The idea and much of the original implementation comes from Nathan Voss njvoss299@gmail.com.
The port to afl++ if by Dominik Maier mail@dmnk.co.
The CompareCoverage and NeverZero counters features by Andrea Fioraldi andreafioraldi@gmail.com.
The code in ./unicorn_mode allows you to build a standalone feature that leverages the Unicorn Engine and allows callers to obtain instrumentation output for black-box, closed-source binary code snippets. This mechanism can be then used by afl-fuzz to stress-test targets that couldn't be built with afl-gcc or used in QEMU mode, or with other extensions such as TriforceAFL.
There is a significant performance penalty compared to native AFL, but at least we're able to use AFL on these binaries, right?
Requirements: you need an installed python2 environment.
First, make afl++ as usual. Once that completes successfully you need to build and add in the Unicorn Mode features:
$ cd unicorn_mode $ ./build_unicorn_support.sh
NOTE: This script downloads a Unicorn Engine commit that has been tested and is stable-ish from the Unicorn github page. If you are offline, you‘ll need to hack up this script a little bit and supply your own copy of Unicorn’s latest stable release. It's not very hard, just check out the beginning of the build_unicorn_support.sh script and adjust as necessary.
Building Unicorn will take a little bit (~5-10 minutes). Once it completes it automatically compiles a sample application and verify that it works.
To really use unicorn-mode effectively you need to prepare the following:
* Relevant binary code to be fuzzed * Knowledge of the memory map and good starting state * Folder containing sample inputs to start fuzzing with + Same ideas as any other AFL inputs + Quality/speed of results will depend greatly on quality of starting samples + See AFL's guidance on how to create a sample corpus * Unicorn-based test harness which: + Adds memory map regions + Loads binary code into memory + Emulates at least one instruction* + Yeah, this is lame. See 'Gotchas' section below for more info + Loads and verifies data to fuzz from a command-line specified file + AFL will provide mutated inputs by changing the file passed to the test harness + Presumably the data to be fuzzed is at a fixed buffer address + If input constraints (size, invalid bytes, etc.) are known they should be checked after the file is loaded. If a constraint fails, just exit the test harness. AFL will treat the input as 'uninteresting' and move on. + Sets up registers and memory state for beginning of test + Emulates the interested code from beginning to end + If a crash is detected, the test harness must 'crash' by throwing a signal (SIGSEGV, SIGKILL, SIGABORT, etc.)
Once you have all those things ready to go you just need to run afl-fuzz in ‘unicorn-mode’ by passing in the ‘-U’ flag:
$ afl-fuzz -U -m none -i /path/to/inputs -o /path/to/results -- ./test_harness @@
The normal afl-fuzz command line format applies to everything here. Refer to AFL's main documentation for more info about how to use afl-fuzz effectively.
For a much clearer vision of what all of this looks like, please refer to the sample provided in the ‘unicorn_mode/samples’ directory. There is also a blog post that goes over the basics at:
https://medium.com/@njvoss299/afl-unicorn-fuzzing-arbitrary-binary-code-563ca28936bf
The ‘helper_scripts’ directory also contains several helper scripts that allow you to dump context from a running process, load it, and hook heap allocations. For details on how to use this check out the follow-up blog post to the one linked above.
A example use of AFL-Unicorn mode is discussed in the Paper Unicorefuzz: https://www.usenix.org/conference/woot19/presentation/maier
As for the QEMU-based instrumentation, the afl-unicorn twist of afl++ comes with a sub-instruction based instrumentation similar in purpose to laf-intel.
The options that enables Unicorn CompareCoverage are the same used for QEMU. AFL_COMPCOV_LEVEL=1 is to instrument comparisons with only immediate values. QEMU_COMPCOV_LEVEL=2 instruments all comparison instructions. Comparison instructions are currently instrumented only on the x86 and x86_64 targets.
To make sure that AFL‘s fork server starts up correctly the Unicorn test harness script must emulate at least one instruction before loading the data that will be fuzzed from the input file. It doesn’t matter what the instruction is, nor if it is valid. This is an artifact of how the fork-server is started and could likely be fixed with some clever re-arranging of the patches applied to Unicorn.
Running the build script builds Unicorn and its python bindings and installs them on your system. This installation will supersede any existing Unicorn installation with the patched afl-unicorn version.
Refer to the unicorn_mode/samples/arm_example/arm_tester.c for an example of how to do this properly! If you don't get this right, AFL will not load any mutated inputs and your fuzzing will be useless!