In visitSTORE, always use FindBetterChain, rather than only when UseAA is enabled.

    Recommiting after fixing X86 inc/dec chain bug.

    * Simplify Consecutive Merge Store Candidate Search

    Now that address aliasing is much less conservative, push through
    simplified store merging search and chain alias analysis which only
    checks for parallel stores through the chain subgraph. This is cleaner
    as the separation of non-interfering loads/stores from the
    store-merging logic.

    When merging stores search up the chain through a single load, and
    finds all possible stores by looking down from through a load and a
    TokenFactor to all stores visited.

    This improves the quality of the output SelectionDAG and the output
    Codegen (save perhaps for some ARM cases where we correctly constructs
    wider loads, but then promotes them to float operations which appear
    but requires more expensive constant generation).

    Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)

    Additional Minor Changes:

      1. Finishes removing unused AliasLoad code

      2. Unifies the chain aggregation in the merged stores across code
         paths

      3. Re-add the Store node to the worklist after calling
         SimplifyDemandedBits.

      4. Increase GatherAllAliasesMaxDepth from 6 to 18. That number is
         arbitrary, but seems sufficient to not cause regressions in
         tests.

      5. Remove Chain dependencies of Memory operations on CopyfromReg
         nodes as these are captured by data dependence

      6. Forward loads-store values through tokenfactors containing
          {CopyToReg,CopyFromReg} Values.

      7. Peephole to convert buildvector of extract_vector_elt to
         extract_subvector if possible (see
         CodeGen/AArch64/store-merge.ll)

      8. Store merging for the ARM target is restricted to 32-bit as
         some in some contexts invalid 64-bit operations are being
         generated. This can be removed once appropriate checks are
         added.

    This finishes the change Matt Arsenault started in r246307 and
    jyknight's original patch.

    Many tests required some changes as memory operations are now
    reorderable, improving load-store forwarding. One test in
    particular is worth noting:

      CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
      forwarding converts a load-store pair into a parallel store and
      a memory-realized bitcast of the same value. However, because we
      lose the sharing of the explicit and implicit store values we
      must create another local store. A similar transformation
      happens before SelectionDAG as well.

    Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle

llvm-svn: 293893

72 files changed, 2046 insertions, 2226 deletions

diff --git a/llvm/include/llvm/Target/TargetLowering.h b/llvm/include/llvm/Target/TargetLowering.h
index 01c5c51f6b1..2f3b51c992f 100644
--- a/llvm/include/llvm/Target/TargetLowering.h
+++ b/llvm/include/llvm/Target/TargetLowering.h
@@ -363,6 +363,9 @@ public:
     return false;
   }
 
+  /// Returns if it's reasonable to merge stores to MemVT size.
+  virtual bool canMergeStoresTo(EVT MemVT) const { return true; }
+
   /// \brief Return true if it is cheap to speculate a call to intrinsic cttz.
   virtual bool isCheapToSpeculateCttz() const {
     return false;
diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 58942469556..315214385bb 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -53,10 +53,6 @@ STATISTIC(SlicedLoads, "Number of load sliced");
 
 namespace {
   static cl::opt<bool>
-    CombinerAA("combiner-alias-analysis", cl::Hidden,
-               cl::desc("Enable DAG combiner alias-analysis heuristics"));
-
-  static cl::opt<bool>
     CombinerGlobalAA("combiner-global-alias-analysis", cl::Hidden,
                cl::desc("Enable DAG combiner's use of IR alias analysis"));
 
@@ -417,15 +413,12 @@ namespace {
     /// Holds a pointer to an LSBaseSDNode as well as information on where it
     /// is located in a sequence of memory operations connected by a chain.
     struct MemOpLink {
-      MemOpLink (LSBaseSDNode *N, int64_t Offset, unsigned Seq):
-      MemNode(N), OffsetFromBase(Offset), SequenceNum(Seq) { }
+      MemOpLink(LSBaseSDNode *N, int64_t Offset)
+          : MemNode(N), OffsetFromBase(Offset) {}
       // Ptr to the mem node.
       LSBaseSDNode *MemNode;
       // Offset from the base ptr.
       int64_t OffsetFromBase;
-      // What is the sequence number of this mem node.
-      // Lowest mem operand in the DAG starts at zero.
-      unsigned SequenceNum;
     };
 
     /// This is a helper function for visitMUL to check the profitability
@@ -436,12 +429,6 @@ namespace {
                                      SDValue &AddNode,
                                      SDValue &ConstNode);
 
-    /// This is a helper function for MergeStoresOfConstantsOrVecElts. Returns a
-    /// constant build_vector of the stored constant values in Stores.
-    SDValue getMergedConstantVectorStore(SelectionDAG &DAG, const SDLoc &SL,
-                                         ArrayRef<MemOpLink> Stores,
-                                         SmallVectorImpl<SDValue> &Chains,
-                                         EVT Ty) const;
 
     /// This is a helper function for visitAND and visitZERO_EXTEND.  Returns
     /// true if the (and (load x) c) pattern matches an extload.  ExtVT returns
@@ -455,18 +442,15 @@ namespace {
     /// This is a helper function for MergeConsecutiveStores. When the source
     /// elements of the consecutive stores are all constants or all extracted
     /// vector elements, try to merge them into one larger store.
-    /// \return number of stores that were merged into a merged store (always
-    /// a prefix of \p StoreNode).
-    bool MergeStoresOfConstantsOrVecElts(
-        SmallVectorImpl<MemOpLink> &StoreNodes, EVT MemVT, unsigned NumStores,
-        bool IsConstantSrc, bool UseVector);
+    /// \return True if a merged store was created.
+    bool MergeStoresOfConstantsOrVecElts(SmallVectorImpl<MemOpLink> &StoreNodes,
+                                         EVT MemVT, unsigned NumStores,
+                                         bool IsConstantSrc, bool UseVector);
 
     /// This is a helper function for MergeConsecutiveStores.
     /// Stores that may be merged are placed in StoreNodes.
-    /// Loads that may alias with those stores are placed in AliasLoadNodes.
-    void getStoreMergeAndAliasCandidates(
-        StoreSDNode* St, SmallVectorImpl<MemOpLink> &StoreNodes,
-        SmallVectorImpl<LSBaseSDNode*> &AliasLoadNodes);
+    void getStoreMergeCandidates(StoreSDNode *St,
+                                 SmallVectorImpl<MemOpLink> &StoreNodes);
 
     /// Helper function for MergeConsecutiveStores. Checks if
     /// Candidate stores have indirect dependency through their
@@ -478,8 +462,7 @@ namespace {
     /// This optimization uses wide integers or vectors when possible.
     /// \return number of stores that were merged into a merged store (the
     /// affected nodes are stored as a prefix in \p StoreNodes).
-    bool MergeConsecutiveStores(StoreSDNode *N,
-                                SmallVectorImpl<MemOpLink> &StoreNodes);
+    bool MergeConsecutiveStores(StoreSDNode *N);
 
     /// \brief Try to transform a truncation where C is a constant:
     ///     (trunc (and X, C)) -> (and (trunc X), (trunc C))
@@ -1582,7 +1565,7 @@ SDValue DAGCombiner::visitTokenFactor(SDNode *N) {
   }
 
   SmallVector<SDNode *, 8> TFs;     // List of token factors to visit.
-  SmallVector<SDValue, 8> Ops;    // Ops for replacing token factor.
+  SmallVector<SDValue, 8> Ops;      // Ops for replacing token factor.
   SmallPtrSet<SDNode*, 16> SeenOps;
   bool Changed = false;             // If we should replace this token factor.
 
@@ -1626,6 +1609,86 @@ SDValue DAGCombiner::visitTokenFactor(SDNode *N) {
     }
   }
 
+  // Remove Nodes that are chained to another node in the list. Do so
+  // by walking up chains breath-first stopping when we've seen
+  // another operand. In general we must climb to the EntryNode, but we can exit
+  // early if we find all remaining work is associated with just one operand as
+  // no further pruning is possible.
+
+  // List of nodes to search through and original Ops from which they originate.
+  SmallVector<std::pair<SDNode *, unsigned>, 8> Worklist;
+  SmallVector<unsigned, 8> OpWorkCount; // Count of work for each Op.
+  SmallPtrSet<SDNode *, 16> SeenChains;
+  bool DidPruneOps = false;
+
+  unsigned NumLeftToConsider = 0;
+  for (const SDValue &Op : Ops) {
+    Worklist.push_back(std::make_pair(Op.getNode(), NumLeftToConsider++));
+    OpWorkCount.push_back(1);
+  }
+
+  auto AddToWorklist = [&](unsigned CurIdx, SDNode *Op, unsigned OpNumber) {
+    // If this is an Op, we can remove the op from the list. Remark any
+    // search associated with it as from the current OpNumber.
+    if (SeenOps.count(Op) != 0) {
+      Changed = true;
+      DidPruneOps = true;
+      unsigned OrigOpNumber = 0;
+      while (Ops[OrigOpNumber].getNode() != Op && OrigOpNumber < Ops.size())
+        OrigOpNumber++;
+      assert((OrigOpNumber != Ops.size()) &&
+             "expected to find TokenFactor Operand");
+      // Re-mark worklist from OrigOpNumber to OpNumber
+      for (unsigned i = CurIdx + 1; i < Worklist.size(); ++i) {
+        if (Worklist[i].second == OrigOpNumber) {
+          Worklist[i].second = OpNumber;
+        }
+      }
+      OpWorkCount[OpNumber] += OpWorkCount[OrigOpNumber];
+      OpWorkCount[OrigOpNumber] = 0;
+      NumLeftToConsider--;
+    }
+    // Add if it's a new chain
+    if (SeenChains.insert(Op).second) {
+      OpWorkCount[OpNumber]++;
+      Worklist.push_back(std::make_pair(Op, OpNumber));
+    }
+  };
+
+  for (unsigned i = 0; i < Worklist.size(); ++i) {
+    // We need at least be consider at least 2 Ops to prune.
+    if (NumLeftToConsider <= 1)
+      break;
+    auto CurNode = Worklist[i].first;
+    auto CurOpNumber = Worklist[i].second;
+    assert((OpWorkCount[CurOpNumber] > 0) &&
+           "Node should not appear in worklist");
+    switch (CurNode->getOpcode()) {
+    case ISD::EntryToken:
+      // Hitting EntryToken is the only way for the search to terminate without
+      // hitting
+      // another operand's search. Prevent us from marking this operand
+      // considered.
+      NumLeftToConsider++;
+      break;
+    case ISD::TokenFactor:
+      for (const SDValue &Op : CurNode->op_values())
+        AddToWorklist(i, Op.getNode(), CurOpNumber);
+      break;
+    case ISD::CopyFromReg:
+    case ISD::CopyToReg:
+      AddToWorklist(i, CurNode->getOperand(0).getNode(), CurOpNumber);
+      break;
+    default:
+      if (auto *MemNode = dyn_cast<MemSDNode>(CurNode))
+        AddToWorklist(i, MemNode->getChain().getNode(), CurOpNumber);
+      break;
+    }
+    OpWorkCount[CurOpNumber]--;
+    if (OpWorkCount[CurOpNumber] == 0)
+      NumLeftToConsider--;
+  }
+
   SDValue Result;
 
   // If we've changed things around then replace token factor.
@@ -1634,15 +1697,22 @@ SDValue DAGCombiner::visitTokenFactor(SDNode *N) {
       // The entry token is the only possible outcome.
       Result = DAG.getEntryNode();
     } else {
-      // New and improved token factor.
-      Result = DAG.getNode(ISD::TokenFactor, SDLoc(N), MVT::Other, Ops);
+      if (DidPruneOps) {
+        SmallVector<SDValue, 8> PrunedOps;
+        //
+        for (const SDValue &Op : Ops) {
+          if (SeenChains.count(Op.getNode()) == 0)
+            PrunedOps.push_back(Op);
+        }
+        Result = DAG.getNode(ISD::TokenFactor, SDLoc(N), MVT::Other, PrunedOps);
+      } else {
+        Result = DAG.getNode(ISD::TokenFactor, SDLoc(N), MVT::Other, Ops);
+      }
     }
 
-    // Add users to worklist if AA is enabled, since it may introduce
-    // a lot of new chained token factors while removing memory deps.
-    bool UseAA = CombinerAA.getNumOccurrences() > 0 ? CombinerAA
-      : DAG.getSubtarget().useAA();
-    return CombineTo(N, Result, UseAA /*add to worklist*/);
+    // Add users to worklist, since we may introduce a lot of new
+    // chained token factors while removing memory deps.
+    return CombineTo(N, Result, true /*add to worklist*/);
   }
 
   return Result;
@@ -6540,6 +6610,9 @@ SDValue DAGCombiner::CombineExtLoad(SDNode *N) {
   SDValue NewChain = DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Chains);
   SDValue NewValue = DAG.getNode(ISD::CONCAT_VECTORS, DL, DstVT, Loads);
 
+  // Simplify TF.
+  AddToWorklist(NewChain.getNode());
+
   CombineTo(N, NewValue);
 
   // Replace uses of the original load (before extension)
@@ -10692,11 +10765,12 @@ SDValue DAGCombiner::visitLOAD(SDNode *N) {
   // TODO: Handle TRUNCSTORE/LOADEXT
   if (OptLevel != CodeGenOpt::None &&
       ISD::isNormalLoad(N) && !LD->isVolatile()) {
+    // We can forward a direct store or a store off of a tokenfactor.
     if (ISD::isNON_TRUNCStore(Chain.getNode())) {
       StoreSDNode *PrevST = cast<StoreSDNode>(Chain);
       if (PrevST->getBasePtr() == Ptr &&
           PrevST->getValue().getValueType() == N->getValueType(0))
-      return CombineTo(N, Chain.getOperand(1), Chain);
+        return CombineTo(N, PrevST->getOperand(1), Chain);
     }
   }
 
@@ -10714,14 +10788,7 @@ SDValue DAGCombiner::visitLOAD(SDNode *N) {
     }
   }
 
-  bool UseAA = CombinerAA.getNumOccurrences() > 0 ? CombinerAA
-                                                  : DAG.getSubtarget().useAA();
-#ifndef NDEBUG
-  if (CombinerAAOnlyFunc.getNumOccurrences() &&
-      CombinerAAOnlyFunc != DAG.getMachineFunction().getName())
-    UseAA = false;
-#endif
-  if (UseAA && LD->isUnindexed()) {
+  if (LD->isUnindexed()) {
     // Walk up chain skipping non-aliasing memory nodes.
     SDValue BetterChain = FindBetterChain(N, Chain);
 
@@ -11303,6 +11370,7 @@ bool DAGCombiner::SliceUpLoad(SDNode *N) {
   SDValue Chain = DAG.getNode(ISD::TokenFactor, SDLoc(LD), MVT::Other,
                               ArgChains);
   DAG.ReplaceAllUsesOfValueWith(SDValue(N, 1), Chain);
+  AddToWorklist(Chain.getNode());
   return true;
 }
 
@@ -11696,20 +11764,6 @@ bool DAGCombiner::isMulAddWithConstProfitable(SDNode *MulNode,
   return false;
 }
 
-SDValue DAGCombiner::getMergedConstantVectorStore(
-    SelectionDAG &DAG, const SDLoc &SL, ArrayRef<MemOpLink> Stores,
-    SmallVectorImpl<SDValue> &Chains, EVT Ty) const {
-  SmallVector<SDValue, 8> BuildVector;
-
-  for (unsigned I = 0, E = Ty.getVectorNumElements(); I != E; ++I) {
-    StoreSDNode *St = cast<StoreSDNode>(Stores[I].MemNode);
-    Chains.push_back(St->getChain());
-    BuildVector.push_back(St->getValue());
-  }
-
-  return DAG.getBuildVector(Ty, SL, BuildVector);
-}
-
 bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
                   SmallVectorImpl<MemOpLink> &StoreNodes, EVT MemVT,
                   unsigned NumStores, bool IsConstantSrc, bool UseVector) {
@@ -11718,22 +11772,8 @@ bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
     return false;
 
   int64_t ElementSizeBytes = MemVT.getSizeInBits() / 8;
-  LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
-  unsigned LatestNodeUsed = 0;
-
-  for (unsigned i=0; i < NumStores; ++i) {
-    // Find a chain for the new wide-store operand. Notice that some
-    // of the store nodes that we found may not be selected for inclusion
-    // in the wide store. The chain we use needs to be the chain of the
-    // latest store node which is *used* and replaced by the wide store.
-    if (StoreNodes[i].SequenceNum < StoreNodes[LatestNodeUsed].SequenceNum)
-      LatestNodeUsed = i;
-  }
-
-  SmallVector<SDValue, 8> Chains;
 
   // The latest Node in the DAG.
-  LSBaseSDNode *LatestOp = StoreNodes[LatestNodeUsed].MemNode;
   SDLoc DL(StoreNodes[0].MemNode);
 
   SDValue StoredVal;
@@ -11749,7 +11789,18 @@ bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
     assert(TLI.isTypeLegal(Ty) && "Illegal vector store");
 
     if (IsConstantSrc) {
-      StoredVal = getMergedConstantVectorStore(DAG, DL, StoreNodes, Chains, Ty);
+      SmallVector<SDValue, 8> BuildVector;
+      for (unsigned I = 0, E = Ty.getVectorNumElements(); I != E; ++I) {
+        StoreSDNode *St = cast<StoreSDNode>(StoreNodes[I].MemNode);
+        SDValue Val = St->getValue();
+        if (MemVT.getScalarType().isInteger())
+          if (auto *CFP = dyn_cast<ConstantFPSDNode>(St->getValue()))
+            Val = DAG.getConstant(
+                (uint32_t)CFP->getValueAPF().bitcastToAPInt().getZExtValue(),
+                SDLoc(CFP), MemVT);
+        BuildVector.push_back(Val);
+      }
+      StoredVal = DAG.getBuildVector(Ty, DL, BuildVector);
     } else {
       SmallVector<SDValue, 8> Ops;
       for (unsigned i = 0; i < NumStores; ++i) {
@@ -11759,7 +11810,6 @@ bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
         if (Val.getValueType() != MemVT)
           return false;
         Ops.push_back(Val);
-        Chains.push_back(St->getChain());
       }
 
       // Build the extracted vector elements back into a vector.
@@ -11779,7 +11829,6 @@ bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
     for (unsigned i = 0; i < NumStores; ++i) {
       unsigned Idx = IsLE ? (NumStores - 1 - i) : i;
       StoreSDNode *St  = cast<StoreSDNode>(StoreNodes[Idx].MemNode);
-      Chains.push_back(St->getChain());
 
       SDValue Val = St->getValue();
       StoreInt <<= ElementSizeBytes * 8;
@@ -11797,54 +11846,36 @@ bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
     StoredVal = DAG.getConstant(StoreInt, DL, StoreTy);
   }
 
-  assert(!Chains.empty());
+  SmallVector<SDValue, 8> Chains;
+
+  // Gather all Chains we're inheriting. As generally all chains are
+  // equal, do minor check to remove obvious redundancies.
+  Chains.push_back(StoreNodes[0].MemNode->getChain());
+  for (unsigned i = 1; i < NumStores; ++i)
+    if (StoreNodes[0].MemNode->getChain() != StoreNodes[i].MemNode->getChain())
+      Chains.push_back(StoreNodes[i].MemNode->getChain());
 
+  LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
   SDValue NewChain = DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Chains);
   SDValue NewStore = DAG.getStore(NewChain, DL, StoredVal,
                                   FirstInChain->getBasePtr(),
                                   FirstInChain->getPointerInfo(),
                                   FirstInChain->getAlignment());
 
-  bool UseAA = CombinerAA.getNumOccurrences() > 0 ? CombinerAA
-                                                  : DAG.getSubtarget().useAA();
-  if (UseAA) {
-    // Replace all merged stores with the new store.
-    for (unsigned i = 0; i < NumStores; ++i)
-      CombineTo(StoreNodes[i].MemNode, NewStore);
-  } else {
-    // Replace the last store with the new store.
-    CombineTo(LatestOp, NewStore);
-    // Erase all other stores.
-    for (unsigned i = 0; i < NumStores; ++i) {
-      if (StoreNodes[i].MemNode == LatestOp)
-        continue;
-      StoreSDNode *St = cast<StoreSDNode>(StoreNodes[i].MemNode);
-      // ReplaceAllUsesWith will replace all uses that existed when it was
-      // called, but graph optimizations may cause new ones to appear. For
-      // example, the case in pr14333 looks like
-      //
-      //  St's chain -> St -> another store -> X
-      //
-      // And the only difference from St to the other store is the chain.
-      // When we change it's chain to be St's chain they become identical,
-      // get CSEed and the net result is that X is now a use of St.
-      // Since we know that St is redundant, just iterate.
-      while (!St->use_empty())
-        DAG.ReplaceAllUsesWith(SDValue(St, 0), St->getChain());
-      deleteAndRecombine(St);
-    }
-  }
+  // Replace all merged stores with the new store.
+  for (unsigned i = 0; i < NumStores; ++i)
+    CombineTo(StoreNodes[i].MemNode, NewStore);
 
-  StoreNodes.erase(StoreNodes.begin() + NumStores, StoreNodes.end());
+  AddToWorklist(NewChain.getNode());
   return true;
 }
 
-void DAGCombiner::getStoreMergeAndAliasCandidates(
-    StoreSDNode* St, SmallVectorImpl<MemOpLink> &StoreNodes,
-    SmallVectorImpl<LSBaseSDNode*> &AliasLoadNodes) {
+void DAGCombiner::getStoreMergeCandidates(
+    StoreSDNode *St, SmallVectorImpl<MemOpLink> &StoreNodes) {
   // This holds the base pointer, index, and the offset in bytes from the base
   // pointer.
   BaseIndexOffset BasePtr = BaseIndexOffset::match(St->getBasePtr(), DAG);
+  EVT MemVT = St->getMemoryVT();
 
   // We must have a base and an offset.
   if (!BasePtr.Base.getNode())
@@ -11854,104 +11885,70 @@ void DAGCombiner::getStoreMergeAndAliasCandidates(
   if (BasePtr.Base.isUndef())
     return;
 
-  // Walk up the chain and look for nodes with offsets from the same
-  // base pointer. Stop when reaching an instruction with a different kind
-  // or instruction which has a different base pointer.
-  EVT MemVT = St->getMemoryVT();
-  unsigned Seq = 0;
-  StoreSDNode *Index = St;
-
-
-  bool UseAA = CombinerAA.getNumOccurrences() > 0 ? CombinerAA
-                                                  : DAG.getSubtarget().useAA();
-
-  if (UseAA) {
-    // Look at other users of the same chain. Stores on the same chain do not
-    // alias. If combiner-aa is enabled, non-aliasing stores are canonicalized
-    // to be on the same chain, so don't bother looking at adjacent chains.
-
-    SDValue Chain = St->getChain();
-    for (auto I = Chain->use_begin(), E = Chain->use_end(); I != E; ++I) {
-      if (StoreSDNode *OtherST = dyn_cast<StoreSDNode>(*I)) {
-        if (I.getOperandNo() != 0)
-          continue;
-
-        if (OtherST->isVolatile() || OtherST->isIndexed())
-          continue;
-
-        if (OtherST->getMemoryVT() != MemVT)
-          continue;
-
-        BaseIndexOffset Ptr = BaseIndexOffset::match(OtherST->getBasePtr(), DAG);
-
-        if (Ptr.equalBaseIndex(BasePtr))
-          StoreNodes.push_back(MemOpLink(OtherST, Ptr.Offset, Seq++));
-      }
-    }
-
-    return;
-  }
-
-  while (Index) {
-    // If the chain has more than one use, then we can't reorder the mem ops.
-    if (Index != St && !SDValue(Index, 0)->hasOneUse())
-      break;
-
-    // Find the base pointer and offset for this memory node.
-    BaseIndexOffset Ptr = BaseIndexOffset::match(Index->getBasePtr(), DAG);
-
-    // Check that the base pointer is the same as the original one.
-    if (!Ptr.equalBaseIndex(BasePtr))
-      break;
-
-    // The memory operands must not be volatile.
-    if (Index->isVolatile() || Index->isIndexed())
-      break;
+  // We looking for a root node which is an ancestor to all mergable
+  // stores. We search up through a load, to our root and then down
+  // through all children. For instance we will find Store{1,2,3} if
+  // St is Store1, Store2. or Store3 where the root is not a load
+  // which always true for nonvolatile ops. TODO: Expand
+  // the search to find all valid candidates through multiple layers of loads.
+  //
+  // Root
+  // |-------|-------|
+  // Load    Load    Store3
+  // |       |
+  // Store1   Store2
+  //
+  // FIXME: We should be able to climb and
+  // descend TokenFactors to find candidates as well.
 
-    // No truncation.
-    if (Index->isTruncatingStore())
-      break;
+  SDNode *RootNode = (St->getChain()).getNode();
 
-    // The stored memory type must be the same.
-    if (Index->getMemoryVT() != MemVT)
-      break;
+  // Set of Parents of Candidates
+  std::set<SDNode *> CandidateParents;
 
-    // We do not allow under-aligned stores in order to prevent
-    // overriding stores. NOTE: this is a bad hack. Alignment SHOULD
-    // be irrelevant here; what MATTERS is that we not move memory
-    // operations that potentially overlap past each-other.
-    if (Index->getAlignment() < MemVT.getStoreSize())
-      break;
-
-    // We found a potential memory operand to merge.
-    StoreNodes.push_back(MemOpLink(Index, Ptr.Offset, Seq++));
+  if (LoadSDNode *Ldn = dyn_cast<LoadSDNode>(RootNode)) {
+    RootNode = Ldn->getChain().getNode();
+    for (auto I = RootNode->use_begin(), E = RootNode->use_end(); I != E; ++I)
+      if (I.getOperandNo() == 0 && isa<LoadSDNode>(*I)) // walk down chain
+        CandidateParents.insert(*I);
+  } else
+    CandidateParents.insert(RootNode);
+
+  bool IsLoadSrc = isa<LoadSDNode>(St->getValue());
+  bool IsConstantSrc = isa<ConstantSDNode>(St->getValue()) ||
+                       isa<ConstantFPSDNode>(St->getValue());
+  bool IsExtractVecSrc =
+      (St->getValue().getOpcode() == ISD::EXTRACT_VECTOR_ELT ||
+       St->getValue().getOpcode() == ISD::EXTRACT_SUBVECTOR);
+  auto CorrectValueKind = [&](StoreSDNode *Other) -> bool {
+    if (IsLoadSrc)
+      return isa<LoadSDNode>(Other->getValue());
+    if (IsConstantSrc)
+      return (isa<ConstantSDNode>(Other->getValue()) ||
+              isa<ConstantFPSDNode>(Other->getValue()));
+    if (IsExtractVecSrc)
+      return (Other->getValue().getOpcode() == ISD::EXTRACT_VECTOR_ELT ||
+              Other->getValue().getOpcode() == ISD::EXTRACT_SUBVECTOR);
+    return false;
+  };
 
-    // Find the next memory operand in the chain. If the next operand in the
-    // chain is a store then move up and continue the scan with the next
-    // memory operand. If the next operand is a load save it and use alias
-    // information to check if it interferes with anything.
-    SDNode *NextInChain = Index->getChain().getNode();
-    while (1) {
-      if (StoreSDNode *STn = dyn_cast<StoreSDNode>(NextInChain)) {
-        // We found a store node. Use it for the next iteration.
-        Index = STn;
-        break;
-      } else if (LoadSDNode *Ldn = dyn_cast<LoadSDNode>(NextInChain)) {
-        if (Ldn->isVolatile()) {
-          Index = nullptr;
-          break;
+  // check all parents of mergable children
+  for (auto P = CandidateParents.begin(); P != CandidateParents.end(); ++P)
+    for (auto I = (*P)->use_begin(), E = (*P)->use_end(); I != E; ++I)
+      if (I.getOperandNo() == 0)
+        if (StoreSDNode *OtherST = dyn_cast<StoreSDNode>(*I)) {
+          if (OtherST->isVolatile() || OtherST->isIndexed())
+            continue;
+          // We can merge constant floats to equivalent integers
+          if (OtherST->getMemoryVT() != MemVT)
+            if (!(MemVT.isInteger() && MemVT.bitsEq(OtherST->getMemoryVT()) &&
+                  isa<ConstantFPSDNode>(OtherST->getValue())))
+              continue;
+          BaseIndexOffset Ptr =
+              BaseIndexOffset::match(OtherST->getBasePtr(), DAG);
+          if (Ptr.equalBaseIndex(BasePtr) && CorrectValueKind(OtherST))
+            StoreNodes.push_back(MemOpLink(OtherST, Ptr.Offset));
         }
-
-        // Save the load node for later. Continue the scan.
-        AliasLoadNodes.push_back(Ldn);
-        NextInChain = Ldn->getChain().getNode();
-        continue;
-      } else {
-        Index = nullptr;
-        break;
-      }
-    }
-  }
 }
 
 // We need to check that merging these stores does not cause a loop
@@ -11978,8 +11975,7 @@ bool DAGCombiner::checkMergeStoreCandidatesForDependencies(
   return true;
 }
 
-bool DAGCombiner::MergeConsecutiveStores(
-    StoreSDNode* St, SmallVectorImpl<MemOpLink> &StoreNodes) {
+bool DAGCombiner::MergeConsecutiveStores(StoreSDNode *St) {
   if (OptLevel == CodeGenOpt::None)
     return false;
 
@@ -12013,145 +12009,136 @@ bool DAGCombiner::MergeConsecutiveStores(
   if (MemVT.isVector() && IsLoadSrc)
     return false;
 
-  // Only look at ends of store sequences.
-  SDValue Chain = SDValue(St, 0);
-  if (Chain->hasOneUse() && Chain->use_begin()->getOpcode() == ISD::STORE)
-    return false;
-
-  // Save the LoadSDNodes that we find in the chain.
-  // We need to make sure that these nodes do not interfere with
-  // any of the store nodes.
-  SmallVector<LSBaseSDNode*, 8> AliasLoadNodes;
-
-  getStoreMergeAndAliasCandidates(St, StoreNodes, AliasLoadNodes);
+  SmallVector<MemOpLink, 8> StoreNodes;
+  // Find potential store merge candidates by searching through chain sub-DAG
+  getStoreMergeCandidates(St, StoreNodes);
 
   // Check if there is anything to merge.
   if (StoreNodes.size() < 2)
     return false;
 
-  // only do dependence check in AA case
-  bool UseAA = CombinerAA.getNumOccurrences() > 0 ? CombinerAA
-                                                  : DAG.getSubtarget().useAA();
-  if (UseAA && !checkMergeStoreCandidatesForDependencies(StoreNodes))
+  // Check that we can merge these candidates without causing a cycle
+  if (!checkMergeStoreCandidatesForDependencies(StoreNodes))
     return false;
 
   // Sort the memory operands according to their distance from the
-  // base pointer.  As a secondary criteria: make sure stores coming
-  // later in the code come first in the list. This is important for
-  // the non-UseAA case, because we're merging stores into the FINAL
-  // store along a chain which potentially contains aliasing stores.
-  // Thus, if there are multiple stores to the same address, the last
-  // one can be considered for merging but not the others.
+  // base pointer.
   std::sort(StoreNodes.begin(), StoreNodes.end(),
             [](MemOpLink LHS, MemOpLink RHS) {
-    return LHS.OffsetFromBase < RHS.OffsetFromBase ||
-           (LHS.OffsetFromBase == RHS.OffsetFromBase &&
-            LHS.SequenceNum < RHS.SequenceNum);
-  });
+              return LHS.OffsetFromBase < RHS.OffsetFromBase;
+            });
 
   // Scan the memory operations on the chain and find the first non-consecutive
   // store memory address.
-  unsigned LastConsecutiveStore = 0;
+  unsigned NumConsecutiveStores = 0;
   int64_t StartAddress = StoreNodes[0].OffsetFromBase;
-  for (unsigned i = 0, e = StoreNodes.size(); i < e; ++i) {
-
-    // Check that the addresses are consecutive starting from the second
-    // element in the list of stores.
-    if (i > 0) {
-      int64_t CurrAddress = StoreNodes[i].OffsetFromBase;
-      if (CurrAddress - StartAddress != (ElementSizeBytes * i))
-        break;
-    }
 
-    // Check if this store interferes with any of the loads that we found.
-    // If we find a load that alias with this store. Stop the sequence.
-    if (any_of(AliasLoadNodes, [&](LSBaseSDNode *Ldn) {
-          return isAlias(Ldn, StoreNodes[i].MemNode);
-        }))
+  // Check that the addresses are consecutive starting from the second
+  // element in the list of stores.
+  for (unsigned i = 1, e = StoreNodes.size(); i < e; ++i) {
+    int64_t CurrAddress = StoreNodes[i].OffsetFromBase;
+    if (CurrAddress - StartAddress != (ElementSizeBytes * i))
       break;
-
-    // Mark this node as useful.
-    LastConsecutiveStore = i;
+    NumConsecutiveStores = i + 1;
   }
 
+  if (NumConsecutiveStores < 2)
+    return false;
+
   // The node with the lowest store address.
-  LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
-  unsigned FirstStoreAS = FirstInChain->getAddressSpace();
-  unsigned FirstStoreAlign = FirstInChain->getAlignment();
   LLVMContext &Context = *DAG.getContext();
   const DataLayout &DL = DAG.getDataLayout();
 
   // Store the constants into memory as one consecutive store.
   if (IsConstantSrc) {
-    unsigned LastLegalType = 0;
-    unsigned LastLegalVectorType = 0;
-    bool NonZero = false;
-    for (unsigned i=0; i<LastConsecutiveStore+1; ++i) {
-      StoreSDNode *St  = cast<StoreSDNode>(StoreNodes[i].MemNode);
-      SDValue StoredVal = St->getValue();
-
-      if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(StoredVal)) {
-        NonZero |= !C->isNullValue();
-      } else if (ConstantFPSDNode *C = dyn_cast<ConstantFPSDNode>(StoredVal)) {
-        NonZero |= !C->getConstantFPValue()->isNullValue();
-      } else {
-        // Non-constant.
-        break;
-      }
+    bool RV = false;
+    while (NumConsecutiveStores > 1) {
+      LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
+      unsigned FirstStoreAS = FirstInChain->getAddressSpace();
+      unsigned FirstStoreAlign = FirstInChain->getAlignment();
+      unsigned LastLegalType = 0;
+      unsigned LastLegalVectorType = 0;
+      bool NonZero = false;
+      for (unsigned i = 0; i < NumConsecutiveStores; ++i) {
+        StoreSDNode *ST = cast<StoreSDNode>(StoreNodes[i].MemNode);
+        SDValue StoredVal = ST->getValue();
+
+        if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(StoredVal)) {
+          NonZero |= !C->isNullValue();
+        } else if (ConstantFPSDNode *C =
+                       dyn_cast<ConstantFPSDNode>(StoredVal)) {
+          NonZero |= !C->getConstantFPValue()->isNullValue();
+        } else {
+          // Non-constant.
+          break;
+        }
 
-      // Find a legal type for the constant store.
-      unsigned SizeInBits = (i+1) * ElementSizeBytes * 8;
-      EVT StoreTy = EVT::getIntegerVT(Context, SizeInBits);
-      bool IsFast;
-      if (TLI.isTypeLegal(StoreTy) &&
-          TLI.allowsMemoryAccess(Context, DL, StoreTy, FirstStoreAS,
-                                 FirstStoreAlign, &IsFast) && IsFast) {
-        LastLegalType = i+1;
-      // Or check whether a truncstore is legal.
-      } else if (TLI.getTypeAction(Context, StoreTy) ==
-                 TargetLowering::TypePromoteInteger) {
-        EVT LegalizedStoredValueTy =
-          TLI.getTypeToTransformTo(Context, StoredVal.getValueType());
-        if (TLI.isTruncStoreLegal(LegalizedStoredValueTy, StoreTy) &&
-            TLI.allowsMemoryAccess(Context, DL, LegalizedStoredValueTy,
-                                   FirstStoreAS, FirstStoreAlign, &IsFast) &&
+        // Find a legal type for the constant store.
+        unsigned SizeInBits = (i + 1) * ElementSizeBytes * 8;
+        EVT StoreTy = EVT::getIntegerVT(Context, SizeInBits);
+        bool IsFast = false;
+        if (TLI.isTypeLegal(StoreTy) &&
+            TLI.allowsMemoryAccess(Context, DL, StoreTy, FirstStoreAS,
+                                   FirstStoreAlign, &IsFast) &&
             IsFast) {
           LastLegalType = i + 1;
+          // Or check whether a truncstore is legal.
+        } else if (TLI.getTypeAction(Context, StoreTy) ==
+                   TargetLowering::TypePromoteInteger) {
+          EVT LegalizedStoredValueTy =
+              TLI.getTypeToTransformTo(Context, StoredVal.getValueType());
+          if (TLI.isTruncStoreLegal(LegalizedStoredValueTy, StoreTy) &&
+              TLI.allowsMemoryAccess(Context, DL, LegalizedStoredValueTy,
+                                     FirstStoreAS, FirstStoreAlign, &IsFast) &&
+              IsFast) {
+            LastLegalType = i + 1;
+          }
         }
-      }
 
-      // We only use vectors if the constant is known to be zero or the target
-      // allows it and the function is not marked with the noimplicitfloat
-      // attribute.
-      if ((!NonZero || TLI.storeOfVectorConstantIsCheap(MemVT, i+1,
-                                                        FirstStoreAS)) &&
-          !NoVectors) {
-        // Find a legal type for the vector store.
-        EVT Ty = EVT::getVectorVT(Context, MemVT, i+1);
-        if (TLI.isTypeLegal(Ty) &&
-            TLI.allowsMemoryAccess(Context, DL, Ty, FirstStoreAS,
-                                   FirstStoreAlign, &IsFast) && IsFast)
-          LastLegalVectorType = i + 1;
+        // We only use vectors if the constant is known to be zero or the target
+        // allows it and the function is not marked with the noimplicitfloat
+        // attribute.
+        if ((!NonZero ||
+             TLI.storeOfVectorConstantIsCheap(MemVT, i + 1, FirstStoreAS)) &&
+            !NoVectors) {
+          // Find a legal type for the vector store.
+          EVT Ty = EVT::getVectorVT(Context, MemVT, i + 1);
+          if (TLI.isTypeLegal(Ty) && TLI.canMergeStoresTo(Ty) &&
+              TLI.allowsMemoryAccess(Context, DL, Ty, FirstStoreAS,
+                                     FirstStoreAlign, &IsFast) &&
+              IsFast)
+            LastLegalVectorType = i + 1;
+        }
       }
-    }
 
-    // Check if we found a legal integer type to store.
-    if (LastLegalType == 0 && LastLegalVectorType == 0)
-      return false;
+      // Check if we found a legal integer type that creates a meaningful merge.
+      if (LastLegalType < 2 && LastLegalVectorType < 2)
+        break;
 
-    bool UseVector = (LastLegalVectorType > LastLegalType) && !NoVectors;
-    unsigned NumElem = UseVector ? LastLegalVectorType : LastLegalType;
+      bool UseVector = (LastLegalVectorType > LastLegalType) && !NoVectors;
+      unsigned NumElem = (UseVector) ? LastLegalVectorType : LastLegalType;
 
-    return MergeStoresOfConstantsOrVecElts(StoreNodes, MemVT, NumElem,
-                                           true, UseVector);
+      bool Merged = MergeStoresOfConstantsOrVecElts(StoreNodes, MemVT, NumElem,
+                                                    true, UseVector);
+      if (!Merged)
+        break;
+      // Remove merged stores for next iteration.
+      StoreNodes.erase(StoreNodes.begin(), StoreNodes.begin() + NumElem);
+      RV = true;
+      NumConsecutiveStores -= NumElem;
+    }
+    return RV;
   }
 
   // When extracting multiple vector elements, try to store them
   // in one vector store rather than a sequence of scalar stores.
   if (IsExtractVecSrc) {
+    LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
+    unsigned FirstStoreAS = FirstInChain->getAddressSpace();
+    unsigned FirstStoreAlign = FirstInChain->getAlignment();
     unsigned NumStoresToMerge = 0;
     bool IsVec = MemVT.isVector();
-    for (unsigned i = 0; i < LastConsecutiveStore + 1; ++i) {
+    for (unsigned i = 0; i < NumConsecutiveStores; ++i) {
       StoreSDNode *St  = cast<StoreSDNode>(StoreNodes[i].MemNode);
       unsigned StoreValOpcode = St->getValue().getOpcode();
       // This restriction could be loosened.
@@ -12191,7 +12178,7 @@ bool DAGCombiner::MergeConsecutiveStores(
   // Find acceptable loads. Loads need to have the same chain (token factor),
   // must not be zext, volatile, indexed, and they must be consecutive.
   BaseIndexOffset LdBasePtr;
-  for (unsigned i=0; i<LastConsecutiveStore+1; ++i) {
+  for (unsigned i = 0; i < NumConsecutiveStores; ++i) {
     StoreSDNode *St  = cast<StoreSDNode>(StoreNodes[i].MemNode);
     LoadSDNode *Ld = dyn_cast<LoadSDNode>(St->getValue());
     if (!Ld) break;
@@ -12224,7 +12211,7 @@ bool DAGCombiner::MergeConsecutiveStores(
     }
 
     // We found a potential memory operand to merge.
-    LoadNodes.push_back(MemOpLink(Ld, LdPtr.Offset, 0));
+    LoadNodes.push_back(MemOpLink(Ld, LdPtr.Offset));
   }
 
   if (LoadNodes.size() < 2)
@@ -12236,7 +12223,9 @@ bool DAGCombiner::MergeConsecutiveStores(
   if (LoadNodes.size() == 2 && TLI.hasPairedLoad(MemVT, RequiredAlignment) &&
       St->getAlignment() >= RequiredAlignment)
     return false;
-
+  LSBaseSDNode *FirstInChain = StoreNodes[0].MemNode;
+  unsigned FirstStoreAS = FirstInChain->getAddressSpace();
+  unsigned FirstStoreAlign = FirstInChain->getAlignment();
   LoadSDNode *FirstLoad = cast<LoadSDNode>(LoadNodes[0].MemNode);
   unsigned FirstLoadAS = FirstLoad->getAddressSpace();
   unsigned FirstLoadAlign = FirstLoad->getAlignment();
@@ -12305,30 +12294,19 @@ bool DAGCombiner::MergeConsecutiveStores(
 
   // We add +1 here because the LastXXX variables refer to location while
   // the NumElem refers to array/index size.
-  unsigned NumElem = std::min(LastConsecutiveStore, LastConsecutiveLoad) + 1;
+  unsigned NumElem = std::min(NumConsecutiveStores, LastConsecutiveLoad + 1);
   NumElem = std::min(LastLegalType, NumElem);
 
   if (NumElem < 2)
     return false;
 
-  // Collect the chains from all merged stores.
+  // Collect the chains from all merged stores. Because the common case
+  // all chains are the same, check if we match the first Chain.
   SmallVector<SDValue, 8> MergeStoreChains;
   MergeStoreChains.push_back(StoreNodes[0].MemNode->getChain());
-
-  // The latest Node in the DAG.
-  unsigned LatestNodeUsed = 0;
-  for (unsigned i=1; i<NumElem; ++i) {
-    // Find a chain for the new wide-store operand. Notice that some
-    // of the store nodes that we found may not be selected for inclusion
-    // in the wide store. The chain we use needs to be the chain of the
-    // latest store node which is *used* and replaced by the wide store.
-    if (StoreNodes[i].SequenceNum < StoreNodes[LatestNodeUsed].SequenceNum)
-      LatestNodeUsed = i;
-
-    MergeStoreChains.push_back(StoreNodes[i].MemNode->getChain());
-  }
-
-  LSBaseSDNode *LatestOp = StoreNodes[LatestNodeUsed].MemNode;
+  for (unsigned i = 1; i < NumElem; ++i)
+    if (StoreNodes[0].MemNode->getChain() != StoreNodes[i].MemNode->getChain())
+      MergeStoreChains.push_back(StoreNodes[i].MemNode->getChain());
 
   // Find if it is better to use vectors or integers to load and store
   // to memory.
@@ -12352,6 +12330,8 @@ bool DAGCombiner::MergeConsecutiveStores(
   SDValue NewStoreChain =
     DAG.getNode(ISD::TokenFactor, StoreDL, MVT::Other, MergeStoreChains);
 
+  AddToWorklist(NewStoreChain.getNode());
+
   SDValue NewStore =
       DAG.getStore(NewStoreChain, StoreDL, NewLoad, FirstInChain->getBasePtr(),
                    FirstInChain->getPointerInfo(), FirstStoreAlign);
@@ -12363,25 +12343,9 @@ bool DAGCombiner::MergeConsecutiveStores(
                                   SDValue(NewLoad.getNode(), 1));
   }
 
-  if (UseAA) {
-    // Replace the all stores with the new store.
-    for (unsigned i = 0; i < NumElem; ++i)
-      CombineTo(StoreNodes[i].MemNode, NewStore);
-  } else {
-    // Replace the last store with the new store.
-    CombineTo(LatestOp, NewStore);
-    // Erase all other stores.
-    for (unsigned i = 0; i < NumElem; ++i) {
-      // Remove all Store nodes.
-      if (StoreNodes[i].MemNode == LatestOp)
-        continue;
-      StoreSDNode *St = cast<StoreSDNode>(StoreNodes[i].MemNode);
-      DAG.ReplaceAllUsesOfValueWith(SDValue(St, 0), St->getChain());
-      deleteAndRecombine(St);
-    }
-  }
-
-  StoreNodes.erase(StoreNodes.begin() + NumElem, StoreNodes.end());
+  // Replace the all stores with the new store.
+  for (unsigned i = 0; i < NumElem; ++i)
+    CombineTo(StoreNodes[i].MemNode, NewStore);
   return true;
 }
 
@@ -12538,19 +12502,7 @@ SDValue DAGCombiner::visitSTORE(SDNode *N) {
   if (SDValue NewST = TransformFPLoadStorePair(N))
     return NewST;
 
-  bool UseAA = CombinerAA.getNumOccurrences() > 0 ? CombinerAA
-                                                  : DAG.getSubtarget().useAA();
-#ifndef NDEBUG
-  if (CombinerAAOnlyFunc.getNumOccurrences() &&
-      CombinerAAOnlyFunc != DAG.getMachineFunction().getName())
-    UseAA = false;
-#endif
-  if (UseAA && ST->isUnindexed()) {
-    // FIXME: We should do this even without AA enabled. AA will just allow
-    // FindBetterChain to work in more situations. The problem with this is that
-    // any combine that expects memory operations to be on consecutive chains
-    // first needs to be updated to look for users of the same chain.
-
+  if (ST->isUnindexed()) {
     // Walk up chain skipping non-aliasing memory nodes, on this store and any
     // adjacent stores.
     if (findBetterNeighborChains(ST)) {
@@ -12584,8 +12536,13 @@ SDValue DAGCombiner::visitSTORE(SDNode *N) {
     if (SimplifyDemandedBits(
             Value,
             APInt::getLowBitsSet(Value.getScalarValueSizeInBits(),
-                                 ST->getMemoryVT().getScalarSizeInBits())))
+                                 ST->getMemoryVT().getScalarSizeInBits()))) {
+      // Re-visit the store if anything changed; SimplifyDemandedBits
+      // will add Value's node back to the worklist if necessary, but
+      // we also need to re-visit the Store node itself.
+      AddToWorklist(N);
       return SDValue(N, 0);
+    }
   }
 
   // If this is a load followed by a store to the same location, then the store
@@ -12629,15 +12586,12 @@ SDValue DAGCombiner::visitSTORE(SDNode *N) {
       // There can be multiple store sequences on the same chain.
       // Keep trying to merge store sequences until we are unable to do so
       // or until we merge the last store on the chain.
-      SmallVector<MemOpLink, 8> StoreNodes;
-      bool Changed = MergeConsecutiveStores(ST, StoreNodes);
+      bool Changed = MergeConsecutiveStores(ST);
       if (!Changed) break;
-
-      if (any_of(StoreNodes,
-                 [ST](const MemOpLink &Link) { return Link.MemNode == ST; })) {
-        // ST has been merged and no longer exists.
+      // Return N as merge only uses CombineTo and no worklist clean
+      // up is necessary.
+      if (N->getOpcode() == ISD::DELETED_NODE || !isa<StoreSDNode>(N))
         return SDValue(N, 0);
-      }
     }
   }
 
@@ -12646,7 +12600,7 @@ SDValue DAGCombiner::visitSTORE(SDNode *N) {
   // Make sure to do this only after attempting to merge stores in order to
   //  avoid changing the types of some subset of stores due to visit order,
   //  preventing their merging.
-  if (isa<ConstantFPSDNode>(Value)) {
+  if (isa<ConstantFPSDNode>(ST->getValue())) {
     if (SDValue NewSt = replaceStoreOfFPConstant(ST))
       return NewSt;
   }
@@ -13581,6 +13535,35 @@ SDValue DAGCombiner::visitBUILD_VECTOR(SDNode *N) {
   if (ISD::allOperandsUndef(N))
     return DAG.getUNDEF(VT);
 
+  // Check if we can express BUILD VECTOR via subvector extract.
+  if (!LegalTypes && (N->getNumOperands() > 1)) {
+    SDValue Op0 = N->getOperand(0);
+    auto checkElem = [&](SDValue Op) -> uint64_t {
+      if ((Op.getOpcode() == ISD::EXTRACT_VECTOR_ELT) &&
+          (Op0.getOperand(0) == Op.getOperand(0)))
+        if (auto CNode = dyn_cast<ConstantSDNode>(Op.getOperand(1)))
+          return CNode->getZExtValue();
+      return -1;
+    };
+
+    int Offset = checkElem(Op0);
+    for (unsigned i = 0; i < N->getNumOperands(); ++i) {
+      if (Offset + i != checkElem(N->getOperand(i))) {
+        Offset = -1;
+        break;
+      }
+    }
+
+    if ((Offset == 0) &&
+        (Op0.getOperand(0).getValueType() == N->getValueType(0)))
+      return Op0.getOperand(0);
+    if ((Offset != -1) &&
+        ((Offset % N->getValueType(0).getVectorNumElements()) ==
+         0)) // IDX must be multiple of output size.
+      return DAG.getNode(ISD::EXTRACT_SUBVECTOR, SDLoc(N), N->getValueType(0),
+                         Op0.getOperand(0), Op0.getOperand(1));
+  }
+
   if (SDValue V = reduceBuildVecExtToExtBuildVec(N))
     return V;
 
@@ -15741,6 +15724,12 @@ void DAGCombiner::GatherAllAliases(SDNode *N, SDValue OriginalChain,
       ++Depth;
       break;
 
+    case ISD::CopyFromReg:
+      // Forward past CopyFromReg.
+      Chains.push_back(Chain.getOperand(0));
+      ++Depth;
+      break;
+
     default:
       // For all other instructions we will just have to take what we can get.
       Aliases.push_back(Chain);
@@ -15769,6 +15758,18 @@ SDValue DAGCombiner::FindBetterChain(SDNode *N, SDValue OldChain) {
   return DAG.getNode(ISD::TokenFactor, SDLoc(N), MVT::Other, Aliases);
 }
 
+// This function tries to collect a bunch of potentially interesting
+// nodes to improve the chains of, all at once. This might seem
+// redundant, as this function gets called when visiting every store
+// node, so why not let the work be done on each store as it's visited?
+//
+// I believe this is mainly important because MergeConsecutiveStores
+// is unable to deal with merging stores of different sizes, so unless
+// we improve the chains of all the potential candidates up-front
+// before running MergeConsecutiveStores, it might only see some of
+// the nodes that will eventually be candidates, and then not be able
+// to go from a partially-merged state to the desired final
+// fully-merged state.
 bool DAGCombiner::findBetterNeighborChains(StoreSDNode *St) {
   // This holds the base pointer, index, and the offset in bytes from the base
   // pointer.
@@ -15804,10 +15805,8 @@ bool DAGCombiner::findBetterNeighborChains(StoreSDNode *St) {
     if (!Ptr.equalBaseIndex(BasePtr))
       break;
 
-    // Find the next memory operand in the chain. If the next operand in the
-    // chain is a store then move up and continue the scan with the next
-    // memory operand. If the next operand is a load save it and use alias
-    // information to check if it interferes with anything.
+    // Walk up the chain to find the next store node, ignoring any
+    // intermediate loads. Any other kind of node will halt the loop.
     SDNode *NextInChain = Index->getChain().getNode();
     while (true) {
       if (StoreSDNode *STn = dyn_cast<StoreSDNode>(NextInChain)) {
@@ -15826,9 +15825,14 @@ bool DAGCombiner::findBetterNeighborChains(StoreSDNode *St) {
         Index = nullptr;
         break;
       }
-    }
+    } // end while
   }
 
+  // At this point, ChainedStores lists all of the Store nodes
+  // reachable by iterating up through chain nodes matching the above
+  // conditions.  For each such store identified, try to find an
+  // earlier chain to attach the store to which won't violate the
+  // required ordering.
   bool MadeChangeToSt = false;
   SmallVector<std::pair<StoreSDNode *, SDValue>, 8> BetterChains;
 
diff --git a/llvm/lib/CodeGen/TargetLoweringBase.cpp b/llvm/lib/CodeGen/TargetLoweringBase.cpp
index f18cdca4861..9b8d48fd4d4 100644
--- a/llvm/lib/CodeGen/TargetLoweringBase.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringBase.cpp
@@ -851,7 +851,7 @@ TargetLoweringBase::TargetLoweringBase(const TargetMachine &tm) : TM(tm) {
   MinFunctionAlignment = 0;
   PrefFunctionAlignment = 0;
   PrefLoopAlignment = 0;
-  GatherAllAliasesMaxDepth = 6;
+  GatherAllAliasesMaxDepth = 18;
   MinStackArgumentAlignment = 1;
   // TODO: the default will be switched to 0 in the next commit, along
   // with the Target-specific changes necessary.
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 8fb9c5c4a38..5929daa2629 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -9254,7 +9254,7 @@ static SDValue performSTORECombine(SDNode *N,
   return SDValue();
 }
 
-  /// This function handles the log2-shuffle pattern produced by the
+/// This function handles the log2-shuffle pattern produced by the
 /// LoopVectorizer for the across vector reduction. It consists of
 /// log2(NumVectorElements) steps and, in each step, 2^(s) elements
 /// are reduced, where s is an induction variable from 0 to
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp b/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
index 3350654daad..d48eac4b0fe 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
@@ -451,16 +451,6 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(const TargetMachine &TM,
 
   PredictableSelectIsExpensive = false;
 
-  // We want to find all load dependencies for long chains of stores to enable
-  // merging into very wide vectors. The problem is with vectors with > 4
-  // elements. MergeConsecutiveStores will attempt to merge these because x8/x16
-  // vectors are a legal type, even though we have to split the loads
-  // usually. When we can more precisely specify load legality per address
-  // space, we should be able to make FindBetterChain/MergeConsecutiveStores
-  // smarter so that they can figure out what to do in 2 iterations without all
-  // N > 4 stores on the same chain.
-  GatherAllAliasesMaxDepth = 16;
-
   // FIXME: Need to really handle these.
   MaxStoresPerMemcpy  = 4096;
   MaxStoresPerMemmove = 4096;
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.h b/llvm/lib/Target/ARM/ARMISelLowering.h
index 84c6eb845bb..2b4cdf92ea9 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.h
+++ b/llvm/lib/Target/ARM/ARMISelLowering.h
@@ -500,6 +500,11 @@ class InstrItineraryData;
     bool canCombineStoreAndExtract(Type *VectorTy, Value *Idx,
                                    unsigned &Cost) const override;
 
+    bool canMergeStoresTo(EVT MemVT) const override {
+      // Do not merge to larger than i32.
+      return (MemVT.getSizeInBits() <= 32);
+    }
+
     bool isCheapToSpeculateCttz() const override;
     bool isCheapToSpeculateCtlz() const override;
 
diff --git a/llvm/test/CodeGen/AArch64/argument-blocks.ll b/llvm/test/CodeGen/AArch64/argument-blocks.ll
index 3169abc2dcb..b5374ca8ced 100644
--- a/llvm/test/CodeGen/AArch64/argument-blocks.ll
+++ b/llvm/test/CodeGen/AArch64/argument-blocks.ll
@@ -59,10 +59,10 @@ define i64 @test_hfa_ignores_gprs([7 x float], [2 x float] %in, i64, i64 %res) {
 }
 
 ; [2 x float] should not be promoted to double by the Darwin varargs handling,
-; but should go in an 8-byte aligned slot.
+; but should go in an 8-byte aligned slot and can be merged as integer stores.
 define void @test_varargs_stackalign() {
 ; CHECK-LABEL: test_varargs_stackalign:
-; CHECK-DARWINPCS: stp {{w[0-9]+}}, {{w[0-9]+}}, [sp, #16]
+; CHECK-DARWINPCS: str {{x[0-9]+}}, [sp, #16]
 
   call void(...) @callee([3 x float] undef, [2 x float] [float 1.0, float 2.0])
   ret void
diff --git a/llvm/test/CodeGen/AArch64/arm64-abi.ll b/llvm/test/CodeGen/AArch64/arm64-abi.ll
index fb52b1d99fc..6cf0ab35b9b 100644
--- a/llvm/test/CodeGen/AArch64/arm64-abi.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-abi.ll
@@ -205,10 +205,7 @@ declare i32 @args_i32(i32, i32, i32, i32, i32, i32, i32, i32, i16 signext, i32,
 define i32 @test8(i32 %argc, i8** nocapture %argv) nounwind {
 entry:
 ; CHECK-LABEL: test8
-; CHECK: strb {{w[0-9]+}}, [sp, #3]
-; CHECK: strb wzr, [sp, #2]
-; CHECK: strb {{w[0-9]+}}, [sp, #1]
-; CHECK: strb wzr, [sp]
+; CHECK: str w8, [sp]
 ; CHECK: bl
 ; FAST-LABEL: test8
 ; FAST: strb {{w[0-9]+}}, [sp]
diff --git a/llvm/test/CodeGen/AArch64/arm64-memset-inline.ll b/llvm/test/CodeGen/AArch64/arm64-memset-inline.ll
index 8f22f97ca08..384aaa8541d 100644
--- a/llvm/test/CodeGen/AArch64/arm64-memset-inline.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-memset-inline.ll
@@ -13,8 +13,8 @@ define void @t2() nounwind ssp {
 entry:
 ; CHECK-LABEL: t2:
 ; CHECK: strh wzr, [sp, #32]
-; CHECK: stp xzr, xzr, [sp, #16]
-; CHECK: str xzr, [sp, #8]
+; CHECK: stp xzr, xzr, [sp, #8]
+; CHECK: str xzr, [sp, #24]
   %buf = alloca [26 x i8], align 1
   %0 = getelementptr inbounds [26 x i8], [26 x i8]* %buf, i32 0, i32 0
   call void @llvm.memset.p0i8.i32(i8* %0, i8 0, i32 26, i32 1, i1 false)
diff --git a/llvm/test/CodeGen/AArch64/arm64-variadic-aapcs.ll b/llvm/test/CodeGen/AArch64/arm64-variadic-aapcs.ll
index 16ddf690fe9..375877c5179 100644
--- a/llvm/test/CodeGen/AArch64/arm64-variadic-aapcs.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-variadic-aapcs.ll
@@ -99,7 +99,7 @@ define void @test_nospare([8 x i64], [8 x float], ...) {
 ; __stack field should point just past them.
 define void @test_offsetstack([8 x i64], [2 x i64], [3 x float], ...) {
 ; CHECK-LABEL: test_offsetstack:
-; CHECK: sub sp, sp, #80
+; CHECK: stp {{q[0-9]+}}, {{q[0-9]+}}, [sp, #-80]!
 ; CHECK: add [[STACK_TOP:x[0-9]+]], sp, #96
 ; CHECK: add x[[VAR:[0-9]+]], {{x[0-9]+}}, :lo12:var
 ; CHECK: str [[STACK_TOP]], [x[[VAR]]]
diff --git a/llvm/test/CodeGen/AArch64/merge-store.ll b/llvm/test/CodeGen/AArch64/merge-store.ll
index 1d0196ad521..1d26e4a42b1 100644
--- a/llvm/test/CodeGen/AArch64/merge-store.ll
+++ b/llvm/test/CodeGen/AArch64/merge-store.ll
@@ -4,8 +4,7 @@
 @g0 = external global <3 x float>, align 16
 @g1 = external global <3 x float>, align 4
 
-; CHECK: ldr s[[R0:[0-9]+]], {{\[}}[[R1:x[0-9]+]]{{\]}}, #4
-; CHECK: ld1{{\.?s?}} { v[[R0]]{{\.?s?}} }[1], {{\[}}[[R1]]{{\]}}
+; CHECK: ldr q[[R0:[0-9]+]], {{\[}}[[R1:x[0-9]+]], :lo12:g0
 ; CHECK: str d[[R0]]
 
 define void @blam() {
diff --git a/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll b/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll
index 9220947e836..e4e64ef8c8d 100644
--- a/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll
+++ b/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll
@@ -1,5 +1,4 @@
-; RUN: llc --combiner-alias-analysis=false < %s | FileCheck %s
-; RUN: llc --combiner-alias-analysis=true  < %s | FileCheck %s
+; RUN: llc < %s | FileCheck %s
 
 ; This test checks that we do not merge stores together which have
 ; dependencies through their non-chain operands (e.g. one store is the
diff --git a/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll b/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll
index 6638f4e2582..7be7d9486a4 100644
--- a/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll
+++ b/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll
@@ -1,13 +1,21 @@
-; RUN: llc -O0 -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=+amdgpu-debugger-insert-nops -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -O0 -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=+amdgpu-debugger-insert-nops -verify-machineinstrs < %s | FileCheck %s --check-prefix=CHECK
+; RUN: llc -O0 -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=+amdgpu-debugger-insert-nops -verify-machineinstrs < %s | FileCheck %s --check-prefix=CHECKNOP
 
-; CHECK: test01.cl:2:{{[0-9]+}}
-; CHECK-NEXT: s_nop 0
+; This test expects that we have one instance for each line in some order with "s_nop 0" instances after each.
 
-; CHECK: test01.cl:3:{{[0-9]+}}
-; CHECK-NEXT: s_nop 0
+; Check that each line appears at least once
+; CHECK-DAG: test01.cl:2:3
+; CHECK-DAG: test01.cl:3:3
+; CHECK-DAG: test01.cl:4:3
 
-; CHECK: test01.cl:4:{{[0-9]+}}
-; CHECK-NEXT: s_nop 0
+
+; Check that each of each of the lines consists of the line output, followed by "s_nop 0"
+; CHECKNOP: test01.cl:{{[234]}}:3
+; CHECKNOP-NEXT: s_nop 0
+; CHECKNOP: test01.cl:{{[234]}}:3
+; CHECKNOP-NEXT: s_nop 0
+; CHECKNOP: test01.cl:{{[234]}}:3
+; CHECKNOP-NEXT: s_nop 0
 
 ; CHECK: test01.cl:5:{{[0-9]+}}
 ; CHECK-NEXT: s_nop 0
@@ -21,7 +29,7 @@ entry:
   call void @llvm.dbg.declare(metadata i32 addrspace(1)** %A.addr, metadata !17, metadata !18), !dbg !19
   %0 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4, !dbg !20
   %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %0, i32 0, !dbg !20
-  store i32 1, i32 addrspace(1)* %arrayidx, align 4, !dbg !21
+  store i32 1, i32 addrspace(1)* %arrayidx, align 4, !dbg !20
   %1 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4, !dbg !22
   %arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %1, i32 1, !dbg !22
   store i32 2, i32 addrspace(1)* %arrayidx1, align 4, !dbg !23
diff --git a/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll b/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
index 65ac693a4f4..08acee0d4be 100644
--- a/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
+++ b/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
@@ -257,9 +257,8 @@ define void @dynamic_insertelement_v2i8(<2 x i8> addrspace(1)* %out, <2 x i8> %a
 
 ; GCN-DAG: buffer_store_byte v{{[0-9]+}}, off, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offset:2
 ; GCN-DAG: buffer_store_byte v{{[0-9]+}}, off, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offset:1
-; GCN-DAG: buffer_store_byte v{{[0-9]+}}, off, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}}{{$}}
-
-; GCN: buffer_store_byte v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offen{{$}}
+; GCN-DAG: buffer_store_byte v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offen{{$}}
+; GCN-DAG: buffer_store_byte v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offen{{$}}
 
 ; GCN: buffer_load_ubyte
 ; GCN: buffer_load_ubyte
diff --git a/llvm/test/CodeGen/AMDGPU/merge-stores.ll b/llvm/test/CodeGen/AMDGPU/merge-stores.ll
index 07104ebc8c9..fd2d8ee1189 100644
--- a/llvm/test/CodeGen/AMDGPU/merge-stores.ll
+++ b/llvm/test/CodeGen/AMDGPU/merge-stores.ll
@@ -1,8 +1,5 @@
-; RUN: llc -march=amdgcn -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-NOAA %s
-; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-NOAA %s
-
-; RUN: llc -march=amdgcn -verify-machineinstrs -combiner-alias-analysis -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
-; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -combiner-alias-analysis -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
+; RUN: llc -march=amdgcn -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
+; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
 
 ; This test is mostly to test DAG store merging, so disable the vectorizer.
 ; Run with devices with different unaligned load restrictions.
@@ -150,12 +147,7 @@ define void @merge_global_store_4_constants_f32(float addrspace(1)* %out) #0 {
 }
 
 ; GCN-LABEL: {{^}}merge_global_store_4_constants_mixed_i32_f32:
-; GCN-NOAA: buffer_store_dwordx4 v
-
-; GCN-AA: buffer_store_dwordx2
-; GCN-AA: buffer_store_dword v
-; GCN-AA: buffer_store_dword v
-
+; GCN-AA: buffer_store_dwordx4 v
 ; GCN: s_endpgm
 define void @merge_global_store_4_constants_mixed_i32_f32(float addrspace(1)* %out) #0 {
   %out.gep.1 = getelementptr float, float addrspace(1)* %out, i32 1
@@ -474,17 +466,9 @@ define void @merge_global_store_4_adjacent_loads_i8_natural_align(i8 addrspace(1
   ret void
 }
 
-; This works once AA is enabled on the subtarget
 ; GCN-LABEL: {{^}}merge_global_store_4_vector_elts_loads_v4i32:
 ; GCN: buffer_load_dwordx4 [[LOAD:v\[[0-9]+:[0-9]+\]]]
-
-; GCN-NOAA: buffer_store_dword v
-; GCN-NOAA: buffer_store_dword v
-; GCN-NOAA: buffer_store_dword v
-; GCN-NOAA: buffer_store_dword v
-
-; GCN-AA: buffer_store_dwordx4 [[LOAD]]
-
+; GCN: buffer_store_dwordx4 [[LOAD]]
 ; GCN: s_endpgm
 define void @merge_global_store_4_vector_elts_loads_v4i32(i32 addrspace(1)* %out, <4 x i32> addrspace(1)* %in) #0 {
   %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
diff --git a/llvm/test/CodeGen/AMDGPU/private-element-size.ll b/llvm/test/CodeGen/AMDGPU/private-element-size.ll
index de9a8f75512..8bbb72fcda4 100644
--- a/llvm/test/CodeGen/AMDGPU/private-element-size.ll
+++ b/llvm/test/CodeGen/AMDGPU/private-element-size.ll
@@ -32,10 +32,10 @@
 ; HSA-ELT4-DAG: buffer_store_dword {{v[0-9]+}}, off, s[0:3], s9 offset:24{{$}}
 ; HSA-ELT4-DAG: buffer_store_dword {{v[0-9]+}}, off, s[0:3], s9 offset:28{{$}}
 
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen{{$}}
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:4{{$}}
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:8{{$}}
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:12{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:4{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:8{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:12{{$}}
 define void @private_elt_size_v4i32(<4 x i32> addrspace(1)* %out, i32 addrspace(1)* %index.array) #0 {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
@@ -130,8 +130,8 @@ entry:
 ; HSA-ELT8: private_element_size = 2
 ; HSA-ELT4: private_element_size = 1
 
-; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, off, s[0:3], s9{{$}}
-; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, off, s[0:3], s9 offset:8
+; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, {{off|v[0-9]}}, s[0:3], s9{{$}}
+; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, {{off|v[0-9]}}, s[0:3], s9 offset:8
 
 ; HSA-ELTGE8: buffer_load_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}}, s[0:3], s9 offen
 
diff --git a/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll b/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
index 4beefb047f2..c1d691fcff8 100644
--- a/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
+++ b/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
@@ -157,9 +157,8 @@ define void @reorder_global_load_local_store_global_load(i32 addrspace(1)* %out,
 
 ; FUNC-LABEL: @reorder_local_offsets
 ; CI: ds_read2_b32 {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}} offset0:100 offset1:102
-; CI: ds_write2_b32 {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}} offset0:3 offset1:100
-; CI: ds_read_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:12
-; CI: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:408
+; CI-DAG: ds_write2_b32 {{v[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset0:3 offset1:100
+; CI-DAG: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:408
 ; CI: buffer_store_dword
 ; CI: s_endpgm
 define void @reorder_local_offsets(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* noalias nocapture readnone %gptr, i32 addrspace(3)* noalias nocapture %ptr0) #0 {
@@ -181,12 +180,12 @@ define void @reorder_local_offsets(i32 addrspace(1)* nocapture %out, i32 addrspa
 }
 
 ; FUNC-LABEL: @reorder_global_offsets
-; CI: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
-; CI: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
-; CI: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
-; CI: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
-; CI: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
-; CI: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
+; CI-DAG: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
+; CI-DAG: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
+; CI-DAG: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
+; CI-DAG: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
+; CI-DAG: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
+; CI: buffer_store_dword
 ; CI: s_endpgm
 define void @reorder_global_offsets(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* noalias nocapture readnone %gptr, i32 addrspace(1)* noalias nocapture %ptr0) #0 {
   %ptr1 = getelementptr inbounds i32, i32 addrspace(1)* %ptr0, i32 3
diff --git a/llvm/test/CodeGen/ARM/2012-10-04-AAPCS-byval-align8.ll b/llvm/test/CodeGen/ARM/2012-10-04-AAPCS-byval-align8.ll
index 4a1341c4d6e..d2ff09a6200 100644
--- a/llvm/test/CodeGen/ARM/2012-10-04-AAPCS-byval-align8.ll
+++ b/llvm/test/CodeGen/ARM/2012-10-04-AAPCS-byval-align8.ll
@@ -12,7 +12,8 @@ define void @test_byval_8_bytes_alignment(i32 %i, ...) {
 entry:
 ; CHECK: sub       sp, sp, #12
 ; CHECK: sub       sp, sp, #4
-; CHECK: stmib     sp, {r1, r2, r3}
+; CHECK: add       r0, sp, #4
+; CHECK: stm       sp, {r0, r1, r2, r3}
   %g = alloca i8*
   %g1 = bitcast i8** %g to i8*
   call void @llvm.va_start(i8* %g1)
diff --git a/llvm/test/CodeGen/ARM/alloc-no-stack-realign.ll b/llvm/test/CodeGen/ARM/alloc-no-stack-realign.ll
index 7d37c83d748..0e077b3aee5 100644
--- a/llvm/test/CodeGen/ARM/alloc-no-stack-realign.ll
+++ b/llvm/test/CodeGen/ARM/alloc-no-stack-realign.ll
@@ -1,5 +1,4 @@
-; RUN: llc < %s -mtriple=armv7-apple-ios -O0 | FileCheck %s -check-prefix=NO-REALIGN
-; RUN: llc < %s -mtriple=armv7-apple-ios -O0 | FileCheck %s -check-prefix=REALIGN
+; RUN: llc < %s -mtriple=armv7-apple-ios -O0 | FileCheck %s
 
 ; rdar://12713765
 ; When realign-stack is set to false, make sure we are not creating stack
@@ -8,29 +7,31 @@
 
 define void @test1(<16 x float>* noalias sret %agg.result) nounwind ssp "no-realign-stack" {
 entry:
-; NO-REALIGN-LABEL: test1
-; NO-REALIGN: mov r[[R2:[0-9]+]], r[[R1:[0-9]+]]
-; NO-REALIGN: vld1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]!
-; NO-REALIGN: vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; NO-REALIGN: add r[[R2:[0-9]+]], r[[R1]], #32
-; NO-REALIGN: vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; NO-REALIGN: add r[[R2:[0-9]+]], r[[R1]], #48
-; NO-REALIGN: vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-
-; NO-REALIGN: add r[[R2:[0-9]+]], r[[R1:[0-9]+]], #48
-; NO-REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; NO-REALIGN: add r[[R2:[0-9]+]], r[[R1]], #32
-; NO-REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; NO-REALIGN: mov	r[[R3:[0-9]+]], r[[R1]]
-; NO-REALIGN: vst1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R3]]:128]!
-; NO-REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R3]]:128]
-
-; NO-REALIGN: add r[[R2:[0-9]+]], r[[R0:0]], #48
-; NO-REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; NO-REALIGN: add r[[R2:[0-9]+]], r[[R0]], #32
-; NO-REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; NO-REALIGN: vst1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R0]]:128]!
-; NO-REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R0]]:128]
+; CHECK-LABEL: test1
+; CHECK:	ldr	r[[R1:[0-9]+]], [pc, r1]
+; CHECK:	add	r[[R2:[0-9]+]], r1, #48
+; CHECK:	vld1.64	 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	mov	r[[R2:[0-9]+]], r[[R1]]
+; CHECK:	vld1.32	 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]!
+; CHECK:	vld1.64	 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	add	r[[R1:[0-9]+]], r[[R1]], #32
+; CHECK:	vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	mov	r[[R1:[0-9]+]], sp
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	add	r[[R2:[0-9]+]], r[[R1]], #32
+; CHECK:	vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	vld1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]!
+; CHECK:	vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	vld1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]!
+; CHECK:	vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	add	r[[R1:[0-9]+]], r0, #48
+; CHECK:	vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	add	r[[R1:[0-9]+]], r0, #32
+; CHECK:	vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	vst1.32 {{{d[0-9]+, d[0-9]+}}}, [r0:128]!
+; CHECK:	vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r0:128]
  %retval = alloca <16 x float>, align 16
  %0 = load <16 x float>, <16 x float>* @T3_retval, align 16
  store <16 x float> %0, <16 x float>* %retval
@@ -41,32 +42,33 @@ entry:
 
 define void @test2(<16 x float>* noalias sret %agg.result) nounwind ssp {
 entry:
-; REALIGN-LABEL: test2
-; REALIGN: bfc sp, #0, #6
-; REALIGN: mov r[[R2:[0-9]+]], r[[R1:[0-9]+]]
-; REALIGN: vld1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]!
-; REALIGN: vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; REALIGN: add r[[R2:[0-9]+]], r[[R1]], #32
-; REALIGN: vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; REALIGN: add r[[R2:[0-9]+]], r[[R1]], #48
-; REALIGN: vld1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	ldr	r[[R1:[0-9]+]], [pc, r1]
+; CHECK:	add	r[[R2:[0-9]+]], r[[R1]], #48
+; CHECK:	vld1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	mov	r[[R2:[0-9]+]], r[[R1]]
+; CHECK:	vld1.32	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]!
+; CHECK:	vld1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	add	r[[R1:[0-9]+]], r[[R1]], #32
+; CHECK:	vld1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	mov	r[[R1:[0-9]+]], sp
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	orr	r[[R2:[0-9]+]], r[[R1]], #32
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	vld1.32	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]!
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	vld1.32	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]!
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	vld1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	vld1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
+; CHECK:	add	r[[R1:[0-9]+]], r0, #48
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	add	r[[R1:[0-9]+]], r0, #32
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
+; CHECK:	vst1.32	{{{d[0-9]+, d[0-9]+}}}, [r0:128]!
+; CHECK:	vst1.64	{{{d[0-9]+, d[0-9]+}}}, [r0:128]
 
 
-; REALIGN: orr r[[R2:[0-9]+]], r[[R1:[0-9]+]], #48
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; REALIGN: orr r[[R2:[0-9]+]], r[[R1]], #32
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; REALIGN: orr r[[R2:[0-9]+]], r[[R1]], #16
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R2]]:128]
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
-
-; REALIGN: add r[[R1:[0-9]+]], r[[R0:0]], #48
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
-; REALIGN: add r[[R1:[0-9]+]], r[[R0]], #32
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R1]]:128]
-; REALIGN: vst1.32 {{{d[0-9]+, d[0-9]+}}}, [r[[R0]]:128]!
-; REALIGN: vst1.64 {{{d[0-9]+, d[0-9]+}}}, [r[[R0]]:128]
- %retval = alloca <16 x float>, align 16
+%retval = alloca <16 x float>, align 16
  %0 = load <16 x float>, <16 x float>* @T3_retval, align 16
  store <16 x float> %0, <16 x float>* %retval
  %1 = load <16 x float>, <16 x float>* %retval
diff --git a/llvm/test/CodeGen/ARM/gpr-paired-spill.ll b/llvm/test/CodeGen/ARM/gpr-paired-spill.ll
index ef3e5a54a2d..797b147d5d0 100644
--- a/llvm/test/CodeGen/ARM/gpr-paired-spill.ll
+++ b/llvm/test/CodeGen/ARM/gpr-paired-spill.ll
@@ -16,22 +16,22 @@ define void @foo(i64* %addr) {
   ; an LDMIA was created with both a FrameIndex and an offset, which
   ; is not allowed.
 
-; CHECK-WITH-LDRD: strd {{r[0-9]+}}, {{r[0-9]+}}, [sp, #8]
-; CHECK-WITH-LDRD: strd {{r[0-9]+}}, {{r[0-9]+}}, [sp]
+; CHECK-WITH-LDRD-DAG: strd {{r[0-9]+}}, {{r[0-9]+}}, [sp, #8]
+; CHECK-WITH-LDRD-DAG: strd {{r[0-9]+}}, {{r[0-9]+}}, [sp]
 
-; CHECK-WITH-LDRD: ldrd {{r[0-9]+}}, {{r[0-9]+}}, [sp, #8]
-; CHECK-WITH-LDRD: ldrd {{r[0-9]+}}, {{r[0-9]+}}, [sp]
+; CHECK-WITH-LDRD-DAG: ldrd {{r[0-9]+}}, {{r[0-9]+}}, [sp, #8]
+; CHECK-WITH-LDRD-DAG: ldrd {{r[0-9]+}}, {{r[0-9]+}}, [sp]
 
   ; We also want to ensure the register scavenger is working (i.e. an
   ; offset from sp can be generated), so we need two spills.
-; CHECK-WITHOUT-LDRD: add [[ADDRREG:[a-z0-9]+]], sp, #{{[0-9]+}}
-; CHECK-WITHOUT-LDRD: stm [[ADDRREG]], {r{{[0-9]+}}, r{{[0-9]+}}}
-; CHECK-WITHOUT-LDRD: stm sp, {r{{[0-9]+}}, r{{[0-9]+}}}
+; CHECK-WITHOUT-LDRD-DAG: add [[ADDRREG:[a-z0-9]+]], sp, #{{[0-9]+}}
+; CHECK-WITHOUT-LDRD-DAG: stm [[ADDRREG]], {r{{[0-9]+}}, r{{[0-9]+}}}
+; CHECK-WITHOUT-LDRD-DAG: stm sp, {r{{[0-9]+}}, r{{[0-9]+}}}
 
   ; In principle LLVM may have to recalculate the offset. At the moment
   ; it reuses the original though.
-; CHECK-WITHOUT-LDRD: ldm [[ADDRREG]], {r{{[0-9]+}}, r{{[0-9]+}}}
-; CHECK-WITHOUT-LDRD: ldm sp, {r{{[0-9]+}}, r{{[0-9]+}}}
+; CHECK-WITHOUT-LDRD-DAG: ldm [[ADDRREG]], {r{{[0-9]+}}, r{{[0-9]+}}}
+; CHECK-WITHOUT-LDRD-DAG: ldm sp, {r{{[0-9]+}}, r{{[0-9]+}}}
 
   store volatile i64 %val1, i64* %addr
   store volatile i64 %val2, i64* %addr
diff --git a/llvm/test/CodeGen/ARM/ifcvt10.ll b/llvm/test/CodeGen/ARM/ifcvt10.ll
index 5725a404c32..c7e18d35dbe 100644
--- a/llvm/test/CodeGen/ARM/ifcvt10.ll
+++ b/llvm/test/CodeGen/ARM/ifcvt10.ll
@@ -10,8 +10,6 @@ entry:
 ; CHECK: vpop {d8}
 ; CHECK-NOT: vpopne
 ; CHECK: pop {r7, pc}
-; CHECK: vpop {d8}
-; CHECK: pop {r7, pc}
   br i1 undef, label %if.else, label %if.then
 
 if.then:                                          ; preds = %entry
diff --git a/llvm/test/CodeGen/ARM/static-addr-hoisting.ll b/llvm/test/CodeGen/ARM/static-addr-hoisting.ll
index 3d47e02f965..683d607936b 100644
--- a/llvm/test/CodeGen/ARM/static-addr-hoisting.ll
+++ b/llvm/test/CodeGen/ARM/static-addr-hoisting.ll
@@ -6,9 +6,9 @@ define void @multiple_store() {
 ; CHECK: movs [[VAL:r[0-9]+]], #42
 ; CHECK: movt r[[BASE1]], #15
 
-; CHECK: str [[VAL]], [r[[BASE1]]]
-; CHECK: str [[VAL]], [r[[BASE1]], #24]
-; CHECK: str.w [[VAL]], [r[[BASE1]], #42]
+; CHECK-DAG: str [[VAL]], [r[[BASE1]]]
+; CHECK-DAG: str [[VAL]], [r[[BASE1]], #24]
+; CHECK-DAG: str.w [[VAL]], [r[[BASE1]], #42]
 
 ; CHECK: movw r[[BASE2:[0-9]+]], #20394
 ; CHECK: movt r[[BASE2]], #18
diff --git a/llvm/test/CodeGen/BPF/undef.ll b/llvm/test/CodeGen/BPF/undef.ll
index 541d81ea07b..de14bfde1ab 100644
--- a/llvm/test/CodeGen/BPF/undef.ll
+++ b/llvm/test/CodeGen/BPF/undef.ll
@@ -13,50 +13,55 @@
 
 ; Function Attrs: nounwind uwtable
 define i32 @ebpf_filter(%struct.__sk_buff* nocapture readnone %ebpf_packet) #0 section "socket1" {
+; CHECK: r2 = r10
+; CHECK: r2 += -2
+; CHECK: r1 = 0
+; CHECK: *(u16 *)(r2 + 6) = r1
+; CHECK: *(u16 *)(r2 + 4) = r1
+; CHECK: *(u16 *)(r2 + 2) = r1
+; CHECK: r2 = 6
+; CHECK: *(u8 *)(r10 - 7) = r2
+; CHECK: r2 = 5
+; CHECK: *(u8 *)(r10 - 8) = r2
+; CHECK: r2 = 7
+; CHECK: *(u8 *)(r10 - 6) = r2
+; CHECK: r2 = 8
+; CHECK: *(u8 *)(r10 - 5) = r2
+; CHECK: r2 = 9
+; CHECK: *(u8 *)(r10 - 4) = r2
+; CHECK: r2 = 10
+; CHECK: *(u8 *)(r10 - 3) = r2
+; CHECK: *(u16 *)(r10 + 24) = r1
+; CHECK: *(u16 *)(r10 + 22) = r1
+; CHECK: *(u16 *)(r10 + 20) = r1
+; CHECK: *(u16 *)(r10 + 18) = r1
+; CHECK: *(u16 *)(r10 + 16) = r1
+; CHECK: *(u16 *)(r10 + 14) = r1
+; CHECK: *(u16 *)(r10 + 12) = r1
+; CHECK: *(u16 *)(r10 + 10) = r1
+; CHECK: *(u16 *)(r10 + 8) = r1
+; CHECK: *(u16 *)(r10 + 6) = r1
+; CHECK: *(u16 *)(r10 - 2) = r1
+; CHECK: *(u16 *)(r10 + 26) = r1
+; CHECK: r2 = r10
+; CHECK: r2 += -8
+; CHECK: r1 = <MCOperand Expr:(routing)>ll
+; CHECK: call bpf_map_lookup_elem
+; CHECK: exit
   %key = alloca %struct.routing_key_2, align 1
   %1 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 0, i32 0, i64 0
-; CHECK: r1 = 5
-; CHECK: *(u8 *)(r10 - 8) = r1
   store i8 5, i8* %1, align 1
   %2 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 0, i32 0, i64 1
-; CHECK: r1 = 6
-; CHECK: *(u8 *)(r10 - 7) = r1
   store i8 6, i8* %2, align 1
   %3 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 0, i32 0, i64 2
-; CHECK: r1 = 7
-; CHECK: *(u8 *)(r10 - 6) = r1
   store i8 7, i8* %3, align 1
   %4 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 0, i32 0, i64 3
-; CHECK: r1 = 8
-; CHECK: *(u8 *)(r10 - 5) = r1
   store i8 8, i8* %4, align 1
   %5 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 0, i32 0, i64 4
-; CHECK: r1 = 9
-; CHECK: *(u8 *)(r10 - 4) = r1
   store i8 9, i8* %5, align 1
   %6 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 0, i32 0, i64 5
-; CHECK: r1 = 10
-; CHECK: *(u8 *)(r10 - 3) = r1
   store i8 10, i8* %6, align 1
   %7 = getelementptr inbounds %struct.routing_key_2, %struct.routing_key_2* %key, i64 1, i32 0, i64 0
-; CHECK: r1 = r10
-; CHECK: r1 += -2
-; CHECK: r2 = 0
-; CHECK: *(u16 *)(r1 + 6) = r2
-; CHECK: *(u16 *)(r1 + 4) = r2
-; CHECK: *(u16 *)(r1 + 2) = r2
-; CHECK: *(u16 *)(r10 + 24) = r2
-; CHECK: *(u16 *)(r10 + 22) = r2
-; CHECK: *(u16 *)(r10 + 20) = r2
-; CHECK: *(u16 *)(r10 + 18) = r2
-; CHECK: *(u16 *)(r10 + 16) = r2
-; CHECK: *(u16 *)(r10 + 14) = r2
-; CHECK: *(u16 *)(r10 + 12) = r2
-; CHECK: *(u16 *)(r10 + 10) = r2
-; CHECK: *(u16 *)(r10 + 8) = r2
-; CHECK: *(u16 *)(r10 + 6) = r2
-; CHECK: *(u16 *)(r10 - 2) = r2
-; CHECK: *(u16 *)(r10 + 26) = r2
   call void @llvm.memset.p0i8.i64(i8* %7, i8 0, i64 30, i32 1, i1 false)
   %8 = call i32 (%struct.bpf_map_def*, %struct.routing_key_2*, ...) bitcast (i32 (...)* @bpf_map_lookup_elem to i32 (%struct.bpf_map_def*, %struct.routing_key_2*, ...)*)(%struct.bpf_map_def* nonnull @routing, %struct.routing_key_2* nonnull %key) #3
   ret i32 undef
diff --git a/llvm/test/CodeGen/MSP430/Inst16mm.ll b/llvm/test/CodeGen/MSP430/Inst16mm.ll
index c75e1beb235..a48d8592c1a 100644
--- a/llvm/test/CodeGen/MSP430/Inst16mm.ll
+++ b/llvm/test/CodeGen/MSP430/Inst16mm.ll
@@ -1,4 +1,4 @@
-; RUN: llc -march=msp430 -combiner-alias-analysis < %s | FileCheck %s
+; RUN: llc -march=msp430 < %s | FileCheck %s
 target datalayout = "e-p:16:8:8-i8:8:8-i16:8:8-i32:8:8"
 target triple = "msp430-generic-generic"
 @foo = common global i16 0, align 2
diff --git a/llvm/test/CodeGen/Mips/cconv/arguments-float.ll b/llvm/test/CodeGen/Mips/cconv/arguments-float.ll
index 7d32992ecb1..004f6d94749 100644
--- a/llvm/test/CodeGen/Mips/cconv/arguments-float.ll
+++ b/llvm/test/CodeGen/Mips/cconv/arguments-float.ll
@@ -63,39 +63,39 @@ entry:
 ; NEW-DAG:           sd $5, 16([[R2]])
 
 ; O32 has run out of argument registers and starts using the stack
-; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 24($sp)
-; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 28($sp)
+; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 16($sp)
+; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 20($sp)
 ; O32-DAG:           sw [[R3]], 24([[R2]])
 ; O32-DAG:           sw [[R4]], 28([[R2]])
 ; NEW-DAG:           sd $6, 24([[R2]])
 
-; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 32($sp)
-; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 36($sp)
+; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 24($sp)
+; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 28($sp)
 ; O32-DAG:           sw [[R3]], 32([[R2]])
 ; O32-DAG:           sw [[R4]], 36([[R2]])
 ; NEW-DAG:           sd $7, 32([[R2]])
 
-; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 40($sp)
-; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 44($sp)
+; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 32($sp)
+; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 36($sp)
 ; O32-DAG:           sw [[R3]], 40([[R2]])
 ; O32-DAG:           sw [[R4]], 44([[R2]])
 ; NEW-DAG:           sd $8, 40([[R2]])
 
-; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 48($sp)
-; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 52($sp)
+; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 40($sp)
+; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 44($sp)
 ; O32-DAG:           sw [[R3]], 48([[R2]])
 ; O32-DAG:           sw [[R4]], 52([[R2]])
 ; NEW-DAG:           sd $9, 48([[R2]])
 
-; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 56($sp)
-; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 60($sp)
+; O32-DAG:           lw [[R3:\$([0-9]+|gp)]], 48($sp)
+; O32-DAG:           lw [[R4:\$([0-9]+|gp)]], 52($sp)
 ; O32-DAG:           sw [[R3]], 56([[R2]])
 ; O32-DAG:           sw [[R4]], 60([[R2]])
 ; NEW-DAG:           sd $10, 56([[R2]])
 
 ; N32/N64 have run out of registers and starts using the stack too
-; O32-DAG:           lw [[R3:\$[0-9]+]], 64($sp)
-; O32-DAG:           lw [[R4:\$[0-9]+]], 68($sp)
+; O32-DAG:           lw [[R3:\$[0-9]+]], 56($sp)
+; O32-DAG:           lw [[R4:\$[0-9]+]], 60($sp)
 ; O32-DAG:           sw [[R3]], 64([[R2]])
 ; O32-DAG:           sw [[R4]], 68([[R2]])
 ; NEW-DAG:           ld [[R3:\$[0-9]+]], 0($sp)
diff --git a/llvm/test/CodeGen/Mips/cconv/arguments-varargs.ll b/llvm/test/CodeGen/Mips/cconv/arguments-varargs.ll
index 785188b3c51..d662128945f 100644
--- a/llvm/test/CodeGen/Mips/cconv/arguments-varargs.ll
+++ b/llvm/test/CodeGen/Mips/cconv/arguments-varargs.ll
@@ -315,12 +315,11 @@ entry:
 ; Big-endian mode for N32/N64 must add an additional 4 to the offset due to byte
 ; order.
 ; O32-DAG:       addiu [[GV:\$[0-9]+]], ${{[0-9]+}}, %lo(dwords)
-; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG1]], 8([[GV]])
-; O32-DAG:       lw [[VA:\$[0-9]+]], 0([[SP]])
-; O32-DAG:       addiu [[VA2:\$[0-9]+]], [[VA]], 4
-; O32-DAG:       sw [[VA2]], 0([[SP]])
-; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       addiu [[VA3:\$[0-9]+]], [[VA2]], 4
+; O32-DAG:       sw [[VA3]], 0([[SP]])
+; O32-DAG:       lw [[ARG1:\$[0-9]+]], 4([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG1]], 12([[GV]])
 
 ; N32-DAG:       addiu [[GV:\$[0-9]+]], ${{[0-9]+}}, %lo(dwords)
@@ -349,10 +348,9 @@ entry:
 ; Load the second argument from the variable portion and copy it to the global.
 ; O32-DAG:       lw [[ARG2:\$[0-9]+]], 0([[VA]])
 ; O32-DAG:       sw [[ARG2]], 16([[GV]])
-; O32-DAG:       lw [[VA:\$[0-9]+]], 0([[SP]])
-; O32-DAG:       addiu [[VA2:\$[0-9]+]], [[VA]], 4
-; O32-DAG:       sw [[VA2]], 0([[SP]])
-; O32-DAG:       lw [[ARG2:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       addiu [[VA3:\$[0-9]+]], [[VA2]], 4
+; O32-DAG:       sw [[VA3]], 0([[SP]])
+; O32-DAG:       lw [[ARG2:\$[0-9]+]], 4([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG2]], 20([[GV]])
 
 ; NEW-DAG:       ld [[ARG2:\$[0-9]+]], 0([[VA2]])
@@ -678,12 +676,11 @@ entry:
 ; Big-endian mode for N32/N64 must add an additional 4 to the offset due to byte
 ; order.
 ; O32-DAG:       addiu [[GV:\$[0-9]+]], ${{[0-9]+}}, %lo(dwords)
-; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG1]], 8([[GV]])
-; O32-DAG:       lw [[VA:\$[0-9]+]], 0([[SP]])
-; O32-DAG:       addiu [[VA2:\$[0-9]+]], [[VA]], 4
-; O32-DAG:       sw [[VA2]], 0([[SP]])
-; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       addiu [[VA3:\$[0-9]+]], [[VA2]], 4
+; O32-DAG:       sw [[VA3]], 0([[SP]])
+; O32-DAG:       lw [[ARG1:\$[0-9]+]], 4([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG1]], 12([[GV]])
 
 ; N32-DAG:       addiu [[GV:\$[0-9]+]], ${{[0-9]+}}, %lo(dwords)
@@ -712,10 +709,9 @@ entry:
 ; Load the second argument from the variable portion and copy it to the global.
 ; O32-DAG:       lw [[ARG2:\$[0-9]+]], 0([[VA]])
 ; O32-DAG:       sw [[ARG2]], 16([[GV]])
-; O32-DAG:       lw [[VA:\$[0-9]+]], 0([[SP]])
-; O32-DAG:       addiu [[VA2:\$[0-9]+]], [[VA]], 4
+; O32-DAG:       addiu [[VA3:\$[0-9]+]], [[VA2]], 4
 ; O32-DAG:       sw [[VA2]], 0([[SP]])
-; O32-DAG:       lw [[ARG2:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       lw [[ARG2:\$[0-9]+]], 4([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG2]], 20([[GV]])
 
 ; NEW-DAG:       ld [[ARG2:\$[0-9]+]], 0([[VA2]])
@@ -1040,10 +1036,9 @@ entry:
 ; O32-DAG:       addiu [[GV:\$[0-9]+]], ${{[0-9]+}}, %lo(dwords)
 ; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA]])
 ; O32-DAG:       sw [[ARG1]], 8([[GV]])
-; O32-DAG:       lw [[VA:\$[0-9]+]], 0([[SP]])
-; O32-DAG:       addiu [[VA2:\$[0-9]+]], [[VA]], 4
-; O32-DAG:       sw [[VA2]], 0([[SP]])
-; O32-DAG:       lw [[ARG1:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       addiu [[VA3:\$[0-9]+]], [[VA2]], 4
+; O32-DAG:       sw [[VA3]], 0([[SP]])
+; O32-DAG:       lw [[ARG1:\$[0-9]+]], 4([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG1]], 12([[GV]])
 
 ; N32-DAG:       addiu [[GV:\$[0-9]+]], ${{[0-9]+}}, %lo(dwords)
@@ -1072,10 +1067,9 @@ entry:
 ; Load the second argument from the variable portion and copy it to the global.
 ; O32-DAG:       lw [[ARG2:\$[0-9]+]], 0([[VA]])
 ; O32-DAG:       sw [[ARG2]], 16([[GV]])
-; O32-DAG:       lw [[VA:\$[0-9]+]], 0([[SP]])
-; O32-DAG:       addiu [[VA2:\$[0-9]+]], [[VA]], 4
-; O32-DAG:       sw [[VA2]], 0([[SP]])
-; O32-DAG:       lw [[ARG2:\$[0-9]+]], 0([[VA]])
+; O32-DAG:       addiu [[VA3:\$[0-9]+]], [[VA2]], 4
+; O32-DAG:       sw [[VA3]], 0([[SP]])
+; O32-DAG:       lw [[ARG2:\$[0-9]+]], 4([[VA_TMP2]])
 ; O32-DAG:       sw [[ARG2]], 20([[GV]])
 
 ; NEW-DAG:       ld [[ARG2:\$[0-9]+]], 0([[VA2]])
diff --git a/llvm/test/CodeGen/Mips/fastcc.ll b/llvm/test/CodeGen/Mips/fastcc.ll
index 13abc20eb3e..fb1bc4d9a8a 100644
--- a/llvm/test/CodeGen/Mips/fastcc.ll
+++ b/llvm/test/CodeGen/Mips/fastcc.ll
@@ -132,20 +132,19 @@ entry:
 define internal fastcc void @callee0(i32 %a0, i32 %a1, i32 %a2, i32 %a3, i32 %a4, i32 %a5, i32 %a6, i32 %a7, i32 %a8, i32 %a9, i32 %a10, i32 %a11, i32 %a12, i32 %a13, i32 %a14, i32 %a15, i32 %a16) nounwind noinline {
 entry:
 ; CHECK: callee0
-; CHECK: sw  $4
-; CHECK: sw  $5
-; CHECK: sw  $6
-; CHECK: sw  $7
-; CHECK: sw  $8
-; CHECK: sw  $9
-; CHECK: sw  $10
-; CHECK: sw  $11
-; CHECK: sw  $12
-; CHECK: sw  $13
-; CHECK: sw  $14
-; CHECK: sw  $15
-; CHECK: sw  $24
-; CHECK: sw  $3
+; CHECK-DAG: sw  $4
+; CHECK-DAG: sw  $5
+; CHECK-DAG: sw  $7
+; CHECK-DAG: sw  $8
+; CHECK-DAG: sw  $9
+; CHECK-DAG: sw  $10
+; CHECK-DAG: sw  $11
+; CHECK-DAG: sw  $12
+; CHECK-DAG: sw  $13
+; CHECK-DAG: sw  $14
+; CHECK-DAG: sw  $15
+; CHECK-DAG: sw  $24
+; CHECK-DAG: sw  $3
 
 ; t6, t7 and t8 are reserved in NaCl and cannot be used for fastcc.
 ; CHECK-NACL-NOT: sw  $14
@@ -223,27 +222,27 @@ entry:
 
 define internal fastcc void @callee1(float %a0, float %a1, float %a2, float %a3, float %a4, float %a5, float %a6, float %a7, float %a8, float %a9, float %a10, float %a11, float %a12, float %a13, float %a14, float %a15, float %a16, float %a17, float %a18, float %a19, float %a20) nounwind noinline {
 entry:
-; CHECK: callee1
-; CHECK: swc1  $f0
-; CHECK: swc1  $f1
-; CHECK: swc1  $f2
-; CHECK: swc1  $f3
-; CHECK: swc1  $f4
-; CHECK: swc1  $f5
-; CHECK: swc1  $f6
-; CHECK: swc1  $f7
-; CHECK: swc1  $f8
-; CHECK: swc1  $f9
-; CHECK: swc1  $f10
-; CHECK: swc1  $f11
-; CHECK: swc1  $f12
-; CHECK: swc1  $f13
-; CHECK: swc1  $f14
-; CHECK: swc1  $f15
-; CHECK: swc1  $f16
-; CHECK: swc1  $f17
-; CHECK: swc1  $f18
-; CHECK: swc1  $f19
+; CHECK-LABEL: callee1:
+; CHECK-DAG: swc1  $f0
+; CHECK-DAG: swc1  $f1
+; CHECK-DAG: swc1  $f2
+; CHECK-DAG: swc1  $f3
+; CHECK-DAG: swc1  $f4
+; CHECK-DAG: swc1  $f5
+; CHECK-DAG: swc1  $f6
+; CHECK-DAG: swc1  $f7
+; CHECK-DAG: swc1  $f8
+; CHECK-DAG: swc1  $f9
+; CHECK-DAG: swc1  $f10
+; CHECK-DAG: swc1  $f11
+; CHECK-DAG: swc1  $f12
+; CHECK-DAG: swc1  $f13
+; CHECK-DAG: swc1  $f14
+; CHECK-DAG: swc1  $f15
+; CHECK-DAG: swc1  $f16
+; CHECK-DAG: swc1  $f17
+; CHECK-DAG: swc1  $f18
+; CHECK-DAG: swc1  $f19
 
   store float %a0, float* @gf0, align 4
   store float %a1, float* @gf1, align 4
@@ -316,8 +315,6 @@ entry:
 
 ; NOODDSPREG-LABEL:  callee2:
 
-; NOODDSPREG:        addiu   $sp, $sp, -[[OFFSET:[0-9]+]]
-
 ; Check that first 10 arguments are received in even float registers
 ; f0, f2, ... , f18. Check that 11th argument is received on stack.
 
@@ -333,7 +330,7 @@ entry:
 ; NOODDSPREG-DAG:    swc1    $f16, 32($[[R0]])
 ; NOODDSPREG-DAG:    swc1    $f18, 36($[[R0]])
 
-; NOODDSPREG-DAG:    lwc1    $[[F0:f[0-9]*[02468]]], [[OFFSET]]($sp)
+; NOODDSPREG-DAG:    lwc1    $[[F0:f[0-9]*[02468]]], 0($sp)
 ; NOODDSPREG-DAG:    swc1    $[[F0]], 40($[[R0]])
 
   store float %a0, float* getelementptr ([11 x float], [11 x float]* @fa, i32 0, i32 0), align 4
@@ -397,7 +394,6 @@ entry:
 
 ; FP64-NOODDSPREG-LABEL:  callee3:
 
-; FP64-NOODDSPREG:        addiu   $sp, $sp, -[[OFFSET:[0-9]+]]
 
 ; Check that first 10 arguments are received in even float registers
 ; f0, f2, ... , f18. Check that 11th argument is received on stack.
@@ -414,7 +410,7 @@ entry:
 ; FP64-NOODDSPREG-DAG:    sdc1    $f16, 64($[[R0]])
 ; FP64-NOODDSPREG-DAG:    sdc1    $f18, 72($[[R0]])
 
-; FP64-NOODDSPREG-DAG:    ldc1    $[[F0:f[0-9]*[02468]]], [[OFFSET]]($sp)
+; FP64-NOODDSPREG-DAG:    ldc1    $[[F0:f[0-9]*[02468]]], 0($sp)
 ; FP64-NOODDSPREG-DAG:    sdc1    $[[F0]], 80($[[R0]])
 
   store double %a0, double* getelementptr ([11 x double], [11 x double]* @da, i32 0, i32 0), align 8
diff --git a/llvm/test/CodeGen/Mips/load-store-left-right.ll b/llvm/test/CodeGen/Mips/load-store-left-right.ll
index 3bd924a8120..6def55cf883 100644
--- a/llvm/test/CodeGen/Mips/load-store-left-right.ll
+++ b/llvm/test/CodeGen/Mips/load-store-left-right.ll
@@ -250,12 +250,18 @@ entry:
 ; MIPS64-EB:     ld $[[PTR:[0-9]+]], %got_disp(struct_s0)(
 ; MIPS64R6:      ld $[[PTR:[0-9]+]], %got_disp(struct_s0)(
 
-; FIXME: We should be able to do better than this on MIPS32r6/MIPS64r6 since
-;        we have unaligned halfword load/store available
-; ALL-DAG:       lbu $[[R1:[0-9]+]], 0($[[PTR]])
-; ALL-DAG:       sb $[[R1]], 2($[[PTR]])
-; ALL-DAG:       lbu $[[R1:[0-9]+]], 1($[[PTR]])
-; ALL-DAG:       sb $[[R1]], 3($[[PTR]])
+; MIPS32-DAG:       lbu $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS32-DAG:       sb $[[R1]], 2($[[PTR]])
+; MIPS32-DAG:       lbu $[[R2:[0-9]+]], 1($[[PTR]])
+; MIPS32-DAG:       sb $[[R2]], 3($[[PTR]])
+
+; MIPS32R6:       lhu $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS32R6:       sh $[[R1]], 2($[[PTR]])
+
+; MIPS64-DAG:       lbu $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64-DAG:       sb $[[R1]], 2($[[PTR]])
+; MIPS64-DAG:       lbu $[[R2:[0-9]+]], 1($[[PTR]])
+; MIPS64-DAG:       sb $[[R2]], 3($[[PTR]])
 
   %0 = load %struct.S0, %struct.S0* getelementptr inbounds (%struct.S0, %struct.S0* @struct_s0, i32 0), align 1
   store %struct.S0 %0, %struct.S0* getelementptr inbounds (%struct.S0, %struct.S0* @struct_s0, i32 1), align 1
@@ -268,37 +274,54 @@ entry:
 
 ; MIPS32-EL:     lw $[[PTR:[0-9]+]], %got(struct_s1)(
 ; MIPS32-EB:     lw $[[PTR:[0-9]+]], %got(struct_s1)(
-; MIPS32-DAG:    lbu $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS32-DAG:    sb $[[R1]], 4($[[PTR]])
-; MIPS32-DAG:    lbu $[[R1:[0-9]+]], 1($[[PTR]])
-; MIPS32-DAG:    sb $[[R1]], 5($[[PTR]])
-; MIPS32-DAG:    lbu $[[R1:[0-9]+]], 2($[[PTR]])
-; MIPS32-DAG:    sb $[[R1]], 6($[[PTR]])
-; MIPS32-DAG:    lbu $[[R1:[0-9]+]], 3($[[PTR]])
-; MIPS32-DAG:    sb $[[R1]], 7($[[PTR]])
+; MIPS32-EL-DAG:    lwl $[[R1:[0-9]+]], 3($[[PTR]])
+; MIPS32-EL-DAG:    lwr $[[R1]], 0($[[PTR]])
+; MIPS32-EL-DAG:    swl $[[R1]], 7($[[PTR]])
+; MIPS32-EL-DAG:    swr $[[R1]], 4($[[PTR]])
+; MIPS32-EB-DAG:    lwl $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS32-EB-DAG:    lwr $[[R1]], 3($[[PTR]])
+; MIPS32-EB-DAG:    swl $[[R1]], 4($[[PTR]])
+; MIPS32-EB-DAG:    swr $[[R1]], 7($[[PTR]])
+
+; MIPS32-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    sb $[[R1]], 4($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 1($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    sb $[[R1]], 5($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 2($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    sb $[[R1]], 6($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 3($[[PTR]])
+; MIPS32-NOLEFTRIGHT-DAG:    sb $[[R1]], 7($[[PTR]])
 
 ; MIPS32R6:      lw $[[PTR:[0-9]+]], %got(struct_s1)(
-; MIPS32R6-DAG:  lhu $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS32R6-DAG:  sh $[[R1]], 4($[[PTR]])
-; MIPS32R6-DAG:  lhu $[[R1:[0-9]+]], 2($[[PTR]])
-; MIPS32R6-DAG:  sh $[[R1]], 6($[[PTR]])
+; MIPS32R6-DAG:  lw $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS32R6-DAG:  sw $[[R1]], 4($[[PTR]])
 
 ; MIPS64-EL:     ld $[[PTR:[0-9]+]], %got_disp(struct_s1)(
 ; MIPS64-EB:     ld $[[PTR:[0-9]+]], %got_disp(struct_s1)(
-; MIPS64-DAG:    lbu $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS64-DAG:    sb $[[R1]], 4($[[PTR]])
-; MIPS64-DAG:    lbu $[[R1:[0-9]+]], 1($[[PTR]])
-; MIPS64-DAG:    sb $[[R1]], 5($[[PTR]])
-; MIPS64-DAG:    lbu $[[R1:[0-9]+]], 2($[[PTR]])
-; MIPS64-DAG:    sb $[[R1]], 6($[[PTR]])
-; MIPS64-DAG:    lbu $[[R1:[0-9]+]], 3($[[PTR]])
-; MIPS64-DAG:    sb $[[R1]], 7($[[PTR]])
+
+; MIPS64-EL-DAG:    lwl $[[R1:[0-9]+]], 3($[[PTR]])
+; MIPS64-EL-DAG:    lwr $[[R1]], 0($[[PTR]])
+; MIPS64-EL-DAG:    swl $[[R1]], 7($[[PTR]])
+; MIPS64-EL-DAG:    swr $[[R1]], 4($[[PTR]])
+
+; MIPS64-EB-DAG:    lwl $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64-EB-DAG:    lwr $[[R1]], 3($[[PTR]])
+; MIPS64-EB-DAG:    swl $[[R1]], 4($[[PTR]])
+; MIPS64-EB-DAG:    swr $[[R1]], 7($[[PTR]])
+
+
+; MIPS64-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    sb $[[R1]], 4($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 1($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    sb $[[R1]], 5($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 2($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    sb $[[R1]], 6($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    lbu $[[R1:[0-9]+]], 3($[[PTR]])
+; MIPS64-NOLEFTRIGHT-DAG:    sb $[[R1]], 7($[[PTR]])
 
 ; MIPS64R6:      ld $[[PTR:[0-9]+]], %got_disp(struct_s1)(
-; MIPS64R6-DAG:  lhu $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS64R6-DAG:  sh $[[R1]], 4($[[PTR]])
-; MIPS64R6-DAG:  lhu $[[R1:[0-9]+]], 2($[[PTR]])
-; MIPS64R6-DAG:  sh $[[R1]], 6($[[PTR]])
+; MIPS64R6-DAG:  lw $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64R6-DAG:  sw $[[R1]], 4($[[PTR]])
 
   %0 = load %struct.S1, %struct.S1* getelementptr inbounds (%struct.S1, %struct.S1* @struct_s1, i32 0), align 1
   store %struct.S1 %0, %struct.S1* getelementptr inbounds (%struct.S1, %struct.S1* @struct_s1, i32 1), align 1
@@ -336,30 +359,21 @@ entry:
 ; MIPS32R6-DAG:  sw $[[R1]],       12($[[PTR]])
 
 ; MIPS64-EL:     ld $[[PTR:[0-9]+]], %got_disp(struct_s2)(
-; MIPS64-EL-DAG: lwl $[[R1:[0-9]+]], 3($[[PTR]])
-; MIPS64-EL-DAG: lwr $[[R1]],        0($[[PTR]])
-; MIPS64-EL-DAG: swl $[[R1]],       11($[[PTR]])
-; MIPS64-EL-DAG: swr $[[R1]],        8($[[PTR]])
-; MIPS64-EL-DAG: lwl $[[R1:[0-9]+]], 7($[[PTR]])
-; MIPS64-EL-DAG: lwr $[[R1]],        4($[[PTR]])
-; MIPS64-EL-DAG: swl $[[R1]],       15($[[PTR]])
-; MIPS64-EL-DAG: swr $[[R1]],       12($[[PTR]])
+
+; MIPS64-EL-DAG: ldl $[[R1:[0-9]+]], 7($[[PTR]])
+; MIPS64-EL-DAG: ldr $[[R1]],        0($[[PTR]])
+; MIPS64-EL-DAG: sdl $[[R1]],       15($[[PTR]])
+; MIPS64-EL-DAG: sdr $[[R1]],        8($[[PTR]])
 
 ; MIPS64-EB:     ld $[[PTR:[0-9]+]], %got_disp(struct_s2)(
-; MIPS64-EB-DAG: lwl $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS64-EB-DAG: lwr $[[R1]],        3($[[PTR]])
-; MIPS64-EB-DAG: swl $[[R1]],        8($[[PTR]])
-; MIPS64-EB-DAG: swr $[[R1]],       11($[[PTR]])
-; MIPS64-EB-DAG: lwl $[[R1:[0-9]+]], 4($[[PTR]])
-; MIPS64-EB-DAG: lwr $[[R1]],        7($[[PTR]])
-; MIPS64-EB-DAG: swl $[[R1]],       12($[[PTR]])
-; MIPS64-EB-DAG: swr $[[R1]],       15($[[PTR]])
+; MIPS64-EB-DAG: ldl $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64-EB-DAG: ldr $[[R1]],        7($[[PTR]])
+; MIPS64-EB-DAG: sdl $[[R1]],        8($[[PTR]])
+; MIPS64-EB-DAG: sdr $[[R1]],       15($[[PTR]])
 
 ; MIPS64R6:      ld $[[PTR:[0-9]+]], %got_disp(struct_s2)(
-; MIPS64R6-DAG:  lw $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS64R6-DAG:  sw $[[R1]],        8($[[PTR]])
-; MIPS64R6-DAG:  lw $[[R1:[0-9]+]], 4($[[PTR]])
-; MIPS64R6-DAG:  sw $[[R1]],       12($[[PTR]])
+; MIPS64R6-DAG:  ld $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64R6-DAG:  sd $[[R1]],        8($[[PTR]])
 
   %0 = load %struct.S2, %struct.S2* getelementptr inbounds (%struct.S2, %struct.S2* @struct_s2, i32 0), align 1
   store %struct.S2 %0, %struct.S2* getelementptr inbounds (%struct.S2, %struct.S2* @struct_s2, i32 1), align 1
@@ -416,17 +430,17 @@ entry:
 ; MIPS64-EL-DAG: lwl $[[R1:[0-9]+]], 3($[[PTR]])
 ; MIPS64-EL-DAG: lwr $[[R1]],   0($[[PTR]])
 
-; MIPS64-EB:     ld $[[SPTR:[0-9]+]], %got_disp(arr)(
-; MIPS64-EB-DAG: lwl  $[[R1:[0-9]+]], 0($[[PTR]])
-; MIPS64-EB-DAG: lwr  $[[R1]],   3($[[PTR]])
-; MIPS64-EB-DAG: dsll $[[R1]], $[[R1]], 32
+; MIPS64-EB: ld $[[SPTR:[0-9]+]], %got_disp(arr)(
 ; MIPS64-EB-DAG: lbu  $[[R2:[0-9]+]], 5($[[PTR]])
 ; MIPS64-EB-DAG: lbu  $[[R3:[0-9]+]], 4($[[PTR]])
 ; MIPS64-EB-DAG: dsll $[[T0:[0-9]+]], $[[R3]], 8
 ; MIPS64-EB-DAG: or   $[[T1:[0-9]+]], $[[T0]], $[[R2]]
-; MIPS64-EB-DAG: dsll $[[T1]], $[[T1]], 16
-; MIPS64-EB-DAG: or   $[[T3:[0-9]+]], $[[R1]], $[[T1]]
 ; MIPS64-EB-DAG: lbu  $[[R4:[0-9]+]], 6($[[PTR]])
+; MIPS64-EB-DAG: dsll $[[T1]], $[[T1]], 16
+; MIPS64-EB-DAG: lwl  $[[R1:[0-9]+]], 0($[[PTR]])
+; MIPS64-EB-DAG: lwr  $[[R1]],   3($[[PTR]])
+; MIPS64-EB-DAG: dsll $[[R5:[0-9]+]], $[[R1]], 32
+; MIPS64-EB-DAG: or   $[[T3:[0-9]+]], $[[R5]], $[[T1]]
 ; MIPS64-EB-DAG: dsll $[[T4:[0-9]+]], $[[R4]], 8
 ; MIPS64-EB-DAG: or   $4, $[[T3]], $[[T4]]
 
diff --git a/llvm/test/CodeGen/Mips/micromips-li.ll b/llvm/test/CodeGen/Mips/micromips-li.ll
index ac315f93825..997f4e9196a 100644
--- a/llvm/test/CodeGen/Mips/micromips-li.ll
+++ b/llvm/test/CodeGen/Mips/micromips-li.ll
@@ -13,6 +13,6 @@ entry:
   ret i32 0
 }
 
-; CHECK: li16   ${{[2-7]|16|17}}, 1
 ; CHECK: addiu  ${{[0-9]+}}, $zero, 2148
+; CHECK: li16   ${{[2-7]|16|17}}, 1
 ; CHECK: ori ${{[0-9]+}}, $zero, 33332
diff --git a/llvm/test/CodeGen/Mips/mips64-f128-call.ll b/llvm/test/CodeGen/Mips/mips64-f128-call.ll
index c59f25ef4af..19fa8fc7524 100644
--- a/llvm/test/CodeGen/Mips/mips64-f128-call.ll
+++ b/llvm/test/CodeGen/Mips/mips64-f128-call.ll
@@ -4,8 +4,8 @@
 @gld1 = external global fp128
 
 ; CHECK: foo0
-; CHECK: sdc1  $f12, %lo(gld0)(${{[0-9]+}})
-; CHECK: sdc1  $f13, 8(${{[0-9]+}})
+; CHECK-DAG: sdc1  $f12, %lo(gld0)(${{[0-9]+}})
+; CHECK-DAG: sdc1  $f13, 8(${{[0-9]+}})
 
 define void @foo0(fp128 %a0) {
 entry:
@@ -14,8 +14,8 @@ entry:
 }
 
 ; CHECK: foo1
-; CHECK: ldc1  $f12, %lo(gld0)(${{[0-9]+}})
-; CHECK: ldc1  $f13, 8(${{[0-9]+}})
+; CHECK-DAG: ldc1  $f12, %lo(gld0)(${{[0-9]+}})
+; CHECK-DAG: ldc1  $f13, 8(${{[0-9]+}})
 
 define void @foo1() {
 entry:
@@ -26,11 +26,11 @@ entry:
 
 declare void @foo2(fp128)
 
+
 ; CHECK: foo3:
-; CHECK: daddiu $[[R0:[0-9]+]], ${{[0-9]+}}, %hi(gld0)
-; CHECK: dsll $[[R1:[0-9]+]], $[[R0]], 16
+
+; CHECK: daddiu $[[R2:[0-9]+]], $[[R1:[0-9]+]], %lo(gld0)
 ; CHECK: sdc1 $f0, %lo(gld0)($[[R1]])
-; CHECK: daddiu $[[R2:[0-9]]], $[[R1]], %lo(gld0)
 ; CHECK: sdc1 $f2, 8($[[R2]])
 ; CHECK: daddiu $[[R3:[0-9]+]], ${{[0-9]+}}, %hi(gld1)
 ; CHECK: dsll $[[R4:[0-9]+]], $[[R3]], 16
@@ -39,7 +39,6 @@ declare void @foo2(fp128)
 ; CHECK: ldc1 $f2, 8($[[R5]])
 
 
-
 define fp128 @foo3() {
 entry:
   %call = tail call fp128 @foo4()
diff --git a/llvm/test/CodeGen/Mips/mips64-f128.ll b/llvm/test/CodeGen/Mips/mips64-f128.ll
index 304ab8b90d3..237b4c5f1ee 100644
--- a/llvm/test/CodeGen/Mips/mips64-f128.ll
+++ b/llvm/test/CodeGen/Mips/mips64-f128.ll
@@ -577,10 +577,10 @@ entry:
 
 ; ALL-LABEL: store_LD_LD:
 ; ALL: ld $[[R0:[0-9]+]], %got_disp(gld1)
-; ALL: ld $[[R1:[0-9]+]], 0($[[R0]])
 ; ALL: ld $[[R2:[0-9]+]], 8($[[R0]])
 ; ALL: ld $[[R3:[0-9]+]], %got_disp(gld0)
 ; ALL: sd $[[R2]], 8($[[R3]])
+; ALL: ld $[[R1:[0-9]+]], 0($[[R0]])
 ; ALL: sd $[[R1]], 0($[[R3]])
 
 define void @store_LD_LD() {
diff --git a/llvm/test/CodeGen/Mips/mno-ldc1-sdc1.ll b/llvm/test/CodeGen/Mips/mno-ldc1-sdc1.ll
index 9663138d4c8..0260afaa186 100644
--- a/llvm/test/CodeGen/Mips/mno-ldc1-sdc1.ll
+++ b/llvm/test/CodeGen/Mips/mno-ldc1-sdc1.ll
@@ -130,12 +130,12 @@
 ; MM-MNO-PIC:         addiu   $[[R1:[0-9]+]], $[[R0]], %lo(_gp_disp)
 ; MM-MNO-PIC:         addu    $[[R2:[0-9]+]], $[[R1]], $25
 ; MM-MNO-PIC:         lw      $[[R3:[0-9]+]], %got(g0)($[[R2]])
-; MM-MNO-PIC:         lw16    $[[R4:[0-9]+]], 0($[[R3]])
-; MM-MNO-PIC:         lw16    $[[R5:[0-9]+]], 4($[[R3]])
-; MM-MNO-LE-PIC:      mtc1    $[[R4]], $f0
-; MM-MNO-LE-PIC:      mthc1   $[[R5]], $f0
-; MM-MNO-BE-PIC:      mtc1    $[[R5]], $f0
-; MM-MNO-BE-PIC:      mthc1   $[[R4]], $f0
+; MM-MNO-PIC-DAG:     lw16    $[[R4:[0-9]+]], 0($[[R3]])
+; MM-MNO-PIC-DAG:     lw16    $[[R5:[0-9]+]], 4($[[R3]])
+; MM-MNO-LE-PIC-DAG:  mtc1    $[[R4]], $f0
+; MM-MNO-LE-PIC-DAG:  mthc1   $[[R5]], $f0
+; MM-MNO-BE-PIC-DAG:  mtc1    $[[R5]], $f0
+; MM-MNO-BE-PIC-DAG:  mthc1   $[[R4]], $f0
 
 ; MM-STATIC-PIC:      lui     $[[R0:[0-9]+]], %hi(g0)
 ; MM-STATIC-PIC:      ldc1    $f0, %lo(g0)($[[R0]])
@@ -214,13 +214,13 @@ entry:
 ; MM-MNO-PIC:         lui     $[[R0:[0-9]+]], %hi(_gp_disp)
 ; MM-MNO-PIC:         addiu   $[[R1:[0-9]+]], $[[R0]], %lo(_gp_disp)
 ; MM-MNO-PIC:         addu    $[[R2:[0-9]+]], $[[R1]], $25
-; MM-MNO-LE-PIC:      mfc1    $[[R3:[0-9]+]], $f12
-; MM-MNO-BE-PIC:      mfhc1   $[[R3:[0-9]+]], $f12
-; MM-MNO-PIC:         lw      $[[R4:[0-9]+]], %got(g0)($[[R2]])
-; MM-MNO-PIC:         sw16    $[[R3]], 0($[[R4]])
-; MM-MNO-LE-PIC:      mfhc1   $[[R5:[0-9]+]], $f12
-; MM-MNO-BE-PIC:      mfc1    $[[R5:[0-9]+]], $f12
-; MM-MNO-PIC:         sw16    $[[R5]], 4($[[R4]])
+; MM-MNO-LE-PIC-DAG:  mfc1    $[[R3:[0-9]+]], $f12
+; MM-MNO-BE-PIC-DAG:  mfhc1   $[[R3:[0-9]+]], $f12
+; MM-MNO-PIC-DAG:     lw      $[[R4:[0-9]+]], %got(g0)($[[R2]])
+; MM-MNO-PIC-DAG:     sw16    $[[R3]], 0($[[R4]])
+; MM-MNO-LE-PIC-DAG:  mfhc1   $[[R5:[0-9]+]], $f12
+; MM-MNO-BE-PIC-DAG:  mfc1    $[[R5:[0-9]+]], $f12
+; MM-MNO-PIC-DAG:     sw16    $[[R5]], 4($[[R4]])
 
 ; MM-STATIC-PIC:      lui     $[[R0:[0-9]+]], %hi(g0)
 ; MM-STATIC-PIC:      sdc1    $f12, %lo(g0)($[[R0]])
@@ -267,8 +267,8 @@ entry:
 
 ; MM-MNO-PIC:    sll16   $[[R0:[0-9]+]], $5, 3
 ; MM-MNO-PIC:    addu16  $[[R1:[0-9]+]], $4, $[[R0]]
-; MM-MNO-PIC:    lw16    $[[R2:[0-9]+]], 0($[[R1]])
-; MM-MNO-PIC:    lw16    $[[R3:[0-9]+]], 4($[[R1]])
+; MM-MNO-PIC-DAG: lw16    $[[R2:[0-9]+]], 0($[[R1]])
+; MM-MNO-PIC-DAG: lw16    $[[R3:[0-9]+]], 4($[[R1]])
 ; MM-MNO-LE-PIC: mtc1    $[[R2]], $f0
 ; MM-MNO-LE-PIC: mthc1   $[[R3]], $f0
 ; MM-MNO-BE-PIC: mtc1    $[[R3]], $f0
@@ -313,14 +313,14 @@ entry:
 ; MM:            addu16  $[[R1:[0-9]+]], $6, $[[R0]]
 ; MM:            sdc1    $f12, 0($[[R1]])
 
-; MM-MNO-PIC:    sll16   $[[R0:[0-9]+]], $7, 3
-; MM-MNO-PIC:    addu16  $[[R1:[0-9]+]], $6, $[[R0]]
-; MM-MNO-LE-PIC: mfc1    $[[R2:[0-9]+]], $f12
-; MM-MNO-BE-PIC: mfhc1   $[[R2:[0-9]+]], $f12
-; MM-MNO-PIC:    sw16    $[[R2]], 0($[[R1]])
-; MM-MNO-LE-PIC: mfhc1   $[[R3:[0-9]+]], $f12
-; MM-MNO-BE-PIC: mfc1    $[[R3:[0-9]+]], $f12
-; MM-MNO-PIC:    sw16    $[[R3]], 4($[[R1]])
+; MM-MNO-PIC:        sll16   $[[R0:[0-9]+]], $7, 3
+; MM-MNO-PIC:        addu16  $[[R1:[0-9]+]], $6, $[[R0]]
+; MM-MNO-LE-PIC-DAG: mfc1    $[[R2:[0-9]+]], $f12
+; MM-MNO-BE-PIC-DAG: mfhc1   $[[R2:[0-9]+]], $f12
+; MM-MNO-PIC-DAG:    sw16    $[[R2]], 0($[[R1]])
+; MM-MNO-LE-PIC-DAG: mfhc1   $[[R3:[0-9]+]], $f12
+; MM-MNO-BE-PIC-DAG: mfc1    $[[R3:[0-9]+]], $f12
+; MM-MNO-PIC-DAG:    sw16    $[[R3]], 4($[[R1]])
 
 ; MM-STATIC-PIC: sll16   $[[R0:[0-9]+]], $7, 3
 ; MM-STATIC-PIC: addu16  $[[R1:[0-9]+]], $6, $[[R0]]
diff --git a/llvm/test/CodeGen/Mips/msa/f16-llvm-ir.ll b/llvm/test/CodeGen/Mips/msa/f16-llvm-ir.ll
index 9957d5be26e..ac69dc913c1 100644
--- a/llvm/test/CodeGen/Mips/msa/f16-llvm-ir.ll
+++ b/llvm/test/CodeGen/Mips/msa/f16-llvm-ir.ll
@@ -234,15 +234,15 @@ entry:
 ; MIPS32:     insert.w $w[[W0]][1], $[[R1]]
 ; MIPS32:     insert.w $w[[W0]][3], $[[R1]]
 
-; MIPS64-N64: ld $[[R3:[0-9]+]], %got_disp(h)
-; MIPS64-N32: lw $[[R3:[0-9]+]], %got_disp(h)
-; MIPS64:     dmfc1 $[[R1:[0-9]+]], $f[[F2]]
-; MIPS64:     fill.d $w[[W0:[0-9]+]], $[[R1]]
+; MIPS64-N64-DAG: ld $[[R3:[0-9]+]], %got_disp(h)
+; MIPS64-N32-DAG: lw $[[R3:[0-9]+]], %got_disp(h)
+; MIPS64-DAG:     dmfc1 $[[R1:[0-9]+]], $f[[F2]]
+; MIPS64-DAG:     fill.d $w[[W0:[0-9]+]], $[[R1]]
 
-; ALL:        fexdo.w $w[[W1:[0-9]+]], $w[[W0]], $w[[W0]]
-; ALL:        fexdo.h $w[[W2:[0-9]+]], $w[[W1]], $w[[W1]]
+; ALL-DAG:        fexdo.w $w[[W1:[0-9]+]], $w[[W0]], $w[[W0]]
+; ALL-DAG:        fexdo.h $w[[W2:[0-9]+]], $w[[W1]], $w[[W1]]
 
-; MIPS32:     lw $[[R3:[0-9]+]], %got(h)
+; MIPS32-DAG:     lw $[[R3:[0-9]+]], %got(h)
 
 ; ALL:        copy_u.h $[[R2:[0-9]+]], $w[[W2]]
 ; ALL:        sh $[[R2]], 0($[[R3]])
diff --git a/llvm/test/CodeGen/Mips/msa/i5_ld_st.ll b/llvm/test/CodeGen/Mips/msa/i5_ld_st.ll
index c644d242a00..812c400d46e 100644
--- a/llvm/test/CodeGen/Mips/msa/i5_ld_st.ll
+++ b/llvm/test/CodeGen/Mips/msa/i5_ld_st.ll
@@ -336,8 +336,8 @@ entry:
 
 ; CHECK: llvm_mips_st_b_valid_range_tests:
 ; CHECK: ld.b
-; CHECK: st.b [[R1:\$w[0-9]+]], -512(
-; CHECK: st.b [[R1:\$w[0-9]+]], 511(
+; CHECK-DAG: st.b [[R1:\$w[0-9]+]], -512(
+; CHECK-DAG: st.b [[R1:\$w[0-9]+]], 511(
 ; CHECK: .size llvm_mips_st_b_valid_range_tests
 ;
 
@@ -351,10 +351,10 @@ entry:
 }
 
 ; CHECK: llvm_mips_st_b_invalid_range_tests:
-; CHECK: addiu $2, $1, -513
+; CHECK: addiu $2, $1, 512
 ; CHECK: ld.b
 ; CHECK: st.b [[R1:\$w[0-9]+]], 0(
-; CHECK: addiu $1, $1, 512
+; CHECK: addiu $1, $1, -513
 ; CHECK: st.b [[R1:\$w[0-9]+]], 0(
 ; CHECK: .size llvm_mips_st_b_invalid_range_tests
 ;
@@ -404,8 +404,8 @@ entry:
 
 ; CHECK: llvm_mips_st_h_valid_range_tests:
 ; CHECK: ld.h
-; CHECK: st.h [[R1:\$w[0-9]+]], -1024(
-; CHECK: st.h [[R1:\$w[0-9]+]], 1022(
+; CHECK-DAG: st.h [[R1:\$w[0-9]+]], -1024(
+; CHECK-DAG: st.h [[R1:\$w[0-9]+]], 1022(
 ; CHECK: .size llvm_mips_st_h_valid_range_tests
 ;
 
@@ -419,10 +419,10 @@ entry:
 }
 
 ; CHECK: llvm_mips_st_h_invalid_range_tests:
-; CHECK: addiu $2, $1, -1026
+; CHECK: addiu $2, $1, 1024
 ; CHECK: ld.h
 ; CHECK: st.h [[R1:\$w[0-9]+]], 0(
-; CHECK: addiu $1, $1, 1024
+; CHECK: addiu $1, $1, -1026
 ; CHECK: st.h [[R1:\$w[0-9]+]], 0(
 ; CHECK: .size llvm_mips_st_h_invalid_range_tests
 ;
@@ -472,8 +472,8 @@ entry:
 
 ; CHECK: llvm_mips_st_w_valid_range_tests:
 ; CHECK: ld.w
-; CHECK: st.w [[R1:\$w[0-9]+]], -2048(
-; CHECK: st.w [[R1:\$w[0-9]+]], 2044(
+; CHECK-DAG: st.w [[R1:\$w[0-9]+]], -2048(
+; CHECK-DAG: st.w [[R1:\$w[0-9]+]], 2044(
 ; CHECK: .size llvm_mips_st_w_valid_range_tests
 ;
 
@@ -487,10 +487,10 @@ entry:
 }
 
 ; CHECK: llvm_mips_st_w_invalid_range_tests:
-; CHECK: addiu $2, $1, -2052
+; CHECK: addiu $2, $1, 2048
 ; CHECK: ld.w
 ; CHECK: st.w [[R1:\$w[0-9]+]], 0(
-; CHECK: addiu $1, $1, 2048
+; CHECK: addiu $1, $1, -2052
 ; CHECK: st.w [[R1:\$w[0-9]+]], 0(
 ; CHECK: .size llvm_mips_st_w_invalid_range_tests
 ;
@@ -540,8 +540,8 @@ entry:
 
 ; CHECK: llvm_mips_st_d_valid_range_tests:
 ; CHECK: ld.d
-; CHECK: st.d [[R1:\$w[0-9]+]], -4096(
-; CHECK: st.d [[R1:\$w[0-9]+]], 4088(
+; CHECK-DAG: st.d [[R1:\$w[0-9]+]], -4096(
+; CHECK-DAG: st.d [[R1:\$w[0-9]+]], 4088(
 ; CHECK: .size llvm_mips_st_d_valid_range_tests
 ;
 
@@ -555,10 +555,10 @@ entry:
 }
 
 ; CHECK: llvm_mips_st_d_invalid_range_tests:
-; CHECK: addiu $2, $1, -4104
+; CHECK: addiu $2, $1, 4096
 ; CHECK: ld.d
 ; CHECK: st.d [[R1:\$w[0-9]+]], 0(
-; CHECK: addiu $1, $1, 4096
+; CHECK: addiu $1, $1, -4104
 ; CHECK: st.d [[R1:\$w[0-9]+]], 0(
 ; CHECK: .size llvm_mips_st_d_invalid_range_tests
 ;
diff --git a/llvm/test/CodeGen/Mips/o32_cc_byval.ll b/llvm/test/CodeGen/Mips/o32_cc_byval.ll
index 33431dba43c..eadf4abfc75 100644
--- a/llvm/test/CodeGen/Mips/o32_cc_byval.ll
+++ b/llvm/test/CodeGen/Mips/o32_cc_byval.ll
@@ -45,20 +45,18 @@ declare void @callee3(float, %struct.S3* byval, %struct.S1* byval)
 define void @f2(float %f, %struct.S1* nocapture byval %s1) nounwind {
 entry:
 ; CHECK: addiu $sp, $sp, -48
-; CHECK: sw  $7, 60($sp)
-; CHECK: sw  $6, 56($sp)
-; CHECK: lw  $4, 80($sp)
-; CHECK: ldc1 $f[[F0:[0-9]+]], 72($sp)
-; CHECK: lw  $[[R3:[0-9]+]], 64($sp)
-; CHECK: lw  $[[R4:[0-9]+]], 68($sp)
-; CHECK: lw  $[[R2:[0-9]+]], 60($sp)
-; CHECK: lh  $[[R1:[0-9]+]], 58($sp)
-; CHECK: lb  $[[R0:[0-9]+]], 56($sp)
-; CHECK: sw  $[[R0]], 32($sp)
-; CHECK: sw  $[[R1]], 28($sp)
-; CHECK: sw  $[[R2]], 24($sp)
-; CHECK: sw  $[[R4]], 20($sp)
-; CHECK: sw  $[[R3]], 16($sp)
+; CHECK-DAG: sw  $7, 60($sp)
+; CHECK-DAG: sw  $6, 56($sp)
+; CHECK-DAG: ldc1 $f[[F0:[0-9]+]], 72($sp)
+; CHECK-DAG: lw  $[[R3:[0-9]+]], 64($sp)
+; CHECK-DAG: lw  $[[R4:[0-9]+]], 68($sp)
+; CHECK-DAG: lh  $[[R1:[0-9]+]], 58($sp)
+; CHECK-DAG: lb  $[[R0:[0-9]+]], 56($sp)
+; CHECK-DAG: sw  $[[R0]], 32($sp)
+; CHECK-DAG: sw  $[[R1]], 28($sp)
+; CHECK-DAG: sw  $[[R4]], 20($sp)
+; CHECK-DAG: sw  $[[R3]], 16($sp)
+; CHECK-DAG: sw  $7, 24($sp)
 ; CHECK: mfc1 $6, $f[[F0]]
 
   %i2 = getelementptr inbounds %struct.S1, %struct.S1* %s1, i32 0, i32 5
@@ -82,13 +80,11 @@ declare void @callee4(i32, double, i64, i32, i16 signext, i8 signext, float)
 define void @f3(%struct.S2* nocapture byval %s2) nounwind {
 entry:
 ; CHECK: addiu $sp, $sp, -48
-; CHECK: sw  $7, 60($sp)
-; CHECK: sw  $6, 56($sp)
-; CHECK: sw  $5, 52($sp)
-; CHECK: sw  $4, 48($sp)
-; CHECK: lw  $4, 48($sp)
-; CHECK: lw  $[[R0:[0-9]+]], 60($sp)
-; CHECK: sw  $[[R0]], 24($sp)
+; CHECK-DAG: sw  $7, 60($sp)
+; CHECK-DAG: sw  $6, 56($sp)
+; CHECK-DAG: sw  $5, 52($sp)
+; CHECK-DAG: sw  $4, 48($sp)
+; CHECK-DAG: sw  $7, 24($sp)
 
   %arrayidx = getelementptr inbounds %struct.S2, %struct.S2* %s2, i32 0, i32 0, i32 0
   %tmp = load i32, i32* %arrayidx, align 4
@@ -101,14 +97,14 @@ entry:
 define void @f4(float %f, %struct.S3* nocapture byval %s3, %struct.S1* nocapture byval %s1) nounwind {
 entry:
 ; CHECK: addiu $sp, $sp, -48
-; CHECK: sw  $7, 60($sp)
-; CHECK: sw  $6, 56($sp)
-; CHECK: sw  $5, 52($sp)
-; CHECK: lw  $4, 60($sp)
-; CHECK: lw  $[[R1:[0-9]+]], 80($sp)
-; CHECK: lb  $[[R0:[0-9]+]], 52($sp)
-; CHECK: sw  $[[R0]], 32($sp)
-; CHECK: sw  $[[R1]], 24($sp)
+; CHECK-DAG: sw  $7, 60($sp)
+; CHECK-DAG: sw  $6, 56($sp)
+; CHECK-DAG: sw  $5, 52($sp)
+; CHECK-DAG: lw  $[[R1:[0-9]+]], 80($sp)
+; CHECK-DAG: lb  $[[R0:[0-9]+]], 52($sp)
+; CHECK-DAG: sw  $[[R0]], 32($sp)
+; CHECK-DAG: sw  $[[R1]], 24($sp)
+; CHECK: move $4, $7
 
   %i = getelementptr inbounds %struct.S1, %struct.S1* %s1, i32 0, i32 2
   %tmp = load i32, i32* %i, align 4
diff --git a/llvm/test/CodeGen/Mips/o32_cc_vararg.ll b/llvm/test/CodeGen/Mips/o32_cc_vararg.ll
index b4597a3214e..80a1c648b78 100644
--- a/llvm/test/CodeGen/Mips/o32_cc_vararg.ll
+++ b/llvm/test/CodeGen/Mips/o32_cc_vararg.ll
@@ -29,9 +29,9 @@ entry:
 
 ; CHECK-LABEL: va1:
 ; CHECK: addiu   $sp, $sp, -16
+; CHECK: sw      $5, 20($sp)
 ; CHECK: sw      $7, 28($sp)
 ; CHECK: sw      $6, 24($sp)
-; CHECK: sw      $5, 20($sp)
 ; CHECK: lw      $2, 20($sp)
 }
 
@@ -83,8 +83,8 @@ entry:
 
 ; CHECK-LABEL: va3:
 ; CHECK: addiu   $sp, $sp, -16
-; CHECK: sw      $7, 28($sp)
 ; CHECK: sw      $6, 24($sp)
+; CHECK: sw      $7, 28($sp)
 ; CHECK: lw      $2, 24($sp)
 }
 
diff --git a/llvm/test/CodeGen/PowerPC/anon_aggr.ll b/llvm/test/CodeGen/PowerPC/anon_aggr.ll
index f4e788849ec..9b32a8f55f3 100644
--- a/llvm/test/CodeGen/PowerPC/anon_aggr.ll
+++ b/llvm/test/CodeGen/PowerPC/anon_aggr.ll
@@ -60,33 +60,34 @@ equal:
 unequal:
   ret i8* %array2_ptr
 }
-
 ; CHECK-LABEL: func2:
-; CHECK: ld [[REG2:[0-9]+]], 72(1)
-; CHECK: cmpld {{([0-9]+,)?}}4, [[REG2]]
-; CHECK-DAG: std [[REG2]], -[[OFFSET1:[0-9]+]]
+; CHECK: cmpld {{([0-9]+,)?}}4, 6
+; CHECK-DAG: std 6, 72(1)
+; CHECK-DAG: std 5, 64(1)
+; CHECK-DAG: std 6, -[[OFFSET1:[0-9]+]]
 ; CHECK-DAG: std 4, -[[OFFSET2:[0-9]+]]
 ; CHECK: ld 3, -[[OFFSET2]](1)
 ; CHECK: ld 3, -[[OFFSET1]](1)
 
-; DARWIN32: _func2:
-; DARWIN32: addi r[[REG1:[0-9]+]], r[[REGSP:[0-9]+]], 36
-; DARWIN32: lwz r[[REG2:[0-9]+]], 44(r[[REGSP]])
+; DARWIN32-LABEL: _func2
+; DARWIN32-DAG: addi r[[REG8:[0-9]+]], r[[REGSP:[0-9]+]], 36
+; DARWIN32-DAG: lwz r[[REG2:[0-9]+]], 44(r[[REGSP]])
 ; DARWIN32: mr
-; DARWIN32: mr r[[REG3:[0-9]+]], r[[REGA:[0-9]+]]
-; DARWIN32: cmplw {{(cr[0-9]+,)?}}r[[REGA]], r[[REG2]]
-; DARWIN32: stw r[[REG3]], -[[OFFSET1:[0-9]+]]
-; DARWIN32: stw r[[REG2]], -[[OFFSET2:[0-9]+]]
-; DARWIN32: lwz r3, -[[OFFSET1]]
-; DARWIN32: lwz r3, -[[OFFSET2]]
+; DARWIN32: mr r[[REG7:[0-9]+]], r5
+; DARWIN32-DAG: cmplw {{(cr[0-9]+,)?}}r5, r[[REG2]]
+; DARWIN32-DAG: stw r[[REG7]], -[[OFFSET1:[0-9]+]]
+; DARWIN32-DAG: stw r[[REG2]], -[[OFFSET2:[0-9]+]]
+; DARWIN32-DAG: lwz r3, -[[OFFSET1]]
+; DARWIN32-DAG: lwz r3, -[[OFFSET2]]
+
 
 ; DARWIN64: _func2:
 ; DARWIN64: ld r[[REG2:[0-9]+]], 72(r1)
 ; DARWIN64: mr
 ; DARWIN64: mr r[[REG3:[0-9]+]], r[[REGA:[0-9]+]]
 ; DARWIN64: cmpld {{(cr[0-9]+,)?}}r[[REGA]], r[[REG2]]
-; DARWIN64: std r[[REG3]], -[[OFFSET1:[0-9]+]]
 ; DARWIN64: std r[[REG2]], -[[OFFSET2:[0-9]+]]
+; DARWIN64: std r[[REG3]], -[[OFFSET1:[0-9]+]]
 ; DARWIN64: ld r3, -[[OFFSET1]]
 ; DARWIN64: ld r3, -[[OFFSET2]]
 
@@ -106,24 +107,24 @@ unequal:
 }
 
 ; CHECK-LABEL: func3:
-; CHECK: ld [[REG3:[0-9]+]], 72(1)
-; CHECK: ld [[REG4:[0-9]+]], 56(1)
-; CHECK: cmpld {{([0-9]+,)?}}[[REG4]], [[REG3]]
-; CHECK: std [[REG3]], -[[OFFSET1:[0-9]+]](1)
-; CHECK: std [[REG4]], -[[OFFSET2:[0-9]+]](1)
+; CHECK: cmpld {{([0-9]+,)?}}4, 6
+; CHECK-DAG: std 4, -[[OFFSET2:[0-9]+]](1)
+; CHECK-DAG: std 6, -[[OFFSET1:[0-9]+]](1)
 ; CHECK: ld 3, -[[OFFSET2]](1)
 ; CHECK: ld 3, -[[OFFSET1]](1)
 
-; DARWIN32: _func3:
-; DARWIN32: addi r[[REG1:[0-9]+]], r[[REGSP:[0-9]+]], 36
-; DARWIN32: addi r[[REG2:[0-9]+]], r[[REGSP]], 24
-; DARWIN32: lwz r[[REG3:[0-9]+]], 44(r[[REGSP]])
-; DARWIN32: lwz r[[REG4:[0-9]+]], 32(r[[REGSP]])
-; DARWIN32: cmplw {{(cr[0-9]+,)?}}r[[REG4]], r[[REG3]]
-; DARWIN32: stw r[[REG3]], -[[OFFSET1:[0-9]+]]
-; DARWIN32: stw r[[REG4]], -[[OFFSET2:[0-9]+]]
-; DARWIN32: lwz r3, -[[OFFSET2]]
-; DARWIN32: lwz r3, -[[OFFSET1]]
+; DARWIN32-LABEL: _func3:
+; DARWIN32-DAG: stw r[[REG8:[0-9]+]], 44(r[[REGSP:[0-9]+]])
+; DARWIN32-DAG: stw r[[REG5:[0-9]+]], 32(r[[REGSP]])
+; DARWIN32-DAG: addi r[[REG5a:[0-9]+]], r[[REGSP:[0-9]+]], 36
+; DARWIN32-DAG: addi r[[REG8a:[0-9]+]], r[[REGSP]], 24
+; DARWIN32-DAG: lwz r[[REG5a:[0-9]+]], 44(r[[REGSP]])
+; DARWIN32-DAG: lwz r[[REG8a:[0-9]+]], 32(r[[REGSP]])
+; DARWIN32-DAG: cmplw {{(cr[0-9]+,)?}}r[[REG8a]], r[[REG5a]]
+; DARWIN32-DAG: stw r[[REG5a]], -[[OFFSET1:[0-9]+]]
+; DARWIN32-DAG: stw r[[REG8a]], -[[OFFSET2:[0-9]+]]
+; DARWIN32-DAG: lwz r3, -[[OFFSET1:[0-9]+]]
+; DARWIN32-DAG: lwz r3, -[[OFFSET2:[0-9]+]]
 
 ; DARWIN64: _func3:
 ; DARWIN64: ld r[[REG3:[0-9]+]], 72(r1)
diff --git a/llvm/test/CodeGen/PowerPC/complex-return.ll b/llvm/test/CodeGen/PowerPC/complex-return.ll
index f6097e65512..ec87a89b110 100644
--- a/llvm/test/CodeGen/PowerPC/complex-return.ll
+++ b/llvm/test/CodeGen/PowerPC/complex-return.ll
@@ -24,10 +24,10 @@ entry:
 }
 
 ; CHECK-LABEL: foo:
-; CHECK: lfd 1
-; CHECK: lfd 2
-; CHECK: lfd 3
-; CHECK: lfd 4
+; CHECK-DAG: lfd 1
+; CHECK-DAG: lfd 2
+; CHECK-DAG: lfd 3
+; CHECK_DAG: lfd 4
 
 define { float, float } @oof() nounwind {
 entry:
@@ -50,6 +50,6 @@ entry:
 }
 
 ; CHECK-LABEL: oof:
-; CHECK: lfs 2
-; CHECK: lfs 1
+; CHECK-DAG: lfs 2
+; CHECK-DAG: lfs 1
 
diff --git a/llvm/test/CodeGen/PowerPC/jaggedstructs.ll b/llvm/test/CodeGen/PowerPC/jaggedstructs.ll
index b28b34d7814..6128316f45f 100644
--- a/llvm/test/CodeGen/PowerPC/jaggedstructs.ll
+++ b/llvm/test/CodeGen/PowerPC/jaggedstructs.ll
@@ -18,31 +18,31 @@ entry:
   ret void
 }
 
-; CHECK: std 6, 184(1)
-; CHECK: std 5, 176(1)
-; CHECK: std 4, 168(1)
-; CHECK: std 3, 160(1)
-; CHECK: lbz {{[0-9]+}}, 167(1)
-; CHECK: lhz {{[0-9]+}}, 165(1)
-; CHECK: stb {{[0-9]+}}, 55(1)
-; CHECK: sth {{[0-9]+}}, 53(1)
-; CHECK: lbz {{[0-9]+}}, 175(1)
-; CHECK: lwz {{[0-9]+}}, 171(1)
-; CHECK: stb {{[0-9]+}}, 63(1)
-; CHECK: stw {{[0-9]+}}, 59(1)
-; CHECK: lhz {{[0-9]+}}, 182(1)
-; CHECK: lwz {{[0-9]+}}, 178(1)
-; CHECK: sth {{[0-9]+}}, 70(1)
-; CHECK: stw {{[0-9]+}}, 66(1)
-; CHECK: lbz {{[0-9]+}}, 191(1)
-; CHECK: lhz {{[0-9]+}}, 189(1)
-; CHECK: lwz {{[0-9]+}}, 185(1)
-; CHECK: stb {{[0-9]+}}, 79(1)
-; CHECK: sth {{[0-9]+}}, 77(1)
-; CHECK: stw {{[0-9]+}}, 73(1)
-; CHECK: ld 6, 72(1)
-; CHECK: ld 5, 64(1)
-; CHECK: ld 4, 56(1)
-; CHECK: ld 3, 48(1)
+; CHECK-DAG: std 3, 160(1)
+; CHECK-DAG: std 6, 184(1)
+; CHECK-DAG: std 5, 176(1)
+; CHECK-DAG: std 4, 168(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 167(1)
+; CHECK-DAG: lhz {{[0-9]+}}, 165(1)
+; CHECK-DAG: stb {{[0-9]+}}, 55(1)
+; CHECK-DAG-DAG: sth {{[0-9]+}}, 53(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 175(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 171(1)
+; CHECK-DAG: stb {{[0-9]+}}, 63(1)
+; CHECK-DAG: stw {{[0-9]+}}, 59(1)
+; CHECK-DAG: lhz {{[0-9]+}}, 182(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 178(1)
+; CHECK-DAG: sth {{[0-9]+}}, 70(1)
+; CHECK-DAG: stw {{[0-9]+}}, 66(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 191(1)
+; CHECK-DAG: lhz {{[0-9]+}}, 189(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 185(1)
+; CHECK-DAG: stb {{[0-9]+}}, 79(1)
+; CHECK-DAG: sth {{[0-9]+}}, 77(1)
+; CHECK-DAG: stw {{[0-9]+}}, 73(1)
+; CHECK-DAG: ld 6, 72(1)
+; CHECK-DAG: ld 5, 64(1)
+; CHECK-DAG: ld 4, 56(1)
+; CHECK-DAG: ld 3, 48(1)
 
 declare void @check(%struct.S3* byval, %struct.S5* byval, %struct.S6* byval, %struct.S7* byval)
diff --git a/llvm/test/CodeGen/PowerPC/ppc64-align-long-double.ll b/llvm/test/CodeGen/PowerPC/ppc64-align-long-double.ll
index c3cccd5b293..d59dc64dcf8 100644
--- a/llvm/test/CodeGen/PowerPC/ppc64-align-long-double.ll
+++ b/llvm/test/CodeGen/PowerPC/ppc64-align-long-double.ll
@@ -1,6 +1,6 @@
 ; RUN: llc -verify-machineinstrs -mcpu=pwr7 -O0 -fast-isel=false -mattr=-vsx < %s | FileCheck %s
 ; RUN: llc -verify-machineinstrs -mcpu=pwr7 -O0 -fast-isel=false -mattr=+vsx < %s | FileCheck -check-prefix=CHECK-VSX %s
-; RUN: llc -verify-machineinstrs -mcpu=pwr9 -O0 -fast-isel=false -mattr=+vsx < %s | FileCheck %s
+; RUN: llc -verify-machineinstrs -mcpu=pwr9 -O0 -fast-isel=false -mattr=+vsx < %s | FileCheck -check-prefix=CHECK-P9 %s
 
 ; Verify internal alignment of long double in a struct.  The double
 ; argument comes in in GPR3; GPR4 is skipped; GPRs 5 and 6 contain
@@ -19,19 +19,44 @@ entry:
   ret ppc_fp128 %0
 }
 
+; The additional stores are caused because we forward the value in the
+; store->load->bitcast path to make a store and bitcast of the same
+; value. Since the target does bitcast through memory and we no longer
+; remember the address we need to do the store in a fresh local
+; address. 
+
 ; CHECK-DAG: std 6, 72(1)
 ; CHECK-DAG: std 5, 64(1)
 ; CHECK-DAG: std 4, 56(1)
 ; CHECK-DAG: std 3, 48(1)
-; CHECK: lfd 1, 64(1)
-; CHECK: lfd 2, 72(1)
+
+; CHECK-DAG: std 5, -16(1)
+; CHECK-DAG: std 6, -8(1)
+; CHECK-DAG: lfd 1, -16(1)
+; CHECK-DAG: lfd 2, -8(1)
+
+; FIXMECHECK: lfd 1, 64(1)
+; FIXMECHECK: lfd 2, 72(1)
 
 ; CHECK-VSX-DAG: std 6, 72(1)
 ; CHECK-VSX-DAG: std 5, 64(1)
 ; CHECK-VSX-DAG: std 4, 56(1)
 ; CHECK-VSX-DAG: std 3, 48(1)
-; CHECK-VSX: li 3, 16
-; CHECK-VSX: addi 4, 1, 48
-; CHECK-VSX: lxsdx 1, 4, 3
-; CHECK-VSX: li 3, 24
-; CHECK-VSX: lxsdx 2, 4, 3
+; CHECK-VSX-DAG: std 5, -16(1)
+; CHECK-VSX-DAG: std 6, -8(1)
+; CHECK-VSX: addi 3, 1, -16
+; CHECK-VSX: lxsdx 1, 0, 3
+; CHECK-VSX: addi 3, 1, -8
+; CHECK-VSX: lxsdx 2, 0, 3
+
+; FIXME-VSX: addi 4, 1, 48
+; FIXME-VSX: lxsdx 1, 4, 3
+; FIXME-VSX: li 3, 24
+; FIXME-VSX: lxsdx 2, 4, 3
+
+; CHECK-P9: std 6, 72(1)
+; CHECK-P9: std 5, 64(1)
+; CHECK-P9: std 4, 56(1)
+; CHECK-P9: std 3, 48(1)
+; CHECK-P9: mtvsrd 1, 5
+; CHECK-P9: mtvsrd 2, 6
diff --git a/llvm/test/CodeGen/PowerPC/structsinmem.ll b/llvm/test/CodeGen/PowerPC/structsinmem.ll
index 3777f3ec5ba..01b0848e707 100644
--- a/llvm/test/CodeGen/PowerPC/structsinmem.ll
+++ b/llvm/test/CodeGen/PowerPC/structsinmem.ll
@@ -113,13 +113,13 @@ entry:
   %add13 = add nsw i32 %add11, %6
   ret i32 %add13
 
-; CHECK: lha {{[0-9]+}}, 126(1)
-; CHECK: lha {{[0-9]+}}, 132(1)
-; CHECK: lbz {{[0-9]+}}, 119(1)
-; CHECK: lwz {{[0-9]+}}, 140(1)
-; CHECK: lwz {{[0-9]+}}, 144(1)
-; CHECK: lwz {{[0-9]+}}, 152(1)
-; CHECK: lwz {{[0-9]+}}, 160(1)
+; CHECK-DAG: lha {{[0-9]+}}, 126(1)
+; CHECK-DAG: lha {{[0-9]+}}, 132(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 119(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 140(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 144(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 152(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 160(1)
 }
 
 define i32 @caller2() nounwind {
@@ -205,11 +205,11 @@ entry:
   %add13 = add nsw i32 %add11, %6
   ret i32 %add13
 
-; CHECK: lha {{[0-9]+}}, 126(1)
-; CHECK: lha {{[0-9]+}}, 133(1)
-; CHECK: lbz {{[0-9]+}}, 119(1)
-; CHECK: lwz {{[0-9]+}}, 140(1)
-; CHECK: lwz {{[0-9]+}}, 147(1)
-; CHECK: lwz {{[0-9]+}}, 154(1)
-; CHECK: lwz {{[0-9]+}}, 161(1)
+; CHECK-DAG: lha {{[0-9]+}}, 126(1)
+; CHECK-DAG: lha {{[0-9]+}}, 133(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 119(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 140(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 147(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 154(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 161(1)
 }
diff --git a/llvm/test/CodeGen/PowerPC/structsinregs.ll b/llvm/test/CodeGen/PowerPC/structsinregs.ll
index e27041dd4c8..54679f259e9 100644
--- a/llvm/test/CodeGen/PowerPC/structsinregs.ll
+++ b/llvm/test/CodeGen/PowerPC/structsinregs.ll
@@ -59,6 +59,7 @@ entry:
   %call = call i32 @callee1(%struct.s1* byval %p1, %struct.s2* byval %p2, %struct.s3* byval %p3, %struct.s4* byval %p4, %struct.s5* byval %p5, %struct.s6* byval %p6, %struct.s7* byval %p7)
   ret i32 %call
 
+; CHECK-LABEL: caller1
 ; CHECK: ld 9, 112(31)
 ; CHECK: ld 8, 120(31)
 ; CHECK: ld 7, 128(31)
@@ -97,20 +98,21 @@ entry:
   %add13 = add nsw i32 %add11, %6
   ret i32 %add13
 
-; CHECK: std 9, 96(1)
-; CHECK: std 8, 88(1)
-; CHECK: std 7, 80(1)
-; CHECK: stw 6, 76(1)
-; CHECK: stw 5, 68(1)
-; CHECK: sth 4, 62(1)
-; CHECK: stb 3, 55(1)
-; CHECK: lha {{[0-9]+}}, 62(1)
-; CHECK: lha {{[0-9]+}}, 68(1)
-; CHECK: lbz {{[0-9]+}}, 55(1)
-; CHECK: lwz {{[0-9]+}}, 76(1)
-; CHECK: lwz {{[0-9]+}}, 80(1)
-; CHECK: lwz {{[0-9]+}}, 88(1)
-; CHECK: lwz {{[0-9]+}}, 96(1)
+; CHECK-LABEL: callee1
+; CHECK-DAG: std 9, 96(1)
+; CHECK-DAG: std 8, 88(1)
+; CHECK-DAG: std 7, 80(1)
+; CHECK-DAG: stw 6, 76(1)
+; CHECK-DAG: stw 5, 68(1)
+; CHECK-DAG: sth 4, 62(1)
+; CHECK-DAG: stb 3, 55(1)
+; CHECK-DAG: lha {{[0-9]+}}, 62(1)
+; CHECK-DAG: lha {{[0-9]+}}, 68(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 55(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 76(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 80(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 88(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 96(1)
 }
 
 define i32 @caller2() nounwind {
@@ -139,6 +141,7 @@ entry:
   %call = call i32 @callee2(%struct.t1* byval %p1, %struct.t2* byval %p2, %struct.t3* byval %p3, %struct.t4* byval %p4, %struct.t5* byval %p5, %struct.t6* byval %p6, %struct.t7* byval %p7)
   ret i32 %call
 
+; CHECK-LABEL: caller2
 ; CHECK: stb {{[0-9]+}}, 71(1)
 ; CHECK: sth {{[0-9]+}}, 69(1)
 ; CHECK: stb {{[0-9]+}}, 87(1)
@@ -184,18 +187,19 @@ entry:
   %add13 = add nsw i32 %add11, %6
   ret i32 %add13
 
-; CHECK: std 9, 96(1)
-; CHECK: std 8, 88(1)
-; CHECK: std 7, 80(1)
-; CHECK: stw 6, 76(1)
-; CHECK: std 5, 64(1)
-; CHECK: sth 4, 62(1)
-; CHECK: stb 3, 55(1)
-; CHECK: lha {{[0-9]+}}, 62(1)
-; CHECK: lha {{[0-9]+}}, 69(1)
-; CHECK: lbz {{[0-9]+}}, 55(1)
-; CHECK: lwz {{[0-9]+}}, 76(1)
-; CHECK: lwz {{[0-9]+}}, 83(1)
-; CHECK: lwz {{[0-9]+}}, 90(1)
-; CHECK: lwz {{[0-9]+}}, 97(1)
+; CHECK-LABEL: callee2
+; CHECK-DAG: std 9, 96(1)
+; CHECK-DAG: std 8, 88(1)
+; CHECK-DAG: std 7, 80(1)
+; CHECK-DAG: stw 6, 76(1)
+; CHECK-DAG: std 5, 64(1)
+; CHECK-DAG: sth 4, 62(1)
+; CHECK-DAG: stb 3, 55(1)
+; CHECK-DAG: lha {{[0-9]+}}, 62(1)
+; CHECK-DAG: lha {{[0-9]+}}, 69(1)
+; CHECK-DAG: lbz {{[0-9]+}}, 55(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 76(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 83(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 90(1)
+; CHECK-DAG: lwz {{[0-9]+}}, 97(1)
 }
diff --git a/llvm/test/CodeGen/SystemZ/unaligned-01.ll b/llvm/test/CodeGen/SystemZ/unaligned-01.ll
index 94cad0e1743..2af1aa79a23 100644
--- a/llvm/test/CodeGen/SystemZ/unaligned-01.ll
+++ b/llvm/test/CodeGen/SystemZ/unaligned-01.ll
@@ -1,10 +1,7 @@
 ; Check that unaligned accesses are allowed in general.  We check the
 ; few exceptions (like CRL) in their respective test files.
 ;
-; FIXME: -combiner-alias-analysis (the default for SystemZ) stops
-;        f1 from being optimized.
-; RUN: llc < %s -mtriple=s390x-linux-gnu -combiner-alias-analysis=false \
-; RUN:   | FileCheck %s
+; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s
 
 ; Check that these four byte stores become a single word store.
 define void @f1(i8 *%ptr) {
diff --git a/llvm/test/CodeGen/Thumb/2010-07-15-debugOrdering.ll b/llvm/test/CodeGen/Thumb/2010-07-15-debugOrdering.ll
index 2f8e36b66b8..08349a31dfa 100644
--- a/llvm/test/CodeGen/Thumb/2010-07-15-debugOrdering.ll
+++ b/llvm/test/CodeGen/Thumb/2010-07-15-debugOrdering.ll
@@ -9,9 +9,9 @@
 
 define void @_Z19getClosestDiagonal3ii(%0* noalias sret, i32, i32) nounwind {
 ; CHECK: bl ___muldf3
-; CHECK: bl ___muldf3
 ; CHECK: beq LBB0
 ; CHECK: bl ___muldf3
+; CHECK: bl ___muldf3
 ; <label>:3
   switch i32 %1, label %4 [
     i32 0, label %5
diff --git a/llvm/test/CodeGen/Thumb/stack-access.ll b/llvm/test/CodeGen/Thumb/stack-access.ll
index fded4104207..44217aba62d 100644
--- a/llvm/test/CodeGen/Thumb/stack-access.ll
+++ b/llvm/test/CodeGen/Thumb/stack-access.ll
@@ -74,15 +74,17 @@ define zeroext i16 @test6() {
 }
 
 ; Accessing the bottom of a large array shouldn't require materializing a base
+; 
+; CHECK: movs [[REG:r[0-9]+]], #1
+; CHECK: str [[REG]], [sp, #16]
+; CHECK: str [[REG]], [sp, #4]
+
 define void @test7() {
   %arr = alloca [200 x i32], align 4
 
-  ; CHECK: movs [[REG:r[0-9]+]], #1
-  ; CHECK: str [[REG]], [sp, #4]
   %arrayidx = getelementptr inbounds [200 x i32], [200 x i32]* %arr, i32 0, i32 1
   store i32 1, i32* %arrayidx, align 4
 
-  ; CHECK: str [[REG]], [sp, #16]
   %arrayidx1 = getelementptr inbounds [200 x i32], [200 x i32]* %arr, i32 0, i32 4
   store i32 1, i32* %arrayidx1, align 4
 
@@ -96,30 +98,36 @@ define void @test8() {
   %arr1 = alloca [224 x i32], align 4
 
 ; CHECK: movs [[REG:r[0-9]+]], #1
-; CHECK: str [[REG]], [sp]
+; CHECK-DAG: str [[REG]], [sp]
   %arr1idx1 = getelementptr inbounds [224 x i32], [224 x i32]* %arr1, i32 0, i32 0
   store i32 1, i32* %arr1idx1, align 4
 
 ; Offset in range for sp-based store, but not for non-sp-based store
-; CHECK: str [[REG]], [sp, #128]
+; CHECK-DAG: str [[REG]], [sp, #128]
   %arr1idx2 = getelementptr inbounds [224 x i32], [224 x i32]* %arr1, i32 0, i32 32
   store i32 1, i32* %arr1idx2, align 4
 
-; CHECK: str [[REG]], [sp, #896]
+; CHECK-DAG: str [[REG]], [sp, #896]
   %arr2idx1 = getelementptr inbounds [224 x i32], [224 x i32]* %arr2, i32 0, i32 0
   store i32 1, i32* %arr2idx1, align 4
 
 ; %arr2 is in range, but this element of it is not
-; CHECK: str [[REG]], [{{r[0-9]+}}]
+; CHECK-DAG: ldr [[RA:r[0-9]+]], .LCPI7_2
+; CHECK-DAG: add [[RA]], sp
+; CHECK-DAG: str [[REG]], [{{r[0-9]+}}]
   %arr2idx2 = getelementptr inbounds [224 x i32], [224 x i32]* %arr2, i32 0, i32 32
   store i32 1, i32* %arr2idx2, align 4
 
 ; %arr3 is not in range
-; CHECK: str [[REG]], [{{r[0-9]+}}]
+; CHECK-DAG: ldr [[RB:r[0-9]+]], .LCPI7_3
+; CHECK-DAG: add [[RB]], sp
+; CHECK-DAG: str [[REG]], [{{r[0-9]+}}]
   %arr3idx1 = getelementptr inbounds [224 x i32], [224 x i32]* %arr3, i32 0, i32 0
   store i32 1, i32* %arr3idx1, align 4
 
-; CHECK: str [[REG]], [{{r[0-9]+}}]
+; CHECK-DAG: ldr [[RC:r[0-9]+]], .LCPI7_4
+; CHECK-DAG: add [[RC]], sp
+; CHECK-DAG: str [[REG]], [{{r[0-9]+}}]
   %arr3idx2 = getelementptr inbounds [224 x i32], [224 x i32]* %arr3, i32 0, i32 32
   store i32 1, i32* %arr3idx2, align 4
 
diff --git a/llvm/test/CodeGen/X86/2010-09-17-SideEffectsInChain.ll b/llvm/test/CodeGen/X86/2010-09-17-SideEffectsInChain.ll
index b7380196bd9..54a7763eb69 100644
--- a/llvm/test/CodeGen/X86/2010-09-17-SideEffectsInChain.ll
+++ b/llvm/test/CodeGen/X86/2010-09-17-SideEffectsInChain.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -combiner-alias-analysis -march=x86-64 -mcpu=core2 | FileCheck %s
+; RUN: llc < %s -march=x86-64 -mcpu=core2 | FileCheck %s
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
 target triple = "x86_64-apple-darwin10.4"
diff --git a/llvm/test/CodeGen/X86/2012-11-28-merge-store-alias.ll b/llvm/test/CodeGen/X86/2012-11-28-merge-store-alias.ll
index c16deeff3d9..2e8206a7591 100644
--- a/llvm/test/CodeGen/X86/2012-11-28-merge-store-alias.ll
+++ b/llvm/test/CodeGen/X86/2012-11-28-merge-store-alias.ll
@@ -3,8 +3,8 @@
 ; CHECK: merge_stores_can
 ; CHECK: callq foo
 ; CHECK: xorps %xmm0, %xmm0
-; CHECK-NEXT: movl 36(%rsp), %ebp
 ; CHECK-NEXT: movups  %xmm0
+; CHECK-NEXT: movl 36(%rsp), %ebp
 ; CHECK: callq foo
 ; CHECK: ret
 declare i32 @foo([10 x i32]* )
diff --git a/llvm/test/CodeGen/X86/MergeConsecutiveStores.ll b/llvm/test/CodeGen/X86/MergeConsecutiveStores.ll
index 4d7cb765d7b..29ecb7838a1 100644
--- a/llvm/test/CodeGen/X86/MergeConsecutiveStores.ll
+++ b/llvm/test/CodeGen/X86/MergeConsecutiveStores.ll
@@ -111,8 +111,7 @@ define void @merge_const_store_vec(i32 %count, %struct.B* nocapture %p) nounwind
 ; CHECK-LABEL: merge_nonconst_store:
 ; CHECK: movl $67305985
 ; CHECK: movb
-; CHECK: movb
-; CHECK: movb
+; CHECK: movw
 ; CHECK: movb
 ; CHECK: ret
 define void @merge_nonconst_store(i32 %count, i8 %zz, %struct.A* nocapture %p) nounwind uwtable noinline ssp {
@@ -292,16 +291,12 @@ block4:                                       ; preds = %4, %.lr.ph
   ret void
 }
 
-;; On x86, even unaligned copies should be merged to vector ops.
-;; TODO: however, this cannot happen at the moment, due to brokenness
-;; in MergeConsecutiveStores. See UseAA FIXME in DAGCombiner.cpp
-;; visitSTORE.
-
+;; On x86, even unaligned copies can be merged to vector ops.
 ; CHECK-LABEL: merge_loads_no_align:
 ;  load:
-; CHECK-NOT: vmovups ;; TODO
+; CHECK: vmovups
 ;  store:
-; CHECK-NOT: vmovups ;; TODO
+; CHECK: vmovups
 ; CHECK: ret
 define void @merge_loads_no_align(i32 %count, %struct.B* noalias nocapture %q, %struct.B* noalias nocapture %p) nounwind uwtable noinline ssp {
   %a1 = icmp sgt i32 %count, 0
@@ -583,8 +578,8 @@ define void @merge_vec_element_and_scalar_load([6 x i64]* %array) {
 
 ; CHECK-LABEL: merge_vec_element_and_scalar_load
 ; CHECK:      movq	(%rdi), %rax
+; CHECK-NEXT: movq	8(%rdi), %rcx
 ; CHECK-NEXT: movq	%rax, 32(%rdi)
-; CHECK-NEXT: movq	8(%rdi), %rax
-; CHECK-NEXT: movq	%rax, 40(%rdi)
+; CHECK-NEXT: movq	%rcx, 40(%rdi)
 ; CHECK-NEXT: retq
 }
diff --git a/llvm/test/CodeGen/X86/avx512-mask-op.ll b/llvm/test/CodeGen/X86/avx512-mask-op.ll
index 41fb19f38e0..7e36828bffb 100644
--- a/llvm/test/CodeGen/X86/avx512-mask-op.ll
+++ b/llvm/test/CodeGen/X86/avx512-mask-op.ll
@@ -1169,10 +1169,6 @@ define void @ktest_2(<32 x float> %in, float * %base) {
 ; KNL-NEXT:    kmovw %k0, %eax
 ; KNL-NEXT:    vpinsrb $15, %eax, %xmm2, %xmm2
 ; KNL-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
-; KNL-NEXT:    vpsllw $7, %ymm2, %ymm2
-; KNL-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
-; KNL-NEXT:    vpxor %ymm3, %ymm3, %ymm3
-; KNL-NEXT:    vpcmpgtb %ymm2, %ymm3, %ymm2
 ; KNL-NEXT:    vmovups 4(%rdi), %zmm3 {%k2} {z}
 ; KNL-NEXT:    vmovups 68(%rdi), %zmm4 {%k1} {z}
 ; KNL-NEXT:    vcmpltps %zmm4, %zmm1, %k0
diff --git a/llvm/test/CodeGen/X86/chain_order.ll b/llvm/test/CodeGen/X86/chain_order.ll
index 8c3aa6e1515..cc48e5b6149 100644
--- a/llvm/test/CodeGen/X86/chain_order.ll
+++ b/llvm/test/CodeGen/X86/chain_order.ll
@@ -11,9 +11,9 @@ define void @cftx020(double* nocapture %a) {
 ; CHECK-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
 ; CHECK-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vmovupd (%rdi), %xmm1
-; CHECK-NEXT:    vsubpd 16(%rdi), %xmm1, %xmm1
 ; CHECK-NEXT:    vmovupd %xmm0, (%rdi)
-; CHECK-NEXT:    vmovupd %xmm1, 16(%rdi)
+; CHECK-NEXT:    vsubpd 16(%rdi), %xmm1, %xmm0
+; CHECK-NEXT:    vmovupd %xmm0, 16(%rdi)
 ; CHECK-NEXT:    retq
 entry:
   %0 = load double, double* %a, align 8
diff --git a/llvm/test/CodeGen/X86/clear_upper_vector_element_bits.ll b/llvm/test/CodeGen/X86/clear_upper_vector_element_bits.ll
index 6d50048ed7c..1956e8fa762 100644
--- a/llvm/test/CodeGen/X86/clear_upper_vector_element_bits.ll
+++ b/llvm/test/CodeGen/X86/clear_upper_vector_element_bits.ll
@@ -137,47 +137,47 @@ define <16 x i8> @_clearupper16xi8a(<16 x i8>) nounwind {
 ; SSE-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
 ; SSE-NEXT:    movd %eax, %xmm0
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %r9d
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %edx
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %esi
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %r8d
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %edi
 ; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
 ; SSE-NEXT:    movd %eax, %xmm1
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE-NEXT:    movd %esi, %xmm0
 ; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %esi
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %ecx
-; SSE-NEXT:    movd %ecx, %xmm2
+; SSE-NEXT:    movd %eax, %xmm0
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm2
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; SSE-NEXT:    movd %edx, %xmm0
-; SSE-NEXT:    movd %esi, %xmm1
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE-NEXT:    movd %edi, %xmm0
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %ecx
-; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %edx
-; SSE-NEXT:    movd %edx, %xmm3
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm0
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm3
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3],xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
-; SSE-NEXT:    movd %r9d, %xmm0
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm0
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
 ; SSE-NEXT:    movd %eax, %xmm1
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE-NEXT:    movd %r8d, %xmm0
-; SSE-NEXT:    movd %ecx, %xmm2
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
 ; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE-NEXT:    movd {{.*#+}} xmm4 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
+; SSE-NEXT:    movd {{.*#+}} xmm3 = mem[0],zero,zero,zero
 ; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm2
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm3
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3],xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm2
+; SSE-NEXT:    movzbl -{{[0-9]+}}(%rsp), %eax
+; SSE-NEXT:    movd %eax, %xmm4
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1],xmm4[2],xmm2[2],xmm4[3],xmm2[3],xmm4[4],xmm2[4],xmm4[5],xmm2[5],xmm4[6],xmm2[6],xmm4[7],xmm2[7]
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3],xmm4[4],xmm3[4],xmm4[5],xmm3[5],xmm4[6],xmm3[6],xmm4[7],xmm3[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
 ; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
 ; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/combiner-aa-0.ll b/llvm/test/CodeGen/X86/combiner-aa-0.ll
deleted file mode 100644
index 403059d90ab..00000000000
--- a/llvm/test/CodeGen/X86/combiner-aa-0.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llc < %s -march=x86-64 -combiner-global-alias-analysis -combiner-alias-analysis
-
-target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
-	%struct.Hash_Key = type { [4 x i32], i32 }
-@g_flipV_hashkey = external global %struct.Hash_Key, align 16		; <%struct.Hash_Key*> [#uses=1]
-
-define void @foo() nounwind {
-	%t0 = load i32, i32* undef, align 16		; <i32> [#uses=1]
-	%t1 = load i32, i32* null, align 4		; <i32> [#uses=1]
-	%t2 = srem i32 %t0, 32		; <i32> [#uses=1]
-	%t3 = shl i32 1, %t2		; <i32> [#uses=1]
-	%t4 = xor i32 %t3, %t1		; <i32> [#uses=1]
-	store i32 %t4, i32* null, align 4
-	%t5 = getelementptr %struct.Hash_Key, %struct.Hash_Key* @g_flipV_hashkey, i64 0, i32 0, i64 0		; <i32*> [#uses=2]
-	%t6 = load i32, i32* %t5, align 4		; <i32> [#uses=1]
-	%t7 = shl i32 1, undef		; <i32> [#uses=1]
-	%t8 = xor i32 %t7, %t6		; <i32> [#uses=1]
-	store i32 %t8, i32* %t5, align 4
-	unreachable
-}
diff --git a/llvm/test/CodeGen/X86/combiner-aa-1.ll b/llvm/test/CodeGen/X86/combiner-aa-1.ll
deleted file mode 100644
index cc3e5ca1260..00000000000
--- a/llvm/test/CodeGen/X86/combiner-aa-1.ll
+++ /dev/null
@@ -1,23 +0,0 @@
-; RUN: llc < %s --combiner-alias-analysis --combiner-global-alias-analysis
-; PR4880
-
-target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32"
-target triple = "i386-pc-linux-gnu"
-
-%struct.alst_node = type { %struct.node }
-%struct.arg_node = type { %struct.node, i8*, %struct.alst_node* }
-%struct.arglst_node = type { %struct.alst_node, %struct.arg_node*, %struct.arglst_node* }
-%struct.lam_node = type { %struct.alst_node, %struct.arg_node*, %struct.alst_node* }
-%struct.node = type { i32 (...)**, %struct.node* }
-
-define i32 @._ZN8lam_node18resolve_name_clashEP8arg_nodeP9alst_node._ZNK8lam_nodeeqERK8exp_node._ZN11arglst_nodeD0Ev(%struct.lam_node* %this.this, %struct.arg_node* %outer_arg, %struct.alst_node* %env.cmp, %struct.arglst_node* %this, i32 %functionID) {
-comb_entry:
-  %.SV59 = alloca %struct.node*                   ; <%struct.node**> [#uses=1]
-  %0 = load i32 (...)**, i32 (...)*** null, align 4            ; <i32 (...)**> [#uses=1]
-  %1 = getelementptr inbounds i32 (...)*, i32 (...)** %0, i32 3 ; <i32 (...)**> [#uses=1]
-  %2 = load i32 (...)*, i32 (...)** %1, align 4               ; <i32 (...)*> [#uses=1]
-  store %struct.node* undef, %struct.node** %.SV59
-  %3 = bitcast i32 (...)* %2 to i32 (%struct.node*)* ; <i32 (%struct.node*)*> [#uses=1]
-  %4 = tail call i32 %3(%struct.node* undef)      ; <i32> [#uses=0]
-  unreachable
-}
diff --git a/llvm/test/CodeGen/X86/copy-eflags.ll b/llvm/test/CodeGen/X86/copy-eflags.ll
index 796c1ecd8c7..d98d8a7839b 100644
--- a/llvm/test/CodeGen/X86/copy-eflags.ll
+++ b/llvm/test/CodeGen/X86/copy-eflags.ll
@@ -9,19 +9,22 @@ target triple = "i686-unknown-linux-gnu"
 @.str = private unnamed_addr constant [4 x i8] c"%d\0A\00", align 1
 
 ; CHECK-LABEL: func:
-; This tests whether eax is properly saved/restored around the lahf/sahf
-; instruction sequences.
+; This tests whether eax is properly saved/restored around the
+; lahf/sahf instruction sequences. We make mem op volatile to prevent
+; their reordering to avoid spills.
+
+
 define i32 @func() {
 entry:
   %bval = load i8, i8* @b
   %inc = add i8 %bval, 1
-  store i8 %inc, i8* @b
-  %cval = load i32, i32* @c
+  store volatile i8 %inc, i8* @b
+  %cval = load volatile i32, i32* @c
   %inc1 = add nsw i32 %cval, 1
-  store i32 %inc1, i32* @c
-  %aval = load i8, i8* @a
+  store volatile i32 %inc1, i32* @c
+  %aval = load volatile i8, i8* @a
   %inc2 = add i8 %aval, 1
-  store i8 %inc2, i8* @a
+  store volatile i8 %inc2, i8* @a
 ; Copy flags produced by the incb of %inc1 to a register, need to save+restore
 ; eax around it. The flags will be reused by %tobool.
 ; CHECK: pushl %eax
diff --git a/llvm/test/CodeGen/X86/dag-merge-fast-accesses.ll b/llvm/test/CodeGen/X86/dag-merge-fast-accesses.ll
index 867881d83d3..e5dfccb278c 100644
--- a/llvm/test/CodeGen/X86/dag-merge-fast-accesses.ll
+++ b/llvm/test/CodeGen/X86/dag-merge-fast-accesses.ll
@@ -51,19 +51,11 @@ define void @merge_vec_element_store(<4 x double> %v, double* %ptr) {
 }
 
 
-;; TODO: FAST *should* be:
-;;    movups (%rdi), %xmm0
-;;    movups %xmm0, 40(%rdi)
-;; ..but is not currently. See the UseAA FIXME in DAGCombiner.cpp
-;; visitSTORE.
-
 define void @merge_vec_load_and_stores(i64 *%ptr) {
 ; FAST-LABEL: merge_vec_load_and_stores:
 ; FAST:       # BB#0:
-; FAST-NEXT:    movq (%rdi), %rax
-; FAST-NEXT:    movq 8(%rdi), %rcx
-; FAST-NEXT:    movq %rax, 40(%rdi)
-; FAST-NEXT:    movq %rcx, 48(%rdi)
+; FAST-NEXT:    movups (%rdi), %xmm0
+; FAST-NEXT:    movups %xmm0, 40(%rdi)
 ; FAST-NEXT:    retq
 ;
 ; SLOW-LABEL: merge_vec_load_and_stores:
diff --git a/llvm/test/CodeGen/X86/dont-trunc-store-double-to-float.ll b/llvm/test/CodeGen/X86/dont-trunc-store-double-to-float.ll
index 8a334d21631..05245d0d9e1 100644
--- a/llvm/test/CodeGen/X86/dont-trunc-store-double-to-float.ll
+++ b/llvm/test/CodeGen/X86/dont-trunc-store-double-to-float.ll
@@ -1,9 +1,9 @@
 ; RUN: llc -march=x86 < %s | FileCheck %s
 
 ; CHECK-LABEL: @bar
-; CHECK: movl $1074339512,
-; CHECK: movl $1374389535,
-; CHECK: movl $1078523331,
+; CHECK-DAG: movl $1074339512,
+; CHECK-DAG: movl $1374389535,
+; CHECK-DAG: movl $1078523331,
 define void @bar() unnamed_addr {
 entry-block:
   %a = alloca double
diff --git a/llvm/test/CodeGen/X86/extractelement-legalization-store-ordering.ll b/llvm/test/CodeGen/X86/extractelement-legalization-store-ordering.ll
index c418e67ecb6..5d5cbc76f92 100644
--- a/llvm/test/CodeGen/X86/extractelement-legalization-store-ordering.ll
+++ b/llvm/test/CodeGen/X86/extractelement-legalization-store-ordering.ll
@@ -16,19 +16,20 @@ target datalayout = "e-m:o-p:32:32-f64:32:64-f80:128-n8:16:32-S128"
 ; CHECK-NEXT: movl 20(%esp), %edx
 ; CHECK-NEXT: paddd (%edx), %xmm0
 ; CHECK-NEXT: movdqa %xmm0, (%edx)
-; CHECK-NEXT: movl (%edx), %esi
-; CHECK-NEXT: movl 12(%edx), %edi
-; CHECK-NEXT: movl 8(%edx), %ebx
-; CHECK-NEXT: movl 4(%edx), %edx
-; CHECK-NEXT: shll	$4, %ecx
+; CHECK-NEXT:	movl	(%edx), %esi
+; CHECK-NEXT:	movl	4(%edx), %edi
+; CHECK-NEXT:	shll	$4, %ecx
+; CHECK-NEXT:	movl	8(%edx), %ebx
+; CHECK-NEXT:	movl	12(%edx), %edx
 ; CHECK-NEXT: movl %esi, 12(%eax,%ecx)
-; CHECK-NEXT: movl %edx, (%eax,%ecx)
+; CHECK-NEXT: movl %edi, (%eax,%ecx)
 ; CHECK-NEXT: movl %ebx, 8(%eax,%ecx)
-; CHECK-NEXT: movl %edi, 4(%eax,%ecx)
+; CHECK-NEXT: movl %edx, 4(%eax,%ecx)
 ; CHECK-NEXT: popl %esi
 ; CHECK-NEXT: popl %edi
 ; CHECK-NEXT: popl %ebx
 ; CHECK-NEXT: retl
+
 define void @test_extractelement_legalization_storereuse(<4 x i32> %a, i32* nocapture %x, i32* nocapture readonly %y, i32 %i) #0 {
 entry:
   %0 = bitcast i32* %y to <4 x i32>*
diff --git a/llvm/test/CodeGen/X86/i256-add.ll b/llvm/test/CodeGen/X86/i256-add.ll
index 6164d898ca1..d2821dc3dce 100644
--- a/llvm/test/CodeGen/X86/i256-add.ll
+++ b/llvm/test/CodeGen/X86/i256-add.ll
@@ -2,17 +2,17 @@
 ; RUN: grep adcl %t | count 7
 ; RUN: grep sbbl %t | count 7
 
-define void @add(i256* %p, i256* %q) nounwind {
+define void @add(i256* %p, i256* %q, i256* %r) nounwind {
   %a = load i256, i256* %p
   %b = load i256, i256* %q
   %c = add i256 %a, %b
-  store i256 %c, i256* %p
+  store i256 %c, i256* %r
   ret void
 }
-define void @sub(i256* %p, i256* %q) nounwind {
+define void @sub(i256* %p, i256* %q, i256* %r) nounwind {
   %a = load i256, i256* %p
   %b = load i256, i256* %q
   %c = sub i256 %a, %b
-  store i256 %c, i256* %p
+  store i256 %c, i256* %r
   ret void
 }
diff --git a/llvm/test/CodeGen/X86/i386-shrink-wrapping.ll b/llvm/test/CodeGen/X86/i386-shrink-wrapping.ll
index 2c3e384b70a..d4e099ac655 100644
--- a/llvm/test/CodeGen/X86/i386-shrink-wrapping.ll
+++ b/llvm/test/CodeGen/X86/i386-shrink-wrapping.ll
@@ -55,8 +55,7 @@ target triple = "i386-apple-macosx10.5"
 ;
 ; CHECK-NEXT: L_e$non_lazy_ptr, [[E:%[a-z]+]]
 ; CHECK-NEXT: movb [[D]], ([[E]])
-; CHECK-NEXT: L_f$non_lazy_ptr, [[F:%[a-z]+]]
-; CHECK-NEXT: movsbl ([[F]]), [[CONV:%[a-z]+]]
+; CHECK-NEXT: movsbl ([[E]]), [[CONV:%[a-z]+]]
 ; CHECK-NEXT: movl $6, [[CONV:%[a-z]+]]
 ; The eflags is used in the next instruction.
 ; If that instruction disappear, we are not exercising the bug
@@ -96,7 +95,7 @@ for.end:                                          ; preds = %for.cond.preheader
   %.b3 = load i1, i1* @d, align 1
   %tmp2 = select i1 %.b3, i8 0, i8 6
   store i8 %tmp2, i8* @e, align 1
-  %tmp3 = load i8, i8* @f, align 1
+  %tmp3 = load i8, i8* @e, align 1
   %conv = sext i8 %tmp3 to i32
   %add = add nsw i32 %conv, 1
   %rem = srem i32 %tmp1, %add
diff --git a/llvm/test/CodeGen/X86/live-range-nosubreg.ll b/llvm/test/CodeGen/X86/live-range-nosubreg.ll
index f28d59237b4..899a375221c 100644
--- a/llvm/test/CodeGen/X86/live-range-nosubreg.ll
+++ b/llvm/test/CodeGen/X86/live-range-nosubreg.ll
@@ -1,7 +1,6 @@
-; RUN: llc -march=x86-64 < %s | FileCheck %s
+; RUN: llc -march=x86-64 < %s
 
-; Check for a sane output. This testcase used to crash. See PR29132.
-; CHECK: leal -1
+; This testcase used to crash. See PR29132.
 
 target triple = "x86_64-unknown-linux-gnu"
 
diff --git a/llvm/test/CodeGen/X86/merge-consecutive-loads-128.ll b/llvm/test/CodeGen/X86/merge-consecutive-loads-128.ll
index 003e2e60521..94994655144 100644
--- a/llvm/test/CodeGen/X86/merge-consecutive-loads-128.ll
+++ b/llvm/test/CodeGen/X86/merge-consecutive-loads-128.ll
@@ -1037,12 +1037,12 @@ define <2 x i64> @merge_2i64_i64_12_volatile(i64* %ptr) nounwind uwtable noinlin
 define <4 x float> @merge_4f32_f32_2345_volatile(float* %ptr) nounwind uwtable noinline ssp {
 ; SSE2-LABEL: merge_4f32_f32_2345_volatile:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movss {{.*#+}} xmm3 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSE2-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; SSE2-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; SSE2-NEXT:    retq
 ;
@@ -1065,13 +1065,13 @@ define <4 x float> @merge_4f32_f32_2345_volatile(float* %ptr) nounwind uwtable n
 ; X32-SSE1-LABEL: merge_4f32_f32_2345_volatile:
 ; X32-SSE1:       # BB#0:
 ; X32-SSE1-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X32-SSE1-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X32-SSE1-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; X32-SSE1-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; X32-SSE1-NEXT:    movss {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; X32-SSE1-NEXT:    unpcklps {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; X32-SSE1-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; X32-SSE1-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; X32-SSE1-DAG:    movss 8(%eax), %[[R0:xmm[0-3]]]  # [[R0]] = mem[0],zero,zero,zero
+; X32-SSE1-DAG:    movss 12(%eax), %[[R1:xmm[0-3]]] # [[R1]] = mem[0],zero,zero,zero
+; X32-SSE1-DAG:    movss 16(%eax), %[[R2:xmm[0-3]]] # [[R2]] = mem[0],zero,zero,zero
+; X32-SSE1-DAG:    movss 20(%eax), %[[R3:xmm[0-3]]] # [[R3]] = mem[0],zero,zero,zero
+; X32-SSE1-DAG:    unpcklps %[[R2]], %[[R0]] # [[R0]] = [[R0]][0],[[R2]][0],[[R0]][1],[[R2]][1]
+; X32-SSE1-DAG:    unpcklps %[[R3]], %[[R1]] # [[R1]] = [[R1]][0],[[R3]][0],[[R1]][1],[[R3]][1]
+; X32-SSE1-DAG:    unpcklps %[[R1]], %[[R0]] # [[R0]] = [[R0]][0],[[R1]][0],[[R0]][1],[[R1]][1]
 ; X32-SSE1-NEXT:    retl
 ;
 ; X32-SSE41-LABEL: merge_4f32_f32_2345_volatile:
diff --git a/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll b/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll
index c18a98e05d1..42650af2cf0 100644
--- a/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll
+++ b/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll
@@ -668,10 +668,10 @@ define <16 x i16> @merge_16i16_i16_0uu3zzuuuuuzCuEF_volatile(i16* %ptr) nounwind
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vpinsrw $0, (%rdi), %xmm0, %xmm1
-; AVX1-NEXT:    vpinsrw $3, 6(%rdi), %xmm1, %xmm1
 ; AVX1-NEXT:    vpinsrw $4, 24(%rdi), %xmm0, %xmm0
 ; AVX1-NEXT:    vpinsrw $6, 28(%rdi), %xmm0, %xmm0
 ; AVX1-NEXT:    vpinsrw $7, 30(%rdi), %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $3, 6(%rdi), %xmm1, %xmm1
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
@@ -679,10 +679,10 @@ define <16 x i16> @merge_16i16_i16_0uu3zzuuuuuzCuEF_volatile(i16* %ptr) nounwind
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpxor %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vpinsrw $0, (%rdi), %xmm0, %xmm1
-; AVX2-NEXT:    vpinsrw $3, 6(%rdi), %xmm1, %xmm1
 ; AVX2-NEXT:    vpinsrw $4, 24(%rdi), %xmm0, %xmm0
 ; AVX2-NEXT:    vpinsrw $6, 28(%rdi), %xmm0, %xmm0
 ; AVX2-NEXT:    vpinsrw $7, 30(%rdi), %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $3, 6(%rdi), %xmm1, %xmm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
@@ -690,10 +690,10 @@ define <16 x i16> @merge_16i16_i16_0uu3zzuuuuuzCuEF_volatile(i16* %ptr) nounwind
 ; AVX512F:       # BB#0:
 ; AVX512F-NEXT:    vpxor %xmm0, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpinsrw $0, (%rdi), %xmm0, %xmm1
-; AVX512F-NEXT:    vpinsrw $3, 6(%rdi), %xmm1, %xmm1
 ; AVX512F-NEXT:    vpinsrw $4, 24(%rdi), %xmm0, %xmm0
 ; AVX512F-NEXT:    vpinsrw $6, 28(%rdi), %xmm0, %xmm0
 ; AVX512F-NEXT:    vpinsrw $7, 30(%rdi), %xmm0, %xmm0
+; AVX512F-NEXT:    vpinsrw $3, 6(%rdi), %xmm1, %xmm1
 ; AVX512F-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX512F-NEXT:    retq
 ;
@@ -702,10 +702,10 @@ define <16 x i16> @merge_16i16_i16_0uu3zzuuuuuzCuEF_volatile(i16* %ptr) nounwind
 ; X32-AVX-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X32-AVX-NEXT:    vpxor %xmm0, %xmm0, %xmm0
 ; X32-AVX-NEXT:    vpinsrw $0, (%eax), %xmm0, %xmm1
-; X32-AVX-NEXT:    vpinsrw $3, 6(%eax), %xmm1, %xmm1
 ; X32-AVX-NEXT:    vpinsrw $4, 24(%eax), %xmm0, %xmm0
 ; X32-AVX-NEXT:    vpinsrw $6, 28(%eax), %xmm0, %xmm0
 ; X32-AVX-NEXT:    vpinsrw $7, 30(%eax), %xmm0, %xmm0
+; X32-AVX-NEXT:    vpinsrw $3, 6(%eax), %xmm1, %xmm1
 ; X32-AVX-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; X32-AVX-NEXT:    retl
   %ptr0 = getelementptr inbounds i16, i16* %ptr, i64 0
diff --git a/llvm/test/CodeGen/X86/merge-store-partially-alias-loads.ll b/llvm/test/CodeGen/X86/merge-store-partially-alias-loads.ll
index 735e64a076d..6ca964be957 100644
--- a/llvm/test/CodeGen/X86/merge-store-partially-alias-loads.ll
+++ b/llvm/test/CodeGen/X86/merge-store-partially-alias-loads.ll
@@ -21,11 +21,11 @@
 ; DBGDAG-DAG: [[LD2:t[0-9]+]]: i16,ch = load<LD2[%tmp81](align=1)> [[ENTRYTOKEN]], [[BASEPTR]], undef:i64
 ; DBGDAG-DAG: [[LD1:t[0-9]+]]: i8,ch = load<LD1[%tmp12]> [[ENTRYTOKEN]], [[ADDPTR]], undef:i64
 
-; DBGDAG: [[LOADTOKEN:t[0-9]+]]: ch = TokenFactor [[LD2]]:1, [[LD1]]:1
-
+; DBGDAG-DAG: [[ST1:t[0-9]+]]: ch = store<ST1[%tmp14]> [[ENTRYTOKEN]], [[LD1]], t{{[0-9]+}}, undef:i64
+; DBGDAG-DAG: [[LOADTOKEN:t[0-9]+]]: ch = TokenFactor [[LD2]]:1, [[LD1]]:1
 ; DBGDAG-DAG: [[ST2:t[0-9]+]]: ch = store<ST2[%tmp10](align=1)> [[LOADTOKEN]], [[LD2]], t{{[0-9]+}}, undef:i64
-; DBGDAG-DAG: [[ST1:t[0-9]+]]: ch = store<ST1[%tmp14]> [[ST2]], [[LD1]], t{{[0-9]+}}, undef:i64
-; DBGDAG: X86ISD::RET_FLAG [[ST1]],
+
+; DBGDAG: X86ISD::RET_FLAG t{{[0-9]+}},
 
 ; DBGDAG: Type-legalized selection DAG: BB#0 'merge_store_partial_overlap_load:'
 define void @merge_store_partial_overlap_load([4 x i8]* %tmp) {
diff --git a/llvm/test/CodeGen/X86/pr18023.ll b/llvm/test/CodeGen/X86/pr18023.ll
deleted file mode 100644
index c7ea20c281b..00000000000
--- a/llvm/test/CodeGen/X86/pr18023.ll
+++ /dev/null
@@ -1,31 +0,0 @@
-; RUN: llc < %s -mtriple x86_64-apple-macosx10.9.0 | FileCheck %s
-; PR18023
-
-; CHECK: movabsq $4294967296, %rcx
-; CHECK: movq  %rcx, (%rax)
-; CHECK: movl  $1, 4(%rax)
-; CHECK: movl  $0, 4(%rax)
-; CHECK: movq  $1, 4(%rax)
-
-@c = common global i32 0, align 4
-@a = common global [3 x i32] zeroinitializer, align 4
-@b = common global i32 0, align 4
-@.str = private unnamed_addr constant [4 x i8] c"%d\0A\00", align 1
-
-define void @func() {
-  store i32 1, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 1), align 4
-  store i32 0, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 0), align 4
-  %1 = load volatile i32, i32* @b, align 4
-  store i32 1, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 1), align 4
-  store i32 0, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 1), align 4
-  %2 = load volatile i32, i32* @b, align 4
-  store i32 1, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 1), align 4
-  store i32 0, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 2), align 4
-  %3 = load volatile i32, i32* @b, align 4
-  store i32 3, i32* @c, align 4
-  %4 = load i32, i32* getelementptr inbounds ([3 x i32], [3 x i32]* @a, i64 0, i64 1), align 4
-  %call = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0), i32 %4)
-  ret void
-}
-
-declare i32 @printf(i8*, ...)
diff --git a/llvm/test/CodeGen/X86/split-store.ll b/llvm/test/CodeGen/X86/split-store.ll
index c2e67fb2527..6e320efb2b2 100644
--- a/llvm/test/CodeGen/X86/split-store.ll
+++ b/llvm/test/CodeGen/X86/split-store.ll
@@ -1,8 +1,8 @@
 ; RUN: llc -mtriple=x86_64-unknown-unknown -force-split-store < %s | FileCheck %s
 
 ; CHECK-LABEL: int32_float_pair
-; CHECK: movl %edi, (%rsi)
-; CHECK: movss %xmm0, 4(%rsi)
+; CHECK-DAG: movl %edi, (%rsi)
+; CHECK-DAG: movss %xmm0, 4(%rsi)
 define void @int32_float_pair(i32 %tmp1, float %tmp2, i64* %ref.tmp) {
 entry:
   %t0 = bitcast float %tmp2 to i32
@@ -15,8 +15,8 @@ entry:
 }
 
 ; CHECK-LABEL: float_int32_pair
-; CHECK: movss %xmm0, (%rsi)
-; CHECK: movl %edi, 4(%rsi)
+; CHECK-DAG: movss %xmm0, (%rsi)
+; CHECK-DAG: movl %edi, 4(%rsi)
 define void @float_int32_pair(float %tmp1, i32 %tmp2, i64* %ref.tmp) {
 entry:
   %t0 = bitcast float %tmp1 to i32
@@ -29,9 +29,9 @@ entry:
 }
 
 ; CHECK-LABEL: int16_float_pair
-; CHECK: movzwl	%di, %eax
-; CHECK: movl %eax, (%rsi)
-; CHECK: movss %xmm0, 4(%rsi)
+; CHECK-DAG: movzwl	%di, %eax
+; CHECK-DAG: movl %eax, (%rsi)
+; CHECK-DAG: movss %xmm0, 4(%rsi)
 define void @int16_float_pair(i16 signext %tmp1, float %tmp2, i64* %ref.tmp) {
 entry:
   %t0 = bitcast float %tmp2 to i32
@@ -44,9 +44,9 @@ entry:
 }
 
 ; CHECK-LABEL: int8_float_pair
-; CHECK: movzbl	%dil, %eax
-; CHECK: movl %eax, (%rsi)
-; CHECK: movss %xmm0, 4(%rsi)
+; CHECK-DAG: movzbl	%dil, %eax
+; CHECK-DAG: movl %eax, (%rsi)
+; CHECK-DAG: movss %xmm0, 4(%rsi)
 define void @int8_float_pair(i8 signext %tmp1, float %tmp2, i64* %ref.tmp) {
 entry:
   %t0 = bitcast float %tmp2 to i32
@@ -146,10 +146,9 @@ entry:
 ; CHECK: movw	%di, (%rdx)
 ; CHECK: shrl	$16, %edi
 ; CHECK: movb	%dil, 2(%rdx)
-; CHECK: movl	%esi, %eax
-; CHECK: shrl	$16, %eax
-; CHECK: movb	%al, 6(%rdx)
-; CHECK: movw	%si, 4(%rdx)
+; CHECK: movw    %si, 4(%rdx)
+; CHECK: shrl    $16, %esi
+; CHECK: movb    %sil, 6(%rdx)
 define void @int24_int24_pair(i24 signext %tmp1, i24 signext %tmp2, i48* %ref.tmp) {
 entry:
   %t1 = zext i24 %tmp2 to i48
diff --git a/llvm/test/CodeGen/X86/stores-merging.ll b/llvm/test/CodeGen/X86/stores-merging.ll
index 9e479bd71b9..dbfb06881d8 100644
--- a/llvm/test/CodeGen/X86/stores-merging.ll
+++ b/llvm/test/CodeGen/X86/stores-merging.ll
@@ -13,9 +13,9 @@ target triple = "x86_64-unknown-linux-gnu"
 ;; the same result in memory in the end.
 
 ; CHECK-LABEL: redundant_stores_merging:
-; CHECK:   movl    $123, e+8(%rip)
-; CHECK:   movabsq $1958505086977, %rax
+; CHECK:   movabsq $528280977409, %rax
 ; CHECK:   movq    %rax, e+4(%rip)
+; CHECK:   movl    $456, e+8(%rip)
 define void @redundant_stores_merging() {
 entry:
   store i32 1, i32* getelementptr inbounds (%structTy, %structTy* @e, i64 0, i32 1), align 4
@@ -26,9 +26,9 @@ entry:
 
 ;; This variant tests PR25154.
 ; CHECK-LABEL: redundant_stores_merging_reverse:
-; CHECK:   movl    $123, e+8(%rip)
-; CHECK:   movabsq $1958505086977, %rax
+; CHECK:   movabsq $528280977409, %rax
 ; CHECK:   movq    %rax, e+4(%rip)
+; CHECK:   movl    $456, e+8(%rip)
 define void @redundant_stores_merging_reverse() {
 entry:
   store i32 123, i32* getelementptr inbounds (%structTy, %structTy* @e, i64 0, i32 2), align 4
@@ -45,9 +45,8 @@ entry:
 ;; a movl, after the store to 3).
 
 ;; CHECK-LABEL: overlapping_stores_merging:
-;; CHECK:  movw    $0, b+2(%rip)
+;; CHECK:  movl    $1, b(%rip)
 ;; CHECK:  movw    $2, b+3(%rip)
-;; CHECK:  movw    $1, b(%rip)
 define void @overlapping_stores_merging() {
 entry:
   store i16 0, i16* bitcast (i8* getelementptr inbounds ([8 x i8], [8 x i8]* @b, i64 0, i64 2) to i16*), align 2
diff --git a/llvm/test/CodeGen/X86/vector-compare-results.ll b/llvm/test/CodeGen/X86/vector-compare-results.ll
index c34f333ef78..5445c0f9515 100644
--- a/llvm/test/CodeGen/X86/vector-compare-results.ll
+++ b/llvm/test/CodeGen/X86/vector-compare-results.ll
@@ -343,98 +343,98 @@ define <32 x i1> @test_cmp_v32i8(<32 x i8> %a0, <32 x i8> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -881,98 +881,98 @@ define <32 x i1> @test_cmp_v32i16(<32 x i16> %a0, <32 x i16> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -1157,196 +1157,196 @@ define <64 x i1> @test_cmp_v64i8(<64 x i8> %a0, <64 x i8> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
-; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
-; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 6(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 4(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -2474,98 +2474,98 @@ define <32 x i1> @test_cmp_v32f32(<32 x float> %a0, <32 x float> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -3977,98 +3977,98 @@ define <32 x i1> @test_cmp_v32i32(<32 x i32> %a0, <32 x i32> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -4938,196 +4938,196 @@ define <64 x i1> @test_cmp_v64i16(<64 x i16> %a0, <64 x i16> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
-; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
-; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 6(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 4(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -6780,392 +6780,392 @@ define <128 x i1> @test_cmp_v128i8(<128 x i8> %a0, <128 x i8> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 14(%rdi)
-; SSE2-NEXT:    movdqa %xmm6, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
-; SSE2-NEXT:    movdqa %xmm5, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
-; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
-; SSE2-NEXT:    movdqa %xmm3, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
-; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
-; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 14(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 14(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movdqa %xmm6, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 12(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 12(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movdqa %xmm5, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 10(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 10(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 8(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 8(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movdqa %xmm3, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 6(%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 4(%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 14(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 12(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 10(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 8(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 6(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 4(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -8584,98 +8584,98 @@ define <32 x i1> @test_cmp_v32f64(<32 x double> %a0, <32 x double> %a1) nounwind
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
@@ -9907,98 +9907,98 @@ define <32 x i1> @test_cmp_v32i64(<32 x i64> %a0, <32 x i64> %a1) nounwind {
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
+; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
+; SSE2-NEXT:    andb $1, %cl
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
 ; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-variable-128.ll b/llvm/test/CodeGen/X86/vector-shuffle-variable-128.ll
index 5ee3908b024..87fd4a7bf6b 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-variable-128.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-variable-128.ll
@@ -42,8 +42,8 @@ define <2 x i64> @var_shuffle_v2i64_v2i64_xx_i64(<2 x i64> %x, i32 %i0, i32 %i1)
 ; SSE-NEXT:    andl $1, %edi
 ; SSE-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE-NEXT:    andl $1, %esi
-; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
 ; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
 ; SSE-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; SSE-NEXT:    retq
 ;
@@ -56,7 +56,7 @@ define <2 x i64> @var_shuffle_v2i64_v2i64_xx_i64(<2 x i64> %x, i32 %i0, i32 %i1)
 ; AVX-NEXT:    andl $1, %esi
 ; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX-NEXT:    retq
   %x0 = extractelement <2 x i64> %x, i32 %i0
   %x1 = extractelement <2 x i64> %x, i32 %i1
@@ -79,10 +79,10 @@ define <4 x float> @var_shuffle_v4f32_v4f32_xxxx_i32(<4 x float> %x, i32 %i0, i3
 ; SSE2-NEXT:    andl $3, %ecx
 ; SSE2-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSE2-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSE2-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSE2-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
@@ -99,10 +99,10 @@ define <4 x float> @var_shuffle_v4f32_v4f32_xxxx_i32(<4 x float> %x, i32 %i0, i3
 ; SSSE3-NEXT:    andl $3, %ecx
 ; SSSE3-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSSE3-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSSE3-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSSE3-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
@@ -164,10 +164,10 @@ define <4 x i32> @var_shuffle_v4i32_v4i32_xxxx_i32(<4 x i32> %x, i32 %i0, i32 %i
 ; SSE2-NEXT:    andl $3, %ecx
 ; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSE2-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
@@ -184,10 +184,10 @@ define <4 x i32> @var_shuffle_v4i32_v4i32_xxxx_i32(<4 x i32> %x, i32 %i0, i32 %i
 ; SSSE3-NEXT:    andl $3, %ecx
 ; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSSE3-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSSE3-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
@@ -255,29 +255,29 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xxxxxxxx_i16(<8 x i16> %x, i16 %i0, i1
 ; SSE2-NEXT:    andl $7, %r10d
 ; SSE2-NEXT:    movzwl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $7, %eax
-; SSE2-NEXT:    movzwl -24(%rsp,%r10,2), %r10d
 ; SSE2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; SSE2-NEXT:    movzwl -24(%rsp,%rdi,2), %edi
-; SSE2-NEXT:    movzwl -24(%rsp,%rsi,2), %esi
-; SSE2-NEXT:    movd %r10d, %xmm0
-; SSE2-NEXT:    movzwl -24(%rsp,%rdx,2), %edx
-; SSE2-NEXT:    movd %edx, %xmm1
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movzwl -24(%rsp,%rcx,2), %ecx
-; SSE2-NEXT:    movd %edi, %xmm0
-; SSE2-NEXT:    movzwl -24(%rsp,%r8,2), %edx
-; SSE2-NEXT:    movd %edx, %xmm2
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    movzwl -24(%rsp,%rcx,2), %eax
 ; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movd %ecx, %xmm2
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSE2-NEXT:    movd %esi, %xmm1
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
 ; SSE2-NEXT:    movzwl -24(%rsp,%r9,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm2
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSE2-NEXT:    movzwl -24(%rsp,%r10,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    movzwl -24(%rsp,%rdx,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm1
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSE2-NEXT:    movzwl -24(%rsp,%r8,2), %eax
 ; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; SSE2-NEXT:    movzwl -24(%rsp,%rdi,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: var_shuffle_v8i16_v8i16_xxxxxxxx_i16:
@@ -299,29 +299,29 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xxxxxxxx_i16(<8 x i16> %x, i16 %i0, i1
 ; SSSE3-NEXT:    andl $7, %r10d
 ; SSSE3-NEXT:    movzwl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $7, %eax
-; SSSE3-NEXT:    movzwl -24(%rsp,%r10,2), %r10d
 ; SSSE3-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; SSSE3-NEXT:    movzwl -24(%rsp,%rdi,2), %edi
-; SSSE3-NEXT:    movzwl -24(%rsp,%rsi,2), %esi
-; SSSE3-NEXT:    movd %r10d, %xmm0
-; SSSE3-NEXT:    movzwl -24(%rsp,%rdx,2), %edx
-; SSSE3-NEXT:    movd %edx, %xmm1
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movzwl -24(%rsp,%rcx,2), %ecx
-; SSSE3-NEXT:    movd %edi, %xmm0
-; SSSE3-NEXT:    movzwl -24(%rsp,%r8,2), %edx
-; SSSE3-NEXT:    movd %edx, %xmm2
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    movzwl -24(%rsp,%rcx,2), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movd %ecx, %xmm2
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSSE3-NEXT:    movd %esi, %xmm1
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
 ; SSSE3-NEXT:    movzwl -24(%rsp,%r9,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm2
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; SSSE3-NEXT:    movzwl -24(%rsp,%r10,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    movzwl -24(%rsp,%rdx,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm1
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSSE3-NEXT:    movzwl -24(%rsp,%r8,2), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; SSSE3-NEXT:    movzwl -24(%rsp,%rdi,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: var_shuffle_v8i16_v8i16_xxxxxxxx_i16:
@@ -343,8 +343,6 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xxxxxxxx_i16(<8 x i16> %x, i16 %i0, i1
 ; SSE41-NEXT:    andl $7, %r10d
 ; SSE41-NEXT:    movzwl {{[0-9]+}}(%rsp), %eax
 ; SSE41-NEXT:    andl $7, %eax
-; SSE41-NEXT:    movzwl -24(%rsp,%r10,2), %r10d
-; SSE41-NEXT:    movzwl -24(%rsp,%rax,2), %eax
 ; SSE41-NEXT:    movzwl -24(%rsp,%rdi,2), %edi
 ; SSE41-NEXT:    movd %edi, %xmm0
 ; SSE41-NEXT:    pinsrw $1, -24(%rsp,%rsi,2), %xmm0
@@ -352,8 +350,8 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xxxxxxxx_i16(<8 x i16> %x, i16 %i0, i1
 ; SSE41-NEXT:    pinsrw $3, -24(%rsp,%rcx,2), %xmm0
 ; SSE41-NEXT:    pinsrw $4, -24(%rsp,%r8,2), %xmm0
 ; SSE41-NEXT:    pinsrw $5, -24(%rsp,%r9,2), %xmm0
-; SSE41-NEXT:    pinsrw $6, %r10d, %xmm0
-; SSE41-NEXT:    pinsrw $7, %eax, %xmm0
+; SSE41-NEXT:    pinsrw $6, -24(%rsp,%r10,2), %xmm0
+; SSE41-NEXT:    pinsrw $7, -24(%rsp,%rax,2), %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: var_shuffle_v8i16_v8i16_xxxxxxxx_i16:
@@ -375,8 +373,6 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xxxxxxxx_i16(<8 x i16> %x, i16 %i0, i1
 ; AVX-NEXT:    andl $7, %r10d
 ; AVX-NEXT:    movzwl {{[0-9]+}}(%rsp), %eax
 ; AVX-NEXT:    andl $7, %eax
-; AVX-NEXT:    movzwl -24(%rsp,%r10,2), %r10d
-; AVX-NEXT:    movzwl -24(%rsp,%rax,2), %eax
 ; AVX-NEXT:    movzwl -24(%rsp,%rdi,2), %edi
 ; AVX-NEXT:    vmovd %edi, %xmm0
 ; AVX-NEXT:    vpinsrw $1, -24(%rsp,%rsi,2), %xmm0, %xmm0
@@ -384,8 +380,8 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xxxxxxxx_i16(<8 x i16> %x, i16 %i0, i1
 ; AVX-NEXT:    vpinsrw $3, -24(%rsp,%rcx,2), %xmm0, %xmm0
 ; AVX-NEXT:    vpinsrw $4, -24(%rsp,%r8,2), %xmm0, %xmm0
 ; AVX-NEXT:    vpinsrw $5, -24(%rsp,%r9,2), %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $6, %r10d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $7, %eax, %xmm0, %xmm0
+; AVX-NEXT:    vpinsrw $6, -24(%rsp,%r10,2), %xmm0, %xmm0
+; AVX-NEXT:    vpinsrw $7, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %x0 = extractelement <8 x i16> %x, i16 %i0
   %x1 = extractelement <8 x i16> %x, i16 %i1
@@ -416,80 +412,80 @@ define <16 x i8> @var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8(<16 x i8> %x, i8 %
 ; SSE2-NEXT:    # kill: %ESI<def> %ESI<kill> %RSI<def>
 ; SSE2-NEXT:    # kill: %EDI<def> %EDI<kill> %RDI<def>
 ; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %r10d
-; SSE2-NEXT:    andl $15, %r10d
-; SSE2-NEXT:    leaq -{{[0-9]+}}(%rsp), %r11
-; SSE2-NEXT:    movzbl (%r10,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm15
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
+; SSE2-NEXT:    leaq -{{[0-9]+}}(%rsp), %r10
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm8
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm15
+; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
+; SSE2-NEXT:    andl $15, %eax
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm9
-; SSE2-NEXT:    andl $15, %edx
-; SSE2-NEXT:    movzbl (%rdx,%r11), %eax
+; SSE2-NEXT:    andl $15, %ecx
+; SSE2-NEXT:    movzbl (%rcx,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm3
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm10
-; SSE2-NEXT:    andl $15, %edi
-; SSE2-NEXT:    movzbl (%rdi,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    andl $15, %r9d
+; SSE2-NEXT:    movzbl (%r9,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm7
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm11
-; SSE2-NEXT:    andl $15, %r8d
-; SSE2-NEXT:    movzbl (%r8,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm7
+; SSE2-NEXT:    andl $15, %esi
+; SSE2-NEXT:    movzbl (%rsi,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm6
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm2
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm12
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm12
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm5
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm13
-; SSE2-NEXT:    andl $15, %ecx
-; SSE2-NEXT:    movzbl (%rcx,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm6
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm4
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
 ; SSE2-NEXT:    movd %eax, %xmm14
-; SSE2-NEXT:    andl $15, %esi
-; SSE2-NEXT:    movzbl (%rsi,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm5
+; SSE2-NEXT:    andl $15, %r8d
+; SSE2-NEXT:    movzbl (%r8,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm1
 ; SSE2-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    andl $15, %r9d
-; SSE2-NEXT:    movzbl (%r9,%r11), %eax
-; SSE2-NEXT:    movd %eax, %xmm1
+; SSE2-NEXT:    movzbl (%rax,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm2
+; SSE2-NEXT:    andl $15, %edi
+; SSE2-NEXT:    movzbl (%rdi,%r10), %eax
+; SSE2-NEXT:    movd %eax, %xmm0
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3],xmm3[4],xmm15[4],xmm3[5],xmm15[5],xmm3[6],xmm15[6],xmm3[7],xmm15[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3],xmm0[4],xmm10[4],xmm0[5],xmm10[5],xmm0[6],xmm10[6],xmm0[7],xmm10[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1],xmm7[2],xmm11[2],xmm7[3],xmm11[3],xmm7[4],xmm11[4],xmm7[5],xmm11[5],xmm7[6],xmm11[6],xmm7[7],xmm11[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1],xmm0[2],xmm7[2],xmm0[3],xmm7[3],xmm0[4],xmm7[4],xmm0[5],xmm7[5],xmm0[6],xmm7[6],xmm0[7],xmm7[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1],xmm2[2],xmm12[2],xmm2[3],xmm12[3],xmm2[4],xmm12[4],xmm2[5],xmm12[5],xmm2[6],xmm12[6],xmm2[7],xmm12[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm13[0],xmm6[1],xmm13[1],xmm6[2],xmm13[2],xmm6[3],xmm13[3],xmm6[4],xmm13[4],xmm6[5],xmm13[5],xmm6[6],xmm13[6],xmm6[7],xmm13[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1],xmm6[2],xmm2[2],xmm6[3],xmm2[3],xmm6[4],xmm2[4],xmm6[5],xmm2[5],xmm6[6],xmm2[6],xmm6[7],xmm2[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm14[0],xmm5[1],xmm14[1],xmm5[2],xmm14[2],xmm5[3],xmm14[3],xmm5[4],xmm14[4],xmm5[5],xmm14[5],xmm5[6],xmm14[6],xmm5[7],xmm14[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1],xmm5[2],xmm1[2],xmm5[3],xmm1[3],xmm5[4],xmm1[4],xmm5[5],xmm1[5],xmm5[6],xmm1[6],xmm5[7],xmm1[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3],xmm5[4],xmm6[4],xmm5[5],xmm6[5],xmm5[6],xmm6[6],xmm5[7],xmm6[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3],xmm6[4],xmm7[4],xmm6[5],xmm7[5],xmm6[6],xmm7[6],xmm6[7],xmm7[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1],xmm6[2],xmm3[2],xmm6[3],xmm3[3],xmm6[4],xmm3[4],xmm6[5],xmm3[5],xmm6[6],xmm3[6],xmm6[7],xmm3[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3],xmm0[4],xmm6[4],xmm0[5],xmm6[5],xmm0[6],xmm6[6],xmm0[7],xmm6[7]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
@@ -501,89 +497,84 @@ define <16 x i8> @var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8(<16 x i8> %x, i8 %
 ; SSSE3-NEXT:    # kill: %ESI<def> %ESI<kill> %RSI<def>
 ; SSSE3-NEXT:    # kill: %EDI<def> %EDI<kill> %RDI<def>
 ; SSSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %r10d
-; SSSE3-NEXT:    andl $15, %r10d
-; SSSE3-NEXT:    leaq -{{[0-9]+}}(%rsp), %r11
-; SSSE3-NEXT:    movzbl (%r10,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm15
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
+; SSSE3-NEXT:    leaq -{{[0-9]+}}(%rsp), %r10
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm8
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm15
+; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
+; SSSE3-NEXT:    andl $15, %eax
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm9
-; SSSE3-NEXT:    andl $15, %edx
-; SSSE3-NEXT:    movzbl (%rdx,%r11), %eax
+; SSSE3-NEXT:    andl $15, %ecx
+; SSSE3-NEXT:    movzbl (%rcx,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm3
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm10
-; SSSE3-NEXT:    andl $15, %edi
-; SSSE3-NEXT:    movzbl (%rdi,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    andl $15, %r9d
+; SSSE3-NEXT:    movzbl (%r9,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm7
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm11
-; SSSE3-NEXT:    andl $15, %r8d
-; SSSE3-NEXT:    movzbl (%r8,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm7
+; SSSE3-NEXT:    andl $15, %esi
+; SSSE3-NEXT:    movzbl (%rsi,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm6
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm12
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm12
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm5
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm13
-; SSSE3-NEXT:    andl $15, %ecx
-; SSSE3-NEXT:    movzbl (%rcx,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm6
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm4
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm14
-; SSSE3-NEXT:    andl $15, %esi
-; SSSE3-NEXT:    movzbl (%rsi,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm5
+; SSSE3-NEXT:    andl $15, %r8d
+; SSSE3-NEXT:    movzbl (%r8,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm1
 ; SSSE3-NEXT:    movzbl {{[0-9]+}}(%rsp), %eax
 ; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    andl $15, %r9d
-; SSSE3-NEXT:    movzbl (%r9,%r11), %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
+; SSSE3-NEXT:    movzbl (%rax,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm2
+; SSSE3-NEXT:    andl $15, %edi
+; SSSE3-NEXT:    movzbl (%rdi,%r10), %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3],xmm3[4],xmm15[4],xmm3[5],xmm15[5],xmm3[6],xmm15[6],xmm3[7],xmm15[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3],xmm0[4],xmm10[4],xmm0[5],xmm10[5],xmm0[6],xmm10[6],xmm0[7],xmm10[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1],xmm7[2],xmm11[2],xmm7[3],xmm11[3],xmm7[4],xmm11[4],xmm7[5],xmm11[5],xmm7[6],xmm11[6],xmm7[7],xmm11[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1],xmm0[2],xmm7[2],xmm0[3],xmm7[3],xmm0[4],xmm7[4],xmm0[5],xmm7[5],xmm0[6],xmm7[6],xmm0[7],xmm7[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1],xmm2[2],xmm12[2],xmm2[3],xmm12[3],xmm2[4],xmm12[4],xmm2[5],xmm12[5],xmm2[6],xmm12[6],xmm2[7],xmm12[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm13[0],xmm6[1],xmm13[1],xmm6[2],xmm13[2],xmm6[3],xmm13[3],xmm6[4],xmm13[4],xmm6[5],xmm13[5],xmm6[6],xmm13[6],xmm6[7],xmm13[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1],xmm6[2],xmm2[2],xmm6[3],xmm2[3],xmm6[4],xmm2[4],xmm6[5],xmm2[5],xmm6[6],xmm2[6],xmm6[7],xmm2[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm14[0],xmm5[1],xmm14[1],xmm5[2],xmm14[2],xmm5[3],xmm14[3],xmm5[4],xmm14[4],xmm5[5],xmm14[5],xmm5[6],xmm14[6],xmm5[7],xmm14[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1],xmm5[2],xmm1[2],xmm5[3],xmm1[3],xmm5[4],xmm1[4],xmm5[5],xmm1[5],xmm5[6],xmm1[6],xmm5[7],xmm1[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3],xmm5[4],xmm6[4],xmm5[5],xmm6[5],xmm5[6],xmm6[6],xmm5[7],xmm6[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3],xmm6[4],xmm7[4],xmm6[5],xmm7[5],xmm6[6],xmm7[6],xmm6[7],xmm7[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1],xmm6[2],xmm3[2],xmm6[3],xmm3[3],xmm6[4],xmm3[4],xmm6[5],xmm3[5],xmm6[6],xmm3[6],xmm6[7],xmm3[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3],xmm0[4],xmm6[4],xmm0[5],xmm6[5],xmm0[6],xmm6[6],xmm0[7],xmm6[7]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
 ; SSE41-NEXT:    # kill: %R9D<def> %R9D<kill> %R9<def>
 ; SSE41-NEXT:    # kill: %R8D<def> %R8D<kill> %R8<def>
 ; SSE41-NEXT:    # kill: %ECX<def> %ECX<kill> %RCX<def>
@@ -591,74 +582,54 @@ define <16 x i8> @var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8(<16 x i8> %x, i8 %
 ; SSE41-NEXT:    # kill: %ESI<def> %ESI<kill> %RSI<def>
 ; SSE41-NEXT:    # kill: %EDI<def> %EDI<kill> %RDI<def>
 ; SSE41-NEXT:    andl $15, %edi
-; SSE41-NEXT:    andl $15, %esi
-; SSE41-NEXT:    andl $15, %edx
-; SSE41-NEXT:    andl $15, %ecx
-; SSE41-NEXT:    andl $15, %r8d
 ; SSE41-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE41-NEXT:    andl $15, %r9d
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %r10d
-; SSE41-NEXT:    andl $15, %r10d
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %r11d
-; SSE41-NEXT:    andl $15, %r11d
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %r14d
-; SSE41-NEXT:    andl $15, %r14d
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %r15d
-; SSE41-NEXT:    andl $15, %r15d
 ; SSE41-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax
 ; SSE41-NEXT:    movzbl (%rdi,%rax), %edi
 ; SSE41-NEXT:    movd %edi, %xmm0
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %r12d
-; SSE41-NEXT:    andl $15, %r12d
-; SSE41-NEXT:    pinsrb $1, (%rsi,%rax), %xmm0
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %esi
 ; SSE41-NEXT:    andl $15, %esi
-; SSE41-NEXT:    pinsrb $2, (%rdx,%rax), %xmm0
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %edx
+; SSE41-NEXT:    pinsrb $1, (%rsi,%rax), %xmm0
 ; SSE41-NEXT:    andl $15, %edx
-; SSE41-NEXT:    pinsrb $3, (%rcx,%rax), %xmm0
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    pinsrb $2, (%rdx,%rax), %xmm0
 ; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $3, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    andl $15, %r8d
 ; SSE41-NEXT:    pinsrb $4, (%r8,%rax), %xmm0
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ebx
-; SSE41-NEXT:    andl $15, %ebx
+; SSE41-NEXT:    andl $15, %r9d
 ; SSE41-NEXT:    pinsrb $5, (%r9,%rax), %xmm0
-; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %edi
-; SSE41-NEXT:    andl $15, %edi
-; SSE41-NEXT:    movzbl (%r10,%rax), %r8d
-; SSE41-NEXT:    movzbl (%r11,%rax), %r9d
-; SSE41-NEXT:    movzbl (%r14,%rax), %r10d
-; SSE41-NEXT:    movzbl (%r15,%rax), %r11d
-; SSE41-NEXT:    movzbl (%r12,%rax), %ebp
-; SSE41-NEXT:    movzbl (%rsi,%rax), %esi
-; SSE41-NEXT:    movzbl (%rdx,%rax), %edx
-; SSE41-NEXT:    movzbl (%rcx,%rax), %ecx
-; SSE41-NEXT:    movzbl (%rbx,%rax), %ebx
-; SSE41-NEXT:    movzbl (%rdi,%rax), %eax
-; SSE41-NEXT:    pinsrb $6, %r8d, %xmm0
-; SSE41-NEXT:    pinsrb $7, %r9d, %xmm0
-; SSE41-NEXT:    pinsrb $8, %r10d, %xmm0
-; SSE41-NEXT:    pinsrb $9, %r11d, %xmm0
-; SSE41-NEXT:    pinsrb $10, %ebp, %xmm0
-; SSE41-NEXT:    pinsrb $11, %esi, %xmm0
-; SSE41-NEXT:    pinsrb $12, %edx, %xmm0
-; SSE41-NEXT:    pinsrb $13, %ecx, %xmm0
-; SSE41-NEXT:    pinsrb $14, %ebx, %xmm0
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $6, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $7, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $8, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $9, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $10, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $11, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $12, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $13, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $14, (%rcx,%rax), %xmm0
+; SSE41-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; SSE41-NEXT:    andl $15, %ecx
+; SSE41-NEXT:    pinsrb $15, (%rcx,%rax), %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
 ; AVX:       # BB#0:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %r15
-; AVX-NEXT:    pushq %r14
-; AVX-NEXT:    pushq %r12
-; AVX-NEXT:    pushq %rbx
 ; AVX-NEXT:    # kill: %R9D<def> %R9D<kill> %R9<def>
 ; AVX-NEXT:    # kill: %R8D<def> %R8D<kill> %R8<def>
 ; AVX-NEXT:    # kill: %ECX<def> %ECX<kill> %RCX<def>
@@ -666,65 +637,50 @@ define <16 x i8> @var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8(<16 x i8> %x, i8 %
 ; AVX-NEXT:    # kill: %ESI<def> %ESI<kill> %RSI<def>
 ; AVX-NEXT:    # kill: %EDI<def> %EDI<kill> %RDI<def>
 ; AVX-NEXT:    andl $15, %edi
-; AVX-NEXT:    andl $15, %esi
-; AVX-NEXT:    andl $15, %edx
-; AVX-NEXT:    andl $15, %ecx
-; AVX-NEXT:    andl $15, %r8d
 ; AVX-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
-; AVX-NEXT:    andl $15, %r9d
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %r10d
-; AVX-NEXT:    andl $15, %r10d
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %r11d
-; AVX-NEXT:    andl $15, %r11d
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %r14d
-; AVX-NEXT:    andl $15, %r14d
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %r15d
-; AVX-NEXT:    andl $15, %r15d
 ; AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax
 ; AVX-NEXT:    movzbl (%rdi,%rax), %edi
 ; AVX-NEXT:    vmovd %edi, %xmm0
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %r12d
-; AVX-NEXT:    andl $15, %r12d
-; AVX-NEXT:    vpinsrb $1, (%rsi,%rax), %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %esi
 ; AVX-NEXT:    andl $15, %esi
-; AVX-NEXT:    vpinsrb $2, (%rdx,%rax), %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %edx
+; AVX-NEXT:    vpinsrb $1, (%rsi,%rax), %xmm0, %xmm0
 ; AVX-NEXT:    andl $15, %edx
-; AVX-NEXT:    vpinsrb $3, (%rcx,%rax), %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    vpinsrb $2, (%rdx,%rax), %xmm0, %xmm0
 ; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $3, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    andl $15, %r8d
 ; AVX-NEXT:    vpinsrb $4, (%r8,%rax), %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ebx
-; AVX-NEXT:    andl $15, %ebx
+; AVX-NEXT:    andl $15, %r9d
 ; AVX-NEXT:    vpinsrb $5, (%r9,%rax), %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %edi
-; AVX-NEXT:    andl $15, %edi
-; AVX-NEXT:    movzbl (%r10,%rax), %r8d
-; AVX-NEXT:    movzbl (%r11,%rax), %r9d
-; AVX-NEXT:    movzbl (%r14,%rax), %r10d
-; AVX-NEXT:    movzbl (%r15,%rax), %r11d
-; AVX-NEXT:    movzbl (%r12,%rax), %ebp
-; AVX-NEXT:    movzbl (%rsi,%rax), %esi
-; AVX-NEXT:    movzbl (%rdx,%rax), %edx
-; AVX-NEXT:    movzbl (%rcx,%rax), %ecx
-; AVX-NEXT:    movzbl (%rbx,%rax), %ebx
-; AVX-NEXT:    movzbl (%rdi,%rax), %eax
-; AVX-NEXT:    vpinsrb $6, %r8d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $7, %r9d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $8, %r10d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $9, %r11d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $10, %ebp, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $11, %esi, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $12, %edx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $13, %ecx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $14, %ebx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %r12
-; AVX-NEXT:    popq %r14
-; AVX-NEXT:    popq %r15
-; AVX-NEXT:    popq %rbp
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $6, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $7, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $8, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $9, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $10, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $11, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $12, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $13, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $14, (%rcx,%rax), %xmm0, %xmm0
+; AVX-NEXT:    movzbl {{[0-9]+}}(%rsp), %ecx
+; AVX-NEXT:    andl $15, %ecx
+; AVX-NEXT:    vpinsrb $15, (%rcx,%rax), %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %x0  = extractelement <16 x i8> %x, i8 %i0
   %x1  = extractelement <16 x i8> %x, i8 %i1
@@ -779,11 +735,11 @@ define <4 x i32> @mem_shuffle_v4i32_v4i32_xxxx_i32(<4 x i32> %x, i32* %i) nounwi
 ; SSE2-NEXT:    andl $3, %esi
 ; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSE2-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: mem_shuffle_v4i32_v4i32_xxxx_i32:
@@ -799,11 +755,11 @@ define <4 x i32> @mem_shuffle_v4i32_v4i32_xxxx_i32(<4 x i32> %x, i32* %i) nounwi
 ; SSSE3-NEXT:    andl $3, %esi
 ; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSSE3-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSSE3-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: mem_shuffle_v4i32_v4i32_xxxx_i32:
@@ -862,341 +818,281 @@ define <16 x i8> @mem_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8(<16 x i8> %x, i8*
 ; SSE2-LABEL: mem_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    movzbl (%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
 ; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movzbl 15(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
 ; SSE2-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl 8(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm8
-; SSE2-NEXT:    movzbl 12(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm9
-; SSE2-NEXT:    movzbl 4(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    movzbl 14(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm10
-; SSE2-NEXT:    movzbl 6(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm5
-; SSE2-NEXT:    movzbl 10(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm11
-; SSE2-NEXT:    movzbl 2(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm7
-; SSE2-NEXT:    movzbl 15(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm12
-; SSE2-NEXT:    movzbl 7(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    movzbl 11(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm13
-; SSE2-NEXT:    movzbl 3(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm6
-; SSE2-NEXT:    movzbl 13(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm14
-; SSE2-NEXT:    movzbl 5(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    movzbl 9(%rdi), %eax
-; SSE2-NEXT:    andl $15, %eax
-; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm15
-; SSE2-NEXT:    movzbl 1(%rdi), %eax
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm8
+; SSE2-NEXT:    movzbl 7(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm15
+; SSE2-NEXT:    movzbl 11(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm9
+; SSE2-NEXT:    movzbl 3(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm3
+; SSE2-NEXT:    movzbl 13(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm10
+; SSE2-NEXT:    movzbl 5(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm7
+; SSE2-NEXT:    movzbl 9(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm11
+; SSE2-NEXT:    movzbl 1(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm6
+; SSE2-NEXT:    movzbl 14(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm12
+; SSE2-NEXT:    movzbl 6(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm5
+; SSE2-NEXT:    movzbl 10(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm13
+; SSE2-NEXT:    movzbl 2(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm4
+; SSE2-NEXT:    movzbl 12(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm14
+; SSE2-NEXT:    movzbl 4(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm1
+; SSE2-NEXT:    movzbl 8(%rdi), %edx
+; SSE2-NEXT:    andl $15, %edx
+; SSE2-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSE2-NEXT:    movd %edx, %xmm2
 ; SSE2-NEXT:    andl $15, %eax
 ; SSE2-NEXT:    movzbl (%rax,%rcx), %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1],xmm0[2],xmm8[2],xmm0[3],xmm8[3],xmm0[4],xmm8[4],xmm0[5],xmm8[5],xmm0[6],xmm8[6],xmm0[7],xmm8[7]
+; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm10[0],xmm5[1],xmm10[1],xmm5[2],xmm10[2],xmm5[3],xmm10[3],xmm5[4],xmm10[4],xmm5[5],xmm10[5],xmm5[6],xmm10[6],xmm5[7],xmm10[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1],xmm7[2],xmm11[2],xmm7[3],xmm11[3],xmm7[4],xmm11[4],xmm7[5],xmm11[5],xmm7[6],xmm11[6],xmm7[7],xmm11[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm5[0],xmm7[1],xmm5[1],xmm7[2],xmm5[2],xmm7[3],xmm5[3],xmm7[4],xmm5[4],xmm7[5],xmm5[5],xmm7[6],xmm5[6],xmm7[7],xmm5[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1],xmm0[2],xmm7[2],xmm0[3],xmm7[3],xmm0[4],xmm7[4],xmm0[5],xmm7[5],xmm0[6],xmm7[6],xmm0[7],xmm7[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1],xmm2[2],xmm12[2],xmm2[3],xmm12[3],xmm2[4],xmm12[4],xmm2[5],xmm12[5],xmm2[6],xmm12[6],xmm2[7],xmm12[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm13[0],xmm6[1],xmm13[1],xmm6[2],xmm13[2],xmm6[3],xmm13[3],xmm6[4],xmm13[4],xmm6[5],xmm13[5],xmm6[6],xmm13[6],xmm6[7],xmm13[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1],xmm6[2],xmm2[2],xmm6[3],xmm2[3],xmm6[4],xmm2[4],xmm6[5],xmm2[5],xmm6[6],xmm2[6],xmm6[7],xmm2[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm14[0],xmm4[1],xmm14[1],xmm4[2],xmm14[2],xmm4[3],xmm14[3],xmm4[4],xmm14[4],xmm4[5],xmm14[5],xmm4[6],xmm14[6],xmm4[7],xmm14[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1],xmm1[2],xmm15[2],xmm1[3],xmm15[3],xmm1[4],xmm15[4],xmm1[5],xmm15[5],xmm1[6],xmm15[6],xmm1[7],xmm15[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1],xmm1[2],xmm6[2],xmm1[3],xmm6[3],xmm1[4],xmm6[4],xmm1[5],xmm6[5],xmm1[6],xmm6[6],xmm1[7],xmm6[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3],xmm3[4],xmm15[4],xmm3[5],xmm15[5],xmm3[6],xmm15[6],xmm3[7],xmm15[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3],xmm6[4],xmm7[4],xmm6[5],xmm7[5],xmm6[6],xmm7[6],xmm6[7],xmm7[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1],xmm6[2],xmm3[2],xmm6[3],xmm3[3],xmm6[4],xmm3[4],xmm6[5],xmm3[5],xmm6[6],xmm3[6],xmm6[7],xmm3[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3],xmm0[4],xmm6[4],xmm0[5],xmm6[5],xmm0[6],xmm6[6],xmm0[7],xmm6[7]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: mem_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    movzbl (%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
 ; SSSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
+; SSSE3-NEXT:    movzbl 15(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
 ; SSSE3-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl 8(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm8
-; SSSE3-NEXT:    movzbl 12(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm9
-; SSSE3-NEXT:    movzbl 4(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    movzbl 14(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm10
-; SSSE3-NEXT:    movzbl 6(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm5
-; SSSE3-NEXT:    movzbl 10(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm11
-; SSSE3-NEXT:    movzbl 2(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm7
-; SSSE3-NEXT:    movzbl 15(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm12
-; SSSE3-NEXT:    movzbl 7(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    movzbl 11(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm13
-; SSSE3-NEXT:    movzbl 3(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm6
-; SSSE3-NEXT:    movzbl 13(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm14
-; SSSE3-NEXT:    movzbl 5(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    movzbl 9(%rdi), %eax
-; SSSE3-NEXT:    andl $15, %eax
-; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm15
-; SSSE3-NEXT:    movzbl 1(%rdi), %eax
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm8
+; SSSE3-NEXT:    movzbl 7(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm15
+; SSSE3-NEXT:    movzbl 11(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm9
+; SSSE3-NEXT:    movzbl 3(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm3
+; SSSE3-NEXT:    movzbl 13(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm10
+; SSSE3-NEXT:    movzbl 5(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm7
+; SSSE3-NEXT:    movzbl 9(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm11
+; SSSE3-NEXT:    movzbl 1(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm6
+; SSSE3-NEXT:    movzbl 14(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm12
+; SSSE3-NEXT:    movzbl 6(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm5
+; SSSE3-NEXT:    movzbl 10(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm13
+; SSSE3-NEXT:    movzbl 2(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm4
+; SSSE3-NEXT:    movzbl 12(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm14
+; SSSE3-NEXT:    movzbl 4(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm1
+; SSSE3-NEXT:    movzbl 8(%rdi), %edx
+; SSSE3-NEXT:    andl $15, %edx
+; SSSE3-NEXT:    movzbl (%rdx,%rcx), %edx
+; SSSE3-NEXT:    movd %edx, %xmm2
 ; SSSE3-NEXT:    andl $15, %eax
 ; SSSE3-NEXT:    movzbl (%rax,%rcx), %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1],xmm0[2],xmm8[2],xmm0[3],xmm8[3],xmm0[4],xmm8[4],xmm0[5],xmm8[5],xmm0[6],xmm8[6],xmm0[7],xmm8[7]
+; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm8[0],xmm15[1],xmm8[1],xmm15[2],xmm8[2],xmm15[3],xmm8[3],xmm15[4],xmm8[4],xmm15[5],xmm8[5],xmm15[6],xmm8[6],xmm15[7],xmm8[7]
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3],xmm3[4],xmm9[4],xmm3[5],xmm9[5],xmm3[6],xmm9[6],xmm3[7],xmm9[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm10[0],xmm5[1],xmm10[1],xmm5[2],xmm10[2],xmm5[3],xmm10[3],xmm5[4],xmm10[4],xmm5[5],xmm10[5],xmm5[6],xmm10[6],xmm5[7],xmm10[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1],xmm7[2],xmm11[2],xmm7[3],xmm11[3],xmm7[4],xmm11[4],xmm7[5],xmm11[5],xmm7[6],xmm11[6],xmm7[7],xmm11[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm5[0],xmm7[1],xmm5[1],xmm7[2],xmm5[2],xmm7[3],xmm5[3],xmm7[4],xmm5[4],xmm7[5],xmm5[5],xmm7[6],xmm5[6],xmm7[7],xmm5[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1],xmm0[2],xmm7[2],xmm0[3],xmm7[3],xmm0[4],xmm7[4],xmm0[5],xmm7[5],xmm0[6],xmm7[6],xmm0[7],xmm7[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1],xmm2[2],xmm12[2],xmm2[3],xmm12[3],xmm2[4],xmm12[4],xmm2[5],xmm12[5],xmm2[6],xmm12[6],xmm2[7],xmm12[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm13[0],xmm6[1],xmm13[1],xmm6[2],xmm13[2],xmm6[3],xmm13[3],xmm6[4],xmm13[4],xmm6[5],xmm13[5],xmm6[6],xmm13[6],xmm6[7],xmm13[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm2[0],xmm6[1],xmm2[1],xmm6[2],xmm2[2],xmm6[3],xmm2[3],xmm6[4],xmm2[4],xmm6[5],xmm2[5],xmm6[6],xmm2[6],xmm6[7],xmm2[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm14[0],xmm4[1],xmm14[1],xmm4[2],xmm14[2],xmm4[3],xmm14[3],xmm4[4],xmm14[4],xmm4[5],xmm14[5],xmm4[6],xmm14[6],xmm4[7],xmm14[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1],xmm1[2],xmm15[2],xmm1[3],xmm15[3],xmm1[4],xmm15[4],xmm1[5],xmm15[5],xmm1[6],xmm15[6],xmm1[7],xmm15[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1],xmm1[2],xmm6[2],xmm1[3],xmm6[3],xmm1[4],xmm6[4],xmm1[5],xmm6[5],xmm1[6],xmm6[6],xmm1[7],xmm6[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm15[0],xmm3[1],xmm15[1],xmm3[2],xmm15[2],xmm3[3],xmm15[3],xmm3[4],xmm15[4],xmm3[5],xmm15[5],xmm3[6],xmm15[6],xmm3[7],xmm15[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3],xmm7[4],xmm10[4],xmm7[5],xmm10[5],xmm7[6],xmm10[6],xmm7[7],xmm10[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1],xmm6[2],xmm7[2],xmm6[3],xmm7[3],xmm6[4],xmm7[4],xmm6[5],xmm7[5],xmm6[6],xmm7[6],xmm6[7],xmm7[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm3[0],xmm6[1],xmm3[1],xmm6[2],xmm3[2],xmm6[3],xmm3[3],xmm6[4],xmm3[4],xmm6[5],xmm3[5],xmm6[6],xmm3[6],xmm6[7],xmm3[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm12[0],xmm5[1],xmm12[1],xmm5[2],xmm12[2],xmm5[3],xmm12[3],xmm5[4],xmm12[4],xmm5[5],xmm12[5],xmm5[6],xmm12[6],xmm5[7],xmm12[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm13[0],xmm4[1],xmm13[1],xmm4[2],xmm13[2],xmm4[3],xmm13[3],xmm4[4],xmm13[4],xmm4[5],xmm13[5],xmm4[6],xmm13[6],xmm4[7],xmm13[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm14[0],xmm1[1],xmm14[1],xmm1[2],xmm14[2],xmm1[3],xmm14[3],xmm1[4],xmm14[4],xmm1[5],xmm14[5],xmm1[6],xmm14[6],xmm1[7],xmm14[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3],xmm0[4],xmm6[4],xmm0[5],xmm6[5],xmm0[6],xmm6[6],xmm0[7],xmm6[7]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: mem_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    movzbl (%rdi), %r11d
-; SSE41-NEXT:    andl $15, %r11d
+; SSE41-NEXT:    movzbl (%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
 ; SSE41-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE41-NEXT:    movzbl 1(%rdi), %r9d
-; SSE41-NEXT:    andl $15, %r9d
+; SSE41-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
+; SSE41-NEXT:    movzbl (%rax,%rcx), %eax
+; SSE41-NEXT:    movd %eax, %xmm0
+; SSE41-NEXT:    movzbl 1(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $1, (%rax,%rcx), %xmm0
 ; SSE41-NEXT:    movzbl 2(%rdi), %eax
 ; SSE41-NEXT:    andl $15, %eax
-; SSE41-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
+; SSE41-NEXT:    pinsrb $2, (%rax,%rcx), %xmm0
 ; SSE41-NEXT:    movzbl 3(%rdi), %eax
 ; SSE41-NEXT:    andl $15, %eax
-; SSE41-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; SSE41-NEXT:    movzbl 4(%rdi), %r14d
-; SSE41-NEXT:    andl $15, %r14d
-; SSE41-NEXT:    movzbl 5(%rdi), %r15d
-; SSE41-NEXT:    andl $15, %r15d
-; SSE41-NEXT:    movzbl 6(%rdi), %r12d
-; SSE41-NEXT:    andl $15, %r12d
-; SSE41-NEXT:    movzbl 7(%rdi), %r13d
-; SSE41-NEXT:    andl $15, %r13d
-; SSE41-NEXT:    movzbl 8(%rdi), %r8d
-; SSE41-NEXT:    andl $15, %r8d
+; SSE41-NEXT:    pinsrb $3, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 4(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $4, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 5(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $5, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 6(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $6, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 7(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $7, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 8(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $8, (%rax,%rcx), %xmm0
 ; SSE41-NEXT:    movzbl 9(%rdi), %eax
 ; SSE41-NEXT:    andl $15, %eax
-; SSE41-NEXT:    movzbl 10(%rdi), %ecx
-; SSE41-NEXT:    andl $15, %ecx
-; SSE41-NEXT:    movzbl 11(%rdi), %edx
-; SSE41-NEXT:    andl $15, %edx
-; SSE41-NEXT:    movzbl 12(%rdi), %esi
-; SSE41-NEXT:    andl $15, %esi
-; SSE41-NEXT:    leaq -{{[0-9]+}}(%rsp), %rbp
-; SSE41-NEXT:    movzbl (%r11,%rbp), %ebx
-; SSE41-NEXT:    movd %ebx, %xmm0
-; SSE41-NEXT:    movzbl 13(%rdi), %r11d
-; SSE41-NEXT:    andl $15, %r11d
-; SSE41-NEXT:    pinsrb $1, (%r9,%rbp), %xmm0
-; SSE41-NEXT:    movzbl 14(%rdi), %ebx
-; SSE41-NEXT:    andl $15, %ebx
-; SSE41-NEXT:    movzbl 15(%rdi), %edi
-; SSE41-NEXT:    andl $15, %edi
-; SSE41-NEXT:    movzbl (%rdi,%rbp), %r10d
-; SSE41-NEXT:    movzbl (%rbx,%rbp), %r9d
-; SSE41-NEXT:    movzbl (%r11,%rbp), %r11d
-; SSE41-NEXT:    movzbl (%rsi,%rbp), %esi
-; SSE41-NEXT:    movzbl (%rdx,%rbp), %edx
-; SSE41-NEXT:    movzbl (%rcx,%rbp), %ecx
-; SSE41-NEXT:    movzbl (%rax,%rbp), %eax
-; SSE41-NEXT:    movzbl (%r8,%rbp), %r8d
-; SSE41-NEXT:    movzbl (%r13,%rbp), %r13d
-; SSE41-NEXT:    movzbl (%r12,%rbp), %r12d
-; SSE41-NEXT:    movzbl (%r15,%rbp), %r15d
-; SSE41-NEXT:    movzbl (%r14,%rbp), %r14d
-; SSE41-NEXT:    movq -{{[0-9]+}}(%rsp), %rdi # 8-byte Reload
-; SSE41-NEXT:    movzbl (%rdi,%rbp), %edi
-; SSE41-NEXT:    movq -{{[0-9]+}}(%rsp), %rbx # 8-byte Reload
-; SSE41-NEXT:    movzbl (%rbx,%rbp), %ebp
-; SSE41-NEXT:    pinsrb $2, %ebp, %xmm0
-; SSE41-NEXT:    pinsrb $3, %edi, %xmm0
-; SSE41-NEXT:    pinsrb $4, %r14d, %xmm0
-; SSE41-NEXT:    pinsrb $5, %r15d, %xmm0
-; SSE41-NEXT:    pinsrb $6, %r12d, %xmm0
-; SSE41-NEXT:    pinsrb $7, %r13d, %xmm0
-; SSE41-NEXT:    pinsrb $8, %r8d, %xmm0
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm0
-; SSE41-NEXT:    pinsrb $10, %ecx, %xmm0
-; SSE41-NEXT:    pinsrb $11, %edx, %xmm0
-; SSE41-NEXT:    pinsrb $12, %esi, %xmm0
-; SSE41-NEXT:    pinsrb $13, %r11d, %xmm0
-; SSE41-NEXT:    pinsrb $14, %r9d, %xmm0
-; SSE41-NEXT:    pinsrb $15, %r10d, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
+; SSE41-NEXT:    pinsrb $9, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 10(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $10, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 11(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $11, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 12(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $12, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 13(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $13, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 14(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $14, (%rax,%rcx), %xmm0
+; SSE41-NEXT:    movzbl 15(%rdi), %eax
+; SSE41-NEXT:    andl $15, %eax
+; SSE41-NEXT:    pinsrb $15, (%rax,%rcx), %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: mem_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8:
 ; AVX:       # BB#0:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %r15
-; AVX-NEXT:    pushq %r14
-; AVX-NEXT:    pushq %r13
-; AVX-NEXT:    pushq %r12
-; AVX-NEXT:    pushq %rbx
-; AVX-NEXT:    movzbl (%rdi), %r11d
-; AVX-NEXT:    andl $15, %r11d
+; AVX-NEXT:    movzbl (%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
 ; AVX-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
-; AVX-NEXT:    movzbl 1(%rdi), %r9d
-; AVX-NEXT:    andl $15, %r9d
+; AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rcx
+; AVX-NEXT:    movzbl (%rax,%rcx), %eax
+; AVX-NEXT:    vmovd %eax, %xmm0
+; AVX-NEXT:    movzbl 1(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $1, (%rax,%rcx), %xmm0, %xmm0
 ; AVX-NEXT:    movzbl 2(%rdi), %eax
 ; AVX-NEXT:    andl $15, %eax
-; AVX-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
+; AVX-NEXT:    vpinsrb $2, (%rax,%rcx), %xmm0, %xmm0
 ; AVX-NEXT:    movzbl 3(%rdi), %eax
 ; AVX-NEXT:    andl $15, %eax
-; AVX-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; AVX-NEXT:    movzbl 4(%rdi), %r14d
-; AVX-NEXT:    andl $15, %r14d
-; AVX-NEXT:    movzbl 5(%rdi), %r15d
-; AVX-NEXT:    andl $15, %r15d
-; AVX-NEXT:    movzbl 6(%rdi), %r12d
-; AVX-NEXT:    andl $15, %r12d
-; AVX-NEXT:    movzbl 7(%rdi), %r13d
-; AVX-NEXT:    andl $15, %r13d
-; AVX-NEXT:    movzbl 8(%rdi), %r8d
-; AVX-NEXT:    andl $15, %r8d
+; AVX-NEXT:    vpinsrb $3, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 4(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $4, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 5(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $5, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 6(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $6, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 7(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $7, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 8(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $8, (%rax,%rcx), %xmm0, %xmm0
 ; AVX-NEXT:    movzbl 9(%rdi), %eax
 ; AVX-NEXT:    andl $15, %eax
-; AVX-NEXT:    movzbl 10(%rdi), %ecx
-; AVX-NEXT:    andl $15, %ecx
-; AVX-NEXT:    movzbl 11(%rdi), %edx
-; AVX-NEXT:    andl $15, %edx
-; AVX-NEXT:    movzbl 12(%rdi), %esi
-; AVX-NEXT:    andl $15, %esi
-; AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rbp
-; AVX-NEXT:    movzbl (%r11,%rbp), %ebx
-; AVX-NEXT:    vmovd %ebx, %xmm0
-; AVX-NEXT:    movzbl 13(%rdi), %r11d
-; AVX-NEXT:    andl $15, %r11d
-; AVX-NEXT:    vpinsrb $1, (%r9,%rbp), %xmm0, %xmm0
-; AVX-NEXT:    movzbl 14(%rdi), %ebx
-; AVX-NEXT:    andl $15, %ebx
-; AVX-NEXT:    movzbl 15(%rdi), %edi
-; AVX-NEXT:    andl $15, %edi
-; AVX-NEXT:    movzbl (%rdi,%rbp), %r10d
-; AVX-NEXT:    movzbl (%rbx,%rbp), %r9d
-; AVX-NEXT:    movzbl (%r11,%rbp), %r11d
-; AVX-NEXT:    movzbl (%rsi,%rbp), %esi
-; AVX-NEXT:    movzbl (%rdx,%rbp), %edx
-; AVX-NEXT:    movzbl (%rcx,%rbp), %ecx
-; AVX-NEXT:    movzbl (%rax,%rbp), %eax
-; AVX-NEXT:    movzbl (%r8,%rbp), %r8d
-; AVX-NEXT:    movzbl (%r13,%rbp), %r13d
-; AVX-NEXT:    movzbl (%r12,%rbp), %r12d
-; AVX-NEXT:    movzbl (%r15,%rbp), %r15d
-; AVX-NEXT:    movzbl (%r14,%rbp), %r14d
-; AVX-NEXT:    movq -{{[0-9]+}}(%rsp), %rdi # 8-byte Reload
-; AVX-NEXT:    movzbl (%rdi,%rbp), %edi
-; AVX-NEXT:    movq -{{[0-9]+}}(%rsp), %rbx # 8-byte Reload
-; AVX-NEXT:    movzbl (%rbx,%rbp), %ebp
-; AVX-NEXT:    vpinsrb $2, %ebp, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $3, %edi, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $4, %r14d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $5, %r15d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $6, %r12d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $7, %r13d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $8, %r8d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $10, %ecx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $11, %edx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $12, %esi, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $13, %r11d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $14, %r9d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrb $15, %r10d, %xmm0, %xmm0
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %r12
-; AVX-NEXT:    popq %r13
-; AVX-NEXT:    popq %r14
-; AVX-NEXT:    popq %r15
-; AVX-NEXT:    popq %rbp
+; AVX-NEXT:    vpinsrb $9, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 10(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $10, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 11(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $11, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 12(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $12, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 13(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $13, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 14(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $14, (%rax,%rcx), %xmm0, %xmm0
+; AVX-NEXT:    movzbl 15(%rdi), %eax
+; AVX-NEXT:    andl $15, %eax
+; AVX-NEXT:    vpinsrb $15, (%rax,%rcx), %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %p0  = getelementptr inbounds i8, i8* %i, i64 0
   %p1  = getelementptr inbounds i8, i8* %i, i64 1
@@ -1331,26 +1227,26 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xyxyxy00_i16(<8 x i16> %x, <8 x i16> %
 ; SSE2-NEXT:    andl $7, %r8d
 ; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    andl $7, %r9d
-; SSE2-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
-; SSE2-NEXT:    movzwl -24(%rsp,%rcx,2), %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
+; SSE2-NEXT:    movzwl -24(%rsp,%rcx,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm0
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSE2-NEXT:    movd %eax, %xmm2
 ; SSE2-NEXT:    movzwl -24(%rsp,%r9,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm2
+; SSE2-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
 ; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    movzwl -40(%rsp,%rdi,2), %eax
-; SSE2-NEXT:    movzwl -40(%rsp,%rdx,2), %ecx
-; SSE2-NEXT:    movd %ecx, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; SSE2-NEXT:    movd %eax, %xmm0
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
+; SSE2-NEXT:    movzwl -40(%rsp,%rdx,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm2
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
 ; SSE2-NEXT:    movzwl -40(%rsp,%r8,2), %eax
 ; SSE2-NEXT:    movd %eax, %xmm1
+; SSE2-NEXT:    movzwl -40(%rsp,%rdi,2), %eax
+; SSE2-NEXT:    movd %eax, %xmm0
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: var_shuffle_v8i16_v8i16_xyxyxy00_i16:
@@ -1369,26 +1265,26 @@ define <8 x i16> @var_shuffle_v8i16_v8i16_xyxyxy00_i16(<8 x i16> %x, <8 x i16> %
 ; SSSE3-NEXT:    andl $7, %r8d
 ; SSSE3-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
 ; SSSE3-NEXT:    andl $7, %r9d
-; SSSE3-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
-; SSSE3-NEXT:    movzwl -24(%rsp,%rcx,2), %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
+; SSSE3-NEXT:    movzwl -24(%rsp,%rcx,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
 ; SSSE3-NEXT:    pxor %xmm1, %xmm1
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSSE3-NEXT:    movd %eax, %xmm2
 ; SSSE3-NEXT:    movzwl -24(%rsp,%r9,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm2
+; SSSE3-NEXT:    movzwl -24(%rsp,%rsi,2), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    movzwl -40(%rsp,%rdi,2), %eax
-; SSSE3-NEXT:    movzwl -40(%rsp,%rdx,2), %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; SSSE3-NEXT:    movd %eax, %xmm0
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
+; SSSE3-NEXT:    movzwl -40(%rsp,%rdx,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm2
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
 ; SSSE3-NEXT:    movzwl -40(%rsp,%r8,2), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm1
+; SSSE3-NEXT:    movzwl -40(%rsp,%rdi,2), %eax
+; SSSE3-NEXT:    movd %eax, %xmm0
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: var_shuffle_v8i16_v8i16_xyxyxy00_i16:
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-variable-256.ll b/llvm/test/CodeGen/X86/vector-shuffle-variable-256.ll
index 42b3c11d3d6..3a2413d2dc0 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-variable-256.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-variable-256.ll
@@ -1,4 +1,5 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; NOTE: Assertions have been autogenerated by update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
 
@@ -13,16 +14,16 @@ define <4 x double> @var_shuffle_v4f64_v4f64_xxxx_i64(<4 x double> %x, i64 %i0,
 ; ALL-NEXT:    movq %rsp, %rbp
 ; ALL-NEXT:    andq $-32, %rsp
 ; ALL-NEXT:    subq $64, %rsp
-; ALL-NEXT:    andl $3, %ecx
-; ALL-NEXT:    andl $3, %edx
 ; ALL-NEXT:    andl $3, %esi
 ; ALL-NEXT:    andl $3, %edi
+; ALL-NEXT:    andl $3, %ecx
+; ALL-NEXT:    andl $3, %edx
 ; ALL-NEXT:    vmovaps %ymm0, (%rsp)
 ; ALL-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero
 ; ALL-NEXT:    vmovhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
 ; ALL-NEXT:    vmovsd {{.*#+}} xmm1 = mem[0],zero
 ; ALL-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; ALL-NEXT:    movq %rbp, %rsp
 ; ALL-NEXT:    popq %rbp
 ; ALL-NEXT:    retq
@@ -68,16 +69,16 @@ define <4 x double> @var_shuffle_v4f64_v4f64_uxx0_i64(<4 x double> %x, i64 %i0,
 define <4 x double> @var_shuffle_v4f64_v2f64_xxxx_i64(<2 x double> %x, i64 %i0, i64 %i1, i64 %i2, i64 %i3) nounwind {
 ; ALL-LABEL: var_shuffle_v4f64_v2f64_xxxx_i64:
 ; ALL:       # BB#0:
-; ALL-NEXT:    andl $1, %ecx
-; ALL-NEXT:    andl $1, %edx
 ; ALL-NEXT:    andl $1, %esi
 ; ALL-NEXT:    andl $1, %edi
+; ALL-NEXT:    andl $1, %ecx
+; ALL-NEXT:    andl $1, %edx
 ; ALL-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
 ; ALL-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero
 ; ALL-NEXT:    vmovhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
 ; ALL-NEXT:    vmovsd {{.*#+}} xmm1 = mem[0],zero
 ; ALL-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; ALL-NEXT:    retq
   %x0 = extractelement <2 x double> %x, i64 %i0
   %x1 = extractelement <2 x double> %x, i64 %i1
@@ -97,18 +98,18 @@ define <4 x i64> @var_shuffle_v4i64_v4i64_xxxx_i64(<4 x i64> %x, i64 %i0, i64 %i
 ; AVX1-NEXT:    movq %rsp, %rbp
 ; AVX1-NEXT:    andq $-32, %rsp
 ; AVX1-NEXT:    subq $64, %rsp
-; AVX1-NEXT:    andl $3, %ecx
-; AVX1-NEXT:    andl $3, %edx
-; AVX1-NEXT:    andl $3, %esi
 ; AVX1-NEXT:    andl $3, %edi
+; AVX1-NEXT:    andl $3, %esi
+; AVX1-NEXT:    andl $3, %edx
+; AVX1-NEXT:    andl $3, %ecx
 ; AVX1-NEXT:    vmovaps %ymm0, (%rsp)
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    movq %rbp, %rsp
 ; AVX1-NEXT:    popq %rbp
 ; AVX1-NEXT:    retq
@@ -119,18 +120,18 @@ define <4 x i64> @var_shuffle_v4i64_v4i64_xxxx_i64(<4 x i64> %x, i64 %i0, i64 %i
 ; AVX2-NEXT:    movq %rsp, %rbp
 ; AVX2-NEXT:    andq $-32, %rsp
 ; AVX2-NEXT:    subq $64, %rsp
-; AVX2-NEXT:    andl $3, %ecx
-; AVX2-NEXT:    andl $3, %edx
-; AVX2-NEXT:    andl $3, %esi
 ; AVX2-NEXT:    andl $3, %edi
+; AVX2-NEXT:    andl $3, %esi
+; AVX2-NEXT:    andl $3, %edx
+; AVX2-NEXT:    andl $3, %ecx
 ; AVX2-NEXT:    vmovaps %ymm0, (%rsp)
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    movq %rbp, %rsp
 ; AVX2-NEXT:    popq %rbp
 ; AVX2-NEXT:    retq
@@ -152,12 +153,12 @@ define <4 x i64> @var_shuffle_v4i64_v4i64_xx00_i64(<4 x i64> %x, i64 %i0, i64 %i
 ; AVX1-NEXT:    movq %rsp, %rbp
 ; AVX1-NEXT:    andq $-32, %rsp
 ; AVX1-NEXT:    subq $64, %rsp
-; AVX1-NEXT:    andl $3, %esi
 ; AVX1-NEXT:    andl $3, %edi
+; AVX1-NEXT:    andl $3, %esi
 ; AVX1-NEXT:    vmovaps %ymm0, (%rsp)
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    movq %rbp, %rsp
@@ -170,12 +171,12 @@ define <4 x i64> @var_shuffle_v4i64_v4i64_xx00_i64(<4 x i64> %x, i64 %i0, i64 %i
 ; AVX2-NEXT:    movq %rsp, %rbp
 ; AVX2-NEXT:    andq $-32, %rsp
 ; AVX2-NEXT:    subq $64, %rsp
-; AVX2-NEXT:    andl $3, %esi
 ; AVX2-NEXT:    andl $3, %edi
+; AVX2-NEXT:    andl $3, %esi
 ; AVX2-NEXT:    vmovaps %ymm0, (%rsp)
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-NEXT:    movq %rbp, %rsp
@@ -195,34 +196,34 @@ define <4 x i64> @var_shuffle_v4i64_v4i64_xx00_i64(<4 x i64> %x, i64 %i0, i64 %i
 define <4 x i64> @var_shuffle_v4i64_v2i64_xxxx_i64(<2 x i64> %x, i64 %i0, i64 %i1, i64 %i2, i64 %i3) nounwind {
 ; AVX1-LABEL: var_shuffle_v4i64_v2i64_xxxx_i64:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    andl $1, %ecx
-; AVX1-NEXT:    andl $1, %edx
-; AVX1-NEXT:    andl $1, %esi
 ; AVX1-NEXT:    andl $1, %edi
+; AVX1-NEXT:    andl $1, %esi
+; AVX1-NEXT:    andl $1, %edx
+; AVX1-NEXT:    andl $1, %ecx
 ; AVX1-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: var_shuffle_v4i64_v2i64_xxxx_i64:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    andl $1, %ecx
-; AVX2-NEXT:    andl $1, %edx
-; AVX2-NEXT:    andl $1, %esi
 ; AVX2-NEXT:    andl $1, %edi
+; AVX2-NEXT:    andl $1, %esi
+; AVX2-NEXT:    andl $1, %edx
+; AVX2-NEXT:    andl $1, %ecx
 ; AVX2-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
   %x0 = extractelement <2 x i64> %x, i64 %i0
   %x1 = extractelement <2 x i64> %x, i64 %i1
@@ -260,16 +261,14 @@ define <8 x float> @var_shuffle_v8f32_v8f32_xxxxxxxx_i32(<8 x float> %x, i32 %i0
 ; AVX1-NEXT:    movl 24(%rbp), %eax
 ; AVX1-NEXT:    andl $7, %eax
 ; AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
+; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
+; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
 ; AVX1-NEXT:    vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; AVX1-NEXT:    vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm2 = xmm2[0],mem[0],xmm2[2,3]
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm2 = xmm2[0,1],mem[0],xmm2[3]
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm2 = xmm2[0,1,2],mem[0]
-; AVX1-NEXT:    vmovss {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm3 = xmm3[0],mem[0],xmm3[2,3]
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm3[0,1],xmm0[0],xmm3[3]
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0],mem[0],xmm1[2,3]
+; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
+; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0,1,2],mem[0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    movq %rbp, %rsp
 ; AVX1-NEXT:    popq %rbp
 ; AVX1-NEXT:    retq
@@ -340,16 +339,14 @@ define <8 x float> @var_shuffle_v8f32_v4f32_xxxxxxxx_i32(<4 x float> %x, i32 %i0
 ; ALL-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; ALL-NEXT:    andl $3, %eax
 ; ALL-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; ALL-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
+; ALL-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
+; ALL-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
 ; ALL-NEXT:    vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; ALL-NEXT:    vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; ALL-NEXT:    vinsertps {{.*#+}} xmm2 = xmm2[0],mem[0],xmm2[2,3]
-; ALL-NEXT:    vinsertps {{.*#+}} xmm2 = xmm2[0,1],mem[0],xmm2[3]
-; ALL-NEXT:    vinsertps {{.*#+}} xmm2 = xmm2[0,1,2],mem[0]
-; ALL-NEXT:    vmovss {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; ALL-NEXT:    vinsertps {{.*#+}} xmm3 = xmm3[0],mem[0],xmm3[2,3]
-; ALL-NEXT:    vinsertps {{.*#+}} xmm0 = xmm3[0,1],xmm0[0],xmm3[3]
-; ALL-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; ALL-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0],mem[0],xmm1[2,3]
+; ALL-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
+; ALL-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0,1,2],mem[0]
+; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; ALL-NEXT:    retq
   %x0 = extractelement <4 x float> %x, i32 %i0
   %x1 = extractelement <4 x float> %x, i32 %i1
@@ -390,32 +387,25 @@ define <16 x i16> @var_shuffle_v16i16_v16i16_xxxxxxxxxxxxxxxx_i16(<16 x i16> %x,
 ; AVX1-NEXT:    vmovd %eax, %xmm0
 ; AVX1-NEXT:    movl 40(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $1, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $1, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl 48(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $2, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl 56(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $3, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl 64(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $4, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $4, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl 72(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $5, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $5, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl 80(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $6, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $6, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl 88(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $7, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $7, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    andl $15, %edi
 ; AVX1-NEXT:    movzwl (%rsp,%rdi,2), %eax
 ; AVX1-NEXT:    vmovd %eax, %xmm1
@@ -431,12 +421,10 @@ define <16 x i16> @var_shuffle_v16i16_v16i16_xxxxxxxxxxxxxxxx_i16(<16 x i16> %x,
 ; AVX1-NEXT:    vpinsrw $5, (%rsp,%r9,2), %xmm1, %xmm1
 ; AVX1-NEXT:    movl 16(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $6, %eax, %xmm1, %xmm1
+; AVX1-NEXT:    vpinsrw $6, (%rsp,%rax,2), %xmm1, %xmm1
 ; AVX1-NEXT:    movl 24(%rbp), %eax
 ; AVX1-NEXT:    andl $15, %eax
-; AVX1-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $7, %eax, %xmm1, %xmm1
+; AVX1-NEXT:    vpinsrw $7, (%rsp,%rax,2), %xmm1, %xmm1
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    movq %rbp, %rsp
 ; AVX1-NEXT:    popq %rbp
@@ -461,32 +449,25 @@ define <16 x i16> @var_shuffle_v16i16_v16i16_xxxxxxxxxxxxxxxx_i16(<16 x i16> %x,
 ; AVX2-NEXT:    vmovd %eax, %xmm0
 ; AVX2-NEXT:    movl 40(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $1, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $1, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl 48(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $2, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl 56(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $3, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl 64(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $4, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $4, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl 72(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $5, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $5, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl 80(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $6, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $6, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl 88(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $7, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $7, (%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    andl $15, %edi
 ; AVX2-NEXT:    movzwl (%rsp,%rdi,2), %eax
 ; AVX2-NEXT:    vmovd %eax, %xmm1
@@ -502,12 +483,10 @@ define <16 x i16> @var_shuffle_v16i16_v16i16_xxxxxxxxxxxxxxxx_i16(<16 x i16> %x,
 ; AVX2-NEXT:    vpinsrw $5, (%rsp,%r9,2), %xmm1, %xmm1
 ; AVX2-NEXT:    movl 16(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $6, %eax, %xmm1, %xmm1
+; AVX2-NEXT:    vpinsrw $6, (%rsp,%rax,2), %xmm1, %xmm1
 ; AVX2-NEXT:    movl 24(%rbp), %eax
 ; AVX2-NEXT:    andl $15, %eax
-; AVX2-NEXT:    movzwl (%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $7, %eax, %xmm1, %xmm1
+; AVX2-NEXT:    vpinsrw $7, (%rsp,%rax,2), %xmm1, %xmm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    movq %rbp, %rsp
 ; AVX2-NEXT:    popq %rbp
@@ -563,32 +542,25 @@ define <16 x i16> @var_shuffle_v16i16_v8i16_xxxxxxxxxxxxxxxx_i16(<8 x i16> %x, i
 ; AVX1-NEXT:    vmovd %eax, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $1, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $1, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $2, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $3, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $4, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $4, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $5, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $5, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $6, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $6, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $7, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpinsrw $7, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX1-NEXT:    andl $7, %edi
 ; AVX1-NEXT:    movzwl -24(%rsp,%rdi,2), %eax
 ; AVX1-NEXT:    vmovd %eax, %xmm1
@@ -604,12 +576,10 @@ define <16 x i16> @var_shuffle_v16i16_v8i16_xxxxxxxxxxxxxxxx_i16(<8 x i16> %x, i
 ; AVX1-NEXT:    vpinsrw $5, -24(%rsp,%r9,2), %xmm1, %xmm1
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $6, %eax, %xmm1, %xmm1
+; AVX1-NEXT:    vpinsrw $6, -24(%rsp,%rax,2), %xmm1, %xmm1
 ; AVX1-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX1-NEXT:    andl $7, %eax
-; AVX1-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX1-NEXT:    vpinsrw $7, %eax, %xmm1, %xmm1
+; AVX1-NEXT:    vpinsrw $7, -24(%rsp,%rax,2), %xmm1, %xmm1
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
@@ -628,32 +598,25 @@ define <16 x i16> @var_shuffle_v16i16_v8i16_xxxxxxxxxxxxxxxx_i16(<8 x i16> %x, i
 ; AVX2-NEXT:    vmovd %eax, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $1, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $1, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $2, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $3, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $4, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $4, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $5, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $5, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $6, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $6, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $7, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpinsrw $7, -24(%rsp,%rax,2), %xmm0, %xmm0
 ; AVX2-NEXT:    andl $7, %edi
 ; AVX2-NEXT:    movzwl -24(%rsp,%rdi,2), %eax
 ; AVX2-NEXT:    vmovd %eax, %xmm1
@@ -669,12 +632,10 @@ define <16 x i16> @var_shuffle_v16i16_v8i16_xxxxxxxxxxxxxxxx_i16(<8 x i16> %x, i
 ; AVX2-NEXT:    vpinsrw $5, -24(%rsp,%r9,2), %xmm1, %xmm1
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $6, %eax, %xmm1, %xmm1
+; AVX2-NEXT:    vpinsrw $6, -24(%rsp,%rax,2), %xmm1, %xmm1
 ; AVX2-NEXT:    movl {{[0-9]+}}(%rsp), %eax
 ; AVX2-NEXT:    andl $7, %eax
-; AVX2-NEXT:    movzwl -24(%rsp,%rax,2), %eax
-; AVX2-NEXT:    vpinsrw $7, %eax, %xmm1, %xmm1
+; AVX2-NEXT:    vpinsrw $7, -24(%rsp,%rax,2), %xmm1, %xmm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
   %x0  = extractelement <8 x i16> %x, i32 %i0
@@ -734,11 +695,11 @@ define <4 x i64> @mem_shuffle_v4i64_v4i64_xxxx_i64(<4 x i64> %x, i64* %i) nounwi
 ; AVX1-NEXT:    vmovaps %ymm0, (%rsp)
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
-; AVX1-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    movq %rbp, %rsp
 ; AVX1-NEXT:    popq %rbp
 ; AVX1-NEXT:    retq
@@ -760,11 +721,11 @@ define <4 x i64> @mem_shuffle_v4i64_v4i64_xxxx_i64(<4 x i64> %x, i64* %i) nounwi
 ; AVX2-NEXT:    vmovaps %ymm0, (%rsp)
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
-; AVX2-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    movq %rbp, %rsp
 ; AVX2-NEXT:    popq %rbp
 ; AVX2-NEXT:    retq
@@ -801,11 +762,11 @@ define <4 x i64> @mem_shuffle_v4i64_v2i64_xxxx_i64(<2 x i64> %x, i64* %i) nounwi
 ; AVX1-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
-; AVX1-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: mem_shuffle_v4i64_v2i64_xxxx_i64:
@@ -821,11 +782,11 @@ define <4 x i64> @mem_shuffle_v4i64_v2i64_xxxx_i64(<2 x i64> %x, i64* %i) nounwi
 ; AVX2-NEXT:    vmovaps %xmm0, -{{[0-9]+}}(%rsp)
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
 ; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
-; AVX2-NEXT:    vmovq {{.*#+}} xmm2 = mem[0],zero
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
   %p0  = getelementptr inbounds i64, i64* %i, i32 0
   %p1  = getelementptr inbounds i64, i64* %i, i32 1
diff --git a/llvm/test/CodeGen/X86/vectorcall.ll b/llvm/test/CodeGen/X86/vectorcall.ll
index 6ba5e10dd21..598a339ee2f 100644
--- a/llvm/test/CodeGen/X86/vectorcall.ll
+++ b/llvm/test/CodeGen/X86/vectorcall.ll
@@ -103,7 +103,7 @@ entry:
 }
 ; CHECK-LABEL: test_mixed_1
 ; CHECK:       movaps	%xmm1, 16(%{{(e|r)}}sp)
-; CHECK:       movaps	16(%{{(e|r)}}sp), %xmm0
+; CHECK:       movaps	%xmm1, %xmm0
 ; CHECK:       ret{{q|l}}
 
 define x86_vectorcallcc <4 x float> @test_mixed_2(%struct.HVA4 inreg %a, %struct.HVA4* %b, <4 x float> %c) {
@@ -149,7 +149,7 @@ entry:
 }
 ; CHECK-LABEL: test_mixed_5
 ; CHECK:       movaps	%xmm5, 16(%{{(e|r)}}sp)
-; CHECK:       movaps	16(%{{(e|r)}}sp), %xmm0
+; CHECK:       movaps	%xmm5, %xmm0
 ; CHECK:       ret{{[ql]}}
 
 define x86_vectorcallcc %struct.HVA4 @test_mixed_6(%struct.HVA4 inreg %a, %struct.HVA4* %b) {
diff --git a/llvm/test/CodeGen/X86/win32-eh.ll b/llvm/test/CodeGen/X86/win32-eh.ll
index 88403c68740..de8464e4f8b 100644
--- a/llvm/test/CodeGen/X86/win32-eh.ll
+++ b/llvm/test/CodeGen/X86/win32-eh.ll
@@ -27,23 +27,26 @@ catch:
 
 ; CHECK-LABEL: _use_except_handler3:
 ; CHECK: pushl %ebp
-; CHECK: movl %esp, %ebp
-; CHECK: pushl %ebx
-; CHECK: pushl %edi
-; CHECK: pushl %esi
-; CHECK: subl ${{[0-9]+}}, %esp
-; CHECK: movl $-1, -16(%ebp)
-; CHECK: movl $L__ehtable$use_except_handler3, -20(%ebp)
-; CHECK: leal -28(%ebp), %[[node:[^ ,]*]]
-; CHECK: movl $__except_handler3, -24(%ebp)
-; CHECK: movl %fs:0, %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], -28(%ebp)
-; CHECK: movl %[[node]], %fs:0
-; CHECK: calll _may_throw_or_crash
+; CHECK-NEXT: movl %esp, %ebp
+; CHECK-NEXT: pushl %ebx
+; CHECK-NEXT: pushl %edi
+; CHECK-NEXT: pushl %esi
+; CHECK-NEXT: subl ${{[0-9]+}}, %esp
+; CHECK-NEXT: movl %esp, -36(%ebp)
+; CHECK-NEXT: movl $-1, -16(%ebp)
+; CHECK-NEXT: movl $L__ehtable$use_except_handler3, -20(%ebp)
+; CHECK-NEXT: leal -28(%ebp), %[[node:[^ ,]*]]
+; CHECK-NEXT: movl $__except_handler3, -24(%ebp)
+; CHECK-NEXT: movl %fs:0, %[[next:[^ ,]*]]
+; CHECK-NEXT: movl %[[next]], -28(%ebp)
+; CHECK-NEXT: movl %[[node]], %fs:0
+; CHECK-NEXT: movl $0, -16(%ebp)
+; CHECK-NEXT: calll _may_throw_or_crash
+
 ; CHECK: movl -28(%ebp), %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], %fs:0
+; CHECK-NEXT: movl %[[next]], %fs:0
 ; CHECK: retl
-; CHECK: LBB1_2: # %catch{{$}}
+; CHECK-NEXT: LBB1_2: # %catch{{$}}
 
 ; CHECK: .section .xdata,"dr"
 ; CHECK-LABEL: L__ehtable$use_except_handler3:
@@ -66,23 +69,37 @@ catch:
 
 ; CHECK-LABEL: _use_except_handler4:
 ; CHECK: pushl %ebp
-; CHECK: movl %esp, %ebp
-; CHECK: subl ${{[0-9]+}}, %esp
-; CHECK: movl %esp, -36(%ebp)
-; CHECK: movl $-2, -16(%ebp)
-; CHECK: movl $L__ehtable$use_except_handler4, %[[lsda:[^ ,]*]]
-; CHECK: xorl ___security_cookie, %[[lsda]]
-; CHECK: movl %[[lsda]], -20(%ebp)
-; CHECK: leal -28(%ebp), %[[node:[^ ,]*]]
-; CHECK: movl $__except_handler4, -24(%ebp)
-; CHECK: movl %fs:0, %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], -28(%ebp)
-; CHECK: movl %[[node]], %fs:0
-; CHECK: calll _may_throw_or_crash
+; CHECK-NEXT: movl %esp, %ebp
+; CHECK-NEXT: pushl %ebx
+; CHECK-NEXT: pushl %edi
+; CHECK-NEXT: pushl %esi
+; CHECK-NEXT: subl ${{[0-9]+}}, %esp
+; CHECK-NEXT: movl %ebp, %eax
+; CHECK-NEXT: movl %esp, -36(%ebp)
+; CHECK-NEXT: movl $-2, -16(%ebp)
+; CHECK-NEXT: movl $L__ehtable$use_except_handler4, %[[lsda:[^ ,]*]]
+; CHECK-NEXT: movl ___security_cookie, %[[seccookie:[^ ,]*]]
+; CHECK-NEXT: xorl %[[seccookie]], %[[lsda]]
+; CHECK-NEXT: movl %[[lsda]], -20(%ebp)
+; CHECK-NEXT: xorl %[[seccookie]], %[[tmp1:[^ ,]*]]
+; CHECK-NEXT: movl %[[tmp1]], -40(%ebp)
+; CHECK-NEXT: leal -28(%ebp), %[[node:[^ ,]*]]
+; CHECK-NEXT: movl $__except_handler4, -24(%ebp)
+; CHECK-NEXT: movl %fs:0, %[[next:[^ ,]*]]
+; CHECK-NEXT: movl %[[next]], -28(%ebp)
+; CHECK-NEXT: movl %[[node]], %fs:0
+; CHECK-NEXT: movl $0, -16(%ebp)
+; CHECK-NEXT: calll _may_throw_or_crash
+
 ; CHECK: movl -28(%ebp), %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], %fs:0
-; CHECK: retl
-; CHECK: LBB2_2: # %catch{{$}}
+; CHECK-NEXT: movl %[[next]], %fs:0
+; CHECK-NEXT: addl $28, %esp
+; CHECK-NEXT: popl %esi
+; CHECK-NEXT: popl %edi
+; CHECK-NEXT: popl %ebx
+; CHECK-NEXT: popl %ebp
+; CHECK-NEXT: retl
+; CHECK-NEXT: LBB2_2: # %catch{{$}}
 
 ; CHECK: .section .xdata,"dr"
 ; CHECK-LABEL: L__ehtable$use_except_handler4:
@@ -109,26 +126,33 @@ catch:
 
 ; CHECK-LABEL: _use_except_handler4_ssp:
 ; CHECK: pushl %ebp
-; CHECK: movl %esp, %ebp
-; CHECK: subl ${{[0-9]+}}, %esp
-; CHECK: movl %ebp, %[[ehguard:[^ ,]*]]
-; CHECK: movl %esp, -36(%ebp)
-; CHECK: movl $-2, -16(%ebp)
-; CHECK: movl $L__ehtable$use_except_handler4_ssp, %[[lsda:[^ ,]*]]
-; CHECK: xorl ___security_cookie, %[[lsda]]
-; CHECK: movl %[[lsda]], -20(%ebp)
-; CHECK: xorl ___security_cookie, %[[ehguard]]
-; CHECK: movl %[[ehguard]], -40(%ebp)
-; CHECK: leal -28(%ebp), %[[node:[^ ,]*]]
-; CHECK: movl $__except_handler4, -24(%ebp)
-; CHECK: movl %fs:0, %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], -28(%ebp)
-; CHECK: movl %[[node]], %fs:0
-; CHECK: calll _may_throw_or_crash
+; CHECK-NEXT: movl %esp, %ebp
+; CHECK-NEXT: pushl %ebx
+; CHECK-NEXT: pushl %edi
+; CHECK-NEXT: pushl %esi
+; CHECK-NEXT: subl ${{[0-9]+}}, %esp
+; CHECK-NEXT: movl %ebp, %[[ehguard:[^ ,]*]]
+; CHECK-NEXT: movl %esp, -36(%ebp)
+; CHECK-NEXT: movl $-2, -16(%ebp)
+; CHECK-NEXT: movl $L__ehtable$use_except_handler4_ssp, %[[lsda:[^ ,]*]]
+; CHECK-NEXT: movl ___security_cookie, %[[seccookie:[^ ,]*]]
+; CHECK-NEXT: xorl %[[seccookie]], %[[lsda]]
+; CHECK-NEXT: movl %[[lsda]], -20(%ebp)
+; CHECK-NEXT: xorl %[[seccookie]], %[[ehguard]]
+; CHECK-NEXT: movl %[[ehguard]], -40(%ebp)
+; CHECK-NEXT: leal -28(%ebp), %[[node:[^ ,]*]]
+; CHECK-NEXT: movl $__except_handler4, -24(%ebp)
+; CHECK-NEXT: movl %fs:0, %[[next:[^ ,]*]]
+; CHECK-NEXT: movl %[[next]], -28(%ebp)
+; CHECK-NEXT: movl %[[node]], %fs:0
+; CHECK-NEXT: movl $0, -16(%ebp)
+; CHECK-NEXT: calll _may_throw_or_crash
 ; CHECK: movl -28(%ebp), %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], %fs:0   
+; CHECK-NEXT: movl %[[next]], %fs:0   
 ; CHECK: retl
-; CHECK: [[catch:[^ ,]*]]: # %catch{{$}}
+; CHECK-NEXT: [[catch:[^ ,]*]]: # %catch{{$}}
+
+
 
 ; CHECK: .section .xdata,"dr"
 ; CHECK-LABEL: L__ehtable$use_except_handler4_ssp:
@@ -155,23 +179,26 @@ catch:
 
 ; CHECK-LABEL: _use_CxxFrameHandler3:
 ; CHECK: pushl %ebp
-; CHECK: movl %esp, %ebp
-; CHECK: subl ${{[0-9]+}}, %esp
-; CHECK: movl %esp, -28(%ebp)
-; CHECK: movl $-1, -16(%ebp)
-; CHECK: leal -24(%ebp), %[[node:[^ ,]*]]
-; CHECK: movl $___ehhandler$use_CxxFrameHandler3, -20(%ebp)
-; CHECK: movl %fs:0, %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], -24(%ebp)
-; CHECK: movl %[[node]], %fs:0
-; CHECK: movl $0, -16(%ebp)
-; CHECK: calll _may_throw_or_crash
+; CHECK-NEXT: movl %esp, %ebp
+; CHECK-NEXT: pushl %ebx
+; CHECK-NEXT: pushl %edi
+; CHECK-NEXT: pushl %esi
+; CHECK-NEXT: subl ${{[0-9]+}}, %esp
+; CHECK-NEXT: movl %esp, -28(%ebp)
+; CHECK-NEXT: movl $-1, -16(%ebp)
+; CHECK-NEXT: leal -24(%ebp), %[[node:[^ ,]*]]
+; CHECK-NEXT: movl $___ehhandler$use_CxxFrameHandler3, -20(%ebp)
+; CHECK-NEXT: movl %fs:0, %[[next:[^ ,]*]]
+; CHECK-NEXT: movl %[[next]], -24(%ebp)
+; CHECK-NEXT: movl %[[node]], %fs:0
+; CHECK-NEXT: movl $0, -16(%ebp)
+; CHECK-NEXT: calll _may_throw_or_crash
 ; CHECK: movl -24(%ebp), %[[next:[^ ,]*]]
-; CHECK: movl %[[next]], %fs:0
+; CHECK-NEXT: movl %[[next]], %fs:0
 ; CHECK: retl
 
 ; CHECK: .section .xdata,"dr"
-; CHECK: .p2align 2
+; CHECK-NEXT: .p2align 2
 ; CHECK-LABEL: L__ehtable$use_CxxFrameHandler3:
 ; CHECK-NEXT:  .long   429065506
 ; CHECK-NEXT:  .long   2
@@ -185,8 +212,8 @@ catch:
 
 ; CHECK-LABEL: ___ehhandler$use_CxxFrameHandler3:
 ; CHECK: movl $L__ehtable$use_CxxFrameHandler3, %eax
-; CHECK: jmp  ___CxxFrameHandler3 # TAILCALL
+; CHECK-NEXT: jmp  ___CxxFrameHandler3 # TAILCALL
 
 ; CHECK: .safeseh __except_handler3
-; CHECK: .safeseh __except_handler4
-; CHECK: .safeseh ___ehhandler$use_CxxFrameHandler3
+; CHECK-NEXT: .safeseh __except_handler4
+; CHECK-NEXT: .safeseh ___ehhandler$use_CxxFrameHandler3
diff --git a/llvm/test/CodeGen/XCore/varargs.ll b/llvm/test/CodeGen/XCore/varargs.ll
index 28c293390c5..2e364b27561 100644
--- a/llvm/test/CodeGen/XCore/varargs.ll
+++ b/llvm/test/CodeGen/XCore/varargs.ll
@@ -26,10 +26,10 @@ entry:
 ; CHECK-LABEL: test_vararg
 ; CHECK: extsp 6
 ; CHECK: stw lr, sp[1]
+; CHECK: stw r3, sp[6]
 ; CHECK: stw r0, sp[3]
 ; CHECK: stw r1, sp[4]
 ; CHECK: stw r2, sp[5]
-; CHECK: stw r3, sp[6]
 ; CHECK: ldaw r0, sp[3]
 ; CHECK: stw r0, sp[2]
   %list = alloca i8*, align 4