Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore a 15-minute conference presentation that unifies and simplifies soundness definitions for predictive concurrency analyses. Learn how researchers from UCLA and SUNY Oswego address the challenge of defining soundness for analyses that detect concurrency bugs without accessing program source code, taking only execution traces as input. Discover their novel graph-based soundness definition that separates thread-local properties from whole-execution properties and effectively handles weak memory executions. Examine the three-step proof recipe developed by the authors and see how it applies to proving the soundness of six existing analyses, including the first soundness proof for a predictive analysis working with weak memory models. Gain insights into dynamic program analysis techniques, concurrency bug detection methods, and the theoretical foundations that ensure these analyses report no false positives while discovering bugs represented as consecutive sequences of events.
