Work

Arithmetizing Shape Analysis

Year: 2025
Type: conference-paper
Abstract: Memory safety is a fundamental correctness property of software. For programs that manipulate linked, heap-allocated data structures, ensuring memory safety requires analyzing their possible shapes. Despite significant advances in shape analysis, existing techniques rely on hand-crafted domains tailored to specific data structures, making them difficult to generalize and extend. This paper presents a novel approach that reduces memory-safety proofs to the verification of heap-less imperative programs, enabling the use of off-the-shelf software verification tools. We achieve this reduction through two complementary program instrumentation techniques: space invariants, which enable symbolic reasoning about unbounded heaps, and flow abstraction, which encodes global heap properties as local flow equations. The approach effectively verifies memory safety across a broad range of programs, including concurrent lists and trees that lie beyond the reach of existing shape analysis tools.
Language: English
Subfield: Software
Open Access status: hybrid
Estimated APC paid: $5,392
OpenAlex ID: W4412533011
FWCI: 1.758
Cites: 80
Cited by: 5
Related to: 10