Index: include/llvm/CodeGen/MachineInstr.h
===================================================================
--- include/llvm/CodeGen/MachineInstr.h
+++ include/llvm/CodeGen/MachineInstr.h
@@ -614,7 +614,6 @@
   /// are not marking copies from and to the same register class with this flag.
   bool isAsCheapAsAMove(QueryType Type = AllInBundle) const {
     // Only returns true for a bundle if all bundled instructions are cheap.
-    // FIXME: This probably requires a target hook.
     return hasProperty(MCID::CheapAsAMove, Type);
   }
 
Index: include/llvm/MC/MCInstrDesc.h
===================================================================
--- include/llvm/MC/MCInstrDesc.h
+++ include/llvm/MC/MCInstrDesc.h
@@ -451,9 +451,12 @@
   }
 
   /// isRematerializable - Returns true if this instruction is a candidate for
-  /// remat.  This flag is deprecated, please don't use it anymore.  If this
-  /// flag is set, the isReallyTriviallyReMaterializable() method is called to
-  /// verify the instruction is really rematable.
+  /// remat. This flag is only used in TargetInstrInfo method
+  /// isTriviallyRematerializable.
+  ///
+  /// If this flag is set, the isReallyTriviallyReMaterializable()
+  /// or isReallyTriviallyReMaterializableGeneric methods are called to verify
+  /// the instruction is really rematable.
   bool isRematerializable() const {
     return Flags & (1 << MCID::Rematerializable);
   }
Index: include/llvm/Target/TargetInstrInfo.h
===================================================================
--- include/llvm/Target/TargetInstrInfo.h
+++ include/llvm/Target/TargetInstrInfo.h
@@ -200,6 +200,15 @@
                                  unsigned &Size, unsigned &Offset,
                                  const TargetMachine *TM) const;
 
+  /// isAsCheapAsAMove - Return true if the instruction is as cheap as a move
+  /// instruction.
+  ///
+  /// Targets for different archs need to override this, and different
+  /// micro-architectures can also be finely tuned inside.
+  virtual bool isAsCheapAsAMove(const MachineInstr *MI) const {
+    return MI->isAsCheapAsAMove();
+  }
+
   /// reMaterialize - Re-issue the specified 'original' instruction at the
   /// specific location targeting a new destination register.
   /// The register in Orig->getOperand(0).getReg() will be substituted by
Index: lib/CodeGen/RegisterCoalescer.cpp
===================================================================
--- lib/CodeGen/RegisterCoalescer.cpp
+++ lib/CodeGen/RegisterCoalescer.cpp
@@ -751,7 +751,7 @@
     IsDefCopy = true;
     return false;
   }
-  if (!DefMI->isAsCheapAsAMove())
+  if (!TII->isAsCheapAsAMove(DefMI))
     return false;
   if (!TII->isTriviallyReMaterializable(DefMI, AA))
     return false;
Index: lib/Target/AArch64/AArch64InstrFormats.td
===================================================================
--- lib/Target/AArch64/AArch64InstrFormats.td
+++ lib/Target/AArch64/AArch64InstrFormats.td
@@ -1612,7 +1612,7 @@
 
 multiclass AddSub<bit isSub, string mnemonic,
                   SDPatternOperator OpNode = null_frag> {
-  let hasSideEffects = 0 in {
+  let hasSideEffects = 0, isReMaterializable = 1 in {
   // Add/Subtract immediate
   def Wri  : BaseAddSubImm<isSub, 0, GPR32sp, GPR32sp, addsub_shifted_imm32,
                            mnemonic, OpNode> {
@@ -1935,7 +1935,7 @@
     : InstAlias<asm#" $dst, $src1, $src2",
                 (inst regtype:$dst, regtype:$src1, regtype:$src2, 0)>;
 
-let AddedComplexity = 6 in
+let AddedComplexity = 6, isReMaterializable = 1 in
 multiclass LogicalImm<bits<2> opc, string mnemonic, SDNode OpNode> {
   def Wri : BaseLogicalImm<opc, GPR32sp, GPR32, logical_imm32, mnemonic,
                            [(set GPR32sp:$Rd, (OpNode GPR32:$Rn,
@@ -1972,8 +1972,10 @@
 // Split from LogicalImm as not all instructions have both.
 multiclass LogicalReg<bits<2> opc, bit N, string mnemonic,
                       SDPatternOperator OpNode> {
+  let isReMaterializable = 1 in {
   def Wrr : BaseLogicalRegPseudo<GPR32, OpNode>;
   def Xrr : BaseLogicalRegPseudo<GPR64, OpNode>;
+  }
 
   def Wrs : BaseLogicalSReg<opc, N, GPR32, logical_shifted_reg32, mnemonic,
                             [(set GPR32:$Rd, (OpNode GPR32:$Rn,
Index: lib/Target/AArch64/AArch64InstrInfo.h
===================================================================
--- lib/Target/AArch64/AArch64InstrInfo.h
+++ lib/Target/AArch64/AArch64InstrInfo.h
@@ -46,6 +46,8 @@
 
   unsigned GetInstSizeInBytes(const MachineInstr *MI) const;
 
+  bool isAsCheapAsAMove(const MachineInstr *MI) const override;
+
   bool isCoalescableExtInstr(const MachineInstr &MI, unsigned &SrcReg,
                              unsigned &DstReg, unsigned &SubIdx) const override;
 
Index: lib/Target/AArch64/AArch64InstrInfo.cpp
===================================================================
--- lib/Target/AArch64/AArch64InstrInfo.cpp
+++ lib/Target/AArch64/AArch64InstrInfo.cpp
@@ -541,6 +541,55 @@
       CC);
 }
 
+// FIXME: this implementation should be micro-architecture dependent, so a
+// micro-architecture target hook should be introduced here in future.
+bool AArch64InstrInfo::isAsCheapAsAMove(const MachineInstr *MI) const {
+  if (Subtarget.isCortexA57() || Subtarget.isCortexA53()) {
+    switch (MI->getOpcode()) {
+    default:
+      return false;
+
+    // add/sub on register without shift
+    case AArch64::ADDWri:
+    case AArch64::ADDXri:
+    case AArch64::SUBWri:
+    case AArch64::SUBXri:
+      if (MI->getOperand(3).getImm() != 0)
+        return false;
+      else
+        return true;
+
+    // logical ops on immediate
+    case AArch64::ANDSWri:
+    case AArch64::ANDSXri:
+    case AArch64::ANDWri:
+    case AArch64::ANDXri:
+    case AArch64::EORWri:
+    case AArch64::EORXri:
+    case AArch64::ORRWri:
+    case AArch64::ORRXri:
+      return true;
+
+    // logical ops on register without shift
+    case AArch64::ANDWrr:
+    case AArch64::ANDXrr:
+    case AArch64::BICWrr:
+    case AArch64::BICXrr:
+    case AArch64::EONWrr:
+    case AArch64::EONXrr:
+    case AArch64::EORWrr:
+    case AArch64::EORXrr:
+    case AArch64::ORNWrr:
+    case AArch64::ORNXrr:
+    case AArch64::ORRWrr:
+    case AArch64::ORRXrr:
+      return true;
+    }
+  }
+
+  return MI->isAsCheapAsAMove();
+}
+
 bool AArch64InstrInfo::isCoalescableExtInstr(const MachineInstr &MI,
                                              unsigned &SrcReg, unsigned &DstReg,
                                              unsigned &SubIdx) const {
Index: lib/Target/AArch64/AArch64Subtarget.h
===================================================================
--- lib/Target/AArch64/AArch64Subtarget.h
+++ lib/Target/AArch64/AArch64Subtarget.h
@@ -100,6 +100,8 @@
   bool isTargetMachO() const { return TargetTriple.isOSBinFormatMachO(); }
 
   bool isCyclone() const { return CPUString == "cyclone"; }
+  bool isCortexA57() const { return CPUString == "cortex-a57"; }
+  bool isCortexA53() const { return CPUString == "cortex-a53"; }
 
   /// getMaxInlineSizeThreshold - Returns the maximum memset / memcpy size
   /// that still makes it profitable to inline the call.
Index: test/CodeGen/AArch64/remat.ll
===================================================================
--- /dev/null
+++ test/CodeGen/AArch64/remat.ll
@@ -0,0 +1,16 @@
+; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=cortex-a57 -o - %s | FileCheck %s
+; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=cortex-a53 -o - %s | FileCheck %s
+
+%X = type { i64, i64, i64 }
+declare void @f(%X*)
+define void @t() {
+entry:
+  %tmp = alloca %X
+  call void @f(%X* %tmp)
+; CHECK: add x0, sp, #8
+; CHECK-NEXT-NOT: mov
+  call void @f(%X* %tmp)               
+; CHECK: add x0, sp, #8
+; CHECK-NEXT-NOT: mov
+  ret void 
+}