blob: 5899c677085b24fa0d31d843a85d48bef603abbb [file] [log] [blame]
Kostya Serebryany6f5a8042016-09-21 01:50:50 +00001//===- FuzzerCorpus.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// fuzzer::InputCorpus
10//===----------------------------------------------------------------------===//
11
12#ifndef LLVM_FUZZER_CORPUS
13#define LLVM_FUZZER_CORPUS
14
Kostya Serebryany20801e12016-09-21 21:41:48 +000015#include <random>
Kostya Serebryany29bb6642016-09-21 22:42:17 +000016#include <unordered_set>
Kostya Serebryany20801e12016-09-21 21:41:48 +000017
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000018#include "FuzzerDefs.h"
Kostya Serebryany20801e12016-09-21 21:41:48 +000019#include "FuzzerRandom.h"
Kostya Serebryany2c556132016-09-30 01:19:56 +000020#include "FuzzerTracePC.h"
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000021
22namespace fuzzer {
23
24struct InputInfo {
25 Unit U; // The actual input data.
Kostya Serebryany20801e12016-09-21 21:41:48 +000026 uint8_t Sha1[kSHA1NumBytes]; // Checksum.
Kostya Serebryany2c556132016-09-30 01:19:56 +000027 // Number of features that this input has and no smaller input has.
28 size_t NumFeatures = 0;
29 size_t Tmp = 0; // Used by ValidateFeatureSet.
Kostya Serebryany29bb6642016-09-21 22:42:17 +000030 // Stats.
Kostya Serebryany2c556132016-09-30 01:19:56 +000031 size_t NumExecutedMutations = 0;
32 size_t NumSuccessfullMutations = 0;
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000033};
34
35class InputCorpus {
36 public:
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000037 static const size_t kFeatureSetSize = 1 << 16;
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000038 InputCorpus() {
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000039 memset(InputSizesPerFeature, 0, sizeof(InputSizesPerFeature));
40 memset(SmallestElementPerFeature, 0, sizeof(SmallestElementPerFeature));
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000041 }
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000042 ~InputCorpus() {
43 for (auto II : Inputs)
44 delete II;
45 }
Kostya Serebryany20801e12016-09-21 21:41:48 +000046 size_t size() const { return Inputs.size(); }
Kostya Serebryany2455f0d2016-10-05 00:25:17 +000047 size_t SizeInBytes() const {
48 size_t Res = 0;
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000049 for (auto II : Inputs)
50 Res += II->U.size();
Kostya Serebryany2455f0d2016-10-05 00:25:17 +000051 return Res;
52 }
53 size_t NumActiveUnits() const {
54 size_t Res = 0;
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000055 for (auto II : Inputs)
56 Res += !II->U.empty();
Kostya Serebryany2455f0d2016-10-05 00:25:17 +000057 return Res;
58 }
Kostya Serebryany20801e12016-09-21 21:41:48 +000059 bool empty() const { return Inputs.empty(); }
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000060 const Unit &operator[] (size_t Idx) const { return Inputs[Idx]->U; }
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000061 void AddToCorpus(const Unit &U, size_t NumFeatures) {
Kostya Serebryany2455f0d2016-10-05 00:25:17 +000062 assert(!U.empty());
Kostya Serebryany624f59f2016-09-22 01:34:58 +000063 uint8_t Hash[kSHA1NumBytes];
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000064 if (FeatureDebug)
65 Printf("ADD_TO_CORPUS %zd NF %zd\n", Inputs.size(), NumFeatures);
Kostya Serebryany624f59f2016-09-22 01:34:58 +000066 ComputeSHA1(U.data(), U.size(), Hash);
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000067 Hashes.insert(Sha1ToString(Hash));
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000068 Inputs.push_back(new InputInfo());
69 InputInfo &II = *Inputs.back();
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000070 II.U = U;
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000071 II.NumFeatures = NumFeatures;
Kostya Serebryany624f59f2016-09-22 01:34:58 +000072 memcpy(II.Sha1, Hash, kSHA1NumBytes);
Kostya Serebryany20801e12016-09-21 21:41:48 +000073 UpdateCorpusDistribution();
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +000074 ValidateFeatureSet();
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000075 }
76
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000077 bool HasUnit(const Unit &U) { return Hashes.count(Hash(U)); }
Kostya Serebryanyd2169222016-10-01 01:04:29 +000078 bool HasUnit(const std::string &H) { return Hashes.count(H); }
Kostya Serebryany29bb6642016-09-21 22:42:17 +000079 InputInfo &ChooseUnitToMutate(Random &Rand) {
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000080 InputInfo &II = *Inputs[ChooseUnitIdxToMutate(Rand)];
Kostya Serebryany2455f0d2016-10-05 00:25:17 +000081 assert(!II.U.empty());
82 return II;
Kostya Serebryany20801e12016-09-21 21:41:48 +000083 };
Kostya Serebryany6f5a8042016-09-21 01:50:50 +000084
Kostya Serebryany20801e12016-09-21 21:41:48 +000085 // Returns an index of random unit from the corpus to mutate.
86 // Hypothesis: units added to the corpus last are more likely to be
87 // interesting. This function gives more weight to the more recent units.
88 size_t ChooseUnitIdxToMutate(Random &Rand) {
Kostya Serebryany29bb6642016-09-21 22:42:17 +000089 size_t Idx = static_cast<size_t>(CorpusDistribution(Rand.Get_mt19937()));
Kostya Serebryany20801e12016-09-21 21:41:48 +000090 assert(Idx < Inputs.size());
91 return Idx;
92 }
93
Kostya Serebryany29bb6642016-09-21 22:42:17 +000094 void PrintStats() {
95 for (size_t i = 0; i < Inputs.size(); i++) {
Kostya Serebryanycd04ec22016-10-08 21:57:48 +000096 const auto &II = *Inputs[i];
Kostya Serebryany16a145f2016-09-23 01:58:51 +000097 Printf(" [%zd %s]\tsz: %zd\truns: %zd\tsucc: %zd\n", i,
Kostya Serebryany29bb6642016-09-21 22:42:17 +000098 Sha1ToString(II.Sha1).c_str(), II.U.size(),
Kostya Serebryany16a145f2016-09-23 01:58:51 +000099 II.NumExecutedMutations, II.NumSuccessfullMutations);
Kostya Serebryany29bb6642016-09-21 22:42:17 +0000100 }
101 }
102
Kostya Serebryany2c556132016-09-30 01:19:56 +0000103 void PrintFeatureSet() {
Kostya Serebryany2c556132016-09-30 01:19:56 +0000104 for (size_t i = 0; i < kFeatureSetSize; i++) {
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000105 if(size_t Sz = GetFeature(i))
106 Printf("[%zd: id %zd sz%zd] ", i, SmallestElementPerFeature[i], Sz);
Kostya Serebryany2c556132016-09-30 01:19:56 +0000107 }
108 Printf("\n\t");
109 for (size_t i = 0; i < Inputs.size(); i++)
Kostya Serebryanycd04ec22016-10-08 21:57:48 +0000110 if (size_t N = Inputs[i]->NumFeatures)
Kostya Serebryany2c556132016-09-30 01:19:56 +0000111 Printf(" %zd=>%zd ", i, N);
112 Printf("\n");
113 }
114
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000115 bool AddFeature(size_t Idx, uint32_t NewSize, bool Shrink) {
116 assert(NewSize);
117 Idx = Idx % kFeatureSetSize;
118 uint32_t OldSize = GetFeature(Idx);
119 if (OldSize == 0 || (Shrink && OldSize > NewSize)) {
120 if (OldSize > 0) {
Kostya Serebryanycd04ec22016-10-08 21:57:48 +0000121 InputInfo &II = *Inputs[SmallestElementPerFeature[Idx]];
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000122 assert(II.NumFeatures > 0);
123 II.NumFeatures--;
124 if (II.NumFeatures == 0) {
Kostya Serebryany936b1e72016-10-06 05:14:00 +0000125 Unit().swap(II.U);
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000126 if (FeatureDebug)
127 Printf("EVICTED %zd\n", SmallestElementPerFeature[Idx]);
128 }
129 }
130 if (FeatureDebug)
131 Printf("ADD FEATURE %zd sz %d\n", Idx, NewSize);
132 SmallestElementPerFeature[Idx] = Inputs.size();
133 InputSizesPerFeature[Idx] = NewSize;
134 CountingFeatures = true;
135 return true;
136 }
137 return false;
138 }
139
140 size_t NumFeatures() const {
141 size_t Res = 0;
142 for (size_t i = 0; i < kFeatureSetSize; i++)
143 Res += GetFeature(i) != 0;
144 return Res;
145 }
146
Kostya Serebryany20801e12016-09-21 21:41:48 +0000147private:
148
Kostya Serebryany2c556132016-09-30 01:19:56 +0000149 static const bool FeatureDebug = false;
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000150
151 size_t GetFeature(size_t Idx) const { return InputSizesPerFeature[Idx]; }
Kostya Serebryany2c556132016-09-30 01:19:56 +0000152
153 void ValidateFeatureSet() {
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000154 if (!CountingFeatures) return;
155 if (FeatureDebug)
156 PrintFeatureSet();
157 for (size_t Idx = 0; Idx < kFeatureSetSize; Idx++)
158 if (GetFeature(Idx))
Kostya Serebryanycd04ec22016-10-08 21:57:48 +0000159 Inputs[SmallestElementPerFeature[Idx]]->Tmp++;
160 for (auto II: Inputs) {
161 if (II->Tmp != II->NumFeatures)
162 Printf("ZZZ %zd %zd\n", II->Tmp, II->NumFeatures);
163 assert(II->Tmp == II->NumFeatures);
164 II->Tmp = 0;
Kostya Serebryany2c556132016-09-30 01:19:56 +0000165 }
166 }
167
Kostya Serebryany20801e12016-09-21 21:41:48 +0000168 // Updates the probability distribution for the units in the corpus.
169 // Must be called whenever the corpus or unit weights are changed.
170 void UpdateCorpusDistribution() {
171 size_t N = Inputs.size();
Kostya Serebryany5a52a112016-10-04 01:51:44 +0000172 Intervals.resize(N + 1);
173 Weights.resize(N);
Kostya Serebryany20801e12016-09-21 21:41:48 +0000174 std::iota(Intervals.begin(), Intervals.end(), 0);
Kostya Serebryany5a52a112016-10-04 01:51:44 +0000175 if (CountingFeatures)
176 for (size_t i = 0; i < N; i++)
Kostya Serebryanycd04ec22016-10-08 21:57:48 +0000177 Weights[i] = Inputs[i]->NumFeatures * (i + 1);
Kostya Serebryany5a52a112016-10-04 01:51:44 +0000178 else
179 std::iota(Weights.begin(), Weights.end(), 1);
Kostya Serebryany20801e12016-09-21 21:41:48 +0000180 CorpusDistribution = std::piecewise_constant_distribution<double>(
181 Intervals.begin(), Intervals.end(), Weights.begin());
182 }
183 std::piecewise_constant_distribution<double> CorpusDistribution;
184
Kostya Serebryany5a52a112016-10-04 01:51:44 +0000185 std::vector<double> Intervals;
186 std::vector<double> Weights;
187
Kostya Serebryany6f5a8042016-09-21 01:50:50 +0000188 std::unordered_set<std::string> Hashes;
Kostya Serebryanycd04ec22016-10-08 21:57:48 +0000189 std::vector<InputInfo*> Inputs;
Kostya Serebryany2c556132016-09-30 01:19:56 +0000190
Kostya Serebryany5a52a112016-10-04 01:51:44 +0000191 bool CountingFeatures = false;
Kostya Serebryany1c73f1b2016-10-05 22:56:21 +0000192 uint32_t InputSizesPerFeature[kFeatureSetSize];
193 uint32_t SmallestElementPerFeature[kFeatureSetSize];
Kostya Serebryany6f5a8042016-09-21 01:50:50 +0000194};
195
196} // namespace fuzzer
197
198#endif // LLVM_FUZZER_CORPUS