trailofbits / mishegos

A differential fuzzer for x86 decoders

Home Page:https://blog.trailofbits.com/2019/10/31/destroying-x86_64-instruction-decoders-with-differential-fuzzing/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

mishegos

CI

A differential fuzzer for x86 decoders.

mishegos

Read more about mishegos in its accompanying blog post and academic publication (paper | recording | slides).

@InProceedings{woodruff21differential,
  author       = "William Woodruff and Niki Carroll and Sebastiaan Peters",
  title        = "Differential analysis of x86-64 instruction decoders",
  booktitle    = "Proceedings of the Seventh Language-Theoretic Security Workshop~({LangSec}) at the {IEEE} Symposium on Security and Privacy",
  year         = "2021",
  month        = "May"
}

Usage

Start with a clone, including submodules:

git clone --recurse-submodules https://github.com/trailofbits/mishegos

Building

mishegos is most easily built within Docker:

docker build -t mishegos .

Alternatively, you can try building it directly.

Make sure you have binutils-dev (or however your system provides libopcodes) installed:

make
# or
make debug

Running

Run the fuzzer for a bit:

./src/mishegos/mishegos ./workers.spec > /tmp/mishegos

mishegos checks for three environment variables:

  • V=1 enables verbose output on stderr
  • D=1 enables the "dummy" mutation mode for debugging purposes
  • M=1 enables the "manual" mutation mode (i.e., read from stdin)
  • MODE=mode can be used to configure the mutation mode in the absence of D and M
    • Valid mutation modes are sliding (default), havoc, and structured

Convert mishegos's raw output into JSONL suitable for analysis:

./src/mish2jsonl/mish2jsonl /tmp/mishegos > /tmp/mishegos.jsonl

mish2jsonl checks for V=1 to enable verbose output on stderr.

Run an analysis/filter pass group on the results:

./src/analysis/analysis -p same-size-different-decodings < /tmp/mishegos.jsonl > /tmp/mishegos.interesting

Generate an ugly pretty visualization of the filtered results:

./src/mishmat/mishmat < /tmp/mishegos.interesting > /tmp/mishegos.html
open /tmp/mishegos.html

Tip: The HTML file that mishmat generates could be hundreds of megabytes large, which will likely result in a bad browser viewing experience. Using the split tool, you can create multiple smaller HTML files with a specified number of entries per file (10,000 in the following example) and load each of them separately:

mkdir /tmp/mishegos-html
split -d --lines=10000 - /tmp/mishegos-html/mishegos_ \
    --additional-suffix='.html' --filter='./src/mishmat/mishmat > $FILE' \
    < /tmp/mishegos.interesting

Contributing

We welcome contributors to mishegos!

A guide for adding new disassembler workers can be found here.

Performance notes

All numbers below correspond to the following run:

V=1 timeout 60s ./src/mishegos/mishegos ./workers.spec > /tmp/mishegos

Outside Docker:

  • On a Linux desktop (Ubuntu 20.04, Ryzen 5 3600, 32GB DDR4):
    • Commit d80063a
    • 8 workers (no udis86) + 1 mishegos fuzzer process
    • 8.7M outputs/minute
    • 9 cores pinned

TODO

  • Performance improvements
    • Break cohort collection out into a separate process (requires re-addition of semaphores)
    • Maybe use a better data structure for input/output/cohort slots
  • Add a scaling factor for workers, e.g. spawn N of each worker
  • Pre-analysis normalization (whitespace, immediate representation, prefixes)
  • Analysis strategies:
    • Filter by length, decode status discrepancies
    • Easy: lexical comparison
    • Easy: reassembly + effects modeling (maybe with microx?)
  • Scoring ideas:
    • Low value: Flag/prefix discrepancies
    • Medium value: Decode success/failure/crash discrepancies
    • High value: Decode discrepancies with differing control flow, operands, maybe some immediates
  • Visualization ideas:
    • Basic but not really basic: some kind of mouse-over differential visualization

License

mishegos is licensed and distributed under the Apache v2.0 license. Contact us if you’re looking for an exception to the terms.

About

A differential fuzzer for x86 decoders

https://blog.trailofbits.com/2019/10/31/destroying-x86_64-instruction-decoders-with-differential-fuzzing/

License:Apache License 2.0


Languages

Language:C++ 39.4%Language:C 33.3%Language:Ruby 13.6%Language:Makefile 7.4%Language:Rust 4.9%Language:Dockerfile 0.7%Language:CMake 0.6%