIM B XO BX TX] xxextractuw xxinsertw
+
+// VSX Scalar Extract Exponent Double-Precision XX2-form 676
+[PO BF DCMX B XO BX /]  
+[PO T XO B XO BX /] xsxexpdp xsxsigdp
+// X-form
+[PO VRT XO VRB XO /] xsxexpqp xsxsigqp
+
+// VSX Vector Extract Exponent Double-Precision XX2-form 784
+[PO T XO B XO BX TX] xvxexpdp xvxexpsp
+
+// VSX Vector Extract Significand Double-Precision XX2-form 785
+[PO T XO B XO BX TX] xvxsigdp xvxsigsp
+
+//--------------------------------------
+// VSX Scalar Test Data Class Double-Precision XX2-form p673
+// VSX Scalar Test Data Class Quad-Precision X-form 674
+// VSX Scalar Test Data Class Single-Precision XX2-form 675
+[PO BF DCMX B XO BX /]  xststdcdp xststdcsp
+[PO BF DCMX VRB XO /]   xststdcqp 
+
+// VSX Vector Test Data Class Double-Precision XX2-form 782 783
+[PO T dx B XO dc XO dm BX TX] xvtstdcdp xvtstdcsp 
+
+//--------------------------------------
+// VSX Scalar Maximum Type-C Double-Precision XX3-form 601 ~ 609
+[PO T A B XO AX BX TX] xsmaxcdp xsmaxjdp xsmincdp xsminjdp
+
+//--------------------------------------
+// VSX Vector Byte-Reverse Doubleword XX2-form 786 787
+[PO T XO B XO BX TX] xxbrd xxbrh xxbrq xxbrw
+
+// VSX Vector Permute XX3-form 794
+[PO T A B XO AX BX TX] xxperm xxpermr
+
+// VSX Vector Splat Immediate Byte 796 x-form
+[PO T EO IMM8 XO TX] xxspltib   <= sign or unsigned?
+
+30
+//--------------------------------------
+// Load VSX Vector DQ-form 511
+[PO T RA DQ TX XO] lxv 
+
+// Store VSX Vector DQ-form 526
+[PO S RA DQ SX XO] stxv
+
+// Load VSX Scalar Doubleword DS-form 499
+// Load VSX Scalar Single DS-form 504
+[PO VRT RA DS XO] lxsd lxssp 
+
+// Store VSX Scalar Doubleword DS-form 517
+// Store VSX Scalar Single DS-form 520
+[PO VRT RA DS XO] stxsd stxssp
+
+
+// Load VSX Vector Indexed X-form 511
+// Load VSX Scalar as Integer Byte & Zero Indexed X-form 501
+// Load VSX Vector Byte*16 Indexed X-form 506
+// Load VSX Vector with Length X-form 508
+// Load VSX Vector Left-justified with Length X-form 510
+// Load VSX Vector Halfword*8 Indexed X-form 514
+// Load VSX Vector Word & Splat Indexed X-form 516
+[PO T RA RB XO TX] lxvx lxsibzx lxsihzx lxvb16x lxvl lxvll lxvh8x lxvwsx
+
+// Store VSX Scalar as Integer Byte Indexed X-form 518
+// Store VSX Scalar as Integer Halfword Indexed X-form 518
+// Store VSX Vector Byte*16 Indexed X-form 522
+// Store VSX Vector Halfword*8 Indexed X-form 524
+// Store VSX Vector with Length X-form 526
+// Store VSX Vector Left-justified with Length X-form 528
+// Store VSX Vector Indexed X-form 529
+[PO S RA RB XO SX] stxsibx stxsihx stxvb16x stxvh8x stxvl stxvll stxvx
+
+21
+
+//--------------------------------------
+". vector instructions"
+
+[1] PowerISA-v3.0 p.933 - Table 1, and Chapter 6. Vector Facility (altivec)
+[2] https://sourceware.org/ml/binutils/2015-11/msg00071.html
+
+//--------------------------------------
+New patch:
+// vector bit, p.367, 6.16 Vector Bit Permute Instruction
+[PO VRT VRA VRB XO] vbpermd, (existing: vbpermq)
+
+// vector permute, p.280
+[PO VRT VRA VRB VRC XO] vpermr
+
+// vector rotate left, p.341
+[PO VRT VRA VRB XO] vrlwnm vrlwmi vrldnm vrldmi
+
+// vector shift, p.285
+[PO VRT VRA VRB XO] vslv vsrv
+
+// vector multiply-by-10, p.375
+[PO VRT VRA /// XO] vmul10cuq vmul10uq
+[PO VRT VRA VRB XO] vmul10ecuq vmul10euq 
+
+12
+//--------------------------------------
+http://reviews.llvm.org/D15887 + ext + neg + prty - vbpermd
+// vector count leading/trailing zero
+. new vx-form: p.31, 1.6.14 VX-FORM
+[PO RT EO VRB XO] vclzlsbb vctzlsbb (p.363)
+
+// Vector Count Trailing Zeros Instructions, 362
+[PO VRT EO VRB XO] vctzb vctzh vctzw vctzd (v16i8 v8i16 v4i32 v2i64)
+
+// vector extend sign (p.314)
+[PO VRT EO VRB XO] vextsb2w vextsh2w vextsb2d vextsh2d vextsw2d
+
+// vector negate, p.313
+[PO VRT EO VRB XO] vnegd vnegw
+
+// vector parity, p.335
+[PO VRT EO VRB XO] vprtybd vprtybq vprtybw
+
+16
+//--------------------------------------
+// vector compare, p.330
+[PO VRT VRA VRB RC XO] vcmpneb vcmpneb. vcmpneh vcmpneh. vcmpnew vcmpnew.
+                       vcmpnezb vcmpnezb. vcmpnezh vcmpnezh. vcmpnezw vcmpnezw.
+12
+//--------------------------------------
+http://reviews.llvm.org/D15917 + insert
+// vector extract (p.287) ref: vspltb (v2.07, p.227)
+// vector insert, p.288
+[PO VRT / UIM VRB XO] vinsertb vinsertd vinserth vinsertw
+
+// Vector Extract Unsigned
+[PO VRT / UIM VRB XO] vextractub vextractuh vextractuw vextractd
+
+// p.364: Vector Extract Unsigned Left/Right-Indexed
+[PO RT RA VRB XO] vextublx vextubrx vextuhlx vextuhrx vextuwlx vextuwrx 
+
+14