[Clang][OpenMP Offload] Move offload registration code to the wrapper

The final list of OpenMP offload targets becomes known only at the link time and since offload registration code depends on the targets list it makes sense to delay offload registration code generation to the link time instead of adding it to the host part of every fat object. This patch moves offload registration code generation from clang to the offload wrapper tool.

This is the last part of the OpenMP linker script elimination patch https://reviews.llvm.org/D64943

Differential Revision: https://reviews.llvm.org/D68746

llvm-svn: 374937

3 files changed, 7 insertions, 191 deletions

diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.cpp b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
index ae2b4853356..42d18b4098e 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.cpp
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.cpp
@@ -4006,157 +4006,6 @@ void CGOpenMPRuntime::OffloadEntriesInfoManagerTy::
     Action(E.getKey(), E.getValue());
 }
 
-llvm::Function *
-CGOpenMPRuntime::createOffloadingBinaryDescriptorRegistration() {
-  // If we don't have entries or if we are emitting code for the device, we
-  // don't need to do anything.
-  if (CGM.getLangOpts().OpenMPIsDevice || OffloadEntriesInfoManager.empty())
-    return nullptr;
-
-  llvm::Module &M = CGM.getModule();
-  ASTContext &C = CGM.getContext();
-
-  // Get list of devices we care about
-  const std::vector<llvm::Triple> &Devices = CGM.getLangOpts().OMPTargetTriples;
-
-  // We should be creating an offloading descriptor only if there are devices
-  // specified.
-  assert(!Devices.empty() && "No OpenMP offloading devices??");
-
-  // Create the external variables that will point to the begin and end of the
-  // host entries section. These will be defined by the linker.
-  llvm::Type *OffloadEntryTy =
-      CGM.getTypes().ConvertTypeForMem(getTgtOffloadEntryQTy());
-  auto *HostEntriesBegin = new llvm::GlobalVariable(
-      M, OffloadEntryTy, /*isConstant=*/true,
-      llvm::GlobalValue::ExternalLinkage, /*Initializer=*/nullptr,
-      "__start_omp_offloading_entries");
-  HostEntriesBegin->setVisibility(llvm::GlobalValue::HiddenVisibility);
-  auto *HostEntriesEnd = new llvm::GlobalVariable(
-      M, OffloadEntryTy, /*isConstant=*/true,
-      llvm::GlobalValue::ExternalLinkage,
-      /*Initializer=*/nullptr, "__stop_omp_offloading_entries");
-  HostEntriesEnd->setVisibility(llvm::GlobalValue::HiddenVisibility);
-
-  // Create all device images
-  auto *DeviceImageTy = cast<llvm::StructType>(
-      CGM.getTypes().ConvertTypeForMem(getTgtDeviceImageQTy()));
-  ConstantInitBuilder DeviceImagesBuilder(CGM);
-  ConstantArrayBuilder DeviceImagesEntries =
-      DeviceImagesBuilder.beginArray(DeviceImageTy);
-
-  for (const llvm::Triple &Device : Devices) {
-    StringRef T = Device.getTriple();
-    std::string BeginName = getName({"omp_offloading", "img_start", ""});
-    auto *ImgBegin = new llvm::GlobalVariable(
-        M, CGM.Int8Ty, /*isConstant=*/true,
-        llvm::GlobalValue::ExternalWeakLinkage,
-        /*Initializer=*/nullptr, Twine(BeginName).concat(T));
-    std::string EndName = getName({"omp_offloading", "img_end", ""});
-    auto *ImgEnd = new llvm::GlobalVariable(
-        M, CGM.Int8Ty, /*isConstant=*/true,
-        llvm::GlobalValue::ExternalWeakLinkage,
-        /*Initializer=*/nullptr, Twine(EndName).concat(T));
-
-    llvm::Constant *Data[] = {ImgBegin, ImgEnd, HostEntriesBegin,
-                              HostEntriesEnd};
-    createConstantGlobalStructAndAddToParent(CGM, getTgtDeviceImageQTy(), Data,
-                                             DeviceImagesEntries);
-  }
-
-  // Create device images global array.
-  std::string ImagesName = getName({"omp_offloading", "device_images"});
-  llvm::GlobalVariable *DeviceImages =
-      DeviceImagesEntries.finishAndCreateGlobal(ImagesName,
-                                                CGM.getPointerAlign(),
-                                                /*isConstant=*/true);
-  DeviceImages->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
-
-  // This is a Zero array to be used in the creation of the constant expressions
-  llvm::Constant *Index[] = {llvm::Constant::getNullValue(CGM.Int32Ty),
-                             llvm::Constant::getNullValue(CGM.Int32Ty)};
-
-  // Create the target region descriptor.
-  llvm::Constant *Data[] = {
-      llvm::ConstantInt::get(CGM.Int32Ty, Devices.size()),
-      llvm::ConstantExpr::getGetElementPtr(DeviceImages->getValueType(),
-                                           DeviceImages, Index),
-      HostEntriesBegin, HostEntriesEnd};
-  std::string Descriptor = getName({"omp_offloading", "descriptor"});
-  llvm::GlobalVariable *Desc = createGlobalStruct(
-      CGM, getTgtBinaryDescriptorQTy(), /*IsConstant=*/true, Data, Descriptor);
-
-  // Emit code to register or unregister the descriptor at execution
-  // startup or closing, respectively.
-
-  llvm::Function *UnRegFn;
-  {
-    FunctionArgList Args;
-    ImplicitParamDecl DummyPtr(C, C.VoidPtrTy, ImplicitParamDecl::Other);
-    Args.push_back(&DummyPtr);
-
-    CodeGenFunction CGF(CGM);
-    // Disable debug info for global (de-)initializer because they are not part
-    // of some particular construct.
-    CGF.disableDebugInfo();
-    const auto &FI =
-        CGM.getTypes().arrangeBuiltinFunctionDeclaration(C.VoidTy, Args);
-    llvm::FunctionType *FTy = CGM.getTypes().GetFunctionType(FI);
-    std::string UnregName = getName({"omp_offloading", "descriptor_unreg"});
-    UnRegFn = CGM.CreateGlobalInitOrDestructFunction(FTy, UnregName, FI);
-    CGF.StartFunction(GlobalDecl(), C.VoidTy, UnRegFn, FI, Args);
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__tgt_unregister_lib),
-                        Desc);
-    CGF.FinishFunction();
-  }
-  llvm::Function *RegFn;
-  {
-    CodeGenFunction CGF(CGM);
-    // Disable debug info for global (de-)initializer because they are not part
-    // of some particular construct.
-    CGF.disableDebugInfo();
-    const auto &FI = CGM.getTypes().arrangeNullaryFunction();
-    llvm::FunctionType *FTy = CGM.getTypes().GetFunctionType(FI);
-
-    // Encode offload target triples into the registration function name. It
-    // will serve as a comdat key for the registration/unregistration code for
-    // this particular combination of offloading targets.
-    SmallVector<StringRef, 4U> RegFnNameParts(Devices.size() + 2U);
-    RegFnNameParts[0] = "omp_offloading";
-    RegFnNameParts[1] = "descriptor_reg";
-    llvm::transform(Devices, std::next(RegFnNameParts.begin(), 2),
-                    [](const llvm::Triple &T) -> const std::string& {
-                      return T.getTriple();
-                    });
-    llvm::sort(std::next(RegFnNameParts.begin(), 2), RegFnNameParts.end());
-    std::string Descriptor = getName(RegFnNameParts);
-    RegFn = CGM.CreateGlobalInitOrDestructFunction(FTy, Descriptor, FI);
-    CGF.StartFunction(GlobalDecl(), C.VoidTy, RegFn, FI, FunctionArgList());
-    CGF.EmitRuntimeCall(createRuntimeFunction(OMPRTL__tgt_register_lib), Desc);
-    // Create a variable to drive the registration and unregistration of the
-    // descriptor, so we can reuse the logic that emits Ctors and Dtors.
-    ImplicitParamDecl RegUnregVar(C, C.getTranslationUnitDecl(),
-                                  SourceLocation(), nullptr, C.CharTy,
-                                  ImplicitParamDecl::Other);
-    CGM.getCXXABI().registerGlobalDtor(CGF, RegUnregVar, UnRegFn, Desc);
-    CGF.FinishFunction();
-  }
-  if (CGM.supportsCOMDAT()) {
-    // It is sufficient to call registration function only once, so create a
-    // COMDAT group for registration/unregistration functions and associated
-    // data. That would reduce startup time and code size. Registration
-    // function serves as a COMDAT group key.
-    llvm::Comdat *ComdatKey = M.getOrInsertComdat(RegFn->getName());
-    RegFn->setLinkage(llvm::GlobalValue::LinkOnceAnyLinkage);
-    RegFn->setVisibility(llvm::GlobalValue::HiddenVisibility);
-    RegFn->setComdat(ComdatKey);
-    UnRegFn->setComdat(ComdatKey);
-    DeviceImages->setComdat(ComdatKey);
-    Desc->setComdat(ComdatKey);
-  }
-  return RegFn;
-}
-
 void CGOpenMPRuntime::createOffloadEntry(
     llvm::Constant *ID, llvm::Constant *Addr, uint64_t Size, int32_t Flags,
     llvm::GlobalValue::LinkageTypes Linkage) {
@@ -4197,8 +4046,9 @@ void CGOpenMPRuntime::createOffloadEntriesAndInfoMetadata() {
   // Right now we only generate metadata for function that contain target
   // regions.
 
-  // If we do not have entries, we don't need to do anything.
-  if (OffloadEntriesInfoManager.empty())
+  // If we are in simd mode or there are no entries, we don't need to do
+  // anything.
+  if (CGM.getLangOpts().OpenMPSimd || OffloadEntriesInfoManager.empty())
     return;
 
   llvm::Module &M = CGM.getModule();
@@ -10031,17 +9881,6 @@ llvm::Function *CGOpenMPRuntime::emitRequiresDirectiveRegFun() {
   return RequiresRegFn;
 }
 
-llvm::Function *CGOpenMPRuntime::emitRegistrationFunction() {
-  // If we have offloading in the current module, we need to emit the entries
-  // now and register the offloading descriptor.
-  createOffloadEntriesAndInfoMetadata();
-
-  // Create and register the offloading binary descriptors. This is the main
-  // entity that captures all the information about offloading in the current
-  // compilation unit.
-  return createOffloadingBinaryDescriptorRegistration();
-}
-
 void CGOpenMPRuntime::emitTeamsCall(CodeGenFunction &CGF,
                                     const OMPExecutableDirective &D,
                                     SourceLocation Loc,
@@ -11534,10 +11373,6 @@ bool CGOpenMPSIMDRuntime::emitTargetGlobal(GlobalDecl GD) {
   return false;
 }
 
-llvm::Function *CGOpenMPSIMDRuntime::emitRegistrationFunction() {
-  return nullptr;
-}
-
 void CGOpenMPSIMDRuntime::emitTeamsCall(CodeGenFunction &CGF,
                                         const OMPExecutableDirective &D,
                                         SourceLocation Loc,
diff --git a/clang/lib/CodeGen/CGOpenMPRuntime.h b/clang/lib/CodeGen/CGOpenMPRuntime.h
index 9215bd666c9..bf8e0ac8090 100644
--- a/clang/lib/CodeGen/CGOpenMPRuntime.h
+++ b/clang/lib/CodeGen/CGOpenMPRuntime.h
@@ -672,14 +672,6 @@ private:
   /// Device routines are specific to the
   bool HasEmittedDeclareTargetRegion = false;
 
-  /// Creates and registers offloading binary descriptor for the current
-  /// compilation unit. The function that does the registration is returned.
-  llvm::Function *createOffloadingBinaryDescriptorRegistration();
-
-  /// Creates all the offload entries in the current compilation unit
-  /// along with the associated metadata.
-  void createOffloadEntriesAndInfoMetadata();
-
   /// Loads all the offload entries information from the host IR
   /// metadata.
   void loadOffloadInfoMetadata();
@@ -1492,10 +1484,9 @@ public:
   /// requires directives was used in the current module.
   llvm::Function *emitRequiresDirectiveRegFun();
 
-  /// Creates the offloading descriptor in the event any target region
-  /// was emitted in the current module and return the function that registers
-  /// it.
-  virtual llvm::Function *emitRegistrationFunction();
+  /// Creates all the offload entries in the current compilation unit
+  /// along with the associated metadata.
+  void createOffloadEntriesAndInfoMetadata();
 
   /// Emits code for teams call of the \a OutlinedFn with
   /// variables captured in a record which address is stored in \a
@@ -2167,11 +2158,6 @@ public:
   /// \param GD Global to scan.
   bool emitTargetGlobal(GlobalDecl GD) override;
 
-  /// Creates the offloading descriptor in the event any target region
-  /// was emitted in the current module and return the function that registers
-  /// it.
-  llvm::Function *emitRegistrationFunction() override;
-
   /// Emits code for teams call of the \a OutlinedFn with
   /// variables captured in a record which address is stored in \a
   /// CapturedStruct.
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index c97e7b22170..8eb2176ca3a 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -414,12 +414,7 @@ void CodeGenModule::Release() {
             OpenMPRuntime->emitRequiresDirectiveRegFun()) {
       AddGlobalCtor(OpenMPRequiresDirectiveRegFun, 0);
     }
-    if (llvm::Function *OpenMPRegistrationFunction =
-            OpenMPRuntime->emitRegistrationFunction()) {
-      auto ComdatKey = OpenMPRegistrationFunction->hasComdat() ?
-        OpenMPRegistrationFunction : nullptr;
-      AddGlobalCtor(OpenMPRegistrationFunction, 0, ComdatKey);
-    }
+    OpenMPRuntime->createOffloadEntriesAndInfoMetadata();
     OpenMPRuntime->clear();
   }
   if (PGOReader) {