中央研究院 資訊科學研究所

In this talk, we will present a framework that tackles the non-determinism and probe effect incurred in execution analysis of embedded software. When porting multi-threaded embedded software to multicore processors, developers may be cautious that synchronization errors and concurrency bugs that are benign on a uniprocessor system may surface in multiprocessor execution. It becomes critical to carry out execution analysis that is deterministic and free of probe effect from instrumentation overhead.

We will first discuss a deterministic replay mechanism. Based on weak determinism model of program execution, the happens-before relation of program execution events is recorded and used to guide any replay operations in which software instrumentation and execution analysis can be performed. Then, a discussion of probe effect is presented and a simulation-based analysis is proposed. The approach allows us to infer the ordering of events in program execution and verify the existence of probe effect resulted from instrumentation. Finally, to ensure the existence of weak determinism, the framework includes an efficient parallel data race detector for multicore machines. In this detector, data access information needed for dynamic detection is collected at application threads and passed to detection threads.  As a consequence, the overhead caused by locking operations in data race detection can be alleviated and multiple cores can be fully utilized to speed up and scale up the detection.

Yann-Hang Lee received his Ph.D. degree in Computer, Information, and Control Engineering from the University of Michigan, Ann Arbor, MI, in 1984. From December 1984 to August 1988, he was a research staff member at the Architecture Analysis and Design Group, IBM Thomas J. Watson Research Center, Yorktown Heights, NY. He joined Computer and Information Sciences Department, University of Florida, Gainesville, FL, in 1988, and is currently a professor in the School of Computing, Informatics, and Decision Systems Engineering, Arizona State University. Since 2008, he has served as the Program Chair of the Computer Science and Engineering Program in the School.

Dr. Lee's research interests are in the areas of real-time computing, embedded systems, software engineering, distributed systems, and performance evaluation. Currently, his research projects are focused on various software issues of mission and safety critical real-time embedded systems, service-oriented distributed embedded systems, and system support for embedded software. He co-edited three special issues in Real-time Systems in IEEE Computer (May 1992), IEEE Proceedings (Jan. 1994), and IEEE Proceedings (July, 2003). He was a program co-chair for the Real-time Systems Symposium, 1995, conference co-chair for the Real-time Systems Symposium, 1996, and a program co-chair of the Sixth IEEE International Symposium on Autonomous Decentralized Systems, 2003.