[PATCH 09/15] iio: adc: at91-sama5d2_adc: adapt the driver for sama7d65
Andy Shevchenko
andy.shevchenko at gmail.com
Mon Aug 4 06:19:55 PDT 2025
On Mon, Aug 4, 2025 at 12:03 PM Varshini Rajendran
<varshini.rajendran at microchip.com> wrote:
>
> Add support to sama7d65 ADC. The differences are highlighted with the
> compatible. The init and parsing of the temperature sensor and
> calibration indexes are the main differences.
...
> +static int at91_sama7d65_adc_temp_sensor_init(struct at91_adc_state *st,
> + struct device *dev);
Again, please try to avoid forward declarations. They make code harder
to maintain.
...
> +enum at91_sama7d65_adc_ts_clb_idx {
> + AT91_SAMA7D65_ADC_TS_CLB_IDX_P1 = 2,
> + AT91_SAMA7D65_ADC_TS_CLB_IDX_P4 = 1,
> + AT91_SAMA7D65_ADC_TS_CLB_IDX_P6 = 4,
> + AT91_SAMA7D65_ADC_TS_CLB_IDX_MAX = 10,
MAX and trailing comma are odd when they go together.
> +};
...
> +static const struct at91_adc_platform sama7d65_platform = {
> + .layout = &sama7g5_layout,
> + .adc_channels = &at91_sama7g5_adc_channels,
> +#define AT91_SAMA7D65_SINGLE_CHAN_CNT 16
> +#define AT91_SAMA7D65_DIFF_CHAN_CNT 8
> +#define AT91_SAMA7D65_TEMP_CHAN_CNT 1
> + .nr_channels = AT91_SAMA7D65_SINGLE_CHAN_CNT +
> + AT91_SAMA7D65_DIFF_CHAN_CNT +
> + AT91_SAMA7D65_TEMP_CHAN_CNT,
> +#define AT91_SAMA7D65_MAX_CHAN_IDX (AT91_SAMA7D65_SINGLE_CHAN_CNT + \
> + AT91_SAMA7D65_DIFF_CHAN_CNT + \
> + AT91_SAMA7D65_TEMP_CHAN_CNT)
> + .max_channels = ARRAY_SIZE(at91_sama7g5_adc_channels),
> + .max_index = AT91_SAMA7D65_MAX_CHAN_IDX,
> +#define AT91_SAMA7G5_HW_TRIG_CNT 3
> + .hw_trig_cnt = AT91_SAMA7G5_HW_TRIG_CNT,
> + .osr_mask = GENMASK(18, 16),
> + .oversampling_avail = { 1, 4, 16, 64, 256, },
> + .oversampling_avail_no = 5,
> + .chan_realbits = 16,
> + .temp_sensor = true,
> + .temp_chan = AT91_SAMA7G5_ADC_TEMP_CHANNEL,
> + .temp_init = at91_sama7d65_adc_temp_sensor_init,
> +};
It's harder to read the static assignment interleaved by the sparse
definitions. Can't you group definitions followed by the initialiser?
...
> +static int at91_sama7d65_adc_temp_sensor_init(struct at91_adc_state *st,
> + struct device *dev)
> +{
> + struct at91_adc_temp_sensor_clb *clb = &st->soc_info.temp_sensor_clb;
> + struct nvmem_cell *temp_calib;
> + u32 *buf = NULL;
What for this NULL initialiser?
> + size_t len;
> + int ret = 0;
This initialisation can be avoided (see below how).
> + if (!st->soc_info.platform->temp_sensor)
> + return 0;
> +
> + /* Get the calibration data from NVMEM. */
> + temp_calib = devm_nvmem_cell_get(dev, "temperature_calib");
> + if (IS_ERR(temp_calib)) {
> + ret = PTR_ERR(temp_calib);
> + if (ret != -ENOENT)
> + dev_err(dev, "Failed to get temperature_calib cell!\n");
> + return ret;
> + }
> +
> + buf = nvmem_cell_read(temp_calib, &len);
> + if (IS_ERR(buf)) {
> + dev_err(dev, "Failed to read calibration data!\n");
> + return PTR_ERR(buf);
> + }
...
> + /*
> + * We prepare here the conversion to milli from micro to avoid
Here we prepare the...
> + * doing it on hotpath.
> + */
...
> +free_buf:
> + kfree(buf);
> + return ret;
Can't use cleanup.h? I.e. __free().
--
With Best Regards,
Andy Shevchenko
More information about the linux-arm-kernel
mailing list