Probabilistic Testing for Weak Memory Concurrency
In the multicore era shared memory concurrency has emerged as a key programming paradigm. These programming models follow weak memory models to exploit the performance of the multicores. In consequence the programs may exhibit additional executions which cannot be explained by thread interleaving or sequential consistency.
Ensuring correctness is a well known problem in concurrency considering the large number of possible executions. The number of executions are exponential under sequential consistency due to possible interleavings and the weak memory models proliferate the challenge further.
Probabilistic concurrency testing (PCT) is an effective approach in identifying concurrency bugs with probabilistic guarantees. In this project we want to extend the PCT techniques for weak memory concurrent programs.
A Randomized Scheduler with Probabilistic Guarantees of Finding Bugs. Burckhardt et al. ASPLOS 2010.
Randomized Testing of Distributed Systems with Probabilistic Guarantees. Ozkan et. al. OOPSLA 2018.
C11Tester: a race detector for C/C++ atomics. Luo et. al. ASPLOS 2021.
Contacts for the project
Supervisor(s): Soham Chakraborty, Burcu Ozkan
Posted: June 07, 2021