• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# Feedback-driven fuzzing #
2
3Honggfuzz is capable of performing feedback-guided (code coverage driven) fuzzing. It can utilize the following sources of data:
4  * (Linux) Hardware-based counters (instructions, branches)
5  * (Linux) Intel BTS code coverage (kernel >= 4.2)
6  * (Linux) Intel PT code coverage (kernel >= 4.2)
7  * Sanitzer-coverage instrumentation (`-fsanitize-coverage=bb`)
8  * Compile-time instrumentation (`-finstrument-functions` or `-fsanitize-coverage=trace-pc[-guard],indirect-calls,trace-cmp` or both)
9
10Developers should provide the initial file corpus which will be gradually improved upon. It can even comprise of a single 1-byte initial file, and honggfuzz will try to generate better inputs starting from there.
11
12---
13# Requirements for software-based coverage-guided fuzzing #
14  * `-fsanitize-coverage=trace-pc-guard,indirect-calls,trace-cmp` - Clang >= 4.0
15  * `-fsanitize-coverage=bb` - Clang >= 3.7
16  * `-finstrument-functions` - GCC or Clang
17  * [older, slower variant] `-fsanitize-coverage=trace-pc,indirect-calls` - Clang >= 3.9
18
19_Note_: The _-fsanitize-coverage=trace-pc-guard,indirect-calls,trace-cmp_ set of flags will be automatically added to clang's command-line switches when using [hfuzz-clang](https://github.com/google/honggfuzz/tree/master/hfuzz_cc) binary.
20
21```
22$ [honggfuzz_dir]/honggfuzz/hfuzz_cc/hfuzz-clang terminal-test.c -o terminal-test
23```
24
25---
26# Requirements for hardware-based counter-based fuzzing #
27  * GNU/Linux OS
28  * Relatively modern Linux kernel (4.2 should suffice)
29  * CPU which is supported by the [perf subsystem](https://perf.wiki.kernel.org/index.php/Main_Page) for hardware-assisted instruction and branch counting
30
31---
32# Requirements for hardware-based coverage-feedback fuzzing (Intel) #
33  * CPU supporting [BTS (Branch Trace Store)](https://software.intel.com/en-us/forums/topic/277868?language=en) for hardware assisted unique pc and edges (branch pairs) counting. Currently it's available only in some newer Intel CPUs (unfortunately no AMD support for now)
34  * CPU supporting [Intel PT (Processor Tracing)](https://software.intel.com/en-us/blogs/2013/09/18/processor-tracing) for hardware assisted unique edge (branch pairs) counting. Currently it's available only in some newer Intel CPUs (since Broadwell architecture)
35  * GNU/Linux OS with a supported CPU; Intel Core 2 for BTS, Intel Broadwell for Intel PT
36  * Intel's [ibipt library](http://packages.ubuntu.com/yakkety/libipt1) for Intel PT
37  * Linux kernel >= v4.2 for perf AUXTRACE
38
39---
40# Fuzzing strategy #
41The implemented strategy is trying to identify files which add new code coverage (or increased instruction/branch counters). Then those inputs are added (dynamically stored in memory) corpus, and reused during following fuzzing rounds
42
43There are 2 phases of feedback-driven the fuzzing:
44  * Honggfuzz goes through each file in the initial corpus directory (-f). It adds files which hit new code coverage to the dynamic input corpus (as well as saving them on the disk, using *COVERAGE_DATA.PID.pid.RND.time.rnd* pattern
45  * Honggfuzz choses randomly files from the dynamic input corpus (in-memory), mutates them, and runs a new fuzzing round (round in persistent mode, exec in non-persistent mode). If the newly created file induces new code path (extends code coverage), it gets added to the dynamic input corpus
46
47# Compile-time instrumentation with clang/gcc (default mode) #
48
49Here you can use the following:
50  * gcc/clang `-finstrument-functions` (less-precise)
51  * clang's (>= 4.0) `-fsanitize-coverage=trace-pc-guard,indirect-calls,trace-cmp`
52    (trace-cmp adds additional comparison map to the instrumentation)
53
54_Note_: The _-fsanitize-coverage=trace-pc-guard,indirect-calls,trace-cmp_ set of flags will be automatically added to clang's command-line switches when using [hfuzz-clang](https://github.com/google/honggfuzz/tree/master/hfuzz_cc) binary. The [hfuzz-clang](https://github.com/google/honggfuzz/tree/master/hfuzz_cc) binary will also link your code with _libhfuzz.a_
55
56Two persistent modes can be used here:
57
58### LLVM-style LLVMFuzzerTestOneInput ###
59
60```c
61$ cat test.c
62#include <inttypes.h>
63#include <testlib.h>  // Our API to test
64
65extern int LLVMFuzzerTestOneInput(uint8_t **buf, size_t *len);
66
67int LLVMFuzzerTestOneInput(uint8_t *buf, size_t len) {
68  _FuncFromFuzzedLib_(buf, len);
69  return 0;
70}
71```
72
73```
74$ [honggfuzz_dir]/honggfuzz/hfuzz_cc/hfuzz-clang -c fuzzedlib.c -o fuzzedlib.o
75$ [honggfuzz_dir]/honggfuzz/hfuzz_cc/hfuzz-clang test.c fuzzedlib.o -o test
76$ [honggfuzz_dir]/honggfuzz -P -f INPUT.corpus -- ./test
77```
78
79`LLVMFuzzerInitialize(int *argc, char **argv)` is supported as well
80
81### Fetching input with HF_ITER() ###
82
83```c
84$ cat test.c
85#include <inttypes.h>
86#include <testlib.h>  // Our API to test
87
88// Get input from the fuzzer
89extern void HF_ITER(uint8_t **buf, size_t *len);
90
91int main(void) {
92  for (;;) {
93    uint8_t *buf;
94    size_t len;
95    HF_ITER(&buf, &len);
96    _FuncFromFuzzedLib_(buf, len);
97  }
98  return 0;
99}
100```
101```
102$ [honggfuzz_dir]/honggfuzz/hfuzz_cc/hfuzz-clang -c fuzzedlib.c -o fuzzedlib.o
103$ [honggfuzz_dir]/honggfuzz/hfuzz_cc/hfuzz-clang test.c fuzzedlib.o -o test
104$ [honggfuzz_dir]/honggfuzz -P -f INPUT.corpus -- ./test
105```
106
107Example:
108```
109$ [honggfuzz_dir]/honggfuzz -P -f IN.server/ -- ./persistent.server.openssl.1.0.2i.asan
110------------------------------[ honggfuzz v0.8 ]------------------------------
111      Iterations : 3,275,169 [3.28M]
112        Run Time : 2 hrs 17 min 16 sec (since: 2016-09-27 07:30:04)
113       Input Dir : 'IN.server/'
114      Fuzzed Cmd : './persistent.server.openssl.1.0.2i.asan'
115 Fuzzing Threads : 2, CPUs: 8, CPU: 759.0% (94.9%/CPU)
116   Speed (Round) : 86/sec (avg: 397)
117         Crashes : 0 (unique: 0, blacklist: 0, verified: 0)
118        Timeouts : 0 [10 sec.]
119     Corpus size : 393 (max file size: 40,000 bytes)
120        Coverage :
121       *** blocks seen:    3,545, comparison map: 204,542
122-----------------------------------[ LOGS ]-----------------------------------
123```
124
125PS. You can also use a non-persistent mode here (without the __-P__ flag), in which case you need to read data either from a file passed at command-line (`___FILE___`), or from the standard input (e.g. with `read(0, buf, sizeof(buf))`. The compile-time instrumentation will still work in such case.
126
127# Hardware-based coverage #
128## Unique branch pair (edges) counting (--linux_perf_bts_edge) ##
129
130This mode will take into consideration pairs (tuples) of jumps, recording unique from-to jump pairs. The data is taken from the Intel BTS CPU registers.
131
132```
133$ [honggfuzz_dir]/honggfuzz --linux_perf_bts_edge -f IN.corpus/ -- /usr/bin/xmllint -format ___FILE___
134============================== STAT ==============================
135Iterations: 1
136Start time: 2016-02-16 18:37:08 (1 seconds elapsed)
137Input file/dir: 'IN/'
138Fuzzed cmd: '/usr/bin/xmllint -format ___FILE___'
139Fuzzing threads: 2
140Execs per second: 1 (avg: 1)
141Crashes: 0 (unique: 0, blacklist: 0, verified: 0)
142Timeouts: 0
143Number of dynamic files: 251
144Coverage (max):
145  - BTS unique edges:   2341
146============================== LOGS ==============================
147[2016-02-16T18:37:09+0100][I][14944] fuzz_perfFeedback():420 New: (Size New,Old): 257,257, Perf (Cur,New): 0/0/0/0/0/0,0/0/0/2341/0/0
148```
149
150## Unique branch points counting (--linux_perf_ipt_block) ##
151
152This mode will utilize Interl's PT (Process Trace) subsystem, which should be way faster than BTS (Branch Trace Store), but will currently produce less precise results.
153
154```
155$ [honggfuzz_dir]/honggfuzz --linux_perf_ipt_block -f IN.corpus/ -- /usr/bin/xmllint -format ___FILE___
156============================== STAT ==============================
157Iterations: 0
158Start time: 2016-02-16 18:38:45 (0 seconds elapsed)
159Input file/dir: 'IN/'
160Fuzzed cmd: '/usr/bin/xmllint -format ___FILE___'
161Fuzzing threads: 2
162Execs per second: 0 (avg: 0)
163Crashes: 0 (unique: 0, blacklist: 0, verified: 0)
164Timeouts: 0
165Number of dynamic files: 251
166Coverage (max):
167  - PT unique blocks: 243
168============================== LOGS ==============================
169```
170
171## Instruction counting (--linux_perf_instr) ##
172
173This mode tries to maximize the number of instructions taken during each process iteration. The counters will be taken from the Linux perf subsystems. Intel, AMD and even other CPU architectures are supported for this mode.
174
175```
176$ [honggfuzz_dir]/honggfuzz --linux_perf_instr -f IN.corpus -- /usr/bin/xmllint -format ___FILE___
177============================== STAT ==============================
178Iterations: 2776
179Start time: 2016-02-16 18:40:51 (3 seconds elapsed)
180Input file/dir: 'CURRENT_BEST'
181Fuzzed cmd: '/usr/bin/xmllint -format ___FILE___'
182Fuzzing threads: 2
183Execs per second: 922 (avg: 925)
184Crashes: 0 (unique: 0, blacklist: 0, verified: 0)
185Timeouts: 0
186Number of dynamic files: 251
187Coverage (max):
188  - cpu instructions:      1369752
189============================== LOGS ==============================
190[2016-02-16T18:40:54+0100][I][17406] fuzz_perfFeedback():420 New: (Size New,Old): 2497,2496, Perf (Cur,New): 1369752/0/0/0/0/0,1371747/0/0/0/0/0
191[2016-02-16T18:40:54+0100][I][17406] fuzz_perfFeedback():420 New: (Size New,Old): 2497,2497, Perf (Cur,New): 1371747/0/0/0/0/0,1372273/0/0/0/0/0
192[2016-02-16T18:40:54+0100][I][17406] fuzz_perfFeedback():420 New: (Size New,Old): 2497,2497, Perf (Cur,New): 1372273/0/0/0/0/0,1372390/0/0/0/0/0
193[2016-02-16T18:40:54+0100][I][17406] fuzz_perfFeedback():420 New: (Size New,Old): 2497,2497, Perf (Cur,New): 1372390/0/0/0/0/0,1372793/0/0/0/0/0
194```
195
196## Branch counting (--linux_perf_branch) ##
197
198As above, it will try to maximize the number of branches taken by CPU on behalf of the fuzzed process (here: djpeg.static) while performing each fuzzing iteration. Intel, AMD and even other CPU architectures are supported for this mode.
199
200```
201$ [honggfuzz_dir]/honggfuzz --linux_perf_branch -f IN/ -F 2500 -- /usr/bin/xmllint -format ___FILE___
202============================== STAT ==============================
203Iterations: 0
204Start time: 2016-02-16 18:39:41 (0 seconds elapsed)
205Input file/dir: 'IN/'
206Fuzzed cmd: '/usr/bin/xmllint -format ___FILE___'
207Fuzzing threads: 2
208Execs per second: 0 (avg: 0)
209Crashes: 0 (unique: 0, blacklist: 0, verified: 0)
210Timeouts: 0
211Number of dynamic files: 251
212Coverage (max):
213  - cpu branches:          0
214============================== LOGS ==============================
215[2016-02-16T18:39:41+0100][I][16738] fuzz_perfFeedback():420 New: (Size New,Old): 2500,2500, Perf (Cur,New): 0/0/0/0/0/0,0/258259/0/0/0/0
216[2016-02-16T18:39:41+0100][I][16738] fuzz_perfFeedback():420 New: (Size New,Old): 2500,2500, Perf (Cur,New): 0/258259/0/0/0/0,0/258260/0/0/0/0
217[2016-02-16T18:39:41+0100][I][16738] fuzz_perfFeedback():420 New: (Size New,Old): 2500,2500, Perf (Cur,New): 0/258260/0/0/0/0,0/258261/0/0/0/0
218[2016-02-16T18:39:41+0100][I][16738] fuzz_perfFeedback():420 New: (Size New,Old): 2500,2500, Perf (Cur,New): 0/258261/0/0/0/0,0/258263/0/0/0/0
219```
220