Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 1 | //===- FuzzerInternal.h - Internal header for the Fuzzer --------*- C++ -* ===// |
| 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 | // Define the main class fuzzer::Fuzzer and most functions. |
| 10 | //===----------------------------------------------------------------------===// |
Yaron Keren | 347663b | 2015-08-10 16:37:40 +0000 | [diff] [blame] | 11 | |
| 12 | #ifndef LLVM_FUZZER_INTERNAL_H |
| 13 | #define LLVM_FUZZER_INTERNAL_H |
| 14 | |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 15 | #include <cassert> |
Kostya Serebryany | 33f8669 | 2015-02-04 22:20:09 +0000 | [diff] [blame] | 16 | #include <climits> |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 17 | #include <chrono> |
| 18 | #include <cstddef> |
| 19 | #include <cstdlib> |
| 20 | #include <string> |
| 21 | #include <vector> |
Kostya Serebryany | 2c1b33b | 2015-01-29 23:01:07 +0000 | [diff] [blame] | 22 | #include <unordered_set> |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 23 | |
Kostya Serebryany | 016852c | 2015-02-19 18:45:37 +0000 | [diff] [blame] | 24 | #include "FuzzerInterface.h" |
| 25 | |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 26 | namespace fuzzer { |
| 27 | typedef std::vector<uint8_t> Unit; |
| 28 | using namespace std::chrono; |
| 29 | |
Kostya Serebryany | 52a788e | 2015-03-31 20:13:20 +0000 | [diff] [blame] | 30 | std::string FileToString(const std::string &Path); |
| 31 | Unit FileToVector(const std::string &Path); |
Kostya Serebryany | 1ac8055 | 2015-05-08 21:30:55 +0000 | [diff] [blame] | 32 | void ReadDirToVectorOfUnits(const char *Path, std::vector<Unit> *V, |
| 33 | long *Epoch); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 34 | void WriteToFile(const Unit &U, const std::string &Path); |
Kostya Serebryany | 5b266a8 | 2015-02-04 19:10:20 +0000 | [diff] [blame] | 35 | void CopyFileToErr(const std::string &Path); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 36 | // Returns "Dir/FileName" or equivalent for the current OS. |
| 37 | std::string DirPlusFile(const std::string &DirPath, |
| 38 | const std::string &FileName); |
| 39 | |
Kostya Serebryany | 20e9bcb | 2015-05-23 01:07:46 +0000 | [diff] [blame] | 40 | void Printf(const char *Fmt, ...); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 41 | void Print(const Unit &U, const char *PrintAfter = ""); |
| 42 | void PrintASCII(const Unit &U, const char *PrintAfter = ""); |
| 43 | std::string Hash(const Unit &U); |
| 44 | void SetTimer(int Seconds); |
Kostya Serebryany | ca6a2a2 | 2015-05-05 21:59:51 +0000 | [diff] [blame] | 45 | void PrintFileAsBase64(const std::string &Path); |
Kostya Serebryany | dc3135d | 2015-11-12 01:02:01 +0000 | [diff] [blame] | 46 | int ExecuteCommand(const std::string &Command); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 47 | |
Kostya Serebryany | 96eab65 | 2015-05-14 22:41:49 +0000 | [diff] [blame] | 48 | // Private copy of SHA1 implementation. |
| 49 | static const int kSHA1NumBytes = 20; |
| 50 | // Computes SHA1 hash of 'Len' bytes in 'Data', writes kSHA1NumBytes to 'Out'. |
| 51 | void ComputeSHA1(const uint8_t *Data, size_t Len, uint8_t *Out); |
| 52 | |
Kostya Serebryany | bc7c0ad | 2015-08-11 01:44:42 +0000 | [diff] [blame] | 53 | // Changes U to contain only ASCII (isprint+isspace) characters. |
| 54 | // Returns true iff U has been changed. |
| 55 | bool ToASCII(Unit &U); |
Kostya Serebryany | a9346c2 | 2015-09-02 19:08:08 +0000 | [diff] [blame] | 56 | bool IsASCII(const Unit &U); |
Kostya Serebryany | bc7c0ad | 2015-08-11 01:44:42 +0000 | [diff] [blame] | 57 | |
Kostya Serebryany | 9690fcf | 2015-05-12 18:51:57 +0000 | [diff] [blame] | 58 | int NumberOfCpuCores(); |
Kostya Serebryany | d6edce9 | 2015-10-16 23:04:31 +0000 | [diff] [blame] | 59 | int GetPid(); |
Kostya Serebryany | 9690fcf | 2015-05-12 18:51:57 +0000 | [diff] [blame] | 60 | |
Kostya Serebryany | 9838b2b | 2015-09-03 20:23:46 +0000 | [diff] [blame] | 61 | // Dictionary. |
| 62 | |
| 63 | // Parses one dictionary entry. |
| 64 | // If successfull, write the enty to Unit and returns true, |
| 65 | // otherwise returns false. |
| 66 | bool ParseOneDictionaryEntry(const std::string &Str, Unit *U); |
| 67 | // Parses the dictionary file, fills Units, returns true iff all lines |
| 68 | // were parsed succesfully. |
| 69 | bool ParseDictionaryFile(const std::string &Text, std::vector<Unit> *Units); |
| 70 | |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 71 | class Fuzzer { |
| 72 | public: |
| 73 | struct FuzzingOptions { |
| 74 | int Verbosity = 1; |
| 75 | int MaxLen = 0; |
Kostya Serebryany | 490bbd6 | 2015-05-19 22:12:57 +0000 | [diff] [blame] | 76 | int UnitTimeoutSec = 300; |
Kostya Serebryany | b85db17 | 2015-10-02 20:47:55 +0000 | [diff] [blame] | 77 | int MaxTotalTimeSec = 0; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 78 | bool DoCrossOver = true; |
Kostya Serebryany | 5b266a8 | 2015-02-04 19:10:20 +0000 | [diff] [blame] | 79 | int MutateDepth = 5; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 80 | bool ExitOnFirst = false; |
Kostya Serebryany | be5e0ed | 2015-03-03 23:27:02 +0000 | [diff] [blame] | 81 | bool UseCounters = false; |
Kostya Serebryany | 2e9fca9 | 2015-10-22 23:55:39 +0000 | [diff] [blame] | 82 | bool UseIndirCalls = true; |
Kostya Serebryany | 5a99ecb | 2015-05-11 20:51:19 +0000 | [diff] [blame] | 83 | bool UseTraces = false; |
Kostya Serebryany | 2c1b33b | 2015-01-29 23:01:07 +0000 | [diff] [blame] | 84 | bool UseFullCoverageSet = false; |
Kostya Serebryany | 1ac8055 | 2015-05-08 21:30:55 +0000 | [diff] [blame] | 85 | bool Reload = true; |
Kostya Serebryany | fed509e | 2015-10-17 04:38:26 +0000 | [diff] [blame] | 86 | bool ShuffleAtStartUp = true; |
Kostya Serebryany | 92e0476 | 2015-02-04 23:42:42 +0000 | [diff] [blame] | 87 | int PreferSmallDuringInitialShuffle = -1; |
Kostya Serebryany | 33f8669 | 2015-02-04 22:20:09 +0000 | [diff] [blame] | 88 | size_t MaxNumberOfRuns = ULONG_MAX; |
Kostya Serebryany | 2da7b84 | 2015-05-18 21:34:20 +0000 | [diff] [blame] | 89 | int SyncTimeout = 600; |
Kostya Serebryany | 70926ae | 2015-08-05 21:43:48 +0000 | [diff] [blame] | 90 | int ReportSlowUnits = 10; |
Kostya Serebryany | bc7c0ad | 2015-08-11 01:44:42 +0000 | [diff] [blame] | 91 | bool OnlyASCII = false; |
Kostya Serebryany | 12c7837 | 2015-08-12 01:55:37 +0000 | [diff] [blame] | 92 | int TBMDepth = 10; |
| 93 | int TBMWidth = 10; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 94 | std::string OutputCorpus; |
Kostya Serebryany | 2da7b84 | 2015-05-18 21:34:20 +0000 | [diff] [blame] | 95 | std::string SyncCommand; |
Kostya Serebryany | bd5d1cd | 2015-10-09 03:57:59 +0000 | [diff] [blame] | 96 | std::string ArtifactPrefix = "./"; |
Kostya Serebryany | 2d0ef14 | 2015-11-25 21:40:46 +0000 | [diff] [blame] | 97 | std::string ExactArtifactPath; |
Kostya Serebryany | b91c62b | 2015-10-16 22:41:47 +0000 | [diff] [blame] | 98 | bool SaveArtifacts = true; |
Kostya Serebryany | dc3135d | 2015-11-12 01:02:01 +0000 | [diff] [blame] | 99 | bool PrintNEW = true; // Print a status line when new units are found; |
Mike Aizatsky | a9c2387 | 2015-11-12 04:38:40 +0000 | [diff] [blame] | 100 | bool OutputCSV = false; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 101 | }; |
Kostya Serebryany | f342459 | 2015-05-22 22:35:31 +0000 | [diff] [blame] | 102 | Fuzzer(UserSuppliedFuzzer &USF, FuzzingOptions Options); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 103 | void AddToCorpus(const Unit &U) { Corpus.push_back(U); } |
Kostya Serebryany | dc3135d | 2015-11-12 01:02:01 +0000 | [diff] [blame] | 104 | size_t ChooseUnitIdxToMutate(); |
| 105 | const Unit &ChooseUnitToMutate() { return Corpus[ChooseUnitIdxToMutate()]; }; |
Kostya Serebryany | 468ed78 | 2015-09-08 17:30:35 +0000 | [diff] [blame] | 106 | void Loop(); |
Kostya Serebryany | dc3135d | 2015-11-12 01:02:01 +0000 | [diff] [blame] | 107 | void Drill(); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 108 | void ShuffleAndMinimize(); |
Kostya Serebryany | 2252625 | 2015-05-11 21:16:27 +0000 | [diff] [blame] | 109 | void InitializeTraceState(); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 110 | size_t CorpusSize() const { return Corpus.size(); } |
Kostya Serebryany | 1ac8055 | 2015-05-08 21:30:55 +0000 | [diff] [blame] | 111 | void ReadDir(const std::string &Path, long *Epoch) { |
Kostya Serebryany | 9cc3b0d | 2015-10-24 01:16:40 +0000 | [diff] [blame] | 112 | Printf("Loading corpus: %s\n", Path.c_str()); |
Kostya Serebryany | 1ac8055 | 2015-05-08 21:30:55 +0000 | [diff] [blame] | 113 | ReadDirToVectorOfUnits(Path.c_str(), &Corpus, Epoch); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 114 | } |
Kostya Serebryany | 1ac8055 | 2015-05-08 21:30:55 +0000 | [diff] [blame] | 115 | void RereadOutputCorpus(); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 116 | // Save the current corpus to OutputCorpus. |
| 117 | void SaveCorpus(); |
| 118 | |
Kostya Serebryany | 92e0476 | 2015-02-04 23:42:42 +0000 | [diff] [blame] | 119 | size_t secondsSinceProcessStartUp() { |
| 120 | return duration_cast<seconds>(system_clock::now() - ProcessStartTime) |
| 121 | .count(); |
| 122 | } |
| 123 | |
| 124 | size_t getTotalNumberOfRuns() { return TotalNumberOfRuns; } |
| 125 | |
Kostya Serebryany | 52a788e | 2015-03-31 20:13:20 +0000 | [diff] [blame] | 126 | static void StaticAlarmCallback(); |
| 127 | |
Ivan Krasin | 95e82d5 | 2015-10-01 23:23:06 +0000 | [diff] [blame] | 128 | void ExecuteCallback(const Unit &U); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 129 | |
Kostya Serebryany | 9cc3b0d | 2015-10-24 01:16:40 +0000 | [diff] [blame] | 130 | // Merge Corpora[1:] into Corpora[0]. |
| 131 | void Merge(const std::vector<std::string> &Corpora); |
| 132 | |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 133 | private: |
Kostya Serebryany | 52a788e | 2015-03-31 20:13:20 +0000 | [diff] [blame] | 134 | void AlarmCallback(); |
Kostya Serebryany | 7d470cf | 2015-05-07 18:32:29 +0000 | [diff] [blame] | 135 | void MutateAndTestOne(Unit *U); |
Kostya Serebryany | 007c9b2 | 2015-10-22 22:50:47 +0000 | [diff] [blame] | 136 | void ReportNewCoverage(const Unit &U); |
| 137 | bool RunOne(const Unit &U); |
Kostya Serebryany | bc7c0ad | 2015-08-11 01:44:42 +0000 | [diff] [blame] | 138 | void RunOneAndUpdateCorpus(Unit &U); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 139 | void WriteToOutputCorpus(const Unit &U); |
Kostya Serebryany | 2b7d2e9 | 2015-07-23 18:37:22 +0000 | [diff] [blame] | 140 | void WriteUnitToFileWithPrefix(const Unit &U, const char *Prefix); |
Kostya Serebryany | 09d2a5f | 2015-10-22 22:56:45 +0000 | [diff] [blame] | 141 | void PrintStats(const char *Where, const char *End = "\n"); |
Kostya Serebryany | dc3135d | 2015-11-12 01:02:01 +0000 | [diff] [blame] | 142 | void PrintStatusForNewUnit(const Unit &U); |
Kostya Serebryany | b360256 | 2015-10-22 21:48:09 +0000 | [diff] [blame] | 143 | void PrintUnitInASCII(const Unit &U, const char *PrintAfter = ""); |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 144 | |
Kostya Serebryany | 2da7b84 | 2015-05-18 21:34:20 +0000 | [diff] [blame] | 145 | void SyncCorpus(); |
| 146 | |
Kostya Serebryany | 007c9b2 | 2015-10-22 22:50:47 +0000 | [diff] [blame] | 147 | size_t RecordBlockCoverage(); |
Kostya Serebryany | 2e9fca9 | 2015-10-22 23:55:39 +0000 | [diff] [blame] | 148 | size_t RecordCallerCalleeCoverage(); |
Kostya Serebryany | 007c9b2 | 2015-10-22 22:50:47 +0000 | [diff] [blame] | 149 | void PrepareCoverageBeforeRun(); |
| 150 | bool CheckCoverageAfterRun(); |
| 151 | |
| 152 | |
Kostya Serebryany | beb24c3 | 2015-05-07 21:02:11 +0000 | [diff] [blame] | 153 | // Trace-based fuzzing: we run a unit with some kind of tracing |
| 154 | // enabled and record potentially useful mutations. Then |
| 155 | // We apply these mutations one by one to the unit and run it again. |
| 156 | |
| 157 | // Start tracing; forget all previously proposed mutations. |
| 158 | void StartTraceRecording(); |
| 159 | // Stop tracing and return the number of proposed mutations. |
| 160 | size_t StopTraceRecording(); |
| 161 | // Apply Idx-th trace-based mutation to U. |
| 162 | void ApplyTraceBasedMutation(size_t Idx, Unit *U); |
| 163 | |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 164 | void SetDeathCallback(); |
Kostya Serebryany | 52a788e | 2015-03-31 20:13:20 +0000 | [diff] [blame] | 165 | static void StaticDeathCallback(); |
| 166 | void DeathCallback(); |
| 167 | Unit CurrentUnit; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 168 | |
| 169 | size_t TotalNumberOfRuns = 0; |
Kostya Serebryany | 12c7837 | 2015-08-12 01:55:37 +0000 | [diff] [blame] | 170 | size_t TotalNumberOfExecutedTraceBasedMutations = 0; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 171 | |
| 172 | std::vector<Unit> Corpus; |
Kostya Serebryany | cbb2334 | 2015-05-19 01:06:07 +0000 | [diff] [blame] | 173 | std::unordered_set<std::string> UnitHashesAddedToCorpus; |
Kostya Serebryany | be5e0ed | 2015-03-03 23:27:02 +0000 | [diff] [blame] | 174 | |
| 175 | // For UseCounters |
| 176 | std::vector<uint8_t> CounterBitmap; |
| 177 | size_t TotalBits() { // Slow. Call it only for printing stats. |
| 178 | size_t Res = 0; |
| 179 | for (auto x : CounterBitmap) Res += __builtin_popcount(x); |
| 180 | return Res; |
| 181 | } |
| 182 | |
Kostya Serebryany | f342459 | 2015-05-22 22:35:31 +0000 | [diff] [blame] | 183 | UserSuppliedFuzzer &USF; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 184 | FuzzingOptions Options; |
| 185 | system_clock::time_point ProcessStartTime = system_clock::now(); |
Kostya Serebryany | 2da7b84 | 2015-05-18 21:34:20 +0000 | [diff] [blame] | 186 | system_clock::time_point LastExternalSync = system_clock::now(); |
Kostya Serebryany | 52a788e | 2015-03-31 20:13:20 +0000 | [diff] [blame] | 187 | system_clock::time_point UnitStartTime; |
Kostya Serebryany | 16901a9 | 2015-03-30 23:04:35 +0000 | [diff] [blame] | 188 | long TimeOfLongestUnitInSeconds = 0; |
Kostya Serebryany | 1ac8055 | 2015-05-08 21:30:55 +0000 | [diff] [blame] | 189 | long EpochOfLastReadOfOutputCorpus = 0; |
Kostya Serebryany | 007c9b2 | 2015-10-22 22:50:47 +0000 | [diff] [blame] | 190 | size_t LastRecordedBlockCoverage = 0; |
Kostya Serebryany | 2e9fca9 | 2015-10-22 23:55:39 +0000 | [diff] [blame] | 191 | size_t LastRecordedCallerCalleeCoverage = 0; |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 192 | }; |
| 193 | |
Kostya Serebryany | f342459 | 2015-05-22 22:35:31 +0000 | [diff] [blame] | 194 | class SimpleUserSuppliedFuzzer: public UserSuppliedFuzzer { |
| 195 | public: |
Kostya Serebryany | 404c69f | 2015-07-24 01:06:40 +0000 | [diff] [blame] | 196 | SimpleUserSuppliedFuzzer(FuzzerRandomBase *Rand, UserCallback Callback) |
| 197 | : UserSuppliedFuzzer(Rand), Callback(Callback) {} |
Kostya Serebryany | 20bb5e7 | 2015-10-02 23:34:06 +0000 | [diff] [blame] | 198 | |
Kostya Serebryany | 20bb5e7 | 2015-10-02 23:34:06 +0000 | [diff] [blame] | 199 | virtual int TargetFunction(const uint8_t *Data, size_t Size) override { |
Kostya Serebryany | 94660b3 | 2015-10-23 18:37:58 +0000 | [diff] [blame] | 200 | return Callback(Data, Size); |
Kostya Serebryany | f342459 | 2015-05-22 22:35:31 +0000 | [diff] [blame] | 201 | } |
| 202 | |
| 203 | private: |
Kostya Serebryany | 20bb5e7 | 2015-10-02 23:34:06 +0000 | [diff] [blame] | 204 | UserCallback Callback = nullptr; |
Kostya Serebryany | f342459 | 2015-05-22 22:35:31 +0000 | [diff] [blame] | 205 | }; |
| 206 | |
Aaron Ballman | ef11698 | 2015-01-29 16:58:29 +0000 | [diff] [blame] | 207 | }; // namespace fuzzer |
Yaron Keren | 347663b | 2015-08-10 16:37:40 +0000 | [diff] [blame] | 208 | |
| 209 | #endif // LLVM_FUZZER_INTERNAL_H |