1 ========================================================
2 LibFuzzer -- a library for coverage-guided fuzz testing.
3 ========================================================
11 This library is intended primarily for in-process coverage-guided fuzz testing
12 (fuzzing) of other libraries. The typical workflow looks like this:
14 * Build the Fuzzer library as a static archive (or just a set of .o files).
15 Note that the Fuzzer contains the main() function.
16 Preferably do *not* use sanitizers while building the Fuzzer.
17 * Build the library you are going to test with
18 `-fsanitize-coverage={bb,edge}[,indirect-calls,8bit-counters]`
19 and one of the sanitizers. We recommend to build the library in several
20 different modes (e.g. asan, msan, lsan, ubsan, etc) and even using different
21 optimizations options (e.g. -O0, -O1, -O2) to diversify testing.
22 * Build a test driver using the same options as the library.
23 The test driver is a C/C++ file containing interesting calls to the library
24 inside a single function ``extern "C" int LLVMFuzzerTestOneInput(const uint8_t *Data, size_t Size);``.
25 Currently, the only expected return value is 0, others are reserved for future.
26 * Link the Fuzzer, the library and the driver together into an executable
27 using the same sanitizer options as for the library.
28 * Collect the initial corpus of inputs for the
29 fuzzer (a directory with test inputs, one file per input).
30 The better your inputs are the faster you will find something interesting.
31 Also try to keep your inputs small, otherwise the Fuzzer will run too slow.
32 By default, the Fuzzer limits the size of every input to 64 bytes
33 (use ``-max_len=N`` to override).
34 * Run the fuzzer with the test corpus. As new interesting test cases are
35 discovered they will be added to the corpus. If a bug is discovered by
36 the sanitizer (asan, etc) it will be reported as usual and the reproducer
37 will be written to disk.
38 Each Fuzzer process is single-threaded (unless the library starts its own
39 threads). You can run the Fuzzer on the same corpus in multiple processes
43 The Fuzzer is similar in concept to AFL_,
44 but uses in-process Fuzzing, which is more fragile, more restrictive, but
45 potentially much faster as it has no overhead for process start-up.
46 It uses LLVM's SanitizerCoverage_ instrumentation to get in-process
49 The code resides in the LLVM repository, requires the fresh Clang compiler to build
50 and is used to fuzz various parts of LLVM,
51 but the Fuzzer itself does not (and should not) depend on any
52 part of LLVM and can be used for other projects w/o requiring the rest of LLVM.
56 The most important flags are::
58 seed 0 Random seed. If 0, seed is generated.
59 runs -1 Number of individual test runs (-1 for infinite runs).
60 max_len 64 Maximum length of the test input.
61 cross_over 1 If 1, cross over inputs.
62 mutate_depth 5 Apply this number of consecutive mutations to each input.
63 timeout 1200 Timeout in seconds (if positive). If one unit runs more than this number of seconds the process will abort.
64 max_total_time 0 If positive, indicates the maximal total time in seconds to run the fuzzer.
66 save_minimized_corpus 0 If 1, the minimized corpus is saved into the first input directory. Example: ./fuzzer -save_minimized_corpus=1 NEW_EMPTY_DIR OLD_CORPUS
67 merge 0 If 1, the 2-nd, 3-rd, etc corpora will be merged into the 1-st corpus. Only interesting units will be taken.
68 jobs 0 Number of jobs to run. If jobs >= 1 we spawn this number of jobs in separate worker processes with stdout/stderr redirected to fuzz-JOB.log.
69 workers 0 Number of simultaneous worker processes to run the jobs. If zero, "min(jobs,NumberOfCpuCores()/2)" is used.
70 sync_command 0 Execute an external command "<sync_command> <test_corpus>" to synchronize the test corpus.
71 sync_timeout 600 Minimum timeout between syncs.
72 use_traces 0 Experimental: use instruction traces
73 only_ascii 0 If 1, generate only ASCII (isprint+isspace) inputs.
74 test_single_input "" Use specified file content as test input. Test will be run only once. Useful for debugging a particular case.
75 artifact_prefix "" Write fuzzing artifacts (crash, timeout, or slow inputs) as $(artifact_prefix)file
76 exact_artifact_path "" Write the single artifact on failure (crash, timeout) as $(exact_artifact_path). This overrides -artifact_prefix and will not use checksum in the file name. Do not use the same path for several parallel processes.
78 For the full list of flags run the fuzzer binary with ``-help=1``.
86 A simple function that does something interesting if it receives the input "HI!"::
88 cat << EOF >> test_fuzzer.cc
89 extern "C" int LLVMFuzzerTestOneInput(const unsigned char *data, unsigned long size) {
90 if (size > 0 && data[0] == 'H')
91 if (size > 1 && data[1] == 'I')
92 if (size > 2 && data[2] == '!')
97 # Get lib/Fuzzer. Assuming that you already have fresh clang in PATH.
98 svn co http://llvm.org/svn/llvm-project/llvm/trunk/lib/Fuzzer
99 # Build lib/Fuzzer files.
100 clang -c -g -O2 -std=c++11 Fuzzer/*.cpp -IFuzzer
101 # Build test_fuzzer.cc with asan and link against lib/Fuzzer.
102 clang++ -fsanitize=address -fsanitize-coverage=edge test_fuzzer.cc Fuzzer*.o
103 # Run the fuzzer with no corpus.
106 You should get ``Illegal instruction (core dumped)`` pretty quickly.
111 Here we show how to use lib/Fuzzer on something real, yet simple: pcre2_::
113 COV_FLAGS=" -fsanitize-coverage=edge,indirect-calls,8bit-counters"
115 svn co svn://vcs.exim.org/pcre2/code/trunk pcre
116 # Get lib/Fuzzer. Assuming that you already have fresh clang in PATH.
117 svn co http://llvm.org/svn/llvm-project/llvm/trunk/lib/Fuzzer
118 # Build PCRE2 with AddressSanitizer and coverage.
119 (cd pcre; ./autogen.sh; CC="clang -fsanitize=address $COV_FLAGS" ./configure --prefix=`pwd`/../inst && make -j && make install)
120 # Build lib/Fuzzer files.
121 clang -c -g -O2 -std=c++11 Fuzzer/*.cpp -IFuzzer
122 # Build the actual function that does something interesting with PCRE2.
123 cat << EOF > pcre_fuzzer.cc
125 #include "pcre2posix.h"
126 extern "C" int LLVMFuzzerTestOneInput(const unsigned char *data, size_t size) {
127 if (size < 1) return 0;
128 char *str = new char[size+1];
129 memcpy(str, data, size);
132 if (0 == regcomp(&preg, str, 0)) {
133 regexec(&preg, str, 0, 0, 0);
140 clang++ -g -fsanitize=address $COV_FLAGS -c -std=c++11 -I inst/include/ pcre_fuzzer.cc
142 clang++ -g -fsanitize=address -Wl,--whole-archive inst/lib/*.a -Wl,-no-whole-archive Fuzzer*.o pcre_fuzzer.o -o pcre_fuzzer
144 This will give you a binary of the fuzzer, called ``pcre_fuzzer``.
145 Now, create a directory that will hold the test corpus::
149 For simple input languages like regular expressions this is all you need.
150 For more complicated inputs populate the directory with some input samples.
151 Now run the fuzzer with the corpus dir as the only parameter::
153 ./pcre_fuzzer ./CORPUS
155 You will see output like this::
158 #0 READ cov 0 bits 0 units 1 exec/s 0
159 #1 pulse cov 3 bits 0 units 1 exec/s 0
160 #1 INITED cov 3 bits 0 units 1 exec/s 0
161 #2 pulse cov 208 bits 0 units 1 exec/s 0
162 #2 NEW cov 208 bits 0 units 2 exec/s 0 L: 64
163 #3 NEW cov 217 bits 0 units 3 exec/s 0 L: 63
164 #4 pulse cov 217 bits 0 units 3 exec/s 0
166 * The ``Seed:`` line shows you the current random seed (you can change it with ``-seed=N`` flag).
167 * The ``READ`` line shows you how many input files were read (since you passed an empty dir there were inputs, but one dummy input was synthesised).
168 * The ``INITED`` line shows you that how many inputs will be fuzzed.
169 * The ``NEW`` lines appear with the fuzzer finds a new interesting input, which is saved to the CORPUS dir. If multiple corpus dirs are given, the first one is used.
170 * The ``pulse`` lines appear periodically to show the current status.
172 Now, interrupt the fuzzer and run it again the same way. You will see::
175 #0 READ cov 0 bits 0 units 564 exec/s 0
176 #1 pulse cov 502 bits 0 units 564 exec/s 0
178 #512 pulse cov 2933 bits 0 units 564 exec/s 512
179 #564 INITED cov 2991 bits 0 units 344 exec/s 564
180 #1024 pulse cov 2991 bits 0 units 344 exec/s 1024
181 #1455 NEW cov 2995 bits 0 units 345 exec/s 1455 L: 49
183 This time you were running the fuzzer with a non-empty input corpus (564 items).
184 As the first step, the fuzzer minimized the set to produce 344 interesting items (the ``INITED`` line)
186 It is quite convenient to store test corpuses in git.
187 As an example, here is a git repository with test inputs for the above PCRE2 fuzzer::
189 git clone https://github.com/kcc/fuzzing-with-sanitizers.git
190 ./pcre_fuzzer ./fuzzing-with-sanitizers/pcre2/C1/
192 You may run ``N`` independent fuzzer jobs in parallel on ``M`` CPUs::
194 N=100; M=4; ./pcre_fuzzer ./CORPUS -jobs=$N -workers=$M
196 By default (``-reload=1``) the fuzzer processes will periodically scan the CORPUS directory
197 and reload any new tests. This way the test inputs found by one process will be picked up
200 If ``-workers=$M`` is not supplied, ``min($N,NumberOfCpuCore/2)`` will be used.
204 Remember Heartbleed_?
205 As it was recently `shown <https://blog.hboeck.de/archives/868-How-Heartbleed-couldve-been-found.html>`_,
206 fuzzing with AddressSanitizer can find Heartbleed. Indeed, here are the step-by-step instructions
207 to find Heartbleed with LibFuzzer::
209 wget https://www.openssl.org/source/openssl-1.0.1f.tar.gz
210 tar xf openssl-1.0.1f.tar.gz
211 COV_FLAGS="-fsanitize-coverage=edge,indirect-calls" # -fsanitize-coverage=8bit-counters
212 (cd openssl-1.0.1f/ && ./config &&
213 make -j 32 CC="clang -g -fsanitize=address $COV_FLAGS")
214 # Get and build LibFuzzer
215 svn co http://llvm.org/svn/llvm-project/llvm/trunk/lib/Fuzzer
216 clang -c -g -O2 -std=c++11 Fuzzer/*.cpp -IFuzzer
217 # Get examples of key/pem files.
218 git clone https://github.com/hannob/selftls
219 cp selftls/server* . -v
220 cat << EOF > handshake-fuzz.cc
221 #include <openssl/ssl.h>
222 #include <openssl/err.h>
227 SSL_load_error_strings();
228 ERR_load_BIO_strings();
229 OpenSSL_add_all_algorithms();
230 assert (sctx = SSL_CTX_new(TLSv1_method()));
231 assert (SSL_CTX_use_certificate_file(sctx, "server.pem", SSL_FILETYPE_PEM));
232 assert (SSL_CTX_use_PrivateKey_file(sctx, "server.key", SSL_FILETYPE_PEM));
235 extern "C" int LLVMFuzzerTestOneInput(unsigned char *Data, size_t Size) {
236 static int unused = Init();
237 SSL *server = SSL_new(sctx);
238 BIO *sinbio = BIO_new(BIO_s_mem());
239 BIO *soutbio = BIO_new(BIO_s_mem());
240 SSL_set_bio(server, sinbio, soutbio);
241 SSL_set_accept_state(server);
242 BIO_write(sinbio, Data, Size);
243 SSL_do_handshake(server);
249 clang++ -g handshake-fuzz.cc -fsanitize=address \
250 openssl-1.0.1f/libssl.a openssl-1.0.1f/libcrypto.a Fuzzer*.o
251 # Run 20 independent fuzzer jobs.
252 ./a.out -jobs=20 -workers=20
256 #1048576 pulse cov 3424 bits 0 units 9 exec/s 24385
257 =================================================================
258 ==17488==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x629000004748 at pc 0x00000048c979 bp 0x7fffe3e864f0 sp 0x7fffe3e85ca8
259 READ of size 60731 at 0x629000004748 thread T0
260 #0 0x48c978 in __asan_memcpy
261 #1 0x4db504 in tls1_process_heartbeat openssl-1.0.1f/ssl/t1_lib.c:2586:3
262 #2 0x580be3 in ssl3_read_bytes openssl-1.0.1f/ssl/s3_pkt.c:1092:4
270 LibFuzzer supports user-supplied dictionaries with input language keywords
271 or other interesting byte sequences (e.g. multi-byte magic values).
272 Use ``-dict=DICTIONARY_FILE``. For some input languages using a dictionary
273 may significantly improve the search speed.
274 The dictionary syntax is similar to that used by AFL_ for its ``-x`` option::
276 # Lines starting with '#' and empty lines are ignored.
278 # Adds "blah" (w/o quotes) to the dictionary.
280 # Use \\ for backslash and \" for quotes.
282 # Use \xAB for hex values
284 # the name of the keyword followed by '=' may be omitted:
287 Data-flow-guided fuzzing
288 ------------------------
291 With an additional compiler flag ``-fsanitize-coverage=trace-cmp`` (see SanitizerCoverageTraceDataFlow_)
292 and extra run-time flag ``-use_traces=1`` the fuzzer will try to apply *data-flow-guided fuzzing*.
293 That is, the fuzzer will record the inputs to comparison instructions, switch statements,
294 and several libc functions (``memcmp``, ``strcmp``, ``strncmp``, etc).
295 It will later use those recorded inputs during mutations.
297 This mode can be combined with DataFlowSanitizer_ to achieve better sensitivity.
301 LibFuzzer can be used in parallel with AFL_ on the same test corpus.
302 Both fuzzers expect the test corpus to reside in a directory, one file per input.
303 You can run both fuzzers on the same corpus in parallel::
305 ./afl-fuzz -i testcase_dir -o findings_dir /path/to/program -r @@
306 ./llvm-fuzz testcase_dir findings_dir # Will write new tests to testcase_dir
308 Periodically restart both fuzzers so that they can use each other's findings.
310 How good is my fuzzer?
311 ----------------------
313 Once you implement your target function ``LLVMFuzzerTestOneInput`` and fuzz it to death,
314 you will want to know whether the function or the corpus can be improved further.
315 One easy to use metric is, of course, code coverage.
316 You can get the coverage for your corpus like this::
318 ASAN_OPTIONS=coverage_pcs=1 ./fuzzer CORPUS_DIR -runs=0
320 This will run all the tests in the CORPUS_DIR but will not generate any new tests
321 and dump covered PCs to disk before exiting.
322 Then you can subtract the set of covered PCs from the set of all instrumented PCs in the binary,
323 see SanitizerCoverage_ for details.
325 User-supplied mutators
326 ----------------------
328 LibFuzzer allows to use custom (user-supplied) mutators,
329 see FuzzerInterface.h_
331 Fuzzing components of LLVM
332 ==========================
336 The inputs are random pieces of C++-like text.
338 Build (make sure to use fresh clang as the host compiler)::
340 cmake -GNinja -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DLLVM_USE_SANITIZER=Address -DLLVM_USE_SANITIZE_COVERAGE=YES -DCMAKE_BUILD_TYPE=Release /path/to/llvm
341 ninja clang-format-fuzzer
343 ./bin/clang-format-fuzzer CORPUS_DIR
345 Optionally build other kinds of binaries (asan+Debug, msan, ubsan, etc).
347 Tracking bug: https://llvm.org/bugs/show_bug.cgi?id=23052
352 The behavior is very similar to ``clang-format-fuzzer``.
354 Tracking bug: https://llvm.org/bugs/show_bug.cgi?id=23057
359 Tracking bug: https://llvm.org/bugs/show_bug.cgi?id=24639
364 This tool fuzzes the MC layer. Currently it is only able to fuzz the
365 disassembler but it is hoped that assembly, and round-trip verification will be
368 When run in dissassembly mode, the inputs are opcodes to be disassembled. The
369 fuzzer will consume as many instructions as possible and will stop when it
370 finds an invalid instruction or runs out of data.
372 Please note that the command line interface differs slightly from that of other
373 fuzzers. The fuzzer arguments should follow ``--fuzzer-args`` and should have
374 a single dash, while other arguments control the operation mode and target in a
375 similar manner to ``llvm-mc`` and should have two dashes. For example::
377 llvm-mc-fuzzer --triple=aarch64-linux-gnu --disassemble --fuzzer-args -max_len=4 -jobs=10
382 We have a buildbot that runs the above fuzzers for LLVM components
383 24/7/365 at http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fuzzer .
385 Pre-fuzzed test inputs in git
386 -----------------------------
388 The buildbot occumulates large test corpuses over time.
389 The corpuses are stored in git on github and can be used like this::
391 git clone https://github.com/kcc/fuzzing-with-sanitizers.git
392 bin/clang-format-fuzzer fuzzing-with-sanitizers/llvm/clang-format/C1
393 bin/clang-fuzzer fuzzing-with-sanitizers/llvm/clang/C1/
394 bin/llvm-as-fuzzer fuzzing-with-sanitizers/llvm/llvm-as/C1 -only_ascii=1
398 =========================
400 Q. Why Fuzzer does not use any of the LLVM support?
401 ---------------------------------------------------
403 There are two reasons.
405 First, we want this library to be used outside of the LLVM w/o users having to
406 build the rest of LLVM. This may sound unconvincing for many LLVM folks,
407 but in practice the need for building the whole LLVM frightens many potential
408 users -- and we want more users to use this code.
410 Second, there is a subtle technical reason not to rely on the rest of LLVM, or
411 any other large body of code (maybe not even STL). When coverage instrumentation
412 is enabled, it will also instrument the LLVM support code which will blow up the
413 coverage set of the process (since the fuzzer is in-process). In other words, by
414 using more external dependencies we will slow down the fuzzer while the main
415 reason for it to exist is extreme speed.
417 Q. What about Windows then? The Fuzzer contains code that does not build on Windows.
418 ------------------------------------------------------------------------------------
420 The sanitizer coverage support does not work on Windows either as of 01/2015.
421 Once it's there, we'll need to re-implement OS-specific parts (I/O, signals).
423 Q. When this Fuzzer is not a good solution for a problem?
424 ---------------------------------------------------------
426 * If the test inputs are validated by the target library and the validator
427 asserts/crashes on invalid inputs, the in-process fuzzer is not applicable
428 (we could use fork() w/o exec, but it comes with extra overhead).
429 * Bugs in the target library may accumulate w/o being detected. E.g. a memory
430 corruption that goes undetected at first and then leads to a crash while
431 testing another input. This is why it is highly recommended to run this
432 in-process fuzzer with all sanitizers to detect most bugs on the spot.
433 * It is harder to protect the in-process fuzzer from excessive memory
434 consumption and infinite loops in the target library (still possible).
435 * The target library should not have significant global state that is not
436 reset between the runs.
437 * Many interesting target libs are not designed in a way that supports
438 the in-process fuzzer interface (e.g. require a file path instead of a
440 * If a single test run takes a considerable fraction of a second (or
441 more) the speed benefit from the in-process fuzzer is negligible.
442 * If the target library runs persistent threads (that outlive
443 execution of one test) the fuzzing results will be unreliable.
445 Q. So, what exactly this Fuzzer is good for?
446 --------------------------------------------
448 This Fuzzer might be a good choice for testing libraries that have relatively
449 small inputs, each input takes < 1ms to run, and the library code is not expected
450 to crash on invalid inputs.
451 Examples: regular expression matchers, text or binary format parsers.
455 * GLIBC: https://sourceware.org/glibc/wiki/FuzzingLibc
459 * http://git.musl-libc.org/cgit/musl/commit/?id=39dfd58417ef642307d90306e1c7e50aaec5a35c
460 * http://www.openwall.com/lists/oss-security/2015/03/30/3
462 * `pugixml <https://github.com/zeux/pugixml/issues/39>`_
464 * PCRE: Search for "LLVM fuzzer" in http://vcs.pcre.org/pcre2/code/trunk/ChangeLog?view=markup;
465 also in `bugzilla <https://bugs.exim.org/buglist.cgi?bug_status=__all__&content=libfuzzer&no_redirect=1&order=Importance&product=PCRE&query_format=specific>`_
467 * `ICU <http://bugs.icu-project.org/trac/ticket/11838>`_
469 * `Freetype <https://savannah.nongnu.org/search/?words=LibFuzzer&type_of_search=bugs&Search=Search&exact=1#options>`_
471 * `Harfbuzz <https://github.com/behdad/harfbuzz/issues/139>`_
473 * `SQLite <http://www3.sqlite.org/cgi/src/info/088009efdd56160b>`_
475 * `Python <http://bugs.python.org/issue25388>`_
478 <https://bugzilla.gnome.org/buglist.cgi?bug_status=__all__&content=libFuzzer&list_id=68957&order=Importance&product=libxml2&query_format=specific>`_
480 * `Linux Kernel's BPF verifier <https://github.com/iovisor/bpf-fuzzer>`_
482 * LLVM: `Clang <https://llvm.org/bugs/show_bug.cgi?id=23057>`_, `Clang-format <https://llvm.org/bugs/show_bug.cgi?id=23052>`_, `libc++ <https://llvm.org/bugs/show_bug.cgi?id=24411>`_, `llvm-as <https://llvm.org/bugs/show_bug.cgi?id=24639>`_, Disassembler: http://reviews.llvm.org/rL247405, http://reviews.llvm.org/rL247414, http://reviews.llvm.org/rL247416, http://reviews.llvm.org/rL247417, http://reviews.llvm.org/rL247420, http://reviews.llvm.org/rL247422.
484 .. _pcre2: http://www.pcre.org/
486 .. _AFL: http://lcamtuf.coredump.cx/afl/
488 .. _SanitizerCoverage: http://clang.llvm.org/docs/SanitizerCoverage.html
489 .. _SanitizerCoverageTraceDataFlow: http://clang.llvm.org/docs/SanitizerCoverage.html#tracing-data-flow
490 .. _DataFlowSanitizer: http://clang.llvm.org/docs/DataFlowSanitizer.html
492 .. _Heartbleed: http://en.wikipedia.org/wiki/Heartbleed
494 .. _FuzzerInterface.h: https://github.com/llvm-mirror/llvm/blob/master/lib/Fuzzer/FuzzerInterface.h