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/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/TargetRegistry.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 asmFD(raw_fd_ostream(uniqueAsmPath.c_str(),
 189                                           /*Binary=*/false, /*Force=*/true,
 190                                           errMsg));
 191       formatted_raw_ostream asmFile(asmFD);
 192       if (!errMsg.empty())
 193         return NULL;
 194       genResult = this->generateAssemblyCode(asmFile, errMsg);
 195     }
 196     if ( genResult ) {
 197         if ( uniqueAsmPath.exists() )
 198             uniqueAsmPath.eraseFromDisk();
 199         return NULL;
 200     }
 201
 202     // make unique temp .o file to put generated object file
 203     sys::PathWithStatus uniqueObjPath("lto-llvm.o");
 204     if ( uniqueObjPath.createTemporaryFileOnDisk(true, &errMsg) ) {
 205         if ( uniqueAsmPath.exists() )
 206             uniqueAsmPath.eraseFromDisk();
 207         return NULL;
 208     }
 209     sys::RemoveFileOnSignal(uniqueObjPath);
 210
 211     // assemble the assembly code
 212     const std::string& uniqueObjStr = uniqueObjPath.toString();
 213     bool asmResult = this->assemble(uniqueAsmPath.toString(),
 214                                                         uniqueObjStr, errMsg);
 215     if ( !asmResult ) {
 216         // remove old buffer if compile() called twice
 217         delete _nativeObjectFile;
 218
 219         // read .o file into memory buffer
 220         _nativeObjectFile = MemoryBuffer::getFile(uniqueObjStr.c_str(),&errMsg);
 221     }
 222
 223     // remove temp files
 224     uniqueAsmPath.eraseFromDisk();
 225     uniqueObjPath.eraseFromDisk();
 226
 227     // return buffer, unless error
 228     if ( _nativeObjectFile == NULL )
 229         return NULL;
 230     *length = _nativeObjectFile->getBufferSize();
 231     return _nativeObjectFile->getBufferStart();
 232 }
 233
 234
 235 bool LTOCodeGenerator::assemble(const std::string& asmPath,
 236                                 const std::string& objPath, std::string& errMsg)
 237 {
 238     sys::Path tool;
 239     bool needsCompilerOptions = true;
 240     if ( _assemblerPath ) {
 241         tool = *_assemblerPath;
 242         needsCompilerOptions = false;
 243     }
 244     else if ( _gccPath ) {
 245         tool = *_gccPath;
 246     } else {
 247         // find compiler driver
 248         tool = sys::Program::FindProgramByName("gcc");
 249         if ( tool.isEmpty() ) {
 250             errMsg = "can't locate gcc";
 251             return true;
 252         }
 253     }
 254
 255     // build argument list
 256     std::vector<const char*> args;
 257     std::string targetTriple = _linker.getModule()->getTargetTriple();
 258     args.push_back(tool.c_str());
 259     if ( targetTriple.find("darwin") != std::string::npos ) {
 260         // darwin specific command line options
 261         if (strncmp(targetTriple.c_str(), "i386-apple-", 11) == 0) {
 262             args.push_back("-arch");
 263             args.push_back("i386");
 264         }
 265         else if (strncmp(targetTriple.c_str(), "x86_64-apple-", 13) == 0) {
 266             args.push_back("-arch");
 267             args.push_back("x86_64");
 268         }
 269         else if (strncmp(targetTriple.c_str(), "powerpc-apple-", 14) == 0) {
 270             args.push_back("-arch");
 271             args.push_back("ppc");
 272         }
 273         else if (strncmp(targetTriple.c_str(), "powerpc64-apple-", 16) == 0) {
 274             args.push_back("-arch");
 275             args.push_back("ppc64");
 276         }
 277         else if (strncmp(targetTriple.c_str(), "arm-apple-", 10) == 0) {
 278             args.push_back("-arch");
 279             args.push_back("arm");
 280         }
 281         else if ((strncmp(targetTriple.c_str(), "armv4t-apple-", 13) == 0) ||
 282                  (strncmp(targetTriple.c_str(), "thumbv4t-apple-", 15) == 0)) {
 283             args.push_back("-arch");
 284             args.push_back("armv4t");
 285         }
 286         else if ((strncmp(targetTriple.c_str(), "armv5-apple-", 12) == 0) ||
 287                  (strncmp(targetTriple.c_str(), "armv5e-apple-", 13) == 0) ||
 288                  (strncmp(targetTriple.c_str(), "thumbv5-apple-", 14) == 0) ||
 289                  (strncmp(targetTriple.c_str(), "thumbv5e-apple-", 15) == 0)) {
 290             args.push_back("-arch");
 291             args.push_back("armv5");
 292         }
 293         else if ((strncmp(targetTriple.c_str(), "armv6-apple-", 12) == 0) ||
 294                  (strncmp(targetTriple.c_str(), "thumbv6-apple-", 14) == 0)) {
 295             args.push_back("-arch");
 296             args.push_back("armv6");
 297         }
 298         else if ((strncmp(targetTriple.c_str(), "armv7-apple-", 12) == 0) ||
 299                  (strncmp(targetTriple.c_str(), "thumbv7-apple-", 14) == 0)) {
 300             args.push_back("-arch");
 301             args.push_back("armv7");
 302         }
 303         // add -static to assembler command line when code model requires
 304         if ( (_assemblerPath != NULL) && (_codeModel == LTO_CODEGEN_PIC_MODEL_STATIC) )
 305             args.push_back("-static");
 306     }
 307     if ( needsCompilerOptions ) {
 308         args.push_back("-c");
 309         args.push_back("-x");
 310         args.push_back("assembler");
 311     }
 312     args.push_back("-o");
 313     args.push_back(objPath.c_str());
 314     args.push_back(asmPath.c_str());
 315     args.push_back(0);
 316
 317     // invoke assembler
 318     if ( sys::Program::ExecuteAndWait(tool, &args[0], 0, 0, 0, 0, &errMsg) ) {
 319         errMsg = "error in assembly";
 320         return true;
 321     }
 322     return false; // success
 323 }
 324
 325
 326
 327 bool LTOCodeGenerator::determineTarget(std::string& errMsg)
 328 {
 329     if ( _target == NULL ) {
 330         // create target machine from info for merged modules
 331         Module* mergedModule = _linker.getModule();
 332         const Target *march =
 333           TargetRegistry::getClosestStaticTargetForModule(*mergedModule,
 334                                                           errMsg);
 335         if ( march == NULL )
 336             return true;
 337
 338         // The relocation model is actually a static member of TargetMachine
 339         // and needs to be set before the TargetMachine is instantiated.
 340         switch( _codeModel ) {
 341         case LTO_CODEGEN_PIC_MODEL_STATIC:
 342             TargetMachine::setRelocationModel(Reloc::Static);
 343             break;
 344         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 345             TargetMachine::setRelocationModel(Reloc::PIC_);
 346             break;
 347         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 348             TargetMachine::setRelocationModel(Reloc::DynamicNoPIC);
 349             break;
 350         }
 351
 352         // construct LTModule, hand over ownership of module and target
 353         std::string FeatureStr =
 354           getFeatureString(_linker.getModule()->getTargetTriple().c_str());
 355         _target = march->createTargetMachine(*mergedModule, FeatureStr.c_str());
 356     }
 357     return false;
 358 }
 359
 360 void LTOCodeGenerator::applyScopeRestrictions()
 361 {
 362     if ( !_scopeRestrictionsDone ) {
 363         Module* mergedModule = _linker.getModule();
 364
 365         // Start off with a verification pass.
 366         PassManager passes;
 367         passes.add(createVerifierPass());
 368
 369         // mark which symbols can not be internalized
 370         if ( !_mustPreserveSymbols.empty() ) {
 371             Mangler mangler(*mergedModule,
 372                                 _target->getTargetAsmInfo()->getGlobalPrefix());
 373             std::vector<const char*> mustPreserveList;
 374             for (Module::iterator f = mergedModule->begin(),
 375                                         e = mergedModule->end(); f != e; ++f) {
 376                 if ( !f->isDeclaration()
 377                   && _mustPreserveSymbols.count(mangler.getMangledName(f)) )
 378                     mustPreserveList.push_back(::strdup(f->getName().c_str()));
 379             }
 380             for (Module::global_iterator v = mergedModule->global_begin(),
 381                                  e = mergedModule->global_end(); v !=  e; ++v) {
 382                 if ( !v->isDeclaration()
 383                   && _mustPreserveSymbols.count(mangler.getMangledName(v)) )
 384                     mustPreserveList.push_back(::strdup(v->getName().c_str()));
 385             }
 386             passes.add(createInternalizePass(mustPreserveList));
 387         }
 388         // apply scope restrictions
 389         passes.run(*mergedModule);
 390
 391         _scopeRestrictionsDone = true;
 392     }
 393 }
 394
 395 /// Optimize merged modules using various IPO passes
 396 bool LTOCodeGenerator::generateAssemblyCode(formatted_raw_ostream& out,
 397                                             std::string& errMsg)
 398 {
 399     if (  this->determineTarget(errMsg) )
 400         return true;
 401
 402     // mark which symbols can not be internalized
 403     this->applyScopeRestrictions();
 404
 405     Module* mergedModule = _linker.getModule();
 406
 407      // If target supports exception handling then enable it now.
 408     if ( _target->getTargetAsmInfo()->doesSupportExceptionHandling() )
 409         llvm::ExceptionHandling = true;
 410
 411     // if options were requested, set them
 412     if ( !_codegenOptions.empty() )
 413         cl::ParseCommandLineOptions(_codegenOptions.size(),
 414                                                 (char**)&_codegenOptions[0]);
 415
 416     // Instantiate the pass manager to organize the passes.
 417     PassManager passes;
 418
 419     // Start off with a verification pass.
 420     passes.add(createVerifierPass());
 421
 422     // Add an appropriate TargetData instance for this module...
 423     passes.add(new TargetData(*_target->getTargetData()));
 424
 425     createStandardLTOPasses(&passes, /*Internalize=*/ false, !DisableInline,
 426                             /*VerifyEach=*/ false);
 427
 428     // Make sure everything is still good.
 429     passes.add(createVerifierPass());
 430
 431     FunctionPassManager* codeGenPasses =
 432             new FunctionPassManager(new ExistingModuleProvider(mergedModule));
 433
 434     codeGenPasses->add(new TargetData(*_target->getTargetData()));
 435
 436     ObjectCodeEmitter* oce = NULL;
 437
 438     switch (_target->addPassesToEmitFile(*codeGenPasses, out,
 439                                          TargetMachine::AssemblyFile,
 440                                          CodeGenOpt::Aggressive)) {
 441         case FileModel::MachOFile:
 442             oce = AddMachOWriter(*codeGenPasses, out, *_target);
 443             break;
 444         case FileModel::ElfFile:
 445             oce = AddELFWriter(*codeGenPasses, out, *_target);
 446             break;
 447         case FileModel::AsmFile:
 448             break;
 449         case FileModel::Error:
 450         case FileModel::None:
 451             errMsg = "target file type not supported";
 452             return true;
 453     }
 454
 455     if (_target->addPassesToEmitFileFinish(*codeGenPasses, oce,
 456                                            CodeGenOpt::Aggressive)) {
 457         errMsg = "target does not support generation of this file type";
 458         return true;
 459     }
 460
 461     // Run our queue of passes all at once now, efficiently.
 462     passes.run(*mergedModule);
 463
 464     // Run the code generator, and write assembly file
 465     codeGenPasses->doInitialization();
 466
 467     for (Module::iterator
 468            it = mergedModule->begin(), e = mergedModule->end(); it != e; ++it)
 469       if (!it->isDeclaration())
 470         codeGenPasses->run(*it);
 471
 472     codeGenPasses->doFinalization();
 473     return false; // success
 474 }
 475
 476
 477 /// Optimize merged modules using various IPO passes
 478 void LTOCodeGenerator::setCodeGenDebugOptions(const char* options)
 479 {
 480     std::string ops(options);
 481     for (std::string o = getToken(ops); !o.empty(); o = getToken(ops)) {
 482         // ParseCommandLineOptions() expects argv[0] to be program name.
 483         // Lazily add that.
 484         if ( _codegenOptions.empty() )
 485             _codegenOptions.push_back("libLTO");
 486         _codegenOptions.push_back(strdup(o.c_str()));
 487     }
 488 }