ConFuzz: Coverage-Guided Property Fuzzing for Event-Driven Programs

Bug-free concurrent programs are hard to write due to non-determinism arising out of concurrency and program inputs. Since concurrency bugs typically manifest under specific inputs and thread schedules, conventional testing methodologies for concurrent programs like stress testing and random testing, which explore random schedules, have a strong chance of missing buggy schedules.In this paper, we introduce a novel technique that combines property-based testing with mutation-based, grey box fuzzer, applied to event-driven OCaml programs. We have implemented this technique in , a directed concurrency bug-finding tool for event-driven OCaml programs. Using , programmers specify high-level program properties as assertions in the concurrent program. uses the popular greybox fuzzer AFL to generate inputs as well as concurrent schedules to maximise the likelihood of finding new schedules and paths in the program so as to make the assertion fail. does not require any modification to the concurrent program, which is free to perform arbitrary I/O operations. Our experimental results show that is easy-to-use, effective, detects concurrency bugs faster than Node.Fz - a random fuzzer for event-driven JavaScript programs, and is able to reproduce known concurrency bugs in widely used OCaml libraries.

Choice Under Conditions of Uncertainty

All behavior varies, and behavior under a wide range of circumstances displays high levels of variability. This is especially true under many concurrent schedules of reinforcement, despite those schedules not being designed to elicit unpredictable behavior. A generalized matching analysis was performed to measure the effects of reinforcement on 5-item choice under two conditions: Probabilistic concurrent schedules and threshold-based operant variability schedules. Behavior was analyzed in terms of conditional probabilities, incorporating trial-by-trial response dynamics into the model. Performing this analysis meant overcoming a major difficulty: Obtained reinforcement is not a valid independent predictor of behavior the two are causally interlinked. The method of instrumental variable estimation is utilized to overcome the “endogeneity” of reinforcement, which permits unbiased estimation of the causal influence of reinforcement on responding. The analysis revealed a simple relationship between choices made and the distances traveled to make them. Subjects were more willing to travel through the chamber under the Concurrent schedule than under the Variability schedule. As a result, with respect to predicting a subject’s next response, Concurrent schedules elicited higher levels of behavioral variability than did Variability schedules. However, longer-term behavior under the Variability schedules better resembled a steady-state random process.