Research And Implementation Of Fault-tolerant Multiprocessor Systems In Parallel

With the development of science and technology, the degree of the industrial production automation is raised day by day; the automatic control system is complicated day by day too. Not only has put forward higher and higher requirement to the processing speeds of the control system, and, requiring the system at the same time has very high safe reliability and accuracy in time. It requires control system not only at hardware, but also especially software should be real-time character, and be very fast in processing the mistakes at the same time. Under the apparent kinds of prerequisite this, parallel compute technology and multi-processors technology are widely used, and merge each other. Fault-tolerant technology means the tolerance of fault. System can detect fault and recover the system automatically when mistake taking place. When presenting some designated hardware troubles or software mistakes, the system can still carry out a group of procedures, the procedure will not discontinue or revise because of trouble in the system in other words, and the result will not be affect by the mistake that is caused by fault in the system. How fault-tolerant theory working in parallel mutli-processors system is main content of this text.This text tells the theories of multi-processors technology, parallel computing and fault-tolerant technology which are needed to analyze the parallel multi-processors system fault-tolerant at first. And then combine the system which is used; expound the hardware foundation of parallel multi-processors system fault-tolerant and the concrete method of software programming. Focal point of this text promptly in how to realize the software programming in the fault-tolerant of parallel multi-processors system. The article is from two respects of the real-time task of the system and backstage supporter's task separately, have expounded separately that how to realize the ways of real-time fault-tolerant and self-examination by oneself.Real-time fault-tolerant detect the function modules of CPU, bus, I/O equipment on bus, peripheral sensor, communication in all CPUs and input/output data working state at any time when system works in normal running. This is especially a requirement for considering systematic real-time character. According to examines the result judges each function module in turn whether appears the breakdown, or will lead to the whole system appear unstable state. Software notes the fault; system monitor read the notes...