An Analytical Approach to Modern Binary Deobfuscation

DATE: 22-25 August 2022

TIME: 09:00 to 17:00 SGT/GMT +8

Code obfuscation has become one of the most prevalent mechanisms aiming to complicate the process of software reverse engineering. It plays a major role on a wide range of domains: from malware threats to protection of intellectual property and digital rights management.

An Analytical approach to Modern Binary Deobfuscation is a curated training that provides an intensive jump-start into the field of code (de)obfuscation. Over the course of this training, students will receive a comprehensive introduction to the most relevant software obfuscation mechanisms as well as existing deobfuscation techniques to analyze, confront and defeat obfuscated code.

Teaching Methodology

Live classes are designed to be dynamic and engaging, making the students get the most out of the training materials and instructor expertise. A clear presentation of the concepts, accompanied by illustrative examples and demos. For each section, there will be practice time allocated. The students will be provided with several exercises and projects to work on, with the continuous support of the instructor.

Students will be provided with

• Access to a VM with all tools, examples and exercises
• Access to a private chat with instructor and other students

Agenda

Day 1

• Introduction, context and motivation
• Data-flow based obfuscation
  • Constant unfolding
  • Dead code insertion
  • Encodings
  • Pattern-based obfuscation
• Control-flow based obfuscation
  • Function inlining/outlining
  • Opaque predicates
  • Control-flow flattening
• Mixing data-flow and control-flow obfuscation
  • VM-based obfuscation
  • Hardening VM-based obfuscation

Exercises
Project – Manually craft a custom obfuscation VM

Day 2

• SMT-based analysis
  • A primer on SMT solvers
  • Translate code conditions into SMT solver constraints
  • Program analysis with SMT solvers
• Mixed Boolean-Arithmetic
  • Preliminary concepts
  • MBA rewriting
  • Insertion of identities
  • Opaque constants

Exercises
Project – Applied MBA to obfuscate the semantics of VM-handlers

Day 3

• Symbolic execution
  • Reasoning about code in a symbolic way
  • Working with native code
  • Working with intermediate representations
  • Data-flow analysis and compiler optimizations
  • Extract symbolic formulas
  • Extract path constraints
  • Plugging an SMT solver
  • Attacking obfuscation schemes

Guided project – Build your own (toy) symbolic execution engine

Exercises
Project – Attack obfuscated VM and explore symbolic execution limits

Day 4

• Program synthesis
  • Code syntax VS Code semantics
  • Specifying program behavior
  • Oracle-based program synthesis
  • Describing semantics through I/O behavior
  • Generating I/O pairs
  • Different synthesis flavors
  • Practical considerations
  • Attacking obfuscation schemes

• Conclusions and research directions

Guided project – Build your own code semantics synthesizer

Exercises
Project – Recover the semantics of MBA-obfuscated VM-handlers

Tools used

• Disassemblers
  • IDA Free/Home/Pro
  • Ghidra
  • Radare2
• Obfuscation
  • Manual obfuscation
  • O-LLVM
  • Tigress
• Symbolic execution
  • Miasm
  • Triton
• Program synthesis
  • Syntia
  • Msynth
  • Custom tooling
• Other tools
  • Z3
  • Other custom tooling