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/Mangler.h"
  34 #include "llvm/Support/MemoryBuffer.h"
  35 #include "llvm/Support/StandardPasses.h"
  36 #include "llvm/Support/SystemUtils.h"
  37 #include "llvm/Support/raw_ostream.h"
  38 #include "llvm/System/Signals.h"
  39 #include "llvm/Target/SubtargetFeature.h"
  40 #include "llvm/Target/TargetOptions.h"
  41 #include "llvm/Target/TargetAsmInfo.h"
  42 #include "llvm/Target/TargetData.h"
  43 #include "llvm/Target/TargetMachine.h"
  44 #include "llvm/Target/TargetMachineRegistry.h"
  45 #include "llvm/Target/TargetSelect.h"
  46 #include "llvm/Transforms/IPO.h"
  47 #include "llvm/Transforms/Scalar.h"
  48 #include "llvm/Config/config.h"
  49
  50
  51 #include <cstdlib>
  52 #include <fstream>
  53 #include <unistd.h>
  54 #include <fcntl.h>
  55
  56
  57 using namespace llvm;
  58
  59 static cl::opt<bool> DisableInline("disable-inlining",
  60   cl::desc("Do not run the inliner pass"));
  61
  62
  63 const char* LTOCodeGenerator::getVersionString()
  64 {
  65 #ifdef LLVM_VERSION_INFO
  66     return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
  67 #else
  68     return PACKAGE_NAME " version " PACKAGE_VERSION;
  69 #endif
  70 }
  71
  72
  73 LTOCodeGenerator::LTOCodeGenerator()
  74     : _context(getGlobalContext()),
  75       _linker("LinkTimeOptimizer", "ld-temp.o", _context), _target(NULL),
  76       _emitDwarfDebugInfo(false), _scopeRestrictionsDone(false),
  77       _codeModel(LTO_CODEGEN_PIC_MODEL_DYNAMIC),
  78       _nativeObjectFile(NULL), _gccPath(NULL), _assemblerPath(NULL)
  79 {
  80   InitializeAllTargets();
  81   InitializeAllAsmPrinters();
  82
  83 }
  84
  85 LTOCodeGenerator::~LTOCodeGenerator()
  86 {
  87     delete _target;
  88     delete _nativeObjectFile;
  89 }
  90
  91
  92
  93 bool LTOCodeGenerator::addModule(LTOModule* mod, std::string& errMsg)
  94 {
  95     return _linker.LinkInModule(mod->getLLVVMModule(), &errMsg);
  96 }
  97
  98
  99 bool LTOCodeGenerator::setDebugInfo(lto_debug_model debug, std::string& errMsg)
 100 {
 101     switch (debug) {
 102         case LTO_DEBUG_MODEL_NONE:
 103             _emitDwarfDebugInfo = false;
 104             return false;
 105
 106         case LTO_DEBUG_MODEL_DWARF:
 107             _emitDwarfDebugInfo = true;
 108             return false;
 109     }
 110     errMsg = "unknown debug format";
 111     return true;
 112 }
 113
 114
 115 bool LTOCodeGenerator::setCodePICModel(lto_codegen_model model,
 116                                        std::string& errMsg)
 117 {
 118     switch (model) {
 119         case LTO_CODEGEN_PIC_MODEL_STATIC:
 120         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 121         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 122             _codeModel = model;
 123             return false;
 124     }
 125     errMsg = "unknown pic model";
 126     return true;
 127 }
 128
 129 void LTOCodeGenerator::setGccPath(const char* path)
 130 {
 131     if ( _gccPath )
 132         delete _gccPath;
 133     _gccPath = new sys::Path(path);
 134 }
 135
 136 void LTOCodeGenerator::setAssemblerPath(const char* path)
 137 {
 138     if ( _assemblerPath )
 139         delete _assemblerPath;
 140     _assemblerPath = new sys::Path(path);
 141 }
 142
 143 void LTOCodeGenerator::addMustPreserveSymbol(const char* sym)
 144 {
 145     _mustPreserveSymbols[sym] = 1;
 146 }
 147
 148
 149 bool LTOCodeGenerator::writeMergedModules(const char* path, std::string& errMsg)
 150 {
 151     if ( this->determineTarget(errMsg) )
 152         return true;
 153
 154     // mark which symbols can not be internalized
 155     this->applyScopeRestrictions();
 156
 157     // create output file
 158     std::ofstream out(path, std::ios_base::out|std::ios::trunc|std::ios::binary);
 159     if ( out.fail() ) {
 160         errMsg = "could not open bitcode file for writing: ";
 161         errMsg += path;
 162         return true;
 163     }
 164
 165     // write bitcode to it
 166     WriteBitcodeToFile(_linker.getModule(), out);
 167     if ( out.fail() ) {
 168         errMsg = "could not write bitcode file: ";
 169         errMsg += path;
 170         return true;
 171     }
 172
 173     return false;
 174 }
 175
 176
 177 const void* LTOCodeGenerator::compile(size_t* length, std::string& errMsg)
 178 {
 179     // make unique temp .s file to put generated assembly code
 180     sys::Path uniqueAsmPath("lto-llvm.s");
 181     if ( uniqueAsmPath.createTemporaryFileOnDisk(true, &errMsg) )
 182         return NULL;
 183     sys::RemoveFileOnSignal(uniqueAsmPath);
 184
 185     // generate assembly code
 186     bool genResult = false;
 187     {
 188       raw_fd_ostream asmFile(uniqueAsmPath.c_str(), false, errMsg);
 189       if (!errMsg.empty())
 190         return NULL;
 191       genResult = this->generateAssemblyCode(asmFile, errMsg);
 192     }
 193     if ( genResult ) {
 194         if ( uniqueAsmPath.exists() )
 195             uniqueAsmPath.eraseFromDisk();
 196         return NULL;
 197     }
 198
 199     // make unique temp .o file to put generated object file
 200     sys::PathWithStatus uniqueObjPath("lto-llvm.o");
 201     if ( uniqueObjPath.createTemporaryFileOnDisk(true, &errMsg) ) {
 202         if ( uniqueAsmPath.exists() )
 203             uniqueAsmPath.eraseFromDisk();
 204         return NULL;
 205     }
 206     sys::RemoveFileOnSignal(uniqueObjPath);
 207
 208     // assemble the assembly code
 209     const std::string& uniqueObjStr = uniqueObjPath.toString();
 210     bool asmResult = this->assemble(uniqueAsmPath.toString(),
 211                                                         uniqueObjStr, errMsg);
 212     if ( !asmResult ) {
 213         // remove old buffer if compile() called twice
 214         delete _nativeObjectFile;
 215
 216         // read .o file into memory buffer
 217         _nativeObjectFile = MemoryBuffer::getFile(uniqueObjStr.c_str(),&errMsg);
 218     }
 219
 220     // remove temp files
 221     uniqueAsmPath.eraseFromDisk();
 222     uniqueObjPath.eraseFromDisk();
 223
 224     // return buffer, unless error
 225     if ( _nativeObjectFile == NULL )
 226         return NULL;
 227     *length = _nativeObjectFile->getBufferSize();
 228     return _nativeObjectFile->getBufferStart();
 229 }
 230
 231
 232 bool LTOCodeGenerator::assemble(const std::string& asmPath,
 233                                 const std::string& objPath, std::string& errMsg)
 234 {
 235     sys::Path tool;
 236     bool needsCompilerOptions = true;
 237     if ( _assemblerPath ) {
 238         tool = *_assemblerPath;
 239         needsCompilerOptions = false;
 240     }
 241     else if ( _gccPath ) {
 242         tool = *_gccPath;
 243     } else {
 244         // find compiler driver
 245         tool = sys::Program::FindProgramByName("gcc");
 246         if ( tool.isEmpty() ) {
 247             errMsg = "can't locate gcc";
 248             return true;
 249         }
 250     }
 251
 252     // build argument list
 253     std::vector<const char*> args;
 254     std::string targetTriple = _linker.getModule()->getTargetTriple();
 255     args.push_back(tool.c_str());
 256     if ( targetTriple.find("darwin") != std::string::npos ) {
 257         // darwin specific command line options
 258         if (strncmp(targetTriple.c_str(), "i386-apple-", 11) == 0) {
 259             args.push_back("-arch");
 260             args.push_back("i386");
 261         }
 262         else if (strncmp(targetTriple.c_str(), "x86_64-apple-", 13) == 0) {
 263             args.push_back("-arch");
 264             args.push_back("x86_64");
 265         }
 266         else if (strncmp(targetTriple.c_str(), "powerpc-apple-", 14) == 0) {
 267             args.push_back("-arch");
 268             args.push_back("ppc");
 269         }
 270         else if (strncmp(targetTriple.c_str(), "powerpc64-apple-", 16) == 0) {
 271             args.push_back("-arch");
 272             args.push_back("ppc64");
 273         }
 274         else if (strncmp(targetTriple.c_str(), "arm-apple-", 10) == 0) {
 275             args.push_back("-arch");
 276             args.push_back("arm");
 277         }
 278         else if ((strncmp(targetTriple.c_str(), "armv4t-apple-", 13) == 0) ||
 279                  (strncmp(targetTriple.c_str(), "thumbv4t-apple-", 15) == 0)) {
 280             args.push_back("-arch");
 281             args.push_back("armv4t");
 282         }
 283         else if ((strncmp(targetTriple.c_str(), "armv5-apple-", 12) == 0) ||
 284                  (strncmp(targetTriple.c_str(), "armv5e-apple-", 13) == 0) ||
 285                  (strncmp(targetTriple.c_str(), "thumbv5-apple-", 14) == 0) ||
 286                  (strncmp(targetTriple.c_str(), "thumbv5e-apple-", 15) == 0)) {
 287             args.push_back("-arch");
 288             args.push_back("armv5");
 289         }
 290         else if ((strncmp(targetTriple.c_str(), "armv6-apple-", 12) == 0) ||
 291                  (strncmp(targetTriple.c_str(), "thumbv6-apple-", 14) == 0)) {
 292             args.push_back("-arch");
 293             args.push_back("armv6");
 294         }
 295         else if ((strncmp(targetTriple.c_str(), "armv7-apple-", 12) == 0) ||
 296                  (strncmp(targetTriple.c_str(), "thumbv7-apple-", 14) == 0)) {
 297             args.push_back("-arch");
 298             args.push_back("armv7");
 299         }
 300         // add -static to assembler command line when code model requires
 301         if ( (_assemblerPath != NULL) && (_codeModel == LTO_CODEGEN_PIC_MODEL_STATIC) )
 302             args.push_back("-static");
 303     }
 304     if ( needsCompilerOptions ) {
 305         args.push_back("-c");
 306         args.push_back("-x");
 307         args.push_back("assembler");
 308     }
 309     args.push_back("-o");
 310     args.push_back(objPath.c_str());
 311     args.push_back(asmPath.c_str());
 312     args.push_back(0);
 313
 314     // invoke assembler
 315     if ( sys::Program::ExecuteAndWait(tool, &args[0], 0, 0, 0, 0, &errMsg) ) {
 316         errMsg = "error in assembly";
 317         return true;
 318     }
 319     return false; // success
 320 }
 321
 322
 323
 324 bool LTOCodeGenerator::determineTarget(std::string& errMsg)
 325 {
 326     if ( _target == NULL ) {
 327         // create target machine from info for merged modules
 328         Module* mergedModule = _linker.getModule();
 329         const TargetMachineRegistry::entry* march =
 330           TargetMachineRegistry::getClosestStaticTargetForModule(
 331                                                        *mergedModule, errMsg);
 332         if ( march == NULL )
 333             return true;
 334
 335         // The relocation model is actually a static member of TargetMachine
 336         // and needs to be set before the TargetMachine is instantiated.
 337         switch( _codeModel ) {
 338         case LTO_CODEGEN_PIC_MODEL_STATIC:
 339             TargetMachine::setRelocationModel(Reloc::Static);
 340             break;
 341         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 342             TargetMachine::setRelocationModel(Reloc::PIC_);
 343             break;
 344         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 345             TargetMachine::setRelocationModel(Reloc::DynamicNoPIC);
 346             break;
 347         }
 348
 349         // construct LTModule, hand over ownership of module and target
 350         std::string FeatureStr =
 351           getFeatureString(_linker.getModule()->getTargetTriple().c_str());
 352         _target = march->CtorFn(*mergedModule, FeatureStr.c_str());
 353     }
 354     return false;
 355 }
 356
 357 void LTOCodeGenerator::applyScopeRestrictions()
 358 {
 359     if ( !_scopeRestrictionsDone ) {
 360         Module* mergedModule = _linker.getModule();
 361
 362         // Start off with a verification pass.
 363         PassManager passes;
 364         passes.add(createVerifierPass());
 365
 366         // mark which symbols can not be internalized
 367         if ( !_mustPreserveSymbols.empty() ) {
 368             Mangler mangler(*mergedModule,
 369                                 _target->getTargetAsmInfo()->getGlobalPrefix());
 370             std::vector<const char*> mustPreserveList;
 371             for (Module::iterator f = mergedModule->begin(),
 372                                         e = mergedModule->end(); f != e; ++f) {
 373                 if ( !f->isDeclaration()
 374                   && _mustPreserveSymbols.count(mangler.getValueName(f)) )
 375                     mustPreserveList.push_back(::strdup(f->getName().c_str()));
 376             }
 377             for (Module::global_iterator v = mergedModule->global_begin(),
 378                                  e = mergedModule->global_end(); v !=  e; ++v) {
 379                 if ( !v->isDeclaration()
 380                   && _mustPreserveSymbols.count(mangler.getValueName(v)) )
 381                     mustPreserveList.push_back(::strdup(v->getName().c_str()));
 382             }
 383             passes.add(createInternalizePass(mustPreserveList));
 384         }
 385         // apply scope restrictions
 386         passes.run(*mergedModule);
 387
 388         _scopeRestrictionsDone = true;
 389     }
 390 }
 391
 392 /// Optimize merged modules using various IPO passes
 393 bool LTOCodeGenerator::generateAssemblyCode(raw_ostream& out,
 394                                             std::string& errMsg)
 395 {
 396     if (  this->determineTarget(errMsg) )
 397         return true;
 398
 399     // mark which symbols can not be internalized
 400     this->applyScopeRestrictions();
 401
 402     Module* mergedModule = _linker.getModule();
 403
 404      // If target supports exception handling then enable it now.
 405     if ( _target->getTargetAsmInfo()->doesSupportExceptionHandling() )
 406         llvm::ExceptionHandling = true;
 407
 408     // if options were requested, set them
 409     if ( !_codegenOptions.empty() )
 410         cl::ParseCommandLineOptions(_codegenOptions.size(),
 411                                                 (char**)&_codegenOptions[0]);
 412
 413     // Instantiate the pass manager to organize the passes.
 414     PassManager passes;
 415
 416     // Start off with a verification pass.
 417     passes.add(createVerifierPass());
 418
 419     // Add an appropriate TargetData instance for this module...
 420     passes.add(new TargetData(*_target->getTargetData()));
 421
 422     createStandardLTOPasses(&passes, /*Internalize=*/ false, !DisableInline,
 423                             /*VerifyEach=*/ false);
 424
 425     // Make sure everything is still good.
 426     passes.add(createVerifierPass());
 427
 428     FunctionPassManager* codeGenPasses =
 429             new FunctionPassManager(new ExistingModuleProvider(mergedModule));
 430
 431     codeGenPasses->add(new TargetData(*_target->getTargetData()));
 432
 433     ObjectCodeEmitter* oce = NULL;
 434
 435     switch (_target->addPassesToEmitFile(*codeGenPasses, out,
 436                                          TargetMachine::AssemblyFile,
 437                                          CodeGenOpt::Aggressive)) {
 438         case FileModel::MachOFile:
 439             oce = AddMachOWriter(*codeGenPasses, out, *_target);
 440             break;
 441         case FileModel::ElfFile:
 442             oce = AddELFWriter(*codeGenPasses, out, *_target);
 443             break;
 444         case FileModel::AsmFile:
 445             break;
 446         case FileModel::Error:
 447         case FileModel::None:
 448             errMsg = "target file type not supported";
 449             return true;
 450     }
 451
 452     if (_target->addPassesToEmitFileFinish(*codeGenPasses, oce,
 453                                            CodeGenOpt::Aggressive)) {
 454         errMsg = "target does not support generation of this file type";
 455         return true;
 456     }
 457
 458     // Run our queue of passes all at once now, efficiently.
 459     passes.run(*mergedModule);
 460
 461     // Run the code generator, and write assembly file
 462     codeGenPasses->doInitialization();
 463
 464     for (Module::iterator
 465            it = mergedModule->begin(), e = mergedModule->end(); it != e; ++it)
 466       if (!it->isDeclaration())
 467         codeGenPasses->run(*it);
 468
 469     codeGenPasses->doFinalization();
 470     return false; // success
 471 }
 472
 473
 474 /// Optimize merged modules using various IPO passes
 475 void LTOCodeGenerator::setCodeGenDebugOptions(const char* options)
 476 {
 477     std::string ops(options);
 478     for (std::string o = getToken(ops); !o.empty(); o = getToken(ops)) {
 479         // ParseCommandLineOptions() expects argv[0] to be program name.
 480         // Lazily add that.
 481         if ( _codegenOptions.empty() )
 482             _codegenOptions.push_back("libLTO");
 483         _codegenOptions.push_back(strdup(o.c_str()));
 484     }
 485 }