[ECOS] DSR not called from ISR (Real Time Clock)

[Satish Kumar <satish at bvt dot sc dot sanyo dot co dot jp>]

Hi,
   Due to ecos scheduler not gettin initialized ; the DSRs/pending DSRs
are not being invoked from the RealTimeClock::ISR routine, which is
entered everytime when a clock interrupt takes place.
           i never enter into unlock_inner() func; my sched_lock
identifier is undefined everytime and hence even the set_sched_lock() 
also fails to write in the new_value.
    can some one tell me what could be the reason for the scheduler not
gettin initialized..my Cyg_Schedule_Implementation::Schedule program in
the mlqueue.cxx is as follows:

Cyg_Thread *
Cyg_Scheduler_Implementation::schedule(void)
{
    CYG_REPORT_FUNCTYPE("returning thread %08x");

    // The run queue may _never_ be empty, there is always
    // an idle thread at the lowest priority.

    CYG_ASSERT( queue_map != 0, "Run queue empty");
    CYG_ASSERT( queue_map & (1<<CYG_THREAD_MIN_PRIORITY), "Idle thread vanished!!!");
    CYG_ASSERT( !run_queue[CYG_THREAD_MIN_PRIORITY].empty(), "Idle thread vanished!!!");

#ifdef CYGPKG_KERNEL_SMP_SUPPORT

    Cyg_Thread *current = get_current_thread();
    register cyg_uint32 index;

    CYG_ASSERT( current->cpu != CYG_KERNEL_CPU_NONE, "Current thread does not have CPU set!");

    // If the current thread is still runnable, return it to pending
    // state so that it can be considered alongside any other threads
    // for execution.
    if( current->get_state() == Cyg_Thread::RUNNING )
    {
        current->cpu = CYG_KERNEL_CPU_NONE;
        pending[current->priority]++;
        pending_map |= (1<<current->priority);
    }
    else
    {
        // Otherwise, ensure that the thread is no longer marked as
        // running.
        current->cpu = CYG_KERNEL_CPU_NONE;        
    }

    
    HAL_LSBIT_INDEX(index, pending_map);

    Cyg_RunQueue *queue = &run_queue[index];
    
    CYG_ASSERT( !queue->empty(), "Queue for index empty");
    CYG_ASSERT( pending[index] > 0, "Pending array and map disagree");

    Cyg_Thread *thread = queue->get_head();

    // We know there is a runnable thread in this queue, If the thread
    // we got is not it, scan until we find it. While not constant time,
    // this search has an upper bound of the number of CPUs in the system.
    
    while( thread->cpu != CYG_KERNEL_CPU_NONE )
        thread = thread->get_next();

    // Take newly scheduled thread out of pending map
    thread->cpu = CYG_KERNEL_CPU_THIS();
    if( --pending[index] == 0 )
        pending_map &= ~(1<<index);
    
#else    

    register cyg_uint32 index;

    HAL_LSBIT_INDEX(index, queue_map);

    Cyg_RunQueue *queue = &run_queue[index];
    
    CYG_ASSERT( !queue->empty(), "Queue for index empty");

    Cyg_Thread *thread = queue->get_head();

#endif
    
    CYG_INSTRUMENT_MLQ( SCHEDULE, thread, index);
    
    CYG_ASSERT( thread != NULL , "No threads in run queue");
    CYG_ASSERT( thread->queue == NULL , "Runnable thread on a queue!");
   
    CYG_REPORT_RETVAL(thread);

    return thread;
}


plz someone tell me how to solve this problem.

thanks
satish

-- 
Satish Kumar <satish@bvt.sc.sanyo.co.jp>


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