Message-ID: <CAPybu_2pk9twU6Wmxng0jcdZyN_1L0jWPwDB266Cs3ZwyMmFLw@mail.gmail.com>
Date:   Sat, 26 Jan 2019 22:34:50 +0100
From:   Ricardo Ribalda Delgado <ricardo@...alda.com>
To:     Jonathan Cameron <jic23@...nel.org>
Cc:     Alexandru Ardelean <ardeleanalex@...il.com>,
        LKML <linux-kernel@...r.kernel.org>,
        alexandru.ardelean@...log.com, linux-iio@...r.kernel.org
Subject: Re: [PATCH v2] iio:dac:ti-dac7612: Add driver for Texas Instruments DAC7612

HI Jonathan

Thanks for your review. I will make the changes and send it back to
you after testing it on Monday on real hardware.

Until then I have pushed my changes to
https://github.com/ribalda/linux/tree/ribalda-iio-v3 in case you want
to see it before I send it to the list.
(I am still missing the dt-bindings)

Some comments inline.

Best regards!

On Sat, Jan 26, 2019 at 9:42 PM Jonathan Cameron <jic23@...nel.org> wrote:
>
> On Fri, 25 Jan 2019 11:00:55 +0100
> Ricardo Ribalda Delgado <ricardo@...alda.com> wrote:
>
> > It is a driver for Texas Instruments Dual, 12-Bit Serial Input
> > Digital-to-Analog Converter.
> >
> > Datasheet of this chip:
> > http://www.ti.com/lit/ds/sbas106/sbas106.pdf
> >
> > Signed-off-by: Ricardo Ribalda Delgado <ricardo@...alda.com>
> Hi Ricardo,
>
> Various bits and pieces inline.
>
> Jonathan
>
> > ---
> >
> > v2: Fix range
> >  MAINTAINERS                  |   6 ++
> >  drivers/iio/dac/Kconfig      |  10 +++
> >  drivers/iio/dac/Makefile     |   1 +
> >  drivers/iio/dac/ti-dac7612.c | 169 +++++++++++++++++++++++++++++++++++
> >  4 files changed, 186 insertions(+)
> >  create mode 100644 drivers/iio/dac/ti-dac7612.c
> >
> > diff --git a/MAINTAINERS b/MAINTAINERS
> > index d039f66a5cef..30ba5435906b 100644
> > --- a/MAINTAINERS
> > +++ b/MAINTAINERS
> > @@ -14877,6 +14877,12 @@ F:   Documentation/devicetree/bindings/clock/ti,sci-clk.txt
> >  F:   drivers/clk/keystone/sci-clk.c
> >  F:   drivers/reset/reset-ti-sci.c
> >
> > +Texas Instruments' DAC7612 DAC Driver
> > +M:   Ricardo Ribalda <ricardo@...alda.com>
> > +L:   linux-iio@...r.kernel.org
> > +S:   Supported
> > +F:   drivers/iio/dac/ti-dac7612.c
> > +
> >  THANKO'S RAREMONO AM/FM/SW RADIO RECEIVER USB DRIVER
> >  M:   Hans Verkuil <hverkuil@...all.nl>
> >  L:   linux-media@...r.kernel.org
> > diff --git a/drivers/iio/dac/Kconfig b/drivers/iio/dac/Kconfig
> > index f28daf67db6a..fbef9107acad 100644
> > --- a/drivers/iio/dac/Kconfig
> > +++ b/drivers/iio/dac/Kconfig
> > @@ -375,6 +375,16 @@ config TI_DAC7311
> >
> >         If compiled as a module, it will be called ti-dac7311.
> >
> > +config TI_DAC7612
> > +     tristate "Texas Instruments 12-bit 2-channel DAC driver"
> > +     depends on SPI_MASTER && GPIOLIB
> > +     help
> > +       Driver for the Texas Instruments DAC7612, DAC7612U, DAC7612UB
> > +       The driver hand drive the load pin automatically, otherwise
> > +       it needs to be toggled manually.
>
> Given the driver doesn't load without that pin, do we need to give
> the otherwise?  I would drop this comment entirely.

I am probing the gpio with devm_gpiod_get_optional() so the driver can
load without the pin

>
> > +
> > +       If compiled as a module, it will be called ti-dac7612.
> > +
> >  config VF610_DAC
> >       tristate "Vybrid vf610 DAC driver"
> >       depends on OF
> > diff --git a/drivers/iio/dac/Makefile b/drivers/iio/dac/Makefile
> > index f0a37c93de8e..1369fa1d2f0e 100644
> > --- a/drivers/iio/dac/Makefile
> > +++ b/drivers/iio/dac/Makefile
> > @@ -41,4 +41,5 @@ obj-$(CONFIG_STM32_DAC) += stm32-dac.o
> >  obj-$(CONFIG_TI_DAC082S085) += ti-dac082s085.o
> >  obj-$(CONFIG_TI_DAC5571) += ti-dac5571.o
> >  obj-$(CONFIG_TI_DAC7311) += ti-dac7311.o
> > +obj-$(CONFIG_TI_DAC7612) += ti-dac7612.o
> >  obj-$(CONFIG_VF610_DAC) += vf610_dac.o
> > diff --git a/drivers/iio/dac/ti-dac7612.c b/drivers/iio/dac/ti-dac7612.c
> > new file mode 100644
> > index 000000000000..278406f69e0c
> > --- /dev/null
> > +++ b/drivers/iio/dac/ti-dac7612.c
> > @@ -0,0 +1,169 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * DAC7612 Dual, 12-Bit Serial input Digital-to-Analog Converter
> > + *
> > + * Copyright 2019 Qtechnology A/S
> > + * 2019 Ricardo Ribalda <ricardo@...alda.com>
> > + *
> > + * Licensed under the GPL-2.
> > + */
> > +#include <linux/kernel.h>
> > +#include <linux/module.h>
> > +#include <linux/spi/spi.h>
> > +#include <linux/gpio/consumer.h>
> > +#include <linux/iio/iio.h>
> > +
> > +#define NUM_CHANS 2
> > +#define RESOLUTION 12
> Please prefix any naming that is generic like this with
> the driver name.  Avoids potential redefined clashes in the future.
> Where they are 'real numbers' rather than Magic numbers I would
> generally look at using the actual number rather than a define.
>
> > +
> > +struct dac7612 {
> > +     struct spi_device *spi;
> > +     struct gpio_desc *nld;
> > +     uint16_t cache[NUM_CHANS];
> > +};
> > +
> > +static int dac7612_cmd_single(struct dac7612 *priv, int channel, u16 val)
> > +{
> > +     uint8_t buffer[2];
> > +     int ret;
> > +
> > +     if (channel >= NUM_CHANS)
> Is there any way this can happen?  Seems a little over paranoid.
> > +             return -EINVAL;
> > +
> > +     buffer[0] = BIT(5) | (channel << 4) | (val >> 8);
> BIT(5) is a magic number so should probably come from a define
> as should the shifts of the other parts.
>
> > +     buffer[1] = val & 0xff;
> > +
> > +     priv->cache[channel] = val;
> > +
> > +     ret = spi_write(priv->spi, buffer, sizeof(buffer));
>
> spi write can potentially do a dma transfer so it needs
> a dma safe buffer.  This one isn't as it is on the stack.
> Given it is a moderately fast path, best option is to put the
> buffer in priv and use the __cacheline_aligned trick (plus the
> fact private data is already cacheline aligned) to get it into
> it's own cacheline.  Wofram Sang's OSS talk on i2c dma is a good
> tutorial on this and is on youtube if you have time.
>
> https://www.youtube.com/watch?v=JDwaMClvV-s
>
> > +     if (ret)
> > +             return ret;
> > +
> > +     gpiod_set_value(priv->nld, 0);
> > +     gpiod_set_value(priv->nld, 1);
> > +
> > +     return 0;
> > +}
> > +
> > +#define dac7612_CHANNEL(chan, name) {                                \
> > +     .type = IIO_VOLTAGE,                                    \
> > +     .channel = (chan),                                      \
> > +     .address = (chan),                                      \
> Not used, so don't set it.
>
> > +     .indexed = true,                                        \
> These are bit fields, so whilst this works, = 1 is more conventional.
>
> > +     .output = true,                                         \
> > +     .datasheet_name = name,                                 \
> > +     .info_mask_separate = BIT(IIO_CHAN_INFO_RAW),           \
> > +     .info_mask_shared_by_type = BIT(IIO_CHAN_INFO_SCALE),   \
> > +     .scan_type = {                                          \
> > +                     .sign = 'u',                            \
> > +                     .realbits = RESOLUTION,                 \
> > +                     .storagebits = RESOLUTION,              \
> No need to provide scan_type as we don't have buffered output in
> mainline.  Also this would be wrong as storagebits needs to be
> 8/16/32/64 etc.
>
> Would prefer a straight forward value like RESOLUTION was just
> expressed as a number.
>
> > +     },                                                      \
> > +}
> > +
> > +static const struct iio_chan_spec dac7612_channels[] = {
> > +     dac7612_CHANNEL(0, "OUTA"),
> > +     dac7612_CHANNEL(1, "OUTB"),
> > +};
> > +
> > +static int dac7612_read_raw(struct iio_dev *iio_dev,
> > +                         const struct iio_chan_spec *chan,
> > +                         int *val, int *val2, long mask)
> > +{
> > +     struct dac7612 *priv;
> > +
> > +     switch (mask) {
> > +     case IIO_CHAN_INFO_RAW:
> > +             priv = iio_priv(iio_dev);
> > +             *val = priv->cache[chan->channel];
> > +             return IIO_VAL_INT;
> > +
> > +     case IIO_CHAN_INFO_SCALE:
> > +             *val = 1000000;
>
> This makes me wonder.  The units of voltage are millivolts, so is
> raw value * 1000000 = value in mv?

I tried using IIO_VAL_FRACTIONAL, with val = 1 and val2=1000000, and
it only returned -EINVAL.
Then I discovered the PLUS_MICRO and I want to remember that cating
the range sysfs file returned:
0,000001, but I will try again on Monday.

>
> That would make this a very high voltage device.
> See Documentation/ABI/testing/sysfs-bus-iio* for
> IIO ABI documentation.
>
> > +             return IIO_VAL_INT_PLUS_MICRO;
> > +
> > +     default:
> > +             return -EINVAL;
> > +     }
> > +}
> > +
> > +static int dac7612_write_raw(struct iio_dev *iio_dev,
> > +                          const struct iio_chan_spec *chan,
> > +                          int val, int val2, long mask)
> > +{
> > +     struct dac7612 *priv;
>
> struct dac7612 *priv = iio_priv(iio_dev);
>
> (it's just macro magic to get the right point so normally considered
> fine to be called like this).
>
>
> > +     int ret;
> > +
> > +     if (mask != IIO_CHAN_INFO_RAW)
> > +             return -EINVAL;
> > +
> > +     if ((val >= BIT(RESOLUTION)) || val < 0)
> > +             return -EINVAL;
>
> As not providing a write_fmt function (which is fine)
> should also sanity check that val2 is 0.
>
> > +
> > +     priv = iio_priv(iio_dev);
> > +     if (val == priv->cache[chan->channel])
> > +             return 0;
> > +
> > +     mutex_lock(&iio_dev->mlock);
> > +     ret = dac7612_cmd_single(priv, chan->channel, val);
> > +     mutex_unlock(&iio_dev->mlock);
> > +
> > +     return ret;
> > +}
> > +
> > +static const struct iio_info dac7612_info = {
> > +     .read_raw = dac7612_read_raw,
> > +     .write_raw = dac7612_write_raw,
> > +};
> > +
> > +static int dac7612_probe(struct spi_device *spi)
> > +{
> > +     struct iio_dev *iio_dev;
> > +     struct dac7612 *priv;
> > +     int i;
> > +     int ret;
> > +
> > +     iio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*priv));
> > +     if (!iio_dev)
> > +             return -ENOMEM;
> > +
> > +     priv = iio_priv(iio_dev);
> > +     priv->nld = devm_gpiod_get_optional(&spi->dev, "nLD", GPIOD_OUT_HIGH);
>
> This isn't a particularly standard gpio name.  Hence this driver definitely
> needs a DT binding doc (remember to cc dt list and maintainers).  Also
> this should almost certainly be prefixed with a vendor prefix.
>
> It's also not the name I'm seeing on the datasheet, so needs some justifying
> comments.
>
> > +     if (IS_ERR(priv->nld))
> > +             return PTR_ERR(priv->nld);
> > +     priv->spi = spi;
> > +     spi_set_drvdata(spi, iio_dev);
> > +     iio_dev->dev.parent = &spi->dev;
> > +     iio_dev->info = &dac7612_info;
> > +     iio_dev->modes = INDIO_DIRECT_MODE;
> > +     iio_dev->channels = dac7612_channels;
> > +     iio_dev->num_channels = NUM_CHANS;
> ARRAY_SIZE.
> > +     iio_dev->name = spi_get_device_id(spi)->name;
> > +
> > +     for (i = 0; i < NUM_CHANS; i++) {
> ARRAY_SIZE
> > +             ret = dac7612_cmd_single(priv, i, 0);
> > +             if (ret)
> > +                     return ret;
> > +     }
> > +
> > +     return devm_iio_device_register(&spi->dev, iio_dev);
> > +}
> > +
> > +static const struct spi_device_id dac7612_id[] = {
> > +     {"ti-dac7612"},
> > +     {}
> > +};
> > +MODULE_DEVICE_TABLE(spi, dac7612_id);
> > +
> > +static struct spi_driver dac7612_driver = {
> > +     .driver = {
> > +                .name = "ti-dac7612",
> > +                },
> > +     .probe = dac7612_probe,
> > +     .id_table = dac7612_id,
> > +};
> > +module_spi_driver(dac7612_driver);
> > +
> > +MODULE_AUTHOR("Ricardo Ribalda <ricardo@...alda.com>");
> > +MODULE_DESCRIPTION("Texas Instruments DAC7612 DAC driver");
> > +MODULE_LICENSE("GPL v2");
>


-- 
Ricardo Ribalda

Powered by blists - more mailing lists