Search based software testing for the generation of synchronization sequences for mutation testing of concurrent programs

• Ano de defesa: 2018
• Autor(a) principal: Silva, Rodolfo Adamshuk
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Concurrent programming; Mutation testing; Programação concorrente; Search based software testing; Teste de mutação; Teste de software baseado em busca
• Link de acesso: http://www.teses.usp.br/teses/disponiveis/55/55134/tde-10102018-141934/

Concurrent programming has become an essential paradigm for reductions in the computational time in many application domains. However, the validation and testing activity is more complex than the testing for sequential programs due to the non-determinism, synchronization and inter-process communication. Mutation testing is based on mistakes produced by software developers and presents a high effectiveness to reveal faults. However, high computational costs limit its applicability even for sequential code, becoming higher for concurrent programs in which each test has to be executed with different (ideally all) thread schedules. To date, only selective mutation have been applied to reduce the number of mutants in concurrent programs, however, the problem of state explosion of thread schedules still remains. This Ph.D. thesis presents the SBBMuT approach that applies deterministic execution and genetic algorithm for the generation and execution of a set of synchronization sequences during the mutation testing of Java multithreaded programs. An experimental study was conducted, and the results showed that the set of synchronization sequences generated by SBBMuT achieved a higher mutation score in comparison with the use of the Java PathFinder model checking tool.