[LITMUS^RT] Questions abiut rt_launch

Björn Brandenburg bbb at mpi-sws.org
Fri Jan 19 08:31:55 CET 2018


> On 19. Jan 2018, at 03:29, Ashraf E. Suyyagh <mrsuyyagh at gmail.com> wrote:
> 
> 
> 1. As it seems, launching these binaries using rt_launch only launches each one of them once. Whereas, the desired result is that these binaries be periodic similar to what is being done using rtspin. Am I missing something? Is there a technique where I can have my binaries run periodically using rt_launch. I consulted the documentation and there is no help there.

Achieving periodic activations (or any other activation pattern you want, or event-driven activation) is the responsibility of the process. rt_launch simply launches the process and moves it into the LITMUS^RT scheduling class. 

So if your binaries contain no periodic main loop, then they will not run periodically. Note that you really do *not* want to launch a new process for every activation (i.e., for every job / loop iteration); that would be way too costly. 

> 
> 2. Also, I am trying to draw the trace of my binaries. Even when they run once. Tracing does not work. It works perfectly with rtspin. It seems as if nothing is being recorded. I followed the same technique to record and draw a trace for both rtspin and rt_launch.

It should work just the same with rt_launch, rtspin, or any real task; the tracing framework is oblivious to how tasks are launched. 

If you want help with this, you’ll need to provide more details; just saying “tracing does not work” does not tell us enough to offer any useful advice.

Cheers,
Björn




More information about the litmus-dev mailing list