Abstract We present a case study on proof-driven software understanding of mature, security-critical infrastructure. While formal methods are traditionally applied during the design phase, we present our experience applying formal reasoning onto a mature industrial C++ codebase. We focus on a formal analysis of the core algorithm that implements the Stellar blockchain’s order book. By combining large language models (LLMs), Prototype Verification System (PVS), and Seahorn , we are able to prove core properties of the production codebase. Our approach also identified an inconsistency in documentation related to the reachability of an exception location. Most importantly, however, we produce artifacts that make it easy for code changes to be checked against established invariants. This work demonstrates how the strategic combination of theorem proving and model checking provides a path for delivering robust assurance to legacy systems.

Groupware applications require secure communication and group-management services. Participants in such applications may have divergent interests and may not fully trust each other. The services provided must then be designed to tolerate possibly misbehaving participants. Enclaves is a software framework for building such group applications. We discuss how the protocols used by Enclaves can be modified to guarantee proper service in the presence of nontrustworthy group members. We show how the improved protocol was formally specified and proven correct.

Modularization and abstraction are the keys to practical verification and analysis of large and complex systems. We present in an incremental methodology for the automatic analysis and verification of concurrent software systems. Our methodology is based on the theory of abstract interpretation. We first propose a compositional data flow analysis algorithm that computes invariants of concurrent systems by composing invariants generated separately for each component. We present a novel compositional rule allowing us to obtain invariants of the whole system as conjunctions of local invariants of each component. We also show how the generated invariants are used to construct, almost for free, finite state abstractions of the original system that preserve safety properties. This reduces dramatically the cost of computing such abstractions as reported in previous work. We finally give a novel refinement algorithm that refines the constructed abstraction until the property of interest is proved or a counterexample is exhibited. Our methodology is implemented in a framework that combines deductive methods supported by theorem proving techniques and algorithmic methods supported by model checking and abstract interpretation techniques.