Discovering and Validating Concurrency Specification from Test Executions

Concurrent systems are notoriously hard to get correct and testingthese systems is challenging as typically only a subset of concurrentbehavior can be covered and observed. Hence, even with a large body oftest cases it often remains unclear if the test cases cover the rightbehavior and, at the same time, if the system behaves as expected. Inthis work we present an automated formal validation approach torediscover concurrency behavior from dynamic (symbolic) executions. Inparticular, we generate formal UPPAAL models based on testobservations, which can be used to validate the intuition of the testengineers and can be formally checked on its own for potential safetyviolations to enhance the overall validation process. We present thekey ideas, our implementation for model generation from testingpthreaded C code using the Forest symbolic execution framework, and wepresent early experimental results.