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/ModuleProvider.h"
  25 #include "llvm/PassManager.h"
  26 #include "llvm/ADT/StringExtras.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/Mangler.h"
  35 #include "llvm/Support/MemoryBuffer.h"
  36 #include "llvm/Support/StandardPasses.h"
  37 #include "llvm/Support/SystemUtils.h"
  38 #include "llvm/System/Host.h"
  39 #include "llvm/System/Program.h"
  40 #include "llvm/System/Signals.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     std::string targetTriple = _linker.getModule()->getTargetTriple();
 246     args.push_back(tool.c_str());
 247     if ( targetTriple.find("darwin") != std::string::npos ) {
 248         // darwin specific command line options
 249         if (strncmp(targetTriple.c_str(), "i386-apple-", 11) == 0) {
 250             args.push_back("-arch");
 251             args.push_back("i386");
 252         }
 253         else if (strncmp(targetTriple.c_str(), "x86_64-apple-", 13) == 0) {
 254             args.push_back("-arch");
 255             args.push_back("x86_64");
 256         }
 257         else if (strncmp(targetTriple.c_str(), "powerpc-apple-", 14) == 0) {
 258             args.push_back("-arch");
 259             args.push_back("ppc");
 260         }
 261         else if (strncmp(targetTriple.c_str(), "powerpc64-apple-", 16) == 0) {
 262             args.push_back("-arch");
 263             args.push_back("ppc64");
 264         }
 265         else if (strncmp(targetTriple.c_str(), "arm-apple-", 10) == 0) {
 266             args.push_back("-arch");
 267             args.push_back("arm");
 268         }
 269         else if ((strncmp(targetTriple.c_str(), "armv4t-apple-", 13) == 0) ||
 270                  (strncmp(targetTriple.c_str(), "thumbv4t-apple-", 15) == 0)) {
 271             args.push_back("-arch");
 272             args.push_back("armv4t");
 273         }
 274         else if ((strncmp(targetTriple.c_str(), "armv5-apple-", 12) == 0) ||
 275                  (strncmp(targetTriple.c_str(), "armv5e-apple-", 13) == 0) ||
 276                  (strncmp(targetTriple.c_str(), "thumbv5-apple-", 14) == 0) ||
 277                  (strncmp(targetTriple.c_str(), "thumbv5e-apple-", 15) == 0)) {
 278             args.push_back("-arch");
 279             args.push_back("armv5");
 280         }
 281         else if ((strncmp(targetTriple.c_str(), "armv6-apple-", 12) == 0) ||
 282                  (strncmp(targetTriple.c_str(), "thumbv6-apple-", 14) == 0)) {
 283             args.push_back("-arch");
 284             args.push_back("armv6");
 285         }
 286         else if ((strncmp(targetTriple.c_str(), "armv7-apple-", 12) == 0) ||
 287                  (strncmp(targetTriple.c_str(), "thumbv7-apple-", 14) == 0)) {
 288             args.push_back("-arch");
 289             args.push_back("armv7");
 290         }
 291         // add -static to assembler command line when code model requires
 292         if ( (_assemblerPath != NULL) && (_codeModel == LTO_CODEGEN_PIC_MODEL_STATIC) )
 293             args.push_back("-static");
 294     }
 295     if ( needsCompilerOptions ) {
 296         args.push_back("-c");
 297         args.push_back("-x");
 298         args.push_back("assembler");
 299     }
 300     args.push_back("-o");
 301     args.push_back(objPath.c_str());
 302     args.push_back(asmPath.c_str());
 303     args.push_back(0);
 304
 305     // invoke assembler
 306     if ( sys::Program::ExecuteAndWait(tool, &args[0], 0, 0, 0, 0, &errMsg) ) {
 307         errMsg = "error in assembly";
 308         return true;
 309     }
 310     return false; // success
 311 }
 312
 313
 314
 315 bool LTOCodeGenerator::determineTarget(std::string& errMsg)
 316 {
 317     if ( _target == NULL ) {
 318         std::string Triple = _linker.getModule()->getTargetTriple();
 319         if (Triple.empty())
 320           Triple = sys::getHostTriple();
 321
 322         // create target machine from info for merged modules
 323         const Target *march = TargetRegistry::lookupTarget(Triple, errMsg);
 324         if ( march == NULL )
 325             return true;
 326
 327         // The relocation model is actually a static member of TargetMachine
 328         // and needs to be set before the TargetMachine is instantiated.
 329         switch( _codeModel ) {
 330         case LTO_CODEGEN_PIC_MODEL_STATIC:
 331             TargetMachine::setRelocationModel(Reloc::Static);
 332             break;
 333         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 334             TargetMachine::setRelocationModel(Reloc::PIC_);
 335             break;
 336         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 337             TargetMachine::setRelocationModel(Reloc::DynamicNoPIC);
 338             break;
 339         }
 340
 341         // construct LTModule, hand over ownership of module and target
 342         std::string FeatureStr = getFeatureString(Triple.c_str());
 343         _target = march->createTargetMachine(Triple, FeatureStr);
 344     }
 345     return false;
 346 }
 347
 348 void LTOCodeGenerator::applyScopeRestrictions()
 349 {
 350     if ( !_scopeRestrictionsDone ) {
 351         Module* mergedModule = _linker.getModule();
 352
 353         // Start off with a verification pass.
 354         PassManager passes;
 355         passes.add(createVerifierPass());
 356
 357         // mark which symbols can not be internalized
 358         if ( !_mustPreserveSymbols.empty() ) {
 359             Mangler mangler(*mergedModule,
 360                                 _target->getMCAsmInfo()->getGlobalPrefix());
 361             std::vector<const char*> mustPreserveList;
 362             for (Module::iterator f = mergedModule->begin(),
 363                                         e = mergedModule->end(); f != e; ++f) {
 364                 if ( !f->isDeclaration()
 365                   && _mustPreserveSymbols.count(mangler.getMangledName(f)) )
 366                   mustPreserveList.push_back(::strdup(f->getNameStr().c_str()));
 367             }
 368             for (Module::global_iterator v = mergedModule->global_begin(),
 369                                  e = mergedModule->global_end(); v !=  e; ++v) {
 370                 if ( !v->isDeclaration()
 371                   && _mustPreserveSymbols.count(mangler.getMangledName(v)) )
 372                   mustPreserveList.push_back(::strdup(v->getNameStr().c_str()));
 373             }
 374             passes.add(createInternalizePass(mustPreserveList));
 375         }
 376         // apply scope restrictions
 377         passes.run(*mergedModule);
 378
 379         _scopeRestrictionsDone = true;
 380     }
 381 }
 382
 383 /// Optimize merged modules using various IPO passes
 384 bool LTOCodeGenerator::generateAssemblyCode(formatted_raw_ostream& out,
 385                                             std::string& errMsg)
 386 {
 387     if ( this->determineTarget(errMsg) )
 388         return true;
 389
 390     // mark which symbols can not be internalized
 391     this->applyScopeRestrictions();
 392
 393     Module* mergedModule = _linker.getModule();
 394
 395     // If target supports exception handling then enable it now.
 396     switch (_target->getMCAsmInfo()->getExceptionHandlingType()) {
 397     case ExceptionHandling::Dwarf:
 398       llvm::DwarfExceptionHandling = true;
 399       break;
 400     case ExceptionHandling::SjLj:
 401       llvm::SjLjExceptionHandling = true;
 402       break;
 403     case ExceptionHandling::None:
 404       break;
 405     default:
 406       assert (0 && "Unknown exception handling model!");
 407     }
 408
 409     // if options were requested, set them
 410     if ( !_codegenOptions.empty() )
 411         cl::ParseCommandLineOptions(_codegenOptions.size(),
 412                                                 (char**)&_codegenOptions[0]);
 413
 414     // Instantiate the pass manager to organize the passes.
 415     PassManager passes;
 416
 417     // Start off with a verification pass.
 418     passes.add(createVerifierPass());
 419
 420     // Add an appropriate TargetData instance for this module...
 421     passes.add(new TargetData(*_target->getTargetData()));
 422
 423     createStandardLTOPasses(&passes, /*Internalize=*/ false, !DisableInline,
 424                             /*VerifyEach=*/ false);
 425
 426     // Make sure everything is still good.
 427     passes.add(createVerifierPass());
 428
 429     FunctionPassManager* codeGenPasses =
 430             new FunctionPassManager(new ExistingModuleProvider(mergedModule));
 431
 432     codeGenPasses->add(new TargetData(*_target->getTargetData()));
 433
 434     ObjectCodeEmitter* oce = NULL;
 435
 436     switch (_target->addPassesToEmitFile(*codeGenPasses, out,
 437                                          TargetMachine::AssemblyFile,
 438                                          CodeGenOpt::Aggressive)) {
 439         case FileModel::MachOFile:
 440             oce = AddMachOWriter(*codeGenPasses, out, *_target);
 441             break;
 442         case FileModel::ElfFile:
 443             oce = AddELFWriter(*codeGenPasses, out, *_target);
 444             break;
 445         case FileModel::AsmFile:
 446             break;
 447         case FileModel::Error:
 448         case FileModel::None:
 449             errMsg = "target file type not supported";
 450             return true;
 451     }
 452
 453     if (_target->addPassesToEmitFileFinish(*codeGenPasses, oce,
 454                                            CodeGenOpt::Aggressive)) {
 455         errMsg = "target does not support generation of this file type";
 456         return true;
 457     }
 458
 459     // Run our queue of passes all at once now, efficiently.
 460     passes.run(*mergedModule);
 461
 462     // Run the code generator, and write assembly file
 463     codeGenPasses->doInitialization();
 464
 465     for (Module::iterator
 466            it = mergedModule->begin(), e = mergedModule->end(); it != e; ++it)
 467       if (!it->isDeclaration())
 468         codeGenPasses->run(*it);
 469
 470     codeGenPasses->doFinalization();
 471
 472     return false; // success
 473 }
 474
 475
 476 /// Optimize merged modules using various IPO passes
 477 void LTOCodeGenerator::setCodeGenDebugOptions(const char* options)
 478 {
 479     std::string ops(options);
 480     for (std::string o = getToken(ops); !o.empty(); o = getToken(ops)) {
 481         // ParseCommandLineOptions() expects argv[0] to be program name.
 482         // Lazily add that.
 483         if ( _codegenOptions.empty() )
 484             _codegenOptions.push_back("libLTO");
 485         _codegenOptions.push_back(strdup(o.c_str()));
 486     }
 487 }