Re: [rtl] Workng with the example

[Emili Badia <ebadia@nil.fut.es>]

Hi. I've been working a little with the scheduler, trying to introduce
deadline control. I'm with the rate monotonic scheduler.
To prevent from power off the system without a shutdown when the rt_tasks
don't leave the processor, i've modified the scheduler introducing a
function that kills all the tasks except the null. With the request_RTirq I
install a handler for the serial line. When any character arrives, the
serial handler is executed and there i call the function that kills the
rt_task in execution.
This way i prevent to stop the computer bye the wrong way and can see the
results with the crono application for debug.
By this way, trying intentionally that a task interrupt another I could see
it has success. 


At 10:57 PM 23/06/98 -0400, you wrote:
>mshsaint wrote:
>> 
>> mshsaint wrote:
>> > > Now I am looking at this 2tasks examples and there are some things I
>> don't
>> > > understand:
>> > > And another thing due to the output I have got is:
>> > > Does the output really have a nice represntation of the premprtion. It
>> > > seemes to me that the lower process priority never gets preempted by
>> the
>> > > higher one, since the higher is asleep...
>> >
>> cecil wrote:
>> > My understanding of the demo is different: each task runs to completion
>> > without interruption, and each task is called on a strict schedule and
>> > runs at a precise time.
>> 
>> so what is all the code regarding the preemption about? The code is in
>> rt_sched.c...
>> 
>> Assuming we have two tasks
>> Task 1,  with a high priority
>> Task 2,  with a lower priority
>> 
>> With the code we have in hand shouldn't task2 be interrupted if the waiting
>> time of task1 is over,
>> say task1 is supposed to run every 10 seconds and task2 takes two seconds
>> to run and it starts running at 9. shouldn't task2 be interrupted by task
>> 1( assuming the last time task1 was executed is at 0)?
>> Thanx
>> M
>
>I've never tried this, since it never occured to me that a task should
>be interrupted while running the applications in which I'm interested.
>Does someone have a practical example?
>
>I'm most interested in running a single small task from 100 to 10,000
>times per second, or possibly two or three tasks of equal priority in
>strict rotation up to 1000 times per second each.
>
>The 2tasks demo is ambiguous on this issue, since each task has plenty
>of time to complete and then go to sleep. The tasks don't need to
>interrupt each other. 
>
>There is danger if a task isn't allowed enough time to complete before
>it has to start again or yield to another. Again, I've never tried it
>intentionally. I did get tired of rebooting my system and waiting for
>fsck to finish when I did it out of ignorance.
>
>Cecil Ogg
>
>--- [rtl] ---
>For more information on Real-Time Linux see:
>http://www.rtlinux.org/
>
Emili
--- [rtl] ---
For more information on Real-Time Linux see:
http://www.rtlinux.org/