Predicting potential deadlocks in multithreaded programs

Abstract

Summary In a multithreaded program, competition of threads for shared resources raises the deadlock possibility, which narrows the system liveness. Because such errors appear in specific schedules of concurrent executions of threads, runtime verification of threads behavior is a significant concern. In this study, we extended our previous approach for prediction of runtime behavior of threads may lead to an impasse. Such a prediction is of importance because of the nondeterministic manner of competing threads. The prediction process tries to forecast future behavior of threads based on their observed behavior. To this end, we map observed behavior of threads into time‐series data sets and use statistical and artificial intelligence methods for forecasting subsequent members of the sets as future behavior of the threads. The deadlock prediction is carried out based on probing the allocation graph obtained from actual and predicted allocation of resources to threads. In our approach, we use an artificial neural network (ANN) because ANNs enjoy the applicable performance and flexibility in predicting complex behavior. Using three case studies, we contrasted results of the current and our previous approaches to demonstrate results. Copyright © 2015 John Wiley & Sons, Ltd.

Latest advances in parallel, distributed, and network‐based processing

“…The solution is carried out by aborting/recovering deadlocked threads, but this has the cost of wasting the resources used by the deadlocked threads till their deletion. Several prediction techniques have been presented to avoid such situations, for example, . Ghorbani and Babamir present a framework composed by two recurrent neural networks for deadlock prediction and resolution in resource allocation presenting an accuracy of more than 80%.…”

Runtime deadlock tracking and prevention of concurrent multithreaded programs: A learning‐based approach

“…A false deadlock prediction is an actual deadlock‐free state predicted as a deadlock‐prone. In this study, we presented an intelligent method to predict and resolve deadlocks where the number of false deadlock predictions was reduced over our previous study . Moreover, we addressed the weaknesses of other deadlock prediction methods.…”

“…In this paper, we extended our previous work to decide suitably on an action for resolving forecasted deadlocks in multithreaded programs. The method presented in the work of Babamir et al used a specific recurrent Neural Network (NN) called Nonlinear Auto Regressive Exogenous (NARX) just to predict future potential deadlocks at runtime. Based on the aforementioned work, we present a new configuration where two NNs are used for deadlock prediction and resolution in resource allocation.…”

“…The method presented in the work of Babamir et al used a specific recurrent Neural Network (NN) called Nonlinear Auto Regressive Exogenous (NARX) just to predict future potential deadlocks at runtime. Based on the aforementioned work, we present a new configuration where two NNs are used for deadlock prediction and resolution in resource allocation. Through this learning ‐based configuration to predict and prevention deadlocks, our method attempts to produce fewer false positive reports in prediction.…”

“…Therefore, the proposed method attempts to prevent deadlocks by an appropriate action just in cases that a deadlock probably occurs; when there is a low probability of a deadlock, it does not impose any additional overhead to prevent the predicted deadlock. In addition to the work of Babamir et al, we have two other experiences on using NARX in prediction of load balancing web servers for execution of concurrent requests.…”