tools/lto/LTOCodeGenerator.cpp

   1 //===-LTOCodeGenerator.cpp - LLVM Link Time Optimizer ---------------------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // This file implements the Link Time Optimization library. This library is
  11 // intended to be used by linker to optimize code at link time.
  12 //
  13 //===----------------------------------------------------------------------===//
  14
  15 #include "LTOModule.h"
  16 #include "LTOCodeGenerator.h"
  17
  18
  19 #include "llvm/Constants.h"
  20 #include "llvm/DerivedTypes.h"
  21 #include "llvm/Linker.h"
  22 #include "llvm/LLVMContext.h"
  23 #include "llvm/Module.h"
  24 #include "llvm/PassManager.h"
  25 #include "llvm/ADT/StringExtras.h"
  26 #include "llvm/ADT/Triple.h"
  27 #include "llvm/Analysis/Passes.h"
  28 #include "llvm/Analysis/LoopPass.h"
  29 #include "llvm/Analysis/Verifier.h"
  30 #include "llvm/Bitcode/ReaderWriter.h"
  31 #include "llvm/CodeGen/FileWriters.h"
  32 #include "llvm/Support/CommandLine.h"
  33 #include "llvm/Support/FormattedStream.h"
  34 #include "llvm/Support/MemoryBuffer.h"
  35 #include "llvm/Support/StandardPasses.h"
  36 #include "llvm/Support/SystemUtils.h"
  37 #include "llvm/System/Host.h"
  38 #include "llvm/System/Program.h"
  39 #include "llvm/System/Signals.h"
  40 #include "llvm/Target/Mangler.h"
  41 #include "llvm/Target/SubtargetFeature.h"
  42 #include "llvm/Target/TargetOptions.h"
  43 #include "llvm/MC/MCAsmInfo.h"
  44 #include "llvm/Target/TargetData.h"
  45 #include "llvm/Target/TargetMachine.h"
  46 #include "llvm/Target/TargetRegistry.h"
  47 #include "llvm/Target/TargetSelect.h"
  48 #include "llvm/Transforms/IPO.h"
  49 #include "llvm/Transforms/Scalar.h"
  50 #include "llvm/Config/config.h"
  51 #include <cstdlib>
  52 #include <unistd.h>
  53 #include <fcntl.h>
  54
  55
  56 using namespace llvm;
  57
  58 static cl::opt<bool> DisableInline("disable-inlining",
  59   cl::desc("Do not run the inliner pass"));
  60
  61
  62 const char* LTOCodeGenerator::getVersionString()
  63 {
  64 #ifdef LLVM_VERSION_INFO
  65     return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
  66 #else
  67     return PACKAGE_NAME " version " PACKAGE_VERSION;
  68 #endif
  69 }
  70
  71
  72 LTOCodeGenerator::LTOCodeGenerator()
  73     : _context(getGlobalContext()),
  74       _linker("LinkTimeOptimizer", "ld-temp.o", _context), _target(NULL),
  75       _emitDwarfDebugInfo(false), _scopeRestrictionsDone(false),
  76       _codeModel(LTO_CODEGEN_PIC_MODEL_DYNAMIC),
  77       _nativeObjectFile(NULL), _assemblerPath(NULL)
  78 {
  79     InitializeAllTargets();
  80     InitializeAllAsmPrinters();
  81 }
  82
  83 LTOCodeGenerator::~LTOCodeGenerator()
  84 {
  85     delete _target;
  86     delete _nativeObjectFile;
  87 }
  88
  89
  90
  91 bool LTOCodeGenerator::addModule(LTOModule* mod, std::string& errMsg)
  92 {
  93     return _linker.LinkInModule(mod->getLLVVMModule(), &errMsg);
  94 }
  95
  96
  97 bool LTOCodeGenerator::setDebugInfo(lto_debug_model debug, std::string& errMsg)
  98 {
  99     switch (debug) {
 100         case LTO_DEBUG_MODEL_NONE:
 101             _emitDwarfDebugInfo = false;
 102             return false;
 103
 104         case LTO_DEBUG_MODEL_DWARF:
 105             _emitDwarfDebugInfo = true;
 106             return false;
 107     }
 108     errMsg = "unknown debug format";
 109     return true;
 110 }
 111
 112
 113 bool LTOCodeGenerator::setCodePICModel(lto_codegen_model model,
 114                                        std::string& errMsg)
 115 {
 116     switch (model) {
 117         case LTO_CODEGEN_PIC_MODEL_STATIC:
 118         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 119         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 120             _codeModel = model;
 121             return false;
 122     }
 123     errMsg = "unknown pic model";
 124     return true;
 125 }
 126
 127 void LTOCodeGenerator::setAssemblerPath(const char* path)
 128 {
 129     if ( _assemblerPath )
 130         delete _assemblerPath;
 131     _assemblerPath = new sys::Path(path);
 132 }
 133
 134 void LTOCodeGenerator::addMustPreserveSymbol(const char* sym)
 135 {
 136     _mustPreserveSymbols[sym] = 1;
 137 }
 138
 139
 140 bool LTOCodeGenerator::writeMergedModules(const char *path,
 141                                           std::string &errMsg) {
 142   if (determineTarget(errMsg))
 143     return true;
 144
 145   // mark which symbols can not be internalized
 146   applyScopeRestrictions();
 147
 148   // create output file
 149   std::string ErrInfo;
 150   raw_fd_ostream Out(path, ErrInfo,
 151                      raw_fd_ostream::F_Binary);
 152   if (!ErrInfo.empty()) {
 153     errMsg = "could not open bitcode file for writing: ";
 154     errMsg += path;
 155     return true;
 156   }
 157
 158   // write bitcode to it
 159   WriteBitcodeToFile(_linker.getModule(), Out);
 160
 161   if (Out.has_error()) {
 162     errMsg = "could not write bitcode file: ";
 163     errMsg += path;
 164     return true;
 165   }
 166
 167   return false;
 168 }
 169
 170
 171 const void* LTOCodeGenerator::compile(size_t* length, std::string& errMsg)
 172 {
 173     // make unique temp .s file to put generated assembly code
 174     sys::Path uniqueAsmPath("lto-llvm.s");
 175     if ( uniqueAsmPath.createTemporaryFileOnDisk(true, &errMsg) )
 176         return NULL;
 177     sys::RemoveFileOnSignal(uniqueAsmPath);
 178
 179     // generate assembly code
 180     bool genResult = false;
 181     {
 182       raw_fd_ostream asmFD(uniqueAsmPath.c_str(), errMsg);
 183       formatted_raw_ostream asmFile(asmFD);
 184       if (!errMsg.empty())
 185         return NULL;
 186       genResult = this->generateAssemblyCode(asmFile, errMsg);
 187     }
 188     if ( genResult ) {
 189         if ( uniqueAsmPath.exists() )
 190             uniqueAsmPath.eraseFromDisk();
 191         return NULL;
 192     }
 193
 194     // make unique temp .o file to put generated object file
 195     sys::PathWithStatus uniqueObjPath("lto-llvm.o");
 196     if ( uniqueObjPath.createTemporaryFileOnDisk(true, &errMsg) ) {
 197         if ( uniqueAsmPath.exists() )
 198             uniqueAsmPath.eraseFromDisk();
 199         return NULL;
 200     }
 201     sys::RemoveFileOnSignal(uniqueObjPath);
 202
 203     // assemble the assembly code
 204     const std::string& uniqueObjStr = uniqueObjPath.str();
 205     bool asmResult = this->assemble(uniqueAsmPath.str(), uniqueObjStr, errMsg);
 206     if ( !asmResult ) {
 207         // remove old buffer if compile() called twice
 208         delete _nativeObjectFile;
 209
 210         // read .o file into memory buffer
 211         _nativeObjectFile = MemoryBuffer::getFile(uniqueObjStr.c_str(),&errMsg);
 212     }
 213
 214     // remove temp files
 215     uniqueAsmPath.eraseFromDisk();
 216     uniqueObjPath.eraseFromDisk();
 217
 218     // return buffer, unless error
 219     if ( _nativeObjectFile == NULL )
 220         return NULL;
 221     *length = _nativeObjectFile->getBufferSize();
 222     return _nativeObjectFile->getBufferStart();
 223 }
 224
 225
 226 bool LTOCodeGenerator::assemble(const std::string& asmPath,
 227                                 const std::string& objPath, std::string& errMsg)
 228 {
 229     sys::Path tool;
 230     bool needsCompilerOptions = true;
 231     if ( _assemblerPath ) {
 232         tool = *_assemblerPath;
 233         needsCompilerOptions = false;
 234     } else {
 235         // find compiler driver
 236         tool = sys::Program::FindProgramByName("gcc");
 237         if ( tool.isEmpty() ) {
 238             errMsg = "can't locate gcc";
 239             return true;
 240         }
 241     }
 242
 243     // build argument list
 244     std::vector<const char*> args;
 245     llvm::Triple targetTriple(_linker.getModule()->getTargetTriple());
 246     const char *arch = targetTriple.getArchNameForAssembler();
 247
 248     args.push_back(tool.c_str());
 249
 250     if (targetTriple.getOS() == Triple::Darwin) {
 251         // darwin specific command line options
 252         if (arch != NULL) {
 253             args.push_back("-arch");
 254             args.push_back(arch);
 255         }
 256         // add -static to assembler command line when code model requires
 257         if ( (_assemblerPath != NULL) && (_codeModel == LTO_CODEGEN_PIC_MODEL_STATIC) )
 258             args.push_back("-static");
 259     }
 260     if ( needsCompilerOptions ) {
 261         args.push_back("-c");
 262         args.push_back("-x");
 263         args.push_back("assembler");
 264     }
 265     args.push_back("-o");
 266     args.push_back(objPath.c_str());
 267     args.push_back(asmPath.c_str());
 268     args.push_back(0);
 269
 270     // invoke assembler
 271     if ( sys::Program::ExecuteAndWait(tool, &args[0], 0, 0, 0, 0, &errMsg) ) {
 272         errMsg = "error in assembly";
 273         return true;
 274     }
 275     return false; // success
 276 }
 277
 278
 279
 280 bool LTOCodeGenerator::determineTarget(std::string& errMsg)
 281 {
 282     if ( _target == NULL ) {
 283         std::string Triple = _linker.getModule()->getTargetTriple();
 284         if (Triple.empty())
 285           Triple = sys::getHostTriple();
 286
 287         // create target machine from info for merged modules
 288         const Target *march = TargetRegistry::lookupTarget(Triple, errMsg);
 289         if ( march == NULL )
 290             return true;
 291
 292         // The relocation model is actually a static member of TargetMachine
 293         // and needs to be set before the TargetMachine is instantiated.
 294         switch( _codeModel ) {
 295         case LTO_CODEGEN_PIC_MODEL_STATIC:
 296             TargetMachine::setRelocationModel(Reloc::Static);
 297             break;
 298         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 299             TargetMachine::setRelocationModel(Reloc::PIC_);
 300             break;
 301         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 302             TargetMachine::setRelocationModel(Reloc::DynamicNoPIC);
 303             break;
 304         }
 305
 306         // construct LTModule, hand over ownership of module and target
 307         const std::string FeatureStr =
 308             SubtargetFeatures::getDefaultSubtargetFeatures(llvm::Triple(Triple));
 309         _target = march->createTargetMachine(Triple, FeatureStr);
 310     }
 311     return false;
 312 }
 313
 314 void LTOCodeGenerator::applyScopeRestrictions()
 315 {
 316     if ( !_scopeRestrictionsDone ) {
 317         Module* mergedModule = _linker.getModule();
 318
 319         // Start off with a verification pass.
 320         PassManager passes;
 321         passes.add(createVerifierPass());
 322
 323         // mark which symbols can not be internalized
 324         if ( !_mustPreserveSymbols.empty() ) {
 325             Mangler mangler(*_target->getMCAsmInfo());
 326             std::vector<const char*> mustPreserveList;
 327             for (Module::iterator f = mergedModule->begin(),
 328                                         e = mergedModule->end(); f != e; ++f) {
 329                 if ( !f->isDeclaration()
 330                   && _mustPreserveSymbols.count(mangler.getNameWithPrefix(f)) )
 331                   mustPreserveList.push_back(::strdup(f->getNameStr().c_str()));
 332             }
 333             for (Module::global_iterator v = mergedModule->global_begin(),
 334                                  e = mergedModule->global_end(); v !=  e; ++v) {
 335                 if ( !v->isDeclaration()
 336                   && _mustPreserveSymbols.count(mangler.getNameWithPrefix(v)) )
 337                   mustPreserveList.push_back(::strdup(v->getNameStr().c_str()));
 338             }
 339             passes.add(createInternalizePass(mustPreserveList));
 340         }
 341         // apply scope restrictions
 342         passes.run(*mergedModule);
 343
 344         _scopeRestrictionsDone = true;
 345     }
 346 }
 347
 348 /// Optimize merged modules using various IPO passes
 349 bool LTOCodeGenerator::generateAssemblyCode(formatted_raw_ostream& out,
 350                                             std::string& errMsg)
 351 {
 352     if ( this->determineTarget(errMsg) )
 353         return true;
 354
 355     // mark which symbols can not be internalized
 356     this->applyScopeRestrictions();
 357
 358     Module* mergedModule = _linker.getModule();
 359
 360     // If target supports exception handling then enable it now.
 361     switch (_target->getMCAsmInfo()->getExceptionHandlingType()) {
 362     case ExceptionHandling::Dwarf:
 363       llvm::DwarfExceptionHandling = true;
 364       break;
 365     case ExceptionHandling::SjLj:
 366       llvm::SjLjExceptionHandling = true;
 367       break;
 368     case ExceptionHandling::None:
 369       break;
 370     default:
 371       assert (0 && "Unknown exception handling model!");
 372     }
 373
 374     // if options were requested, set them
 375     if ( !_codegenOptions.empty() )
 376         cl::ParseCommandLineOptions(_codegenOptions.size(),
 377                                                 (char**)&_codegenOptions[0]);
 378
 379     // Instantiate the pass manager to organize the passes.
 380     PassManager passes;
 381
 382     // Start off with a verification pass.
 383     passes.add(createVerifierPass());
 384
 385     // Add an appropriate TargetData instance for this module...
 386     passes.add(new TargetData(*_target->getTargetData()));
 387
 388     createStandardLTOPasses(&passes, /*Internalize=*/ false, !DisableInline,
 389                             /*VerifyEach=*/ false);
 390
 391     // Make sure everything is still good.
 392     passes.add(createVerifierPass());
 393
 394     FunctionPassManager* codeGenPasses = new FunctionPassManager(mergedModule);
 395
 396     codeGenPasses->add(new TargetData(*_target->getTargetData()));
 397
 398     ObjectCodeEmitter* oce = NULL;
 399
 400     switch (_target->addPassesToEmitFile(*codeGenPasses, out,
 401                                          TargetMachine::AssemblyFile,
 402                                          CodeGenOpt::Aggressive)) {
 403         case FileModel::ElfFile:
 404             oce = AddELFWriter(*codeGenPasses, out, *_target);
 405             break;
 406         case FileModel::AsmFile:
 407             break;
 408         case FileModel::MachOFile:
 409         case FileModel::Error:
 410         case FileModel::None:
 411             errMsg = "target file type not supported";
 412             return true;
 413     }
 414
 415     if (_target->addPassesToEmitFileFinish(*codeGenPasses, oce,
 416                                            CodeGenOpt::Aggressive)) {
 417         errMsg = "target does not support generation of this file type";
 418         return true;
 419     }
 420
 421     // Run our queue of passes all at once now, efficiently.
 422     passes.run(*mergedModule);
 423
 424     // Run the code generator, and write assembly file
 425     codeGenPasses->doInitialization();
 426
 427     for (Module::iterator
 428            it = mergedModule->begin(), e = mergedModule->end(); it != e; ++it)
 429       if (!it->isDeclaration())
 430         codeGenPasses->run(*it);
 431
 432     codeGenPasses->doFinalization();
 433
 434     return false; // success
 435 }
 436
 437
 438 /// Optimize merged modules using various IPO passes
 439 void LTOCodeGenerator::setCodeGenDebugOptions(const char* options)
 440 {
 441     for (std::pair<StringRef, StringRef> o = getToken(options);
 442          !o.first.empty(); o = getToken(o.second)) {
 443         // ParseCommandLineOptions() expects argv[0] to be program name.
 444         // Lazily add that.
 445         if ( _codegenOptions.empty() )
 446             _codegenOptions.push_back("libLTO");
 447         _codegenOptions.push_back(strdup(o.first.str().c_str()));
 448     }
 449 }