[Paper Review] A survey of Hardware-based Control Flow Integrity (CFI)

Ruan de Clercq, Ingrid Verbauwhede|arXiv (Cornell University)|Jun 22, 2017

Security and Verification in Computing46 references51 citations

TL;DR

This survey analyzes 21 hardware-based CFI architectures to assess their security policies, limitations, hardware cost, performance, and practicality. It argues that many policies are not suitable for widespread adoption.

ABSTRACT

CFI is a computer security technique that detects runtime attacks by monitoring a program's branching behavior. This work presents a detailed analysis of the security policies enforced by 21 recent hardware-based CFI architectures. The goal is to evaluate the security, limitations, hardware cost, performance, and practicality of using these policies. We show that many architectures are not suitable for widespread adoption, since they have practical issues, such as relying on accurate control flow model (which is difficult to obtain) or they implement policies which provide only limited security.

Motivation & Objective

• Evaluate security policies enforced by 21 recent hardware-based CFI architectures.
• Assess limitations and practical challenges of these policies.
• Analyze hardware cost and performance impacts of hardware-based CFI implementations.
• Determine the practicality of adopting hardware-based CFI in real systems.

Proposed method

• Review and analyze 21 hardware-based CFI architectures.
• Characterize each architecture's control-flow policy and security guarantees.
• Evaluate practicality including hardware cost and performance impact.
• Identify common limitations such as reliance on accurate control-flow models.

Experimental results

Research questions

• RQ1 What security policies do current hardware-based CFI architectures enforce?
• RQ2 What are the limitations and practical challenges of these policies for real-world deployment?
• RQ3 How do hardware costs and performance overheads affect the viability of hardware-based CFI?
• RQ4 Are the control-flow models used by these architectures accurate and feasible to obtain?

Key findings

• Many architectures rely on accurate control-flow models, which are difficult to obtain.
• Several policies provide only limited security guarantees.
• A substantial portion of architectures face practical issues hindering widespread adoption.
• Hardware cost and performance overhead are significant factors in evaluating practicality.
• The surveyed policies vary in robustness and applicability across scenarios.