lto.cpp revision e3611871cb5430f9c958cab1be7ef2e0778b3241 
1//===-lto.cpp - LLVM Link Time Optimizer ----------------------------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file was developed by Devang Patel and is distributed under
6// the University of Illinois Open Source 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 "llvm/Module.h"
16#include "llvm/PassManager.h"
17#include "llvm/Linker.h"
18#include "llvm/Constants.h"
19#include "llvm/DerivedTypes.h"
20#include "llvm/ModuleProvider.h"
21#include "llvm/Bitcode/ReaderWriter.h"
22#include "llvm/Support/CommandLine.h"
23#include "llvm/Support/FileUtilities.h"
24#include "llvm/Support/SystemUtils.h"
25#include "llvm/Support/Mangler.h"
26#include "llvm/Support/MemoryBuffer.h"
27#include "llvm/System/Program.h"
28#include "llvm/System/Signals.h"
29#include "llvm/Analysis/Passes.h"
30#include "llvm/Analysis/LoopPass.h"
31#include "llvm/Analysis/Verifier.h"
32#include "llvm/CodeGen/FileWriters.h"
33#include "llvm/Target/SubtargetFeature.h"
34#include "llvm/Target/TargetData.h"
35#include "llvm/Target/TargetMachine.h"
36#include "llvm/Target/TargetMachineRegistry.h"
37#include "llvm/Target/TargetAsmInfo.h"
38#include "llvm/Transforms/IPO.h"
39#include "llvm/Transforms/Scalar.h"
40#include "llvm/Analysis/LoadValueNumbering.h"
41#include "llvm/Support/MathExtras.h"
42#include "llvm/LinkTimeOptimizer.h"
43#include <fstream>
44#include <ostream>
45using namespace llvm;
46
47extern "C"
48llvm::LinkTimeOptimizer *createLLVMOptimizer(unsigned VERSION)
49{
50  if (VERSION != LLVM_LTO_VERSION)
51    return NULL;
52
53  llvm::LTO *l = new llvm::LTO();
54  return l;
55}
56
57/// If symbol is not used then make it internal and let optimizer takes
58/// care of it.
59void LLVMSymbol::mayBeNotUsed() {
60  gv->setLinkage(GlobalValue::InternalLinkage);
61}
62
63// Map LLVM LinkageType to LTO LinakgeType
64static LTOLinkageTypes
65getLTOLinkageType(GlobalValue *v)
66{
67  LTOLinkageTypes lt;
68  if (v->hasExternalLinkage())
69    lt = LTOExternalLinkage;
70  else if (v->hasLinkOnceLinkage())
71    lt = LTOLinkOnceLinkage;
72  else if (v->hasWeakLinkage())
73    lt = LTOWeakLinkage;
74  else
75    // Otherwise it is internal linkage for link time optimizer
76    lt = LTOInternalLinkage;
77  return lt;
78}
79
80// Find exeternal symbols referenced by VALUE. This is a recursive function.
81static void
82findExternalRefs(Value *value, std::set<std::string> &references,
83                 Mangler &mangler) {
84
85  if (GlobalValue *gv = dyn_cast<GlobalValue>(value)) {
86    LTOLinkageTypes lt = getLTOLinkageType(gv);
87    if (lt != LTOInternalLinkage && strncmp (gv->getName().c_str(), "llvm.", 5))
88      references.insert(mangler.getValueName(gv));
89  }
90
91  // GlobalValue, even with InternalLinkage type, may have operands with
92  // ExternalLinkage type. Do not ignore these operands.
93  if (Constant *c = dyn_cast<Constant>(value))
94    // Handle ConstantExpr, ConstantStruct, ConstantArry etc..
95    for (unsigned i = 0, e = c->getNumOperands(); i != e; ++i)
96      findExternalRefs(c->getOperand(i), references, mangler);
97}
98
99/// If Module with InputFilename is available then remove it from allModules
100/// and call delete on it.
101void
102LTO::removeModule (const std::string &InputFilename)
103{
104  NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
105  if (pos == allModules.end())
106    return;
107
108  Module *m = pos->second;
109  allModules.erase(pos);
110  delete m;
111}
112
113/// InputFilename is a LLVM bitcode file. If Module with InputFilename is
114/// available then return it. Otherwise parseInputFilename.
115Module *
116LTO::getModule(const std::string &InputFilename)
117{
118  Module *m = NULL;
119
120  NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
121  if (pos != allModules.end())
122    m = allModules[InputFilename.c_str()];
123  else {
124    if (MemoryBuffer *Buffer
125        = MemoryBuffer::getFile(&InputFilename[0], InputFilename.size())) {
126      m = ParseBitcodeFile(Buffer);
127      delete Buffer;
128    }
129    allModules[InputFilename.c_str()] = m;
130  }
131  return m;
132}
133
134/// InputFilename is a LLVM bitcode file. Reade this bitcode file and
135/// set corresponding target triplet string.
136void
137LTO::getTargetTriple(const std::string &InputFilename,
138                     std::string &targetTriple)
139{
140  Module *m = getModule(InputFilename);
141  if (m)
142    targetTriple = m->getTargetTriple();
143}
144
145/// InputFilename is a LLVM bitcode file. Read it using bitcode reader.
146/// Collect global functions and symbol names in symbols vector.
147/// Collect external references in references vector.
148/// Return LTO_READ_SUCCESS if there is no error.
149enum LTOStatus
150LTO::readLLVMObjectFile(const std::string &InputFilename,
151                        NameToSymbolMap &symbols,
152                        std::set<std::string> &references)
153{
154  Module *m = getModule(InputFilename);
155  if (!m)
156    return LTO_READ_FAILURE;
157
158  // Collect Target info
159  getTarget(m);
160
161  if (!Target)
162    return LTO_READ_FAILURE;
163
164  // Use mangler to add GlobalPrefix to names to match linker names.
165  // FIXME : Instead of hard coding "-" use GlobalPrefix.
166  Mangler mangler(*m, Target->getTargetAsmInfo()->getGlobalPrefix());
167  modules.push_back(m);
168
169  for (Module::iterator f = m->begin(), e = m->end(); f != e; ++f) {
170
171    LTOLinkageTypes lt = getLTOLinkageType(f);
172
173    if (!f->isDeclaration() && lt != LTOInternalLinkage
174        && strncmp (f->getName().c_str(), "llvm.", 5)) {
175      int alignment = ( 16 > f->getAlignment() ? 16 : f->getAlignment());
176      LLVMSymbol *newSymbol = new LLVMSymbol(lt, f, f->getName(),
177                                             mangler.getValueName(f),
178                                             Log2_32(alignment));
179      symbols[newSymbol->getMangledName()] = newSymbol;
180      allSymbols[newSymbol->getMangledName()] = newSymbol;
181    }
182
183    // Collect external symbols referenced by this function.
184    for (Function::iterator b = f->begin(), fe = f->end(); b != fe; ++b)
185      for (BasicBlock::iterator i = b->begin(), be = b->end();
186           i != be; ++i)
187        for (unsigned count = 0, total = i->getNumOperands();
188             count != total; ++count)
189          findExternalRefs(i->getOperand(count), references, mangler);
190  }
191
192  for (Module::global_iterator v = m->global_begin(), e = m->global_end();
193       v !=  e; ++v) {
194    LTOLinkageTypes lt = getLTOLinkageType(v);
195    if (!v->isDeclaration() && lt != LTOInternalLinkage
196        && strncmp (v->getName().c_str(), "llvm.", 5)) {
197      const TargetData *TD = Target->getTargetData();
198      LLVMSymbol *newSymbol = new LLVMSymbol(lt, v, v->getName(),
199                                             mangler.getValueName(v),
200                                             TD->getPreferredAlignmentLog(v));
201      symbols[newSymbol->getMangledName()] = newSymbol;
202      allSymbols[newSymbol->getMangledName()] = newSymbol;
203
204      for (unsigned count = 0, total = v->getNumOperands();
205           count != total; ++count)
206        findExternalRefs(v->getOperand(count), references, mangler);
207
208    }
209  }
210
211  return LTO_READ_SUCCESS;
212}
213
214/// Get TargetMachine.
215/// Use module M to find appropriate Target.
216void
217LTO::getTarget (Module *M) {
218
219  if (Target)
220    return;
221
222  std::string Err;
223  const TargetMachineRegistry::entry* March =
224    TargetMachineRegistry::getClosestStaticTargetForModule(*M, Err);
225
226  if (March == 0)
227    return;
228
229  // Create target
230  std::string Features;
231  Target = March->CtorFn(*M, Features);
232}
233
234/// Optimize module M using various IPO passes. Use exportList to
235/// internalize selected symbols. Target platform is selected
236/// based on information available to module M. No new target
237/// features are selected.
238enum LTOStatus
239LTO::optimize(Module *M, std::ostream &Out,
240              std::vector<const char *> &exportList)
241{
242  // Instantiate the pass manager to organize the passes.
243  PassManager Passes;
244
245  // Collect Target info
246  getTarget(M);
247
248  if (!Target)
249    return LTO_NO_TARGET;
250
251  // Start off with a verification pass.
252  Passes.add(createVerifierPass());
253
254  // Add an appropriate TargetData instance for this module...
255  Passes.add(new TargetData(*Target->getTargetData()));
256
257  // Internalize symbols if export list is nonemty
258  if (!exportList.empty())
259    Passes.add(createInternalizePass(exportList));
260
261  // Now that we internalized some globals, see if we can hack on them!
262  Passes.add(createGlobalOptimizerPass());
263
264  // Linking modules together can lead to duplicated global constants, only
265  // keep one copy of each constant...
266  Passes.add(createConstantMergePass());
267
268  // If the -s command line option was specified, strip the symbols out of the
269  // resulting program to make it smaller.  -s is a GLD option that we are
270  // supporting.
271  Passes.add(createStripSymbolsPass());
272
273  // Propagate constants at call sites into the functions they call.
274  Passes.add(createIPConstantPropagationPass());
275
276  // Remove unused arguments from functions...
277  Passes.add(createDeadArgEliminationPass());
278
279  Passes.add(createFunctionInliningPass()); // Inline small functions
280
281  Passes.add(createPruneEHPass());            // Remove dead EH info
282
283  Passes.add(createGlobalDCEPass());          // Remove dead functions
284
285  // If we didn't decide to inline a function, check to see if we can
286  // transform it to pass arguments by value instead of by reference.
287  Passes.add(createArgumentPromotionPass());
288
289  // The IPO passes may leave cruft around.  Clean up after them.
290  Passes.add(createInstructionCombiningPass());
291
292  Passes.add(createScalarReplAggregatesPass()); // Break up allocas
293
294  // Run a few AA driven optimizations here and now, to cleanup the code.
295  Passes.add(createGlobalsModRefPass());      // IP alias analysis
296
297  Passes.add(createLICMPass());               // Hoist loop invariants
298  Passes.add(createLoadValueNumberingPass()); // GVN for load instrs
299  Passes.add(createGCSEPass());               // Remove common subexprs
300  Passes.add(createDeadStoreEliminationPass()); // Nuke dead stores
301
302  // Cleanup and simplify the code after the scalar optimizations.
303  Passes.add(createInstructionCombiningPass());
304
305  // Delete basic blocks, which optimization passes may have killed...
306  Passes.add(createCFGSimplificationPass());
307
308  // Now that we have optimized the program, discard unreachable functions...
309  Passes.add(createGlobalDCEPass());
310
311  // Make sure everything is still good.
312  Passes.add(createVerifierPass());
313
314  FunctionPassManager *CodeGenPasses =
315    new FunctionPassManager(new ExistingModuleProvider(M));
316
317  CodeGenPasses->add(new TargetData(*Target->getTargetData()));
318
319  MachineCodeEmitter *MCE = 0;
320
321  switch (Target->addPassesToEmitFile(*CodeGenPasses, Out,
322                                      TargetMachine::AssemblyFile, true)) {
323  default:
324  case FileModel::Error:
325    return LTO_WRITE_FAILURE;
326  case FileModel::AsmFile:
327    break;
328  case FileModel::MachOFile:
329    MCE = AddMachOWriter(*CodeGenPasses, Out, *Target);
330    break;
331  case FileModel::ElfFile:
332    MCE = AddELFWriter(*CodeGenPasses, Out, *Target);
333    break;
334  }
335
336  if (Target->addPassesToEmitFileFinish(*CodeGenPasses, MCE, true))
337    return LTO_WRITE_FAILURE;
338
339  // Run our queue of passes all at once now, efficiently.
340  Passes.run(*M);
341
342  // Run the code generator, if present.
343  CodeGenPasses->doInitialization();
344  for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) {
345    if (!I->isDeclaration())
346      CodeGenPasses->run(*I);
347  }
348  CodeGenPasses->doFinalization();
349
350  return LTO_OPT_SUCCESS;
351}
352
353///Link all modules together and optimize them using IPO. Generate
354/// native object file using OutputFilename
355/// Return appropriate LTOStatus.
356enum LTOStatus
357LTO::optimizeModules(const std::string &OutputFilename,
358                     std::vector<const char *> &exportList,
359                     std::string &targetTriple,
360                     bool saveTemps,
361                     const char *FinalOutputFilename)
362{
363  if (modules.empty())
364    return LTO_NO_WORK;
365
366  std::ios::openmode io_mode =
367    std::ios::out | std::ios::trunc | std::ios::binary;
368  std::string *errMsg = NULL;
369  Module *bigOne = modules[0];
370  Linker theLinker("LinkTimeOptimizer", bigOne, false);
371  for (unsigned i = 1, e = modules.size(); i != e; ++i)
372    if (theLinker.LinkModules(bigOne, modules[i], errMsg))
373      return LTO_MODULE_MERGE_FAILURE;
374  //  all modules have been handed off to the linker.
375  modules.clear();
376
377  sys::Path FinalOutputPath(FinalOutputFilename);
378  FinalOutputPath.eraseSuffix();
379
380  if (saveTemps) {
381    std::string tempFileName(FinalOutputPath.c_str());
382    tempFileName += "0.bc";
383    std::ofstream Out(tempFileName.c_str(), io_mode);
384    WriteBitcodeToFile(bigOne, Out);
385  }
386
387  // Strip leading underscore because it was added to match names
388  // seen by linker.
389  for (unsigned i = 0, e = exportList.size(); i != e; ++i) {
390    const char *name = exportList[i];
391    NameToSymbolMap::iterator itr = allSymbols.find(name);
392    if (itr != allSymbols.end())
393      exportList[i] = allSymbols[name]->getName();
394  }
395
396
397  std::string ErrMsg;
398  sys::Path TempDir = sys::Path::GetTemporaryDirectory(&ErrMsg);
399  if (TempDir.isEmpty()) {
400    cerr << "lto: " << ErrMsg << "\n";
401    return LTO_WRITE_FAILURE;
402  }
403  sys::Path tmpAsmFilePath(TempDir);
404  if (!tmpAsmFilePath.appendComponent("lto")) {
405    cerr << "lto: " << ErrMsg << "\n";
406    TempDir.eraseFromDisk(true);
407    return LTO_WRITE_FAILURE;
408  }
409  if (tmpAsmFilePath.createTemporaryFileOnDisk(true, &ErrMsg)) {
410    cerr << "lto: " << ErrMsg << "\n";
411    TempDir.eraseFromDisk(true);
412    return LTO_WRITE_FAILURE;
413  }
414  sys::RemoveFileOnSignal(tmpAsmFilePath);
415
416  std::ofstream asmFile(tmpAsmFilePath.c_str(), io_mode);
417  if (!asmFile.is_open() || asmFile.bad()) {
418    if (tmpAsmFilePath.exists()) {
419      tmpAsmFilePath.eraseFromDisk();
420      TempDir.eraseFromDisk(true);
421    }
422    return LTO_WRITE_FAILURE;
423  }
424
425  enum LTOStatus status = optimize(bigOne, asmFile, exportList);
426  asmFile.close();
427  if (status != LTO_OPT_SUCCESS) {
428    tmpAsmFilePath.eraseFromDisk();
429    TempDir.eraseFromDisk(true);
430    return status;
431  }
432
433  if (saveTemps) {
434    std::string tempFileName(FinalOutputPath.c_str());
435    tempFileName += "1.bc";
436    std::ofstream Out(tempFileName.c_str(), io_mode);
437    WriteBitcodeToFile(bigOne, Out);
438  }
439
440  targetTriple = bigOne->getTargetTriple();
441
442  // Run GCC to assemble and link the program into native code.
443  //
444  // Note:
445  //  We can't just assemble and link the file with the system assembler
446  //  and linker because we don't know where to put the _start symbol.
447  //  GCC mysteriously knows how to do it.
448  const sys::Path gcc = sys::Program::FindProgramByName("gcc");
449  if (gcc.isEmpty()) {
450    tmpAsmFilePath.eraseFromDisk();
451    TempDir.eraseFromDisk(true);
452    return LTO_ASM_FAILURE;
453  }
454
455  std::vector<const char*> args;
456  args.push_back(gcc.c_str());
457  args.push_back("-c");
458  args.push_back("-x");
459  args.push_back("assembler");
460  args.push_back("-o");
461  args.push_back(OutputFilename.c_str());
462  args.push_back(tmpAsmFilePath.c_str());
463  args.push_back(0);
464
465  if (sys::Program::ExecuteAndWait(gcc, &args[0], 0, 0, 1, 0, &ErrMsg)) {
466    cerr << "lto: " << ErrMsg << "\n";
467    return LTO_ASM_FAILURE;
468  }
469
470  tmpAsmFilePath.eraseFromDisk();
471  TempDir.eraseFromDisk(true);
472
473  return LTO_OPT_SUCCESS;
474}
475
476void LTO::printVersion() {
477    cl::PrintVersionMessage();
478}
479
480/// Unused pure-virtual destructor. Must remain empty.
481LinkTimeOptimizer::~LinkTimeOptimizer() {}
482
483/// Destruct LTO. Delete all modules, symbols and target.
484LTO::~LTO() {
485
486  for (std::vector<Module *>::iterator itr = modules.begin(), e = modules.end();
487       itr != e; ++itr)
488    delete *itr;
489
490  modules.clear();
491
492  for (NameToSymbolMap::iterator itr = allSymbols.begin(), e = allSymbols.end();
493       itr != e; ++itr)
494    delete itr->second;
495
496  allSymbols.clear();
497
498  delete Target;
499}
500