中断驱动的嵌入式软件性能动态评测方法

发布时间：2018-03-13 01:00

本文选题：嵌入式系统　切入点：动态测试　出处：《大连理工大学》2013年硕士论文　论文类型：学位论文

【摘要】：嵌入式系统的三个基本特征：资源紧凑有限、时间约束严格、成本极度敏感,并且嵌入式软件本身由于需要处理大量中断,许多嵌入式软件性能指标(系统吞吐量、任务处理能力、堆栈深度等)与中断密切相关,是衡量软件优劣、指导系统设计的重要依据。中断的不确定性和动态性使得这些动态性能指标测试困难。本文围绕当前嵌入式软件开发设计和可信度保障实际问题以及未来技术需求在嵌入式软件动态仿真验证系统的前期工作基础上展开：构建带有抢占优先权的中断排队模型,采用排队论方法对嵌入式软件进行性能评估,通过讨论排队系统稳态时的特点,分析系统吞吐率、中断丢失率、系统的平均中断处理时间、中断处理时间、CPU安全占有率等关键性能指标,对嵌入式软件动态性能进行评价；将遗传算法、决策树和状态变迁矩阵应用到堆栈深度检测中,从寻求产生最大堆栈深度的软件执行路径的角度详细分析堆栈使用原因和中断类型,建立中断调度模型,提出基于遗传算法的WCSD (Worst Case Stack Depth)动态检测方法,以降低最大堆栈深度检测的误差；研究更加稳定的WCSD检测方法,分析堆栈在软件指令区间的分布情况,构建多层中断叠加模型并基于该模型设计堆栈动态检测方法,以获得嵌入式软件堆栈深度上限。最后,在高性能集群平台和SPARC仿真技术上,有效集成上述研究成果,建立全数字仿真验证平台,快速准确评测嵌入式软件关键的性能指标,提高嵌入式软件的安全性。经过实验验证可知,上述研究成果具有很好的可信度和可用性以及实用意义。
[Abstract]:The three basic characteristics of embedded system are: limited resources, strict time constraints, extremely sensitive cost, and many embedded software performance indicators (system throughput) due to the need to deal with a large number of interrupts. Task processing ability, stack depth and so on are closely related to interrupt, which is an important basis for evaluating software quality and guiding system design. The uncertainty and dynamics of interrupt make it difficult to test these dynamic performance indexes. On the basis of the previous work of embedded software dynamic simulation verification system, this paper focuses on the actual problems of embedded software development, design and credibility assurance and the future technical requirements. It constructs an interrupt queue model with preemptive priority. The performance of embedded software is evaluated by queueing theory method. The system throughput, interrupt loss rate and average interrupt processing time are analyzed by discussing the characteristics of the steady state of the queueing system. The dynamic performance of embedded software is evaluated by interrupt processing time and CPU security share, and genetic algorithm, decision tree and state transition matrix are applied to stack depth detection. From the point of seeking the software execution path that produces the maximum stack depth, the reasons and types of stacks are analyzed in detail, the interrupt scheduling model is established, and the dynamic detection method of WCSD Worst Case Stack based on genetic algorithm is proposed. In order to reduce the error of maximum stack depth detection, a more stable WCSD detection method is studied, the distribution of stack in the software instruction interval is analyzed, and the stack dynamic detection method is designed based on the multi-layer interrupt superposition model. To get the upper limit of embedded software stack depth. Finally, in the aspect of high performance cluster platform and SPARC simulation technology, the above research results are effectively integrated, the full digital simulation verification platform is established, and the key performance indexes of embedded software are evaluated quickly and accurately. The experimental results show that the above research results have good reliability, usability and practical significance.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP311.53;TP368.1

【参考文献】