[OPENMP] Fix emission of the __kmpc_global_thread_num.

Fixed emission of the __kmpc_global_thread_num() so that it is not
messed up with alloca instructions anymore. Plus, fixes emission of the
__kmpc_global_thread_num() functions in the target outlined regions so
that they are not called before runtime is initialized.

llvm-svn: 343856

3 files changed, 44 insertions, 4 deletions

diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.cpp b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
index d421729faa2..c00f66dd030 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -1485,6 +1485,31 @@ Address CGOpenMPRuntime::getOrCreateDefaultLocation(unsigned Flags) {
   return Address(Entry, Align);
 }
 
+void CGOpenMPRuntime::setLocThreadIdInsertPt(CodeGenFunction &CGF,
+                                             bool AtCurrentPoint) {
+  auto &Elem = OpenMPLocThreadIDMap.FindAndConstruct(CGF.CurFn);
+  assert(!Elem.second.ServiceInsertPt && "Insert point is set already.");
+
+  llvm::Value *Undef = llvm::UndefValue::get(CGF.Int32Ty);
+  if (AtCurrentPoint) {
+    Elem.second.ServiceInsertPt = new llvm::BitCastInst(
+        Undef, CGF.Int32Ty, "svcpt", CGF.Builder.GetInsertBlock());
+  } else {
+    Elem.second.ServiceInsertPt =
+        new llvm::BitCastInst(Undef, CGF.Int32Ty, "svcpt");
+    Elem.second.ServiceInsertPt->insertAfter(CGF.AllocaInsertPt);
+  }
+}
+
+void CGOpenMPRuntime::clearLocThreadIdInsertPt(CodeGenFunction &CGF) {
+  auto &Elem = OpenMPLocThreadIDMap.FindAndConstruct(CGF.CurFn);
+  if (Elem.second.ServiceInsertPt) {
+    llvm::Instruction *Ptr = Elem.second.ServiceInsertPt;
+    Elem.second.ServiceInsertPt = nullptr;
+    Ptr->eraseFromParent();
+  }
+}
+
 llvm::Value *CGOpenMPRuntime::emitUpdateLocation(CodeGenFunction &CGF,
                                                  SourceLocation Loc,
                                                  unsigned Flags) {
@@ -1511,8 +1536,10 @@ llvm::Value *CGOpenMPRuntime::emitUpdateLocation(CodeGenFunction &CGF,
     Elem.second.DebugLoc = AI.getPointer();
     LocValue = AI;
 
+    if (!Elem.second.ServiceInsertPt)
+      setLocThreadIdInsertPt(CGF);
     CGBuilderTy::InsertPointGuard IPG(CGF.Builder);
-    CGF.Builder.SetInsertPoint(CGF.AllocaInsertPt);
+    CGF.Builder.SetInsertPoint(Elem.second.ServiceInsertPt);
     CGF.Builder.CreateMemCpy(LocValue, getOrCreateDefaultLocation(Flags),
                              CGF.getTypeSize(IdentQTy));
   }
@@ -1582,21 +1609,25 @@ llvm::Value *CGOpenMPRuntime::getThreadID(CodeGenFunction &CGF,
   // kmpc_global_thread_num(ident_t *loc).
   // Generate thread id value and cache this value for use across the
   // function.
+  auto &Elem = OpenMPLocThreadIDMap.FindAndConstruct(CGF.CurFn);
+  if (!Elem.second.ServiceInsertPt)
+    setLocThreadIdInsertPt(CGF);
   CGBuilderTy::InsertPointGuard IPG(CGF.Builder);
-  CGF.Builder.SetInsertPoint(CGF.AllocaInsertPt);
+  CGF.Builder.SetInsertPoint(Elem.second.ServiceInsertPt);
   llvm::CallInst *Call = CGF.Builder.CreateCall(
       createRuntimeFunction(OMPRTL__kmpc_global_thread_num),
       emitUpdateLocation(CGF, Loc));
   Call->setCallingConv(CGF.getRuntimeCC());
-  auto &Elem = OpenMPLocThreadIDMap.FindAndConstruct(CGF.CurFn);
   Elem.second.ThreadID = Call;
   return Call;
 }
 
 void CGOpenMPRuntime::functionFinished(CodeGenFunction &CGF) {
   assert(CGF.CurFn && "No function in current CodeGenFunction.");
-  if (OpenMPLocThreadIDMap.count(CGF.CurFn))
+  if (OpenMPLocThreadIDMap.count(CGF.CurFn)) {
+    clearLocThreadIdInsertPt(CGF);
     OpenMPLocThreadIDMap.erase(CGF.CurFn);
+  }
   if (FunctionUDRMap.count(CGF.CurFn) > 0) {
     for(auto *D : FunctionUDRMap[CGF.CurFn])
       UDRMap.erase(D);
diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.h b/clang/lib/CodeGen/CGOpenMPRuntime.h
index 35f75a9ec08..3fb1a1538fa 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.h
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.h
@@ -278,6 +278,10 @@ protected:
   /// stored.
   virtual Address emitThreadIDAddress(CodeGenFunction &CGF, SourceLocation Loc);
 
+  void setLocThreadIdInsertPt(CodeGenFunction &CGF,
+                              bool AtCurrentPoint = false);
+  void clearLocThreadIdInsertPt(CodeGenFunction &CGF);
+
 private:
   /// Default const ident_t object used for initialization of all other
   /// ident_t objects.
@@ -300,6 +304,8 @@ private:
   struct DebugLocThreadIdTy {
     llvm::Value *DebugLoc;
     llvm::Value *ThreadID;
+    /// Insert point for the service instructions.
+    llvm::AssertingVH<llvm::Instruction> ServiceInsertPt = nullptr;
   };
   /// Map of local debug location, ThreadId and functions.
   typedef llvm::DenseMap<llvm::Function *, DebugLocThreadIdTy>
diff --git a/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp b/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
index a923232e817..f0f0a735f8a 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntimeNVPTX.cpp
@@ -1197,8 +1197,11 @@ void CGOpenMPRuntimeNVPTX::emitSPMDKernel(const OMPExecutableDirective &D,
         : RT(RT), EST(EST), D(D) {}
     void Enter(CodeGenFunction &CGF) override {
       RT.emitSPMDEntryHeader(CGF, EST, D);
+      // Skip target region initialization.
+      RT.setLocThreadIdInsertPt(CGF, /*AtCurrentPoint=*/true);
     }
     void Exit(CodeGenFunction &CGF) override {
+      RT.clearLocThreadIdInsertPt(CGF);
       RT.emitSPMDEntryFooter(CGF, EST);
     }
   } Action(*this, EST, D);