Real-Time

Summary by Bruce Alan Wynn

In order to support realtime data gathering and event handling, the authors have developed an enhanced Linux kernel. This kernel differs from the standard primarily by not allowing interrupts to be completely disabled; as a result, time-critical processes or tasks can access system resources when they are scheduled or required.

After an initial introduction to the concept of "realtime," this presentation focused on the methods which can be used with this new kernel to create time critical tasks as loadable kernel modules. The authors suggested that developers separate their application into those parts which are truly time critical and those which are not.

When a developer uses this model, realtime tasks must consume system resources in order to guarantee that they are not swapped out or otherwise unavailable when they need system resources. By keeping this in mind when designing an application, the developer can take advantage of the predictability of the realtime kernel.

Among other future goals, the authors hope one day to include the realtime kernel with the standard Linux distribution.

Additional information on the realtime Linux project can be found at https://luz.cs.nmt.edu/~rtlinux/.

Originally published in ;login: Vol. 22, No.2, April 1997.