Re: [Xenomai] Hilscher driver for cifX boards

[Jerome Poncin <JPoncin@hilscher.com>]

Hello Jan,

Thank you for your answer. I checked /proc/interrupts and lspci -v, and I saw that the IRQ is shared.

"If yes, resolve that conflict."

=> What must I do ? I'm not sure to have understood ?

/*****************************************************************************/
/*!
*   \brief  cifx_handler
*
*   \param  [in]    irq      Resource task pointer
*
*   \return RTDM_IRQ_NONE or RTDM_IRQ_HANDLED
*
*   \note   cifx interrupt handler
*/
/*****************************************************************************/
static /*inline*/ int cifx_handler(rtdm_irq_t *irq)
{
     struct rtdm_device    *info             = (struct rtdm_device 
*)rtdm_irq_get_arg(irq, void);
     io_info_t             *device_data     = (io_info_t *) 
info->device_data;

     if ((device_data->priv != NULL) || (device_data->irq_registered == 
FALSE) || (device_data->irq_enable == FALSE))
     {
         /* This is a PLX device and cannot produce an IRQ */
         return RTDM_IRQ_NONE;
     }
     else
     {
         void __iomem *int_enable_reg = 
device_data->mem[DPM_INDEX].internal_addr + DPM_HOST_INT_EN0;
         void __iomem *int_status_reg = 
device_data->mem[DPM_INDEX].internal_addr + DPM_HOST_INT_STAT0;

         /* Is one of our interrupts enabled and active ? */
         if (!(ioread32(int_enable_reg) & ioread32(int_status_reg) & 
DPM_HOST_INT_MASK))
             return RTDM_IRQ_NONE;

         /* Disable interrupt */
         iowrite32(ioread32(int_enable_reg) & ~DPM_HOST_INT_GLOBAL_EN, 
int_enable_reg);

         return RTDM_IRQ_HANDLED;
     }
}

If IRQ is not mine, I return RTDM_IRQ_NONE, it's not correct ?

Thank you very much for your help,

Hilscher France
12, rue du 35ème Régiment d'Aviation
Miniparc du Chêne
69500 BRON
France
Tél. : +33 (0) 4 72 37 98 40
Fax  : +33 (0) 4 78 26 83 27
http://www.hilscher.fr
HILSCHER FRANCE 	Jérôme Poncin
jponcin@hilscher.com
Ingénieur Développement Logiciel
Tél. : +33 (0) 4 72 37 98 44





Le 26/02/2013 12:37, Jan Kiszka a écrit :
> On 2013-02-26 10:29, Jerome Poncin wrote:
>> Hello,
>>
>> I have a problem with driver for Hilscher cifx with IRQ.
>>
>> All driver run good (I verified in polling mode) but in case I activate
>> IRQ the system freeze.
>>
>> I checked parameters of "rtdm_irq_request" function without success (all
>> seems good) ! I tried to see on internet to get some information about
>> the use of "rtdm_irq_request", or compare my code with other driver but
>> it's not easy...
>> I give you in attached file my code because I have no idea about my
>> problem, and it's really difficult to have some information or trace
>> after system freeze...
>> Do you have an idea about my problem ?
>> static inline int cifx_handler(rtdm_irq_t *irq)
>           ^^^^^^
> But I suspect the compiler is smart enough to ignore this.
>
> Is the line you are registering for shared with other devices? Check
> /proc/interrupts and lspci -v. If yes, resolve that conflict.
>
> If not, add printk to your IRQ handler, redirect the kernel messages to
> a serial console and try to find out how far and how ofter you get into
> the handler.
>
> Well, and then there is the option of source-level debugging. kgdb for
> I-pipe patches aren't merged yet, but more convenient is KVM-based
> debugging anyway. See [1] for the concept. Just requires VT-d or AMD
> IOMMU on your target.
>
> Jan
>
> PS: Still no kernel coding style. Specifically the long lines makes it
> hard to read.
>
> [1]
> http://chemnitzer.linux-tage.de/2012/vortraege/folien/1061-VirtualDebugging.pdf
>