基于二型模糊逻辑的多核程序数据竞争与死锁检测方法研究

With the prevalence of multi-core platforms, it attracts more and more attention to improve the reliability of multi-core programs. Common defects, e.g. data race and deadlock, may affect the reliability of multi-threaded programs on multi-core platforms. It is not easy for previous approaches to detect these kinds of defects. This project starts from the characteristics of data race and deadlock, focuses on runtime environment of multi-core platforms, and presents a Type-2 Fuzzy Logic based data race and deadlock detection approach. The research contents of this proposal include following parts: (1) This approach provides a static analysis pretreatment step for the target multi-threaded program to detect potential defect positions and potential defect conditions. (2) It takes important runtime environment factors as observations, and analyzes the probability distribution of the time sequences and the corresponding observation sequences of this program. (3) It constructs an interval Type-2 fuzzy logic based Hidden Markov Model for this program, and optimizes this time sequence analysis model. (4) It detects data race and deadlock problems based on Type-2 fuzzy logic. This project integrates Type-2 fuzzy logic with data race and deadlock detection to increase the accuracy, which is a new exploration in this area and a significant supplement of the previous works.

随着多核平台的迅速普及，提高多核软件可信性的研究日益受到重视。数据竞争和死锁是影响多核平台上多线程程序可信性的常见缺陷类型，传统方法对于这两类缺陷的检测仍存在不足。本项目从数据竞争和死锁问题的特点出发，关注多核平台上程序运行时的环境状态，提出基于二型模糊逻辑的多线程数据竞争和死锁检测方法。本项目的具体研究内容包括：(1)采用静态分析技术对待检测多线程程序进行预处理，确定潜在缺陷位置和缺陷发生条件；(2)评估运行时的环境因素并确定观察值，对不同观察值序列影响下执行时序的概率分布建模；(3)基于二型模糊逻辑构建待检测多线程程序的时序分析模型，研究模型状态空间优化技术；(4)基于二型模糊逻辑对待检测多线程程序进行缺陷检测。本项目将二型模糊系统理论应用于数据竞争和死锁检测中，以期显著地提高缺陷检测的精确度。该方法是该领域中的一项创新探索，也是对传统方法的一项有益补充。

随着多核平台的迅速普及，提高多核软件可信性的研究日益受到重视。数据竞争和死锁是影响多核平台上多线程程序可信性的常见缺陷类型，传统方法对于这两类缺陷的检测仍存在不足。本项目从数据竞争和死锁问题的特点出发，关注多核平台上程序运行时的环境状态，提出基于二型模糊逻辑的多线程数据竞争和死锁检测方法。本项目的具体研究内容包括：(1)采用静态分析技术对待检测多线程程序进行预处理，确定潜在缺陷位置和缺陷发生条件；(2)评估运行时的环境因素并确定观察值，对不同观察值序列影响下执行时序的概率分布建模；(3)基于二型模糊逻辑构建待检测多线程程序的时序分析模型，研究模型状态空间优化技术；(4)基于二型模糊逻辑对待检测多线程程序进行缺陷检测。本项目将二型模糊系统理论应用于数据竞争和死锁检测中，以期显著地提高缺陷检测的精确度。该方法是该领域中的一项创新探索，也是对传统方法的一项有益补充。