LTOCodeGenerator.cpp revision 439661395fd2a2a832dba01c65bc88718528313c 
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#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/ADT/StringExtras.h"
24#include "llvm/ADT/Triple.h"
25#include "llvm/Analysis/Passes.h"
26#include "llvm/Bitcode/ReaderWriter.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/Target/TargetRegistry.h"
36#include "llvm/Target/TargetSelect.h"
37#include "llvm/Support/CommandLine.h"
38#include "llvm/Support/FormattedStream.h"
39#include "llvm/Support/MemoryBuffer.h"
40#include "llvm/Support/PassManagerBuilder.h"
41#include "llvm/Support/SystemUtils.h"
42#include "llvm/Support/ToolOutputFile.h"
43#include "llvm/Support/Host.h"
44#include "llvm/Support/Program.h"
45#include "llvm/Support/Signals.h"
46#include "llvm/Support/system_error.h"
47#include "llvm/Config/config.h"
48#include <cstdlib>
49#include <unistd.h>
50#include <fcntl.h>
51
52
53using namespace llvm;
54
55static cl::opt<bool> DisableInline("disable-inlining",
56  cl::desc("Do not run the inliner pass"));
57
58
59const char* LTOCodeGenerator::getVersionString()
60{
61#ifdef LLVM_VERSION_INFO
62    return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
63#else
64    return PACKAGE_NAME " version " PACKAGE_VERSION;
65#endif
66}
67
68
69LTOCodeGenerator::LTOCodeGenerator()
70    : _context(getGlobalContext()),
71      _linker("LinkTimeOptimizer", "ld-temp.o", _context), _target(NULL),
72      _emitDwarfDebugInfo(false), _scopeRestrictionsDone(false),
73      _codeModel(LTO_CODEGEN_PIC_MODEL_DYNAMIC),
74      _nativeObjectFile(NULL)
75{
76    InitializeAllTargets();
77    InitializeAllMCAsmInfos();
78    InitializeAllMCCodeGenInfos();
79    InitializeAllMCRegisterInfos();
80    InitializeAllMCSubtargetInfos();
81    InitializeAllAsmPrinters();
82}
83
84LTOCodeGenerator::~LTOCodeGenerator()
85{
86    delete _target;
87    delete _nativeObjectFile;
88}
89
90
91
92bool LTOCodeGenerator::addModule(LTOModule* mod, std::string& errMsg)
93{
94
95  if(mod->getLLVVMModule()->MaterializeAllPermanently(&errMsg))
96    return true;
97
98  bool ret = _linker.LinkInModule(mod->getLLVVMModule(), &errMsg);
99
100  const std::vector<const char*> &undefs = mod->getAsmUndefinedRefs();
101  for (int i = 0, e = undefs.size(); i != e; ++i)
102    _asmUndefinedRefs[undefs[i]] = 1;
103
104  return ret;
105}
106
107
108bool LTOCodeGenerator::setDebugInfo(lto_debug_model debug, std::string& errMsg)
109{
110    switch (debug) {
111        case LTO_DEBUG_MODEL_NONE:
112            _emitDwarfDebugInfo = false;
113            return false;
114
115        case LTO_DEBUG_MODEL_DWARF:
116            _emitDwarfDebugInfo = true;
117            return false;
118    }
119    errMsg = "unknown debug format";
120    return true;
121}
122
123
124bool LTOCodeGenerator::setCodePICModel(lto_codegen_model model,
125                                       std::string& errMsg)
126{
127    switch (model) {
128        case LTO_CODEGEN_PIC_MODEL_STATIC:
129        case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
130        case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
131            _codeModel = model;
132            return false;
133    }
134    errMsg = "unknown pic model";
135    return true;
136}
137
138void LTOCodeGenerator::setCpu(const char* mCpu)
139{
140  _mCpu = mCpu;
141}
142
143void LTOCodeGenerator::addMustPreserveSymbol(const char* sym)
144{
145    _mustPreserveSymbols[sym] = 1;
146}
147
148
149bool LTOCodeGenerator::writeMergedModules(const char *path,
150                                          std::string &errMsg) {
151  if (determineTarget(errMsg))
152    return true;
153
154  // mark which symbols can not be internalized
155  applyScopeRestrictions();
156
157  // create output file
158  std::string ErrInfo;
159  tool_output_file Out(path, ErrInfo,
160                       raw_fd_ostream::F_Binary);
161  if (!ErrInfo.empty()) {
162    errMsg = "could not open bitcode file for writing: ";
163    errMsg += path;
164    return true;
165  }
166
167  // write bitcode to it
168  WriteBitcodeToFile(_linker.getModule(), Out.os());
169  Out.os().close();
170
171  if (Out.os().has_error()) {
172    errMsg = "could not write bitcode file: ";
173    errMsg += path;
174    Out.os().clear_error();
175    return true;
176  }
177
178  Out.keep();
179  return false;
180}
181
182
183bool LTOCodeGenerator::compile_to_file(const char** name, std::string& errMsg)
184{
185  // make unique temp .o file to put generated object file
186  sys::PathWithStatus uniqueObjPath("lto-llvm.o");
187  if ( uniqueObjPath.createTemporaryFileOnDisk(false, &errMsg) ) {
188    uniqueObjPath.eraseFromDisk();
189    return true;
190  }
191  sys::RemoveFileOnSignal(uniqueObjPath);
192
193  // generate object file
194  bool genResult = false;
195  tool_output_file objFile(uniqueObjPath.c_str(), errMsg);
196  if (!errMsg.empty())
197    return NULL;
198  genResult = this->generateObjectFile(objFile.os(), errMsg);
199  objFile.os().close();
200  if (objFile.os().has_error()) {
201    objFile.os().clear_error();
202    return true;
203  }
204  objFile.keep();
205  if ( genResult ) {
206    uniqueObjPath.eraseFromDisk();
207    return true;
208  }
209
210  _nativeObjectPath = uniqueObjPath.str();
211  *name = _nativeObjectPath.c_str();
212  return false;
213}
214
215const void* LTOCodeGenerator::compile(size_t* length, std::string& errMsg)
216{
217  const char *name;
218  if (compile_to_file(&name, errMsg))
219    return NULL;
220
221  // remove old buffer if compile() called twice
222  delete _nativeObjectFile;
223
224  // read .o file into memory buffer
225  OwningPtr<MemoryBuffer> BuffPtr;
226  if (error_code ec = MemoryBuffer::getFile(name, BuffPtr, -1, false)) {
227    errMsg = ec.message();
228    return NULL;
229  }
230  _nativeObjectFile = BuffPtr.take();
231
232  // remove temp files
233  sys::Path(_nativeObjectPath).eraseFromDisk();
234
235  // return buffer, unless error
236  if ( _nativeObjectFile == NULL )
237    return NULL;
238  *length = _nativeObjectFile->getBufferSize();
239  return _nativeObjectFile->getBufferStart();
240}
241
242bool LTOCodeGenerator::determineTarget(std::string& errMsg)
243{
244    if ( _target == NULL ) {
245        std::string Triple = _linker.getModule()->getTargetTriple();
246        if (Triple.empty())
247          Triple = sys::getHostTriple();
248
249        // create target machine from info for merged modules
250        const Target *march = TargetRegistry::lookupTarget(Triple, errMsg);
251        if ( march == NULL )
252            return true;
253
254        // The relocation model is actually a static member of TargetMachine
255        // and needs to be set before the TargetMachine is instantiated.
256        Reloc::Model RelocModel = Reloc::Default;
257        switch( _codeModel ) {
258        case LTO_CODEGEN_PIC_MODEL_STATIC:
259            RelocModel = Reloc::Static;
260            break;
261        case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
262            RelocModel = Reloc::PIC_;
263            break;
264        case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
265            RelocModel = Reloc::DynamicNoPIC;
266            break;
267        }
268
269        // construct LTModule, hand over ownership of module and target
270        SubtargetFeatures Features;
271        Features.getDefaultSubtargetFeatures(llvm::Triple(Triple));
272        std::string FeatureStr = Features.getString();
273        _target = march->createTargetMachine(Triple, _mCpu, FeatureStr,
274                                             RelocModel);
275    }
276    return false;
277}
278
279void LTOCodeGenerator::applyRestriction(GlobalValue &GV,
280                                     std::vector<const char*> &mustPreserveList,
281                                        SmallPtrSet<GlobalValue*, 8> &asmUsed,
282                                        Mangler &mangler) {
283  SmallString<64> Buffer;
284  mangler.getNameWithPrefix(Buffer, &GV, false);
285
286  if (GV.isDeclaration())
287    return;
288  if (_mustPreserveSymbols.count(Buffer))
289    mustPreserveList.push_back(GV.getName().data());
290  if (_asmUndefinedRefs.count(Buffer))
291    asmUsed.insert(&GV);
292}
293
294static void findUsedValues(GlobalVariable *LLVMUsed,
295                           SmallPtrSet<GlobalValue*, 8> &UsedValues) {
296  if (LLVMUsed == 0) return;
297
298  ConstantArray *Inits = dyn_cast<ConstantArray>(LLVMUsed->getInitializer());
299  if (Inits == 0) return;
300
301  for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
302    if (GlobalValue *GV =
303          dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
304      UsedValues.insert(GV);
305}
306
307void LTOCodeGenerator::applyScopeRestrictions() {
308  if (_scopeRestrictionsDone) return;
309  Module *mergedModule = _linker.getModule();
310
311  // Start off with a verification pass.
312  PassManager passes;
313  passes.add(createVerifierPass());
314
315  // mark which symbols can not be internalized
316  MCContext Context(*_target->getMCAsmInfo(), *_target->getRegisterInfo(),
317                    NULL);
318  Mangler mangler(Context, *_target->getTargetData());
319  std::vector<const char*> mustPreserveList;
320  SmallPtrSet<GlobalValue*, 8> asmUsed;
321
322  for (Module::iterator f = mergedModule->begin(),
323         e = mergedModule->end(); f != e; ++f)
324    applyRestriction(*f, mustPreserveList, asmUsed, mangler);
325  for (Module::global_iterator v = mergedModule->global_begin(),
326         e = mergedModule->global_end(); v !=  e; ++v)
327    applyRestriction(*v, mustPreserveList, asmUsed, mangler);
328  for (Module::alias_iterator a = mergedModule->alias_begin(),
329         e = mergedModule->alias_end(); a != e; ++a)
330    applyRestriction(*a, mustPreserveList, asmUsed, mangler);
331
332  GlobalVariable *LLVMCompilerUsed =
333    mergedModule->getGlobalVariable("llvm.compiler.used");
334  findUsedValues(LLVMCompilerUsed, asmUsed);
335  if (LLVMCompilerUsed)
336    LLVMCompilerUsed->eraseFromParent();
337
338  llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(_context);
339  std::vector<Constant*> asmUsed2;
340  for (SmallPtrSet<GlobalValue*, 16>::const_iterator i = asmUsed.begin(),
341         e = asmUsed.end(); i !=e; ++i) {
342    GlobalValue *GV = *i;
343    Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
344    asmUsed2.push_back(c);
345  }
346
347  llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
348  LLVMCompilerUsed =
349    new llvm::GlobalVariable(*mergedModule, ATy, false,
350                             llvm::GlobalValue::AppendingLinkage,
351                             llvm::ConstantArray::get(ATy, asmUsed2),
352                             "llvm.compiler.used");
353
354  LLVMCompilerUsed->setSection("llvm.metadata");
355
356  passes.add(createInternalizePass(mustPreserveList));
357
358  // apply scope restrictions
359  passes.run(*mergedModule);
360
361  _scopeRestrictionsDone = true;
362}
363
364/// Optimize merged modules using various IPO passes
365bool LTOCodeGenerator::generateObjectFile(raw_ostream &out,
366                                          std::string &errMsg) {
367    if ( this->determineTarget(errMsg) )
368        return true;
369
370    // mark which symbols can not be internalized
371    this->applyScopeRestrictions();
372
373    Module* mergedModule = _linker.getModule();
374
375    // if options were requested, set them
376    if ( !_codegenOptions.empty() )
377        cl::ParseCommandLineOptions(_codegenOptions.size(),
378                                    const_cast<char **>(&_codegenOptions[0]));
379
380    // Instantiate the pass manager to organize the passes.
381    PassManager passes;
382
383    // Start off with a verification pass.
384    passes.add(createVerifierPass());
385
386    // Add an appropriate TargetData instance for this module...
387    passes.add(new TargetData(*_target->getTargetData()));
388
389    PassManagerBuilder().populateLTOPassManager(passes, /*Internalize=*/ false,
390                                                !DisableInline);
391
392    // Make sure everything is still good.
393    passes.add(createVerifierPass());
394
395    FunctionPassManager *codeGenPasses = new FunctionPassManager(mergedModule);
396
397    codeGenPasses->add(new TargetData(*_target->getTargetData()));
398
399    formatted_raw_ostream Out(out);
400
401    if (_target->addPassesToEmitFile(*codeGenPasses, Out,
402                                     TargetMachine::CGFT_ObjectFile,
403                                     CodeGenOpt::Aggressive)) {
404      errMsg = "target file type not supported";
405      return true;
406    }
407
408    // Run our queue of passes all at once now, efficiently.
409    passes.run(*mergedModule);
410
411    // Run the code generator, and write assembly file
412    codeGenPasses->doInitialization();
413
414    for (Module::iterator
415           it = mergedModule->begin(), e = mergedModule->end(); it != e; ++it)
416      if (!it->isDeclaration())
417        codeGenPasses->run(*it);
418
419    codeGenPasses->doFinalization();
420    delete codeGenPasses;
421
422    return false; // success
423}
424
425
426/// Optimize merged modules using various IPO passes
427void LTOCodeGenerator::setCodeGenDebugOptions(const char* options)
428{
429    for (std::pair<StringRef, StringRef> o = getToken(options);
430         !o.first.empty(); o = getToken(o.second)) {
431        // ParseCommandLineOptions() expects argv[0] to be program name.
432        // Lazily add that.
433        if ( _codegenOptions.empty() )
434            _codegenOptions.push_back("libLTO");
435        _codegenOptions.push_back(strdup(o.first.str().c_str()));
436    }
437}
438