Finding deadlocks in large concurrent java programs using genetic algorithms

Abstract

Model checking is a fully automatic technique for checking concurrent software properties in which the states of a concurrent system are explored in an explicit or implicit way. However, the state explosion problem limits the size of the models that are possible to check. Genetic Algorithms (GAs) are metaheuristic techniques that have obtained good results in problems in which exhaustive techniques fail due to the size of the search space. Unlike exact techniques, metaheuristic techniques can not be used to verify that a program satisfies a given property, but they can find errors on the software using a lower amount of resources than exact techniques. In this paper, we compare a GA against classical exact techniques and we propose a new operator for this problem, called memory operator, that allows the GA to explore even larger search spaces. We implemented our ideas in the Java Pathfinder (JPF) model checker to validate them and present our results. To the best of our knowledge, this is the first implementation of a Genetic Algorithm in this model checker.

Generating Unit Tests for Concurrent Classes

“…A similar approach is also followed by Alba et al [1]. These approaches search for individual program inputs rather than sequences of calls, and they are not driven by structural coverage criteria.…”

“…For example, Figure 1 shows the entrySet method of the HashMultimap.AsMap class, containing bug #339 1 . If two threads access the same hashmap and call entrySet at the same time, a race condition can lead to erroneous 1 http://code.google.com/p/guava-libraries/issues/detail?id=339 1 public Set entrySet() { 2 Set result = entrySet; 3 return (entrySet == null) ?…”

“…If two threads access the same hashmap and call entrySet at the same time, a race condition can lead to erroneous 1 http://code.google.com/p/guava-libraries/issues/detail?id=339 1 public Set entrySet() { 2 Set result = entrySet; 3 return (entrySet == null) ? behavior: If entrySet is null initially, then the first thread assigns null to result and should then check if entrySet was null.…”

Using Contracts to Guide the Search-Based Verification of Concurrent Programs

“…For example, Godefroid and Khurshid [4], Alba et al [1], and Shousha et al [7], each describe the use of a genetic algorithm to locate deadlocks in models of concurrent programs; Staunton and Clark [8] describe an estimation of distribution algorithm using N-grams with the same objective. The fitness function used by Godefroid and Khurshid utilises the total number of transitions out of model states visited on the path to the current state.…”

Comparing Metaheuristic Algorithms for Error Detection in Java Programs

“…Our objective is to find deadlocks in the programs and we prefer short walks. As such, our fitness function f is defined as follows: [11] BFS yes yes no A * [11] A * yes yes yes Genetic Algorithm [13] GA no no yes Genetic Algorithm [13] GAMO no no yes with Memory Operator Particle Swarm Optimization [14] PSO no no yes Ant Colony Optimization [9] ACOhg no no yes Simulated Annealing SA no no yes Random Search RS no no no Beam Search [11] BS no yes yes…”