Re: [PATCH v4 4/5] iio: pressure: bmp280: Add support for BMP380 sensor family

[Angel Iglesias <ang.iglesiasg@gmail.com>]

On lun, 2022-07-25 at 23:15 +0200, Andy Shevchenko wrote:
> On Sat, Jul 23, 2022 at 7:40 PM Angel Iglesias <ang.iglesiasg@gmail.com>
> wrote:
> > 
> > Adds compatibility with the new generation of this sensor, the BMP380
> > 
> > Includes basic sensor initialization to do pressure and temp
> > measurements and allows tuning oversampling settings for each channel.
> > 
> > The compensation algorithms are adapted from the device datasheet and
> > the repository https://github.com/BoschSensortec/BMP3-Sensor-API
> 
> ...
> 
> > +       /* wait for 2ms for command to be proccessed */
> 
> processed
> 
> > +       dev_dbg(data->dev, "Command 0x%X proccessed successfully\n", cmd);
> 
> Ditto. Can you run a spell checker? Kernel has a little one called codespell.

Once again, my typos strike back! I didn't know of this tool, I only checked
format issues using the script check_patch.pl

> > +
> > +       return 0;
> > +}
> 
> ...
> 
> > +static s32 bmp380_compensate_temp(struct bmp280_data *data, u32 adc_temp)
> > +{
> > +       s64 var1, var2, var3, var4, var5, var6, comp_temp;
> > +       struct bmp380_calib *calib = &data->calib.bmp380;
> > +
> > +       var1 = ((s64) adc_temp) - (((s64) calib->T1) << 8);
> > +       var2 = var1 * ((s64) calib->T2);
> > +       var3 = var1 * var1;
> > +       var4 = var3 * ((s64) calib->T3);
> > +       var5 = (var2 << 18) + var4;
> > +       var6 = var5 >> 32;
> > +       data->t_fine = (s32) var6;
> > +       comp_temp = (var6 * 25) >> 14;
> > +
> > +       comp_temp = clamp_val(comp_temp, BMP380_MIN_TEMP, BMP380_MAX_TEMP);
> > +       return (s32) comp_temp;
> > +}
> 
> ...
> 
> > +       s64 var1, var2, var3, var4, var5, var6, offset, sensitivity;
> > +       u64 comp_press;
> > +       struct bmp380_calib *calib = &data->calib.bmp380;
> > +
> > +       var1 = ((s64)data->t_fine) * ((s64)data->t_fine);
> > +       var2 = var1 >> 6;
> > +       var3 = (var2 * ((s64) data->t_fine)) >> 8;
> > +       var4 = (((s64)calib->P8) * var3) >> 5;
> > +       var5 = (((s64) calib->P7) * var1) << 4;
> > +       var6 = (((s64) calib->P6) * ((s64)data->t_fine)) << 22;
> > +       offset = (((s64)calib->P5) << 47) + var4 + var5 + var6;
> > +       var2 = (((s64)calib->P4) * var3) >> 5;
> > +       var4 = (((s64) calib->P3) * var1) << 2;
> > +       var5 = (((s64) calib->P2) - ((s64) 1<<14)) *
> > +               (((s64)data->t_fine) << 21);
> > +       sensitivity = ((((s64) calib->P1) - ((s64) 1 << 14)) << 46) +
> > +                       var2 + var4 + var5;
> > +       var1 = (sensitivity >> 24) * ((s64)adc_press);
> > +       var2 = ((s64)calib->P10) * ((s64) data->t_fine);
> > +       var3 = var2 + (((s64) calib->P9) << 16);
> > +       var4 = (var3 * ((s64)adc_press)) >> 13;
> > +
> > +       /*
> > +        * Dividing by 10 followed by multiplying by 10 to avoid
> > +        * possible overflow caused by (uncomp_data->pressure *
> > partial_data4)
> > +        */
> > +       var5 = (((s64)adc_press) * (var4 / 10)) >> 9;
> > +       var5 *= 10;
> > +       var6 = ((s64)adc_press) * ((s64)adc_press);
> > +       var2 = (((s64)calib->P11) * var6) >> 16;
> > +       var3 = (var2 * ((s64)adc_press)) >> 7;
> > +       var4 = (offset >> 2) + var1 + var5 + var3;
> > +       comp_press = ((u64)var4 * 25) >> 40;
> 
> 
> Kbuild bot is right, you forgot to (compile-)test for a 32-bit machine.

Oh I see. Thanks for heads up, I only tested arm64 and x86_64 archs.

> ...
> 
> > +       ret = regmap_bulk_read(data->regmap, BMP380_REG_TEMP_XLSB, data-
> > >buf, 3);
> 
> sizeof() ?
> 
> ...
> 
> > +       /* Read and compensate temperature so we get a reading of t_fine. */
> 
> for temperature
> 
> ...
> 
> > +       ret = regmap_bulk_read(data->regmap, BMP380_REG_PRESS_XLSB, data-
> > >buf, 3);
> 
> sizeof() ?
> 
> ...
> 
> > +       .oversampling_temp_default = ilog2(1),
> 
> > +       .oversampling_press_default = ilog2(4),
> 
> BIT()

I preserved previous committer use of ilog2() to recover the index value as it
is done on bmp280_write_oversampling_* functions, but you're right, is better to
initialize using BIT() macros.

> ...
> ´
> > +#define BMP380_REG_CMD                 0x7E
> > +#define BMP380_REG_CONFIG              0x1F
> > +#define BMP380_REG_ODR                 0X1D
> > +#define BMP380_REG_OSR                 0X1C
> > +#define BMP380_REG_POWER_CONTROL       0X1B
> > +#define BMP380_REG_IF_CONFIG           0X1A
> > +#define BMP380_REG_INT_CONTROL         0X19
> > +#define BMP380_REG_INT_STATUS          0X11
> > +#define BMP380_REG_EVENT               0X10
> > +#define BMP380_REG_STATUS              0X03
> > +#define BMP380_REG_ERROR               0X02
> > +#define BMP380_REG_ID                  0X00
> > +
> > +#define BMP380_REG_FIFO_CONFIG_1       0X18
> > +#define BMP380_REG_FIFO_CONFIG_2       0X17
> > +#define BMP380_REG_FIFO_WATERMARK_MSB  0X16
> > +#define BMP380_REG_FIFO_WATERMARK_LSB  0X15
> > +#define BMP380_REG_FIFO_DATA           0X14
> > +#define BMP380_REG_FIFO_LENGTH_MSB     0X13
> > +#define BMP380_REG_FIFO_LENGTH_LSB     0X12
> > +
> > +#define BMP380_REG_SENSOR_TIME_MSB     0X0E
> > +#define BMP380_REG_SENSOR_TIME_LSB     0X0D
> > +#define BMP380_REG_SENSOR_TIME_XLSB    0X0C
> > +
> > +#define BMP380_REG_TEMP_MSB            0X09
> > +#define BMP380_REG_TEMP_LSB            0X08
> > +#define BMP380_REG_TEMP_XLSB           0X07
> > +
> > +#define BMP380_REG_PRESS_MSB           0X06
> > +#define BMP380_REG_PRESS_LSB           0X05
> > +#define BMP380_REG_PRESS_XLSB          0X04
> > +
> > +#define BMP380_REG_CALIB_TEMP_START    0x31
> 
> Be consistent x vs X (we prefer x).
> 
Sorry about that


As always, thank you for your comments,

Kind regards,
Angel