Contact - RTEMS tests

[Gedare Bloom]

On Thu, Mar 14, 2013 at 2:19 PM, Gustavo Luiz Pasqualini
<pasqualini.gustavo at gmail.com> wrote:
> Hello,
>
> I am a student of Masters in Mechatronics by IFSC (Federal Institute of
> Science and Technology).
>
> My research includes testing different RTOS (hard and soft)
> The ideia is to load(port) in the same hardware these compatibles RTOS. The
> plataform is the Kit Luminary lm3s8962(Stellaris Arm Cortex - Texas) and
> after make the tests.
>
> Do you have any material / documentation / articles / monographies about
> RTOS for i attach in my work?
>
Comparisons between RTOSs can be subtle. For example, some RTOS
support fewer tasks than others, or have different sets of features
that are implemented. Keep an eye out for large differences and try to
be clear when explaining them.

> I need help to prepare the documentation of this work!
>
> Best Regards
>
> About the Tests:
>
> My idea is to load this RTOS in the Kit Luminary lm3s8962(Stellaris Arm
> Cortex - Texas) and make a lot of tests.
>
> The tests include a lot of things how to verify the latency of a RTOS, how
What do you mean by "latency of a RTOS"? In RTEMS we do have a suite
of tests located at rtems/testsuites/tmtests that are designed to
check the timing through various paths in the kernel.

> much non deterministic a RTOS is,
Similarly, what do you mean by non-deterministic? Some behavior is
inherently non-deterministic, e.g. interacting with I/O or standard
library calls that can sleep such as malloc.

> control a servo motor and simultaneously
> make calls for get the status of the interfaces of the kit and verify if the
> time of these processes(threads) are served, execute simultaneously
Do you plan to implement the servo motor driver as a part of your
application, or as a device in each of the RTOSs? Probably each RTOS
will have a different driver interface that may affect timing and
functionality.

> processes and verify the scheduler of the RTOS selecting the process by
What do you mean by "verify the scheduler"? I think just ensuring that
the right priority tasks run is not interesting... Maybe testing the
limits of the schedulers though, such as pushing the utilization
bounds for periodic schedulers such as rate monotonic and EDF.
Eventually the system overhead will cause the theoretical bound to be
unreachable.

> priorities, allocate memory(instances of objects) and verify the velocity of
> this job, etc.
>
What kind of allocators do you intend to examine? For example, RTEMS
provides a few different allocators such as Region and Partition. I'm
not sure if there are any standard real-time allocators, and comparing
different kinds of allocators with each other may not be informative.

-Gedare

> I will consider the same activity for different RTOSes!
>
> My research is public!!
>
> Thanks!!
> Gustavo Pasqualini
>
> _______________________________________________
> rtems-users mailing list
> rtems-users at rtems.org
> http://www.rtems.org/mailman/listinfo/rtems-users
>