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/Signals.h"
  40 #include "llvm/Target/SubtargetFeature.h"
  41 #include "llvm/Target/TargetOptions.h"
  42 #include "llvm/Target/TargetAsmInfo.h"
  43 #include "llvm/Target/TargetData.h"
  44 #include "llvm/Target/TargetMachine.h"
  45 #include "llvm/Target/TargetRegistry.h"
  46 #include "llvm/Target/TargetSelect.h"
  47 #include "llvm/Transforms/IPO.h"
  48 #include "llvm/Transforms/Scalar.h"
  49 #include "llvm/Config/config.h"
  50
  51
  52 #include <cstdlib>
  53 #include <fstream>
  54 #include <unistd.h>
  55 #include <fcntl.h>
  56
  57
  58 using namespace llvm;
  59
  60 static cl::opt<bool> DisableInline("disable-inlining",
  61   cl::desc("Do not run the inliner pass"));
  62
  63
  64 const char* LTOCodeGenerator::getVersionString()
  65 {
  66 #ifdef LLVM_VERSION_INFO
  67     return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
  68 #else
  69     return PACKAGE_NAME " version " PACKAGE_VERSION;
  70 #endif
  71 }
  72
  73
  74 LTOCodeGenerator::LTOCodeGenerator()
  75     : _context(getGlobalContext()),
  76       _linker("LinkTimeOptimizer", "ld-temp.o", _context), _target(NULL),
  77       _emitDwarfDebugInfo(false), _scopeRestrictionsDone(false),
  78       _codeModel(LTO_CODEGEN_PIC_MODEL_DYNAMIC),
  79       _nativeObjectFile(NULL), _assemblerPath(NULL)
  80 {
  81     InitializeAllTargets();
  82     InitializeAllAsmPrinters();
  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::setAssemblerPath(const char* path)
 130 {
 131     if ( _assemblerPath )
 132         delete _assemblerPath;
 133     _assemblerPath = new sys::Path(path);
 134 }
 135
 136 void LTOCodeGenerator::addMustPreserveSymbol(const char* sym)
 137 {
 138     _mustPreserveSymbols[sym] = 1;
 139 }
 140
 141
 142 bool LTOCodeGenerator::writeMergedModules(const char* path, std::string& errMsg)
 143 {
 144     if ( this->determineTarget(errMsg) )
 145         return true;
 146
 147     // mark which symbols can not be internalized
 148     this->applyScopeRestrictions();
 149
 150     // create output file
 151     std::ofstream out(path, std::ios_base::out|std::ios::trunc|std::ios::binary);
 152     if ( out.fail() ) {
 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     if ( out.fail() ) {
 161         errMsg = "could not write bitcode file: ";
 162         errMsg += path;
 163         return true;
 164     }
 165
 166     return false;
 167 }
 168
 169
 170 const void* LTOCodeGenerator::compile(size_t* length, std::string& errMsg)
 171 {
 172     // make unique temp .s file to put generated assembly code
 173     sys::Path uniqueAsmPath("lto-llvm.s");
 174     if ( uniqueAsmPath.createTemporaryFileOnDisk(true, &errMsg) )
 175         return NULL;
 176     sys::RemoveFileOnSignal(uniqueAsmPath);
 177
 178     // generate assembly code
 179     bool genResult = false;
 180     {
 181       raw_fd_ostream asmFD(uniqueAsmPath.c_str(),
 182                            /*Binary=*/false, /*Force=*/true,
 183                            errMsg);
 184       formatted_raw_ostream asmFile(asmFD);
 185       if (!errMsg.empty())
 186         return NULL;
 187       genResult = this->generateAssemblyCode(asmFile, errMsg);
 188     }
 189     if ( genResult ) {
 190         if ( uniqueAsmPath.exists() )
 191             uniqueAsmPath.eraseFromDisk();
 192         return NULL;
 193     }
 194
 195     // make unique temp .o file to put generated object file
 196     sys::PathWithStatus uniqueObjPath("lto-llvm.o");
 197     if ( uniqueObjPath.createTemporaryFileOnDisk(true, &errMsg) ) {
 198         if ( uniqueAsmPath.exists() )
 199             uniqueAsmPath.eraseFromDisk();
 200         return NULL;
 201     }
 202     sys::RemoveFileOnSignal(uniqueObjPath);
 203
 204     // assemble the assembly code
 205     const std::string& uniqueObjStr = uniqueObjPath.toString();
 206     bool asmResult = this->assemble(uniqueAsmPath.toString(),
 207                                                         uniqueObjStr, errMsg);
 208     if ( !asmResult ) {
 209         // remove old buffer if compile() called twice
 210         delete _nativeObjectFile;
 211
 212         // read .o file into memory buffer
 213         _nativeObjectFile = MemoryBuffer::getFile(uniqueObjStr.c_str(),&errMsg);
 214     }
 215
 216     // remove temp files
 217     uniqueAsmPath.eraseFromDisk();
 218     uniqueObjPath.eraseFromDisk();
 219
 220     // return buffer, unless error
 221     if ( _nativeObjectFile == NULL )
 222         return NULL;
 223     *length = _nativeObjectFile->getBufferSize();
 224     return _nativeObjectFile->getBufferStart();
 225 }
 226
 227
 228 bool LTOCodeGenerator::assemble(const std::string& asmPath,
 229                                 const std::string& objPath, std::string& errMsg)
 230 {
 231     sys::Path tool;
 232     bool needsCompilerOptions = true;
 233     if ( _assemblerPath ) {
 234         tool = *_assemblerPath;
 235         needsCompilerOptions = false;
 236     } else {
 237         // find compiler driver
 238         tool = sys::Program::FindProgramByName("gcc");
 239         if ( tool.isEmpty() ) {
 240             errMsg = "can't locate gcc";
 241             return true;
 242         }
 243     }
 244
 245     // build argument list
 246     std::vector<const char*> args;
 247     std::string targetTriple = _linker.getModule()->getTargetTriple();
 248     args.push_back(tool.c_str());
 249     if ( targetTriple.find("darwin") != std::string::npos ) {
 250         // darwin specific command line options
 251         if (strncmp(targetTriple.c_str(), "i386-apple-", 11) == 0) {
 252             args.push_back("-arch");
 253             args.push_back("i386");
 254         }
 255         else if (strncmp(targetTriple.c_str(), "x86_64-apple-", 13) == 0) {
 256             args.push_back("-arch");
 257             args.push_back("x86_64");
 258         }
 259         else if (strncmp(targetTriple.c_str(), "powerpc-apple-", 14) == 0) {
 260             args.push_back("-arch");
 261             args.push_back("ppc");
 262         }
 263         else if (strncmp(targetTriple.c_str(), "powerpc64-apple-", 16) == 0) {
 264             args.push_back("-arch");
 265             args.push_back("ppc64");
 266         }
 267         else if (strncmp(targetTriple.c_str(), "arm-apple-", 10) == 0) {
 268             args.push_back("-arch");
 269             args.push_back("arm");
 270         }
 271         else if ((strncmp(targetTriple.c_str(), "armv4t-apple-", 13) == 0) ||
 272                  (strncmp(targetTriple.c_str(), "thumbv4t-apple-", 15) == 0)) {
 273             args.push_back("-arch");
 274             args.push_back("armv4t");
 275         }
 276         else if ((strncmp(targetTriple.c_str(), "armv5-apple-", 12) == 0) ||
 277                  (strncmp(targetTriple.c_str(), "armv5e-apple-", 13) == 0) ||
 278                  (strncmp(targetTriple.c_str(), "thumbv5-apple-", 14) == 0) ||
 279                  (strncmp(targetTriple.c_str(), "thumbv5e-apple-", 15) == 0)) {
 280             args.push_back("-arch");
 281             args.push_back("armv5");
 282         }
 283         else if ((strncmp(targetTriple.c_str(), "armv6-apple-", 12) == 0) ||
 284                  (strncmp(targetTriple.c_str(), "thumbv6-apple-", 14) == 0)) {
 285             args.push_back("-arch");
 286             args.push_back("armv6");
 287         }
 288         else if ((strncmp(targetTriple.c_str(), "armv7-apple-", 12) == 0) ||
 289                  (strncmp(targetTriple.c_str(), "thumbv7-apple-", 14) == 0)) {
 290             args.push_back("-arch");
 291             args.push_back("armv7");
 292         }
 293         // add -static to assembler command line when code model requires
 294         if ( (_assemblerPath != NULL) && (_codeModel == LTO_CODEGEN_PIC_MODEL_STATIC) )
 295             args.push_back("-static");
 296     }
 297     if ( needsCompilerOptions ) {
 298         args.push_back("-c");
 299         args.push_back("-x");
 300         args.push_back("assembler");
 301     }
 302     args.push_back("-o");
 303     args.push_back(objPath.c_str());
 304     args.push_back(asmPath.c_str());
 305     args.push_back(0);
 306
 307     // invoke assembler
 308     if ( sys::Program::ExecuteAndWait(tool, &args[0], 0, 0, 0, 0, &errMsg) ) {
 309         errMsg = "error in assembly";
 310         return true;
 311     }
 312     return false; // success
 313 }
 314
 315
 316
 317 bool LTOCodeGenerator::determineTarget(std::string& errMsg)
 318 {
 319     if ( _target == NULL ) {
 320         std::string Triple = _linker.getModule()->getTargetTriple();
 321         if (Triple.empty())
 322           Triple = sys::getHostTriple();
 323
 324         // create target machine from info for merged modules
 325         const Target *march = TargetRegistry::lookupTarget(Triple, errMsg);
 326         if ( march == NULL )
 327             return true;
 328
 329         // The relocation model is actually a static member of TargetMachine
 330         // and needs to be set before the TargetMachine is instantiated.
 331         switch( _codeModel ) {
 332         case LTO_CODEGEN_PIC_MODEL_STATIC:
 333             TargetMachine::setRelocationModel(Reloc::Static);
 334             break;
 335         case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
 336             TargetMachine::setRelocationModel(Reloc::PIC_);
 337             break;
 338         case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
 339             TargetMachine::setRelocationModel(Reloc::DynamicNoPIC);
 340             break;
 341         }
 342
 343         // construct LTModule, hand over ownership of module and target
 344         std::string FeatureStr = getFeatureString(Triple.c_str());
 345         _target = march->createTargetMachine(Triple, FeatureStr);
 346     }
 347     return false;
 348 }
 349
 350 void LTOCodeGenerator::applyScopeRestrictions()
 351 {
 352     if ( !_scopeRestrictionsDone ) {
 353         Module* mergedModule = _linker.getModule();
 354
 355         // Start off with a verification pass.
 356         PassManager passes;
 357         passes.add(createVerifierPass());
 358
 359         // mark which symbols can not be internalized
 360         if ( !_mustPreserveSymbols.empty() ) {
 361             Mangler mangler(*mergedModule,
 362                                 _target->getTargetAsmInfo()->getGlobalPrefix());
 363             std::vector<const char*> mustPreserveList;
 364             for (Module::iterator f = mergedModule->begin(),
 365                                         e = mergedModule->end(); f != e; ++f) {
 366                 if ( !f->isDeclaration()
 367                   && _mustPreserveSymbols.count(mangler.getMangledName(f)) )
 368                   mustPreserveList.push_back(::strdup(f->getNameStr().c_str()));
 369             }
 370             for (Module::global_iterator v = mergedModule->global_begin(),
 371                                  e = mergedModule->global_end(); v !=  e; ++v) {
 372                 if ( !v->isDeclaration()
 373                   && _mustPreserveSymbols.count(mangler.getMangledName(v)) )
 374                   mustPreserveList.push_back(::strdup(v->getNameStr().c_str()));
 375             }
 376             passes.add(createInternalizePass(mustPreserveList));
 377         }
 378         // apply scope restrictions
 379         passes.run(*mergedModule);
 380
 381         _scopeRestrictionsDone = true;
 382     }
 383 }
 384
 385 /// Optimize merged modules using various IPO passes
 386 bool LTOCodeGenerator::generateAssemblyCode(formatted_raw_ostream& out,
 387                                             std::string& errMsg)
 388 {
 389     if ( this->determineTarget(errMsg) )
 390         return true;
 391
 392     // mark which symbols can not be internalized
 393     this->applyScopeRestrictions();
 394
 395     Module* mergedModule = _linker.getModule();
 396
 397     // If target supports exception handling then enable it now.
 398     switch (_target->getTargetAsmInfo()->getExceptionHandlingType()) {
 399     case ExceptionHandling::Dwarf:
 400       llvm::DwarfExceptionHandling = true;
 401       break;
 402     case ExceptionHandling::SjLj:
 403       llvm::SjLjExceptionHandling = true;
 404       break;
 405     case ExceptionHandling::None:
 406       break;
 407     default:
 408       assert (0 && "Unknown exception handling model!");
 409     }
 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
 474     out.flush();
 475
 476     return false; // success
 477 }
 478
 479
 480 /// Optimize merged modules using various IPO passes
 481 void LTOCodeGenerator::setCodeGenDebugOptions(const char* options)
 482 {
 483     std::string ops(options);
 484     for (std::string o = getToken(ops); !o.empty(); o = getToken(ops)) {
 485         // ParseCommandLineOptions() expects argv[0] to be program name.
 486         // Lazily add that.
 487         if ( _codegenOptions.empty() )
 488             _codegenOptions.push_back("libLTO");
 489         _codegenOptions.push_back(strdup(o.c_str()));
 490     }
 491 }