Re: [PATCH v5 2/2] iio: adc: Add rtq6056 support

[Jonathan Cameron <jic23@kernel.org>]

> > > > +
> > > > +static int rtq6056_probe(struct i2c_client *i2c)
> > > > +{
> > > > +     struct iio_dev *indio_dev;
> > > > +     struct rtq6056_priv *priv;
> > > > +     struct device *dev = &i2c->dev;
> > > > +     struct regmap *regmap;
> > > > +     unsigned int vendor_id, shunt_resistor_uohm;
> > > > +     int ret;
> > > > +
> > > > +     if (!i2c_check_functionality(i2c->adapter, I2C_FUNC_SMBUS_WORD_DATA))
> > > > +             return -EOPNOTSUPP;
> > > > +
> > > > +     indio_dev = devm_iio_device_alloc(dev, sizeof(*priv));
> > > > +     if (!indio_dev)
> > > > +             return -ENOMEM;
> > > > +
> > > > +     priv = iio_priv(indio_dev);
> > > > +     priv->dev = dev;
> > > > +     priv->vshuntct_us = priv->vbusct_us = 1037;
> > > > +     priv->avg_sample = 1;
> > > > +     i2c_set_clientdata(i2c, priv);
> > > > +
> > > > +     regmap = devm_regmap_init_i2c(i2c, &rtq6056_regmap_config);
> > > > +     if (IS_ERR(regmap))
> > > > +             return dev_err_probe(dev, PTR_ERR(regmap),
> > > > +                                  "Failed to init regmap\n");
> > > > +
> > > > +     priv->regmap = regmap;
> > > > +
> > > > +     ret = regmap_read(regmap, RTQ6056_REG_MANUFACTID, &vendor_id);
> > > > +     if (ret)
> > > > +             return dev_err_probe(dev, ret,
> > > > +                                  "Failed to get manufacturer info\n");
> > > > +
> > > > +     if (vendor_id != RTQ6056_VENDOR_ID)
> > > > +             return dev_err_probe(dev, -ENODEV,
> > > > +                                  "Invalid vendor id 0x%04x\n", vendor_id);
> > > > +
> > > > +     ret = devm_regmap_field_bulk_alloc(dev, regmap, priv->rm_fields,
> > > > +                                        rtq6056_reg_fields, F_MAX_FIELDS);
> > > > +     if (ret)
> > > > +             return dev_err_probe(dev, ret, "Failed to init regmap field\n");
> > > > +
> > > > +     /*
> > > > +      * By default, configure average sample as 1, bus and shunt conversion
> > > > +      * timea as 1037 microsecond, and operating mode to all on.
> > > > +      */
> > > > +     ret = regmap_write(regmap, RTQ6056_REG_CONFIG, RTQ6056_DEFAULT_CONFIG);
> > > > +     if (ret)
> > > > +             return dev_err_probe(dev, ret,
> > > > +                                  "Failed to enable continuous sensing\n");
> > > > +
> > > > +     pm_runtime_set_autosuspend_delay(dev, MSEC_PER_SEC);
> > > > +     pm_runtime_use_autosuspend(dev);
> > > > +     pm_runtime_set_active(dev);
> > > > +     pm_runtime_mark_last_busy(dev);
> > > > +     pm_runtime_enable(dev);  
> > >
> > > Look at whether you can use devm_pm_runtime_enable()
> > > Note it handles disabling autosuspend for you.
> > >
> > > When using runtime_pm() you want to ensure that the device works without
> > > runtime pm support being enabled.  As such, you turn the device on before
> > > enabling runtime_pm() and (this is missing I think) turn it off after disabling
> > > runtime pm.  So I'd expect a devm_add_action_or_reset() call to unwind
> > > setting the device into continuous sending above.
> > >  
> > If so, I think it's better to configure the device keep in off state
> > in probe stage.
> > The calling order may need to be changed as below
> > devm_add_action_or_reset...
> >
> > pm_runtime_set_autosuspend_delay
> > pm_runtime_use_auto_suspend
> > devm_pm_runtime_enable
> >  
> Ah, not correct. How about if 'PM_RUNTIME' is not enabled?
> Do we need to consider about this case?
> 
> If yes, the original flow about 'pm_runtime' is correct.
I don't follow.  Perhaps next version will make it clear what you mean.

Jonathan