Re: [PATCH v6 3/3] iio: adc: mt6360: Add ADC driver for MT6360

[Andy Shevchenko <andy.shevchenko@gmail.com>]

On Wed, Sep 30, 2020 at 11:55 AM Gene Chen <gene.chen.richtek@gmail.com> wrote:
>
> From: Gene Chen <gene_chen@richtek.com>
>
> Add MT6360 ADC driver include Charger Current, Voltage, and

including

> Temperature.

...

> +#define MT6360_AICR_MASK       GENMASK(7, 2)

bits.h is missed.

...

> +#define MT6360_ADCEN_MASK      0x8000

BIT() ?

...

> +static int mt6360_adc_read_channel(struct mt6360_adc_data *mad, int channel, int *val)
> +{

> +       start_t = ktime_get();
> +       predict_end_t = ktime_add_ms(mad->last_off_timestamps[channel], 2 * ADC_WAIT_TIME_MS);
> +
> +       if (ktime_after(start_t, predict_end_t))
> +               pre_wait_time = ADC_WAIT_TIME_MS;
> +       else
> +               pre_wait_time = 3 * ADC_WAIT_TIME_MS;
> +
> +       msleep(pre_wait_time);

This looks like NIH of wait_interruptible() or so.

...

> +       while (true) {

Oh, please avoid infinite loops.

> +               ret = regmap_raw_read(mad->regmap, MT6360_REG_PMUADCRPT1, rpt, sizeof(rpt));
> +               if (ret)
> +                       goto out_adc_conv;
> +
> +               /*
> +                * There are two functions, ZCV and TypeC OTP, running ADC VBAT and TS in
> +                * background, and ADC samples are taken on a fixed frequency no matter read the
> +                * previous one or not.
> +                * To avoid conflict, We set minimum time threshold after enable ADC and
> +                * check report channel is the same.
> +                * The worst case is run the same ADC twice and background function is also running,
> +                * ADC conversion sequence is desire channel before start ADC, background ADC,
> +                * desire channel after start ADC.
> +                * So the minimum correct data is three times of typical conversion time.
> +                */
> +               if ((rpt[0] & MT6360_RPTCH_MASK) == channel)
> +                       break;
> +
> +               msleep(ADC_WAIT_TIME_MS);
> +       }

So, a new NIH of regmap_read_poll_timeout() ?

...

> +       /* rpt[1]: ADC_HI_BYTE, rpt[2]: ADC_LOW_BYTE */
> +       *val = rpt[1] << 8 | rpt[2];

get_unalined_be16() and hence redundant comment.

...

> +static int mt6360_adc_read_offset(struct mt6360_adc_data *mad, int channel, int *val)
> +{
> +       *val = (channel == MT6360_CHAN_TEMP_JC) ? -80 : 0;
> +       return IIO_VAL_INT;

> +

Misplaced blank line.

> +}

> +static const char *mt6360_channel_labels[MT6360_CHAN_MAX] = {
> +       "usbid", "vbusdiv5", "vbusdiv2", "vsys", "vbat", "ibus", "ibat", "chg_vddp", "temp_jc",
> +       "vref_ts", "ts"

Leave comma.
And split  either by power of two or equally between the lines (seems
like 8 + 3 here).

> +};

...

> +       /* Reset all channel off time to the current one */
> +       all_off_time = ktime_get();
> +       for (i = 0; i < MT6360_CHAN_MAX; i++)
> +               info->last_off_timestamps[i] = all_off_time;

memset32() / memset64() / memset_l()
(Not applicable, but JFYI: memset_p() also present)

-- 
With Best Regards,
Andy Shevchenko

_______________________________________________
Linux-mediatek mailing list
Linux-mediatek@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-mediatek