MFC 261991:

Upgrade our copy of llvm/clang to 3.4 release.  This version supports
all of the features in the current working draft of the upcoming C++
standard, provisionally named C++1y.

The code generator's performance is greatly increased, and the loop
auto-vectorizer is now enabled at -Os and -O2 in addition to -O3.  The
PowerPC backend has made several major improvements to code generation
quality and compile time, and the X86, SPARC, ARM32, Aarch64 and SystemZ
backends have all seen major feature work.

Release notes for llvm and clang can be found here:
<http://llvm.org/releases/3.4/docs/ReleaseNotes.html>
<http://llvm.org/releases/3.4/tools/clang/docs/ReleaseNotes.html>

MFC 262121 (by emaste):

Update lldb for clang/llvm 3.4 import

This commit largely restores the lldb source to the upstream r196259
snapshot with the addition of threaded inferior support and a few bug
fixes.

Specific upstream lldb revisions restored include:
   SVN      git
  181387  779e6ac
  181703  7bef4e2
  182099  b31044e
  182650  f2dcf35
  182683  0d91b80
  183862  15c1774
  183929  99447a6
  184177  0b2934b
  184948  4dc3761
  184954  007e7bc
  186990  eebd175

Sponsored by:	DARPA, AFRL

MFC 262186 (by emaste):

Fix mismerge in r262121

A break statement was lost in the merge.  The error had no functional
impact, but restore it to reduce the diff against upstream.

MFC 262303:

Pull in r197521 from upstream clang trunk (by rdivacky):

  Use the integrated assembler by default on FreeBSD/ppc and ppc64.

Requested by:	jhibbits

MFC 262611:

Pull in r196874 from upstream llvm trunk:

  Fix a crash that occurs when PWD is invalid.

  MCJIT needs to be able to run in hostile environments, even when PWD
  is invalid. There's no need to crash MCJIT in this case.

  The obvious fix is to simply leave MCContext's CompilationDir empty
  when PWD can't be determined. This way, MCJIT clients,
  and other clients that link with LLVM don't need a valid working directory.

  If we do want to guarantee valid CompilationDir, that should be done
  only for clients of getCompilationDir(). This is as simple as checking
  for an empty string.

  The only current use of getCompilationDir is EmitGenDwarfInfo, which
  won't conceivably run with an invalid working dir. However, in the
  purely hypothetically and untestable case that this happens, the
  AT_comp_dir will be omitted from the compilation_unit DIE.

This should help fix assertions occurring with ports-mgmt/tinderbox,
when it is using jails, and sometimes invalidates clang's current
working directory.

Reported by:	decke

MFC 262809:

Pull in r203007 from upstream clang trunk:

  Don't produce an alias between destructors with different calling conventions.

  Fixes pr19007.

(Please note that is an LLVM PR identifier, not a FreeBSD one.)

This should fix Firefox and/or libxul crashes (due to problems with
regparm/stdcall calling conventions) on i386.

Reported by:	multiple users on freebsd-current
PR:		bin/187103

MFC 263048:

Repair recognition of "CC" as an alias for the C++ compiler, since it
was silently broken by upstream for a Windows-specific use-case.

Apparently some versions of CMake still rely on this archaic feature...

Reported by:	rakuco

MFC 263049:

Garbage collect the old way of adding the libstdc++ include directories
in clang's InitHeaderSearch.cpp.  This has been superseded by David
Chisnall's commit in r255321.

Moreover, if libc++ is used, the libstdc++ include directories should
not be in the search path at all.  These directories are now only used
if you pass -stdlib=libstdc++.

1 files changed, 381 insertions, 22 deletions

diff --git a/contrib/llvm/lib/Target/X86/MCTargetDesc/X86AsmBackend.cpp b/contrib/llvm/lib/Target/X86/MCTargetDesc/X86AsmBackend.cpp
index c995aad..f8e359b 100644
--- a/contrib/llvm/lib/Target/X86/MCTargetDesc/X86AsmBackend.cpp
+++ b/contrib/llvm/lib/Target/X86/MCTargetDesc/X86AsmBackend.cpp
@@ -9,6 +9,7 @@
 
 #include "MCTargetDesc/X86BaseInfo.h"
 #include "MCTargetDesc/X86FixupKinds.h"
+#include "llvm/ADT/StringSwitch.h"
 #include "llvm/MC/MCAsmBackend.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCELFObjectWriter.h"
@@ -19,10 +20,10 @@
 #include "llvm/MC/MCSectionCOFF.h"
 #include "llvm/MC/MCSectionELF.h"
 #include "llvm/MC/MCSectionMachO.h"
-#include "llvm/Object/MachOFormat.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/ELF.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/MachO.h"
 #include "llvm/Support/TargetRegistry.h"
 #include "llvm/Support/raw_ostream.h"
 using namespace llvm;
@@ -67,9 +68,16 @@ public:
 
 class X86AsmBackend : public MCAsmBackend {
   StringRef CPU;
+  bool HasNopl;
 public:
   X86AsmBackend(const Target &T, StringRef _CPU)
-    : MCAsmBackend(), CPU(_CPU) {}
+    : MCAsmBackend(), CPU(_CPU) {
+    HasNopl = CPU != "generic" && CPU != "i386" && CPU != "i486" &&
+              CPU != "i586" && CPU != "pentium" && CPU != "pentium-mmx" &&
+              CPU != "i686" && CPU != "k6" && CPU != "k6-2" && CPU != "k6-3" &&
+              CPU != "geode" && CPU != "winchip-c6" && CPU != "winchip2" &&
+              CPU != "c3" && CPU != "c3-2";
+  }
 
   unsigned getNumFixupKinds() const {
     return X86::NumTargetFixupKinds;
@@ -309,11 +317,7 @@ bool X86AsmBackend::writeNopData(uint64_t Count, MCObjectWriter *OW) const {
   // This CPU doesnt support long nops. If needed add more.
   // FIXME: Can we get this from the subtarget somehow?
   // FIXME: We could generated something better than plain 0x90.
-  if (CPU == "generic" || CPU == "i386" || CPU == "i486" || CPU == "i586" ||
-      CPU == "pentium" || CPU == "pentium-mmx" || CPU == "i686" ||
-      CPU == "k6" || CPU == "k6-2" || CPU == "k6-3" || CPU == "geode" ||
-      CPU == "winchip-c6" || CPU == "winchip2" || CPU == "c3" ||
-      CPU == "c3-2") {
+  if (!HasNopl) {
     for (uint64_t i = 0; i < Count; ++i)
       OW->Write8(0x90);
     return true;
@@ -338,6 +342,7 @@ bool X86AsmBackend::writeNopData(uint64_t Count, MCObjectWriter *OW) const {
 /* *** */
 
 namespace {
+
 class ELFX86AsmBackend : public X86AsmBackend {
 public:
   uint8_t OSABI;
@@ -386,35 +391,368 @@ public:
   }
 };
 
+namespace CU {
+
+  /// Compact unwind encoding values.
+  enum CompactUnwindEncodings {
+    /// [RE]BP based frame where [RE]BP is pused on the stack immediately after
+    /// the return address, then [RE]SP is moved to [RE]BP.
+    UNWIND_MODE_BP_FRAME                   = 0x01000000,
+
+    /// A frameless function with a small constant stack size.
+    UNWIND_MODE_STACK_IMMD                 = 0x02000000,
+
+    /// A frameless function with a large constant stack size.
+    UNWIND_MODE_STACK_IND                  = 0x03000000,
+
+    /// No compact unwind encoding is available.
+    UNWIND_MODE_DWARF                      = 0x04000000,
+
+    /// Mask for encoding the frame registers.
+    UNWIND_BP_FRAME_REGISTERS              = 0x00007FFF,
+
+    /// Mask for encoding the frameless registers.
+    UNWIND_FRAMELESS_STACK_REG_PERMUTATION = 0x000003FF
+  };
+
+} // end CU namespace
+
 class DarwinX86AsmBackend : public X86AsmBackend {
+  const MCRegisterInfo &MRI;
+
+  /// \brief Number of registers that can be saved in a compact unwind encoding.
+  enum { CU_NUM_SAVED_REGS = 6 };
+
+  mutable unsigned SavedRegs[CU_NUM_SAVED_REGS];
+  bool Is64Bit;
+
+  unsigned OffsetSize;                   ///< Offset of a "push" instruction.
+  unsigned PushInstrSize;                ///< Size of a "push" instruction.
+  unsigned MoveInstrSize;                ///< Size of a "move" instruction.
+  unsigned StackDivide;                  ///< Amount to adjust stack stize by.
+protected:
+  /// \brief Implementation of algorithm to generate the compact unwind encoding
+  /// for the CFI instructions.
+  uint32_t
+  generateCompactUnwindEncodingImpl(ArrayRef<MCCFIInstruction> Instrs) const {
+    if (Instrs.empty()) return 0;
+
+    // Reset the saved registers.
+    unsigned SavedRegIdx = 0;
+    memset(SavedRegs, 0, sizeof(SavedRegs));
+
+    bool HasFP = false;
+
+    // Encode that we are using EBP/RBP as the frame pointer.
+    uint32_t CompactUnwindEncoding = 0;
+
+    unsigned SubtractInstrIdx = Is64Bit ? 3 : 2;
+    unsigned InstrOffset = 0;
+    unsigned StackAdjust = 0;
+    unsigned StackSize = 0;
+    unsigned PrevStackSize = 0;
+    unsigned NumDefCFAOffsets = 0;
+
+    for (unsigned i = 0, e = Instrs.size(); i != e; ++i) {
+      const MCCFIInstruction &Inst = Instrs[i];
+
+      switch (Inst.getOperation()) {
+      default:
+        // Any other CFI directives indicate a frame that we aren't prepared
+        // to represent via compact unwind, so just bail out.
+        return 0;
+      case MCCFIInstruction::OpDefCfaRegister: {
+        // Defines a frame pointer. E.g.
+        //
+        //     movq %rsp, %rbp
+        //  L0:
+        //     .cfi_def_cfa_register %rbp
+        //
+        HasFP = true;
+        assert(MRI.getLLVMRegNum(Inst.getRegister(), true) ==
+               (Is64Bit ? X86::RBP : X86::EBP) && "Invalid frame pointer!");
+
+        // Reset the counts.
+        memset(SavedRegs, 0, sizeof(SavedRegs));
+        StackAdjust = 0;
+        SavedRegIdx = 0;
+        InstrOffset += MoveInstrSize;
+        break;
+      }
+      case MCCFIInstruction::OpDefCfaOffset: {
+        // Defines a new offset for the CFA. E.g.
+        //
+        //  With frame:
+        //  
+        //     pushq %rbp
+        //  L0:
+        //     .cfi_def_cfa_offset 16
+        //
+        //  Without frame:
+        //
+        //     subq $72, %rsp
+        //  L0:
+        //     .cfi_def_cfa_offset 80
+        //
+        PrevStackSize = StackSize;
+        StackSize = std::abs(Inst.getOffset()) / StackDivide;
+        ++NumDefCFAOffsets;
+        break;
+      }
+      case MCCFIInstruction::OpOffset: {
+        // Defines a "push" of a callee-saved register. E.g.
+        //
+        //     pushq %r15
+        //     pushq %r14
+        //     pushq %rbx
+        //  L0:
+        //     subq $120, %rsp
+        //  L1:
+        //     .cfi_offset %rbx, -40
+        //     .cfi_offset %r14, -32
+        //     .cfi_offset %r15, -24
+        //
+        if (SavedRegIdx == CU_NUM_SAVED_REGS)
+          // If there are too many saved registers, we cannot use a compact
+          // unwind encoding.
+          return CU::UNWIND_MODE_DWARF;
+
+        unsigned Reg = MRI.getLLVMRegNum(Inst.getRegister(), true);
+        SavedRegs[SavedRegIdx++] = Reg;
+        StackAdjust += OffsetSize;
+        InstrOffset += PushInstrSize;
+        break;
+      }
+      }
+    }
+
+    StackAdjust /= StackDivide;
+
+    if (HasFP) {
+      if ((StackAdjust & 0xFF) != StackAdjust)
+        // Offset was too big for a compact unwind encoding.
+        return CU::UNWIND_MODE_DWARF;
+
+      // Get the encoding of the saved registers when we have a frame pointer.
+      uint32_t RegEnc = encodeCompactUnwindRegistersWithFrame();
+      if (RegEnc == ~0U) return CU::UNWIND_MODE_DWARF;
+
+      CompactUnwindEncoding |= CU::UNWIND_MODE_BP_FRAME;
+      CompactUnwindEncoding |= (StackAdjust & 0xFF) << 16;
+      CompactUnwindEncoding |= RegEnc & CU::UNWIND_BP_FRAME_REGISTERS;
+    } else {
+      // If the amount of the stack allocation is the size of a register, then
+      // we "push" the RAX/EAX register onto the stack instead of adjusting the
+      // stack pointer with a SUB instruction. We don't support the push of the
+      // RAX/EAX register with compact unwind. So we check for that situation
+      // here.
+      if ((NumDefCFAOffsets == SavedRegIdx + 1 &&
+           StackSize - PrevStackSize == 1) ||
+          (Instrs.size() == 1 && NumDefCFAOffsets == 1 && StackSize == 2))
+        return CU::UNWIND_MODE_DWARF;
+
+      SubtractInstrIdx += InstrOffset;
+      ++StackAdjust;
+
+      if ((StackSize & 0xFF) == StackSize) {
+        // Frameless stack with a small stack size.
+        CompactUnwindEncoding |= CU::UNWIND_MODE_STACK_IMMD;
+
+        // Encode the stack size.
+        CompactUnwindEncoding |= (StackSize & 0xFF) << 16;
+      } else {
+        if ((StackAdjust & 0x7) != StackAdjust)
+          // The extra stack adjustments are too big for us to handle.
+          return CU::UNWIND_MODE_DWARF;
+
+        // Frameless stack with an offset too large for us to encode compactly.
+        CompactUnwindEncoding |= CU::UNWIND_MODE_STACK_IND;
+
+        // Encode the offset to the nnnnnn value in the 'subl $nnnnnn, ESP'
+        // instruction.
+        CompactUnwindEncoding |= (SubtractInstrIdx & 0xFF) << 16;
+
+        // Encode any extra stack stack adjustments (done via push
+        // instructions).
+        CompactUnwindEncoding |= (StackAdjust & 0x7) << 13;
+      }
+
+      // Encode the number of registers saved. (Reverse the list first.)
+      std::reverse(&SavedRegs[0], &SavedRegs[SavedRegIdx]);
+      CompactUnwindEncoding |= (SavedRegIdx & 0x7) << 10;
+
+      // Get the encoding of the saved registers when we don't have a frame
+      // pointer.
+      uint32_t RegEnc = encodeCompactUnwindRegistersWithoutFrame(SavedRegIdx);
+      if (RegEnc == ~0U) return CU::UNWIND_MODE_DWARF;
+
+      // Encode the register encoding.
+      CompactUnwindEncoding |=
+        RegEnc & CU::UNWIND_FRAMELESS_STACK_REG_PERMUTATION;
+    }
+
+    return CompactUnwindEncoding;
+  }
+
+private:
+  /// \brief Get the compact unwind number for a given register. The number
+  /// corresponds to the enum lists in compact_unwind_encoding.h.
+  int getCompactUnwindRegNum(unsigned Reg) const {
+    static const uint16_t CU32BitRegs[7] = {
+      X86::EBX, X86::ECX, X86::EDX, X86::EDI, X86::ESI, X86::EBP, 0
+    };
+    static const uint16_t CU64BitRegs[] = {
+      X86::RBX, X86::R12, X86::R13, X86::R14, X86::R15, X86::RBP, 0
+    };
+    const uint16_t *CURegs = Is64Bit ? CU64BitRegs : CU32BitRegs;
+    for (int Idx = 1; *CURegs; ++CURegs, ++Idx)
+      if (*CURegs == Reg)
+        return Idx;
+
+    return -1;
+  }
+
+  /// \brief Return the registers encoded for a compact encoding with a frame
+  /// pointer.
+  uint32_t encodeCompactUnwindRegistersWithFrame() const {
+    // Encode the registers in the order they were saved --- 3-bits per
+    // register. The list of saved registers is assumed to be in reverse
+    // order. The registers are numbered from 1 to CU_NUM_SAVED_REGS.
+    uint32_t RegEnc = 0;
+    for (int i = 0, Idx = 0; i != CU_NUM_SAVED_REGS; ++i) {
+      unsigned Reg = SavedRegs[i];
+      if (Reg == 0) break;
+
+      int CURegNum = getCompactUnwindRegNum(Reg);
+      if (CURegNum == -1) return ~0U;
+
+      // Encode the 3-bit register number in order, skipping over 3-bits for
+      // each register.
+      RegEnc |= (CURegNum & 0x7) << (Idx++ * 3);
+    }
+
+    assert((RegEnc & 0x3FFFF) == RegEnc &&
+           "Invalid compact register encoding!");
+    return RegEnc;
+  }
+
+  /// \brief Create the permutation encoding used with frameless stacks. It is
+  /// passed the number of registers to be saved and an array of the registers
+  /// saved.
+  uint32_t encodeCompactUnwindRegistersWithoutFrame(unsigned RegCount) const {
+    // The saved registers are numbered from 1 to 6. In order to encode the
+    // order in which they were saved, we re-number them according to their
+    // place in the register order. The re-numbering is relative to the last
+    // re-numbered register. E.g., if we have registers {6, 2, 4, 5} saved in
+    // that order:
+    //
+    //    Orig  Re-Num
+    //    ----  ------
+    //     6       6
+    //     2       2
+    //     4       3
+    //     5       3
+    //
+    for (unsigned i = 0; i != CU_NUM_SAVED_REGS; ++i) {
+      int CUReg = getCompactUnwindRegNum(SavedRegs[i]);
+      if (CUReg == -1) return ~0U;
+      SavedRegs[i] = CUReg;
+    }
+
+    // Reverse the list.
+    std::reverse(&SavedRegs[0], &SavedRegs[CU_NUM_SAVED_REGS]);
+
+    uint32_t RenumRegs[CU_NUM_SAVED_REGS];
+    for (unsigned i = CU_NUM_SAVED_REGS - RegCount; i < CU_NUM_SAVED_REGS; ++i){
+      unsigned Countless = 0;
+      for (unsigned j = CU_NUM_SAVED_REGS - RegCount; j < i; ++j)
+        if (SavedRegs[j] < SavedRegs[i])
+          ++Countless;
+
+      RenumRegs[i] = SavedRegs[i] - Countless - 1;
+    }
+
+    // Take the renumbered values and encode them into a 10-bit number.
+    uint32_t permutationEncoding = 0;
+    switch (RegCount) {
+    case 6:
+      permutationEncoding |= 120 * RenumRegs[0] + 24 * RenumRegs[1]
+                             + 6 * RenumRegs[2] +  2 * RenumRegs[3]
+                             +     RenumRegs[4];
+      break;
+    case 5:
+      permutationEncoding |= 120 * RenumRegs[1] + 24 * RenumRegs[2]
+                             + 6 * RenumRegs[3] +  2 * RenumRegs[4]
+                             +     RenumRegs[5];
+      break;
+    case 4:
+      permutationEncoding |=  60 * RenumRegs[2] + 12 * RenumRegs[3]
+                             + 3 * RenumRegs[4] +      RenumRegs[5];
+      break;
+    case 3:
+      permutationEncoding |=  20 * RenumRegs[3] +  4 * RenumRegs[4]
+                             +     RenumRegs[5];
+      break;
+    case 2:
+      permutationEncoding |=   5 * RenumRegs[4] +      RenumRegs[5];
+      break;
+    case 1:
+      permutationEncoding |=       RenumRegs[5];
+      break;
+    }
+
+    assert((permutationEncoding & 0x3FF) == permutationEncoding &&
+           "Invalid compact register encoding!");
+    return permutationEncoding;
+  }
+
 public:
-  DarwinX86AsmBackend(const Target &T, StringRef CPU)
-    : X86AsmBackend(T, CPU) { }
+  DarwinX86AsmBackend(const Target &T, const MCRegisterInfo &MRI, StringRef CPU,
+                      bool Is64Bit)
+    : X86AsmBackend(T, CPU), MRI(MRI), Is64Bit(Is64Bit) {
+    memset(SavedRegs, 0, sizeof(SavedRegs));
+    OffsetSize = Is64Bit ? 8 : 4;
+    MoveInstrSize = Is64Bit ? 3 : 2;
+    StackDivide = Is64Bit ? 8 : 4;
+    PushInstrSize = 1;
+  }
 };
 
 class DarwinX86_32AsmBackend : public DarwinX86AsmBackend {
+  bool SupportsCU;
 public:
-  DarwinX86_32AsmBackend(const Target &T, StringRef CPU)
-    : DarwinX86AsmBackend(T, CPU) {}
+  DarwinX86_32AsmBackend(const Target &T, const MCRegisterInfo &MRI,
+                         StringRef CPU, bool SupportsCU)
+    : DarwinX86AsmBackend(T, MRI, CPU, false), SupportsCU(SupportsCU) {}
 
   MCObjectWriter *createObjectWriter(raw_ostream &OS) const {
     return createX86MachObjectWriter(OS, /*Is64Bit=*/false,
-                                     object::mach::CTM_i386,
-                                     object::mach::CSX86_ALL);
+                                     MachO::CPU_TYPE_I386,
+                                     MachO::CPU_SUBTYPE_I386_ALL);
+  }
+
+  /// \brief Generate the compact unwind encoding for the CFI instructions.
+  virtual uint32_t
+  generateCompactUnwindEncoding(ArrayRef<MCCFIInstruction> Instrs) const {
+    return SupportsCU ? generateCompactUnwindEncodingImpl(Instrs) : 0;
   }
 };
 
 class DarwinX86_64AsmBackend : public DarwinX86AsmBackend {
+  bool SupportsCU;
+  const MachO::CPUSubTypeX86 Subtype;
 public:
-  DarwinX86_64AsmBackend(const Target &T, StringRef CPU)
-    : DarwinX86AsmBackend(T, CPU) {
+  DarwinX86_64AsmBackend(const Target &T, const MCRegisterInfo &MRI,
+                         StringRef CPU, bool SupportsCU,
+                         MachO::CPUSubTypeX86 st)
+    : DarwinX86AsmBackend(T, MRI, CPU, true), SupportsCU(SupportsCU),
+      Subtype(st) {
     HasReliableSymbolDifference = true;
   }
 
   MCObjectWriter *createObjectWriter(raw_ostream &OS) const {
     return createX86MachObjectWriter(OS, /*Is64Bit=*/true,
-                                     object::mach::CTM_x86_64,
-                                     object::mach::CSX86_ALL);
+                                     MachO::CPU_TYPE_X86_64, Subtype);
   }
 
   virtual bool doesSectionRequireSymbols(const MCSection &Section) const {
@@ -449,15 +787,26 @@ public:
       return false;
     }
   }
+
+  /// \brief Generate the compact unwind encoding for the CFI instructions.
+  virtual uint32_t
+  generateCompactUnwindEncoding(ArrayRef<MCCFIInstruction> Instrs) const {
+    return SupportsCU ? generateCompactUnwindEncodingImpl(Instrs) : 0;
+  }
 };
 
 } // end anonymous namespace
 
-MCAsmBackend *llvm::createX86_32AsmBackend(const Target &T, StringRef TT, StringRef CPU) {
+MCAsmBackend *llvm::createX86_32AsmBackend(const Target &T,
+                                           const MCRegisterInfo &MRI,
+                                           StringRef TT,
+                                           StringRef CPU) {
   Triple TheTriple(TT);
 
   if (TheTriple.isOSDarwin() || TheTriple.getEnvironment() == Triple::MachO)
-    return new DarwinX86_32AsmBackend(T, CPU);
+    return new DarwinX86_32AsmBackend(T, MRI, CPU,
+                                      TheTriple.isMacOSX() &&
+                                      !TheTriple.isMacOSXVersionLT(10, 7));
 
   if (TheTriple.isOSWindows() && TheTriple.getEnvironment() != Triple::ELF)
     return new WindowsX86AsmBackend(T, false, CPU);
@@ -466,11 +815,21 @@ MCAsmBackend *llvm::createX86_32AsmBackend(const Target &T, StringRef TT, String
   return new ELFX86_32AsmBackend(T, OSABI, CPU);
 }
 
-MCAsmBackend *llvm::createX86_64AsmBackend(const Target &T, StringRef TT, StringRef CPU) {
+MCAsmBackend *llvm::createX86_64AsmBackend(const Target &T,
+                                           const MCRegisterInfo &MRI,
+                                           StringRef TT,
+                                           StringRef CPU) {
   Triple TheTriple(TT);
 
-  if (TheTriple.isOSDarwin() || TheTriple.getEnvironment() == Triple::MachO)
-    return new DarwinX86_64AsmBackend(T, CPU);
+  if (TheTriple.isOSDarwin() || TheTriple.getEnvironment() == Triple::MachO) {
+    MachO::CPUSubTypeX86 CS =
+        StringSwitch<MachO::CPUSubTypeX86>(TheTriple.getArchName())
+            .Case("x86_64h", MachO::CPU_SUBTYPE_X86_64_H)
+            .Default(MachO::CPU_SUBTYPE_X86_64_ALL);
+    return new DarwinX86_64AsmBackend(T, MRI, CPU,
+                                      TheTriple.isMacOSX() &&
+                                      !TheTriple.isMacOSXVersionLT(10, 7), CS);
+  }
 
   if (TheTriple.isOSWindows() && TheTriple.getEnvironment() != Triple::ELF)
     return new WindowsX86AsmBackend(T, true, CPU);