Date:   Sat, 29 Aug 2020 17:46:23 +0100
From:   Jonathan Cameron <jic23@...nel.org>
To:     Nishant Malpani <nish.malpani25@...il.com>
Cc:     robh+dt@...nel.org, dragos.bogdan@...log.com,
        darius.berghe@...log.com, linux-kernel@...r.kernel.org,
        linux-iio@...r.kernel.org, devicetree@...r.kernel.org
Subject: Re: [PATCH 1/3] iio: gyro: adxrs290: Add triggered buffer support

On Tue, 25 Aug 2020 18:17:09 +0530
Nishant Malpani <nish.malpani25@...il.com> wrote:

> Provide a way for continuous data capture by setting up buffer support. The
> data ready signal exposed at the SYNC pin of the ADXRS290 is exploited as
> a hardware interrupt which triggers to fill the buffer.
> 
> Triggered buffer setup was tested with both hardware trigger (DATA_RDY) and
> software triggers (sysfs-trig & hrtimer).
> 
> Signed-off-by: Nishant Malpani <nish.malpani25@...il.com>

I've tried not to overlap with Andy's comments but might have hit on a few.
Anyhow, comments inline.

Jonathan


> ---
>  drivers/iio/gyro/Kconfig    |   2 +
>  drivers/iio/gyro/adxrs290.c | 272 ++++++++++++++++++++++++++++++++++--
>  2 files changed, 260 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/iio/gyro/Kconfig b/drivers/iio/gyro/Kconfig
> index 024a34139875..5824f2edf975 100644
> --- a/drivers/iio/gyro/Kconfig
> +++ b/drivers/iio/gyro/Kconfig
> @@ -44,6 +44,8 @@ config ADIS16260
>  config ADXRS290
>  	tristate "Analog Devices ADXRS290 Dual-Axis MEMS Gyroscope SPI driver"
>  	depends on SPI
> +	select IIO_BUFFER
> +	select IIO_TRIGGERED_BUFFER
>  	help
>  	  Say yes here to build support for Analog Devices ADXRS290 programmable
>  	  digital output gyroscope.
> diff --git a/drivers/iio/gyro/adxrs290.c b/drivers/iio/gyro/adxrs290.c
> index ff989536d2fb..25046590761e 100644
> --- a/drivers/iio/gyro/adxrs290.c
> +++ b/drivers/iio/gyro/adxrs290.c
> @@ -13,8 +13,12 @@
>  #include <linux/module.h>
>  #include <linux/spi/spi.h>
>  
> +#include <linux/iio/buffer.h>
>  #include <linux/iio/iio.h>
>  #include <linux/iio/sysfs.h>
> +#include <linux/iio/trigger.h>
> +#include <linux/iio/triggered_buffer.h>
> +#include <linux/iio/trigger_consumer.h>
>  
>  #define ADXRS290_ADI_ID		0xAD
>  #define ADXRS290_MEMS_ID	0x1D
> @@ -35,7 +39,9 @@
>  #define ADXRS290_READ		BIT(7)
>  #define ADXRS290_TSM		BIT(0)
>  #define ADXRS290_MEASUREMENT	BIT(1)
> -#define ADXRS290_SYNC		GENMASK(1, 0)
> +#define ADXRS290_DATA_RDY_OUT	BIT(0)
> +#define ADXRS290_SYNC_MASK	GENMASK(1, 0)
> +#define ADXRS290_SYNC(x)	FIELD_PREP(ADXRS290_SYNC_MASK, x)
>  #define ADXRS290_LPF_MASK	GENMASK(2, 0)
>  #define ADXRS290_LPF(x)		FIELD_PREP(ADXRS290_LPF_MASK, x)
>  #define ADXRS290_HPF_MASK	GENMASK(7, 4)
> @@ -50,6 +56,13 @@ enum adxrs290_mode {
>  	ADXRS290_MODE_MEASUREMENT,
>  };
>  
> +enum adxrs290_scan_index {
> +	ADXRS290_IDX_X,
> +	ADXRS290_IDX_Y,
> +	ADXRS290_IDX_TEMP,
> +	ADXRS290_IDX_TS,
> +};
> +
>  struct adxrs290_state {
>  	struct spi_device	*spi;
>  	/* Serialize reads and their subsequent processing */
> @@ -57,6 +70,12 @@ struct adxrs290_state {
>  	enum adxrs290_mode	mode;
>  	unsigned int		lpf_3db_freq_idx;
>  	unsigned int		hpf_3db_freq_idx;
> +	struct iio_trigger      *dready_trig;
> +	/* Ensure correct alignment of timestamp when present */
> +	struct {
> +		s16 channels[3];
> +		s64 ts __aligned(8);
> +	}                       buffer;
        } buffer;

I can see why you did the alignment like you have, but it looks really odd
in this case.

>  };
>  

...

>  
> +static void adxrs290_chip_off_action(void *data)
> +{
> +	struct iio_dev *indio_dev = data;
> +
> +	adxrs290_set_mode(indio_dev, ADXRS290_MODE_STANDBY);
> +}
> +
>  static int adxrs290_read_raw(struct iio_dev *indio_dev,
>  			     struct iio_chan_spec const *chan,
>  			     int *val,
> @@ -215,24 +287,34 @@ static int adxrs290_read_raw(struct iio_dev *indio_dev,
>  
>  	switch (mask) {
>  	case IIO_CHAN_INFO_RAW:
> +		ret = iio_device_claim_direct_mode(indio_dev);
> +		if (ret)
> +			return ret;
> +
>  		switch (chan->type) {
>  		case IIO_ANGL_VEL:
>  			ret = adxrs290_get_rate_data(indio_dev,
>  						     ADXRS290_READ_REG(chan->address),
>  						     val);
>  			if (ret < 0)
> -				return ret;
> +				goto err_release;
>  
> -			return IIO_VAL_INT;
> +			ret = IIO_VAL_INT;
> +			break;
>  		case IIO_TEMP:
>  			ret = adxrs290_get_temp_data(indio_dev, val);
>  			if (ret < 0)
> -				return ret;
> +				goto err_release;
>  
> -			return IIO_VAL_INT;
> +			ret = IIO_VAL_INT;
> +			break;
>  		default:
> -			return -EINVAL;
> +			ret = -EINVAL;
> +			break;
>  		}
> +err_release:
> +		iio_device_release_direct_mode(indio_dev);
> +		return ret;
>  	case IIO_CHAN_INFO_SCALE:
>  		switch (chan->type) {
>  		case IIO_ANGL_VEL:
> @@ -279,34 +361,53 @@ static int adxrs290_write_raw(struct iio_dev *indio_dev,
>  			      long mask)
>  {
>  	struct adxrs290_state *st = iio_priv(indio_dev);
> -	int lpf_idx, hpf_idx;
> +	int ret, lpf_idx, hpf_idx;
> +
> +	ret = iio_device_claim_direct_mode(indio_dev);

Is there anything in the datasheet that says we can't change the filters
whilst doing captures?  Might get crazy data but if the datasheet doesn't
want against a particular action, then we tend not to try and prevent it.

> +	if (ret)
> +		return ret;
>  
>  	switch (mask) {
>  	case IIO_CHAN_INFO_LOW_PASS_FILTER_3DB_FREQUENCY:
>  		lpf_idx = adxrs290_find_match(adxrs290_lpf_3db_freq_hz_table,
>  					      ARRAY_SIZE(adxrs290_lpf_3db_freq_hz_table),
>  					      val, val2);
> -		if (lpf_idx < 0)
> -			return -EINVAL;
> +		if (lpf_idx < 0) {
> +			ret = -EINVAL;
> +			goto err_release;
> +		}
> +
>  		/* caching the updated state of the low-pass filter */
>  		st->lpf_3db_freq_idx = lpf_idx;
>  		/* retrieving the current state of the high-pass filter */
>  		hpf_idx = st->hpf_3db_freq_idx;
> -		return adxrs290_set_filter_freq(indio_dev, lpf_idx, hpf_idx);
> +		ret = adxrs290_set_filter_freq(indio_dev, lpf_idx, hpf_idx);
> +		break;
> +
>  	case IIO_CHAN_INFO_HIGH_PASS_FILTER_3DB_FREQUENCY:
>  		hpf_idx = adxrs290_find_match(adxrs290_hpf_3db_freq_hz_table,
>  					      ARRAY_SIZE(adxrs290_hpf_3db_freq_hz_table),
>  					      val, val2);
> -		if (hpf_idx < 0)
> -			return -EINVAL;
> +		if (hpf_idx < 0) {
> +			ret = -EINVAL;
> +			goto err_release;
> +		}
> +
>  		/* caching the updated state of the high-pass filter */
>  		st->hpf_3db_freq_idx = hpf_idx;
>  		/* retrieving the current state of the low-pass filter */
>  		lpf_idx = st->lpf_3db_freq_idx;
> -		return adxrs290_set_filter_freq(indio_dev, lpf_idx, hpf_idx);
> +		ret = adxrs290_set_filter_freq(indio_dev, lpf_idx, hpf_idx);
> +		break;
> +
> +	default:
> +		ret = -EINVAL;
> +		break;
>  	}
>  
> -	return -EINVAL;
> +err_release:
I think Andy may have mentioned this, but a break would work just as well
to get you here

> +	iio_device_release_direct_mode(indio_dev);
> +	return ret;
>  }
>  
>  static int adxrs290_read_avail(struct iio_dev *indio_dev,
> @@ -334,6 +435,68 @@ static int adxrs290_read_avail(struct iio_dev *indio_dev,
>  	}
>  }
>  
> +static int adxrs290_data_rdy_trigger_set_state(struct iio_trigger *trig, bool state)
> +{
> +	struct iio_dev *indio_dev = iio_trigger_get_drvdata(trig);
> +	struct adxrs290_state *st = iio_priv(indio_dev);
> +	int ret;
> +	u8 val;
> +
> +	val = (state ? ADXRS290_SYNC(ADXRS290_DATA_RDY_OUT) : 0);

Why the outer set of brackets?

> +
> +	ret = adxrs290_spi_write_reg(st->spi, ADXRS290_REG_DATA_RDY, val);
> +	if (ret < 0)
> +		dev_err(&st->spi->dev, "failed to start data ready interrupt\n");
> +
> +	return ret;
> +}
> +
> +static int adxrs290_reset_trig(struct iio_trigger *trig)
> +{
> +	struct iio_dev *indio_dev = iio_trigger_get_drvdata(trig);
> +	int val;
> +
> +	/*
> +	 * Data ready interrupt is reset after a read of the data registers.
> +	 * Here, we only read the 16b DATAY registers as that marks the end of
> +	 * a read of the data registers and initiates a reset for the interrupt line.
> +	 */
> +	return adxrs290_get_rate_data(indio_dev, ADXRS290_READ_REG(ADXRS290_REG_DATAY0), &val);

We have a bug around this at the moment btw.

A try_reenable callback shouldn't return an error.  The return value is
meant to indicate that we need to go round again or not
(though that's broken currently anwyay). Hence safer to just not
return errors.

return 0;

> +}
> +
> +static const struct iio_trigger_ops adxrs290_trigger_ops = {
> +	.set_trigger_state = &adxrs290_data_rdy_trigger_set_state,
> +	.validate_device = &iio_trigger_validate_own_device,
> +	.try_reenable = &adxrs290_reset_trig,
> +};
> +
> +static irqreturn_t adxrs290_trigger_handler(int irq, void *p)
> +{
> +	struct iio_poll_func *pf = p;
> +	struct iio_dev *indio_dev = pf->indio_dev;
> +	struct adxrs290_state *st = iio_priv(indio_dev);
> +	struct spi_device *spi = st->spi;

As you only use spi once, I'd just use st->spi inline.

> +	u8 tx = ADXRS290_READ_REG(ADXRS290_REG_DATAX0);
> +	int ret;
> +
> +	mutex_lock(&st->lock);
> +
> +	/* exercise a bulk data capture starting from reg DATAX0... */
> +	ret = spi_write_then_read(spi, &tx, sizeof(tx), st->buffer.channels,
> +				  sizeof(st->buffer.channels));
> +	if (ret < 0)
> +		goto done;
> +
> +	iio_push_to_buffers_with_timestamp(indio_dev, &st->buffer,
> +					   pf->timestamp);
> +
> +done:
> +	mutex_unlock(&st->lock);
> +	iio_trigger_notify_done(indio_dev->trig);
> +
> +	return IRQ_HANDLED;
> +}
> +
>  #define ADXRS290_ANGL_VEL_CHANNEL(reg, axis) {				\
>  	.type = IIO_ANGL_VEL,						\
>  	.address = reg,							\
> @@ -346,6 +509,13 @@ static int adxrs290_read_avail(struct iio_dev *indio_dev,
>  	.info_mask_shared_by_type_available =				\
>  	BIT(IIO_CHAN_INFO_LOW_PASS_FILTER_3DB_FREQUENCY) |		\
>  	BIT(IIO_CHAN_INFO_HIGH_PASS_FILTER_3DB_FREQUENCY),		\
> +	.scan_index = ADXRS290_IDX_##axis,				\
> +	.scan_type = {                                                  \
> +		.sign = 's',                                            \
> +		.realbits = 16,                                         \
> +		.storagebits = 16,                                      \
> +		.endianness = IIO_LE,					\
> +	},                                                              \
>  }
>  
>  static const struct iio_chan_spec adxrs290_channels[] = {
> @@ -356,7 +526,21 @@ static const struct iio_chan_spec adxrs290_channels[] = {
>  		.address = ADXRS290_REG_TEMP0,
>  		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) |
>  		BIT(IIO_CHAN_INFO_SCALE),
> +		.scan_index = ADXRS290_IDX_TEMP,
> +		.scan_type = {
> +			.sign = 's',
> +			.realbits = 12,
> +			.storagebits = 16,
> +			.shift = 0,

No need to specify shift if it is 0 as that's the obvious default.

> +			.endianness = IIO_LE,
> +		},
>  	},
> +	IIO_CHAN_SOFT_TIMESTAMP(ADXRS290_IDX_TS),
> +};
> +
> +static const unsigned long adxrs290_avail_scan_masks[] = {
> +	BIT(ADXRS290_IDX_X) | BIT(ADXRS290_IDX_Y) | BIT(ADXRS290_IDX_TEMP),
> +	0
>  };
>  
>  static const struct iio_info adxrs290_info = {
> @@ -365,6 +549,47 @@ static const struct iio_info adxrs290_info = {
>  	.read_avail = &adxrs290_read_avail,
>  };
>  
> +static int adxrs290_probe_trigger(struct iio_dev *indio_dev)
> +{
> +	struct adxrs290_state *st = iio_priv(indio_dev);
> +	int ret;
> +
> +	if (!st->spi->irq) {
> +		dev_info(&st->spi->dev, "no irq, using polling\n");
> +		return 0;
> +	}
> +
> +	st->dready_trig = devm_iio_trigger_alloc(&st->spi->dev,
> +						 "%s-dev%d",
> +						 indio_dev->name,
> +						 indio_dev->id);

As below, this could be sensibly done on fewer lines without any real loss
of readability (staying under advised 80 chars)

> +	if (!st->dready_trig)
> +		return -ENOMEM;
> +
> +	st->dready_trig->dev.parent = &st->spi->dev;
> +	st->dready_trig->ops = &adxrs290_trigger_ops;
> +	iio_trigger_set_drvdata(st->dready_trig, indio_dev);
> +
> +	ret = devm_request_irq(&st->spi->dev, st->spi->irq,
> +			       &iio_trigger_generic_data_rdy_poll,
> +			       IRQF_ONESHOT,
> +			       "adxrs290_irq", st->dready_trig);
> +	if (ret < 0) {
> +		dev_err(&st->spi->dev, "request irq %d failed\n", st->spi->irq);
> +		return ret;
> +	}
> +
> +	ret = devm_iio_trigger_register(&st->spi->dev, st->dready_trig);
> +	if (ret) {
> +		dev_err(&st->spi->dev, "iio trigger register failed\n");
> +		return ret;
> +	}
> +
> +	indio_dev->trig = iio_trigger_get(st->dready_trig);
> +
> +	return 0;
> +}
> +
>  static int adxrs290_probe(struct spi_device *spi)
>  {
>  	struct iio_dev *indio_dev;
> @@ -384,6 +609,7 @@ static int adxrs290_probe(struct spi_device *spi)
>  	indio_dev->channels = adxrs290_channels;
>  	indio_dev->num_channels = ARRAY_SIZE(adxrs290_channels);
>  	indio_dev->info = &adxrs290_info;
> +	indio_dev->available_scan_masks = adxrs290_avail_scan_masks;
>  
>  	mutex_init(&st->lock);
>  
> @@ -416,6 +642,12 @@ static int adxrs290_probe(struct spi_device *spi)
>  	/* max transition time to measurement mode */
>  	msleep(ADXRS290_MAX_TRANSITION_TIME_MS);
>  
> +	ret = devm_add_action_or_reset(&spi->dev,
> +				       adxrs290_chip_off_action,
> +				       indio_dev);

What is this unwinding?  We should only be using devm to undo things
that have been 'done' in probe and the association should be clear.

You have been a bit in consistent with wrapping.  Personally I would
go for as many parameters as fit on each line under 80 chars as possible.
The only exception being when there is some clear 'pairing' of parameters
or similar.

> +	if (ret < 0)
> +		return ret;
> +
>  	ret = adxrs290_get_3db_freq(indio_dev, &val, &val2);
>  	if (ret < 0)
>  		return ret;
> @@ -423,6 +655,18 @@ static int adxrs290_probe(struct spi_device *spi)
>  	st->lpf_3db_freq_idx = val;
>  	st->hpf_3db_freq_idx = val2;
>  
> +	ret = devm_iio_triggered_buffer_setup(&spi->dev, indio_dev,
> +					      &iio_pollfunc_store_time,
> +					      &adxrs290_trigger_handler, NULL);
> +	if (ret < 0) {
> +		dev_err(&spi->dev, "iio triggered buffer setup failed\n");
> +		return ret;
> +	}
> +
> +	ret = adxrs290_probe_trigger(indio_dev);
> +	if (ret < 0)
> +		return ret;
> +
>  	return devm_iio_device_register(&spi->dev, indio_dev);
>  }
>  

Powered by blists - more mailing lists