Surreptitious Software

Surreptitious Software: Obfuscation,

Watermarking, and Tamperproofing for
Software Protection

Table of Contents

Contents
Preface
About the Authors
Acknowledgments
1 What Is Surreptitious Software?
1.1 Setting the Scene
1.2 Attack and Defense
1.3 Program Analysis
1.3.1 A Simple Reverse Engineering Example
1.4 Code Obfuscation
1.4.1 Applications of Code Obfuscation
1.4.2 Obfuscating Transformations
1.4.3 Black Hat Code Obfuscation
1.5 Tamperproofing
1.5.1 Applications of Tamperproofing
1.5.2 An Example
1.6 Software Watermarking
1.6.1 An Example
1.6.2 Attacks on Watermarking Systems
1.7 Software Similarity
1.7.1 Plagiarism
 Table of Contents
1.7.2 Software Forensics
1.7.3 Birthmarking
1.7.4 A Birthmarking Example
1.8 Hardware-Based Protection Techniques
1.8.1 Distribution with Physical Token
1.8.2 Tying the Program to the CPU
1.8.3 Ensuring Safe Execution Environment
1.8.4 Encrypted Execution
1.8.5 Physical Barriers
1.9 Discussion
1.9.1 Reasons to Use Software Protection . . .
1.9.2 . . . and Reasons Not To
1.9.3 So Which Algorithms Should I Use?
1.10 Notation
2 Methods of Attack and Defense
2.1 Attack Strategies
2.1.1 A Prototypical Cracking Target
2.1.2 What's the Adversary's Motivation?
2.1.3 What Does the Adversary Get to Crack?
2.1.4 What's the Adversary's Attack Methodology?
2.1.5 What Tools Does the Adversary Use?
2.1.6 What Techniques Does the Adversary Use?
2.1.7 Discussion
2.2 Defense Strategies
2.2.1 Notation
2.2.2 The cover Primitive
2.2.3 The duplicate Primitive
2.2.4 The split and merge Primitives
2.2.5 The reorder Primitive
 Table of Contents
2.2.6 The map Primitive
2.2.7 The indirect Primitive
2.2.8 The mimic Primitive
2.2.9 The advertise Primitive
2.2.10 The detect-respond Primitive
2.2.11 The dynamic Primitive
2.2.12 Discussion
2.3 Discussion
2.3.1 What Do We Need from Attack and Defense Models?
2.3.2 How Do We Use the Models to Devise Algorithms?

3 Program Analysis
3.1 Static Analysis
3.1.1 Control Flow Analysis
3.1.2 Data Flow Analysis
3.1.3 Data Dependence Analysis
3.1.4 Alias Analysis
3.1.5 Slicing
3.1.6 Abstract Interpretation
3.2 Dynamic Analysis
3.2.1 Debugging
3.2.2 Profiling
3.2.3 Tracing
3.2.4 Emulation
3.3 Reconstituting Source
3.3.1 Disassembly
3.3.2 Decompilation
3.4 Pragmatic Analysis
3.4.1 Style Metrics
3.4.2 Software Complexity Metrics
 Table of Contents
3.4.3 Software Visualization
3.5 Discussion
4 Code Obfuscation
4.1 Semantics-Preserving Obfuscating Transformations
4.1.1 Algorithm OBFCF: Diversifying Transformations
4.1.2 Algorithm OBFTP: Identifier Renaming
4.1.3 Obfuscation Executives
4.2 Definitions
4.2.1 Potent Obfuscating Transformations
4.2.2 Efficient Obfuscating Transformations
4.2.3 Stealth
4.2.4 Other Definitions
4.3 Complicating Control Flow
4.3.1 Opaque Expressions
4.3.2 Algorithm OBFWHKD: Control-Flow Flattening
4.3.3 Introducing Aliasing
4.3.4 Algorithm OBFCTJ[sub(bogus)]: Inserting Bogus Control Flow
4.3.5 Algorithm OBFLDK: Jumps Through Branch Functions
4.3.6 Attacks
4.4 Opaque Predicates
4.4.1 Algorithm OBFCTJ[sub(pointer)]: Opaque Predicates from Pointer
Aliasing
4.4.2 OBFWHKD[sub(opaque)]: Opaque Values from Array Aliasing
4.4.3 Algorithm OBFCTJ[sub(thread)]: Opaque Predicates from
Concurrency
4.4.4 Breaking Opaque Predicates
4.5 Data Encodings
4.5.1 Encoding Integers
4.5.2 Encoding Booleans
 Table of Contents
4.5.3 Encoding Literal Data
4.5.4 Encoding Arrays
4.6 Breaking Abstractions
4.6.1 Algorithm OBFWC[sub(sig)]: Merging Function Signatures
4.6.2 Algorithm OBFCTJ[sub(class)]: Splitting and Merging Classes
4.6.3 Algorithm OBFDMRVSL: Destroying High-Level Structures
4.6.4 Algorithm OBFAJV: Modifying Instruction Encodings
4.7 Discussion
5 Obfuscation Theory
5.1 Definitions
5.2 Provably Secure Obfuscation: Possible or Impossible?
5.2.1 Turing's Halting Problem
5.2.2 Algorithm REAA: De-obfuscating Programs
5.3 Provably Secure Obfuscation: It's Possible (Sometimes)!
5.3.1 Algorithm OBFLBS: Obfuscating with Point Functions
5.3.2 Algorithm OBFNS: Obfuscating Databases
5.3.3 Algorithm OBFPP: Homomorphic Encryption
5.3.4 Algorithm OBFCEJO: Whitebox DES
5.4 Provably Secure Obfuscation: It's Impossible (Sometimes)!
5.4.1 A General Obfuscator
5.4.2 Obfuscating Learnable Functions
5.4.3 Proving that Obfuscation Is Impossible
5.4.4 Discussion
5.5 Provably Secure Obfuscation: Can It Be Saved?
5.5.1 Overcoming Impossibility
5.5.2 Definitions Revisited: Make Obfuscation Interactive
5.5.3 Definition Revisited: Make Obfuscation Non-Semantics Preserving
5.6 Discussion
 Table of Contents

6 Dynamic Obfuscation
6.1 Definitions
6.2 Moving Code Around
6.2.1 Algorithm OBFKMNM: Replacing Instructions
6.2.2 OBFAG[sub(swap)]: Self-Modifying State Machine
6.2.3 OBFMAMDSB: Dynamic Code Merging
6.3 Encryption
6.3.1 OBFCKSP: Code as Key Material
6.3.2 OBFAG[sub(crypt)]: Combining Self-Modification and Encryption
6.4 Discussion
7 Software Tamperproofing
7.1 Definitions
7.1.1 Checking for Tampering
7.1.2 Responding to Tampering
7.1.3 System Design
7.2 Introspection
7.2.1 Algorithm TPCA: Checker Network
7.2.2 Generating Hash Functions
7.2.3 Algorithm TPHMST: Hiding Hash Values
7.2.4 The Skype Obfuscated Protocol
7.2.5 Algorithm REWOS: Attacking Self-Hashing Algorithms
7.2.6 Discussion
7.3 Algorithm TPTCJ: Response Mechanisms
7.4 State Inspection
7.4.1 Algorithm TPCVCPSJ: Oblivious Hash Functions
7.4.2 Algorithm TPJJV: Overlapping Instructions
7.5 Remote Tamperproofing
7.5.1 Distributed Check and Respond
 Table of Contents
7.5.2 Solution Strategies
7.5.3 Algorithm TPZG: Slicing Functions
7.5.4 Algorithm TPSLSPDK: Measuring Remote Hardware
7.5.5 TPCNS: Continuous Replacement
7.6 Discussion
8 Software Watermarking
8.1 History and Applications
8.1.1 Applications
8.1.2 Embedding a Mark in Audio
8.1.3 Embedding a Mark in an Image
8.1.4 Embedding a Mark in Natural-Language Text
8.2 Watermarking Software
8.3 Definitions
8.3.1 Watermark Credibility
8.3.2 Attacks
8.3.3 Watermarking vs. Fingerprinting
8.4 Watermarking by Permutation
8.4.1 Algorithm WMDM: Reordering Basic Blocks
8.4.2 Renumbering
8.4.3 Algorithm WMQP: Improving Credibility
8.5 Tamperproofing Watermarks
8.5.1 Algorithm WMMC: Embedding Media Watermarks
8.6 Improving Resilience
8.6.1 Algorithm WMSHKQ: Statistical Watermarking
8.7 Improving Stealth
8.7.1 Algorithm WMMIMIT: Mapping Instructions
8.7.2 Algorithm WMVVS: Watermarks in CFGs
8.7.3 Algorithm WMCC: Abstract Interpretation
 Table of Contents

8.8 Steganographic Embeddings

8.8.1 Algorithm WMASB: The Compiler as Embedder
8.9 Splitting Watermark Integers
8.9.1 Splitting a Large Mark into Small Pieces
8.9.2 Redundant Watermark Pieces
8.9.3 Sparse Codes for Increased Credibility
8.10 Graph Codecs
8.10.1 Oriented Parent-Pointer Tree
8.10.2 Radix Graphs
8.10.3 Permutation Graphs
8.10.4 Planted Plane Cubic Trees
8.10.5 Reducible Permutation Graphs
8.11 Discussion
8.11.1 Embedding Techniques
8.11.2 Attack Models

9 Dynamic Watermarking
9.1 Algorithm WMCT: Exploiting Aliasing
9.1.1 A Simple Example
9.1.2 Recognition Problems
9.1.3 Increasing Bitrate
9.1.4 Increasing Resilience to Attack
9.1.5 Increasing Stealth
9.1.6 Discussion
9.2 Algorithm WMNT: Exploiting Parallelism
9.2.1 Embedding Watermarking Widgets
9.2.2 Embedding Example
9.2.3 Recognition
9.2.4 Avoiding Pattern-Matching Attacks
9.2.5 Tamperproofing Widgets
 Table of Contents
9.2.6 Discussion
9.3 Algorithm WMCCDKHLS[sub(paths)]: Expanding Execution
Paths
9.3.1 Encoding and Embedding
9.3.2 Recognition
9.3.3 Discussion
9.4 Algorithm WMCCDKHLS[sub(bf)] : Tamperproofing Execution
Paths
9.4.1 Embedding
9.4.2 Recognition
9.4.3 Tamperproofing the Branches
9.4.4 Discussion
9.5 Discussion
10 Software Similarity Analysis
10.1 Applications
10.1.1 Clone Detection
10.1.2 Software Forensics
10.1.3 Plagiarism Detection
10.1.4 Birthmark Detection
10.2 Definitions
10.2.1 Similarity Measures
10.3 k-gram-Based Analysis
10.3.1 SSSWA[sub(WINNOW)]: Selecting k-gram Hashes
10.3.2 SSSWA[sub(MOSS)]: Software Plagiarism Detection
10.3.3 SSMC[sub(kgram)]: k-gram Java Bytecode Birthmarks
10.4 API-Based Analysis
10.4.1 SSTNMM: Object-Oriented Birthmarks
10.4.2 SSTONMM: Dynamic Function Call Birthmarks
 Table of Contents
10.4.3 SSSDL: Dynamic k-gram API Birthmarks
10.5 Tree-Based Analysis
10.5.1 SSEFM: AST-Based Clone Detection
10.6 Graph-Based Analysis
10.6.1 SSKH: PDG-Based Clone Detection
10.6.2 SSLCHY: PDG-Based Plagiarism Detection
10.6.3 SSMC[sub(wpp)]: Dynamic Whole Program Birthmarks
10.7 Metrics-Based Analysis
10.7.1 SSKK: Metrics-Based Clone Detection
10.7.2 SSLM: Metrics-Based Authorship Analysis
10.8 Discussion
11 Hardware for Protecting Software
11.1 Anti-Piracy by Physical Distribution
11.1.1 Distribution Disk Protection
11.1.2 Dongles and Tokens
11.2 Authenticated Boot Using a Trusted Platform Module
11.2.1 Trusted Boot
11.2.2 Taking Measurements
11.2.3 The TPM
11.2.4 The Challenge
11.2.5 Social Trust and Privacy Issues
11.2.6 Applications and Controversies
11.3 Encrypted Execution
11.3.1 The XOM Architecture
11.3.2 Preventing Replay Attacks
11.3.3 Fixing a Leaky Address Bus
11.3.4 Fixing a Leaky Data Bus
11.3.5 Discussion
 Table of Contents

11.4 Attacks on Tamperproof Devices

11.4.1 Tapping the BusThe Microsoft XBOX Hack
11.4.2 Injecting CiphertextDallas Semiconductor DS5002FP
11.4.3 Hacking Smartcards
11.4.4 Non-Invasive Attacks
11.4.5 Board-Level Protection
11.5 Discussion
Bibliography
Index