A UML/MARTE model analysis method for detection of data races in concurrent systems

The earlier concurrency problems are identified, the less costly they are to fix. As larger, more complex concurrent systems are developed, early detection of problems is made increasingly difficult. We have developed a general approach meant to be used in the context of Model Driven Development. Our approach is based on the analysis of design models expressed in the Unified Modeling Language (UML) and uses specifically designed genetic algorithms to detect concurrency problems. Our main motivation is to devise practical solutions that are applicable in the context of UML design of concurrent systems without requiring additional modeling. All relevant concurrency information is extracted from UML models that comply with the UML Modeling and Analysis of Real-Time and Embedded Systems (MARTE) profile. Our approach was shown to work for both deadlocks and starvation. The current paper addresses data race detection, further illustrating how our approach can be tailored to other concurrency issues. Results on a case study inspired from the Therac-25 radiation machine show that our approach is effective in the detection of data races