Vendor import of clang trunk r291012:vendor/clang/clang-trunk-r291012

https://llvm.org/svn/llvm-project/cfe/trunk@291012

31 files changed, 565 insertions, 274 deletions

diff --git a/lib/AST/ExprConstant.cpp b/lib/AST/ExprConstant.cpp
index b3f8925b6464..6dcb705c44d3 100644
--- a/lib/AST/ExprConstant.cpp
+++ b/lib/AST/ExprConstant.cpp
@@ -7192,6 +7192,12 @@ static bool determineEndOffset(EvalInfo &Info, SourceLocation ExprLoc,
                                CharUnits &EndOffset) {
   bool DetermineForCompleteObject = refersToCompleteObject(LVal);
 
+  auto CheckedHandleSizeof = [&](QualType Ty, CharUnits &Result) {
+    if (Ty.isNull() || Ty->isIncompleteType() || Ty->isFunctionType())
+      return false;
+    return HandleSizeof(Info, ExprLoc, Ty, Result);
+  };
+
   // We want to evaluate the size of the entire object. This is a valid fallback
   // for when Type=1 and the designator is invalid, because we're asked for an
   // upper-bound.
@@ -7209,7 +7215,7 @@ static bool determineEndOffset(EvalInfo &Info, SourceLocation ExprLoc,
       return false;
 
     QualType BaseTy = getObjectType(LVal.getLValueBase());
-    return !BaseTy.isNull() && HandleSizeof(Info, ExprLoc, BaseTy, EndOffset);
+    return CheckedHandleSizeof(BaseTy, EndOffset);
   }
 
   // We want to evaluate the size of a subobject.
@@ -7238,7 +7244,7 @@ static bool determineEndOffset(EvalInfo &Info, SourceLocation ExprLoc,
   }
 
   CharUnits BytesPerElem;
-  if (!HandleSizeof(Info, ExprLoc, Designator.MostDerivedType, BytesPerElem))
+  if (!CheckedHandleSizeof(Designator.MostDerivedType, BytesPerElem))
     return false;
 
   // According to the GCC documentation, we want the size of the subobject
diff --git a/lib/AST/StmtOpenMP.cpp b/lib/AST/StmtOpenMP.cpp
index 0a90740162b9..a7c71bb5f45c 100644
--- a/lib/AST/StmtOpenMP.cpp
+++ b/lib/AST/StmtOpenMP.cpp
@@ -1659,3 +1659,64 @@ OMPTargetTeamsDistributeParallelForDirective::CreateEmpty(const ASTContext &C,
   return new (Mem)
       OMPTargetTeamsDistributeParallelForDirective(CollapsedNum, NumClauses);
 }
+
+OMPTargetTeamsDistributeParallelForSimdDirective *
+OMPTargetTeamsDistributeParallelForSimdDirective::Create(
+    const ASTContext &C, SourceLocation StartLoc, SourceLocation EndLoc,
+    unsigned CollapsedNum, ArrayRef<OMPClause *> Clauses, Stmt *AssociatedStmt,
+    const HelperExprs &Exprs) {
+  auto Size =
+      llvm::alignTo(sizeof(OMPTargetTeamsDistributeParallelForSimdDirective),
+                    alignof(OMPClause *));
+  void *Mem = C.Allocate(
+      Size + sizeof(OMPClause *) * Clauses.size() +
+      sizeof(Stmt *) *
+          numLoopChildren(CollapsedNum,
+                          OMPD_target_teams_distribute_parallel_for_simd));
+  OMPTargetTeamsDistributeParallelForSimdDirective *Dir =
+      new (Mem) OMPTargetTeamsDistributeParallelForSimdDirective(
+           StartLoc, EndLoc, CollapsedNum, Clauses.size());
+  Dir->setClauses(Clauses);
+  Dir->setAssociatedStmt(AssociatedStmt);
+  Dir->setIterationVariable(Exprs.IterationVarRef);
+  Dir->setLastIteration(Exprs.LastIteration);
+  Dir->setCalcLastIteration(Exprs.CalcLastIteration);
+  Dir->setPreCond(Exprs.PreCond);
+  Dir->setCond(Exprs.Cond);
+  Dir->setInit(Exprs.Init);
+  Dir->setInc(Exprs.Inc);
+  Dir->setIsLastIterVariable(Exprs.IL);
+  Dir->setLowerBoundVariable(Exprs.LB);
+  Dir->setUpperBoundVariable(Exprs.UB);
+  Dir->setStrideVariable(Exprs.ST);
+  Dir->setEnsureUpperBound(Exprs.EUB);
+  Dir->setNextLowerBound(Exprs.NLB);
+  Dir->setNextUpperBound(Exprs.NUB);
+  Dir->setNumIterations(Exprs.NumIterations);
+  Dir->setPrevLowerBoundVariable(Exprs.PrevLB);
+  Dir->setPrevUpperBoundVariable(Exprs.PrevUB);
+  Dir->setCounters(Exprs.Counters);
+  Dir->setPrivateCounters(Exprs.PrivateCounters);
+  Dir->setInits(Exprs.Inits);
+  Dir->setUpdates(Exprs.Updates);
+  Dir->setFinals(Exprs.Finals);
+  Dir->setPreInits(Exprs.PreInits);
+  return Dir;
+}
+
+OMPTargetTeamsDistributeParallelForSimdDirective *
+OMPTargetTeamsDistributeParallelForSimdDirective::CreateEmpty(
+    const ASTContext &C, unsigned NumClauses, unsigned CollapsedNum,
+    EmptyShell) {
+  auto Size =
+      llvm::alignTo(sizeof(OMPTargetTeamsDistributeParallelForSimdDirective),
+                    alignof(OMPClause *));
+  void *Mem = C.Allocate(
+      Size + sizeof(OMPClause *) * NumClauses +
+      sizeof(Stmt *) *
+          numLoopChildren(CollapsedNum,
+                          OMPD_target_teams_distribute_parallel_for_simd));
+  return new (Mem) OMPTargetTeamsDistributeParallelForSimdDirective(
+      CollapsedNum, NumClauses);
+}
+
diff --git a/lib/AST/StmtPrinter.cpp b/lib/AST/StmtPrinter.cpp
index a8f493dca07d..a9c64c3ba6ae 100644
--- a/lib/AST/StmtPrinter.cpp
+++ b/lib/AST/StmtPrinter.cpp
@@ -1244,6 +1244,12 @@ void StmtPrinter::VisitOMPTargetTeamsDistributeParallelForDirective(
   PrintOMPExecutableDirective(Node);
 }
 
+void StmtPrinter::VisitOMPTargetTeamsDistributeParallelForSimdDirective(
+    OMPTargetTeamsDistributeParallelForSimdDirective *Node) {
+  Indent() << "#pragma omp target teams distribute parallel for simd ";
+  PrintOMPExecutableDirective(Node);
+}
+
 //===----------------------------------------------------------------------===//
 //  Expr printing methods.
 //===----------------------------------------------------------------------===//
diff --git a/lib/AST/StmtProfile.cpp b/lib/AST/StmtProfile.cpp
index dd59a9b96c98..df36bf06b843 100644
--- a/lib/AST/StmtProfile.cpp
+++ b/lib/AST/StmtProfile.cpp
@@ -763,6 +763,11 @@ void StmtProfiler::VisitOMPTargetTeamsDistributeParallelForDirective(
   VisitOMPLoopDirective(S);
 }
 
+void StmtProfiler::VisitOMPTargetTeamsDistributeParallelForSimdDirective(
+    const OMPTargetTeamsDistributeParallelForSimdDirective *S) {
+  VisitOMPLoopDirective(S);
+}
+
 void StmtProfiler::VisitExpr(const Expr *S) {
   VisitStmt(S);
 }
diff --git a/lib/Analysis/CFG.cpp b/lib/Analysis/CFG.cpp
index bf3cc05cdb6e..a1a463f1d037 100644
--- a/lib/Analysis/CFG.cpp
+++ b/lib/Analysis/CFG.cpp
@@ -1690,15 +1690,19 @@ CFGBuilder::VisitLogicalOperator(BinaryOperator *B,
     // we have been provided.
     ExitBlock = RHSBlock = createBlock(false);
 
+    // Even though KnownVal is only used in the else branch of the next
+    // conditional, tryEvaluateBool performs additional checking on the
+    // Expr, so it should be called unconditionally.
+    TryResult KnownVal = tryEvaluateBool(RHS);
+    if (!KnownVal.isKnown())
+      KnownVal = tryEvaluateBool(B);
+
     if (!Term) {
       assert(TrueBlock == FalseBlock);
       addSuccessor(RHSBlock, TrueBlock);
     }
     else {
       RHSBlock->setTerminator(Term);
-      TryResult KnownVal = tryEvaluateBool(RHS);
-      if (!KnownVal.isKnown())
-        KnownVal = tryEvaluateBool(B);
       addSuccessor(RHSBlock, TrueBlock, !KnownVal.isFalse());
       addSuccessor(RHSBlock, FalseBlock, !KnownVal.isTrue());
     }
diff --git a/lib/Basic/OpenMPKinds.cpp b/lib/Basic/OpenMPKinds.cpp
index 4675995ea722..7bd1f8762bff 100644
--- a/lib/Basic/OpenMPKinds.cpp
+++ b/lib/Basic/OpenMPKinds.cpp
@@ -690,6 +690,16 @@ bool clang::isAllowedClauseForDirective(OpenMPDirectiveKind DKind,
       break;
     }
     break;
+  case OMPD_target_teams_distribute_parallel_for_simd:
+    switch (CKind) {
+#define OPENMP_TARGET_TEAMS_DISTRIBUTE_PARALLEL_FOR_SIMD_CLAUSE(Name)          \
+  case OMPC_##Name:                                                            \
+    return true;
+#include "clang/Basic/OpenMPKinds.def"
+    default:
+      break;
+    }
+    break;
   case OMPD_declare_target:
   case OMPD_end_declare_target:
   case OMPD_unknown:
@@ -721,7 +731,8 @@ bool clang::isOpenMPLoopDirective(OpenMPDirectiveKind DKind) {
          DKind == OMPD_teams_distribute_parallel_for_simd ||
          DKind == OMPD_teams_distribute_parallel_for ||
          DKind == OMPD_target_teams_distribute ||
-         DKind == OMPD_target_teams_distribute_parallel_for;
+         DKind == OMPD_target_teams_distribute_parallel_for ||
+         DKind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPWorksharingDirective(OpenMPDirectiveKind DKind) {
@@ -735,8 +746,8 @@ bool clang::isOpenMPWorksharingDirective(OpenMPDirectiveKind DKind) {
          DKind == OMPD_target_parallel_for_simd ||
          DKind == OMPD_teams_distribute_parallel_for_simd ||
          DKind == OMPD_teams_distribute_parallel_for ||
-         DKind == OMPD_target_teams_distribute_parallel_for;
-  // TODO add next directives.
+         DKind == OMPD_target_teams_distribute_parallel_for ||
+         DKind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPTaskLoopDirective(OpenMPDirectiveKind DKind) {
@@ -752,8 +763,8 @@ bool clang::isOpenMPParallelDirective(OpenMPDirectiveKind DKind) {
          DKind == OMPD_target_parallel_for_simd ||
          DKind == OMPD_teams_distribute_parallel_for ||
          DKind == OMPD_teams_distribute_parallel_for_simd ||
-         DKind == OMPD_target_teams_distribute_parallel_for;
-  // TODO add next directives.
+         DKind == OMPD_target_teams_distribute_parallel_for ||
+         DKind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPTargetExecutionDirective(OpenMPDirectiveKind DKind) {
@@ -761,7 +772,8 @@ bool clang::isOpenMPTargetExecutionDirective(OpenMPDirectiveKind DKind) {
          DKind == OMPD_target_parallel_for || 
          DKind == OMPD_target_parallel_for_simd || DKind == OMPD_target_simd ||
          DKind == OMPD_target_teams || DKind == OMPD_target_teams_distribute ||
-         DKind == OMPD_target_teams_distribute_parallel_for;
+         DKind == OMPD_target_teams_distribute_parallel_for ||
+         DKind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPTargetDataManagementDirective(OpenMPDirectiveKind DKind) {
@@ -779,7 +791,8 @@ bool clang::isOpenMPNestingTeamsDirective(OpenMPDirectiveKind DKind) {
 bool clang::isOpenMPTeamsDirective(OpenMPDirectiveKind DKind) {
   return isOpenMPNestingTeamsDirective(DKind) ||
          DKind == OMPD_target_teams || DKind == OMPD_target_teams_distribute ||
-         DKind == OMPD_target_teams_distribute_parallel_for;
+         DKind == OMPD_target_teams_distribute_parallel_for ||
+         DKind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPSimdDirective(OpenMPDirectiveKind DKind) {
@@ -788,8 +801,8 @@ bool clang::isOpenMPSimdDirective(OpenMPDirectiveKind DKind) {
          DKind == OMPD_distribute_parallel_for_simd ||
          DKind == OMPD_distribute_simd || DKind == OMPD_target_simd ||
          DKind == OMPD_teams_distribute_simd ||
-         DKind == OMPD_teams_distribute_parallel_for_simd;
-  // TODO add next directives.
+         DKind == OMPD_teams_distribute_parallel_for_simd ||
+         DKind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPNestingDistributeDirective(OpenMPDirectiveKind Kind) {
@@ -805,7 +818,8 @@ bool clang::isOpenMPDistributeDirective(OpenMPDirectiveKind Kind) {
          Kind == OMPD_teams_distribute_parallel_for_simd ||
          Kind == OMPD_teams_distribute_parallel_for ||
          Kind == OMPD_target_teams_distribute ||
-         Kind == OMPD_target_teams_distribute_parallel_for;
+         Kind == OMPD_target_teams_distribute_parallel_for ||
+         Kind == OMPD_target_teams_distribute_parallel_for_simd;
 }
 
 bool clang::isOpenMPPrivate(OpenMPClauseKind Kind) {
@@ -830,5 +844,6 @@ bool clang::isOpenMPLoopBoundSharingDirective(OpenMPDirectiveKind Kind) {
          Kind == OMPD_teams_distribute_parallel_for_simd ||
          Kind == OMPD_teams_distribute_parallel_for ||
          Kind == OMPD_target_teams_distribute ||
-         Kind == OMPD_target_teams_distribute_parallel_for;
+         Kind == OMPD_target_teams_distribute_parallel_for ||
+         Kind == OMPD_target_teams_distribute_parallel_for_simd;
 }
diff --git a/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp b/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
index 451f9e9221ad..fe0e2acdfdbf 100644
--- a/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ b/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -20,53 +20,64 @@
 using namespace clang;
 using namespace CodeGen;
 
-/// \brief Get the GPU warp size.
-llvm::Value *CGOpenMPRuntimeNVPTX::getNVPTXWarpSize(CodeGenFunction &CGF) {
+namespace {
+enum OpenMPRTLFunctionNVPTX {
+  /// \brief Call to void __kmpc_kernel_init(kmp_int32 omp_handle,
+  /// kmp_int32 thread_limit);
+  OMPRTL_NVPTX__kmpc_kernel_init,
+};
+
+// NVPTX Address space
+enum AddressSpace {
+  AddressSpaceShared = 3,
+};
+} // namespace
+
+/// Get the GPU warp size.
+static llvm::Value *getNVPTXWarpSize(CodeGenFunction &CGF) {
   CGBuilderTy &Bld = CGF.Builder;
   return Bld.CreateCall(
       llvm::Intrinsic::getDeclaration(
-          &CGM.getModule(), llvm::Intrinsic::nvvm_read_ptx_sreg_warpsize),
+          &CGF.CGM.getModule(), llvm::Intrinsic::nvvm_read_ptx_sreg_warpsize),
       llvm::None, "nvptx_warp_size");
 }
 
-/// \brief Get the id of the current thread on the GPU.
-llvm::Value *CGOpenMPRuntimeNVPTX::getNVPTXThreadID(CodeGenFunction &CGF) {
+/// Get the id of the current thread on the GPU.
+static llvm::Value *getNVPTXThreadID(CodeGenFunction &CGF) {
   CGBuilderTy &Bld = CGF.Builder;
   return Bld.CreateCall(
       llvm::Intrinsic::getDeclaration(
-          &CGM.getModule(), llvm::Intrinsic::nvvm_read_ptx_sreg_tid_x),
+          &CGF.CGM.getModule(), llvm::Intrinsic::nvvm_read_ptx_sreg_tid_x),
       llvm::None, "nvptx_tid");
 }
 
-// \brief Get the maximum number of threads in a block of the GPU.
-llvm::Value *CGOpenMPRuntimeNVPTX::getNVPTXNumThreads(CodeGenFunction &CGF) {
+/// Get the maximum number of threads in a block of the GPU.
+static llvm::Value *getNVPTXNumThreads(CodeGenFunction &CGF) {
   CGBuilderTy &Bld = CGF.Builder;
   return Bld.CreateCall(
       llvm::Intrinsic::getDeclaration(
-          &CGM.getModule(), llvm::Intrinsic::nvvm_read_ptx_sreg_ntid_x),
+          &CGF.CGM.getModule(), llvm::Intrinsic::nvvm_read_ptx_sreg_ntid_x),
       llvm::None, "nvptx_num_threads");
 }
 
-/// \brief Get barrier to synchronize all threads in a block.
-void CGOpenMPRuntimeNVPTX::getNVPTXCTABarrier(CodeGenFunction &CGF) {
+/// Get barrier to synchronize all threads in a block.
+static void getNVPTXCTABarrier(CodeGenFunction &CGF) {
   CGBuilderTy &Bld = CGF.Builder;
   Bld.CreateCall(llvm::Intrinsic::getDeclaration(
-      &CGM.getModule(), llvm::Intrinsic::nvvm_barrier0));
+      &CGF.CGM.getModule(), llvm::Intrinsic::nvvm_barrier0));
 }
 
-// \brief Synchronize all GPU threads in a block.
-void CGOpenMPRuntimeNVPTX::syncCTAThreads(CodeGenFunction &CGF) {
-  getNVPTXCTABarrier(CGF);
-}
+/// Synchronize all GPU threads in a block.
+static void syncCTAThreads(CodeGenFunction &CGF) { getNVPTXCTABarrier(CGF); }
 
-/// \brief Get the thread id of the OMP master thread.
+/// Get the thread id of the OMP master thread.
 /// The master thread id is the first thread (lane) of the last warp in the
 /// GPU block.  Warp size is assumed to be some power of 2.
 /// Thread id is 0 indexed.
 /// E.g: If NumThreads is 33, master id is 32.
 ///      If NumThreads is 64, master id is 32.
 ///      If NumThreads is 1024, master id is 992.
-llvm::Value *CGOpenMPRuntimeNVPTX::getMasterThreadID(CodeGenFunction &CGF) {
+static llvm::Value *getMasterThreadID(CodeGenFunction &CGF) {
   CGBuilderTy &Bld = CGF.Builder;
   llvm::Value *NumThreads = getNVPTXNumThreads(CGF);
 
@@ -77,19 +88,6 @@ llvm::Value *CGOpenMPRuntimeNVPTX::getMasterThreadID(CodeGenFunction &CGF) {
                        Bld.CreateNot(Mask), "master_tid");
 }
 
-namespace {
-enum OpenMPRTLFunctionNVPTX {
-  /// \brief Call to void __kmpc_kernel_init(kmp_int32 omp_handle,
-  /// kmp_int32 thread_limit);
-  OMPRTL_NVPTX__kmpc_kernel_init,
-};
-
-// NVPTX Address space
-enum ADDRESS_SPACE {
-  ADDRESS_SPACE_SHARED = 3,
-};
-} // namespace
-
 CGOpenMPRuntimeNVPTX::WorkerFunctionState::WorkerFunctionState(
     CodeGenModule &CGM)
     : WorkerFn(nullptr), CGFI(nullptr) {
@@ -119,14 +117,14 @@ void CGOpenMPRuntimeNVPTX::initializeEnvironment() {
       CGM.getModule(), CGM.Int32Ty, /*isConstant=*/false,
       llvm::GlobalValue::CommonLinkage,
       llvm::Constant::getNullValue(CGM.Int32Ty), "__omp_num_threads", 0,
-      llvm::GlobalVariable::NotThreadLocal, ADDRESS_SPACE_SHARED);
+      llvm::GlobalVariable::NotThreadLocal, AddressSpaceShared);
   ActiveWorkers->setAlignment(DL.getPrefTypeAlignment(CGM.Int32Ty));
 
   WorkID = new llvm::GlobalVariable(
       CGM.getModule(), CGM.Int64Ty, /*isConstant=*/false,
       llvm::GlobalValue::CommonLinkage,
       llvm::Constant::getNullValue(CGM.Int64Ty), "__tgt_work_id", 0,
-      llvm::GlobalVariable::NotThreadLocal, ADDRESS_SPACE_SHARED);
+      llvm::GlobalVariable::NotThreadLocal, AddressSpaceShared);
   WorkID->setAlignment(DL.getPrefTypeAlignment(CGM.Int64Ty));
 }
 
diff --git a/lib/CodeGen/CGOpenMPRuntimeNVPTX.h b/lib/CodeGen/CGOpenMPRuntimeNVPTX.h
index e18d28cdda9f..a33fb27579f6 100644
--- a/lib/CodeGen/CGOpenMPRuntimeNVPTX.h
+++ b/lib/CodeGen/CGOpenMPRuntimeNVPTX.h
@@ -50,38 +50,6 @@ public:
 
 private:
   //
-  // NVPTX calls.
-  //
-
-  /// \brief Get the GPU warp size.
-  llvm::Value *getNVPTXWarpSize(CodeGenFunction &CGF);
-
-  /// \brief Get the id of the current thread on the GPU.
-  llvm::Value *getNVPTXThreadID(CodeGenFunction &CGF);
-
-  // \brief Get the maximum number of threads in a block of the GPU.
-  llvm::Value *getNVPTXNumThreads(CodeGenFunction &CGF);
-
-  /// \brief Get barrier to synchronize all threads in a block.
-  void getNVPTXCTABarrier(CodeGenFunction &CGF);
-
-  // \brief Synchronize all GPU threads in a block.
-  void syncCTAThreads(CodeGenFunction &CGF);
-
-  //
-  // OMP calls.
-  //
-
-  /// \brief Get the thread id of the OMP master thread.
-  /// The master thread id is the first thread (lane) of the last warp in the
-  /// GPU block.  Warp size is assumed to be some power of 2.
-  /// Thread id is 0 indexed.
-  /// E.g: If NumThreads is 33, master id is 32.
-  ///      If NumThreads is 64, master id is 32.
-  ///      If NumThreads is 1024, master id is 992.
-  llvm::Value *getMasterThreadID(CodeGenFunction &CGF);
-
-  //
   // Private state and methods.
   //
 
diff --git a/lib/CodeGen/CGStmt.cpp b/lib/CodeGen/CGStmt.cpp
index f2acb798b881..8d391f95d9f7 100644
--- a/lib/CodeGen/CGStmt.cpp
+++ b/lib/CodeGen/CGStmt.cpp
@@ -326,6 +326,10 @@ void CodeGenFunction::EmitStmt(const Stmt *S) {
     EmitOMPTargetTeamsDistributeParallelForDirective(
         cast<OMPTargetTeamsDistributeParallelForDirective>(*S));
     break;
+  case Stmt::OMPTargetTeamsDistributeParallelForSimdDirectiveClass:
+    EmitOMPTargetTeamsDistributeParallelForSimdDirective(
+        cast<OMPTargetTeamsDistributeParallelForSimdDirective>(*S));
+    break;
   }
 }
 
diff --git a/lib/CodeGen/CGStmtOpenMP.cpp b/lib/CodeGen/CGStmtOpenMP.cpp
index ba39e1fbd41f..386c4f0fe69c 100644
--- a/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/lib/CodeGen/CGStmtOpenMP.cpp
@@ -2032,6 +2032,16 @@ void CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForDirective(
       });
 }
 
+void CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForSimdDirective(
+    const OMPTargetTeamsDistributeParallelForSimdDirective &S) {
+  CGM.getOpenMPRuntime().emitInlinedDirective(
+      *this, OMPD_target_teams_distribute_parallel_for_simd,
+      [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+        CGF.EmitStmt(
+            cast<CapturedStmt>(S.getAssociatedStmt())->getCapturedStmt());
+      });
+}
+
 /// \brief Emit a helper variable and return corresponding lvalue.
 static LValue EmitOMPHelperVar(CodeGenFunction &CGF,
                                const DeclRefExpr *Helper) {
@@ -2760,6 +2770,7 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPDistributeDirective &S) {
 
   auto &RT = CGM.getOpenMPRuntime();
 
+  bool HasLastprivateClause = false;
   // Check pre-condition.
   {
     OMPLoopScope PreInitScope(*this, S);
@@ -2793,6 +2804,16 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPDistributeDirective &S) {
           EmitOMPHelperVar(*this, cast<DeclRefExpr>(S.getIsLastIterVariable()));
 
       OMPPrivateScope LoopScope(*this);
+      if (EmitOMPFirstprivateClause(S, LoopScope)) {
+        // Emit implicit barrier to synchronize threads and avoid data races on
+        // initialization of firstprivate variables and post-update of
+        // lastprivate variables.
+        CGM.getOpenMPRuntime().emitBarrierCall(
+          *this, S.getLocStart(), OMPD_unknown, /*EmitChecks=*/false,
+          /*ForceSimpleCall=*/true);
+      }
+      EmitOMPPrivateClause(S, LoopScope);
+      HasLastprivateClause = EmitOMPLastprivateClauseInit(S, LoopScope);
       EmitOMPPrivateLoopCounters(S, LoopScope);
       (void)LoopScope.Privatize();
 
@@ -2849,6 +2870,13 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPDistributeDirective &S) {
                             LB.getAddress(), UB.getAddress(), ST.getAddress(),
                             IL.getAddress(), Chunk);
       }
+
+      // Emit final copy of the lastprivate variables if IsLastIter != 0.
+      if (HasLastprivateClause)
+        EmitOMPLastprivateClauseFinal(
+            S, /*NoFinals=*/false,
+            Builder.CreateIsNotNull(
+                EmitLoadOfScalar(IL, S.getLocStart())));
     }
 
     // We're now done with the loop, so jump to the continuation block.
diff --git a/lib/CodeGen/CodeGenFunction.cpp b/lib/CodeGen/CodeGenFunction.cpp
index a954f487d1e4..7cab13de923b 100644
--- a/lib/CodeGen/CodeGenFunction.cpp
+++ b/lib/CodeGen/CodeGenFunction.cpp
@@ -1049,6 +1049,19 @@ QualType CodeGenFunction::BuildFunctionArgList(GlobalDecl GD,
   return ResTy;
 }
 
+static bool
+shouldUseUndefinedBehaviorReturnOptimization(const FunctionDecl *FD,
+                                             const ASTContext &Context) {
+  QualType T = FD->getReturnType();
+  // Avoid the optimization for functions that return a record type with a
+  // trivial destructor or another trivially copyable type.
+  if (const RecordType *RT = T.getCanonicalType()->getAs<RecordType>()) {
+    if (const auto *ClassDecl = dyn_cast<CXXRecordDecl>(RT->getDecl()))
+      return !ClassDecl->hasTrivialDestructor();
+  }
+  return !T.isTriviallyCopyableType(Context);
+}
+
 void CodeGenFunction::GenerateCode(GlobalDecl GD, llvm::Function *Fn,
                                    const CGFunctionInfo &FnInfo) {
   const FunctionDecl *FD = cast<FunctionDecl>(GD.getDecl());
@@ -1127,17 +1140,23 @@ void CodeGenFunction::GenerateCode(GlobalDecl GD, llvm::Function *Fn,
   //   function call is used by the caller, the behavior is undefined.
   if (getLangOpts().CPlusPlus && !FD->hasImplicitReturnZero() && !SawAsmBlock &&
       !FD->getReturnType()->isVoidType() && Builder.GetInsertBlock()) {
+    bool ShouldEmitUnreachable =
+        CGM.getCodeGenOpts().StrictReturn ||
+        shouldUseUndefinedBehaviorReturnOptimization(FD, getContext());
     if (SanOpts.has(SanitizerKind::Return)) {
       SanitizerScope SanScope(this);
       llvm::Value *IsFalse = Builder.getFalse();
       EmitCheck(std::make_pair(IsFalse, SanitizerKind::Return),
                 SanitizerHandler::MissingReturn,
                 EmitCheckSourceLocation(FD->getLocation()), None);
-    } else if (CGM.getCodeGenOpts().OptimizationLevel == 0) {
-      EmitTrapCall(llvm::Intrinsic::trap);
+    } else if (ShouldEmitUnreachable) {
+      if (CGM.getCodeGenOpts().OptimizationLevel == 0)
+        EmitTrapCall(llvm::Intrinsic::trap);
+    }
+    if (SanOpts.has(SanitizerKind::Return) || ShouldEmitUnreachable) {
+      Builder.CreateUnreachable();
+      Builder.ClearInsertionPoint();
     }
-    Builder.CreateUnreachable();
-    Builder.ClearInsertionPoint();
   }
 
   // Emit the standard function epilogue.
diff --git a/lib/CodeGen/CodeGenFunction.h b/lib/CodeGen/CodeGenFunction.h
index 222d0e97968a..1347f54df9ac 100644
--- a/lib/CodeGen/CodeGenFunction.h
+++ b/lib/CodeGen/CodeGenFunction.h
@@ -2699,6 +2699,8 @@ public:
       const OMPTargetTeamsDistributeDirective &S);
   void EmitOMPTargetTeamsDistributeParallelForDirective(
       const OMPTargetTeamsDistributeParallelForDirective &S);
+  void EmitOMPTargetTeamsDistributeParallelForSimdDirective(
+      const OMPTargetTeamsDistributeParallelForSimdDirective &S);
 
   /// Emit outlined function for the target directive.
   static std::pair<llvm::Function * /*OutlinedFn*/,
@@ -3569,7 +3571,7 @@ public:
 
     // If we still have any arguments, emit them using the type of the argument.
     for (auto *A : llvm::make_range(Arg, ArgRange.end()))
-      ArgTypes.push_back(getVarArgType(A));
+      ArgTypes.push_back(CallArgTypeInfo ? getVarArgType(A) : A->getType());
 
     EmitCallArgs(Args, ArgTypes, ArgRange, CalleeDecl, ParamsToSkip, Order);
   }
diff --git a/lib/Driver/Tools.cpp b/lib/Driver/Tools.cpp
index 2a367bb29aa5..ea5ad7d051b6 100644
--- a/lib/Driver/Tools.cpp
+++ b/lib/Driver/Tools.cpp
@@ -2235,6 +2235,15 @@ static void AddGoldPlugin(const ToolChain &ToolChain, const ArgList &Args,
                    UseSeparateSections)) {
     CmdArgs.push_back("-plugin-opt=-data-sections");
   }
+
+  if (Arg *A = Args.getLastArg(options::OPT_fprofile_sample_use_EQ)) {
+    StringRef FName = A->getValue();
+    if (!llvm::sys::fs::exists(FName))
+      D.Diag(diag::err_drv_no_such_file) << FName;
+    else
+      CmdArgs.push_back(
+          Args.MakeArgString(Twine("-plugin-opt=sample-profile=") + FName));
+  }
 }
 
 /// This is a helper function for validating the optional refinement step
@@ -3058,6 +3067,10 @@ static void CollectArgsForIntegratedAssembler(Compilation &C,
         continue;
       }
 
+      if (C.getDefaultToolChain().getTriple().isOSBinFormatCOFF() &&
+          Value == "-mbig-obj")
+        continue; // LLVM handles bigobj automatically
+
       switch (C.getDefaultToolChain().getArch()) {
       default:
         break;
@@ -4453,6 +4466,9 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
   if (Args.hasFlag(options::OPT_fstrict_enums, options::OPT_fno_strict_enums,
                    false))
     CmdArgs.push_back("-fstrict-enums");
+  if (!Args.hasFlag(options::OPT_fstrict_return, options::OPT_fno_strict_return,
+                    true))
+    CmdArgs.push_back("-fno-strict-return");
   if (Args.hasFlag(options::OPT_fstrict_vtable_pointers,
                    options::OPT_fno_strict_vtable_pointers,
                    false))
diff --git a/lib/Format/Format.cpp b/lib/Format/Format.cpp
index 70b90d6fa14e..389761d48249 100644
--- a/lib/Format/Format.cpp
+++ b/lib/Format/Format.cpp
@@ -638,6 +638,9 @@ FormatStyle getChromiumStyle(FormatStyle::LanguageKind Language) {
     ChromiumStyle.BreakAfterJavaFieldAnnotations = true;
     ChromiumStyle.ContinuationIndentWidth = 8;
     ChromiumStyle.IndentWidth = 4;
+  } else if (Language == FormatStyle::LK_JavaScript) {
+    ChromiumStyle.AllowShortIfStatementsOnASingleLine = false;
+    ChromiumStyle.AllowShortLoopsOnASingleLine = false;
   } else {
     ChromiumStyle.AllowAllParametersOfDeclarationOnNextLine = false;
     ChromiumStyle.AllowShortFunctionsOnASingleLine = FormatStyle::SFS_Inline;
diff --git a/lib/Format/UnwrappedLineParser.cpp b/lib/Format/UnwrappedLineParser.cpp
index 84e06d05c739..370cf7afa330 100644
--- a/lib/Format/UnwrappedLineParser.cpp
+++ b/lib/Format/UnwrappedLineParser.cpp
@@ -1255,10 +1255,13 @@ void UnwrappedLineParser::tryToParseJSFunction() {
     if (FormatTok->is(tok::l_brace))
       tryToParseBracedList();
     else
-      while (FormatTok->isNot(tok::l_brace) && !eof())
+      while (!FormatTok->isOneOf(tok::l_brace, tok::semi) && !eof())
         nextToken();
   }
 
+  if (FormatTok->is(tok::semi))
+    return;
+
   parseChildBlock();
 }
 
diff --git a/lib/Frontend/ASTConsumers.cpp b/lib/Frontend/ASTConsumers.cpp
index bd2ee06d1653..d8118cb30f63 100644
--- a/lib/Frontend/ASTConsumers.cpp
+++ b/lib/Frontend/ASTConsumers.cpp
@@ -370,6 +370,26 @@ void DeclContextPrinter::PrintDeclContext(const DeclContext* DC,
     break;
   }
 
+  case Decl::ClassTemplateSpecialization: {
+    const auto *CTSD = cast<ClassTemplateSpecializationDecl>(DC);
+    if (CTSD->isCompleteDefinition())
+      Out << "[class template specialization] ";
+    else
+      Out << "<class template specialization> ";
+    Out << *CTSD;
+    break;
+  }
+
+  case Decl::ClassTemplatePartialSpecialization: {
+    const auto *CTPSD = cast<ClassTemplatePartialSpecializationDecl>(DC);
+    if (CTPSD->isCompleteDefinition())
+      Out << "[class template partial specialization] ";
+    else
+      Out << "<class template partial specialization> ";
+    Out << *CTPSD;
+    break;
+  }
+
   default:
     llvm_unreachable("a decl that inherits DeclContext isn't handled");
   }
@@ -400,7 +420,8 @@ void DeclContextPrinter::PrintDeclContext(const DeclContext* DC,
     case Decl::CXXConstructor:
     case Decl::CXXDestructor:
     case Decl::CXXConversion:
-    {
+    case Decl::ClassTemplateSpecialization:
+    case Decl::ClassTemplatePartialSpecialization: {
       DeclContext* DC = cast<DeclContext>(I);
       PrintDeclContext(DC, Indentation+2);
       break;
@@ -478,6 +499,37 @@ void DeclContextPrinter::PrintDeclContext(const DeclContext* DC,
       Out << "<omp threadprivate> " << '"' << I << "\"\n";
       break;
     }
+    case Decl::Friend: {
+      Out << "<friend>";
+      if (const NamedDecl *ND = cast<FriendDecl>(I)->getFriendDecl())
+        Out << ' ' << *ND;
+      Out << "\n";
+      break;
+    }
+    case Decl::Using: {
+      Out << "<using> " << *cast<UsingDecl>(I) << "\n";
+      break;
+    }
+    case Decl::UsingShadow: {
+      Out << "<using shadow> " << *cast<UsingShadowDecl>(I) << "\n";
+      break;
+    }
+    case Decl::Empty: {
+      Out << "<empty>\n";
+      break;
+    }
+    case Decl::AccessSpec: {
+      Out << "<access specifier>\n";
+      break;
+    }
+    case Decl::VarTemplate: {
+      Out << "<var template> " << *cast<VarTemplateDecl>(I) << "\n";
+      break;
+    }
+    case Decl::StaticAssert: {
+      Out << "<static assert>\n";
+      break;
+    }
     default:
       Out << "DeclKind: " << DK << '"' << I << "\"\n";
       llvm_unreachable("decl unhandled");
diff --git a/lib/Frontend/CompilerInvocation.cpp b/lib/Frontend/CompilerInvocation.cpp
index a0682e26e702..ca4a7655a37d 100644
--- a/lib/Frontend/CompilerInvocation.cpp
+++ b/lib/Frontend/CompilerInvocation.cpp
@@ -602,6 +602,7 @@ static bool ParseCodeGenArgs(CodeGenOptions &Opts, ArgList &Args, InputKind IK,
   Opts.NoDwarfDirectoryAsm = Args.hasArg(OPT_fno_dwarf_directory_asm);
   Opts.SoftFloat = Args.hasArg(OPT_msoft_float);
   Opts.StrictEnums = Args.hasArg(OPT_fstrict_enums);
+  Opts.StrictReturn = !Args.hasArg(OPT_fno_strict_return);
   Opts.StrictVTablePointers = Args.hasArg(OPT_fstrict_vtable_pointers);
   Opts.UnsafeFPMath = Args.hasArg(OPT_menable_unsafe_fp_math) ||
                       Args.hasArg(OPT_cl_unsafe_math_optimizations) ||
diff --git a/lib/Headers/__clang_cuda_runtime_wrapper.h b/lib/Headers/__clang_cuda_runtime_wrapper.h
index 205e15b40b5d..931d44b6965b 100644
--- a/lib/Headers/__clang_cuda_runtime_wrapper.h
+++ b/lib/Headers/__clang_cuda_runtime_wrapper.h
@@ -232,6 +232,11 @@ static inline __device__ void __brkpt(int __c) { __brkpt(); }
 // hardware, seems to generate faster machine code because ptxas can more easily
 // reason about our code.
 
+#if CUDA_VERSION >= 8000
+#include "sm_60_atomic_functions.hpp"
+#include "sm_61_intrinsics.hpp"
+#endif
+
 #undef __MATH_FUNCTIONS_HPP__
 
 // math_functions.hpp defines ::signbit as a __host__ __device__ function.  This
diff --git a/lib/Parse/ParseOpenMP.cpp b/lib/Parse/ParseOpenMP.cpp
index 061721dfb8da..f9ea8af00f50 100644
--- a/lib/Parse/ParseOpenMP.cpp
+++ b/lib/Parse/ParseOpenMP.cpp
@@ -119,7 +119,8 @@ static OpenMPDirectiveKind ParseOpenMPDirectiveKind(Parser &P) {
     { OMPD_target, OMPD_teams, OMPD_target_teams },
     { OMPD_target_teams, OMPD_distribute, OMPD_target_teams_distribute },
     { OMPD_target_teams_distribute, OMPD_parallel, OMPD_target_teams_distribute_parallel },
-    { OMPD_target_teams_distribute_parallel, OMPD_for, OMPD_target_teams_distribute_parallel_for }
+    { OMPD_target_teams_distribute_parallel, OMPD_for, OMPD_target_teams_distribute_parallel_for },
+    { OMPD_target_teams_distribute_parallel_for, OMPD_simd, OMPD_target_teams_distribute_parallel_for_simd }
   };
   enum { CancellationPoint = 0, DeclareReduction = 1, TargetData = 2 };
   auto Tok = P.getCurToken();
@@ -758,6 +759,7 @@ Parser::DeclGroupPtrTy Parser::ParseOpenMPDeclarativeDirectiveWithExtDecl(
   case OMPD_target_teams:
   case OMPD_target_teams_distribute:
   case OMPD_target_teams_distribute_parallel_for:
+  case OMPD_target_teams_distribute_parallel_for_simd:
     Diag(Tok, diag::err_omp_unexpected_directive)
         << getOpenMPDirectiveName(DKind);
     break;
@@ -796,7 +798,8 @@ Parser::DeclGroupPtrTy Parser::ParseOpenMPDeclarativeDirectiveWithExtDecl(
 ///         'teams distribute parallel for simd' |
 ///         'teams distribute parallel for' | 'target teams' |
 ///         'target teams distribute' |
-///         'target teams distribute parallel for' {clause}
+///         'target teams distribute parallel for' |
+///         'target teams distribute parallel for simd' {clause}
 ///         annot_pragma_openmp_end
 ///
 StmtResult Parser::ParseOpenMPDeclarativeOrExecutableDirective(
@@ -912,7 +915,8 @@ StmtResult Parser::ParseOpenMPDeclarativeOrExecutableDirective(
   case OMPD_teams_distribute_parallel_for:
   case OMPD_target_teams:
   case OMPD_target_teams_distribute:
-  case OMPD_target_teams_distribute_parallel_for: {
+  case OMPD_target_teams_distribute_parallel_for:
+  case OMPD_target_teams_distribute_parallel_for_simd: {
     ConsumeToken();
     // Parse directive name of the 'critical' directive if any.
     if (DKind == OMPD_critical) {
diff --git a/lib/Sema/SemaCUDA.cpp b/lib/Sema/SemaCUDA.cpp
index 6f272ec839f5..282633bbc9e1 100644
--- a/lib/Sema/SemaCUDA.cpp
+++ b/lib/Sema/SemaCUDA.cpp
@@ -228,10 +228,8 @@ void Sema::EraseUnwantedCUDAMatches(
       [&](const Pair &M1, const Pair &M2) { return GetCFP(M1) < GetCFP(M2); }));
 
   // Erase all functions with lower priority.
-  Matches.erase(
-      llvm::remove_if(
-          Matches, [&](const Pair &Match) { return GetCFP(Match) < BestCFP; }),
-      Matches.end());
+  llvm::erase_if(Matches,
+                 [&](const Pair &Match) { return GetCFP(Match) < BestCFP; });
 }
 
 /// When an implicitly-declared special member has to invoke more than one
diff --git a/lib/Sema/SemaCodeComplete.cpp b/lib/Sema/SemaCodeComplete.cpp
index d76bde574677..3eef366b75b3 100644
--- a/lib/Sema/SemaCodeComplete.cpp
+++ b/lib/Sema/SemaCodeComplete.cpp
@@ -7471,6 +7471,23 @@ void Sema::CodeCompleteObjCMethodDeclSelector(Scope *S,
   }
   
   Results.ExitScope();
+
+  if (!AtParameterName && !SelIdents.empty() &&
+      SelIdents.front()->getName().startswith("init")) {
+    for (const auto &M : PP.macros()) {
+      if (M.first->getName() != "NS_DESIGNATED_INITIALIZER")
+        continue;
+      Results.EnterNewScope();
+      CodeCompletionBuilder Builder(Results.getAllocator(),
+                                    Results.getCodeCompletionTUInfo());
+      Builder.AddTypedTextChunk(
+          Builder.getAllocator().CopyString(M.first->getName()));
+      Results.AddResult(CodeCompletionResult(Builder.TakeString(), CCP_Macro,
+                                             CXCursor_MacroDefinition));
+      Results.ExitScope();
+    }
+  }
+
   HandleCodeCompleteResults(this, CodeCompleter, 
                             CodeCompletionContext::CCC_Other,
                             Results.data(),Results.size());
diff --git a/lib/Sema/SemaOpenMP.cpp b/lib/Sema/SemaOpenMP.cpp
index 804aadc0ff77..edceb537df75 100644
--- a/lib/Sema/SemaOpenMP.cpp
+++ b/lib/Sema/SemaOpenMP.cpp
@@ -1700,7 +1700,8 @@ void Sema::ActOnOpenMPRegionStart(OpenMPDirectiveKind DKind, Scope *CurScope) {
   case OMPD_teams_distribute_parallel_for_simd:
   case OMPD_teams_distribute_parallel_for:
   case OMPD_target_teams_distribute:
-  case OMPD_target_teams_distribute_parallel_for: {
+  case OMPD_target_teams_distribute_parallel_for:
+  case OMPD_target_teams_distribute_parallel_for_simd: {
     QualType KmpInt32Ty = Context.getIntTypeForBitwidth(32, 1);
     QualType KmpInt32PtrTy =
         Context.getPointerType(KmpInt32Ty).withConst().withRestrict();
@@ -2439,6 +2440,12 @@ StmtResult Sema::ActOnOpenMPExecutableDirective(
     AllowedNameModifiers.push_back(OMPD_target);
     AllowedNameModifiers.push_back(OMPD_parallel);
     break;
+  case OMPD_target_teams_distribute_parallel_for_simd:
+    Res = ActOnOpenMPTargetTeamsDistributeParallelForSimdDirective(
+        ClausesWithImplicit, AStmt, StartLoc, EndLoc, VarsWithInheritedDSA);
+    AllowedNameModifiers.push_back(OMPD_target);
+    AllowedNameModifiers.push_back(OMPD_parallel);
+    break;
   case OMPD_declare_target:
   case OMPD_end_declare_target:
   case OMPD_threadprivate:
@@ -6375,6 +6382,52 @@ StmtResult Sema::ActOnOpenMPTargetTeamsDistributeParallelForDirective(
       Context, StartLoc, EndLoc, NestedLoopCount, Clauses, AStmt, B);
 }
 
+StmtResult Sema::ActOnOpenMPTargetTeamsDistributeParallelForSimdDirective(
+    ArrayRef<OMPClause *> Clauses, Stmt *AStmt, SourceLocation StartLoc,
+    SourceLocation EndLoc,
+    llvm::DenseMap<ValueDecl *, Expr *> &VarsWithImplicitDSA) {
+  if (!AStmt)
+    return StmtError();
+
+  CapturedStmt *CS = cast<CapturedStmt>(AStmt);
+  // 1.2.2 OpenMP Language Terminology
+  // Structured block - An executable statement with a single entry at the
+  // top and a single exit at the bottom.
+  // The point of exit cannot be a branch out of the structured block.
+  // longjmp() and throw() must not violate the entry/exit criteria.
+  CS->getCapturedDecl()->setNothrow();
+
+  OMPLoopDirective::HelperExprs B;
+  // In presence of clause 'collapse' with number of loops, it will
+  // define the nested loops number.
+  auto NestedLoopCount = CheckOpenMPLoop(
+      OMPD_target_teams_distribute_parallel_for_simd,
+      getCollapseNumberExpr(Clauses),
+      nullptr /*ordered not a clause on distribute*/, AStmt, *this, *DSAStack,
+      VarsWithImplicitDSA, B);
+  if (NestedLoopCount == 0)
+    return StmtError();
+
+  assert((CurContext->isDependentContext() || B.builtAll()) &&
+         "omp target teams distribute parallel for simd loop exprs were not "
+         "built");
+
+  if (!CurContext->isDependentContext()) {
+    // Finalize the clauses that need pre-built expressions for CodeGen.
+    for (auto C : Clauses) {
+      if (auto *LC = dyn_cast<OMPLinearClause>(C))
+        if (FinishOpenMPLinearClause(*LC, cast<DeclRefExpr>(B.IterationVarRef),
+                                     B.NumIterations, *this, CurScope,
+                                     DSAStack))
+          return StmtError();
+    }
+  }
+
+  getCurFunction()->setHasBranchProtectedScope();
+  return OMPTargetTeamsDistributeParallelForSimdDirective::Create(
+      Context, StartLoc, EndLoc, NestedLoopCount, Clauses, AStmt, B);
+}
+
 OMPClause *Sema::ActOnOpenMPSingleExprClause(OpenMPClauseKind Kind, Expr *Expr,
                                              SourceLocation StartLoc,
                                              SourceLocation LParenLoc,
@@ -7397,7 +7450,8 @@ OMPClause *Sema::ActOnOpenMPPrivateClause(ArrayRef<Expr *> VarList,
     if (CurrDir == OMPD_target || CurrDir == OMPD_target_parallel ||
         CurrDir == OMPD_target_teams ||
         CurrDir == OMPD_target_teams_distribute ||
-        CurrDir == OMPD_target_teams_distribute_parallel_for) {
+        CurrDir == OMPD_target_teams_distribute_parallel_for ||
+        CurrDir == OMPD_target_teams_distribute_parallel_for_simd) {
       OpenMPClauseKind ConflictKind;
       if (DSAStack->checkMappableExprComponentListsForDecl(
               VD, /*CurrentRegionOnly=*/true,
@@ -7657,7 +7711,8 @@ OMPClause *Sema::ActOnOpenMPFirstprivateClause(ArrayRef<Expr *> VarList,
       if (CurrDir == OMPD_target || CurrDir == OMPD_target_parallel ||
           CurrDir == OMPD_target_teams ||
           CurrDir == OMPD_target_teams_distribute ||
-          CurrDir == OMPD_target_teams_distribute_parallel_for) {
+          CurrDir == OMPD_target_teams_distribute_parallel_for ||
+          CurrDir == OMPD_target_teams_distribute_parallel_for_simd) {
         OpenMPClauseKind ConflictKind;
         if (DSAStack->checkMappableExprComponentListsForDecl(
                 VD, /*CurrentRegionOnly=*/true,
@@ -10175,7 +10230,8 @@ checkMappableExpressionList(Sema &SemaRef, DSAStackTy *DSAS,
       // attribute clause on the same construct
       if ((DKind == OMPD_target || DKind == OMPD_target_teams ||
            DKind == OMPD_target_teams_distribute ||
-           DKind == OMPD_target_teams_distribute_parallel_for) && VD) {
+           DKind == OMPD_target_teams_distribute_parallel_for ||
+           DKind == OMPD_target_teams_distribute_parallel_for_simd) && VD) {
         auto DVar = DSAS->getTopDSA(VD, false);
         if (isOpenMPPrivate(DVar.CKind)) {
           SemaRef.Diag(ELoc, diag::err_omp_variable_in_given_clause_and_dsa)
diff --git a/lib/Sema/SemaOverload.cpp b/lib/Sema/SemaOverload.cpp
index 47e3df20d911..b5c0e634fa50 100644
--- a/lib/Sema/SemaOverload.cpp
+++ b/lib/Sema/SemaOverload.cpp
@@ -8958,9 +8958,7 @@ OverloadCandidateSet::BestViableFunction(Sema &S, SourceLocation Loc,
                S.IdentifyCUDAPreference(Caller, Cand->Function) ==
                    Sema::CFP_WrongSide;
       };
-      Candidates.erase(std::remove_if(Candidates.begin(), Candidates.end(),
-                                      IsWrongSideCandidate),
-                       Candidates.end());
+      llvm::erase_if(Candidates, IsWrongSideCandidate);
     }
   }
 
diff --git a/lib/Sema/SemaTemplate.cpp b/lib/Sema/SemaTemplate.cpp
index facc5d1b375b..66a10ef7993e 100644
--- a/lib/Sema/SemaTemplate.cpp
+++ b/lib/Sema/SemaTemplate.cpp
@@ -4244,7 +4244,7 @@ namespace {
     UnnamedLocalNoLinkageFinder(Sema &S, SourceRange SR) : S(S), SR(SR) { }
 
     bool Visit(QualType T) {
-      return inherited::Visit(T.getTypePtr());
+      return T.isNull() ? false : inherited::Visit(T.getTypePtr());
     }
 
 #define TYPE(Class, Parent) \
@@ -4497,17 +4497,7 @@ bool Sema::CheckTemplateArgument(TemplateTypeParmDecl *Param,
   //
   // C++11 allows these, and even in C++03 we allow them as an extension with
   // a warning.
-  bool NeedsCheck;
-  if (LangOpts.CPlusPlus11)
-    NeedsCheck =
-        !Diags.isIgnored(diag::warn_cxx98_compat_template_arg_unnamed_type,
-                         SR.getBegin()) ||
-        !Diags.isIgnored(diag::warn_cxx98_compat_template_arg_local_type,
-                         SR.getBegin());
-  else
-    NeedsCheck = Arg->hasUnnamedOrLocalType();
-
-  if (NeedsCheck) {
+  if (LangOpts.CPlusPlus11 || Arg->hasUnnamedOrLocalType()) {
     UnnamedLocalNoLinkageFinder Finder(*this, SR);
     (void)Finder.Visit(Context.getCanonicalType(Arg));
   }
diff --git a/lib/Sema/SemaTemplateDeduction.cpp b/lib/Sema/SemaTemplateDeduction.cpp
index 0bc85a2f2635..7f1fd91c46f0 100644
--- a/lib/Sema/SemaTemplateDeduction.cpp
+++ b/lib/Sema/SemaTemplateDeduction.cpp
@@ -288,19 +288,22 @@ checkDeducedTemplateArguments(ASTContext &Context,
         X.pack_size() != Y.pack_size())
       return DeducedTemplateArgument();
 
+    llvm::SmallVector<TemplateArgument, 8> NewPack;
     for (TemplateArgument::pack_iterator XA = X.pack_begin(),
                                       XAEnd = X.pack_end(),
                                          YA = Y.pack_begin();
          XA != XAEnd; ++XA, ++YA) {
-      // FIXME: Do we need to merge the results together here?
-      if (checkDeducedTemplateArguments(Context,
-                    DeducedTemplateArgument(*XA, X.wasDeducedFromArrayBound()),
-                    DeducedTemplateArgument(*YA, Y.wasDeducedFromArrayBound()))
-            .isNull())
+      TemplateArgument Merged = checkDeducedTemplateArguments(
+          Context, DeducedTemplateArgument(*XA, X.wasDeducedFromArrayBound()),
+          DeducedTemplateArgument(*YA, Y.wasDeducedFromArrayBound()));
+      if (Merged.isNull())
         return DeducedTemplateArgument();
+      NewPack.push_back(Merged);
     }
 
-    return X;
+    return DeducedTemplateArgument(
+        TemplateArgument::CreatePackCopy(Context, NewPack),
+        X.wasDeducedFromArrayBound() && Y.wasDeducedFromArrayBound());
   }
 
   llvm_unreachable("Invalid TemplateArgument Kind!");
@@ -672,17 +675,20 @@ public:
     // for that pack, then clear out the deduced argument.
     for (auto &Pack : Packs) {
       DeducedTemplateArgument &DeducedArg = Deduced[Pack.Index];
-      if (!DeducedArg.isNull()) {
+      if (!Pack.New.empty() || !DeducedArg.isNull()) {
+        while (Pack.New.size() < PackElements)
+          Pack.New.push_back(DeducedTemplateArgument());
         Pack.New.push_back(DeducedArg);
         DeducedArg = DeducedTemplateArgument();
       }
     }
+    ++PackElements;
   }
 
   /// \brief Finish template argument deduction for a set of argument packs,
   /// producing the argument packs and checking for consistency with prior
   /// deductions.
-  Sema::TemplateDeductionResult finish(bool HasAnyArguments) {
+  Sema::TemplateDeductionResult finish() {
     // Build argument packs for each of the parameter packs expanded by this
     // pack expansion.
     for (auto &Pack : Packs) {
@@ -691,7 +697,7 @@ public:
 
       // Build or find a new value for this pack.
       DeducedTemplateArgument NewPack;
-      if (HasAnyArguments && Pack.New.empty()) {
+      if (PackElements && Pack.New.empty()) {
         if (Pack.DeferredDeduction.isNull()) {
           // We were not able to deduce anything for this parameter pack
           // (because it only appeared in non-deduced contexts), so just
@@ -758,6 +764,7 @@ private:
   TemplateParameterList *TemplateParams;
   SmallVectorImpl<DeducedTemplateArgument> &Deduced;
   TemplateDeductionInfo &Info;
+  unsigned PackElements = 0;
 
   SmallVector<DeducedPack, 2> Packs;
 };
@@ -861,10 +868,7 @@ DeduceTemplateArguments(Sema &S,
     QualType Pattern = Expansion->getPattern();
     PackDeductionScope PackScope(S, TemplateParams, Deduced, Info, Pattern);
 
-    bool HasAnyArguments = false;
     for (; ArgIdx < NumArgs; ++ArgIdx) {
-      HasAnyArguments = true;
-
       // Deduce template arguments from the pattern.
       if (Sema::TemplateDeductionResult Result
             = DeduceTemplateArgumentsByTypeMatch(S, TemplateParams, Pattern,
@@ -877,7 +881,7 @@ DeduceTemplateArguments(Sema &S,
 
     // Build argument packs for each of the parameter packs expanded by this
     // pack expansion.
-    if (auto Result = PackScope.finish(HasAnyArguments))
+    if (auto Result = PackScope.finish())
       return Result;
   }
 
@@ -1935,10 +1939,7 @@ DeduceTemplateArguments(Sema &S, TemplateParameterList *TemplateParams,
     // Keep track of the deduced template arguments for each parameter pack
     // expanded by this pack expansion (the outer index) and for each
     // template argument (the inner SmallVectors).
-    bool HasAnyArguments = false;
     for (; hasTemplateArgumentForDeduction(Args, ArgIdx); ++ArgIdx) {
-      HasAnyArguments = true;
-
       // Deduce template arguments from the pattern.
       if (Sema::TemplateDeductionResult Result
             = DeduceTemplateArguments(S, TemplateParams, Pattern, Args[ArgIdx],
@@ -1950,7 +1951,7 @@ DeduceTemplateArguments(Sema &S, TemplateParameterList *TemplateParams,
 
     // Build argument packs for each of the parameter packs expanded by this
     // pack expansion.
-    if (auto Result = PackScope.finish(HasAnyArguments))
+    if (auto Result = PackScope.finish())
       return Result;
   }
 
@@ -2145,6 +2146,16 @@ ConvertDeducedTemplateArgument(Sema &S, NamedDecl *Param,
       InnerArg.setDeducedFromArrayBound(Arg.wasDeducedFromArrayBound());
       assert(InnerArg.getKind() != TemplateArgument::Pack &&
              "deduced nested pack");
+      if (P.isNull()) {
+        // We deduced arguments for some elements of this pack, but not for
+        // all of them. This happens if we get a conditionally-non-deduced
+        // context in a pack expansion (such as an overload set in one of the
+        // arguments).
+        S.Diag(Param->getLocation(),
+               diag::err_template_arg_deduced_incomplete_pack)
+          << Arg << Param;
+        return true;
+      }
       if (ConvertArg(InnerArg, PackedArgsBuilder.size()))
         return true;
 
@@ -3192,67 +3203,59 @@ static Sema::TemplateDeductionResult DeduceTemplateArgumentByListElement(
 
 /// \brief Attempt template argument deduction from an initializer list
 ///        deemed to be an argument in a function call.
-static bool
+static Sema::TemplateDeductionResult
 DeduceFromInitializerList(Sema &S, TemplateParameterList *TemplateParams,
                           QualType AdjustedParamType, InitListExpr *ILE,
                           TemplateDeductionInfo &Info,
                           SmallVectorImpl<DeducedTemplateArgument> &Deduced,
-                          unsigned TDF, Sema::TemplateDeductionResult &Result) {
-
-  // [temp.deduct.call] p1 (post CWG-1591)
-  // If removing references and cv-qualifiers from P gives
-  // std::initializer_list<P0> or P0[N] for some P0 and N and the argument is a
-  // non-empty initializer list (8.5.4), then deduction is performed instead for
-  // each element of the initializer list, taking P0 as a function template
-  // parameter type and the initializer element as its argument, and in the
-  // P0[N] case, if N is a non-type template parameter, N is deduced from the
-  // length of the initializer list. Otherwise, an initializer list argument
-  // causes the parameter to be considered a non-deduced context
-
-  const bool IsConstSizedArray = AdjustedParamType->isConstantArrayType();
-
-  const bool IsDependentSizedArray =
-      !IsConstSizedArray && AdjustedParamType->isDependentSizedArrayType();
-
-  QualType ElTy;  // The element type of the std::initializer_list or the array.
-
-  const bool IsSTDList = !IsConstSizedArray && !IsDependentSizedArray &&
-                         S.isStdInitializerList(AdjustedParamType, &ElTy);
-
-  if (!IsConstSizedArray && !IsDependentSizedArray && !IsSTDList)
-    return false;
-
-  Result = Sema::TDK_Success;
-  // If we are not deducing against the 'T' in a std::initializer_list<T> then
-  // deduce against the 'T' in T[N].
-  if (ElTy.isNull()) {
-    assert(!IsSTDList);
-    ElTy = S.Context.getAsArrayType(AdjustedParamType)->getElementType();
+                          unsigned TDF) {
+  // C++ [temp.deduct.call]p1: (CWG 1591)
+  //   If removing references and cv-qualifiers from P gives
+  //   std::initializer_list<P0> or P0[N] for some P0 and N and the argument is
+  //   a non-empty initializer list, then deduction is performed instead for
+  //   each element of the initializer list, taking P0 as a function template
+  //   parameter type and the initializer element as its argument
+  //
+  // FIXME: Remove references and cv-qualifiers here? Consider
+  //          std::initializer_list<std::initializer_list<T>&&>
+  QualType ElTy;
+  auto *ArrTy = S.Context.getAsArrayType(AdjustedParamType);
+  if (ArrTy)
+    ElTy = ArrTy->getElementType();
+  else if (!S.isStdInitializerList(AdjustedParamType, &ElTy)) {
+    //   Otherwise, an initializer list argument causes the parameter to be
+    //   considered a non-deduced context
+    return Sema::TDK_Success;
   }
+
   // Deduction only needs to be done for dependent types.
   if (ElTy->isDependentType()) {
     for (Expr *E : ILE->inits()) {
-      if ((Result = DeduceTemplateArgumentByListElement(S, TemplateParams, ElTy,
-                                                        E, Info, Deduced, TDF)))
-        return true;
+      if (auto Result = DeduceTemplateArgumentByListElement(
+              S, TemplateParams, ElTy, E, Info, Deduced, TDF))
+        return Result;
     }
   }
-  if (IsDependentSizedArray) {
-    const DependentSizedArrayType *ArrTy =
-        S.Context.getAsDependentSizedArrayType(AdjustedParamType);
+
+  //   in the P0[N] case, if N is a non-type template parameter, N is deduced
+  //   from the length of the initializer list.
+  // FIXME: We're not supposed to get here if N would be deduced as 0.
+  if (auto *DependentArrTy = dyn_cast_or_null<DependentSizedArrayType>(ArrTy)) {
     // Determine the array bound is something we can deduce.
     if (NonTypeTemplateParmDecl *NTTP =
-            getDeducedParameterFromExpr(Info, ArrTy->getSizeExpr())) {
+            getDeducedParameterFromExpr(Info, DependentArrTy->getSizeExpr())) {
       // We can perform template argument deduction for the given non-type
       // template parameter.
       llvm::APInt Size(S.Context.getIntWidth(NTTP->getType()),
                        ILE->getNumInits());
-      Result = DeduceNonTypeTemplateArgument(
-          S, TemplateParams, NTTP, llvm::APSInt(Size), NTTP->getType(),
-          /*ArrayBound=*/true, Info, Deduced);
+      if (auto Result = DeduceNonTypeTemplateArgument(
+              S, TemplateParams, NTTP, llvm::APSInt(Size), NTTP->getType(),
+              /*ArrayBound=*/true, Info, Deduced))
+        return Result;
     }
   }
-  return true;
+
+  return Sema::TDK_Success;
 }
 
 /// \brief Perform template argument deduction by matching a parameter type
@@ -3268,15 +3271,10 @@ DeduceTemplateArgumentByListElement(Sema &S,
                                     unsigned TDF) {
   // Handle the case where an init list contains another init list as the
   // element.
-  if (InitListExpr *ILE = dyn_cast<InitListExpr>(Arg)) {
-    Sema::TemplateDeductionResult Result;
-    if (!DeduceFromInitializerList(S, TemplateParams,
-                                   ParamType.getNonReferenceType(), ILE, Info,
-                                   Deduced, TDF, Result))
-      return Sema::TDK_Success; // Just ignore this expression.
-
-    return Result;
-  }
+  if (InitListExpr *ILE = dyn_cast<InitListExpr>(Arg))
+    return DeduceFromInitializerList(S, TemplateParams,
+                                     ParamType.getNonReferenceType(), ILE, Info,
+                                     Deduced, TDF);
 
   // For all other cases, just match by type.
   QualType ArgType = Arg->getType();
@@ -3363,58 +3361,51 @@ Sema::TemplateDeductionResult Sema::DeduceTemplateArguments(
       ParamTypes.push_back(Function->getParamDecl(I)->getType());
   }
 
-  // Deduce template arguments from the function parameters.
-  Deduced.resize(TemplateParams->size());
-  unsigned ArgIdx = 0;
   SmallVector<OriginalCallArg, 4> OriginalCallArgs;
-  for (unsigned ParamIdx = 0, NumParamTypes = ParamTypes.size();
-       ParamIdx != NumParamTypes; ++ParamIdx) {
-    QualType OrigParamType = ParamTypes[ParamIdx];
-    QualType ParamType = OrigParamType;
 
-    const PackExpansionType *ParamExpansion
-      = dyn_cast<PackExpansionType>(ParamType);
-    if (!ParamExpansion) {
-      // Simple case: matching a function parameter to a function argument.
-      if (ArgIdx >= CheckArgs)
-        break;
+  // Deduce an argument of type ParamType from an expression with index ArgIdx.
+  auto DeduceCallArgument = [&](QualType ParamType, unsigned ArgIdx) {
+    Expr *Arg = Args[ArgIdx];
+    QualType ArgType = Arg->getType();
+    QualType OrigParamType = ParamType;
 
-      Expr *Arg = Args[ArgIdx++];
-      QualType ArgType = Arg->getType();
+    unsigned TDF = 0;
+    if (AdjustFunctionParmAndArgTypesForDeduction(*this, TemplateParams,
+                                                  ParamType, ArgType, Arg,
+                                                  TDF))
+      return Sema::TDK_Success;
 
-      unsigned TDF = 0;
-      if (AdjustFunctionParmAndArgTypesForDeduction(*this, TemplateParams,
-                                                    ParamType, ArgType, Arg,
-                                                    TDF))
-        continue;
+    // If we have nothing to deduce, we're done.
+    if (!hasDeducibleTemplateParameters(*this, FunctionTemplate, ParamType))
+      return Sema::TDK_Success;
 
-      // If we have nothing to deduce, we're done.
-      if (!hasDeducibleTemplateParameters(*this, FunctionTemplate, ParamType))
-        continue;
+    // If the argument is an initializer list ...
+    if (InitListExpr *ILE = dyn_cast<InitListExpr>(Arg))
+      return DeduceFromInitializerList(*this, TemplateParams, ParamType, ILE,
+                                       Info, Deduced, TDF);
 
-      // If the argument is an initializer list ...
-      if (InitListExpr *ILE = dyn_cast<InitListExpr>(Arg)) {
-        TemplateDeductionResult Result;
-        // Removing references was already done.
-        if (!DeduceFromInitializerList(*this, TemplateParams, ParamType, ILE,
-                                       Info, Deduced, TDF, Result))
-          continue;
+    // Keep track of the argument type and corresponding parameter index,
+    // so we can check for compatibility between the deduced A and A.
+    OriginalCallArgs.push_back(OriginalCallArg(OrigParamType, ArgIdx, ArgType));
 
-        if (Result)
-          return Result;
-        // Don't track the argument type, since an initializer list has none.
-        continue;
-      }
+    return DeduceTemplateArgumentsByTypeMatch(*this, TemplateParams, ParamType,
+                                              ArgType, Info, Deduced, TDF);
+  };
 
-      // Keep track of the argument type and corresponding parameter index,
-      // so we can check for compatibility between the deduced A and A.
-      OriginalCallArgs.push_back(OriginalCallArg(OrigParamType, ArgIdx-1,
-                                                 ArgType));
+  // Deduce template arguments from the function parameters.
+  Deduced.resize(TemplateParams->size());
+  for (unsigned ParamIdx = 0, NumParamTypes = ParamTypes.size(), ArgIdx = 0;
+       ParamIdx != NumParamTypes; ++ParamIdx) {
+    QualType ParamType = ParamTypes[ParamIdx];
 
-      if (TemplateDeductionResult Result
-            = DeduceTemplateArgumentsByTypeMatch(*this, TemplateParams,
-                                                 ParamType, ArgType,
-                                                 Info, Deduced, TDF))
+    const PackExpansionType *ParamExpansion =
+        dyn_cast<PackExpansionType>(ParamType);
+    if (!ParamExpansion) {
+      // Simple case: matching a function parameter to a function argument.
+      if (ArgIdx >= CheckArgs)
+        break;
+
+      if (auto Result = DeduceCallArgument(ParamType, ArgIdx++))
         return Result;
 
       continue;
@@ -3429,6 +3420,9 @@ Sema::TemplateDeductionResult Sema::DeduceTemplateArguments(
     //   the function parameter pack. For a function parameter pack that does
     //   not occur at the end of the parameter-declaration-list, the type of
     //   the parameter pack is a non-deduced context.
+    // FIXME: This does not say that subsequent parameters are also non-deduced.
+    // See also DR1388 / DR1399, which effectively says we should keep deducing
+    // after the pack.
     if (ParamIdx + 1 < NumParamTypes)
       break;
 
@@ -3436,57 +3430,13 @@ Sema::TemplateDeductionResult Sema::DeduceTemplateArguments(
     PackDeductionScope PackScope(*this, TemplateParams, Deduced, Info,
                                  ParamPattern);
 
-    bool HasAnyArguments = false;
-    for (; ArgIdx < Args.size(); ++ArgIdx) {
-      HasAnyArguments = true;
-
-      QualType OrigParamType = ParamPattern;
-      ParamType = OrigParamType;
-      Expr *Arg = Args[ArgIdx];
-      QualType ArgType = Arg->getType();
-
-      unsigned TDF = 0;
-      if (AdjustFunctionParmAndArgTypesForDeduction(*this, TemplateParams,
-                                                    ParamType, ArgType, Arg,
-                                                    TDF)) {
-        // We can't actually perform any deduction for this argument, so stop
-        // deduction at this point.
-        ++ArgIdx;
-        break;
-      }
-
-      // As above, initializer lists need special handling.
-      if (InitListExpr *ILE = dyn_cast<InitListExpr>(Arg)) {
-        TemplateDeductionResult Result;
-        if (!DeduceFromInitializerList(*this, TemplateParams, ParamType, ILE,
-                                       Info, Deduced, TDF, Result)) {
-          ++ArgIdx;
-          break;
-        }
-
-        if (Result)
-          return Result;
-      } else {
-
-        // Keep track of the argument type and corresponding argument index,
-        // so we can check for compatibility between the deduced A and A.
-        if (hasDeducibleTemplateParameters(*this, FunctionTemplate, ParamType))
-          OriginalCallArgs.push_back(OriginalCallArg(OrigParamType, ArgIdx,
-                                                     ArgType));
-
-        if (TemplateDeductionResult Result
-            = DeduceTemplateArgumentsByTypeMatch(*this, TemplateParams,
-                                                 ParamType, ArgType, Info,
-                                                 Deduced, TDF))
-          return Result;
-      }
-
-      PackScope.nextPackElement();
-    }
+    for (; ArgIdx < Args.size(); PackScope.nextPackElement(), ++ArgIdx)
+      if (auto Result = DeduceCallArgument(ParamPattern, ArgIdx))
+        return Result;
 
     // Build argument packs for each of the parameter packs expanded by this
     // pack expansion.
-    if (auto Result = PackScope.finish(HasAnyArguments))
+    if (auto Result = PackScope.finish())
       return Result;
 
     // After we've matching against a parameter pack, we're done.
diff --git a/lib/Sema/TreeTransform.h b/lib/Sema/TreeTransform.h
index 3ab6019f0ec3..66892936e573 100644
--- a/lib/Sema/TreeTransform.h
+++ b/lib/Sema/TreeTransform.h
@@ -7766,6 +7766,20 @@ TreeTransform<Derived>::TransformOMPTargetTeamsDistributeParallelForDirective(
   return Res;
 }
 
+template <typename Derived>
+StmtResult TreeTransform<Derived>::
+    TransformOMPTargetTeamsDistributeParallelForSimdDirective(
+        OMPTargetTeamsDistributeParallelForSimdDirective *D) {
+  DeclarationNameInfo DirName;
+  getDerived().getSema().StartOpenMPDSABlock(
+      OMPD_target_teams_distribute_parallel_for_simd, DirName, nullptr,
+      D->getLocStart());
+  auto Res = getDerived().TransformOMPExecutableDirective(D);
+  getDerived().getSema().EndOpenMPDSABlock(Res.get());
+  return Res;
+}
+
+
 //===----------------------------------------------------------------------===//
 // OpenMP clause transformation
 //===----------------------------------------------------------------------===//
diff --git a/lib/Serialization/ASTReaderStmt.cpp b/lib/Serialization/ASTReaderStmt.cpp
index 5607f764a9c3..19fac55664ae 100644
--- a/lib/Serialization/ASTReaderStmt.cpp
+++ b/lib/Serialization/ASTReaderStmt.cpp
@@ -2860,6 +2860,11 @@ void ASTStmtReader::VisitOMPTargetTeamsDistributeParallelForDirective(
   VisitOMPLoopDirective(D);
 }
 
+void ASTStmtReader::VisitOMPTargetTeamsDistributeParallelForSimdDirective(
+    OMPTargetTeamsDistributeParallelForSimdDirective *D) {
+  VisitOMPLoopDirective(D);
+}
+
 //===----------------------------------------------------------------------===//
 // ASTReader Implementation
 //===----------------------------------------------------------------------===//
@@ -3638,6 +3643,14 @@ Stmt *ASTReader::ReadStmtFromStream(ModuleFile &F) {
       break;
     }
 
+    case STMT_OMP_TARGET_TEAMS_DISTRIBUTE_PARALLEL_FOR_SIMD_DIRECTIVE: {
+      auto NumClauses = Record[ASTStmtReader::NumStmtFields];
+      auto CollapsedNum = Record[ASTStmtReader::NumStmtFields + 1];
+      S = OMPTargetTeamsDistributeParallelForSimdDirective::CreateEmpty(
+          Context, NumClauses, CollapsedNum, Empty);
+      break;
+    }
+
     case EXPR_CXX_OPERATOR_CALL:
       S = new (Context) CXXOperatorCallExpr(Context, Empty);
       break;
diff --git a/lib/Serialization/ASTWriterStmt.cpp b/lib/Serialization/ASTWriterStmt.cpp
index 3993be146edf..162b2bd25260 100644
--- a/lib/Serialization/ASTWriterStmt.cpp
+++ b/lib/Serialization/ASTWriterStmt.cpp
@@ -2556,6 +2556,13 @@ void ASTStmtWriter::VisitOMPTargetTeamsDistributeParallelForDirective(
   Code = serialization::STMT_OMP_TARGET_TEAMS_DISTRIBUTE_PARALLEL_FOR_DIRECTIVE;
 }
 
+void ASTStmtWriter::VisitOMPTargetTeamsDistributeParallelForSimdDirective(
+    OMPTargetTeamsDistributeParallelForSimdDirective *D) {
+  VisitOMPLoopDirective(D);
+  Code = serialization::
+      STMT_OMP_TARGET_TEAMS_DISTRIBUTE_PARALLEL_FOR_SIMD_DIRECTIVE;
+}
+
 //===----------------------------------------------------------------------===//
 // ASTWriter Implementation
 //===----------------------------------------------------------------------===//
diff --git a/lib/StaticAnalyzer/Core/ExprEngine.cpp b/lib/StaticAnalyzer/Core/ExprEngine.cpp
index 5b2119aeda27..707168b4de0a 100644
--- a/lib/StaticAnalyzer/Core/ExprEngine.cpp
+++ b/lib/StaticAnalyzer/Core/ExprEngine.cpp
@@ -869,6 +869,7 @@ void ExprEngine::Visit(const Stmt *S, ExplodedNode *Pred,
     case Stmt::OMPTargetTeamsDirectiveClass:
     case Stmt::OMPTargetTeamsDistributeDirectiveClass:
     case Stmt::OMPTargetTeamsDistributeParallelForDirectiveClass:
+    case Stmt::OMPTargetTeamsDistributeParallelForSimdDirectiveClass:
       llvm_unreachable("Stmt should not be in analyzer evaluation loop");
 
     case Stmt::ObjCSubscriptRefExprClass:
diff --git a/lib/Tooling/Core/CMakeLists.txt b/lib/Tooling/Core/CMakeLists.txt
index f6348cbf80ed..e2b0dd424d10 100644
--- a/lib/Tooling/Core/CMakeLists.txt
+++ b/lib/Tooling/Core/CMakeLists.txt
@@ -4,6 +4,7 @@ add_clang_library(clangToolingCore
   Lookup.cpp
   Replacement.cpp
   QualTypeNames.cpp
+  Diagnostic.cpp
 
   LINK_LIBS
   clangAST
diff --git a/lib/Tooling/Core/Diagnostic.cpp b/lib/Tooling/Core/Diagnostic.cpp
new file mode 100644
index 000000000000..3bbc2b901e38
--- /dev/null
+++ b/lib/Tooling/Core/Diagnostic.cpp
@@ -0,0 +1,46 @@
+//===--- Diagnostic.cpp - Framework for clang diagnostics tools ----------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+//  Implements classes to support/store diagnostics refactoring.
+//
+//===----------------------------------------------------------------------===//
+
+#include "clang/Tooling/Core/Diagnostic.h"
+#include "clang/Basic/SourceManager.h"
+
+namespace clang {
+namespace tooling {
+
+DiagnosticMessage::DiagnosticMessage(llvm::StringRef Message)
+    : Message(Message), FileOffset(0) {}
+
+DiagnosticMessage::DiagnosticMessage(llvm::StringRef Message,
+                                     const SourceManager &Sources,
+                                     SourceLocation Loc)
+    : Message(Message) {
+  assert(Loc.isValid() && Loc.isFileID());
+  FilePath = Sources.getFilename(Loc);
+  FileOffset = Sources.getFileOffset(Loc);
+}
+
+Diagnostic::Diagnostic(llvm::StringRef DiagnosticName,
+                       Diagnostic::Level DiagLevel, StringRef BuildDirectory)
+    : DiagnosticName(DiagnosticName), DiagLevel(DiagLevel),
+      BuildDirectory(BuildDirectory) {}
+
+Diagnostic::Diagnostic(llvm::StringRef DiagnosticName,
+                       DiagnosticMessage &Message,
+                       llvm::StringMap<Replacements> &Fix,
+                       SmallVector<DiagnosticMessage, 1> &Notes,
+                       Level DiagLevel, llvm::StringRef BuildDirectory)
+    : DiagnosticName(DiagnosticName), Message(Message), Fix(Fix), Notes(Notes),
+      DiagLevel(DiagLevel), BuildDirectory(BuildDirectory) {}
+
+} // end namespace tooling
+} // end namespace clang