RTEMS simple example

Joel Sherrill <joel@OARcorp.com> joel.sherrill at OARcorp.com
Thu Sep 29 18:14:46 UTC 2005


Aleix Conchillo Flaque wrote:
> Hi,
> 
> I've been playing a bit with RTEMS and thought that someone could be
> interested in another example of RTEMS usage. I have taken the algorithms
> found in a public file around called "stanford.c" and separated them into
> tasks. The example uses quite a few features of RTEMS, such as semaphores,
> message queues, events, memory regions and partitions and some other
> things.

Neat.  Would you mind if it got merged into RTEMS examples?

> This is my first attempt at using RTEMS so I'm sure a lot of things could be
> done better and I would really appreciate if someone could point them
> out, and  thus make the example better and useful for other people
> wanting to learn RTEMS.

You might want to consider adding a simple menu to the bottom of the 
Init task which lets you print out the CPU usage or generate a stack 
report if you have that turned on.

> I have copied below the README file so you don't have to open the file to
> see the idea.

The ERC32 specific addresses is easy to solve.  We have this issue on
the RTEMS tests and solve it by declaring an array that is the memory
so the linker can easily account for it.  It's just a test/demo so not
a problem.  This is a hack at changing this.

static unsigned char partition_buffer[ 100000 ];
static void *partition_address = (void *) partition_buffer;
static unsigned int const partition_length = 100000;
static unsigned int const partition_buf_size = 10000;

static unsigned char region_buffer[ 20480 ];
static void *region_address = (void *) region_buffer;
static unsigned int const region_length = sizeof( region_buffer );
static unsigned int const region_page_size = 1024;




> The latest source tree can be found at:
> 
> http://arch.hacks-galore.org/cgi-bin/archzoom.cgi/aconchillo@ieec.fcr.es--rtems/stanford--dev--0--LATEST
>>From there you can download a tarball. Or, you can download the tarball
> directly from:
> http://hacks-galore.org/aleix/rtems/rtems-stanford.tar.gz
> 
> Thanks in advance for all the comments, critiques or whatever.
> 
> aleix
> 
> --
> Aleix Conchillo Flaque   Software Engineer
> aconchillo at ieec.fcr.es   Institut d'Estudis Espacials de Catalunya (IEEC)
>                          http://www.ieec.fcr.es
> 
> ----------------------------------------
> 
> RTEMS stanford example
> ----------------------
> 
> This is a  simple example to show some of the  features of RTEMS.
> 
> The application is based on three kinds of tasks:
> 
>         - wait task: this task  waits for two events, RTEMS_EVENT_0 or
>           RTEMS_EVENT_1 (see  RTEMS documentation for  event details).
>           RTEMS_EVENT_0 is  sent by application  tasks indicating that
>           they have finished.  RTEMS_EVENT_1 can only be sent from the
>           error  task and indicates  a fatal  error causing  the whole
>           application to stop. When RTEMS_EVENT_0 is received the wait
>           task  also decrements the  counter of  running tasks  in the
>           system. This counter is  controlled by a sempahore that with
>           the current  implementation is not  needed at all (as  it is
>           only used by  the wait task), but has been  kept to show the
>           RTEMS semaphore API.
> 
>         - error task:  the error task  is in charge of  receving error
>           messages  on a  queue. Two  types of  error messages  can be
>           received,  normal or fatal  errors. When  a normal  error is
>           received, the error tasks prints the acompanying description
>           in the  message (only if  debug is activated). If  the error
>           task  receives a fatal  error from  an application  task, it
>           automatically sends an EVENT_1 to the wait task.
> 
>         - application  tasks:  these  tasks  perform some  well  known
>           algorithms.   The  algorithms have  been  obtained from  the
>           public file stanford.c (which  is also included in the 'etc'
>           directory).   All of  the  application tasks  have the  same
>           priority, this means that if a task have the RTEMS_TIMESLICE
>           mode activated, it will share  the CPU with other tasks with
>           the same priority and with RTEMS_TIMESLICE mode activated.
> 
> Note that  this example has only  been tested in  an ERC32 environment
> and may not work in other  systems. For instance, the memory task uses
> specific  ERC32  memory  addresses  which  should be  changed  to  the
> specific ones in another system.
> 
> ***
> 
> The stanford.c benchmarks were  gathered by John Hennessy and modified
> by Peter Nye.
> 
> --
> Aleix Conchillo Flaque <aconchillo at ieec.fcr.es>
> 
> 
> 


-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
    Support Available             (256) 722-9985




More information about the users mailing list