Future of Cyber Lecture Notes

Final Exam Details

• Date: May 15th for the paper exam, and May 15th to May 18th for the practical exam.
• Format: Similar choice between practical (online/remote) and in-person.

Effective Cybersecurity Policies

• Policies provide guidance for non-technical people.
• They often target low-hanging fruit, but are not comprehensive.

New Approach: Making Failure Impossible

• Security challenges often stem from system design complexities.
• No single action can fix all security issues.

Memory Safe Languages

• Problems in C/C++:
  • Undefined behaviors like out-of-bounds reads, buffer overflows.
  • Usage of uninitialized or freed memory.
• Solutions:
  • Address Sanitizer (ASAN): Detects but doesn't prevent all memory issues.
  • Memory Safe Languages (e.g., Go, Rust): Prevents memory safety issues (
    • Prevention of out-of-bounds indexing, buffer overflows, and use-after-free scenarios.
  • Rust has a borrow checker to ensure memory safety.
  • Go’s garbage collection automates memory management.
• Limitations:
  • Both languages allow "unsafe" operations for specific needs like interfacing with C libraries.
  • C code still prevalent, prone to logic errors rather than memory corruption.

Sandboxing Techniques

• Goals:
  • Prevent unauthorized actions.
  • Prevent resource exhaustion.
• System Calls: Key interface between programs and the kernel.
• ptrace: Used in debugging (e.g., GDB) and for sandboxing; performance can be a concern.
• seccomp:
  • Filters syscalls to enforce secure modes.
  • Example code provided for setting up seccomp filters.

Additional Security Measures

• AppArmor/SELinux/Landlock: Kernel-level security modules to control application permissions.
• Namespaces: Enables Docker-style container isolation.
• cgroups: Controls resource allocation (e.g., CPU, memory) for processes.

Program Analysis: Formal Verification

• Objective: Verify that software does not perform unwanted actions.
• Tools:
  • angr: Symbolic execution to explore software states.
  • cbmc: Overapproximation to check for potential buffer overflows.
  • TLA+: Mathematical approach to verify algorithms.

Zero Trust Model

• Controls access to systems and services based on user credentials.
• Technologies like Tailscale, Yggdrasil, Zerotier implement zero trust by limiting access to approved users.

Assignment 10

• Available online, but does not include any flags.

Summary

• Future cybersecurity efforts focus on eliminating vulnerabilities through technological advancements rather than relying solely on policy.
• Next session features a career guest speaker.