[Xenomai-help] Detect switch in secondary mode + mlockall = bug?

[Xenomai-help] Detect switch in secondary mode + mlockall = bug?

[Charles Clerdan]

[-- Attachment #1: Type: text/plain, Size: 1056 bytes --]

Hi,

I'm trying to catch the signal SIGXCPU, in order to know when a task goes in
secondary mode.
I would like to know, why I have to comment the fonction mlockall.

PS : The problem is not related to the Code, but, there is a piece of it :

#include "config.h"
#include <native/task.h>
#include <stdio.h>
#include <assert.h>
#include <signal.h>


void WorkingTask (void* cookie)
{
   while (1) {};
}

static RT_TASK Rt_Work_Task;

void go (void * cookie)
{
   printf("Test for switch : %d\r\n",(int)rt_task_self());
}

  void warn_upon_switch(int sig __attribute__((unused)))
{
   printf("/!\\ Secondary Mode /!\\ : ");
}

int main (void)
{
   RT_TASK Rt_main_task;
   int erreur;

   /* Pagination Locked */
  // mlockall( MCL_CURRENT | MCL_FUTURE );   <- This is the line which makes
my question


    signal(SIGXCPU, warn_upon_switch);

    /* Real-Time Task */
    rt_task_spawn (&Rt_main_task, "go", 2048, 20, T_FPU | T_JOINABLE,
&go,&Rt_main_task);

    /* Waiting for the end */
    rt_task_join(&Rt_main_task);

    return 0;
}

Regards,

Charles

[-- Attachment #2: Type: text/html, Size: 1350 bytes --]

Re: [Xenomai-help] Detect switch in secondary mode + mlockall = bug?

[Gilles Chanteperdrix]

Charles Clerdan wrote:
> Hi,
> 
> I'm trying to catch the signal SIGXCPU, in order to know when a task goes in
> secondary mode.
> I would like to know, why I have to comment the fonction mlockall.

http://www.xenomai.org/index.php/Request_for_information

-- 
                                                                Gilles.

Re: [Xenomai-help] Detect switch in secondary mode + mlockall = bug?

[Gilles Chanteperdrix]

Gilles Chanteperdrix wrote:
> Charles Clerdan wrote:
>> Hi,
>>
>> I'm trying to catch the signal SIGXCPU, in order to know when a task goes in
>> secondary mode.
>> I would like to know, why I have to comment the fonction mlockall.
> 
> http://www.xenomai.org/index.php/Request_for_information

Also, does the latency test run on your platform? Because if it does, it
shows that the bug is elsewhere: latency contains a call to mlockall and
registers a signal handler for SIGXCPU.

And if you want us to help you, you should tell us what happens if you
do not comment mlockall, because we can not guess it. See, we run
latency all the time, and obviously, we have no issue.

-- 
                                                                Gilles.