Java Code Obfuscation

1
Java Code Obfuscation

• Neerja Bhatnagar

2
Reverse Engineering

• Figuring out source code corresponding to a given
  byte code
• Source code intellectual property, needs
  protection
• Money, intelligence, hard work, time put to work
  to develop code

3
Problem of Reverse Engineering

• Java code especially prone to reverse
  engineering, because
• Java byte code well documented and well
  structured (freely available in JVM specs)
• Most of the information contained in source code
  also present in byte code.
• Software engineers have little control over
  distribution of Java applications and their
  source code because most Java applications
  distributed over the Internet (easy downloads)

4
Code Obfuscation

• Technique to discourage and deter reverse
  engineering
• Scrambles byte code to make it unintelligible,
  difficult to understand
• Alters structure and appearance of code
• Attempts to make in uneconomical for an attacker
  to reverse engineer code
• Does not affect function of original program
• Does not alter source code, alters byte code

5
Code Obfuscation Techniques

• Low-Level
• Layout Obfuscations
• Control Obfuscations
• Data Obfuscations
• High-Level or Design Obfuscations
• Class Coalescing Obfuscation
• Class Splitting Obfuscation
• Type Hiding

6
Design Obfuscations

• Low-level obfuscation techniques not sufficient
  because they do not hide design information
• Design information can facilitate an attacker
  from gaining general understanding of code and
  what it does
• Design obfuscations aim at hiding valuable
  high-level design information

7
Class Coalescing Obfuscation

• Combines two or more classes into a single class
• If the classes to be coalesced
• have attributes with same name, these are renamed
  to distinguish among them
• have non-constructor methods with same signature,
  except for one non-constructor methods, all
  others are renamed and altered

8
Class Coalescing Obfuscation Example
9
public class Animal String name int age
public Animal() name Goofy age 1
public int calcAge() return 2004
1985 public void comments()
System.out.println(Fluffy)
public class A String name String model
int age int age2 public Animal()
name Goofy age 1 public Car()
name VW age 2 public int
calcAge() return 2004 1985
public int calcAge(int make) return 2004
make public void comments()
System.out.println(Fluffy) public void
commentsFromCar() System.out.println(Smoot
h)
public class Car String model int age
public Car() name VW age 2
public int calcAge(int make) return 2004
make public void comments()
System.out.println(Smooth)
10
Class Coalescing Obfuscation Shortcomings

• Becomes very complicated when classes extend
  other classes or implement interfaces because
• superclasses and interfaces need to be coalesced
  too
• variable names and method signatures cannot be
  altered in any way
• Cannot be performed when classes extend classes
  from Java standard library
• Java standard library classes need to be
  coalesced too
• Makes code non-portable
• Classes with native methods cannot be coalesced
  because analyzing native code very difficult

11
Class Splitting Obfuscation

• Splits a class into multiple classes
• Several ways to split a class
• Valid split function preserves dependencies among
  classs methods and fields
• Valid split functions include
• Split class A into classes A1 and A2, where A2
  extends A1
• Split methods defined in class A between classes
  A1 and A2

12
Class Splitting Obfuscation Example
13
public class Animal String name int
age public Animal() name Bruno
age 5 public void sleep()
System.out.println(zzzzzz)
public class Dog String name int age
String sound public Dog() name
Bruno age 5 sound woof
public void bark() System.out.println(Woof
! Woof!) public void sleep()
System.out.println(zzzzzz)
public class Doggie extends Animal String
sound public Doggie() super()
sound woof public void bark()
System.out.println(Woof! Woof!)
14
Class Splitting Obfuscation Shortcomings

• Splitting a class very difficult unless initial
  design faulty
• May have reverse effect
• Splitting a class may improve overall code
  design, and get rid of spaghetti code!
• Might actually help the attacker, rather than
  deter him

15
Type Hiding

• Uses the concept of Java interface
• Transforms a concrete class into several
  interfaces
• Each interface contains a random subset of the
  concrete classs methods and fields
• Obscures by declaring a concrete class that
  implements those interfaces

16
Type Hiding Example
17
interface LivingCreature public void
bark()
public class Dog String name int age
String sound public Dog() name
Bruno age 5 sound woof
public void bark() System.out.println(Woof
! Woof!) public void sleep()
System.out.println(zzzzzz)
interface Animal public void sleep()
public class Doggie String name int age
String sound public Dog() name
Bruno age 5 sound woof
public void bark() System.out.println(Woof
! Woof!) public void sleep()
System.out.println(zzzzzz)
18
Type Hiding Shortcomings

• Type hiding done the way described in previous
  slide especially vulnerable to reverse
  engineering
• More effective when combined with low-level
  obfuscation techniques
• Can be made stronger when randomly selected
  classes go through type hiding, instead of all
  classes

19
Obfuscation Products

• RetroGuard by Retrologic Systems
• Klassmaster by Zelix
• JObfuscator by Duckware
• Semantic Designs

20
Conclusion

• None of the obfuscation techniques sufficient by
  themselves
• A combination of these techniques might provide
  strong resistance to reverse engineering
• Some claim code obfuscation slows down an
  application
• Impact on performance depends on a particular
  technique
• Impact usually minimal or none at all
• Decision to obfuscate or not is a trade-off
  between protecting code or taking a slight
  performance hit