[OPENMP] Codegen for `distribute simd` directive.

Initial codegen support for `distribute simd` directive.

llvm-svn: 319661

1 files changed, 54 insertions, 21 deletions

diff --git a/clang/lib/CodeGen/CGStmtOpenMP.cpp b/clang/lib/CodeGen/CGStmtOpenMP.cpp
index bb987d87548..96f5a655956 100644
--- a/clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ b/clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -1093,7 +1093,8 @@ void CodeGenFunction::EmitOMPReductionClauseFinal(
     bool WithNowait = D.getSingleClause<OMPNowaitClause>() ||
                       isOpenMPParallelDirective(D.getDirectiveKind()) ||
                       D.getDirectiveKind() == OMPD_simd;
-    bool SimpleReduction = D.getDirectiveKind() == OMPD_simd;
+    bool SimpleReduction = D.getDirectiveKind() == OMPD_simd ||
+                           D.getDirectiveKind() == OMPD_distribute_simd;
     // Emit nowait reduction if nowait clause is present or directive is a
     // parallel directive (it always has implicit barrier).
     CGM.getOpenMPRuntime().emitReduction(
@@ -1215,9 +1216,13 @@ void CodeGenFunction::EmitOMPLoopBody(const OMPLoopDirective &D,
     EmitIgnoredExpr(I);
   }
   // Update the linear variables.
-  for (const auto *C : D.getClausesOfKind<OMPLinearClause>()) {
-    for (auto *U : C->updates())
-      EmitIgnoredExpr(U);
+  // In distribute directives only loop counters may be marked as linear, no
+  // need to generate the code for them.
+  if (!isOpenMPDistributeDirective(D.getDirectiveKind())) {
+    for (const auto *C : D.getClausesOfKind<OMPLinearClause>()) {
+      for (auto *U : C->updates())
+        EmitIgnoredExpr(U);
+    }
   }
 
   // On a continue in the body, jump to the end.
@@ -2054,14 +2059,11 @@ void CodeGenFunction::EmitOMPDistributeParallelForSimdDirective(
 
 void CodeGenFunction::EmitOMPDistributeSimdDirective(
     const OMPDistributeSimdDirective &S) {
+  auto &&CodeGen = [&S](CodeGenFunction &CGF, PrePostActionTy &) {
+    CGF.EmitOMPDistributeLoop(S, emitOMPLoopBodyWithStopPoint, S.getInc());
+  };
   OMPLexicalScope Scope(*this, S, /*AsInlined=*/true);
-  CGM.getOpenMPRuntime().emitInlinedDirective(
-      *this, OMPD_distribute_simd,
-      [&S](CodeGenFunction &CGF, PrePostActionTy &) {
-        OMPLoopScope PreInitScope(CGF, S);
-        CGF.EmitStmt(
-            cast<CapturedStmt>(S.getAssociatedStmt())->getCapturedStmt());
-      });
+  CGM.getOpenMPRuntime().emitInlinedDirective(*this, OMPD_simd, CodeGen);
 }
 
 void CodeGenFunction::EmitOMPTargetSimdDeviceFunction(
@@ -3075,6 +3077,7 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPLoopDirective &S,
       incrementProfileCounter(&S);
     }
 
+    emitAlignedClause(*this, S);
     // Emit 'then' code.
     {
       // Emit helper vars inits.
@@ -3096,14 +3099,17 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPLoopDirective &S,
 
       OMPPrivateScope LoopScope(*this);
       if (EmitOMPFirstprivateClause(S, LoopScope)) {
-        // Emit implicit barrier to synchronize threads and avoid data races on
-        // initialization of firstprivate variables and post-update of
+        // Emit implicit barrier to synchronize threads and avoid data races
+        // on initialization of firstprivate variables and post-update of
         // lastprivate variables.
         CGM.getOpenMPRuntime().emitBarrierCall(
-          *this, S.getLocStart(), OMPD_unknown, /*EmitChecks=*/false,
-          /*ForceSimpleCall=*/true);
+            *this, S.getLocStart(), OMPD_unknown, /*EmitChecks=*/false,
+            /*ForceSimpleCall=*/true);
       }
       EmitOMPPrivateClause(S, LoopScope);
+      if (isOpenMPSimdDirective(S.getDirectiveKind()) &&
+          !isOpenMPParallelDirective(S.getDirectiveKind()))
+        EmitOMPReductionClauseInit(S, LoopScope);
       HasLastprivateClause = EmitOMPLastprivateClauseInit(S, LoopScope);
       EmitOMPPrivateLoopCounters(S, LoopScope);
       (void)LoopScope.Privatize();
@@ -3116,8 +3122,8 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPLoopDirective &S,
         if (const auto *Ch = C->getChunkSize()) {
           Chunk = EmitScalarExpr(Ch);
           Chunk = EmitScalarConversion(Chunk, Ch->getType(),
-          S.getIterationVariable()->getType(),
-          S.getLocStart());
+                                       S.getIterationVariable()->getType(),
+                                       S.getLocStart());
         }
       }
       const unsigned IVSize = getContext().getTypeSize(IVExpr->getType());
@@ -3133,6 +3139,8 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPLoopDirective &S,
       // league. The size of the chunks is unspecified in this case.
       if (RT.isStaticNonchunked(ScheduleKind,
                                 /* Chunked */ Chunk != nullptr)) {
+        if (isOpenMPSimdDirective(S.getDirectiveKind()))
+          EmitOMPSimdInit(S, /*IsMonotonic=*/true);
         CGOpenMPRuntime::StaticRTInput StaticInit(
             IVSize, IVSigned, /* Ordered = */ false, IL.getAddress(),
             LB.getAddress(), UB.getAddress(), ST.getAddress());
@@ -3174,13 +3182,38 @@ void CodeGenFunction::EmitOMPDistributeLoop(const OMPLoopDirective &S,
         EmitOMPDistributeOuterLoop(ScheduleKind, S, LoopScope, LoopArguments,
                                    CodeGenLoop);
       }
-
+      if (isOpenMPSimdDirective(S.getDirectiveKind())) {
+        EmitOMPSimdFinal(S, [&](CodeGenFunction &CGF) -> llvm::Value * {
+          return CGF.Builder.CreateIsNotNull(
+              CGF.EmitLoadOfScalar(IL, S.getLocStart()));
+        });
+      }
+      OpenMPDirectiveKind ReductionKind = OMPD_unknown;
+      if (isOpenMPParallelDirective(S.getDirectiveKind()) &&
+          isOpenMPSimdDirective(S.getDirectiveKind())) {
+        ReductionKind = OMPD_parallel_for_simd;
+      } else if (isOpenMPParallelDirective(S.getDirectiveKind())) {
+        ReductionKind = OMPD_parallel_for;
+      } else if (isOpenMPSimdDirective(S.getDirectiveKind())) {
+        ReductionKind = OMPD_simd;
+      } else if (!isOpenMPTeamsDirective(S.getDirectiveKind()) &&
+                 S.hasClausesOfKind<OMPReductionClause>()) {
+        llvm_unreachable(
+            "No reduction clauses is allowed in distribute directive.");
+      }
+      EmitOMPReductionClauseFinal(S, ReductionKind);
+      // Emit post-update of the reduction variables if IsLastIter != 0.
+      emitPostUpdateForReductionClause(
+          *this, S, [&](CodeGenFunction &CGF) -> llvm::Value * {
+            return CGF.Builder.CreateIsNotNull(
+                CGF.EmitLoadOfScalar(IL, S.getLocStart()));
+          });
       // Emit final copy of the lastprivate variables if IsLastIter != 0.
-      if (HasLastprivateClause)
+      if (HasLastprivateClause) {
         EmitOMPLastprivateClauseFinal(
             S, /*NoFinals=*/false,
-            Builder.CreateIsNotNull(
-                EmitLoadOfScalar(IL, S.getLocStart())));
+            Builder.CreateIsNotNull(EmitLoadOfScalar(IL, S.getLocStart())));
+      }
     }
 
     // We're now done with the loop, so jump to the continuation block.