Understanding _io_read Errors

ttocsmi — Tue, 23 Sep 2014 20:34:08 GMT

I'm using MQX 4.0 and am using _io_read to read values from one the internal ADC modules on a MK10 mirco. My read logic is similar to:

if ( _io_read(...) ) {

do something

} else {

write error message to console

}

Every so often, the _io_read function returns an error value and I get an error message on the console output. I'm having trouble trying to understand what's causing the error & am hoping for some help. How can I translate the return value of the function to a pass/fail value and learn more about why the read failed?

Here's more info:

I'm reading from multiple ADC inputs, and it's not always the same channel which errors out
Only one task reads from the ADCs
I have higher-priority tasks running, so it's possible my ADC task could be pre-empted during the _io_read process. If so, how would I know this and how could I prevent it?
Is there a way to monitor task stack usage, or tell if a task is being continually pre-empted?

Thanks for the help!

Re: Understanding _io_read Errors

Alice_Yang — Thu, 25 Sep 2014 03:08:47 GMT

Hello Scott,

Can you show on the configuration about the ADC and the task , also the error ?

And you can refer to the adc sample under C:\Freescale\Freescale_MQX_4_1_1\mqx\examples (change 4.1 to 4.0, I use MQX4.1, ).

Hope it helps

Alice

Re: Re: Understanding _io_read Errors

ttocsmi — Tue, 07 Oct 2014 13:55:54 GMT

Thanks for the reply, Alice.

I'm not sure exactly what the "error" is, as I'm having difficulty reading and interpreting the return value of the _io_read function. When the "error" occurs, my code enters the "else" branch and I see the message I have within.

This looks to be the relevant code from the adc_demo.c file:

        /* channel 1 sample ready? */
        if (read(f_ch1, &data, sizeof(data) ))
            printf("ADC ch 1: %4d ", data.result);
        else
            printf("              ");

Looking at the comment line, if a sample cannot be read, it's because it is not "ready". Does this imply that there is not necessarily an error, just that it hasn't been acquired/processed yet? If so, then my interpretation of a "failed" read as an "error" is incorrect. Like I stated previously, this does not happen regularly or consistently on any one channel.

Is there a way to determine the reason why the sample isn't ready? What can I do to ensure that when requested, the sample will be ready?

Here are the configuration/initialization commands for my ADC:

#define ADC1 "adc1:" /* adc1 must be #1 as the inputs are wired to ADC 1 */

// ADC configuration parameters

#define ADC_HARDWARE_AVERAGING      16
#define NO_SAMPLE_IN_ONE_RUN     16 // number of samples in one run sequence (orig 32)
#define START_OFFSET    4400 // time offset from trigger point in us (orig 100); manually tweaked for improved accuracy
#define PERIOD      30000   // period in us (orig 30000)
#define SCALE_RANGE     0x10000 // scale range of result (not used now)
#define CIRCULAR_BUFFER_SIZE 16 // circular buffer size (sample count) (orig 32)
#define ADC_LONG_SAMPLE    2 // long sample time = 2 usec
#define ADC_REF_VREF    3300    // Vref = 3.3 volts
#define MY_TRIGGER ADC_PDB_TRIGGER    // logical trigger ID that starts this ADC channel

// global variables
MQX_FILE_PTR f0, f1;

// ADC device initialization structure
const ADC_INIT_STRUCT adc_init = { ADC_RESOLUTION_DEFAULT, /* resolution */
};

//
// AI_MUX_DA and AI_MUX_DA read together
//

const ADC_INIT_CHANNEL_STRUCT ai_mux_da_param = { AI_MUX_DA, // physical ADC channel
ADC_CHANNEL_MEASURE_ONCE | ADC_CHANNEL_START_TRIGGERED,
NO_SAMPLE_IN_ONE_RUN,
START_OFFSET,
PERIOD,
SCALE_RANGE,
CIRCULAR_BUFFER_SIZE,
MY_TRIGGER
};

const ADC_INIT_CHANNEL_STRUCT ai_mux_db_param = { AI_MUX_DB, // physical ADC channel
ADC_CHANNEL_MEASURE_ONCE | ADC_CHANNEL_START_TRIGGERED,
NO_SAMPLE_IN_ONE_RUN,
START_OFFSET+100, // problems occur if both reads occur simultaneously
PERIOD,
SCALE_RANGE,
CIRCULAR_BUFFER_SIZE,
MY_TRIGGER
};

Within the main task loop, I'm opening the ADCs

_mqx_int error_code = MQX_OK;

f1 = _io_fopen(ADC1, (const char*) &adc_init);
if (f1 != NULL) {
// long sample time
error_code = _io_ioctl(f1, ADC_IOCTL_SET_LONG_SAMPLE, (pointer)ADC_LONG_SAMPLE);
// hardware averaging
error_code = _io_ioctl(f1, ADC_IOCTL_SET_HW_AVERAGING, (pointer)ADC_HARDWARE_AVERAGING);
// reference voltage; not sure how this is used
error_code = _io_ioctl(f1, ADC_IOCTL_SET_REFERENCE,(pointer)ADC_REF_VREF);
// acquisition speed (was previously HIGH speed
error_code = _io_ioctl(f1, ADC_IOCTL_SET_LOW_SPEED, NULL);
} else {
_task_block();
}
_time_delay(OPEN_DELAY);

Then opening the individual channels

ch_ai_mux_da = _io_fopen(ADC1 "MUX_DA", (const char*) &ai_mux_da_param);
if (ch_ai_mux_da != NULL) {
} else {
_io_printf("Failed to open MUX_DA\n");
_task_block();
}
ch_ai_mux_db = _io_fopen(ADC1 "MUX_DB", (const char*) &ai_mux_db_param);
if (ch_ai_mux_db != NULL) {
} else {
_io_printf("Failed to open MUX_DB\n");
_task_block();
}

Then triggering

_io_ioctl(f1, ADC_IOCTL_FIRE_TRIGGER, (pointer) MY_TRIGGER);
_time_delay(TRIGGER_TIME_DELAY);

Then reading (analogLimitCheck is my limit check routine, the BYPASS... statement is the one which started this discussion)

if (_io_read(ch_ai_mux_da, &data, sizeof(data))) {
analogLimitCheck(...);
} else {
_io_printf("BYPASS reading ch_ai_mux_da\n");
}
if (_io_read(ch_ai_mux_db, &data, sizeof(data))) {
analogLimitCheck(...);
} else {
_io_printf("BYPASS reading ch_ai_mux_da\n");
}

Then closing the channels

_io_fclose(ch_ai_mux_da);
_io_fclose(ch_ai_mux_db);

And closing the ADC

_io_fclose(f1);
_time_delay(CLOSE_DELAY);

========

Please let me know if I'm missing anything. I think I got it all.

Thanks!

topic Re: Understanding _io_read Errors in MQX Software Solutions

Understanding _io_read Errors

Re: Understanding _io_read Errors

Re: Re: Understanding _io_read Errors