Home | History | Annotate | Download | only in Fuzzer
      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 //===----------------------------------------------------------------------===//
     11 
     12 #ifndef LLVM_FUZZER_INTERNAL_H
     13 #define LLVM_FUZZER_INTERNAL_H
     14 
     15 #include <cassert>
     16 #include <climits>
     17 #include <chrono>
     18 #include <cstddef>
     19 #include <cstdlib>
     20 #include <string>
     21 #include <vector>
     22 #include <unordered_set>
     23 
     24 #include "FuzzerInterface.h"
     25 
     26 namespace fuzzer {
     27 using namespace std::chrono;
     28 
     29 std::string FileToString(const std::string &Path);
     30 Unit FileToVector(const std::string &Path);
     31 void ReadDirToVectorOfUnits(const char *Path, std::vector<Unit> *V,
     32                             long *Epoch);
     33 void WriteToFile(const Unit &U, const std::string &Path);
     34 void CopyFileToErr(const std::string &Path);
     35 // Returns "Dir/FileName" or equivalent for the current OS.
     36 std::string DirPlusFile(const std::string &DirPath,
     37                         const std::string &FileName);
     38 
     39 void Printf(const char *Fmt, ...);
     40 void Print(const Unit &U, const char *PrintAfter = "");
     41 void PrintASCII(const Unit &U, const char *PrintAfter = "");
     42 std::string Hash(const Unit &U);
     43 void SetTimer(int Seconds);
     44 std::string Base64(const Unit &U);
     45 int ExecuteCommand(const std::string &Command);
     46 
     47 // Private copy of SHA1 implementation.
     48 static const int kSHA1NumBytes = 20;
     49 // Computes SHA1 hash of 'Len' bytes in 'Data', writes kSHA1NumBytes to 'Out'.
     50 void ComputeSHA1(const uint8_t *Data, size_t Len, uint8_t *Out);
     51 
     52 // Changes U to contain only ASCII (isprint+isspace) characters.
     53 // Returns true iff U has been changed.
     54 bool ToASCII(Unit &U);
     55 bool IsASCII(const Unit &U);
     56 
     57 int NumberOfCpuCores();
     58 int GetPid();
     59 
     60 // Dictionary.
     61 
     62 // Parses one dictionary entry.
     63 // If successfull, write the enty to Unit and returns true,
     64 // otherwise returns false.
     65 bool ParseOneDictionaryEntry(const std::string &Str, Unit *U);
     66 // Parses the dictionary file, fills Units, returns true iff all lines
     67 // were parsed succesfully.
     68 bool ParseDictionaryFile(const std::string &Text, std::vector<Unit> *Units);
     69 
     70 class Fuzzer {
     71  public:
     72   struct FuzzingOptions {
     73     int Verbosity = 1;
     74     int MaxLen = 0;
     75     int UnitTimeoutSec = 300;
     76     int MaxTotalTimeSec = 0;
     77     bool DoCrossOver = true;
     78     int  MutateDepth = 5;
     79     bool ExitOnFirst = false;
     80     bool UseCounters = false;
     81     bool UseIndirCalls = true;
     82     bool UseTraces = false;
     83     bool UseFullCoverageSet  = false;
     84     bool Reload = true;
     85     bool ShuffleAtStartUp = true;
     86     int PreferSmallDuringInitialShuffle = -1;
     87     size_t MaxNumberOfRuns = ULONG_MAX;
     88     int SyncTimeout = 600;
     89     int ReportSlowUnits = 10;
     90     bool OnlyASCII = false;
     91     int TBMDepth = 10;
     92     int TBMWidth = 10;
     93     std::string OutputCorpus;
     94     std::string SyncCommand;
     95     std::string ArtifactPrefix = "./";
     96     std::string ExactArtifactPath;
     97     bool SaveArtifacts = true;
     98     bool PrintNEW = true;  // Print a status line when new units are found;
     99     bool OutputCSV = false;
    100   };
    101   Fuzzer(UserSuppliedFuzzer &USF, FuzzingOptions Options);
    102   void AddToCorpus(const Unit &U) { Corpus.push_back(U); }
    103   size_t ChooseUnitIdxToMutate();
    104   const Unit &ChooseUnitToMutate() { return Corpus[ChooseUnitIdxToMutate()]; };
    105   void Loop();
    106   void Drill();
    107   void ShuffleAndMinimize();
    108   void InitializeTraceState();
    109   size_t CorpusSize() const { return Corpus.size(); }
    110   void ReadDir(const std::string &Path, long *Epoch) {
    111     Printf("Loading corpus: %s\n", Path.c_str());
    112     ReadDirToVectorOfUnits(Path.c_str(), &Corpus, Epoch);
    113   }
    114   void RereadOutputCorpus();
    115   // Save the current corpus to OutputCorpus.
    116   void SaveCorpus();
    117 
    118   size_t secondsSinceProcessStartUp() {
    119     return duration_cast<seconds>(system_clock::now() - ProcessStartTime)
    120         .count();
    121   }
    122 
    123   size_t getTotalNumberOfRuns() { return TotalNumberOfRuns; }
    124 
    125   static void StaticAlarmCallback();
    126 
    127   void ExecuteCallback(const Unit &U);
    128 
    129   // Merge Corpora[1:] into Corpora[0].
    130   void Merge(const std::vector<std::string> &Corpora);
    131 
    132  private:
    133   void AlarmCallback();
    134   void MutateAndTestOne();
    135   void ReportNewCoverage(const Unit &U);
    136   bool RunOne(const Unit &U);
    137   void RunOneAndUpdateCorpus(Unit &U);
    138   void WriteToOutputCorpus(const Unit &U);
    139   void WriteUnitToFileWithPrefix(const Unit &U, const char *Prefix);
    140   void PrintStats(const char *Where, const char *End = "\n");
    141   void PrintStatusForNewUnit(const Unit &U);
    142   void PrintUnitInASCII(const Unit &U, const char *PrintAfter = "");
    143 
    144   void SyncCorpus();
    145 
    146   size_t RecordBlockCoverage();
    147   size_t RecordCallerCalleeCoverage();
    148   void PrepareCoverageBeforeRun();
    149   bool CheckCoverageAfterRun();
    150 
    151 
    152   // Trace-based fuzzing: we run a unit with some kind of tracing
    153   // enabled and record potentially useful mutations. Then
    154   // We apply these mutations one by one to the unit and run it again.
    155 
    156   // Start tracing; forget all previously proposed mutations.
    157   void StartTraceRecording();
    158   // Stop tracing and return the number of proposed mutations.
    159   size_t StopTraceRecording();
    160   // Apply Idx-th trace-based mutation to U.
    161   void ApplyTraceBasedMutation(size_t Idx, Unit *U);
    162 
    163   void SetDeathCallback();
    164   static void StaticDeathCallback();
    165   void DeathCallback();
    166   Unit CurrentUnit;
    167 
    168   size_t TotalNumberOfRuns = 0;
    169   size_t TotalNumberOfExecutedTraceBasedMutations = 0;
    170 
    171   std::vector<Unit> Corpus;
    172   std::unordered_set<std::string> UnitHashesAddedToCorpus;
    173 
    174   // For UseCounters
    175   std::vector<uint8_t> CounterBitmap;
    176   size_t TotalBits() {  // Slow. Call it only for printing stats.
    177     size_t Res = 0;
    178     for (auto x : CounterBitmap) Res += __builtin_popcount(x);
    179     return Res;
    180   }
    181 
    182   UserSuppliedFuzzer &USF;
    183   FuzzingOptions Options;
    184   system_clock::time_point ProcessStartTime = system_clock::now();
    185   system_clock::time_point LastExternalSync = system_clock::now();
    186   system_clock::time_point UnitStartTime;
    187   long TimeOfLongestUnitInSeconds = 0;
    188   long EpochOfLastReadOfOutputCorpus = 0;
    189   size_t LastRecordedBlockCoverage = 0;
    190   size_t LastRecordedCallerCalleeCoverage = 0;
    191 };
    192 
    193 class SimpleUserSuppliedFuzzer: public UserSuppliedFuzzer {
    194  public:
    195   SimpleUserSuppliedFuzzer(FuzzerRandomBase *Rand, UserCallback Callback)
    196       : UserSuppliedFuzzer(Rand), Callback(Callback) {}
    197 
    198   virtual int TargetFunction(const uint8_t *Data, size_t Size) override {
    199     return Callback(Data, Size);
    200   }
    201 
    202  private:
    203   UserCallback Callback = nullptr;
    204 };
    205 
    206 };  // namespace fuzzer
    207 
    208 #endif // LLVM_FUZZER_INTERNAL_H
    209