From Thread to Pool: Modeling and Static Detection of Thread Pool Misuses in Java Programs
TL;DR
PoolGuard, a static analysis framework for detecting both executor-generic lifecycle/result-handling defects and ThreadPoolExecutor-specific configuration misuses, is developed and applied to large-scale open-source systems.
Abstract
Asynchronous programming is fundamental to modern software performance, and Java ExecutorService-based executors have become a standard abstraction for managing concurrent tasks. While existing research has extensively studied traditional concurrency bugs, management misuses of executors remain under-analyzed. These defects arise not from the business logic of asynchronous tasks, but from how executors are configured, used, and terminated. We present a taxonomy of seven executor-management misuse patterns and formalize them using a resource-management model. Based on this, we develop PoolGuard, a static analysis framework for detecting both executor-generic lifecycle/result-handling defects and ThreadPoolExecutor-specific configuration misuses. Extensive evaluation on a benchmark suite with real-world defects shows that PoolGuard is better than specialized async detectors as well as general-purpose analyzers such as SpotBugs and Infer. Applying PoolGuard to large-scale open-source systems, we identified 50 previously unknown issues, 24 of which have already been fixed or confirmed by developers, demonstrating the practical value of semantic analysis for executor-management defects.