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 "llvm/LTO/LTOCodeGenerator.h"
16#include "llvm/ADT/StringExtras.h"
17#include "llvm/Analysis/Passes.h"
18#include "llvm/Bitcode/ReaderWriter.h"
19#include "llvm/CodeGen/RuntimeLibcalls.h"
20#include "llvm/Config/config.h"
21#include "llvm/IR/Constants.h"
22#include "llvm/IR/DataLayout.h"
23#include "llvm/IR/DerivedTypes.h"
24#include "llvm/IR/DiagnosticInfo.h"
25#include "llvm/IR/DiagnosticPrinter.h"
26#include "llvm/IR/LLVMContext.h"
27#include "llvm/IR/Mangler.h"
28#include "llvm/IR/Module.h"
29#include "llvm/IR/Verifier.h"
30#include "llvm/InitializePasses.h"
31#include "llvm/LTO/LTOModule.h"
32#include "llvm/Linker/Linker.h"
33#include "llvm/MC/MCAsmInfo.h"
34#include "llvm/MC/MCContext.h"
35#include "llvm/MC/SubtargetFeature.h"
36#include "llvm/PassManager.h"
37#include "llvm/Support/CommandLine.h"
38#include "llvm/Support/FileSystem.h"
39#include "llvm/Support/FormattedStream.h"
40#include "llvm/Support/Host.h"
41#include "llvm/Support/MemoryBuffer.h"
42#include "llvm/Support/Signals.h"
43#include "llvm/Support/TargetRegistry.h"
44#include "llvm/Support/TargetSelect.h"
45#include "llvm/Support/ToolOutputFile.h"
46#include "llvm/Support/raw_ostream.h"
47#include "llvm/Target/TargetLibraryInfo.h"
48#include "llvm/Target/TargetLowering.h"
49#include "llvm/Target/TargetOptions.h"
50#include "llvm/Target/TargetRegisterInfo.h"
51#include "llvm/Transforms/IPO.h"
52#include "llvm/Transforms/IPO/PassManagerBuilder.h"
53#include "llvm/Transforms/ObjCARC.h"
54#include <system_error>
55using namespace llvm;
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()), IRLinker(new Module("ld-temp.o", Context)),
67      TargetMach(nullptr), EmitDwarfDebugInfo(false),
68      ScopeRestrictionsDone(false), CodeModel(LTO_CODEGEN_PIC_MODEL_DEFAULT),
69      NativeObjectFile(nullptr), DiagHandler(nullptr), DiagContext(nullptr) {
70  initializeLTOPasses();
71}
72
73LTOCodeGenerator::~LTOCodeGenerator() {
74  delete TargetMach;
75  delete NativeObjectFile;
76  TargetMach = nullptr;
77  NativeObjectFile = nullptr;
78
79  IRLinker.deleteModule();
80
81  for (std::vector<char *>::iterator I = CodegenOptions.begin(),
82                                     E = CodegenOptions.end();
83       I != E; ++I)
84    free(*I);
85}
86
87// Initialize LTO passes. Please keep this funciton in sync with
88// PassManagerBuilder::populateLTOPassManager(), and make sure all LTO
89// passes are initialized.
90void LTOCodeGenerator::initializeLTOPasses() {
91  PassRegistry &R = *PassRegistry::getPassRegistry();
92
93  initializeInternalizePassPass(R);
94  initializeIPSCCPPass(R);
95  initializeGlobalOptPass(R);
96  initializeConstantMergePass(R);
97  initializeDAHPass(R);
98  initializeInstCombinerPass(R);
99  initializeSimpleInlinerPass(R);
100  initializePruneEHPass(R);
101  initializeGlobalDCEPass(R);
102  initializeArgPromotionPass(R);
103  initializeJumpThreadingPass(R);
104  initializeSROAPass(R);
105  initializeSROA_DTPass(R);
106  initializeSROA_SSAUpPass(R);
107  initializeFunctionAttrsPass(R);
108  initializeGlobalsModRefPass(R);
109  initializeLICMPass(R);
110  initializeGVNPass(R);
111  initializeMemCpyOptPass(R);
112  initializeDCEPass(R);
113  initializeCFGSimplifyPassPass(R);
114}
115
116bool LTOCodeGenerator::addModule(LTOModule* mod, std::string& errMsg) {
117  bool ret = IRLinker.linkInModule(&mod->getModule(), &errMsg);
118
119  const std::vector<const char*> &undefs = mod->getAsmUndefinedRefs();
120  for (int i = 0, e = undefs.size(); i != e; ++i)
121    AsmUndefinedRefs[undefs[i]] = 1;
122
123  return !ret;
124}
125
126void LTOCodeGenerator::setTargetOptions(TargetOptions options) {
127  Options = options;
128}
129
130void LTOCodeGenerator::setDebugInfo(lto_debug_model debug) {
131  switch (debug) {
132  case LTO_DEBUG_MODEL_NONE:
133    EmitDwarfDebugInfo = false;
134    return;
135
136  case LTO_DEBUG_MODEL_DWARF:
137    EmitDwarfDebugInfo = true;
138    return;
139  }
140  llvm_unreachable("Unknown debug format!");
141}
142
143void LTOCodeGenerator::setCodePICModel(lto_codegen_model model) {
144  switch (model) {
145  case LTO_CODEGEN_PIC_MODEL_STATIC:
146  case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
147  case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
148  case LTO_CODEGEN_PIC_MODEL_DEFAULT:
149    CodeModel = model;
150    return;
151  }
152  llvm_unreachable("Unknown PIC model!");
153}
154
155bool LTOCodeGenerator::writeMergedModules(const char *path,
156                                          std::string &errMsg) {
157  if (!determineTarget(errMsg))
158    return false;
159
160  // mark which symbols can not be internalized
161  applyScopeRestrictions();
162
163  // create output file
164  std::string ErrInfo;
165  tool_output_file Out(path, ErrInfo, sys::fs::F_None);
166  if (!ErrInfo.empty()) {
167    errMsg = "could not open bitcode file for writing: ";
168    errMsg += path;
169    return false;
170  }
171
172  // write bitcode to it
173  WriteBitcodeToFile(IRLinker.getModule(), Out.os());
174  Out.os().close();
175
176  if (Out.os().has_error()) {
177    errMsg = "could not write bitcode file: ";
178    errMsg += path;
179    Out.os().clear_error();
180    return false;
181  }
182
183  Out.keep();
184  return true;
185}
186
187bool LTOCodeGenerator::compile_to_file(const char** name,
188                                       bool disableOpt,
189                                       bool disableInline,
190                                       bool disableGVNLoadPRE,
191                                       std::string& errMsg) {
192  // make unique temp .o file to put generated object file
193  SmallString<128> Filename;
194  int FD;
195  std::error_code EC =
196      sys::fs::createTemporaryFile("lto-llvm", "o", FD, Filename);
197  if (EC) {
198    errMsg = EC.message();
199    return false;
200  }
201
202  // generate object file
203  tool_output_file objFile(Filename.c_str(), FD);
204
205  bool genResult = generateObjectFile(objFile.os(), disableOpt, disableInline,
206                                      disableGVNLoadPRE, errMsg);
207  objFile.os().close();
208  if (objFile.os().has_error()) {
209    objFile.os().clear_error();
210    sys::fs::remove(Twine(Filename));
211    return false;
212  }
213
214  objFile.keep();
215  if (!genResult) {
216    sys::fs::remove(Twine(Filename));
217    return false;
218  }
219
220  NativeObjectPath = Filename.c_str();
221  *name = NativeObjectPath.c_str();
222  return true;
223}
224
225const void* LTOCodeGenerator::compile(size_t* length,
226                                      bool disableOpt,
227                                      bool disableInline,
228                                      bool disableGVNLoadPRE,
229                                      std::string& errMsg) {
230  const char *name;
231  if (!compile_to_file(&name, disableOpt, disableInline, disableGVNLoadPRE,
232                       errMsg))
233    return nullptr;
234
235  // remove old buffer if compile() called twice
236  delete NativeObjectFile;
237
238  // read .o file into memory buffer
239  ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
240      MemoryBuffer::getFile(name, -1, false);
241  if (std::error_code EC = BufferOrErr.getError()) {
242    errMsg = EC.message();
243    sys::fs::remove(NativeObjectPath);
244    return nullptr;
245  }
246  NativeObjectFile = BufferOrErr.get().release();
247
248  // remove temp files
249  sys::fs::remove(NativeObjectPath);
250
251  // return buffer, unless error
252  if (!NativeObjectFile)
253    return nullptr;
254  *length = NativeObjectFile->getBufferSize();
255  return NativeObjectFile->getBufferStart();
256}
257
258bool LTOCodeGenerator::determineTarget(std::string &errMsg) {
259  if (TargetMach)
260    return true;
261
262  std::string TripleStr = IRLinker.getModule()->getTargetTriple();
263  if (TripleStr.empty())
264    TripleStr = sys::getDefaultTargetTriple();
265  llvm::Triple Triple(TripleStr);
266
267  // create target machine from info for merged modules
268  const Target *march = TargetRegistry::lookupTarget(TripleStr, errMsg);
269  if (!march)
270    return false;
271
272  // The relocation model is actually a static member of TargetMachine and
273  // needs to be set before the TargetMachine is instantiated.
274  Reloc::Model RelocModel = Reloc::Default;
275  switch (CodeModel) {
276  case LTO_CODEGEN_PIC_MODEL_STATIC:
277    RelocModel = Reloc::Static;
278    break;
279  case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
280    RelocModel = Reloc::PIC_;
281    break;
282  case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
283    RelocModel = Reloc::DynamicNoPIC;
284    break;
285  case LTO_CODEGEN_PIC_MODEL_DEFAULT:
286    // RelocModel is already the default, so leave it that way.
287    break;
288  }
289
290  // Construct LTOModule, hand over ownership of module and target. Use MAttr as
291  // the default set of features.
292  SubtargetFeatures Features(MAttr);
293  Features.getDefaultSubtargetFeatures(Triple);
294  std::string FeatureStr = Features.getString();
295  // Set a default CPU for Darwin triples.
296  if (MCpu.empty() && Triple.isOSDarwin()) {
297    if (Triple.getArch() == llvm::Triple::x86_64)
298      MCpu = "core2";
299    else if (Triple.getArch() == llvm::Triple::x86)
300      MCpu = "yonah";
301    else if (Triple.getArch() == llvm::Triple::arm64 ||
302             Triple.getArch() == llvm::Triple::aarch64)
303      MCpu = "cyclone";
304  }
305
306  TargetMach = march->createTargetMachine(TripleStr, MCpu, FeatureStr, Options,
307                                          RelocModel, CodeModel::Default,
308                                          CodeGenOpt::Aggressive);
309  return true;
310}
311
312void LTOCodeGenerator::
313applyRestriction(GlobalValue &GV,
314                 const ArrayRef<StringRef> &Libcalls,
315                 std::vector<const char*> &MustPreserveList,
316                 SmallPtrSet<GlobalValue*, 8> &AsmUsed,
317                 Mangler &Mangler) {
318  // There are no restrictions to apply to declarations.
319  if (GV.isDeclaration())
320    return;
321
322  // There is nothing more restrictive than private linkage.
323  if (GV.hasPrivateLinkage())
324    return;
325
326  SmallString<64> Buffer;
327  TargetMach->getNameWithPrefix(Buffer, &GV, Mangler);
328
329  if (MustPreserveSymbols.count(Buffer))
330    MustPreserveList.push_back(GV.getName().data());
331  if (AsmUndefinedRefs.count(Buffer))
332    AsmUsed.insert(&GV);
333
334  // Conservatively append user-supplied runtime library functions to
335  // llvm.compiler.used.  These could be internalized and deleted by
336  // optimizations like -globalopt, causing problems when later optimizations
337  // add new library calls (e.g., llvm.memset => memset and printf => puts).
338  // Leave it to the linker to remove any dead code (e.g. with -dead_strip).
339  if (isa<Function>(GV) &&
340      std::binary_search(Libcalls.begin(), Libcalls.end(), GV.getName()))
341    AsmUsed.insert(&GV);
342}
343
344static void findUsedValues(GlobalVariable *LLVMUsed,
345                           SmallPtrSet<GlobalValue*, 8> &UsedValues) {
346  if (!LLVMUsed) return;
347
348  ConstantArray *Inits = cast<ConstantArray>(LLVMUsed->getInitializer());
349  for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
350    if (GlobalValue *GV =
351        dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
352      UsedValues.insert(GV);
353}
354
355static void accumulateAndSortLibcalls(std::vector<StringRef> &Libcalls,
356                                      const TargetLibraryInfo& TLI,
357                                      const TargetLowering *Lowering)
358{
359  // TargetLibraryInfo has info on C runtime library calls on the current
360  // target.
361  for (unsigned I = 0, E = static_cast<unsigned>(LibFunc::NumLibFuncs);
362       I != E; ++I) {
363    LibFunc::Func F = static_cast<LibFunc::Func>(I);
364    if (TLI.has(F))
365      Libcalls.push_back(TLI.getName(F));
366  }
367
368  // TargetLowering has info on library calls that CodeGen expects to be
369  // available, both from the C runtime and compiler-rt.
370  if (Lowering)
371    for (unsigned I = 0, E = static_cast<unsigned>(RTLIB::UNKNOWN_LIBCALL);
372         I != E; ++I)
373      if (const char *Name
374          = Lowering->getLibcallName(static_cast<RTLIB::Libcall>(I)))
375        Libcalls.push_back(Name);
376
377  array_pod_sort(Libcalls.begin(), Libcalls.end());
378  Libcalls.erase(std::unique(Libcalls.begin(), Libcalls.end()),
379                 Libcalls.end());
380}
381
382void LTOCodeGenerator::applyScopeRestrictions() {
383  if (ScopeRestrictionsDone)
384    return;
385  Module *mergedModule = IRLinker.getModule();
386
387  // Start off with a verification pass.
388  PassManager passes;
389  passes.add(createVerifierPass());
390  passes.add(createDebugInfoVerifierPass());
391
392  // mark which symbols can not be internalized
393  Mangler Mangler(TargetMach->getDataLayout());
394  std::vector<const char*> MustPreserveList;
395  SmallPtrSet<GlobalValue*, 8> AsmUsed;
396  std::vector<StringRef> Libcalls;
397  TargetLibraryInfo TLI(Triple(TargetMach->getTargetTriple()));
398  accumulateAndSortLibcalls(Libcalls, TLI, TargetMach->getTargetLowering());
399
400  for (Module::iterator f = mergedModule->begin(),
401         e = mergedModule->end(); f != e; ++f)
402    applyRestriction(*f, Libcalls, MustPreserveList, AsmUsed, Mangler);
403  for (Module::global_iterator v = mergedModule->global_begin(),
404         e = mergedModule->global_end(); v !=  e; ++v)
405    applyRestriction(*v, Libcalls, MustPreserveList, AsmUsed, Mangler);
406  for (Module::alias_iterator a = mergedModule->alias_begin(),
407         e = mergedModule->alias_end(); a != e; ++a)
408    applyRestriction(*a, Libcalls, MustPreserveList, AsmUsed, Mangler);
409
410  GlobalVariable *LLVMCompilerUsed =
411    mergedModule->getGlobalVariable("llvm.compiler.used");
412  findUsedValues(LLVMCompilerUsed, AsmUsed);
413  if (LLVMCompilerUsed)
414    LLVMCompilerUsed->eraseFromParent();
415
416  if (!AsmUsed.empty()) {
417    llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(Context);
418    std::vector<Constant*> asmUsed2;
419    for (auto *GV : AsmUsed) {
420      Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
421      asmUsed2.push_back(c);
422    }
423
424    llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
425    LLVMCompilerUsed =
426      new llvm::GlobalVariable(*mergedModule, ATy, false,
427                               llvm::GlobalValue::AppendingLinkage,
428                               llvm::ConstantArray::get(ATy, asmUsed2),
429                               "llvm.compiler.used");
430
431    LLVMCompilerUsed->setSection("llvm.metadata");
432  }
433
434  passes.add(createInternalizePass(MustPreserveList));
435
436  // apply scope restrictions
437  passes.run(*mergedModule);
438
439  ScopeRestrictionsDone = true;
440}
441
442/// Optimize merged modules using various IPO passes
443bool LTOCodeGenerator::generateObjectFile(raw_ostream &out,
444                                          bool DisableOpt,
445                                          bool DisableInline,
446                                          bool DisableGVNLoadPRE,
447                                          std::string &errMsg) {
448  if (!this->determineTarget(errMsg))
449    return false;
450
451  Module *mergedModule = IRLinker.getModule();
452
453  // Mark which symbols can not be internalized
454  this->applyScopeRestrictions();
455
456  // Instantiate the pass manager to organize the passes.
457  PassManager passes;
458
459  // Start off with a verification pass.
460  passes.add(createVerifierPass());
461  passes.add(createDebugInfoVerifierPass());
462
463  // Add an appropriate DataLayout instance for this module...
464  mergedModule->setDataLayout(TargetMach->getDataLayout());
465  passes.add(new DataLayoutPass(mergedModule));
466
467  // Add appropriate TargetLibraryInfo for this module.
468  passes.add(new TargetLibraryInfo(Triple(TargetMach->getTargetTriple())));
469
470  TargetMach->addAnalysisPasses(passes);
471
472  // Enabling internalize here would use its AllButMain variant. It
473  // keeps only main if it exists and does nothing for libraries. Instead
474  // we create the pass ourselves with the symbol list provided by the linker.
475  if (!DisableOpt)
476    PassManagerBuilder().populateLTOPassManager(passes,
477                                              /*Internalize=*/false,
478                                              !DisableInline,
479                                              DisableGVNLoadPRE);
480
481  // Make sure everything is still good.
482  passes.add(createVerifierPass());
483  passes.add(createDebugInfoVerifierPass());
484
485  PassManager codeGenPasses;
486
487  codeGenPasses.add(new DataLayoutPass(mergedModule));
488
489  formatted_raw_ostream Out(out);
490
491  // If the bitcode files contain ARC code and were compiled with optimization,
492  // the ObjCARCContractPass must be run, so do it unconditionally here.
493  codeGenPasses.add(createObjCARCContractPass());
494
495  if (TargetMach->addPassesToEmitFile(codeGenPasses, Out,
496                                      TargetMachine::CGFT_ObjectFile)) {
497    errMsg = "target file type not supported";
498    return false;
499  }
500
501  // Run our queue of passes all at once now, efficiently.
502  passes.run(*mergedModule);
503
504  // Run the code generator, and write assembly file
505  codeGenPasses.run(*mergedModule);
506
507  return true;
508}
509
510/// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging
511/// LTO problems.
512void LTOCodeGenerator::setCodeGenDebugOptions(const char *options) {
513  for (std::pair<StringRef, StringRef> o = getToken(options);
514       !o.first.empty(); o = getToken(o.second)) {
515    // ParseCommandLineOptions() expects argv[0] to be program name. Lazily add
516    // that.
517    if (CodegenOptions.empty())
518      CodegenOptions.push_back(strdup("libLLVMLTO"));
519    CodegenOptions.push_back(strdup(o.first.str().c_str()));
520  }
521}
522
523void LTOCodeGenerator::parseCodeGenDebugOptions() {
524  // if options were requested, set them
525  if (!CodegenOptions.empty())
526    cl::ParseCommandLineOptions(CodegenOptions.size(),
527                                const_cast<char **>(&CodegenOptions[0]));
528}
529
530void LTOCodeGenerator::DiagnosticHandler(const DiagnosticInfo &DI,
531                                         void *Context) {
532  ((LTOCodeGenerator *)Context)->DiagnosticHandler2(DI);
533}
534
535void LTOCodeGenerator::DiagnosticHandler2(const DiagnosticInfo &DI) {
536  // Map the LLVM internal diagnostic severity to the LTO diagnostic severity.
537  lto_codegen_diagnostic_severity_t Severity;
538  switch (DI.getSeverity()) {
539  case DS_Error:
540    Severity = LTO_DS_ERROR;
541    break;
542  case DS_Warning:
543    Severity = LTO_DS_WARNING;
544    break;
545  case DS_Remark:
546    Severity = LTO_DS_REMARK;
547    break;
548  case DS_Note:
549    Severity = LTO_DS_NOTE;
550    break;
551  }
552  // Create the string that will be reported to the external diagnostic handler.
553  std::string MsgStorage;
554  raw_string_ostream Stream(MsgStorage);
555  DiagnosticPrinterRawOStream DP(Stream);
556  DI.print(DP);
557  Stream.flush();
558
559  // If this method has been called it means someone has set up an external
560  // diagnostic handler. Assert on that.
561  assert(DiagHandler && "Invalid diagnostic handler");
562  (*DiagHandler)(Severity, MsgStorage.c_str(), DiagContext);
563}
564
565void
566LTOCodeGenerator::setDiagnosticHandler(lto_diagnostic_handler_t DiagHandler,
567                                       void *Ctxt) {
568  this->DiagHandler = DiagHandler;
569  this->DiagContext = Ctxt;
570  if (!DiagHandler)
571    return Context.setDiagnosticHandler(nullptr, nullptr);
572  // Register the LTOCodeGenerator stub in the LLVMContext to forward the
573  // diagnostic to the external DiagHandler.
574  Context.setDiagnosticHandler(LTOCodeGenerator::DiagnosticHandler, this);
575}
576