[Xenomai-core] [bug] don't try this at home...

[Jan Kiszka <kiszka@domain.hid>]

[-- Attachment #1.1: Type: text/plain, Size: 691 bytes --]

Hi Philippe,

I'm afraid this one is serious: let the attached migration stress test
run on likely any Xenomai since 2.0, preferably with
CONFIG_XENO_OPT_DEBUG on. Will give a nice crash sooner or later (I'm
trying to set up a serial console now). To be more precisely:

Setup 1:
    xeno 2.0, kernel 2.6.13.5, ipipe-1.0-09
    on Thinkpad R40

Setup 2:
    xeno svn-trunk, kernel 2.6.14.3, ipipe-1.0-11
    on Thinkpad R40 & P-III embedded PC

I have no clue yet what happens, I just nailed various strange
behaviours and latencies down to this issue.

ALARM! ;)

Jan


PS: I suggest including this test for future regression checks into the
upcoming(?) testsuite.

[-- Attachment #1.2: nat_migration.c --]
[-- Type: text/plain, Size: 756 bytes --]

#include <stdio.h>
#include <sys/mman.h>
#include <semaphore.h>

#include <native/task.h>
#include <native/sem.h>

sem_t sem;


static void simple_thread(void *arg)
{
    sem_post(&sem);
}


int main (int ac, char **av)
{
    RT_TASK task;
    int     ret;


    mlockall(MCL_CURRENT | MCL_FUTURE);

    sem_init(&sem, 0, 0);

    while (1) {
        ret = rt_task_create(&task, "migration", 0, 1, 0);
        if (ret != 0)
            printf("(%s) error creating task: %d\n", __FUNCTION__, ret);

        ret = rt_task_start(&task, &simple_thread, NULL);
        if (ret != 0)
            printf("(%s) error starting task: %d\n", __FUNCTION__, ret);

        sem_wait(&sem);

        rt_task_delete(&task);
    }
}

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]