Message-ID: <20180819185246.GA5217@gmail.com>
Date:   Sun, 19 Aug 2018 20:52:46 +0200
From:   Marcus Folkesson <marcus.folkesson@...il.com>
To:     Jonathan Cameron <jic23@...nel.org>
Cc:     Hartmut Knaack <knaack.h@....de>,
        Lars-Peter Clausen <lars@...afoo.de>,
        Peter Meerwald-Stadler <pmeerw@...erw.net>,
        Rob Herring <robh+dt@...nel.org>,
        Mark Rutland <mark.rutland@....com>, linux-iio@...r.kernel.org,
        devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/3] iio: dac: add support for ltc166x

Hi Jonathan,

Thanks for your comments!

On Sun, Aug 19, 2018 at 05:38:50PM +0100, Jonathan Cameron wrote:
> On Sat, 11 Aug 2018 22:02:24 +0200
> Marcus Folkesson <marcus.folkesson@...il.com> wrote:
> 
> > LTC1665/LTC1660 is a 8/10-bit Digital-to-Analog Converter
> > (DAC) with eight individual channels.
> > 
> > Signed-off-by: Marcus Folkesson <marcus.folkesson@...il.com>
> 
> So first rule we try to stick to that this breaks is never use wild cards
> in drivers.  You probably wouldn't believe how often this has gone wrong with
> a manufacturer deciding to use other values that wild card covers...

Hah, I can imagine. I will switch do use ltc1660 all over the place.
> 
> Please pick a part and name everything after that.  Either one is fine.
> 
> A few really minor things inline.  Nice little driver.
> 
> Thanks,
> 
> Jonathan
> 
> 
> > ---
> >  drivers/iio/dac/Kconfig   |  10 ++
> >  drivers/iio/dac/Makefile  |   1 +
> >  drivers/iio/dac/ltc166x.c | 244 ++++++++++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 255 insertions(+)
> >  create mode 100644 drivers/iio/dac/ltc166x.c
> > 
> > diff --git a/drivers/iio/dac/Kconfig b/drivers/iio/dac/Kconfig
> > index 76db0768e454..04cfa6bb9dc1 100644
> > --- a/drivers/iio/dac/Kconfig
> > +++ b/drivers/iio/dac/Kconfig
> > @@ -120,6 +120,16 @@ config AD5624R_SPI
> >  	  Say yes here to build support for Analog Devices AD5624R, AD5644R and
> >  	  AD5664R converters (DAC). This driver uses the common SPI interface.
> >  
> > +config LTC166X
> > +	tristate "Linear Technology LTC1660/LTC1665 DAC SPI driver"
> > +	depends on SPI
> > +	help
> > +	  Say yes here to build support for Linear Technology
> > +	  LTC1660 and LTC1665 Digital to Analog Converters.
> > +
> > +	  To compile this driver as a module, choose M here: the
> > +	  module will be called ltc166x.
> > +
> >  config LTC2632
> >  	tristate "Linear Technology LTC2632-12/10/8 DAC spi driver"
> >  	depends on SPI
> > diff --git a/drivers/iio/dac/Makefile b/drivers/iio/dac/Makefile
> > index 81e710ed7491..380749c87c26 100644
> > --- a/drivers/iio/dac/Makefile
> > +++ b/drivers/iio/dac/Makefile
> > @@ -26,6 +26,7 @@ obj-$(CONFIG_CIO_DAC) += cio-dac.o
> >  obj-$(CONFIG_DPOT_DAC) += dpot-dac.o
> >  obj-$(CONFIG_DS4424) += ds4424.o
> >  obj-$(CONFIG_LPC18XX_DAC) += lpc18xx_dac.o
> > +obj-$(CONFIG_LTC166X) += ltc166x.o
> >  obj-$(CONFIG_LTC2632) += ltc2632.o
> >  obj-$(CONFIG_M62332) += m62332.o
> >  obj-$(CONFIG_MAX517) += max517.o
> > diff --git a/drivers/iio/dac/ltc166x.c b/drivers/iio/dac/ltc166x.c
> > new file mode 100644
> > index 000000000000..0031f2b50f14
> > --- /dev/null
> > +++ b/drivers/iio/dac/ltc166x.c
> > @@ -0,0 +1,244 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Driver for Linear Technology LTC1665/LTC1660, 8 channels DAC
> > + *
> > + * Copyright (C) 2018 Marcus Folkesson <marcus.folkesson@...il.com>
> > + */
> > +#include <linux/bitops.h>
> > +#include <linux/iio/iio.h>
> > +#include <linux/iio/sysfs.h>
> > +#include <linux/init.h>
> > +#include <linux/module.h>
> > +#include <linux/regulator/consumer.h>
> > +#include <linux/regmap.h>
> > +#include <linux/spi/spi.h>
> > +
> > +#define LTC166X_REG_WAKE	0x0
> > +#define LTC166X_REG_DAC_A	0x1
> > +#define LTC166X_REG_DAC_B	0x2
> > +#define LTC166X_REG_DAC_C	0x3
> > +#define LTC166X_REG_DAC_D	0x4
> > +#define LTC166X_REG_DAC_E	0x5
> > +#define LTC166X_REG_DAC_F	0x6
> > +#define LTC166X_REG_DAC_G	0x7
> > +#define LTC166X_REG_DAC_H	0x8
> > +#define LTC166X_REG_SLEEP	0xe
> > +
> > +#define LTC166X_NUM_CHANNELS	8
> > +
> > +static const struct regmap_config ltc166x_regmap_config = {
> > +	.reg_bits = 4,
> > +	.val_bits = 12,
> > +};
> > +
> > +enum ltc166x_supported_device_ids {
> > +	ID_LTC1660,
> > +	ID_LTC1665,
> > +};
> > +
> > +struct ltc166x_priv {
> > +	struct spi_device *spi;
> > +	struct regmap *regmap;
> > +	struct regulator *vref_reg;
> > +	unsigned int value[LTC166X_NUM_CHANNELS];
> > +	unsigned int vref_mv;
> > +};
> > +
> > +static int ltc166x_read_raw(struct iio_dev *indio_dev,
> > +		struct iio_chan_spec const *chan,
> > +		int *val,
> > +		int *val2,
> > +		long mask)
> > +{
> > +	struct ltc166x_priv *priv = iio_priv(indio_dev);
> > +
> > +	switch (mask) {
> > +	case IIO_CHAN_INFO_RAW:
> > +		*val = priv->value[chan->channel];
> > +		return IIO_VAL_INT;
> > +	case IIO_CHAN_INFO_SCALE:
> > +		*val = priv->vref_mv;
> > +		*val2 = chan->scan_type.realbits;
> > +		return IIO_VAL_FRACTIONAL_LOG2;
> > +	default:
> > +		return -EINVAL;
> > +	}
> > +}
> > +
> > +static int ltc166x_write_raw(struct iio_dev *indio_dev,
> > +		struct iio_chan_spec const *chan,
> > +		int val,
> > +		int val2,
> > +		long mask)
> > +{
> > +	struct ltc166x_priv *priv = iio_priv(indio_dev);
> > +
> > +	switch (mask) {
> > +	case IIO_CHAN_INFO_RAW:
> > +		if (val2 != 0)
> > +			return -EINVAL;
> > +		if (val > GENMASK(chan->scan_type.realbits-1, 0))
> > +			return -EINVAL;
> 
> nothing makes val positive..  Also spaces around the '-'.
> 
> 

You are right.
I will go for
	val &=  GENMASK(chan->scan_type.realbits - 1, 0);
instead.

> > +		priv->value[chan->channel] = val;
> > +		val <<= chan->scan_type.shift;
> > +		return regmap_write(priv->regmap, chan->channel, val);
> 
> Given that this could fail, usual convention is set the cached value
> priv->value[] only after we know the write succeeded.  

I will switch the order, thanks.

> 
> > +	default:
> > +		return -EINVAL;
> > +	}
> > +}
> > +
> > +#define LTC166X_CHAN(chan, bits) {			\
> > +	.type = IIO_VOLTAGE,				\
> > +	.indexed = 1,					\
> > +	.output = 1,					\
> > +	.channel = chan,				\
> > +	.info_mask_separate = BIT(IIO_CHAN_INFO_RAW),	\
> > +	.info_mask_shared_by_type = BIT(IIO_CHAN_INFO_SCALE),	\
> > +	.scan_type = {					\
> > +		.sign = 'u',				\
> 
> Supplying this for a DAC is unusual given scan_type only becomes relevant
> (in the IIO core) when we have buffers involved.  For output devices
> we don't currently have buffered support.
> 
> Hmm. You are using it as a convenient stashing location though so perhaps
> fair enough.. You could only specify the realbits and shift, but that seems
> a bit odd so perhaps best to leave this as it is.

Yeah, I was actually struggling with how I should handle the different
bit-lengts, then I saw that other drivers (most of all DAC drivers for Analog
Devices) was using the "buffer variables" for stashing these values.

> 
> 
> > +		.realbits = (bits),			\
> > +		.storagebits = 16,			\
> > +		.shift = 12 - (bits),			\
> > +	},						\
> > +}
> > +
> > +#define  LTC166X_OCTAL_CHANNELS(bits) {			\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_A, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_B, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_C, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_D, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_E, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_F, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_G, bits),	\
> > +		LTC166X_CHAN(LTC166X_REG_DAC_H, bits),	\
> > +}
> > +
> > +static const struct iio_chan_spec ltc166x_channels[][LTC166X_NUM_CHANNELS] = {
> > +	[ID_LTC1660] = LTC166X_OCTAL_CHANNELS(10),
> > +	[ID_LTC1665] = LTC166X_OCTAL_CHANNELS(8),
> > +};
> > +
> > +static const struct iio_info ltc166x_info = {
> > +	.read_raw = &ltc166x_read_raw,
> > +	.write_raw = &ltc166x_write_raw,
> > +};
> > +
> > +static int __maybe_unused ltc166x_suspend(struct device *dev)
> > +{
> > +	struct ltc166x_priv *priv = iio_priv(spi_get_drvdata(
> > +						to_spi_device(dev)));
> > +	return regmap_write(priv->regmap, LTC166X_REG_SLEEP, 0x00);
> > +}
> > +
> > +static int __maybe_unused ltc166x_resume(struct device *dev)
> > +{
> > +	struct ltc166x_priv *priv = iio_priv(spi_get_drvdata(
> > +						to_spi_device(dev)));
> > +	return regmap_write(priv->regmap, LTC166X_REG_WAKE, 0x00);
> > +}
> > +static SIMPLE_DEV_PM_OPS(ltc166x_pm_ops, ltc166x_suspend, ltc166x_resume);
> > +
> > +static int ltc166x_probe(struct spi_device *spi)
> > +{
> > +	struct iio_dev *indio_dev;
> > +	struct ltc166x_priv *priv;
> > +	const struct spi_device_id *id = spi_get_device_id(spi);
> > +	int ret;
> > +
> > +	indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*priv));
> > +	if (indio_dev == NULL)
> > +		return -ENOMEM;
> > +
> > +	priv = iio_priv(indio_dev);
> > +	priv->regmap = devm_regmap_init_spi(spi, &ltc166x_regmap_config);
> > +	if (IS_ERR(priv->regmap)) {
> > +		dev_err(&spi->dev, "failed to register spi regmap %ld\n",
> > +			PTR_ERR(priv->regmap));
> > +		return PTR_ERR(priv->regmap);
> > +	}
> > +
> > +	priv->vref_reg = devm_regulator_get(&spi->dev, "vref");
> > +	if (IS_ERR(priv->vref_reg)) {
> > +		dev_err(&spi->dev, "vref regulator not specified\n");
> > +		return PTR_ERR(priv->vref_reg);
> > +	}
> > +
> > +	ret = regulator_enable(priv->vref_reg);
> > +	if (ret) {
> > +		dev_err(&spi->dev, "failed to enable vref regulator: %d\n",
> > +				ret);
> > +		return ret;
> > +	}
> > +
> > +	ret = regulator_get_voltage(priv->vref_reg);
> > +	if (ret < 0) {
> > +		dev_err(&spi->dev, "failed to read vref regulator: %d\n",
> > +				ret);
> > +		goto error_disable_reg;
> > +	}
> > +	priv->vref_mv = ret / 1000;
> 
> It is admittedly fairly unlikely that someone might share this regulator
> across multiple devices, but in theory this might change.
> 
> So I'd slightly prefer this read being done where we actually need
> the answer, so in the read_raw callback.
> 
> Lots of drivers do it in the wrong place and given how unlikely such
> as setup is, I'm not that fussy about it.  However as you need to respin
> anyway for the naming, we might as well make this perfect.

Will do, thanks!

> 
> > +
> > +	priv->spi = spi;
> > +	spi_set_drvdata(spi, indio_dev);
> > +	indio_dev->dev.parent = &spi->dev;
> > +	indio_dev->info = &ltc166x_info;
> > +	indio_dev->modes = INDIO_DIRECT_MODE;
> > +	indio_dev->channels = ltc166x_channels[id->driver_data];
> > +	indio_dev->num_channels = LTC166X_NUM_CHANNELS;
> > +	indio_dev->name = id->name;
> > +
> > +	ret = iio_device_register(indio_dev);
> > +	if (ret) {
> > +		dev_err(&spi->dev, "failed to register iio device: %d\n",
> > +				ret);
> > +		goto error_disable_reg;
> > +	}
> > +
> > +	return 0;
> > +
> > +error_disable_reg:
> > +	regulator_disable(priv->vref_reg);
> > +
> > +	return ret;
> > +}
> > +
> > +static int ltc166x_remove(struct spi_device *spi)
> > +{
> > +	struct iio_dev *indio_dev = spi_get_drvdata(spi);
> > +	struct ltc166x_priv *priv = iio_priv(indio_dev);
> > +
> > +	iio_device_unregister(indio_dev);
> > +	regulator_disable(priv->vref_reg);
> > +
> > +	return 0;
> > +}
> > +
> > +static const struct of_device_id ltc166x_dt_ids[] = {
> > +	{ .compatible = "lltc,ltc1660", .data = (void *)ID_LTC1660 },
> > +	{ .compatible = "lltc,ltc1665", .data = (void *)ID_LTC1665 },
> > +	{ /* sentinel */ }
> > +};
> > +MODULE_DEVICE_TABLE(of, ltc166x_dt_ids);
> > +
> > +static const struct spi_device_id ltc166x_id[] = {
> > +	{"ltc1660", ID_LTC1660},
> > +	{"ltc1665", ID_LTC1665},
> > +	{ /* sentinel */ }
> > +};
> > +MODULE_DEVICE_TABLE(spi, ltc166x_id);
> > +
> > +static struct spi_driver ltc166x_driver = {
> > +	.driver = {
> > +		.name = "ltc166x",
> > +		.of_match_table = ltc166x_dt_ids,
> > +		.pm = &ltc166x_pm_ops,
> > +	},
> > +	.probe	= ltc166x_probe,
> > +	.remove = ltc166x_remove,
> > +	.id_table = ltc166x_id,
> > +};
> > +module_spi_driver(ltc166x_driver);
> > +
> > +MODULE_AUTHOR("Marcus Folkesson <marcus.folkesson@...il.com>");
> > +MODULE_DESCRIPTION("Linear Technology LTC166X DAC");
> > +MODULE_LICENSE("GPL v2");
> 

Best regards
Marcus Folkesson

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)

Powered by blists - more mailing lists