diff --git a/lib/Target/X86/X86InstrInfo.cpp b/lib/Target/X86/X86InstrInfo.cpp
index e585ca1434c..17c6375c4a1 100644
--- a/lib/Target/X86/X86InstrInfo.cpp
+++ b/lib/Target/X86/X86InstrInfo.cpp
@@ -868,6 +868,7 @@ X86InstrInfo::isReallyTriviallyReMaterializable(const MachineInstr *MI,
     case X86::MOVSSrm:
     case X86::MOVSDrm:
     case X86::MOVAPSrm:
+    case X86::MOVUPSrm:
     case X86::MOVAPDrm:
     case X86::MOVDQArm:
     case X86::MMX_MOVD64rm:
@@ -1966,8 +1967,7 @@ void X86InstrInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg,
                                   MachineInstr::mmo_iterator MMOBegin,
                                   MachineInstr::mmo_iterator MMOEnd,
                                   SmallVectorImpl<MachineInstr*> &NewMIs) const {
-  bool isAligned = (RI.getStackAlignment() >= 16) ||
-    RI.needsStackRealignment(MF);
+  bool isAligned = (*MMOBegin)->getAlignment() >= 16;
   unsigned Opc = getStoreRegOpcode(SrcReg, RC, isAligned, TM);
   DebugLoc DL = DebugLoc::getUnknownLoc();
   MachineInstrBuilder MIB = BuildMI(MF, DL, get(Opc));
@@ -2060,8 +2060,7 @@ void X86InstrInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg,
                                  MachineInstr::mmo_iterator MMOBegin,
                                  MachineInstr::mmo_iterator MMOEnd,
                                  SmallVectorImpl<MachineInstr*> &NewMIs) const {
-  bool isAligned = (RI.getStackAlignment() >= 16) ||
-    RI.needsStackRealignment(MF);
+  bool isAligned = (*MMOBegin)->getAlignment() >= 16;
   unsigned Opc = getLoadRegOpcode(DestReg, RC, isAligned, TM);
   DebugLoc DL = DebugLoc::getUnknownLoc();
   MachineInstrBuilder MIB = BuildMI(MF, DL, get(Opc), DestReg);
@@ -2638,17 +2637,16 @@ X86InstrInfo::unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N,
   MachineFunction &MF = DAG.getMachineFunction();
   if (FoldedLoad) {
     EVT VT = *RC->vt_begin();
-    bool isAligned = (RI.getStackAlignment() >= 16) ||
-      RI.needsStackRealignment(MF);
+    std::pair<MachineInstr::mmo_iterator,
+              MachineInstr::mmo_iterator> MMOs =
+      MF.extractLoadMemRefs(cast<MachineSDNode>(N)->memoperands_begin(),
+                            cast<MachineSDNode>(N)->memoperands_end());
+    bool isAligned = (*MMOs.first)->getAlignment() >= 16;
     Load = DAG.getMachineNode(getLoadRegOpcode(0, RC, isAligned, TM), dl,
                               VT, MVT::Other, &AddrOps[0], AddrOps.size());
     NewNodes.push_back(Load);
 
     // Preserve memory reference information.
-    std::pair<MachineInstr::mmo_iterator,
-              MachineInstr::mmo_iterator> MMOs =
-      MF.extractLoadMemRefs(cast<MachineSDNode>(N)->memoperands_begin(),
-                            cast<MachineSDNode>(N)->memoperands_end());
     cast<MachineSDNode>(Load)->setMemRefs(MMOs.first, MMOs.second);
   }
 
@@ -2676,8 +2674,11 @@ X86InstrInfo::unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N,
     AddrOps.pop_back();
     AddrOps.push_back(SDValue(NewNode, 0));
     AddrOps.push_back(Chain);
-    bool isAligned = (RI.getStackAlignment() >= 16) ||
-      RI.needsStackRealignment(MF);
+    std::pair<MachineInstr::mmo_iterator,
+              MachineInstr::mmo_iterator> MMOs =
+      MF.extractStoreMemRefs(cast<MachineSDNode>(N)->memoperands_begin(),
+                             cast<MachineSDNode>(N)->memoperands_end());
+    bool isAligned = (*MMOs.first)->getAlignment() >= 16;
     SDNode *Store = DAG.getMachineNode(getStoreRegOpcode(0, DstRC,
                                                          isAligned, TM),
                                        dl, MVT::Other,
@@ -2685,10 +2686,6 @@ X86InstrInfo::unfoldMemoryOperand(SelectionDAG &DAG, SDNode *N,
     NewNodes.push_back(Store);
 
     // Preserve memory reference information.
-    std::pair<MachineInstr::mmo_iterator,
-              MachineInstr::mmo_iterator> MMOs =
-      MF.extractStoreMemRefs(cast<MachineSDNode>(N)->memoperands_begin(),
-                             cast<MachineSDNode>(N)->memoperands_end());
     cast<MachineSDNode>(Load)->setMemRefs(MMOs.first, MMOs.second);
   }
 
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 7d60507476a..83e7088f37a 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -706,7 +706,7 @@ def MOVAPSmr : PSI<0x29, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
 let neverHasSideEffects = 1 in
 def MOVUPSrr : PSI<0x10, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
                    "movups\t{$src, $dst|$dst, $src}", []>;
-let canFoldAsLoad = 1 in
+let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in
 def MOVUPSrm : PSI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
                    "movups\t{$src, $dst|$dst, $src}",
                    [(set VR128:$dst, (loadv4f32 addr:$src))]>;
diff --git a/test/CodeGen/X86/2009-11-16-UnfoldMemOpBug.ll b/test/CodeGen/X86/2009-11-16-UnfoldMemOpBug.ll
new file mode 100644
index 00000000000..3ce9edb5db4
--- /dev/null
+++ b/test/CodeGen/X86/2009-11-16-UnfoldMemOpBug.ll
@@ -0,0 +1,28 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s
+; rdar://7396984
+
+@str = private constant [28 x i8] c"xxxxxxxxxxxxxxxxxxxxxxxxxxx\00", align 1
+
+define void @t(i32 %count) ssp nounwind {
+entry:
+; CHECK: t:
+; CHECK: movq ___stack_chk_guard@GOTPCREL(%rip)
+; CHECK: movups L_str(%rip), %xmm0
+  %tmp0 = alloca [60 x i8], align 1
+  %tmp1 = getelementptr inbounds [60 x i8]* %tmp0, i64 0, i64 0
+  br label %bb1
+
+bb1:
+; CHECK: LBB1_1:
+; CHECK: movaps %xmm0, (%rsp)
+  %tmp2 = phi i32 [ %tmp3, %bb1 ], [ 0, %entry ]
+  call void @llvm.memcpy.i64(i8* %tmp1, i8* getelementptr inbounds ([28 x i8]* @str, i64 0, i64 0), i64 28, i32 1)
+  %tmp3 = add i32 %tmp2, 1
+  %tmp4 = icmp eq i32 %tmp3, %count
+  br i1 %tmp4, label %bb2, label %bb1
+
+bb2:
+  ret void
+}
+
+declare void @llvm.memcpy.i64(i8* nocapture, i8* nocapture, i64, i32) nounwind