hw/bsp/dwm1001-dev Add LIS2DH12 accelerometer support

[seijikun]

Hey guys,

thanks for this awesome piece of software!
This PR is an attempt at adding support for the LIS2DH12 accelerometer sensor, which is integrated with DWM1001-DEV boards.
I'm fairly new to mynewt, so this PR is more like a shot in the dark, and mostly copied from other BSPs with the same accelerometer.
The sensor_itf.si_ints values are chosen, such that it does not crash - which does not mean that it's working ;). Would be good if somebody with more hardware knowledge could take a look at them (the .active = true certainly isn't true, this should be something like HAL_GPIO_TRIG_RISING or HAL_GPIO_TRIG_FALLING).

Here is an example application:

#include <assert.h>
#include <string.h>
#include <stdio.h>
#include <math.h>
#include "sysinit/sysinit.h"
#include "os/os.h"
#include "bsp/bsp.h"
#include "hal/hal_gpio.h"
#include "hal/hal_bsp.h"
#include "hal/hal_i2c.h"
#include "mcu/nrf52_hal.h"
#ifdef ARCH_sim
#include "mcu/mcu_sim.h"
#endif

#include <config/config.h>

#include <sensor/sensor.h>
#include "sensor/accel.h"
#include <lis2dh12/lis2dh12.h>

#if MYNEWT_VAL(DW1000_DEVICE_0)
#include <dw1000/dw1000_hal.h>
#endif

int accel_data_cb(struct sensor* sensor, void* a, void* databuf, sensor_type_t sensorType) {
    if(!databuf) { return SYS_EINVAL; }
    struct sensor_accel_data* sad = (struct sensor_accel_data*) databuf;
    if (!sad->sad_x_is_valid || !sad->sad_y_is_valid || !sad->sad_z_is_valid) { return SYS_EINVAL; }
    printf("%s: [ secs: %ld usecs: %d cputime: %u ]\n",
                   "LISTENER_CB",
                   (long int)sensor->s_sts.st_ostv.tv_sec,
                   (int)sensor->s_sts.st_ostv.tv_usec,
                   (unsigned int)sensor->s_sts.st_cputime);

    printf("x = %f, y = %f, z = %f\n", sad->sad_x, sad->sad_y, sad->sad_z);
    return 0;
}

static struct sensor_listener sensor_listener = {
    .sl_sensor_type = SENSOR_TYPE_ACCELEROMETER,
    .sl_func = accel_data_cb,  
    .sl_arg = NULL
};

 

int main(int argc, char **argv){
    sysinit();

    struct sensor* accel = sensor_mgr_find_next_bydevname("lis2dh12_0", NULL);
    assert(accel != NULL);

    assert(0 == lis2dh12_set_full_scale(&accel->s_itf, LIS2DH12_FS_2G));
    assert(0 == lis2dh12_set_op_mode(&accel->s_itf, LIS2DH12_OM_HIGH_RESOLUTION));

    sensor_set_poll_rate_ms(accel->s_dev->od_name, 1000 / 50);
    sensor_register_listener(accel, &sensor_listener);

    while (1) {
        os_eventq_run(os_eventq_dflt_get());
    }
    assert(0);
    return 0;
}

[mlaz]

Sounds like you need to look at the module's (DWM1001, not DWM1001-dev) schematic there you should find all the pinout info.
I'd advise that if you need any help it might be easier if you reach out on slack before posting PRs.

[kasjer]

Minor style (extra spaces and line) changes would be nice.
restriction should be changed to allow turning sensor on and off.

[kasjer]

extra empty line

[kasjer]

in mynewt code it is custom not to put space after finishing ) in casts like this

dev = (struct os_dev *)os_dev_open(....

[kasjer]

extra space after cast expression

[kasjer]

if it was:

syscfg.restrictions:
    - "!LIS2DH12_ONB || (I2C_0 == 1 && I2C_0_PIN_SCL == 28 && I2C_0_PIN_SDA == 29)"

sensor could be turned on and off as syscfg value implies it could.
notice extra I2C_0 == 1 that is also needed I guess

[kasjer]

I would remove this second table entry, first one seems to be correct

[kasjer]

if it was changed to:

pkg.init.LIS2DH12_ONB:

whole config_lis2dh12_sensor() function could be ifdef'ed-out not just its body

[kasjer]

Fist line of commit message would also be more consistent with what is used in mynewt if it looked like text that was chosen for PR title:

hw/bsp/dwm1001-dev: Add LIS2DH12 accelerometer support

instead of "Add LIS2DH12 accelerometer support for DWM1001-dev boards"

[seijikun]

Sorry, I don't have access to the hardware anymore, so I can't continue on this with testing.
I tried having only one entry in the interrupts array, that crashed on initialization.
I'm going to close this. Sorry again!