基于模拟器的故障注入冗余优化和执行加速技术研究
发布时间：2019-02-26 16:33
【摘要】：日益增长的软硬件系统规模和系统可靠性需求对软件开发测试提出了更高的要求。其中故障注入是衡量系统容错能力和抗干扰能力的一种重要的测试手段,在系统中引入故障值或信号可以模拟真实环境下的外部干扰和系统扰动,达到检测异常状态下系统容错机制或恢复机制的目的。故障注入技术不断成熟的今天,特别是软件实现或模拟实现的故障注入技术中,却仍存在着注入精度低、注入性能差、测试周期长的问题。本文借助模拟器容易对程序进行运行控制和跟踪的特点,在模拟器环境下对故障注入进行了研究和优化,来实现在高注入精度下的高性能故障注入,达到提高故障注入测试效率,缩短测试开发周期的目的。本文的研究工作如下:1)研究了故障注入点上的读写行为特点,发现了故障注入过程中存在的冗余注入现象,针对该现象提出了指令级故障注入模型,以及在故障规则设计阶段进行优化的最小化注入原则;2)基于指令级故障注入模型,研究实现了在模拟器环境下的运行时和静态去冗优化注入方式,设计实现了基于共享内存的日志记录方式来解决程序因故障注入而崩溃时日志丢失的问题;3)研究实现了模拟器环境下针对软件故障的、基于指令注入和编译技术的故障注入执行加速技术;4)对上述功能进行了系统性的测试,验证了本文工作的有效性。
[Abstract]:The increasing software and hardware system scale and system reliability requirements put forward higher requirements for software development and testing. Fault injection is an important testing method to measure the fault tolerance and anti-jamming ability of the system. The introduction of fault value or signal into the system can simulate the external disturbance and system disturbance in the real environment. The system fault-tolerant mechanism or recovery mechanism under abnormal state detection is achieved. Nowadays, the fault injection technology is becoming more and more mature, especially in the fault injection technology realized by software or simulation, there are still some problems, such as low injection precision, poor injection performance and long test period. With the help of the characteristic that the simulator is easy to control and track the program, this paper studies and optimizes the fault injection under the simulator environment, so as to realize the high performance fault injection under the high injection precision. To improve the efficiency of fault injection testing, shorten the test development cycle. The research work in this paper is as follows: 1) the characteristics of read-write behavior at fault injection point are studied, and the redundant injection phenomenon in fault injection process is found, and a command-level fault injection model is proposed for this phenomenon. And the minimum injection principle of optimization in the design phase of fault rules; 2) based on the instruction-level fault injection model, the runtime and static de-redundancy optimization injection methods in simulator environment are studied and implemented. The method of log recording based on shared memory is designed and implemented to solve the problem of log loss when the program crashes due to fault injection. 3) the fault injection execution acceleration technology based on instruction injection and compilation technology for software fault in simulator environment is studied and implemented. 4) the above functions are systematically tested to verify the effectiveness of the work in this paper.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.5
本文编号：2430949
[Abstract]:The increasing software and hardware system scale and system reliability requirements put forward higher requirements for software development and testing. Fault injection is an important testing method to measure the fault tolerance and anti-jamming ability of the system. The introduction of fault value or signal into the system can simulate the external disturbance and system disturbance in the real environment. The system fault-tolerant mechanism or recovery mechanism under abnormal state detection is achieved. Nowadays, the fault injection technology is becoming more and more mature, especially in the fault injection technology realized by software or simulation, there are still some problems, such as low injection precision, poor injection performance and long test period. With the help of the characteristic that the simulator is easy to control and track the program, this paper studies and optimizes the fault injection under the simulator environment, so as to realize the high performance fault injection under the high injection precision. To improve the efficiency of fault injection testing, shorten the test development cycle. The research work in this paper is as follows: 1) the characteristics of read-write behavior at fault injection point are studied, and the redundant injection phenomenon in fault injection process is found, and a command-level fault injection model is proposed for this phenomenon. And the minimum injection principle of optimization in the design phase of fault rules; 2) based on the instruction-level fault injection model, the runtime and static de-redundancy optimization injection methods in simulator environment are studied and implemented. The method of log recording based on shared memory is designed and implemented to solve the problem of log loss when the program crashes due to fault injection. 3) the fault injection execution acceleration technology based on instruction injection and compilation technology for software fault in simulator environment is studied and implemented. 4) the above functions are systematically tested to verify the effectiveness of the work in this paper.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.5
【参考文献】
相关期刊论文 前3条
1 李华旺,刘海涛,杨根庆;航天单粒子事件故障注入系统研究[J];量子电子学报;2002年01期
2 孙峻朝,王建莹,杨孝宗;故障注入方法与工具的研究现状[J];宇航学报;2001年01期
3 王建莹,孙峻朝,杨孝宗;容错计算机系统可靠性评估工具:HFI-2故障注入器[J];电子学报;1999年11期
相关博士学位论文 前1条
1 潘庆和;软件故障注入关键技术研究[D];哈尔滨工业大学;2011年
相关硕士学位论文 前1条
1 赵旭林;基于DSP模拟器的外设并行与运行监控模型研究[D];浙江大学;2015年
，本文编号：2430949
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2430949.html
