Finding Concurrency Bugs

Mahdi Eslamimehr
Visiting Project Scientist at UCLA Compilers Lab, USA

Abstract
Concurrent programming with shared memory can lead to a variety of concurrency bugs such as deadlocks and data races. How can we find such bugs? I will survey the vast literature on solutions to this problem, and then I will present a new approach that does even better. Our approach combines concolic execution and constraint solving into a new technique that drives an execution towards a concurrency bug candidate. In 4.5 million lines of Java, our tool found substantially more real concurrency bugs than many previous techniques combined. This a joint work with Jens Palsberg from UCLA, Mohsen Lesani from MIT, and George Edwards from USC.

Short Bio
Dr. Mahdi Eslamimehr received his Bachelor of Engineering in computer engineering from Sharif University, and his Master of Science in computer science from Linkoping University. He received his PhD from UCLA. He was a postdoctoral researcher at Compiler Construction Lab at UCLA, and later at SAP Labs/ Viewpoints Research Institute, and JPL NASA. He was formerly an engineer at Ericsson R&D, staff researcher at Samsung R&D, and research faculty at UCLA. Dr. Eslamimehr is now a senior researcher at YCombinarors.
his research activities, can be found on his website at www.mah-d.com and also his research lab (Quandary Peak) at www.quandarypeak.com.

Time: 15-16:30, Monday 10th of Khordad, 1395
Place: Main Amphitheatre, 2nd level, Department of Computer Engineering and IT, Amirkabir
University of Technology.