How to Rapidly Prototype a Real-Time Scheduler

Abstract

Implementing a new scheduling algorithm in an OS kernel is often an important step in scheduling research because it permits evaluation of the algorithm's performance on real workloads. However, developing a new scheduler is not a trivial task because it requires sophisticated programming skills and a deep knowledge of kernel internals. In this paper we show how to use the HLS scheduling framework to develop new schedulers in a user-level simulator, where advanced debugging tools can be used to achieve a high level of robustness before the scheduler is converted to a loadable kernel module simply by recompiling it. Besides facilitating debugging and porting, the HLS abstraction has the benefit of bringing the programming model very close to what, in our experience, scheduler developers want.

Presented as a work in progress at the 23rd IEEE Real-Time Systems Symposium (RTSS 2002), Austin, TX, December 3-5 2002.

• PDF
• PS.gz
• source code

Copyright 2002 IEEE.