[OPENMP] Codegen for 'taskwait' directive.

[ABataev]

Summary

Emit the following code for 'taskwait' directive within tied task:

call i32 @__kmpc_omp_taskwait(<loc>, i32 <thread_id>);

[rjmccall]

LGTM.

cfe/trunk/lib/CodeGen/CGOpenMPRuntime.h

enum OpenMPRTLFunction {
     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,
   };
 
   /// \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,
public:
   /// or 'operator binop(LHS, RHS)'.
   /// \param WithNowait true if parent directive has also nowait clause, false
   /// otherwise.
   virtual void emitReduction(CodeGenFunction &CGF, SourceLocation Loc,
                              ArrayRef<const Expr *> LHSExprs,
                              ArrayRef<const Expr *> RHSExprs,
                              ArrayRef<const Expr *> ReductionOps,
                              bool WithNowait);
+
+  /// \brief Emit code for 'taskwait' directive.
+  virtual void emitTaskwaitCall(CodeGenFunction &CGF, SourceLocation Loc);
 };
 
 } // namespace CodeGen
 } // namespace clang
 
 #endif

cfe/trunk/lib/CodeGen/CGOpenMPRuntime.cpp

CGOpenMPRuntime::createRuntimeFunction(OpenMPRTLFunction Function) {
   case OMPRTL__kmpc_end_ordered: {
     // Build void __kmpc_ordered(ident_t *loc, kmp_int32 global_tid);
     llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty};
     llvm::FunctionType *FnTy =
         llvm::FunctionType::get(CGM.VoidTy, TypeParams, /*isVarArg=*/false);
     RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_end_ordered");
     break;
   }
+  case OMPRTL__kmpc_omp_taskwait: {
+    // Build kmp_int32 __kmpc_omp_taskwait(ident_t *loc, kmp_int32 global_tid);
+    llvm::Type *TypeParams[] = {getIdentTyPointerTy(), CGM.Int32Ty};
+    llvm::FunctionType *FnTy =
+        llvm::FunctionType::get(CGM.Int32Ty, TypeParams, /*isVarArg=*/false);
+    RTLFn = CGM.CreateRuntimeFunction(FnTy, "__kmpc_omp_taskwait");
+    break;
+  }
   }
   return RTLFn;
 }
 
 llvm::Constant *CGOpenMPRuntime::createForStaticInitFunction(unsigned IVSize,
                                                              bool IVSigned) {
   assert((IVSize == 32 || IVSize == 64) &&
          "IV size is not compatible with the omp runtime");
for (auto *E : ReductionOps) {
       ++I;
     }
   }
 
   CGF.EmitBranch(DefaultBB);
   CGF.EmitBlock(DefaultBB, /*IsFinished=*/true);
 }
 
+void CGOpenMPRuntime::emitTaskwaitCall(CodeGenFunction &CGF,
+                                       SourceLocation Loc) {
+  // Build call kmp_int32 __kmpc_omp_taskwait(ident_t *loc, kmp_int32
+  // global_tid);
+  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);
+}
+
 void CGOpenMPRuntime::emitInlinedDirective(CodeGenFunction &CGF,
                                            const RegionCodeGenTy &CodeGen) {
   InlinedOpenMPRegionRAII Region(CGF, CodeGen);
   CGF.CapturedStmtInfo->EmitBody(CGF, /*S=*/nullptr);
 }
 

cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp

void CodeGenFunction::EmitOMPTaskyieldDirective(
     const OMPTaskyieldDirective &S) {
   CGM.getOpenMPRuntime().emitTaskyieldCall(*this, S.getLocStart());
 }
 
 void CodeGenFunction::EmitOMPBarrierDirective(const OMPBarrierDirective &S) {
   CGM.getOpenMPRuntime().emitBarrierCall(*this, S.getLocStart(), OMPD_barrier);
 }
 
-void CodeGenFunction::EmitOMPTaskwaitDirective(const OMPTaskwaitDirective &) {
-  llvm_unreachable("CodeGen for 'omp taskwait' is not supported yet.");
+void CodeGenFunction::EmitOMPTaskwaitDirective(const OMPTaskwaitDirective &S) {
+  CGM.getOpenMPRuntime().emitTaskwaitCall(*this, S.getLocStart());
 }
 
 void CodeGenFunction::EmitOMPFlushDirective(const OMPFlushDirective &S) {
   CGM.getOpenMPRuntime().emitFlush(*this, [&]() -> ArrayRef<const Expr *> {
     if (auto C = S.getSingleClause(/*K*/ OMPC_flush)) {
       auto FlushClause = cast<OMPFlushClause>(C);
       return llvm::makeArrayRef(FlushClause->varlist_begin(),
                                 FlushClause->varlist_end());

cfe/trunk/test/OpenMP/taskwait_codegen.cpp

+// RUN: %clang_cc1 -verify -triple x86_64-apple-darwin10 -fopenmp=libiomp5 -x c++ -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -fopenmp=libiomp5 -x c++ -triple x86_64-apple-darwin10 -emit-pch -o %t %s
+// RUN: %clang_cc1 -fopenmp=libiomp5 -x c++ -triple x86_64-apple-darwin10 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s
+// expected-no-diagnostics
+
+#ifndef HEADER
+#define HEADER
+
+void foo() {}
+
+template <class T>
+T tmain(T argc) {
+  static T a;
+#pragma omp taskwait
+  return a + argc;
+}
+int main(int argc, char **argv) {
+#pragma omp taskwait
+  return tmain(argc);
+}
+
+// CHECK-LABEL: @main
+// CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num(%{{.+}}* @{{.+}})
+// CHECK: call i32 @__kmpc_omp_taskwait(%{{.+}}* @{{.+}}, i32 [[GTID]])
+
+// CHECK-LABEL: tmain
+// CHECK: [[GTID:%.+]] = call i32 @__kmpc_global_thread_num(%{{.+}}* @{{.+}})
+// CHECK: call i32 @__kmpc_omp_taskwait(%{{.+}}* @{{.+}}, i32 [[GTID]])
+
+
+#endif