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/Analysis/TargetLibraryInfo.h"
19 #include "llvm/Analysis/TargetTransformInfo.h"
20 #include "llvm/Bitcode/ReaderWriter.h"
21 #include "llvm/CodeGen/RuntimeLibcalls.h"
22 #include "llvm/Config/config.h"
23 #include "llvm/IR/Constants.h"
24 #include "llvm/IR/DataLayout.h"
25 #include "llvm/IR/DerivedTypes.h"
26 #include "llvm/IR/DiagnosticInfo.h"
27 #include "llvm/IR/DiagnosticPrinter.h"
28 #include "llvm/IR/LLVMContext.h"
29 #include "llvm/IR/Mangler.h"
30 #include "llvm/IR/Module.h"
31 #include "llvm/IR/Verifier.h"
32 #include "llvm/InitializePasses.h"
33 #include "llvm/LTO/LTOModule.h"
34 #include "llvm/Linker/Linker.h"
35 #include "llvm/MC/MCAsmInfo.h"
36 #include "llvm/MC/MCContext.h"
37 #include "llvm/MC/SubtargetFeature.h"
38 #include "llvm/PassManager.h"
39 #include "llvm/Support/CommandLine.h"
40 #include "llvm/Support/FileSystem.h"
41 #include "llvm/Support/FormattedStream.h"
42 #include "llvm/Support/Host.h"
43 #include "llvm/Support/MemoryBuffer.h"
44 #include "llvm/Support/Signals.h"
45 #include "llvm/Support/TargetRegistry.h"
46 #include "llvm/Support/TargetSelect.h"
47 #include "llvm/Support/ToolOutputFile.h"
48 #include "llvm/Support/raw_ostream.h"
49 #include "llvm/Target/TargetLowering.h"
50 #include "llvm/Target/TargetOptions.h"
51 #include "llvm/Target/TargetRegisterInfo.h"
52 #include "llvm/Target/TargetSubtargetInfo.h"
53 #include "llvm/Transforms/IPO.h"
54 #include "llvm/Transforms/IPO/PassManagerBuilder.h"
55 #include "llvm/Transforms/ObjCARC.h"
56 #include <system_error>
57 using namespace llvm;
58 
59 const char* LTOCodeGenerator::getVersionString() {
60 #ifdef LLVM_VERSION_INFO
61   return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
62 #else
63   return PACKAGE_NAME " version " PACKAGE_VERSION;
64 #endif
65 }
66 
67 LTOCodeGenerator::LTOCodeGenerator()
68     : Context(getGlobalContext()), IRLinker(new Module("ld-temp.o", Context)) {
69   initialize();
70 }
71 
72 LTOCodeGenerator::LTOCodeGenerator(std::unique_ptr<LLVMContext> Context)
73     : OwnedContext(std::move(Context)), Context(*OwnedContext),
74       IRLinker(new Module("ld-temp.o", *OwnedContext)) {
75   initialize();
76 }
77 
78 void LTOCodeGenerator::initialize() {
79   TargetMach = nullptr;
80   EmitDwarfDebugInfo = false;
81   ScopeRestrictionsDone = false;
82   CodeModel = LTO_CODEGEN_PIC_MODEL_DEFAULT;
83   DiagHandler = nullptr;
84   DiagContext = nullptr;
85 
86   initializeLTOPasses();
87 }
88 
89 LTOCodeGenerator::~LTOCodeGenerator() {
90   delete TargetMach;
91   TargetMach = nullptr;
92 
93   IRLinker.deleteModule();
94 
95   for (std::vector<char *>::iterator I = CodegenOptions.begin(),
96                                      E = CodegenOptions.end();
97        I != E; ++I)
98     free(*I);
99 }
100 
101 // Initialize LTO passes. Please keep this funciton in sync with
102 // PassManagerBuilder::populateLTOPassManager(), and make sure all LTO
103 // passes are initialized.
104 void LTOCodeGenerator::initializeLTOPasses() {
105   PassRegistry &R = *PassRegistry::getPassRegistry();
106 
107   initializeInternalizePassPass(R);
108   initializeIPSCCPPass(R);
109   initializeGlobalOptPass(R);
110   initializeConstantMergePass(R);
111   initializeDAHPass(R);
112   initializeInstructionCombiningPassPass(R);
113   initializeSimpleInlinerPass(R);
114   initializePruneEHPass(R);
115   initializeGlobalDCEPass(R);
116   initializeArgPromotionPass(R);
117   initializeJumpThreadingPass(R);
118   initializeSROAPass(R);
119   initializeSROA_DTPass(R);
120   initializeSROA_SSAUpPass(R);
121   initializeFunctionAttrsPass(R);
122   initializeGlobalsModRefPass(R);
123   initializeLICMPass(R);
124   initializeMergedLoadStoreMotionPass(R);
125   initializeGVNPass(R);
126   initializeMemCpyOptPass(R);
127   initializeDCEPass(R);
128   initializeCFGSimplifyPassPass(R);
129 }
130 
131 bool LTOCodeGenerator::addModule(LTOModule *mod) {
132   assert(&mod->getModule().getContext() == &Context &&
133          "Expected module in same context");
134 
135   bool ret = IRLinker.linkInModule(&mod->getModule());
136 
137   const std::vector<const char*> &undefs = mod->getAsmUndefinedRefs();
138   for (int i = 0, e = undefs.size(); i != e; ++i)
139     AsmUndefinedRefs[undefs[i]] = 1;
140 
141   return !ret;
142 }
143 
144 void LTOCodeGenerator::setTargetOptions(TargetOptions options) {
145   Options = options;
146 }
147 
148 void LTOCodeGenerator::setDebugInfo(lto_debug_model debug) {
149   switch (debug) {
150   case LTO_DEBUG_MODEL_NONE:
151     EmitDwarfDebugInfo = false;
152     return;
153 
154   case LTO_DEBUG_MODEL_DWARF:
155     EmitDwarfDebugInfo = true;
156     return;
157   }
158   llvm_unreachable("Unknown debug format!");
159 }
160 
161 void LTOCodeGenerator::setCodePICModel(lto_codegen_model model) {
162   switch (model) {
163   case LTO_CODEGEN_PIC_MODEL_STATIC:
164   case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
165   case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
166   case LTO_CODEGEN_PIC_MODEL_DEFAULT:
167     CodeModel = model;
168     return;
169   }
170   llvm_unreachable("Unknown PIC model!");
171 }
172 
173 bool LTOCodeGenerator::writeMergedModules(const char *path,
174                                           std::string &errMsg) {
175   if (!determineTarget(errMsg))
176     return false;
177 
178   // mark which symbols can not be internalized
179   applyScopeRestrictions();
180 
181   // create output file
182   std::error_code EC;
183   tool_output_file Out(path, EC, sys::fs::F_None);
184   if (EC) {
185     errMsg = "could not open bitcode file for writing: ";
186     errMsg += path;
187     return false;
188   }
189 
190   // write bitcode to it
191   WriteBitcodeToFile(IRLinker.getModule(), Out.os());
192   Out.os().close();
193 
194   if (Out.os().has_error()) {
195     errMsg = "could not write bitcode file: ";
196     errMsg += path;
197     Out.os().clear_error();
198     return false;
199   }
200 
201   Out.keep();
202   return true;
203 }
204 
205 bool LTOCodeGenerator::compile_to_file(const char** name,
206                                        bool disableOpt,
207                                        bool disableInline,
208                                        bool disableGVNLoadPRE,
209                                        bool disableVectorization,
210                                        std::string& errMsg) {
211   // make unique temp .o file to put generated object file
212   SmallString<128> Filename;
213   int FD;
214   std::error_code EC =
215       sys::fs::createTemporaryFile("lto-llvm", "o", FD, Filename);
216   if (EC) {
217     errMsg = EC.message();
218     return false;
219   }
220 
221   // generate object file
222   tool_output_file objFile(Filename.c_str(), FD);
223 
224   bool genResult =
225       generateObjectFile(objFile.os(), disableOpt, disableInline,
226                          disableGVNLoadPRE, disableVectorization, errMsg);
227   objFile.os().close();
228   if (objFile.os().has_error()) {
229     objFile.os().clear_error();
230     sys::fs::remove(Twine(Filename));
231     return false;
232   }
233 
234   objFile.keep();
235   if (!genResult) {
236     sys::fs::remove(Twine(Filename));
237     return false;
238   }
239 
240   NativeObjectPath = Filename.c_str();
241   *name = NativeObjectPath.c_str();
242   return true;
243 }
244 
245 const void* LTOCodeGenerator::compile(size_t* length,
246                                       bool disableOpt,
247                                       bool disableInline,
248                                       bool disableGVNLoadPRE,
249                                       bool disableVectorization,
250                                       std::string& errMsg) {
251   const char *name;
252   if (!compile_to_file(&name, disableOpt, disableInline, disableGVNLoadPRE,
253                        disableVectorization, errMsg))
254     return nullptr;
255 
256   // read .o file into memory buffer
257   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
258       MemoryBuffer::getFile(name, -1, false);
259   if (std::error_code EC = BufferOrErr.getError()) {
260     errMsg = EC.message();
261     sys::fs::remove(NativeObjectPath);
262     return nullptr;
263   }
264   NativeObjectFile = std::move(*BufferOrErr);
265 
266   // remove temp files
267   sys::fs::remove(NativeObjectPath);
268 
269   // return buffer, unless error
270   if (!NativeObjectFile)
271     return nullptr;
272   *length = NativeObjectFile->getBufferSize();
273   return NativeObjectFile->getBufferStart();
274 }
275 
276 bool LTOCodeGenerator::determineTarget(std::string &errMsg) {
277   if (TargetMach)
278     return true;
279 
280   std::string TripleStr = IRLinker.getModule()->getTargetTriple();
281   if (TripleStr.empty())
282     TripleStr = sys::getDefaultTargetTriple();
283   llvm::Triple Triple(TripleStr);
284 
285   // create target machine from info for merged modules
286   const Target *march = TargetRegistry::lookupTarget(TripleStr, errMsg);
287   if (!march)
288     return false;
289 
290   // The relocation model is actually a static member of TargetMachine and
291   // needs to be set before the TargetMachine is instantiated.
292   Reloc::Model RelocModel = Reloc::Default;
293   switch (CodeModel) {
294   case LTO_CODEGEN_PIC_MODEL_STATIC:
295     RelocModel = Reloc::Static;
296     break;
297   case LTO_CODEGEN_PIC_MODEL_DYNAMIC:
298     RelocModel = Reloc::PIC_;
299     break;
300   case LTO_CODEGEN_PIC_MODEL_DYNAMIC_NO_PIC:
301     RelocModel = Reloc::DynamicNoPIC;
302     break;
303   case LTO_CODEGEN_PIC_MODEL_DEFAULT:
304     // RelocModel is already the default, so leave it that way.
305     break;
306   }
307 
308   // Construct LTOModule, hand over ownership of module and target. Use MAttr as
309   // the default set of features.
310   SubtargetFeatures Features(MAttr);
311   Features.getDefaultSubtargetFeatures(Triple);
312   std::string FeatureStr = Features.getString();
313   // Set a default CPU for Darwin triples.
314   if (MCpu.empty() && Triple.isOSDarwin()) {
315     if (Triple.getArch() == llvm::Triple::x86_64)
316       MCpu = "core2";
317     else if (Triple.getArch() == llvm::Triple::x86)
318       MCpu = "yonah";
319     else if (Triple.getArch() == llvm::Triple::aarch64)
320       MCpu = "cyclone";
321   }
322 
323   TargetMach = march->createTargetMachine(TripleStr, MCpu, FeatureStr, Options,
324                                           RelocModel, CodeModel::Default,
325                                           CodeGenOpt::Aggressive);
326   return true;
327 }
328 
329 void LTOCodeGenerator::
330 applyRestriction(GlobalValue &GV,
331                  ArrayRef<StringRef> Libcalls,
332                  std::vector<const char*> &MustPreserveList,
333                  SmallPtrSetImpl<GlobalValue*> &AsmUsed,
334                  Mangler &Mangler) {
335   // There are no restrictions to apply to declarations.
336   if (GV.isDeclaration())
337     return;
338 
339   // There is nothing more restrictive than private linkage.
340   if (GV.hasPrivateLinkage())
341     return;
342 
343   SmallString<64> Buffer;
344   TargetMach->getNameWithPrefix(Buffer, &GV, Mangler);
345 
346   if (MustPreserveSymbols.count(Buffer))
347     MustPreserveList.push_back(GV.getName().data());
348   if (AsmUndefinedRefs.count(Buffer))
349     AsmUsed.insert(&GV);
350 
351   // Conservatively append user-supplied runtime library functions to
352   // llvm.compiler.used.  These could be internalized and deleted by
353   // optimizations like -globalopt, causing problems when later optimizations
354   // add new library calls (e.g., llvm.memset => memset and printf => puts).
355   // Leave it to the linker to remove any dead code (e.g. with -dead_strip).
356   if (isa<Function>(GV) &&
357       std::binary_search(Libcalls.begin(), Libcalls.end(), GV.getName()))
358     AsmUsed.insert(&GV);
359 }
360 
361 static void findUsedValues(GlobalVariable *LLVMUsed,
362                            SmallPtrSetImpl<GlobalValue*> &UsedValues) {
363   if (!LLVMUsed) return;
364 
365   ConstantArray *Inits = cast<ConstantArray>(LLVMUsed->getInitializer());
366   for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
367     if (GlobalValue *GV =
368         dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
369       UsedValues.insert(GV);
370 }
371 
372 // Collect names of runtime library functions. User-defined functions with the
373 // same names are added to llvm.compiler.used to prevent them from being
374 // deleted by optimizations.
375 static void accumulateAndSortLibcalls(std::vector<StringRef> &Libcalls,
376                                       const TargetLibraryInfo& TLI,
377                                       const Module &Mod,
378                                       const TargetMachine &TM) {
379   // TargetLibraryInfo has info on C runtime library calls on the current
380   // target.
381   for (unsigned I = 0, E = static_cast<unsigned>(LibFunc::NumLibFuncs);
382        I != E; ++I) {
383     LibFunc::Func F = static_cast<LibFunc::Func>(I);
384     if (TLI.has(F))
385       Libcalls.push_back(TLI.getName(F));
386   }
387 
388   SmallPtrSet<const TargetLowering *, 1> TLSet;
389 
390   for (const Function &F : Mod) {
391     const TargetLowering *Lowering =
392         TM.getSubtargetImpl(F)->getTargetLowering();
393 
394     if (Lowering && TLSet.insert(Lowering).second)
395       // TargetLowering has info on library calls that CodeGen expects to be
396       // available, both from the C runtime and compiler-rt.
397       for (unsigned I = 0, E = static_cast<unsigned>(RTLIB::UNKNOWN_LIBCALL);
398            I != E; ++I)
399         if (const char *Name =
400                 Lowering->getLibcallName(static_cast<RTLIB::Libcall>(I)))
401           Libcalls.push_back(Name);
402   }
403 
404   array_pod_sort(Libcalls.begin(), Libcalls.end());
405   Libcalls.erase(std::unique(Libcalls.begin(), Libcalls.end()),
406                  Libcalls.end());
407 }
408 
409 void LTOCodeGenerator::applyScopeRestrictions() {
410   if (ScopeRestrictionsDone)
411     return;
412   Module *mergedModule = IRLinker.getModule();
413 
414   // Start off with a verification pass.
415   PassManager passes;
416   passes.add(createVerifierPass());
417   passes.add(createDebugInfoVerifierPass());
418 
419   // mark which symbols can not be internalized
420   Mangler Mangler(TargetMach->getDataLayout());
421   std::vector<const char*> MustPreserveList;
422   SmallPtrSet<GlobalValue*, 8> AsmUsed;
423   std::vector<StringRef> Libcalls;
424   TargetLibraryInfoImpl TLII(Triple(TargetMach->getTargetTriple()));
425   TargetLibraryInfo TLI(TLII);
426 
427   accumulateAndSortLibcalls(Libcalls, TLI, *mergedModule, *TargetMach);
428 
429   for (Module::iterator f = mergedModule->begin(),
430          e = mergedModule->end(); f != e; ++f)
431     applyRestriction(*f, Libcalls, MustPreserveList, AsmUsed, Mangler);
432   for (Module::global_iterator v = mergedModule->global_begin(),
433          e = mergedModule->global_end(); v !=  e; ++v)
434     applyRestriction(*v, Libcalls, MustPreserveList, AsmUsed, Mangler);
435   for (Module::alias_iterator a = mergedModule->alias_begin(),
436          e = mergedModule->alias_end(); a != e; ++a)
437     applyRestriction(*a, Libcalls, MustPreserveList, AsmUsed, Mangler);
438 
439   GlobalVariable *LLVMCompilerUsed =
440     mergedModule->getGlobalVariable("llvm.compiler.used");
441   findUsedValues(LLVMCompilerUsed, AsmUsed);
442   if (LLVMCompilerUsed)
443     LLVMCompilerUsed->eraseFromParent();
444 
445   if (!AsmUsed.empty()) {
446     llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(Context);
447     std::vector<Constant*> asmUsed2;
448     for (auto *GV : AsmUsed) {
449       Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
450       asmUsed2.push_back(c);
451     }
452 
453     llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
454     LLVMCompilerUsed =
455       new llvm::GlobalVariable(*mergedModule, ATy, false,
456                                llvm::GlobalValue::AppendingLinkage,
457                                llvm::ConstantArray::get(ATy, asmUsed2),
458                                "llvm.compiler.used");
459 
460     LLVMCompilerUsed->setSection("llvm.metadata");
461   }
462 
463   passes.add(createInternalizePass(MustPreserveList));
464 
465   // apply scope restrictions
466   passes.run(*mergedModule);
467 
468   ScopeRestrictionsDone = true;
469 }
470 
471 /// Optimize merged modules using various IPO passes
472 bool LTOCodeGenerator::generateObjectFile(raw_ostream &out,
473                                           bool DisableOpt,
474                                           bool DisableInline,
475                                           bool DisableGVNLoadPRE,
476                                           bool DisableVectorization,
477                                           std::string &errMsg) {
478   if (!this->determineTarget(errMsg))
479     return false;
480 
481   Module *mergedModule = IRLinker.getModule();
482 
483   // Mark which symbols can not be internalized
484   this->applyScopeRestrictions();
485 
486   // Instantiate the pass manager to organize the passes.
487   PassManager passes;
488 
489   // Add an appropriate DataLayout instance for this module...
490   mergedModule->setDataLayout(TargetMach->getDataLayout());
491 
492   passes.add(new DataLayoutPass());
493   passes.add(
494       createTargetTransformInfoWrapperPass(TargetMach->getTargetIRAnalysis()));
495 
496   Triple TargetTriple(TargetMach->getTargetTriple());
497   PassManagerBuilder PMB;
498   PMB.DisableGVNLoadPRE = DisableGVNLoadPRE;
499   PMB.LoopVectorize = !DisableVectorization;
500   PMB.SLPVectorize = !DisableVectorization;
501   if (!DisableInline)
502     PMB.Inliner = createFunctionInliningPass();
503   PMB.LibraryInfo = new TargetLibraryInfoImpl(TargetTriple);
504   if (DisableOpt)
505     PMB.OptLevel = 0;
506   PMB.VerifyInput = true;
507   PMB.VerifyOutput = true;
508 
509   PMB.populateLTOPassManager(passes);
510 
511   PassManager codeGenPasses;
512 
513   codeGenPasses.add(new DataLayoutPass());
514 
515   formatted_raw_ostream Out(out);
516 
517   // If the bitcode files contain ARC code and were compiled with optimization,
518   // the ObjCARCContractPass must be run, so do it unconditionally here.
519   codeGenPasses.add(createObjCARCContractPass());
520 
521   if (TargetMach->addPassesToEmitFile(codeGenPasses, Out,
522                                       TargetMachine::CGFT_ObjectFile)) {
523     errMsg = "target file type not supported";
524     return false;
525   }
526 
527   // Run our queue of passes all at once now, efficiently.
528   passes.run(*mergedModule);
529 
530   // Run the code generator, and write assembly file
531   codeGenPasses.run(*mergedModule);
532 
533   return true;
534 }
535 
536 /// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging
537 /// LTO problems.
538 void LTOCodeGenerator::setCodeGenDebugOptions(const char *options) {
539   for (std::pair<StringRef, StringRef> o = getToken(options);
540        !o.first.empty(); o = getToken(o.second)) {
541     // ParseCommandLineOptions() expects argv[0] to be program name. Lazily add
542     // that.
543     if (CodegenOptions.empty())
544       CodegenOptions.push_back(strdup("libLLVMLTO"));
545     CodegenOptions.push_back(strdup(o.first.str().c_str()));
546   }
547 }
548 
549 void LTOCodeGenerator::parseCodeGenDebugOptions() {
550   // if options were requested, set them
551   if (!CodegenOptions.empty())
552     cl::ParseCommandLineOptions(CodegenOptions.size(),
553                                 const_cast<char **>(&CodegenOptions[0]));
554 }
555 
556 void LTOCodeGenerator::DiagnosticHandler(const DiagnosticInfo &DI,
557                                          void *Context) {
558   ((LTOCodeGenerator *)Context)->DiagnosticHandler2(DI);
559 }
560 
561 void LTOCodeGenerator::DiagnosticHandler2(const DiagnosticInfo &DI) {
562   // Map the LLVM internal diagnostic severity to the LTO diagnostic severity.
563   lto_codegen_diagnostic_severity_t Severity;
564   switch (DI.getSeverity()) {
565   case DS_Error:
566     Severity = LTO_DS_ERROR;
567     break;
568   case DS_Warning:
569     Severity = LTO_DS_WARNING;
570     break;
571   case DS_Remark:
572     Severity = LTO_DS_REMARK;
573     break;
574   case DS_Note:
575     Severity = LTO_DS_NOTE;
576     break;
577   }
578   // Create the string that will be reported to the external diagnostic handler.
579   std::string MsgStorage;
580   raw_string_ostream Stream(MsgStorage);
581   DiagnosticPrinterRawOStream DP(Stream);
582   DI.print(DP);
583   Stream.flush();
584 
585   // If this method has been called it means someone has set up an external
586   // diagnostic handler. Assert on that.
587   assert(DiagHandler && "Invalid diagnostic handler");
588   (*DiagHandler)(Severity, MsgStorage.c_str(), DiagContext);
589 }
590 
591 void
592 LTOCodeGenerator::setDiagnosticHandler(lto_diagnostic_handler_t DiagHandler,
593                                        void *Ctxt) {
594   this->DiagHandler = DiagHandler;
595   this->DiagContext = Ctxt;
596   if (!DiagHandler)
597     return Context.setDiagnosticHandler(nullptr, nullptr);
598   // Register the LTOCodeGenerator stub in the LLVMContext to forward the
599   // diagnostic to the external DiagHandler.
600   Context.setDiagnosticHandler(LTOCodeGenerator::DiagnosticHandler, this,
601                                /* RespectFilters */ true);
602 }
603