基于S698芯片的软件故障注入技术的研究

发布时间：2019-07-08 20:49

【摘要】：在诸如航空航天、医疗卫生、工业生产、国防军事等对于计算机可靠性有着严格要求的领域，研究设计容错计算机以及带有容错机制的软件应用具有非常重要的意义。而故障注入技术通过引入故障，加速系统失效，能够在短时间内有效地评估容错计算机系统的容错性能，正是辅助容错计算机系统设计、实现必不可少的有力工具。本文通过对S698芯片以及当前软件故障注入技术的研究，设计实现了一款能够对运行于S698芯片之上的软件系统进行关键寄存器修改、代码段破坏、进入非法工作区、进入死循环、软件不喂狗、接口器件工作异常等类型故障的注入工具。该工具利用串口通信的方式，根据用户在普通PC机上通过控制界面配置的故障注入参数形成故障注入命令，经由串口发送给以S698芯片为处理器的嵌入式系统之上，在该系统的串口中断处理程序中，根据接收成功的故障注入命令，执行所需类型故障的注入，并通过该串口发回相应的反馈。此外，该工具利用定时器的定时功能，实现了对寄存器被修改等具有间歇故障意义的故障类型进行间歇性的故障注入。在进行寄存器故障注入时，用到了Sparc处理器特有的寄存器窗口旋转的特性，通过旋转窗口，找到了在中断过程中被自动保存的寄存器的存储位置，通过修改其内容成功地实现了故障值的注入，，同时对于没有被自动保存的寄存器，通过转到其所在窗口的方式成功地实现了故障注入。最后，本文使用已实现的软件故障注入工具对一个运行于S698芯片之上的矩阵运算程序进行了故障注入实验，并根据实验结果分析了各类型故障对目标系统的影响程度。
[Abstract]:In areas such as aerospace, medical and health, industrial production, national defense and military, which have strict requirements for computer reliability, it is of great significance to study and design fault-tolerant computers and software applications with fault-tolerant mechanisms. By introducing fault and accelerating system failure, fault injection technology can effectively evaluate the fault-tolerant performance of fault-tolerant computer system in a short period of time, which is an indispensable and powerful tool to assist fault-tolerant computer system design and realize it. Based on the research of S698 chip and the current software fault injection technology, this paper designs and implements an injection tool which can modify the key register, destroy the code segment, enter the illegal working area, enter the dead cycle, the software does not feed the dog, and the interface device works abnormally. By means of serial communication, the tool forms a fault injection command according to the fault injection parameters configured by the user on the ordinary PC, and sends it to the embedded system with S698 chip as the processor through serial port. In the serial port interrupt processing program of the system, according to the successful fault injection command, the required type of fault injection is executed, and the corresponding feedback is sent back through the serial port. In addition, the tool makes use of the timing function of timer to realize intermittent fault injection to fault types with intermittent fault significance, such as register modification. When the register fault injection is carried out, the characteristic of register window rotation unique to Sparc processor is used. Through the rotating window, the storage location of the register which is automatically saved in the interrupt process is found, and the fault value injection is successfully realized by modifying its content. At the same time, for the register that is not automatically saved, the fault injection is successfully realized by going to its window. Finally, the fault injection experiment of a matrix operation program running on S698 chip is carried out by using the implemented software fault injection tool, and the influence degree of each type of fault on the target system is analyzed according to the experimental results.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06

【参考文献】