MFC r309124:

Upgrade our copies of clang, llvm, lldb, compiler-rt and libc++ to 3.9.0
release, and add lld 3.9.0.  Also completely revamp the build system for
clang, llvm, lldb and their related tools.

Please note that from 3.5.0 onwards, clang, llvm and lldb require C++11
support to build; see UPDATING for more information.

Release notes for llvm, clang and lld are available here:
<http://llvm.org/releases/3.9.0/docs/ReleaseNotes.html>
<http://llvm.org/releases/3.9.0/tools/clang/docs/ReleaseNotes.html>
<http://llvm.org/releases/3.9.0/tools/lld/docs/ReleaseNotes.html>

Thanks to Ed Maste, Bryan Drewery, Andrew Turner, Antoine Brodin and Jan
Beich for their help.

Relnotes:	yes

MFC r309147:

Pull in r282174 from upstream llvm trunk (by Krzysztof Parzyszek):

  [PPC] Set SP after loading data from stack frame, if no red zone is
  present

  Follow-up to r280705: Make sure that the SP is only restored after
  all data is loaded from the stack frame, if there is no red zone.

  This completes the fix for
  https://llvm.org/bugs/show_bug.cgi?id=26519.

  Differential Revision: https://reviews.llvm.org/D24466

Reported by:    Mark Millard
PR:             214433

MFC r309149:

Pull in r283060 from upstream llvm trunk (by Hal Finkel):

  [PowerPC] Refactor soft-float support, and enable PPC64 soft float

  This change enables soft-float for PowerPC64, and also makes
  soft-float disable all vector instruction sets for both 32-bit and
  64-bit modes. This latter part is necessary because the PPC backend
  canonicalizes many Altivec vector types to floating-point types, and
  so soft-float breaks scalarization support for many operations. Both
  for embedded targets and for operating-system kernels desiring
  soft-float support, it seems reasonable that disabling hardware
  floating-point also disables vector instructions (embedded targets
  without hardware floating point support are unlikely to have Altivec,
  etc. and operating system kernels desiring not to use floating-point
  registers to lower syscall cost are unlikely to want to use vector
  registers either). If someone needs this to work, we'll need to
  change the fact that we promote many Altivec operations to act on
  v4f32. To make it possible to disable Altivec when soft-float is
  enabled, hardware floating-point support needs to be expressed as a
  positive feature, like the others, and not a negative feature,
  because target features cannot have dependencies on the disabling of
  some other feature. So +soft-float has now become -hard-float.

  Fixes PR26970.

Pull in r283061 from upstream clang trunk (by Hal Finkel):

  [PowerPC] Enable soft-float for PPC64, and +soft-float -> -hard-float

  Enable soft-float support on PPC64, as the backend now supports it.
  Also, the backend now uses -hard-float instead of +soft-float, so set
  the target features accordingly.

  Fixes PR26970.

Reported by:	Mark Millard
PR:		214433

MFC r309212:

Add a few missed clang 3.9.0 files to OptionalObsoleteFiles.

MFC r309262:

Fix packaging for clang, lldb and lld 3.9.0

During the upgrade of clang/llvm etc to 3.9.0 in r309124, the PACKAGE
directive in the usr.bin/clang/*.mk files got dropped accidentally.

Restore it, with a few minor changes and additions:
* Correct license in clang.ucl to NCSA
* Add PACKAGE=clang for clang and most of the "ll" tools
* Put lldb in its own package
* Put lld in its own package

Reviewed by:	gjb, jmallett
Differential Revision: https://reviews.freebsd.org/D8666

MFC r309656:

During the bootstrap phase, when building the minimal llvm library on
PowerPC, add lib/Support/Atomic.cpp.  This is needed because upstream
llvm revision r271821 disabled the use of std::call_once, which causes
some fallback functions from Atomic.cpp to be used instead.

Reported by:	Mark Millard
PR:		214902

MFC r309835:

Tentatively apply https://reviews.llvm.org/D18730 to work around gcc PR
70528 (bogus error: constructor required before non-static data member).
This should fix buildworld with the external gcc package.

Reported by:	https://jenkins.freebsd.org/job/FreeBSD_HEAD_amd64_gcc/

MFC r310194:

Upgrade our copies of clang, llvm, lld, lldb, compiler-rt and libc++ to
3.9.1 release.

Please note that from 3.5.0 onwards, clang, llvm and lldb require C++11
support to build; see UPDATING for more information.

Release notes for llvm, clang and lld will be available here:
<http://releases.llvm.org/3.9.1/docs/ReleaseNotes.html>
<http://releases.llvm.org/3.9.1/tools/clang/docs/ReleaseNotes.html>
<http://releases.llvm.org/3.9.1/tools/lld/docs/ReleaseNotes.html>

Relnotes:	yes

1 files changed, 3258 insertions, 914 deletions

diff --git a/contrib/llvm/tools/clang/lib/CodeGen/CGOpenMPRuntime.cpp b/contrib/llvm/tools/clang/lib/CodeGen/CGOpenMPRuntime.cpp
index 5cfacac..6a0edbe 100644
--- a/contrib/llvm/tools/clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/contrib/llvm/tools/clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -72,6 +72,8 @@ public:
   /// \return LValue for thread id variable. This LValue always has type int32*.
   virtual LValue getThreadIDVariableLValue(CodeGenFunction &CGF);
 
+  virtual void emitUntiedSwitch(CodeGenFunction & /*CGF*/) {}
+
   CGOpenMPRegionKind getRegionKind() const { return RegionKind; }
 
   OpenMPDirectiveKind getDirectiveKind() const { return Kind; }
@@ -82,6 +84,8 @@ public:
     return Info->getKind() == CR_OpenMP;
   }
 
+  ~CGOpenMPRegionInfo() override = default;
+
 protected:
   CGOpenMPRegionKind RegionKind;
   RegionCodeGenTy CodeGen;
@@ -90,7 +94,7 @@ protected:
 };
 
 /// \brief API for captured statement code generation in OpenMP constructs.
-class CGOpenMPOutlinedRegionInfo : public CGOpenMPRegionInfo {
+class CGOpenMPOutlinedRegionInfo final : public CGOpenMPRegionInfo {
 public:
   CGOpenMPOutlinedRegionInfo(const CapturedStmt &CS, const VarDecl *ThreadIDVar,
                              const RegionCodeGenTy &CodeGen,
@@ -100,6 +104,7 @@ public:
         ThreadIDVar(ThreadIDVar) {
     assert(ThreadIDVar != nullptr && "No ThreadID in OpenMP region.");
   }
+
   /// \brief Get a variable or parameter for storing global thread id
   /// inside OpenMP construct.
   const VarDecl *getThreadIDVariable() const override { return ThreadIDVar; }
@@ -120,16 +125,65 @@ private:
 };
 
 /// \brief API for captured statement code generation in OpenMP constructs.
-class CGOpenMPTaskOutlinedRegionInfo : public CGOpenMPRegionInfo {
+class CGOpenMPTaskOutlinedRegionInfo final : public CGOpenMPRegionInfo {
 public:
+  class UntiedTaskActionTy final : public PrePostActionTy {
+    bool Untied;
+    const VarDecl *PartIDVar;
+    const RegionCodeGenTy UntiedCodeGen;
+    llvm::SwitchInst *UntiedSwitch = nullptr;
+
+  public:
+    UntiedTaskActionTy(bool Tied, const VarDecl *PartIDVar,
+                       const RegionCodeGenTy &UntiedCodeGen)
+        : Untied(!Tied), PartIDVar(PartIDVar), UntiedCodeGen(UntiedCodeGen) {}
+    void Enter(CodeGenFunction &CGF) override {
+      if (Untied) {
+        // Emit task switching point.
+        auto PartIdLVal = CGF.EmitLoadOfPointerLValue(
+            CGF.GetAddrOfLocalVar(PartIDVar),
+            PartIDVar->getType()->castAs<PointerType>());
+        auto *Res = CGF.EmitLoadOfScalar(PartIdLVal, SourceLocation());
+        auto *DoneBB = CGF.createBasicBlock(".untied.done.");
+        UntiedSwitch = CGF.Builder.CreateSwitch(Res, DoneBB);
+        CGF.EmitBlock(DoneBB);
+        CGF.EmitBranchThroughCleanup(CGF.ReturnBlock);
+        CGF.EmitBlock(CGF.createBasicBlock(".untied.jmp."));
+        UntiedSwitch->addCase(CGF.Builder.getInt32(0),
+                              CGF.Builder.GetInsertBlock());
+        emitUntiedSwitch(CGF);
+      }
+    }
+    void emitUntiedSwitch(CodeGenFunction &CGF) const {
+      if (Untied) {
+        auto PartIdLVal = CGF.EmitLoadOfPointerLValue(
+            CGF.GetAddrOfLocalVar(PartIDVar),
+            PartIDVar->getType()->castAs<PointerType>());
+        CGF.EmitStoreOfScalar(CGF.Builder.getInt32(UntiedSwitch->getNumCases()),
+                              PartIdLVal);
+        UntiedCodeGen(CGF);
+        CodeGenFunction::JumpDest CurPoint =
+            CGF.getJumpDestInCurrentScope(".untied.next.");
+        CGF.EmitBranchThroughCleanup(CGF.ReturnBlock);
+        CGF.EmitBlock(CGF.createBasicBlock(".untied.jmp."));
+        UntiedSwitch->addCase(CGF.Builder.getInt32(UntiedSwitch->getNumCases()),
+                              CGF.Builder.GetInsertBlock());
+        CGF.EmitBranchThroughCleanup(CurPoint);
+        CGF.EmitBlock(CurPoint.getBlock());
+      }
+    }
+    unsigned getNumberOfParts() const { return UntiedSwitch->getNumCases(); }
+  };
   CGOpenMPTaskOutlinedRegionInfo(const CapturedStmt &CS,
                                  const VarDecl *ThreadIDVar,
                                  const RegionCodeGenTy &CodeGen,
-                                 OpenMPDirectiveKind Kind, bool HasCancel)
+                                 OpenMPDirectiveKind Kind, bool HasCancel,
+                                 const UntiedTaskActionTy &Action)
       : CGOpenMPRegionInfo(CS, TaskOutlinedRegion, CodeGen, Kind, HasCancel),
-        ThreadIDVar(ThreadIDVar) {
+        ThreadIDVar(ThreadIDVar), Action(Action) {
     assert(ThreadIDVar != nullptr && "No ThreadID in OpenMP region.");
   }
+
   /// \brief Get a variable or parameter for storing global thread id
   /// inside OpenMP construct.
   const VarDecl *getThreadIDVariable() const override { return ThreadIDVar; }
@@ -140,6 +194,10 @@ public:
   /// \brief Get the name of the capture helper.
   StringRef getHelperName() const override { return ".omp_outlined."; }
 
+  void emitUntiedSwitch(CodeGenFunction &CGF) override {
+    Action.emitUntiedSwitch(CGF);
+  }
+
   static bool classof(const CGCapturedStmtInfo *Info) {
     return CGOpenMPRegionInfo::classof(Info) &&
            cast<CGOpenMPRegionInfo>(Info)->getRegionKind() ==
@@ -150,6 +208,8 @@ private:
   /// \brief A variable or parameter storing global thread id for OpenMP
   /// constructs.
   const VarDecl *ThreadIDVar;
+  /// Action for emitting code for untied tasks.
+  const UntiedTaskActionTy &Action;
 };
 
 /// \brief API for inlined captured statement code generation in OpenMP
@@ -162,12 +222,14 @@ public:
       : CGOpenMPRegionInfo(InlinedRegion, CodeGen, Kind, HasCancel),
         OldCSI(OldCSI),
         OuterRegionInfo(dyn_cast_or_null<CGOpenMPRegionInfo>(OldCSI)) {}
+
   // \brief Retrieve the value of the context parameter.
   llvm::Value *getContextValue() const override {
     if (OuterRegionInfo)
       return OuterRegionInfo->getContextValue();
     llvm_unreachable("No context value for inlined OpenMP region");
   }
+
   void setContextValue(llvm::Value *V) override {
     if (OuterRegionInfo) {
       OuterRegionInfo->setContextValue(V);
@@ -175,6 +237,7 @@ public:
     }
     llvm_unreachable("No context value for inlined OpenMP region");
   }
+
   /// \brief Lookup the captured field decl for a variable.
   const FieldDecl *lookup(const VarDecl *VD) const override {
     if (OuterRegionInfo)
@@ -183,11 +246,13 @@ public:
     // captured variables, we can use the original one.
     return nullptr;
   }
+
   FieldDecl *getThisFieldDecl() const override {
     if (OuterRegionInfo)
       return OuterRegionInfo->getThisFieldDecl();
     return nullptr;
   }
+
   /// \brief Get a variable or parameter for storing global thread id
   /// inside OpenMP construct.
   const VarDecl *getThreadIDVariable() const override {
@@ -203,6 +268,11 @@ public:
     llvm_unreachable("No helper name for inlined OpenMP construct");
   }
 
+  void emitUntiedSwitch(CodeGenFunction &CGF) override {
+    if (OuterRegionInfo)
+      OuterRegionInfo->emitUntiedSwitch(CGF);
+  }
+
   CodeGenFunction::CGCapturedStmtInfo *getOldCSI() const { return OldCSI; }
 
   static bool classof(const CGCapturedStmtInfo *Info) {
@@ -210,6 +280,8 @@ public:
            cast<CGOpenMPRegionInfo>(Info)->getRegionKind() == InlinedRegion;
   }
 
+  ~CGOpenMPInlinedRegionInfo() override = default;
+
 private:
   /// \brief CodeGen info about outer OpenMP region.
   CodeGenFunction::CGCapturedStmtInfo *OldCSI;
@@ -221,7 +293,7 @@ private:
 /// captured fields. The name of the target region has to be unique in a given
 /// application so it is provided by the client, because only the client has
 /// the information to generate that.
-class CGOpenMPTargetRegionInfo : public CGOpenMPRegionInfo {
+class CGOpenMPTargetRegionInfo final : public CGOpenMPRegionInfo {
 public:
   CGOpenMPTargetRegionInfo(const CapturedStmt &CS,
                            const RegionCodeGenTy &CodeGen, StringRef HelperName)
@@ -245,9 +317,75 @@ private:
   StringRef HelperName;
 };
 
+static void EmptyCodeGen(CodeGenFunction &, PrePostActionTy &) {
+  llvm_unreachable("No codegen for expressions");
+}
+/// \brief API for generation of expressions captured in a innermost OpenMP
+/// region.
+class CGOpenMPInnerExprInfo final : public CGOpenMPInlinedRegionInfo {
+public:
+  CGOpenMPInnerExprInfo(CodeGenFunction &CGF, const CapturedStmt &CS)
+      : CGOpenMPInlinedRegionInfo(CGF.CapturedStmtInfo, EmptyCodeGen,
+                                  OMPD_unknown,
+                                  /*HasCancel=*/false),
+        PrivScope(CGF) {
+    // Make sure the globals captured in the provided statement are local by
+    // using the privatization logic. We assume the same variable is not
+    // captured more than once.
+    for (auto &C : CS.captures()) {
+      if (!C.capturesVariable() && !C.capturesVariableByCopy())
+        continue;
+
+      const VarDecl *VD = C.getCapturedVar();
+      if (VD->isLocalVarDeclOrParm())
+        continue;
+
+      DeclRefExpr DRE(const_cast<VarDecl *>(VD),
+                      /*RefersToEnclosingVariableOrCapture=*/false,
+                      VD->getType().getNonReferenceType(), VK_LValue,
+                      SourceLocation());
+      PrivScope.addPrivate(VD, [&CGF, &DRE]() -> Address {
+        return CGF.EmitLValue(&DRE).getAddress();
+      });
+    }
+    (void)PrivScope.Privatize();
+  }
+
+  /// \brief Lookup the captured field decl for a variable.
+  const FieldDecl *lookup(const VarDecl *VD) const override {
+    if (auto *FD = CGOpenMPInlinedRegionInfo::lookup(VD))
+      return FD;
+    return nullptr;
+  }
+
+  /// \brief Emit the captured statement body.
+  void EmitBody(CodeGenFunction &CGF, const Stmt *S) override {
+    llvm_unreachable("No body for expressions");
+  }
+
+  /// \brief Get a variable or parameter for storing global thread id
+  /// inside OpenMP construct.
+  const VarDecl *getThreadIDVariable() const override {
+    llvm_unreachable("No thread id for expressions");
+  }
+
+  /// \brief Get the name of the capture helper.
+  StringRef getHelperName() const override {
+    llvm_unreachable("No helper name for expressions");
+  }
+
+  static bool classof(const CGCapturedStmtInfo *Info) { return false; }
+
+private:
+  /// Private scope to capture global variables.
+  CodeGenFunction::OMPPrivateScope PrivScope;
+};
+
 /// \brief RAII for emitting code of OpenMP constructs.
 class InlinedOpenMPRegionRAII {
   CodeGenFunction &CGF;
+  llvm::DenseMap<const VarDecl *, FieldDecl *> LambdaCaptureFields;
+  FieldDecl *LambdaThisCaptureField = nullptr;
 
 public:
   /// \brief Constructs region for combined constructs.
@@ -260,30 +398,306 @@ public:
     // Start emission for the construct.
     CGF.CapturedStmtInfo = new CGOpenMPInlinedRegionInfo(
         CGF.CapturedStmtInfo, CodeGen, Kind, HasCancel);
+    std::swap(CGF.LambdaCaptureFields, LambdaCaptureFields);
+    LambdaThisCaptureField = CGF.LambdaThisCaptureField;
+    CGF.LambdaThisCaptureField = nullptr;
   }
+
   ~InlinedOpenMPRegionRAII() {
     // Restore original CapturedStmtInfo only if we're done with code emission.
     auto *OldCSI =
         cast<CGOpenMPInlinedRegionInfo>(CGF.CapturedStmtInfo)->getOldCSI();
     delete CGF.CapturedStmtInfo;
     CGF.CapturedStmtInfo = OldCSI;
+    std::swap(CGF.LambdaCaptureFields, LambdaCaptureFields);
+    CGF.LambdaThisCaptureField = LambdaThisCaptureField;
+  }
+};
+
+/// \brief Values for bit flags used in the ident_t to describe the fields.
+/// All enumeric elements are named and described in accordance with the code
+/// from http://llvm.org/svn/llvm-project/openmp/trunk/runtime/src/kmp.h
+enum OpenMPLocationFlags {
+  /// \brief Use trampoline for internal microtask.
+  OMP_IDENT_IMD = 0x01,
+  /// \brief Use c-style ident structure.
+  OMP_IDENT_KMPC = 0x02,
+  /// \brief Atomic reduction option for kmpc_reduce.
+  OMP_ATOMIC_REDUCE = 0x10,
+  /// \brief Explicit 'barrier' directive.
+  OMP_IDENT_BARRIER_EXPL = 0x20,
+  /// \brief Implicit barrier in code.
+  OMP_IDENT_BARRIER_IMPL = 0x40,
+  /// \brief Implicit barrier in 'for' directive.
+  OMP_IDENT_BARRIER_IMPL_FOR = 0x40,
+  /// \brief Implicit barrier in 'sections' directive.
+  OMP_IDENT_BARRIER_IMPL_SECTIONS = 0xC0,
+  /// \brief Implicit barrier in 'single' directive.
+  OMP_IDENT_BARRIER_IMPL_SINGLE = 0x140
+};
+
+/// \brief Describes ident structure that describes a source location.
+/// All descriptions are taken from
+/// http://llvm.org/svn/llvm-project/openmp/trunk/runtime/src/kmp.h
+/// Original structure:
+/// typedef struct ident {
+///    kmp_int32 reserved_1;   /**<  might be used in Fortran;
+///                                  see above  */
+///    kmp_int32 flags;        /**<  also f.flags; KMP_IDENT_xxx flags;
+///                                  KMP_IDENT_KMPC identifies this union
+///                                  member  */
+///    kmp_int32 reserved_2;   /**<  not really used in Fortran any more;
+///                                  see above */
+///#if USE_ITT_BUILD
+///                            /*  but currently used for storing
+///                                region-specific ITT */
+///                            /*  contextual information. */
+///#endif /* USE_ITT_BUILD */
+///    kmp_int32 reserved_3;   /**< source[4] in Fortran, do not use for
+///                                 C++  */
+///    char const *psource;    /**< String describing the source location.
+///                            The string is composed of semi-colon separated
+//                             fields which describe the source file,
+///                            the function and a pair of line numbers that
+///                            delimit the construct.
+///                             */
+/// } ident_t;
+enum IdentFieldIndex {
+  /// \brief might be used in Fortran
+  IdentField_Reserved_1,
+  /// \brief OMP_IDENT_xxx flags; OMP_IDENT_KMPC identifies this union member.
+  IdentField_Flags,
+  /// \brief Not really used in Fortran any more
+  IdentField_Reserved_2,
+  /// \brief Source[4] in Fortran, do not use for C++
+  IdentField_Reserved_3,
+  /// \brief String describing the source location. The string is composed of
+  /// semi-colon separated fields which describe the source file, the function
+  /// and a pair of line numbers that delimit the construct.
+  IdentField_PSource
+};
+
+/// \brief Schedule types for 'omp for' loops (these enumerators are taken from
+/// the enum sched_type in kmp.h).
+enum OpenMPSchedType {
+  /// \brief Lower bound for default (unordered) versions.
+  OMP_sch_lower = 32,
+  OMP_sch_static_chunked = 33,
+  OMP_sch_static = 34,
+  OMP_sch_dynamic_chunked = 35,
+  OMP_sch_guided_chunked = 36,
+  OMP_sch_runtime = 37,
+  OMP_sch_auto = 38,
+  /// static with chunk adjustment (e.g., simd)
+  OMP_sch_static_balanced_chunked   = 45,
+  /// \brief Lower bound for 'ordered' versions.
+  OMP_ord_lower = 64,
+  OMP_ord_static_chunked = 65,
+  OMP_ord_static = 66,
+  OMP_ord_dynamic_chunked = 67,
+  OMP_ord_guided_chunked = 68,
+  OMP_ord_runtime = 69,
+  OMP_ord_auto = 70,
+  OMP_sch_default = OMP_sch_static,
+  /// \brief dist_schedule types
+  OMP_dist_sch_static_chunked = 91,
+  OMP_dist_sch_static = 92,
+  /// Support for OpenMP 4.5 monotonic and nonmonotonic schedule modifiers.
+  /// Set if the monotonic schedule modifier was present.
+  OMP_sch_modifier_monotonic = (1 << 29),
+  /// Set if the nonmonotonic schedule modifier was present.
+  OMP_sch_modifier_nonmonotonic = (1 << 30),
+};
+
+enum OpenMPRTLFunction {
+  /// \brief Call to void __kmpc_fork_call(ident_t *loc, kmp_int32 argc,
+  /// kmpc_micro microtask, ...);
+  OMPRTL__kmpc_fork_call,
+  /// \brief Call to void *__kmpc_threadprivate_cached(ident_t *loc,
+  /// kmp_int32 global_tid, void *data, size_t size, void ***cache);
+  OMPRTL__kmpc_threadprivate_cached,
+  /// \brief Call to void __kmpc_threadprivate_register( ident_t *,
+  /// void *data, kmpc_ctor ctor, kmpc_cctor cctor, kmpc_dtor dtor);
+  OMPRTL__kmpc_threadprivate_register,
+  // Call to __kmpc_int32 kmpc_global_thread_num(ident_t *loc);
+  OMPRTL__kmpc_global_thread_num,
+  // Call to void __kmpc_critical(ident_t *loc, kmp_int32 global_tid,
+  // kmp_critical_name *crit);
+  OMPRTL__kmpc_critical,
+  // Call to void __kmpc_critical_with_hint(ident_t *loc, kmp_int32
+  // global_tid, kmp_critical_name *crit, uintptr_t hint);
+  OMPRTL__kmpc_critical_with_hint,
+  // Call to void __kmpc_end_critical(ident_t *loc, kmp_int32 global_tid,
+  // kmp_critical_name *crit);
+  OMPRTL__kmpc_end_critical,
+  // Call to kmp_int32 __kmpc_cancel_barrier(ident_t *loc, kmp_int32
+  // global_tid);
+  OMPRTL__kmpc_cancel_barrier,
+  // Call to void __kmpc_barrier(ident_t *loc, kmp_int32 global_tid);
+  OMPRTL__kmpc_barrier,
+  // Call to void __kmpc_for_static_fini(ident_t *loc, kmp_int32 global_tid);
+  OMPRTL__kmpc_for_static_fini,
+  // Call to void __kmpc_serialized_parallel(ident_t *loc, kmp_int32
+  // global_tid);
+  OMPRTL__kmpc_serialized_parallel,
+  // Call to void __kmpc_end_serialized_parallel(ident_t *loc, kmp_int32
+  // global_tid);
+  OMPRTL__kmpc_end_serialized_parallel,
+  // Call to void __kmpc_push_num_threads(ident_t *loc, kmp_int32 global_tid,
+  // kmp_int32 num_threads);
+  OMPRTL__kmpc_push_num_threads,
+  // Call to void __kmpc_flush(ident_t *loc);
+  OMPRTL__kmpc_flush,
+  // Call to kmp_int32 __kmpc_master(ident_t *, kmp_int32 global_tid);
+  OMPRTL__kmpc_master,
+  // Call to void __kmpc_end_master(ident_t *, kmp_int32 global_tid);
+  OMPRTL__kmpc_end_master,
+  // Call to kmp_int32 __kmpc_omp_taskyield(ident_t *, kmp_int32 global_tid,
+  // int end_part);
+  OMPRTL__kmpc_omp_taskyield,
+  // Call to kmp_int32 __kmpc_single(ident_t *, kmp_int32 global_tid);
+  OMPRTL__kmpc_single,
+  // Call to void __kmpc_end_single(ident_t *, kmp_int32 global_tid);
+  OMPRTL__kmpc_end_single,
+  // Call to kmp_task_t * __kmpc_omp_task_alloc(ident_t *, kmp_int32 gtid,
+  // kmp_int32 flags, size_t sizeof_kmp_task_t, size_t sizeof_shareds,
+  // kmp_routine_entry_t *task_entry);
+  OMPRTL__kmpc_omp_task_alloc,
+  // Call to kmp_int32 __kmpc_omp_task(ident_t *, kmp_int32 gtid, kmp_task_t *
+  // new_task);
+  OMPRTL__kmpc_omp_task,
+  // Call to void __kmpc_copyprivate(ident_t *loc, kmp_int32 global_tid,
+  // size_t cpy_size, void *cpy_data, void(*cpy_func)(void *, void *),
+  // kmp_int32 didit);
+  OMPRTL__kmpc_copyprivate,
+  // Call to kmp_int32 __kmpc_reduce(ident_t *loc, kmp_int32 global_tid,
+  // kmp_int32 num_vars, size_t reduce_size, void *reduce_data, void
+  // (*reduce_func)(void *lhs_data, void *rhs_data), kmp_critical_name *lck);
+  OMPRTL__kmpc_reduce,
+  // Call to kmp_int32 __kmpc_reduce_nowait(ident_t *loc, kmp_int32
+  // global_tid, kmp_int32 num_vars, size_t reduce_size, void *reduce_data,
+  // void (*reduce_func)(void *lhs_data, void *rhs_data), kmp_critical_name
+  // *lck);
+  OMPRTL__kmpc_reduce_nowait,
+  // Call to void __kmpc_end_reduce(ident_t *loc, kmp_int32 global_tid,
+  // kmp_critical_name *lck);
+  OMPRTL__kmpc_end_reduce,
+  // Call to void __kmpc_end_reduce_nowait(ident_t *loc, kmp_int32 global_tid,
+  // kmp_critical_name *lck);
+  OMPRTL__kmpc_end_reduce_nowait,
+  // Call to void __kmpc_omp_task_begin_if0(ident_t *, kmp_int32 gtid,
+  // kmp_task_t * new_task);
+  OMPRTL__kmpc_omp_task_begin_if0,
+  // Call to void __kmpc_omp_task_complete_if0(ident_t *, kmp_int32 gtid,
+  // kmp_task_t * new_task);
+  OMPRTL__kmpc_omp_task_complete_if0,
+  // Call to void __kmpc_ordered(ident_t *loc, kmp_int32 global_tid);
+  OMPRTL__kmpc_ordered,
+  // Call to void __kmpc_end_ordered(ident_t *loc, kmp_int32 global_tid);
+  OMPRTL__kmpc_end_ordered,
+  // Call to kmp_int32 __kmpc_omp_taskwait(ident_t *loc, kmp_int32
+  // global_tid);
+  OMPRTL__kmpc_omp_taskwait,
+  // Call to void __kmpc_taskgroup(ident_t *loc, kmp_int32 global_tid);
+  OMPRTL__kmpc_taskgroup,
+  // Call to void __kmpc_end_taskgroup(ident_t *loc, kmp_int32 global_tid);
+  OMPRTL__kmpc_end_taskgroup,
+  // Call to void __kmpc_push_proc_bind(ident_t *loc, kmp_int32 global_tid,
+  // int proc_bind);
+  OMPRTL__kmpc_push_proc_bind,
+  // Call to kmp_int32 __kmpc_omp_task_with_deps(ident_t *loc_ref, kmp_int32
+  // gtid, kmp_task_t * new_task, kmp_int32 ndeps, kmp_depend_info_t
+  // *dep_list, kmp_int32 ndeps_noalias, kmp_depend_info_t *noalias_dep_list);
+  OMPRTL__kmpc_omp_task_with_deps,
+  // Call to void __kmpc_omp_wait_deps(ident_t *loc_ref, kmp_int32
+  // gtid, kmp_int32 ndeps, kmp_depend_info_t *dep_list, kmp_int32
+  // ndeps_noalias, kmp_depend_info_t *noalias_dep_list);
+  OMPRTL__kmpc_omp_wait_deps,
+  // Call to kmp_int32 __kmpc_cancellationpoint(ident_t *loc, kmp_int32
+  // global_tid, kmp_int32 cncl_kind);
+  OMPRTL__kmpc_cancellationpoint,
+  // Call to kmp_int32 __kmpc_cancel(ident_t *loc, kmp_int32 global_tid,
+  // kmp_int32 cncl_kind);
+  OMPRTL__kmpc_cancel,
+  // Call to void __kmpc_push_num_teams(ident_t *loc, kmp_int32 global_tid,
+  // kmp_int32 num_teams, kmp_int32 thread_limit);
+  OMPRTL__kmpc_push_num_teams,
+  // Call to void __kmpc_fork_teams(ident_t *loc, kmp_int32 argc, kmpc_micro
+  // microtask, ...);
+  OMPRTL__kmpc_fork_teams,
+  // Call to void __kmpc_taskloop(ident_t *loc, int gtid, kmp_task_t *task, int
+  // if_val, kmp_uint64 *lb, kmp_uint64 *ub, kmp_int64 st, int nogroup, int
+  // sched, kmp_uint64 grainsize, void *task_dup);
+  OMPRTL__kmpc_taskloop,
+  // Call to void __kmpc_doacross_init(ident_t *loc, kmp_int32 gtid, kmp_int32
+  // num_dims, struct kmp_dim *dims);
+  OMPRTL__kmpc_doacross_init,
+  // Call to void __kmpc_doacross_fini(ident_t *loc, kmp_int32 gtid);
+  OMPRTL__kmpc_doacross_fini,
+  // Call to void __kmpc_doacross_post(ident_t *loc, kmp_int32 gtid, kmp_int64
+  // *vec);
+  OMPRTL__kmpc_doacross_post,
+  // Call to void __kmpc_doacross_wait(ident_t *loc, kmp_int32 gtid, kmp_int64
+  // *vec);
+  OMPRTL__kmpc_doacross_wait,
+
+  //
+  // Offloading related calls
+  //
+  // Call to int32_t __tgt_target(int32_t device_id, void *host_ptr, int32_t
+  // arg_num, void** args_base, void **args, size_t *arg_sizes, int32_t
+  // *arg_types);
+  OMPRTL__tgt_target,
+  // Call to int32_t __tgt_target_teams(int32_t device_id, void *host_ptr,
+  // int32_t arg_num, void** args_base, void **args, size_t *arg_sizes,
+  // int32_t *arg_types, int32_t num_teams, int32_t thread_limit);
+  OMPRTL__tgt_target_teams,
+  // Call to void __tgt_register_lib(__tgt_bin_desc *desc);
+  OMPRTL__tgt_register_lib,
+  // Call to void __tgt_unregister_lib(__tgt_bin_desc *desc);
+  OMPRTL__tgt_unregister_lib,
+  // Call to void __tgt_target_data_begin(int32_t device_id, int32_t arg_num,
+  // void** args_base, void **args, size_t *arg_sizes, int32_t *arg_types);
+  OMPRTL__tgt_target_data_begin,
+  // Call to void __tgt_target_data_end(int32_t device_id, int32_t arg_num,
+  // void** args_base, void **args, size_t *arg_sizes, int32_t *arg_types);
+  OMPRTL__tgt_target_data_end,
+  // Call to void __tgt_target_data_update(int32_t device_id, int32_t arg_num,
+  // void** args_base, void **args, size_t *arg_sizes, int32_t *arg_types);
+  OMPRTL__tgt_target_data_update,
+};
+
+/// A basic class for pre|post-action for advanced codegen sequence for OpenMP
+/// region.
+class CleanupTy final : public EHScopeStack::Cleanup {
+  PrePostActionTy *Action;
+
+public:
+  explicit CleanupTy(PrePostActionTy *Action) : Action(Action) {}
+  void Emit(CodeGenFunction &CGF, Flags /*flags*/) override {
+    if (!CGF.HaveInsertPoint())
+      return;
+    Action->Exit(CGF);
   }
 };
 
 } // anonymous namespace
 
-static LValue emitLoadOfPointerLValue(CodeGenFunction &CGF, Address PtrAddr,
-                                      QualType Ty) {
-  AlignmentSource Source;
-  CharUnits Align = CGF.getNaturalPointeeTypeAlignment(Ty, &Source);
-  return CGF.MakeAddrLValue(Address(CGF.Builder.CreateLoad(PtrAddr), Align),
-                            Ty->getPointeeType(), Source);
+void RegionCodeGenTy::operator()(CodeGenFunction &CGF) const {
+  CodeGenFunction::RunCleanupsScope Scope(CGF);
+  if (PrePostAction) {
+    CGF.EHStack.pushCleanup<CleanupTy>(NormalAndEHCleanup, PrePostAction);
+    Callback(CodeGen, CGF, *PrePostAction);
+  } else {
+    PrePostActionTy Action;
+    Callback(CodeGen, CGF, Action);
+  }
 }
 
 LValue CGOpenMPRegionInfo::getThreadIDVariableLValue(CodeGenFunction &CGF) {
-  return emitLoadOfPointerLValue(CGF,
-                                 CGF.GetAddrOfLocalVar(getThreadIDVariable()),
-                                 getThreadIDVariable()->getType());
+  return CGF.EmitLoadOfPointerLValue(
+      CGF.GetAddrOfLocalVar(getThreadIDVariable()),
+      getThreadIDVariable()->getType()->castAs<PointerType>());
 }
 
 void CGOpenMPRegionInfo::EmitBody(CodeGenFunction &CGF, const Stmt * /*S*/) {
@@ -295,10 +709,7 @@ void CGOpenMPRegionInfo::EmitBody(CodeGenFunction &CGF, const Stmt * /*S*/) {
   // The point of exit cannot be a branch out of the structured block.
   // longjmp() and throw() must not violate the entry/exit criteria.
   CGF.EHStack.pushTerminate();
-  {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    CodeGen(CGF);
-  }
+  CodeGen(CGF);
   CGF.EHStack.popTerminate();
 }
 
@@ -310,16 +721,11 @@ LValue CGOpenMPTaskOutlinedRegionInfo::getThreadIDVariableLValue(
 }
 
 CGOpenMPRuntime::CGOpenMPRuntime(CodeGenModule &CGM)
-    : CGM(CGM), DefaultOpenMPPSource(nullptr), KmpRoutineEntryPtrTy(nullptr),
-      OffloadEntriesInfoManager(CGM) {
+    : CGM(CGM), OffloadEntriesInfoManager(CGM) {
   IdentTy = llvm::StructType::create(
       "ident_t", CGM.Int32Ty /* reserved_1 */, CGM.Int32Ty /* flags */,
       CGM.Int32Ty /* reserved_2 */, CGM.Int32Ty /* reserved_3 */,
       CGM.Int8PtrTy /* psource */, nullptr);
-  // Build void (*kmpc_micro)(kmp_int32 *global_tid, kmp_int32 *bound_tid,...)
-  llvm::Type *MicroParams[] = {llvm::PointerType::getUnqual(CGM.Int32Ty),
-                               llvm::PointerType::getUnqual(CGM.Int32Ty)};
-  Kmpc_MicroTy = llvm::FunctionType::get(CGM.VoidTy, MicroParams, true);
   KmpCriticalNameTy = llvm::ArrayType::get(CGM.Int32Ty, /*NumElements*/ 8);
 
   loadOffloadInfoMetadata();
@@ -329,6 +735,90 @@ void CGOpenMPRuntime::clear() {
   InternalVars.clear();
 }
 
+static llvm::Function *
+emitCombinerOrInitializer(CodeGenModule &CGM, QualType Ty,
+                          const Expr *CombinerInitializer, const VarDecl *In,
+                          const VarDecl *Out, bool IsCombiner) {
+  // void .omp_combiner.(Ty *in, Ty *out);
+  auto &C = CGM.getContext();
+  QualType PtrTy = C.getPointerType(Ty).withRestrict();
+  FunctionArgList Args;
+  ImplicitParamDecl OmpOutParm(C, /*DC=*/nullptr, Out->getLocation(),
+                               /*Id=*/nullptr, PtrTy);
+  ImplicitParamDecl OmpInParm(C, /*DC=*/nullptr, In->getLocation(),
+                              /*Id=*/nullptr, PtrTy);
+  Args.push_back(&OmpOutParm);
+  Args.push_back(&OmpInParm);
+  auto &FnInfo =
+      CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
+  auto *FnTy = CGM.getTypes().GetFunctionType(FnInfo);
+  auto *Fn = llvm::Function::Create(
+      FnTy, llvm::GlobalValue::InternalLinkage,
+      IsCombiner ? ".omp_combiner." : ".omp_initializer.", &CGM.getModule());
+  CGM.SetInternalFunctionAttributes(/*D=*/nullptr, Fn, FnInfo);
+  Fn->addFnAttr(llvm::Attribute::AlwaysInline);
+  CodeGenFunction CGF(CGM);
+  // Map "T omp_in;" variable to "*omp_in_parm" value in all expressions.
+  // Map "T omp_out;" variable to "*omp_out_parm" value in all expressions.
+  CGF.StartFunction(GlobalDecl(), C.VoidTy, Fn, FnInfo, Args);
+  CodeGenFunction::OMPPrivateScope Scope(CGF);
+  Address AddrIn = CGF.GetAddrOfLocalVar(&OmpInParm);
+  Scope.addPrivate(In, [&CGF, AddrIn, PtrTy]() -> Address {
+    return CGF.EmitLoadOfPointerLValue(AddrIn, PtrTy->castAs<PointerType>())
+        .getAddress();
+  });
+  Address AddrOut = CGF.GetAddrOfLocalVar(&OmpOutParm);
+  Scope.addPrivate(Out, [&CGF, AddrOut, PtrTy]() -> Address {
+    return CGF.EmitLoadOfPointerLValue(AddrOut, PtrTy->castAs<PointerType>())
+        .getAddress();
+  });
+  (void)Scope.Privatize();
+  CGF.EmitIgnoredExpr(CombinerInitializer);
+  Scope.ForceCleanup();
+  CGF.FinishFunction();
+  return Fn;
+}
+
+void CGOpenMPRuntime::emitUserDefinedReduction(
+    CodeGenFunction *CGF, const OMPDeclareReductionDecl *D) {
+  if (UDRMap.count(D) > 0)
+    return;
+  auto &C = CGM.getContext();
+  if (!In || !Out) {
+    In = &C.Idents.get("omp_in");
+    Out = &C.Idents.get("omp_out");
+  }
+  llvm::Function *Combiner = emitCombinerOrInitializer(
+      CGM, D->getType(), D->getCombiner(), cast<VarDecl>(D->lookup(In).front()),
+      cast<VarDecl>(D->lookup(Out).front()),
+      /*IsCombiner=*/true);
+  llvm::Function *Initializer = nullptr;
+  if (auto *Init = D->getInitializer()) {
+    if (!Priv || !Orig) {
+      Priv = &C.Idents.get("omp_priv");
+      Orig = &C.Idents.get("omp_orig");
+    }
+    Initializer = emitCombinerOrInitializer(
+        CGM, D->getType(), Init, cast<VarDecl>(D->lookup(Orig).front()),
+        cast<VarDecl>(D->lookup(Priv).front()),
+        /*IsCombiner=*/false);
+  }
+  UDRMap.insert(std::make_pair(D, std::make_pair(Combiner, Initializer)));
+  if (CGF) {
+    auto &Decls = FunctionUDRMap.FindAndConstruct(CGF->CurFn);
+    Decls.second.push_back(D);
+  }
+}
+
+std::pair<llvm::Function *, llvm::Function *>
+CGOpenMPRuntime::getUserDefinedReduction(const OMPDeclareReductionDecl *D) {
+  auto I = UDRMap.find(D);
+  if (I != UDRMap.end())
+    return I->second;
+  emitUserDefinedReduction(/*CGF=*/nullptr, D);
+  return UDRMap.lookup(D);
+}
+
 // Layout information for ident_t.
 static CharUnits getIdentAlign(CodeGenModule &CGM) {
   return CGM.getPointerAlign();
@@ -337,18 +827,18 @@ static CharUnits getIdentSize(CodeGenModule &CGM) {
   assert((4 * CGM.getPointerSize()).isMultipleOf(CGM.getPointerAlign()));
   return CharUnits::fromQuantity(16) + CGM.getPointerSize();
 }
-static CharUnits getOffsetOfIdentField(CGOpenMPRuntime::IdentFieldIndex Field) {
+static CharUnits getOffsetOfIdentField(IdentFieldIndex Field) {
   // All the fields except the last are i32, so this works beautifully.
   return unsigned(Field) * CharUnits::fromQuantity(4);
 }
 static Address createIdentFieldGEP(CodeGenFunction &CGF, Address Addr,
-                                   CGOpenMPRuntime::IdentFieldIndex Field,
+                                   IdentFieldIndex Field,
                                    const llvm::Twine &Name = "") {
   auto Offset = getOffsetOfIdentField(Field);
   return CGF.Builder.CreateStructGEP(Addr, Field, Offset, Name);
 }
 
-llvm::Value *CGOpenMPRuntime::emitParallelOutlinedFunction(
+llvm::Value *CGOpenMPRuntime::emitParallelOrTeamsOutlinedFunction(
     const OMPExecutableDirective &D, const VarDecl *ThreadIDVar,
     OpenMPDirectiveKind InnermostKind, const RegionCodeGenTy &CodeGen) {
   assert(ThreadIDVar->getType()->isPointerType() &&
@@ -370,19 +860,39 @@ llvm::Value *CGOpenMPRuntime::emitParallelOutlinedFunction(
 
 llvm::Value *CGOpenMPRuntime::emitTaskOutlinedFunction(
     const OMPExecutableDirective &D, const VarDecl *ThreadIDVar,
-    OpenMPDirectiveKind InnermostKind, const RegionCodeGenTy &CodeGen) {
+    const VarDecl *PartIDVar, const VarDecl *TaskTVar,
+    OpenMPDirectiveKind InnermostKind, const RegionCodeGenTy &CodeGen,
+    bool Tied, unsigned &NumberOfParts) {
+  auto &&UntiedCodeGen = [this, &D, TaskTVar](CodeGenFunction &CGF,
+                                              PrePostActionTy &) {
+    auto *ThreadID = getThreadID(CGF, D.getLocStart());
+    auto *UpLoc = emitUpdateLocation(CGF, D.getLocStart());
+    llvm::Value *TaskArgs[] = {
+        UpLoc, ThreadID,
+        CGF.EmitLoadOfPointerLValue(CGF.GetAddrOfLocalVar(TaskTVar),
+                                    TaskTVar->getType()->castAs<PointerType>())
+            .getPointer()};
+    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_task), TaskArgs);
+  };
+  CGOpenMPTaskOutlinedRegionInfo::UntiedTaskActionTy Action(Tied, PartIDVar,
+                                                            UntiedCodeGen);
+  CodeGen.setAction(Action);
   assert(!ThreadIDVar->getType()->isPointerType() &&
          "thread id variable must be of type kmp_int32 for tasks");
   auto *CS = cast<CapturedStmt>(D.getAssociatedStmt());
+  auto *TD = dyn_cast<OMPTaskDirective>(&D);
   CodeGenFunction CGF(CGM, true);
   CGOpenMPTaskOutlinedRegionInfo CGInfo(*CS, ThreadIDVar, CodeGen,
                                         InnermostKind,
-                                        cast<OMPTaskDirective>(D).hasCancel());
+                                        TD ? TD->hasCancel() : false, Action);
   CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
-  return CGF.GenerateCapturedStmtFunction(*CS);
+  auto *Res = CGF.GenerateCapturedStmtFunction(*CS);
+  if (!Tied)
+    NumberOfParts = Action.getNumberOfParts();
+  return Res;
 }
 
-Address CGOpenMPRuntime::getOrCreateDefaultLocation(OpenMPLocationFlags Flags) {
+Address CGOpenMPRuntime::getOrCreateDefaultLocation(unsigned Flags) {
   CharUnits Align = getIdentAlign(CGM);
   llvm::Value *Entry = OpenMPDefaultLocMap.lookup(Flags);
   if (!Entry) {
@@ -399,7 +909,7 @@ Address CGOpenMPRuntime::getOrCreateDefaultLocation(OpenMPLocationFlags Flags) {
     auto DefaultOpenMPLocation = new llvm::GlobalVariable(
         CGM.getModule(), IdentTy, /*isConstant*/ true,
         llvm::GlobalValue::PrivateLinkage, /*Initializer*/ nullptr);
-    DefaultOpenMPLocation->setUnnamedAddr(true);
+    DefaultOpenMPLocation->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
     DefaultOpenMPLocation->setAlignment(Align.getQuantity());
 
     llvm::Constant *Zero = llvm::ConstantInt::get(CGM.Int32Ty, 0, true);
@@ -415,9 +925,10 @@ Address CGOpenMPRuntime::getOrCreateDefaultLocation(OpenMPLocationFlags Flags) {
 
 llvm::Value *CGOpenMPRuntime::emitUpdateLocation(CodeGenFunction &CGF,
                                                  SourceLocation Loc,
-                                                 OpenMPLocationFlags Flags) {
+                                                 unsigned Flags) {
+  Flags |= OMP_IDENT_KMPC;
   // If no debug info is generated - return global default location.
-  if (CGM.getCodeGenOpts().getDebugInfo() == CodeGenOptions::NoDebugInfo ||
+  if (CGM.getCodeGenOpts().getDebugInfo() == codegenoptions::NoDebugInfo ||
       Loc.isInvalid())
     return getOrCreateDefaultLocation(Flags).getPointer();
 
@@ -517,20 +1028,34 @@ void CGOpenMPRuntime::functionFinished(CodeGenFunction &CGF) {
   assert(CGF.CurFn && "No function in current CodeGenFunction.");
   if (OpenMPLocThreadIDMap.count(CGF.CurFn))
     OpenMPLocThreadIDMap.erase(CGF.CurFn);
+  if (FunctionUDRMap.count(CGF.CurFn) > 0) {
+    for(auto *D : FunctionUDRMap[CGF.CurFn]) {
+      UDRMap.erase(D);
+    }
+    FunctionUDRMap.erase(CGF.CurFn);
+  }
 }
 
 llvm::Type *CGOpenMPRuntime::getIdentTyPointerTy() {
+  if (!IdentTy) {
+  }
   return llvm::PointerType::getUnqual(IdentTy);
 }
 
 llvm::Type *CGOpenMPRuntime::getKmpc_MicroPointerTy() {
+  if (!Kmpc_MicroTy) {
+    // Build void (*kmpc_micro)(kmp_int32 *global_tid, kmp_int32 *bound_tid,...)
+    llvm::Type *MicroParams[] = {llvm::PointerType::getUnqual(CGM.Int32Ty),
+                                 llvm::PointerType::getUnqual(CGM.Int32Ty)};
+    Kmpc_MicroTy = llvm::FunctionType::get(CGM.VoidTy, MicroParams, true);
+  }
   return llvm::PointerType::getUnqual(Kmpc_MicroTy);
 }
 
 llvm::Constant *
-CGOpenMPRuntime::createRuntimeFunction(OpenMPRTLFunction Function) {
+CGOpenMPRuntime::createRuntimeFunction(unsigned Function) {
   llvm::Constant *RTLFn = nullptr;
-  switch (Function) {
+  switch (static_cast<OpenMPRTLFunction>(Function)) {
   case OMPRTL__kmpc_fork_call: {
     // Build void __kmpc_fork_call(ident_t *loc, kmp_int32 argc, kmpc_micro
     // microtask, ...);
@@ -927,6 +1452,86 @@ CGOpenMPRuntime::createRuntimeFunction(OpenMPRTLFunction Function) {
     RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_cancel");
     break;
   }
+  case OMPRTL__kmpc_push_num_teams: {
+    // Build void kmpc_push_num_teams (ident_t loc, kmp_int32 global_tid,
+    // kmp_int32 num_teams, kmp_int32 num_threads)
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty, CGM.Int32Ty,
+        CGM.Int32Ty};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.Int32Ty, TypeParams, /*isVarArg*/ false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_push_num_teams");
+    break;
+  }
+  case OMPRTL__kmpc_fork_teams: {
+    // Build void __kmpc_fork_teams(ident_t *loc, kmp_int32 argc, kmpc_micro
+    // microtask, ...);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty,
+                                getKmpc_MicroPointerTy()};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg*/ true);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_fork_teams");
+    break;
+  }
+  case OMPRTL__kmpc_taskloop: {
+    // Build void __kmpc_taskloop(ident_t *loc, int gtid, kmp_task_t *task, int
+    // if_val, kmp_uint64 *lb, kmp_uint64 *ub, kmp_int64 st, int nogroup, int
+    // sched, kmp_uint64 grainsize, void *task_dup);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(),
+                                CGM.IntTy,
+                                CGM.VoidPtrTy,
+                                CGM.IntTy,
+                                CGM.Int64Ty->getPointerTo(),
+                                CGM.Int64Ty->getPointerTo(),
+                                CGM.Int64Ty,
+                                CGM.IntTy,
+                                CGM.IntTy,
+                                CGM.Int64Ty,
+                                CGM.VoidPtrTy};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg=*/false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, /*Name=*/"__kmpc_taskloop");
+    break;
+  }
+  case OMPRTL__kmpc_doacross_init: {
+    // Build void __kmpc_doacross_init(ident_t *loc, kmp_int32 gtid, kmp_int32
+    // num_dims, struct kmp_dim *dims);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(),
+                                CGM.Int32Ty,
+                                CGM.Int32Ty,
+                                CGM.VoidPtrTy};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg=*/false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, /*Name=*/"__kmpc_doacross_init");
+    break;
+  }
+  case OMPRTL__kmpc_doacross_fini: {
+    // Build void __kmpc_doacross_fini(ident_t *loc, kmp_int32 gtid);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg=*/false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, /*Name=*/"__kmpc_doacross_fini");
+    break;
+  }
+  case OMPRTL__kmpc_doacross_post: {
+    // Build void __kmpc_doacross_post(ident_t *loc, kmp_int32 gtid, kmp_int64
+    // *vec);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty,
+                                CGM.Int64Ty->getPointerTo()};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg=*/false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, /*Name=*/"__kmpc_doacross_post");
+    break;
+  }
+  case OMPRTL__kmpc_doacross_wait: {
+    // Build void __kmpc_doacross_wait(ident_t *loc, kmp_int32 gtid, kmp_int64
+    // *vec);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty,
+                                CGM.Int64Ty->getPointerTo()};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg=*/false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, /*Name=*/"__kmpc_doacross_wait");
+    break;
+  }
   case OMPRTL__tgt_target: {
     // Build int32_t __tgt_target(int32_t device_id, void *host_ptr, int32_t
     // arg_num, void** args_base, void **args, size_t *arg_sizes, int32_t
@@ -943,6 +1548,24 @@ CGOpenMPRuntime::createRuntimeFunction(OpenMPRTLFunction Function) {
     RTLFn = CGM.CreateRuntimeFunction(FnTy, "__tgt_target");
     break;
   }
+  case OMPRTL__tgt_target_teams: {
+    // Build int32_t __tgt_target_teams(int32_t device_id, void *host_ptr,
+    // int32_t arg_num, void** args_base, void **args, size_t *arg_sizes,
+    // int32_t *arg_types, int32_t num_teams, int32_t thread_limit);
+    llvm::Type *TypeParams[] = {CGM.Int32Ty,
+                                CGM.VoidPtrTy,
+                                CGM.Int32Ty,
+                                CGM.VoidPtrPtrTy,
+                                CGM.VoidPtrPtrTy,
+                                CGM.SizeTy->getPointerTo(),
+                                CGM.Int32Ty->getPointerTo(),
+                                CGM.Int32Ty,
+                                CGM.Int32Ty};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.Int32Ty, TypeParams, /*isVarArg*/ false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__tgt_target_teams");
+    break;
+  }
   case OMPRTL__tgt_register_lib: {
     // Build void __tgt_register_lib(__tgt_bin_desc *desc);
     QualType ParamTy =
@@ -963,30 +1586,53 @@ CGOpenMPRuntime::createRuntimeFunction(OpenMPRTLFunction Function) {
     RTLFn = CGM.CreateRuntimeFunction(FnTy, "__tgt_unregister_lib");
     break;
   }
+  case OMPRTL__tgt_target_data_begin: {
+    // Build void __tgt_target_data_begin(int32_t device_id, int32_t arg_num,
+    // void** args_base, void **args, size_t *arg_sizes, int32_t *arg_types);
+    llvm::Type *TypeParams[] = {CGM.Int32Ty,
+                                CGM.Int32Ty,
+                                CGM.VoidPtrPtrTy,
+                                CGM.VoidPtrPtrTy,
+                                CGM.SizeTy->getPointerTo(),
+                                CGM.Int32Ty->getPointerTo()};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg*/ false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__tgt_target_data_begin");
+    break;
+  }
+  case OMPRTL__tgt_target_data_end: {
+    // Build void __tgt_target_data_end(int32_t device_id, int32_t arg_num,
+    // void** args_base, void **args, size_t *arg_sizes, int32_t *arg_types);
+    llvm::Type *TypeParams[] = {CGM.Int32Ty,
+                                CGM.Int32Ty,
+                                CGM.VoidPtrPtrTy,
+                                CGM.VoidPtrPtrTy,
+                                CGM.SizeTy->getPointerTo(),
+                                CGM.Int32Ty->getPointerTo()};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg*/ false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__tgt_target_data_end");
+    break;
+  }
+  case OMPRTL__tgt_target_data_update: {
+    // Build void __tgt_target_data_update(int32_t device_id, int32_t arg_num,
+    // void** args_base, void **args, size_t *arg_sizes, int32_t *arg_types);
+    llvm::Type *TypeParams[] = {CGM.Int32Ty,
+                                CGM.Int32Ty,
+                                CGM.VoidPtrPtrTy,
+                                CGM.VoidPtrPtrTy,
+                                CGM.SizeTy->getPointerTo(),
+                                CGM.Int32Ty->getPointerTo()};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg*/ false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__tgt_target_data_update");
+    break;
+  }
   }
+  assert(RTLFn && "Unable to find OpenMP runtime function");
   return RTLFn;
 }
 
-static llvm::Value *getTypeSize(CodeGenFunction &CGF, QualType Ty) {
-  auto &C = CGF.getContext();
-  llvm::Value *Size = nullptr;
-  auto SizeInChars = C.getTypeSizeInChars(Ty);
-  if (SizeInChars.isZero()) {
-    // getTypeSizeInChars() returns 0 for a VLA.
-    while (auto *VAT = C.getAsVariableArrayType(Ty)) {
-      llvm::Value *ArraySize;
-      std::tie(ArraySize, Ty) = CGF.getVLASize(VAT);
-      Size = Size ? CGF.Builder.CreateNUWMul(Size, ArraySize) : ArraySize;
-    }
-    SizeInChars = C.getTypeSizeInChars(Ty);
-    assert(!SizeInChars.isZero());
-    Size = CGF.Builder.CreateNUWMul(
-        Size, llvm::ConstantInt::get(CGF.SizeTy, SizeInChars.getQuantity()));
-  } else
-    Size = llvm::ConstantInt::get(CGF.SizeTy, SizeInChars.getQuantity());
-  return Size;
-}
-
 llvm::Constant *CGOpenMPRuntime::createForStaticInitFunction(unsigned IVSize,
                                                              bool IVSigned) {
   assert((IVSize == 32 || IVSize == 64) &&
@@ -1144,9 +1790,8 @@ llvm::Function *CGOpenMPRuntime::emitThreadPrivateVarDefinition(
                             /*Id=*/nullptr, CGM.getContext().VoidPtrTy);
       Args.push_back(&Dst);
 
-      auto &FI = CGM.getTypes().arrangeFreeFunctionDeclaration(
-          CGM.getContext().VoidPtrTy, Args, FunctionType::ExtInfo(),
-          /*isVariadic=*/false);
+      auto &FI = CGM.getTypes().arrangeBuiltinFunctionDeclaration(
+          CGM.getContext().VoidPtrTy, Args);
       auto FTy = CGM.getTypes().GetFunctionType(FI);
       auto Fn = CGM.CreateGlobalInitOrDestructFunction(
           FTy, ".__kmpc_global_ctor_.", FI, Loc);
@@ -1176,14 +1821,16 @@ llvm::Function *CGOpenMPRuntime::emitThreadPrivateVarDefinition(
                             /*Id=*/nullptr, CGM.getContext().VoidPtrTy);
       Args.push_back(&Dst);
 
-      auto &FI = CGM.getTypes().arrangeFreeFunctionDeclaration(
-          CGM.getContext().VoidTy, Args, FunctionType::ExtInfo(),
-          /*isVariadic=*/false);
+      auto &FI = CGM.getTypes().arrangeBuiltinFunctionDeclaration(
+          CGM.getContext().VoidTy, Args);
       auto FTy = CGM.getTypes().GetFunctionType(FI);
       auto Fn = CGM.CreateGlobalInitOrDestructFunction(
           FTy, ".__kmpc_global_dtor_.", FI, Loc);
+      auto NL = ApplyDebugLocation::CreateEmpty(DtorCGF);
       DtorCGF.StartFunction(GlobalDecl(), CGM.getContext().VoidTy, Fn, FI, Args,
                             SourceLocation());
+      // Create a scope with an artificial location for the body of this function.
+      auto AL = ApplyDebugLocation::CreateArtificial(DtorCGF);
       auto ArgVal = DtorCGF.EmitLoadOfScalar(
           DtorCGF.GetAddrOfLocalVar(&Dst),
           /*Volatile=*/false, CGM.getContext().VoidPtrTy, Dst.getLocation());
@@ -1251,12 +1898,10 @@ static void emitOMPIfClause(CodeGenFunction &CGF, const Expr *Cond,
   // the condition and the dead arm of the if/else.
   bool CondConstant;
   if (CGF.ConstantFoldsToSimpleInteger(Cond, CondConstant)) {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    if (CondConstant) {
+    if (CondConstant)
       ThenGen(CGF);
-    } else {
+    else
       ElseGen(CGF);
-    }
     return;
   }
 
@@ -1269,26 +1914,16 @@ static void emitOMPIfClause(CodeGenFunction &CGF, const Expr *Cond,
 
   // Emit the 'then' code.
   CGF.EmitBlock(ThenBlock);
-  {
-    CodeGenFunction::RunCleanupsScope ThenScope(CGF);
-    ThenGen(CGF);
-  }
+  ThenGen(CGF);
   CGF.EmitBranch(ContBlock);
   // Emit the 'else' code if present.
-  {
-    // There is no need to emit line number for unconditional branch.
-    auto NL = ApplyDebugLocation::CreateEmpty(CGF);
-    CGF.EmitBlock(ElseBlock);
-  }
-  {
-    CodeGenFunction::RunCleanupsScope ThenScope(CGF);
-    ElseGen(CGF);
-  }
-  {
-    // There is no need to emit line number for unconditional branch.
-    auto NL = ApplyDebugLocation::CreateEmpty(CGF);
-    CGF.EmitBranch(ContBlock);
-  }
+  // There is no need to emit line number for unconditional branch.
+  (void)ApplyDebugLocation::CreateEmpty(CGF);
+  CGF.EmitBlock(ElseBlock);
+  ElseGen(CGF);
+  // There is no need to emit line number for unconditional branch.
+  (void)ApplyDebugLocation::CreateEmpty(CGF);
+  CGF.EmitBranch(ContBlock);
   // Emit the continuation block for code after the if.
   CGF.EmitBlock(ContBlock, /*IsFinished=*/true);
 }
@@ -1300,34 +1935,36 @@ void CGOpenMPRuntime::emitParallelCall(CodeGenFunction &CGF, SourceLocation Loc,
   if (!CGF.HaveInsertPoint())
     return;
   auto *RTLoc = emitUpdateLocation(CGF, Loc);
-  auto &&ThenGen = [this, OutlinedFn, CapturedVars,
-                    RTLoc](CodeGenFunction &CGF) {
+  auto &&ThenGen = [OutlinedFn, CapturedVars, RTLoc](CodeGenFunction &CGF,
+                                                     PrePostActionTy &) {
     // Build call __kmpc_fork_call(loc, n, microtask, var1, .., varn);
+    auto &RT = CGF.CGM.getOpenMPRuntime();
     llvm::Value *Args[] = {
         RTLoc,
         CGF.Builder.getInt32(CapturedVars.size()), // Number of captured vars
-        CGF.Builder.CreateBitCast(OutlinedFn, getKmpc_MicroPointerTy())};
+        CGF.Builder.CreateBitCast(OutlinedFn, RT.getKmpc_MicroPointerTy())};
     llvm::SmallVector<llvm::Value *, 16> RealArgs;
     RealArgs.append(std::begin(Args), std::end(Args));
     RealArgs.append(CapturedVars.begin(), CapturedVars.end());
 
-    auto RTLFn = createRuntimeFunction(OMPRTL__kmpc_fork_call);
+    auto RTLFn = RT.createRuntimeFunction(OMPRTL__kmpc_fork_call);
     CGF.EmitRuntimeCall(RTLFn, RealArgs);
   };
-  auto &&ElseGen = [this, OutlinedFn, CapturedVars, RTLoc,
-                    Loc](CodeGenFunction &CGF) {
-    auto ThreadID = getThreadID(CGF, Loc);
+  auto &&ElseGen = [OutlinedFn, CapturedVars, RTLoc, Loc](CodeGenFunction &CGF,
+                                                          PrePostActionTy &) {
+    auto &RT = CGF.CGM.getOpenMPRuntime();
+    auto ThreadID = RT.getThreadID(CGF, Loc);
     // Build calls:
     // __kmpc_serialized_parallel(&Loc, GTid);
     llvm::Value *Args[] = {RTLoc, ThreadID};
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_serialized_parallel),
-                        Args);
+    CGF.EmitRuntimeCall(
+        RT.createRuntimeFunction(OMPRTL__kmpc_serialized_parallel), Args);
 
     // OutlinedFn(&GTid, &zero, CapturedStruct);
-    auto ThreadIDAddr = emitThreadIDAddress(CGF, Loc);
+    auto ThreadIDAddr = RT.emitThreadIDAddress(CGF, Loc);
     Address ZeroAddr =
-      CGF.CreateTempAlloca(CGF.Int32Ty, CharUnits::fromQuantity(4),
-                           /*Name*/ ".zero.addr");
+        CGF.CreateTempAlloca(CGF.Int32Ty, CharUnits::fromQuantity(4),
+                             /*Name*/ ".zero.addr");
     CGF.InitTempAlloca(ZeroAddr, CGF.Builder.getInt32(/*C*/ 0));
     llvm::SmallVector<llvm::Value *, 16> OutlinedFnArgs;
     OutlinedFnArgs.push_back(ThreadIDAddr.getPointer());
@@ -1336,15 +1973,16 @@ void CGOpenMPRuntime::emitParallelCall(CodeGenFunction &CGF, SourceLocation Loc,
     CGF.EmitCallOrInvoke(OutlinedFn, OutlinedFnArgs);
 
     // __kmpc_end_serialized_parallel(&Loc, GTid);
-    llvm::Value *EndArgs[] = {emitUpdateLocation(CGF, Loc), ThreadID};
+    llvm::Value *EndArgs[] = {RT.emitUpdateLocation(CGF, Loc), ThreadID};
     CGF.EmitRuntimeCall(
-        createRuntimeFunction(OMPRTL__kmpc_end_serialized_parallel), EndArgs);
+        RT.createRuntimeFunction(OMPRTL__kmpc_end_serialized_parallel),
+        EndArgs);
   };
-  if (IfCond) {
+  if (IfCond)
     emitOMPIfClause(CGF, IfCond, ThenGen, ElseGen);
-  } else {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    ThenGen(CGF);
+  else {
+    RegionCodeGenTy ThenRCG(ThenGen);
+    ThenRCG(CGF);
   }
 }
 
@@ -1397,20 +2035,39 @@ llvm::Value *CGOpenMPRuntime::getCriticalRegionLock(StringRef CriticalName) {
 }
 
 namespace {
-template <size_t N> class CallEndCleanup final : public EHScopeStack::Cleanup {
-  llvm::Value *Callee;
-  llvm::Value *Args[N];
+/// Common pre(post)-action for different OpenMP constructs.
+class CommonActionTy final : public PrePostActionTy {
+  llvm::Value *EnterCallee;
+  ArrayRef<llvm::Value *> EnterArgs;
+  llvm::Value *ExitCallee;
+  ArrayRef<llvm::Value *> ExitArgs;
+  bool Conditional;
+  llvm::BasicBlock *ContBlock = nullptr;
 
 public:
-  CallEndCleanup(llvm::Value *Callee, ArrayRef<llvm::Value *> CleanupArgs)
-      : Callee(Callee) {
-    assert(CleanupArgs.size() == N);
-    std::copy(CleanupArgs.begin(), CleanupArgs.end(), std::begin(Args));
+  CommonActionTy(llvm::Value *EnterCallee, ArrayRef<llvm::Value *> EnterArgs,
+                 llvm::Value *ExitCallee, ArrayRef<llvm::Value *> ExitArgs,
+                 bool Conditional = false)
+      : EnterCallee(EnterCallee), EnterArgs(EnterArgs), ExitCallee(ExitCallee),
+        ExitArgs(ExitArgs), Conditional(Conditional) {}
+  void Enter(CodeGenFunction &CGF) override {
+    llvm::Value *EnterRes = CGF.EmitRuntimeCall(EnterCallee, EnterArgs);
+    if (Conditional) {
+      llvm::Value *CallBool = CGF.Builder.CreateIsNotNull(EnterRes);
+      auto *ThenBlock = CGF.createBasicBlock("omp_if.then");
+      ContBlock = CGF.createBasicBlock("omp_if.end");
+      // Generate the branch (If-stmt)
+      CGF.Builder.CreateCondBr(CallBool, ThenBlock, ContBlock);
+      CGF.EmitBlock(ThenBlock);
+    }
   }
-  void Emit(CodeGenFunction &CGF, Flags /*flags*/) override {
-    if (!CGF.HaveInsertPoint())
-      return;
-    CGF.EmitRuntimeCall(Callee, Args);
+  void Done(CodeGenFunction &CGF) {
+    // Emit the rest of blocks/branches
+    CGF.EmitBranch(ContBlock);
+    CGF.EmitBlock(ContBlock, true);
+  }
+  void Exit(CodeGenFunction &CGF) override {
+    CGF.EmitRuntimeCall(ExitCallee, ExitArgs);
   }
 };
 } // anonymous namespace
@@ -1425,45 +2082,22 @@ void CGOpenMPRuntime::emitCriticalRegion(CodeGenFunction &CGF,
   // Prepare arguments and build a call to __kmpc_critical
   if (!CGF.HaveInsertPoint())
     return;
-  CodeGenFunction::RunCleanupsScope Scope(CGF);
   llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc),
                          getCriticalRegionLock(CriticalName)};
+  llvm::SmallVector<llvm::Value *, 4> EnterArgs(std::begin(Args),
+                                                std::end(Args));
   if (Hint) {
-    llvm::SmallVector<llvm::Value *, 8> ArgsWithHint(std::begin(Args),
-                                                     std::end(Args));
-    auto *HintVal = CGF.EmitScalarExpr(Hint);
-    ArgsWithHint.push_back(
-        CGF.Builder.CreateIntCast(HintVal, CGM.IntPtrTy, /*isSigned=*/false));
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_critical_with_hint),
-                        ArgsWithHint);
-  } else
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_critical), Args);
-  // Build a call to __kmpc_end_critical
-  CGF.EHStack.pushCleanup<CallEndCleanup<std::extent<decltype(Args)>::value>>(
-      NormalAndEHCleanup, createRuntimeFunction(OMPRTL__kmpc_end_critical),
-      llvm::makeArrayRef(Args));
+    EnterArgs.push_back(CGF.Builder.CreateIntCast(
+        CGF.EmitScalarExpr(Hint), CGM.IntPtrTy, /*isSigned=*/false));
+  }
+  CommonActionTy Action(
+      createRuntimeFunction(Hint ? OMPRTL__kmpc_critical_with_hint
+                                 : OMPRTL__kmpc_critical),
+      EnterArgs, createRuntimeFunction(OMPRTL__kmpc_end_critical), Args);
+  CriticalOpGen.setAction(Action);
   emitInlinedDirective(CGF, OMPD_critical, CriticalOpGen);
 }
 
-static void emitIfStmt(CodeGenFunction &CGF, llvm::Value *IfCond,
-                       OpenMPDirectiveKind Kind, SourceLocation Loc,
-                       const RegionCodeGenTy &BodyOpGen) {
-  llvm::Value *CallBool = CGF.EmitScalarConversion(
-      IfCond,
-      CGF.getContext().getIntTypeForBitwidth(/*DestWidth=*/32, /*Signed=*/true),
-      CGF.getContext().BoolTy, Loc);
-
-  auto *ThenBlock = CGF.createBasicBlock("omp_if.then");
-  auto *ContBlock = CGF.createBasicBlock("omp_if.end");
-  // Generate the branch (If-stmt)
-  CGF.Builder.CreateCondBr(CallBool, ThenBlock, ContBlock);
-  CGF.EmitBlock(ThenBlock);
-  CGF.CGM.getOpenMPRuntime().emitInlinedDirective(CGF, Kind, BodyOpGen);
-  // Emit the rest of bblocks/branches
-  CGF.EmitBranch(ContBlock);
-  CGF.EmitBlock(ContBlock, true);
-}
-
 void CGOpenMPRuntime::emitMasterRegion(CodeGenFunction &CGF,
                                        const RegionCodeGenTy &MasterOpGen,
                                        SourceLocation Loc) {
@@ -1475,18 +2109,12 @@ void CGOpenMPRuntime::emitMasterRegion(CodeGenFunction &CGF,
   // }
   // Prepare arguments and build a call to __kmpc_master
   llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
-  auto *IsMaster =
-      CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_master), Args);
-  typedef CallEndCleanup<std::extent<decltype(Args)>::value>
-      MasterCallEndCleanup;
-  emitIfStmt(
-      CGF, IsMaster, OMPD_master, Loc, [&](CodeGenFunction &CGF) -> void {
-        CodeGenFunction::RunCleanupsScope Scope(CGF);
-        CGF.EHStack.pushCleanup<MasterCallEndCleanup>(
-            NormalAndEHCleanup, createRuntimeFunction(OMPRTL__kmpc_end_master),
-            llvm::makeArrayRef(Args));
-        MasterOpGen(CGF);
-      });
+  CommonActionTy Action(createRuntimeFunction(OMPRTL__kmpc_master), Args,
+                        createRuntimeFunction(OMPRTL__kmpc_end_master), Args,
+                        /*Conditional=*/true);
+  MasterOpGen.setAction(Action);
+  emitInlinedDirective(CGF, OMPD_master, MasterOpGen);
+  Action.Done(CGF);
 }
 
 void CGOpenMPRuntime::emitTaskyieldCall(CodeGenFunction &CGF,
@@ -1498,6 +2126,8 @@ void CGOpenMPRuntime::emitTaskyieldCall(CodeGenFunction &CGF,
       emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc),
       llvm::ConstantInt::get(CGM.IntTy, /*V=*/0, /*isSigned=*/true)};
   CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_taskyield), Args);
+  if (auto *Region = dyn_cast_or_null<CGOpenMPRegionInfo>(CGF.CapturedStmtInfo))
+    Region->emitUntiedSwitch(CGF);
 }
 
 void CGOpenMPRuntime::emitTaskgroupRegion(CodeGenFunction &CGF,
@@ -1509,16 +2139,12 @@ void CGOpenMPRuntime::emitTaskgroupRegion(CodeGenFunction &CGF,
   // TaskgroupOpGen();
   // __kmpc_end_taskgroup(ident_t *, gtid);
   // Prepare arguments and build a call to __kmpc_taskgroup
-  {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_taskgroup), Args);
-    // Build a call to __kmpc_end_taskgroup
-    CGF.EHStack.pushCleanup<CallEndCleanup<std::extent<decltype(Args)>::value>>(
-        NormalAndEHCleanup, createRuntimeFunction(OMPRTL__kmpc_end_taskgroup),
-        llvm::makeArrayRef(Args));
-    emitInlinedDirective(CGF, OMPD_taskgroup, TaskgroupOpGen);
-  }
+  llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
+  CommonActionTy Action(createRuntimeFunction(OMPRTL__kmpc_taskgroup), Args,
+                        createRuntimeFunction(OMPRTL__kmpc_end_taskgroup),
+                        Args);
+  TaskgroupOpGen.setAction(Action);
+  emitInlinedDirective(CGF, OMPD_taskgroup, TaskgroupOpGen);
 }
 
 /// Given an array of pointers to variables, project the address of a
@@ -1549,9 +2175,7 @@ static llvm::Value *emitCopyprivateCopyFunction(
                            C.VoidPtrTy);
   Args.push_back(&LHSArg);
   Args.push_back(&RHSArg);
-  FunctionType::ExtInfo EI;
-  auto &CGFI = CGM.getTypes().arrangeFreeFunctionDeclaration(
-      C.VoidTy, Args, EI, /*isVariadic=*/false);
+  auto &CGFI = CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
   auto *Fn = llvm::Function::Create(
       CGM.getTypes().GetFunctionType(CGFI), llvm::GlobalValue::InternalLinkage,
       ".omp.copyprivate.copy_func", &CGM.getModule());
@@ -1616,22 +2240,16 @@ void CGOpenMPRuntime::emitSingleRegion(CodeGenFunction &CGF,
   }
   // Prepare arguments and build a call to __kmpc_single
   llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
-  auto *IsSingle =
-      CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_single), Args);
-  typedef CallEndCleanup<std::extent<decltype(Args)>::value>
-      SingleCallEndCleanup;
-  emitIfStmt(
-      CGF, IsSingle, OMPD_single, Loc, [&](CodeGenFunction &CGF) -> void {
-        CodeGenFunction::RunCleanupsScope Scope(CGF);
-        CGF.EHStack.pushCleanup<SingleCallEndCleanup>(
-            NormalAndEHCleanup, createRuntimeFunction(OMPRTL__kmpc_end_single),
-            llvm::makeArrayRef(Args));
-        SingleOpGen(CGF);
-        if (DidIt.isValid()) {
-          // did_it = 1;
-          CGF.Builder.CreateStore(CGF.Builder.getInt32(1), DidIt);
-        }
-      });
+  CommonActionTy Action(createRuntimeFunction(OMPRTL__kmpc_single), Args,
+                        createRuntimeFunction(OMPRTL__kmpc_end_single), Args,
+                        /*Conditional=*/true);
+  SingleOpGen.setAction(Action);
+  emitInlinedDirective(CGF, OMPD_single, SingleOpGen);
+  if (DidIt.isValid()) {
+    // did_it = 1;
+    CGF.Builder.CreateStore(CGF.Builder.getInt32(1), DidIt);
+  }
+  Action.Done(CGF);
   // call __kmpc_copyprivate(ident_t *, gtid, <buf_size>, <copyprivate list>,
   // <copy_func>, did_it);
   if (DidIt.isValid()) {
@@ -1655,7 +2273,7 @@ void CGOpenMPRuntime::emitSingleRegion(CodeGenFunction &CGF,
     auto *CpyFn = emitCopyprivateCopyFunction(
         CGM, CGF.ConvertTypeForMem(CopyprivateArrayTy)->getPointerTo(),
         CopyprivateVars, SrcExprs, DstExprs, AssignmentOps);
-    auto *BufSize = getTypeSize(CGF, CopyprivateArrayTy);
+    auto *BufSize = CGF.getTypeSize(CopyprivateArrayTy);
     Address CL =
       CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(CopyprivateList,
                                                       CGF.VoidPtrTy);
@@ -1681,14 +2299,14 @@ void CGOpenMPRuntime::emitOrderedRegion(CodeGenFunction &CGF,
   // OrderedOpGen();
   // __kmpc_end_ordered(ident_t *, gtid);
   // Prepare arguments and build a call to __kmpc_ordered
-  CodeGenFunction::RunCleanupsScope Scope(CGF);
   if (IsThreads) {
     llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_ordered), Args);
-    // Build a call to __kmpc_end_ordered
-    CGF.EHStack.pushCleanup<CallEndCleanup<std::extent<decltype(Args)>::value>>(
-        NormalAndEHCleanup, createRuntimeFunction(OMPRTL__kmpc_end_ordered),
-        llvm::makeArrayRef(Args));
+    CommonActionTy Action(createRuntimeFunction(OMPRTL__kmpc_ordered), Args,
+                          createRuntimeFunction(OMPRTL__kmpc_end_ordered),
+                          Args);
+    OrderedOpGen.setAction(Action);
+    emitInlinedDirective(CGF, OMPD_ordered, OrderedOpGen);
+    return;
   }
   emitInlinedDirective(CGF, OMPD_ordered, OrderedOpGen);
 }
@@ -1700,21 +2318,17 @@ void CGOpenMPRuntime::emitBarrierCall(CodeGenFunction &CGF, SourceLocation Loc,
     return;
   // Build call __kmpc_cancel_barrier(loc, thread_id);
   // Build call __kmpc_barrier(loc, thread_id);
-  OpenMPLocationFlags Flags = OMP_IDENT_KMPC;
-  if (Kind == OMPD_for) {
-    Flags =
-        static_cast<OpenMPLocationFlags>(Flags | OMP_IDENT_BARRIER_IMPL_FOR);
-  } else if (Kind == OMPD_sections) {
-    Flags = static_cast<OpenMPLocationFlags>(Flags |
-                                             OMP_IDENT_BARRIER_IMPL_SECTIONS);
-  } else if (Kind == OMPD_single) {
-    Flags =
-        static_cast<OpenMPLocationFlags>(Flags | OMP_IDENT_BARRIER_IMPL_SINGLE);
-  } else if (Kind == OMPD_barrier) {
-    Flags = static_cast<OpenMPLocationFlags>(Flags | OMP_IDENT_BARRIER_EXPL);
-  } else {
-    Flags = static_cast<OpenMPLocationFlags>(Flags | OMP_IDENT_BARRIER_IMPL);
-  }
+  unsigned Flags;
+  if (Kind == OMPD_for)
+    Flags = OMP_IDENT_BARRIER_IMPL_FOR;
+  else if (Kind == OMPD_sections)
+    Flags = OMP_IDENT_BARRIER_IMPL_SECTIONS;
+  else if (Kind == OMPD_single)
+    Flags = OMP_IDENT_BARRIER_IMPL_SINGLE;
+  else if (Kind == OMPD_barrier)
+    Flags = OMP_IDENT_BARRIER_EXPL;
+  else
+    Flags = OMP_IDENT_BARRIER_IMPL;
   // Build call __kmpc_cancel_barrier(loc, thread_id) or __kmpc_barrier(loc,
   // thread_id);
   llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc, Flags),
@@ -1745,28 +2359,6 @@ void CGOpenMPRuntime::emitBarrierCall(CodeGenFunction &CGF, SourceLocation Loc,
   CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_barrier), Args);
 }
 
-/// \brief Schedule types for 'omp for' loops (these enumerators are taken from
-/// the enum sched_type in kmp.h).
-enum OpenMPSchedType {
-  /// \brief Lower bound for default (unordered) versions.
-  OMP_sch_lower = 32,
-  OMP_sch_static_chunked = 33,
-  OMP_sch_static = 34,
-  OMP_sch_dynamic_chunked = 35,
-  OMP_sch_guided_chunked = 36,
-  OMP_sch_runtime = 37,
-  OMP_sch_auto = 38,
-  /// \brief Lower bound for 'ordered' versions.
-  OMP_ord_lower = 64,
-  OMP_ord_static_chunked = 65,
-  OMP_ord_static = 66,
-  OMP_ord_dynamic_chunked = 67,
-  OMP_ord_guided_chunked = 68,
-  OMP_ord_runtime = 69,
-  OMP_ord_auto = 70,
-  OMP_sch_default = OMP_sch_static,
-};
-
 /// \brief Map the OpenMP loop schedule to the runtime enumeration.
 static OpenMPSchedType getRuntimeSchedule(OpenMPScheduleClauseKind ScheduleKind,
                                           bool Chunked, bool Ordered) {
@@ -1789,12 +2381,26 @@ static OpenMPSchedType getRuntimeSchedule(OpenMPScheduleClauseKind ScheduleKind,
   llvm_unreachable("Unexpected runtime schedule");
 }
 
+/// \brief Map the OpenMP distribute schedule to the runtime enumeration.
+static OpenMPSchedType
+getRuntimeSchedule(OpenMPDistScheduleClauseKind ScheduleKind, bool Chunked) {
+  // only static is allowed for dist_schedule
+  return Chunked ? OMP_dist_sch_static_chunked : OMP_dist_sch_static;
+}
+
 bool CGOpenMPRuntime::isStaticNonchunked(OpenMPScheduleClauseKind ScheduleKind,
                                          bool Chunked) const {
   auto Schedule = getRuntimeSchedule(ScheduleKind, Chunked, /*Ordered=*/false);
   return Schedule == OMP_sch_static;
 }
 
+bool CGOpenMPRuntime::isStaticNonchunked(
+    OpenMPDistScheduleClauseKind ScheduleKind, bool Chunked) const {
+  auto Schedule = getRuntimeSchedule(ScheduleKind, Chunked);
+  return Schedule == OMP_dist_sch_static;
+}
+
+
 bool CGOpenMPRuntime::isDynamic(OpenMPScheduleClauseKind ScheduleKind) const {
   auto Schedule =
       getRuntimeSchedule(ScheduleKind, /*Chunked=*/false, /*Ordered=*/false);
@@ -1802,19 +2408,57 @@ bool CGOpenMPRuntime::isDynamic(OpenMPScheduleClauseKind ScheduleKind) const {
   return Schedule != OMP_sch_static;
 }
 
+static int addMonoNonMonoModifier(OpenMPSchedType Schedule,
+                                  OpenMPScheduleClauseModifier M1,
+                                  OpenMPScheduleClauseModifier M2) {
+  int Modifier = 0;
+  switch (M1) {
+  case OMPC_SCHEDULE_MODIFIER_monotonic:
+    Modifier = OMP_sch_modifier_monotonic;
+    break;
+  case OMPC_SCHEDULE_MODIFIER_nonmonotonic:
+    Modifier = OMP_sch_modifier_nonmonotonic;
+    break;
+  case OMPC_SCHEDULE_MODIFIER_simd:
+    if (Schedule == OMP_sch_static_chunked)
+      Schedule = OMP_sch_static_balanced_chunked;
+    break;
+  case OMPC_SCHEDULE_MODIFIER_last:
+  case OMPC_SCHEDULE_MODIFIER_unknown:
+    break;
+  }
+  switch (M2) {
+  case OMPC_SCHEDULE_MODIFIER_monotonic:
+    Modifier = OMP_sch_modifier_monotonic;
+    break;
+  case OMPC_SCHEDULE_MODIFIER_nonmonotonic:
+    Modifier = OMP_sch_modifier_nonmonotonic;
+    break;
+  case OMPC_SCHEDULE_MODIFIER_simd:
+    if (Schedule == OMP_sch_static_chunked)
+      Schedule = OMP_sch_static_balanced_chunked;
+    break;
+  case OMPC_SCHEDULE_MODIFIER_last:
+  case OMPC_SCHEDULE_MODIFIER_unknown:
+    break;
+  }
+  return Schedule | Modifier;
+}
+
 void CGOpenMPRuntime::emitForDispatchInit(CodeGenFunction &CGF,
                                           SourceLocation Loc,
-                                          OpenMPScheduleClauseKind ScheduleKind,
+                                          const OpenMPScheduleTy &ScheduleKind,
                                           unsigned IVSize, bool IVSigned,
                                           bool Ordered, llvm::Value *UB,
                                           llvm::Value *Chunk) {
   if (!CGF.HaveInsertPoint())
     return;
   OpenMPSchedType Schedule =
-      getRuntimeSchedule(ScheduleKind, Chunk != nullptr, Ordered);
+      getRuntimeSchedule(ScheduleKind.Schedule, Chunk != nullptr, Ordered);
   assert(Ordered ||
          (Schedule != OMP_sch_static && Schedule != OMP_sch_static_chunked &&
-          Schedule != OMP_ord_static && Schedule != OMP_ord_static_chunked));
+          Schedule != OMP_ord_static && Schedule != OMP_ord_static_chunked &&
+          Schedule != OMP_sch_static_balanced_chunked));
   // Call __kmpc_dispatch_init(
   //          ident_t *loc, kmp_int32 tid, kmp_int32 schedule,
   //          kmp_int[32|64] lower, kmp_int[32|64] upper,
@@ -1824,59 +2468,94 @@ void CGOpenMPRuntime::emitForDispatchInit(CodeGenFunction &CGF,
   if (Chunk == nullptr)
     Chunk = CGF.Builder.getIntN(IVSize, 1);
   llvm::Value *Args[] = {
-    emitUpdateLocation(CGF, Loc, OMP_IDENT_KMPC),
-    getThreadID(CGF, Loc),
-    CGF.Builder.getInt32(Schedule), // Schedule type
-    CGF.Builder.getIntN(IVSize, 0), // Lower
-    UB,                             // Upper
-    CGF.Builder.getIntN(IVSize, 1), // Stride
-    Chunk                           // Chunk
+      emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc),
+      CGF.Builder.getInt32(addMonoNonMonoModifier(
+          Schedule, ScheduleKind.M1, ScheduleKind.M2)), // Schedule type
+      CGF.Builder.getIntN(IVSize, 0),                   // Lower
+      UB,                                               // Upper
+      CGF.Builder.getIntN(IVSize, 1),                   // Stride
+      Chunk                                             // Chunk
   };
   CGF.EmitRuntimeCall(createDispatchInitFunction(IVSize, IVSigned), Args);
 }
 
+static void emitForStaticInitCall(
+    CodeGenFunction &CGF, llvm::Value *UpdateLocation, llvm::Value *ThreadId,
+    llvm::Constant *ForStaticInitFunction, OpenMPSchedType Schedule,
+    OpenMPScheduleClauseModifier M1, OpenMPScheduleClauseModifier M2,
+    unsigned IVSize, bool Ordered, Address IL, Address LB, Address UB,
+    Address ST, llvm::Value *Chunk) {
+  if (!CGF.HaveInsertPoint())
+     return;
+
+   assert(!Ordered);
+   assert(Schedule == OMP_sch_static || Schedule == OMP_sch_static_chunked ||
+          Schedule == OMP_sch_static_balanced_chunked ||
+          Schedule == OMP_ord_static || Schedule == OMP_ord_static_chunked ||
+          Schedule == OMP_dist_sch_static ||
+          Schedule == OMP_dist_sch_static_chunked);
+
+   // Call __kmpc_for_static_init(
+   //          ident_t *loc, kmp_int32 tid, kmp_int32 schedtype,
+   //          kmp_int32 *p_lastiter, kmp_int[32|64] *p_lower,
+   //          kmp_int[32|64] *p_upper, kmp_int[32|64] *p_stride,
+   //          kmp_int[32|64] incr, kmp_int[32|64] chunk);
+   if (Chunk == nullptr) {
+     assert((Schedule == OMP_sch_static || Schedule == OMP_ord_static ||
+             Schedule == OMP_dist_sch_static) &&
+            "expected static non-chunked schedule");
+     // If the Chunk was not specified in the clause - use default value 1.
+       Chunk = CGF.Builder.getIntN(IVSize, 1);
+   } else {
+     assert((Schedule == OMP_sch_static_chunked ||
+             Schedule == OMP_sch_static_balanced_chunked ||
+             Schedule == OMP_ord_static_chunked ||
+             Schedule == OMP_dist_sch_static_chunked) &&
+            "expected static chunked schedule");
+   }
+   llvm::Value *Args[] = {
+       UpdateLocation, ThreadId, CGF.Builder.getInt32(addMonoNonMonoModifier(
+                                     Schedule, M1, M2)), // Schedule type
+       IL.getPointer(),                                  // &isLastIter
+       LB.getPointer(),                                  // &LB
+       UB.getPointer(),                                  // &UB
+       ST.getPointer(),                                  // &Stride
+       CGF.Builder.getIntN(IVSize, 1),                   // Incr
+       Chunk                                             // Chunk
+   };
+   CGF.EmitRuntimeCall(ForStaticInitFunction, Args);
+}
+
 void CGOpenMPRuntime::emitForStaticInit(CodeGenFunction &CGF,
                                         SourceLocation Loc,
-                                        OpenMPScheduleClauseKind ScheduleKind,
+                                        const OpenMPScheduleTy &ScheduleKind,
                                         unsigned IVSize, bool IVSigned,
                                         bool Ordered, Address IL, Address LB,
                                         Address UB, Address ST,
                                         llvm::Value *Chunk) {
-  if (!CGF.HaveInsertPoint())
-    return;
-  OpenMPSchedType Schedule =
-    getRuntimeSchedule(ScheduleKind, Chunk != nullptr, Ordered);
-  assert(!Ordered);
-  assert(Schedule == OMP_sch_static || Schedule == OMP_sch_static_chunked ||
-         Schedule == OMP_ord_static || Schedule == OMP_ord_static_chunked);
-
-  // Call __kmpc_for_static_init(
-  //          ident_t *loc, kmp_int32 tid, kmp_int32 schedtype,
-  //          kmp_int32 *p_lastiter, kmp_int[32|64] *p_lower,
-  //          kmp_int[32|64] *p_upper, kmp_int[32|64] *p_stride,
-  //          kmp_int[32|64] incr, kmp_int[32|64] chunk);
-  if (Chunk == nullptr) {
-    assert((Schedule == OMP_sch_static || Schedule == OMP_ord_static) &&
-           "expected static non-chunked schedule");
-    // If the Chunk was not specified in the clause - use default value 1.
-      Chunk = CGF.Builder.getIntN(IVSize, 1);
-  } else {
-    assert((Schedule == OMP_sch_static_chunked ||
-            Schedule == OMP_ord_static_chunked) &&
-           "expected static chunked schedule");
-  }
-  llvm::Value *Args[] = {
-    emitUpdateLocation(CGF, Loc, OMP_IDENT_KMPC),
-    getThreadID(CGF, Loc),
-    CGF.Builder.getInt32(Schedule), // Schedule type
-    IL.getPointer(),                // &isLastIter
-    LB.getPointer(),                // &LB
-    UB.getPointer(),                // &UB
-    ST.getPointer(),                // &Stride
-    CGF.Builder.getIntN(IVSize, 1), // Incr
-    Chunk                           // Chunk
-  };
-  CGF.EmitRuntimeCall(createForStaticInitFunction(IVSize, IVSigned), Args);
+  OpenMPSchedType ScheduleNum =
+      getRuntimeSchedule(ScheduleKind.Schedule, Chunk != nullptr, Ordered);
+  auto *UpdatedLocation = emitUpdateLocation(CGF, Loc);
+  auto *ThreadId = getThreadID(CGF, Loc);
+  auto *StaticInitFunction = createForStaticInitFunction(IVSize, IVSigned);
+  emitForStaticInitCall(CGF, UpdatedLocation, ThreadId, StaticInitFunction,
+                        ScheduleNum, ScheduleKind.M1, ScheduleKind.M2, IVSize,
+                        Ordered, IL, LB, UB, ST, Chunk);
+}
+
+void CGOpenMPRuntime::emitDistributeStaticInit(
+    CodeGenFunction &CGF, SourceLocation Loc,
+    OpenMPDistScheduleClauseKind SchedKind, unsigned IVSize, bool IVSigned,
+    bool Ordered, Address IL, Address LB, Address UB, Address ST,
+    llvm::Value *Chunk) {
+  OpenMPSchedType ScheduleNum = getRuntimeSchedule(SchedKind, Chunk != nullptr);
+  auto *UpdatedLocation = emitUpdateLocation(CGF, Loc);
+  auto *ThreadId = getThreadID(CGF, Loc);
+  auto *StaticInitFunction = createForStaticInitFunction(IVSize, IVSigned);
+  emitForStaticInitCall(CGF, UpdatedLocation, ThreadId, StaticInitFunction,
+                        ScheduleNum, OMPC_SCHEDULE_MODIFIER_unknown,
+                        OMPC_SCHEDULE_MODIFIER_unknown, IVSize, Ordered, IL, LB,
+                        UB, ST, Chunk);
 }
 
 void CGOpenMPRuntime::emitForStaticFinish(CodeGenFunction &CGF,
@@ -1884,8 +2563,7 @@ void CGOpenMPRuntime::emitForStaticFinish(CodeGenFunction &CGF,
   if (!CGF.HaveInsertPoint())
     return;
   // Call __kmpc_for_static_fini(ident_t *loc, kmp_int32 tid);
-  llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc, OMP_IDENT_KMPC),
-                         getThreadID(CGF, Loc)};
+  llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
   CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_for_static_fini),
                       Args);
 }
@@ -1897,8 +2575,7 @@ void CGOpenMPRuntime::emitForOrderedIterationEnd(CodeGenFunction &CGF,
   if (!CGF.HaveInsertPoint())
     return;
   // Call __kmpc_for_dynamic_fini_(4|8)[u](ident_t *loc, kmp_int32 tid);
-  llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc, OMP_IDENT_KMPC),
-                         getThreadID(CGF, Loc)};
+  llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
   CGF.EmitRuntimeCall(createDispatchFiniFunction(IVSize, IVSigned), Args);
 }
 
@@ -1912,7 +2589,8 @@ llvm::Value *CGOpenMPRuntime::emitForNext(CodeGenFunction &CGF,
   //          kmp_int[32|64] *p_lower, kmp_int[32|64] *p_upper,
   //          kmp_int[32|64] *p_stride);
   llvm::Value *Args[] = {
-      emitUpdateLocation(CGF, Loc, OMP_IDENT_KMPC), getThreadID(CGF, Loc),
+      emitUpdateLocation(CGF, Loc),
+      getThreadID(CGF, Loc),
       IL.getPointer(), // &isLastIter
       LB.getPointer(), // &Lower
       UB.getPointer(), // &Upper
@@ -1991,8 +2669,18 @@ enum KmpTaskTFields {
   KmpTaskTRoutine,
   /// \brief Partition id for the untied tasks.
   KmpTaskTPartId,
-  /// \brief Function with call of destructors for private variables.
-  KmpTaskTDestructors,
+  /// Function with call of destructors for private variables.
+  Data1,
+  /// Task priority.
+  Data2,
+  /// (Taskloops only) Lower bound.
+  KmpTaskTLowerBound,
+  /// (Taskloops only) Upper bound.
+  KmpTaskTUpperBound,
+  /// (Taskloops only) Stride.
+  KmpTaskTStride,
+  /// (Taskloops only) Is last iteration flag.
+  KmpTaskTLastIter,
 };
 } // anonymous namespace
 
@@ -2005,11 +2693,11 @@ bool CGOpenMPRuntime::OffloadEntriesInfoManagerTy::empty() const {
 void CGOpenMPRuntime::OffloadEntriesInfoManagerTy::
     initializeTargetRegionEntryInfo(unsigned DeviceID, unsigned FileID,
                                     StringRef ParentName, unsigned LineNum,
-                                    unsigned ColNum, unsigned Order) {
+                                    unsigned Order) {
   assert(CGM.getLangOpts().OpenMPIsDevice && "Initialization of entries is "
                                              "only required for the device "
                                              "code generation.");
-  OffloadEntriesTargetRegion[DeviceID][FileID][ParentName][LineNum][ColNum] =
+  OffloadEntriesTargetRegion[DeviceID][FileID][ParentName][LineNum] =
       OffloadEntryInfoTargetRegion(Order, /*Addr=*/nullptr, /*ID=*/nullptr);
   ++OffloadingEntriesNum;
 }
@@ -2017,30 +2705,27 @@ void CGOpenMPRuntime::OffloadEntriesInfoManagerTy::
 void CGOpenMPRuntime::OffloadEntriesInfoManagerTy::
     registerTargetRegionEntryInfo(unsigned DeviceID, unsigned FileID,
                                   StringRef ParentName, unsigned LineNum,
-                                  unsigned ColNum, llvm::Constant *Addr,
-                                  llvm::Constant *ID) {
+                                  llvm::Constant *Addr, llvm::Constant *ID) {
   // If we are emitting code for a target, the entry is already initialized,
   // only has to be registered.
   if (CGM.getLangOpts().OpenMPIsDevice) {
-    assert(hasTargetRegionEntryInfo(DeviceID, FileID, ParentName, LineNum,
-                                    ColNum) &&
+    assert(hasTargetRegionEntryInfo(DeviceID, FileID, ParentName, LineNum) &&
            "Entry must exist.");
-    auto &Entry = OffloadEntriesTargetRegion[DeviceID][FileID][ParentName]
-                                            [LineNum][ColNum];
+    auto &Entry =
+        OffloadEntriesTargetRegion[DeviceID][FileID][ParentName][LineNum];
     assert(Entry.isValid() && "Entry not initialized!");
     Entry.setAddress(Addr);
     Entry.setID(ID);
     return;
   } else {
     OffloadEntryInfoTargetRegion Entry(OffloadingEntriesNum++, Addr, ID);
-    OffloadEntriesTargetRegion[DeviceID][FileID][ParentName][LineNum][ColNum] =
-        Entry;
+    OffloadEntriesTargetRegion[DeviceID][FileID][ParentName][LineNum] = Entry;
   }
 }
 
 bool CGOpenMPRuntime::OffloadEntriesInfoManagerTy::hasTargetRegionEntryInfo(
-    unsigned DeviceID, unsigned FileID, StringRef ParentName, unsigned LineNum,
-    unsigned ColNum) const {
+    unsigned DeviceID, unsigned FileID, StringRef ParentName,
+    unsigned LineNum) const {
   auto PerDevice = OffloadEntriesTargetRegion.find(DeviceID);
   if (PerDevice == OffloadEntriesTargetRegion.end())
     return false;
@@ -2053,11 +2738,8 @@ bool CGOpenMPRuntime::OffloadEntriesInfoManagerTy::hasTargetRegionEntryInfo(
   auto PerLine = PerParentName->second.find(LineNum);
   if (PerLine == PerParentName->second.end())
     return false;
-  auto PerColumn = PerLine->second.find(ColNum);
-  if (PerColumn == PerLine->second.end())
-    return false;
   // Fail if this entry is already registered.
-  if (PerColumn->second.getAddress() || PerColumn->second.getID())
+  if (PerLine->second.getAddress() || PerLine->second.getID())
     return false;
   return true;
 }
@@ -2069,8 +2751,7 @@ void CGOpenMPRuntime::OffloadEntriesInfoManagerTy::actOnTargetRegionEntriesInfo(
     for (auto &F : D.second)
       for (auto &P : F.second)
         for (auto &L : P.second)
-          for (auto &C : L.second)
-            Action(D.first, F.first, P.first(), L.first, C.first, C.second);
+          Action(D.first, F.first, P.first(), L.first, L.second);
 }
 
 /// \brief Create a Ctor/Dtor-like function whose body is emitted through
@@ -2087,9 +2768,7 @@ createOffloadingBinaryDescriptorFunction(CodeGenModule &CGM, StringRef Name,
 
   CodeGenFunction CGF(CGM);
   GlobalDecl();
-  auto &FI = CGM.getTypes().arrangeFreeFunctionDeclaration(
-      C.VoidTy, Args, FunctionType::ExtInfo(),
-      /*isVariadic=*/false);
+  auto &FI = CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
   auto FTy = CGM.getTypes().GetFunctionType(FI);
   auto *Fn =
       CGM.CreateGlobalInitOrDestructFunction(FTy, Name, FI, SourceLocation());
@@ -2123,11 +2802,11 @@ CGOpenMPRuntime::createOffloadingBinaryDescriptorRegistration() {
       CGM.getTypes().ConvertTypeForMem(getTgtOffloadEntryQTy());
   llvm::GlobalVariable *HostEntriesBegin = new llvm::GlobalVariable(
       M, OffloadEntryTy, /*isConstant=*/true,
-      llvm::GlobalValue::ExternalLinkage, /*Initializer=*/0,
+      llvm::GlobalValue::ExternalLinkage, /*Initializer=*/nullptr,
       ".omp_offloading.entries_begin");
   llvm::GlobalVariable *HostEntriesEnd = new llvm::GlobalVariable(
       M, OffloadEntryTy, /*isConstant=*/true,
-      llvm::GlobalValue::ExternalLinkage, /*Initializer=*/0,
+      llvm::GlobalValue::ExternalLinkage, /*Initializer=*/nullptr,
       ".omp_offloading.entries_end");
 
   // Create all device images
@@ -2139,10 +2818,11 @@ CGOpenMPRuntime::createOffloadingBinaryDescriptorRegistration() {
     StringRef T = Devices[i].getTriple();
     auto *ImgBegin = new llvm::GlobalVariable(
         M, CGM.Int8Ty, /*isConstant=*/true, llvm::GlobalValue::ExternalLinkage,
-        /*Initializer=*/0, Twine(".omp_offloading.img_start.") + Twine(T));
+        /*Initializer=*/nullptr,
+        Twine(".omp_offloading.img_start.") + Twine(T));
     auto *ImgEnd = new llvm::GlobalVariable(
         M, CGM.Int8Ty, /*isConstant=*/true, llvm::GlobalValue::ExternalLinkage,
-        /*Initializer=*/0, Twine(".omp_offloading.img_end.") + Twine(T));
+        /*Initializer=*/nullptr, Twine(".omp_offloading.img_end.") + Twine(T));
 
     llvm::Constant *Dev =
         llvm::ConstantStruct::get(DeviceImageTy, ImgBegin, ImgEnd,
@@ -2160,7 +2840,7 @@ CGOpenMPRuntime::createOffloadingBinaryDescriptorRegistration() {
       M, DeviceImagesInitTy, /*isConstant=*/true,
       llvm::GlobalValue::InternalLinkage, DeviceImagesInit,
       ".omp_offloading.device_images");
-  DeviceImages->setUnnamedAddr(true);
+  DeviceImages->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
 
   // This is a Zero array to be used in the creation of the constant expressions
   llvm::Constant *Index[] = {llvm::Constant::getNullValue(CGM.Int32Ty),
@@ -2190,12 +2870,14 @@ CGOpenMPRuntime::createOffloadingBinaryDescriptorRegistration() {
                                 IdentInfo, C.CharTy);
 
   auto *UnRegFn = createOffloadingBinaryDescriptorFunction(
-      CGM, ".omp_offloading.descriptor_unreg", [&](CodeGenFunction &CGF) {
+      CGM, ".omp_offloading.descriptor_unreg",
+      [&](CodeGenFunction &CGF, PrePostActionTy &) {
         CGF.EmitCallOrInvoke(createRuntimeFunction(OMPRTL__tgt_unregister_lib),
                              Desc);
       });
   auto *RegFn = createOffloadingBinaryDescriptorFunction(
-      CGM, ".omp_offloading.descriptor_reg", [&](CodeGenFunction &CGF) {
+      CGM, ".omp_offloading.descriptor_reg",
+      [&](CodeGenFunction &CGF, PrePostActionTy &) {
         CGF.EmitCallOrInvoke(createRuntimeFunction(OMPRTL__tgt_register_lib),
                              Desc);
         CGM.getCXXABI().registerGlobalDtor(CGF, RegUnregVar, UnRegFn, Desc);
@@ -2203,15 +2885,16 @@ CGOpenMPRuntime::createOffloadingBinaryDescriptorRegistration() {
   return RegFn;
 }
 
-void CGOpenMPRuntime::createOffloadEntry(llvm::Constant *Addr, StringRef Name,
-                                         uint64_t Size) {
+void CGOpenMPRuntime::createOffloadEntry(llvm::Constant *ID,
+                                         llvm::Constant *Addr, uint64_t Size) {
+  StringRef Name = Addr->getName();
   auto *TgtOffloadEntryType = cast<llvm::StructType>(
       CGM.getTypes().ConvertTypeForMem(getTgtOffloadEntryQTy()));
   llvm::LLVMContext &C = CGM.getModule().getContext();
   llvm::Module &M = CGM.getModule();
 
   // Make sure the address has the right type.
-  llvm::Constant *AddrPtr = llvm::ConstantExpr::getBitCast(Addr, CGM.VoidPtrTy);
+  llvm::Constant *AddrPtr = llvm::ConstantExpr::getBitCast(ID, CGM.VoidPtrTy);
 
   // Create constant string with the name.
   llvm::Constant *StrPtrInit = llvm::ConstantDataArray::getString(C, Name);
@@ -2220,7 +2903,7 @@ void CGOpenMPRuntime::createOffloadEntry(llvm::Constant *Addr, StringRef Name,
       new llvm::GlobalVariable(M, StrPtrInit->getType(), /*isConstant=*/true,
                                llvm::GlobalValue::InternalLinkage, StrPtrInit,
                                ".omp_offloading.entry_name");
-  Str->setUnnamedAddr(true);
+  Str->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
   llvm::Constant *StrPtr = llvm::ConstantExpr::getBitCast(Str, CGM.Int8PtrTy);
 
   // Create the entry struct.
@@ -2236,7 +2919,6 @@ void CGOpenMPRuntime::createOffloadEntry(llvm::Constant *Addr, StringRef Name,
   // We can't have any padding between symbols, so we need to have 1-byte
   // alignment.
   Entry->setAlignment(1);
-  return;
 }
 
 void CGOpenMPRuntime::createOffloadEntriesAndInfoMetadata() {
@@ -2272,7 +2954,6 @@ void CGOpenMPRuntime::createOffloadEntriesAndInfoMetadata() {
   // Create function that emits metadata for each target region entry;
   auto &&TargetRegionMetadataEmitter = [&](
       unsigned DeviceID, unsigned FileID, StringRef ParentName, unsigned Line,
-      unsigned Column,
       OffloadEntriesInfoManagerTy::OffloadEntryInfoTargetRegion &E) {
     llvm::SmallVector<llvm::Metadata *, 32> Ops;
     // Generate metadata for target regions. Each entry of this metadata
@@ -2282,15 +2963,13 @@ void CGOpenMPRuntime::createOffloadEntriesAndInfoMetadata() {
     // - Entry 2 -> File ID of the file where the entry was identified.
     // - Entry 3 -> Mangled name of the function where the entry was identified.
     // - Entry 4 -> Line in the file where the entry was identified.
-    // - Entry 5 -> Column in the file where the entry was identified.
-    // - Entry 6 -> Order the entry was created.
+    // - Entry 5 -> Order the entry was created.
     // The first element of the metadata node is the kind.
     Ops.push_back(getMDInt(E.getKind()));
     Ops.push_back(getMDInt(DeviceID));
     Ops.push_back(getMDInt(FileID));
     Ops.push_back(getMDString(ParentName));
     Ops.push_back(getMDInt(Line));
-    Ops.push_back(getMDInt(Column));
     Ops.push_back(getMDInt(E.getOrder()));
 
     // Save this entry in the right position of the ordered entries array.
@@ -2310,7 +2989,7 @@ void CGOpenMPRuntime::createOffloadEntriesAndInfoMetadata() {
                 E)) {
       assert(CE->getID() && CE->getAddress() &&
              "Entry ID and Addr are invalid!");
-      createOffloadEntry(CE->getID(), CE->getAddress()->getName(), /*Size=*/0);
+      createOffloadEntry(CE->getID(), CE->getAddress(), /*Size=*/0);
     } else
       llvm_unreachable("Unsupported entry kind.");
   }
@@ -2365,7 +3044,7 @@ void CGOpenMPRuntime::loadOffloadInfoMetadata() {
       OffloadEntriesInfoManager.initializeTargetRegionEntryInfo(
           /*DeviceID=*/getMDInt(1), /*FileID=*/getMDInt(2),
           /*ParentName=*/getMDString(3), /*Line=*/getMDInt(4),
-          /*Column=*/getMDInt(5), /*Order=*/getMDInt(6));
+          /*Order=*/getMDInt(5));
       break;
     }
   }
@@ -2509,21 +3188,45 @@ createPrivatesRecordDecl(CodeGenModule &CGM, ArrayRef<PrivateDataTy> Privates) {
 }
 
 static RecordDecl *
-createKmpTaskTRecordDecl(CodeGenModule &CGM, QualType KmpInt32Ty,
+createKmpTaskTRecordDecl(CodeGenModule &CGM, OpenMPDirectiveKind Kind,
+                         QualType KmpInt32Ty,
                          QualType KmpRoutineEntryPointerQTy) {
   auto &C = CGM.getContext();
   // Build struct kmp_task_t {
   //         void *              shareds;
   //         kmp_routine_entry_t routine;
   //         kmp_int32           part_id;
-  //         kmp_routine_entry_t destructors;
+  //         kmp_cmplrdata_t data1;
+  //         kmp_cmplrdata_t data2;
+  // For taskloops additional fields:
+  //         kmp_uint64          lb;
+  //         kmp_uint64          ub;
+  //         kmp_int64           st;
+  //         kmp_int32           liter;
   //       };
+  auto *UD = C.buildImplicitRecord("kmp_cmplrdata_t", TTK_Union);
+  UD->startDefinition();
+  addFieldToRecordDecl(C, UD, KmpInt32Ty);
+  addFieldToRecordDecl(C, UD, KmpRoutineEntryPointerQTy);
+  UD->completeDefinition();
+  QualType KmpCmplrdataTy = C.getRecordType(UD);
   auto *RD = C.buildImplicitRecord("kmp_task_t");
   RD->startDefinition();
   addFieldToRecordDecl(C, RD, C.VoidPtrTy);
   addFieldToRecordDecl(C, RD, KmpRoutineEntryPointerQTy);
   addFieldToRecordDecl(C, RD, KmpInt32Ty);
-  addFieldToRecordDecl(C, RD, KmpRoutineEntryPointerQTy);
+  addFieldToRecordDecl(C, RD, KmpCmplrdataTy);
+  addFieldToRecordDecl(C, RD, KmpCmplrdataTy);
+  if (isOpenMPTaskLoopDirective(Kind)) {
+    QualType KmpUInt64Ty =
+        CGM.getContext().getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/0);
+    QualType KmpInt64Ty =
+        CGM.getContext().getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/1);
+    addFieldToRecordDecl(C, RD, KmpUInt64Ty);
+    addFieldToRecordDecl(C, RD, KmpUInt64Ty);
+    addFieldToRecordDecl(C, RD, KmpInt64Ty);
+    addFieldToRecordDecl(C, RD, KmpInt32Ty);
+  }
   RD->completeDefinition();
   return RD;
 }
@@ -2550,14 +3253,17 @@ createKmpTaskTWithPrivatesRecordDecl(CodeGenModule &CGM, QualType KmpTaskTQTy,
 /// argument.
 /// \code
 /// kmp_int32 .omp_task_entry.(kmp_int32 gtid, kmp_task_t *tt) {
-///   TaskFunction(gtid, tt->part_id, &tt->privates, task_privates_map,
+///   TaskFunction(gtid, tt->part_id, &tt->privates, task_privates_map, tt,
+///   For taskloops:
+///   tt->task_data.lb, tt->task_data.ub, tt->task_data.st, tt->task_data.liter,
 ///   tt->shareds);
 ///   return 0;
 /// }
 /// \endcode
 static llvm::Value *
 emitProxyTaskFunction(CodeGenModule &CGM, SourceLocation Loc,
-                      QualType KmpInt32Ty, QualType KmpTaskTWithPrivatesPtrQTy,
+                      OpenMPDirectiveKind Kind, QualType KmpInt32Ty,
+                      QualType KmpTaskTWithPrivatesPtrQTy,
                       QualType KmpTaskTWithPrivatesQTy, QualType KmpTaskTQTy,
                       QualType SharedsPtrTy, llvm::Value *TaskFunction,
                       llvm::Value *TaskPrivatesMap) {
@@ -2569,10 +3275,8 @@ emitProxyTaskFunction(CodeGenModule &CGM, SourceLocation Loc,
                                 KmpTaskTWithPrivatesPtrQTy.withRestrict());
   Args.push_back(&GtidArg);
   Args.push_back(&TaskTypeArg);
-  FunctionType::ExtInfo Info;
   auto &TaskEntryFnInfo =
-      CGM.getTypes().arrangeFreeFunctionDeclaration(KmpInt32Ty, Args, Info,
-                                                    /*isVariadic=*/false);
+      CGM.getTypes().arrangeBuiltinFunctionDeclaration(KmpInt32Ty, Args);
   auto *TaskEntryTy = CGM.getTypes().GetFunctionType(TaskEntryFnInfo);
   auto *TaskEntry =
       llvm::Function::Create(TaskEntryTy, llvm::GlobalValue::InternalLinkage,
@@ -2583,11 +3287,15 @@ emitProxyTaskFunction(CodeGenModule &CGM, SourceLocation Loc,
   CGF.StartFunction(GlobalDecl(), KmpInt32Ty, TaskEntry, TaskEntryFnInfo, Args);
 
   // TaskFunction(gtid, tt->task_data.part_id, &tt->privates, task_privates_map,
+  // tt,
+  // For taskloops:
+  // tt->task_data.lb, tt->task_data.ub, tt->task_data.st, tt->task_data.liter,
   // tt->task_data.shareds);
   auto *GtidParam = CGF.EmitLoadOfScalar(
       CGF.GetAddrOfLocalVar(&GtidArg), /*Volatile=*/false, KmpInt32Ty, Loc);
-  LValue TDBase = emitLoadOfPointerLValue(
-      CGF, CGF.GetAddrOfLocalVar(&TaskTypeArg), KmpTaskTWithPrivatesPtrQTy);
+  LValue TDBase = CGF.EmitLoadOfPointerLValue(
+      CGF.GetAddrOfLocalVar(&TaskTypeArg),
+      KmpTaskTWithPrivatesPtrQTy->castAs<PointerType>());
   auto *KmpTaskTWithPrivatesQTyRD =
       cast<RecordDecl>(KmpTaskTWithPrivatesQTy->getAsTagDecl());
   LValue Base =
@@ -2595,7 +3303,7 @@ emitProxyTaskFunction(CodeGenModule &CGM, SourceLocation Loc,
   auto *KmpTaskTQTyRD = cast<RecordDecl>(KmpTaskTQTy->getAsTagDecl());
   auto PartIdFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTPartId);
   auto PartIdLVal = CGF.EmitLValueForField(Base, *PartIdFI);
-  auto *PartidParam = CGF.EmitLoadOfLValue(PartIdLVal, Loc).getScalarVal();
+  auto *PartidParam = PartIdLVal.getPointer();
 
   auto SharedsFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTShareds);
   auto SharedsLVal = CGF.EmitLValueForField(Base, *SharedsFI);
@@ -2609,12 +3317,37 @@ emitProxyTaskFunction(CodeGenModule &CGM, SourceLocation Loc,
     auto PrivatesLVal = CGF.EmitLValueForField(TDBase, *PrivatesFI);
     PrivatesParam = CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
         PrivatesLVal.getPointer(), CGF.VoidPtrTy);
-  } else {
+  } else
     PrivatesParam = llvm::ConstantPointerNull::get(CGF.VoidPtrTy);
+
+  llvm::Value *CommonArgs[] = {GtidParam, PartidParam, PrivatesParam,
+                               TaskPrivatesMap,
+                               CGF.Builder
+                                   .CreatePointerBitCastOrAddrSpaceCast(
+                                       TDBase.getAddress(), CGF.VoidPtrTy)
+                                   .getPointer()};
+  SmallVector<llvm::Value *, 16> CallArgs(std::begin(CommonArgs),
+                                          std::end(CommonArgs));
+  if (isOpenMPTaskLoopDirective(Kind)) {
+    auto LBFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTLowerBound);
+    auto LBLVal = CGF.EmitLValueForField(Base, *LBFI);
+    auto *LBParam = CGF.EmitLoadOfLValue(LBLVal, Loc).getScalarVal();
+    auto UBFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTUpperBound);
+    auto UBLVal = CGF.EmitLValueForField(Base, *UBFI);
+    auto *UBParam = CGF.EmitLoadOfLValue(UBLVal, Loc).getScalarVal();
+    auto StFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTStride);
+    auto StLVal = CGF.EmitLValueForField(Base, *StFI);
+    auto *StParam = CGF.EmitLoadOfLValue(StLVal, Loc).getScalarVal();
+    auto LIFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTLastIter);
+    auto LILVal = CGF.EmitLValueForField(Base, *LIFI);
+    auto *LIParam = CGF.EmitLoadOfLValue(LILVal, Loc).getScalarVal();
+    CallArgs.push_back(LBParam);
+    CallArgs.push_back(UBParam);
+    CallArgs.push_back(StParam);
+    CallArgs.push_back(LIParam);
   }
+  CallArgs.push_back(SharedsParam);
 
-  llvm::Value *CallArgs[] = {GtidParam, PartidParam, PrivatesParam,
-                             TaskPrivatesMap, SharedsParam};
   CGF.EmitCallOrInvoke(TaskFunction, CallArgs);
   CGF.EmitStoreThroughLValue(
       RValue::get(CGF.Builder.getInt32(/*C=*/0)),
@@ -2638,8 +3371,7 @@ static llvm::Value *emitDestructorsFunction(CodeGenModule &CGM,
   Args.push_back(&TaskTypeArg);
   FunctionType::ExtInfo Info;
   auto &DestructorFnInfo =
-      CGM.getTypes().arrangeFreeFunctionDeclaration(KmpInt32Ty, Args, Info,
-                                                    /*isVariadic=*/false);
+      CGM.getTypes().arrangeBuiltinFunctionDeclaration(KmpInt32Ty, Args);
   auto *DestructorFnTy = CGM.getTypes().GetFunctionType(DestructorFnInfo);
   auto *DestructorFn =
       llvm::Function::Create(DestructorFnTy, llvm::GlobalValue::InternalLinkage,
@@ -2651,8 +3383,9 @@ static llvm::Value *emitDestructorsFunction(CodeGenModule &CGM,
   CGF.StartFunction(GlobalDecl(), KmpInt32Ty, DestructorFn, DestructorFnInfo,
                     Args);
 
-  LValue Base = emitLoadOfPointerLValue(
-      CGF, CGF.GetAddrOfLocalVar(&TaskTypeArg), KmpTaskTWithPrivatesPtrQTy);
+  LValue Base = CGF.EmitLoadOfPointerLValue(
+      CGF.GetAddrOfLocalVar(&TaskTypeArg),
+      KmpTaskTWithPrivatesPtrQTy->castAs<PointerType>());
   auto *KmpTaskTWithPrivatesQTyRD =
       cast<RecordDecl>(KmpTaskTWithPrivatesQTy->getAsTagDecl());
   auto FI = std::next(KmpTaskTWithPrivatesQTyRD->field_begin());
@@ -2682,6 +3415,7 @@ static llvm::Value *
 emitTaskPrivateMappingFunction(CodeGenModule &CGM, SourceLocation Loc,
                                ArrayRef<const Expr *> PrivateVars,
                                ArrayRef<const Expr *> FirstprivateVars,
+                               ArrayRef<const Expr *> LastprivateVars,
                                QualType PrivatesQTy,
                                ArrayRef<PrivateDataTy> Privates) {
   auto &C = CGM.getContext();
@@ -2712,10 +3446,18 @@ emitTaskPrivateMappingFunction(CodeGenModule &CGM, SourceLocation Loc,
     PrivateVarsPos[VD] = Counter;
     ++Counter;
   }
-  FunctionType::ExtInfo Info;
+  for (auto *E: LastprivateVars) {
+    Args.push_back(ImplicitParamDecl::Create(
+        C, /*DC=*/nullptr, Loc,
+        /*Id=*/nullptr, C.getPointerType(C.getPointerType(E->getType()))
+                            .withConst()
+                            .withRestrict()));
+    auto *VD = cast<VarDecl>(cast<DeclRefExpr>(E)->getDecl());
+    PrivateVarsPos[VD] = Counter;
+    ++Counter;
+  }
   auto &TaskPrivatesMapFnInfo =
-      CGM.getTypes().arrangeFreeFunctionDeclaration(C.VoidTy, Args, Info,
-                                                    /*isVariadic=*/false);
+      CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
   auto *TaskPrivatesMapTy =
       CGM.getTypes().GetFunctionType(TaskPrivatesMapFnInfo);
   auto *TaskPrivatesMap = llvm::Function::Create(
@@ -2730,16 +3472,17 @@ emitTaskPrivateMappingFunction(CodeGenModule &CGM, SourceLocation Loc,
                     TaskPrivatesMapFnInfo, Args);
 
   // *privi = &.privates.privi;
-  LValue Base = emitLoadOfPointerLValue(
-      CGF, CGF.GetAddrOfLocalVar(&TaskPrivatesArg), TaskPrivatesArg.getType());
+  LValue Base = CGF.EmitLoadOfPointerLValue(
+      CGF.GetAddrOfLocalVar(&TaskPrivatesArg),
+      TaskPrivatesArg.getType()->castAs<PointerType>());
   auto *PrivatesQTyRD = cast<RecordDecl>(PrivatesQTy->getAsTagDecl());
   Counter = 0;
   for (auto *Field : PrivatesQTyRD->fields()) {
     auto FieldLVal = CGF.EmitLValueForField(Base, Field);
     auto *VD = Args[PrivateVarsPos[Privates[Counter].second.Original]];
     auto RefLVal = CGF.MakeAddrLValue(CGF.GetAddrOfLocalVar(VD), VD->getType());
-    auto RefLoadLVal =
-        emitLoadOfPointerLValue(CGF, RefLVal.getAddress(), RefLVal.getType());
+    auto RefLoadLVal = CGF.EmitLoadOfPointerLValue(
+        RefLVal.getAddress(), RefLVal.getType()->castAs<PointerType>());
     CGF.EmitStoreOfScalar(FieldLVal.getPointer(), RefLoadLVal);
     ++Counter;
   }
@@ -2752,23 +3495,199 @@ static int array_pod_sort_comparator(const PrivateDataTy *P1,
   return P1->first < P2->first ? 1 : (P2->first < P1->first ? -1 : 0);
 }
 
-void CGOpenMPRuntime::emitTaskCall(
-    CodeGenFunction &CGF, SourceLocation Loc, const OMPExecutableDirective &D,
-    bool Tied, llvm::PointerIntPair<llvm::Value *, 1, bool> Final,
-    llvm::Value *TaskFunction, QualType SharedsTy, Address Shareds,
-    const Expr *IfCond, ArrayRef<const Expr *> PrivateVars,
-    ArrayRef<const Expr *> PrivateCopies,
-    ArrayRef<const Expr *> FirstprivateVars,
-    ArrayRef<const Expr *> FirstprivateCopies,
-    ArrayRef<const Expr *> FirstprivateInits,
-    ArrayRef<std::pair<OpenMPDependClauseKind, const Expr *>> Dependences) {
-  if (!CGF.HaveInsertPoint())
-    return;
+/// Emit initialization for private variables in task-based directives.
+static void emitPrivatesInit(CodeGenFunction &CGF,
+                             const OMPExecutableDirective &D,
+                             Address KmpTaskSharedsPtr, LValue TDBase,
+                             const RecordDecl *KmpTaskTWithPrivatesQTyRD,
+                             QualType SharedsTy, QualType SharedsPtrTy,
+                             const OMPTaskDataTy &Data,
+                             ArrayRef<PrivateDataTy> Privates, bool ForDup) {
+  auto &C = CGF.getContext();
+  auto FI = std::next(KmpTaskTWithPrivatesQTyRD->field_begin());
+  LValue PrivatesBase = CGF.EmitLValueForField(TDBase, *FI);
+  LValue SrcBase;
+  if (!Data.FirstprivateVars.empty()) {
+    SrcBase = CGF.MakeAddrLValue(
+        CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
+            KmpTaskSharedsPtr, CGF.ConvertTypeForMem(SharedsPtrTy)),
+        SharedsTy);
+  }
+  CodeGenFunction::CGCapturedStmtInfo CapturesInfo(
+      cast<CapturedStmt>(*D.getAssociatedStmt()));
+  FI = cast<RecordDecl>(FI->getType()->getAsTagDecl())->field_begin();
+  for (auto &&Pair : Privates) {
+    auto *VD = Pair.second.PrivateCopy;
+    auto *Init = VD->getAnyInitializer();
+    if (Init && (!ForDup || (isa<CXXConstructExpr>(Init) &&
+                             !CGF.isTrivialInitializer(Init)))) {
+      LValue PrivateLValue = CGF.EmitLValueForField(PrivatesBase, *FI);
+      if (auto *Elem = Pair.second.PrivateElemInit) {
+        auto *OriginalVD = Pair.second.Original;
+        auto *SharedField = CapturesInfo.lookup(OriginalVD);
+        auto SharedRefLValue = CGF.EmitLValueForField(SrcBase, SharedField);
+        SharedRefLValue = CGF.MakeAddrLValue(
+            Address(SharedRefLValue.getPointer(), C.getDeclAlign(OriginalVD)),
+            SharedRefLValue.getType(), AlignmentSource::Decl);
+        QualType Type = OriginalVD->getType();
+        if (Type->isArrayType()) {
+          // Initialize firstprivate array.
+          if (!isa<CXXConstructExpr>(Init) || CGF.isTrivialInitializer(Init)) {
+            // Perform simple memcpy.
+            CGF.EmitAggregateAssign(PrivateLValue.getAddress(),
+                                    SharedRefLValue.getAddress(), Type);
+          } else {
+            // Initialize firstprivate array using element-by-element
+            // intialization.
+            CGF.EmitOMPAggregateAssign(
+                PrivateLValue.getAddress(), SharedRefLValue.getAddress(), Type,
+                [&CGF, Elem, Init, &CapturesInfo](Address DestElement,
+                                                  Address SrcElement) {
+                  // Clean up any temporaries needed by the initialization.
+                  CodeGenFunction::OMPPrivateScope InitScope(CGF);
+                  InitScope.addPrivate(
+                      Elem, [SrcElement]() -> Address { return SrcElement; });
+                  (void)InitScope.Privatize();
+                  // Emit initialization for single element.
+                  CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(
+                      CGF, &CapturesInfo);
+                  CGF.EmitAnyExprToMem(Init, DestElement,
+                                       Init->getType().getQualifiers(),
+                                       /*IsInitializer=*/false);
+                });
+          }
+        } else {
+          CodeGenFunction::OMPPrivateScope InitScope(CGF);
+          InitScope.addPrivate(Elem, [SharedRefLValue]() -> Address {
+            return SharedRefLValue.getAddress();
+          });
+          (void)InitScope.Privatize();
+          CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CapturesInfo);
+          CGF.EmitExprAsInit(Init, VD, PrivateLValue,
+                             /*capturedByInit=*/false);
+        }
+      } else
+        CGF.EmitExprAsInit(Init, VD, PrivateLValue, /*capturedByInit=*/false);
+    }
+    ++FI;
+  }
+}
+
+/// Check if duplication function is required for taskloops.
+static bool checkInitIsRequired(CodeGenFunction &CGF,
+                                ArrayRef<PrivateDataTy> Privates) {
+  bool InitRequired = false;
+  for (auto &&Pair : Privates) {
+    auto *VD = Pair.second.PrivateCopy;
+    auto *Init = VD->getAnyInitializer();
+    InitRequired = InitRequired || (Init && isa<CXXConstructExpr>(Init) &&
+                                    !CGF.isTrivialInitializer(Init));
+  }
+  return InitRequired;
+}
+
+
+/// Emit task_dup function (for initialization of
+/// private/firstprivate/lastprivate vars and last_iter flag)
+/// \code
+/// void __task_dup_entry(kmp_task_t *task_dst, const kmp_task_t *task_src, int
+/// lastpriv) {
+/// // setup lastprivate flag
+///    task_dst->last = lastpriv;
+/// // could be constructor calls here...
+/// }
+/// \endcode
+static llvm::Value *
+emitTaskDupFunction(CodeGenModule &CGM, SourceLocation Loc,
+                    const OMPExecutableDirective &D,
+                    QualType KmpTaskTWithPrivatesPtrQTy,
+                    const RecordDecl *KmpTaskTWithPrivatesQTyRD,
+                    const RecordDecl *KmpTaskTQTyRD, QualType SharedsTy,
+                    QualType SharedsPtrTy, const OMPTaskDataTy &Data,
+                    ArrayRef<PrivateDataTy> Privates, bool WithLastIter) {
+  auto &C = CGM.getContext();
+  FunctionArgList Args;
+  ImplicitParamDecl DstArg(C, /*DC=*/nullptr, Loc,
+                           /*Id=*/nullptr, KmpTaskTWithPrivatesPtrQTy);
+  ImplicitParamDecl SrcArg(C, /*DC=*/nullptr, Loc,
+                           /*Id=*/nullptr, KmpTaskTWithPrivatesPtrQTy);
+  ImplicitParamDecl LastprivArg(C, /*DC=*/nullptr, Loc,
+                                /*Id=*/nullptr, C.IntTy);
+  Args.push_back(&DstArg);
+  Args.push_back(&SrcArg);
+  Args.push_back(&LastprivArg);
+  auto &TaskDupFnInfo =
+      CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
+  auto *TaskDupTy = CGM.getTypes().GetFunctionType(TaskDupFnInfo);
+  auto *TaskDup =
+      llvm::Function::Create(TaskDupTy, llvm::GlobalValue::InternalLinkage,
+                             ".omp_task_dup.", &CGM.getModule());
+  CGM.SetInternalFunctionAttributes(/*D=*/nullptr, TaskDup, TaskDupFnInfo);
+  CodeGenFunction CGF(CGM);
+  CGF.disableDebugInfo();
+  CGF.StartFunction(GlobalDecl(), C.VoidTy, TaskDup, TaskDupFnInfo, Args);
+
+  LValue TDBase = CGF.EmitLoadOfPointerLValue(
+      CGF.GetAddrOfLocalVar(&DstArg),
+      KmpTaskTWithPrivatesPtrQTy->castAs<PointerType>());
+  // task_dst->liter = lastpriv;
+  if (WithLastIter) {
+    auto LIFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTLastIter);
+    LValue Base = CGF.EmitLValueForField(
+        TDBase, *KmpTaskTWithPrivatesQTyRD->field_begin());
+    LValue LILVal = CGF.EmitLValueForField(Base, *LIFI);
+    llvm::Value *Lastpriv = CGF.EmitLoadOfScalar(
+        CGF.GetAddrOfLocalVar(&LastprivArg), /*Volatile=*/false, C.IntTy, Loc);
+    CGF.EmitStoreOfScalar(Lastpriv, LILVal);
+  }
+
+  // Emit initial values for private copies (if any).
+  assert(!Privates.empty());
+  Address KmpTaskSharedsPtr = Address::invalid();
+  if (!Data.FirstprivateVars.empty()) {
+    LValue TDBase = CGF.EmitLoadOfPointerLValue(
+        CGF.GetAddrOfLocalVar(&SrcArg),
+        KmpTaskTWithPrivatesPtrQTy->castAs<PointerType>());
+    LValue Base = CGF.EmitLValueForField(
+        TDBase, *KmpTaskTWithPrivatesQTyRD->field_begin());
+    KmpTaskSharedsPtr = Address(
+        CGF.EmitLoadOfScalar(CGF.EmitLValueForField(
+                                 Base, *std::next(KmpTaskTQTyRD->field_begin(),
+                                                  KmpTaskTShareds)),
+                             Loc),
+        CGF.getNaturalTypeAlignment(SharedsTy));
+  }
+  emitPrivatesInit(CGF, D, KmpTaskSharedsPtr, TDBase, KmpTaskTWithPrivatesQTyRD,
+                   SharedsTy, SharedsPtrTy, Data, Privates, /*ForDup=*/true);
+  CGF.FinishFunction();
+  return TaskDup;
+}
+
+/// Checks if destructor function is required to be generated.
+/// \return true if cleanups are required, false otherwise.
+static bool
+checkDestructorsRequired(const RecordDecl *KmpTaskTWithPrivatesQTyRD) {
+  bool NeedsCleanup = false;
+  auto FI = std::next(KmpTaskTWithPrivatesQTyRD->field_begin());
+  auto *PrivateRD = cast<RecordDecl>(FI->getType()->getAsTagDecl());
+  for (auto *FD : PrivateRD->fields()) {
+    NeedsCleanup = NeedsCleanup || FD->getType().isDestructedType();
+    if (NeedsCleanup)
+      break;
+  }
+  return NeedsCleanup;
+}
+
+CGOpenMPRuntime::TaskResultTy
+CGOpenMPRuntime::emitTaskInit(CodeGenFunction &CGF, SourceLocation Loc,
+                              const OMPExecutableDirective &D,
+                              llvm::Value *TaskFunction, QualType SharedsTy,
+                              Address Shareds, const OMPTaskDataTy &Data) {
   auto &C = CGM.getContext();
-  llvm::SmallVector<PrivateDataTy, 8> Privates;
+  llvm::SmallVector<PrivateDataTy, 4> Privates;
   // Aggregate privates and sort them by the alignment.
-  auto I = PrivateCopies.begin();
-  for (auto *E : PrivateVars) {
+  auto I = Data.PrivateCopies.begin();
+  for (auto *E : Data.PrivateVars) {
     auto *VD = cast<VarDecl>(cast<DeclRefExpr>(E)->getDecl());
     Privates.push_back(std::make_pair(
         C.getDeclAlign(VD),
@@ -2776,16 +3695,26 @@ void CGOpenMPRuntime::emitTaskCall(
                          /*PrivateElemInit=*/nullptr)));
     ++I;
   }
-  I = FirstprivateCopies.begin();
-  auto IElemInitRef = FirstprivateInits.begin();
-  for (auto *E : FirstprivateVars) {
+  I = Data.FirstprivateCopies.begin();
+  auto IElemInitRef = Data.FirstprivateInits.begin();
+  for (auto *E : Data.FirstprivateVars) {
     auto *VD = cast<VarDecl>(cast<DeclRefExpr>(E)->getDecl());
     Privates.push_back(std::make_pair(
         C.getDeclAlign(VD),
         PrivateHelpersTy(
             VD, cast<VarDecl>(cast<DeclRefExpr>(*I)->getDecl()),
             cast<VarDecl>(cast<DeclRefExpr>(*IElemInitRef)->getDecl()))));
-    ++I, ++IElemInitRef;
+    ++I;
+    ++IElemInitRef;
+  }
+  I = Data.LastprivateCopies.begin();
+  for (auto *E : Data.LastprivateVars) {
+    auto *VD = cast<VarDecl>(cast<DeclRefExpr>(E)->getDecl());
+    Privates.push_back(std::make_pair(
+        C.getDeclAlign(VD),
+        PrivateHelpersTy(VD, cast<VarDecl>(cast<DeclRefExpr>(*I)->getDecl()),
+                         /*PrivateElemInit=*/nullptr)));
+    ++I;
   }
   llvm::array_pod_sort(Privates.begin(), Privates.end(),
                        array_pod_sort_comparator);
@@ -2794,8 +3723,8 @@ void CGOpenMPRuntime::emitTaskCall(
   emitKmpRoutineEntryT(KmpInt32Ty);
   // Build type kmp_task_t (if not built yet).
   if (KmpTaskTQTy.isNull()) {
-    KmpTaskTQTy = C.getRecordType(
-        createKmpTaskTRecordDecl(CGM, KmpInt32Ty, KmpRoutineEntryPtrQTy));
+    KmpTaskTQTy = C.getRecordType(createKmpTaskTRecordDecl(
+        CGM, D.getDirectiveKind(), KmpInt32Ty, KmpRoutineEntryPtrQTy));
   }
   auto *KmpTaskTQTyRD = cast<RecordDecl>(KmpTaskTQTy->getAsTagDecl());
   // Build particular struct kmp_task_t for the given task.
@@ -2806,7 +3735,7 @@ void CGOpenMPRuntime::emitTaskCall(
       C.getPointerType(KmpTaskTWithPrivatesQTy);
   auto *KmpTaskTWithPrivatesTy = CGF.ConvertType(KmpTaskTWithPrivatesQTy);
   auto *KmpTaskTWithPrivatesPtrTy = KmpTaskTWithPrivatesTy->getPointerTo();
-  auto *KmpTaskTWithPrivatesTySize = getTypeSize(CGF, KmpTaskTWithPrivatesQTy);
+  auto *KmpTaskTWithPrivatesTySize = CGF.getTypeSize(KmpTaskTWithPrivatesQTy);
   QualType SharedsPtrTy = C.getPointerType(SharedsTy);
 
   // Emit initial values for private copies (if any).
@@ -2818,7 +3747,8 @@ void CGOpenMPRuntime::emitTaskCall(
   if (!Privates.empty()) {
     auto FI = std::next(KmpTaskTWithPrivatesQTyRD->field_begin());
     TaskPrivatesMap = emitTaskPrivateMappingFunction(
-        CGM, Loc, PrivateVars, FirstprivateVars, FI->getType(), Privates);
+        CGM, Loc, Data.PrivateVars, Data.FirstprivateVars, Data.LastprivateVars,
+        FI->getType(), Privates);
     TaskPrivatesMap = CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
         TaskPrivatesMap, TaskPrivatesMapTy);
   } else {
@@ -2828,8 +3758,9 @@ void CGOpenMPRuntime::emitTaskCall(
   // Build a proxy function kmp_int32 .omp_task_entry.(kmp_int32 gtid,
   // kmp_task_t *tt);
   auto *TaskEntry = emitProxyTaskFunction(
-      CGM, Loc, KmpInt32Ty, KmpTaskTWithPrivatesPtrQTy, KmpTaskTWithPrivatesQTy,
-      KmpTaskTQTy, SharedsPtrTy, TaskFunction, TaskPrivatesMap);
+      CGM, Loc, D.getDirectiveKind(), KmpInt32Ty, KmpTaskTWithPrivatesPtrQTy,
+      KmpTaskTWithPrivatesQTy, KmpTaskTQTy, SharedsPtrTy, TaskFunction,
+      TaskPrivatesMap);
 
   // Build call kmp_task_t * __kmpc_omp_task_alloc(ident_t *, kmp_int32 gtid,
   // kmp_int32 flags, size_t sizeof_kmp_task_t, size_t sizeof_shareds,
@@ -2837,15 +3768,27 @@ void CGOpenMPRuntime::emitTaskCall(
   // Task flags. Format is taken from
   // http://llvm.org/svn/llvm-project/openmp/trunk/runtime/src/kmp.h,
   // description of kmp_tasking_flags struct.
-  const unsigned TiedFlag = 0x1;
-  const unsigned FinalFlag = 0x2;
-  unsigned Flags = Tied ? TiedFlag : 0;
+  enum {
+    TiedFlag = 0x1,
+    FinalFlag = 0x2,
+    DestructorsFlag = 0x8,
+    PriorityFlag = 0x20
+  };
+  unsigned Flags = Data.Tied ? TiedFlag : 0;
+  bool NeedsCleanup = false;
+  if (!Privates.empty()) {
+    NeedsCleanup = checkDestructorsRequired(KmpTaskTWithPrivatesQTyRD);
+    if (NeedsCleanup)
+      Flags = Flags | DestructorsFlag;
+  }
+  if (Data.Priority.getInt())
+    Flags = Flags | PriorityFlag;
   auto *TaskFlags =
-      Final.getPointer()
-          ? CGF.Builder.CreateSelect(Final.getPointer(),
+      Data.Final.getPointer()
+          ? CGF.Builder.CreateSelect(Data.Final.getPointer(),
                                      CGF.Builder.getInt32(FinalFlag),
                                      CGF.Builder.getInt32(/*C=*/0))
-          : CGF.Builder.getInt32(Final.getInt() ? FinalFlag : 0);
+          : CGF.Builder.getInt32(Data.Final.getInt() ? FinalFlag : 0);
   TaskFlags = CGF.Builder.CreateOr(TaskFlags, CGF.Builder.getInt32(Flags));
   auto *SharedsSize = CGM.getSize(C.getTypeSizeInChars(SharedsTy));
   llvm::Value *AllocArgs[] = {emitUpdateLocation(CGF, Loc),
@@ -2875,96 +3818,71 @@ void CGOpenMPRuntime::emitTaskCall(
     CGF.EmitAggregateCopy(KmpTaskSharedsPtr, Shareds, SharedsTy);
   }
   // Emit initial values for private copies (if any).
-  bool NeedsCleanup = false;
+  TaskResultTy Result;
   if (!Privates.empty()) {
-    auto FI = std::next(KmpTaskTWithPrivatesQTyRD->field_begin());
-    auto PrivatesBase = CGF.EmitLValueForField(Base, *FI);
-    FI = cast<RecordDecl>(FI->getType()->getAsTagDecl())->field_begin();
-    LValue SharedsBase;
-    if (!FirstprivateVars.empty()) {
-      SharedsBase = CGF.MakeAddrLValue(
-          CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
-              KmpTaskSharedsPtr, CGF.ConvertTypeForMem(SharedsPtrTy)),
-          SharedsTy);
-    }
-    CodeGenFunction::CGCapturedStmtInfo CapturesInfo(
-        cast<CapturedStmt>(*D.getAssociatedStmt()));
-    for (auto &&Pair : Privates) {
-      auto *VD = Pair.second.PrivateCopy;
-      auto *Init = VD->getAnyInitializer();
-      LValue PrivateLValue = CGF.EmitLValueForField(PrivatesBase, *FI);
-      if (Init) {
-        if (auto *Elem = Pair.second.PrivateElemInit) {
-          auto *OriginalVD = Pair.second.Original;
-          auto *SharedField = CapturesInfo.lookup(OriginalVD);
-          auto SharedRefLValue =
-              CGF.EmitLValueForField(SharedsBase, SharedField);
-          SharedRefLValue = CGF.MakeAddrLValue(
-              Address(SharedRefLValue.getPointer(), C.getDeclAlign(OriginalVD)),
-              SharedRefLValue.getType(), AlignmentSource::Decl);
-          QualType Type = OriginalVD->getType();
-          if (Type->isArrayType()) {
-            // Initialize firstprivate array.
-            if (!isa<CXXConstructExpr>(Init) ||
-                CGF.isTrivialInitializer(Init)) {
-              // Perform simple memcpy.
-              CGF.EmitAggregateAssign(PrivateLValue.getAddress(),
-                                      SharedRefLValue.getAddress(), Type);
-            } else {
-              // Initialize firstprivate array using element-by-element
-              // intialization.
-              CGF.EmitOMPAggregateAssign(
-                  PrivateLValue.getAddress(), SharedRefLValue.getAddress(),
-                  Type, [&CGF, Elem, Init, &CapturesInfo](
-                            Address DestElement, Address SrcElement) {
-                    // Clean up any temporaries needed by the initialization.
-                    CodeGenFunction::OMPPrivateScope InitScope(CGF);
-                    InitScope.addPrivate(Elem, [SrcElement]() -> Address {
-                      return SrcElement;
-                    });
-                    (void)InitScope.Privatize();
-                    // Emit initialization for single element.
-                    CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(
-                        CGF, &CapturesInfo);
-                    CGF.EmitAnyExprToMem(Init, DestElement,
-                                         Init->getType().getQualifiers(),
-                                         /*IsInitializer=*/false);
-                  });
-            }
-          } else {
-            CodeGenFunction::OMPPrivateScope InitScope(CGF);
-            InitScope.addPrivate(Elem, [SharedRefLValue]() -> Address {
-              return SharedRefLValue.getAddress();
-            });
-            (void)InitScope.Privatize();
-            CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CapturesInfo);
-            CGF.EmitExprAsInit(Init, VD, PrivateLValue,
-                               /*capturedByInit=*/false);
-          }
-        } else {
-          CGF.EmitExprAsInit(Init, VD, PrivateLValue, /*capturedByInit=*/false);
-        }
-      }
-      NeedsCleanup = NeedsCleanup || FI->getType().isDestructedType();
-      ++FI;
+    emitPrivatesInit(CGF, D, KmpTaskSharedsPtr, Base, KmpTaskTWithPrivatesQTyRD,
+                     SharedsTy, SharedsPtrTy, Data, Privates,
+                     /*ForDup=*/false);
+    if (isOpenMPTaskLoopDirective(D.getDirectiveKind()) &&
+        (!Data.LastprivateVars.empty() || checkInitIsRequired(CGF, Privates))) {
+      Result.TaskDupFn = emitTaskDupFunction(
+          CGM, Loc, D, KmpTaskTWithPrivatesPtrQTy, KmpTaskTWithPrivatesQTyRD,
+          KmpTaskTQTyRD, SharedsTy, SharedsPtrTy, Data, Privates,
+          /*WithLastIter=*/!Data.LastprivateVars.empty());
     }
   }
+  // Fields of union "kmp_cmplrdata_t" for destructors and priority.
+  enum { Priority = 0, Destructors = 1 };
   // Provide pointer to function with destructors for privates.
-  llvm::Value *DestructorFn =
-      NeedsCleanup ? emitDestructorsFunction(CGM, Loc, KmpInt32Ty,
-                                             KmpTaskTWithPrivatesPtrQTy,
-                                             KmpTaskTWithPrivatesQTy)
-                   : llvm::ConstantPointerNull::get(
-                         cast<llvm::PointerType>(KmpRoutineEntryPtrTy));
-  LValue Destructor = CGF.EmitLValueForField(
-      TDBase, *std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTDestructors));
-  CGF.EmitStoreOfScalar(CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
-                            DestructorFn, KmpRoutineEntryPtrTy),
-                        Destructor);
+  auto FI = std::next(KmpTaskTQTyRD->field_begin(), Data1);
+  auto *KmpCmplrdataUD = (*FI)->getType()->getAsUnionType()->getDecl();
+  if (NeedsCleanup) {
+    llvm::Value *DestructorFn = emitDestructorsFunction(
+        CGM, Loc, KmpInt32Ty, KmpTaskTWithPrivatesPtrQTy,
+        KmpTaskTWithPrivatesQTy);
+    LValue Data1LV = CGF.EmitLValueForField(TDBase, *FI);
+    LValue DestructorsLV = CGF.EmitLValueForField(
+        Data1LV, *std::next(KmpCmplrdataUD->field_begin(), Destructors));
+    CGF.EmitStoreOfScalar(CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
+                              DestructorFn, KmpRoutineEntryPtrTy),
+                          DestructorsLV);
+  }
+  // Set priority.
+  if (Data.Priority.getInt()) {
+    LValue Data2LV = CGF.EmitLValueForField(
+        TDBase, *std::next(KmpTaskTQTyRD->field_begin(), Data2));
+    LValue PriorityLV = CGF.EmitLValueForField(
+        Data2LV, *std::next(KmpCmplrdataUD->field_begin(), Priority));
+    CGF.EmitStoreOfScalar(Data.Priority.getPointer(), PriorityLV);
+  }
+  Result.NewTask = NewTask;
+  Result.TaskEntry = TaskEntry;
+  Result.NewTaskNewTaskTTy = NewTaskNewTaskTTy;
+  Result.TDBase = TDBase;
+  Result.KmpTaskTQTyRD = KmpTaskTQTyRD;
+  return Result;
+}
+
+void CGOpenMPRuntime::emitTaskCall(CodeGenFunction &CGF, SourceLocation Loc,
+                                   const OMPExecutableDirective &D,
+                                   llvm::Value *TaskFunction,
+                                   QualType SharedsTy, Address Shareds,
+                                   const Expr *IfCond,
+                                   const OMPTaskDataTy &Data) {
+  if (!CGF.HaveInsertPoint())
+    return;
 
+  TaskResultTy Result =
+      emitTaskInit(CGF, Loc, D, TaskFunction, SharedsTy, Shareds, Data);
+  llvm::Value *NewTask = Result.NewTask;
+  llvm::Value *TaskEntry = Result.TaskEntry;
+  llvm::Value *NewTaskNewTaskTTy = Result.NewTaskNewTaskTTy;
+  LValue TDBase = Result.TDBase;
+  RecordDecl *KmpTaskTQTyRD = Result.KmpTaskTQTyRD;
+  auto &C = CGM.getContext();
   // Process list of dependences.
   Address DependenciesArray = Address::invalid();
-  unsigned NumDependencies = Dependences.size();
+  unsigned NumDependencies = Data.Dependences.size();
   if (NumDependencies) {
     // Dependence kind for RTL.
     enum RTLDependenceKindTy { DepIn = 0x01, DepInOut = 0x3 };
@@ -2981,18 +3899,18 @@ void CGOpenMPRuntime::emitTaskCall(
       addFieldToRecordDecl(C, KmpDependInfoRD, FlagsTy);
       KmpDependInfoRD->completeDefinition();
       KmpDependInfoTy = C.getRecordType(KmpDependInfoRD);
-    } else {
+    } else
       KmpDependInfoRD = cast<RecordDecl>(KmpDependInfoTy->getAsTagDecl());
-    }
     CharUnits DependencySize = C.getTypeSizeInChars(KmpDependInfoTy);
     // Define type kmp_depend_info[<Dependences.size()>];
     QualType KmpDependInfoArrayTy = C.getConstantArrayType(
         KmpDependInfoTy, llvm::APInt(/*numBits=*/64, NumDependencies),
         ArrayType::Normal, /*IndexTypeQuals=*/0);
     // kmp_depend_info[<Dependences.size()>] deps;
-    DependenciesArray = CGF.CreateMemTemp(KmpDependInfoArrayTy);
+    DependenciesArray =
+        CGF.CreateMemTemp(KmpDependInfoArrayTy, ".dep.arr.addr");
     for (unsigned i = 0; i < NumDependencies; ++i) {
-      const Expr *E = Dependences[i].second;
+      const Expr *E = Data.Dependences[i].second;
       auto Addr = CGF.EmitLValue(E);
       llvm::Value *Size;
       QualType Ty = E->getType();
@@ -3006,7 +3924,7 @@ void CGOpenMPRuntime::emitTaskCall(
         llvm::Value *UpIntPtr = CGF.Builder.CreatePtrToInt(UpAddr, CGM.SizeTy);
         Size = CGF.Builder.CreateNUWSub(UpIntPtr, LowIntPtr);
       } else
-        Size = getTypeSize(CGF, Ty);
+        Size = CGF.getTypeSize(Ty);
       auto Base = CGF.MakeAddrLValue(
           CGF.Builder.CreateConstArrayGEP(DependenciesArray, i, DependencySize),
           KmpDependInfoTy);
@@ -3022,7 +3940,7 @@ void CGOpenMPRuntime::emitTaskCall(
       CGF.EmitStoreOfScalar(Size, LenLVal);
       // deps[i].flags = <Dependences[i].first>;
       RTLDependenceKindTy DepKind;
-      switch (Dependences[i].first) {
+      switch (Data.Dependences[i].first) {
       case OMPC_DEPEND_in:
         DepKind = DepIn;
         break;
@@ -3048,8 +3966,6 @@ void CGOpenMPRuntime::emitTaskCall(
 
   // NOTE: routine and part_id fields are intialized by __kmpc_omp_task_alloc()
   // libcall.
-  // Build kmp_int32 __kmpc_omp_task(ident_t *, kmp_int32 gtid, kmp_task_t
-  // *new_task);
   // Build kmp_int32 __kmpc_omp_task_with_deps(ident_t *, kmp_int32 gtid,
   // kmp_task_t *new_task, kmp_int32 ndeps, kmp_depend_info_t *dep_list,
   // kmp_int32 ndeps_noalias, kmp_depend_info_t *noalias_dep_list) if dependence
@@ -3067,19 +3983,26 @@ void CGOpenMPRuntime::emitTaskCall(
     DepTaskArgs[5] = CGF.Builder.getInt32(0);
     DepTaskArgs[6] = llvm::ConstantPointerNull::get(CGF.VoidPtrTy);
   }
-  auto &&ThenCodeGen = [this, NumDependencies,
-                        &TaskArgs, &DepTaskArgs](CodeGenFunction &CGF) {
-    // TODO: add check for untied tasks.    
+  auto &&ThenCodeGen = [this, Loc, &Data, TDBase, KmpTaskTQTyRD,
+                        NumDependencies, &TaskArgs,
+                        &DepTaskArgs](CodeGenFunction &CGF, PrePostActionTy &) {
+    if (!Data.Tied) {
+      auto PartIdFI = std::next(KmpTaskTQTyRD->field_begin(), KmpTaskTPartId);
+      auto PartIdLVal = CGF.EmitLValueForField(TDBase, *PartIdFI);
+      CGF.EmitStoreOfScalar(CGF.Builder.getInt32(0), PartIdLVal);
+    }
     if (NumDependencies) {
-      CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_task_with_deps),
-                          DepTaskArgs);
+      CGF.EmitRuntimeCall(
+          createRuntimeFunction(OMPRTL__kmpc_omp_task_with_deps), DepTaskArgs);
     } else {
       CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_task),
                           TaskArgs);
     }
+    // Check if parent region is untied and build return for untied task;
+    if (auto *Region =
+            dyn_cast_or_null<CGOpenMPRegionInfo>(CGF.CapturedStmtInfo))
+      Region->emitUntiedSwitch(CGF);
   };
-  typedef CallEndCleanup<std::extent<decltype(TaskArgs)>::value>
-      IfCallEndCleanup;
 
   llvm::Value *DepWaitTaskArgs[6];
   if (NumDependencies) {
@@ -3090,40 +4013,111 @@ void CGOpenMPRuntime::emitTaskCall(
     DepWaitTaskArgs[4] = CGF.Builder.getInt32(0);
     DepWaitTaskArgs[5] = llvm::ConstantPointerNull::get(CGF.VoidPtrTy);
   }
-  auto &&ElseCodeGen = [this, &TaskArgs, ThreadID, NewTaskNewTaskTTy, TaskEntry,
-                        NumDependencies, &DepWaitTaskArgs](CodeGenFunction &CGF) {
+  auto &&ElseCodeGen = [&TaskArgs, ThreadID, NewTaskNewTaskTTy, TaskEntry,
+                        NumDependencies, &DepWaitTaskArgs](CodeGenFunction &CGF,
+                                                           PrePostActionTy &) {
+    auto &RT = CGF.CGM.getOpenMPRuntime();
     CodeGenFunction::RunCleanupsScope LocalScope(CGF);
     // Build void __kmpc_omp_wait_deps(ident_t *, kmp_int32 gtid,
     // kmp_int32 ndeps, kmp_depend_info_t *dep_list, kmp_int32
     // ndeps_noalias, kmp_depend_info_t *noalias_dep_list); if dependence info
     // is specified.
     if (NumDependencies)
-      CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_wait_deps),
+      CGF.EmitRuntimeCall(RT.createRuntimeFunction(OMPRTL__kmpc_omp_wait_deps),
                           DepWaitTaskArgs);
+    // Call proxy_task_entry(gtid, new_task);
+    auto &&CodeGen = [TaskEntry, ThreadID, NewTaskNewTaskTTy](
+        CodeGenFunction &CGF, PrePostActionTy &Action) {
+      Action.Enter(CGF);
+      llvm::Value *OutlinedFnArgs[] = {ThreadID, NewTaskNewTaskTTy};
+      CGF.EmitCallOrInvoke(TaskEntry, OutlinedFnArgs);
+    };
+
     // Build void __kmpc_omp_task_begin_if0(ident_t *, kmp_int32 gtid,
     // kmp_task_t *new_task);
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_task_begin_if0),
-                        TaskArgs);
     // Build void __kmpc_omp_task_complete_if0(ident_t *, kmp_int32 gtid,
     // kmp_task_t *new_task);
-    CGF.EHStack.pushCleanup<IfCallEndCleanup>(
-        NormalAndEHCleanup,
-        createRuntimeFunction(OMPRTL__kmpc_omp_task_complete_if0),
-        llvm::makeArrayRef(TaskArgs));
-
-    // Call proxy_task_entry(gtid, new_task);
-    llvm::Value *OutlinedFnArgs[] = {ThreadID, NewTaskNewTaskTTy};
-    CGF.EmitCallOrInvoke(TaskEntry, OutlinedFnArgs);
+    RegionCodeGenTy RCG(CodeGen);
+    CommonActionTy Action(
+        RT.createRuntimeFunction(OMPRTL__kmpc_omp_task_begin_if0), TaskArgs,
+        RT.createRuntimeFunction(OMPRTL__kmpc_omp_task_complete_if0), TaskArgs);
+    RCG.setAction(Action);
+    RCG(CGF);
   };
 
-  if (IfCond) {
+  if (IfCond)
     emitOMPIfClause(CGF, IfCond, ThenCodeGen, ElseCodeGen);
-  } else {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    ThenCodeGen(CGF);
+  else {
+    RegionCodeGenTy ThenRCG(ThenCodeGen);
+    ThenRCG(CGF);
   }
 }
 
+void CGOpenMPRuntime::emitTaskLoopCall(CodeGenFunction &CGF, SourceLocation Loc,
+                                       const OMPLoopDirective &D,
+                                       llvm::Value *TaskFunction,
+                                       QualType SharedsTy, Address Shareds,
+                                       const Expr *IfCond,
+                                       const OMPTaskDataTy &Data) {
+  if (!CGF.HaveInsertPoint())
+    return;
+  TaskResultTy Result =
+      emitTaskInit(CGF, Loc, D, TaskFunction, SharedsTy, Shareds, Data);
+  // NOTE: routine and part_id fields are intialized by __kmpc_omp_task_alloc()
+  // libcall.
+  // Call to void __kmpc_taskloop(ident_t *loc, int gtid, kmp_task_t *task, int
+  // if_val, kmp_uint64 *lb, kmp_uint64 *ub, kmp_int64 st, int nogroup, int
+  // sched, kmp_uint64 grainsize, void *task_dup);
+  llvm::Value *ThreadID = getThreadID(CGF, Loc);
+  llvm::Value *UpLoc = emitUpdateLocation(CGF, Loc);
+  llvm::Value *IfVal;
+  if (IfCond) {
+    IfVal = CGF.Builder.CreateIntCast(CGF.EvaluateExprAsBool(IfCond), CGF.IntTy,
+                                      /*isSigned=*/true);
+  } else
+    IfVal = llvm::ConstantInt::getSigned(CGF.IntTy, /*V=*/1);
+
+  LValue LBLVal = CGF.EmitLValueForField(
+      Result.TDBase,
+      *std::next(Result.KmpTaskTQTyRD->field_begin(), KmpTaskTLowerBound));
+  auto *LBVar =
+      cast<VarDecl>(cast<DeclRefExpr>(D.getLowerBoundVariable())->getDecl());
+  CGF.EmitAnyExprToMem(LBVar->getInit(), LBLVal.getAddress(), LBLVal.getQuals(),
+                       /*IsInitializer=*/true);
+  LValue UBLVal = CGF.EmitLValueForField(
+      Result.TDBase,
+      *std::next(Result.KmpTaskTQTyRD->field_begin(), KmpTaskTUpperBound));
+  auto *UBVar =
+      cast<VarDecl>(cast<DeclRefExpr>(D.getUpperBoundVariable())->getDecl());
+  CGF.EmitAnyExprToMem(UBVar->getInit(), UBLVal.getAddress(), UBLVal.getQuals(),
+                       /*IsInitializer=*/true);
+  LValue StLVal = CGF.EmitLValueForField(
+      Result.TDBase,
+      *std::next(Result.KmpTaskTQTyRD->field_begin(), KmpTaskTStride));
+  auto *StVar =
+      cast<VarDecl>(cast<DeclRefExpr>(D.getStrideVariable())->getDecl());
+  CGF.EmitAnyExprToMem(StVar->getInit(), StLVal.getAddress(), StLVal.getQuals(),
+                       /*IsInitializer=*/true);
+  enum { NoSchedule = 0, Grainsize = 1, NumTasks = 2 };
+  llvm::Value *TaskArgs[] = {
+      UpLoc, ThreadID, Result.NewTask, IfVal, LBLVal.getPointer(),
+      UBLVal.getPointer(), CGF.EmitLoadOfScalar(StLVal, SourceLocation()),
+      llvm::ConstantInt::getSigned(CGF.IntTy, Data.Nogroup ? 1 : 0),
+      llvm::ConstantInt::getSigned(
+          CGF.IntTy, Data.Schedule.getPointer()
+                         ? Data.Schedule.getInt() ? NumTasks : Grainsize
+                         : NoSchedule),
+      Data.Schedule.getPointer()
+          ? CGF.Builder.CreateIntCast(Data.Schedule.getPointer(), CGF.Int64Ty,
+                                      /*isSigned=*/false)
+          : llvm::ConstantInt::get(CGF.Int64Ty, /*V=*/0),
+      Result.TaskDupFn
+          ? CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(Result.TaskDupFn,
+                                                            CGF.VoidPtrTy)
+          : llvm::ConstantPointerNull::get(CGF.VoidPtrTy)};
+  CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_taskloop), TaskArgs);
+}
+
 /// \brief Emit reduction operation for each element of array (required for
 /// array sections) LHS op = RHS.
 /// \param Type Type of array.
@@ -3204,6 +4198,26 @@ static void EmitOMPAggregateReduction(
   CGF.EmitBlock(DoneBB, /*IsFinished=*/true);
 }
 
+/// Emit reduction combiner. If the combiner is a simple expression emit it as
+/// is, otherwise consider it as combiner of UDR decl and emit it as a call of
+/// UDR combiner function.
+static void emitReductionCombiner(CodeGenFunction &CGF,
+                                  const Expr *ReductionOp) {
+  if (auto *CE = dyn_cast<CallExpr>(ReductionOp))
+    if (auto *OVE = dyn_cast<OpaqueValueExpr>(CE->getCallee()))
+      if (auto *DRE =
+              dyn_cast<DeclRefExpr>(OVE->getSourceExpr()->IgnoreImpCasts()))
+        if (auto *DRD = dyn_cast<OMPDeclareReductionDecl>(DRE->getDecl())) {
+          std::pair<llvm::Function *, llvm::Function *> Reduction =
+              CGF.CGM.getOpenMPRuntime().getUserDefinedReduction(DRD);
+          RValue Func = RValue::get(Reduction.first);
+          CodeGenFunction::OpaqueValueMapping Map(CGF, OVE, Func);
+          CGF.EmitIgnoredExpr(ReductionOp);
+          return;
+        }
+  CGF.EmitIgnoredExpr(ReductionOp);
+}
+
 static llvm::Value *emitReductionFunction(CodeGenModule &CGM,
                                           llvm::Type *ArgsType,
                                           ArrayRef<const Expr *> Privates,
@@ -3220,9 +4234,7 @@ static llvm::Value *emitReductionFunction(CodeGenModule &CGM,
                            C.VoidPtrTy);
   Args.push_back(&LHSArg);
   Args.push_back(&RHSArg);
-  FunctionType::ExtInfo EI;
-  auto &CGFI = CGM.getTypes().arrangeFreeFunctionDeclaration(
-      C.VoidTy, Args, EI, /*isVariadic=*/false);
+  auto &CGFI = CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
   auto *Fn = llvm::Function::Create(
       CGM.getTypes().GetFunctionType(CGFI), llvm::GlobalValue::InternalLinkage,
       ".omp.reduction.reduction_func", &CGM.getModule());
@@ -3255,17 +4267,16 @@ static llvm::Value *emitReductionFunction(CodeGenModule &CGM,
       return emitAddrOfVarFromArray(CGF, LHS, Idx, LHSVar);
     });
     QualType PrivTy = (*IPriv)->getType();
-    if (PrivTy->isArrayType()) {
+    if (PrivTy->isVariablyModifiedType()) {
       // Get array size and emit VLA type.
       ++Idx;
       Address Elem =
           CGF.Builder.CreateConstArrayGEP(LHS, Idx, CGF.getPointerSize());
       llvm::Value *Ptr = CGF.Builder.CreateLoad(Elem);
+      auto *VLA = CGF.getContext().getAsVariableArrayType(PrivTy);
+      auto *OVE = cast<OpaqueValueExpr>(VLA->getSizeExpr());
       CodeGenFunction::OpaqueValueMapping OpaqueMap(
-          CGF,
-          cast<OpaqueValueExpr>(
-              CGF.getContext().getAsVariableArrayType(PrivTy)->getSizeExpr()),
-          RValue::get(CGF.Builder.CreatePtrToInt(Ptr, CGF.SizeTy)));
+          CGF, OVE, RValue::get(CGF.Builder.CreatePtrToInt(Ptr, CGF.SizeTy)));
       CGF.EmitVariablyModifiedType(PrivTy);
     }
   }
@@ -3278,20 +4289,42 @@ static llvm::Value *emitReductionFunction(CodeGenModule &CGM,
       // Emit reduction for array section.
       auto *LHSVar = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
       auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
-      EmitOMPAggregateReduction(CGF, (*IPriv)->getType(), LHSVar, RHSVar,
-                                [=](CodeGenFunction &CGF, const Expr *,
-                                    const Expr *,
-                                    const Expr *) { CGF.EmitIgnoredExpr(E); });
+      EmitOMPAggregateReduction(
+          CGF, (*IPriv)->getType(), LHSVar, RHSVar,
+          [=](CodeGenFunction &CGF, const Expr *, const Expr *, const Expr *) {
+            emitReductionCombiner(CGF, E);
+          });
     } else
       // Emit reduction for array subscript or single variable.
-      CGF.EmitIgnoredExpr(E);
-    ++IPriv, ++ILHS, ++IRHS;
+      emitReductionCombiner(CGF, E);
+    ++IPriv;
+    ++ILHS;
+    ++IRHS;
   }
   Scope.ForceCleanup();
   CGF.FinishFunction();
   return Fn;
 }
 
+static void emitSingleReductionCombiner(CodeGenFunction &CGF,
+                                        const Expr *ReductionOp,
+                                        const Expr *PrivateRef,
+                                        const DeclRefExpr *LHS,
+                                        const DeclRefExpr *RHS) {
+  if (PrivateRef->getType()->isArrayType()) {
+    // Emit reduction for array section.
+    auto *LHSVar = cast<VarDecl>(LHS->getDecl());
+    auto *RHSVar = cast<VarDecl>(RHS->getDecl());
+    EmitOMPAggregateReduction(
+        CGF, PrivateRef->getType(), LHSVar, RHSVar,
+        [=](CodeGenFunction &CGF, const Expr *, const Expr *, const Expr *) {
+          emitReductionCombiner(CGF, ReductionOp);
+        });
+  } else
+    // Emit reduction for array subscript or single variable.
+    emitReductionCombiner(CGF, ReductionOp);
+}
+
 void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
                                     ArrayRef<const Expr *> Privates,
                                     ArrayRef<const Expr *> LHSExprs,
@@ -3343,16 +4376,11 @@ void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
     auto ILHS = LHSExprs.begin();
     auto IRHS = RHSExprs.begin();
     for (auto *E : ReductionOps) {
-      if ((*IPriv)->getType()->isArrayType()) {
-        auto *LHSVar = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
-        auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
-        EmitOMPAggregateReduction(
-            CGF, (*IPriv)->getType(), LHSVar, RHSVar,
-            [=](CodeGenFunction &CGF, const Expr *, const Expr *,
-                const Expr *) { CGF.EmitIgnoredExpr(E); });
-      } else
-        CGF.EmitIgnoredExpr(E);
-      ++IPriv, ++ILHS, ++IRHS;
+      emitSingleReductionCombiner(CGF, E, *IPriv, cast<DeclRefExpr>(*ILHS),
+                                  cast<DeclRefExpr>(*IRHS));
+      ++IPriv;
+      ++ILHS;
+      ++IRHS;
     }
     return;
   }
@@ -3361,7 +4389,7 @@ void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
   // void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};
   auto Size = RHSExprs.size();
   for (auto *E : Privates) {
-    if (E->getType()->isArrayType())
+    if (E->getType()->isVariablyModifiedType())
       // Reserve place for array size.
       ++Size;
   }
@@ -3380,20 +4408,18 @@ void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
         CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
             CGF.EmitLValue(RHSExprs[I]).getPointer(), CGF.VoidPtrTy),
         Elem);
-    if ((*IPriv)->getType()->isArrayType()) {
+    if ((*IPriv)->getType()->isVariablyModifiedType()) {
       // Store array size.
       ++Idx;
       Elem = CGF.Builder.CreateConstArrayGEP(ReductionList, Idx,
                                              CGF.getPointerSize());
-      CGF.Builder.CreateStore(
-          CGF.Builder.CreateIntToPtr(
-              CGF.Builder.CreateIntCast(
-                  CGF.getVLASize(CGF.getContext().getAsVariableArrayType(
-                                     (*IPriv)->getType()))
-                      .first,
-                  CGF.SizeTy, /*isSigned=*/false),
-              CGF.VoidPtrTy),
-          Elem);
+      llvm::Value *Size = CGF.Builder.CreateIntCast(
+          CGF.getVLASize(
+                 CGF.getContext().getAsVariableArrayType((*IPriv)->getType()))
+              .first,
+          CGF.SizeTy, /*isSigned=*/false);
+      CGF.Builder.CreateStore(CGF.Builder.CreateIntToPtr(Size, CGF.VoidPtrTy),
+                              Elem);
     }
   }
 
@@ -3407,11 +4433,9 @@ void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
 
   // 4. Build res = __kmpc_reduce{_nowait}(<loc>, <gtid>, <n>, sizeof(RedList),
   // RedList, reduce_func, &<lock>);
-  auto *IdentTLoc = emitUpdateLocation(
-      CGF, Loc,
-      static_cast<OpenMPLocationFlags>(OMP_IDENT_KMPC | OMP_ATOMIC_REDUCE));
+  auto *IdentTLoc = emitUpdateLocation(CGF, Loc, OMP_ATOMIC_REDUCE);
   auto *ThreadId = getThreadID(CGF, Loc);
-  auto *ReductionArrayTySize = getTypeSize(CGF, ReductionArrayTy);
+  auto *ReductionArrayTySize = CGF.getTypeSize(ReductionArrayTy);
   auto *RL =
     CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(ReductionList.getPointer(),
                                                     CGF.VoidPtrTy);
@@ -3443,38 +4467,33 @@ void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
   SwInst->addCase(CGF.Builder.getInt32(1), Case1BB);
   CGF.EmitBlock(Case1BB);
 
-  {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    // Add emission of __kmpc_end_reduce{_nowait}(<loc>, <gtid>, &<lock>);
-    llvm::Value *EndArgs[] = {
-        IdentTLoc, // ident_t *<loc>
-        ThreadId,  // i32 <gtid>
-        Lock       // kmp_critical_name *&<lock>
-    };
-    CGF.EHStack
-        .pushCleanup<CallEndCleanup<std::extent<decltype(EndArgs)>::value>>(
-            NormalAndEHCleanup,
-            createRuntimeFunction(WithNowait ? OMPRTL__kmpc_end_reduce_nowait
-                                             : OMPRTL__kmpc_end_reduce),
-            llvm::makeArrayRef(EndArgs));
+  // Add emission of __kmpc_end_reduce{_nowait}(<loc>, <gtid>, &<lock>);
+  llvm::Value *EndArgs[] = {
+      IdentTLoc, // ident_t *<loc>
+      ThreadId,  // i32 <gtid>
+      Lock       // kmp_critical_name *&<lock>
+  };
+  auto &&CodeGen = [&Privates, &LHSExprs, &RHSExprs, &ReductionOps](
+      CodeGenFunction &CGF, PrePostActionTy &Action) {
     auto IPriv = Privates.begin();
     auto ILHS = LHSExprs.begin();
     auto IRHS = RHSExprs.begin();
     for (auto *E : ReductionOps) {
-      if ((*IPriv)->getType()->isArrayType()) {
-        // Emit reduction for array section.
-        auto *LHSVar = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
-        auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
-        EmitOMPAggregateReduction(
-            CGF, (*IPriv)->getType(), LHSVar, RHSVar,
-            [=](CodeGenFunction &CGF, const Expr *, const Expr *,
-                const Expr *) { CGF.EmitIgnoredExpr(E); });
-      } else
-        // Emit reduction for array subscript or single variable.
-        CGF.EmitIgnoredExpr(E);
-      ++IPriv, ++ILHS, ++IRHS;
+      emitSingleReductionCombiner(CGF, E, *IPriv, cast<DeclRefExpr>(*ILHS),
+                                  cast<DeclRefExpr>(*IRHS));
+      ++IPriv;
+      ++ILHS;
+      ++IRHS;
     }
-  }
+  };
+  RegionCodeGenTy RCG(CodeGen);
+  CommonActionTy Action(
+      nullptr, llvm::None,
+      createRuntimeFunction(WithNowait ? OMPRTL__kmpc_end_reduce_nowait
+                                       : OMPRTL__kmpc_end_reduce),
+      EndArgs);
+  RCG.setAction(Action);
+  RCG(CGF);
 
   CGF.EmitBranch(DefaultBB);
 
@@ -3487,103 +4506,113 @@ void CGOpenMPRuntime::emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
   SwInst->addCase(CGF.Builder.getInt32(2), Case2BB);
   CGF.EmitBlock(Case2BB);
 
-  {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    if (!WithNowait) {
-      // Add emission of __kmpc_end_reduce(<loc>, <gtid>, &<lock>);
-      llvm::Value *EndArgs[] = {
-          IdentTLoc, // ident_t *<loc>
-          ThreadId,  // i32 <gtid>
-          Lock       // kmp_critical_name *&<lock>
-      };
-      CGF.EHStack
-          .pushCleanup<CallEndCleanup<std::extent<decltype(EndArgs)>::value>>(
-              NormalAndEHCleanup,
-              createRuntimeFunction(OMPRTL__kmpc_end_reduce),
-              llvm::makeArrayRef(EndArgs));
-    }
+  auto &&AtomicCodeGen = [Loc, &Privates, &LHSExprs, &RHSExprs, &ReductionOps](
+      CodeGenFunction &CGF, PrePostActionTy &Action) {
     auto ILHS = LHSExprs.begin();
     auto IRHS = RHSExprs.begin();
     auto IPriv = Privates.begin();
     for (auto *E : ReductionOps) {
-        const Expr *XExpr = nullptr;
-        const Expr *EExpr = nullptr;
-        const Expr *UpExpr = nullptr;
-        BinaryOperatorKind BO = BO_Comma;
-        if (auto *BO = dyn_cast<BinaryOperator>(E)) {
-          if (BO->getOpcode() == BO_Assign) {
-            XExpr = BO->getLHS();
-            UpExpr = BO->getRHS();
-          }
+      const Expr *XExpr = nullptr;
+      const Expr *EExpr = nullptr;
+      const Expr *UpExpr = nullptr;
+      BinaryOperatorKind BO = BO_Comma;
+      if (auto *BO = dyn_cast<BinaryOperator>(E)) {
+        if (BO->getOpcode() == BO_Assign) {
+          XExpr = BO->getLHS();
+          UpExpr = BO->getRHS();
         }
-        // Try to emit update expression as a simple atomic.
-        auto *RHSExpr = UpExpr;
-        if (RHSExpr) {
-          // Analyze RHS part of the whole expression.
-          if (auto *ACO = dyn_cast<AbstractConditionalOperator>(
-                  RHSExpr->IgnoreParenImpCasts())) {
-            // If this is a conditional operator, analyze its condition for
-            // min/max reduction operator.
-            RHSExpr = ACO->getCond();
-          }
-          if (auto *BORHS =
-                  dyn_cast<BinaryOperator>(RHSExpr->IgnoreParenImpCasts())) {
-            EExpr = BORHS->getRHS();
-            BO = BORHS->getOpcode();
-          }
+      }
+      // Try to emit update expression as a simple atomic.
+      auto *RHSExpr = UpExpr;
+      if (RHSExpr) {
+        // Analyze RHS part of the whole expression.
+        if (auto *ACO = dyn_cast<AbstractConditionalOperator>(
+                RHSExpr->IgnoreParenImpCasts())) {
+          // If this is a conditional operator, analyze its condition for
+          // min/max reduction operator.
+          RHSExpr = ACO->getCond();
         }
-        if (XExpr) {
-          auto *VD = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
-          auto &&AtomicRedGen = [this, BO, VD, IPriv,
-                                 Loc](CodeGenFunction &CGF, const Expr *XExpr,
-                                      const Expr *EExpr, const Expr *UpExpr) {
-            LValue X = CGF.EmitLValue(XExpr);
-            RValue E;
-            if (EExpr)
-              E = CGF.EmitAnyExpr(EExpr);
-            CGF.EmitOMPAtomicSimpleUpdateExpr(
-                X, E, BO, /*IsXLHSInRHSPart=*/true, llvm::Monotonic, Loc,
-                [&CGF, UpExpr, VD, IPriv, Loc](RValue XRValue) {
-                  CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
-                  PrivateScope.addPrivate(
-                      VD, [&CGF, VD, XRValue, Loc]() -> Address {
-                        Address LHSTemp = CGF.CreateMemTemp(VD->getType());
-                        CGF.emitOMPSimpleStore(
-                            CGF.MakeAddrLValue(LHSTemp, VD->getType()), XRValue,
-                            VD->getType().getNonReferenceType(), Loc);
-                        return LHSTemp;
-                      });
-                  (void)PrivateScope.Privatize();
-                  return CGF.EmitAnyExpr(UpExpr);
-                });
-          };
-          if ((*IPriv)->getType()->isArrayType()) {
-            // Emit atomic reduction for array section.
-            auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
-            EmitOMPAggregateReduction(CGF, (*IPriv)->getType(), VD, RHSVar,
-                                      AtomicRedGen, XExpr, EExpr, UpExpr);
-          } else
-            // Emit atomic reduction for array subscript or single variable.
-            AtomicRedGen(CGF, XExpr, EExpr, UpExpr);
-        } else {
-          // Emit as a critical region.
-          auto &&CritRedGen = [this, E, Loc](CodeGenFunction &CGF, const Expr *,
-                                             const Expr *, const Expr *) {
-            emitCriticalRegion(
-                CGF, ".atomic_reduction",
-                [E](CodeGenFunction &CGF) { CGF.EmitIgnoredExpr(E); }, Loc);
-          };
-          if ((*IPriv)->getType()->isArrayType()) {
-            auto *LHSVar = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
-            auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
-            EmitOMPAggregateReduction(CGF, (*IPriv)->getType(), LHSVar, RHSVar,
-                                      CritRedGen);
-          } else
-            CritRedGen(CGF, nullptr, nullptr, nullptr);
+        if (auto *BORHS =
+                dyn_cast<BinaryOperator>(RHSExpr->IgnoreParenImpCasts())) {
+          EExpr = BORHS->getRHS();
+          BO = BORHS->getOpcode();
         }
-      ++ILHS, ++IRHS, ++IPriv;
+      }
+      if (XExpr) {
+        auto *VD = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
+        auto &&AtomicRedGen = [BO, VD, IPriv,
+                               Loc](CodeGenFunction &CGF, const Expr *XExpr,
+                                    const Expr *EExpr, const Expr *UpExpr) {
+          LValue X = CGF.EmitLValue(XExpr);
+          RValue E;
+          if (EExpr)
+            E = CGF.EmitAnyExpr(EExpr);
+          CGF.EmitOMPAtomicSimpleUpdateExpr(
+              X, E, BO, /*IsXLHSInRHSPart=*/true,
+              llvm::AtomicOrdering::Monotonic, Loc,
+              [&CGF, UpExpr, VD, IPriv, Loc](RValue XRValue) {
+                CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
+                PrivateScope.addPrivate(
+                    VD, [&CGF, VD, XRValue, Loc]() -> Address {
+                      Address LHSTemp = CGF.CreateMemTemp(VD->getType());
+                      CGF.emitOMPSimpleStore(
+                          CGF.MakeAddrLValue(LHSTemp, VD->getType()), XRValue,
+                          VD->getType().getNonReferenceType(), Loc);
+                      return LHSTemp;
+                    });
+                (void)PrivateScope.Privatize();
+                return CGF.EmitAnyExpr(UpExpr);
+              });
+        };
+        if ((*IPriv)->getType()->isArrayType()) {
+          // Emit atomic reduction for array section.
+          auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
+          EmitOMPAggregateReduction(CGF, (*IPriv)->getType(), VD, RHSVar,
+                                    AtomicRedGen, XExpr, EExpr, UpExpr);
+        } else
+          // Emit atomic reduction for array subscript or single variable.
+          AtomicRedGen(CGF, XExpr, EExpr, UpExpr);
+      } else {
+        // Emit as a critical region.
+        auto &&CritRedGen = [E, Loc](CodeGenFunction &CGF, const Expr *,
+                                     const Expr *, const Expr *) {
+          auto &RT = CGF.CGM.getOpenMPRuntime();
+          RT.emitCriticalRegion(
+              CGF, ".atomic_reduction",
+              [=](CodeGenFunction &CGF, PrePostActionTy &Action) {
+                Action.Enter(CGF);
+                emitReductionCombiner(CGF, E);
+              },
+              Loc);
+        };
+        if ((*IPriv)->getType()->isArrayType()) {
+          auto *LHSVar = cast<VarDecl>(cast<DeclRefExpr>(*ILHS)->getDecl());
+          auto *RHSVar = cast<VarDecl>(cast<DeclRefExpr>(*IRHS)->getDecl());
+          EmitOMPAggregateReduction(CGF, (*IPriv)->getType(), LHSVar, RHSVar,
+                                    CritRedGen);
+        } else
+          CritRedGen(CGF, nullptr, nullptr, nullptr);
+      }
+      ++ILHS;
+      ++IRHS;
+      ++IPriv;
     }
-  }
+  };
+  RegionCodeGenTy AtomicRCG(AtomicCodeGen);
+  if (!WithNowait) {
+    // Add emission of __kmpc_end_reduce(<loc>, <gtid>, &<lock>);
+    llvm::Value *EndArgs[] = {
+        IdentTLoc, // ident_t *<loc>
+        ThreadId,  // i32 <gtid>
+        Lock       // kmp_critical_name *&<lock>
+    };
+    CommonActionTy Action(nullptr, llvm::None,
+                          createRuntimeFunction(OMPRTL__kmpc_end_reduce),
+                          EndArgs);
+    AtomicRCG.setAction(Action);
+    AtomicRCG(CGF);
+  } else
+    AtomicRCG(CGF);
 
   CGF.EmitBranch(DefaultBB);
   CGF.EmitBlock(DefaultBB, /*IsFinished=*/true);
@@ -3598,6 +4627,8 @@ void CGOpenMPRuntime::emitTaskwaitCall(CodeGenFunction &CGF,
   llvm::Value *Args[] = {emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc)};
   // Ignore return result until untied tasks are supported.
   CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_omp_taskwait), Args);
+  if (auto *Region = dyn_cast_or_null<CGOpenMPRegionInfo>(CGF.CapturedStmtInfo))
+    Region->emitUntiedSwitch(CGF);
 }
 
 void CGOpenMPRuntime::emitInlinedDirective(CodeGenFunction &CGF,
@@ -3618,7 +4649,7 @@ enum RTCancelKind {
   CancelSections = 3,
   CancelTaskgroup = 4
 };
-}
+} // anonymous namespace
 
 static RTCancelKind getCancellationKind(OpenMPDirectiveKind CancelRegion) {
   RTCancelKind CancelKind = CancelNoreq;
@@ -3680,14 +4711,15 @@ void CGOpenMPRuntime::emitCancelCall(CodeGenFunction &CGF, SourceLocation Loc,
   // kmp_int32 cncl_kind);
   if (auto *OMPRegionInfo =
           dyn_cast_or_null<CGOpenMPRegionInfo>(CGF.CapturedStmtInfo)) {
-    auto &&ThenGen = [this, Loc, CancelRegion,
-                      OMPRegionInfo](CodeGenFunction &CGF) {
+    auto &&ThenGen = [Loc, CancelRegion, OMPRegionInfo](CodeGenFunction &CGF,
+                                                        PrePostActionTy &) {
+      auto &RT = CGF.CGM.getOpenMPRuntime();
       llvm::Value *Args[] = {
-          emitUpdateLocation(CGF, Loc), getThreadID(CGF, Loc),
+          RT.emitUpdateLocation(CGF, Loc), RT.getThreadID(CGF, Loc),
           CGF.Builder.getInt32(getCancellationKind(CancelRegion))};
       // Ignore return result until untied tasks are supported.
-      auto *Result =
-          CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_cancel), Args);
+      auto *Result = CGF.EmitRuntimeCall(
+          RT.createRuntimeFunction(OMPRTL__kmpc_cancel), Args);
       // if (__kmpc_cancel()) {
       //  __kmpc_cancel_barrier();
       //   exit from construct;
@@ -3698,7 +4730,7 @@ void CGOpenMPRuntime::emitCancelCall(CodeGenFunction &CGF, SourceLocation Loc,
       CGF.Builder.CreateCondBr(Cmp, ExitBB, ContBB);
       CGF.EmitBlock(ExitBB);
       // __kmpc_cancel_barrier();
-      emitBarrierCall(CGF, Loc, OMPD_unknown, /*EmitChecks=*/false);
+      RT.emitBarrierCall(CGF, Loc, OMPD_unknown, /*EmitChecks=*/false);
       // exit from construct;
       auto CancelDest =
           CGF.getOMPCancelDestination(OMPRegionInfo->getDirectiveKind());
@@ -3706,18 +4738,21 @@ void CGOpenMPRuntime::emitCancelCall(CodeGenFunction &CGF, SourceLocation Loc,
       CGF.EmitBlock(ContBB, /*IsFinished=*/true);
     };
     if (IfCond)
-      emitOMPIfClause(CGF, IfCond, ThenGen, [](CodeGenFunction &) {});
-    else
-      ThenGen(CGF);
+      emitOMPIfClause(CGF, IfCond, ThenGen,
+                      [](CodeGenFunction &, PrePostActionTy &) {});
+    else {
+      RegionCodeGenTy ThenRCG(ThenGen);
+      ThenRCG(CGF);
+    }
   }
 }
 
 /// \brief Obtain information that uniquely identifies a target entry. This
-/// consists of the file and device IDs as well as line and column numbers
-/// associated with the relevant entry source location.
+/// consists of the file and device IDs as well as line number associated with
+/// the relevant entry source location.
 static void getTargetEntryUniqueInfo(ASTContext &C, SourceLocation Loc,
                                      unsigned &DeviceID, unsigned &FileID,
-                                     unsigned &LineNum, unsigned &ColumnNum) {
+                                     unsigned &LineNum) {
 
   auto &SM = C.getSourceManager();
 
@@ -3737,49 +4772,45 @@ static void getTargetEntryUniqueInfo(ASTContext &C, SourceLocation Loc,
   DeviceID = ID.getDevice();
   FileID = ID.getFile();
   LineNum = PLoc.getLine();
-  ColumnNum = PLoc.getColumn();
-  return;
 }
 
 void CGOpenMPRuntime::emitTargetOutlinedFunction(
     const OMPExecutableDirective &D, StringRef ParentName,
     llvm::Function *&OutlinedFn, llvm::Constant *&OutlinedFnID,
-    bool IsOffloadEntry) {
-
+    bool IsOffloadEntry, const RegionCodeGenTy &CodeGen) {
   assert(!ParentName.empty() && "Invalid target region parent name!");
 
-  const CapturedStmt &CS = *cast<CapturedStmt>(D.getAssociatedStmt());
-
-  // Emit target region as a standalone region.
-  auto &&CodeGen = [&CS](CodeGenFunction &CGF) {
-    CGF.EmitStmt(CS.getCapturedStmt());
-  };
+  emitTargetOutlinedFunctionHelper(D, ParentName, OutlinedFn, OutlinedFnID,
+                                   IsOffloadEntry, CodeGen);
+}
 
-  // Create a unique name for the proxy/entry function that using the source
-  // location information of the current target region. The name will be
-  // something like:
+void CGOpenMPRuntime::emitTargetOutlinedFunctionHelper(
+    const OMPExecutableDirective &D, StringRef ParentName,
+    llvm::Function *&OutlinedFn, llvm::Constant *&OutlinedFnID,
+    bool IsOffloadEntry, const RegionCodeGenTy &CodeGen) {
+  // Create a unique name for the entry function using the source location
+  // information of the current target region. The name will be something like:
   //
-  // .omp_offloading.DD_FFFF.PP.lBB.cCC
+  // __omp_offloading_DD_FFFF_PP_lBB
   //
   // where DD_FFFF is an ID unique to the file (device and file IDs), PP is the
-  // mangled name of the function that encloses the target region, BB is the
-  // line number of the target region, and CC is the column number of the target
-  // region.
+  // mangled name of the function that encloses the target region and BB is the
+  // line number of the target region.
 
   unsigned DeviceID;
   unsigned FileID;
   unsigned Line;
-  unsigned Column;
   getTargetEntryUniqueInfo(CGM.getContext(), D.getLocStart(), DeviceID, FileID,
-                           Line, Column);
+                           Line);
   SmallString<64> EntryFnName;
   {
     llvm::raw_svector_ostream OS(EntryFnName);
-    OS << ".omp_offloading" << llvm::format(".%x", DeviceID)
-       << llvm::format(".%x.", FileID) << ParentName << ".l" << Line << ".c"
-       << Column;
+    OS << "__omp_offloading" << llvm::format("_%x", DeviceID)
+       << llvm::format("_%x_", FileID) << ParentName << "_l" << Line;
   }
 
+  const CapturedStmt &CS = *cast<CapturedStmt>(D.getAssociatedStmt());
+
   CodeGenFunction CGF(CGM, true);
   CGOpenMPTargetRegionInfo CGInfo(CS, CodeGen, EntryFnName);
   CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
@@ -3813,18 +4844,122 @@ void CGOpenMPRuntime::emitTargetOutlinedFunction(
 
   // Register the information for the entry associated with this target region.
   OffloadEntriesInfoManager.registerTargetRegionEntryInfo(
-      DeviceID, FileID, ParentName, Line, Column, OutlinedFn, OutlinedFnID);
-  return;
+      DeviceID, FileID, ParentName, Line, OutlinedFn, OutlinedFnID);
 }
 
-void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
-                                     const OMPExecutableDirective &D,
-                                     llvm::Value *OutlinedFn,
-                                     llvm::Value *OutlinedFnID,
-                                     const Expr *IfCond, const Expr *Device,
-                                     ArrayRef<llvm::Value *> CapturedVars) {
-  if (!CGF.HaveInsertPoint())
-    return;
+/// discard all CompoundStmts intervening between two constructs
+static const Stmt *ignoreCompoundStmts(const Stmt *Body) {
+  while (auto *CS = dyn_cast_or_null<CompoundStmt>(Body))
+    Body = CS->body_front();
+
+  return Body;
+}
+
+/// \brief Emit the num_teams clause of an enclosed teams directive at the
+/// target region scope. If there is no teams directive associated with the
+/// target directive, or if there is no num_teams clause associated with the
+/// enclosed teams directive, return nullptr.
+static llvm::Value *
+emitNumTeamsClauseForTargetDirective(CGOpenMPRuntime &OMPRuntime,
+                                     CodeGenFunction &CGF,
+                                     const OMPExecutableDirective &D) {
+
+  assert(!CGF.getLangOpts().OpenMPIsDevice && "Clauses associated with the "
+                                              "teams directive expected to be "
+                                              "emitted only for the host!");
+
+  // FIXME: For the moment we do not support combined directives with target and
+  // teams, so we do not expect to get any num_teams clause in the provided
+  // directive. Once we support that, this assertion can be replaced by the
+  // actual emission of the clause expression.
+  assert(D.getSingleClause<OMPNumTeamsClause>() == nullptr &&
+         "Not expecting clause in directive.");
+
+  // If the current target region has a teams region enclosed, we need to get
+  // the number of teams to pass to the runtime function call. This is done
+  // by generating the expression in a inlined region. This is required because
+  // the expression is captured in the enclosing target environment when the
+  // teams directive is not combined with target.
+
+  const CapturedStmt &CS = *cast<CapturedStmt>(D.getAssociatedStmt());
+
+  // FIXME: Accommodate other combined directives with teams when they become
+  // available.
+  if (auto *TeamsDir = dyn_cast_or_null<OMPTeamsDirective>(
+          ignoreCompoundStmts(CS.getCapturedStmt()))) {
+    if (auto *NTE = TeamsDir->getSingleClause<OMPNumTeamsClause>()) {
+      CGOpenMPInnerExprInfo CGInfo(CGF, CS);
+      CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
+      llvm::Value *NumTeams = CGF.EmitScalarExpr(NTE->getNumTeams());
+      return CGF.Builder.CreateIntCast(NumTeams, CGF.Int32Ty,
+                                       /*IsSigned=*/true);
+    }
+
+    // If we have an enclosed teams directive but no num_teams clause we use
+    // the default value 0.
+    return CGF.Builder.getInt32(0);
+  }
+
+  // No teams associated with the directive.
+  return nullptr;
+}
+
+/// \brief Emit the thread_limit clause of an enclosed teams directive at the
+/// target region scope. If there is no teams directive associated with the
+/// target directive, or if there is no thread_limit clause associated with the
+/// enclosed teams directive, return nullptr.
+static llvm::Value *
+emitThreadLimitClauseForTargetDirective(CGOpenMPRuntime &OMPRuntime,
+                                        CodeGenFunction &CGF,
+                                        const OMPExecutableDirective &D) {
+
+  assert(!CGF.getLangOpts().OpenMPIsDevice && "Clauses associated with the "
+                                              "teams directive expected to be "
+                                              "emitted only for the host!");
+
+  // FIXME: For the moment we do not support combined directives with target and
+  // teams, so we do not expect to get any thread_limit clause in the provided
+  // directive. Once we support that, this assertion can be replaced by the
+  // actual emission of the clause expression.
+  assert(D.getSingleClause<OMPThreadLimitClause>() == nullptr &&
+         "Not expecting clause in directive.");
+
+  // If the current target region has a teams region enclosed, we need to get
+  // the thread limit to pass to the runtime function call. This is done
+  // by generating the expression in a inlined region. This is required because
+  // the expression is captured in the enclosing target environment when the
+  // teams directive is not combined with target.
+
+  const CapturedStmt &CS = *cast<CapturedStmt>(D.getAssociatedStmt());
+
+  // FIXME: Accommodate other combined directives with teams when they become
+  // available.
+  if (auto *TeamsDir = dyn_cast_or_null<OMPTeamsDirective>(
+          ignoreCompoundStmts(CS.getCapturedStmt()))) {
+    if (auto *TLE = TeamsDir->getSingleClause<OMPThreadLimitClause>()) {
+      CGOpenMPInnerExprInfo CGInfo(CGF, CS);
+      CodeGenFunction::CGCapturedStmtRAII CapInfoRAII(CGF, &CGInfo);
+      llvm::Value *ThreadLimit = CGF.EmitScalarExpr(TLE->getThreadLimit());
+      return CGF.Builder.CreateIntCast(ThreadLimit, CGF.Int32Ty,
+                                       /*IsSigned=*/true);
+    }
+
+    // If we have an enclosed teams directive but no thread_limit clause we use
+    // the default value 0.
+    return CGF.Builder.getInt32(0);
+  }
+
+  // No teams associated with the directive.
+  return nullptr;
+}
+
+namespace {
+// \brief Utility to handle information from clauses associated with a given
+// construct that use mappable expressions (e.g. 'map' clause, 'to' clause).
+// It provides a convenient interface to obtain the information and generate
+// code for that information.
+class MappableExprsHandler {
+public:
   /// \brief Values for bit flags used to specify the mapping type for
   /// offloading.
   enum OpenMPOffloadMappingFlags {
@@ -3832,105 +4967,806 @@ void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
     OMP_MAP_TO = 0x01,
     /// \brief Allocate memory on the device and move data from device to host.
     OMP_MAP_FROM = 0x02,
-    /// \brief The element passed to the device is a pointer.
-    OMP_MAP_PTR = 0x20,
+    /// \brief Always perform the requested mapping action on the element, even
+    /// if it was already mapped before.
+    OMP_MAP_ALWAYS = 0x04,
+    /// \brief Delete the element from the device environment, ignoring the
+    /// current reference count associated with the element.
+    OMP_MAP_DELETE = 0x08,
+    /// \brief The element being mapped is a pointer, therefore the pointee
+    /// should be mapped as well.
+    OMP_MAP_IS_PTR = 0x10,
+    /// \brief This flags signals that an argument is the first one relating to
+    /// a map/private clause expression. For some cases a single
+    /// map/privatization results in multiple arguments passed to the runtime
+    /// library.
+    OMP_MAP_FIRST_REF = 0x20,
+    /// \brief This flag signals that the reference being passed is a pointer to
+    /// private data.
+    OMP_MAP_PRIVATE_PTR = 0x80,
     /// \brief Pass the element to the device by value.
-    OMP_MAP_BYCOPY = 0x80,
+    OMP_MAP_PRIVATE_VAL = 0x100,
   };
 
-  enum OpenMPOffloadingReservedDeviceIDs {
-    /// \brief Device ID if the device was not defined, runtime should get it
-    /// from environment variables in the spec.
-    OMP_DEVICEID_UNDEF = -1,
-  };
+  typedef SmallVector<llvm::Value *, 16> MapValuesArrayTy;
+  typedef SmallVector<unsigned, 16> MapFlagsArrayTy;
+
+private:
+  /// \brief Directive from where the map clauses were extracted.
+  const OMPExecutableDirective &Directive;
+
+  /// \brief Function the directive is being generated for.
+  CodeGenFunction &CGF;
+
+  /// \brief Set of all first private variables in the current directive.
+  llvm::SmallPtrSet<const VarDecl *, 8> FirstPrivateDecls;
+
+  llvm::Value *getExprTypeSize(const Expr *E) const {
+    auto ExprTy = E->getType().getCanonicalType();
+
+    // Reference types are ignored for mapping purposes.
+    if (auto *RefTy = ExprTy->getAs<ReferenceType>())
+      ExprTy = RefTy->getPointeeType().getCanonicalType();
+
+    // Given that an array section is considered a built-in type, we need to
+    // do the calculation based on the length of the section instead of relying
+    // on CGF.getTypeSize(E->getType()).
+    if (const auto *OAE = dyn_cast<OMPArraySectionExpr>(E)) {
+      QualType BaseTy = OMPArraySectionExpr::getBaseOriginalType(
+                            OAE->getBase()->IgnoreParenImpCasts())
+                            .getCanonicalType();
+
+      // If there is no length associated with the expression, that means we
+      // are using the whole length of the base.
+      if (!OAE->getLength() && OAE->getColonLoc().isValid())
+        return CGF.getTypeSize(BaseTy);
+
+      llvm::Value *ElemSize;
+      if (auto *PTy = BaseTy->getAs<PointerType>())
+        ElemSize = CGF.getTypeSize(PTy->getPointeeType().getCanonicalType());
+      else {
+        auto *ATy = cast<ArrayType>(BaseTy.getTypePtr());
+        assert(ATy && "Expecting array type if not a pointer type.");
+        ElemSize = CGF.getTypeSize(ATy->getElementType().getCanonicalType());
+      }
+
+      // If we don't have a length at this point, that is because we have an
+      // array section with a single element.
+      if (!OAE->getLength())
+        return ElemSize;
+
+      auto *LengthVal = CGF.EmitScalarExpr(OAE->getLength());
+      LengthVal =
+          CGF.Builder.CreateIntCast(LengthVal, CGF.SizeTy, /*isSigned=*/false);
+      return CGF.Builder.CreateNUWMul(LengthVal, ElemSize);
+    }
+    return CGF.getTypeSize(ExprTy);
+  }
+
+  /// \brief Return the corresponding bits for a given map clause modifier. Add
+  /// a flag marking the map as a pointer if requested. Add a flag marking the
+  /// map as the first one of a series of maps that relate to the same map
+  /// expression.
+  unsigned getMapTypeBits(OpenMPMapClauseKind MapType,
+                          OpenMPMapClauseKind MapTypeModifier, bool AddPtrFlag,
+                          bool AddIsFirstFlag) const {
+    unsigned Bits = 0u;
+    switch (MapType) {
+    case OMPC_MAP_alloc:
+    case OMPC_MAP_release:
+      // alloc and release is the default behavior in the runtime library,  i.e.
+      // if we don't pass any bits alloc/release that is what the runtime is
+      // going to do. Therefore, we don't need to signal anything for these two
+      // type modifiers.
+      break;
+    case OMPC_MAP_to:
+      Bits = OMP_MAP_TO;
+      break;
+    case OMPC_MAP_from:
+      Bits = OMP_MAP_FROM;
+      break;
+    case OMPC_MAP_tofrom:
+      Bits = OMP_MAP_TO | OMP_MAP_FROM;
+      break;
+    case OMPC_MAP_delete:
+      Bits = OMP_MAP_DELETE;
+      break;
+    default:
+      llvm_unreachable("Unexpected map type!");
+      break;
+    }
+    if (AddPtrFlag)
+      Bits |= OMP_MAP_IS_PTR;
+    if (AddIsFirstFlag)
+      Bits |= OMP_MAP_FIRST_REF;
+    if (MapTypeModifier == OMPC_MAP_always)
+      Bits |= OMP_MAP_ALWAYS;
+    return Bits;
+  }
+
+  /// \brief Return true if the provided expression is a final array section. A
+  /// final array section, is one whose length can't be proved to be one.
+  bool isFinalArraySectionExpression(const Expr *E) const {
+    auto *OASE = dyn_cast<OMPArraySectionExpr>(E);
+
+    // It is not an array section and therefore not a unity-size one.
+    if (!OASE)
+      return false;
+
+    // An array section with no colon always refer to a single element.
+    if (OASE->getColonLoc().isInvalid())
+      return false;
+
+    auto *Length = OASE->getLength();
+
+    // If we don't have a length we have to check if the array has size 1
+    // for this dimension. Also, we should always expect a length if the
+    // base type is pointer.
+    if (!Length) {
+      auto BaseQTy = OMPArraySectionExpr::getBaseOriginalType(
+                         OASE->getBase()->IgnoreParenImpCasts())
+                         .getCanonicalType();
+      if (auto *ATy = dyn_cast<ConstantArrayType>(BaseQTy.getTypePtr()))
+        return ATy->getSize().getSExtValue() != 1;
+      // If we don't have a constant dimension length, we have to consider
+      // the current section as having any size, so it is not necessarily
+      // unitary. If it happen to be unity size, that's user fault.
+      return true;
+    }
+
+    // Check if the length evaluates to 1.
+    llvm::APSInt ConstLength;
+    if (!Length->EvaluateAsInt(ConstLength, CGF.getContext()))
+      return true; // Can have more that size 1.
+
+    return ConstLength.getSExtValue() != 1;
+  }
+
+  /// \brief Generate the base pointers, section pointers, sizes and map type
+  /// bits for the provided map type, map modifier, and expression components.
+  /// \a IsFirstComponent should be set to true if the provided set of
+  /// components is the first associated with a capture.
+  void generateInfoForComponentList(
+      OpenMPMapClauseKind MapType, OpenMPMapClauseKind MapTypeModifier,
+      OMPClauseMappableExprCommon::MappableExprComponentListRef Components,
+      MapValuesArrayTy &BasePointers, MapValuesArrayTy &Pointers,
+      MapValuesArrayTy &Sizes, MapFlagsArrayTy &Types,
+      bool IsFirstComponentList) const {
+
+    // The following summarizes what has to be generated for each map and the
+    // types bellow. The generated information is expressed in this order:
+    // base pointer, section pointer, size, flags
+    // (to add to the ones that come from the map type and modifier).
+    //
+    // double d;
+    // int i[100];
+    // float *p;
+    //
+    // struct S1 {
+    //   int i;
+    //   float f[50];
+    // }
+    // struct S2 {
+    //   int i;
+    //   float f[50];
+    //   S1 s;
+    //   double *p;
+    //   struct S2 *ps;
+    // }
+    // S2 s;
+    // S2 *ps;
+    //
+    // map(d)
+    // &d, &d, sizeof(double), noflags
+    //
+    // map(i)
+    // &i, &i, 100*sizeof(int), noflags
+    //
+    // map(i[1:23])
+    // &i(=&i[0]), &i[1], 23*sizeof(int), noflags
+    //
+    // map(p)
+    // &p, &p, sizeof(float*), noflags
+    //
+    // map(p[1:24])
+    // p, &p[1], 24*sizeof(float), noflags
+    //
+    // map(s)
+    // &s, &s, sizeof(S2), noflags
+    //
+    // map(s.i)
+    // &s, &(s.i), sizeof(int), noflags
+    //
+    // map(s.s.f)
+    // &s, &(s.i.f), 50*sizeof(int), noflags
+    //
+    // map(s.p)
+    // &s, &(s.p), sizeof(double*), noflags
+    //
+    // map(s.p[:22], s.a s.b)
+    // &s, &(s.p), sizeof(double*), noflags
+    // &(s.p), &(s.p[0]), 22*sizeof(double), ptr_flag + extra_flag
+    //
+    // map(s.ps)
+    // &s, &(s.ps), sizeof(S2*), noflags
+    //
+    // map(s.ps->s.i)
+    // &s, &(s.ps), sizeof(S2*), noflags
+    // &(s.ps), &(s.ps->s.i), sizeof(int), ptr_flag + extra_flag
+    //
+    // map(s.ps->ps)
+    // &s, &(s.ps), sizeof(S2*), noflags
+    // &(s.ps), &(s.ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    //
+    // map(s.ps->ps->ps)
+    // &s, &(s.ps), sizeof(S2*), noflags
+    // &(s.ps), &(s.ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    // &(s.ps->ps), &(s.ps->ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    //
+    // map(s.ps->ps->s.f[:22])
+    // &s, &(s.ps), sizeof(S2*), noflags
+    // &(s.ps), &(s.ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    // &(s.ps->ps), &(s.ps->ps->s.f[0]), 22*sizeof(float), ptr_flag + extra_flag
+    //
+    // map(ps)
+    // &ps, &ps, sizeof(S2*), noflags
+    //
+    // map(ps->i)
+    // ps, &(ps->i), sizeof(int), noflags
+    //
+    // map(ps->s.f)
+    // ps, &(ps->s.f[0]), 50*sizeof(float), noflags
+    //
+    // map(ps->p)
+    // ps, &(ps->p), sizeof(double*), noflags
+    //
+    // map(ps->p[:22])
+    // ps, &(ps->p), sizeof(double*), noflags
+    // &(ps->p), &(ps->p[0]), 22*sizeof(double), ptr_flag + extra_flag
+    //
+    // map(ps->ps)
+    // ps, &(ps->ps), sizeof(S2*), noflags
+    //
+    // map(ps->ps->s.i)
+    // ps, &(ps->ps), sizeof(S2*), noflags
+    // &(ps->ps), &(ps->ps->s.i), sizeof(int), ptr_flag + extra_flag
+    //
+    // map(ps->ps->ps)
+    // ps, &(ps->ps), sizeof(S2*), noflags
+    // &(ps->ps), &(ps->ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    //
+    // map(ps->ps->ps->ps)
+    // ps, &(ps->ps), sizeof(S2*), noflags
+    // &(ps->ps), &(ps->ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    // &(ps->ps->ps), &(ps->ps->ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    //
+    // map(ps->ps->ps->s.f[:22])
+    // ps, &(ps->ps), sizeof(S2*), noflags
+    // &(ps->ps), &(ps->ps->ps), sizeof(S2*), ptr_flag + extra_flag
+    // &(ps->ps->ps), &(ps->ps->ps->s.f[0]), 22*sizeof(float), ptr_flag +
+    // extra_flag
+
+    // Track if the map information being generated is the first for a capture.
+    bool IsCaptureFirstInfo = IsFirstComponentList;
+
+    // Scan the components from the base to the complete expression.
+    auto CI = Components.rbegin();
+    auto CE = Components.rend();
+    auto I = CI;
+
+    // Track if the map information being generated is the first for a list of
+    // components.
+    bool IsExpressionFirstInfo = true;
+    llvm::Value *BP = nullptr;
+
+    if (auto *ME = dyn_cast<MemberExpr>(I->getAssociatedExpression())) {
+      // The base is the 'this' pointer. The content of the pointer is going
+      // to be the base of the field being mapped.
+      BP = CGF.EmitScalarExpr(ME->getBase());
+    } else {
+      // The base is the reference to the variable.
+      // BP = &Var.
+      BP = CGF.EmitLValue(cast<DeclRefExpr>(I->getAssociatedExpression()))
+               .getPointer();
+
+      // If the variable is a pointer and is being dereferenced (i.e. is not
+      // the last component), the base has to be the pointer itself, not its
+      // reference.
+      if (I->getAssociatedDeclaration()->getType()->isAnyPointerType() &&
+          std::next(I) != CE) {
+        auto PtrAddr = CGF.MakeNaturalAlignAddrLValue(
+            BP, I->getAssociatedDeclaration()->getType());
+        BP = CGF.EmitLoadOfPointerLValue(PtrAddr.getAddress(),
+                                         I->getAssociatedDeclaration()
+                                             ->getType()
+                                             ->getAs<PointerType>())
+                 .getPointer();
+
+        // We do not need to generate individual map information for the
+        // pointer, it can be associated with the combined storage.
+        ++I;
+      }
+    }
+
+    for (; I != CE; ++I) {
+      auto Next = std::next(I);
+
+      // We need to generate the addresses and sizes if this is the last
+      // component, if the component is a pointer or if it is an array section
+      // whose length can't be proved to be one. If this is a pointer, it
+      // becomes the base address for the following components.
+
+      // A final array section, is one whose length can't be proved to be one.
+      bool IsFinalArraySection =
+          isFinalArraySectionExpression(I->getAssociatedExpression());
+
+      // Get information on whether the element is a pointer. Have to do a
+      // special treatment for array sections given that they are built-in
+      // types.
+      const auto *OASE =
+          dyn_cast<OMPArraySectionExpr>(I->getAssociatedExpression());
+      bool IsPointer =
+          (OASE &&
+           OMPArraySectionExpr::getBaseOriginalType(OASE)
+               .getCanonicalType()
+               ->isAnyPointerType()) ||
+          I->getAssociatedExpression()->getType()->isAnyPointerType();
+
+      if (Next == CE || IsPointer || IsFinalArraySection) {
+
+        // If this is not the last component, we expect the pointer to be
+        // associated with an array expression or member expression.
+        assert((Next == CE ||
+                isa<MemberExpr>(Next->getAssociatedExpression()) ||
+                isa<ArraySubscriptExpr>(Next->getAssociatedExpression()) ||
+                isa<OMPArraySectionExpr>(Next->getAssociatedExpression())) &&
+               "Unexpected expression");
+
+        // Save the base we are currently using.
+        BasePointers.push_back(BP);
+
+        auto *LB = CGF.EmitLValue(I->getAssociatedExpression()).getPointer();
+        auto *Size = getExprTypeSize(I->getAssociatedExpression());
+
+        Pointers.push_back(LB);
+        Sizes.push_back(Size);
+        // We need to add a pointer flag for each map that comes from the
+        // same expression except for the first one. We also need to signal
+        // this map is the first one that relates with the current capture
+        // (there is a set of entries for each capture).
+        Types.push_back(getMapTypeBits(MapType, MapTypeModifier,
+                                       !IsExpressionFirstInfo,
+                                       IsCaptureFirstInfo));
+
+        // If we have a final array section, we are done with this expression.
+        if (IsFinalArraySection)
+          break;
+
+        // The pointer becomes the base for the next element.
+        if (Next != CE)
+          BP = LB;
+
+        IsExpressionFirstInfo = false;
+        IsCaptureFirstInfo = false;
+        continue;
+      }
+    }
+  }
+
+  /// \brief Return the adjusted map modifiers if the declaration a capture
+  /// refers to appears in a first-private clause. This is expected to be used
+  /// only with directives that start with 'target'.
+  unsigned adjustMapModifiersForPrivateClauses(const CapturedStmt::Capture &Cap,
+                                               unsigned CurrentModifiers) {
+    assert(Cap.capturesVariable() && "Expected capture by reference only!");
+
+    // A first private variable captured by reference will use only the
+    // 'private ptr' and 'map to' flag. Return the right flags if the captured
+    // declaration is known as first-private in this handler.
+    if (FirstPrivateDecls.count(Cap.getCapturedVar()))
+      return MappableExprsHandler::OMP_MAP_PRIVATE_PTR |
+             MappableExprsHandler::OMP_MAP_TO;
+
+    // We didn't modify anything.
+    return CurrentModifiers;
+  }
+
+public:
+  MappableExprsHandler(const OMPExecutableDirective &Dir, CodeGenFunction &CGF)
+      : Directive(Dir), CGF(CGF) {
+    // Extract firstprivate clause information.
+    for (const auto *C : Dir.getClausesOfKind<OMPFirstprivateClause>())
+      for (const auto *D : C->varlists())
+        FirstPrivateDecls.insert(
+            cast<VarDecl>(cast<DeclRefExpr>(D)->getDecl())->getCanonicalDecl());
+  }
+
+  /// \brief Generate all the base pointers, section pointers, sizes and map
+  /// types for the extracted mappable expressions.
+  void generateAllInfo(MapValuesArrayTy &BasePointers,
+                       MapValuesArrayTy &Pointers, MapValuesArrayTy &Sizes,
+                       MapFlagsArrayTy &Types) const {
+    BasePointers.clear();
+    Pointers.clear();
+    Sizes.clear();
+    Types.clear();
+
+    struct MapInfo {
+      OMPClauseMappableExprCommon::MappableExprComponentListRef Components;
+      OpenMPMapClauseKind MapType;
+      OpenMPMapClauseKind MapTypeModifier;
+    };
+
+    // We have to process the component lists that relate with the same
+    // declaration in a single chunk so that we can generate the map flags
+    // correctly. Therefore, we organize all lists in a map.
+    llvm::DenseMap<const ValueDecl *, SmallVector<MapInfo, 8>> Info;
+
+    // Helper function to fill the information map for the different supported
+    // clauses.
+    auto &&InfoGen =
+        [&Info](const ValueDecl *D,
+                OMPClauseMappableExprCommon::MappableExprComponentListRef L,
+                OpenMPMapClauseKind MapType, OpenMPMapClauseKind MapModifier) {
+          const ValueDecl *VD =
+              D ? cast<ValueDecl>(D->getCanonicalDecl()) : nullptr;
+          Info[VD].push_back({L, MapType, MapModifier});
+        };
+
+    for (auto *C : Directive.getClausesOfKind<OMPMapClause>())
+      for (auto L : C->component_lists())
+        InfoGen(L.first, L.second, C->getMapType(), C->getMapTypeModifier());
+    for (auto *C : Directive.getClausesOfKind<OMPToClause>())
+      for (auto L : C->component_lists())
+        InfoGen(L.first, L.second, OMPC_MAP_to, OMPC_MAP_unknown);
+    for (auto *C : Directive.getClausesOfKind<OMPFromClause>())
+      for (auto L : C->component_lists())
+        InfoGen(L.first, L.second, OMPC_MAP_from, OMPC_MAP_unknown);
+
+    for (auto &M : Info) {
+      // We need to know when we generate information for the first component
+      // associated with a capture, because the mapping flags depend on it.
+      bool IsFirstComponentList = true;
+      for (MapInfo &L : M.second) {
+        assert(!L.Components.empty() &&
+               "Not expecting declaration with no component lists.");
+        generateInfoForComponentList(L.MapType, L.MapTypeModifier, L.Components,
+                                     BasePointers, Pointers, Sizes, Types,
+                                     IsFirstComponentList);
+        IsFirstComponentList = false;
+      }
+    }
+  }
+
+  /// \brief Generate the base pointers, section pointers, sizes and map types
+  /// associated to a given capture.
+  void generateInfoForCapture(const CapturedStmt::Capture *Cap,
+                              MapValuesArrayTy &BasePointers,
+                              MapValuesArrayTy &Pointers,
+                              MapValuesArrayTy &Sizes,
+                              MapFlagsArrayTy &Types) const {
+    assert(!Cap->capturesVariableArrayType() &&
+           "Not expecting to generate map info for a variable array type!");
+
+    BasePointers.clear();
+    Pointers.clear();
+    Sizes.clear();
+    Types.clear();
+
+    const ValueDecl *VD =
+        Cap->capturesThis()
+            ? nullptr
+            : cast<ValueDecl>(Cap->getCapturedVar()->getCanonicalDecl());
+
+    // We need to know when we generating information for the first component
+    // associated with a capture, because the mapping flags depend on it.
+    bool IsFirstComponentList = true;
+    for (auto *C : Directive.getClausesOfKind<OMPMapClause>())
+      for (auto L : C->decl_component_lists(VD)) {
+        assert(L.first == VD &&
+               "We got information for the wrong declaration??");
+        assert(!L.second.empty() &&
+               "Not expecting declaration with no component lists.");
+        generateInfoForComponentList(C->getMapType(), C->getMapTypeModifier(),
+                                     L.second, BasePointers, Pointers, Sizes,
+                                     Types, IsFirstComponentList);
+        IsFirstComponentList = false;
+      }
+
+    return;
+  }
+
+  /// \brief Generate the default map information for a given capture \a CI,
+  /// record field declaration \a RI and captured value \a CV.
+  void generateDefaultMapInfo(
+      const CapturedStmt::Capture &CI, const FieldDecl &RI, llvm::Value *CV,
+      MappableExprsHandler::MapValuesArrayTy &CurBasePointers,
+      MappableExprsHandler::MapValuesArrayTy &CurPointers,
+      MappableExprsHandler::MapValuesArrayTy &CurSizes,
+      MappableExprsHandler::MapFlagsArrayTy &CurMapTypes) {
+
+    // Do the default mapping.
+    if (CI.capturesThis()) {
+      CurBasePointers.push_back(CV);
+      CurPointers.push_back(CV);
+      const PointerType *PtrTy = cast<PointerType>(RI.getType().getTypePtr());
+      CurSizes.push_back(CGF.getTypeSize(PtrTy->getPointeeType()));
+      // Default map type.
+      CurMapTypes.push_back(MappableExprsHandler::OMP_MAP_TO |
+                            MappableExprsHandler::OMP_MAP_FROM);
+    } else if (CI.capturesVariableByCopy()) {
+      CurBasePointers.push_back(CV);
+      CurPointers.push_back(CV);
+      if (!RI.getType()->isAnyPointerType()) {
+        // We have to signal to the runtime captures passed by value that are
+        // not pointers.
+        CurMapTypes.push_back(MappableExprsHandler::OMP_MAP_PRIVATE_VAL);
+        CurSizes.push_back(CGF.getTypeSize(RI.getType()));
+      } else {
+        // Pointers are implicitly mapped with a zero size and no flags
+        // (other than first map that is added for all implicit maps).
+        CurMapTypes.push_back(0u);
+        CurSizes.push_back(llvm::Constant::getNullValue(CGF.SizeTy));
+      }
+    } else {
+      assert(CI.capturesVariable() && "Expected captured reference.");
+      CurBasePointers.push_back(CV);
+      CurPointers.push_back(CV);
+
+      const ReferenceType *PtrTy =
+          cast<ReferenceType>(RI.getType().getTypePtr());
+      QualType ElementType = PtrTy->getPointeeType();
+      CurSizes.push_back(CGF.getTypeSize(ElementType));
+      // The default map type for a scalar/complex type is 'to' because by
+      // default the value doesn't have to be retrieved. For an aggregate
+      // type, the default is 'tofrom'.
+      CurMapTypes.push_back(ElementType->isAggregateType()
+                                ? (MappableExprsHandler::OMP_MAP_TO |
+                                   MappableExprsHandler::OMP_MAP_FROM)
+                                : MappableExprsHandler::OMP_MAP_TO);
+
+      // If we have a capture by reference we may need to add the private
+      // pointer flag if the base declaration shows in some first-private
+      // clause.
+      CurMapTypes.back() =
+          adjustMapModifiersForPrivateClauses(CI, CurMapTypes.back());
+    }
+    // Every default map produces a single argument, so, it is always the
+    // first one.
+    CurMapTypes.back() |= MappableExprsHandler::OMP_MAP_FIRST_REF;
+  }
+};
+
+enum OpenMPOffloadingReservedDeviceIDs {
+  /// \brief Device ID if the device was not defined, runtime should get it
+  /// from environment variables in the spec.
+  OMP_DEVICEID_UNDEF = -1,
+};
+} // anonymous namespace
+
+/// \brief Emit the arrays used to pass the captures and map information to the
+/// offloading runtime library. If there is no map or capture information,
+/// return nullptr by reference.
+static void
+emitOffloadingArrays(CodeGenFunction &CGF, llvm::Value *&BasePointersArray,
+                     llvm::Value *&PointersArray, llvm::Value *&SizesArray,
+                     llvm::Value *&MapTypesArray,
+                     MappableExprsHandler::MapValuesArrayTy &BasePointers,
+                     MappableExprsHandler::MapValuesArrayTy &Pointers,
+                     MappableExprsHandler::MapValuesArrayTy &Sizes,
+                     MappableExprsHandler::MapFlagsArrayTy &MapTypes) {
+  auto &CGM = CGF.CGM;
+  auto &Ctx = CGF.getContext();
+
+  BasePointersArray = PointersArray = SizesArray = MapTypesArray = nullptr;
+
+  if (unsigned PointerNumVal = BasePointers.size()) {
+    // Detect if we have any capture size requiring runtime evaluation of the
+    // size so that a constant array could be eventually used.
+    bool hasRuntimeEvaluationCaptureSize = false;
+    for (auto *S : Sizes)
+      if (!isa<llvm::Constant>(S)) {
+        hasRuntimeEvaluationCaptureSize = true;
+        break;
+      }
+
+    llvm::APInt PointerNumAP(32, PointerNumVal, /*isSigned=*/true);
+    QualType PointerArrayType =
+        Ctx.getConstantArrayType(Ctx.VoidPtrTy, PointerNumAP, ArrayType::Normal,
+                                 /*IndexTypeQuals=*/0);
+
+    BasePointersArray =
+        CGF.CreateMemTemp(PointerArrayType, ".offload_baseptrs").getPointer();
+    PointersArray =
+        CGF.CreateMemTemp(PointerArrayType, ".offload_ptrs").getPointer();
+
+    // If we don't have any VLA types or other types that require runtime
+    // evaluation, we can use a constant array for the map sizes, otherwise we
+    // need to fill up the arrays as we do for the pointers.
+    if (hasRuntimeEvaluationCaptureSize) {
+      QualType SizeArrayType = Ctx.getConstantArrayType(
+          Ctx.getSizeType(), PointerNumAP, ArrayType::Normal,
+          /*IndexTypeQuals=*/0);
+      SizesArray =
+          CGF.CreateMemTemp(SizeArrayType, ".offload_sizes").getPointer();
+    } else {
+      // We expect all the sizes to be constant, so we collect them to create
+      // a constant array.
+      SmallVector<llvm::Constant *, 16> ConstSizes;
+      for (auto S : Sizes)
+        ConstSizes.push_back(cast<llvm::Constant>(S));
+
+      auto *SizesArrayInit = llvm::ConstantArray::get(
+          llvm::ArrayType::get(CGM.SizeTy, ConstSizes.size()), ConstSizes);
+      auto *SizesArrayGbl = new llvm::GlobalVariable(
+          CGM.getModule(), SizesArrayInit->getType(),
+          /*isConstant=*/true, llvm::GlobalValue::PrivateLinkage,
+          SizesArrayInit, ".offload_sizes");
+      SizesArrayGbl->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
+      SizesArray = SizesArrayGbl;
+    }
+
+    // The map types are always constant so we don't need to generate code to
+    // fill arrays. Instead, we create an array constant.
+    llvm::Constant *MapTypesArrayInit =
+        llvm::ConstantDataArray::get(CGF.Builder.getContext(), MapTypes);
+    auto *MapTypesArrayGbl = new llvm::GlobalVariable(
+        CGM.getModule(), MapTypesArrayInit->getType(),
+        /*isConstant=*/true, llvm::GlobalValue::PrivateLinkage,
+        MapTypesArrayInit, ".offload_maptypes");
+    MapTypesArrayGbl->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
+    MapTypesArray = MapTypesArrayGbl;
+
+    for (unsigned i = 0; i < PointerNumVal; ++i) {
+      llvm::Value *BPVal = BasePointers[i];
+      if (BPVal->getType()->isPointerTy())
+        BPVal = CGF.Builder.CreateBitCast(BPVal, CGM.VoidPtrTy);
+      else {
+        assert(BPVal->getType()->isIntegerTy() &&
+               "If not a pointer, the value type must be an integer.");
+        BPVal = CGF.Builder.CreateIntToPtr(BPVal, CGM.VoidPtrTy);
+      }
+      llvm::Value *BP = CGF.Builder.CreateConstInBoundsGEP2_32(
+          llvm::ArrayType::get(CGM.VoidPtrTy, PointerNumVal), BasePointersArray,
+          0, i);
+      Address BPAddr(BP, Ctx.getTypeAlignInChars(Ctx.VoidPtrTy));
+      CGF.Builder.CreateStore(BPVal, BPAddr);
+
+      llvm::Value *PVal = Pointers[i];
+      if (PVal->getType()->isPointerTy())
+        PVal = CGF.Builder.CreateBitCast(PVal, CGM.VoidPtrTy);
+      else {
+        assert(PVal->getType()->isIntegerTy() &&
+               "If not a pointer, the value type must be an integer.");
+        PVal = CGF.Builder.CreateIntToPtr(PVal, CGM.VoidPtrTy);
+      }
+      llvm::Value *P = CGF.Builder.CreateConstInBoundsGEP2_32(
+          llvm::ArrayType::get(CGM.VoidPtrTy, PointerNumVal), PointersArray, 0,
+          i);
+      Address PAddr(P, Ctx.getTypeAlignInChars(Ctx.VoidPtrTy));
+      CGF.Builder.CreateStore(PVal, PAddr);
+
+      if (hasRuntimeEvaluationCaptureSize) {
+        llvm::Value *S = CGF.Builder.CreateConstInBoundsGEP2_32(
+            llvm::ArrayType::get(CGM.SizeTy, PointerNumVal), SizesArray,
+            /*Idx0=*/0,
+            /*Idx1=*/i);
+        Address SAddr(S, Ctx.getTypeAlignInChars(Ctx.getSizeType()));
+        CGF.Builder.CreateStore(
+            CGF.Builder.CreateIntCast(Sizes[i], CGM.SizeTy, /*isSigned=*/true),
+            SAddr);
+      }
+    }
+  }
+}
+/// \brief Emit the arguments to be passed to the runtime library based on the
+/// arrays of pointers, sizes and map types.
+static void emitOffloadingArraysArgument(
+    CodeGenFunction &CGF, llvm::Value *&BasePointersArrayArg,
+    llvm::Value *&PointersArrayArg, llvm::Value *&SizesArrayArg,
+    llvm::Value *&MapTypesArrayArg, llvm::Value *BasePointersArray,
+    llvm::Value *PointersArray, llvm::Value *SizesArray,
+    llvm::Value *MapTypesArray, unsigned NumElems) {
+  auto &CGM = CGF.CGM;
+  if (NumElems) {
+    BasePointersArrayArg = CGF.Builder.CreateConstInBoundsGEP2_32(
+        llvm::ArrayType::get(CGM.VoidPtrTy, NumElems), BasePointersArray,
+        /*Idx0=*/0, /*Idx1=*/0);
+    PointersArrayArg = CGF.Builder.CreateConstInBoundsGEP2_32(
+        llvm::ArrayType::get(CGM.VoidPtrTy, NumElems), PointersArray,
+        /*Idx0=*/0,
+        /*Idx1=*/0);
+    SizesArrayArg = CGF.Builder.CreateConstInBoundsGEP2_32(
+        llvm::ArrayType::get(CGM.SizeTy, NumElems), SizesArray,
+        /*Idx0=*/0, /*Idx1=*/0);
+    MapTypesArrayArg = CGF.Builder.CreateConstInBoundsGEP2_32(
+        llvm::ArrayType::get(CGM.Int32Ty, NumElems), MapTypesArray,
+        /*Idx0=*/0,
+        /*Idx1=*/0);
+  } else {
+    BasePointersArrayArg = llvm::ConstantPointerNull::get(CGM.VoidPtrPtrTy);
+    PointersArrayArg = llvm::ConstantPointerNull::get(CGM.VoidPtrPtrTy);
+    SizesArrayArg = llvm::ConstantPointerNull::get(CGM.SizeTy->getPointerTo());
+    MapTypesArrayArg =
+        llvm::ConstantPointerNull::get(CGM.Int32Ty->getPointerTo());
+  }
+}
+
+void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
+                                     const OMPExecutableDirective &D,
+                                     llvm::Value *OutlinedFn,
+                                     llvm::Value *OutlinedFnID,
+                                     const Expr *IfCond, const Expr *Device,
+                                     ArrayRef<llvm::Value *> CapturedVars) {
+  if (!CGF.HaveInsertPoint())
+    return;
 
   assert(OutlinedFn && "Invalid outlined function!");
 
   auto &Ctx = CGF.getContext();
 
-  // Fill up the arrays with the all the captured variables.
-  SmallVector<llvm::Value *, 16> BasePointers;
-  SmallVector<llvm::Value *, 16> Pointers;
-  SmallVector<llvm::Value *, 16> Sizes;
-  SmallVector<unsigned, 16> MapTypes;
+  // Fill up the arrays with all the captured variables.
+  MappableExprsHandler::MapValuesArrayTy KernelArgs;
+  MappableExprsHandler::MapValuesArrayTy BasePointers;
+  MappableExprsHandler::MapValuesArrayTy Pointers;
+  MappableExprsHandler::MapValuesArrayTy Sizes;
+  MappableExprsHandler::MapFlagsArrayTy MapTypes;
 
-  bool hasVLACaptures = false;
+  MappableExprsHandler::MapValuesArrayTy CurBasePointers;
+  MappableExprsHandler::MapValuesArrayTy CurPointers;
+  MappableExprsHandler::MapValuesArrayTy CurSizes;
+  MappableExprsHandler::MapFlagsArrayTy CurMapTypes;
+
+  // Get mappable expression information.
+  MappableExprsHandler MEHandler(D, CGF);
 
   const CapturedStmt &CS = *cast<CapturedStmt>(D.getAssociatedStmt());
   auto RI = CS.getCapturedRecordDecl()->field_begin();
-  // auto II = CS.capture_init_begin();
   auto CV = CapturedVars.begin();
   for (CapturedStmt::const_capture_iterator CI = CS.capture_begin(),
                                             CE = CS.capture_end();
        CI != CE; ++CI, ++RI, ++CV) {
     StringRef Name;
     QualType Ty;
-    llvm::Value *BasePointer;
-    llvm::Value *Pointer;
-    llvm::Value *Size;
-    unsigned MapType;
 
-    // VLA sizes are passed to the outlined region by copy.
+    CurBasePointers.clear();
+    CurPointers.clear();
+    CurSizes.clear();
+    CurMapTypes.clear();
+
+    // VLA sizes are passed to the outlined region by copy and do not have map
+    // information associated.
     if (CI->capturesVariableArrayType()) {
-      BasePointer = Pointer = *CV;
-      Size = getTypeSize(CGF, RI->getType());
+      CurBasePointers.push_back(*CV);
+      CurPointers.push_back(*CV);
+      CurSizes.push_back(CGF.getTypeSize(RI->getType()));
       // Copy to the device as an argument. No need to retrieve it.
-      MapType = OMP_MAP_BYCOPY;
-      hasVLACaptures = true;
-    } else if (CI->capturesThis()) {
-      BasePointer = Pointer = *CV;
-      const PointerType *PtrTy = cast<PointerType>(RI->getType().getTypePtr());
-      Size = getTypeSize(CGF, PtrTy->getPointeeType());
-      // Default map type.
-      MapType = OMP_MAP_TO | OMP_MAP_FROM;
-    } else if (CI->capturesVariableByCopy()) {
-      MapType = OMP_MAP_BYCOPY;
-      if (!RI->getType()->isAnyPointerType()) {
-        // If the field is not a pointer, we need to save the actual value and
-        // load it as a void pointer.
-        auto DstAddr = CGF.CreateMemTemp(
-            Ctx.getUIntPtrType(),
-            Twine(CI->getCapturedVar()->getName()) + ".casted");
-        LValue DstLV = CGF.MakeAddrLValue(DstAddr, Ctx.getUIntPtrType());
-
-        auto *SrcAddrVal = CGF.EmitScalarConversion(
-            DstAddr.getPointer(), Ctx.getPointerType(Ctx.getUIntPtrType()),
-            Ctx.getPointerType(RI->getType()), SourceLocation());
-        LValue SrcLV =
-            CGF.MakeNaturalAlignAddrLValue(SrcAddrVal, RI->getType());
-
-        // Store the value using the source type pointer.
-        CGF.EmitStoreThroughLValue(RValue::get(*CV), SrcLV);
-
-        // Load the value using the destination type pointer.
-        BasePointer = Pointer =
-            CGF.EmitLoadOfLValue(DstLV, SourceLocation()).getScalarVal();
-      } else {
-        MapType |= OMP_MAP_PTR;
-        BasePointer = Pointer = *CV;
-      }
-      Size = getTypeSize(CGF, RI->getType());
+      CurMapTypes.push_back(MappableExprsHandler::OMP_MAP_PRIVATE_VAL |
+                            MappableExprsHandler::OMP_MAP_FIRST_REF);
     } else {
-      assert(CI->capturesVariable() && "Expected captured reference.");
-      BasePointer = Pointer = *CV;
-
-      const ReferenceType *PtrTy =
-          cast<ReferenceType>(RI->getType().getTypePtr());
-      QualType ElementType = PtrTy->getPointeeType();
-      Size = getTypeSize(CGF, ElementType);
-      // The default map type for a scalar/complex type is 'to' because by
-      // default the value doesn't have to be retrieved. For an aggregate type,
-      // the default is 'tofrom'.
-      MapType = ElementType->isAggregateType() ? (OMP_MAP_TO | OMP_MAP_FROM)
-                                               : OMP_MAP_TO;
-      if (ElementType->isAnyPointerType())
-        MapType |= OMP_MAP_PTR;
+      // If we have any information in the map clause, we use it, otherwise we
+      // just do a default mapping.
+      MEHandler.generateInfoForCapture(CI, CurBasePointers, CurPointers,
+                                       CurSizes, CurMapTypes);
+      if (CurBasePointers.empty())
+        MEHandler.generateDefaultMapInfo(*CI, **RI, *CV, CurBasePointers,
+                                         CurPointers, CurSizes, CurMapTypes);
     }
-
-    BasePointers.push_back(BasePointer);
-    Pointers.push_back(Pointer);
-    Sizes.push_back(Size);
-    MapTypes.push_back(MapType);
+    // We expect to have at least an element of information for this capture.
+    assert(!CurBasePointers.empty() && "Non-existing map pointer for capture!");
+    assert(CurBasePointers.size() == CurPointers.size() &&
+           CurBasePointers.size() == CurSizes.size() &&
+           CurBasePointers.size() == CurMapTypes.size() &&
+           "Inconsistent map information sizes!");
+
+    // The kernel args are always the first elements of the base pointers
+    // associated with a capture.
+    KernelArgs.push_back(CurBasePointers.front());
+    // We need to append the results of this capture to what we already have.
+    BasePointers.append(CurBasePointers.begin(), CurBasePointers.end());
+    Pointers.append(CurPointers.begin(), CurPointers.end());
+    Sizes.append(CurSizes.begin(), CurSizes.end());
+    MapTypes.append(CurMapTypes.begin(), CurMapTypes.end());
   }
 
   // Keep track on whether the host function has to be executed.
@@ -3943,128 +5779,22 @@ void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
                         OffloadError);
 
   // Fill up the pointer arrays and transfer execution to the device.
-  auto &&ThenGen = [this, &Ctx, &BasePointers, &Pointers, &Sizes, &MapTypes,
-                    hasVLACaptures, Device, OutlinedFnID, OffloadError,
-                    OffloadErrorQType](CodeGenFunction &CGF) {
-    unsigned PointerNumVal = BasePointers.size();
-    llvm::Value *PointerNum = CGF.Builder.getInt32(PointerNumVal);
+  auto &&ThenGen = [&Ctx, &BasePointers, &Pointers, &Sizes, &MapTypes, Device,
+                    OutlinedFnID, OffloadError, OffloadErrorQType,
+                    &D](CodeGenFunction &CGF, PrePostActionTy &) {
+    auto &RT = CGF.CGM.getOpenMPRuntime();
+    // Emit the offloading arrays.
     llvm::Value *BasePointersArray;
     llvm::Value *PointersArray;
     llvm::Value *SizesArray;
     llvm::Value *MapTypesArray;
-
-    if (PointerNumVal) {
-      llvm::APInt PointerNumAP(32, PointerNumVal, /*isSigned=*/true);
-      QualType PointerArrayType = Ctx.getConstantArrayType(
-          Ctx.VoidPtrTy, PointerNumAP, ArrayType::Normal,
-          /*IndexTypeQuals=*/0);
-
-      BasePointersArray =
-          CGF.CreateMemTemp(PointerArrayType, ".offload_baseptrs").getPointer();
-      PointersArray =
-          CGF.CreateMemTemp(PointerArrayType, ".offload_ptrs").getPointer();
-
-      // If we don't have any VLA types, we can use a constant array for the map
-      // sizes, otherwise we need to fill up the arrays as we do for the
-      // pointers.
-      if (hasVLACaptures) {
-        QualType SizeArrayType = Ctx.getConstantArrayType(
-            Ctx.getSizeType(), PointerNumAP, ArrayType::Normal,
-            /*IndexTypeQuals=*/0);
-        SizesArray =
-            CGF.CreateMemTemp(SizeArrayType, ".offload_sizes").getPointer();
-      } else {
-        // We expect all the sizes to be constant, so we collect them to create
-        // a constant array.
-        SmallVector<llvm::Constant *, 16> ConstSizes;
-        for (auto S : Sizes)
-          ConstSizes.push_back(cast<llvm::Constant>(S));
-
-        auto *SizesArrayInit = llvm::ConstantArray::get(
-            llvm::ArrayType::get(CGM.SizeTy, ConstSizes.size()), ConstSizes);
-        auto *SizesArrayGbl = new llvm::GlobalVariable(
-            CGM.getModule(), SizesArrayInit->getType(),
-            /*isConstant=*/true, llvm::GlobalValue::PrivateLinkage,
-            SizesArrayInit, ".offload_sizes");
-        SizesArrayGbl->setUnnamedAddr(true);
-        SizesArray = SizesArrayGbl;
-      }
-
-      // The map types are always constant so we don't need to generate code to
-      // fill arrays. Instead, we create an array constant.
-      llvm::Constant *MapTypesArrayInit =
-          llvm::ConstantDataArray::get(CGF.Builder.getContext(), MapTypes);
-      auto *MapTypesArrayGbl = new llvm::GlobalVariable(
-          CGM.getModule(), MapTypesArrayInit->getType(),
-          /*isConstant=*/true, llvm::GlobalValue::PrivateLinkage,
-          MapTypesArrayInit, ".offload_maptypes");
-      MapTypesArrayGbl->setUnnamedAddr(true);
-      MapTypesArray = MapTypesArrayGbl;
-
-      for (unsigned i = 0; i < PointerNumVal; ++i) {
-
-        llvm::Value *BPVal = BasePointers[i];
-        if (BPVal->getType()->isPointerTy())
-          BPVal = CGF.Builder.CreateBitCast(BPVal, CGM.VoidPtrTy);
-        else {
-          assert(BPVal->getType()->isIntegerTy() &&
-                 "If not a pointer, the value type must be an integer.");
-          BPVal = CGF.Builder.CreateIntToPtr(BPVal, CGM.VoidPtrTy);
-        }
-        llvm::Value *BP = CGF.Builder.CreateConstInBoundsGEP2_32(
-            llvm::ArrayType::get(CGM.VoidPtrTy, PointerNumVal),
-            BasePointersArray, 0, i);
-        Address BPAddr(BP, Ctx.getTypeAlignInChars(Ctx.VoidPtrTy));
-        CGF.Builder.CreateStore(BPVal, BPAddr);
-
-        llvm::Value *PVal = Pointers[i];
-        if (PVal->getType()->isPointerTy())
-          PVal = CGF.Builder.CreateBitCast(PVal, CGM.VoidPtrTy);
-        else {
-          assert(PVal->getType()->isIntegerTy() &&
-                 "If not a pointer, the value type must be an integer.");
-          PVal = CGF.Builder.CreateIntToPtr(PVal, CGM.VoidPtrTy);
-        }
-        llvm::Value *P = CGF.Builder.CreateConstInBoundsGEP2_32(
-            llvm::ArrayType::get(CGM.VoidPtrTy, PointerNumVal), PointersArray,
-            0, i);
-        Address PAddr(P, Ctx.getTypeAlignInChars(Ctx.VoidPtrTy));
-        CGF.Builder.CreateStore(PVal, PAddr);
-
-        if (hasVLACaptures) {
-          llvm::Value *S = CGF.Builder.CreateConstInBoundsGEP2_32(
-              llvm::ArrayType::get(CGM.SizeTy, PointerNumVal), SizesArray,
-              /*Idx0=*/0,
-              /*Idx1=*/i);
-          Address SAddr(S, Ctx.getTypeAlignInChars(Ctx.getSizeType()));
-          CGF.Builder.CreateStore(CGF.Builder.CreateIntCast(
-                                      Sizes[i], CGM.SizeTy, /*isSigned=*/true),
-                                  SAddr);
-        }
-      }
-
-      BasePointersArray = CGF.Builder.CreateConstInBoundsGEP2_32(
-          llvm::ArrayType::get(CGM.VoidPtrTy, PointerNumVal), BasePointersArray,
-          /*Idx0=*/0, /*Idx1=*/0);
-      PointersArray = CGF.Builder.CreateConstInBoundsGEP2_32(
-          llvm::ArrayType::get(CGM.VoidPtrTy, PointerNumVal), PointersArray,
-          /*Idx0=*/0,
-          /*Idx1=*/0);
-      SizesArray = CGF.Builder.CreateConstInBoundsGEP2_32(
-          llvm::ArrayType::get(CGM.SizeTy, PointerNumVal), SizesArray,
-          /*Idx0=*/0, /*Idx1=*/0);
-      MapTypesArray = CGF.Builder.CreateConstInBoundsGEP2_32(
-          llvm::ArrayType::get(CGM.Int32Ty, PointerNumVal), MapTypesArray,
-          /*Idx0=*/0,
-          /*Idx1=*/0);
-
-    } else {
-      BasePointersArray = llvm::ConstantPointerNull::get(CGM.VoidPtrPtrTy);
-      PointersArray = llvm::ConstantPointerNull::get(CGM.VoidPtrPtrTy);
-      SizesArray = llvm::ConstantPointerNull::get(CGM.SizeTy->getPointerTo());
-      MapTypesArray =
-          llvm::ConstantPointerNull::get(CGM.Int32Ty->getPointerTo());
-    }
+    emitOffloadingArrays(CGF, BasePointersArray, PointersArray, SizesArray,
+                         MapTypesArray, BasePointers, Pointers, Sizes,
+                         MapTypes);
+    emitOffloadingArraysArgument(CGF, BasePointersArray, PointersArray,
+                                 SizesArray, MapTypesArray, BasePointersArray,
+                                 PointersArray, SizesArray, MapTypesArray,
+                                 BasePointers.size());
 
     // On top of the arrays that were filled up, the target offloading call
     // takes as arguments the device id as well as the host pointer. The host
@@ -4082,23 +5812,48 @@ void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
     llvm::Value *DeviceID;
     if (Device)
       DeviceID = CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(Device),
-                                           CGM.Int32Ty, /*isSigned=*/true);
+                                           CGF.Int32Ty, /*isSigned=*/true);
     else
       DeviceID = CGF.Builder.getInt32(OMP_DEVICEID_UNDEF);
 
-    llvm::Value *OffloadingArgs[] = {
-        DeviceID,      OutlinedFnID, PointerNum,   BasePointersArray,
-        PointersArray, SizesArray,   MapTypesArray};
-    auto Return = CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__tgt_target),
-                                      OffloadingArgs);
+    // Emit the number of elements in the offloading arrays.
+    llvm::Value *PointerNum = CGF.Builder.getInt32(BasePointers.size());
+
+    // Return value of the runtime offloading call.
+    llvm::Value *Return;
+
+    auto *NumTeams = emitNumTeamsClauseForTargetDirective(RT, CGF, D);
+    auto *ThreadLimit = emitThreadLimitClauseForTargetDirective(RT, CGF, D);
+
+    // If we have NumTeams defined this means that we have an enclosed teams
+    // region. Therefore we also expect to have ThreadLimit defined. These two
+    // values should be defined in the presence of a teams directive, regardless
+    // of having any clauses associated. If the user is using teams but no
+    // clauses, these two values will be the default that should be passed to
+    // the runtime library - a 32-bit integer with the value zero.
+    if (NumTeams) {
+      assert(ThreadLimit && "Thread limit expression should be available along "
+                            "with number of teams.");
+      llvm::Value *OffloadingArgs[] = {
+          DeviceID,          OutlinedFnID,  PointerNum,
+          BasePointersArray, PointersArray, SizesArray,
+          MapTypesArray,     NumTeams,      ThreadLimit};
+      Return = CGF.EmitRuntimeCall(
+          RT.createRuntimeFunction(OMPRTL__tgt_target_teams), OffloadingArgs);
+    } else {
+      llvm::Value *OffloadingArgs[] = {
+          DeviceID,      OutlinedFnID, PointerNum,   BasePointersArray,
+          PointersArray, SizesArray,   MapTypesArray};
+      Return = CGF.EmitRuntimeCall(RT.createRuntimeFunction(OMPRTL__tgt_target),
+                                   OffloadingArgs);
+    }
 
     CGF.EmitStoreOfScalar(Return, OffloadError);
   };
 
   // Notify that the host version must be executed.
-  auto &&ElseGen = [this, OffloadError,
-                    OffloadErrorQType](CodeGenFunction &CGF) {
-    CGF.EmitStoreOfScalar(llvm::ConstantInt::get(CGM.Int32Ty, /*V=*/-1u),
+  auto &&ElseGen = [OffloadError](CodeGenFunction &CGF, PrePostActionTy &) {
+    CGF.EmitStoreOfScalar(llvm::ConstantInt::get(CGF.Int32Ty, /*V=*/-1u),
                           OffloadError);
   };
 
@@ -4107,15 +5862,15 @@ void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
   // regardless of the conditional in the if clause if, e.g., the user do not
   // specify target triples.
   if (OutlinedFnID) {
-    if (IfCond) {
+    if (IfCond)
       emitOMPIfClause(CGF, IfCond, ThenGen, ElseGen);
-    } else {
-      CodeGenFunction::RunCleanupsScope Scope(CGF);
-      ThenGen(CGF);
+    else {
+      RegionCodeGenTy ThenRCG(ThenGen);
+      ThenRCG(CGF);
     }
   } else {
-    CodeGenFunction::RunCleanupsScope Scope(CGF);
-    ElseGen(CGF);
+    RegionCodeGenTy ElseRCG(ElseGen);
+    ElseRCG(CGF);
   }
 
   // Check the error code and execute the host version if required.
@@ -4126,11 +5881,10 @@ void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
   CGF.Builder.CreateCondBr(Failed, OffloadFailedBlock, OffloadContBlock);
 
   CGF.EmitBlock(OffloadFailedBlock);
-  CGF.Builder.CreateCall(OutlinedFn, BasePointers);
+  CGF.Builder.CreateCall(OutlinedFn, KernelArgs);
   CGF.EmitBranch(OffloadContBlock);
 
   CGF.EmitBlock(OffloadContBlock, /*IsFinished=*/true);
-  return;
 }
 
 void CGOpenMPRuntime::scanForTargetRegionsFunctions(const Stmt *S,
@@ -4148,26 +5902,27 @@ void CGOpenMPRuntime::scanForTargetRegionsFunctions(const Stmt *S,
     unsigned DeviceID;
     unsigned FileID;
     unsigned Line;
-    unsigned Column;
     getTargetEntryUniqueInfo(CGM.getContext(), E->getLocStart(), DeviceID,
-                             FileID, Line, Column);
+                             FileID, Line);
 
     // Is this a target region that should not be emitted as an entry point? If
     // so just signal we are done with this target region.
-    if (!OffloadEntriesInfoManager.hasTargetRegionEntryInfo(
-            DeviceID, FileID, ParentName, Line, Column))
+    if (!OffloadEntriesInfoManager.hasTargetRegionEntryInfo(DeviceID, FileID,
+                                                            ParentName, Line))
       return;
 
     llvm::Function *Fn;
     llvm::Constant *Addr;
-    emitTargetOutlinedFunction(*E, ParentName, Fn, Addr,
-                               /*isOffloadEntry=*/true);
+    std::tie(Fn, Addr) =
+        CodeGenFunction::EmitOMPTargetDirectiveOutlinedFunction(
+            CGM, cast<OMPTargetDirective>(*E), ParentName,
+            /*isOffloadEntry=*/true);
     assert(Fn && Addr && "Target region emission failed.");
     return;
   }
 
   if (const OMPExecutableDirective *E = dyn_cast<OMPExecutableDirective>(S)) {
-    if (!E->getAssociatedStmt())
+    if (!E->hasAssociatedStmt())
       return;
 
     scanForTargetRegionsFunctions(
@@ -4183,8 +5938,6 @@ void CGOpenMPRuntime::scanForTargetRegionsFunctions(const Stmt *S,
   // Keep looking for target regions recursively.
   for (auto *II : S->children())
     scanForTargetRegionsFunctions(II, ParentName);
-
-  return;
 }
 
 bool CGOpenMPRuntime::emitTargetFunctions(GlobalDecl GD) {
@@ -4249,3 +6002,594 @@ llvm::Function *CGOpenMPRuntime::emitRegistrationFunction() {
   // compilation unit.
   return createOffloadingBinaryDescriptorRegistration();
 }
+
+void CGOpenMPRuntime::emitTeamsCall(CodeGenFunction &CGF,
+                                    const OMPExecutableDirective &D,
+                                    SourceLocation Loc,
+                                    llvm::Value *OutlinedFn,
+                                    ArrayRef<llvm::Value *> CapturedVars) {
+  if (!CGF.HaveInsertPoint())
+    return;
+
+  auto *RTLoc = emitUpdateLocation(CGF, Loc);
+  CodeGenFunction::RunCleanupsScope Scope(CGF);
+
+  // Build call __kmpc_fork_teams(loc, n, microtask, var1, .., varn);
+  llvm::Value *Args[] = {
+      RTLoc,
+      CGF.Builder.getInt32(CapturedVars.size()), // Number of captured vars
+      CGF.Builder.CreateBitCast(OutlinedFn, getKmpc_MicroPointerTy())};
+  llvm::SmallVector<llvm::Value *, 16> RealArgs;
+  RealArgs.append(std::begin(Args), std::end(Args));
+  RealArgs.append(CapturedVars.begin(), CapturedVars.end());
+
+  auto RTLFn = createRuntimeFunction(OMPRTL__kmpc_fork_teams);
+  CGF.EmitRuntimeCall(RTLFn, RealArgs);
+}
+
+void CGOpenMPRuntime::emitNumTeamsClause(CodeGenFunction &CGF,
+                                         const Expr *NumTeams,
+                                         const Expr *ThreadLimit,
+                                         SourceLocation Loc) {
+  if (!CGF.HaveInsertPoint())
+    return;
+
+  auto *RTLoc = emitUpdateLocation(CGF, Loc);
+
+  llvm::Value *NumTeamsVal =
+      (NumTeams)
+          ? CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(NumTeams),
+                                      CGF.CGM.Int32Ty, /* isSigned = */ true)
+          : CGF.Builder.getInt32(0);
+
+  llvm::Value *ThreadLimitVal =
+      (ThreadLimit)
+          ? CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(ThreadLimit),
+                                      CGF.CGM.Int32Ty, /* isSigned = */ true)
+          : CGF.Builder.getInt32(0);
+
+  // Build call __kmpc_push_num_teamss(&loc, global_tid, num_teams, thread_limit)
+  llvm::Value *PushNumTeamsArgs[] = {RTLoc, getThreadID(CGF, Loc), NumTeamsVal,
+                                     ThreadLimitVal};
+  CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__kmpc_push_num_teams),
+                      PushNumTeamsArgs);
+}
+
+void CGOpenMPRuntime::emitTargetDataCalls(CodeGenFunction &CGF,
+                                          const OMPExecutableDirective &D,
+                                          const Expr *IfCond,
+                                          const Expr *Device,
+                                          const RegionCodeGenTy &CodeGen) {
+
+  if (!CGF.HaveInsertPoint())
+    return;
+
+  llvm::Value *BasePointersArray = nullptr;
+  llvm::Value *PointersArray = nullptr;
+  llvm::Value *SizesArray = nullptr;
+  llvm::Value *MapTypesArray = nullptr;
+  unsigned NumOfPtrs = 0;
+
+  // Generate the code for the opening of the data environment. Capture all the
+  // arguments of the runtime call by reference because they are used in the
+  // closing of the region.
+  auto &&BeginThenGen = [&D, &CGF, &BasePointersArray, &PointersArray,
+                         &SizesArray, &MapTypesArray, Device,
+                         &NumOfPtrs](CodeGenFunction &CGF, PrePostActionTy &) {
+    // Fill up the arrays with all the mapped variables.
+    MappableExprsHandler::MapValuesArrayTy BasePointers;
+    MappableExprsHandler::MapValuesArrayTy Pointers;
+    MappableExprsHandler::MapValuesArrayTy Sizes;
+    MappableExprsHandler::MapFlagsArrayTy MapTypes;
+
+    // Get map clause information.
+    MappableExprsHandler MCHandler(D, CGF);
+    MCHandler.generateAllInfo(BasePointers, Pointers, Sizes, MapTypes);
+    NumOfPtrs = BasePointers.size();
+
+    // Fill up the arrays and create the arguments.
+    emitOffloadingArrays(CGF, BasePointersArray, PointersArray, SizesArray,
+                         MapTypesArray, BasePointers, Pointers, Sizes,
+                         MapTypes);
+
+    llvm::Value *BasePointersArrayArg = nullptr;
+    llvm::Value *PointersArrayArg = nullptr;
+    llvm::Value *SizesArrayArg = nullptr;
+    llvm::Value *MapTypesArrayArg = nullptr;
+    emitOffloadingArraysArgument(CGF, BasePointersArrayArg, PointersArrayArg,
+                                 SizesArrayArg, MapTypesArrayArg,
+                                 BasePointersArray, PointersArray, SizesArray,
+                                 MapTypesArray, NumOfPtrs);
+
+    // Emit device ID if any.
+    llvm::Value *DeviceID = nullptr;
+    if (Device)
+      DeviceID = CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(Device),
+                                           CGF.Int32Ty, /*isSigned=*/true);
+    else
+      DeviceID = CGF.Builder.getInt32(OMP_DEVICEID_UNDEF);
+
+    // Emit the number of elements in the offloading arrays.
+    auto *PointerNum = CGF.Builder.getInt32(NumOfPtrs);
+
+    llvm::Value *OffloadingArgs[] = {
+        DeviceID,         PointerNum,    BasePointersArrayArg,
+        PointersArrayArg, SizesArrayArg, MapTypesArrayArg};
+    auto &RT = CGF.CGM.getOpenMPRuntime();
+    CGF.EmitRuntimeCall(RT.createRuntimeFunction(OMPRTL__tgt_target_data_begin),
+                        OffloadingArgs);
+  };
+
+  // Generate code for the closing of the data region.
+  auto &&EndThenGen = [&CGF, &BasePointersArray, &PointersArray, &SizesArray,
+                       &MapTypesArray, Device,
+                       &NumOfPtrs](CodeGenFunction &CGF, PrePostActionTy &) {
+    assert(BasePointersArray && PointersArray && SizesArray && MapTypesArray &&
+           NumOfPtrs && "Invalid data environment closing arguments.");
+
+    llvm::Value *BasePointersArrayArg = nullptr;
+    llvm::Value *PointersArrayArg = nullptr;
+    llvm::Value *SizesArrayArg = nullptr;
+    llvm::Value *MapTypesArrayArg = nullptr;
+    emitOffloadingArraysArgument(CGF, BasePointersArrayArg, PointersArrayArg,
+                                 SizesArrayArg, MapTypesArrayArg,
+                                 BasePointersArray, PointersArray, SizesArray,
+                                 MapTypesArray, NumOfPtrs);
+
+    // Emit device ID if any.
+    llvm::Value *DeviceID = nullptr;
+    if (Device)
+      DeviceID = CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(Device),
+                                           CGF.Int32Ty, /*isSigned=*/true);
+    else
+      DeviceID = CGF.Builder.getInt32(OMP_DEVICEID_UNDEF);
+
+    // Emit the number of elements in the offloading arrays.
+    auto *PointerNum = CGF.Builder.getInt32(NumOfPtrs);
+
+    llvm::Value *OffloadingArgs[] = {
+        DeviceID,         PointerNum,    BasePointersArrayArg,
+        PointersArrayArg, SizesArrayArg, MapTypesArrayArg};
+    auto &RT = CGF.CGM.getOpenMPRuntime();
+    CGF.EmitRuntimeCall(RT.createRuntimeFunction(OMPRTL__tgt_target_data_end),
+                        OffloadingArgs);
+  };
+
+  // In the event we get an if clause, we don't have to take any action on the
+  // else side.
+  auto &&ElseGen = [](CodeGenFunction &CGF, PrePostActionTy &) {};
+
+  if (IfCond) {
+    emitOMPIfClause(CGF, IfCond, BeginThenGen, ElseGen);
+  } else {
+    RegionCodeGenTy BeginThenRCG(BeginThenGen);
+    BeginThenRCG(CGF);
+  }
+
+  CGM.getOpenMPRuntime().emitInlinedDirective(CGF, OMPD_target_data, CodeGen);
+
+  if (IfCond) {
+    emitOMPIfClause(CGF, IfCond, EndThenGen, ElseGen);
+  } else {
+    RegionCodeGenTy EndThenRCG(EndThenGen);
+    EndThenRCG(CGF);
+  }
+}
+
+void CGOpenMPRuntime::emitTargetDataStandAloneCall(
+    CodeGenFunction &CGF, const OMPExecutableDirective &D, const Expr *IfCond,
+    const Expr *Device) {
+  if (!CGF.HaveInsertPoint())
+    return;
+
+  assert((isa<OMPTargetEnterDataDirective>(D) ||
+          isa<OMPTargetExitDataDirective>(D) ||
+          isa<OMPTargetUpdateDirective>(D)) &&
+         "Expecting either target enter, exit data, or update directives.");
+
+  // Generate the code for the opening of the data environment.
+  auto &&ThenGen = [&D, &CGF, Device](CodeGenFunction &CGF, PrePostActionTy &) {
+    // Fill up the arrays with all the mapped variables.
+    MappableExprsHandler::MapValuesArrayTy BasePointers;
+    MappableExprsHandler::MapValuesArrayTy Pointers;
+    MappableExprsHandler::MapValuesArrayTy Sizes;
+    MappableExprsHandler::MapFlagsArrayTy MapTypes;
+
+    // Get map clause information.
+    MappableExprsHandler MEHandler(D, CGF);
+    MEHandler.generateAllInfo(BasePointers, Pointers, Sizes, MapTypes);
+
+    llvm::Value *BasePointersArrayArg = nullptr;
+    llvm::Value *PointersArrayArg = nullptr;
+    llvm::Value *SizesArrayArg = nullptr;
+    llvm::Value *MapTypesArrayArg = nullptr;
+
+    // Fill up the arrays and create the arguments.
+    emitOffloadingArrays(CGF, BasePointersArrayArg, PointersArrayArg,
+                         SizesArrayArg, MapTypesArrayArg, BasePointers,
+                         Pointers, Sizes, MapTypes);
+    emitOffloadingArraysArgument(
+        CGF, BasePointersArrayArg, PointersArrayArg, SizesArrayArg,
+        MapTypesArrayArg, BasePointersArrayArg, PointersArrayArg, SizesArrayArg,
+        MapTypesArrayArg, BasePointers.size());
+
+    // Emit device ID if any.
+    llvm::Value *DeviceID = nullptr;
+    if (Device)
+      DeviceID = CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(Device),
+                                           CGF.Int32Ty, /*isSigned=*/true);
+    else
+      DeviceID = CGF.Builder.getInt32(OMP_DEVICEID_UNDEF);
+
+    // Emit the number of elements in the offloading arrays.
+    auto *PointerNum = CGF.Builder.getInt32(BasePointers.size());
+
+    llvm::Value *OffloadingArgs[] = {
+        DeviceID,         PointerNum,    BasePointersArrayArg,
+        PointersArrayArg, SizesArrayArg, MapTypesArrayArg};
+
+    auto &RT = CGF.CGM.getOpenMPRuntime();
+    // Select the right runtime function call for each expected standalone
+    // directive.
+    OpenMPRTLFunction RTLFn;
+    switch (D.getDirectiveKind()) {
+    default:
+      llvm_unreachable("Unexpected standalone target data directive.");
+      break;
+    case OMPD_target_enter_data:
+      RTLFn = OMPRTL__tgt_target_data_begin;
+      break;
+    case OMPD_target_exit_data:
+      RTLFn = OMPRTL__tgt_target_data_end;
+      break;
+    case OMPD_target_update:
+      RTLFn = OMPRTL__tgt_target_data_update;
+      break;
+    }
+    CGF.EmitRuntimeCall(RT.createRuntimeFunction(RTLFn), OffloadingArgs);
+  };
+
+  // In the event we get an if clause, we don't have to take any action on the
+  // else side.
+  auto &&ElseGen = [](CodeGenFunction &CGF, PrePostActionTy &) {};
+
+  if (IfCond) {
+    emitOMPIfClause(CGF, IfCond, ThenGen, ElseGen);
+  } else {
+    RegionCodeGenTy ThenGenRCG(ThenGen);
+    ThenGenRCG(CGF);
+  }
+}
+
+namespace {
+  /// Kind of parameter in a function with 'declare simd' directive.
+  enum ParamKindTy { LinearWithVarStride, Linear, Uniform, Vector };
+  /// Attribute set of the parameter.
+  struct ParamAttrTy {
+    ParamKindTy Kind = Vector;
+    llvm::APSInt StrideOrArg;
+    llvm::APSInt Alignment;
+  };
+} // namespace
+
+static unsigned evaluateCDTSize(const FunctionDecl *FD,
+                                ArrayRef<ParamAttrTy> ParamAttrs) {
+  // Every vector variant of a SIMD-enabled function has a vector length (VLEN).
+  // If OpenMP clause "simdlen" is used, the VLEN is the value of the argument
+  // of that clause. The VLEN value must be power of 2.
+  // In other case the notion of the function`s "characteristic data type" (CDT)
+  // is used to compute the vector length.
+  // CDT is defined in the following order:
+  //   a) For non-void function, the CDT is the return type.
+  //   b) If the function has any non-uniform, non-linear parameters, then the
+  //   CDT is the type of the first such parameter.
+  //   c) If the CDT determined by a) or b) above is struct, union, or class
+  //   type which is pass-by-value (except for the type that maps to the
+  //   built-in complex data type), the characteristic data type is int.
+  //   d) If none of the above three cases is applicable, the CDT is int.
+  // The VLEN is then determined based on the CDT and the size of vector
+  // register of that ISA for which current vector version is generated. The
+  // VLEN is computed using the formula below:
+  //   VLEN  = sizeof(vector_register) / sizeof(CDT),
+  // where vector register size specified in section 3.2.1 Registers and the
+  // Stack Frame of original AMD64 ABI document.
+  QualType RetType = FD->getReturnType();
+  if (RetType.isNull())
+    return 0;
+  ASTContext &C = FD->getASTContext();
+  QualType CDT;
+  if (!RetType.isNull() && !RetType->isVoidType())
+    CDT = RetType;
+  else {
+    unsigned Offset = 0;
+    if (auto *MD = dyn_cast<CXXMethodDecl>(FD)) {
+      if (ParamAttrs[Offset].Kind == Vector)
+        CDT = C.getPointerType(C.getRecordType(MD->getParent()));
+      ++Offset;
+    }
+    if (CDT.isNull()) {
+      for (unsigned I = 0, E = FD->getNumParams(); I < E; ++I) {
+        if (ParamAttrs[I + Offset].Kind == Vector) {
+          CDT = FD->getParamDecl(I)->getType();
+          break;
+        }
+      }
+    }
+  }
+  if (CDT.isNull())
+    CDT = C.IntTy;
+  CDT = CDT->getCanonicalTypeUnqualified();
+  if (CDT->isRecordType() || CDT->isUnionType())
+    CDT = C.IntTy;
+  return C.getTypeSize(CDT);
+}
+
+static void
+emitX86DeclareSimdFunction(const FunctionDecl *FD, llvm::Function *Fn,
+                           llvm::APSInt VLENVal,
+                           ArrayRef<ParamAttrTy> ParamAttrs,
+                           OMPDeclareSimdDeclAttr::BranchStateTy State) {
+  struct ISADataTy {
+    char ISA;
+    unsigned VecRegSize;
+  };
+  ISADataTy ISAData[] = {
+      {
+          'b', 128
+      }, // SSE
+      {
+          'c', 256
+      }, // AVX
+      {
+          'd', 256
+      }, // AVX2
+      {
+          'e', 512
+      }, // AVX512
+  };
+  llvm::SmallVector<char, 2> Masked;
+  switch (State) {
+  case OMPDeclareSimdDeclAttr::BS_Undefined:
+    Masked.push_back('N');
+    Masked.push_back('M');
+    break;
+  case OMPDeclareSimdDeclAttr::BS_Notinbranch:
+    Masked.push_back('N');
+    break;
+  case OMPDeclareSimdDeclAttr::BS_Inbranch:
+    Masked.push_back('M');
+    break;
+  }
+  for (auto Mask : Masked) {
+    for (auto &Data : ISAData) {
+      SmallString<256> Buffer;
+      llvm::raw_svector_ostream Out(Buffer);
+      Out << "_ZGV" << Data.ISA << Mask;
+      if (!VLENVal) {
+        Out << llvm::APSInt::getUnsigned(Data.VecRegSize /
+                                         evaluateCDTSize(FD, ParamAttrs));
+      } else
+        Out << VLENVal;
+      for (auto &ParamAttr : ParamAttrs) {
+        switch (ParamAttr.Kind){
+        case LinearWithVarStride:
+          Out << 's' << ParamAttr.StrideOrArg;
+          break;
+        case Linear:
+          Out << 'l';
+          if (!!ParamAttr.StrideOrArg)
+            Out << ParamAttr.StrideOrArg;
+          break;
+        case Uniform:
+          Out << 'u';
+          break;
+        case Vector:
+          Out << 'v';
+          break;
+        }
+        if (!!ParamAttr.Alignment)
+          Out << 'a' << ParamAttr.Alignment;
+      }
+      Out << '_' << Fn->getName();
+      Fn->addFnAttr(Out.str());
+    }
+  }
+}
+
+void CGOpenMPRuntime::emitDeclareSimdFunction(const FunctionDecl *FD,
+                                              llvm::Function *Fn) {
+  ASTContext &C = CGM.getContext();
+  FD = FD->getCanonicalDecl();
+  // Map params to their positions in function decl.
+  llvm::DenseMap<const Decl *, unsigned> ParamPositions;
+  if (isa<CXXMethodDecl>(FD))
+    ParamPositions.insert({FD, 0});
+  unsigned ParamPos = ParamPositions.size();
+  for (auto *P : FD->parameters()) {
+    ParamPositions.insert({P->getCanonicalDecl(), ParamPos});
+    ++ParamPos;
+  }
+  for (auto *Attr : FD->specific_attrs<OMPDeclareSimdDeclAttr>()) {
+    llvm::SmallVector<ParamAttrTy, 8> ParamAttrs(ParamPositions.size());
+    // Mark uniform parameters.
+    for (auto *E : Attr->uniforms()) {
+      E = E->IgnoreParenImpCasts();
+      unsigned Pos;
+      if (isa<CXXThisExpr>(E))
+        Pos = ParamPositions[FD];
+      else {
+        auto *PVD = cast<ParmVarDecl>(cast<DeclRefExpr>(E)->getDecl())
+                        ->getCanonicalDecl();
+        Pos = ParamPositions[PVD];
+      }
+      ParamAttrs[Pos].Kind = Uniform;
+    }
+    // Get alignment info.
+    auto NI = Attr->alignments_begin();
+    for (auto *E : Attr->aligneds()) {
+      E = E->IgnoreParenImpCasts();
+      unsigned Pos;
+      QualType ParmTy;
+      if (isa<CXXThisExpr>(E)) {
+        Pos = ParamPositions[FD];
+        ParmTy = E->getType();
+      } else {
+        auto *PVD = cast<ParmVarDecl>(cast<DeclRefExpr>(E)->getDecl())
+                        ->getCanonicalDecl();
+        Pos = ParamPositions[PVD];
+        ParmTy = PVD->getType();
+      }
+      ParamAttrs[Pos].Alignment =
+          (*NI) ? (*NI)->EvaluateKnownConstInt(C)
+                : llvm::APSInt::getUnsigned(
+                      C.toCharUnitsFromBits(C.getOpenMPDefaultSimdAlign(ParmTy))
+                          .getQuantity());
+      ++NI;
+    }
+    // Mark linear parameters.
+    auto SI = Attr->steps_begin();
+    auto MI = Attr->modifiers_begin();
+    for (auto *E : Attr->linears()) {
+      E = E->IgnoreParenImpCasts();
+      unsigned Pos;
+      if (isa<CXXThisExpr>(E))
+        Pos = ParamPositions[FD];
+      else {
+        auto *PVD = cast<ParmVarDecl>(cast<DeclRefExpr>(E)->getDecl())
+                        ->getCanonicalDecl();
+        Pos = ParamPositions[PVD];
+      }
+      auto &ParamAttr = ParamAttrs[Pos];
+      ParamAttr.Kind = Linear;
+      if (*SI) {
+        if (!(*SI)->EvaluateAsInt(ParamAttr.StrideOrArg, C,
+                                  Expr::SE_AllowSideEffects)) {
+          if (auto *DRE = cast<DeclRefExpr>((*SI)->IgnoreParenImpCasts())) {
+            if (auto *StridePVD = cast<ParmVarDecl>(DRE->getDecl())) {
+              ParamAttr.Kind = LinearWithVarStride;
+              ParamAttr.StrideOrArg = llvm::APSInt::getUnsigned(
+                  ParamPositions[StridePVD->getCanonicalDecl()]);
+            }
+          }
+        }
+      }
+      ++SI;
+      ++MI;
+    }
+    llvm::APSInt VLENVal;
+    if (const Expr *VLEN = Attr->getSimdlen())
+      VLENVal = VLEN->EvaluateKnownConstInt(C);
+    OMPDeclareSimdDeclAttr::BranchStateTy State = Attr->getBranchState();
+    if (CGM.getTriple().getArch() == llvm::Triple::x86 ||
+        CGM.getTriple().getArch() == llvm::Triple::x86_64)
+      emitX86DeclareSimdFunction(FD, Fn, VLENVal, ParamAttrs, State);
+  }
+}
+
+namespace {
+/// Cleanup action for doacross support.
+class DoacrossCleanupTy final : public EHScopeStack::Cleanup {
+public:
+  static const int DoacrossFinArgs = 2;
+
+private:
+  llvm::Value *RTLFn;
+  llvm::Value *Args[DoacrossFinArgs];
+
+public:
+  DoacrossCleanupTy(llvm::Value *RTLFn, ArrayRef<llvm::Value *> CallArgs)
+      : RTLFn(RTLFn) {
+    assert(CallArgs.size() == DoacrossFinArgs);
+    std::copy(CallArgs.begin(), CallArgs.end(), std::begin(Args));
+  }
+  void Emit(CodeGenFunction &CGF, Flags /*flags*/) override {
+    if (!CGF.HaveInsertPoint())
+      return;
+    CGF.EmitRuntimeCall(RTLFn, Args);
+  }
+};
+} // namespace
+
+void CGOpenMPRuntime::emitDoacrossInit(CodeGenFunction &CGF,
+                                       const OMPLoopDirective &D) {
+  if (!CGF.HaveInsertPoint())
+    return;
+
+  ASTContext &C = CGM.getContext();
+  QualType Int64Ty = C.getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/true);
+  RecordDecl *RD;
+  if (KmpDimTy.isNull()) {
+    // Build struct kmp_dim {  // loop bounds info casted to kmp_int64
+    //  kmp_int64 lo; // lower
+    //  kmp_int64 up; // upper
+    //  kmp_int64 st; // stride
+    // };
+    RD = C.buildImplicitRecord("kmp_dim");
+    RD->startDefinition();
+    addFieldToRecordDecl(C, RD, Int64Ty);
+    addFieldToRecordDecl(C, RD, Int64Ty);
+    addFieldToRecordDecl(C, RD, Int64Ty);
+    RD->completeDefinition();
+    KmpDimTy = C.getRecordType(RD);
+  } else
+    RD = cast<RecordDecl>(KmpDimTy->getAsTagDecl());
+
+  Address DimsAddr = CGF.CreateMemTemp(KmpDimTy, "dims");
+  CGF.EmitNullInitialization(DimsAddr, KmpDimTy);
+  enum { LowerFD = 0, UpperFD, StrideFD };
+  // Fill dims with data.
+  LValue DimsLVal = CGF.MakeAddrLValue(DimsAddr, KmpDimTy);
+  // dims.upper = num_iterations;
+  LValue UpperLVal =
+      CGF.EmitLValueForField(DimsLVal, *std::next(RD->field_begin(), UpperFD));
+  llvm::Value *NumIterVal = CGF.EmitScalarConversion(
+      CGF.EmitScalarExpr(D.getNumIterations()), D.getNumIterations()->getType(),
+      Int64Ty, D.getNumIterations()->getExprLoc());
+  CGF.EmitStoreOfScalar(NumIterVal, UpperLVal);
+  // dims.stride = 1;
+  LValue StrideLVal =
+      CGF.EmitLValueForField(DimsLVal, *std::next(RD->field_begin(), StrideFD));
+  CGF.EmitStoreOfScalar(llvm::ConstantInt::getSigned(CGM.Int64Ty, /*V=*/1),
+                        StrideLVal);
+
+  // Build call void __kmpc_doacross_init(ident_t *loc, kmp_int32 gtid,
+  // kmp_int32 num_dims, struct kmp_dim * dims);
+  llvm::Value *Args[] = {emitUpdateLocation(CGF, D.getLocStart()),
+                         getThreadID(CGF, D.getLocStart()),
+                         llvm::ConstantInt::getSigned(CGM.Int32Ty, 1),
+                         CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(
+                             DimsAddr.getPointer(), CGM.VoidPtrTy)};
+
+  llvm::Value *RTLFn = createRuntimeFunction(OMPRTL__kmpc_doacross_init);
+  CGF.EmitRuntimeCall(RTLFn, Args);
+  llvm::Value *FiniArgs[DoacrossCleanupTy::DoacrossFinArgs] = {
+      emitUpdateLocation(CGF, D.getLocEnd()), getThreadID(CGF, D.getLocEnd())};
+  llvm::Value *FiniRTLFn = createRuntimeFunction(OMPRTL__kmpc_doacross_fini);
+  CGF.EHStack.pushCleanup<DoacrossCleanupTy>(NormalAndEHCleanup, FiniRTLFn,
+                                             llvm::makeArrayRef(FiniArgs));
+}
+
+void CGOpenMPRuntime::emitDoacrossOrdered(CodeGenFunction &CGF,
+                                          const OMPDependClause *C) {
+  QualType Int64Ty =
+      CGM.getContext().getIntTypeForBitwidth(/*DestWidth=*/64, /*Signed=*/1);
+  const Expr *CounterVal = C->getCounterValue();
+  assert(CounterVal);
+  llvm::Value *CntVal = CGF.EmitScalarConversion(CGF.EmitScalarExpr(CounterVal),
+                                                 CounterVal->getType(), Int64Ty,
+                                                 CounterVal->getExprLoc());
+  Address CntAddr = CGF.CreateMemTemp(Int64Ty, ".cnt.addr");
+  CGF.EmitStoreOfScalar(CntVal, CntAddr, /*Volatile=*/false, Int64Ty);
+  llvm::Value *Args[] = {emitUpdateLocation(CGF, C->getLocStart()),
+                         getThreadID(CGF, C->getLocStart()),
+                         CntAddr.getPointer()};
+  llvm::Value *RTLFn;
+  if (C->getDependencyKind() == OMPC_DEPEND_source)
+    RTLFn = createRuntimeFunction(OMPRTL__kmpc_doacross_post);
+  else {
+    assert(C->getDependencyKind() == OMPC_DEPEND_sink);
+    RTLFn = createRuntimeFunction(OMPRTL__kmpc_doacross_wait);
+  }
+  CGF.EmitRuntimeCall(RTLFn, Args);
+}
+