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/ParallelCG.h"
22 #include "llvm/CodeGen/RuntimeLibcalls.h"
23 #include "llvm/Config/config.h"
24 #include "llvm/IR/Constants.h"
25 #include "llvm/IR/DataLayout.h"
26 #include "llvm/IR/DerivedTypes.h"
27 #include "llvm/IR/DiagnosticInfo.h"
28 #include "llvm/IR/DiagnosticPrinter.h"
29 #include "llvm/IR/LLVMContext.h"
30 #include "llvm/IR/LegacyPassManager.h"
31 #include "llvm/IR/Mangler.h"
32 #include "llvm/IR/Module.h"
33 #include "llvm/IR/Verifier.h"
34 #include "llvm/InitializePasses.h"
35 #include "llvm/LTO/LTOModule.h"
36 #include "llvm/Linker/Linker.h"
37 #include "llvm/MC/MCAsmInfo.h"
38 #include "llvm/MC/MCContext.h"
39 #include "llvm/MC/SubtargetFeature.h"
40 #include "llvm/Support/CommandLine.h"
41 #include "llvm/Support/FileSystem.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 static void handleLTODiagnostic(const DiagnosticInfo &DI) {
68   DiagnosticPrinterRawOStream DP(errs());
69   DI.print(DP);
70   errs() << "\n";
71 }
72 
73 LTOCodeGenerator::LTOCodeGenerator()
74     : Context(getGlobalContext()),
75       MergedModule(new Module("ld-temp.o", Context)),
76       IRLinker(MergedModule.get(), handleLTODiagnostic) {
77   initializeLTOPasses();
78 }
79 
80 LTOCodeGenerator::LTOCodeGenerator(std::unique_ptr<LLVMContext> Context)
81     : OwnedContext(std::move(Context)), Context(*OwnedContext),
82       MergedModule(new Module("ld-temp.o", *OwnedContext)),
83       IRLinker(MergedModule.get(), handleLTODiagnostic) {
84   initializeLTOPasses();
85 }
86 
87 LTOCodeGenerator::~LTOCodeGenerator() {}
88 
89 // Initialize LTO passes. Please keep this function in sync with
90 // PassManagerBuilder::populateLTOPassManager(), and make sure all LTO
91 // passes are initialized.
92 void LTOCodeGenerator::initializeLTOPasses() {
93   PassRegistry &R = *PassRegistry::getPassRegistry();
94 
95   initializeInternalizePassPass(R);
96   initializeIPSCCPPass(R);
97   initializeGlobalOptPass(R);
98   initializeConstantMergePass(R);
99   initializeDAHPass(R);
100   initializeInstructionCombiningPassPass(R);
101   initializeSimpleInlinerPass(R);
102   initializePruneEHPass(R);
103   initializeGlobalDCEPass(R);
104   initializeArgPromotionPass(R);
105   initializeJumpThreadingPass(R);
106   initializeSROALegacyPassPass(R);
107   initializeSROA_DTPass(R);
108   initializeSROA_SSAUpPass(R);
109   initializeFunctionAttrsPass(R);
110   initializeGlobalsAAWrapperPassPass(R);
111   initializeLICMPass(R);
112   initializeMergedLoadStoreMotionPass(R);
113   initializeGVNPass(R);
114   initializeMemCpyOptPass(R);
115   initializeDCEPass(R);
116   initializeCFGSimplifyPassPass(R);
117 }
118 
119 bool LTOCodeGenerator::addModule(LTOModule *Mod) {
120   assert(&Mod->getModule().getContext() == &Context &&
121          "Expected module in same context");
122 
123   bool ret = IRLinker.linkInModule(&Mod->getModule());
124 
125   const std::vector<const char *> &undefs = Mod->getAsmUndefinedRefs();
126   for (int i = 0, e = undefs.size(); i != e; ++i)
127     AsmUndefinedRefs[undefs[i]] = 1;
128 
129   return !ret;
130 }
131 
132 void LTOCodeGenerator::setModule(std::unique_ptr<LTOModule> Mod) {
133   assert(&Mod->getModule().getContext() == &Context &&
134          "Expected module in same context");
135 
136   AsmUndefinedRefs.clear();
137 
138   MergedModule = Mod->takeModule();
139   IRLinker.setModule(MergedModule.get());
140 
141   const std::vector<const char*> &Undefs = Mod->getAsmUndefinedRefs();
142   for (int I = 0, E = Undefs.size(); I != E; ++I)
143     AsmUndefinedRefs[Undefs[I]] = 1;
144 }
145 
146 void LTOCodeGenerator::setTargetOptions(TargetOptions Options) {
147   this->Options = Options;
148 }
149 
150 void LTOCodeGenerator::setDebugInfo(lto_debug_model Debug) {
151   switch (Debug) {
152   case LTO_DEBUG_MODEL_NONE:
153     EmitDwarfDebugInfo = false;
154     return;
155 
156   case LTO_DEBUG_MODEL_DWARF:
157     EmitDwarfDebugInfo = true;
158     return;
159   }
160   llvm_unreachable("Unknown debug format!");
161 }
162 
163 void LTOCodeGenerator::setOptLevel(unsigned Level) {
164   OptLevel = Level;
165   switch (OptLevel) {
166   case 0:
167     CGOptLevel = CodeGenOpt::None;
168     break;
169   case 1:
170     CGOptLevel = CodeGenOpt::Less;
171     break;
172   case 2:
173     CGOptLevel = CodeGenOpt::Default;
174     break;
175   case 3:
176     CGOptLevel = CodeGenOpt::Aggressive;
177     break;
178   }
179 }
180 
181 bool LTOCodeGenerator::writeMergedModules(const char *Path,
182                                           std::string &ErrMsg) {
183   if (!determineTarget(ErrMsg))
184     return false;
185 
186   // mark which symbols can not be internalized
187   applyScopeRestrictions();
188 
189   // create output file
190   std::error_code EC;
191   tool_output_file Out(Path, EC, sys::fs::F_None);
192   if (EC) {
193     ErrMsg = "could not open bitcode file for writing: ";
194     ErrMsg += Path;
195     return false;
196   }
197 
198   // write bitcode to it
199   WriteBitcodeToFile(MergedModule.get(), Out.os(), ShouldEmbedUselists);
200   Out.os().close();
201 
202   if (Out.os().has_error()) {
203     ErrMsg = "could not write bitcode file: ";
204     ErrMsg += Path;
205     Out.os().clear_error();
206     return false;
207   }
208 
209   Out.keep();
210   return true;
211 }
212 
213 bool LTOCodeGenerator::compileOptimizedToFile(const char **Name,
214                                               std::string &ErrMsg) {
215   // make unique temp .o file to put generated object file
216   SmallString<128> Filename;
217   int FD;
218   std::error_code EC =
219       sys::fs::createTemporaryFile("lto-llvm", "o", FD, Filename);
220   if (EC) {
221     ErrMsg = EC.message();
222     return false;
223   }
224 
225   // generate object file
226   tool_output_file objFile(Filename.c_str(), FD);
227 
228   bool genResult = compileOptimized(&objFile.os(), ErrMsg);
229   objFile.os().close();
230   if (objFile.os().has_error()) {
231     objFile.os().clear_error();
232     sys::fs::remove(Twine(Filename));
233     return false;
234   }
235 
236   objFile.keep();
237   if (!genResult) {
238     sys::fs::remove(Twine(Filename));
239     return false;
240   }
241 
242   NativeObjectPath = Filename.c_str();
243   *Name = NativeObjectPath.c_str();
244   return true;
245 }
246 
247 std::unique_ptr<MemoryBuffer>
248 LTOCodeGenerator::compileOptimized(std::string &ErrMsg) {
249   const char *name;
250   if (!compileOptimizedToFile(&name, ErrMsg))
251     return nullptr;
252 
253   // read .o file into memory buffer
254   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
255       MemoryBuffer::getFile(name, -1, false);
256   if (std::error_code EC = BufferOrErr.getError()) {
257     ErrMsg = EC.message();
258     sys::fs::remove(NativeObjectPath);
259     return nullptr;
260   }
261 
262   // remove temp files
263   sys::fs::remove(NativeObjectPath);
264 
265   return std::move(*BufferOrErr);
266 }
267 
268 bool LTOCodeGenerator::compile_to_file(const char **Name, bool DisableVerify,
269                                        bool DisableInline,
270                                        bool DisableGVNLoadPRE,
271                                        bool DisableVectorization,
272                                        std::string &ErrMsg) {
273   if (!optimize(DisableVerify, DisableInline, DisableGVNLoadPRE,
274                 DisableVectorization, ErrMsg))
275     return false;
276 
277   return compileOptimizedToFile(Name, ErrMsg);
278 }
279 
280 std::unique_ptr<MemoryBuffer>
281 LTOCodeGenerator::compile(bool DisableVerify, bool DisableInline,
282                           bool DisableGVNLoadPRE, bool DisableVectorization,
283                           std::string &ErrMsg) {
284   if (!optimize(DisableVerify, DisableInline, DisableGVNLoadPRE,
285                 DisableVectorization, ErrMsg))
286     return nullptr;
287 
288   return compileOptimized(ErrMsg);
289 }
290 
291 bool LTOCodeGenerator::determineTarget(std::string &ErrMsg) {
292   if (TargetMach)
293     return true;
294 
295   std::string TripleStr = MergedModule->getTargetTriple();
296   if (TripleStr.empty()) {
297     TripleStr = sys::getDefaultTargetTriple();
298     MergedModule->setTargetTriple(TripleStr);
299   }
300   llvm::Triple Triple(TripleStr);
301 
302   // create target machine from info for merged modules
303   const Target *march = TargetRegistry::lookupTarget(TripleStr, ErrMsg);
304   if (!march)
305     return false;
306 
307   // Construct LTOModule, hand over ownership of module and target. Use MAttr as
308   // the default set of features.
309   SubtargetFeatures Features(MAttr);
310   Features.getDefaultSubtargetFeatures(Triple);
311   FeatureStr = Features.getString();
312   // Set a default CPU for Darwin triples.
313   if (MCpu.empty() && Triple.isOSDarwin()) {
314     if (Triple.getArch() == llvm::Triple::x86_64)
315       MCpu = "core2";
316     else if (Triple.getArch() == llvm::Triple::x86)
317       MCpu = "yonah";
318     else if (Triple.getArch() == llvm::Triple::aarch64)
319       MCpu = "cyclone";
320   }
321 
322   TargetMach.reset(march->createTargetMachine(TripleStr, MCpu, FeatureStr,
323                                               Options, RelocModel,
324                                               CodeModel::Default, CGOptLevel));
325   return true;
326 }
327 
328 void LTOCodeGenerator::
329 applyRestriction(GlobalValue &GV,
330                  ArrayRef<StringRef> Libcalls,
331                  std::vector<const char*> &MustPreserveList,
332                  SmallPtrSetImpl<GlobalValue*> &AsmUsed,
333                  Mangler &Mangler) {
334   // There are no restrictions to apply to declarations.
335   if (GV.isDeclaration())
336     return;
337 
338   // There is nothing more restrictive than private linkage.
339   if (GV.hasPrivateLinkage())
340     return;
341 
342   SmallString<64> Buffer;
343   TargetMach->getNameWithPrefix(Buffer, &GV, Mangler);
344 
345   if (MustPreserveSymbols.count(Buffer))
346     MustPreserveList.push_back(GV.getName().data());
347   if (AsmUndefinedRefs.count(Buffer))
348     AsmUsed.insert(&GV);
349 
350   // Conservatively append user-supplied runtime library functions to
351   // llvm.compiler.used.  These could be internalized and deleted by
352   // optimizations like -globalopt, causing problems when later optimizations
353   // add new library calls (e.g., llvm.memset => memset and printf => puts).
354   // Leave it to the linker to remove any dead code (e.g. with -dead_strip).
355   if (isa<Function>(GV) &&
356       std::binary_search(Libcalls.begin(), Libcalls.end(), GV.getName()))
357     AsmUsed.insert(&GV);
358 }
359 
360 static void findUsedValues(GlobalVariable *LLVMUsed,
361                            SmallPtrSetImpl<GlobalValue*> &UsedValues) {
362   if (!LLVMUsed) return;
363 
364   ConstantArray *Inits = cast<ConstantArray>(LLVMUsed->getInitializer());
365   for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
366     if (GlobalValue *GV =
367         dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
368       UsedValues.insert(GV);
369 }
370 
371 // Collect names of runtime library functions. User-defined functions with the
372 // same names are added to llvm.compiler.used to prevent them from being
373 // deleted by optimizations.
374 static void accumulateAndSortLibcalls(std::vector<StringRef> &Libcalls,
375                                       const TargetLibraryInfo& TLI,
376                                       const Module &Mod,
377                                       const TargetMachine &TM) {
378   // TargetLibraryInfo has info on C runtime library calls on the current
379   // target.
380   for (unsigned I = 0, E = static_cast<unsigned>(LibFunc::NumLibFuncs);
381        I != E; ++I) {
382     LibFunc::Func F = static_cast<LibFunc::Func>(I);
383     if (TLI.has(F))
384       Libcalls.push_back(TLI.getName(F));
385   }
386 
387   SmallPtrSet<const TargetLowering *, 1> TLSet;
388 
389   for (const Function &F : Mod) {
390     const TargetLowering *Lowering =
391         TM.getSubtargetImpl(F)->getTargetLowering();
392 
393     if (Lowering && TLSet.insert(Lowering).second)
394       // TargetLowering has info on library calls that CodeGen expects to be
395       // available, both from the C runtime and compiler-rt.
396       for (unsigned I = 0, E = static_cast<unsigned>(RTLIB::UNKNOWN_LIBCALL);
397            I != E; ++I)
398         if (const char *Name =
399                 Lowering->getLibcallName(static_cast<RTLIB::Libcall>(I)))
400           Libcalls.push_back(Name);
401   }
402 
403   array_pod_sort(Libcalls.begin(), Libcalls.end());
404   Libcalls.erase(std::unique(Libcalls.begin(), Libcalls.end()),
405                  Libcalls.end());
406 }
407 
408 void LTOCodeGenerator::applyScopeRestrictions() {
409   if (ScopeRestrictionsDone || !ShouldInternalize)
410     return;
411 
412   // Start off with a verification pass.
413   legacy::PassManager passes;
414   passes.add(createVerifierPass());
415 
416   // mark which symbols can not be internalized
417   Mangler Mangler;
418   std::vector<const char*> MustPreserveList;
419   SmallPtrSet<GlobalValue*, 8> AsmUsed;
420   std::vector<StringRef> Libcalls;
421   TargetLibraryInfoImpl TLII(Triple(TargetMach->getTargetTriple()));
422   TargetLibraryInfo TLI(TLII);
423 
424   accumulateAndSortLibcalls(Libcalls, TLI, *MergedModule, *TargetMach);
425 
426   for (Function &f : *MergedModule)
427     applyRestriction(f, Libcalls, MustPreserveList, AsmUsed, Mangler);
428   for (GlobalVariable &v : MergedModule->globals())
429     applyRestriction(v, Libcalls, MustPreserveList, AsmUsed, Mangler);
430   for (GlobalAlias &a : MergedModule->aliases())
431     applyRestriction(a, Libcalls, MustPreserveList, AsmUsed, Mangler);
432 
433   GlobalVariable *LLVMCompilerUsed =
434     MergedModule->getGlobalVariable("llvm.compiler.used");
435   findUsedValues(LLVMCompilerUsed, AsmUsed);
436   if (LLVMCompilerUsed)
437     LLVMCompilerUsed->eraseFromParent();
438 
439   if (!AsmUsed.empty()) {
440     llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(Context);
441     std::vector<Constant*> asmUsed2;
442     for (auto *GV : AsmUsed) {
443       Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
444       asmUsed2.push_back(c);
445     }
446 
447     llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
448     LLVMCompilerUsed =
449       new llvm::GlobalVariable(*MergedModule, ATy, false,
450                                llvm::GlobalValue::AppendingLinkage,
451                                llvm::ConstantArray::get(ATy, asmUsed2),
452                                "llvm.compiler.used");
453 
454     LLVMCompilerUsed->setSection("llvm.metadata");
455   }
456 
457   passes.add(createInternalizePass(MustPreserveList));
458 
459   // apply scope restrictions
460   passes.run(*MergedModule);
461 
462   ScopeRestrictionsDone = true;
463 }
464 
465 /// Optimize merged modules using various IPO passes
466 bool LTOCodeGenerator::optimize(bool DisableVerify, bool DisableInline,
467                                 bool DisableGVNLoadPRE,
468                                 bool DisableVectorization,
469                                 std::string &ErrMsg) {
470   if (!this->determineTarget(ErrMsg))
471     return false;
472 
473   // Mark which symbols can not be internalized
474   this->applyScopeRestrictions();
475 
476   // Instantiate the pass manager to organize the passes.
477   legacy::PassManager passes;
478 
479   // Add an appropriate DataLayout instance for this module...
480   MergedModule->setDataLayout(TargetMach->createDataLayout());
481 
482   passes.add(
483       createTargetTransformInfoWrapperPass(TargetMach->getTargetIRAnalysis()));
484 
485   Triple TargetTriple(TargetMach->getTargetTriple());
486   PassManagerBuilder PMB;
487   PMB.DisableGVNLoadPRE = DisableGVNLoadPRE;
488   PMB.LoopVectorize = !DisableVectorization;
489   PMB.SLPVectorize = !DisableVectorization;
490   if (!DisableInline)
491     PMB.Inliner = createFunctionInliningPass();
492   PMB.LibraryInfo = new TargetLibraryInfoImpl(TargetTriple);
493   PMB.OptLevel = OptLevel;
494   PMB.VerifyInput = !DisableVerify;
495   PMB.VerifyOutput = !DisableVerify;
496 
497   PMB.populateLTOPassManager(passes);
498 
499   // Run our queue of passes all at once now, efficiently.
500   passes.run(*MergedModule);
501 
502   return true;
503 }
504 
505 bool LTOCodeGenerator::compileOptimized(ArrayRef<raw_pwrite_stream *> Out,
506                                         std::string &ErrMsg) {
507   if (!this->determineTarget(ErrMsg))
508     return false;
509 
510   legacy::PassManager preCodeGenPasses;
511 
512   // If the bitcode files contain ARC code and were compiled with optimization,
513   // the ObjCARCContractPass must be run, so do it unconditionally here.
514   preCodeGenPasses.add(createObjCARCContractPass());
515   preCodeGenPasses.run(*MergedModule);
516 
517   // Do code generation. We need to preserve the module in case the client calls
518   // writeMergedModules() after compilation, but we only need to allow this at
519   // parallelism level 1. This is achieved by having splitCodeGen return the
520   // original module at parallelism level 1 which we then assign back to
521   // MergedModule.
522   MergedModule =
523       splitCodeGen(std::move(MergedModule), Out, MCpu, FeatureStr, Options,
524                    RelocModel, CodeModel::Default, CGOptLevel);
525 
526   return true;
527 }
528 
529 /// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging
530 /// LTO problems.
531 void LTOCodeGenerator::setCodeGenDebugOptions(const char *Options) {
532   for (std::pair<StringRef, StringRef> o = getToken(Options); !o.first.empty();
533        o = getToken(o.second))
534     CodegenOptions.push_back(o.first);
535 }
536 
537 void LTOCodeGenerator::parseCodeGenDebugOptions() {
538   // if options were requested, set them
539   if (!CodegenOptions.empty()) {
540     // ParseCommandLineOptions() expects argv[0] to be program name.
541     std::vector<const char *> CodegenArgv(1, "libLLVMLTO");
542     for (std::string &Arg : CodegenOptions)
543       CodegenArgv.push_back(Arg.c_str());
544     cl::ParseCommandLineOptions(CodegenArgv.size(), CodegenArgv.data());
545   }
546 }
547 
548 void LTOCodeGenerator::DiagnosticHandler(const DiagnosticInfo &DI,
549                                          void *Context) {
550   ((LTOCodeGenerator *)Context)->DiagnosticHandler2(DI);
551 }
552 
553 void LTOCodeGenerator::DiagnosticHandler2(const DiagnosticInfo &DI) {
554   // Map the LLVM internal diagnostic severity to the LTO diagnostic severity.
555   lto_codegen_diagnostic_severity_t Severity;
556   switch (DI.getSeverity()) {
557   case DS_Error:
558     Severity = LTO_DS_ERROR;
559     break;
560   case DS_Warning:
561     Severity = LTO_DS_WARNING;
562     break;
563   case DS_Remark:
564     Severity = LTO_DS_REMARK;
565     break;
566   case DS_Note:
567     Severity = LTO_DS_NOTE;
568     break;
569   }
570   // Create the string that will be reported to the external diagnostic handler.
571   std::string MsgStorage;
572   raw_string_ostream Stream(MsgStorage);
573   DiagnosticPrinterRawOStream DP(Stream);
574   DI.print(DP);
575   Stream.flush();
576 
577   // If this method has been called it means someone has set up an external
578   // diagnostic handler. Assert on that.
579   assert(DiagHandler && "Invalid diagnostic handler");
580   (*DiagHandler)(Severity, MsgStorage.c_str(), DiagContext);
581 }
582 
583 void
584 LTOCodeGenerator::setDiagnosticHandler(lto_diagnostic_handler_t DiagHandler,
585                                        void *Ctxt) {
586   this->DiagHandler = DiagHandler;
587   this->DiagContext = Ctxt;
588   if (!DiagHandler)
589     return Context.setDiagnosticHandler(nullptr, nullptr);
590   // Register the LTOCodeGenerator stub in the LLVMContext to forward the
591   // diagnostic to the external DiagHandler.
592   Context.setDiagnosticHandler(LTOCodeGenerator::DiagnosticHandler, this,
593                                /* RespectFilters */ true);
594 }
595