diff --git a/llvm/lib/CodeGen/ExpandMemCmp.cpp b/llvm/lib/CodeGen/ExpandMemCmp.cpp
--- a/llvm/lib/CodeGen/ExpandMemCmp.cpp
+++ b/llvm/lib/CodeGen/ExpandMemCmp.cpp
@@ -103,8 +103,12 @@
   Value *getMemCmpExpansionZeroCase();
   Value *getMemCmpEqZeroOneBlock();
   Value *getMemCmpOneBlock();
-  Value *getPtrToElementAtOffset(Value *Source, Type *LoadSizeType,
-                                 uint64_t OffsetBytes);
+  struct LoadPair {
+    Value *Lhs = nullptr;
+    Value *Rhs = nullptr;
+  };
+  LoadPair getLoadPair(Type *LoadSizeType, bool NeedsBSwap, Type *CmpSizeType,
+                       unsigned OffsetBytes);
 
   static LoadEntryVector
   computeGreedyLoadSequence(uint64_t Size, llvm::ArrayRef<unsigned> LoadSizes,
@@ -261,18 +265,42 @@
                                    EndBlock->getParent(), EndBlock);
 }
 
-/// Return a pointer to an element of type `LoadSizeType` at offset
-/// `OffsetBytes`.
-Value *MemCmpExpansion::getPtrToElementAtOffset(Value *Source,
-                                                Type *LoadSizeType,
-                                                uint64_t OffsetBytes) {
-  if (OffsetBytes > 0) {
-    auto *ByteType = Type::getInt8Ty(CI->getContext());
-    Source = Builder.CreateConstGEP1_64(
-        ByteType, Builder.CreateBitCast(Source, ByteType->getPointerTo()),
-        OffsetBytes);
-  }
-  return Builder.CreateBitCast(Source, LoadSizeType->getPointerTo());
+MemCmpExpansion::LoadPair MemCmpExpansion::getLoadPair(Type *LoadSizeType,
+                                                       bool NeedsBSwap,
+                                                       Type *CmpSizeType,
+                                                       unsigned OffsetBytes) {
+  const auto MakeValue = [this, LoadSizeType, NeedsBSwap, CmpSizeType,
+                          OffsetBytes](int Index) {
+    // Get the memory source at offset `OffsetBytes`.
+    Value *Source = CI->getArgOperand(Index);
+    if (OffsetBytes > 0) {
+      auto *ByteType = Type::getInt8Ty(CI->getContext());
+      Source = Builder.CreateConstGEP1_64(
+          ByteType, Builder.CreateBitCast(Source, ByteType->getPointerTo()),
+          OffsetBytes);
+    }
+    Source = Builder.CreateBitCast(Source, LoadSizeType->getPointerTo());
+
+    // Create a constant or a load from the source.
+    Value *V = nullptr;
+    if (auto *C = dyn_cast<Constant>(Source))
+      V = ConstantFoldLoadFromConstPtr(C, LoadSizeType, DL);
+    if (!V)
+      V = Builder.CreateLoad(LoadSizeType, Source);
+
+    // Swap bytes if required.
+    if (NeedsBSwap) {
+      Function *Bswap = Intrinsic::getDeclaration(
+          CI->getModule(), Intrinsic::bswap, LoadSizeType);
+      V = Builder.CreateCall(Bswap, V);
+    }
+
+    // Zero extend if required.
+    if (CmpSizeType != nullptr && CmpSizeType != LoadSizeType)
+      V = Builder.CreateZExt(V, CmpSizeType);
+    return V;
+  };
+  return {MakeValue(0), MakeValue(1)};
 }
 
 // This function creates the IR instructions for loading and comparing 1 byte.
@@ -282,18 +310,10 @@
 void MemCmpExpansion::emitLoadCompareByteBlock(unsigned BlockIndex,
                                                unsigned OffsetBytes) {
   Builder.SetInsertPoint(LoadCmpBlocks[BlockIndex]);
-  Type *LoadSizeType = Type::getInt8Ty(CI->getContext());
-  Value *Source1 =
-      getPtrToElementAtOffset(CI->getArgOperand(0), LoadSizeType, OffsetBytes);
-  Value *Source2 =
-      getPtrToElementAtOffset(CI->getArgOperand(1), LoadSizeType, OffsetBytes);
-
-  Value *LoadSrc1 = Builder.CreateLoad(LoadSizeType, Source1);
-  Value *LoadSrc2 = Builder.CreateLoad(LoadSizeType, Source2);
-
-  LoadSrc1 = Builder.CreateZExt(LoadSrc1, Type::getInt32Ty(CI->getContext()));
-  LoadSrc2 = Builder.CreateZExt(LoadSrc2, Type::getInt32Ty(CI->getContext()));
-  Value *Diff = Builder.CreateSub(LoadSrc1, LoadSrc2);
+  const LoadPair Loads =
+      getLoadPair(Type::getInt8Ty(CI->getContext()), /*NeedsBSwap=*/false,
+                  Type::getInt32Ty(CI->getContext()), OffsetBytes);
+  Value *Diff = Builder.CreateSub(Loads.Lhs, Loads.Rhs);
 
   PhiRes->addIncoming(Diff, LoadCmpBlocks[BlockIndex]);
 
@@ -340,41 +360,19 @@
                     : IntegerType::get(CI->getContext(), MaxLoadSize * 8);
   for (unsigned i = 0; i < NumLoads; ++i, ++LoadIndex) {
     const LoadEntry &CurLoadEntry = LoadSequence[LoadIndex];
-
-    IntegerType *LoadSizeType =
-        IntegerType::get(CI->getContext(), CurLoadEntry.LoadSize * 8);
-
-    Value *Source1 = getPtrToElementAtOffset(CI->getArgOperand(0), LoadSizeType,
-                                             CurLoadEntry.Offset);
-    Value *Source2 = getPtrToElementAtOffset(CI->getArgOperand(1), LoadSizeType,
-                                             CurLoadEntry.Offset);
-
-    // Get a constant or load a value for each source address.
-    Value *LoadSrc1 = nullptr;
-    if (auto *Source1C = dyn_cast<Constant>(Source1))
-      LoadSrc1 = ConstantFoldLoadFromConstPtr(Source1C, LoadSizeType, DL);
-    if (!LoadSrc1)
-      LoadSrc1 = Builder.CreateLoad(LoadSizeType, Source1);
-
-    Value *LoadSrc2 = nullptr;
-    if (auto *Source2C = dyn_cast<Constant>(Source2))
-      LoadSrc2 = ConstantFoldLoadFromConstPtr(Source2C, LoadSizeType, DL);
-    if (!LoadSrc2)
-      LoadSrc2 = Builder.CreateLoad(LoadSizeType, Source2);
+    const LoadPair Loads = getLoadPair(
+        IntegerType::get(CI->getContext(), CurLoadEntry.LoadSize * 8),
+        /*NeedsBSwap=*/false, MaxLoadType, CurLoadEntry.Offset);
 
     if (NumLoads != 1) {
-      if (LoadSizeType != MaxLoadType) {
-        LoadSrc1 = Builder.CreateZExt(LoadSrc1, MaxLoadType);
-        LoadSrc2 = Builder.CreateZExt(LoadSrc2, MaxLoadType);
-      }
       // If we have multiple loads per block, we need to generate a composite
       // comparison using xor+or.
-      Diff = Builder.CreateXor(LoadSrc1, LoadSrc2);
+      Diff = Builder.CreateXor(Loads.Lhs, Loads.Rhs);
       Diff = Builder.CreateZExt(Diff, MaxLoadType);
       XorList.push_back(Diff);
     } else {
       // If there's only one load per block, we just compare the loaded values.
-      Cmp = Builder.CreateICmpNE(LoadSrc1, LoadSrc2);
+      Cmp = Builder.CreateICmpNE(Loads.Lhs, Loads.Rhs);
     }
   }
 
@@ -451,35 +449,18 @@
 
   Builder.SetInsertPoint(LoadCmpBlocks[BlockIndex]);
 
-  Value *Source1 = getPtrToElementAtOffset(CI->getArgOperand(0), LoadSizeType,
-                                           CurLoadEntry.Offset);
-  Value *Source2 = getPtrToElementAtOffset(CI->getArgOperand(1), LoadSizeType,
-                                           CurLoadEntry.Offset);
-
-  // Load LoadSizeType from the base address.
-  Value *LoadSrc1 = Builder.CreateLoad(LoadSizeType, Source1);
-  Value *LoadSrc2 = Builder.CreateLoad(LoadSizeType, Source2);
-
-  if (DL.isLittleEndian()) {
-    Function *Bswap = Intrinsic::getDeclaration(CI->getModule(),
-                                                Intrinsic::bswap, LoadSizeType);
-    LoadSrc1 = Builder.CreateCall(Bswap, LoadSrc1);
-    LoadSrc2 = Builder.CreateCall(Bswap, LoadSrc2);
-  }
-
-  if (LoadSizeType != MaxLoadType) {
-    LoadSrc1 = Builder.CreateZExt(LoadSrc1, MaxLoadType);
-    LoadSrc2 = Builder.CreateZExt(LoadSrc2, MaxLoadType);
-  }
+  const LoadPair Loads =
+      getLoadPair(LoadSizeType, /*NeedsBSwap=*/DL.isLittleEndian(), MaxLoadType,
+                  CurLoadEntry.Offset);
 
   // Add the loaded values to the phi nodes for calculating memcmp result only
   // if result is not used in a zero equality.
   if (!IsUsedForZeroCmp) {
-    ResBlock.PhiSrc1->addIncoming(LoadSrc1, LoadCmpBlocks[BlockIndex]);
-    ResBlock.PhiSrc2->addIncoming(LoadSrc2, LoadCmpBlocks[BlockIndex]);
+    ResBlock.PhiSrc1->addIncoming(Loads.Lhs, LoadCmpBlocks[BlockIndex]);
+    ResBlock.PhiSrc2->addIncoming(Loads.Rhs, LoadCmpBlocks[BlockIndex]);
   }
 
-  Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, LoadSrc1, LoadSrc2);
+  Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Loads.Lhs, Loads.Rhs);
   BasicBlock *NextBB = (BlockIndex == (LoadCmpBlocks.size() - 1))
                            ? EndBlock
                            : LoadCmpBlocks[BlockIndex + 1];
@@ -568,42 +549,27 @@
 /// the compare, branch, and phi IR that is required in the general case.
 Value *MemCmpExpansion::getMemCmpOneBlock() {
   Type *LoadSizeType = IntegerType::get(CI->getContext(), Size * 8);
-  Value *Source1 = CI->getArgOperand(0);
-  Value *Source2 = CI->getArgOperand(1);
-
-  // Cast source to LoadSizeType*.
-  if (Source1->getType() != LoadSizeType)
-    Source1 = Builder.CreateBitCast(Source1, LoadSizeType->getPointerTo());
-  if (Source2->getType() != LoadSizeType)
-    Source2 = Builder.CreateBitCast(Source2, LoadSizeType->getPointerTo());
-
-  // Load LoadSizeType from the base address.
-  Value *LoadSrc1 = Builder.CreateLoad(LoadSizeType, Source1);
-  Value *LoadSrc2 = Builder.CreateLoad(LoadSizeType, Source2);
-
-  if (DL.isLittleEndian() && Size != 1) {
-    Function *Bswap = Intrinsic::getDeclaration(CI->getModule(),
-                                                Intrinsic::bswap, LoadSizeType);
-    LoadSrc1 = Builder.CreateCall(Bswap, LoadSrc1);
-    LoadSrc2 = Builder.CreateCall(Bswap, LoadSrc2);
-  }
+  bool NeedsBSwap = DL.isLittleEndian() && Size != 1;
 
+  // The i8 and i16 cases don't need compares. We zext the loaded values and
+  // subtract them to get the suitable negative, zero, or positive i32 result.
   if (Size < 4) {
-    // The i8 and i16 cases don't need compares. We zext the loaded values and
-    // subtract them to get the suitable negative, zero, or positive i32 result.
-    LoadSrc1 = Builder.CreateZExt(LoadSrc1, Builder.getInt32Ty());
-    LoadSrc2 = Builder.CreateZExt(LoadSrc2, Builder.getInt32Ty());
-    return Builder.CreateSub(LoadSrc1, LoadSrc2);
+    const LoadPair Loads =
+        getLoadPair(LoadSizeType, NeedsBSwap, Builder.getInt32Ty(),
+                    /*Offset*/ 0);
+    return Builder.CreateSub(Loads.Lhs, Loads.Rhs);
   }
 
+  const LoadPair Loads = getLoadPair(LoadSizeType, NeedsBSwap, LoadSizeType,
+                                     /*Offset*/ 0);
   // The result of memcmp is negative, zero, or positive, so produce that by
   // subtracting 2 extended compare bits: sub (ugt, ult).
   // If a target prefers to use selects to get -1/0/1, they should be able
   // to transform this later. The inverse transform (going from selects to math)
   // may not be possible in the DAG because the selects got converted into
   // branches before we got there.
-  Value *CmpUGT = Builder.CreateICmpUGT(LoadSrc1, LoadSrc2);
-  Value *CmpULT = Builder.CreateICmpULT(LoadSrc1, LoadSrc2);
+  Value *CmpUGT = Builder.CreateICmpUGT(Loads.Lhs, Loads.Rhs);
+  Value *CmpULT = Builder.CreateICmpULT(Loads.Lhs, Loads.Rhs);
   Value *ZextUGT = Builder.CreateZExt(CmpUGT, Builder.getInt32Ty());
   Value *ZextULT = Builder.CreateZExt(CmpULT, Builder.getInt32Ty());
   return Builder.CreateSub(ZextUGT, ZextULT);
diff --git a/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll b/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll
--- a/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll
+++ b/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll
@@ -235,8 +235,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movzwl (%eax), %edx
-; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %dx
+; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %si
 ; X86-NEXT:    cmpw %si, %dx
 ; X86-NEXT:    jne .LBB9_1
@@ -256,8 +256,8 @@
 ; X64-LABEL: length3:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
+; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    cmpw %cx, %ax
 ; X64-NEXT:    jne .LBB9_1
@@ -450,8 +450,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl (%eax), %edx
-; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    cmpl %esi, %edx
 ; X86-NEXT:    jne .LBB16_1
@@ -471,8 +471,8 @@
 ; X64-LABEL: length5:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    cmpl %ecx, %eax
 ; X64-NEXT:    jne .LBB16_1
@@ -526,8 +526,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl (%eax), %edx
-; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    cmpl %esi, %edx
 ; X86-NEXT:    jne .LBB18_1
@@ -549,8 +549,8 @@
 ; X64-LABEL: length5_lt:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    cmpl %ecx, %eax
 ; X64-NEXT:    jne .LBB18_1
@@ -607,15 +607,15 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    cmpl %edx, %ecx
 ; X86-NEXT:    jne .LBB20_2
 ; X86-NEXT:  # %bb.1: # %loadbb1
 ; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
@@ -830,22 +830,22 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    cmpl %edx, %ecx
 ; X86-NEXT:    jne .LBB27_3
 ; X86-NEXT:  # %bb.1: # %loadbb1
 ; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    cmpl %edx, %ecx
 ; X86-NEXT:    jne .LBB27_3
 ; X86-NEXT:  # %bb.2: # %loadbb2
 ; X86-NEXT:    movl 8(%esi), %ecx
-; X86-NEXT:    movl 8(%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl 8(%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
@@ -862,15 +862,15 @@
 ; X64-LABEL: length12:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB27_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
+; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1000,38 +1000,38 @@
 ; X86:       # %bb.0:
 ; X86-NEXT:    pushl %esi
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NEXT:    movl (%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    movl (%eax), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    jne .LBB31_4
 ; X86-NEXT:  # %bb.1: # %loadbb1
-; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
+; X86-NEXT:    movl 4(%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    movl 4(%eax), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    jne .LBB31_4
 ; X86-NEXT:  # %bb.2: # %loadbb2
-; X86-NEXT:    movl 8(%esi), %ecx
-; X86-NEXT:    movl 8(%eax), %edx
+; X86-NEXT:    movl 8(%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    movl 8(%eax), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    jne .LBB31_4
 ; X86-NEXT:  # %bb.3: # %loadbb3
-; X86-NEXT:    movl 12(%esi), %ecx
-; X86-NEXT:    movl 12(%eax), %edx
+; X86-NEXT:    movl 12(%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl 12(%eax), %esi
+; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    je .LBB31_5
 ; X86-NEXT:  .LBB31_4: # %res_block
 ; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    setae %al
 ; X86-NEXT:    leal -1(%eax,%eax), %eax
 ; X86-NEXT:  .LBB31_5: # %endblock
@@ -1041,15 +1041,15 @@
 ; X64-LABEL: length16:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB31_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1175,38 +1175,38 @@
 ; X86:       # %bb.0:
 ; X86-NEXT:    pushl %esi
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; X86-NEXT:    movl (%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    movl (%eax), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    jne .LBB33_4
 ; X86-NEXT:  # %bb.1: # %loadbb1
-; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
+; X86-NEXT:    movl 4(%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    movl 4(%eax), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    jne .LBB33_4
 ; X86-NEXT:  # %bb.2: # %loadbb2
-; X86-NEXT:    movl 8(%esi), %ecx
-; X86-NEXT:    movl 8(%eax), %edx
+; X86-NEXT:    movl 8(%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    movl 8(%eax), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    jne .LBB33_4
 ; X86-NEXT:  # %bb.3: # %loadbb3
-; X86-NEXT:    movl 12(%esi), %ecx
-; X86-NEXT:    movl 12(%eax), %edx
+; X86-NEXT:    movl 12(%edx), %ecx
 ; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl 12(%eax), %esi
+; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    je .LBB33_5
 ; X86-NEXT:  .LBB33_4: # %res_block
 ; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    cmpl %edx, %ecx
+; X86-NEXT:    cmpl %esi, %ecx
 ; X86-NEXT:    setae %al
 ; X86-NEXT:    leal -1(%eax,%eax), %eax
 ; X86-NEXT:  .LBB33_5: # %endblock
@@ -1218,15 +1218,15 @@
 ; X64-LABEL: length16_lt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB33_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1249,43 +1249,43 @@
 ; X86-LABEL: length16_gt:
 ; X86:       # %bb.0:
 ; X86-NEXT:    pushl %esi
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %edx
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-NEXT:    movl (%esi), %eax
-; X86-NEXT:    movl (%edx), %ecx
+; X86-NEXT:    movl (%edx), %eax
 ; X86-NEXT:    bswapl %eax
-; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    cmpl %ecx, %eax
+; X86-NEXT:    movl (%ecx), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %eax
 ; X86-NEXT:    jne .LBB34_4
 ; X86-NEXT:  # %bb.1: # %loadbb1
-; X86-NEXT:    movl 4(%esi), %eax
-; X86-NEXT:    movl 4(%edx), %ecx
+; X86-NEXT:    movl 4(%edx), %eax
 ; X86-NEXT:    bswapl %eax
-; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    cmpl %ecx, %eax
+; X86-NEXT:    movl 4(%ecx), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %eax
 ; X86-NEXT:    jne .LBB34_4
 ; X86-NEXT:  # %bb.2: # %loadbb2
-; X86-NEXT:    movl 8(%esi), %eax
-; X86-NEXT:    movl 8(%edx), %ecx
+; X86-NEXT:    movl 8(%edx), %eax
 ; X86-NEXT:    bswapl %eax
-; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    cmpl %ecx, %eax
+; X86-NEXT:    movl 8(%ecx), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    cmpl %esi, %eax
 ; X86-NEXT:    jne .LBB34_4
 ; X86-NEXT:  # %bb.3: # %loadbb3
-; X86-NEXT:    movl 12(%esi), %eax
-; X86-NEXT:    movl 12(%edx), %ecx
+; X86-NEXT:    movl 12(%edx), %eax
 ; X86-NEXT:    bswapl %eax
-; X86-NEXT:    bswapl %ecx
-; X86-NEXT:    xorl %edx, %edx
-; X86-NEXT:    cmpl %ecx, %eax
+; X86-NEXT:    movl 12(%ecx), %esi
+; X86-NEXT:    bswapl %esi
+; X86-NEXT:    xorl %ecx, %ecx
+; X86-NEXT:    cmpl %esi, %eax
 ; X86-NEXT:    je .LBB34_5
 ; X86-NEXT:  .LBB34_4: # %res_block
-; X86-NEXT:    xorl %edx, %edx
-; X86-NEXT:    cmpl %ecx, %eax
-; X86-NEXT:    setae %dl
-; X86-NEXT:    leal -1(%edx,%edx), %edx
+; X86-NEXT:    xorl %ecx, %ecx
+; X86-NEXT:    cmpl %esi, %eax
+; X86-NEXT:    setae %cl
+; X86-NEXT:    leal -1(%ecx,%ecx), %ecx
 ; X86-NEXT:  .LBB34_5: # %endblock
-; X86-NEXT:    testl %edx, %edx
+; X86-NEXT:    testl %ecx, %ecx
 ; X86-NEXT:    setg %al
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    retl
@@ -1293,15 +1293,15 @@
 ; X64-LABEL: length16_gt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB34_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rax
-; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    xorl %edx, %edx
 ; X64-NEXT:    cmpq %rcx, %rax
@@ -1433,22 +1433,22 @@
 ; X64-LABEL: length24:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB36_3
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB36_3
 ; X64-NEXT:  # %bb.2: # %loadbb2
 ; X64-NEXT:    movq 16(%rdi), %rcx
-; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1592,22 +1592,22 @@
 ; X64-LABEL: length24_lt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB38_3
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB38_3
 ; X64-NEXT:  # %bb.2: # %loadbb2
 ; X64-NEXT:    movq 16(%rdi), %rcx
-; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1642,22 +1642,22 @@
 ; X64-LABEL: length24_gt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB39_3
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rax
-; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB39_3
 ; X64-NEXT:  # %bb.2: # %loadbb2
 ; X64-NEXT:    movq 16(%rdi), %rax
-; X64-NEXT:    movq 16(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 16(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    xorl %edx, %edx
 ; X64-NEXT:    cmpq %rcx, %rax
@@ -2187,29 +2187,29 @@
 ; X64-LABEL: length32:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB47_4
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB47_4
 ; X64-NEXT:  # %bb.2: # %loadbb2
 ; X64-NEXT:    movq 16(%rdi), %rcx
-; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB47_4
 ; X64-NEXT:  # %bb.3: # %loadbb3
 ; X64-NEXT:    movq 24(%rdi), %rcx
-; X64-NEXT:    movq 24(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 24(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -2367,29 +2367,29 @@
 ; X64-LABEL: length32_lt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB49_4
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB49_4
 ; X64-NEXT:  # %bb.2: # %loadbb2
 ; X64-NEXT:    movq 16(%rdi), %rcx
-; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 16(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB49_4
 ; X64-NEXT:  # %bb.3: # %loadbb3
 ; X64-NEXT:    movq 24(%rdi), %rcx
-; X64-NEXT:    movq 24(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 24(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -2424,29 +2424,29 @@
 ; X64-LABEL: length32_gt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB50_4
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rax
-; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB50_4
 ; X64-NEXT:  # %bb.2: # %loadbb2
 ; X64-NEXT:    movq 16(%rdi), %rax
-; X64-NEXT:    movq 16(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 16(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB50_4
 ; X64-NEXT:  # %bb.3: # %loadbb3
 ; X64-NEXT:    movq 24(%rdi), %rax
-; X64-NEXT:    movq 24(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 24(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    xorl %edx, %edx
 ; X64-NEXT:    cmpq %rcx, %rax
diff --git a/llvm/test/CodeGen/X86/memcmp-optsize.ll b/llvm/test/CodeGen/X86/memcmp-optsize.ll
--- a/llvm/test/CodeGen/X86/memcmp-optsize.ll
+++ b/llvm/test/CodeGen/X86/memcmp-optsize.ll
@@ -116,8 +116,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movzwl (%eax), %edx
-; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %dx
+; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %si
 ; X86-NEXT:    cmpw %si, %dx
 ; X86-NEXT:    jne .LBB4_1
@@ -137,8 +137,8 @@
 ; X64-LABEL: length3:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
+; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    cmpw %cx, %ax
 ; X64-NEXT:    jne .LBB4_1
@@ -261,8 +261,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl (%eax), %edx
-; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    cmpl %esi, %edx
 ; X86-NEXT:    jne .LBB9_1
@@ -282,8 +282,8 @@
 ; X64-LABEL: length5:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    cmpl %ecx, %eax
 ; X64-NEXT:    jne .LBB9_1
@@ -337,15 +337,15 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    cmpl %edx, %ecx
 ; X86-NEXT:    jne .LBB11_2
 ; X86-NEXT:  # %bb.1: # %loadbb1
 ; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
@@ -462,15 +462,15 @@
 ; X64-LABEL: length12:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB15_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
+; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -502,15 +502,15 @@
 ; X64-LABEL: length16:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB16_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
diff --git a/llvm/test/CodeGen/X86/memcmp-pgso.ll b/llvm/test/CodeGen/X86/memcmp-pgso.ll
--- a/llvm/test/CodeGen/X86/memcmp-pgso.ll
+++ b/llvm/test/CodeGen/X86/memcmp-pgso.ll
@@ -116,38 +116,38 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movzwl (%eax), %edx
-; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %dx
+; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %si
 ; X86-NEXT:    cmpw %si, %dx
-; X86-NEXT:    jne .LBB4_1
+; X86-NEXT:    jne .LBB5_1
 ; X86-NEXT:  # %bb.2: # %loadbb1
 ; X86-NEXT:    movzbl 2(%eax), %eax
 ; X86-NEXT:    movzbl 2(%ecx), %ecx
 ; X86-NEXT:    subl %ecx, %eax
-; X86-NEXT:    jmp .LBB4_3
-; X86-NEXT:  .LBB4_1: # %res_block
+; X86-NEXT:    jmp .LBB5_3
+; X86-NEXT:  .LBB5_1: # %res_block
 ; X86-NEXT:    setae %al
 ; X86-NEXT:    movzbl %al, %eax
 ; X86-NEXT:    leal -1(%eax,%eax), %eax
-; X86-NEXT:  .LBB4_3: # %endblock
+; X86-NEXT:  .LBB5_3: # %endblock
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: length3:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
+; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    cmpw %cx, %ax
-; X64-NEXT:    jne .LBB4_1
+; X64-NEXT:    jne .LBB5_1
 ; X64-NEXT:  # %bb.2: # %loadbb1
 ; X64-NEXT:    movzbl 2(%rdi), %eax
 ; X64-NEXT:    movzbl 2(%rsi), %ecx
 ; X64-NEXT:    subl %ecx, %eax
 ; X64-NEXT:    retq
-; X64-NEXT:  .LBB4_1: # %res_block
+; X64-NEXT:  .LBB5_1: # %res_block
 ; X64-NEXT:    setae %al
 ; X64-NEXT:    movzbl %al, %eax
 ; X64-NEXT:    leal -1(%rax,%rax), %eax
@@ -261,38 +261,38 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl (%eax), %edx
-; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    cmpl %esi, %edx
-; X86-NEXT:    jne .LBB9_1
+; X86-NEXT:    jne .LBB10_1
 ; X86-NEXT:  # %bb.2: # %loadbb1
 ; X86-NEXT:    movzbl 4(%eax), %eax
 ; X86-NEXT:    movzbl 4(%ecx), %ecx
 ; X86-NEXT:    subl %ecx, %eax
-; X86-NEXT:    jmp .LBB9_3
-; X86-NEXT:  .LBB9_1: # %res_block
+; X86-NEXT:    jmp .LBB10_3
+; X86-NEXT:  .LBB10_1: # %res_block
 ; X86-NEXT:    setae %al
 ; X86-NEXT:    movzbl %al, %eax
 ; X86-NEXT:    leal -1(%eax,%eax), %eax
-; X86-NEXT:  .LBB9_3: # %endblock
+; X86-NEXT:  .LBB10_3: # %endblock
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: length5:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    cmpl %ecx, %eax
-; X64-NEXT:    jne .LBB9_1
+; X64-NEXT:    jne .LBB10_1
 ; X64-NEXT:  # %bb.2: # %loadbb1
 ; X64-NEXT:    movzbl 4(%rdi), %eax
 ; X64-NEXT:    movzbl 4(%rsi), %ecx
 ; X64-NEXT:    subl %ecx, %eax
 ; X64-NEXT:    retq
-; X64-NEXT:  .LBB9_1: # %res_block
+; X64-NEXT:  .LBB10_1: # %res_block
 ; X64-NEXT:    setae %al
 ; X64-NEXT:    movzbl %al, %eax
 ; X64-NEXT:    leal -1(%rax,%rax), %eax
@@ -337,25 +337,25 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    cmpl %edx, %ecx
-; X86-NEXT:    jne .LBB11_2
+; X86-NEXT:    jne .LBB12_2
 ; X86-NEXT:  # %bb.1: # %loadbb1
 ; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
-; X86-NEXT:    je .LBB11_3
-; X86-NEXT:  .LBB11_2: # %res_block
+; X86-NEXT:    je .LBB12_3
+; X86-NEXT:  .LBB12_2: # %res_block
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
 ; X86-NEXT:    setae %al
 ; X86-NEXT:    leal -1(%eax,%eax), %eax
-; X86-NEXT:  .LBB11_3: # %endblock
+; X86-NEXT:  .LBB12_3: # %endblock
 ; X86-NEXT:    popl %esi
 ; X86-NEXT:    retl
 ;
@@ -462,25 +462,25 @@
 ; X64-LABEL: length12:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
-; X64-NEXT:    jne .LBB15_2
+; X64-NEXT:    jne .LBB16_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
+; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
-; X64-NEXT:    je .LBB15_3
-; X64-NEXT:  .LBB15_2: # %res_block
+; X64-NEXT:    je .LBB16_3
+; X64-NEXT:  .LBB16_2: # %res_block
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    setae %al
 ; X64-NEXT:    leal -1(%rax,%rax), %eax
-; X64-NEXT:  .LBB15_3: # %endblock
+; X64-NEXT:  .LBB16_3: # %endblock
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(i8* %X, i8* %Y, i64 12) nounwind
   ret i32 %m
@@ -502,25 +502,25 @@
 ; X64-LABEL: length16:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
-; X64-NEXT:    jne .LBB16_2
+; X64-NEXT:    jne .LBB17_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
-; X64-NEXT:    je .LBB16_3
-; X64-NEXT:  .LBB16_2: # %res_block
+; X64-NEXT:    je .LBB17_3
+; X64-NEXT:  .LBB17_2: # %res_block
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    setae %al
 ; X64-NEXT:    leal -1(%rax,%rax), %eax
-; X64-NEXT:  .LBB16_3: # %endblock
+; X64-NEXT:  .LBB17_3: # %endblock
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(i8* %X, i8* %Y, i64 16) nounwind
   ret i32 %m
diff --git a/llvm/test/CodeGen/X86/memcmp.ll b/llvm/test/CodeGen/X86/memcmp.ll
--- a/llvm/test/CodeGen/X86/memcmp.ll
+++ b/llvm/test/CodeGen/X86/memcmp.ll
@@ -68,24 +68,24 @@
 define i32 @length2(i8* %X, i8* %Y) nounwind {
 ; X86-LABEL: length2:
 ; X86:       # %bb.0:
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT:    movzwl (%eax), %eax
+; X86-NEXT:    rolw $8, %ax
+; X86-NEXT:    movzwl %ax, %eax
 ; X86-NEXT:    movzwl (%ecx), %ecx
-; X86-NEXT:    movzwl (%eax), %edx
 ; X86-NEXT:    rolw $8, %cx
-; X86-NEXT:    rolw $8, %dx
-; X86-NEXT:    movzwl %cx, %eax
-; X86-NEXT:    movzwl %dx, %ecx
+; X86-NEXT:    movzwl %cx, %ecx
 ; X86-NEXT:    subl %ecx, %eax
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: length2:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
-; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %ax, %eax
+; X64-NEXT:    movzwl (%rsi), %ecx
+; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %cx, %ecx
 ; X64-NEXT:    subl %ecx, %eax
 ; X64-NEXT:    retq
@@ -98,23 +98,17 @@
 ; X86:       # %bb.0:
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movzwl (%eax), %eax
-; X86-NEXT:    movzwl .L.str+1, %ecx
 ; X86-NEXT:    rolw $8, %ax
-; X86-NEXT:    rolw $8, %cx
 ; X86-NEXT:    movzwl %ax, %eax
-; X86-NEXT:    movzwl %cx, %ecx
-; X86-NEXT:    subl %ecx, %eax
+; X86-NEXT:    addl $-12594, %eax # imm = 0xCECE
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: length2_const:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl .L.str+{{.*}}(%rip), %ecx
 ; X64-NEXT:    rolw $8, %ax
-; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %ax, %eax
-; X64-NEXT:    movzwl %cx, %ecx
-; X64-NEXT:    subl %ecx, %eax
+; X64-NEXT:    addl $-12594, %eax # imm = 0xCECE
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(i8* %X, i8* getelementptr inbounds ([513 x i8], [513 x i8]* @.str, i32 0, i32 1), i64 2) nounwind
   ret i32 %m
@@ -125,12 +119,9 @@
 ; X86:       # %bb.0:
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movzwl (%eax), %eax
-; X86-NEXT:    movzwl .L.str+1, %ecx
 ; X86-NEXT:    rolw $8, %ax
-; X86-NEXT:    rolw $8, %cx
 ; X86-NEXT:    movzwl %ax, %eax
-; X86-NEXT:    movzwl %cx, %ecx
-; X86-NEXT:    subl %ecx, %eax
+; X86-NEXT:    addl $-12594, %eax # imm = 0xCECE
 ; X86-NEXT:    testl %eax, %eax
 ; X86-NEXT:    setg %al
 ; X86-NEXT:    retl
@@ -138,12 +129,9 @@
 ; X64-LABEL: length2_gt_const:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl .L.str+{{.*}}(%rip), %ecx
 ; X64-NEXT:    rolw $8, %ax
-; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %ax, %eax
-; X64-NEXT:    movzwl %cx, %ecx
-; X64-NEXT:    subl %ecx, %eax
+; X64-NEXT:    addl $-12594, %eax # imm = 0xCECE
 ; X64-NEXT:    testl %eax, %eax
 ; X64-NEXT:    setg %al
 ; X64-NEXT:    retq
@@ -176,14 +164,14 @@
 define i1 @length2_lt(i8* %X, i8* %Y) nounwind {
 ; X86-LABEL: length2_lt:
 ; X86:       # %bb.0:
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT:    movzwl (%eax), %eax
+; X86-NEXT:    rolw $8, %ax
+; X86-NEXT:    movzwl %ax, %eax
 ; X86-NEXT:    movzwl (%ecx), %ecx
-; X86-NEXT:    movzwl (%eax), %edx
 ; X86-NEXT:    rolw $8, %cx
-; X86-NEXT:    rolw $8, %dx
-; X86-NEXT:    movzwl %cx, %eax
-; X86-NEXT:    movzwl %dx, %ecx
+; X86-NEXT:    movzwl %cx, %ecx
 ; X86-NEXT:    subl %ecx, %eax
 ; X86-NEXT:    shrl $31, %eax
 ; X86-NEXT:    # kill: def $al killed $al killed $eax
@@ -192,10 +180,10 @@
 ; X64-LABEL: length2_lt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
-; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %ax, %eax
+; X64-NEXT:    movzwl (%rsi), %ecx
+; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %cx, %ecx
 ; X64-NEXT:    subl %ecx, %eax
 ; X64-NEXT:    shrl $31, %eax
@@ -212,10 +200,10 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movzwl (%ecx), %ecx
-; X86-NEXT:    movzwl (%eax), %eax
 ; X86-NEXT:    rolw $8, %cx
-; X86-NEXT:    rolw $8, %ax
 ; X86-NEXT:    movzwl %cx, %ecx
+; X86-NEXT:    movzwl (%eax), %eax
+; X86-NEXT:    rolw $8, %ax
 ; X86-NEXT:    movzwl %ax, %eax
 ; X86-NEXT:    subl %eax, %ecx
 ; X86-NEXT:    testl %ecx, %ecx
@@ -225,10 +213,10 @@
 ; X64-LABEL: length2_gt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
-; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %ax, %eax
+; X64-NEXT:    movzwl (%rsi), %ecx
+; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    movzwl %cx, %ecx
 ; X64-NEXT:    subl %ecx, %eax
 ; X64-NEXT:    testl %eax, %eax
@@ -293,8 +281,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movzwl (%eax), %edx
-; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %dx
+; X86-NEXT:    movzwl (%ecx), %esi
 ; X86-NEXT:    rolw $8, %si
 ; X86-NEXT:    cmpw %si, %dx
 ; X86-NEXT:    jne .LBB11_1
@@ -314,8 +302,8 @@
 ; X64-LABEL: length3:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
+; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %cx
 ; X64-NEXT:    cmpw %cx, %ax
 ; X64-NEXT:    jne .LBB11_1
@@ -368,8 +356,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl (%ecx), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
@@ -380,8 +368,8 @@
 ; X64-LABEL: length4:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl (%rdi), %ecx
-; X64-NEXT:    movl (%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
+; X64-NEXT:    movl (%rsi), %edx
 ; X64-NEXT:    bswapl %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpl %edx, %ecx
@@ -419,8 +407,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl (%ecx), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
@@ -433,8 +421,8 @@
 ; X64-LABEL: length4_lt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl (%rdi), %ecx
-; X64-NEXT:    movl (%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
+; X64-NEXT:    movl (%rsi), %edx
 ; X64-NEXT:    bswapl %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpl %edx, %ecx
@@ -454,8 +442,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl (%ecx), %ecx
-; X86-NEXT:    movl (%eax), %eax
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %eax
 ; X86-NEXT:    bswapl %eax
 ; X86-NEXT:    xorl %edx, %edx
 ; X86-NEXT:    cmpl %eax, %ecx
@@ -468,8 +456,8 @@
 ; X64-LABEL: length4_gt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    xorl %edx, %edx
 ; X64-NEXT:    cmpl %ecx, %eax
@@ -508,8 +496,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl (%eax), %edx
-; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    cmpl %esi, %edx
 ; X86-NEXT:    jne .LBB18_1
@@ -529,8 +517,8 @@
 ; X64-LABEL: length5:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    cmpl %ecx, %eax
 ; X64-NEXT:    jne .LBB18_1
@@ -584,8 +572,8 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl (%eax), %edx
-; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %edx
+; X86-NEXT:    movl (%ecx), %esi
 ; X86-NEXT:    bswapl %esi
 ; X86-NEXT:    cmpl %esi, %edx
 ; X86-NEXT:    jne .LBB20_1
@@ -607,8 +595,8 @@
 ; X64-LABEL: length5_lt:
 ; X64:       # %bb.0: # %loadbb
 ; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rsi), %ecx
 ; X64-NEXT:    bswapl %ecx
 ; X64-NEXT:    cmpl %ecx, %eax
 ; X64-NEXT:    jne .LBB20_1
@@ -665,15 +653,15 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; X86-NEXT:    movl (%esi), %ecx
-; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    cmpl %edx, %ecx
 ; X86-NEXT:    jne .LBB22_2
 ; X86-NEXT:  # %bb.1: # %loadbb1
 ; X86-NEXT:    movl 4(%esi), %ecx
-; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %ecx
+; X86-NEXT:    movl 4(%eax), %edx
 ; X86-NEXT:    bswapl %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    cmpl %edx, %ecx
@@ -690,8 +678,8 @@
 ; X64-LABEL: length8:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -873,15 +861,15 @@
 ; X64-LABEL: length12:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB29_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
+; X64-NEXT:    movl 8(%rsi), %edx
 ; X64-NEXT:    bswapl %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1042,15 +1030,15 @@
 ; X64-LABEL: length16:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB35_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1171,15 +1159,15 @@
 ; X64-LABEL: length16_lt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rcx
-; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    cmpq %rdx, %rcx
 ; X64-NEXT:    jne .LBB37_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rcx
-; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
+; X64-NEXT:    movq 8(%rsi), %rdx
 ; X64-NEXT:    bswapq %rdx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    cmpq %rdx, %rcx
@@ -1214,15 +1202,15 @@
 ; X64-LABEL: length16_gt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq (%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    cmpq %rcx, %rax
 ; X64-NEXT:    jne .LBB38_2
 ; X64-NEXT:  # %bb.1: # %loadbb1
 ; X64-NEXT:    movq 8(%rdi), %rax
-; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq 8(%rsi), %rcx
 ; X64-NEXT:    bswapq %rcx
 ; X64-NEXT:    xorl %edx, %edx
 ; X64-NEXT:    cmpq %rcx, %rax