lto.cpp revision ef9b9a793949469cdaa4ab6d0173136229dcab7b 
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/Bytecode/Reader.h"
21#include "llvm/Bytecode/Writer.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/System/Program.h"
27#include "llvm/System/Signals.h"
28#include "llvm/Analysis/Passes.h"
29#include "llvm/Analysis/Verifier.h"
30#include "llvm/Target/SubtargetFeature.h"
31#include "llvm/Target/TargetData.h"
32#include "llvm/Target/TargetMachine.h"
33#include "llvm/Target/TargetMachineRegistry.h"
34#include "llvm/Target/TargetAsmInfo.h"
35#include "llvm/Transforms/IPO.h"
36#include "llvm/Transforms/Scalar.h"
37#include "llvm/Analysis/LoadValueNumbering.h"
38#include "llvm/Support/MathExtras.h"
39#include "llvm/Support/Streams.h"
40#include "llvm/LinkTimeOptimizer.h"
41#include <fstream>
42#include <ostream>
43using namespace llvm;
44
45extern "C"
46llvm::LinkTimeOptimizer *createLLVMOptimizer()
47{
48  llvm::LTO *l = new llvm::LTO();
49  return l;
50}
51
52
53
54/// If symbol is not used then make it internal and let optimizer takes
55/// care of it.
56void LLVMSymbol::mayBeNotUsed() {
57  gv->setLinkage(GlobalValue::InternalLinkage);
58}
59
60// Map LLVM LinkageType to LTO LinakgeType
61static LTOLinkageTypes
62getLTOLinkageType(GlobalValue *v)
63{
64  LTOLinkageTypes lt;
65  if (v->hasExternalLinkage())
66    lt = LTOExternalLinkage;
67  else if (v->hasLinkOnceLinkage())
68    lt = LTOLinkOnceLinkage;
69  else if (v->hasWeakLinkage())
70    lt = LTOWeakLinkage;
71  else
72    // Otherwise it is internal linkage for link time optimizer
73    lt = LTOInternalLinkage;
74  return lt;
75}
76
77// Find exeternal symbols referenced by VALUE. This is a recursive function.
78static void
79findExternalRefs(Value *value, std::set<std::string> &references,
80                 Mangler &mangler) {
81
82  if (GlobalValue *gv = dyn_cast<GlobalValue>(value)) {
83    LTOLinkageTypes lt = getLTOLinkageType(gv);
84    if (lt != LTOInternalLinkage && strncmp (gv->getName().c_str(), "llvm.", 5))
85      references.insert(mangler.getValueName(gv));
86  }
87
88  // GlobalValue, even with InternalLinkage type, may have operands with
89  // ExternalLinkage type. Do not ignore these operands.
90  if (Constant *c = dyn_cast<Constant>(value))
91    // Handle ConstantExpr, ConstantStruct, ConstantArry etc..
92    for (unsigned i = 0, e = c->getNumOperands(); i != e; ++i)
93      findExternalRefs(c->getOperand(i), references, mangler);
94}
95
96/// If Module with InputFilename is available then remove it from allModules
97/// and call delete on it.
98void
99LTO::removeModule (const std::string &InputFilename)
100{
101  NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
102  if (pos == allModules.end())
103    return;
104
105  Module *m = pos->second;
106  allModules.erase(pos);
107  delete m;
108}
109
110/// InputFilename is a LLVM bytecode file. If Module with InputFilename is
111/// available then return it. Otherwise parseInputFilename.
112Module *
113LTO::getModule(const std::string &InputFilename)
114{
115  Module *m = NULL;
116
117  NameToModuleMap::iterator pos = allModules.find(InputFilename.c_str());
118  if (pos != allModules.end())
119    m = allModules[InputFilename.c_str()];
120  else {
121    m = ParseBytecodeFile(InputFilename);
122    allModules[InputFilename.c_str()] = m;
123  }
124  return m;
125}
126
127/// InputFilename is a LLVM bytecode file. Reade this bytecode file and
128/// set corresponding target triplet string.
129void
130LTO::getTargetTriple(const std::string &InputFilename,
131                     std::string &targetTriple)
132{
133  Module *m = getModule(InputFilename);
134  if (m)
135    targetTriple = m->getTargetTriple();
136}
137
138/// InputFilename is a LLVM bytecode file. Read it using bytecode reader.
139/// Collect global functions and symbol names in symbols vector.
140/// Collect external references in references vector.
141/// Return LTO_READ_SUCCESS if there is no error.
142enum LTOStatus
143LTO::readLLVMObjectFile(const std::string &InputFilename,
144                        NameToSymbolMap &symbols,
145                        std::set<std::string> &references)
146{
147  Module *m = getModule(InputFilename);
148  if (!m)
149    return LTO_READ_FAILURE;
150
151  // Collect Target info
152  getTarget(m);
153
154  if (!Target)
155    return LTO_READ_FAILURE;
156
157  // Use mangler to add GlobalPrefix to names to match linker names.
158  // FIXME : Instead of hard coding "-" use GlobalPrefix.
159  Mangler mangler(*m, Target->getTargetAsmInfo()->getGlobalPrefix());
160  modules.push_back(m);
161
162  for (Module::iterator f = m->begin(), e = m->end(); f != e; ++f) {
163
164    LTOLinkageTypes lt = getLTOLinkageType(f);
165
166    if (!f->isDeclaration() && lt != LTOInternalLinkage
167        && strncmp (f->getName().c_str(), "llvm.", 5)) {
168      int alignment = ( 16 > f->getAlignment() ? 16 : f->getAlignment());
169      LLVMSymbol *newSymbol = new LLVMSymbol(lt, f, f->getName(),
170                                             mangler.getValueName(f),
171                                             Log2_32(alignment));
172      symbols[newSymbol->getMangledName()] = newSymbol;
173      allSymbols[newSymbol->getMangledName()] = newSymbol;
174    }
175
176    // Collect external symbols referenced by this function.
177    for (Function::iterator b = f->begin(), fe = f->end(); b != fe; ++b)
178      for (BasicBlock::iterator i = b->begin(), be = b->end();
179           i != be; ++i)
180        for (unsigned count = 0, total = i->getNumOperands();
181             count != total; ++count)
182          findExternalRefs(i->getOperand(count), references, mangler);
183  }
184
185  for (Module::global_iterator v = m->global_begin(), e = m->global_end();
186       v !=  e; ++v) {
187    LTOLinkageTypes lt = getLTOLinkageType(v);
188    if (!v->isDeclaration() && lt != LTOInternalLinkage
189        && strncmp (v->getName().c_str(), "llvm.", 5)) {
190      const TargetData *TD = Target->getTargetData();
191      LLVMSymbol *newSymbol = new LLVMSymbol(lt, v, v->getName(),
192                                             mangler.getValueName(v),
193                                             TD->getPreferredAlignmentLog(v));
194      symbols[newSymbol->getMangledName()] = newSymbol;
195      allSymbols[newSymbol->getMangledName()] = newSymbol;
196
197      for (unsigned count = 0, total = v->getNumOperands();
198           count != total; ++count)
199        findExternalRefs(v->getOperand(count), references, mangler);
200
201    }
202  }
203
204  return LTO_READ_SUCCESS;
205}
206
207/// Get TargetMachine.
208/// Use module M to find appropriate Target.
209void
210LTO::getTarget (Module *M) {
211
212  if (Target)
213    return;
214
215  std::string Err;
216  const TargetMachineRegistry::Entry* March =
217    TargetMachineRegistry::getClosestStaticTargetForModule(*M, Err);
218
219  if (March == 0)
220    return;
221
222  // Create target
223  std::string Features;
224  Target = March->CtorFn(*M, Features);
225}
226
227/// Optimize module M using various IPO passes. Use exportList to
228/// internalize selected symbols. Target platform is selected
229/// based on information available to module M. No new target
230/// features are selected.
231enum LTOStatus
232LTO::optimize(Module *M, std::ostream &Out,
233              std::vector<const char *> &exportList)
234{
235  // Instantiate the pass manager to organize the passes.
236  PassManager Passes;
237
238  // Collect Target info
239  getTarget(M);
240
241  if (!Target)
242    return LTO_NO_TARGET;
243
244  // Start off with a verification pass.
245  Passes.add(createVerifierPass());
246
247  // Add an appropriate TargetData instance for this module...
248  Passes.add(new TargetData(*Target->getTargetData()));
249
250  // Internalize symbols if export list is nonemty
251  if (!exportList.empty())
252    Passes.add(createInternalizePass(exportList));
253
254  // Now that we internalized some globals, see if we can hack on them!
255  Passes.add(createGlobalOptimizerPass());
256
257  // Linking modules together can lead to duplicated global constants, only
258  // keep one copy of each constant...
259  Passes.add(createConstantMergePass());
260
261  // If the -s command line option was specified, strip the symbols out of the
262  // resulting program to make it smaller.  -s is a GLD option that we are
263  // supporting.
264  Passes.add(createStripSymbolsPass());
265
266  // Propagate constants at call sites into the functions they call.
267  Passes.add(createIPConstantPropagationPass());
268
269  // Remove unused arguments from functions...
270  Passes.add(createDeadArgEliminationPass());
271
272  Passes.add(createFunctionInliningPass()); // Inline small functions
273
274  Passes.add(createPruneEHPass());            // Remove dead EH info
275
276  Passes.add(createGlobalDCEPass());          // Remove dead functions
277
278  // If we didn't decide to inline a function, check to see if we can
279  // transform it to pass arguments by value instead of by reference.
280  Passes.add(createArgumentPromotionPass());
281
282  // The IPO passes may leave cruft around.  Clean up after them.
283  Passes.add(createInstructionCombiningPass());
284
285  Passes.add(createScalarReplAggregatesPass()); // Break up allocas
286
287  // Run a few AA driven optimizations here and now, to cleanup the code.
288  Passes.add(createGlobalsModRefPass());      // IP alias analysis
289
290  Passes.add(createLICMPass());               // Hoist loop invariants
291  Passes.add(createLoadValueNumberingPass()); // GVN for load instrs
292  Passes.add(createGCSEPass());               // Remove common subexprs
293  Passes.add(createDeadStoreEliminationPass()); // Nuke dead stores
294
295  // Cleanup and simplify the code after the scalar optimizations.
296  Passes.add(createInstructionCombiningPass());
297
298  // Delete basic blocks, which optimization passes may have killed...
299  Passes.add(createCFGSimplificationPass());
300
301  // Now that we have optimized the program, discard unreachable functions...
302  Passes.add(createGlobalDCEPass());
303
304  // Make sure everything is still good.
305  Passes.add(createVerifierPass());
306
307  FunctionPassManager *CodeGenPasses =
308    new FunctionPassManager(new ExistingModuleProvider(M));
309
310  CodeGenPasses->add(new TargetData(*Target->getTargetData()));
311  Target->addPassesToEmitFile(*CodeGenPasses, Out, TargetMachine::AssemblyFile,
312                              true);
313
314  // Run our queue of passes all at once now, efficiently.
315  Passes.run(*M);
316
317  // Run the code generator, if present.
318  CodeGenPasses->doInitialization();
319  for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) {
320    if (!I->isDeclaration())
321      CodeGenPasses->run(*I);
322  }
323  CodeGenPasses->doFinalization();
324
325  return LTO_OPT_SUCCESS;
326}
327
328///Link all modules together and optimize them using IPO. Generate
329/// native object file using OutputFilename
330/// Return appropriate LTOStatus.
331enum LTOStatus
332LTO::optimizeModules(const std::string &OutputFilename,
333                     std::vector<const char *> &exportList,
334                     std::string &targetTriple,
335                     bool saveTemps,
336                     const char *FinalOutputFilename)
337{
338  if (modules.empty())
339    return LTO_NO_WORK;
340
341  std::ios::openmode io_mode =
342    std::ios::out | std::ios::trunc | std::ios::binary;
343  std::string *errMsg = NULL;
344  Module *bigOne = modules[0];
345  Linker theLinker("LinkTimeOptimizer", bigOne, false);
346  for (unsigned i = 1, e = modules.size(); i != e; ++i)
347    if (theLinker.LinkModules(bigOne, modules[i], errMsg))
348      return LTO_MODULE_MERGE_FAILURE;
349  //  all modules have been handed off to the linker.
350  modules.clear();
351
352  sys::Path FinalOutputPath(FinalOutputFilename);
353  FinalOutputPath.eraseSuffix();
354
355  if (saveTemps) {
356    std::string tempFileName(FinalOutputPath.c_str());
357    tempFileName += "0.bc";
358    std::ofstream Out(tempFileName.c_str(), io_mode);
359    OStream L(Out);
360    WriteBytecodeToFile(bigOne, L);
361  }
362
363  // Strip leading underscore because it was added to match names
364  // seen by linker.
365  for (unsigned i = 0, e = exportList.size(); i != e; ++i) {
366    const char *name = exportList[i];
367    NameToSymbolMap::iterator itr = allSymbols.find(name);
368    if (itr != allSymbols.end())
369      exportList[i] = allSymbols[name]->getName();
370  }
371
372
373  std::string ErrMsg;
374  sys::Path TempDir = sys::Path::GetTemporaryDirectory(&ErrMsg);
375  if (TempDir.isEmpty()) {
376    cerr << "lto: " << ErrMsg << "\n";
377    return LTO_WRITE_FAILURE;
378  }
379  sys::Path tmpAsmFilePath(TempDir);
380  if (!tmpAsmFilePath.appendComponent("lto")) {
381    cerr << "lto: " << ErrMsg << "\n";
382    TempDir.eraseFromDisk(true);
383    return LTO_WRITE_FAILURE;
384  }
385  if (tmpAsmFilePath.createTemporaryFileOnDisk(&ErrMsg)) {
386    cerr << "lto: " << ErrMsg << "\n";
387    TempDir.eraseFromDisk(true);
388    return LTO_WRITE_FAILURE;
389  }
390  sys::RemoveFileOnSignal(tmpAsmFilePath);
391
392  std::ofstream asmFile(tmpAsmFilePath.c_str(), io_mode);
393  if (!asmFile.is_open() || asmFile.bad()) {
394    if (tmpAsmFilePath.exists()) {
395      tmpAsmFilePath.eraseFromDisk();
396      TempDir.eraseFromDisk(true);
397    }
398    return LTO_WRITE_FAILURE;
399  }
400
401  enum LTOStatus status = optimize(bigOne, asmFile, exportList);
402  asmFile.close();
403  if (status != LTO_OPT_SUCCESS) {
404    tmpAsmFilePath.eraseFromDisk();
405    TempDir.eraseFromDisk(true);
406    return status;
407  }
408
409  if (saveTemps) {
410    std::string tempFileName(FinalOutputPath.c_str());
411    tempFileName += "1.bc";
412    std::ofstream Out(tempFileName.c_str(), io_mode);
413    OStream L(Out);
414    WriteBytecodeToFile(bigOne, L);
415  }
416
417  targetTriple = bigOne->getTargetTriple();
418
419  // Run GCC to assemble and link the program into native code.
420  //
421  // Note:
422  //  We can't just assemble and link the file with the system assembler
423  //  and linker because we don't know where to put the _start symbol.
424  //  GCC mysteriously knows how to do it.
425  const sys::Path gcc = sys::Program::FindProgramByName("gcc");
426  if (gcc.isEmpty()) {
427    tmpAsmFilePath.eraseFromDisk();
428    TempDir.eraseFromDisk(true);
429    return LTO_ASM_FAILURE;
430  }
431
432  std::vector<const char*> args;
433  args.push_back(gcc.c_str());
434  args.push_back("-c");
435  args.push_back("-x");
436  args.push_back("assembler");
437  args.push_back("-o");
438  args.push_back(OutputFilename.c_str());
439  args.push_back(tmpAsmFilePath.c_str());
440  args.push_back(0);
441
442  if (sys::Program::ExecuteAndWait(gcc, &args[0], 0, 0, 1, &ErrMsg)) {
443    cerr << "lto: " << ErrMsg << "\n";
444    return LTO_ASM_FAILURE;
445  }
446
447  tmpAsmFilePath.eraseFromDisk();
448  TempDir.eraseFromDisk(true);
449
450  return LTO_OPT_SUCCESS;
451}
452
453void LTO::printVersion() {
454    cl::PrintVersionMessage();
455}
456
457/// Unused pure-virtual destructor. Must remain empty.
458LinkTimeOptimizer::~LinkTimeOptimizer() {}
459
460/// Destruct LTO. Delete all modules, symbols and target.
461LTO::~LTO() {
462
463  for (std::vector<Module *>::iterator itr = modules.begin(), e = modules.end();
464       itr != e; ++itr)
465    delete *itr;
466
467  modules.clear();
468
469  for (NameToSymbolMap::iterator itr = allSymbols.begin(), e = allSymbols.end();
470       itr != e; ++itr)
471    delete itr->second;
472
473  allSymbols.clear();
474
475  delete Target;
476}
477