[Orc] Add support classes for inspecting and running C++ static ctor/dtors, and

use these to add support for C++ static ctors/dtors to the Orc-lazy JIT in LLI.

Replace the trivial_retval_1 regression test - the new 'hello' test is covering
strictly more code. 

llvm-svn: 233885

llvm/include/llvm/ExecutionEngine/Orc/ExecutionUtils.h

@@ -0,0 +1,182 @@
+//===-- ExecutionUtils.h - Utilities for executing code in Orc --*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Contains utilities for executing code in Orc.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_EXECUTIONENGINE_ORC_EXECUTIONUTILS_H
+#define LLVM_EXECUTIONENGINE_ORC_EXECUTIONUTILS_H
+
+#include "JITSymbol.h"
+#include "llvm/ADT/iterator_range.h"
+#include "llvm/ADT/StringMap.h"
+#include "llvm/ExecutionEngine/RuntimeDyld.h"
+#include <vector>
+
+namespace llvm {
+
+class ConstantArray;
+class GlobalVariable;
+class Function;
+class Module;
+class Value;
+
+namespace orc {
+
+/// @brief This iterator provides a convenient way to iterate over the elements
+///        of an llvm.global_ctors/llvm.global_dtors instance.
+///
+///   The easiest way to get hold of instances of this class is to use the
+/// getConstructors/getDestructors functions.
+class CtorDtorIterator {
+public:
+
+  /// @brief Accessor for an element of the global_ctors/global_dtors array.
+  ///
+  ///   This class provides a read-only view of the element with any casts on
+  /// the function stripped away.
+  struct Element {
+    Element(unsigned Priority, const Function *Func, const Value *Data)
+      : Priority(Priority), Func(Func), Data(Data) {}
+
+    unsigned Priority;
+    const Function *Func;
+    const Value *Data;
+  };
+
+  /// @brief Construct an iterator instance. If End is true then this iterator
+  ///        acts as the end of the range, otherwise it is the beginning.
+  CtorDtorIterator(const GlobalVariable *GV, bool End);
+
+  /// @brief Test iterators for equality.
+  bool operator==(const CtorDtorIterator &Other) const;
+
+  /// @brief Test iterators for inequality.
+  bool operator!=(const CtorDtorIterator &Other) const;
+
+  /// @brief Pre-increment iterator.
+  CtorDtorIterator& operator++();
+
+  /// @brief Post-increment iterator.
+  CtorDtorIterator operator++(int);
+
+  /// @brief Dereference iterator. The resulting value provides a read-only view
+  ///        of this element of the global_ctors/global_dtors list.
+  Element operator*() const;
+
+private:
+  const ConstantArray *InitList;
+  unsigned I;
+};
+
+/// @brief Create an iterator range over the entries of the llvm.global_ctors
+///        array.
+iterator_range<CtorDtorIterator> getConstructors(const Module &M);
+
+/// @brief Create an iterator range over the entries of the llvm.global_ctors
+///        array.
+iterator_range<CtorDtorIterator> getDestructors(const Module &M);
+
+/// @brief Convenience class for recording constructor/destructor names for
+///        later execution.
+template <typename JITLayerT>
+class CtorDtorRunner {
+public:
+
+  /// @brief Construct a CtorDtorRunner for the given range using the given
+  ///        name mangling function.
+  CtorDtorRunner(std::vector<std::string> CtorDtorNames,
+                 typename JITLayerT::ModuleSetHandleT H)
+      : CtorDtorNames(std::move(CtorDtorNames)), H(H) {}
+
+  /// @brief Run the recorded constructors/destructors through the given JIT
+  ///        layer.
+  bool runViaLayer(JITLayerT &JITLayer) const {
+    typedef void (*CtorDtorTy)();
+
+    bool Error = false;
+    for (const auto &CtorDtorName : CtorDtorNames)
+      if (auto CtorDtorSym = JITLayer.findSymbolIn(H, CtorDtorName, false)) {
+        CtorDtorTy CtorDtor =
+          reinterpret_cast<CtorDtorTy>(
+            static_cast<uintptr_t>(CtorDtorSym.getAddress()));
+        CtorDtor();
+      } else
+        Error = true;
+    return !Error;
+  }
+
+private:
+  std::vector<std::string> CtorDtorNames;
+  typename JITLayerT::ModuleSetHandleT H;
+};
+
+/// @brief Support class for static dtor execution. For hosted (in-process) JITs
+///        only!
+///
+///   If a __cxa_atexit function isn't found C++ programs that use static
+/// destructors will fail to link. However, we don't want to use the host
+/// process's __cxa_atexit, because it will schedule JIT'd destructors to run
+/// after the JIT has been torn down, which is no good. This class makes it easy
+/// to override __cxa_atexit (and the related __dso_handle).
+///
+///   To use, clients should manually call searchOverrides from their symbol
+/// resolver. This should generally be done after attempting symbol resolution
+/// inside the JIT, but before searching the host process's symbol table. When
+/// the client determines that destructors should be run (generally at JIT
+/// teardown or after a return from main), the runDestructors method should be
+/// called.
+class LocalCXXRuntimeOverrides {
+public:
+
+  /// Create a runtime-overrides class.
+  template <typename MangleFtorT>
+  LocalCXXRuntimeOverrides(const MangleFtorT &Mangle) {
+    addOverride(Mangle("__dso_handle"), toTargetAddress(&DSOHandleOverride));
+    addOverride(Mangle("__cxa_atexit"), toTargetAddress(&CXAAtExitOverride));
+  }
+
+  /// Search overrided symbols.
+  RuntimeDyld::SymbolInfo searchOverrides(const std::string &Name) {
+    auto I = CXXRuntimeOverrides.find(Name);
+    if (I != CXXRuntimeOverrides.end())
+      return RuntimeDyld::SymbolInfo(I->second, JITSymbolFlags::Exported);
+    return nullptr;
+  }
+
+  /// Run any destructors recorded by the overriden __cxa_atexit function
+  /// (CXAAtExitOverride).
+  void runDestructors();
+
+private:
+
+  template <typename PtrTy>
+  TargetAddress toTargetAddress(PtrTy* P) {
+    return static_cast<TargetAddress>(reinterpret_cast<uintptr_t>(P));
+  }
+
+  void addOverride(const std::string &Name, TargetAddress Addr) {
+    CXXRuntimeOverrides.insert(std::make_pair(Name, Addr));
+  }
+
+  StringMap<TargetAddress> CXXRuntimeOverrides;
+
+  typedef void (*DestructorPtr)(void*);
+  typedef std::pair<DestructorPtr, void*> CXXDestructorDataPair;
+  typedef std::vector<CXXDestructorDataPair> CXXDestructorDataPairList;
+  CXXDestructorDataPairList DSOHandleOverride;
+  static int CXAAtExitOverride(DestructorPtr Destructor, void *Arg,
+                               void *DSOHandle);
+};
+
+} // End namespace orc.
+} // End namespace llvm.
+
+#endif // LLVM_EXECUTIONENGINE_ORC_EXECUTIONUTILS_H

llvm/lib/ExecutionEngine/Orc/CMakeLists.txt

@@ -1,5 +1,6 @@
 add_llvm_library(LLVMOrcJIT
   CloneSubModule.cpp
+  ExecutionUtils.cpp
   IndirectionUtils.cpp
   OrcMCJITReplacement.cpp
   OrcTargetSupport.cpp

llvm/lib/ExecutionEngine/Orc/ExecutionUtils.cpp

@@ -0,0 +1,102 @@
+//===---- ExecutionUtils.cpp - Utilities for executing functions in Orc ---===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/ExecutionEngine/Orc/ExecutionUtils.h"
+
+#include "llvm/IR/Constants.h"
+#include "llvm/IR/Function.h"
+#include "llvm/IR/GlobalVariable.h"
+#include "llvm/IR/Module.h"
+
+namespace llvm {
+namespace orc {
+
+CtorDtorIterator::CtorDtorIterator(const GlobalVariable *GV, bool End)
+  : InitList(
+      GV ? dyn_cast_or_null<ConstantArray>(GV->getInitializer()) : nullptr),
+    I((InitList && End) ? InitList->getNumOperands() : 0) {
+}
+
+bool CtorDtorIterator::operator==(const CtorDtorIterator &Other) const {
+  assert(InitList == Other.InitList && "Incomparable iterators.");
+  return I == Other.I;
+}
+
+bool CtorDtorIterator::operator!=(const CtorDtorIterator &Other) const {
+  return !(*this == Other);
+}
+
+CtorDtorIterator& CtorDtorIterator::operator++() {
+  ++I;
+  return *this;
+}
+
+CtorDtorIterator CtorDtorIterator::operator++(int) {
+  CtorDtorIterator Temp = *this;
+  ++I;
+  return Temp;
+}
+
+CtorDtorIterator::Element CtorDtorIterator::operator*() const {
+  ConstantStruct *CS = dyn_cast<ConstantStruct>(InitList->getOperand(I));
+  assert(CS && "Unrecognized type in llvm.global_ctors/llvm.global_dtors");
+
+  Constant *FuncC = CS->getOperand(1);
+  Function *Func = nullptr;
+
+  // Extract function pointer, pulling off any casts.
+  while (FuncC) {
+    if (Function *F = dyn_cast_or_null<Function>(FuncC)) {
+      Func = F;
+      break;
+    } else if (ConstantExpr *CE = dyn_cast_or_null<ConstantExpr>(FuncC)) {
+      if (CE->isCast())
+        FuncC = dyn_cast_or_null<ConstantExpr>(CE->getOperand(0));
+      else
+        break;
+    } else {
+      // This isn't anything we recognize. Bail out with Func left set to null.
+      break;
+    }
+  }
+
+  ConstantInt *Priority = dyn_cast<ConstantInt>(CS->getOperand(0));
+  Value *Data = CS->getOperand(2);
+  return Element(Priority->getZExtValue(), Func, Data);
+}
+
+iterator_range<CtorDtorIterator> getConstructors(const Module &M) {
+  const GlobalVariable *CtorsList = M.getNamedGlobal("llvm.global_ctors");
+  return make_range(CtorDtorIterator(CtorsList, false),
+                    CtorDtorIterator(CtorsList, true));
+}
+
+iterator_range<CtorDtorIterator> getDestructors(const Module &M) {
+  const GlobalVariable *DtorsList = M.getNamedGlobal("llvm.global_dtors");
+  return make_range(CtorDtorIterator(DtorsList, false),
+                    CtorDtorIterator(DtorsList, true));
+}
+
+void LocalCXXRuntimeOverrides::runDestructors() {
+  auto& CXXDestructorDataPairs = DSOHandleOverride;
+  for (auto &P : CXXDestructorDataPairs)
+    P.first(P.second);
+  CXXDestructorDataPairs.clear();
+}
+
+int LocalCXXRuntimeOverrides::CXAAtExitOverride(DestructorPtr Destructor,
+                                                void *Arg, void *DSOHandle) {
+  auto& CXXDestructorDataPairs =
+    *reinterpret_cast<CXXDestructorDataPairList*>(DSOHandle);
+  CXXDestructorDataPairs.push_back(std::make_pair(Destructor, Arg));
+  return 0;
+}
+
+} // End namespace orc.
+} // End namespace llvm.

llvm/test/ExecutionEngine/OrcLazy/hello.ll

@@ -0,0 +1,34 @@
+; RUN: lli -jit-kind=orc-lazy %s | FileCheck %s
+;
+; CHECK: Hello
+; CHECK-NEXT: Goodbye
+
+%class.Foo = type { i8 }
+
+@f = global %class.Foo zeroinitializer, align 1
+@__dso_handle = external global i8
+@llvm.global_ctors = appending global [1 x { i32, void ()*, i8* }] [{ i32, void ()*, i8* } { i32 65535, void ()* @_GLOBAL__sub_I_hello.cpp, i8* null }]
+@str = private unnamed_addr constant [6 x i8] c"Hello\00"
+@str2 = private unnamed_addr constant [8 x i8] c"Goodbye\00"
+
+define linkonce_odr void @_ZN3FooD1Ev(%class.Foo* nocapture readnone %this) unnamed_addr align 2 {
+entry:
+  %puts.i = tail call i32 @puts(i8* getelementptr inbounds ([8 x i8], [8 x i8]* @str2, i64 0, i64 0))
+  ret void
+}
+
+declare i32 @__cxa_atexit(void (i8*)*, i8*, i8*)
+
+define i32 @main(i32 %argc, i8** nocapture readnone %argv) {
+entry:
+  ret i32 0
+}
+
+define internal void @_GLOBAL__sub_I_hello.cpp() {
+entry:
+  %puts.i.i.i = tail call i32 @puts(i8* getelementptr inbounds ([6 x i8], [6 x i8]* @str, i64 0, i64 0))
+  %0 = tail call i32 @__cxa_atexit(void (i8*)* bitcast (void (%class.Foo*)* @_ZN3FooD1Ev to void (i8*)*), i8* getelementptr inbounds (%class.Foo, %class.Foo* @f, i64 0, i32 0), i8* @__dso_handle)
+  ret void
+}
+
+declare i32 @puts(i8* nocapture readonly)

llvm/test/ExecutionEngine/OrcLazy/trivial_retval_1.ll

@@ -1,26 +0,0 @@
-; RUN: sh -c 'lli -jit-kind=orc-lazy %s; echo $?' | FileCheck %s
-; CHECK: {{^30$}}
-define i32 @baz() {
-entry:
-  ret i32 2
-}
-
-define i32 @bar() {
-entry:
-  %call = call i32 @baz()
-  %mul = mul nsw i32 3, %call
-  ret i32 %mul
-}
-
-define i32 @foo() {
-entry:
-  %call = call i32 @bar()
-  %mul = mul nsw i32 5, %call
-  ret i32 %mul
-}
-
-define i32 @main(i32 %argc, i8** %argv) {
-entry:
-  %call = call i32 @foo()
-  ret i32 %call
-}

llvm/tools/lli/OrcLazyJIT.cpp

@@ -64,8 +64,6 @@ int llvm::runOrcLazyJIT(std::unique_ptr<Module> M, int ArgC, char* ArgV[]) {
   }
 
   typedef int (*MainFnPtr)(int, char*[]);
-  auto Main = reinterpret_cast<MainFnPtr>(
-                static_cast<uintptr_t>(MainSym.getAddress()));
-
+  auto Main = OrcLazyJIT::fromTargetAddress<MainFnPtr>(MainSym.getAddress());
   return Main(ArgC, ArgV);
 }

llvm/tools/lli/OrcLazyJIT.h

@@ -18,6 +18,7 @@
 #include "llvm/ADT/Triple.h"
 #include "llvm/ExecutionEngine/Orc/CompileOnDemandLayer.h"
 #include "llvm/ExecutionEngine/Orc/CompileUtils.h"
+#include "llvm/ExecutionEngine/Orc/ExecutionUtils.h"
 #include "llvm/ExecutionEngine/Orc/IRCompileLayer.h"
 #include "llvm/ExecutionEngine/Orc/LazyEmittingLayer.h"
 #include "llvm/ExecutionEngine/Orc/ObjectLinkingLayer.h"
@@ -53,22 +54,68 @@ public:
       CompileLayer(ObjectLayer, orc::SimpleCompiler(*this->TM)),
       LazyEmitLayer(CompileLayer),
       CCMgr(BuildCallbackMgr(CompileLayer, CCMgrMemMgr, Context)),
-      CODLayer(LazyEmitLayer, *CCMgr) { }
+      CODLayer(LazyEmitLayer, *CCMgr),
+      CXXRuntimeOverrides([this](const std::string &S) { return mangle(S); }) {}
+
+  ~OrcLazyJIT() {
+    // Run any destructors registered with __cxa_atexit.
+    CXXRuntimeOverrides.runDestructors();
+    // Run any IR destructors.
+    for (auto &DtorRunner : IRStaticDestructorRunners)
+      DtorRunner.runViaLayer(CODLayer);
+  }
+
+  template <typename PtrTy>
+  static PtrTy fromTargetAddress(orc::TargetAddress Addr) {
+    return reinterpret_cast<PtrTy>(static_cast<uintptr_t>(Addr));
+  }
 
   ModuleHandleT addModule(std::unique_ptr<Module> M) {
     // Attach a data-layout if one isn't already present.
     if (M->getDataLayout().isDefault())
       M->setDataLayout(*TM->getDataLayout());
 
-    std::vector<std::unique_ptr<Module>> S;
-    S.push_back(std::move(M));
+    // Record the static constructors and destructors. We have to do this before
+    // we hand over ownership of the module to the JIT.
+    std::vector<std::string> CtorNames, DtorNames;
+    for (auto Ctor : orc::getConstructors(*M))
+      CtorNames.push_back(mangle(Ctor.Func->getName()));
+    for (auto Dtor : orc::getDestructors(*M))
+      DtorNames.push_back(mangle(Dtor.Func->getName()));
+
+    // Symbol resolution order:
+    //   1) Search the JIT symbols.
+    //   2) Check for C++ runtime overrides.
+    //   3) Search the host process (LLI)'s symbol table.
     auto FallbackLookup =
-      [](const std::string &Name) {
+      [this](const std::string &Name) {
+
+        if (auto Sym = CODLayer.findSymbol(Name, true))
+          return RuntimeDyld::SymbolInfo(Sym.getAddress(), Sym.getFlags());
+
+        if (auto Sym = CXXRuntimeOverrides.searchOverrides(Name))
+          return Sym;
+
         if (auto Addr = RTDyldMemoryManager::getSymbolAddressInProcess(Name))
           return RuntimeDyld::SymbolInfo(Addr, JITSymbolFlags::Exported);
+
         return RuntimeDyld::SymbolInfo(nullptr);
       };
-    return CODLayer.addModuleSet(std::move(S), std::move(FallbackLookup));
+
+    // Add the module to the JIT.
+    std::vector<std::unique_ptr<Module>> S;
+    S.push_back(std::move(M));
+    auto H = CODLayer.addModuleSet(std::move(S), std::move(FallbackLookup));
+
+    // Run the static constructors, and save the static destructor runner for
+    // execution when the JIT is torn down.
+    orc::CtorDtorRunner<CODLayerT> CtorRunner(std::move(CtorNames), H);
+    CtorRunner.runViaLayer(CODLayer);
+
+    IRStaticDestructorRunners.push_back(
+        orc::CtorDtorRunner<CODLayerT>(std::move(DtorNames), H));
+
+    return H;
   }
 
   orc::JITSymbol findSymbol(const std::string &Name) {
@@ -99,6 +146,9 @@ private:
   LazyEmitLayerT LazyEmitLayer;
   std::unique_ptr<CompileCallbackMgr> CCMgr;
   CODLayerT CODLayer;
+
+  orc::LocalCXXRuntimeOverrides CXXRuntimeOverrides;
+  std::vector<orc::CtorDtorRunner<CODLayerT>> IRStaticDestructorRunners;
 };
 
 int runOrcLazyJIT(std::unique_ptr<Module> M, int ArgC, char* ArgV[]);