[Kaleidoscope][BuildingAJIT] Add code for Chapter 5 - remote JITing.

This chapter demonstrates lazily JITing from ASTs with the expressions being
executed on a remote machine via a TCP connection. It needs some polish, but is
substantially complete.

Currently x86-64 SysV ABI (Darwin and Linux) only, but other architectures
can be supported by changing the server code to use alternative ABI support
classes from llvm/include/llvm/ExecutionEngine/Orc/OrcABISupport.h.

llvm-svn: 271193

llvm/examples/Kaleidoscope/BuildingAJIT/CMakeLists.txt

@@ -2,4 +2,5 @@ add_subdirectory(Chapter1)
 add_subdirectory(Chapter2)
 add_subdirectory(Chapter3)
 add_subdirectory(Chapter4)
+add_subdirectory(Chapter5)
 

llvm/examples/Kaleidoscope/BuildingAJIT/Chapter5/CMakeLists.txt

@@ -0,0 +1,21 @@
+add_subdirectory(Server)
+
+set(LLVM_LINK_COMPONENTS
+  Analysis
+  Core
+  ExecutionEngine
+  InstCombine
+  Object
+  OrcJIT
+  RuntimeDyld
+  ScalarOpts
+  Support
+  TransformUtils
+  native
+  )
+
+add_kaleidoscope_chapter(BuildingAJIT-Ch5
+  toy.cpp
+  )
+
+export_executable_symbols(BuildingAJIT-Ch5)

llvm/examples/Kaleidoscope/BuildingAJIT/Chapter5/KaleidoscopeJIT.h

@@ -0,0 +1,263 @@
+//===----- KaleidoscopeJIT.h - A simple JIT for Kaleidoscope ----*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Contains a simple JIT definition for use in the kaleidoscope tutorials.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_EXECUTIONENGINE_ORC_KALEIDOSCOPEJIT_H
+#define LLVM_EXECUTIONENGINE_ORC_KALEIDOSCOPEJIT_H
+
+#include "RemoteJITUtils.h"
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/ExecutionEngine/ExecutionEngine.h"
+#include "llvm/ExecutionEngine/RuntimeDyld.h"
+#include "llvm/ExecutionEngine/SectionMemoryManager.h"
+#include "llvm/ExecutionEngine/Orc/CompileOnDemandLayer.h"
+#include "llvm/ExecutionEngine/Orc/CompileUtils.h"
+#include "llvm/ExecutionEngine/Orc/JITSymbol.h"
+#include "llvm/ExecutionEngine/Orc/IRCompileLayer.h"
+#include "llvm/ExecutionEngine/Orc/IRTransformLayer.h"
+#include "llvm/ExecutionEngine/Orc/LambdaResolver.h"
+#include "llvm/ExecutionEngine/Orc/ObjectLinkingLayer.h"
+#include "llvm/ExecutionEngine/Orc/OrcRemoteTargetClient.h"
+#include "llvm/IR/DataLayout.h"
+#include "llvm/IR/Mangler.h"
+#include "llvm/Support/DynamicLibrary.h"
+#include "llvm/Support/raw_ostream.h"
+#include "llvm/Target/TargetMachine.h"
+#include <algorithm>
+#include <memory>
+#include <string>
+#include <vector>
+
+class PrototypeAST;
+class ExprAST;
+
+/// FunctionAST - This class represents a function definition itself.
+class FunctionAST {
+  std::unique_ptr<PrototypeAST> Proto;
+  std::unique_ptr<ExprAST> Body;
+
+public:
+  FunctionAST(std::unique_ptr<PrototypeAST> Proto,
+              std::unique_ptr<ExprAST> Body)
+      : Proto(std::move(Proto)), Body(std::move(Body)) {}
+  const PrototypeAST& getProto() const;
+  const std::string& getName() const;
+  llvm::Function *codegen();
+};
+
+/// This will compile FnAST to IR, rename the function to add the given
+/// suffix (needed to prevent a name-clash with the function's stub),
+/// and then take ownership of the module that the function was compiled
+/// into.
+std::unique_ptr<llvm::Module>
+irgenAndTakeOwnership(FunctionAST &FnAST, const std::string &Suffix);
+
+namespace llvm {
+namespace orc {
+
+// Typedef the remote-client API.
+typedef remote::OrcRemoteTargetClient<FDRPCChannel> MyRemote;
+
+class KaleidoscopeJIT {
+private:
+  MyRemote &Remote;
+  std::unique_ptr<TargetMachine> TM;
+  const DataLayout DL;
+  JITCompileCallbackManager *CompileCallbackMgr;
+  std::unique_ptr<IndirectStubsManager> IndirectStubsMgr;
+  ObjectLinkingLayer<> ObjectLayer;
+  IRCompileLayer<decltype(ObjectLayer)> CompileLayer;
+
+  typedef std::function<std::unique_ptr<Module>(std::unique_ptr<Module>)>
+    OptimizeFunction;
+
+  IRTransformLayer<decltype(CompileLayer), OptimizeFunction> OptimizeLayer;
+
+public:
+  typedef decltype(OptimizeLayer)::ModuleSetHandleT ModuleHandle;
+
+  KaleidoscopeJIT(MyRemote &Remote)
+      : Remote(Remote),
+        TM(EngineBuilder().selectTarget()),
+        DL(TM->createDataLayout()),
+        CompileLayer(ObjectLayer, SimpleCompiler(*TM)),
+        OptimizeLayer(CompileLayer,
+                      [this](std::unique_ptr<Module> M) {
+                        return optimizeModule(std::move(M));
+                      }) {
+    auto CCMgrOrErr = Remote.enableCompileCallbacks(0);
+    if (!CCMgrOrErr) {
+      logAllUnhandledErrors(CCMgrOrErr.takeError(), errs(),
+                            "Error enabling remote compile callbacks:");
+      exit(1);
+    }
+    CompileCallbackMgr = &*CCMgrOrErr;
+    std::unique_ptr<MyRemote::RCIndirectStubsManager> ISM;
+    if (auto Err = Remote.createIndirectStubsManager(ISM)) {
+      logAllUnhandledErrors(std::move(Err), errs(),
+                            "Error creating indirect stubs manager:");
+      exit(1);
+    }
+    IndirectStubsMgr = std::move(ISM);
+    llvm::sys::DynamicLibrary::LoadLibraryPermanently(nullptr);
+  }
+
+  TargetMachine &getTargetMachine() { return *TM; }
+
+  ModuleHandle addModule(std::unique_ptr<Module> M) {
+
+    // Build our symbol resolver:
+    // Lambda 1: Look back into the JIT itself to find symbols that are part of
+    //           the same "logical dylib".
+    // Lambda 2: Search for external symbols in the host process.
+    auto Resolver = createLambdaResolver(
+        [&](const std::string &Name) {
+          if (auto Sym = IndirectStubsMgr->findStub(Name, false))
+            return RuntimeDyld::SymbolInfo(Sym.getAddress(), Sym.getFlags());
+          if (auto Sym = OptimizeLayer.findSymbol(Name, false))
+            return RuntimeDyld::SymbolInfo(Sym.getAddress(), Sym.getFlags());
+          return RuntimeDyld::SymbolInfo(nullptr);
+        },
+        [&](const std::string &Name) {
+          if (auto AddrOrErr = Remote.getSymbolAddress(Name))
+            return RuntimeDyld::SymbolInfo(*AddrOrErr,
+                                           JITSymbolFlags::Exported);
+          else {
+            logAllUnhandledErrors(AddrOrErr.takeError(), errs(),
+                                  "Error resolving remote symbol:");
+            exit(1);
+          }
+          return RuntimeDyld::SymbolInfo(nullptr);
+        });
+
+    std::unique_ptr<MyRemote::RCMemoryManager> MemMgr;
+    if (auto Err = Remote.createRemoteMemoryManager(MemMgr)) {
+      logAllUnhandledErrors(std::move(Err), errs(),
+                            "Error creating remote memory manager:");
+      exit(1);
+    }
+
+    // Build a singlton module set to hold our module.
+    std::vector<std::unique_ptr<Module>> Ms;
+    Ms.push_back(std::move(M));
+
+    // Add the set to the JIT with the resolver we created above and a newly
+    // created SectionMemoryManager.
+    return OptimizeLayer.addModuleSet(std::move(Ms),
+                                      std::move(MemMgr),
+                                      std::move(Resolver));
+  }
+
+  Error addFunctionAST(std::unique_ptr<FunctionAST> FnAST) {
+    // Create a CompileCallback - this is the re-entry point into the compiler
+    // for functions that haven't been compiled yet.
+    auto CCInfo = CompileCallbackMgr->getCompileCallback();
+
+    // Create an indirect stub. This serves as the functions "canonical
+    // definition" - an unchanging (constant address) entry point to the
+    // function implementation.
+    // Initially we point the stub's function-pointer at the compile callback
+    // that we just created. In the compile action for the callback (see below)
+    // we will update the stub's function pointer to point at the function
+    // implementation that we just implemented.
+    if (auto Err = IndirectStubsMgr->createStub(mangle(FnAST->getName()),
+                                                CCInfo.getAddress(),
+                                                JITSymbolFlags::Exported))
+      return Err;
+
+    // Move ownership of FnAST to a shared pointer - C++11 lambdas don't support
+    // capture-by-move, which is be required for unique_ptr.
+    auto SharedFnAST = std::shared_ptr<FunctionAST>(std::move(FnAST));
+
+    // Set the action to compile our AST. This lambda will be run if/when
+    // execution hits the compile callback (via the stub).
+    //
+    // The steps to compile are:
+    // (1) IRGen the function.
+    // (2) Add the IR module to the JIT to make it executable like any other
+    //     module.
+    // (3) Use findSymbol to get the address of the compiled function.
+    // (4) Update the stub pointer to point at the implementation so that
+    ///    subsequent calls go directly to it and bypass the compiler.
+    // (5) Return the address of the implementation: this lambda will actually
+    //     be run inside an attempted call to the function, and we need to
+    //     continue on to the implementation to complete the attempted call.
+    //     The JIT runtime (the resolver block) will use the return address of
+    //     this function as the address to continue at once it has reset the
+    //     CPU state to what it was immediately before the call.
+    CCInfo.setCompileAction(
+      [this, SharedFnAST]() {
+        auto M = irgenAndTakeOwnership(*SharedFnAST, "$impl");
+        addModule(std::move(M));
+        auto Sym = findSymbol(SharedFnAST->getName() + "$impl");
+        assert(Sym && "Couldn't find compiled function?");
+        TargetAddress SymAddr = Sym.getAddress();
+        if (auto Err =
+              IndirectStubsMgr->updatePointer(mangle(SharedFnAST->getName()),
+                                              SymAddr)) {
+          logAllUnhandledErrors(std::move(Err), errs(),
+                                "Error updating function pointer: ");
+          exit(1);
+        }
+
+        return SymAddr;
+      });
+
+    return Error::success();
+  }
+
+  Error executeRemoteExpr(TargetAddress ExprAddr) {
+    return Remote.callVoidVoid(ExprAddr);
+  }
+
+  JITSymbol findSymbol(const std::string Name) {
+    return OptimizeLayer.findSymbol(mangle(Name), true);
+  }
+
+  void removeModule(ModuleHandle H) {
+    OptimizeLayer.removeModuleSet(H);
+  }
+
+private:
+
+  std::string mangle(const std::string &Name) {
+    std::string MangledName;
+    raw_string_ostream MangledNameStream(MangledName);
+    Mangler::getNameWithPrefix(MangledNameStream, Name, DL);
+    return MangledNameStream.str();
+  }
+
+  std::unique_ptr<Module> optimizeModule(std::unique_ptr<Module> M) {
+    // Create a function pass manager.
+    auto FPM = llvm::make_unique<legacy::FunctionPassManager>(M.get());
+
+    // Add some optimizations.
+    FPM->add(createInstructionCombiningPass());
+    FPM->add(createReassociatePass());
+    FPM->add(createGVNPass());
+    FPM->add(createCFGSimplificationPass());
+    FPM->doInitialization();
+
+    // Run the optimizations over all functions in the module being added to
+    // the JIT.
+    for (auto &F : *M)
+      FPM->run(F);
+
+    return M;
+  }
+
+};
+
+} // end namespace orc
+} // end namespace llvm
+
+#endif // LLVM_EXECUTIONENGINE_ORC_KALEIDOSCOPEJIT_H

llvm/examples/Kaleidoscope/BuildingAJIT/Chapter5/RemoteJITUtils.h

@@ -0,0 +1,74 @@
+//===-- RemoteJITUtils.h - Utilities for remote-JITing with LLI -*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Utilities for remote-JITing with LLI.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_TOOLS_LLI_REMOTEJITUTILS_H
+#define LLVM_TOOLS_LLI_REMOTEJITUTILS_H
+
+#include "llvm/ExecutionEngine/Orc/RPCChannel.h"
+#include "llvm/ExecutionEngine/RTDyldMemoryManager.h"
+#include <mutex>
+
+#if !defined(_MSC_VER) && !defined(__MINGW32__)
+#include <unistd.h>
+#else
+#include <io.h>
+#endif
+
+/// RPC channel that reads from and writes from file descriptors.
+class FDRPCChannel final : public llvm::orc::remote::RPCChannel {
+public:
+  FDRPCChannel(int InFD, int OutFD) : InFD(InFD), OutFD(OutFD) {}
+
+  llvm::Error readBytes(char *Dst, unsigned Size) override {
+    assert(Dst && "Attempt to read into null.");
+    ssize_t Completed = 0;
+    while (Completed < static_cast<ssize_t>(Size)) {
+      ssize_t Read = ::read(InFD, Dst + Completed, Size - Completed);
+      if (Read <= 0) {
+        auto ErrNo = errno;
+        if (ErrNo == EAGAIN || ErrNo == EINTR)
+          continue;
+        else
+          return llvm::errorCodeToError(
+                   std::error_code(errno, std::generic_category()));
+      }
+      Completed += Read;
+    }
+    return llvm::Error::success();
+  }
+
+  llvm::Error appendBytes(const char *Src, unsigned Size) override {
+    assert(Src && "Attempt to append from null.");
+    ssize_t Completed = 0;
+    while (Completed < static_cast<ssize_t>(Size)) {
+      ssize_t Written = ::write(OutFD, Src + Completed, Size - Completed);
+      if (Written < 0) {
+        auto ErrNo = errno;
+        if (ErrNo == EAGAIN || ErrNo == EINTR)
+          continue;
+        else
+          return llvm::errorCodeToError(
+                   std::error_code(errno, std::generic_category()));
+      }
+      Completed += Written;
+    }
+    return llvm::Error::success();
+  }
+
+  llvm::Error send() override { return llvm::Error::success(); }
+
+private:
+  int InFD, OutFD;
+};
+
+#endif

llvm/examples/Kaleidoscope/BuildingAJIT/Chapter5/Server/CMakeLists.txt

@@ -0,0 +1,17 @@
+set(LLVM_LINK_COMPONENTS
+  Analysis
+  Core
+  ExecutionEngine
+  InstCombine
+  Object
+  OrcJIT
+  RuntimeDyld
+  ScalarOpts
+  Support
+  TransformUtils
+  native
+  )
+
+add_kaleidoscope_chapter(BuildingAJIT-Ch5-Server
+  server.cpp
+  )

llvm/examples/Kaleidoscope/BuildingAJIT/Chapter5/Server/server.cpp

@@ -0,0 +1,119 @@
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/DynamicLibrary.h"
+#include "llvm/Support/TargetSelect.h"
+#include "llvm/ExecutionEngine/Orc/OrcRemoteTargetServer.h"
+#include "llvm/ExecutionEngine/Orc/OrcABISupport.h"
+
+#include "../RemoteJITUtils.h"
+
+#include <cstring>
+#include <unistd.h>
+#include <netinet/in.h>
+#include <sys/socket.h>
+
+
+using namespace llvm;
+using namespace llvm::orc;
+
+// Command line argument for TCP port.
+cl::opt<uint32_t> Port("port",
+                       cl::desc("TCP port to listen on"),
+                       cl::init(20000));
+
+ExitOnError ExitOnErr;
+
+typedef int (*MainFun)(int, const char*[]);
+
+template <typename NativePtrT>
+NativePtrT MakeNative(uint64_t P) {
+  return reinterpret_cast<NativePtrT>(static_cast<uintptr_t>(P));
+}
+
+extern "C"
+void printExprResult(double Val) {
+  printf("Expression evaluated to: %f\n", Val);
+}
+
+// --- LAZY COMPILE TEST ---
+int main(int argc, char* argv[]) {
+
+  if (argc == 0)
+    ExitOnErr.setBanner("jit_server: ");
+  else
+    ExitOnErr.setBanner(std::string(argv[0]) + ": ");
+
+  // --- Initialize LLVM ---
+  cl::ParseCommandLineOptions(argc, argv, "LLVM lazy JIT example.\n");
+
+  InitializeNativeTarget();
+  InitializeNativeTargetAsmPrinter();
+  InitializeNativeTargetAsmParser();
+
+  if (sys::DynamicLibrary::LoadLibraryPermanently(nullptr)) {
+    errs() << "Error loading program symbols.\n";
+    return 1;
+  }
+
+  // --- Initialize remote connection ---
+
+  int sockfd = socket(PF_INET, SOCK_STREAM, 0);
+  sockaddr_in servAddr, clientAddr;
+  socklen_t clientAddrLen = sizeof(clientAddr);
+  bzero(&servAddr, sizeof(servAddr));
+  servAddr.sin_family = PF_INET;
+  servAddr.sin_family = INADDR_ANY;
+  servAddr.sin_port = htons(Port);
+
+  {
+    // avoid "Address already in use" error.
+    int yes=1;
+    if (setsockopt(sockfd,SOL_SOCKET,SO_REUSEADDR,&yes,sizeof(int)) == -1) {
+      errs() << "Error calling setsockopt.\n";
+      return 1;
+    }
+  }
+
+  if (bind(sockfd, reinterpret_cast<sockaddr*>(&servAddr),
+           sizeof(servAddr)) < 0) {
+    errs() << "Error on binding.\n";
+    return 1;
+  }
+  listen(sockfd, 1);
+  int newsockfd = accept(sockfd, reinterpret_cast<sockaddr*>(&clientAddr),
+                         &clientAddrLen);
+
+  auto SymbolLookup =
+    [](const std::string &Name) {
+      return RTDyldMemoryManager::getSymbolAddressInProcess(Name);
+    };
+
+  auto RegisterEHFrames =
+    [](uint8_t *Addr, uint32_t Size) {
+      RTDyldMemoryManager::registerEHFramesInProcess(Addr, Size);
+    };
+
+  auto DeregisterEHFrames =
+    [](uint8_t *Addr, uint32_t Size) {
+      RTDyldMemoryManager::deregisterEHFramesInProcess(Addr, Size);
+    };
+
+  FDRPCChannel TCPChannel(newsockfd, newsockfd);
+  typedef remote::OrcRemoteTargetServer<FDRPCChannel, OrcX86_64_SysV> MyServerT;
+
+  MyServerT Server(TCPChannel, SymbolLookup, RegisterEHFrames, DeregisterEHFrames);
+
+  while (1) {
+    MyServerT::JITFuncId Id = MyServerT::InvalidId;
+    ExitOnErr(Server.startReceivingFunction(TCPChannel, (uint32_t&)Id));
+    switch (Id) {
+    case MyServerT::TerminateSessionId:
+      ExitOnErr(Server.handleTerminateSession());
+      return 0;
+    default:
+      ExitOnErr(Server.handleKnownFunction(Id));
+      break;
+    }
+  }
+
+  llvm_unreachable("Fell through server command loop.");
+}