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/LegacyPassManager.h"
30 #include "llvm/IR/Mangler.h"
31 #include "llvm/IR/Module.h"
32 #include "llvm/IR/Verifier.h"
33 #include "llvm/InitializePasses.h"
34 #include "llvm/LTO/LTOModule.h"
35 #include "llvm/Linker/Linker.h"
36 #include "llvm/MC/MCAsmInfo.h"
37 #include "llvm/MC/MCContext.h"
38 #include "llvm/MC/SubtargetFeature.h"
39 #include "llvm/Support/CommandLine.h"
40 #include "llvm/Support/FileSystem.h"
41 #include "llvm/Support/Host.h"
42 #include "llvm/Support/MemoryBuffer.h"
43 #include "llvm/Support/Signals.h"
44 #include "llvm/Support/TargetRegistry.h"
45 #include "llvm/Support/TargetSelect.h"
46 #include "llvm/Support/ToolOutputFile.h"
47 #include "llvm/Support/raw_ostream.h"
48 #include "llvm/Target/TargetLowering.h"
49 #include "llvm/Target/TargetOptions.h"
50 #include "llvm/Target/TargetRegisterInfo.h"
51 #include "llvm/Target/TargetSubtargetInfo.h"
52 #include "llvm/Transforms/IPO.h"
53 #include "llvm/Transforms/IPO/PassManagerBuilder.h"
54 #include "llvm/Transforms/ObjCARC.h"
55 #include <system_error>
56 using namespace llvm;
57 
58 const char* LTOCodeGenerator::getVersionString() {
59 #ifdef LLVM_VERSION_INFO
60   return PACKAGE_NAME " version " PACKAGE_VERSION ", " LLVM_VERSION_INFO;
61 #else
62   return PACKAGE_NAME " version " PACKAGE_VERSION;
63 #endif
64 }
65 
66 LTOCodeGenerator::LTOCodeGenerator()
67     : Context(getGlobalContext()),
68       MergedModule(new Module("ld-temp.o", Context)),
69       IRLinker(MergedModule.get()) {
70   initializeLTOPasses();
71 }
72 
73 LTOCodeGenerator::LTOCodeGenerator(std::unique_ptr<LLVMContext> Context)
74     : OwnedContext(std::move(Context)), Context(*OwnedContext),
75       MergedModule(new Module("ld-temp.o", *OwnedContext)),
76       IRLinker(MergedModule.get()) {
77   initializeLTOPasses();
78 }
79 
80 LTOCodeGenerator::~LTOCodeGenerator() {}
81 
82 // Initialize LTO passes. Please keep this funciton in sync with
83 // PassManagerBuilder::populateLTOPassManager(), and make sure all LTO
84 // passes are initialized.
85 void LTOCodeGenerator::initializeLTOPasses() {
86   PassRegistry &R = *PassRegistry::getPassRegistry();
87 
88   initializeInternalizePassPass(R);
89   initializeIPSCCPPass(R);
90   initializeGlobalOptPass(R);
91   initializeConstantMergePass(R);
92   initializeDAHPass(R);
93   initializeInstructionCombiningPassPass(R);
94   initializeSimpleInlinerPass(R);
95   initializePruneEHPass(R);
96   initializeGlobalDCEPass(R);
97   initializeArgPromotionPass(R);
98   initializeJumpThreadingPass(R);
99   initializeSROAPass(R);
100   initializeSROA_DTPass(R);
101   initializeSROA_SSAUpPass(R);
102   initializeFunctionAttrsPass(R);
103   initializeGlobalsModRefPass(R);
104   initializeLICMPass(R);
105   initializeMergedLoadStoreMotionPass(R);
106   initializeGVNPass(R);
107   initializeMemCpyOptPass(R);
108   initializeDCEPass(R);
109   initializeCFGSimplifyPassPass(R);
110 }
111 
112 bool LTOCodeGenerator::addModule(LTOModule *mod) {
113   assert(&mod->getModule().getContext() == &Context &&
114          "Expected module in same context");
115 
116   bool ret = IRLinker.linkInModule(&mod->getModule());
117 
118   const std::vector<const char*> &undefs = mod->getAsmUndefinedRefs();
119   for (int i = 0, e = undefs.size(); i != e; ++i)
120     AsmUndefinedRefs[undefs[i]] = 1;
121 
122   return !ret;
123 }
124 
125 void LTOCodeGenerator::setModule(std::unique_ptr<LTOModule> Mod) {
126   assert(&Mod->getModule().getContext() == &Context &&
127          "Expected module in same context");
128 
129   AsmUndefinedRefs.clear();
130 
131   MergedModule = Mod->takeModule();
132   IRLinker.setModule(MergedModule.get());
133 
134   const std::vector<const char*> &Undefs = Mod->getAsmUndefinedRefs();
135   for (int I = 0, E = Undefs.size(); I != E; ++I)
136     AsmUndefinedRefs[Undefs[I]] = 1;
137 }
138 
139 void LTOCodeGenerator::setTargetOptions(TargetOptions options) {
140   Options = options;
141 }
142 
143 void LTOCodeGenerator::setDebugInfo(lto_debug_model debug) {
144   switch (debug) {
145   case LTO_DEBUG_MODEL_NONE:
146     EmitDwarfDebugInfo = false;
147     return;
148 
149   case LTO_DEBUG_MODEL_DWARF:
150     EmitDwarfDebugInfo = true;
151     return;
152   }
153   llvm_unreachable("Unknown debug format!");
154 }
155 
156 void LTOCodeGenerator::setOptLevel(unsigned level) {
157   OptLevel = level;
158   switch (OptLevel) {
159   case 0:
160     CGOptLevel = CodeGenOpt::None;
161     break;
162   case 1:
163     CGOptLevel = CodeGenOpt::Less;
164     break;
165   case 2:
166     CGOptLevel = CodeGenOpt::Default;
167     break;
168   case 3:
169     CGOptLevel = CodeGenOpt::Aggressive;
170     break;
171   }
172 }
173 
174 bool LTOCodeGenerator::writeMergedModules(const char *path,
175                                           std::string &errMsg) {
176   if (!determineTarget(errMsg))
177     return false;
178 
179   // mark which symbols can not be internalized
180   applyScopeRestrictions();
181 
182   // create output file
183   std::error_code EC;
184   tool_output_file Out(path, EC, sys::fs::F_None);
185   if (EC) {
186     errMsg = "could not open bitcode file for writing: ";
187     errMsg += path;
188     return false;
189   }
190 
191   // write bitcode to it
192   WriteBitcodeToFile(MergedModule.get(), Out.os(), ShouldEmbedUselists);
193   Out.os().close();
194 
195   if (Out.os().has_error()) {
196     errMsg = "could not write bitcode file: ";
197     errMsg += path;
198     Out.os().clear_error();
199     return false;
200   }
201 
202   Out.keep();
203   return true;
204 }
205 
206 bool LTOCodeGenerator::compileOptimizedToFile(const char **name,
207                                               std::string &errMsg) {
208   // make unique temp .o file to put generated object file
209   SmallString<128> Filename;
210   int FD;
211   std::error_code EC =
212       sys::fs::createTemporaryFile("lto-llvm", "o", FD, Filename);
213   if (EC) {
214     errMsg = EC.message();
215     return false;
216   }
217 
218   // generate object file
219   tool_output_file objFile(Filename.c_str(), FD);
220 
221   bool genResult = compileOptimized(objFile.os(), errMsg);
222   objFile.os().close();
223   if (objFile.os().has_error()) {
224     objFile.os().clear_error();
225     sys::fs::remove(Twine(Filename));
226     return false;
227   }
228 
229   objFile.keep();
230   if (!genResult) {
231     sys::fs::remove(Twine(Filename));
232     return false;
233   }
234 
235   NativeObjectPath = Filename.c_str();
236   *name = NativeObjectPath.c_str();
237   return true;
238 }
239 
240 std::unique_ptr<MemoryBuffer>
241 LTOCodeGenerator::compileOptimized(std::string &errMsg) {
242   const char *name;
243   if (!compileOptimizedToFile(&name, errMsg))
244     return nullptr;
245 
246   // read .o file into memory buffer
247   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
248       MemoryBuffer::getFile(name, -1, false);
249   if (std::error_code EC = BufferOrErr.getError()) {
250     errMsg = EC.message();
251     sys::fs::remove(NativeObjectPath);
252     return nullptr;
253   }
254 
255   // remove temp files
256   sys::fs::remove(NativeObjectPath);
257 
258   return std::move(*BufferOrErr);
259 }
260 
261 
262 bool LTOCodeGenerator::compile_to_file(const char **name,
263                                        bool disableInline,
264                                        bool disableGVNLoadPRE,
265                                        bool disableVectorization,
266                                        std::string &errMsg) {
267   if (!optimize(disableInline, disableGVNLoadPRE,
268                 disableVectorization, errMsg))
269     return false;
270 
271   return compileOptimizedToFile(name, errMsg);
272 }
273 
274 std::unique_ptr<MemoryBuffer>
275 LTOCodeGenerator::compile(bool disableInline, bool disableGVNLoadPRE,
276                           bool disableVectorization, std::string &errMsg) {
277   if (!optimize(disableInline, disableGVNLoadPRE,
278                 disableVectorization, errMsg))
279     return nullptr;
280 
281   return compileOptimized(errMsg);
282 }
283 
284 bool LTOCodeGenerator::determineTarget(std::string &errMsg) {
285   if (TargetMach)
286     return true;
287 
288   std::string TripleStr = MergedModule->getTargetTriple();
289   if (TripleStr.empty()) {
290     TripleStr = sys::getDefaultTargetTriple();
291     MergedModule->setTargetTriple(TripleStr);
292   }
293   llvm::Triple Triple(TripleStr);
294 
295   // create target machine from info for merged modules
296   const Target *march = TargetRegistry::lookupTarget(TripleStr, errMsg);
297   if (!march)
298     return false;
299 
300   // Construct LTOModule, hand over ownership of module and target. Use MAttr as
301   // the default set of features.
302   SubtargetFeatures Features(MAttr);
303   Features.getDefaultSubtargetFeatures(Triple);
304   FeatureStr = Features.getString();
305   // Set a default CPU for Darwin triples.
306   if (MCpu.empty() && Triple.isOSDarwin()) {
307     if (Triple.getArch() == llvm::Triple::x86_64)
308       MCpu = "core2";
309     else if (Triple.getArch() == llvm::Triple::x86)
310       MCpu = "yonah";
311     else if (Triple.getArch() == llvm::Triple::aarch64)
312       MCpu = "cyclone";
313   }
314 
315   TargetMach.reset(march->createTargetMachine(TripleStr, MCpu, FeatureStr,
316                                               Options, RelocModel,
317                                               CodeModel::Default, CGOptLevel));
318   return true;
319 }
320 
321 void LTOCodeGenerator::
322 applyRestriction(GlobalValue &GV,
323                  ArrayRef<StringRef> Libcalls,
324                  std::vector<const char*> &MustPreserveList,
325                  SmallPtrSetImpl<GlobalValue*> &AsmUsed,
326                  Mangler &Mangler) {
327   // There are no restrictions to apply to declarations.
328   if (GV.isDeclaration())
329     return;
330 
331   // There is nothing more restrictive than private linkage.
332   if (GV.hasPrivateLinkage())
333     return;
334 
335   SmallString<64> Buffer;
336   TargetMach->getNameWithPrefix(Buffer, &GV, Mangler);
337 
338   if (MustPreserveSymbols.count(Buffer))
339     MustPreserveList.push_back(GV.getName().data());
340   if (AsmUndefinedRefs.count(Buffer))
341     AsmUsed.insert(&GV);
342 
343   // Conservatively append user-supplied runtime library functions to
344   // llvm.compiler.used.  These could be internalized and deleted by
345   // optimizations like -globalopt, causing problems when later optimizations
346   // add new library calls (e.g., llvm.memset => memset and printf => puts).
347   // Leave it to the linker to remove any dead code (e.g. with -dead_strip).
348   if (isa<Function>(GV) &&
349       std::binary_search(Libcalls.begin(), Libcalls.end(), GV.getName()))
350     AsmUsed.insert(&GV);
351 }
352 
353 static void findUsedValues(GlobalVariable *LLVMUsed,
354                            SmallPtrSetImpl<GlobalValue*> &UsedValues) {
355   if (!LLVMUsed) return;
356 
357   ConstantArray *Inits = cast<ConstantArray>(LLVMUsed->getInitializer());
358   for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
359     if (GlobalValue *GV =
360         dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
361       UsedValues.insert(GV);
362 }
363 
364 // Collect names of runtime library functions. User-defined functions with the
365 // same names are added to llvm.compiler.used to prevent them from being
366 // deleted by optimizations.
367 static void accumulateAndSortLibcalls(std::vector<StringRef> &Libcalls,
368                                       const TargetLibraryInfo& TLI,
369                                       const Module &Mod,
370                                       const TargetMachine &TM) {
371   // TargetLibraryInfo has info on C runtime library calls on the current
372   // target.
373   for (unsigned I = 0, E = static_cast<unsigned>(LibFunc::NumLibFuncs);
374        I != E; ++I) {
375     LibFunc::Func F = static_cast<LibFunc::Func>(I);
376     if (TLI.has(F))
377       Libcalls.push_back(TLI.getName(F));
378   }
379 
380   SmallPtrSet<const TargetLowering *, 1> TLSet;
381 
382   for (const Function &F : Mod) {
383     const TargetLowering *Lowering =
384         TM.getSubtargetImpl(F)->getTargetLowering();
385 
386     if (Lowering && TLSet.insert(Lowering).second)
387       // TargetLowering has info on library calls that CodeGen expects to be
388       // available, both from the C runtime and compiler-rt.
389       for (unsigned I = 0, E = static_cast<unsigned>(RTLIB::UNKNOWN_LIBCALL);
390            I != E; ++I)
391         if (const char *Name =
392                 Lowering->getLibcallName(static_cast<RTLIB::Libcall>(I)))
393           Libcalls.push_back(Name);
394   }
395 
396   array_pod_sort(Libcalls.begin(), Libcalls.end());
397   Libcalls.erase(std::unique(Libcalls.begin(), Libcalls.end()),
398                  Libcalls.end());
399 }
400 
401 void LTOCodeGenerator::applyScopeRestrictions() {
402   if (ScopeRestrictionsDone || !ShouldInternalize)
403     return;
404 
405   // Start off with a verification pass.
406   legacy::PassManager passes;
407   passes.add(createVerifierPass());
408 
409   // mark which symbols can not be internalized
410   Mangler Mangler;
411   std::vector<const char*> MustPreserveList;
412   SmallPtrSet<GlobalValue*, 8> AsmUsed;
413   std::vector<StringRef> Libcalls;
414   TargetLibraryInfoImpl TLII(Triple(TargetMach->getTargetTriple()));
415   TargetLibraryInfo TLI(TLII);
416 
417   accumulateAndSortLibcalls(Libcalls, TLI, *MergedModule, *TargetMach);
418 
419   for (Function &f : *MergedModule)
420     applyRestriction(f, Libcalls, MustPreserveList, AsmUsed, Mangler);
421   for (GlobalVariable &v : MergedModule->globals())
422     applyRestriction(v, Libcalls, MustPreserveList, AsmUsed, Mangler);
423   for (GlobalAlias &a : MergedModule->aliases())
424     applyRestriction(a, Libcalls, MustPreserveList, AsmUsed, Mangler);
425 
426   GlobalVariable *LLVMCompilerUsed =
427     MergedModule->getGlobalVariable("llvm.compiler.used");
428   findUsedValues(LLVMCompilerUsed, AsmUsed);
429   if (LLVMCompilerUsed)
430     LLVMCompilerUsed->eraseFromParent();
431 
432   if (!AsmUsed.empty()) {
433     llvm::Type *i8PTy = llvm::Type::getInt8PtrTy(Context);
434     std::vector<Constant*> asmUsed2;
435     for (auto *GV : AsmUsed) {
436       Constant *c = ConstantExpr::getBitCast(GV, i8PTy);
437       asmUsed2.push_back(c);
438     }
439 
440     llvm::ArrayType *ATy = llvm::ArrayType::get(i8PTy, asmUsed2.size());
441     LLVMCompilerUsed =
442       new llvm::GlobalVariable(*MergedModule, ATy, false,
443                                llvm::GlobalValue::AppendingLinkage,
444                                llvm::ConstantArray::get(ATy, asmUsed2),
445                                "llvm.compiler.used");
446 
447     LLVMCompilerUsed->setSection("llvm.metadata");
448   }
449 
450   passes.add(createInternalizePass(MustPreserveList));
451 
452   // apply scope restrictions
453   passes.run(*MergedModule);
454 
455   ScopeRestrictionsDone = true;
456 }
457 
458 /// Optimize merged modules using various IPO passes
459 bool LTOCodeGenerator::optimize(bool DisableInline,
460                                 bool DisableGVNLoadPRE,
461                                 bool DisableVectorization,
462                                 std::string &errMsg) {
463   if (!this->determineTarget(errMsg))
464     return false;
465 
466   // Mark which symbols can not be internalized
467   this->applyScopeRestrictions();
468 
469   // Instantiate the pass manager to organize the passes.
470   legacy::PassManager passes;
471 
472   // Add an appropriate DataLayout instance for this module...
473   MergedModule->setDataLayout(TargetMach->createDataLayout());
474 
475   passes.add(
476       createTargetTransformInfoWrapperPass(TargetMach->getTargetIRAnalysis()));
477 
478   Triple TargetTriple(TargetMach->getTargetTriple());
479   PassManagerBuilder PMB;
480   PMB.DisableGVNLoadPRE = DisableGVNLoadPRE;
481   PMB.LoopVectorize = !DisableVectorization;
482   PMB.SLPVectorize = !DisableVectorization;
483   if (!DisableInline)
484     PMB.Inliner = createFunctionInliningPass();
485   PMB.LibraryInfo = new TargetLibraryInfoImpl(TargetTriple);
486   PMB.OptLevel = OptLevel;
487   PMB.VerifyInput = true;
488   PMB.VerifyOutput = true;
489 
490   PMB.populateLTOPassManager(passes);
491 
492   // Run our queue of passes all at once now, efficiently.
493   passes.run(*MergedModule);
494 
495   return true;
496 }
497 
498 bool LTOCodeGenerator::compileOptimized(raw_pwrite_stream &out,
499                                         std::string &errMsg) {
500   if (!this->determineTarget(errMsg))
501     return false;
502 
503   legacy::PassManager codeGenPasses;
504 
505   // If the bitcode files contain ARC code and were compiled with optimization,
506   // the ObjCARCContractPass must be run, so do it unconditionally here.
507   codeGenPasses.add(createObjCARCContractPass());
508 
509   if (TargetMach->addPassesToEmitFile(codeGenPasses, out,
510                                       TargetMachine::CGFT_ObjectFile)) {
511     errMsg = "target file type not supported";
512     return false;
513   }
514 
515   // Run the code generator, and write object file
516   codeGenPasses.run(*MergedModule);
517 
518   return true;
519 }
520 
521 /// setCodeGenDebugOptions - Set codegen debugging options to aid in debugging
522 /// LTO problems.
523 void LTOCodeGenerator::setCodeGenDebugOptions(const char *options) {
524   for (std::pair<StringRef, StringRef> o = getToken(options);
525        !o.first.empty(); o = getToken(o.second))
526     CodegenOptions.push_back(o.first);
527 }
528 
529 void LTOCodeGenerator::parseCodeGenDebugOptions() {
530   // if options were requested, set them
531   if (!CodegenOptions.empty()) {
532     // ParseCommandLineOptions() expects argv[0] to be program name.
533     std::vector<const char *> CodegenArgv(1, "libLLVMLTO");
534     for (std::string &Arg : CodegenOptions)
535       CodegenArgv.push_back(Arg.c_str());
536     cl::ParseCommandLineOptions(CodegenArgv.size(), CodegenArgv.data());
537   }
538 }
539 
540 void LTOCodeGenerator::DiagnosticHandler(const DiagnosticInfo &DI,
541                                          void *Context) {
542   ((LTOCodeGenerator *)Context)->DiagnosticHandler2(DI);
543 }
544 
545 void LTOCodeGenerator::DiagnosticHandler2(const DiagnosticInfo &DI) {
546   // Map the LLVM internal diagnostic severity to the LTO diagnostic severity.
547   lto_codegen_diagnostic_severity_t Severity;
548   switch (DI.getSeverity()) {
549   case DS_Error:
550     Severity = LTO_DS_ERROR;
551     break;
552   case DS_Warning:
553     Severity = LTO_DS_WARNING;
554     break;
555   case DS_Remark:
556     Severity = LTO_DS_REMARK;
557     break;
558   case DS_Note:
559     Severity = LTO_DS_NOTE;
560     break;
561   }
562   // Create the string that will be reported to the external diagnostic handler.
563   std::string MsgStorage;
564   raw_string_ostream Stream(MsgStorage);
565   DiagnosticPrinterRawOStream DP(Stream);
566   DI.print(DP);
567   Stream.flush();
568 
569   // If this method has been called it means someone has set up an external
570   // diagnostic handler. Assert on that.
571   assert(DiagHandler && "Invalid diagnostic handler");
572   (*DiagHandler)(Severity, MsgStorage.c_str(), DiagContext);
573 }
574 
575 void
576 LTOCodeGenerator::setDiagnosticHandler(lto_diagnostic_handler_t DiagHandler,
577                                        void *Ctxt) {
578   this->DiagHandler = DiagHandler;
579   this->DiagContext = Ctxt;
580   if (!DiagHandler)
581     return Context.setDiagnosticHandler(nullptr, nullptr);
582   // Register the LTOCodeGenerator stub in the LLVMContext to forward the
583   // diagnostic to the external DiagHandler.
584   Context.setDiagnosticHandler(LTOCodeGenerator::DiagnosticHandler, this,
585                                /* RespectFilters */ true);
586 }
587