To quickly get started fuzzing Bitcoin Core using libFuzzer:
$ git clone https://github.com/bitcoin/bitcoin
$ cd bitcoin/
$ cmake --preset=libfuzzer
# macOS users: If you have problem with this step then make sure to read "macOS hints for
# libFuzzer" on https://github.com/bitcoin/bitcoin/blob/master/doc/fuzzing.md#macos-hints-for-libfuzzer
$ cmake --build build_fuzz
$ FUZZ=process_message build_fuzz/src/test/fuzz/fuzz
# abort fuzzing using ctrl-cOne can use --prefix=libfuzzer-nosan to do the same without common sanitizers enabled.
See further for more information.
There is also a runner script to execute all fuzz targets. Refer to
./test/fuzz/test_runner.py --help for more details.
Google has a good overview of fuzzing in general, with contributions from key architects of some of the most-used fuzzers. This paper includes an external overview of the status of Bitcoin Core fuzzing, as of summer 2021. John Regehr provides good advice on writing code that assists fuzzers in finding bugs, which is useful for developers to keep in mind.
process_message is a fuzzing harness for the ProcessMessage(...) function (net_processing). The available fuzzing harnesses are found in src/test/fuzz/.
The fuzzer will output NEW every time it has created a test input that covers new areas of the code under test. For more information on how to interpret the fuzzer output, see the libFuzzer documentation.
If you specify a corpus directory then any new coverage increasing inputs will be saved there:
$ mkdir -p process_message-seeded-from-thin-air/
$ FUZZ=process_message build_fuzz/src/test/fuzz/fuzz process_message-seeded-from-thin-air/
INFO: Seed: 840522292
INFO: Loaded 1 modules (424174 inline 8-bit counters): 424174 [0x55e121ef9ab8, 0x55e121f613a6),
INFO: Loaded 1 PC tables (424174 PCs): 424174 [0x55e121f613a8,0x55e1225da288),
INFO: 0 files found in process_message-seeded-from-thin-air/
INFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes
INFO: A corpus is not provided, starting from an empty corpus
#2 INITED cov: 94 ft: 95 corp: 1/1b exec/s: 0 rss: 150Mb
#3 NEW cov: 95 ft: 96 corp: 2/3b lim: 4 exec/s: 0 rss: 150Mb L: 2/2 MS: 1 InsertByte-
#4 NEW cov: 96 ft: 98 corp: 3/7b lim: 4 exec/s: 0 rss: 150Mb L: 4/4 MS: 1 CrossOver-
#21 NEW cov: 96 ft: 100 corp: 4/11b lim: 4 exec/s: 0 rss: 150Mb L: 4/4 MS: 2 ChangeBit-CrossOver-
#324 NEW cov: 101 ft: 105 corp: 5/12b lim: 6 exec/s: 0 rss: 150Mb L: 6/6 MS: 5 CrossOver-ChangeBit-CopyPart-ChangeBit-ChangeBinInt-
#1239 REDUCE cov: 102 ft: 106 corp: 6/24b lim: 14 exec/s: 0 rss: 150Mb L: 13/13 MS: 5 ChangeBit-CrossOver-EraseBytes-ChangeBit-InsertRepeatedBytes-
#1272 REDUCE cov: 102 ft: 106 corp: 6/23b lim: 14 exec/s: 0 rss: 150Mb L: 12/12 MS: 3 ChangeBinInt-ChangeBit-EraseBytes-
NEW_FUNC[1/677]: 0x55e11f456690 in std::_Function_base::~_Function_base() /usr/lib/gcc/x86_64-linux-gnu/8/../../../../include/c++/8/bits/std_function.h:255
NEW_FUNC[2/677]: 0x55e11f465800 in CDataStream::CDataStream(std::vector<unsigned char, std::allocator<unsigned char> > const&, int, int) src/./streams.h:248
#2125 REDUCE cov: 4820 ft: 4867 corp: 7/29b lim: 21 exec/s: 0 rss: 155Mb L: 6/12 MS: 2 CopyPart-CMP- DE: "block"-
NEW_FUNC[1/9]: 0x55e11f64d790 in std::_Rb_tree<uint256, std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > >, std::_Select1st<std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > > >, std::less<uint256>, std::allocator<std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > > > >::~_Rb_tree() /usr/lib/gcc/x86_64-linux-gnu/8/../../../../include/c++/8/bits/stl_tree.h:972
NEW_FUNC[2/9]: 0x55e11f64d870 in std::_Rb_tree<uint256, std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > >, std::_Select1st<std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > > >, std::less<uint256>, std::allocator<std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > > > >::_M_erase(std::_Rb_tree_node<std::pair<uint256 const, std::chrono::duration<long, std::ratio<1l, 1000000l> > > >*) /usr/lib/gcc/x86_64-linux-gnu/8/../../../../include/c++/8/bits/stl_tree.h:1875
#2228 NEW cov: 4898 ft: 4971 corp: 8/35b lim: 21 exec/s: 0 rss: 156Mb L: 6/12 MS: 3 EraseBytes-CopyPart-PersAutoDict- DE: "block"-
NEW_FUNC[1/5]: 0x55e11f46df70 in std::enable_if<__and_<std::allocator_traits<zero_after_free_allocator<char> >::__construct_helper<char, unsigned char const&>::type>::value, void>::type std::allocator_traits<zero_after_free_allocator<char> >::_S_construct<char, unsigned char const&>(zero_after_free_allocator<char>&, char*, unsigned char const&) /usr/lib/gcc/x86_64-linux-gnu/8/../../../../include/c++/8/bits/alloc_traits.h:243
NEW_FUNC[2/5]: 0x55e11f477390 in std::vector<unsigned char, std::allocator<unsigned char> >::data() /usr/lib/gcc/x86_64-linux-gnu/8/../../../../include/c++/8/bits/stl_vector.h:1056
#2456 NEW cov: 4933 ft: 5042 corp: 9/55b lim: 21 exec/s: 0 rss: 160Mb L: 20/20 MS: 3 ChangeByte-InsertRepeatedBytes-PersAutoDict- DE: "block"-
#2467 NEW cov: 4933 ft: 5043 corp: 10/76b lim: 21 exec/s: 0 rss: 161Mb L: 21/21 MS: 1 InsertByte-
#4215 NEW cov: 4941 ft: 5129 corp: 17/205b lim: 29 exec/s: 4215 rss: 350Mb L: 29/29 MS: 5 InsertByte-ChangeBit-CopyPart-InsertRepeatedBytes-CrossOver-
#4567 REDUCE cov: 4941 ft: 5129 corp: 17/204b lim: 29 exec/s: 4567 rss: 404Mb L: 24/29 MS: 2 ChangeByte-EraseBytes-
#6642 NEW cov: 4941 ft: 5138 corp: 18/244b lim: 43 exec/s: 2214 rss: 450Mb L: 43/43 MS: 3 CopyPart-CMP-CrossOver- DE: "verack"-
# abort fuzzing using ctrl-c
$ ls process_message-seeded-from-thin-air/
349ac589fc66a09abc0b72bb4ae445a7a19e2cd8 4df479f1f421f2ea64b383cd4919a272604087a7
a640312c98dcc55d6744730c33e41c5168c55f09 b135de16e4709558c0797c15f86046d31c5d86d7
c000f7b41b05139de8b63f4cbf7d1ad4c6e2aa7f fc52cc00ec1eb1c08470e69f809ae4993fa70082
$ cat --show-nonprinting process_message-seeded-from-thin-air/349ac589fc66a09abc0b72bb4ae445a7a19e2cd8
block^@M-^?M-^?M-^?M-^?M-^?nM-^?M-^?In this case the fuzzer managed to create a block message which when passed to ProcessMessage(...) increased coverage.
It is possible to specify bitcoind arguments to the fuzz executable.
Depending on the test, they may be ignored or consumed and alter the behavior
of the test. Just make sure to use double-dash to distinguish them from the
fuzzer's own arguments:
$ FUZZ=address_deserialize_v2 build_fuzz/src/test/fuzz/fuzz -runs=1 fuzz_corpora/address_deserialize_v2 --checkaddrman=5 --printtoconsole=1The project's collection of seed corpora is found in the bitcoin-core/qa-assets repo.
To fuzz process_message using the bitcoin-core/qa-assets seed corpus:
$ git clone https://github.com/bitcoin-core/qa-assets
$ FUZZ=process_message build_fuzz/src/test/fuzz/fuzz qa-assets/fuzz_corpora/process_message/
INFO: Seed: 1346407872
INFO: Loaded 1 modules (424174 inline 8-bit counters): 424174 [0x55d8a9004ab8, 0x55d8a906c3a6),
INFO: Loaded 1 PC tables (424174 PCs): 424174 [0x55d8a906c3a8,0x55d8a96e5288),
INFO: 991 files found in qa-assets/fuzz_corpora/process_message/
INFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes
INFO: seed corpus: files: 991 min: 1b max: 1858b total: 288291b rss: 150Mb
#993 INITED cov: 7063 ft: 8236 corp: 25/3821b exec/s: 0 rss: 181Mb
…Fuzzing on a harness compiled with -DSANITIZERS=address,fuzzer,undefined is
good for finding bugs. However, the very slow execution even under libFuzzer
will limit the ability to find new coverage. A good approach is to perform
occasional long runs without the additional bug-detectors
(--preset=libfuzzer-nosan) and then merge new inputs into a corpus as described in
the qa-assets repo
(https://github.com/bitcoin-core/qa-assets/blob/main/.github/PULL_REQUEST_TEMPLATE.md).
Patience is useful; even with improved throughput, libFuzzer may need days and
10s of millions of executions to reach deep/hard targets.
cdinto theqa-assetsdirectory and update it withgit pull qa-assets- locate the crash case described in the CI output, e.g.
Test unit written to ./crash-1bc91feec9fc00b107d97dc225a9f2cdaa078eb6 - make sure to compile with all sanitizers, if they are needed (fuzzing runs more slowly with sanitizers enabled, but a crash should be reproducible very quickly from a crash case)
- run the fuzzer with the case number appended to the seed corpus path:
FUZZ=process_message build_fuzz/src/test/fuzz/fuzz qa-assets/fuzz_corpora/process_message/1bc91feec9fc00b107d97dc225a9f2cdaa078eb6
If you find coverage increasing inputs when fuzzing you are highly encouraged to submit them for inclusion in the bitcoin-core/qa-assets repo.
Every single pull request submitted against the Bitcoin Core repo is automatically tested against all inputs in the bitcoin-core/qa-assets repo. Contributing new coverage increasing inputs is an easy way to help make Bitcoin Core more robust.
The default Clang/LLVM version supplied by Apple on macOS does not include
fuzzing libraries, so macOS users will need to install a full version, for
example using brew install llvm.
You may also need to take care of giving the correct path for clang and
clang++, like CC=/path/to/clang CXX=/path/to/clang++ if the non-systems
clang does not come first in your path.
Full configuration step that was tested on macOS with brew installed llvm:
$ cmake --preset=libfuzzer \
-DCMAKE_C_COMPILER="$(brew --prefix llvm)/bin/clang" \
-DCMAKE_CXX_COMPILER="$(brew --prefix llvm)/bin/clang++" \
-DAPPEND_LDFLAGS=-Wl,-no_warn_duplicate_librariesRead the libFuzzer documentation for more information. This libFuzzer tutorial might also be of interest.
To quickly get started fuzzing Bitcoin Core using afl++:
$ git clone https://github.com/bitcoin/bitcoin
$ cd bitcoin/
$ git clone https://github.com/AFLplusplus/AFLplusplus
$ make -C AFLplusplus/ source-only
# If afl-clang-lto is not available, see
# https://github.com/AFLplusplus/AFLplusplus#a-selecting-the-best-afl-compiler-for-instrumenting-the-target
$ cmake -B build_fuzz \
-DCMAKE_C_COMPILER="$(pwd)/AFLplusplus/afl-clang-lto" \
-DCMAKE_CXX_COMPILER="$(pwd)/AFLplusplus/afl-clang-lto++" \
-DBUILD_FOR_FUZZING=ON
$ cmake --build build_fuzz
# For macOS you may need to ignore x86 compilation checks when running "cmake --build". If so,
# try compiling using: AFL_NO_X86=1 cmake --build build_fuzz
$ mkdir -p inputs/ outputs/
$ echo A > inputs/thin-air-input
$ FUZZ=bech32 ./AFLplusplus/afl-fuzz -i inputs/ -o outputs/ -- build_fuzz/src/test/fuzz/fuzz
# You may have to change a few kernel parameters to test optimally - afl-fuzz
# will print an error and suggestion if so.Read the afl++ documentation for more information.
To quickly get started fuzzing Bitcoin Core using Honggfuzz:
$ git clone https://github.com/bitcoin/bitcoin
$ cd bitcoin/
$ git clone https://github.com/google/honggfuzz
$ cd honggfuzz/
$ make
$ cd ..
$ cmake -B build_fuzz \
-DCMAKE_C_COMPILER="$(pwd)/honggfuzz/hfuzz_cc/hfuzz-clang" \
-DCMAKE_CXX_COMPILER="$(pwd)/honggfuzz/hfuzz_cc/hfuzz-clang++" \
-DBUILD_FOR_FUZZING=ON \
-DSANITIZERS=address,undefined
$ cmake --build build_fuzz
$ mkdir -p inputs/
$ FUZZ=process_message ./honggfuzz/honggfuzz -i inputs/ -- build_fuzz/src/test/fuzz/fuzzRead the Honggfuzz documentation for more information.
Honggfuzz NetDriver allows for very easy fuzzing of TCP servers such as Bitcoin
Core without having to write any custom fuzzing harness. The bitcoind server
process is largely fuzzed without modification.
This makes the fuzzing highly realistic: a bug reachable by the fuzzer is likely also remotely triggerable by an untrusted peer.
To quickly get started fuzzing the P2P layer using Honggfuzz NetDriver:
$ mkdir bitcoin-honggfuzz-p2p/
$ cd bitcoin-honggfuzz-p2p/
$ git clone https://github.com/bitcoin/bitcoin
$ cd bitcoin/
$ git clone https://github.com/google/honggfuzz
$ cd honggfuzz/
$ make
$ cd ..
$ git apply << "EOF"
diff --git a/src/compat/compat.h b/src/compat/compat.h
index 8195bceaec..cce2b31ff0 100644
--- a/src/compat/compat.h
+++ b/src/compat/compat.h
@@ -90,8 +90,12 @@ typedef char* sockopt_arg_type;
// building with a binutils < 2.36 is subject to this ld bug.
#define MAIN_FUNCTION __declspec(dllexport) int main(int argc, char* argv[])
#else
+#ifdef HFND_FUZZING_ENTRY_FUNCTION_CXX
+#define MAIN_FUNCTION HFND_FUZZING_ENTRY_FUNCTION_CXX(int argc, char* argv[])
+#else
#define MAIN_FUNCTION int main(int argc, char* argv[])
#endif
+#endif
// Note these both should work with the current usage of poll, but best to be safe
// WIN32 poll is broken https://daniel.haxx.se/blog/2012/10/10/wsapoll-is-broken/
diff --git a/src/net.cpp b/src/net.cpp
index 7601a6ea84..702d0f56ce 100644
--- a/src/net.cpp
+++ b/src/net.cpp
@@ -727,7 +727,7 @@ int V1TransportDeserializer::readHeader(Span<const uint8_t> msg_bytes)
}
// Check start string, network magic
- if (memcmp(hdr.pchMessageStart, m_chain_params.MessageStart(), CMessageHeader::MESSAGE_START_SIZE) != 0) {
+ if (false && memcmp(hdr.pchMessageStart, m_chain_params.MessageStart(), CMessageHeader::MESSAGE_START_SIZE) != 0) { // skip network magic checking
LogDebug(BCLog::NET, "Header error: Wrong MessageStart %s received, peer=%d\n", HexStr(hdr.pchMessageStart), m_node_id);
return -1;
}
@@ -788,7 +788,7 @@ CNetMessage V1TransportDeserializer::GetMessage(const std::chrono::microseconds
RandAddEvent(ReadLE32(hash.begin()));
// Check checksum and header message type string
- if (memcmp(hash.begin(), hdr.pchChecksum, CMessageHeader::CHECKSUM_SIZE) != 0) {
+ if (false && memcmp(hash.begin(), hdr.pchChecksum, CMessageHeader::CHECKSUM_SIZE) != 0) { // skip checksum checking
LogDebug(BCLog::NET, "Header error: Wrong checksum (%s, %u bytes), expected %s was %s, peer=%d\n",
SanitizeString(msg.m_type), msg.m_message_size,
HexStr(Span{hash}.first(CMessageHeader::CHECKSUM_SIZE)),
EOF
$ cmake -B build_fuzz \
-DCMAKE_C_COMPILER="$(pwd)/honggfuzz/hfuzz_cc/hfuzz-clang" \
-DCMAKE_CXX_COMPILER="$(pwd)/honggfuzz/hfuzz_cc/hfuzz-clang++" \
-DENABLE_WALLET=OFF \
-DBUILD_GUI=OFF \
-DSANITIZERS=address,undefined
$ cmake --build build_fuzz --target bitcoind
$ mkdir -p inputs/
$ ./honggfuzz/honggfuzz --exit_upon_crash --quiet --timeout 4 -n 1 -Q \
-E HFND_TCP_PORT=18444 -f inputs/ -- \
build_fuzz/src/bitcoind -regtest -discover=0 -dns=0 -dnsseed=0 -listenonion=0 \
-nodebuglogfile -bind=127.0.0.1:18444 -logthreadnames \
-debugBitcoin Core participates in Google's OSS-Fuzz program, which includes a dashboard of publicly disclosed vulnerabilities.
Bitcoin Core follows its security disclosure policy, which may differ from Google's standard 90-day disclosure window .
OSS-Fuzz also produces a fuzzing coverage report.