[Xenomai-help] Kernel access of bad area with rtdm_event_wait() (MPC8347)

[Xenomai-help] Kernel access of bad area with rtdm_event_wait() (MPC8347)

[Perrine Martignoni]

[-- Attachment #1: Type: text/plain, Size: 3248 bytes --]

Hello,

I work on a MPC8347 processor and I found out a problem with
rtdm_event_wait(). I reproduce the problem with a simple code :

*driver.c* :
#include <rtdm/rtdm_driver.h>

MODULE_AUTHOR("P.Martignoni");
MODULE_LICENSE("GPL");

#define DEV_FILE_NAME       "DspDev"
#define DRV_NAME            "DspDrv"

#define DEV_FILE_DSP            "dspdev0"

static rtdm_event_t s_ReadEvent;

static int Open(struct rtdm_dev_context    *p_Context,
                rtdm_user_info_t           *p_UserInfo,
                int                        Oflags)
{
    int ret = 0;

    rtdm_event_init(&s_ReadEvent,0);

    // generates an error : Oops : Kernel access of bad area, sig : 11 [#1]
    rtdm_event_wait(&s_ReadEvent);


    return ret;

}

static int Close(struct rtdm_dev_context   *p_Context,
                 rtdm_user_info_t          *p_UserInfo)
{

    rtdm_event_destroy(&s_ReadEvent);

    return 0;
}

/*****************************************************/
/*               Structures rtdm_device              */
/*****************************************************/
static struct rtdm_device t_IdspDevice = {
    struct_version:     RTDM_DEVICE_STRUCT_VER,
    device_flags:       RTDM_NAMED_DEVICE,
    context_size:       0,
    device_name:        DEV_FILE_DSP,

    open_rt:        NULL,
    open_nrt:       Open,

    ops:{
        close_rt:   NULL,
        close_nrt:  Close,
        ioctl_rt:   NULL,
        ioctl_nrt:  NULL,
        read_rt:    NULL,
        read_nrt:   NULL,
        write_rt:   NULL,
        write_rt:   NULL,
        recvmsg_rt: NULL,
        recvmsg_nrt:NULL,
        sendmsg_rt: NULL,
        sendmsg_nrt:NULL,
    },
    device_class:       RTDM_CLASS_EXPERIMENTAL,
    device_sub_class:   222,
    driver_name:        DRV_NAME,
    peripheral_name:    DEV_FILE_NAME,
    provider_name:      "-",
    proc_name:          t_IdspDevice.device_name,
};

/**************************************/
/*          Init Driver               */
/**************************************/
static int ddsp_Init(void)
{

    if(rtdm_dev_register(&t_IdspDevice)!=0)
        rtdm_printk("Enregistrement impossible\n");
    else
        rtdm_printk("Initialisation OK\n");

    return 0;
}

/*************************************/
/*           Exit Driver             */
/*************************************/
static void ddsp_Exit(void)
{

    if(rtdm_dev_unregister(&t_IdspDevice,0))
        rtdm_printk("Desabonnement KO\n");
    else
        rtdm_printk("Desabonnement OK\n");
}

module_init(ddsp_Init);
module_exit(ddsp_Exit);

*App.c* :
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include <sys/mman.h>
#include <native/task.h>

#define DEV_FILE_DSP            "dspdev0"
#define TASK_PRIO             50

static RT_TASK s_TaskTest;

int main(int argc, char * argv[])
{
    int ret = 0;

    mlockall(MCL_CURRENT | MCL_FUTURE);

    // set the currrent task in rt mode
    ret = rt_task_shadow(&s_TaskTest,NULL,1,0);

    rt_dev_open(DEV_FILE_DSP, 0);

    return ret;
}

When I launch this code, I have this error : *Oops : Kernel access of bad
area, sig 11 [#1]

*I notice that if the rtdm_event_wait() is called in a kernel task, there is
no problem.

Have you any idea for my issue ?

Thanks in advance.

[-- Attachment #2: Type: text/html, Size: 6500 bytes --]

Re: [Xenomai-help] Kernel access of bad area with rtdm_event_wait() (MPC8347)

[Jan Kiszka]

Perrine Martignoni wrote:
> Hello,
> 
> I work on a MPC8347 processor and I found out a problem with
> rtdm_event_wait(). I reproduce the problem with a simple code :
> 
> *driver.c* :
> #include <rtdm/rtdm_driver.h>
> 
> MODULE_AUTHOR("P.Martignoni");
> MODULE_LICENSE("GPL");
> 
> #define DEV_FILE_NAME       "DspDev"
> #define DRV_NAME            "DspDrv"
> 
> #define DEV_FILE_DSP            "dspdev0"
> 
> static rtdm_event_t s_ReadEvent;
> 
> static int Open(struct rtdm_dev_context    *p_Context,
>                 rtdm_user_info_t           *p_UserInfo,
>                 int                        Oflags)
> {
>     int ret = 0;
> 
>     rtdm_event_init(&s_ReadEvent,0);
> 
>     // generates an error : Oops : Kernel access of bad area, sig : 11 [#1]
>     rtdm_event_wait(&s_ReadEvent);

General advice: switch on XENO_OPT_DEBUG_RTDM (and IPIPE_DEBUG_CONTEXT)
to catch service invocations from a wrong context, like here.

Your mistake: calling a blocking Xenomai services from non-rt context
(see your registration of Open in t_IdspDevice).

Jan

-- 
Siemens AG, Corporate Technology, CT SE 2
Corporate Competence Center Embedded Linux

Re: [Xenomai-help] Kernel access of bad area with rtdm_event_wait() (MPC8347)

[Perrine Martignoni]

[-- Attachment #1: Type: text/plain, Size: 1522 bytes --]

Indeed, I have bad reproduced the problem in the code I suggest to you. But
your remark makes me understand  that I was in secondary mode and I never
switched to primary mode so ....
My issue is understood and solved.
Thanks

On Thu, Feb 21, 2008 at 4:23 PM, Jan Kiszka <jan.kiszka@domain.hid> wrote:

> Perrine Martignoni wrote:
> > Hello,
> >
> > I work on a MPC8347 processor and I found out a problem with
> > rtdm_event_wait(). I reproduce the problem with a simple code :
> >
> > *driver.c* :
> > #include <rtdm/rtdm_driver.h>
> >
> > MODULE_AUTHOR("P.Martignoni");
> > MODULE_LICENSE("GPL");
> >
> > #define DEV_FILE_NAME       "DspDev"
> > #define DRV_NAME            "DspDrv"
> >
> > #define DEV_FILE_DSP            "dspdev0"
> >
> > static rtdm_event_t s_ReadEvent;
> >
> > static int Open(struct rtdm_dev_context    *p_Context,
> >                 rtdm_user_info_t           *p_UserInfo,
> >                 int                        Oflags)
> > {
> >     int ret = 0;
> >
> >     rtdm_event_init(&s_ReadEvent,0);
> >
> >     // generates an error : Oops : Kernel access of bad area, sig : 11
> [#1]
> >     rtdm_event_wait(&s_ReadEvent);
>
> General advice: switch on XENO_OPT_DEBUG_RTDM (and IPIPE_DEBUG_CONTEXT)
> to catch service invocations from a wrong context, like here.
>
> Your mistake: calling a blocking Xenomai services from non-rt context
> (see your registration of Open in t_IdspDevice).
>
> Jan
>
> --
> Siemens AG, Corporate Technology, CT SE 2
> Corporate Competence Center Embedded Linux
>

[-- Attachment #2: Type: text/html, Size: 2345 bytes --]