实时系统时间分析的首要任务是估计程序的最坏情况执行时间(worst—caseexecutiontime,简称WCET).程序的WCET通常受到硬件体系结构的影响,Cache则是其中最为突出的因素之一.对面向WCET计算的Cache分析研究进行了综述,介绍了经典Cache分析框架与Cache分析核心技术,并从循环结构分析、数据Cache分析、多级Cache分析、多核共享Cache分析、非LRU替换策略分析等角度介绍了Cache分析在不同维度上的研究问题与主要挑战,总结了现有技术的优缺点,展望了Cache分析研究的未来发展方向.
The main task of real-time system design is to analyze the timing behaviors of a system at design time in order to guarantee that the given timing constraints are met at run time. The key issue is to estimate the Worst-Case Execution Time (WCET) of a program. Typically the WCET is heavily influenced by the hardware features of the target processor, among which Cache is the most influential factor. This article presents a survey on Cache analysis for WCET estimation. It introduces main research problems and challenges in different dimensions, such as the analysis of loops, data caches, multi-level caches, multi-core shared caches, non-LRU replacement policies, etc. The mainstream analysis techniques with their pros and cons are evaluated. An outlook for future research directions of Cache analysis is given in the end.