X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FX86%2FX86ISelLowering.cpp;h=2f14d790742f2f77a8852af3c996a0492ecb86ce;hb=22f779d1fd0d7a7e878a3c28bc5d9910ae655df3;hp=7d195fcdae9be4390e6de83d02d3b622afa0238c;hpb=9e93e47b7f196893b5779090897210d750aa1c6b;p=oota-llvm.git

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 7d195fcdae9..2f14d790742 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -22,6 +22,7 @@
 #include "llvm/ADT/SmallSet.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/StringExtras.h"
+#include "llvm/ADT/StringSwitch.h"
 #include "llvm/ADT/VariadicFunction.h"
 #include "llvm/CodeGen/IntrinsicLowering.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
@@ -85,7 +86,8 @@ static SDValue ExtractSubVector(SDValue Vec, unsigned IdxVal,
   // If the input is a buildvector just emit a smaller one.
   if (Vec.getOpcode() == ISD::BUILD_VECTOR)
     return DAG.getNode(ISD::BUILD_VECTOR, dl, ResultVT,
-                       Vec->op_begin()+NormalizedIdxVal, ElemsPerChunk);
+                       makeArrayRef(Vec->op_begin()+NormalizedIdxVal,
+                                    ElemsPerChunk));
 
   SDValue VecIdx = DAG.getIntPtrConstant(NormalizedIdxVal);
   SDValue Result = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, ResultVT, Vec,
@@ -826,7 +828,9 @@ void X86TargetLowering::resetOperationActions() {
     setOperationAction(ISD::FRINT, VT, Expand);
     setOperationAction(ISD::FNEARBYINT, VT, Expand);
     setOperationAction(ISD::SMUL_LOHI, VT, Expand);
+    setOperationAction(ISD::MULHS, VT, Expand);
     setOperationAction(ISD::UMUL_LOHI, VT, Expand);
+    setOperationAction(ISD::MULHU, VT, Expand);
     setOperationAction(ISD::SDIVREM, VT, Expand);
     setOperationAction(ISD::UDIVREM, VT, Expand);
     setOperationAction(ISD::FPOW, VT, Expand);
@@ -938,6 +942,10 @@ void X86TargetLowering::resetOperationActions() {
     setOperationAction(ISD::ADD,                MVT::v2i64, Legal);
     setOperationAction(ISD::MUL,                MVT::v4i32, Custom);
     setOperationAction(ISD::MUL,                MVT::v2i64, Custom);
+    setOperationAction(ISD::UMUL_LOHI,          MVT::v4i32, Custom);
+    setOperationAction(ISD::SMUL_LOHI,          MVT::v4i32, Custom);
+    setOperationAction(ISD::MULHU,              MVT::v8i16, Legal);
+    setOperationAction(ISD::MULHS,              MVT::v8i16, Legal);
     setOperationAction(ISD::SUB,                MVT::v16i8, Legal);
     setOperationAction(ISD::SUB,                MVT::v8i16, Legal);
     setOperationAction(ISD::SUB,                MVT::v4i32, Legal);
@@ -1105,9 +1113,6 @@ void X86TargetLowering::resetOperationActions() {
     setOperationAction(ISD::SHL,               MVT::v4i32, Custom);
 
     setOperationAction(ISD::SRA,               MVT::v4i32, Custom);
-
-    setOperationAction(ISD::SDIV,              MVT::v8i16, Custom);
-    setOperationAction(ISD::SDIV,              MVT::v4i32, Custom);
   }
 
   if (!TM.Options.UseSoftFloat && Subtarget->hasFp256()) {
@@ -1172,8 +1177,6 @@ void X86TargetLowering::resetOperationActions() {
     setOperationAction(ISD::SRA,               MVT::v16i16, Custom);
     setOperationAction(ISD::SRA,               MVT::v32i8, Custom);
 
-    setOperationAction(ISD::SDIV,              MVT::v16i16, Custom);
-
     setOperationAction(ISD::SETCC,             MVT::v32i8, Custom);
     setOperationAction(ISD::SETCC,             MVT::v16i16, Custom);
     setOperationAction(ISD::SETCC,             MVT::v8i32, Custom);
@@ -1226,9 +1229,12 @@ void X86TargetLowering::resetOperationActions() {
       setOperationAction(ISD::MUL,             MVT::v16i16, Legal);
       // Don't lower v32i8 because there is no 128-bit byte mul
 
-      setOperationAction(ISD::VSELECT,         MVT::v32i8, Legal);
+      setOperationAction(ISD::UMUL_LOHI,       MVT::v8i32, Custom);
+      setOperationAction(ISD::SMUL_LOHI,       MVT::v8i32, Custom);
+      setOperationAction(ISD::MULHU,           MVT::v16i16, Legal);
+      setOperationAction(ISD::MULHS,           MVT::v16i16, Legal);
 
-      setOperationAction(ISD::SDIV,            MVT::v8i32, Custom);
+      setOperationAction(ISD::VSELECT,         MVT::v32i8, Legal);
     } else {
       setOperationAction(ISD::ADD,             MVT::v4i64, Custom);
       setOperationAction(ISD::ADD,             MVT::v8i32, Custom);
@@ -1337,7 +1343,6 @@ void X86TargetLowering::resetOperationActions() {
     setOperationAction(ISD::FNEG,               MVT::v8f64, Custom);
     setOperationAction(ISD::FMA,                MVT::v8f64, Legal);
     setOperationAction(ISD::FMA,                MVT::v16f32, Legal);
-    setOperationAction(ISD::SDIV,               MVT::v16i32, Custom);
 
     setOperationAction(ISD::FP_TO_SINT,         MVT::i32, Legal);
     setOperationAction(ISD::FP_TO_UINT,         MVT::i32, Legal);
@@ -1516,6 +1521,15 @@ void X86TargetLowering::resetOperationActions() {
     }
   }
 
+  if (Subtarget->isTargetWin64()) {
+    setOperationAction(ISD::SDIV, MVT::i128, Custom);
+    setOperationAction(ISD::UDIV, MVT::i128, Custom);
+    setOperationAction(ISD::SREM, MVT::i128, Custom);
+    setOperationAction(ISD::UREM, MVT::i128, Custom);
+    setOperationAction(ISD::SDIVREM, MVT::i128, Custom);
+    setOperationAction(ISD::UDIVREM, MVT::i128, Custom);
+  }
+
   // We have target-specific dag combine patterns for the following nodes:
   setTargetDAGCombine(ISD::VECTOR_SHUFFLE);
   setTargetDAGCombine(ISD::EXTRACT_VECTOR_ELT);
@@ -1930,8 +1944,7 @@ X86TargetLowering::LowerReturn(SDValue Chain,
   if (Flag.getNode())
     RetOps.push_back(Flag);
 
-  return DAG.getNode(X86ISD::RET_FLAG, dl,
-                     MVT::Other, &RetOps[0], RetOps.size());
+  return DAG.getNode(X86ISD::RET_FLAG, dl, MVT::Other, RetOps);
 }
 
 bool X86TargetLowering::isUsedByReturnOnly(SDNode *N, SDValue &Chain) const {
@@ -2424,13 +2437,11 @@ X86TargetLowering::LowerFormalArguments(SDValue Chain,
           SaveXMMOps.push_back(Val);
         }
         MemOps.push_back(DAG.getNode(X86ISD::VASTART_SAVE_XMM_REGS, dl,
-                                     MVT::Other,
-                                     &SaveXMMOps[0], SaveXMMOps.size()));
+                                     MVT::Other, SaveXMMOps));
       }
 
       if (!MemOps.empty())
-        Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
-                            &MemOps[0], MemOps.size());
+        Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, MemOps);
     }
   }
 
@@ -2497,10 +2508,10 @@ X86TargetLowering::EmitTailCallLoadRetAddr(SelectionDAG &DAG,
 
 /// EmitTailCallStoreRetAddr - Emit a store of the return address if tail call
 /// optimization is performed and it is required (FPDiff!=0).
-static SDValue
-EmitTailCallStoreRetAddr(SelectionDAG & DAG, MachineFunction &MF,
-                         SDValue Chain, SDValue RetAddrFrIdx, EVT PtrVT,
-                         unsigned SlotSize, int FPDiff, SDLoc dl) {
+static SDValue EmitTailCallStoreRetAddr(SelectionDAG &DAG, MachineFunction &MF,
+                                        SDValue Chain, SDValue RetAddrFrIdx,
+                                        EVT PtrVT, unsigned SlotSize,
+                                        int FPDiff, SDLoc dl) {
   // Store the return address to the appropriate stack slot.
   if (!FPDiff) return Chain;
   // Calculate the new stack slot for the return address.
@@ -2537,17 +2548,19 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   if (MF.getTarget().Options.DisableTailCalls)
     isTailCall = false;
 
-  if (isTailCall) {
+  bool IsMustTail = CLI.CS && CLI.CS->isMustTailCall();
+  if (IsMustTail) {
+    // Force this to be a tail call.  The verifier rules are enough to ensure
+    // that we can lower this successfully without moving the return address
+    // around.
+    isTailCall = true;
+  } else if (isTailCall) {
     // Check if it's really possible to do a tail call.
     isTailCall = IsEligibleForTailCallOptimization(Callee, CallConv,
                     isVarArg, SR != NotStructReturn,
                     MF.getFunction()->hasStructRetAttr(), CLI.RetTy,
                     Outs, OutVals, Ins, DAG);
 
-    if (!isTailCall && CLI.CS && CLI.CS->isMustTailCall())
-      report_fatal_error("failed to perform tail call elimination on a call "
-                         "site marked musttail");
-
     // Sibcalls are automatically detected tailcalls which do not require
     // ABI changes.
     if (!MF.getTarget().Options.GuaranteedTailCallOpt && isTailCall)
@@ -2582,7 +2595,7 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
     NumBytes = GetAlignedArgumentStackSize(NumBytes, DAG);
 
   int FPDiff = 0;
-  if (isTailCall && !IsSibcall) {
+  if (isTailCall && !IsSibcall && !IsMustTail) {
     // Lower arguments at fp - stackoffset + fpdiff.
     X86MachineFunctionInfo *X86Info = MF.getInfo<X86MachineFunctionInfo>();
     unsigned NumBytesCallerPushed = X86Info->getBytesToPopOnReturn();
@@ -2696,8 +2709,7 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
   }
 
   if (!MemOpChains.empty())
-    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
-                        &MemOpChains[0], MemOpChains.size());
+    Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, MemOpChains);
 
   if (Subtarget->isPICStyleGOT()) {
     // ELF / PIC requires GOT in the EBX register before function calls via PLT
@@ -2746,8 +2758,10 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
                                         DAG.getConstant(NumXMMRegs, MVT::i8)));
   }
 
-  // For tail calls lower the arguments to the 'real' stack slot.
-  if (isTailCall) {
+  // For tail calls lower the arguments to the 'real' stack slots.  Sibcalls
+  // don't need this because the eligibility check rejects calls that require
+  // shuffling arguments passed in memory.
+  if (!IsSibcall && isTailCall) {
     // Force all the incoming stack arguments to be loaded from the stack
     // before any new outgoing arguments are stored to the stack, because the
     // outgoing stack slots may alias the incoming argument stack slots, and
@@ -2759,45 +2773,45 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
     SmallVector<SDValue, 8> MemOpChains2;
     SDValue FIN;
     int FI = 0;
-    if (getTargetMachine().Options.GuaranteedTailCallOpt) {
-      for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
-        CCValAssign &VA = ArgLocs[i];
-        if (VA.isRegLoc())
-          continue;
-        assert(VA.isMemLoc());
-        SDValue Arg = OutVals[i];
-        ISD::ArgFlagsTy Flags = Outs[i].Flags;
-        // Create frame index.
-        int32_t Offset = VA.getLocMemOffset()+FPDiff;
-        uint32_t OpSize = (VA.getLocVT().getSizeInBits()+7)/8;
-        FI = MF.getFrameInfo()->CreateFixedObject(OpSize, Offset, true);
-        FIN = DAG.getFrameIndex(FI, getPointerTy());
-
-        if (Flags.isByVal()) {
-          // Copy relative to framepointer.
-          SDValue Source = DAG.getIntPtrConstant(VA.getLocMemOffset());
-          if (!StackPtr.getNode())
-            StackPtr = DAG.getCopyFromReg(Chain, dl,
-                                          RegInfo->getStackRegister(),
-                                          getPointerTy());
-          Source = DAG.getNode(ISD::ADD, dl, getPointerTy(), StackPtr, Source);
-
-          MemOpChains2.push_back(CreateCopyOfByValArgument(Source, FIN,
-                                                           ArgChain,
-                                                           Flags, DAG, dl));
-        } else {
-          // Store relative to framepointer.
-          MemOpChains2.push_back(
-            DAG.getStore(ArgChain, dl, Arg, FIN,
-                         MachinePointerInfo::getFixedStack(FI),
-                         false, false, 0));
-        }
+    for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
+      CCValAssign &VA = ArgLocs[i];
+      if (VA.isRegLoc())
+        continue;
+      assert(VA.isMemLoc());
+      SDValue Arg = OutVals[i];
+      ISD::ArgFlagsTy Flags = Outs[i].Flags;
+      // Skip inalloca arguments.  They don't require any work.
+      if (Flags.isInAlloca())
+        continue;
+      // Create frame index.
+      int32_t Offset = VA.getLocMemOffset()+FPDiff;
+      uint32_t OpSize = (VA.getLocVT().getSizeInBits()+7)/8;
+      FI = MF.getFrameInfo()->CreateFixedObject(OpSize, Offset, true);
+      FIN = DAG.getFrameIndex(FI, getPointerTy());
+
+      if (Flags.isByVal()) {
+        // Copy relative to framepointer.
+        SDValue Source = DAG.getIntPtrConstant(VA.getLocMemOffset());
+        if (!StackPtr.getNode())
+          StackPtr = DAG.getCopyFromReg(Chain, dl,
+                                        RegInfo->getStackRegister(),
+                                        getPointerTy());
+        Source = DAG.getNode(ISD::ADD, dl, getPointerTy(), StackPtr, Source);
+
+        MemOpChains2.push_back(CreateCopyOfByValArgument(Source, FIN,
+                                                         ArgChain,
+                                                         Flags, DAG, dl));
+      } else {
+        // Store relative to framepointer.
+        MemOpChains2.push_back(
+          DAG.getStore(ArgChain, dl, Arg, FIN,
+                       MachinePointerInfo::getFixedStack(FI),
+                       false, false, 0));
       }
     }
 
     if (!MemOpChains2.empty())
-      Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other,
-                          &MemOpChains2[0], MemOpChains2.size());
+      Chain = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, MemOpChains2);
 
     // Store the return address to the appropriate stack slot.
     Chain = EmitTailCallStoreRetAddr(DAG, MF, Chain, RetAddrFrIdx,
@@ -2934,10 +2948,10 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
     // This isn't right, although it's probably harmless on x86; liveouts
     // should be computed from returns not tail calls.  Consider a void
     // function making a tail call to a function returning int.
-    return DAG.getNode(X86ISD::TC_RETURN, dl, NodeTys, &Ops[0], Ops.size());
+    return DAG.getNode(X86ISD::TC_RETURN, dl, NodeTys, Ops);
   }
 
-  Chain = DAG.getNode(X86ISD::CALL, dl, NodeTys, &Ops[0], Ops.size());
+  Chain = DAG.getNode(X86ISD::CALL, dl, NodeTys, Ops);
   InFlag = Chain.getValue(1);
 
   // Create the CALLSEQ_END node.
@@ -4173,6 +4187,29 @@ static bool isUNPCKH_v_undef_Mask(ArrayRef<int> Mask, MVT VT, bool HasInt256) {
   return true;
 }
 
+// Match for INSERTI64x4 INSERTF64x4 instructions (src0[0], src1[0]) or
+// (src1[0], src0[1]), manipulation with 256-bit sub-vectors
+static bool isINSERT64x4Mask(ArrayRef<int> Mask, MVT VT, unsigned int *Imm) {
+  if (!VT.is512BitVector())
+    return false;
+
+  unsigned NumElts = VT.getVectorNumElements();
+  unsigned HalfSize = NumElts/2;
+  if (isSequentialOrUndefInRange(Mask, 0, HalfSize, 0)) {
+    if (isSequentialOrUndefInRange(Mask, HalfSize, HalfSize, NumElts)) {
+      *Imm = 1;
+      return true;
+    }
+  }
+  if (isSequentialOrUndefInRange(Mask, 0, HalfSize, NumElts)) {
+    if (isSequentialOrUndefInRange(Mask, HalfSize, HalfSize, HalfSize)) {
+      *Imm = 0;
+      return true;
+    }
+  }
+  return false;
+}
+
 /// isMOVLMask - Return true if the specified VECTOR_SHUFFLE operand
 /// specifies a shuffle of elements that is suitable for input to MOVSS,
 /// MOVSD, and MOVD, i.e. setting the lowest element.
@@ -4830,28 +4867,24 @@ static SDValue getZeroVector(EVT VT, const X86Subtarget *Subtarget,
     if (Subtarget->hasInt256()) { // AVX2
       SDValue Cst = DAG.getTargetConstant(0, MVT::i32);
       SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32, Ops,
-                        array_lengthof(Ops));
+      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32, Ops);
     } else {
       // 256-bit logic and arithmetic instructions in AVX are all
       // floating-point, no support for integer ops. Emit fp zeroed vectors.
       SDValue Cst = DAG.getTargetConstantFP(+0.0, MVT::f32);
       SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8f32, Ops,
-                        array_lengthof(Ops));
+      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8f32, Ops);
     }
   } else if (VT.is512BitVector()) { // AVX-512
       SDValue Cst = DAG.getTargetConstant(0, MVT::i32);
       SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst,
                         Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v16i32, Ops, 16);
+      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v16i32, Ops);
   } else if (VT.getScalarType() == MVT::i1) {
     assert(VT.getVectorNumElements() <= 16 && "Unexpected vector type");
     SDValue Cst = DAG.getTargetConstant(0, MVT::i1);
-    SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst,
-                      Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT,
-                       Ops, VT.getVectorNumElements());
+    SmallVector<SDValue, 16> Ops(VT.getVectorNumElements(), Cst);
+    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Ops);
   } else
     llvm_unreachable("Unexpected vector type");
 
@@ -4871,8 +4904,7 @@ static SDValue getOnesVector(MVT VT, bool HasInt256, SelectionDAG &DAG,
   if (VT.is256BitVector()) {
     if (HasInt256) { // AVX2
       SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32, Ops,
-                        array_lengthof(Ops));
+      Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v8i32, Ops);
     } else { // AVX
       Vec = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, Cst, Cst, Cst, Cst);
       Vec = Concat128BitVectors(Vec, Vec, MVT::v8i32, 8, DAG, dl);
@@ -5572,8 +5604,7 @@ static SDValue EltsFromConsecutiveLoads(EVT VT, SmallVectorImpl<SDValue> &Elts,
     SDVTList Tys = DAG.getVTList(MVT::v2i64, MVT::Other);
     SDValue Ops[] = { LDBase->getChain(), LDBase->getBasePtr() };
     SDValue ResNode =
-        DAG.getMemIntrinsicNode(X86ISD::VZEXT_LOAD, DL, Tys, Ops,
-                                array_lengthof(Ops), MVT::i64,
+        DAG.getMemIntrinsicNode(X86ISD::VZEXT_LOAD, DL, Tys, Ops, MVT::i64,
                                 LDBase->getPointerInfo(),
                                 LDBase->getAlignment(),
                                 false/*isVolatile*/, true/*ReadMem*/,
@@ -5851,18 +5882,14 @@ X86TargetLowering::LowerBUILD_VECTORvXi1(SDValue Op, SelectionDAG &DAG) const {
   SDLoc dl(Op);
   if (ISD::isBuildVectorAllZeros(Op.getNode())) {
     SDValue Cst = DAG.getTargetConstant(0, MVT::i1);
-    SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst,
-                      Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT,
-                       Ops, VT.getVectorNumElements());
+    SmallVector<SDValue, 16> Ops(VT.getVectorNumElements(), Cst);
+    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Ops);
   }
 
   if (ISD::isBuildVectorAllOnes(Op.getNode())) {
     SDValue Cst = DAG.getTargetConstant(1, MVT::i1);
-    SDValue Ops[] = { Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst,
-                      Cst, Cst, Cst, Cst, Cst, Cst, Cst, Cst };
-    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT,
-                       Ops, VT.getVectorNumElements());
+    SmallVector<SDValue, 16> Ops(VT.getVectorNumElements(), Cst);
+    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Ops);
   }
 
   bool AllContants = true;
@@ -6122,9 +6149,10 @@ X86TargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG) const {
     EVT HVT = EVT::getVectorVT(*DAG.getContext(), ExtVT, NumElems/2);
 
     // Build both the lower and upper subvector.
-    SDValue Lower = DAG.getNode(ISD::BUILD_VECTOR, dl, HVT, &V[0], NumElems/2);
-    SDValue Upper = DAG.getNode(ISD::BUILD_VECTOR, dl, HVT, &V[NumElems / 2],
-                                NumElems/2);
+    SDValue Lower = DAG.getNode(ISD::BUILD_VECTOR, dl, HVT,
+                                makeArrayRef(&V[0], NumElems/2));
+    SDValue Upper = DAG.getNode(ISD::BUILD_VECTOR, dl, HVT,
+                                makeArrayRef(&V[NumElems / 2], NumElems/2));
 
     // Recreate the wider vector with the lower and upper part.
     if (VT.is256BitVector())
@@ -6411,8 +6439,7 @@ static SDValue getPSHUFB(ArrayRef<int> MaskVals, SDValue V1, SDLoc &dl,
   if (ShufVT != VT)
     V1 = DAG.getNode(ISD::BITCAST, dl, ShufVT, V1);
   return DAG.getNode(X86ISD::PSHUFB, dl, ShufVT, V1,
-                     DAG.getNode(ISD::BUILD_VECTOR, dl, ShufVT,
-                                 PshufbMask.data(), PshufbMask.size()));
+                     DAG.getNode(ISD::BUILD_VECTOR, dl, ShufVT, PshufbMask));
 }
 
 // v8i16 shuffles - Prefer shuffles in the following order:
@@ -6714,7 +6741,7 @@ static SDValue LowerVECTOR_SHUFFLEv16i8(ShuffleVectorSDNode *SVOp,
     }
     V1 = DAG.getNode(X86ISD::PSHUFB, dl, MVT::v16i8, V1,
                      DAG.getNode(ISD::BUILD_VECTOR, dl,
-                                 MVT::v16i8, &pshufbMask[0], 16));
+                                 MVT::v16i8, pshufbMask));
 
     // As PSHUFB will zero elements with negative indices, it's safe to ignore
     // the 2nd operand if it's undefined or zero.
@@ -6732,7 +6759,7 @@ static SDValue LowerVECTOR_SHUFFLEv16i8(ShuffleVectorSDNode *SVOp,
     }
     V2 = DAG.getNode(X86ISD::PSHUFB, dl, MVT::v16i8, V2,
                      DAG.getNode(ISD::BUILD_VECTOR, dl,
-                                 MVT::v16i8, &pshufbMask[0], 16));
+                                 MVT::v16i8, pshufbMask));
     return DAG.getNode(ISD::OR, dl, MVT::v16i8, V1, V2);
   }
 
@@ -7003,8 +7030,7 @@ LowerVECTOR_SHUFFLE_256(ShuffleVectorSDNode *SVOp, SelectionDAG &DAG) {
       }
 
       // Construct the output using a BUILD_VECTOR.
-      Output[l] = DAG.getNode(ISD::BUILD_VECTOR, dl, NVT, &SVOps[0],
-                              SVOps.size());
+      Output[l] = DAG.getNode(ISD::BUILD_VECTOR, dl, NVT, SVOps);
     } else if (InputUsed[0] < 0) {
       // No input vectors were used! The result is undefined.
       Output[l] = DAG.getUNDEF(NVT);
@@ -7766,6 +7792,11 @@ X86TargetLowering::LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) const {
                                 getShuffleSHUFImmediate(SVOp), DAG);
   }
 
+  unsigned Idx;
+  if (VT.is512BitVector() && isINSERT64x4Mask(M, VT, &Idx))
+    return Insert256BitVector(V1, Extract256BitVector(V2, 0, DAG, dl),
+                              Idx*(NumElems/2), DAG, dl);
+
   // Handle VPERM2F128/VPERM2I128 permutations
   if (isVPERM2X128Mask(M, VT, HasFp256))
     return getTargetShuffleNode(X86ISD::VPERM2X128, dl, VT, V1,
@@ -7791,8 +7822,7 @@ X86TargetLowering::LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) const {
       permclMask.push_back(DAG.getConstant((M[i]>=0) ? M[i] : 0, MaskEltVT));
     }
 
-    SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, MaskVectorVT,
-                                &permclMask[0], NumElems);
+    SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, MaskVectorVT, permclMask);
     if (V2IsUndef)
       // Bitcast is for VPERMPS since mask is v8i32 but node takes v8f32
       return DAG.getNode(X86ISD::VPERMV, dl, VT,
@@ -8491,10 +8521,10 @@ GetTLSADDR(SelectionDAG &DAG, SDValue Chain, GlobalAddressSDNode *GA,
 
   if (InFlag) {
     SDValue Ops[] = { Chain,  TGA, *InFlag };
-    Chain = DAG.getNode(CallType, dl, NodeTys, Ops, array_lengthof(Ops));
+    Chain = DAG.getNode(CallType, dl, NodeTys, Ops);
   } else {
     SDValue Ops[]  = { Chain, TGA };
-    Chain = DAG.getNode(CallType, dl, NodeTys, Ops, array_lengthof(Ops));
+    Chain = DAG.getNode(CallType, dl, NodeTys, Ops);
   }
 
   // TLSADDR will be codegen'ed as call. Inform MFI that function has calls.
@@ -8678,7 +8708,7 @@ X86TargetLowering::LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const {
     SDValue Chain = DAG.getEntryNode();
     SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
     SDValue Args[] = { Chain, Offset };
-    Chain = DAG.getNode(X86ISD::TLSCALL, DL, NodeTys, Args, 2);
+    Chain = DAG.getNode(X86ISD::TLSCALL, DL, NodeTys, Args);
 
     // TLSCALL will be codegen'ed as call. Inform MFI that function has calls.
     MachineFrameInfo *MFI = DAG.getMachineFunction().getFrameInfo();
@@ -8806,15 +8836,15 @@ static SDValue LowerShiftParts(SDValue Op, SelectionDAG &DAG) {
   SDValue Ops1[4] = { Tmp3, Tmp1, CC, Cond };
 
   if (Op.getOpcode() == ISD::SHL_PARTS) {
-    Hi = DAG.getNode(X86ISD::CMOV, dl, VT, Ops0, 4);
-    Lo = DAG.getNode(X86ISD::CMOV, dl, VT, Ops1, 4);
+    Hi = DAG.getNode(X86ISD::CMOV, dl, VT, Ops0);
+    Lo = DAG.getNode(X86ISD::CMOV, dl, VT, Ops1);
   } else {
-    Lo = DAG.getNode(X86ISD::CMOV, dl, VT, Ops0, 4);
-    Hi = DAG.getNode(X86ISD::CMOV, dl, VT, Ops1, 4);
+    Lo = DAG.getNode(X86ISD::CMOV, dl, VT, Ops0);
+    Hi = DAG.getNode(X86ISD::CMOV, dl, VT, Ops1);
   }
 
   SDValue Ops[2] = { Lo, Hi };
-  return DAG.getMergeValues(Ops, array_lengthof(Ops), dl);
+  return DAG.getMergeValues(Ops, dl);
 }
 
 SDValue X86TargetLowering::LowerSINT_TO_FP(SDValue Op,
@@ -8877,8 +8907,7 @@ SDValue X86TargetLowering::BuildFILD(SDValue Op, EVT SrcVT, SDValue Chain,
   SDValue Ops[] = { Chain, StackSlot, DAG.getValueType(SrcVT) };
   SDValue Result = DAG.getMemIntrinsicNode(useSSE ? X86ISD::FILD_FLAG :
                                            X86ISD::FILD, DL,
-                                           Tys, Ops, array_lengthof(Ops),
-                                           SrcVT, MMO);
+                                           Tys, Ops, SrcVT, MMO);
 
   if (useSSE) {
     Chain = Result.getValue(1);
@@ -8901,8 +8930,7 @@ SDValue X86TargetLowering::BuildFILD(SDValue Op, EVT SrcVT, SDValue Chain,
                             MachineMemOperand::MOStore, SSFISize, SSFISize);
 
     Chain = DAG.getMemIntrinsicNode(X86ISD::FST, DL, Tys,
-                                    Ops, array_lengthof(Ops),
-                                    Op.getValueType(), MMO);
+                                    Ops, Op.getValueType(), MMO);
     Result = DAG.getLoad(Op.getValueType(), DL, Chain, StackSlot,
                          MachinePointerInfo::getFixedStack(SSFI),
                          false, false, false, 0);
@@ -9097,7 +9125,7 @@ SDValue X86TargetLowering::LowerUINT_TO_FP(SDValue Op,
   SDVTList Tys = DAG.getVTList(MVT::f80, MVT::Other);
   SDValue Ops[] = { Store, StackSlot, DAG.getValueType(MVT::i64) };
   SDValue Fild = DAG.getMemIntrinsicNode(X86ISD::FILD, dl, Tys, Ops,
-                                         array_lengthof(Ops), MVT::i64, MMO);
+                                         MVT::i64, MMO);
 
   APInt FF(32, 0x5F800000ULL);
 
@@ -9190,8 +9218,7 @@ X86TargetLowering:: FP_TO_INTHelper(SDValue Op, SelectionDAG &DAG,
     MachineMemOperand *MMO =
       MF.getMachineMemOperand(MachinePointerInfo::getFixedStack(SSFI),
                               MachineMemOperand::MOLoad, MemSize, MemSize);
-    Value = DAG.getMemIntrinsicNode(X86ISD::FLD, DL, Tys, Ops,
-                                    array_lengthof(Ops), DstTy, MMO);
+    Value = DAG.getMemIntrinsicNode(X86ISD::FLD, DL, Tys, Ops, DstTy, MMO);
     Chain = Value.getValue(1);
     SSFI = MF.getFrameInfo()->CreateStackObject(MemSize, MemSize, false);
     StackSlot = DAG.getFrameIndex(SSFI, getPointerTy());
@@ -9205,8 +9232,7 @@ X86TargetLowering:: FP_TO_INTHelper(SDValue Op, SelectionDAG &DAG,
     // Build the FP_TO_INT*_IN_MEM
     SDValue Ops[] = { Chain, Value, StackSlot };
     SDValue FIST = DAG.getMemIntrinsicNode(Opc, DL, DAG.getVTList(MVT::Other),
-                                           Ops, array_lengthof(Ops), DstTy,
-                                           MMO);
+                                           Ops, DstTy, MMO);
     return std::make_pair(FIST, StackSlot);
   } else {
     SDValue ftol = DAG.getNode(X86ISD::WIN_FTOL, DL,
@@ -9218,8 +9244,8 @@ X86TargetLowering:: FP_TO_INTHelper(SDValue Op, SelectionDAG &DAG,
       MVT::i32, eax.getValue(2));
     SDValue Ops[] = { eax, edx };
     SDValue pair = IsReplace
-      ? DAG.getNode(ISD::BUILD_PAIR, DL, MVT::i64, Ops, array_lengthof(Ops))
-      : DAG.getMergeValues(Ops, array_lengthof(Ops), DL);
+      ? DAG.getNode(ISD::BUILD_PAIR, DL, MVT::i64, Ops)
+      : DAG.getMergeValues(Ops, DL);
     return std::make_pair(pair, SDValue());
   }
 }
@@ -9414,8 +9440,7 @@ SDValue X86TargetLowering::LowerTRUNCATE(SDValue Op, SelectionDAG &DAG) const {
         for (unsigned j = 0; j < 8; ++j)
           pshufbMask.push_back(DAG.getConstant(0x80, MVT::i8));
       }
-      SDValue BV = DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v32i8,
-                               &pshufbMask[0], 32);
+      SDValue BV = DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v32i8, pshufbMask);
       In = DAG.getNode(X86ISD::PSHUFB, DL, MVT::v32i8, In, BV);
       In = DAG.getNode(ISD::BITCAST, DL, MVT::v4i64, In);
 
@@ -10012,7 +10037,7 @@ SDValue X86TargetLowering::EmitTest(SDValue Op, unsigned X86CC, SDLoc dl,
   for (unsigned i = 0; i != NumOperands; ++i)
     Ops.push_back(Op.getOperand(i));
 
-  SDValue New = DAG.getNode(Opcode, dl, VTs, &Ops[0], NumOperands);
+  SDValue New = DAG.getNode(Opcode, dl, VTs, Ops);
   DAG.ReplaceAllUsesWith(Op, New);
   return SDValue(New.getNode(), 1);
 }
@@ -10295,7 +10320,7 @@ static SDValue ChangeVSETULTtoVSETULE(SDLoc dl, SDValue Op1, SelectionDAG &DAG)
     ULTOp1.push_back(DAG.getConstant(Val - 1, EVT));
   }
 
-  return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, ULTOp1.data(), ULTOp1.size());
+  return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, ULTOp1);
 }
 
 static SDValue LowerVSETCC(SDValue Op, const X86Subtarget *Subtarget,
@@ -10862,7 +10887,7 @@ SDValue X86TargetLowering::LowerSELECT(SDValue Op, SelectionDAG &DAG) const {
   // condition is true.
   SDVTList VTs = DAG.getVTList(Op.getValueType(), MVT::Glue);
   SDValue Ops[] = { Op2, Op1, CC, Cond };
-  return DAG.getNode(X86ISD::CMOV, DL, VTs, Ops, array_lengthof(Ops));
+  return DAG.getNode(X86ISD::CMOV, DL, VTs, Ops);
 }
 
 static SDValue LowerSIGN_EXTEND_AVX512(SDValue Op, SelectionDAG &DAG) {
@@ -11299,7 +11324,7 @@ X86TargetLowering::LowerDYNAMIC_STACKALLOC(SDValue Op,
         SDLoc(Node));
 
     SDValue Ops[2] = { Tmp1, Tmp2 };
-    return DAG.getMergeValues(Ops, 2, dl);
+    return DAG.getMergeValues(Ops, dl);
   }
 
   // Get the inputs.
@@ -11333,7 +11358,7 @@ X86TargetLowering::LowerDYNAMIC_STACKALLOC(SDValue Op,
     SDValue Value = DAG.getNode(X86ISD::SEG_ALLOCA, dl, SPTy, Chain,
                                 DAG.getRegister(Vreg, SPTy));
     SDValue Ops1[2] = { Value, Chain };
-    return DAG.getMergeValues(Ops1, 2, dl);
+    return DAG.getMergeValues(Ops1, dl);
   } else {
     SDValue Flag;
     unsigned Reg = (Subtarget->is64Bit() ? X86::RAX : X86::EAX);
@@ -11357,7 +11382,7 @@ X86TargetLowering::LowerDYNAMIC_STACKALLOC(SDValue Op,
     }
 
     SDValue Ops1[2] = { SP, Chain };
-    return DAG.getMergeValues(Ops1, 2, dl);
+    return DAG.getMergeValues(Ops1, dl);
   }
 }
 
@@ -11418,8 +11443,7 @@ SDValue X86TargetLowering::LowerVASTART(SDValue Op, SelectionDAG &DAG) const {
   Store = DAG.getStore(Op.getOperand(0), DL, RSFIN, FIN,
                        MachinePointerInfo(SV, 16), false, false, 0);
   MemOps.push_back(Store);
-  return DAG.getNode(ISD::TokenFactor, DL, MVT::Other,
-                     &MemOps[0], MemOps.size());
+  return DAG.getNode(ISD::TokenFactor, DL, MVT::Other, MemOps);
 }
 
 SDValue X86TargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG) const {
@@ -11473,8 +11497,7 @@ SDValue X86TargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG) const {
   InstOps.push_back(DAG.getConstant(Align, MVT::i32));
   SDVTList VTs = DAG.getVTList(getPointerTy(), MVT::Other);
   SDValue VAARG = DAG.getMemIntrinsicNode(X86ISD::VAARG_64, dl,
-                                          VTs, &InstOps[0], InstOps.size(),
-                                          MVT::i64,
+                                          VTs, InstOps, MVT::i64,
                                           MachinePointerInfo(SV),
                                           /*Align=*/0,
                                           /*Volatile=*/false,
@@ -11534,7 +11557,7 @@ static SDValue getTargetVShiftByConstNode(unsigned Opc, SDLoc dl, MVT VT,
     ConstantSDNode *ND;
 
     switch(Opc) {
-    default: llvm_unreachable(0);
+    default: llvm_unreachable(nullptr);
     case X86ISD::VSHLI:
       for (unsigned i=0; i!=NumElts; ++i) {
         SDValue CurrentOp = SrcOp->getOperand(i);
@@ -11573,7 +11596,7 @@ static SDValue getTargetVShiftByConstNode(unsigned Opc, SDLoc dl, MVT VT,
       break;
     }
 
-    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Elts[0], NumElts);
+    return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Elts);
   }
 
   return DAG.getNode(Opc, dl, VT, SrcOp, DAG.getConstant(ShiftAmt, MVT::i8));
@@ -11605,7 +11628,7 @@ static SDValue getTargetVShiftNode(unsigned Opc, SDLoc dl, MVT VT,
   ShOps[0] = ShAmt;
   ShOps[1] = DAG.getConstant(0, MVT::i32);
   ShOps[2] = ShOps[3] = DAG.getUNDEF(MVT::i32);
-  ShAmt = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, &ShOps[0], 4);
+  ShAmt = DAG.getNode(ISD::BUILD_VECTOR, dl, MVT::v4i32, ShOps);
 
   // The return type has to be a 128-bit type with the same element
   // type as the input type.
@@ -11728,6 +11751,21 @@ static SDValue LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG) {
     return DAG.getNode(X86ISD::PMULUDQ, dl, Op.getValueType(),
                        Op.getOperand(1), Op.getOperand(2));
 
+  case Intrinsic::x86_sse41_pmuldq:
+  case Intrinsic::x86_avx2_pmul_dq:
+    return DAG.getNode(X86ISD::PMULDQ, dl, Op.getValueType(),
+                       Op.getOperand(1), Op.getOperand(2));
+
+  case Intrinsic::x86_sse2_pmulhu_w:
+  case Intrinsic::x86_avx2_pmulhu_w:
+    return DAG.getNode(ISD::MULHU, dl, Op.getValueType(),
+                       Op.getOperand(1), Op.getOperand(2));
+
+  case Intrinsic::x86_sse2_pmulh_w:
+  case Intrinsic::x86_avx2_pmulh_w:
+    return DAG.getNode(ISD::MULHS, dl, Op.getValueType(),
+                       Op.getOperand(1), Op.getOperand(2));
+
   // SSE2/AVX2 sub with unsigned saturation intrinsics
   case Intrinsic::x86_sse2_psubus_b:
   case Intrinsic::x86_sse2_psubus_w:
@@ -12179,7 +12217,7 @@ static SDValue LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG) {
     }
     SmallVector<SDValue, 5> NewOps(Op->op_begin()+1, Op->op_end());
     SDVTList VTs = DAG.getVTList(Op.getValueType(), MVT::i32);
-    SDValue PCMP = DAG.getNode(Opcode, dl, VTs, NewOps.data(), NewOps.size());
+    SDValue PCMP = DAG.getNode(Opcode, dl, VTs, NewOps);
     SDValue SetCC = DAG.getNode(X86ISD::SETCC, dl, MVT::i8,
                                 DAG.getConstant(X86CC, MVT::i8),
                                 SDValue(PCMP.getNode(), 1));
@@ -12196,7 +12234,7 @@ static SDValue LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG) {
 
     SmallVector<SDValue, 5> NewOps(Op->op_begin()+1, Op->op_end());
     SDVTList VTs = DAG.getVTList(Op.getValueType(), MVT::i32);
-    return DAG.getNode(Opcode, dl, VTs, NewOps.data(), NewOps.size());
+    return DAG.getNode(Opcode, dl, VTs, NewOps);
   }
   case Intrinsic::x86_fma_vfmadd_ps:
   case Intrinsic::x86_fma_vfmadd_pd:
@@ -12311,7 +12349,7 @@ static SDValue getGatherNode(unsigned Opc, SDValue Op, SelectionDAG &DAG,
   SDValue Ops[] = {Src, MaskInReg, Base, Scale, Index, Disp, Segment, Chain};
   SDNode *Res = DAG.getMachineNode(Opc, dl, VTs, Ops);
   SDValue RetOps[] = { SDValue(Res, 0), SDValue(Res, 2) };
-  return DAG.getMergeValues(RetOps, array_lengthof(RetOps), dl);
+  return DAG.getMergeValues(RetOps, dl);
 }
 
 static SDValue getMGatherNode(unsigned Opc, SDValue Op, SelectionDAG &DAG,
@@ -12333,7 +12371,7 @@ static SDValue getMGatherNode(unsigned Opc, SDValue Op, SelectionDAG &DAG,
   SDValue Ops[] = {Src, MaskInReg, Base, Scale, Index, Disp, Segment, Chain};
   SDNode *Res = DAG.getMachineNode(Opc, dl, VTs, Ops);
   SDValue RetOps[] = { SDValue(Res, 0), SDValue(Res, 2) };
-  return DAG.getMergeValues(RetOps, array_lengthof(RetOps), dl);
+  return DAG.getMergeValues(RetOps, dl);
 }
 
 static SDValue getScatterNode(unsigned Opc, SDValue Op, SelectionDAG &DAG,
@@ -12379,8 +12417,7 @@ static void getReadTimeStampCounter(SDNode *N, SDLoc DL, unsigned Opcode,
                               SelectionDAG &DAG, const X86Subtarget *Subtarget,
                               SmallVectorImpl<SDValue> &Results) {
   SDVTList Tys = DAG.getVTList(MVT::Other, MVT::Glue);
-  SDValue TheChain = N->getOperand(0);
-  SDValue rd = DAG.getNode(Opcode, DL, Tys, &TheChain, 1);
+  SDValue rd = DAG.getNode(Opcode, DL, Tys, N->getOperand(0));
   SDValue LO, HI;
 
   // The processor's time-stamp counter (a 64-bit MSR) is stored into the
@@ -12422,8 +12459,7 @@ static void getReadTimeStampCounter(SDNode *N, SDLoc DL, unsigned Opcode,
 
   // Use a buildpair to merge the two 32-bit values into a 64-bit one.
   SDValue Ops[] = { LO, HI };
-  SDValue Pair = DAG.getNode(ISD::BUILD_PAIR, DL, MVT::i64, Ops,
-                             array_lengthof(Ops));
+  SDValue Pair = DAG.getNode(ISD::BUILD_PAIR, DL, MVT::i64, Ops);
   Results.push_back(Pair);
   Results.push_back(Chain);
 }
@@ -12434,7 +12470,7 @@ static SDValue LowerREADCYCLECOUNTER(SDValue Op, const X86Subtarget *Subtarget,
   SDLoc DL(Op);
   getReadTimeStampCounter(Op.getNode(), DL, X86ISD::RDTSC_DAG, DAG, Subtarget,
                           Results);
-  return DAG.getMergeValues(&Results[0], Results.size(), DL);
+  return DAG.getMergeValues(Results, DL);
 }
 
 static SDValue LowerINTRINSIC_W_CHAIN(SDValue Op, const X86Subtarget *Subtarget,
@@ -12467,7 +12503,7 @@ static SDValue LowerINTRINSIC_W_CHAIN(SDValue Op, const X86Subtarget *Subtarget,
                       SDValue(Result.getNode(), 1) };
     SDValue isValid = DAG.getNode(X86ISD::CMOV, dl,
                                   DAG.getVTList(Op->getValueType(1), MVT::Glue),
-                                  Ops, array_lengthof(Ops));
+                                  Ops);
 
     // Return { result, isValid, chain }.
     return DAG.getNode(ISD::MERGE_VALUES, dl, Op->getVTList(), Result, isValid,
@@ -12625,7 +12661,7 @@ static SDValue LowerINTRINSIC_W_CHAIN(SDValue Op, const X86Subtarget *Subtarget,
     }
     SmallVector<SDValue, 2> Results;
     getReadTimeStampCounter(Op.getNode(), dl, Opc, DAG, Subtarget, Results);
-    return DAG.getMergeValues(&Results[0], Results.size(), dl);
+    return DAG.getMergeValues(Results, dl);
   }
   // XTEST intrinsics.
   case Intrinsic::x86_xtest: {
@@ -12691,6 +12727,18 @@ SDValue X86TargetLowering::LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const {
   return FrameAddr;
 }
 
+// FIXME? Maybe this could be a TableGen attribute on some registers and
+// this table could be generated automatically from RegInfo.
+unsigned X86TargetLowering::getRegisterByName(const char* RegName) const {
+  unsigned Reg = StringSwitch<unsigned>(RegName)
+                       .Case("esp", X86::ESP)
+                       .Case("rsp", X86::RSP)
+                       .Default(0);
+  if (Reg)
+    return Reg;
+  report_fatal_error("Invalid register name global variable");
+}
+
 SDValue X86TargetLowering::LowerFRAME_TO_ARGS_OFFSET(SDValue Op,
                                                      SelectionDAG &DAG) const {
   const X86RegisterInfo *RegInfo =
@@ -12810,7 +12858,7 @@ SDValue X86TargetLowering::LowerINIT_TRAMPOLINE(SDValue Op,
                                 MachinePointerInfo(TrmpAddr, 22),
                                 false, false, 0);
 
-    return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, OutChains, 6);
+    return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, OutChains);
   } else {
     const Function *Func =
       cast<Function>(cast<SrcValueSDNode>(Op.getOperand(5))->getValue());
@@ -12890,7 +12938,7 @@ SDValue X86TargetLowering::LowerINIT_TRAMPOLINE(SDValue Op,
                                 MachinePointerInfo(TrmpAddr, 6),
                                 false, false, 1);
 
-    return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, OutChains, 4);
+    return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, OutChains);
   }
 }
 
@@ -12933,8 +12981,7 @@ SDValue X86TargetLowering::LowerFLT_ROUNDS_(SDValue Op,
   SDValue Ops[] = { DAG.getEntryNode(), StackSlot };
   SDValue Chain = DAG.getMemIntrinsicNode(X86ISD::FNSTCW16m, DL,
                                           DAG.getVTList(MVT::Other),
-                                          Ops, array_lengthof(Ops), MVT::i16,
-                                          MMO);
+                                          Ops, MVT::i16, MMO);
 
   // Load FP Control Word from stack slot
   SDValue CWD = DAG.getLoad(MVT::i16, DL, Chain, StackSlot,
@@ -12987,7 +13034,7 @@ static SDValue LowerCTLZ(SDValue Op, SelectionDAG &DAG) {
     DAG.getConstant(X86::COND_E, MVT::i8),
     Op.getValue(1)
   };
-  Op = DAG.getNode(X86ISD::CMOV, dl, OpVT, Ops, array_lengthof(Ops));
+  Op = DAG.getNode(X86ISD::CMOV, dl, OpVT, Ops);
 
   // Finally xor with NumBits-1.
   Op = DAG.getNode(ISD::XOR, dl, OpVT, Op, DAG.getConstant(NumBits-1, OpVT));
@@ -13039,7 +13086,7 @@ static SDValue LowerCTTZ(SDValue Op, SelectionDAG &DAG) {
     DAG.getConstant(X86::COND_E, MVT::i8),
     Op.getValue(1)
   };
-  return DAG.getNode(X86ISD::CMOV, dl, VT, Ops, array_lengthof(Ops));
+  return DAG.getNode(X86ISD::CMOV, dl, VT, Ops);
 }
 
 // Lower256IntArith - Break a 256-bit integer operation into two new 128-bit
@@ -13157,59 +13204,104 @@ static SDValue LowerMUL(SDValue Op, const X86Subtarget *Subtarget,
   return DAG.getNode(ISD::ADD, dl, VT, Res, AhiBlo);
 }
 
-static SDValue LowerSDIV(SDValue Op, SelectionDAG &DAG) {
-  MVT VT = Op.getSimpleValueType();
-  MVT EltTy = VT.getVectorElementType();
-  unsigned NumElts = VT.getVectorNumElements();
-  SDValue N0 = Op.getOperand(0);
+SDValue X86TargetLowering::LowerWin64_i128OP(SDValue Op, SelectionDAG &DAG) const {
+  assert(Subtarget->isTargetWin64() && "Unexpected target");
+  EVT VT = Op.getValueType();
+  assert(VT.isInteger() && VT.getSizeInBits() == 128 &&
+         "Unexpected return type for lowering");
+
+  RTLIB::Libcall LC;
+  bool isSigned;
+  switch (Op->getOpcode()) {
+  default: llvm_unreachable("Unexpected request for libcall!");
+  case ISD::SDIV:      isSigned = true;  LC = RTLIB::SDIV_I128;    break;
+  case ISD::UDIV:      isSigned = false; LC = RTLIB::UDIV_I128;    break;
+  case ISD::SREM:      isSigned = true;  LC = RTLIB::SREM_I128;    break;
+  case ISD::UREM:      isSigned = false; LC = RTLIB::UREM_I128;    break;
+  case ISD::SDIVREM:   isSigned = true;  LC = RTLIB::SDIVREM_I128; break;
+  case ISD::UDIVREM:   isSigned = false; LC = RTLIB::UDIVREM_I128; break;
+  }
+
   SDLoc dl(Op);
+  SDValue InChain = DAG.getEntryNode();
 
-  // Lower sdiv X, pow2-const.
-  BuildVectorSDNode *C = dyn_cast<BuildVectorSDNode>(Op.getOperand(1));
-  if (!C)
-    return SDValue();
+  TargetLowering::ArgListTy Args;
+  TargetLowering::ArgListEntry Entry;
+  for (unsigned i = 0, e = Op->getNumOperands(); i != e; ++i) {
+    EVT ArgVT = Op->getOperand(i).getValueType();
+    assert(ArgVT.isInteger() && ArgVT.getSizeInBits() == 128 &&
+           "Unexpected argument type for lowering");
+    SDValue StackPtr = DAG.CreateStackTemporary(ArgVT, 16);
+    Entry.Node = StackPtr;
+    InChain = DAG.getStore(InChain, dl, Op->getOperand(i), StackPtr, MachinePointerInfo(),
+                           false, false, 16);
+    Type *ArgTy = ArgVT.getTypeForEVT(*DAG.getContext());
+    Entry.Ty = PointerType::get(ArgTy,0);
+    Entry.isSExt = false;
+    Entry.isZExt = false;
+    Args.push_back(Entry);
+  }
+
+  SDValue Callee = DAG.getExternalSymbol(getLibcallName(LC),
+                                         getPointerTy());
 
-  APInt SplatValue, SplatUndef;
-  unsigned SplatBitSize;
-  bool HasAnyUndefs;
-  if (!C->isConstantSplat(SplatValue, SplatUndef, SplatBitSize,
-                          HasAnyUndefs) ||
-      EltTy.getSizeInBits() < SplatBitSize)
-    return SDValue();
+  TargetLowering::CallLoweringInfo CLI(
+      InChain, static_cast<EVT>(MVT::v2i64).getTypeForEVT(*DAG.getContext()),
+      isSigned, !isSigned, false, true, 0, getLibcallCallingConv(LC),
+      /*isTailCall=*/false,
+      /*doesNotReturn=*/false, /*isReturnValueUsed=*/true, Callee, Args, DAG,
+      dl);
+  std::pair<SDValue, SDValue> CallInfo = LowerCallTo(CLI);
+
+  return DAG.getNode(ISD::BITCAST, dl, VT, CallInfo.first);
+}
+
+static SDValue LowerMUL_LOHI(SDValue Op, const X86Subtarget *Subtarget,
+                             SelectionDAG &DAG) {
+  SDValue Op0 = Op.getOperand(0), Op1 = Op.getOperand(1);
+  EVT VT = Op0.getValueType();
+  SDLoc dl(Op);
+
+  assert((VT == MVT::v4i32 && Subtarget->hasSSE2()) ||
+         (VT == MVT::v8i32 && Subtarget->hasInt256()));
+
+  // Get the high parts.
+  const int Mask[] = {1, 2, 3, 4, 5, 6, 7, 8};
+  SDValue Hi0 = DAG.getVectorShuffle(VT, dl, Op0, Op0, Mask);
+  SDValue Hi1 = DAG.getVectorShuffle(VT, dl, Op1, Op1, Mask);
 
-  if ((SplatValue != 0) &&
-      (SplatValue.isPowerOf2() || (-SplatValue).isPowerOf2())) {
-    unsigned Lg2 = SplatValue.countTrailingZeros();
-    // Splat the sign bit.
-    SmallVector<SDValue, 16> Sz(NumElts,
-                                DAG.getConstant(EltTy.getSizeInBits() - 1,
-                                                EltTy));
-    SDValue SGN = DAG.getNode(ISD::SRA, dl, VT, N0,
-                              DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Sz[0],
-                                          NumElts));
-    // Add (N0 < 0) ? abs2 - 1 : 0;
-    SmallVector<SDValue, 16> Amt(NumElts,
-                                 DAG.getConstant(EltTy.getSizeInBits() - Lg2,
-                                                 EltTy));
-    SDValue SRL = DAG.getNode(ISD::SRL, dl, VT, SGN,
-                              DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Amt[0],
-                                          NumElts));
-    SDValue ADD = DAG.getNode(ISD::ADD, dl, VT, N0, SRL);
-    SmallVector<SDValue, 16> Lg2Amt(NumElts, DAG.getConstant(Lg2, EltTy));
-    SDValue SRA = DAG.getNode(ISD::SRA, dl, VT, ADD,
-                              DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Lg2Amt[0],
-                                          NumElts));
-
-    // If we're dividing by a positive value, we're done.  Otherwise, we must
-    // negate the result.
-    if (SplatValue.isNonNegative())
-      return SRA;
-
-    SmallVector<SDValue, 16> V(NumElts, DAG.getConstant(0, EltTy));
-    SDValue Zero = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], NumElts);
-    return DAG.getNode(ISD::SUB, dl, VT, Zero, SRA);
+  // Emit two multiplies, one for the lower 2 ints and one for the higher 2
+  // ints.
+  MVT MulVT = VT == MVT::v4i32 ? MVT::v2i64 : MVT::v4i64;
+  bool IsSigned = Op->getOpcode() == ISD::SMUL_LOHI;
+  unsigned Opcode =
+      (!IsSigned || !Subtarget->hasSSE41()) ? X86ISD::PMULUDQ : X86ISD::PMULDQ;
+  SDValue Mul1 = DAG.getNode(ISD::BITCAST, dl, VT,
+                             DAG.getNode(Opcode, dl, MulVT, Op0, Op1));
+  SDValue Mul2 = DAG.getNode(ISD::BITCAST, dl, VT,
+                             DAG.getNode(Opcode, dl, MulVT, Hi0, Hi1));
+
+  // Shuffle it back into the right order.
+  const int HighMask[] = {1, 5, 3, 7, 9, 13, 11, 15};
+  SDValue Highs = DAG.getVectorShuffle(VT, dl, Mul1, Mul2, HighMask);
+  const int LowMask[] = {0, 4, 2, 6, 8, 12, 10, 14};
+  SDValue Lows = DAG.getVectorShuffle(VT, dl, Mul1, Mul2, LowMask);
+
+  // If we have a signed multiply but no PMULDQ fix up the high parts of a
+  // unsigned multiply.
+  if (IsSigned && !Subtarget->hasSSE41()) {
+    SDValue ShAmt =
+        DAG.getConstant(31, DAG.getTargetLoweringInfo().getShiftAmountTy(VT));
+    SDValue T1 = DAG.getNode(ISD::AND, dl, VT,
+                             DAG.getNode(ISD::SRA, dl, VT, Op0, ShAmt), Op1);
+    SDValue T2 = DAG.getNode(ISD::AND, dl, VT,
+                             DAG.getNode(ISD::SRA, dl, VT, Op1, ShAmt), Op0);
+
+    SDValue Fixup = DAG.getNode(ISD::ADD, dl, VT, T1, T2);
+    Highs = DAG.getNode(ISD::SUB, dl, VT, Highs, Fixup);
   }
-  return SDValue();
+
+  return DAG.getNode(ISD::MERGE_VALUES, dl, Op.getValueType(), Highs, Lows);
 }
 
 static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
@@ -13253,7 +13345,7 @@ static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
                                      DAG.getConstant(uint8_t(-1U << ShiftAmt),
                                                      MVT::i8));
           return DAG.getNode(ISD::AND, dl, VT, SHL,
-                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], 16));
+                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, V));
         }
         if (Op.getOpcode() == ISD::SRL) {
           // Make a large shift.
@@ -13266,7 +13358,7 @@ static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
                                      DAG.getConstant(uint8_t(-1U) >> ShiftAmt,
                                                      MVT::i8));
           return DAG.getNode(ISD::AND, dl, VT, SRL,
-                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], 16));
+                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, V));
         }
         if (Op.getOpcode() == ISD::SRA) {
           if (ShiftAmt == 7) {
@@ -13279,7 +13371,7 @@ static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
           SDValue Res = DAG.getNode(ISD::SRL, dl, VT, R, Amt);
           SmallVector<SDValue, 16> V(16, DAG.getConstant(128 >> ShiftAmt,
                                                          MVT::i8));
-          SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], 16);
+          SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, V);
           Res = DAG.getNode(ISD::XOR, dl, VT, Res, Mask);
           Res = DAG.getNode(ISD::SUB, dl, VT, Res, Mask);
           return Res;
@@ -13299,7 +13391,7 @@ static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
                                      DAG.getConstant(uint8_t(-1U << ShiftAmt),
                                                      MVT::i8));
           return DAG.getNode(ISD::AND, dl, VT, SHL,
-                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], 32));
+                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, V));
         }
         if (Op.getOpcode() == ISD::SRL) {
           // Make a large shift.
@@ -13312,7 +13404,7 @@ static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
                                      DAG.getConstant(uint8_t(-1U) >> ShiftAmt,
                                                      MVT::i8));
           return DAG.getNode(ISD::AND, dl, VT, SRL,
-                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], 32));
+                             DAG.getNode(ISD::BUILD_VECTOR, dl, VT, V));
         }
         if (Op.getOpcode() == ISD::SRA) {
           if (ShiftAmt == 7) {
@@ -13325,7 +13417,7 @@ static SDValue LowerScalarImmediateShift(SDValue Op, SelectionDAG &DAG,
           SDValue Res = DAG.getNode(ISD::SRL, dl, VT, R, Amt);
           SmallVector<SDValue, 32> V(32, DAG.getConstant(128 >> ShiftAmt,
                                                          MVT::i8));
-          SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &V[0], 32);
+          SDValue Mask = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, V);
           Res = DAG.getNode(ISD::XOR, dl, VT, Res, Mask);
           Res = DAG.getNode(ISD::SUB, dl, VT, Res, Mask);
           return Res;
@@ -13593,7 +13685,7 @@ static SDValue LowerShift(SDValue Op, const X86Subtarget* Subtarget,
       }
       Elts.push_back(DAG.getConstant(One.shl(ShAmt), SVT));
     }
-    SDValue BV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Elts[0], NumElems);
+    SDValue BV = DAG.getNode(ISD::BUILD_VECTOR, dl, VT, Elts);
     return DAG.getNode(ISD::MUL, dl, VT, R, BV);
   }
 
@@ -13757,10 +13849,8 @@ static SDValue LowerShift(SDValue Op, const X86Subtarget* Subtarget,
       for (unsigned i = NumElems/2; i != NumElems; ++i)
         Amt2Csts.push_back(Amt->getOperand(i));
 
-      Amt1 = DAG.getNode(ISD::BUILD_VECTOR, dl, NewVT,
-                                 &Amt1Csts[0], NumElems/2);
-      Amt2 = DAG.getNode(ISD::BUILD_VECTOR, dl, NewVT,
-                                 &Amt2Csts[0], NumElems/2);
+      Amt1 = DAG.getNode(ISD::BUILD_VECTOR, dl, NewVT, Amt1Csts);
+      Amt2 = DAG.getNode(ISD::BUILD_VECTOR, dl, NewVT, Amt2Csts);
     } else {
       // Variable shift amount
       Amt1 = Extract128BitVector(Amt, 0, DAG, dl);
@@ -13991,7 +14081,7 @@ static SDValue LowerCMP_SWAP(SDValue Op, const X86Subtarget *Subtarget,
   SDVTList Tys = DAG.getVTList(MVT::Other, MVT::Glue);
   MachineMemOperand *MMO = cast<AtomicSDNode>(Op)->getMemOperand();
   SDValue Result = DAG.getMemIntrinsicNode(X86ISD::LCMPXCHG_DAG, DL, Tys,
-                                           Ops, array_lengthof(Ops), T, MMO);
+                                           Ops, T, MMO);
   SDValue cpOut =
     DAG.getCopyFromReg(Result.getValue(0), DL, Reg, T, Result.getValue(1));
   return cpOut;
@@ -14201,6 +14291,8 @@ SDValue X86TargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::CTLZ_ZERO_UNDEF:    return LowerCTLZ_ZERO_UNDEF(Op, DAG);
   case ISD::CTTZ:               return LowerCTTZ(Op, DAG);
   case ISD::MUL:                return LowerMUL(Op, Subtarget, DAG);
+  case ISD::UMUL_LOHI:
+  case ISD::SMUL_LOHI:          return LowerMUL_LOHI(Op, Subtarget, DAG);
   case ISD::SRA:
   case ISD::SRL:
   case ISD::SHL:                return LowerShift(Op, Subtarget, DAG);
@@ -14218,7 +14310,6 @@ SDValue X86TargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::SUBE:               return LowerADDC_ADDE_SUBC_SUBE(Op, DAG);
   case ISD::ADD:                return LowerADD(Op, DAG);
   case ISD::SUB:                return LowerSUB(Op, DAG);
-  case ISD::SDIV:               return LowerSDIV(Op, DAG);
   case ISD::FSINCOS:            return LowerFSINCOS(Op, Subtarget, DAG);
   }
 }
@@ -14261,10 +14352,10 @@ ReplaceATOMIC_BINARY_64(SDNode *Node, SmallVectorImpl<SDValue>&Results,
   SDValue Ops[] = { Chain, In1, In2L, In2H };
   SDVTList Tys = DAG.getVTList(MVT::i32, MVT::i32, MVT::Other);
   SDValue Result =
-    DAG.getMemIntrinsicNode(NewOp, dl, Tys, Ops, array_lengthof(Ops), MVT::i64,
+    DAG.getMemIntrinsicNode(NewOp, dl, Tys, Ops, MVT::i64,
                             cast<MemSDNode>(Node)->getMemOperand());
   SDValue OpsF[] = { Result.getValue(0), Result.getValue(1)};
-  Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, MVT::i64, OpsF, 2));
+  Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, MVT::i64, OpsF));
   Results.push_back(Result.getValue(2));
 }
 
@@ -14285,6 +14376,16 @@ void X86TargetLowering::ReplaceNodeResults(SDNode *N,
   case ISD::SUBE:
     // We don't want to expand or promote these.
     return;
+  case ISD::SDIV:
+  case ISD::UDIV:
+  case ISD::SREM:
+  case ISD::UREM:
+  case ISD::SDIVREM:
+  case ISD::UDIVREM: {
+    SDValue V = LowerWin64_i128OP(SDValue(N,0), DAG);
+    Results.push_back(V);
+    return;
+  }
   case ISD::FP_TO_SINT:
   case ISD::FP_TO_UINT: {
     bool IsSigned = N->getOpcode() == ISD::FP_TO_SINT;
@@ -14382,8 +14483,7 @@ void X86TargetLowering::ReplaceNodeResults(SDNode *N,
     MachineMemOperand *MMO = cast<AtomicSDNode>(N)->getMemOperand();
     unsigned Opcode = Regs64bit ? X86ISD::LCMPXCHG16_DAG :
                                   X86ISD::LCMPXCHG8_DAG;
-    SDValue Result = DAG.getMemIntrinsicNode(Opcode, dl, Tys,
-                                             Ops, array_lengthof(Ops), T, MMO);
+    SDValue Result = DAG.getMemIntrinsicNode(Opcode, dl, Tys, Ops, T, MMO);
     SDValue cpOutL = DAG.getCopyFromReg(Result.getValue(0), dl,
                                         Regs64bit ? X86::RAX : X86::EAX,
                                         HalfT, Result.getValue(1));
@@ -14391,7 +14491,7 @@ void X86TargetLowering::ReplaceNodeResults(SDNode *N,
                                         Regs64bit ? X86::RDX : X86::EDX,
                                         HalfT, cpOutL.getValue(2));
     SDValue OpsF[] = { cpOutL.getValue(0), cpOutH.getValue(0)};
-    Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, T, OpsF, 2));
+    Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, T, OpsF));
     Results.push_back(cpOutH.getValue(1));
     return;
   }
@@ -14598,6 +14698,7 @@ const char *X86TargetLowering::getTargetNodeName(unsigned Opcode) const {
   case X86ISD::VPERMIV3:           return "X86ISD::VPERMIV3";
   case X86ISD::VPERMI:             return "X86ISD::VPERMI";
   case X86ISD::PMULUDQ:            return "X86ISD::PMULUDQ";
+  case X86ISD::PMULDQ:             return "X86ISD::PMULDQ";
   case X86ISD::VASTART_SAVE_XMM_REGS: return "X86ISD::VASTART_SAVE_XMM_REGS";
   case X86ISD::VAARG_64:           return "X86ISD::VAARG_64";
   case X86ISD::WIN_ALLOCA:         return "X86ISD::WIN_ALLOCA";
@@ -17069,7 +17170,6 @@ static SDValue PerformShuffleCombine256(SDNode *N, SelectionDAG &DAG,
         SDValue Ops[] = { Ld->getChain(), Ld->getBasePtr() };
         SDValue ResNode =
           DAG.getMemIntrinsicNode(X86ISD::VZEXT_LOAD, dl, Tys, Ops,
-                                  array_lengthof(Ops),
                                   Ld->getMemoryVT(),
                                   Ld->getPointerInfo(),
                                   Ld->getAlignment(),
@@ -18130,8 +18230,7 @@ static SDValue PerformCMOVCombine(SDNode *N, SelectionDAG &DAG,
       (FalseOp.getValueType() != MVT::f80 || hasFPCMov(CC))) {
     SDValue Ops[] = { FalseOp, TrueOp,
                       DAG.getConstant(CC, MVT::i8), Flags };
-    return DAG.getNode(X86ISD::CMOV, DL, N->getVTList(),
-                       Ops, array_lengthof(Ops));
+    return DAG.getNode(X86ISD::CMOV, DL, N->getVTList(), Ops);
   }
 
   // If this is a select between two integer constants, try to do some
@@ -18261,8 +18360,7 @@ static SDValue PerformCMOVCombine(SDNode *N, SelectionDAG &DAG,
           CmpAgainst == dyn_cast<ConstantSDNode>(TrueOp)) {
         SDValue Ops[] = { FalseOp, Cond.getOperand(0),
                           DAG.getConstant(CC, MVT::i8), Cond };
-        return DAG.getNode(X86ISD::CMOV, DL, N->getVTList (), Ops,
-                           array_lengthof(Ops));
+        return DAG.getNode(X86ISD::CMOV, DL, N->getVTList (), Ops);
       }
     }
   }
@@ -18613,7 +18711,7 @@ static SDValue WidenMaskArithmetic(SDNode *N, SelectionDAG &DAG,
     N1 = DAG.getNode(ISD::ZERO_EXTEND, DL, WideVT.getScalarType(),
                      N1->getOperand(0));
     SmallVector<SDValue, 8> C(WideVT.getVectorNumElements(), N1);
-    N1 = DAG.getNode(ISD::BUILD_VECTOR, DL, WideVT, &C[0], C.size());
+    N1 = DAG.getNode(ISD::BUILD_VECTOR, DL, WideVT, C);
   } else if (RHSTrunc) {
     N1 = N1->getOperand(0);
   }
@@ -18896,8 +18994,7 @@ static SDValue performIntegerAbsCombine(SDNode *N, SelectionDAG &DAG) {
         SDValue Ops[] = { N0.getOperand(0), Neg,
                           DAG.getConstant(X86::COND_GE, MVT::i8),
                           SDValue(Neg.getNode(), 1) };
-        return DAG.getNode(X86ISD::CMOV, DL, DAG.getVTList(VT, MVT::Glue),
-                           Ops, array_lengthof(Ops));
+        return DAG.getNode(X86ISD::CMOV, DL, DAG.getVTList(VT, MVT::Glue), Ops);
       }
   return SDValue();
 }
@@ -19054,8 +19151,7 @@ static SDValue PerformLOADCombine(SDNode *N, SelectionDAG &DAG,
       Ptr = DAG.getNode(ISD::ADD, dl, Ptr.getValueType(), Ptr, Increment);
     }
 
-    SDValue TF = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &Chains[0],
-                               Chains.size());
+    SDValue TF = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Chains);
 
     // Bitcast the loaded value to a vector of the original element type, in
     // the size of the target vector type.
@@ -19230,8 +19326,7 @@ static SDValue PerformSTORECombine(SDNode *N, SelectionDAG &DAG,
       Chains.push_back(Ch);
     }
 
-    return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, &Chains[0],
-                               Chains.size());
+    return DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Chains);
   }
 
   // Turn load->store of MMX types into GPR load/stores.  This avoids clobbering
@@ -19297,8 +19392,7 @@ static SDValue PerformSTORECombine(SDNode *N, SelectionDAG &DAG,
       SDValue NewChain = NewLd.getValue(1);
       if (TokenFactorIndex != -1) {
         Ops.push_back(NewChain);
-        NewChain = DAG.getNode(ISD::TokenFactor, LdDL, MVT::Other, &Ops[0],
-                               Ops.size());
+        NewChain = DAG.getNode(ISD::TokenFactor, LdDL, MVT::Other, Ops);
       }
       return DAG.getStore(NewChain, StDL, NewLd, St->getBasePtr(),
                           St->getPointerInfo(),
@@ -19325,8 +19419,7 @@ static SDValue PerformSTORECombine(SDNode *N, SelectionDAG &DAG,
     if (TokenFactorIndex != -1) {
       Ops.push_back(LoLd);
       Ops.push_back(HiLd);
-      NewChain = DAG.getNode(ISD::TokenFactor, LdDL, MVT::Other, &Ops[0],
-                             Ops.size());
+      NewChain = DAG.getNode(ISD::TokenFactor, LdDL, MVT::Other, Ops);
     }
 
     LoAddr = St->getBasePtr();
@@ -20856,8 +20949,22 @@ int X86TargetLowering::getScalingFactorCost(const AddrMode &AM,
                                             Type *Ty) const {
   // Scaling factors are not free at all.
   // An indexed folded instruction, i.e., inst (reg1, reg2, scale),
-  // will take 2 allocations instead of 1 for plain addressing mode,
-  // i.e. inst (reg1).
+  // will take 2 allocations in the out of order engine instead of 1
+  // for plain addressing mode, i.e. inst (reg1).
+  // E.g.,
+  // vaddps (%rsi,%drx), %ymm0, %ymm1
+  // Requires two allocations (one for the load, one for the computation)
+  // whereas:
+  // vaddps (%rsi), %ymm0, %ymm1
+  // Requires just 1 allocation, i.e., freeing allocations for other operations
+  // and having less micro operations to execute.
+  //
+  // For some X86 architectures, this is even worse because for instance for
+  // stores, the complex addressing mode forces the instruction to use the
+  // "load" ports instead of the dedicated "store" port.
+  // E.g., on Haswell:
+  // vmovaps %ymm1, (%r8, %rdi) can use port 2 or 3.
+  // vmovaps %ymm1, (%r8) can use port 2, 3, or 7.   
   if (isLegalAddressingMode(AM, Ty))
     // Scale represents reg2 * scale, thus account for 1
     // as soon as we use a second register.