Real-Time Performance Modification Of Windows CE .NET

Windows CE .NET from Microsoft Corp. is a very popular embedded Operated System (OS). It can use the applicaiton developping tool eVisual C++ and the great developping platform Platform Builder, whose modularizing components can satisfy different environment according to different needs. With a wide applicattion field, Windows CE .NET occupys a highest position in embedded OS market. However, there is still some shortage in it. Its kernel is so huge and complicated that it can not be stored into small spaces. The real-time performance of its kernel is also too week to be competent in some high real-time performance required area.Aimed at these problems, we provide a project to cut down a tiny kernel of Windows CE .NET, which is named as "Mini Windows CE". This tiny kernel needs a smaller size, higher real-time performance and more adapt to those special fields with small spaces and high needs of real time .Firstly, we have a discussion about the key-problems of embedded OS's real-time performance. Avoiding some lack of Windows CE .NET and improving its real-time performance, we present two methods: modifying interrupt mechanism and modifying threadscheduling machanism.Secondly, several relative modules about real-time performance of Windows CE .NET have been anatomized. Based on a deep understanding of the compiling system Platform Builder, we organize the whole original code about the system kernel, especially having an analysis of both interrupt machnism and thread sheduling machanism.Then, we respectively describe and implement both of the above modifying methods. When system finds an interrupt, the processing will lead to two parts: kernel ISR (Interrupt Service Routine) and user IST (Interrupt Service Thread). Optimizing the protection in the time of interrupt will reduce ISR delay and IST delay. Improving the timer definition will shorten the thread's waiting time from 2ms to 1ms, which means reducing ISR delay. Wnidows CE .NET schedules threads with a higher priority to run first. Threads at the same priority level run in a round-robin fashion with each thread receiving a quantum. We break those...