[LITMUS^RT] Handling aperiodic tasks

[Jonathan Herman]

Hi John,

Have you considered using the "phase" parameter to rtspin? This specifies a
delay for the first release of an rtspin process. You could start a ton of
rtspin processes with random phases over some test interval (say 20
seconds), call release_ts,  and the tasks would join the system at random
points over those 20 secons.


On Thu, Nov 15, 2012 at 5:43 AM, John Gamboa <vaulttech at gmail.com> wrote:

> Hello!
>
> I'm sure this question has a very simple answer, but I didn't find a
> solution yet. I'm trying to implement a "Slot Shifting plugin" for Litmus.
> In the plugin, I should be able to handle aperiodic tasks, which, as
> expected, should come at any time during the schedule.
>
> In order to start these tasks, I was planning to create a program to run
> along with the schedule. It would sleep for a certain amount of time and
> then start a new "rtspin". I initially planned to make this program "non
> Real Time", but then I though "what if it should wake up at some point in
> time, but it doesn't because the processor is so loaded that only Real Time
> tasks are running then?". Then I though "I'll make the program Real Time",
> but then I realized that the rtspins wouldn't be Real Time until they reach
> a certain point in the program (where they make the
> "task_mode(LITMUS_RT_TASK)" call).
>
> I believe the children of a Real Time task are not Real Time, right? (I
> didn't find a way to test it, but it would complicate a lot my "admission
> tests" :-))
>
> Do you have any suggestion? I though maybe my Real Time program could
> itself change its children tasks to Real Time, but then wouldn't the
> "task_mode()" call fail? (anyway, this is probably what I'll try, if I
> don't find an easier solution)
>
> Thanks for your attention,
>
> --
> John Gamboa
> rabanetescebolas.blogspot.com
>
> _______________________________________________
> litmus-dev mailing list
> litmus-dev at lists.litmus-rt.org
> https://lists.litmus-rt.org/listinfo/litmus-dev
>
>


-- 
Jonathan Herman
Department of Computer Science at UNC Chapel Hill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20121115/9a802ca0/attachment.html>