Flow-Sensitive Points-to Analysis for Rust: Enhancing Rupta for Improved Precision

open_access
Open Access
Author:
Butcher, Quinn
Area of Honors:
Computer Science
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
G. Gary Tan, Thesis Supervisor
Mohamed Khaled Almekkawy, Thesis Honors Advisor
Keywords:
Rust
MIR
Pointer Analysis
Dominance Frontier
Dominator Tree
Rupta
Static Single Assignment
Pointer-to
Abstract:
This thesis presents an approach to enhancing the Rupta program analysis tool with flow-sensitive points-to analysis for Rust. Flow sensitivity in points-to analysis enables more precise tracking of program behavior by considering control flow during analysis, a critical feature for improving accuracy in languages like Rust. Initially, a Global Value Numbering (GVN) approach was implemented in the tool to introduce flow sensitivity; however, this method had superfluous complexity, such was not implemented. The approach was subsequently revised to use a dominator tree and dominance frontier technique for introducing phi-functions into the Rust Intermediate Representation (MIR). This method allows for more accurate flow-sensitive analysis by incorporating control flow structures and ensuring that phi-functions correctly handle control flow edges. A custom translation function was developed to convert MIR into a form that supports these changes, leading to more accurate points-to relationships. The improved Rupta tool was evaluated on Rust programs, showing enhanced precision in points-to analysis. This work contributes to the development of more efficient and accurate static analysis tools for Rust and provides a foundation for future improvements in flow-sensitive analysis.