The real-time operating systems are wide used in every fields,and they are taken more seriously.In numerous real-time operating systems,operating systems base on Linux are produce a favourable response from the public for opened source codes and stable.But linux is not real real-time operating system,so we should improve real-time capability of the Linux.improving real-time capability of the Linux is hot at national and international.There are two ways to improve Linux's real-time performance.One way,you can modify Linux kernel directly and this way you make it so-called soft real time.the other way,you can adopt the so called sub-kernel idea,that is to implement an extra real-time kernel besides Linux kernel.The real-time kernel is in charge of managing and scheduling all the real-time tasks's and Linux kernel is treated as a real-time kernel's task with the lowest priority.This way,Linux can gain hard real-time performance.This thesis presents the author's work in research of the theory and method of real-time scheduling for his master degree.Most of the thesis focuses the method of real-time scheduling which can supports both hard real-time and soft real-time tasks.Firstly,thorough research real-time attemper arithmetic,put forward the EDF's improvement on this foundation,introduce function of the LST's abortion,make up the EDF's bug.Then a scheme is designed for implementing the integrated scheduling level in a kind of real-time Linux kernel.This scheme is based on a double kernel structure and basic priority scheduling.Consequently,it can support the scheduling of real-time systems.In the last part of this thesis,we designed an experiment scheme using the implemented system in Linux environment.the results from the experiment show that the real-time and soft real-time tasks can be integrated and scheduled,and the aim of supporting many kinds of real-time applications for the real-time systems in one processor is achieved. |