Re: [ECOS] Re: Réf. : Twothreads.c without cyg_thread_delay

[Gary Thomas <gary at mlbassoc dot com>]

Michele Portolan wrote:
> I found the answer: simply the example twothread.c does not explicetly 
> calls up cyg_scheduler_start inside cyg_user_start.
> After adding teh command scheduling works just fine!
Sounds like your platform has some problem (perhaps with tools or 
otherwise).  There should not need to be
any explicit call to start the scheduler - this happens when 
cyg_user_start() returns.

Certainly when I tested it last week (from the latest sources), there 
was no need to change the example program
other than the modifications I posted to light LEDs instead of prints 
and delays.
> Gary Thomas wrote:
> > Michele Portolan wrote:
> > > I have the same problems for eCos working on Leon2: I have no 
> > > context switches if the process is not suspended....
> > > Which interrupt whoudl I check? Thanks,
> >
> >   CYGNUM_HAL_INTERRUPT_RTC
> >
> > > Michele
> > >
> > > Gary Thomas wrote:
> > > > David Luca wrote:
> > > > > Timeslicing is enabled by default in eCos kernel.
> > > > > This is what I'm interested, to see two threads of
> > > > > same priority working. Using cyg_thread_delay doesn't
> > > > > mean much advantage over infinite loops because the
> > > > > tasks can be very long in working time before it
> > > > > reaches cyg_thread_delay. Its just a cooperative
> > > > > multitasking. Timeslicing is the reason I must use
> > > > > RTOS. If timeslicing is not working, then eCos is not
> > > > > a true RTOS. Anyone tested timeslicing?
> > > >
> > > > Of course!
> > > >
> > > > I just tried it.  I changed the thread function to simply
> > > > put a unique value into the platform LEDs - no delays:
> > > >
> > > > /* this is a simple program which runs in a thread */
> > > > void simple_program(cyg_addrword_t data)
> > > > {
> > > >   int message = (int) data;
> > > >
> > > >   printf("Beginning execution; thread data is %d\n", message);
> > > >   while (true) {
> > > >       HAL_SET_LEDS(1<<data);
> > > >   }
> > > > }
> > > >
> > > > Using the default scheduler (MLQUEUE), I changed the timeslice count
> > > > to be something that I can see (500 ticks = 5 seconds).  Running this,
> > > > I see first one value in the LEDS then a different value 5 seconds 
> > > > later.
> > > > The pattern repeats, ad nauseum.
> > > >
> > > > Timeslicing definitely works, as do eCos' other scheduling mechanisms.
> > > >
> > > > I believe that you're having trouble with your platform - 
> > > > interrupts aren't
> > > > working for your system clock.  Have you tried running the various 
> > > > kernel
> > > > tests?
> > > >
> > > > > --- Jean-Marc.Lopez@faiveleytransport.com wrote:
> > > > >
> > > > >
> > > > > ---------------------------------
> > > > > Hi David.
> > > > >  
> > > > >
> > > > > Excuse me but with your explanations, it's looks like
> > > > > you did not manage properly  the threads priorities.
> > > > >  
> > > > >
> > > > > It's just  a problem of Real time OS.
> > > > >  
> > > > >
> > > > > If your 2 threads had the same priority, and Thread_1
> > > > > never "give the hand" to the kernel, the Thread 2 will
> > > > > never run. …Except if you configure the timeSlicing
> > > > > option in your kernel. ( never done at this moment).
> > > > >  
> > > > >
> > > > > By using a cyg_thread_delay function, we said to the kernel to 
> > > > > stop the current Thread ( Thread 1 ) and to
> > > > > run the Thread 2 ( or the thread who have the higtest
> > > > > priority after the Thread1).
> > > > >  
> > > > >
> > > > > By setting the timeslicing option, the kernel will
> > > > > commute between the 2 threads every x ms ( depending
> > > > > of your kernel configuration) even if threads have the
> > > > > same priority. (but not experimented yet).
> > > > >  
> > > > >
> > > > >  
> > > > > LPZ
> > > > >
> > > > >  
> > > > > -----David Luca <davidluca3000@yahoo.com> a écrit :
> > > > > -----
> > > > >
> > > > > Pour : ecos-devel <ecos-devel@sourceware.org>,
> > > > > ecos-discuss <ecos-discuss@sourceware.org>
> > > > > De : David Luca <davidluca3000@yahoo.com>
> > > > > Date : 24/11/2006 14:29
> > > > > Objet : Twothreads.c without cyg_thread_delay
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Hello,
> > > > > Today I started twothreads.c application, but I
> > > > > replaced printf(not working yet on my platform) with
> > > > > leds. My question is, without the cyg_thread_delay, is
> > > > > it possible to run concurrently the two threads?
> > > > > Because if I comment this function, only one task is
> > > > > running, I can see that only one led is turn on and
> > > > > off.
> > > > > Thank you in advance,
> > > > > David Luca.


--
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------


--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss