MQX hangs in _esdhc_status_wait

[Fabi]

Hello,

From time to time I noticed, that MQX blocks in the following loop:

static uint_32 _esdhc_status_wait

    (

        /* [IN] Module registry pointer */

        SDHC_MemMapPtr esdhc_ptr,

        /* [IN] Mask of IRQSTAT bits to wait for */

        uint_32         mask

    )

{

    uint_32             result;

    do

    {

        result = esdhc_ptr->IRQSTAT & mask;

    }

    while (0 == result);

Because IQRSTAT == 0 and mask == 2. The problem is, tha all tasks are stopped and no external diagnostic is possible if this comes up.

Is it a SD card or a MQX issue? Any ideas?

[soledad]

Hi F. Bilz,

Did you test the sdcard example? This is located at  the path: Freescale_MQX_4_1/mfs/examples/

Regards

Sol

[Fabi]

We have activated the HW watchdog as work around. Here's the code snipped:

void init_hw_wdt ( void )

{

    _int_disable();

    // Write sequence within 20 clock cycles to unlock register

    WDOG_UNLOCK   = 0xC520;

    WDOG_UNLOCK   = 0xD928;

    WDOG_TOVALH   = 1000;   // about 7 seconds

    WDOG_STCTRLH |= WDOG_STCTRLH_WDOGEN_MASK;

    WDOG_REFRESH = 0xA602;

    WDOG_REFRESH = 0xB480;

    _int_enable();

}

void hw_wdt_reset ( void )

{

    _int_disable();

    WDOG_REFRESH = 0xA602;

    WDOG_REFRESH = 0xB480;

    _int_enable();

}

@freescale: Could you integrate controller specific code for the hardware watchdog into future MQX releases?

[matthewkendall]

Not sure what processor you are on but you will likely find there is a Processor Expert driver for the watchdog that integrates with MQX. On a K60 we are using the Processor Expert WatchDog_LDD driver for the EWM peripheral.

[matthewkendall]

I too have seen it hang on that line when there is a hardware error on the bus (we suspect an EMI-caused glitch). If you look carefully at the eSDHC code you will be horrified at the number of such infinite loops with no timeout or other error recovery mechanism.