Algorithmic Verification of Procedural Programs in the Presence of Code Variability

Abstract

We present a generic framework for verifying temporal safety properties of procedural programs that are dynamically or statically configured by replacing, adapting, or adding new components. To deal with such a variability of a program, we require programmers to provide local specifications for its variable components, and verify the global properties by replacing these specifications with maximal models. Our framework is a generalization of a previously developed framework that abstracts from all program data. In this work, we capture program data and thus significantly increase the range of properties that can be verified. Our framework is generic by being parametric on the set of observed program events and their semantics. We separate program structure from the behavior it induces to facilitate independent component specification and verification. To exemplify its use, we instantiate our framework to develop a compositional verification technique for programs written in a procedural language with pointers as the only datatype. We also adapt our previously developed toolset to provide support for compositional verification of such programs.

Testing automation systems by means of model checking

“…Finally, we consider the scenario of testing simultaneously several controllers with small differences. This idea has been inspired by the work [23]. The application of this scenario in real life is testing product lines, where different products are composed of common modules.…”

Provably correct control flow graphs from Java bytecode programs with exceptions

“…Still, our extraction algorithm can benefit from finer data flow analyses such as null pointer analysis [37] or symbolic execution [25], provided that these analyses are also proven to be sound. Furthermore, the latter technique is envisaged in [34], by one of the present authors, as the means to extract CFGs with symbolic data.…”