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