Sound One-Phase Shape Analysis with Biabduction

07/12/2023

∙

by   Florian Sextl, et al.

∙

0

∙

share

Biabduction-based shape analysis is a static analysis technique that can find bugs and ensure memory safety in the presence of complex, linked data structures. As such, this analysis has proven to be scalable and is implemented in state-of-the-art industrial strength analyzers. However, standard biabduction-based shape analysis requires two analysis phases to guarantee that all computed function contracts are sound. We introduce two novel techniques shared learning and shape extrapolation which tackle the soundness problems of biabduction-based shape analysis and allow for a sound one-phase analysis in many real-world programs. We formally prove that both techniques guarantee soundness and verify their effectiveness in a case study based on a prototype implementation.