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 "LTOCodeGenerator.h"
16#include "LTOModule.h"
17#include "llvm/Constants.h"
18#include "llvm/DerivedTypes.h"
19#include "llvm/Linker.h"
20#include "llvm/LLVMContext.h"
21#include "llvm/Module.h"
22#include "llvm/PassManager.h"
23#include "llvm/Analysis/Passes.h"
24#include "llvm/Analysis/Verifier.h"
25#include "llvm/Bitcode/ReaderWriter.h"
26#include "llvm/Config/config.h"
27#include "llvm/MC/MCAsmInfo.h"
28#include "llvm/MC/MCContext.h"
29#include "llvm/MC/SubtargetFeature.h"
30#include "llvm/Target/Mangler.h"
31#include "llvm/Target/TargetOptions.h"
32#include "llvm/Target/TargetData.h"
33#include "llvm/Target/TargetMachine.h"
34#include "llvm/Target/TargetRegisterInfo.h"
35#include "llvm/Transforms/IPO.h"
36#include "llvm/Transforms/IPO/PassManagerBuilder.h"
37#include "llvm/Support/CommandLine.h"
38#include "llvm/Support/FormattedStream.h"
39#include "llvm/Support/MemoryBuffer.h"
40#include "llvm/Support/ToolOutputFile.h"
41#include "llvm/Support/Host.h"
42#include "llvm/Support/Signals.h"
43#include "llvm/Support/TargetRegistry.h"
44#include "llvm/Support/TargetSelect.h"
45#include "llvm/Support/system_error.h"
46#include "llvm/ADT/StringExtras.h"
47using namespace llvm;
48
49static cl::opt<bool>
50DisableInline("disable-inlining", cl::init(false),
51  cl::desc("Do not run the inliner pass"));
52
53static cl::opt<bool>
54DisableGVNLoadPRE("disable-gvn-loadpre", cl::init(false),
55  cl::desc("Do not run the GVN load PRE pass"));
56
57const char* LTOCodeGenerator::getVersionString() {
58#ifdef LLVM_VERSION_INFO
59  return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
60#else
61  return PACKAGE_NAME " version " PACKAGE_VERSION;
62#endif
63}
64
65LTOCodeGenerator::LTOCodeGenerator()
66  : _context(getGlobalContext()),
67    _linker("LinkTimeOptimizer", "ld-temp.o", _context), _target(NULL),
68    _emitDwarfDebugInfo(false), _scopeRestrictionsDone(false),
69    _codeModel(LTO_CODEGEN_PIC_MODEL_DYNAMIC),
70    _nativeObjectFile(NULL) {
71  InitializeAllTargets();
72  InitializeAllTargetMCs();
73  InitializeAllAsmPrinters();
74}
75
76LTOCodeGenerator::~LTOCodeGenerator() {
77  delete _target;
78  delete _nativeObjectFile;
79
80  for (std::vector<char*>::iterator I = _codegenOptions.begin(),
81         E = _codegenOptions.end(); I != E; ++I)
82    free(*I);
83}
84
85bool LTOCodeGenerator::addModule(LTOModule* mod, std::string& errMsg) {
86  bool ret = _linker.LinkInModule(mod->getLLVVMModule(), &errMsg);
87
88  const std::vector<const char*> &undefs = mod->getAsmUndefinedRefs();
89  for (int i = 0, e = undefs.size(); i != e; ++i)
90    _asmUndefinedRefs[undefs[i]] = 1;
91
92  return ret;
93}
94
95bool LTOCodeGenerator::setDebugInfo(lto_debug_model debug,
96                                    std::string& errMsg) {
97  switch (debug) {
98  case LTO_DEBUG_MODEL_NONE:
99    _emitDwarfDebugInfo = false;
100    return false;
101
102  case LTO_DEBUG_MODEL_DWARF:
103    _emitDwarfDebugInfo = true;
104    return false;
105  }
106  llvm_unreachable("Unknown debug format!");
107}
108
109bool LTOCodeGenerator::setCodePICModel(lto_codegen_model model,
110                                       std::string& errMsg) {
111  switch (model) {
112  case LTO_CODEGEN_PIC_MODEL_STATIC:
113  case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
114  case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
115    _codeModel = model;
116    return false;
117  }
118  llvm_unreachable("Unknown PIC model!");
119}
120
121bool LTOCodeGenerator::writeMergedModules(const char *path,
122                                          std::string &errMsg) {
123  if (determineTarget(errMsg))
124    return true;
125
126  // mark which symbols can not be internalized
127  applyScopeRestrictions();
128
129  // create output file
130  std::string ErrInfo;
131  tool_output_file Out(path, ErrInfo,
132                       raw_fd_ostream::F_Binary);
133  if (!ErrInfo.empty()) {
134    errMsg = "could not open bitcode file for writing: ";
135    errMsg += path;
136    return true;
137  }
138
139  // write bitcode to it
140  WriteBitcodeToFile(_linker.getModule(), Out.os());
141  Out.os().close();
142
143  if (Out.os().has_error()) {
144    errMsg = "could not write bitcode file: ";
145    errMsg += path;
146    Out.os().clear_error();
147    return true;
148  }
149
150  Out.keep();
151  return false;
152}
153
154bool LTOCodeGenerator::compile_to_file(const char** name, std::string& errMsg) {
155  // make unique temp .o file to put generated object file
156  sys::PathWithStatus uniqueObjPath("lto-llvm.o");
157  if (uniqueObjPath.createTemporaryFileOnDisk(false, &errMsg)) {
158    uniqueObjPath.eraseFromDisk();
159    return true;
160  }
161  sys::RemoveFileOnSignal(uniqueObjPath);
162
163  // generate object file
164  bool genResult = false;
165  tool_output_file objFile(uniqueObjPath.c_str(), errMsg);
166  if (!errMsg.empty()) {
167    uniqueObjPath.eraseFromDisk();
168    return true;
169  }
170
171  genResult = this->generateObjectFile(objFile.os(), errMsg);
172  objFile.os().close();
173  if (objFile.os().has_error()) {
174    objFile.os().clear_error();
175    uniqueObjPath.eraseFromDisk();
176    return true;
177  }
178
179  objFile.keep();
180  if (genResult) {
181    uniqueObjPath.eraseFromDisk();
182    return true;
183  }
184
185  _nativeObjectPath = uniqueObjPath.str();
186  *name = _nativeObjectPath.c_str();
187  return false;
188}
189
190const void* LTOCodeGenerator::compile(size_t* length, std::string& errMsg) {
191  const char *name;
192  if (compile_to_file(&name, errMsg))
193    return NULL;
194
195  // remove old buffer if compile() called twice
196  delete _nativeObjectFile;
197
198  // read .o file into memory buffer
199  OwningPtr<MemoryBuffer> BuffPtr;
200  if (error_code ec = MemoryBuffer::getFile(name, BuffPtr, -1, false)) {
201    errMsg = ec.message();
202    sys::Path(_nativeObjectPath).eraseFromDisk();
203    return NULL;
204  }
205  _nativeObjectFile = BuffPtr.take();
206
207  // remove temp files
208  sys::Path(_nativeObjectPath).eraseFromDisk();
209
210  // return buffer, unless error
211  if (_nativeObjectFile == NULL)
212    return NULL;
213  *length = _nativeObjectFile->getBufferSize();
214  return _nativeObjectFile->getBufferStart();
215}
216
217bool LTOCodeGenerator::determineTarget(std::string& errMsg) {
218  if (_target != NULL)
219    return false;
220
221  std::string TripleStr = _linker.getModule()->getTargetTriple();
222  if (TripleStr.empty())
223    TripleStr = sys::getDefaultTargetTriple();
224  llvm::Triple Triple(TripleStr);
225
226  // create target machine from info for merged modules
227  const Target *march = TargetRegistry::lookupTarget(TripleStr, errMsg);
228  if (march == NULL)
229    return true;
230
231  // The relocation model is actually a static member of TargetMachine and
232  // needs to be set before the TargetMachine is instantiated.
233  Reloc::Model RelocModel = Reloc::Default;
234  switch (_codeModel) {
235  case LTO_CODEGEN_PIC_MODEL_STATIC:
236    RelocModel = Reloc::Static;
237    break;
238  case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
239    RelocModel = Reloc::PIC_;
240    break;
241  case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
242    RelocModel = Reloc::DynamicNoPIC;
243    break;
244  }
245
246  // construct LTOModule, hand over ownership of module and target
247  SubtargetFeatures Features;
248  Features.getDefaultSubtargetFeatures(Triple);
249  std::string FeatureStr = Features.getString();
250  // Set a default CPU for Darwin triples.
251  if (_mCpu.empty() && Triple.isOSDarwin()) {
252    if (Triple.getArch() == llvm::Triple::x86_64)
253      _mCpu = "core2";
254    else if (Triple.getArch() == llvm::Triple::x86)
255      _mCpu = "yonah";
256  }
257  TargetOptions Options;
258  LTOModule::getTargetOptions(Options);
259  _target = march->createTargetMachine(TripleStr, _mCpu, FeatureStr, Options,
260                                       RelocModel, CodeModel::Default,
261                                       CodeGenOpt::Aggressive);
262  return false;
263}
264
265void LTOCodeGenerator::
266applyRestriction(GlobalValue &GV,
267                 std::vector<const char*> &mustPreserveList,
268                 SmallPtrSet<GlobalValue*, 8> &asmUsed,
269                 Mangler &mangler) {
270  SmallString<64> Buffer;
271  mangler.getNameWithPrefix(Buffer, &GV, false);
272
273  if (GV.isDeclaration())
274    return;
275  if (_mustPreserveSymbols.count(Buffer))
276    mustPreserveList.push_back(GV.getName().data());
277  if (_asmUndefinedRefs.count(Buffer))
278    asmUsed.insert(&GV);
279}
280
281static void findUsedValues(GlobalVariable *LLVMUsed,
282                           SmallPtrSet<GlobalValue*, 8> &UsedValues) {
283  if (LLVMUsed == 0) return;
284
285  ConstantArray *Inits = dyn_cast<ConstantArray>(LLVMUsed->getInitializer());
286  if (Inits == 0) return;
287
288  for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
289    if (GlobalValue *GV =
290        dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
291      UsedValues.insert(GV);
292}
293
294void LTOCodeGenerator::applyScopeRestrictions() {
295  if (_scopeRestrictionsDone) return;
296  Module *mergedModule = _linker.getModule();
297
298  // Start off with a verification pass.
299  PassManager passes;
300  passes.add(createVerifierPass());
301
302  // mark which symbols can not be internalized
303  MCContext Context(*_target->getMCAsmInfo(), *_target->getRegisterInfo(),NULL);
304  Mangler mangler(Context, *_target->getTargetData());
305  std::vector<const char*> mustPreserveList;
306  SmallPtrSet<GlobalValue*, 8> asmUsed;
307
308  for (Module::iterator f = mergedModule->begin(),
309         e = mergedModule->end(); f != e; ++f)
310    applyRestriction(*f, mustPreserveList, asmUsed, mangler);
311  for (Module::global_iterator v = mergedModule->global_begin(),
312         e = mergedModule->global_end(); v !=  e; ++v)
313    applyRestriction(*v, mustPreserveList, asmUsed, mangler);
314  for (Module::alias_iterator a = mergedModule->alias_begin(),
315         e = mergedModule->alias_end(); a != e; ++a)
316    applyRestriction(*a, mustPreserveList, asmUsed, mangler);
317
318  GlobalVariable *LLVMCompilerUsed =
319    mergedModule->getGlobalVariable("llvm.compiler.used");
320  findUsedValues(LLVMCompilerUsed, asmUsed);
321  if (LLVMCompilerUsed)
322    LLVMCompilerUsed->eraseFromParent();
323
324  llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(_context);
325  std::vector<Constant*> asmUsed2;
326  for (SmallPtrSet<GlobalValue*, 16>::const_iterator i = asmUsed.begin(),
327         e = asmUsed.end(); i !=e; ++i) {
328    GlobalValue *GV = *i;
329    Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
330    asmUsed2.push_back(c);
331  }
332
333  llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
334  LLVMCompilerUsed =
335    new llvm::GlobalVariable(*mergedModule, ATy, false,
336                             llvm::GlobalValue::AppendingLinkage,
337                             llvm::ConstantArray::get(ATy, asmUsed2),
338                             "llvm.compiler.used");
339
340  LLVMCompilerUsed->setSection("llvm.metadata");
341
342  passes.add(createInternalizePass(mustPreserveList));
343
344  // apply scope restrictions
345  passes.run(*mergedModule);
346
347  _scopeRestrictionsDone = true;
348}
349
350/// Optimize merged modules using various IPO passes
351bool LTOCodeGenerator::generateObjectFile(raw_ostream &out,
352                                          std::string &errMsg) {
353  if (this->determineTarget(errMsg))
354    return true;
355
356  Module* mergedModule = _linker.getModule();
357
358  // if options were requested, set them
359  if (!_codegenOptions.empty())
360    cl::ParseCommandLineOptions(_codegenOptions.size(),
361                                const_cast<char **>(&_codegenOptions[0]));
362
363  // mark which symbols can not be internalized
364  this->applyScopeRestrictions();
365
366  // Instantiate the pass manager to organize the passes.
367  PassManager passes;
368
369  // Start off with a verification pass.
370  passes.add(createVerifierPass());
371
372  // Add an appropriate TargetData instance for this module...
373  passes.add(new TargetData(*_target->getTargetData()));
374
375  // Enabling internalize here would use its AllButMain variant. It
376  // keeps only main if it exists and does nothing for libraries. Instead
377  // we create the pass ourselves with the symbol list provided by the linker.
378  PassManagerBuilder().populateLTOPassManager(passes, /*Internalize=*/false,
379                                              !DisableInline,
380                                              DisableGVNLoadPRE);
381
382  // Make sure everything is still good.
383  passes.add(createVerifierPass());
384
385  FunctionPassManager *codeGenPasses = new FunctionPassManager(mergedModule);
386
387  codeGenPasses->add(new TargetData(*_target->getTargetData()));
388
389  formatted_raw_ostream Out(out);
390
391  if (_target->addPassesToEmitFile(*codeGenPasses, Out,
392                                   TargetMachine::CGFT_ObjectFile)) {
393    errMsg = "target file type not supported";
394    return true;
395  }
396
397  // Run our queue of passes all at once now, efficiently.
398  passes.run(*mergedModule);
399
400  // Run the code generator, and write assembly file
401  codeGenPasses->doInitialization();
402
403  for (Module::iterator
404         it = mergedModule->begin(), e = mergedModule->end(); it != e; ++it)
405    if (!it->isDeclaration())
406      codeGenPasses->run(*it);
407
408  codeGenPasses->doFinalization();
409  delete codeGenPasses;
410
411  return false; // success
412}
413
414/// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging
415/// LTO problems.
416void LTOCodeGenerator::setCodeGenDebugOptions(const char *options) {
417  for (std::pair<StringRef, StringRef> o = getToken(options);
418       !o.first.empty(); o = getToken(o.second)) {
419    // ParseCommandLineOptions() expects argv[0] to be program name. Lazily add
420    // that.
421    if (_codegenOptions.empty())
422      _codegenOptions.push_back(strdup("libLTO"));
423    _codegenOptions.push_back(strdup(o.first.str().c_str()));
424  }
425}
426