摘要
提出了一种延迟锁方法,以减小实时内核对周期性实时任务的响应时间。该技术的基本策略为:只让那些不影响实时任务的非实时任务进入非抢占临界区。为实现该策略,引入了定时器中断预测和锁持有时间的概念。此外,还设计了一种用于该策略的紧急定时器。最后将该策略原型在Linux2.6内核中实施,实验结果显示,在最差的情况下,Linux对实时任务的响应时间比原来降低了23%,同时,对非实时任务的响应时间比原来增加20%。
A new method called delayed lock is proposed,to reduce the latency of real-time task in real-time kernel.The basic rule of this method is that a non-real-time task entering a critical section is allowed only if the operation does not disturb the future execution of the real-time application.To execute this rule,this paper introduced the concepts of timer interrupt prediction and lock hold time.In addition,the paper also designed and implemented a new urgent timer for this technique,and the prototype on Linux 2.6.Experimental results show that the worst-case Linux＇s latency to real-time process is reduced to 23% ,at the expense of slowdown of the non-real-time process by 20%.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第16期90-92,196,共4页
Computer Engineering and Applications
