Updating branches/google/stable to r169803

git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/google/stable@170212 91177308-0d34-0410-b5e6-96231b3b80d8

24 files changed, 284 insertions, 108 deletions

diff --git a/lib/Target/NVPTX/NVPTX.h b/lib/Target/NVPTX/NVPTX.h
index a8d082a4d8..b29a2962aa 100644
--- a/lib/Target/NVPTX/NVPTX.h
+++ b/lib/Target/NVPTX/NVPTX.h
@@ -15,11 +15,11 @@
 #ifndef LLVM_TARGET_NVPTX_H
 #define LLVM_TARGET_NVPTX_H
 
-#include "llvm/Value.h"
+#include "MCTargetDesc/NVPTXBaseInfo.h"
 #include "llvm/Module.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Target/TargetMachine.h"
-#include "MCTargetDesc/NVPTXBaseInfo.h"
+#include "llvm/Value.h"
 #include <cassert>
 #include <iosfwd>
 
diff --git a/lib/Target/NVPTX/NVPTX.td b/lib/Target/NVPTX/NVPTX.td
index ae7710e54f..7aee3595c6 100644
--- a/lib/Target/NVPTX/NVPTX.td
+++ b/lib/Target/NVPTX/NVPTX.td
@@ -24,7 +24,30 @@ include "NVPTXInstrInfo.td"
 // - Need at least one feature to avoid generating zero sized array by
 //   TableGen in NVPTXGenSubtarget.inc.
 //===----------------------------------------------------------------------===//
-def FeatureDummy  : SubtargetFeature<"dummy", "dummy", "true", "">;
+
+// SM Versions
+def SM10 : SubtargetFeature<"sm_10", "SmVersion", "10",
+                            "Target SM 1.0">;
+def SM11 : SubtargetFeature<"sm_11", "SmVersion", "11",
+                            "Target SM 1.1">;
+def SM12 : SubtargetFeature<"sm_12", "SmVersion", "12",
+                            "Target SM 1.2">;
+def SM13 : SubtargetFeature<"sm_13", "SmVersion", "13",
+                            "Target SM 1.3">;
+def SM20 : SubtargetFeature<"sm_20", "SmVersion", "20",
+                            "Target SM 2.0">;
+def SM21 : SubtargetFeature<"sm_21", "SmVersion", "21",
+                            "Target SM 2.1">;
+def SM30 : SubtargetFeature<"sm_30", "SmVersion", "30",
+                            "Target SM 3.0">;
+def SM35 : SubtargetFeature<"sm_35", "SmVersion", "35",
+                            "Target SM 3.5">;
+
+// PTX Versions
+def PTX30 : SubtargetFeature<"ptx30", "PTXVersion", "30",
+                             "Use PTX version 3.0">;
+def PTX31 : SubtargetFeature<"ptx31", "PTXVersion", "31",
+                             "Use PTX version 3.1">;
 
 //===----------------------------------------------------------------------===//
 // NVPTX supported processors.
@@ -33,7 +56,14 @@ def FeatureDummy  : SubtargetFeature<"dummy", "dummy", "true", "">;
 class Proc<string Name, list<SubtargetFeature> Features>
  : Processor<Name, NoItineraries, Features>;
 
-def : Proc<"sm_10", [FeatureDummy]>;
+def : Proc<"sm_10", [SM10]>;
+def : Proc<"sm_11", [SM11]>;
+def : Proc<"sm_12", [SM12]>;
+def : Proc<"sm_13", [SM13]>;
+def : Proc<"sm_20", [SM20]>;
+def : Proc<"sm_21", [SM21]>;
+def : Proc<"sm_30", [SM30]>;
+def : Proc<"sm_35", [SM35]>;
 
 
 def NVPTXInstrInfo : InstrInfo {
diff --git a/lib/Target/NVPTX/NVPTXAllocaHoisting.cpp b/lib/Target/NVPTX/NVPTXAllocaHoisting.cpp
index 668c39308f..38c7083fea 100644
--- a/lib/Target/NVPTX/NVPTXAllocaHoisting.cpp
+++ b/lib/Target/NVPTX/NVPTXAllocaHoisting.cpp
@@ -11,10 +11,10 @@
 //
 //===----------------------------------------------------------------------===//
 
+#include "NVPTXAllocaHoisting.h"
+#include "llvm/Constants.h"
 #include "llvm/Function.h"
 #include "llvm/Instructions.h"
-#include "llvm/Constants.h"
-#include "NVPTXAllocaHoisting.h"
 
 namespace llvm {
 
diff --git a/lib/Target/NVPTX/NVPTXAllocaHoisting.h b/lib/Target/NVPTX/NVPTXAllocaHoisting.h
index c7cabf6953..768d514dbb 100644
--- a/lib/Target/NVPTX/NVPTXAllocaHoisting.h
+++ b/lib/Target/NVPTX/NVPTXAllocaHoisting.h
@@ -15,8 +15,8 @@
 #define NVPTX_ALLOCA_HOISTING_H_
 
 #include "llvm/CodeGen/MachineFunctionAnalysis.h"
-#include "llvm/Pass.h"
 #include "llvm/DataLayout.h"
+#include "llvm/Pass.h"
 
 namespace llvm {
 
diff --git a/lib/Target/NVPTX/NVPTXAsmPrinter.cpp b/lib/Target/NVPTX/NVPTXAsmPrinter.cpp
index d3dfb35e26..af07576d59 100644
--- a/lib/Target/NVPTX/NVPTXAsmPrinter.cpp
+++ b/lib/Target/NVPTX/NVPTXAsmPrinter.cpp
@@ -13,36 +13,36 @@
 //===----------------------------------------------------------------------===//
 
 #include "NVPTXAsmPrinter.h"
+#include "MCTargetDesc/NVPTXMCAsmInfo.h"
 #include "NVPTX.h"
 #include "NVPTXInstrInfo.h"
-#include "NVPTXTargetMachine.h"
+#include "NVPTXNumRegisters.h"
 #include "NVPTXRegisterInfo.h"
+#include "NVPTXTargetMachine.h"
 #include "NVPTXUtilities.h"
-#include "MCTargetDesc/NVPTXMCAsmInfo.h"
-#include "NVPTXNumRegisters.h"
+#include "cl_common_defines.h"
 #include "llvm/ADT/StringExtras.h"
-#include "llvm/DebugInfo.h"
-#include "llvm/Function.h"
-#include "llvm/GlobalVariable.h"
-#include "llvm/Module.h"
+#include "llvm/Analysis/ConstantFolding.h"
+#include "llvm/Assembly/Writer.h"
 #include "llvm/CodeGen/Analysis.h"
-#include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineModuleInfo.h"
+#include "llvm/CodeGen/MachineRegisterInfo.h"
+#include "llvm/DebugInfo.h"
+#include "llvm/DerivedTypes.h"
+#include "llvm/Function.h"
+#include "llvm/GlobalVariable.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSymbol.h"
-#include "llvm/Target/Mangler.h"
-#include "llvm/Target/TargetLoweringObjectFile.h"
-#include "llvm/Support/TargetRegistry.h"
+#include "llvm/Module.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/FormattedStream.h"
-#include "llvm/DerivedTypes.h"
-#include "llvm/Support/TimeValue.h"
-#include "llvm/Support/CommandLine.h"
-#include "llvm/Analysis/ConstantFolding.h"
 #include "llvm/Support/Path.h"
-#include "llvm/Assembly/Writer.h"
-#include "cl_common_defines.h"
+#include "llvm/Support/TargetRegistry.h"
+#include "llvm/Support/TimeValue.h"
+#include "llvm/Target/Mangler.h"
+#include "llvm/Target/TargetLoweringObjectFile.h"
 #include <sstream>
 using namespace llvm;
 
@@ -68,7 +68,54 @@ static cl::opt<bool, true>InterleaveSrc("nvptx-emit-src",
                                         cl::location(llvm::InterleaveSrcInPtx));
 
 
+namespace {
+/// DiscoverDependentGlobals - Return a set of GlobalVariables on which \p V
+/// depends.
+void DiscoverDependentGlobals(Value *V,
+                              DenseSet<GlobalVariable*> &Globals) {
+  if (GlobalVariable *GV = dyn_cast<GlobalVariable>(V))
+    Globals.insert(GV);
+  else {
+    if (User *U = dyn_cast<User>(V)) {
+      for (unsigned i = 0, e = U->getNumOperands(); i != e; ++i) {
+        DiscoverDependentGlobals(U->getOperand(i), Globals);
+      }
+    }
+  }
+}
 
+/// VisitGlobalVariableForEmission - Add \p GV to the list of GlobalVariable
+/// instances to be emitted, but only after any dependents have been added
+/// first.
+void VisitGlobalVariableForEmission(GlobalVariable *GV,
+                                    SmallVectorImpl<GlobalVariable*> &Order,
+                                    DenseSet<GlobalVariable*> &Visited,
+                                    DenseSet<GlobalVariable*> &Visiting) {
+  // Have we already visited this one?
+  if (Visited.count(GV)) return;
+
+  // Do we have a circular dependency?
+  if (Visiting.count(GV))
+    report_fatal_error("Circular dependency found in global variable set");
+
+  // Start visiting this global
+  Visiting.insert(GV);
+
+  // Make sure we visit all dependents first
+  DenseSet<GlobalVariable*> Others;
+  for (unsigned i = 0, e = GV->getNumOperands(); i != e; ++i)
+    DiscoverDependentGlobals(GV->getOperand(i), Others);
+  
+  for (DenseSet<GlobalVariable*>::iterator I = Others.begin(),
+       E = Others.end(); I != E; ++I)
+    VisitGlobalVariableForEmission(*I, Order, Visited, Visiting);
+
+  // Now we can visit ourself
+  Order.push_back(GV);
+  Visited.insert(GV);
+  Visiting.erase(GV);
+}
+}
 
 // @TODO: This is a copy from AsmPrinter.cpp.  The function is static, so we
 // cannot just link to the existing version.
@@ -631,7 +678,7 @@ void NVPTXAsmPrinter::printLdStCode(const MachineInstr *MI, int opNum,
           O << ".global";
         break;
       default:
-        assert("wrong value");
+        llvm_unreachable("Wrong Address Space");
       }
     }
     else if (!strcmp(Modifier, "sign")) {
@@ -649,10 +696,10 @@ void NVPTXAsmPrinter::printLdStCode(const MachineInstr *MI, int opNum,
         O << ".v4";
     }
     else
-      assert("unknown modifier");
+      llvm_unreachable("Unknown Modifier");
   }
   else
-    assert("unknown modifier");
+    llvm_unreachable("Empty Modifier");
 }
 
 void NVPTXAsmPrinter::emitDeclaration (const Function *F, raw_ostream &O) {
@@ -893,10 +940,27 @@ bool NVPTXAsmPrinter::doInitialization (Module &M) {
 
   emitDeclarations(M, OS2);
 
-  // Print out module-level global variables here.
+  // As ptxas does not support forward references of globals, we need to first
+  // sort the list of module-level globals in def-use order. We visit each
+  // global variable in order, and ensure that we emit it *after* its dependent
+  // globals. We use a little extra memory maintaining both a set and a list to
+  // have fast searches while maintaining a strict ordering.
+  SmallVector<GlobalVariable*,8> Globals;
+  DenseSet<GlobalVariable*> GVVisited;
+  DenseSet<GlobalVariable*> GVVisiting;
+
+  // Visit each global variable, in order
   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
-      I != E; ++I)
-    printModuleLevelGV(I, OS2);
+       I != E; ++I)
+    VisitGlobalVariableForEmission(I, Globals, GVVisited, GVVisiting);
+
+  assert(GVVisited.size() == M.getGlobalList().size() && 
+         "Missed a global variable");
+  assert(GVVisiting.size() == 0 && "Did not fully process a global variable");
+
+  // Print out module-level global variables in proper order
+  for (unsigned i = 0, e = Globals.size(); i != e; ++i)
+    printModuleLevelGV(Globals[i], OS2);
 
   OS2 << '\n';
 
@@ -910,7 +974,8 @@ void NVPTXAsmPrinter::emitHeader (Module &M, raw_ostream &O) {
   O << "//\n";
   O << "\n";
 
-  O << ".version 3.0\n";
+  unsigned PTXVersion = nvptxSubtarget.getPTXVersion();
+  O << ".version " << (PTXVersion / 10) << "." << (PTXVersion % 10) << "\n";
 
   O << ".target ";
   O << nvptxSubtarget.getTargetName();
@@ -1422,7 +1487,7 @@ void NVPTXAsmPrinter::printParamName(int paramIndex, raw_ostream &O) {
 void NVPTXAsmPrinter::emitFunctionParamList(const Function *F,
                                             raw_ostream &O) {
   const DataLayout *TD = TM.getDataLayout();
-  const AttrListPtr &PAL = F->getAttributes();
+  const AttributeSet &PAL = F->getAttributes();
   const TargetLowering *TLI = TM.getTargetLowering();
   Function::const_arg_iterator I, E;
   unsigned paramIndex = 0;
@@ -1525,6 +1590,9 @@ void NVPTXAsmPrinter::emitFunctionParamList(const Function *F,
       // <a> = PAL.getparamalignment
       // size = typeallocsize of element type
       unsigned align = PAL.getParamAlignment(paramIndex+1);
+      if (align == 0)
+        align = TD->getABITypeAlignment(ETy);
+
       unsigned sz = TD->getTypeAllocSize(ETy);
       O << "\t.param .align " << align
           << " .b8 ";
diff --git a/lib/Target/NVPTX/NVPTXAsmPrinter.h b/lib/Target/NVPTX/NVPTXAsmPrinter.h
index 6488b14425..29db4abe6c 100644
--- a/lib/Target/NVPTX/NVPTXAsmPrinter.h
+++ b/lib/Target/NVPTX/NVPTXAsmPrinter.h
@@ -16,19 +16,19 @@
 #define NVPTXASMPRINTER_H
 
 #include "NVPTX.h"
-#include "NVPTXTargetMachine.h"
 #include "NVPTXSubtarget.h"
-#include "llvm/Function.h"
+#include "NVPTXTargetMachine.h"
+#include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/StringExtras.h"
 #include "llvm/CodeGen/AsmPrinter.h"
-#include "llvm/Support/CommandLine.h"
+#include "llvm/Function.h"
 #include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/MCExpr.h"
-#include "llvm/Target/TargetMachine.h"
 #include "llvm/MC/MCSymbol.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/FormattedStream.h"
 #include "llvm/Target/Mangler.h"
-#include "llvm/ADT/SmallString.h"
-#include "llvm/ADT/StringExtras.h"
+#include "llvm/Target/TargetMachine.h"
 #include <fstream>
 
 // The ptx syntax and format is very different from that usually seem in a .s
diff --git a/lib/Target/NVPTX/NVPTXFrameLowering.cpp b/lib/Target/NVPTX/NVPTXFrameLowering.cpp
index a9abc00bf3..50072c55f6 100644
--- a/lib/Target/NVPTX/NVPTXFrameLowering.cpp
+++ b/lib/Target/NVPTX/NVPTXFrameLowering.cpp
@@ -17,9 +17,9 @@
 #include "NVPTXSubtarget.h"
 #include "NVPTXTargetMachine.h"
 #include "llvm/ADT/BitVector.h"
-#include "llvm/CodeGen/MachineInstrBuilder.h"
-#include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
+#include "llvm/CodeGen/MachineFunction.h"
+#include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/MC/MachineLocation.h"
 #include "llvm/Target/TargetInstrInfo.h"
 
diff --git a/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp b/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
index 4e92f0e785..d68e3b6871 100644
--- a/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
+++ b/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
@@ -12,14 +12,14 @@
 //===----------------------------------------------------------------------===//
 
 
-#include "llvm/Instructions.h"
-#include "llvm/Support/raw_ostream.h"
 #include "NVPTXISelDAGToDAG.h"
+#include "llvm/GlobalValue.h"
+#include "llvm/Instructions.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
-#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetIntrinsicInfo.h"
-#include "llvm/GlobalValue.h"
 
 #undef DEBUG_TYPE
 #define DEBUG_TYPE "nvptx-isel"
diff --git a/lib/Target/NVPTX/NVPTXISelDAGToDAG.h b/lib/Target/NVPTX/NVPTXISelDAGToDAG.h
index ccd69b29dd..29e4f17962 100644
--- a/lib/Target/NVPTX/NVPTXISelDAGToDAG.h
+++ b/lib/Target/NVPTX/NVPTXISelDAGToDAG.h
@@ -18,8 +18,8 @@
 #include "NVPTXRegisterInfo.h"
 #include "NVPTXTargetMachine.h"
 #include "llvm/CodeGen/SelectionDAGISel.h"
-#include "llvm/Support/Compiler.h"
 #include "llvm/Intrinsics.h"
+#include "llvm/Support/Compiler.h"
 using namespace llvm;
 
 namespace {
diff --git a/lib/Target/NVPTX/NVPTXISelLowering.cpp b/lib/Target/NVPTX/NVPTXISelLowering.cpp
index be771e3567..4e9d68687a 100644
--- a/lib/Target/NVPTX/NVPTXISelLowering.cpp
+++ b/lib/Target/NVPTX/NVPTXISelLowering.cpp
@@ -12,29 +12,29 @@
 //===----------------------------------------------------------------------===//
 
 
-#include "NVPTX.h"
 #include "NVPTXISelLowering.h"
+#include "NVPTX.h"
 #include "NVPTXTargetMachine.h"
 #include "NVPTXTargetObjectFile.h"
 #include "NVPTXUtilities.h"
-#include "llvm/Intrinsics.h"
-#include "llvm/IntrinsicInst.h"
-#include "llvm/Support/CommandLine.h"
-#include "llvm/DerivedTypes.h"
-#include "llvm/GlobalValue.h"
-#include "llvm/Module.h"
-#include "llvm/Function.h"
 #include "llvm/CodeGen/Analysis.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
+#include "llvm/CodeGen/TargetLoweringObjectFileImpl.h"
+#include "llvm/DerivedTypes.h"
+#include "llvm/Function.h"
+#include "llvm/GlobalValue.h"
+#include "llvm/IntrinsicInst.h"
+#include "llvm/Intrinsics.h"
+#include "llvm/MC/MCSectionELF.h"
+#include "llvm/Module.h"
 #include "llvm/Support/CallSite.h"
-#include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
+#include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/raw_ostream.h"
-#include "llvm/CodeGen/TargetLoweringObjectFileImpl.h"
-#include "llvm/MC/MCSectionELF.h"
 #include <sstream>
 
 #undef DEBUG_TYPE
@@ -174,10 +174,11 @@ NVPTXTargetLowering::NVPTXTargetLowering(NVPTXTargetMachine &TM)
   setTruncStoreAction(MVT::f64, MVT::f32, Expand);
 
   // PTX does not support load / store predicate registers
-  setOperationAction(ISD::LOAD, MVT::i1, Expand);
+  setOperationAction(ISD::LOAD, MVT::i1, Custom);
+  setOperationAction(ISD::STORE, MVT::i1, Custom);
+
   setLoadExtAction(ISD::SEXTLOAD, MVT::i1, Promote);
   setLoadExtAction(ISD::ZEXTLOAD, MVT::i1, Promote);
-  setOperationAction(ISD::STORE, MVT::i1, Expand);
   setTruncStoreAction(MVT::i64, MVT::i1, Expand);
   setTruncStoreAction(MVT::i32, MVT::i1, Expand);
   setTruncStoreAction(MVT::i16, MVT::i1, Expand);
@@ -270,6 +271,9 @@ const char *NVPTXTargetLowering::getTargetNodeName(unsigned Opcode) const {
   }
 }
 
+bool NVPTXTargetLowering::shouldSplitVectorElementType(EVT VT) const {
+  return VT == MVT::i1;
+}
 
 SDValue
 NVPTXTargetLowering::LowerGlobalAddress(SDValue Op, SelectionDAG &DAG) const {
@@ -856,11 +860,64 @@ LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::EXTRACT_SUBVECTOR:
     return Op;
   case ISD::CONCAT_VECTORS: return LowerCONCAT_VECTORS(Op, DAG);
+  case ISD::STORE: return LowerSTORE(Op, DAG);
+  case ISD::LOAD: return LowerLOAD(Op, DAG);
   default:
     llvm_unreachable("Custom lowering not defined for operation");
   }
 }
 
+
+// v = ld i1* addr
+//   =>
+// v1 = ld i8* addr
+// v = trunc v1 to i1
+SDValue NVPTXTargetLowering::
+LowerLOAD(SDValue Op, SelectionDAG &DAG) const {
+  SDNode *Node = Op.getNode();
+  LoadSDNode *LD = cast<LoadSDNode>(Node);
+  DebugLoc dl = Node->getDebugLoc();
+  assert(LD->getExtensionType() == ISD::NON_EXTLOAD) ;
+  assert(Node->getValueType(0) == MVT::i1 &&
+         "Custom lowering for i1 load only");
+  SDValue newLD = DAG.getLoad(MVT::i8, dl, LD->getChain(), LD->getBasePtr(),
+                              LD->getPointerInfo(),
+                              LD->isVolatile(), LD->isNonTemporal(),
+                              LD->isInvariant(),
+                              LD->getAlignment());
+  SDValue result = DAG.getNode(ISD::TRUNCATE, dl, MVT::i1, newLD);
+  // The legalizer (the caller) is expecting two values from the legalized
+  // load, so we build a MergeValues node for it. See ExpandUnalignedLoad()
+  // in LegalizeDAG.cpp which also uses MergeValues.
+  SDValue Ops[] = {result, LD->getChain()};
+  return DAG.getMergeValues(Ops, 2, dl);
+}
+
+// st i1 v, addr
+//    =>
+// v1 = zxt v to i8
+// st i8, addr
+SDValue NVPTXTargetLowering::
+LowerSTORE(SDValue Op, SelectionDAG &DAG) const {
+  SDNode *Node = Op.getNode();
+  DebugLoc dl = Node->getDebugLoc();
+  StoreSDNode *ST = cast<StoreSDNode>(Node);
+  SDValue Tmp1 = ST->getChain();
+  SDValue Tmp2 = ST->getBasePtr();
+  SDValue Tmp3 = ST->getValue();
+  assert(Tmp3.getValueType() == MVT::i1 && "Custom lowering for i1 store only");
+  unsigned Alignment = ST->getAlignment();
+  bool isVolatile = ST->isVolatile();
+  bool isNonTemporal = ST->isNonTemporal();
+  Tmp3 = DAG.getNode(ISD::ZERO_EXTEND, dl,
+                     MVT::i8, Tmp3);
+  SDValue Result = DAG.getStore(Tmp1, dl, Tmp3, Tmp2,
+                                ST->getPointerInfo(), isVolatile,
+                                isNonTemporal, Alignment);
+  return Result;
+}
+
+
 SDValue
 NVPTXTargetLowering::getExtSymb(SelectionDAG &DAG, const char *inname, int idx,
                                 EVT v) const {
@@ -900,7 +957,7 @@ bool llvm::isImageOrSamplerVal(const Value *arg, const Module *context) {
     return false;
 
   const StructType *STy = dyn_cast<StructType>(PTy->getElementType());
-  const std::string TypeName = STy ? STy->getName() : "";
+  const std::string TypeName = STy && !STy->isLiteral() ? STy->getName() : "";
 
   for (int i = 0, e = array_lengthof(specialTypes); i != e; ++i)
     if (TypeName == specialTypes[i])
@@ -919,7 +976,7 @@ NVPTXTargetLowering::LowerFormalArguments(SDValue Chain,
   const DataLayout *TD = getDataLayout();
 
   const Function *F = MF.getFunction();
-  const AttrListPtr &PAL = F->getAttributes();
+  const AttributeSet &PAL = F->getAttributes();
 
   SDValue Root = DAG.getRoot();
   std::vector<SDValue> OutChains;
diff --git a/lib/Target/NVPTX/NVPTXISelLowering.h b/lib/Target/NVPTX/NVPTXISelLowering.h
index 86246e6449..0a1833a7c9 100644
--- a/lib/Target/NVPTX/NVPTXISelLowering.h
+++ b/lib/Target/NVPTX/NVPTXISelLowering.h
@@ -92,6 +92,8 @@ public:
   virtual unsigned getFunctionAlignment(const Function *F) const;
 
   virtual EVT getSetCCResultType(EVT VT) const {
+    if (VT.isVector())
+      return MVT::getVectorVT(MVT::i1, VT.getVectorNumElements());
     return MVT::i1;
   }
 
@@ -129,6 +131,8 @@ public:
     return MVT::i32;
   }
 
+  virtual bool shouldSplitVectorElementType(EVT VT) const;
+
 private:
   const NVPTXSubtarget &nvptxSubtarget;  // cache the subtarget here
 
@@ -138,6 +142,9 @@ private:
   SDValue getParamHelpSymbol(SelectionDAG &DAG, int idx);
 
   SDValue LowerCONCAT_VECTORS(SDValue Op, SelectionDAG &DAG) const;
+
+  SDValue LowerSTORE(SDValue Op, SelectionDAG &DAG) const;
+  SDValue LowerLOAD(SDValue Op, SelectionDAG &DAG) const;
 };
 } // namespace llvm
 
diff --git a/lib/Target/NVPTX/NVPTXLowerAggrCopies.cpp b/lib/Target/NVPTX/NVPTXLowerAggrCopies.cpp
index 9273931e99..8e40c965bf 100644
--- a/lib/Target/NVPTX/NVPTXLowerAggrCopies.cpp
+++ b/lib/Target/NVPTX/NVPTXLowerAggrCopies.cpp
@@ -13,6 +13,7 @@
 
 #include "NVPTXLowerAggrCopies.h"
 #include "llvm/Constants.h"
+#include "llvm/DataLayout.h"
 #include "llvm/Function.h"
 #include "llvm/IRBuilder.h"
 #include "llvm/Instructions.h"
@@ -21,7 +22,6 @@
 #include "llvm/LLVMContext.h"
 #include "llvm/Module.h"
 #include "llvm/Support/InstIterator.h"
-#include "llvm/DataLayout.h"
 
 using namespace llvm;
 
diff --git a/lib/Target/NVPTX/NVPTXLowerAggrCopies.h b/lib/Target/NVPTX/NVPTXLowerAggrCopies.h
index b150c69815..9bafce274e 100644
--- a/lib/Target/NVPTX/NVPTXLowerAggrCopies.h
+++ b/lib/Target/NVPTX/NVPTXLowerAggrCopies.h
@@ -15,9 +15,9 @@
 #ifndef NVPTX_LOWER_AGGR_COPIES_H
 #define NVPTX_LOWER_AGGR_COPIES_H
 
-#include "llvm/Pass.h"
 #include "llvm/CodeGen/MachineFunctionAnalysis.h"
 #include "llvm/DataLayout.h"
+#include "llvm/Pass.h"
 
 namespace llvm {
 
diff --git a/lib/Target/NVPTX/NVPTXRegisterInfo.cpp b/lib/Target/NVPTX/NVPTXRegisterInfo.cpp
index e3cd46f063..08be917bb7 100644
--- a/lib/Target/NVPTX/NVPTXRegisterInfo.cpp
+++ b/lib/Target/NVPTX/NVPTXRegisterInfo.cpp
@@ -13,13 +13,13 @@
 
 #define DEBUG_TYPE "nvptx-reg-info"
 
-#include "NVPTX.h"
 #include "NVPTXRegisterInfo.h"
+#include "NVPTX.h"
 #include "NVPTXSubtarget.h"
 #include "llvm/ADT/BitVector.h"
-#include "llvm/CodeGen/MachineInstrBuilder.h"
-#include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
+#include "llvm/CodeGen/MachineFunction.h"
+#include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/MC/MachineLocation.h"
 #include "llvm/Target/TargetInstrInfo.h"
 
diff --git a/lib/Target/NVPTX/NVPTXSection.h b/lib/Target/NVPTX/NVPTXSection.h
index f1ca466266..9c832e1b51 100644
--- a/lib/Target/NVPTX/NVPTXSection.h
+++ b/lib/Target/NVPTX/NVPTXSection.h
@@ -14,8 +14,8 @@
 #ifndef LLVM_NVPTXSECTION_H
 #define LLVM_NVPTXSECTION_H
 
-#include "llvm/MC/MCSection.h"
 #include "llvm/GlobalVariable.h"
+#include "llvm/MC/MCSection.h"
 #include <vector>
 
 namespace llvm {
diff --git a/lib/Target/NVPTX/NVPTXSplitBBatBar.cpp b/lib/Target/NVPTX/NVPTXSplitBBatBar.cpp
index 2836cad4f0..6171292eeb 100644
--- a/lib/Target/NVPTX/NVPTXSplitBBatBar.cpp
+++ b/lib/Target/NVPTX/NVPTXSplitBBatBar.cpp
@@ -11,13 +11,13 @@
 //
 //===----------------------------------------------------------------------===//
 
+#include "NVPTXSplitBBatBar.h"
+#include "NVPTXUtilities.h"
 #include "llvm/Function.h"
 #include "llvm/Instructions.h"
-#include "llvm/Intrinsics.h"
 #include "llvm/IntrinsicInst.h"
+#include "llvm/Intrinsics.h"
 #include "llvm/Support/InstIterator.h"
-#include "NVPTXUtilities.h"
-#include "NVPTXSplitBBatBar.h"
 
 using namespace llvm;
 
diff --git a/lib/Target/NVPTX/NVPTXSplitBBatBar.h b/lib/Target/NVPTX/NVPTXSplitBBatBar.h
index 9e4d5a066d..bdafba9075 100644
--- a/lib/Target/NVPTX/NVPTXSplitBBatBar.h
+++ b/lib/Target/NVPTX/NVPTXSplitBBatBar.h
@@ -15,8 +15,8 @@
 #ifndef NVPTX_SPLIT_BB_AT_BAR_H
 #define NVPTX_SPLIT_BB_AT_BAR_H
 
-#include "llvm/Pass.h"
 #include "llvm/CodeGen/MachineFunctionAnalysis.h"
+#include "llvm/Pass.h"
 
 namespace llvm {
 
diff --git a/lib/Target/NVPTX/NVPTXSubtarget.cpp b/lib/Target/NVPTX/NVPTXSubtarget.cpp
index 6aadd43e94..7b62cce2c6 100644
--- a/lib/Target/NVPTX/NVPTXSubtarget.cpp
+++ b/lib/Target/NVPTX/NVPTXSubtarget.cpp
@@ -34,16 +34,18 @@ DriverInterface(cl::desc("Choose driver interface:"),
 
 NVPTXSubtarget::NVPTXSubtarget(const std::string &TT, const std::string &CPU,
                                const std::string &FS, bool is64Bit)
-:NVPTXGenSubtargetInfo(TT, "", FS), // Don't pass CPU to subtarget,
- // because we don't register all
- // nvptx targets.
- Is64Bit(is64Bit) {
+: NVPTXGenSubtargetInfo(TT, CPU, FS),
+  Is64Bit(is64Bit),
+  PTXVersion(0),
+  SmVersion(10) {
 
   drvInterface = DriverInterface;
 
   // Provide the default CPU if none
   std::string defCPU = "sm_10";
 
+  ParseSubtargetFeatures((CPU.empty() ? defCPU : CPU), FS);
+
   // Get the TargetName from the FS if available
   if (FS.empty() && CPU.empty())
     TargetName = defCPU;
@@ -52,6 +54,12 @@ NVPTXSubtarget::NVPTXSubtarget(const std::string &TT, const std::string &CPU,
   else
     llvm_unreachable("we are not using FeatureStr");
 
-  // Set up the SmVersion
-  SmVersion = atoi(TargetName.c_str()+3);
+  // We default to PTX 3.1, but we cannot just default to it in the initializer
+  // since the attribute parser checks if the given option is >= the default.
+  // So if we set ptx31 as the default, the ptx30 attribute would never match.
+  // Instead, we use 0 as the default and manually set 31 if the default is
+  // used.
+  if (PTXVersion == 0) {
+    PTXVersion = 31;
+  }
 }
diff --git a/lib/Target/NVPTX/NVPTXSubtarget.h b/lib/Target/NVPTX/NVPTXSubtarget.h
index 8f2a629d22..e6cb7c2def 100644
--- a/lib/Target/NVPTX/NVPTXSubtarget.h
+++ b/lib/Target/NVPTX/NVPTXSubtarget.h
@@ -14,8 +14,8 @@
 #ifndef NVPTXSUBTARGET_H
 #define NVPTXSUBTARGET_H
 
-#include "llvm/Target/TargetSubtargetInfo.h"
 #include "NVPTX.h"
+#include "llvm/Target/TargetSubtargetInfo.h"
 
 #define GET_SUBTARGETINFO_HEADER
 #include "NVPTXGenSubtargetInfo.inc"
@@ -25,13 +25,17 @@
 namespace llvm {
 
 class NVPTXSubtarget : public NVPTXGenSubtargetInfo {
-
-  unsigned int SmVersion;
+  
   std::string TargetName;
   NVPTX::DrvInterface drvInterface;
-  bool dummy; // For the 'dummy' feature, see NVPTX.td
   bool Is64Bit;
 
+  // PTX version x.y is represented as 10*x+y, e.g. 3.1 == 31
+  unsigned PTXVersion;
+
+  // SM version x.y is represented as 10*x+y, e.g. 3.1 == 31
+  unsigned int SmVersion;
+
 public:
   /// This constructor initializes the data members to match that
   /// of the specified module.
@@ -69,6 +73,8 @@ public:
   NVPTX::DrvInterface getDrvInterface() const { return drvInterface; }
   std::string getTargetName() const { return TargetName; }
 
+  unsigned getPTXVersion() const { return PTXVersion; }
+
   void ParseSubtargetFeatures(StringRef CPU, StringRef FS);
 
   std::string getDataLayout() const {
diff --git a/lib/Target/NVPTX/NVPTXTargetMachine.cpp b/lib/Target/NVPTX/NVPTXTargetMachine.cpp
index cbb490003d..0867a4e01e 100644
--- a/lib/Target/NVPTX/NVPTXTargetMachine.cpp
+++ b/lib/Target/NVPTX/NVPTXTargetMachine.cpp
@@ -12,27 +12,31 @@
 //===----------------------------------------------------------------------===//
 
 #include "NVPTXTargetMachine.h"
-#include "NVPTX.h"
-#include "NVPTXSplitBBatBar.h"
-#include "NVPTXLowerAggrCopies.h"
 #include "MCTargetDesc/NVPTXMCAsmInfo.h"
+#include "NVPTX.h"
 #include "NVPTXAllocaHoisting.h"
-#include "llvm/PassManager.h"
+#include "NVPTXLowerAggrCopies.h"
+#include "NVPTXSplitBBatBar.h"
+#include "llvm/ADT/OwningPtr.h"
 #include "llvm/Analysis/Passes.h"
 #include "llvm/Analysis/Verifier.h"
 #include "llvm/Assembly/PrintModulePass.h"
-#include "llvm/ADT/OwningPtr.h"
 #include "llvm/CodeGen/AsmPrinter.h"
 #include "llvm/CodeGen/MachineFunctionAnalysis.h"
 #include "llvm/CodeGen/MachineModuleInfo.h"
 #include "llvm/CodeGen/Passes.h"
+#include "llvm/DataLayout.h"
 #include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/MCInstrInfo.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/MC/MCSubtargetInfo.h"
+#include "llvm/PassManager.h"
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Support/FormattedStream.h"
+#include "llvm/Support/TargetRegistry.h"
 #include "llvm/Support/TargetRegistry.h"
 #include "llvm/Support/raw_ostream.h"
-#include "llvm/DataLayout.h"
 #include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetLowering.h"
 #include "llvm/Target/TargetLoweringObjectFile.h"
@@ -41,10 +45,6 @@
 #include "llvm/Target/TargetRegisterInfo.h"
 #include "llvm/Target/TargetSubtargetInfo.h"
 #include "llvm/Transforms/Scalar.h"
-#include "llvm/Support/CommandLine.h"
-#include "llvm/Support/Debug.h"
-#include "llvm/Support/FormattedStream.h"
-#include "llvm/Support/TargetRegistry.h"
 
 
 using namespace llvm;
diff --git a/lib/Target/NVPTX/NVPTXTargetMachine.h b/lib/Target/NVPTX/NVPTXTargetMachine.h
index 11bc9d4fa6..b4763a5bae 100644
--- a/lib/Target/NVPTX/NVPTXTargetMachine.h
+++ b/lib/Target/NVPTX/NVPTXTargetMachine.h
@@ -15,12 +15,12 @@
 #ifndef NVPTX_TARGETMACHINE_H
 #define NVPTX_TARGETMACHINE_H
 
-#include "NVPTXInstrInfo.h"
+#include "ManagedStringPool.h"
+#include "NVPTXFrameLowering.h"
 #include "NVPTXISelLowering.h"
+#include "NVPTXInstrInfo.h"
 #include "NVPTXRegisterInfo.h"
 #include "NVPTXSubtarget.h"
-#include "NVPTXFrameLowering.h"
-#include "ManagedStringPool.h"
 #include "llvm/DataLayout.h"
 #include "llvm/Target/TargetFrameLowering.h"
 #include "llvm/Target/TargetMachine.h"
diff --git a/lib/Target/NVPTX/NVPTXUtilities.cpp b/lib/Target/NVPTX/NVPTXUtilities.cpp
index 3f52251cc1..fff3f43bb0 100644
--- a/lib/Target/NVPTX/NVPTXUtilities.cpp
+++ b/lib/Target/NVPTX/NVPTXUtilities.cpp
@@ -12,10 +12,10 @@
 
 #include "NVPTXUtilities.h"
 #include "NVPTX.h"
-#include "llvm/GlobalVariable.h"
+#include "llvm/Constants.h"
 #include "llvm/Function.h"
+#include "llvm/GlobalVariable.h"
 #include "llvm/Module.h"
-#include "llvm/Constants.h"
 #include "llvm/Operator.h"
 #include <algorithm>
 #include <cstring>
diff --git a/lib/Target/NVPTX/NVPTXUtilities.h b/lib/Target/NVPTX/NVPTXUtilities.h
index fe6ad559e9..5ea7bc8640 100644
--- a/lib/Target/NVPTX/NVPTXUtilities.h
+++ b/lib/Target/NVPTX/NVPTXUtilities.h
@@ -14,10 +14,10 @@
 #ifndef NVPTXUTILITIES_H
 #define NVPTXUTILITIES_H
 
-#include "llvm/Value.h"
-#include "llvm/GlobalVariable.h"
 #include "llvm/Function.h"
+#include "llvm/GlobalVariable.h"
 #include "llvm/IntrinsicInst.h"
+#include "llvm/Value.h"
 #include <cstdarg>
 #include <set>
 #include <string>
diff --git a/lib/Target/NVPTX/VectorElementize.cpp b/lib/Target/NVPTX/VectorElementize.cpp
index 8043e2de09..e67e2e42b3 100644
--- a/lib/Target/NVPTX/VectorElementize.cpp
+++ b/lib/Target/NVPTX/VectorElementize.cpp
@@ -34,24 +34,24 @@
 //
 //===----------------------------------------------------------------------===//
 
+#include "NVPTX.h"
+#include "NVPTXTargetMachine.h"
+#include "llvm/ADT/DepthFirstIterator.h"
+#include "llvm/ADT/SmallPtrSet.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
+#include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/CodeGen/MachineModuleInfo.h"
+#include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/Passes.h"
 #include "llvm/Constant.h"
-#include "llvm/Instructions.h"
 #include "llvm/Function.h"
+#include "llvm/Instructions.h"
 #include "llvm/Pass.h"
-#include "llvm/Type.h"
-#include "llvm/Support/CommandLine.h"
-#include "llvm/CodeGen/MachineFunctionPass.h"
-#include "llvm/CodeGen/MachineModuleInfo.h"
-#include "llvm/CodeGen/MachineRegisterInfo.h"
-#include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/Support/CFG.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Target/TargetInstrInfo.h"
-#include "llvm/ADT/DepthFirstIterator.h"
-#include "llvm/ADT/SmallPtrSet.h"
-#include "NVPTX.h"
-#include "NVPTXTargetMachine.h"
+#include "llvm/Type.h"
 
 using namespace llvm;