Binaryonly Fuzzing

Fuzzing binary-only programs with AFL++

AFL++, libfuzzer and others are great if you have the source code, and it allows for very fast and coverage guided fuzzing.

However, if there is only the binary program and no source code available, then standard afl-fuzz -n (non-instrumented mode) is not effective.

The following is a description of how these binaries can be fuzzed with AFL++.

TL;DR:

qemu_mode in persistent mode is the fastest - if the stability is high enough. Otherwise try retrowrite, afl-dyninst and if these fail too then try standard qemu_mode with AFL_ENTRYPOINT to where you need it.

If your target is a library use utils/afl_frida/.

If your target is non-linux then use unicorn_mode/.

QEMU

Qemu is the “native” solution to the program. It is available in the ./qemu_mode/ directory and once compiled it can be accessed by the afl-fuzz -Q command line option. It is the easiest to use alternative and even works for cross-platform binaries.

The speed decrease is at about 50%. However various options exist to increase the speed:

  • using AFL_ENTRYPOINT to move the forkserver entry to a later basic block in the binary (+5-10% speed)
  • using persistent mode qemu_mode/README.persistent.md this will result in 150-300% overall speed increase - so 3-8x the original qemu_mode speed!
  • using AFL_CODE_START/AFL_CODE_END to only instrument specific parts

Note that there is also honggfuzz: https://github.com/google/honggfuzz which now has a qemu_mode, but its performance is just 1.5% …

As it is included in AFL++ this needs no URL.

If you like to code a customized fuzzer without much work, we highly recommend to check out our sister project libafl which will support QEMU too: https://github.com/AFLplusplus/LibAFL

AFL FRIDA

In frida_mode you can fuzz binary-only targets easily like with QEMU, with the advantage that frida_mode also works on MacOS (both intel and M1).

If you want to fuzz a binary-only library then you can fuzz it with frida-gum via utils/afl_frida/, you will have to write a harness to call the target function in the library, use afl-frida.c as a template.

Both come with AFL++ so this needs no URL.

You can also perform remote fuzzing with frida, e.g. if you want to fuzz on iPhone or Android devices, for this you can use https://github.com/ttdennis/fpicker/ as an intermediate that uses AFL++ for fuzzing.

If you like to code a customized fuzzer without much work, we highly recommend to check out our sister project libafl which supports Frida too: https://github.com/AFLplusplus/LibAFL Working examples already exist :-)

WINE+QEMU

Wine mode can run Win32 PE binaries with the QEMU instrumentation. It needs Wine, python3 and the pefile python package installed.

As it is included in AFL++ this needs no URL.

UNICORN

Unicorn is a fork of QEMU. The instrumentation is, therefore, very similar. In contrast to QEMU, Unicorn does not offer a full system or even userland emulation. Runtime environment and/or loaders have to be written from scratch, if needed. On top, block chaining has been removed. This means the speed boost introduced in the patched QEMU Mode of AFL++ cannot simply be ported over to Unicorn. For further information, check out unicorn_mode/README.md.

As it is included in AFL++ this needs no URL.

AFL UNTRACER

If you want to fuzz a binary-only shared library then you can fuzz it with utils/afl_untracer/, use afl-untracer.c as a template. It is slower than AFL FRIDA (see above).

DYNINST

Dyninst is a binary instrumentation framework similar to Pintool and Dynamorio (see far below). However whereas Pintool and Dynamorio work at runtime, dyninst instruments the target at load time, and then let it run - or save the binary with the changes. This is great for some things, e.g. fuzzing, and not so effective for others, e.g. malware analysis.

So what we can do with dyninst is taking every basic block, and put afl’s instrumention code in there - and then save the binary. Afterwards we can just fuzz the newly saved target binary with afl-fuzz. Sounds great? It is. The issue though - it is a non-trivial problem to insert instructions, which change addresses in the process space, so that everything is still working afterwards. Hence more often than not binaries crash when they are run.

The speed decrease is about 15-35%, depending on the optimization options used with afl-dyninst.

So if Dyninst works, it is the best option available. Otherwise it just doesn’t work well.

https://github.com/vanhauser-thc/afl-dyninst

RETROWRITE, ZAFL, … other binary rewriter

If you have an x86/x86_64 binary that still has its symbols, is compiled with position independant code (PIC/PIE) and does not use most of the C++ features then the retrowrite solution might be for you. It decompiles to ASM files which can then be instrumented with afl-gcc.

It is at about 80-85% performance.

https://git.zephyr-software.com/opensrc/zafl https://github.com/HexHive/retrowrite

MCSEMA

Theoretically you can also decompile to llvm IR with mcsema, and then use llvm_mode to instrument the binary. Good luck with that.

https://github.com/lifting-bits/mcsema

INTEL-PT

If you have a newer Intel CPU, you can make use of Intels processor trace. The big issue with Intel’s PT is the small buffer size and the complex encoding of the debug information collected through PT. This makes the decoding very CPU intensive and hence slow. As a result, the overall speed decrease is about 70-90% (depending on the implementation and other factors).

There are two AFL intel-pt implementations:

  1. https://github.com/junxzm1990/afl-pt => this needs Ubuntu 14.04.05 without any updates and the 4.4 kernel.

  2. https://github.com/hunter-ht-2018/ptfuzzer => this needs a 4.14 or 4.15 kernel. the “nopti” kernel boot option must be used. This one is faster than the other.

Note that there is also honggfuzz: https://github.com/google/honggfuzz But its IPT performance is just 6%!

CORESIGHT

Coresight is ARM’s answer to Intel’s PT. There is no implementation so far which handles coresight and getting it working on an ARM Linux is very difficult due to custom kernel building on embedded systems is difficult. And finding one that has coresight in the ARM chip is difficult too. My guess is that it is slower than Qemu, but faster than Intel PT.

If anyone finds any coresight implementation for AFL please ping me: vh@thc.org

PIN & DYNAMORIO

Pintool and Dynamorio are dynamic instrumentation engines, and they can be used for getting basic block information at runtime. Pintool is only available for Intel x32/x64 on Linux, Mac OS and Windows, whereas Dynamorio is additionally available for ARM and AARCH64. Dynamorio is also 10x faster than Pintool.

The big issue with Dynamorio (and therefore Pintool too) is speed. Dynamorio has a speed decrease of 98-99% Pintool has a speed decrease of 99.5%

Hence Dynamorio is the option to go for if everything else fails, and Pintool only if Dynamorio fails too.

Dynamorio solutions:

Pintool solutions:

Non-AFL solutions

There are many binary-only fuzzing frameworks. Some are great for CTFs but don’t work with large binaries, others are very slow but have good path discovery, some are very hard to set-up …

Closing words

That’s it! News, corrections, updates? Send an email to vh@thc.org