Re: [PATCH 3/3] iio: adc: Add support for PMIC7 ADC

From: Andy Shevchenko
Date: Tue Mar 24 2020 - 12:33:42 EST


On Tue, Mar 24, 2020 at 5:46 PM Jishnu Prakash <jprakash@xxxxxxxxxxxxxx> wrote:
>
> The ADC architecture on PMIC7 is changed as compared to PMIC5. The
> major change from PMIC5 is that all SW communication to ADC goes through
> PMK8350, which communicates with other PMICs through PBS when the ADC
> on PMK8350 works in master mode. The SID register is used to identify the
> PMICs with which the PBS needs to communicate. Add support for the same.
>
> In addition, add definitions for ADC channels and virtual channel
> definitions per PMIC, to be used by ADC clients for PMIC7.

...

> +#define ADC_CHANNEL_OFFSET 0x8
> +#define ADC_CHANNEL_MASK 0xff

GENMASK()

...

> +#define ADC_APP_SID 0x40
> +#define ADC_APP_SID_MASK 0xf

GENMASK()

> +#define ADC7_CONV_TIMEOUT msecs_to_jiffies(10)

Useless.

...

> + buf[1] &= (u8) ~ADC5_USR_FAST_AVG_CTL_SAMPLES_MASK;

Use '0xFF ^ _MASK' instead of casting.

...

> + buf[3] &= (u8) ~ADC5_USR_HW_SETTLE_DELAY_MASK;

Ditto.

...

> + ret = adc5_write(adc, ADC5_USR_CONV_REQ, &conv_req, 1);
> +
> + return ret;

return adc5_write(...);

...

> + pr_err("ADC configure failed with %d\n", ret);

Use dev_*() instead everywhere.

...

> + /* No support for polling mode at present*/
> + wait_for_completion_timeout(&adc->complete,
> + ADC7_CONV_TIMEOUT);

One line. The limit is 80 and it can be bend in some cases a little bit.

...

> + v_channel = ((adc->chan_props[i].sid << ADC_CHANNEL_OFFSET) |
> + adc->chan_props[i].channel);

Too many parentheses.

...

> + sid = (chan >> ADC_CHANNEL_OFFSET);
> + chan = (chan & ADC_CHANNEL_MASK);

Ditto.

...

> + (adc->is_pmic7))

Ditto.

...

> + if (of_device_is_compatible(node, "qcom,spmi-adc7")) {
> + indio_dev->info = &adc7_info;
> + adc->is_pmic7 = true;
> + } else {
> + indio_dev->info = &adc5_info;
> + }

Hmm... I would rather put this as driver_data in ID structure(s).

...

> +static int adc5_exit(struct platform_device *pdev)
> +{
> + struct adc5_chip *adc = platform_get_drvdata(pdev);
> +

> + mutex_destroy(&adc->lock);

Are you sure you will have no race conditions? Does this driver use IRQs?

> + return 0;
> +}

...

> + s64 resistance = 0;

= adc_code // or sign extended variant if needed.

> + /* (ADC code * R_PULLUP (100Kohm)) / (full_scale_code - ADC code)*/

> + resistance = (s64) adc_code * R_PU_100K;
> + resistance = div64_s64(resistance, (RATIO_MAX_ADC7 - adc_code));

resistance *= R_PU_100K;
resistance = div64_s64(resistance, RATIO_MAX_ADC7 - adc_code);

...

> + int voltage, vtemp0, temp, i = 0;

> + while (i < ARRAY_SIZE(adcmap7_die_temp)) {
> + if (adcmap7_die_temp[i].x > voltage)
> + break;
> + i++;
> + }

for loop (one line less, more explicit initial value assignment)?

--
With Best Regards,
Andy Shevchenko