[OpenMP] Initial implementation of code generation for pragma 'target teams distribute parallel for' on host

https://reviews.llvm.org/D41709

This patch includes code generation and testing for offloading when target device is host.

llvm-svn: 321759

3 files changed, 58 insertions, 11 deletions

diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.cpp b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
index 784c90a9eb9..3c594994ffe 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -7292,6 +7292,11 @@ void CGOpenMPRuntime::scanForTargetRegionsFunctions(const Stmt *S,
       CodeGenFunction::EmitOMPTargetSimdDeviceFunction(
           CGM, ParentName, cast<OMPTargetSimdDirective>(*S));
       break;
+    case Stmt::OMPTargetTeamsDistributeParallelForDirectiveClass:
+      CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForDeviceFunction(
+          CGM, ParentName,
+          cast<OMPTargetTeamsDistributeParallelForDirective>(*S));
+      break;
     default:
       llvm_unreachable("Unknown target directive for OpenMP device codegen.");
     }
diff --git a/clang/lib/CodeGen/CGStmtOpenMP.cpp b/clang/lib/CodeGen/CGStmtOpenMP.cpp
index 26817ac330b..07c1e3aa515 100644
--- a/clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -2167,17 +2167,6 @@ void CodeGenFunction::EmitOMPTargetSimdDirective(
   emitCommonOMPTargetDirective(*this, S, CodeGen);
 }
 
-void CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForDirective(
-    const OMPTargetTeamsDistributeParallelForDirective &S) {
-  OMPLexicalScope Scope(*this, S, /*AsInlined=*/true);
-  CGM.getOpenMPRuntime().emitInlinedDirective(
-      *this, OMPD_target_teams_distribute_parallel_for,
-      [&S](CodeGenFunction &CGF, PrePostActionTy &) {
-        CGF.EmitStmt(
-            cast<CapturedStmt>(S.getAssociatedStmt())->getCapturedStmt());
-      });
-}
-
 void CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForSimdDirective(
     const OMPTargetTeamsDistributeParallelForSimdDirective &S) {
   OMPLexicalScope Scope(*this, S, /*AsInlined=*/true);
@@ -4266,6 +4255,55 @@ void CodeGenFunction::EmitOMPTeamsDistributeParallelForSimdDirective(
                                    [](CodeGenFunction &) { return nullptr; });
 }
 
+static void emitTargetTeamsDistributeParallelForRegion(
+    CodeGenFunction &CGF, const OMPTargetTeamsDistributeParallelForDirective &S,
+    PrePostActionTy &Action) {
+  auto &&CodeGenDistribute = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+    CGF.EmitOMPDistributeLoop(S, emitInnerParallelForWhenCombined,
+                              S.getDistInc());
+  };
+
+  // Emit teams region as a standalone region.
+  auto &&CodeGenTeams = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
+                                                 PrePostActionTy &) {
+    CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
+    CGF.EmitOMPReductionClauseInit(S, PrivateScope);
+    (void)PrivateScope.Privatize();
+    CGF.CGM.getOpenMPRuntime().emitInlinedDirective(
+        CGF, OMPD_distribute, CodeGenDistribute, /*HasCancel=*/false);
+    CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
+  };
+
+  emitCommonOMPTeamsDirective(CGF, S, OMPD_distribute_parallel_for,
+                              CodeGenTeams);
+  emitPostUpdateForReductionClause(CGF, S,
+                                   [](CodeGenFunction &) { return nullptr; });
+}
+
+void CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForDeviceFunction(
+    CodeGenModule &CGM, StringRef ParentName,
+    const OMPTargetTeamsDistributeParallelForDirective &S) {
+  // Emit SPMD target teams distribute parallel for region as a standalone
+  // region.
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    emitTargetTeamsDistributeParallelForRegion(CGF, S, Action);
+  };
+  llvm::Function *Fn;
+  llvm::Constant *Addr;
+  // Emit target region as a standalone region.
+  CGM.getOpenMPRuntime().emitTargetOutlinedFunction(
+      S, ParentName, Fn, Addr, /*IsOffloadEntry=*/true, CodeGen);
+  assert(Fn && Addr && "Target device function emission failed.");
+}
+
+void CodeGenFunction::EmitOMPTargetTeamsDistributeParallelForDirective(
+    const OMPTargetTeamsDistributeParallelForDirective &S) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    emitTargetTeamsDistributeParallelForRegion(CGF, S, Action);
+  };
+  emitCommonOMPTargetDirective(*this, S, CodeGen);
+}
+
 void CodeGenFunction::EmitOMPCancellationPointDirective(
     const OMPCancellationPointDirective &S) {
   CGM.getOpenMPRuntime().emitCancellationPointCall(*this, S.getLocStart(),
diff --git a/clang/lib/CodeGen/CodeGenFunction.h b/clang/lib/CodeGen/CodeGenFunction.h
index 3e89586be3a..74750c4d28a 100644
--- a/clang/lib/CodeGen/CodeGenFunction.h
+++ b/clang/lib/CodeGen/CodeGenFunction.h
@@ -2941,6 +2941,10 @@ public:
   static void EmitOMPTargetSimdDeviceFunction(CodeGenModule &CGM,
                                               StringRef ParentName,
                                               const OMPTargetSimdDirective &S);
+
+  static void EmitOMPTargetTeamsDistributeParallelForDeviceFunction(
+      CodeGenModule &CGM, StringRef ParentName,
+      const OMPTargetTeamsDistributeParallelForDirective &S);
   /// \brief Emit inner loop of the worksharing/simd construct.
   ///
   /// \param S Directive, for which the inner loop must be emitted.