[OPENMP, NVPTX] Improve globalization of the variables captured by value.

If the variable is captured by value and the corresponding parameter in
the outlined function escapes its declaration context, this parameter
must be globalized. To globalize it we need to get the address of the
original parameter, load the value, store it to the global address and
use this global address instead of the original.

Patch improves globalization for parallel|teams regions + functions in
declare target regions.

llvm-svn: 327654

1 files changed, 34 insertions, 17 deletions

diff --git a/clang/lib/CodeGen/CGStmtOpenMP.cpp b/clang/lib/CodeGen/CGStmtOpenMP.cpp
index 9d077a2be16..02bb8ae88a1 100644
--- a/clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -1262,7 +1262,7 @@ static void emitEmptyBoundParameters(CodeGenFunction &,
 
 void CodeGenFunction::EmitOMPParallelDirective(const OMPParallelDirective &S) {
   // Emit parallel region as a standalone region.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
     OMPPrivateScope PrivateScope(CGF);
     bool Copyins = CGF.EmitOMPCopyinClause(S);
     (void)CGF.EmitOMPFirstprivateClause(S, PrivateScope);
@@ -1277,6 +1277,7 @@ void CodeGenFunction::EmitOMPParallelDirective(const OMPParallelDirective &S) {
     CGF.EmitOMPPrivateClause(S, PrivateScope);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.EmitStmt(S.getCapturedStmt(OMPD_parallel)->getCapturedStmt());
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_parallel);
   };
@@ -2100,7 +2101,8 @@ emitInnerParallelForWhenCombined(CodeGenFunction &CGF,
                                  const OMPLoopDirective &S,
                                  CodeGenFunction::JumpDest LoopExit) {
   auto &&CGInlinedWorksharingLoop = [&S](CodeGenFunction &CGF,
-                                         PrePostActionTy &) {
+                                         PrePostActionTy &Action) {
+    Action.Enter(CGF);
     bool HasCancel = false;
     if (!isOpenMPSimdDirective(S.getDirectiveKind())) {
       if (const auto *D = dyn_cast<OMPTeamsDistributeParallelForDirective>(&S))
@@ -2682,7 +2684,8 @@ void CodeGenFunction::EmitOMPParallelForDirective(
     const OMPParallelForDirective &S) {
   // Emit directive as a combined directive that consists of two implicit
   // directives: 'parallel' with 'for' directive.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    Action.Enter(CGF);
     OMPCancelStackRAII CancelRegion(CGF, OMPD_parallel_for, S.hasCancel());
     CGF.EmitOMPWorksharingLoop(S, S.getEnsureUpperBound(), emitForLoopBounds,
                                emitDispatchForLoopBounds);
@@ -2695,7 +2698,8 @@ void CodeGenFunction::EmitOMPParallelForSimdDirective(
     const OMPParallelForSimdDirective &S) {
   // Emit directive as a combined directive that consists of two implicit
   // directives: 'parallel' with 'for' directive.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    Action.Enter(CGF);
     CGF.EmitOMPWorksharingLoop(S, S.getEnsureUpperBound(), emitForLoopBounds,
                                emitDispatchForLoopBounds);
   };
@@ -2707,7 +2711,8 @@ void CodeGenFunction::EmitOMPParallelSectionsDirective(
     const OMPParallelSectionsDirective &S) {
   // Emit directive as a combined directive that consists of two implicit
   // directives: 'parallel' with 'sections' directive.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    Action.Enter(CGF);
     CGF.EmitSections(S);
   };
   emitCommonOMPParallelDirective(*this, S, OMPD_sections, CodeGen,
@@ -4033,12 +4038,13 @@ static void emitCommonOMPTeamsDirective(CodeGenFunction &CGF,
 
 void CodeGenFunction::EmitOMPTeamsDirective(const OMPTeamsDirective &S) {
   // Emit teams region as a standalone region.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
     OMPPrivateScope PrivateScope(CGF);
     (void)CGF.EmitOMPFirstprivateClause(S, PrivateScope);
     CGF.EmitOMPPrivateClause(S, PrivateScope);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.EmitStmt(S.getCapturedStmt(OMPD_teams)->getCapturedStmt());
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
   };
@@ -4099,10 +4105,11 @@ emitTargetTeamsDistributeRegion(CodeGenFunction &CGF, PrePostActionTy &Action,
 
   // Emit teams region as a standalone region.
   auto &&CodeGen = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                            PrePostActionTy &) {
+                                            PrePostActionTy &Action) {
     CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(CGF, OMPD_distribute,
                                                     CodeGenDistribute);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4144,10 +4151,11 @@ static void emitTargetTeamsDistributeSimdRegion(
 
   // Emit teams region as a standalone region.
   auto &&CodeGen = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                            PrePostActionTy &) {
+                                            PrePostActionTy &Action) {
     CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(CGF, OMPD_distribute,
                                                     CodeGenDistribute);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4188,10 +4196,11 @@ void CodeGenFunction::EmitOMPTeamsDistributeDirective(
 
   // Emit teams region as a standalone region.
   auto &&CodeGen = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                            PrePostActionTy &) {
+                                            PrePostActionTy &Action) {
     OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(CGF, OMPD_distribute,
                                                     CodeGenDistribute);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4209,10 +4218,11 @@ void CodeGenFunction::EmitOMPTeamsDistributeSimdDirective(
 
   // Emit teams region as a standalone region.
   auto &&CodeGen = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                            PrePostActionTy &) {
+                                            PrePostActionTy &Action) {
     OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(CGF, OMPD_simd,
                                                     CodeGenDistribute);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4231,10 +4241,11 @@ void CodeGenFunction::EmitOMPTeamsDistributeParallelForDirective(
 
   // Emit teams region as a standalone region.
   auto &&CodeGen = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                            PrePostActionTy &) {
+                                            PrePostActionTy &Action) {
     OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(CGF, OMPD_distribute,
                                                     CodeGenDistribute);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4253,10 +4264,11 @@ void CodeGenFunction::EmitOMPTeamsDistributeParallelForSimdDirective(
 
   // Emit teams region as a standalone region.
   auto &&CodeGen = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                            PrePostActionTy &) {
+                                            PrePostActionTy &Action) {
     OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(
         CGF, OMPD_distribute, CodeGenDistribute, /*HasCancel=*/false);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4277,10 +4289,11 @@ static void emitTargetTeamsDistributeParallelForRegion(
 
   // Emit teams region as a standalone region.
   auto &&CodeGenTeams = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                                 PrePostActionTy &) {
+                                                 PrePostActionTy &Action) {
     CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(
         CGF, OMPD_distribute, CodeGenDistribute, /*HasCancel=*/false);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4328,10 +4341,11 @@ static void emitTargetTeamsDistributeParallelForSimdRegion(
 
   // Emit teams region as a standalone region.
   auto &&CodeGenTeams = [&S, &CodeGenDistribute](CodeGenFunction &CGF,
-                                                 PrePostActionTy &) {
+                                                 PrePostActionTy &Action) {
     CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     CGF.CGM.getOpenMPRuntime().emitInlinedDirective(
         CGF, OMPD_distribute, CodeGenDistribute, /*HasCancel=*/false);
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_teams);
@@ -4599,12 +4613,13 @@ static void emitTargetParallelRegion(CodeGenFunction &CGF,
   // Get the captured statement associated with the 'parallel' region.
   auto *CS = S.getCapturedStmt(OMPD_parallel);
   Action.Enter(CGF);
-  auto &&CodeGen = [&S, CS](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S, CS](CodeGenFunction &CGF, PrePostActionTy &Action) {
     CodeGenFunction::OMPPrivateScope PrivateScope(CGF);
     (void)CGF.EmitOMPFirstprivateClause(S, PrivateScope);
     CGF.EmitOMPPrivateClause(S, PrivateScope);
     CGF.EmitOMPReductionClauseInit(S, PrivateScope);
     (void)PrivateScope.Privatize();
+    Action.Enter(CGF);
     // TODO: Add support for clauses.
     CGF.EmitStmt(CS->getCapturedStmt());
     CGF.EmitOMPReductionClauseFinal(S, /*ReductionKind=*/OMPD_parallel);
@@ -4643,7 +4658,8 @@ static void emitTargetParallelForRegion(CodeGenFunction &CGF,
   Action.Enter(CGF);
   // Emit directive as a combined directive that consists of two implicit
   // directives: 'parallel' with 'for' directive.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    Action.Enter(CGF);
     CodeGenFunction::OMPCancelStackRAII CancelRegion(
         CGF, OMPD_target_parallel_for, S.hasCancel());
     CGF.EmitOMPWorksharingLoop(S, S.getEnsureUpperBound(), emitForLoopBounds,
@@ -4683,7 +4699,8 @@ emitTargetParallelForSimdRegion(CodeGenFunction &CGF,
   Action.Enter(CGF);
   // Emit directive as a combined directive that consists of two implicit
   // directives: 'parallel' with 'for' directive.
-  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &Action) {
+    Action.Enter(CGF);
     CGF.EmitOMPWorksharingLoop(S, S.getEnsureUpperBound(), emitForLoopBounds,
                                emitDispatchForLoopBounds);
   };