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