1 //===- GCOVProfiling.cpp - Insert edge counters for gcov profiling --------===//
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 pass implements GCOV-style profiling. When this pass is run it emits
11 // "gcno" files next to the existing source, and instruments the code that runs
12 // to records the edges between blocks that run and emit a complementary "gcda"
13 // file on exit.
14 //
15 //===----------------------------------------------------------------------===//
16 
17 #include "llvm/ADT/DenseMap.h"
18 #include "llvm/ADT/Hashing.h"
19 #include "llvm/ADT/STLExtras.h"
20 #include "llvm/ADT/Sequence.h"
21 #include "llvm/ADT/Statistic.h"
22 #include "llvm/ADT/StringExtras.h"
23 #include "llvm/ADT/StringMap.h"
24 #include "llvm/Analysis/EHPersonalities.h"
25 #include "llvm/Analysis/TargetLibraryInfo.h"
26 #include "llvm/IR/CFG.h"
27 #include "llvm/IR/DebugInfo.h"
28 #include "llvm/IR/DebugLoc.h"
29 #include "llvm/IR/IRBuilder.h"
30 #include "llvm/IR/InstIterator.h"
31 #include "llvm/IR/Instructions.h"
32 #include "llvm/IR/IntrinsicInst.h"
33 #include "llvm/IR/Module.h"
34 #include "llvm/Pass.h"
35 #include "llvm/Support/CommandLine.h"
36 #include "llvm/Support/Debug.h"
37 #include "llvm/Support/FileSystem.h"
38 #include "llvm/Support/Path.h"
39 #include "llvm/Support/raw_ostream.h"
40 #include "llvm/Transforms/Instrumentation.h"
41 #include "llvm/Transforms/Instrumentation/GCOVProfiler.h"
42 #include "llvm/Transforms/Utils/ModuleUtils.h"
43 #include <algorithm>
44 #include <memory>
45 #include <string>
46 #include <utility>
47 using namespace llvm;
48 
49 #define DEBUG_TYPE "insert-gcov-profiling"
50 
51 static cl::opt<std::string>
52 DefaultGCOVVersion("default-gcov-version", cl::init("402*"), cl::Hidden,
53                    cl::ValueRequired);
54 static cl::opt<bool> DefaultExitBlockBeforeBody("gcov-exit-block-before-body",
55                                                 cl::init(false), cl::Hidden);
56 
57 GCOVOptions GCOVOptions::getDefault() {
58   GCOVOptions Options;
59   Options.EmitNotes = true;
60   Options.EmitData = true;
61   Options.UseCfgChecksum = false;
62   Options.NoRedZone = false;
63   Options.FunctionNamesInData = true;
64   Options.ExitBlockBeforeBody = DefaultExitBlockBeforeBody;
65 
66   if (DefaultGCOVVersion.size() != 4) {
67     llvm::report_fatal_error(std::string("Invalid -default-gcov-version: ") +
68                              DefaultGCOVVersion);
69   }
70   memcpy(Options.Version, DefaultGCOVVersion.c_str(), 4);
71   return Options;
72 }
73 
74 namespace {
75 class GCOVFunction;
76 
77 class GCOVProfiler {
78 public:
79   GCOVProfiler() : GCOVProfiler(GCOVOptions::getDefault()) {}
80   GCOVProfiler(const GCOVOptions &Opts) : Options(Opts) {
81     assert((Options.EmitNotes || Options.EmitData) &&
82            "GCOVProfiler asked to do nothing?");
83     ReversedVersion[0] = Options.Version[3];
84     ReversedVersion[1] = Options.Version[2];
85     ReversedVersion[2] = Options.Version[1];
86     ReversedVersion[3] = Options.Version[0];
87     ReversedVersion[4] = '\0';
88   }
89   bool runOnModule(Module &M, const TargetLibraryInfo &TLI);
90 
91 private:
92   // Create the .gcno files for the Module based on DebugInfo.
93   void emitProfileNotes();
94 
95   // Modify the program to track transitions along edges and call into the
96   // profiling runtime to emit .gcda files when run.
97   bool emitProfileArcs();
98 
99   // Get pointers to the functions in the runtime library.
100   Constant *getStartFileFunc();
101   Constant *getEmitFunctionFunc();
102   Constant *getEmitArcsFunc();
103   Constant *getSummaryInfoFunc();
104   Constant *getEndFileFunc();
105 
106   // Add the function to write out all our counters to the global destructor
107   // list.
108   Function *
109   insertCounterWriteout(ArrayRef<std::pair<GlobalVariable *, MDNode *>>);
110   Function *insertFlush(ArrayRef<std::pair<GlobalVariable *, MDNode *>>);
111 
112   enum class GCovFileType { GCNO, GCDA };
113   std::string mangleName(const DICompileUnit *CU, GCovFileType FileType);
114 
115   GCOVOptions Options;
116 
117   // Reversed, NUL-terminated copy of Options.Version.
118   char ReversedVersion[5];
119   // Checksum, produced by hash of EdgeDestinations
120   SmallVector<uint32_t, 4> FileChecksums;
121 
122   Module *M;
123   const TargetLibraryInfo *TLI;
124   LLVMContext *Ctx;
125   SmallVector<std::unique_ptr<GCOVFunction>, 16> Funcs;
126 };
127 
128 class GCOVProfilerLegacyPass : public ModulePass {
129 public:
130   static char ID;
131   GCOVProfilerLegacyPass()
132       : GCOVProfilerLegacyPass(GCOVOptions::getDefault()) {}
133   GCOVProfilerLegacyPass(const GCOVOptions &Opts)
134       : ModulePass(ID), Profiler(Opts) {
135     initializeGCOVProfilerLegacyPassPass(*PassRegistry::getPassRegistry());
136   }
137   StringRef getPassName() const override { return "GCOV Profiler"; }
138 
139   bool runOnModule(Module &M) override {
140     auto &TLI = getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
141     return Profiler.runOnModule(M, TLI);
142   }
143 
144   void getAnalysisUsage(AnalysisUsage &AU) const override {
145     AU.addRequired<TargetLibraryInfoWrapperPass>();
146   }
147 
148 private:
149   GCOVProfiler Profiler;
150 };
151 }
152 
153 char GCOVProfilerLegacyPass::ID = 0;
154 INITIALIZE_PASS_BEGIN(
155     GCOVProfilerLegacyPass, "insert-gcov-profiling",
156     "Insert instrumentation for GCOV profiling", false, false)
157 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
158 INITIALIZE_PASS_END(
159     GCOVProfilerLegacyPass, "insert-gcov-profiling",
160     "Insert instrumentation for GCOV profiling", false, false)
161 
162 ModulePass *llvm::createGCOVProfilerPass(const GCOVOptions &Options) {
163   return new GCOVProfilerLegacyPass(Options);
164 }
165 
166 static StringRef getFunctionName(const DISubprogram *SP) {
167   if (!SP->getLinkageName().empty())
168     return SP->getLinkageName();
169   return SP->getName();
170 }
171 
172 namespace {
173   class GCOVRecord {
174    protected:
175     static const char *const LinesTag;
176     static const char *const FunctionTag;
177     static const char *const BlockTag;
178     static const char *const EdgeTag;
179 
180     GCOVRecord() = default;
181 
182     void writeBytes(const char *Bytes, int Size) {
183       os->write(Bytes, Size);
184     }
185 
186     void write(uint32_t i) {
187       writeBytes(reinterpret_cast<char*>(&i), 4);
188     }
189 
190     // Returns the length measured in 4-byte blocks that will be used to
191     // represent this string in a GCOV file
192     static unsigned lengthOfGCOVString(StringRef s) {
193       // A GCOV string is a length, followed by a NUL, then between 0 and 3 NULs
194       // padding out to the next 4-byte word. The length is measured in 4-byte
195       // words including padding, not bytes of actual string.
196       return (s.size() / 4) + 1;
197     }
198 
199     void writeGCOVString(StringRef s) {
200       uint32_t Len = lengthOfGCOVString(s);
201       write(Len);
202       writeBytes(s.data(), s.size());
203 
204       // Write 1 to 4 bytes of NUL padding.
205       assert((unsigned)(4 - (s.size() % 4)) > 0);
206       assert((unsigned)(4 - (s.size() % 4)) <= 4);
207       writeBytes("\0\0\0\0", 4 - (s.size() % 4));
208     }
209 
210     raw_ostream *os;
211   };
212   const char *const GCOVRecord::LinesTag = "\0\0\x45\x01";
213   const char *const GCOVRecord::FunctionTag = "\0\0\0\1";
214   const char *const GCOVRecord::BlockTag = "\0\0\x41\x01";
215   const char *const GCOVRecord::EdgeTag = "\0\0\x43\x01";
216 
217   class GCOVFunction;
218   class GCOVBlock;
219 
220   // Constructed only by requesting it from a GCOVBlock, this object stores a
221   // list of line numbers and a single filename, representing lines that belong
222   // to the block.
223   class GCOVLines : public GCOVRecord {
224    public:
225     void addLine(uint32_t Line) {
226       assert(Line != 0 && "Line zero is not a valid real line number.");
227       Lines.push_back(Line);
228     }
229 
230     uint32_t length() const {
231       // Here 2 = 1 for string length + 1 for '0' id#.
232       return lengthOfGCOVString(Filename) + 2 + Lines.size();
233     }
234 
235     void writeOut() {
236       write(0);
237       writeGCOVString(Filename);
238       for (int i = 0, e = Lines.size(); i != e; ++i)
239         write(Lines[i]);
240     }
241 
242     GCOVLines(StringRef F, raw_ostream *os)
243       : Filename(F) {
244       this->os = os;
245     }
246 
247    private:
248     StringRef Filename;
249     SmallVector<uint32_t, 32> Lines;
250   };
251 
252 
253   // Represent a basic block in GCOV. Each block has a unique number in the
254   // function, number of lines belonging to each block, and a set of edges to
255   // other blocks.
256   class GCOVBlock : public GCOVRecord {
257    public:
258     GCOVLines &getFile(StringRef Filename) {
259       return LinesByFile.try_emplace(Filename, Filename, os).first->second;
260     }
261 
262     void addEdge(GCOVBlock &Successor) {
263       OutEdges.push_back(&Successor);
264     }
265 
266     void writeOut() {
267       uint32_t Len = 3;
268       SmallVector<StringMapEntry<GCOVLines> *, 32> SortedLinesByFile;
269       for (auto &I : LinesByFile) {
270         Len += I.second.length();
271         SortedLinesByFile.push_back(&I);
272       }
273 
274       writeBytes(LinesTag, 4);
275       write(Len);
276       write(Number);
277 
278       llvm::sort(SortedLinesByFile, [](StringMapEntry<GCOVLines> *LHS,
279                                        StringMapEntry<GCOVLines> *RHS) {
280         return LHS->getKey() < RHS->getKey();
281       });
282       for (auto &I : SortedLinesByFile)
283         I->getValue().writeOut();
284       write(0);
285       write(0);
286     }
287 
288     GCOVBlock(const GCOVBlock &RHS) : GCOVRecord(RHS), Number(RHS.Number) {
289       // Only allow copy before edges and lines have been added. After that,
290       // there are inter-block pointers (eg: edges) that won't take kindly to
291       // blocks being copied or moved around.
292       assert(LinesByFile.empty());
293       assert(OutEdges.empty());
294     }
295 
296    private:
297     friend class GCOVFunction;
298 
299     GCOVBlock(uint32_t Number, raw_ostream *os)
300         : Number(Number) {
301       this->os = os;
302     }
303 
304     uint32_t Number;
305     StringMap<GCOVLines> LinesByFile;
306     SmallVector<GCOVBlock *, 4> OutEdges;
307   };
308 
309   // A function has a unique identifier, a checksum (we leave as zero) and a
310   // set of blocks and a map of edges between blocks. This is the only GCOV
311   // object users can construct, the blocks and lines will be rooted here.
312   class GCOVFunction : public GCOVRecord {
313    public:
314      GCOVFunction(const DISubprogram *SP, Function *F, raw_ostream *os,
315                   uint32_t Ident, bool UseCfgChecksum, bool ExitBlockBeforeBody)
316          : SP(SP), Ident(Ident), UseCfgChecksum(UseCfgChecksum), CfgChecksum(0),
317            ReturnBlock(1, os) {
318       this->os = os;
319 
320       LLVM_DEBUG(dbgs() << "Function: " << getFunctionName(SP) << "\n");
321 
322       uint32_t i = 0;
323       for (auto &BB : *F) {
324         // Skip index 1 if it's assigned to the ReturnBlock.
325         if (i == 1 && ExitBlockBeforeBody)
326           ++i;
327         Blocks.insert(std::make_pair(&BB, GCOVBlock(i++, os)));
328       }
329       if (!ExitBlockBeforeBody)
330         ReturnBlock.Number = i;
331 
332       std::string FunctionNameAndLine;
333       raw_string_ostream FNLOS(FunctionNameAndLine);
334       FNLOS << getFunctionName(SP) << SP->getLine();
335       FNLOS.flush();
336       FuncChecksum = hash_value(FunctionNameAndLine);
337     }
338 
339     GCOVBlock &getBlock(BasicBlock *BB) {
340       return Blocks.find(BB)->second;
341     }
342 
343     GCOVBlock &getReturnBlock() {
344       return ReturnBlock;
345     }
346 
347     std::string getEdgeDestinations() {
348       std::string EdgeDestinations;
349       raw_string_ostream EDOS(EdgeDestinations);
350       Function *F = Blocks.begin()->first->getParent();
351       for (BasicBlock &I : *F) {
352         GCOVBlock &Block = getBlock(&I);
353         for (int i = 0, e = Block.OutEdges.size(); i != e; ++i)
354           EDOS << Block.OutEdges[i]->Number;
355       }
356       return EdgeDestinations;
357     }
358 
359     uint32_t getFuncChecksum() {
360       return FuncChecksum;
361     }
362 
363     void setCfgChecksum(uint32_t Checksum) {
364       CfgChecksum = Checksum;
365     }
366 
367     void writeOut() {
368       writeBytes(FunctionTag, 4);
369       uint32_t BlockLen = 1 + 1 + 1 + lengthOfGCOVString(getFunctionName(SP)) +
370                           1 + lengthOfGCOVString(SP->getFilename()) + 1;
371       if (UseCfgChecksum)
372         ++BlockLen;
373       write(BlockLen);
374       write(Ident);
375       write(FuncChecksum);
376       if (UseCfgChecksum)
377         write(CfgChecksum);
378       writeGCOVString(getFunctionName(SP));
379       writeGCOVString(SP->getFilename());
380       write(SP->getLine());
381 
382       // Emit count of blocks.
383       writeBytes(BlockTag, 4);
384       write(Blocks.size() + 1);
385       for (int i = 0, e = Blocks.size() + 1; i != e; ++i) {
386         write(0);  // No flags on our blocks.
387       }
388       LLVM_DEBUG(dbgs() << Blocks.size() << " blocks.\n");
389 
390       // Emit edges between blocks.
391       if (Blocks.empty()) return;
392       Function *F = Blocks.begin()->first->getParent();
393       for (BasicBlock &I : *F) {
394         GCOVBlock &Block = getBlock(&I);
395         if (Block.OutEdges.empty()) continue;
396 
397         writeBytes(EdgeTag, 4);
398         write(Block.OutEdges.size() * 2 + 1);
399         write(Block.Number);
400         for (int i = 0, e = Block.OutEdges.size(); i != e; ++i) {
401           LLVM_DEBUG(dbgs() << Block.Number << " -> "
402                             << Block.OutEdges[i]->Number << "\n");
403           write(Block.OutEdges[i]->Number);
404           write(0);  // no flags
405         }
406       }
407 
408       // Emit lines for each block.
409       for (BasicBlock &I : *F)
410         getBlock(&I).writeOut();
411     }
412 
413    private:
414      const DISubprogram *SP;
415     uint32_t Ident;
416     uint32_t FuncChecksum;
417     bool UseCfgChecksum;
418     uint32_t CfgChecksum;
419     DenseMap<BasicBlock *, GCOVBlock> Blocks;
420     GCOVBlock ReturnBlock;
421   };
422 }
423 
424 std::string GCOVProfiler::mangleName(const DICompileUnit *CU,
425                                      GCovFileType OutputType) {
426   bool Notes = OutputType == GCovFileType::GCNO;
427 
428   if (NamedMDNode *GCov = M->getNamedMetadata("llvm.gcov")) {
429     for (int i = 0, e = GCov->getNumOperands(); i != e; ++i) {
430       MDNode *N = GCov->getOperand(i);
431       bool ThreeElement = N->getNumOperands() == 3;
432       if (!ThreeElement && N->getNumOperands() != 2)
433         continue;
434       if (dyn_cast<MDNode>(N->getOperand(ThreeElement ? 2 : 1)) != CU)
435         continue;
436 
437       if (ThreeElement) {
438         // These nodes have no mangling to apply, it's stored mangled in the
439         // bitcode.
440         MDString *NotesFile = dyn_cast<MDString>(N->getOperand(0));
441         MDString *DataFile = dyn_cast<MDString>(N->getOperand(1));
442         if (!NotesFile || !DataFile)
443           continue;
444         return Notes ? NotesFile->getString() : DataFile->getString();
445       }
446 
447       MDString *GCovFile = dyn_cast<MDString>(N->getOperand(0));
448       if (!GCovFile)
449         continue;
450 
451       SmallString<128> Filename = GCovFile->getString();
452       sys::path::replace_extension(Filename, Notes ? "gcno" : "gcda");
453       return Filename.str();
454     }
455   }
456 
457   SmallString<128> Filename = CU->getFilename();
458   sys::path::replace_extension(Filename, Notes ? "gcno" : "gcda");
459   StringRef FName = sys::path::filename(Filename);
460   SmallString<128> CurPath;
461   if (sys::fs::current_path(CurPath)) return FName;
462   sys::path::append(CurPath, FName);
463   return CurPath.str();
464 }
465 
466 bool GCOVProfiler::runOnModule(Module &M, const TargetLibraryInfo &TLI) {
467   this->M = &M;
468   this->TLI = &TLI;
469   Ctx = &M.getContext();
470 
471   if (Options.EmitNotes) emitProfileNotes();
472   if (Options.EmitData) return emitProfileArcs();
473   return false;
474 }
475 
476 PreservedAnalyses GCOVProfilerPass::run(Module &M,
477                                         ModuleAnalysisManager &AM) {
478 
479   GCOVProfiler Profiler(GCOVOpts);
480 
481   auto &TLI = AM.getResult<TargetLibraryAnalysis>(M);
482   if (!Profiler.runOnModule(M, TLI))
483     return PreservedAnalyses::all();
484 
485   return PreservedAnalyses::none();
486 }
487 
488 static bool functionHasLines(Function &F) {
489   // Check whether this function actually has any source lines. Not only
490   // do these waste space, they also can crash gcov.
491   for (auto &BB : F) {
492     for (auto &I : BB) {
493       // Debug intrinsic locations correspond to the location of the
494       // declaration, not necessarily any statements or expressions.
495       if (isa<DbgInfoIntrinsic>(&I)) continue;
496 
497       const DebugLoc &Loc = I.getDebugLoc();
498       if (!Loc)
499         continue;
500 
501       // Artificial lines such as calls to the global constructors.
502       if (Loc.getLine() == 0) continue;
503 
504       return true;
505     }
506   }
507   return false;
508 }
509 
510 static bool isUsingScopeBasedEH(Function &F) {
511   if (!F.hasPersonalityFn()) return false;
512 
513   EHPersonality Personality = classifyEHPersonality(F.getPersonalityFn());
514   return isScopedEHPersonality(Personality);
515 }
516 
517 static bool shouldKeepInEntry(BasicBlock::iterator It) {
518 	if (isa<AllocaInst>(*It)) return true;
519 	if (isa<DbgInfoIntrinsic>(*It)) return true;
520 	if (auto *II = dyn_cast<IntrinsicInst>(It)) {
521 		if (II->getIntrinsicID() == llvm::Intrinsic::localescape) return true;
522 	}
523 
524 	return false;
525 }
526 
527 void GCOVProfiler::emitProfileNotes() {
528   NamedMDNode *CU_Nodes = M->getNamedMetadata("llvm.dbg.cu");
529   if (!CU_Nodes) return;
530 
531   for (unsigned i = 0, e = CU_Nodes->getNumOperands(); i != e; ++i) {
532     // Each compile unit gets its own .gcno file. This means that whether we run
533     // this pass over the original .o's as they're produced, or run it after
534     // LTO, we'll generate the same .gcno files.
535 
536     auto *CU = cast<DICompileUnit>(CU_Nodes->getOperand(i));
537 
538     // Skip module skeleton (and module) CUs.
539     if (CU->getDWOId())
540       continue;
541 
542     std::error_code EC;
543     raw_fd_ostream out(mangleName(CU, GCovFileType::GCNO), EC, sys::fs::F_None);
544     if (EC) {
545       Ctx->emitError(Twine("failed to open coverage notes file for writing: ") +
546                      EC.message());
547       continue;
548     }
549 
550     std::string EdgeDestinations;
551 
552     unsigned FunctionIdent = 0;
553     for (auto &F : M->functions()) {
554       DISubprogram *SP = F.getSubprogram();
555       if (!SP) continue;
556       if (!functionHasLines(F)) continue;
557       // TODO: Functions using scope-based EH are currently not supported.
558       if (isUsingScopeBasedEH(F)) continue;
559 
560       // gcov expects every function to start with an entry block that has a
561       // single successor, so split the entry block to make sure of that.
562       BasicBlock &EntryBlock = F.getEntryBlock();
563       BasicBlock::iterator It = EntryBlock.begin();
564       while (shouldKeepInEntry(It))
565         ++It;
566       EntryBlock.splitBasicBlock(It);
567 
568       Funcs.push_back(make_unique<GCOVFunction>(SP, &F, &out, FunctionIdent++,
569                                                 Options.UseCfgChecksum,
570                                                 Options.ExitBlockBeforeBody));
571       GCOVFunction &Func = *Funcs.back();
572 
573       // Add the function line number to the lines of the entry block
574       // to have a counter for the function definition.
575       uint32_t Line = SP->getLine();
576       Func.getBlock(&EntryBlock).getFile(SP->getFilename()).addLine(Line);
577 
578       for (auto &BB : F) {
579         GCOVBlock &Block = Func.getBlock(&BB);
580         Instruction *TI = BB.getTerminator();
581         if (int successors = TI->getNumSuccessors()) {
582           for (int i = 0; i != successors; ++i) {
583             Block.addEdge(Func.getBlock(TI->getSuccessor(i)));
584           }
585         } else if (isa<ReturnInst>(TI)) {
586           Block.addEdge(Func.getReturnBlock());
587         }
588 
589         for (auto &I : BB) {
590           // Debug intrinsic locations correspond to the location of the
591           // declaration, not necessarily any statements or expressions.
592           if (isa<DbgInfoIntrinsic>(&I)) continue;
593 
594           const DebugLoc &Loc = I.getDebugLoc();
595           if (!Loc)
596             continue;
597 
598           // Artificial lines such as calls to the global constructors.
599           if (Loc.getLine() == 0 || Loc.isImplicitCode())
600             continue;
601 
602           if (Line == Loc.getLine()) continue;
603           Line = Loc.getLine();
604           if (SP != getDISubprogram(Loc.getScope()))
605             continue;
606 
607           GCOVLines &Lines = Block.getFile(SP->getFilename());
608           Lines.addLine(Loc.getLine());
609         }
610         Line = 0;
611       }
612       EdgeDestinations += Func.getEdgeDestinations();
613     }
614 
615     FileChecksums.push_back(hash_value(EdgeDestinations));
616     out.write("oncg", 4);
617     out.write(ReversedVersion, 4);
618     out.write(reinterpret_cast<char*>(&FileChecksums.back()), 4);
619 
620     for (auto &Func : Funcs) {
621       Func->setCfgChecksum(FileChecksums.back());
622       Func->writeOut();
623     }
624 
625     out.write("\0\0\0\0\0\0\0\0", 8);  // EOF
626     out.close();
627   }
628 }
629 
630 bool GCOVProfiler::emitProfileArcs() {
631   NamedMDNode *CU_Nodes = M->getNamedMetadata("llvm.dbg.cu");
632   if (!CU_Nodes) return false;
633 
634   bool Result = false;
635   for (unsigned i = 0, e = CU_Nodes->getNumOperands(); i != e; ++i) {
636     SmallVector<std::pair<GlobalVariable *, MDNode *>, 8> CountersBySP;
637     for (auto &F : M->functions()) {
638       DISubprogram *SP = F.getSubprogram();
639       if (!SP) continue;
640       if (!functionHasLines(F)) continue;
641       // TODO: Functions using scope-based EH are currently not supported.
642       if (isUsingScopeBasedEH(F)) continue;
643       if (!Result) Result = true;
644 
645       DenseMap<std::pair<BasicBlock *, BasicBlock *>, unsigned> EdgeToCounter;
646       unsigned Edges = 0;
647       for (auto &BB : F) {
648         Instruction *TI = BB.getTerminator();
649         if (isa<ReturnInst>(TI)) {
650           EdgeToCounter[{&BB, nullptr}] = Edges++;
651         } else {
652           for (BasicBlock *Succ : successors(TI)) {
653             EdgeToCounter[{&BB, Succ}] = Edges++;
654           }
655         }
656       }
657 
658       ArrayType *CounterTy =
659         ArrayType::get(Type::getInt64Ty(*Ctx), Edges);
660       GlobalVariable *Counters =
661         new GlobalVariable(*M, CounterTy, false,
662                            GlobalValue::InternalLinkage,
663                            Constant::getNullValue(CounterTy),
664                            "__llvm_gcov_ctr");
665       CountersBySP.push_back(std::make_pair(Counters, SP));
666 
667       // If a BB has several predecessors, use a PHINode to select
668       // the correct counter.
669       for (auto &BB : F) {
670         const unsigned EdgeCount =
671             std::distance(pred_begin(&BB), pred_end(&BB));
672         if (EdgeCount) {
673           // The phi node must be at the begin of the BB.
674           IRBuilder<> BuilderForPhi(&*BB.begin());
675           Type *Int64PtrTy = Type::getInt64PtrTy(*Ctx);
676           PHINode *Phi = BuilderForPhi.CreatePHI(Int64PtrTy, EdgeCount);
677           for (BasicBlock *Pred : predecessors(&BB)) {
678             auto It = EdgeToCounter.find({Pred, &BB});
679             assert(It != EdgeToCounter.end());
680             const unsigned Edge = It->second;
681             Value *EdgeCounter =
682                 BuilderForPhi.CreateConstInBoundsGEP2_64(Counters, 0, Edge);
683             Phi->addIncoming(EdgeCounter, Pred);
684           }
685 
686           // Skip phis, landingpads.
687           IRBuilder<> Builder(&*BB.getFirstInsertionPt());
688           Value *Count = Builder.CreateLoad(Phi);
689           Count = Builder.CreateAdd(Count, Builder.getInt64(1));
690           Builder.CreateStore(Count, Phi);
691 
692           Instruction *TI = BB.getTerminator();
693           if (isa<ReturnInst>(TI)) {
694             auto It = EdgeToCounter.find({&BB, nullptr});
695             assert(It != EdgeToCounter.end());
696             const unsigned Edge = It->second;
697             Value *Counter =
698                 Builder.CreateConstInBoundsGEP2_64(Counters, 0, Edge);
699             Value *Count = Builder.CreateLoad(Counter);
700             Count = Builder.CreateAdd(Count, Builder.getInt64(1));
701             Builder.CreateStore(Count, Counter);
702           }
703         }
704       }
705     }
706 
707     Function *WriteoutF = insertCounterWriteout(CountersBySP);
708     Function *FlushF = insertFlush(CountersBySP);
709 
710     // Create a small bit of code that registers the "__llvm_gcov_writeout" to
711     // be executed at exit and the "__llvm_gcov_flush" function to be executed
712     // when "__gcov_flush" is called.
713     FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), false);
714     Function *F = Function::Create(FTy, GlobalValue::InternalLinkage,
715                                    "__llvm_gcov_init", M);
716     F->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
717     F->setLinkage(GlobalValue::InternalLinkage);
718     F->addFnAttr(Attribute::NoInline);
719     if (Options.NoRedZone)
720       F->addFnAttr(Attribute::NoRedZone);
721 
722     BasicBlock *BB = BasicBlock::Create(*Ctx, "entry", F);
723     IRBuilder<> Builder(BB);
724 
725     FTy = FunctionType::get(Type::getVoidTy(*Ctx), false);
726     Type *Params[] = {
727       PointerType::get(FTy, 0),
728       PointerType::get(FTy, 0)
729     };
730     FTy = FunctionType::get(Builder.getVoidTy(), Params, false);
731 
732     // Initialize the environment and register the local writeout and flush
733     // functions.
734     Constant *GCOVInit = M->getOrInsertFunction("llvm_gcov_init", FTy);
735     Builder.CreateCall(GCOVInit, {WriteoutF, FlushF});
736     Builder.CreateRetVoid();
737 
738     appendToGlobalCtors(*M, F, 0);
739   }
740 
741   return Result;
742 }
743 
744 Constant *GCOVProfiler::getStartFileFunc() {
745   Type *Args[] = {
746     Type::getInt8PtrTy(*Ctx),  // const char *orig_filename
747     Type::getInt8PtrTy(*Ctx),  // const char version[4]
748     Type::getInt32Ty(*Ctx),    // uint32_t checksum
749   };
750   FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), Args, false);
751   auto *Res = M->getOrInsertFunction("llvm_gcda_start_file", FTy);
752   if (Function *FunRes = dyn_cast<Function>(Res))
753     if (auto AK = TLI->getExtAttrForI32Param(false))
754       FunRes->addParamAttr(2, AK);
755   return Res;
756 
757 }
758 
759 Constant *GCOVProfiler::getEmitFunctionFunc() {
760   Type *Args[] = {
761     Type::getInt32Ty(*Ctx),    // uint32_t ident
762     Type::getInt8PtrTy(*Ctx),  // const char *function_name
763     Type::getInt32Ty(*Ctx),    // uint32_t func_checksum
764     Type::getInt8Ty(*Ctx),     // uint8_t use_extra_checksum
765     Type::getInt32Ty(*Ctx),    // uint32_t cfg_checksum
766   };
767   FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), Args, false);
768   auto *Res = M->getOrInsertFunction("llvm_gcda_emit_function", FTy);
769   if (Function *FunRes = dyn_cast<Function>(Res))
770     if (auto AK = TLI->getExtAttrForI32Param(false)) {
771       FunRes->addParamAttr(0, AK);
772       FunRes->addParamAttr(2, AK);
773       FunRes->addParamAttr(3, AK);
774       FunRes->addParamAttr(4, AK);
775     }
776   return Res;
777 }
778 
779 Constant *GCOVProfiler::getEmitArcsFunc() {
780   Type *Args[] = {
781     Type::getInt32Ty(*Ctx),     // uint32_t num_counters
782     Type::getInt64PtrTy(*Ctx),  // uint64_t *counters
783   };
784   FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), Args, false);
785   auto *Res = M->getOrInsertFunction("llvm_gcda_emit_arcs", FTy);
786   if (Function *FunRes = dyn_cast<Function>(Res))
787     if (auto AK = TLI->getExtAttrForI32Param(false))
788       FunRes->addParamAttr(0, AK);
789   return Res;
790 }
791 
792 Constant *GCOVProfiler::getSummaryInfoFunc() {
793   FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), false);
794   return M->getOrInsertFunction("llvm_gcda_summary_info", FTy);
795 }
796 
797 Constant *GCOVProfiler::getEndFileFunc() {
798   FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), false);
799   return M->getOrInsertFunction("llvm_gcda_end_file", FTy);
800 }
801 
802 Function *GCOVProfiler::insertCounterWriteout(
803     ArrayRef<std::pair<GlobalVariable *, MDNode *> > CountersBySP) {
804   FunctionType *WriteoutFTy = FunctionType::get(Type::getVoidTy(*Ctx), false);
805   Function *WriteoutF = M->getFunction("__llvm_gcov_writeout");
806   if (!WriteoutF)
807     WriteoutF = Function::Create(WriteoutFTy, GlobalValue::InternalLinkage,
808                                  "__llvm_gcov_writeout", M);
809   WriteoutF->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
810   WriteoutF->addFnAttr(Attribute::NoInline);
811   if (Options.NoRedZone)
812     WriteoutF->addFnAttr(Attribute::NoRedZone);
813 
814   BasicBlock *BB = BasicBlock::Create(*Ctx, "entry", WriteoutF);
815   IRBuilder<> Builder(BB);
816 
817   Constant *StartFile = getStartFileFunc();
818   Constant *EmitFunction = getEmitFunctionFunc();
819   Constant *EmitArcs = getEmitArcsFunc();
820   Constant *SummaryInfo = getSummaryInfoFunc();
821   Constant *EndFile = getEndFileFunc();
822 
823   NamedMDNode *CUNodes = M->getNamedMetadata("llvm.dbg.cu");
824   if (!CUNodes) {
825     Builder.CreateRetVoid();
826     return WriteoutF;
827   }
828 
829   // Collect the relevant data into a large constant data structure that we can
830   // walk to write out everything.
831   StructType *StartFileCallArgsTy = StructType::create(
832       {Builder.getInt8PtrTy(), Builder.getInt8PtrTy(), Builder.getInt32Ty()});
833   StructType *EmitFunctionCallArgsTy = StructType::create(
834       {Builder.getInt32Ty(), Builder.getInt8PtrTy(), Builder.getInt32Ty(),
835        Builder.getInt8Ty(), Builder.getInt32Ty()});
836   StructType *EmitArcsCallArgsTy = StructType::create(
837       {Builder.getInt32Ty(), Builder.getInt64Ty()->getPointerTo()});
838   StructType *FileInfoTy =
839       StructType::create({StartFileCallArgsTy, Builder.getInt32Ty(),
840                           EmitFunctionCallArgsTy->getPointerTo(),
841                           EmitArcsCallArgsTy->getPointerTo()});
842 
843   Constant *Zero32 = Builder.getInt32(0);
844   // Build an explicit array of two zeros for use in ConstantExpr GEP building.
845   Constant *TwoZero32s[] = {Zero32, Zero32};
846 
847   SmallVector<Constant *, 8> FileInfos;
848   for (int i : llvm::seq<int>(0, CUNodes->getNumOperands())) {
849     auto *CU = cast<DICompileUnit>(CUNodes->getOperand(i));
850 
851     // Skip module skeleton (and module) CUs.
852     if (CU->getDWOId())
853       continue;
854 
855     std::string FilenameGcda = mangleName(CU, GCovFileType::GCDA);
856     uint32_t CfgChecksum = FileChecksums.empty() ? 0 : FileChecksums[i];
857     auto *StartFileCallArgs = ConstantStruct::get(
858         StartFileCallArgsTy, {Builder.CreateGlobalStringPtr(FilenameGcda),
859                               Builder.CreateGlobalStringPtr(ReversedVersion),
860                               Builder.getInt32(CfgChecksum)});
861 
862     SmallVector<Constant *, 8> EmitFunctionCallArgsArray;
863     SmallVector<Constant *, 8> EmitArcsCallArgsArray;
864     for (int j : llvm::seq<int>(0, CountersBySP.size())) {
865       auto *SP = cast_or_null<DISubprogram>(CountersBySP[j].second);
866       uint32_t FuncChecksum = Funcs.empty() ? 0 : Funcs[j]->getFuncChecksum();
867       EmitFunctionCallArgsArray.push_back(ConstantStruct::get(
868           EmitFunctionCallArgsTy,
869           {Builder.getInt32(j),
870            Options.FunctionNamesInData
871                ? Builder.CreateGlobalStringPtr(getFunctionName(SP))
872                : Constant::getNullValue(Builder.getInt8PtrTy()),
873            Builder.getInt32(FuncChecksum),
874            Builder.getInt8(Options.UseCfgChecksum),
875            Builder.getInt32(CfgChecksum)}));
876 
877       GlobalVariable *GV = CountersBySP[j].first;
878       unsigned Arcs = cast<ArrayType>(GV->getValueType())->getNumElements();
879       EmitArcsCallArgsArray.push_back(ConstantStruct::get(
880           EmitArcsCallArgsTy,
881           {Builder.getInt32(Arcs), ConstantExpr::getInBoundsGetElementPtr(
882                                        GV->getValueType(), GV, TwoZero32s)}));
883     }
884     // Create global arrays for the two emit calls.
885     int CountersSize = CountersBySP.size();
886     assert(CountersSize == (int)EmitFunctionCallArgsArray.size() &&
887            "Mismatched array size!");
888     assert(CountersSize == (int)EmitArcsCallArgsArray.size() &&
889            "Mismatched array size!");
890     auto *EmitFunctionCallArgsArrayTy =
891         ArrayType::get(EmitFunctionCallArgsTy, CountersSize);
892     auto *EmitFunctionCallArgsArrayGV = new GlobalVariable(
893         *M, EmitFunctionCallArgsArrayTy, /*isConstant*/ true,
894         GlobalValue::InternalLinkage,
895         ConstantArray::get(EmitFunctionCallArgsArrayTy,
896                            EmitFunctionCallArgsArray),
897         Twine("__llvm_internal_gcov_emit_function_args.") + Twine(i));
898     auto *EmitArcsCallArgsArrayTy =
899         ArrayType::get(EmitArcsCallArgsTy, CountersSize);
900     EmitFunctionCallArgsArrayGV->setUnnamedAddr(
901         GlobalValue::UnnamedAddr::Global);
902     auto *EmitArcsCallArgsArrayGV = new GlobalVariable(
903         *M, EmitArcsCallArgsArrayTy, /*isConstant*/ true,
904         GlobalValue::InternalLinkage,
905         ConstantArray::get(EmitArcsCallArgsArrayTy, EmitArcsCallArgsArray),
906         Twine("__llvm_internal_gcov_emit_arcs_args.") + Twine(i));
907     EmitArcsCallArgsArrayGV->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
908 
909     FileInfos.push_back(ConstantStruct::get(
910         FileInfoTy,
911         {StartFileCallArgs, Builder.getInt32(CountersSize),
912          ConstantExpr::getInBoundsGetElementPtr(EmitFunctionCallArgsArrayTy,
913                                                 EmitFunctionCallArgsArrayGV,
914                                                 TwoZero32s),
915          ConstantExpr::getInBoundsGetElementPtr(
916              EmitArcsCallArgsArrayTy, EmitArcsCallArgsArrayGV, TwoZero32s)}));
917   }
918 
919   // If we didn't find anything to actually emit, bail on out.
920   if (FileInfos.empty()) {
921     Builder.CreateRetVoid();
922     return WriteoutF;
923   }
924 
925   // To simplify code, we cap the number of file infos we write out to fit
926   // easily in a 32-bit signed integer. This gives consistent behavior between
927   // 32-bit and 64-bit systems without requiring (potentially very slow) 64-bit
928   // operations on 32-bit systems. It also seems unreasonable to try to handle
929   // more than 2 billion files.
930   if ((int64_t)FileInfos.size() > (int64_t)INT_MAX)
931     FileInfos.resize(INT_MAX);
932 
933   // Create a global for the entire data structure so we can walk it more
934   // easily.
935   auto *FileInfoArrayTy = ArrayType::get(FileInfoTy, FileInfos.size());
936   auto *FileInfoArrayGV = new GlobalVariable(
937       *M, FileInfoArrayTy, /*isConstant*/ true, GlobalValue::InternalLinkage,
938       ConstantArray::get(FileInfoArrayTy, FileInfos),
939       "__llvm_internal_gcov_emit_file_info");
940   FileInfoArrayGV->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
941 
942   // Create the CFG for walking this data structure.
943   auto *FileLoopHeader =
944       BasicBlock::Create(*Ctx, "file.loop.header", WriteoutF);
945   auto *CounterLoopHeader =
946       BasicBlock::Create(*Ctx, "counter.loop.header", WriteoutF);
947   auto *FileLoopLatch = BasicBlock::Create(*Ctx, "file.loop.latch", WriteoutF);
948   auto *ExitBB = BasicBlock::Create(*Ctx, "exit", WriteoutF);
949 
950   // We always have at least one file, so just branch to the header.
951   Builder.CreateBr(FileLoopHeader);
952 
953   // The index into the files structure is our loop induction variable.
954   Builder.SetInsertPoint(FileLoopHeader);
955   PHINode *IV =
956       Builder.CreatePHI(Builder.getInt32Ty(), /*NumReservedValues*/ 2);
957   IV->addIncoming(Builder.getInt32(0), BB);
958   auto *FileInfoPtr =
959       Builder.CreateInBoundsGEP(FileInfoArrayGV, {Builder.getInt32(0), IV});
960   auto *StartFileCallArgsPtr = Builder.CreateStructGEP(FileInfoPtr, 0);
961   auto *StartFileCall = Builder.CreateCall(
962       StartFile,
963       {Builder.CreateLoad(Builder.CreateStructGEP(StartFileCallArgsPtr, 0)),
964        Builder.CreateLoad(Builder.CreateStructGEP(StartFileCallArgsPtr, 1)),
965        Builder.CreateLoad(Builder.CreateStructGEP(StartFileCallArgsPtr, 2))});
966   if (auto AK = TLI->getExtAttrForI32Param(false))
967     StartFileCall->addParamAttr(2, AK);
968   auto *NumCounters =
969       Builder.CreateLoad(Builder.CreateStructGEP(FileInfoPtr, 1));
970   auto *EmitFunctionCallArgsArray =
971       Builder.CreateLoad(Builder.CreateStructGEP(FileInfoPtr, 2));
972   auto *EmitArcsCallArgsArray =
973       Builder.CreateLoad(Builder.CreateStructGEP(FileInfoPtr, 3));
974   auto *EnterCounterLoopCond =
975       Builder.CreateICmpSLT(Builder.getInt32(0), NumCounters);
976   Builder.CreateCondBr(EnterCounterLoopCond, CounterLoopHeader, FileLoopLatch);
977 
978   Builder.SetInsertPoint(CounterLoopHeader);
979   auto *JV = Builder.CreatePHI(Builder.getInt32Ty(), /*NumReservedValues*/ 2);
980   JV->addIncoming(Builder.getInt32(0), FileLoopHeader);
981   auto *EmitFunctionCallArgsPtr =
982       Builder.CreateInBoundsGEP(EmitFunctionCallArgsArray, {JV});
983   auto *EmitFunctionCall = Builder.CreateCall(
984       EmitFunction,
985       {Builder.CreateLoad(Builder.CreateStructGEP(EmitFunctionCallArgsPtr, 0)),
986        Builder.CreateLoad(Builder.CreateStructGEP(EmitFunctionCallArgsPtr, 1)),
987        Builder.CreateLoad(Builder.CreateStructGEP(EmitFunctionCallArgsPtr, 2)),
988        Builder.CreateLoad(Builder.CreateStructGEP(EmitFunctionCallArgsPtr, 3)),
989        Builder.CreateLoad(
990            Builder.CreateStructGEP(EmitFunctionCallArgsPtr, 4))});
991   if (auto AK = TLI->getExtAttrForI32Param(false)) {
992     EmitFunctionCall->addParamAttr(0, AK);
993     EmitFunctionCall->addParamAttr(2, AK);
994     EmitFunctionCall->addParamAttr(3, AK);
995     EmitFunctionCall->addParamAttr(4, AK);
996   }
997   auto *EmitArcsCallArgsPtr =
998       Builder.CreateInBoundsGEP(EmitArcsCallArgsArray, {JV});
999   auto *EmitArcsCall = Builder.CreateCall(
1000       EmitArcs,
1001       {Builder.CreateLoad(Builder.CreateStructGEP(EmitArcsCallArgsPtr, 0)),
1002        Builder.CreateLoad(Builder.CreateStructGEP(EmitArcsCallArgsPtr, 1))});
1003   if (auto AK = TLI->getExtAttrForI32Param(false))
1004     EmitArcsCall->addParamAttr(0, AK);
1005   auto *NextJV = Builder.CreateAdd(JV, Builder.getInt32(1));
1006   auto *CounterLoopCond = Builder.CreateICmpSLT(NextJV, NumCounters);
1007   Builder.CreateCondBr(CounterLoopCond, CounterLoopHeader, FileLoopLatch);
1008   JV->addIncoming(NextJV, CounterLoopHeader);
1009 
1010   Builder.SetInsertPoint(FileLoopLatch);
1011   Builder.CreateCall(SummaryInfo, {});
1012   Builder.CreateCall(EndFile, {});
1013   auto *NextIV = Builder.CreateAdd(IV, Builder.getInt32(1));
1014   auto *FileLoopCond =
1015       Builder.CreateICmpSLT(NextIV, Builder.getInt32(FileInfos.size()));
1016   Builder.CreateCondBr(FileLoopCond, FileLoopHeader, ExitBB);
1017   IV->addIncoming(NextIV, FileLoopLatch);
1018 
1019   Builder.SetInsertPoint(ExitBB);
1020   Builder.CreateRetVoid();
1021 
1022   return WriteoutF;
1023 }
1024 
1025 Function *GCOVProfiler::
1026 insertFlush(ArrayRef<std::pair<GlobalVariable*, MDNode*> > CountersBySP) {
1027   FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx), false);
1028   Function *FlushF = M->getFunction("__llvm_gcov_flush");
1029   if (!FlushF)
1030     FlushF = Function::Create(FTy, GlobalValue::InternalLinkage,
1031                               "__llvm_gcov_flush", M);
1032   else
1033     FlushF->setLinkage(GlobalValue::InternalLinkage);
1034   FlushF->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
1035   FlushF->addFnAttr(Attribute::NoInline);
1036   if (Options.NoRedZone)
1037     FlushF->addFnAttr(Attribute::NoRedZone);
1038 
1039   BasicBlock *Entry = BasicBlock::Create(*Ctx, "entry", FlushF);
1040 
1041   // Write out the current counters.
1042   Constant *WriteoutF = M->getFunction("__llvm_gcov_writeout");
1043   assert(WriteoutF && "Need to create the writeout function first!");
1044 
1045   IRBuilder<> Builder(Entry);
1046   Builder.CreateCall(WriteoutF, {});
1047 
1048   // Zero out the counters.
1049   for (const auto &I : CountersBySP) {
1050     GlobalVariable *GV = I.first;
1051     Constant *Null = Constant::getNullValue(GV->getValueType());
1052     Builder.CreateStore(Null, GV);
1053   }
1054 
1055   Type *RetTy = FlushF->getReturnType();
1056   if (RetTy == Type::getVoidTy(*Ctx))
1057     Builder.CreateRetVoid();
1058   else if (RetTy->isIntegerTy())
1059     // Used if __llvm_gcov_flush was implicitly declared.
1060     Builder.CreateRet(ConstantInt::get(RetTy, 0));
1061   else
1062     report_fatal_error("invalid return type for __llvm_gcov_flush");
1063 
1064   return FlushF;
1065 }
1066