MIT OpenCourseWare Security Course Notes

Introduction

• Instructor: Led by a professor, co-lecturer: James Mickens from Microsoft Research.
• TAs: Stephen, Webb, [INAUDIBLE], James.

Course Structure

• Goal: Understand how to build secure systems, why they fail, and how to fix them.
• Textbook: No recommended textbook; based on research papers.
• Reading Assignments: Read and submit answers/questions by 10:00 PM before lecture.
• Lectures: Discussion-based around the assigned papers.

Class Components

• Lectures: Cover various security topics, discuss assigned research papers.
• Lab Assignments: Practical assignments throughout the semester, covering a range of security problems in web servers.
  • Lab 1: Buffer overflows in C and Assembly.
  • Lab 2: Python-based.
  • Lab 3: Different language.
  • Lab 5: JavaScript.
  • Expectation: Learn multiple programming languages.

Administrative Details

• Office Hours: TAs available for help.
• Piazza: Platform for submitting questions online.
• Lecture Recordings: Available online.

Ethical Guidelines

• Conduct security research responsibly.
• Don't perform illegal activities or misuse knowledge.
• Guidelines available online.

What is Security?

• Definition: Achieving goals despite adversaries.
• Policies: Confidentiality, Integrity, Availability.

Threat Model

• Assumptions: Define adversary's capabilities (e.g., do not have passwords, physical access).
• Importance: Conservatively assume worst-case scenarios.

Mechanisms

• Definition: Software/hardware enforcing security policies.
• Example Mechanisms: Cryptography, Authentication.

Challenges in Security

• Negative Goal: Ensuring nothing breaks despite all possible attacks.
• Changing Threat Models: Assumptions may evolve over time.
• Complex Systems: Composed of many parts, error-prone.
• Iterative Process: Continuous improvement needed.

Examples of Security Failures

Policy Failures

• Account Recovery: Weakens the system, case of Sarah Palin's Yahoo account.
• Interconnected Systems: Mat Honan's case, weak policies across services (Apple, Amazon).

Threat Model Failures

• Human Factors: Weak passwords, phishing.
• Technological Evolution: Old assumptions may no longer hold true (e.g., Kerberos key length).
• Government Capability: Backdoors in hardware (NSA).

Mechanism Failures

• Programming Bugs: E.g., Apple's iCloud login issue.
• Implementation Errors: E.g., Android Bitcoin SecureRandom() bug.

Mechanism Bug Examples

• Stack Overflows: Buffer overflows overwrite return addresses, leading to exploits.
  • Lab 1 focuses on this topic.
• Examples in History: DARPA's red team attacks, poor source code protection.
• Modern Concerns: SSL names encoding and null-termination in C.

Mitigating Mechanism Failures

• Defensive Programming: Avoid unsafe functions like getS().
• Modern CPU Features: Non-executable stacks.
• Segmentation and Permissions: Prevent code execution from data segments.
• Minimize Trusted Computing Base: Fewer components enforcing security policies.

Conclusion

• Important Note: Start early on labs, seek help if needed.
• Look Forward: To more specific details in subsequent lectures and assignments.