blob: 10eab0183c3d9d1c4be901d775cb9fa5eaa3f28c [file] [log] [blame]
Kostya Serebryany22526252015-05-11 21:16:27 +00001//===- FuzzerTraceState.cpp - Trace-based fuzzer mutator ------------------===//
Kostya Serebryany16d03bd2015-03-30 22:09:51 +00002//
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//===----------------------------------------------------------------------===//
Kostya Serebryany22526252015-05-11 21:16:27 +00009// This file implements a mutation algorithm based on instruction traces and
10// on taint analysis feedback from DFSan.
11//
12// Instruction traces are special hooks inserted by the compiler around
13// interesting instructions. Currently supported traces:
14// * __sanitizer_cov_trace_cmp -- inserted before every ICMP instruction,
15// receives the type, size and arguments of ICMP.
16//
17// Every time a traced event is intercepted we analyse the data involved
18// in the event and suggest a mutation for future executions.
19// For example if 4 bytes of data that derive from input bytes {4,5,6,7}
20// are compared with a constant 12345,
21// we try to insert 12345, 12344, 12346 into bytes
22// {4,5,6,7} of the next fuzzed inputs.
23//
24// The fuzzer can work only with the traces, or with both traces and DFSan.
25//
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000026// DataFlowSanitizer (DFSan) is a tool for
27// generalised dynamic data flow (taint) analysis:
28// http://clang.llvm.org/docs/DataFlowSanitizer.html .
29//
Kostya Serebryany22526252015-05-11 21:16:27 +000030// The approach with DFSan-based fuzzing has some similarity to
31// "Taint-based Directed Whitebox Fuzzing"
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000032// by Vijay Ganesh & Tim Leek & Martin Rinard:
33// http://dspace.mit.edu/openaccess-disseminate/1721.1/59320,
34// but it uses a full blown LLVM IR taint analysis and separate instrumentation
35// to analyze all of the "attack points" at once.
36//
Kostya Serebryany22526252015-05-11 21:16:27 +000037// Workflow with DFSan:
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000038// * lib/Fuzzer/Fuzzer*.cpp is compiled w/o any instrumentation.
Kostya Serebryany22526252015-05-11 21:16:27 +000039// * The code under test is compiled with DFSan *and* with instruction traces.
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000040// * Every call to HOOK(a,b) is replaced by DFSan with
41// __dfsw_HOOK(a, b, label(a), label(b)) so that __dfsw_HOOK
42// gets all the taint labels for the arguments.
43// * At the Fuzzer startup we assign a unique DFSan label
44// to every byte of the input string (Fuzzer::CurrentUnit) so that for any
45// chunk of data we know which input bytes it has derived from.
46// * The __dfsw_* functions (implemented in this file) record the
47// parameters (i.e. the application data and the corresponding taint labels)
48// in a global state.
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000049//
Kostya Serebryany22526252015-05-11 21:16:27 +000050// Parts of this code will not function when DFSan is not linked in.
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000051// Instead of using ifdefs and thus requiring a separate build of lib/Fuzzer
52// we redeclare the dfsan_* interface functions as weak and check if they
53// are nullptr before calling.
54// If this approach proves to be useful we may add attribute(weak) to the
55// dfsan declarations in dfsan_interface.h
56//
57// This module is in the "proof of concept" stage.
58// It is capable of solving only the simplest puzzles
59// like test/dfsan/DFSanSimpleCmpTest.cpp.
60//===----------------------------------------------------------------------===//
61
Kostya Serebryany22526252015-05-11 21:16:27 +000062/* Example of manual usage (-fsanitize=dataflow is optional):
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000063(
64 cd $LLVM/lib/Fuzzer/
65 clang -fPIC -c -g -O2 -std=c++11 Fuzzer*.cpp
Alexey Samsonov21a33812015-05-07 23:33:24 +000066 clang++ -O0 -std=c++11 -fsanitize-coverage=edge,trace-cmp \
Kostya Serebryany3befe942015-05-06 22:47:24 +000067 -fsanitize=dataflow \
Kostya Serebryany65f50862015-09-10 18:48:38 +000068 test/SimpleCmpTest.cpp Fuzzer*.o
69 ./a.out -use_traces=1
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000070)
71*/
72
Kostya Serebryany65f50862015-09-10 18:48:38 +000073#include "FuzzerDFSan.h"
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000074#include "FuzzerInternal.h"
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000075
Kostya Serebryanybeb24c32015-05-07 21:02:11 +000076#include <algorithm>
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000077#include <cstring>
Kostya Serebryany226b7342016-01-06 00:03:35 +000078#include <thread>
Kostya Serebryany859e86d2016-01-12 02:08:37 +000079#include <map>
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000080
Kostya Serebryany65f50862015-09-10 18:48:38 +000081#if !LLVM_FUZZER_SUPPORTS_DFSAN
82// Stubs for dfsan for platforms where dfsan does not exist and weak
83// functions don't work.
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000084extern "C" {
Kostya Serebryany65f50862015-09-10 18:48:38 +000085dfsan_label dfsan_create_label(const char *desc, void *userdata) { return 0; }
86void dfsan_set_label(dfsan_label label, void *addr, size_t size) {}
87void dfsan_add_label(dfsan_label label, void *addr, size_t size) {}
88const struct dfsan_label_info *dfsan_get_label_info(dfsan_label label) {
89 return nullptr;
90}
91dfsan_label dfsan_read_label(const void *addr, size_t size) { return 0; }
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000092} // extern "C"
Kostya Serebryany65f50862015-09-10 18:48:38 +000093#endif // !LLVM_FUZZER_SUPPORTS_DFSAN
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000094
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +000095namespace fuzzer {
96
Kostya Serebryany16d03bd2015-03-30 22:09:51 +000097// These values are copied from include/llvm/IR/InstrTypes.h.
98// We do not include the LLVM headers here to remain independent.
99// If these values ever change, an assertion in ComputeCmp will fail.
100enum Predicate {
101 ICMP_EQ = 32, ///< equal
102 ICMP_NE = 33, ///< not equal
103 ICMP_UGT = 34, ///< unsigned greater than
104 ICMP_UGE = 35, ///< unsigned greater or equal
105 ICMP_ULT = 36, ///< unsigned less than
106 ICMP_ULE = 37, ///< unsigned less or equal
107 ICMP_SGT = 38, ///< signed greater than
108 ICMP_SGE = 39, ///< signed greater or equal
109 ICMP_SLT = 40, ///< signed less than
110 ICMP_SLE = 41, ///< signed less or equal
111};
112
113template <class U, class S>
114bool ComputeCmp(size_t CmpType, U Arg1, U Arg2) {
115 switch(CmpType) {
116 case ICMP_EQ : return Arg1 == Arg2;
117 case ICMP_NE : return Arg1 != Arg2;
118 case ICMP_UGT: return Arg1 > Arg2;
119 case ICMP_UGE: return Arg1 >= Arg2;
120 case ICMP_ULT: return Arg1 < Arg2;
121 case ICMP_ULE: return Arg1 <= Arg2;
122 case ICMP_SGT: return (S)Arg1 > (S)Arg2;
123 case ICMP_SGE: return (S)Arg1 >= (S)Arg2;
124 case ICMP_SLT: return (S)Arg1 < (S)Arg2;
125 case ICMP_SLE: return (S)Arg1 <= (S)Arg2;
126 default: assert(0 && "unsupported CmpType");
127 }
128 return false;
129}
130
131static bool ComputeCmp(size_t CmpSize, size_t CmpType, uint64_t Arg1,
132 uint64_t Arg2) {
133 if (CmpSize == 8) return ComputeCmp<uint64_t, int64_t>(CmpType, Arg1, Arg2);
134 if (CmpSize == 4) return ComputeCmp<uint32_t, int32_t>(CmpType, Arg1, Arg2);
135 if (CmpSize == 2) return ComputeCmp<uint16_t, int16_t>(CmpType, Arg1, Arg2);
136 if (CmpSize == 1) return ComputeCmp<uint8_t, int8_t>(CmpType, Arg1, Arg2);
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000137 // Other size, ==
138 if (CmpType == ICMP_EQ) return Arg1 == Arg2;
Kostya Serebryany8ce74242015-08-01 01:42:51 +0000139 // assert(0 && "unsupported cmp and type size combination");
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000140 return true;
141}
142
143// As a simplification we use the range of input bytes instead of a set of input
144// bytes.
145struct LabelRange {
146 uint16_t Beg, End; // Range is [Beg, End), thus Beg==End is an empty range.
147
148 LabelRange(uint16_t Beg = 0, uint16_t End = 0) : Beg(Beg), End(End) {}
149
150 static LabelRange Join(LabelRange LR1, LabelRange LR2) {
151 if (LR1.Beg == LR1.End) return LR2;
152 if (LR2.Beg == LR2.End) return LR1;
153 return {std::min(LR1.Beg, LR2.Beg), std::max(LR1.End, LR2.End)};
154 }
155 LabelRange &Join(LabelRange LR) {
156 return *this = Join(*this, LR);
157 }
158 static LabelRange Singleton(const dfsan_label_info *LI) {
159 uint16_t Idx = (uint16_t)(uintptr_t)LI->userdata;
160 assert(Idx > 0);
161 return {(uint16_t)(Idx - 1), Idx};
162 }
163};
164
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000165// For now, very simple: put Size bytes of Data at position Pos.
166struct TraceBasedMutation {
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000167 static const size_t kMaxSize = 28;
168 uint32_t Pos : 24;
169 uint32_t Size : 8;
170 uint8_t Data[kMaxSize];
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000171};
172
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000173const size_t TraceBasedMutation::kMaxSize;
174
Kostya Serebryany22526252015-05-11 21:16:27 +0000175class TraceState {
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000176 public:
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000177 TraceState(UserSuppliedFuzzer &USF,
178 const Fuzzer::FuzzingOptions &Options, const Unit &CurrentUnit)
179 : USF(USF), Options(Options), CurrentUnit(CurrentUnit) {
Kostya Serebryany226b7342016-01-06 00:03:35 +0000180 // Current trace collection is not thread-friendly and it probably
181 // does not have to be such, but at least we should not crash in presence
182 // of threads. So, just ignore all traces coming from all threads but one.
183 IsMyThread = true;
184 }
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000185
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000186 LabelRange GetLabelRange(dfsan_label L);
187 void DFSanCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
188 uint64_t Arg1, uint64_t Arg2, dfsan_label L1,
189 dfsan_label L2);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000190 void DFSanMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
191 const uint8_t *Data2, dfsan_label L1,
192 dfsan_label L2);
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000193 void DFSanSwitchCallback(uint64_t PC, size_t ValSizeInBits, uint64_t Val,
194 size_t NumCases, uint64_t *Cases, dfsan_label L);
Kostya Serebryanye641dd62015-09-04 22:32:25 +0000195 void TraceCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
196 uint64_t Arg1, uint64_t Arg2);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000197 void TraceMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
198 const uint8_t *Data2);
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000199
200 void TraceSwitchCallback(uintptr_t PC, size_t ValSizeInBits, uint64_t Val,
201 size_t NumCases, uint64_t *Cases);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000202 int TryToAddDesiredData(uint64_t PresentData, uint64_t DesiredData,
203 size_t DataSize);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000204 int TryToAddDesiredData(const uint8_t *PresentData,
205 const uint8_t *DesiredData, size_t DataSize);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000206
207 void StartTraceRecording() {
Kostya Serebryany8817e862015-05-11 23:25:28 +0000208 if (!Options.UseTraces) return;
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000209 RecordingTraces = true;
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000210 NumMutations = 0;
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000211 USF.GetMD().ClearAutoDictionary();
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000212 }
213
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000214 void StopTraceRecording() {
215 if (!RecordingTraces) return;
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000216 RecordingTraces = false;
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000217 for (size_t i = 0; i < NumMutations; i++) {
218 auto &M = Mutations[i];
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000219 Unit U(M.Data, M.Data + M.Size);
220 if (Options.Verbosity >= 2) {
221 AutoDictUnitCounts[U]++;
222 AutoDictAdds++;
223 if ((AutoDictAdds & (AutoDictAdds - 1)) == 0) {
224 typedef std::pair<size_t, Unit> CU;
225 std::vector<CU> CountedUnits;
226 for (auto &I : AutoDictUnitCounts)
227 CountedUnits.push_back(std::make_pair(I.second, I.first));
228 std::sort(CountedUnits.begin(), CountedUnits.end(),
229 [](const CU &a, const CU &b) { return a.first > b.first; });
230 Printf("AutoDict:\n");
231 for (auto &I : CountedUnits) {
232 Printf(" %zd ", I.first);
Kostya Serebryany41740052016-01-12 02:36:59 +0000233 PrintASCII(I.second);
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000234 Printf("\n");
235 }
236 }
237 }
238 USF.GetMD().AddWordToAutoDictionary(U, M.Pos);
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000239 }
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000240 }
241
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000242 void AddMutation(uint32_t Pos, uint32_t Size, const uint8_t *Data) {
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000243 if (NumMutations >= kMaxMutations) return;
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000244 assert(Size <= TraceBasedMutation::kMaxSize);
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000245 auto &M = Mutations[NumMutations++];
246 M.Pos = Pos;
247 M.Size = Size;
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000248 memcpy(M.Data, Data, Size);
249 }
250
251 void AddMutation(uint32_t Pos, uint32_t Size, uint64_t Data) {
252 assert(Size <= sizeof(Data));
253 AddMutation(Pos, Size, reinterpret_cast<uint8_t*>(&Data));
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000254 }
255
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000256 private:
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000257 bool IsTwoByteData(uint64_t Data) {
258 int64_t Signed = static_cast<int64_t>(Data);
259 Signed >>= 16;
260 return Signed == 0 || Signed == -1L;
261 }
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000262 bool RecordingTraces = false;
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000263 static const size_t kMaxMutations = 1 << 16;
264 size_t NumMutations;
265 TraceBasedMutation Mutations[kMaxMutations];
Kostya Serebryany4d623222015-11-18 01:08:30 +0000266 LabelRange LabelRanges[1 << (sizeof(dfsan_label) * 8)];
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000267 UserSuppliedFuzzer &USF;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000268 const Fuzzer::FuzzingOptions &Options;
269 const Unit &CurrentUnit;
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000270 std::map<Unit, size_t> AutoDictUnitCounts;
271 size_t AutoDictAdds = 0;
Kostya Serebryany226b7342016-01-06 00:03:35 +0000272 static thread_local bool IsMyThread;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000273};
274
Kostya Serebryany226b7342016-01-06 00:03:35 +0000275thread_local bool TraceState::IsMyThread;
276
Kostya Serebryany22526252015-05-11 21:16:27 +0000277LabelRange TraceState::GetLabelRange(dfsan_label L) {
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000278 LabelRange &LR = LabelRanges[L];
279 if (LR.Beg < LR.End || L == 0)
280 return LR;
281 const dfsan_label_info *LI = dfsan_get_label_info(L);
282 if (LI->l1 || LI->l2)
283 return LR = LabelRange::Join(GetLabelRange(LI->l1), GetLabelRange(LI->l2));
284 return LR = LabelRange::Singleton(LI);
285}
286
Kostya Serebryany22526252015-05-11 21:16:27 +0000287void TraceState::DFSanCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000288 uint64_t Arg1, uint64_t Arg2, dfsan_label L1,
289 dfsan_label L2) {
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000290 assert(ReallyHaveDFSan());
Kostya Serebryany226b7342016-01-06 00:03:35 +0000291 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000292 if (L1 == 0 && L2 == 0)
293 return; // Not actionable.
294 if (L1 != 0 && L2 != 0)
295 return; // Probably still actionable.
296 bool Res = ComputeCmp(CmpSize, CmpType, Arg1, Arg2);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000297 uint64_t Data = L1 ? Arg2 : Arg1;
298 LabelRange LR = L1 ? GetLabelRange(L1) : GetLabelRange(L2);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000299
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000300 for (size_t Pos = LR.Beg; Pos + CmpSize <= LR.End; Pos++) {
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000301 AddMutation(Pos, CmpSize, Data);
302 AddMutation(Pos, CmpSize, Data + 1);
303 AddMutation(Pos, CmpSize, Data - 1);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000304 }
305
306 if (CmpSize > LR.End - LR.Beg)
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000307 AddMutation(LR.Beg, (unsigned)(LR.End - LR.Beg), Data);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000308
309
310 if (Options.Verbosity >= 3)
Kostya Serebryanyae7df1c2015-07-28 01:25:00 +0000311 Printf("DFSanCmpCallback: PC %lx S %zd T %zd A1 %llx A2 %llx R %d L1 %d L2 "
312 "%d MU %zd\n",
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000313 PC, CmpSize, CmpType, Arg1, Arg2, Res, L1, L2, NumMutations);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000314}
315
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000316void TraceState::DFSanMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
317 const uint8_t *Data2, dfsan_label L1,
318 dfsan_label L2) {
319
320 assert(ReallyHaveDFSan());
321 if (!RecordingTraces || !IsMyThread) return;
322 if (L1 == 0 && L2 == 0)
323 return; // Not actionable.
324 if (L1 != 0 && L2 != 0)
325 return; // Probably still actionable.
326
327 const uint8_t *Data = L1 ? Data2 : Data1;
328 LabelRange LR = L1 ? GetLabelRange(L1) : GetLabelRange(L2);
329 for (size_t Pos = LR.Beg; Pos + CmpSize <= LR.End; Pos++) {
330 AddMutation(Pos, CmpSize, Data);
331 if (Options.Verbosity >= 3)
332 Printf("DFSanMemcmpCallback: Pos %d Size %d\n", Pos, CmpSize);
333 }
334}
335
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000336void TraceState::DFSanSwitchCallback(uint64_t PC, size_t ValSizeInBits,
337 uint64_t Val, size_t NumCases,
338 uint64_t *Cases, dfsan_label L) {
339 assert(ReallyHaveDFSan());
Kostya Serebryany226b7342016-01-06 00:03:35 +0000340 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000341 if (!L) return; // Not actionable.
342 LabelRange LR = GetLabelRange(L);
343 size_t ValSize = ValSizeInBits / 8;
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000344 bool TryShort = IsTwoByteData(Val);
345 for (size_t i = 0; i < NumCases; i++)
346 TryShort &= IsTwoByteData(Cases[i]);
347
348 for (size_t Pos = LR.Beg; Pos + ValSize <= LR.End; Pos++)
349 for (size_t i = 0; i < NumCases; i++)
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000350 AddMutation(Pos, ValSize, Cases[i]);
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000351
352 if (TryShort)
353 for (size_t Pos = LR.Beg; Pos + 2 <= LR.End; Pos++)
354 for (size_t i = 0; i < NumCases; i++)
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000355 AddMutation(Pos, 2, Cases[i]);
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000356
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000357 if (Options.Verbosity >= 3)
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000358 Printf("DFSanSwitchCallback: PC %lx Val %zd SZ %zd # %zd L %d: {%d, %d} "
359 "TryShort %d\n",
360 PC, Val, ValSize, NumCases, L, LR.Beg, LR.End, TryShort);
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000361}
362
Kostya Serebryany22526252015-05-11 21:16:27 +0000363int TraceState::TryToAddDesiredData(uint64_t PresentData, uint64_t DesiredData,
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000364 size_t DataSize) {
365 int Res = 0;
366 const uint8_t *Beg = CurrentUnit.data();
367 const uint8_t *End = Beg + CurrentUnit.size();
Kostya Serebryanye641dd62015-09-04 22:32:25 +0000368 for (const uint8_t *Cur = Beg; Cur < End; Cur++) {
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000369 Cur = (uint8_t *)memmem(Cur, End - Cur, &PresentData, DataSize);
370 if (!Cur)
371 break;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000372 size_t Pos = Cur - Beg;
373 assert(Pos < CurrentUnit.size());
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000374 AddMutation(Pos, DataSize, DesiredData);
375 AddMutation(Pos, DataSize, DesiredData + 1);
376 AddMutation(Pos, DataSize, DesiredData - 1);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000377 Res++;
378 }
379 return Res;
380}
381
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000382int TraceState::TryToAddDesiredData(const uint8_t *PresentData,
383 const uint8_t *DesiredData,
384 size_t DataSize) {
385 int Res = 0;
386 const uint8_t *Beg = CurrentUnit.data();
387 const uint8_t *End = Beg + CurrentUnit.size();
388 for (const uint8_t *Cur = Beg; Cur < End; Cur++) {
389 Cur = (uint8_t *)memmem(Cur, End - Cur, PresentData, DataSize);
390 if (!Cur)
391 break;
392 size_t Pos = Cur - Beg;
393 assert(Pos < CurrentUnit.size());
394 AddMutation(Pos, DataSize, DesiredData);
395 Res++;
396 }
397 return Res;
398}
399
Kostya Serebryanye641dd62015-09-04 22:32:25 +0000400void TraceState::TraceCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
401 uint64_t Arg1, uint64_t Arg2) {
Kostya Serebryany226b7342016-01-06 00:03:35 +0000402 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000403 if ((CmpType == ICMP_EQ || CmpType == ICMP_NE) && Arg1 == Arg2)
404 return; // No reason to mutate.
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000405 int Added = 0;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000406 Added += TryToAddDesiredData(Arg1, Arg2, CmpSize);
407 Added += TryToAddDesiredData(Arg2, Arg1, CmpSize);
408 if (!Added && CmpSize == 4 && IsTwoByteData(Arg1) && IsTwoByteData(Arg2)) {
409 Added += TryToAddDesiredData(Arg1, Arg2, 2);
410 Added += TryToAddDesiredData(Arg2, Arg1, 2);
411 }
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000412 if (Options.Verbosity >= 3 && Added)
413 Printf("TraceCmp %zd/%zd: %p %zd %zd\n", CmpSize, CmpType, PC, Arg1, Arg2);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000414}
415
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000416void TraceState::TraceMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
417 const uint8_t *Data2) {
418 if (!RecordingTraces || !IsMyThread) return;
419 CmpSize = std::min(CmpSize, TraceBasedMutation::kMaxSize);
Kostya Serebryany1f9c40d2016-01-09 03:46:08 +0000420 int Added2 = TryToAddDesiredData(Data1, Data2, CmpSize);
421 int Added1 = TryToAddDesiredData(Data2, Data1, CmpSize);
422 if ((Added1 || Added2) && Options.Verbosity >= 3) {
423 Printf("MemCmp Added %d%d: ", Added1, Added2);
Kostya Serebryany41740052016-01-12 02:36:59 +0000424 if (Added1) PrintASCII(Data1, CmpSize);
425 if (Added2) PrintASCII(Data2, CmpSize);
Kostya Serebryany1f9c40d2016-01-09 03:46:08 +0000426 Printf("\n");
427 }
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000428}
429
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000430void TraceState::TraceSwitchCallback(uintptr_t PC, size_t ValSizeInBits,
431 uint64_t Val, size_t NumCases,
432 uint64_t *Cases) {
Kostya Serebryany226b7342016-01-06 00:03:35 +0000433 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000434 size_t ValSize = ValSizeInBits / 8;
435 bool TryShort = IsTwoByteData(Val);
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000436 for (size_t i = 0; i < NumCases; i++)
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000437 TryShort &= IsTwoByteData(Cases[i]);
438
439 if (Options.Verbosity >= 3)
440 Printf("TraceSwitch: %p %zd # %zd; TryShort %d\n", PC, Val, NumCases,
441 TryShort);
442
443 for (size_t i = 0; i < NumCases; i++) {
444 TryToAddDesiredData(Val, Cases[i], ValSize);
445 if (TryShort)
446 TryToAddDesiredData(Val, Cases[i], 2);
447 }
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000448}
449
Kostya Serebryany22526252015-05-11 21:16:27 +0000450static TraceState *TS;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000451
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000452void Fuzzer::StartTraceRecording() {
Kostya Serebryany22526252015-05-11 21:16:27 +0000453 if (!TS) return;
454 TS->StartTraceRecording();
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000455}
456
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000457void Fuzzer::StopTraceRecording() {
458 if (!TS) return;
459 TS->StopTraceRecording();
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000460}
461
Kostya Serebryanyd50a3ee2016-01-13 23:02:30 +0000462void Fuzzer::AssignTaintLabels(uint8_t *Data, size_t Size) {
463 if (!Options.UseTraces) return;
464 if (!ReallyHaveDFSan()) return;
465 for (size_t i = 0; i < Size; i++)
466 dfsan_set_label(i + 1, &Data[i], 1);
467}
468
Kostya Serebryany22526252015-05-11 21:16:27 +0000469void Fuzzer::InitializeTraceState() {
Kostya Serebryanyd8c54722015-05-12 01:58:34 +0000470 if (!Options.UseTraces) return;
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000471 TS = new TraceState(USF, Options, CurrentUnit);
Kostya Serebryanyd50a3ee2016-01-13 23:02:30 +0000472 if (ReallyHaveDFSan()) {
473 for (size_t i = 0; i < static_cast<size_t>(Options.MaxLen); i++) {
474 dfsan_label L = dfsan_create_label("input", (void *)(i + 1));
475 // We assume that no one else has called dfsan_create_label before.
476 if (L != i + 1) {
477 Printf("DFSan labels are not starting from 1, exiting\n");
478 exit(1);
479 }
Kostya Serebryanyd46369d2015-08-05 23:44:42 +0000480 }
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000481 }
482}
483
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000484static size_t InternalStrnlen(const char *S, size_t MaxLen) {
485 size_t Len = 0;
486 for (; Len < MaxLen && S[Len]; Len++) {}
487 return Len;
488}
489
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000490} // namespace fuzzer
491
Kostya Serebryany22526252015-05-11 21:16:27 +0000492using fuzzer::TS;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000493
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000494extern "C" {
495void __dfsw___sanitizer_cov_trace_cmp(uint64_t SizeAndType, uint64_t Arg1,
496 uint64_t Arg2, dfsan_label L0,
497 dfsan_label L1, dfsan_label L2) {
Kostya Serebryany3fe76822015-05-29 20:31:17 +0000498 if (!TS) return;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000499 assert(L0 == 0);
500 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
501 uint64_t CmpSize = (SizeAndType >> 32) / 8;
502 uint64_t Type = (SizeAndType << 32) >> 32;
Kostya Serebryany22526252015-05-11 21:16:27 +0000503 TS->DFSanCmpCallback(PC, CmpSize, Type, Arg1, Arg2, L1, L2);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000504}
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000505
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000506void __dfsw___sanitizer_cov_trace_switch(uint64_t Val, uint64_t *Cases,
507 dfsan_label L1, dfsan_label L2) {
508 if (!TS) return;
509 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
510 TS->DFSanSwitchCallback(PC, Cases[1], Val, Cases[0], Cases+2, L1);
511}
512
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000513void dfsan_weak_hook_memcmp(void *caller_pc, const void *s1, const void *s2,
514 size_t n, dfsan_label s1_label,
515 dfsan_label s2_label, dfsan_label n_label) {
Kostya Serebryany3fe76822015-05-29 20:31:17 +0000516 if (!TS) return;
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000517 dfsan_label L1 = dfsan_read_label(s1, n);
518 dfsan_label L2 = dfsan_read_label(s2, n);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000519 TS->DFSanMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
520 reinterpret_cast<const uint8_t *>(s2), L1, L2);
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000521}
Kostya Serebryany7d470cf2015-05-07 18:32:29 +0000522
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000523void dfsan_weak_hook_strncmp(void *caller_pc, const char *s1, const char *s2,
524 size_t n, dfsan_label s1_label,
525 dfsan_label s2_label, dfsan_label n_label) {
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000526 if (!TS) return;
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000527 n = std::min(n, fuzzer::InternalStrnlen(s1, n));
528 n = std::min(n, fuzzer::InternalStrnlen(s2, n));
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000529 dfsan_label L1 = dfsan_read_label(s1, n);
530 dfsan_label L2 = dfsan_read_label(s2, n);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000531 TS->DFSanMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
532 reinterpret_cast<const uint8_t *>(s2), L1, L2);
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000533}
534
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000535void dfsan_weak_hook_strcmp(void *caller_pc, const char *s1, const char *s2,
536 dfsan_label s1_label, dfsan_label s2_label) {
537 if (!TS) return;
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000538 size_t Len1 = strlen(s1);
539 size_t Len2 = strlen(s2);
540 size_t N = std::min(Len1, Len2);
541 if (N <= 1) return; // Not interesting.
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000542 dfsan_label L1 = dfsan_read_label(s1, Len1);
543 dfsan_label L2 = dfsan_read_label(s2, Len2);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000544 TS->DFSanMemcmpCallback(N, reinterpret_cast<const uint8_t *>(s1),
545 reinterpret_cast<const uint8_t *>(s2), L1, L2);
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000546}
547
Kostya Serebryany4b83a4f2016-01-12 16:50:18 +0000548// We may need to avoid defining weak hooks to stay compatible with older clang.
549#ifndef LLVM_FUZZER_DEFINES_SANITIZER_WEAK_HOOOKS
550# define LLVM_FUZZER_DEFINES_SANITIZER_WEAK_HOOOKS 1
551#endif
552
553#if LLVM_FUZZER_DEFINES_SANITIZER_WEAK_HOOOKS
Kostya Serebryany0e776a22015-07-30 01:34:58 +0000554void __sanitizer_weak_hook_memcmp(void *caller_pc, const void *s1,
Kostya Serebryanye3580952016-01-12 00:43:42 +0000555 const void *s2, size_t n, int result) {
Kostya Serebryany0e776a22015-07-30 01:34:58 +0000556 if (!TS) return;
Kostya Serebryanye3580952016-01-12 00:43:42 +0000557 if (result == 0) return; // No reason to mutate.
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000558 if (n <= 1) return; // Not interesting.
559 TS->TraceMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
560 reinterpret_cast<const uint8_t *>(s2));
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000561}
562
563void __sanitizer_weak_hook_strncmp(void *caller_pc, const char *s1,
Kostya Serebryanye3580952016-01-12 00:43:42 +0000564 const char *s2, size_t n, int result) {
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000565 if (!TS) return;
Kostya Serebryanye3580952016-01-12 00:43:42 +0000566 if (result == 0) return; // No reason to mutate.
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000567 size_t Len1 = fuzzer::InternalStrnlen(s1, n);
568 size_t Len2 = fuzzer::InternalStrnlen(s2, n);
569 n = std::min(n, Len1);
570 n = std::min(n, Len2);
571 if (n <= 1) return; // Not interesting.
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000572 TS->TraceMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
573 reinterpret_cast<const uint8_t *>(s2));
Kostya Serebryany0e776a22015-07-30 01:34:58 +0000574}
575
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000576void __sanitizer_weak_hook_strcmp(void *caller_pc, const char *s1,
Kostya Serebryanye3580952016-01-12 00:43:42 +0000577 const char *s2, int result) {
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000578 if (!TS) return;
Kostya Serebryanye3580952016-01-12 00:43:42 +0000579 if (result == 0) return; // No reason to mutate.
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000580 size_t Len1 = strlen(s1);
581 size_t Len2 = strlen(s2);
582 size_t N = std::min(Len1, Len2);
583 if (N <= 1) return; // Not interesting.
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000584 TS->TraceMemcmpCallback(N, reinterpret_cast<const uint8_t *>(s1),
585 reinterpret_cast<const uint8_t *>(s2));
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000586}
587
Kostya Serebryany4b83a4f2016-01-12 16:50:18 +0000588#endif // LLVM_FUZZER_DEFINES_SANITIZER_WEAK_HOOOKS
589
Kostya Serebryany3287d7a2015-09-30 22:22:37 +0000590__attribute__((visibility("default")))
Kostya Serebryany7d470cf2015-05-07 18:32:29 +0000591void __sanitizer_cov_trace_cmp(uint64_t SizeAndType, uint64_t Arg1,
592 uint64_t Arg2) {
Kostya Serebryany22526252015-05-11 21:16:27 +0000593 if (!TS) return;
Kostya Serebryany35959592015-07-28 00:59:53 +0000594 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000595 uint64_t CmpSize = (SizeAndType >> 32) / 8;
596 uint64_t Type = (SizeAndType << 32) >> 32;
Kostya Serebryany35959592015-07-28 00:59:53 +0000597 TS->TraceCmpCallback(PC, CmpSize, Type, Arg1, Arg2);
Kostya Serebryany7d470cf2015-05-07 18:32:29 +0000598}
599
Kostya Serebryany3287d7a2015-09-30 22:22:37 +0000600__attribute__((visibility("default")))
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000601void __sanitizer_cov_trace_switch(uint64_t Val, uint64_t *Cases) {
602 if (!TS) return;
603 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
604 TS->TraceSwitchCallback(PC, Cases[1], Val, Cases[0], Cases + 2);
605}
606
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000607} // extern "C"