[OpenMP] Codegen for target update directive.

[sfantao]

Summary

This patch implements the code generation for the target update directive. The implemntation relies on the logic already in place for target data standalone directives, i.e. target enter/exit data.

[ABataev]

LG

[sfantao]

• Adapt target update tests bits to reflect latest changes in the trunk.

lib/CodeGen/CGOpenMPRuntime.h

public:
   /// \param Device Expression evaluated in device clause associated with the
   /// target directive, or null if no device clause is used.
   /// \param CodeGen Function that emits the enclosed region.
   virtual void emitTargetDataCalls(CodeGenFunction &CGF,
                                    const OMPExecutableDirective &D,
                                    const Expr *IfCond, const Expr *Device,
                                    const RegionCodeGenTy &CodeGen);
 
-  /// \brief Emit the target enter or exit data mapping code associated with
-  /// directive \a D.
+  /// \brief Emit the data mapping/movement code associated with the directive
+  /// \a D that should be of the form 'target [{enter|exit} data | update]'.
   /// \param D Directive to emit.
   /// \param IfCond Expression evaluated in if clause associated with the target
   /// directive, or null if no if clause is used.
   /// \param Device Expression evaluated in device clause associated with the
   /// target directive, or null if no device clause is used.
-  virtual void emitTargetEnterOrExitDataCall(CodeGenFunction &CGF,
-                                             const OMPExecutableDirective &D,
-                                             const Expr *IfCond,
-                                             const Expr *Device);
+  virtual void emitTargetDataStandAloneCall(CodeGenFunction &CGF,
 
   /// Marks function \a Fn with properly mangled versions of vector functions.
   /// \param FD Function marked as 'declare simd'.
   /// \param Fn LLVM function that must be marked with 'declare simd'
   /// attributes.
   virtual void emitDeclareSimdFunction(const FunctionDecl *FD,
                                        llvm::Function *Fn);
 };
 
 } // namespace CodeGen
 } // namespace clang
 
 #endif

lib/CodeGen/CGOpenMPRuntime.cpp

enum OpenMPRTLFunction {
   // 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:
llvm::Type *TypeParams[] = {CGM.Int32Ty,
                                 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;
 }
 
 llvm::Constant *CGOpenMPRuntime::createForStaticInitFunction(unsigned IVSize,
                                                              bool IVSigned) {
   assert((IVSize == 32 || IVSize == 64) &&
struct MapInfo {
       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;
-    for (auto *C : Directive.getClausesOfKind<OMPMapClause>())
-      for (auto L : C->component_lists()) {
+
+    // 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 =
-            L.first ? cast<ValueDecl>(L.first->getCanonicalDecl()) : nullptr;
-        Info[VD].push_back(
-            {L.second, C->getMapType(), C->getMapTypeModifier()});
-      }
+              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.");
void CGOpenMPRuntime::emitTargetDataCalls(CodeGenFunction &CGF,
   if (IfCond) {
     emitOMPIfClause(CGF, IfCond, EndThenGen, ElseGen);
   } else {
     RegionCodeGenTy EndThenRCG(EndThenGen);
     EndThenRCG(CGF);
   }
 }
 
-void CGOpenMPRuntime::emitTargetEnterOrExitDataCall(
+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)) &&
-         "Expecting either target enter or exit data directives.");
+          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 MCHandler(D, CGF);
-    MCHandler.generateAllInfo(BasePointers, Pointers, Sizes, MapTypes);
+    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,
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();
-    CGF.EmitRuntimeCall(
-        RT.createRuntimeFunction(isa<OMPTargetEnterDataDirective>(D)
-                                     ? OMPRTL__tgt_target_data_begin
-                                     : OMPRTL__tgt_target_data_end),
-        OffloadingArgs);
+    // 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);

lib/CodeGen/CGStmtOpenMP.cpp

void CodeGenFunction::EmitOMPTargetEnterDataDirective(
   if (auto *C = S.getSingleClause<OMPIfClause>())
     IfCond = C->getCondition();
 
   // Check if we have any device clause associated with the directive.
   const Expr *Device = nullptr;
   if (auto *C = S.getSingleClause<OMPDeviceClause>())
     Device = C->getDevice();
 
-  CGM.getOpenMPRuntime().emitTargetEnterOrExitDataCall(*this, S, IfCond,
-                                                       Device);
+  CGM.getOpenMPRuntime().emitTargetDataStandAloneCall(*this, S, IfCond, Device);
 }
 
 void CodeGenFunction::EmitOMPTargetExitDataDirective(
     const OMPTargetExitDataDirective &S) {
   // If we don't have target devices, don't bother emitting the data mapping
   // code.
   if (CGM.getLangOpts().OMPTargetTriples.empty())
     return;
 
   // Check if we have any if clause associated with the directive.
   const Expr *IfCond = nullptr;
   if (auto *C = S.getSingleClause<OMPIfClause>())
     IfCond = C->getCondition();
 
   // Check if we have any device clause associated with the directive.
   const Expr *Device = nullptr;
   if (auto *C = S.getSingleClause<OMPDeviceClause>())
     Device = C->getDevice();
 
-  CGM.getOpenMPRuntime().emitTargetEnterOrExitDataCall(*this, S, IfCond,
-                                                       Device);
+  CGM.getOpenMPRuntime().emitTargetDataStandAloneCall(*this, S, IfCond, Device);
 }
 
 void CodeGenFunction::EmitOMPTargetParallelDirective(
     const OMPTargetParallelDirective &S) {
   // TODO: codegen for target parallel.
 }
 
 void CodeGenFunction::EmitOMPTargetParallelForDirective(

 void CodeGenFunction::EmitOMPTaskLoopSimdDirective(
     const OMPTaskLoopSimdDirective &S) {
   EmitOMPTaskLoopBasedDirective(S);
 }
 
 // Generate the instructions for '#pragma omp target update' directive.
 void CodeGenFunction::EmitOMPTargetUpdateDirective(
     const OMPTargetUpdateDirective &S) {
-  // TODO: codegen for target update
+  // If we don't have target devices, don't bother emitting the data mapping
+  // code.
+  if (CGM.getLangOpts().OMPTargetTriples.empty())
+    return;
+
+  // Check if we have any if clause associated with the directive.
+  const Expr *IfCond = nullptr;
+  if (auto *C = S.getSingleClause<OMPIfClause>())
+    IfCond = C->getCondition();
+
+  // Check if we have any device clause associated with the directive.
+  const Expr *Device = nullptr;
+  if (auto *C = S.getSingleClause<OMPDeviceClause>())
+    Device = C->getDevice();
+
+  CGM.getOpenMPRuntime().emitTargetDataStandAloneCall(*this, S, IfCond, Device);
 }

test/OpenMP/target_update_codegen.cpp

+// expected-no-diagnostics
+#ifndef HEADER
+#define HEADER
+
+///==========================================================================///
+// RUN: %clang_cc1 -DCK1 -verify -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK1 --check-prefix CK1-64
+// RUN: %clang_cc1 -DCK1 -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK1 --check-prefix CK1-64
+// RUN: %clang_cc1 -DCK1 -verify -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK1 --check-prefix CK1-32
+// RUN: %clang_cc1 -DCK1 -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK1 --check-prefix CK1-32
+#ifdef CK1
+
+// CK1: [[ST:%.+]] = type { i32, double* }
+template <typename T>
+struct ST {
+  T a;
+  double *b;
+};
+
+ST<int> gb;
+double gc[100];
+
+// CK1: [[SIZE00:@.+]] = {{.+}}constant [1 x i[[sz:64|32]]] [i{{64|32}} 800]
+// CK1: [[MTYPE00:@.+]] = {{.+}}constant [1 x i32] [i32 2]
+
+// CK1: [[SIZE02:@.+]] = {{.+}}constant [1 x i[[sz]]] [i[[sz]] 4]
+// CK1: [[MTYPE02:@.+]] = {{.+}}constant [1 x i32] [i32 1]
+
+// CK1: [[MTYPE03:@.+]] = {{.+}}constant [1 x i32] [i32 2]
+
+// CK1: [[SIZE04:@.+]] = {{.+}}constant [2 x i[[sz]]] [i[[sz]] {{8|4}}, i[[sz]] 24]
+// CK1: [[MTYPE04:@.+]] = {{.+}}constant [2 x i32] [i32 1, i32 97]
+
+// CK1-LABEL: _Z3fooi
+void foo(int arg) {
+  int la;
+  float lb[arg];
+
+  // Region 00
+  // CK1-DAG: call void @__tgt_target_data_update(i32 [[DEV:%[^,]+]], i32 1, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[SIZE00]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE00]]{{.+}})
+  // CK1-DAG: [[DEV]] = load i32, i32* %{{[^,]+}},
+  // CK1-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
+  // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
+
+  // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: store i8* bitcast ([100 x double]* @gc to i8*), i8** [[BP0]]
+  // CK1-DAG: store i8* bitcast ([100 x double]* @gc to i8*), i8** [[P0]]
+
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  #pragma omp target update if(1+3-5) device(arg) from(gc)
+  {++arg;}
+
+  // Region 01
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  #pragma omp target update to(la) if(1+3-4)
+  {++arg;}
+
+  // Region 02
+  // CK1: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
+  // CK1: [[IFTHEN]]
+  // CK1-DAG: call void @__tgt_target_data_update(i32 4, i32 1, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[SIZE02]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE02]]{{.+}})
+  // CK1-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
+  // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
+
+  // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: store i8* [[CBPVAL0:%[^,]+]], i8** [[BP0]]
+  // CK1-DAG: store i8* [[CPVAL0:%[^,]+]], i8** [[P0]]
+  // CK1-DAG: [[CBPVAL0]] = bitcast i32* [[VAR0:%.+]] to i8*
+  // CK1-DAG: [[CPVAL0]] = bitcast i32* [[VAR0]] to i8*
+  // CK1: br label %[[IFEND:[^,]+]]
+
+  // CK1: [[IFELSE]]
+  // CK1: br label %[[IFEND]]
+  // CK1: [[IFEND]]
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  #pragma omp target update to(arg) if(arg) device(4)
+  {++arg;}
+
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  {++arg;}
+
+  // Region 03
+  // CK1-DAG: call void @__tgt_target_data_update(i32 -1, i32 1, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], i[[sz]]* [[GEPS:%.+]], {{.+}}getelementptr {{.+}}[1 x i{{.+}}]* [[MTYPE03]]{{.+}})
+  // CK1-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
+  // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
+  // CK1-DAG: [[GEPS]] = getelementptr inbounds {{.+}}[[S:%[^,]+]]
+
+  // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: [[S0:%.+]] = getelementptr inbounds {{.+}}[[S]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: store i8* [[CBPVAL0:%[^,]+]], i8** [[BP0]]
+  // CK1-DAG: store i8* [[CPVAL0:%[^,]+]], i8** [[P0]]
+  // CK1-DAG: store i[[sz]] [[CSVAL0:%[^,]+]], i[[sz]]* [[S0]]
+  // CK1-DAG: [[CBPVAL0]] = bitcast float* [[VAR0:%.+]] to i8*
+  // CK1-DAG: [[CPVAL0]] = bitcast float* [[VAR0]] to i8*
+  // CK1-DAG: [[CSVAL0]] = mul nuw i[[sz]] %{{[^,]+}}, 4
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  // CK1-NOT: __tgt_target_data_end
+  #pragma omp target update from(lb)
+  {++arg;}
+
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  {++arg;}
+
+  // Region 04
+  // CK1-DAG: call void @__tgt_target_data_update(i32 -1, i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE04]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE04]]{{.+}})
+  // CK1-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
+  // CK1-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
+
+  // CK1-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
+  // CK1-DAG: store i8* bitcast ([[ST]]* @gb to i8*), i8** [[BP0]]
+  // CK1-DAG: store i8* bitcast (double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1) to i8*), i8** [[P0]]
+
+
+  // CK1-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
+  // CK1-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
+  // CK1-DAG: store i8* bitcast (double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1) to i8*), i8** [[BP1]]
+  // CK1-DAG: store i8* [[CPVAL1:%[^,]+]], i8** [[P1]]
+  // CK1-DAG: [[CPVAL1]] = bitcast double* [[SEC1:%.+]] to i8*
+  // CK1-DAG: [[SEC1]] = getelementptr inbounds {{.+}}double* [[SEC11:%[^,]+]], i{{.+}} 0
+  // CK1-DAG: [[SEC11]] = load double*, double** getelementptr inbounds ([[ST]], [[ST]]* @gb, i32 0, i32 1),
+
+  // CK1: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  // CK1-NOT: __tgt_target_data_end
+  #pragma omp target update to(gb.b[:3])
+  {++arg;}
+}
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK2 -verify -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK2 --check-prefix CK2-64
+// RUN: %clang_cc1 -DCK2 -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK2 --check-prefix CK2-64
+// RUN: %clang_cc1 -DCK2 -verify -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK2 --check-prefix CK2-32
+// RUN: %clang_cc1 -DCK2 -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK2 --check-prefix CK2-32
+#ifdef CK2
+
+// CK2: [[ST:%.+]] = type { i32, double* }
+template <typename T>
+struct ST {
+  T a;
+  double *b;
+
+  T foo(T arg) {
+    // Region 00
+    #pragma omp target update from(b[1:3]) if(a>123) device(arg)
+    {arg++;}
+    return arg;
+  }
+};
+
+// CK2: [[SIZE00:@.+]] = {{.+}}constant [2 x i[[sz:64|32]]] [i{{64|32}} {{8|4}}, i{{64|32}} 24]
+// CK2: [[MTYPE00:@.+]] = {{.+}}constant [2 x i32] [i32 2, i32 98]
+
+// CK2-LABEL: _Z3bari
+int bar(int arg){
+  ST<int> A;
+  return A.foo(arg);
+}
+
+// Region 00
+// CK2: br i1 %{{[^,]+}}, label %[[IFTHEN:[^,]+]], label %[[IFELSE:[^,]+]]
+// CK2: [[IFTHEN]]
+// CK2-DAG: call void @__tgt_target_data_update(i32 [[DEV:%[^,]+]], i32 2, i8** [[GEPBP:%.+]], i8** [[GEPP:%.+]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[SIZE00]], {{.+}}getelementptr {{.+}}[2 x i{{.+}}]* [[MTYPE00]]{{.+}})
+// CK2-DAG: [[DEV]] = load i32, i32* %{{[^,]+}},
+// CK2-DAG: [[GEPBP]] = getelementptr inbounds {{.+}}[[BP:%[^,]+]]
+// CK2-DAG: [[GEPP]] = getelementptr inbounds {{.+}}[[P:%[^,]+]]
+
+// CK2-DAG: [[BP0:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 0
+// CK2-DAG: [[P0:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 0
+// CK2-DAG: store i8* [[CBPVAL0:%[^,]+]], i8** [[BP0]]
+// CK2-DAG: store i8* [[CPVAL0:%[^,]+]], i8** [[P0]]
+// CK2-DAG: [[CBPVAL0]] = bitcast [[ST]]* [[VAR0:%.+]] to i8*
+// CK2-DAG: [[CPVAL0]] = bitcast double** [[SEC0:%[^,]+]] to i8*
+// CK2-DAG: [[SEC0]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
+
+
+// CK2-DAG: [[BP1:%.+]] = getelementptr inbounds {{.+}}[[BP]], i{{.+}} 0, i{{.+}} 1
+// CK2-DAG: [[P1:%.+]] = getelementptr inbounds {{.+}}[[P]], i{{.+}} 0, i{{.+}} 1
+// CK2-DAG: store i8* [[CBPVAL1:%[^,]+]], i8** [[BP1]]
+// CK2-DAG: store i8* [[CPVAL1:%[^,]+]], i8** [[P1]]
+// CK2-DAG: [[CBPVAL1]] = bitcast double** [[SEC0]] to i8*
+// CK2-DAG: [[CPVAL1]] = bitcast double* [[SEC1:%[^,]+]] to i8*
+// CK2-DAG: [[SEC1]] = getelementptr inbounds {{.*}}double* [[SEC11:%[^,]+]], i{{.+}} 1
+// CK2-DAG: [[SEC11]] = load double*, double** [[SEC111:%[^,]+]],
+// CK2-DAG: [[SEC111]] = getelementptr inbounds {{.*}}[[ST]]* [[VAR0]], i32 0, i32 1
+
+// CK2: br label %[[IFEND:[^,]+]]
+
+// CK2: [[IFELSE]]
+// CK2: br label %[[IFEND]]
+// CK2: [[IFEND]]
+// CK2: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK3 --check-prefix CK3-64
+// RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK3 --check-prefix CK3-64
+// RUN: %clang_cc1 -DCK3 -verify -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK3 --check-prefix CK3-32
+// RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK3 --check-prefix CK3-32
+#ifdef CK3
+
+// CK3-LABEL: no_target_devices
+void no_target_devices(int arg) {
+  // CK3-NOT: tgt_target_data_update
+  // CK3: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  // CK3: ret
+  #pragma omp target update to(arg) if(arg) device(4)
+  {++arg;}
+}
+#endif
+///==========================================================================///
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -emit-llvm %s -o - | FileCheck %s --check-prefix CK4 --check-prefix CK4-64
+// RUN: %clang_cc1 -DCK4 -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fomptargets=powerpc64le-ibm-linux-gnu -x c++ -triple powerpc64le-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK4 --check-prefix CK4-64
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -emit-llvm %s -o - | FileCheck %s  --check-prefix CK4 --check-prefix CK4-32
+// RUN: %clang_cc1 -DCK4 -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -std=c++11 -triple i386-unknown-unknown -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp -fomptargets=i386-pc-linux-gnu -x c++ -triple i386-unknown-unknown -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s  --check-prefix CK4 --check-prefix CK4-32
+
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fomptargets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fomptargets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fomp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix TCK4 --check-prefix TCK4-64
+// RUN: %clang_cc1 -DCK4 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fomptargets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fomp-host-ir-file-path %t-ppc-host.bc -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fomptargets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fomp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix TCK4 --check-prefix TCK4-64
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fomptargets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
+// RUN: %clang_cc1 -DCK4 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fomptargets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fomp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix TCK4 --check-prefix TCK4-32
+// RUN: %clang_cc1 -DCK4 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fomptargets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fomp-host-ir-file-path %t-x86-host.bc -o %t %s
+// RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -fomptargets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fomp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix TCK4 --check-prefix TCK4-32
+#ifdef CK4
+
+// CK4-LABEL: device_side_scan
+void device_side_scan(int arg) {
+  // CK4: tgt_target_data_update
+  // CK4: %{{.+}} = add nsw i32 %{{[^,]+}}, 1
+  // CK4: ret
+  // TCK4-NOT: tgt_target_data_update
+  #pragma omp target update from(arg) if(arg) device(4)
+  {++arg;}
+}
+#endif
+#endif