Index: lib/Target/AArch64/AArch64FrameLowering.h
===================================================================
--- lib/Target/AArch64/AArch64FrameLowering.h
+++ lib/Target/AArch64/AArch64FrameLowering.h
@@ -19,10 +19,12 @@
 namespace llvm {
 
 class AArch64FrameLowering : public TargetFrameLowering {
+  mutable bool stackRealigned;
 public:
   explicit AArch64FrameLowering()
       : TargetFrameLowering(StackGrowsDown, 16, 0, 16,
-                            true /*StackRealignable*/) {}
+                            true /*StackRealignable*/), stackRealigned (false)
+  {}
 
   void emitCalleeSavedFrameMoves(MachineBasicBlock &MBB,
                                  MachineBasicBlock::iterator MBBI) const;
Index: lib/Target/AArch64/AArch64FrameLowering.cpp
===================================================================
--- lib/Target/AArch64/AArch64FrameLowering.cpp
+++ lib/Target/AArch64/AArch64FrameLowering.cpp
@@ -285,6 +285,7 @@
 
 void AArch64FrameLowering::emitPrologue(MachineFunction &MF,
                                         MachineBasicBlock &MBB) const {
+  stackRealigned = false;
   MachineBasicBlock::iterator MBBI = MBB.begin();
   const MachineFrameInfo *MFI = MF.getFrameInfo();
   const Function *Fn = MF.getFunction();
@@ -396,6 +397,7 @@
     BuildMI(MBB, MBBI, DL, TII->get(AArch64::ANDXri), AArch64::SP)
       .addReg(scratchSPReg, RegState::Kill)
       .addImm(andMaskEncoded);
+    stackRealigned = true;
   }
 
   // If we need a base pointer, set it up here. It's whatever the value of the
@@ -607,10 +609,13 @@
   // FIXME: Rather than doing the math here, we should instead just use
   // non-post-indexed loads for the restores if we aren't actually going to
   // be able to save any instructions.
-  if (NumBytes || MFI->hasVarSizedObjects())
+  if (MFI->hasVarSizedObjects() || stackRealigned)
     emitFrameOffset(MBB, LastPopI, DL, AArch64::SP, AArch64::FP,
                     -AFI->getCalleeSavedStackSize() + 16, TII,
                     MachineInstr::FrameDestroy);
+  else if (NumBytes)
+    emitFrameOffset(MBB, LastPopI, DL, AArch64::SP, AArch64::SP,
+                    NumBytes, TII, MachineInstr::FrameDestroy);
 
   // This must be placed after the callee-save restore code because that code
   // assumes the SP is at the same location as it was after the callee-save save
Index: test/CodeGen/AArch64/arm64-hello.ll
===================================================================
--- test/CodeGen/AArch64/arm64-hello.ll
+++ test/CodeGen/AArch64/arm64-hello.ll
@@ -9,7 +9,7 @@
 ; CHECK:	adrp	x0, L_.str@PAGE
 ; CHECK:	add	x0, x0, L_.str@PAGEOFF
 ; CHECK-NEXT:	bl	_puts
-; CHECK-NEXT:	mov	sp, x29
+; CHECK-NEXT:	add	sp, sp, #16
 ; CHECK-NEXT:	ldp	x29, x30, [sp], #16
 ; CHECK-NEXT:	ret
 
Index: test/CodeGen/AArch64/arm64-patchpoint.ll
===================================================================
--- test/CodeGen/AArch64/arm64-patchpoint.ll
+++ test/CodeGen/AArch64/arm64-patchpoint.ll
@@ -29,7 +29,7 @@
 ; CHECK:       mov x29, sp
 ; CHECK-NEXT:  sub sp, sp, #32
 ; CHECK:       Ltmp
-; CHECK:       mov sp, x29
+; CHECK:       add sp, sp, #32
 ; CHECK:       ret
 
 define void @caller_meta_leaf() {
Index: test/CodeGen/AArch64/arm64-shrink-wrapping.ll
===================================================================
--- test/CodeGen/AArch64/arm64-shrink-wrapping.ll
+++ test/CodeGen/AArch64/arm64-shrink-wrapping.ll
@@ -33,8 +33,8 @@
 ; Without shrink-wrapping, epilogue is in the exit block.
 ; DISABLE: [[EXIT_LABEL]]:
 ; Epilogue code.
-; CHECK-NEXT: mov sp, [[SAVE_SP]]
-; CHECK-NEXT: ldp [[SAVE_SP]], [[CSR]], [sp], #16
+; CHECK-NEXT: add sp, sp, #16
+; CHECK-NEXT: ldp x{{[0-9]+}}, [[CSR]], [sp], #16
 ;
 ; With shrink-wrapping, exit block is a simple return.
 ; ENABLE: [[EXIT_LABEL]]:
@@ -473,7 +473,7 @@
 ; DISABLE: [[IFEND_LABEL]]: ; %if.end
 ;
 ; Epilogue code.
-; CHECK: mov sp, [[NEW_SP]]
+; CHECK: add sp, sp, #48
 ; CHECK-NEXT: ldp [[CSR1]], [[CSR2]], [sp], #16
 ; CHECK-NEXT: ret
 ;
Index: test/CodeGen/AArch64/fastcc.ll
===================================================================
--- test/CodeGen/AArch64/fastcc.ll
+++ test/CodeGen/AArch64/fastcc.ll
@@ -42,12 +42,12 @@
 ; CHECK-TAIL-NOT: sub sp, sp
 
   ret void
-; CHECK: mov sp, x29
+; CHECK: add sp, sp, #32
 ; CHECK-NEXT: ldp     x29, x30, [sp], #16
 ; CHECK-NEXT: ret
 
 
-; CHECK-TAIL: mov sp, x29
+; CHECK-TAIL: add sp, sp, #32
 ; CHECK-TAIL-NEXT: ldp     x29, x30, [sp], #16
 ; CHECK-TAIL-NEXT: ret
 }
@@ -91,12 +91,12 @@
 ; CHECK-TAIL-NOT: sub sp, sp
 
   ret void
-; CHECK: mov sp, x29
+; CHECK: add sp, sp, #32
 ; CHECK-NEXT: ldp     x29, x30, [sp], #16
 ; CHECK-NEXT: ret
 
 
-; CHECK-TAIL: mov sp, x29
+; CHECK-TAIL: add sp, sp, #32
 ; CHECK-TAIL-NEXT: ldp     x29, x30, [sp], #16
 ; CHECK-TAIL-NEXT: add     sp, sp, #16
 ; CHECK-TAIL-NEXT: ret
@@ -136,11 +136,11 @@
 ; CHECK-TAIL-NOT: sub sp, sp
 
   ret void
-; CHECK: mov sp, x29
+; CHECK: add sp, sp, #32
 ; CHECK-NEXT: ldp     x29, x30, [sp], #16
 ; CHECK-NEXT: ret
 
-; CHECK-TAIL: mov sp, x29
+; CHECK-TAIL: add sp, sp, #32
 ; CHECK-TAIL-NEXT: ldp     x29, x30, [sp], #16
 ; CHECK-TAIL-NEXT: add     sp, sp, #32
 ; CHECK-TAIL-NEXT: ret