blob: 975cfbdf1d1b3a12fa9210300980b41dedcc14e1 [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 Serebryany1f9c40d2016-01-09 03:46:08 +0000173static void PrintDataByte(uint8_t Byte) {
174 if (Byte == '\\')
175 Printf("\\\\");
176 else if (Byte == '"')
177 Printf("\\\"");
178 else if (Byte >= 32 && Byte < 127)
179 Printf("%c", Byte);
180 else
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000181 Printf("\\x%02x", Byte);
Kostya Serebryany1f9c40d2016-01-09 03:46:08 +0000182}
183
184static void PrintData(const uint8_t *Data, size_t Size) {
185 Printf("\"");
186 for (size_t i = 0; i < Size; i++) {
187 PrintDataByte(Data[i]);
188 }
189 Printf("\"");
190}
191
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000192const size_t TraceBasedMutation::kMaxSize;
193
Kostya Serebryany22526252015-05-11 21:16:27 +0000194class TraceState {
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000195 public:
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000196 TraceState(UserSuppliedFuzzer &USF,
197 const Fuzzer::FuzzingOptions &Options, const Unit &CurrentUnit)
198 : USF(USF), Options(Options), CurrentUnit(CurrentUnit) {
Kostya Serebryany226b7342016-01-06 00:03:35 +0000199 // Current trace collection is not thread-friendly and it probably
200 // does not have to be such, but at least we should not crash in presence
201 // of threads. So, just ignore all traces coming from all threads but one.
202 IsMyThread = true;
203 }
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000204
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000205 LabelRange GetLabelRange(dfsan_label L);
206 void DFSanCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
207 uint64_t Arg1, uint64_t Arg2, dfsan_label L1,
208 dfsan_label L2);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000209 void DFSanMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
210 const uint8_t *Data2, dfsan_label L1,
211 dfsan_label L2);
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000212 void DFSanSwitchCallback(uint64_t PC, size_t ValSizeInBits, uint64_t Val,
213 size_t NumCases, uint64_t *Cases, dfsan_label L);
Kostya Serebryanye641dd62015-09-04 22:32:25 +0000214 void TraceCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
215 uint64_t Arg1, uint64_t Arg2);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000216 void TraceMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
217 const uint8_t *Data2);
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000218
219 void TraceSwitchCallback(uintptr_t PC, size_t ValSizeInBits, uint64_t Val,
220 size_t NumCases, uint64_t *Cases);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000221 int TryToAddDesiredData(uint64_t PresentData, uint64_t DesiredData,
222 size_t DataSize);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000223 int TryToAddDesiredData(const uint8_t *PresentData,
224 const uint8_t *DesiredData, size_t DataSize);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000225
226 void StartTraceRecording() {
Kostya Serebryany8817e862015-05-11 23:25:28 +0000227 if (!Options.UseTraces) return;
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000228 RecordingTraces = true;
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000229 NumMutations = 0;
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000230 USF.GetMD().ClearAutoDictionary();
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000231 }
232
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000233 void StopTraceRecording() {
234 if (!RecordingTraces) return;
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000235 RecordingTraces = false;
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000236 for (size_t i = 0; i < NumMutations; i++) {
237 auto &M = Mutations[i];
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000238 Unit U(M.Data, M.Data + M.Size);
239 if (Options.Verbosity >= 2) {
240 AutoDictUnitCounts[U]++;
241 AutoDictAdds++;
242 if ((AutoDictAdds & (AutoDictAdds - 1)) == 0) {
243 typedef std::pair<size_t, Unit> CU;
244 std::vector<CU> CountedUnits;
245 for (auto &I : AutoDictUnitCounts)
246 CountedUnits.push_back(std::make_pair(I.second, I.first));
247 std::sort(CountedUnits.begin(), CountedUnits.end(),
248 [](const CU &a, const CU &b) { return a.first > b.first; });
249 Printf("AutoDict:\n");
250 for (auto &I : CountedUnits) {
251 Printf(" %zd ", I.first);
252 PrintData(I.second.data(), I.second.size());
253 Printf("\n");
254 }
255 }
256 }
257 USF.GetMD().AddWordToAutoDictionary(U, M.Pos);
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000258 }
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000259 }
260
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000261 void AddMutation(uint32_t Pos, uint32_t Size, const uint8_t *Data) {
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000262 if (NumMutations >= kMaxMutations) return;
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000263 assert(Size <= TraceBasedMutation::kMaxSize);
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000264 auto &M = Mutations[NumMutations++];
265 M.Pos = Pos;
266 M.Size = Size;
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000267 memcpy(M.Data, Data, Size);
268 }
269
270 void AddMutation(uint32_t Pos, uint32_t Size, uint64_t Data) {
271 assert(Size <= sizeof(Data));
272 AddMutation(Pos, Size, reinterpret_cast<uint8_t*>(&Data));
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000273 }
274
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000275 private:
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000276 bool IsTwoByteData(uint64_t Data) {
277 int64_t Signed = static_cast<int64_t>(Data);
278 Signed >>= 16;
279 return Signed == 0 || Signed == -1L;
280 }
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000281 bool RecordingTraces = false;
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000282 static const size_t kMaxMutations = 1 << 16;
283 size_t NumMutations;
284 TraceBasedMutation Mutations[kMaxMutations];
Kostya Serebryany4d623222015-11-18 01:08:30 +0000285 LabelRange LabelRanges[1 << (sizeof(dfsan_label) * 8)];
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000286 UserSuppliedFuzzer &USF;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000287 const Fuzzer::FuzzingOptions &Options;
288 const Unit &CurrentUnit;
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000289 std::map<Unit, size_t> AutoDictUnitCounts;
290 size_t AutoDictAdds = 0;
Kostya Serebryany226b7342016-01-06 00:03:35 +0000291 static thread_local bool IsMyThread;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000292};
293
Kostya Serebryany226b7342016-01-06 00:03:35 +0000294thread_local bool TraceState::IsMyThread;
295
Kostya Serebryany22526252015-05-11 21:16:27 +0000296LabelRange TraceState::GetLabelRange(dfsan_label L) {
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000297 LabelRange &LR = LabelRanges[L];
298 if (LR.Beg < LR.End || L == 0)
299 return LR;
300 const dfsan_label_info *LI = dfsan_get_label_info(L);
301 if (LI->l1 || LI->l2)
302 return LR = LabelRange::Join(GetLabelRange(LI->l1), GetLabelRange(LI->l2));
303 return LR = LabelRange::Singleton(LI);
304}
305
Kostya Serebryany22526252015-05-11 21:16:27 +0000306void TraceState::DFSanCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000307 uint64_t Arg1, uint64_t Arg2, dfsan_label L1,
308 dfsan_label L2) {
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000309 assert(ReallyHaveDFSan());
Kostya Serebryany226b7342016-01-06 00:03:35 +0000310 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000311 if (L1 == 0 && L2 == 0)
312 return; // Not actionable.
313 if (L1 != 0 && L2 != 0)
314 return; // Probably still actionable.
315 bool Res = ComputeCmp(CmpSize, CmpType, Arg1, Arg2);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000316 uint64_t Data = L1 ? Arg2 : Arg1;
317 LabelRange LR = L1 ? GetLabelRange(L1) : GetLabelRange(L2);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000318
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000319 for (size_t Pos = LR.Beg; Pos + CmpSize <= LR.End; Pos++) {
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000320 AddMutation(Pos, CmpSize, Data);
321 AddMutation(Pos, CmpSize, Data + 1);
322 AddMutation(Pos, CmpSize, Data - 1);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000323 }
324
325 if (CmpSize > LR.End - LR.Beg)
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000326 AddMutation(LR.Beg, (unsigned)(LR.End - LR.Beg), Data);
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000327
328
329 if (Options.Verbosity >= 3)
Kostya Serebryanyae7df1c2015-07-28 01:25:00 +0000330 Printf("DFSanCmpCallback: PC %lx S %zd T %zd A1 %llx A2 %llx R %d L1 %d L2 "
331 "%d MU %zd\n",
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000332 PC, CmpSize, CmpType, Arg1, Arg2, Res, L1, L2, NumMutations);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000333}
334
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000335void TraceState::DFSanMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
336 const uint8_t *Data2, dfsan_label L1,
337 dfsan_label L2) {
338
339 assert(ReallyHaveDFSan());
340 if (!RecordingTraces || !IsMyThread) return;
341 if (L1 == 0 && L2 == 0)
342 return; // Not actionable.
343 if (L1 != 0 && L2 != 0)
344 return; // Probably still actionable.
345
346 const uint8_t *Data = L1 ? Data2 : Data1;
347 LabelRange LR = L1 ? GetLabelRange(L1) : GetLabelRange(L2);
348 for (size_t Pos = LR.Beg; Pos + CmpSize <= LR.End; Pos++) {
349 AddMutation(Pos, CmpSize, Data);
350 if (Options.Verbosity >= 3)
351 Printf("DFSanMemcmpCallback: Pos %d Size %d\n", Pos, CmpSize);
352 }
353}
354
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000355void TraceState::DFSanSwitchCallback(uint64_t PC, size_t ValSizeInBits,
356 uint64_t Val, size_t NumCases,
357 uint64_t *Cases, dfsan_label L) {
358 assert(ReallyHaveDFSan());
Kostya Serebryany226b7342016-01-06 00:03:35 +0000359 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000360 if (!L) return; // Not actionable.
361 LabelRange LR = GetLabelRange(L);
362 size_t ValSize = ValSizeInBits / 8;
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000363 bool TryShort = IsTwoByteData(Val);
364 for (size_t i = 0; i < NumCases; i++)
365 TryShort &= IsTwoByteData(Cases[i]);
366
367 for (size_t Pos = LR.Beg; Pos + ValSize <= LR.End; Pos++)
368 for (size_t i = 0; i < NumCases; i++)
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000369 AddMutation(Pos, ValSize, Cases[i]);
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000370
371 if (TryShort)
372 for (size_t Pos = LR.Beg; Pos + 2 <= LR.End; Pos++)
373 for (size_t i = 0; i < NumCases; i++)
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000374 AddMutation(Pos, 2, Cases[i]);
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000375
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000376 if (Options.Verbosity >= 3)
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000377 Printf("DFSanSwitchCallback: PC %lx Val %zd SZ %zd # %zd L %d: {%d, %d} "
378 "TryShort %d\n",
379 PC, Val, ValSize, NumCases, L, LR.Beg, LR.End, TryShort);
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000380}
381
Kostya Serebryany22526252015-05-11 21:16:27 +0000382int TraceState::TryToAddDesiredData(uint64_t PresentData, uint64_t DesiredData,
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000383 size_t DataSize) {
384 int Res = 0;
385 const uint8_t *Beg = CurrentUnit.data();
386 const uint8_t *End = Beg + CurrentUnit.size();
Kostya Serebryanye641dd62015-09-04 22:32:25 +0000387 for (const uint8_t *Cur = Beg; Cur < End; Cur++) {
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000388 Cur = (uint8_t *)memmem(Cur, End - Cur, &PresentData, DataSize);
389 if (!Cur)
390 break;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000391 size_t Pos = Cur - Beg;
392 assert(Pos < CurrentUnit.size());
Kostya Serebryanye7583d22016-01-09 00:38:40 +0000393 AddMutation(Pos, DataSize, DesiredData);
394 AddMutation(Pos, DataSize, DesiredData + 1);
395 AddMutation(Pos, DataSize, DesiredData - 1);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000396 Res++;
397 }
398 return Res;
399}
400
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000401int TraceState::TryToAddDesiredData(const uint8_t *PresentData,
402 const uint8_t *DesiredData,
403 size_t DataSize) {
404 int Res = 0;
405 const uint8_t *Beg = CurrentUnit.data();
406 const uint8_t *End = Beg + CurrentUnit.size();
407 for (const uint8_t *Cur = Beg; Cur < End; Cur++) {
408 Cur = (uint8_t *)memmem(Cur, End - Cur, PresentData, DataSize);
409 if (!Cur)
410 break;
411 size_t Pos = Cur - Beg;
412 assert(Pos < CurrentUnit.size());
413 AddMutation(Pos, DataSize, DesiredData);
414 Res++;
415 }
416 return Res;
417}
418
Kostya Serebryanye641dd62015-09-04 22:32:25 +0000419void TraceState::TraceCmpCallback(uintptr_t PC, size_t CmpSize, size_t CmpType,
420 uint64_t Arg1, uint64_t Arg2) {
Kostya Serebryany226b7342016-01-06 00:03:35 +0000421 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000422 if ((CmpType == ICMP_EQ || CmpType == ICMP_NE) && Arg1 == Arg2)
423 return; // No reason to mutate.
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000424 int Added = 0;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000425 Added += TryToAddDesiredData(Arg1, Arg2, CmpSize);
426 Added += TryToAddDesiredData(Arg2, Arg1, CmpSize);
427 if (!Added && CmpSize == 4 && IsTwoByteData(Arg1) && IsTwoByteData(Arg2)) {
428 Added += TryToAddDesiredData(Arg1, Arg2, 2);
429 Added += TryToAddDesiredData(Arg2, Arg1, 2);
430 }
Kostya Serebryany859e86d2016-01-12 02:08:37 +0000431 if (Options.Verbosity >= 3 && Added)
432 Printf("TraceCmp %zd/%zd: %p %zd %zd\n", CmpSize, CmpType, PC, Arg1, Arg2);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000433}
434
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000435void TraceState::TraceMemcmpCallback(size_t CmpSize, const uint8_t *Data1,
436 const uint8_t *Data2) {
437 if (!RecordingTraces || !IsMyThread) return;
438 CmpSize = std::min(CmpSize, TraceBasedMutation::kMaxSize);
Kostya Serebryany1f9c40d2016-01-09 03:46:08 +0000439 int Added2 = TryToAddDesiredData(Data1, Data2, CmpSize);
440 int Added1 = TryToAddDesiredData(Data2, Data1, CmpSize);
441 if ((Added1 || Added2) && Options.Verbosity >= 3) {
442 Printf("MemCmp Added %d%d: ", Added1, Added2);
443 if (Added1) PrintData(Data1, CmpSize);
444 if (Added2) PrintData(Data2, CmpSize);
445 Printf("\n");
446 }
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000447}
448
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000449void TraceState::TraceSwitchCallback(uintptr_t PC, size_t ValSizeInBits,
450 uint64_t Val, size_t NumCases,
451 uint64_t *Cases) {
Kostya Serebryany226b7342016-01-06 00:03:35 +0000452 if (!RecordingTraces || !IsMyThread) return;
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000453 size_t ValSize = ValSizeInBits / 8;
454 bool TryShort = IsTwoByteData(Val);
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000455 for (size_t i = 0; i < NumCases; i++)
Kostya Serebryanyfe7e41e2015-07-31 20:58:55 +0000456 TryShort &= IsTwoByteData(Cases[i]);
457
458 if (Options.Verbosity >= 3)
459 Printf("TraceSwitch: %p %zd # %zd; TryShort %d\n", PC, Val, NumCases,
460 TryShort);
461
462 for (size_t i = 0; i < NumCases; i++) {
463 TryToAddDesiredData(Val, Cases[i], ValSize);
464 if (TryShort)
465 TryToAddDesiredData(Val, Cases[i], 2);
466 }
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000467}
468
Kostya Serebryany22526252015-05-11 21:16:27 +0000469static TraceState *TS;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000470
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000471void Fuzzer::StartTraceRecording() {
Kostya Serebryany22526252015-05-11 21:16:27 +0000472 if (!TS) return;
Kostya Serebryany4cc10d42015-08-05 23:02:57 +0000473 if (ReallyHaveDFSan())
474 for (size_t i = 0; i < static_cast<size_t>(Options.MaxLen); i++)
475 dfsan_set_label(i + 1, &CurrentUnit[i], 1);
Kostya Serebryany22526252015-05-11 21:16:27 +0000476 TS->StartTraceRecording();
Kostya Serebryanybeb24c32015-05-07 21:02:11 +0000477}
478
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000479void Fuzzer::StopTraceRecording() {
480 if (!TS) return;
481 TS->StopTraceRecording();
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000482}
483
Kostya Serebryany22526252015-05-11 21:16:27 +0000484void Fuzzer::InitializeTraceState() {
Kostya Serebryanyd8c54722015-05-12 01:58:34 +0000485 if (!Options.UseTraces) return;
Kostya Serebryanyb65805a2016-01-09 03:08:58 +0000486 TS = new TraceState(USF, Options, CurrentUnit);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000487 CurrentUnit.resize(Options.MaxLen);
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000488 // The rest really requires DFSan.
Kostya Serebryanyd8c54722015-05-12 01:58:34 +0000489 if (!ReallyHaveDFSan()) return;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000490 for (size_t i = 0; i < static_cast<size_t>(Options.MaxLen); i++) {
491 dfsan_label L = dfsan_create_label("input", (void*)(i + 1));
492 // We assume that no one else has called dfsan_create_label before.
Kostya Serebryanyd46369d2015-08-05 23:44:42 +0000493 if (L != i + 1) {
494 Printf("DFSan labels are not starting from 1, exiting\n");
495 exit(1);
496 }
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000497 }
498}
499
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000500static size_t InternalStrnlen(const char *S, size_t MaxLen) {
501 size_t Len = 0;
502 for (; Len < MaxLen && S[Len]; Len++) {}
503 return Len;
504}
505
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000506} // namespace fuzzer
507
Kostya Serebryany22526252015-05-11 21:16:27 +0000508using fuzzer::TS;
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000509
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000510extern "C" {
511void __dfsw___sanitizer_cov_trace_cmp(uint64_t SizeAndType, uint64_t Arg1,
512 uint64_t Arg2, dfsan_label L0,
513 dfsan_label L1, dfsan_label L2) {
Kostya Serebryany3fe76822015-05-29 20:31:17 +0000514 if (!TS) return;
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000515 assert(L0 == 0);
516 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
517 uint64_t CmpSize = (SizeAndType >> 32) / 8;
518 uint64_t Type = (SizeAndType << 32) >> 32;
Kostya Serebryany22526252015-05-11 21:16:27 +0000519 TS->DFSanCmpCallback(PC, CmpSize, Type, Arg1, Arg2, L1, L2);
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000520}
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000521
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000522void __dfsw___sanitizer_cov_trace_switch(uint64_t Val, uint64_t *Cases,
523 dfsan_label L1, dfsan_label L2) {
524 if (!TS) return;
525 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
526 TS->DFSanSwitchCallback(PC, Cases[1], Val, Cases[0], Cases+2, L1);
527}
528
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000529void dfsan_weak_hook_memcmp(void *caller_pc, const void *s1, const void *s2,
530 size_t n, dfsan_label s1_label,
531 dfsan_label s2_label, dfsan_label n_label) {
Kostya Serebryany3fe76822015-05-29 20:31:17 +0000532 if (!TS) return;
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000533 dfsan_label L1 = dfsan_read_label(s1, n);
534 dfsan_label L2 = dfsan_read_label(s2, n);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000535 TS->DFSanMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
536 reinterpret_cast<const uint8_t *>(s2), L1, L2);
Kostya Serebryanya407dde2015-05-07 00:11:33 +0000537}
Kostya Serebryany7d470cf2015-05-07 18:32:29 +0000538
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000539void dfsan_weak_hook_strncmp(void *caller_pc, const char *s1, const char *s2,
540 size_t n, dfsan_label s1_label,
541 dfsan_label s2_label, dfsan_label n_label) {
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000542 if (!TS) return;
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000543 n = std::min(n, fuzzer::InternalStrnlen(s1, n));
544 n = std::min(n, fuzzer::InternalStrnlen(s2, n));
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000545 dfsan_label L1 = dfsan_read_label(s1, n);
546 dfsan_label L2 = dfsan_read_label(s2, n);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000547 TS->DFSanMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
548 reinterpret_cast<const uint8_t *>(s2), L1, L2);
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000549}
550
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000551void dfsan_weak_hook_strcmp(void *caller_pc, const char *s1, const char *s2,
552 dfsan_label s1_label, dfsan_label s2_label) {
553 if (!TS) return;
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000554 size_t Len1 = strlen(s1);
555 size_t Len2 = strlen(s2);
556 size_t N = std::min(Len1, Len2);
557 if (N <= 1) return; // Not interesting.
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000558 dfsan_label L1 = dfsan_read_label(s1, Len1);
559 dfsan_label L2 = dfsan_read_label(s2, Len2);
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000560 TS->DFSanMemcmpCallback(N, reinterpret_cast<const uint8_t *>(s1),
561 reinterpret_cast<const uint8_t *>(s2), L1, L2);
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000562}
563
Kostya Serebryany0e776a22015-07-30 01:34:58 +0000564void __sanitizer_weak_hook_memcmp(void *caller_pc, const void *s1,
Kostya Serebryanye3580952016-01-12 00:43:42 +0000565 const void *s2, size_t n, int result) {
Kostya Serebryany0e776a22015-07-30 01:34:58 +0000566 if (!TS) return;
Kostya Serebryanye3580952016-01-12 00:43:42 +0000567 if (result == 0) return; // No reason to mutate.
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000568 if (n <= 1) return; // Not interesting.
569 TS->TraceMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
570 reinterpret_cast<const uint8_t *>(s2));
Kostya Serebryanyb74ba422015-07-30 02:33:45 +0000571}
572
573void __sanitizer_weak_hook_strncmp(void *caller_pc, const char *s1,
Kostya Serebryanye3580952016-01-12 00:43:42 +0000574 const char *s2, size_t n, int result) {
Kostya Serebryanyc9dc96b2015-07-30 21:22:22 +0000575 if (!TS) return;
Kostya Serebryanye3580952016-01-12 00:43:42 +0000576 if (result == 0) return; // No reason to mutate.
Kostya Serebryanycd6a4662015-07-31 17:05:05 +0000577 size_t Len1 = fuzzer::InternalStrnlen(s1, n);
578 size_t Len2 = fuzzer::InternalStrnlen(s2, n);
579 n = std::min(n, Len1);
580 n = std::min(n, Len2);
581 if (n <= 1) return; // Not interesting.
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000582 TS->TraceMemcmpCallback(n, reinterpret_cast<const uint8_t *>(s1),
583 reinterpret_cast<const uint8_t *>(s2));
Kostya Serebryany0e776a22015-07-30 01:34:58 +0000584}
585
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000586void __sanitizer_weak_hook_strcmp(void *caller_pc, const char *s1,
Kostya Serebryanye3580952016-01-12 00:43:42 +0000587 const char *s2, int result) {
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000588 if (!TS) return;
Kostya Serebryanye3580952016-01-12 00:43:42 +0000589 if (result == 0) return; // No reason to mutate.
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000590 size_t Len1 = strlen(s1);
591 size_t Len2 = strlen(s2);
592 size_t N = std::min(Len1, Len2);
593 if (N <= 1) return; // Not interesting.
Kostya Serebryanyc5733162016-01-09 01:39:55 +0000594 TS->TraceMemcmpCallback(N, reinterpret_cast<const uint8_t *>(s1),
595 reinterpret_cast<const uint8_t *>(s2));
Kostya Serebryany7f4227d2015-08-05 18:23:01 +0000596}
597
Kostya Serebryany3287d7a2015-09-30 22:22:37 +0000598__attribute__((visibility("default")))
Kostya Serebryany7d470cf2015-05-07 18:32:29 +0000599void __sanitizer_cov_trace_cmp(uint64_t SizeAndType, uint64_t Arg1,
600 uint64_t Arg2) {
Kostya Serebryany22526252015-05-11 21:16:27 +0000601 if (!TS) return;
Kostya Serebryany35959592015-07-28 00:59:53 +0000602 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
Kostya Serebryany5a99ecb2015-05-11 20:51:19 +0000603 uint64_t CmpSize = (SizeAndType >> 32) / 8;
604 uint64_t Type = (SizeAndType << 32) >> 32;
Kostya Serebryany35959592015-07-28 00:59:53 +0000605 TS->TraceCmpCallback(PC, CmpSize, Type, Arg1, Arg2);
Kostya Serebryany7d470cf2015-05-07 18:32:29 +0000606}
607
Kostya Serebryany3287d7a2015-09-30 22:22:37 +0000608__attribute__((visibility("default")))
Kostya Serebryanyfb7d8d92015-07-31 01:33:06 +0000609void __sanitizer_cov_trace_switch(uint64_t Val, uint64_t *Cases) {
610 if (!TS) return;
611 uintptr_t PC = reinterpret_cast<uintptr_t>(__builtin_return_address(0));
612 TS->TraceSwitchCallback(PC, Cases[1], Val, Cases[0], Cases + 2);
613}
614
Kostya Serebryany16d03bd2015-03-30 22:09:51 +0000615} // extern "C"