linux/drivers/iio/adc/ti-ads7950.c

541 lines
14 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* Texas Instruments ADS7950 SPI ADC driver
*
* Copyright 2016 David Lechner <david@lechnology.com>
*
* Based on iio/ad7923.c:
* Copyright 2011 Analog Devices Inc
* Copyright 2012 CS Systemes d'Information
*
* And also on hwmon/ads79xx.c
* Copyright (C) 2013 Texas Instruments Incorporated - http://www.ti.com/
* Nishanth Menon
*/
#include <linux/acpi.h>
#include <linux/bitops.h>
#include <linux/device.h>
#include <linux/err.h>
#include <linux/interrupt.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/regulator/consumer.h>
#include <linux/slab.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_consumer.h>
#include <linux/iio/triggered_buffer.h>
/*
* In case of ACPI, we use the 5000 mV as default for the reference pin.
* Device tree users encode that via the vref-supply regulator.
*/
#define TI_ADS7950_VA_MV_ACPI_DEFAULT 5000
#define TI_ADS7950_CR_MANUAL BIT(12)
#define TI_ADS7950_CR_WRITE BIT(11)
#define TI_ADS7950_CR_CHAN(ch) ((ch) << 7)
#define TI_ADS7950_CR_RANGE_5V BIT(6)
#define TI_ADS7950_MAX_CHAN 16
#define TI_ADS7950_TIMESTAMP_SIZE (sizeof(int64_t) / sizeof(__be16))
/* val = value, dec = left shift, bits = number of bits of the mask */
#define TI_ADS7950_EXTRACT(val, dec, bits) \
(((val) >> (dec)) & ((1 << (bits)) - 1))
struct ti_ads7950_state {
struct spi_device *spi;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
struct spi_transfer ring_xfer;
struct spi_transfer scan_single_xfer[3];
struct spi_message ring_msg;
struct spi_message scan_single_msg;
/* Lock to protect the spi xfer buffers */
struct mutex slock;
struct regulator *reg;
unsigned int vref_mv;
unsigned int settings;
/*
* DMA (thus cache coherency maintenance) requires the
* transfer buffers to live in their own cache lines.
*/
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
u16 rx_buf[TI_ADS7950_MAX_CHAN + 2 + TI_ADS7950_TIMESTAMP_SIZE]
____cacheline_aligned;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
u16 tx_buf[TI_ADS7950_MAX_CHAN + 2];
u16 single_tx;
u16 single_rx;
};
struct ti_ads7950_chip_info {
const struct iio_chan_spec *channels;
unsigned int num_channels;
};
enum ti_ads7950_id {
TI_ADS7950,
TI_ADS7951,
TI_ADS7952,
TI_ADS7953,
TI_ADS7954,
TI_ADS7955,
TI_ADS7956,
TI_ADS7957,
TI_ADS7958,
TI_ADS7959,
TI_ADS7960,
TI_ADS7961,
};
#define TI_ADS7950_V_CHAN(index, bits) \
{ \
.type = IIO_VOLTAGE, \
.indexed = 1, \
.channel = index, \
.info_mask_separate = BIT(IIO_CHAN_INFO_RAW), \
.info_mask_shared_by_type = BIT(IIO_CHAN_INFO_SCALE), \
.address = index, \
.datasheet_name = "CH##index", \
.scan_index = index, \
.scan_type = { \
.sign = 'u', \
.realbits = bits, \
.storagebits = 16, \
.shift = 12 - (bits), \
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
.endianness = IIO_CPU, \
}, \
}
#define DECLARE_TI_ADS7950_4_CHANNELS(name, bits) \
const struct iio_chan_spec name ## _channels[] = { \
TI_ADS7950_V_CHAN(0, bits), \
TI_ADS7950_V_CHAN(1, bits), \
TI_ADS7950_V_CHAN(2, bits), \
TI_ADS7950_V_CHAN(3, bits), \
IIO_CHAN_SOFT_TIMESTAMP(4), \
}
#define DECLARE_TI_ADS7950_8_CHANNELS(name, bits) \
const struct iio_chan_spec name ## _channels[] = { \
TI_ADS7950_V_CHAN(0, bits), \
TI_ADS7950_V_CHAN(1, bits), \
TI_ADS7950_V_CHAN(2, bits), \
TI_ADS7950_V_CHAN(3, bits), \
TI_ADS7950_V_CHAN(4, bits), \
TI_ADS7950_V_CHAN(5, bits), \
TI_ADS7950_V_CHAN(6, bits), \
TI_ADS7950_V_CHAN(7, bits), \
IIO_CHAN_SOFT_TIMESTAMP(8), \
}
#define DECLARE_TI_ADS7950_12_CHANNELS(name, bits) \
const struct iio_chan_spec name ## _channels[] = { \
TI_ADS7950_V_CHAN(0, bits), \
TI_ADS7950_V_CHAN(1, bits), \
TI_ADS7950_V_CHAN(2, bits), \
TI_ADS7950_V_CHAN(3, bits), \
TI_ADS7950_V_CHAN(4, bits), \
TI_ADS7950_V_CHAN(5, bits), \
TI_ADS7950_V_CHAN(6, bits), \
TI_ADS7950_V_CHAN(7, bits), \
TI_ADS7950_V_CHAN(8, bits), \
TI_ADS7950_V_CHAN(9, bits), \
TI_ADS7950_V_CHAN(10, bits), \
TI_ADS7950_V_CHAN(11, bits), \
IIO_CHAN_SOFT_TIMESTAMP(12), \
}
#define DECLARE_TI_ADS7950_16_CHANNELS(name, bits) \
const struct iio_chan_spec name ## _channels[] = { \
TI_ADS7950_V_CHAN(0, bits), \
TI_ADS7950_V_CHAN(1, bits), \
TI_ADS7950_V_CHAN(2, bits), \
TI_ADS7950_V_CHAN(3, bits), \
TI_ADS7950_V_CHAN(4, bits), \
TI_ADS7950_V_CHAN(5, bits), \
TI_ADS7950_V_CHAN(6, bits), \
TI_ADS7950_V_CHAN(7, bits), \
TI_ADS7950_V_CHAN(8, bits), \
TI_ADS7950_V_CHAN(9, bits), \
TI_ADS7950_V_CHAN(10, bits), \
TI_ADS7950_V_CHAN(11, bits), \
TI_ADS7950_V_CHAN(12, bits), \
TI_ADS7950_V_CHAN(13, bits), \
TI_ADS7950_V_CHAN(14, bits), \
TI_ADS7950_V_CHAN(15, bits), \
IIO_CHAN_SOFT_TIMESTAMP(16), \
}
static DECLARE_TI_ADS7950_4_CHANNELS(ti_ads7950, 12);
static DECLARE_TI_ADS7950_8_CHANNELS(ti_ads7951, 12);
static DECLARE_TI_ADS7950_12_CHANNELS(ti_ads7952, 12);
static DECLARE_TI_ADS7950_16_CHANNELS(ti_ads7953, 12);
static DECLARE_TI_ADS7950_4_CHANNELS(ti_ads7954, 10);
static DECLARE_TI_ADS7950_8_CHANNELS(ti_ads7955, 10);
static DECLARE_TI_ADS7950_12_CHANNELS(ti_ads7956, 10);
static DECLARE_TI_ADS7950_16_CHANNELS(ti_ads7957, 10);
static DECLARE_TI_ADS7950_4_CHANNELS(ti_ads7958, 8);
static DECLARE_TI_ADS7950_8_CHANNELS(ti_ads7959, 8);
static DECLARE_TI_ADS7950_12_CHANNELS(ti_ads7960, 8);
static DECLARE_TI_ADS7950_16_CHANNELS(ti_ads7961, 8);
static const struct ti_ads7950_chip_info ti_ads7950_chip_info[] = {
[TI_ADS7950] = {
.channels = ti_ads7950_channels,
.num_channels = ARRAY_SIZE(ti_ads7950_channels),
},
[TI_ADS7951] = {
.channels = ti_ads7951_channels,
.num_channels = ARRAY_SIZE(ti_ads7951_channels),
},
[TI_ADS7952] = {
.channels = ti_ads7952_channels,
.num_channels = ARRAY_SIZE(ti_ads7952_channels),
},
[TI_ADS7953] = {
.channels = ti_ads7953_channels,
.num_channels = ARRAY_SIZE(ti_ads7953_channels),
},
[TI_ADS7954] = {
.channels = ti_ads7954_channels,
.num_channels = ARRAY_SIZE(ti_ads7954_channels),
},
[TI_ADS7955] = {
.channels = ti_ads7955_channels,
.num_channels = ARRAY_SIZE(ti_ads7955_channels),
},
[TI_ADS7956] = {
.channels = ti_ads7956_channels,
.num_channels = ARRAY_SIZE(ti_ads7956_channels),
},
[TI_ADS7957] = {
.channels = ti_ads7957_channels,
.num_channels = ARRAY_SIZE(ti_ads7957_channels),
},
[TI_ADS7958] = {
.channels = ti_ads7958_channels,
.num_channels = ARRAY_SIZE(ti_ads7958_channels),
},
[TI_ADS7959] = {
.channels = ti_ads7959_channels,
.num_channels = ARRAY_SIZE(ti_ads7959_channels),
},
[TI_ADS7960] = {
.channels = ti_ads7960_channels,
.num_channels = ARRAY_SIZE(ti_ads7960_channels),
},
[TI_ADS7961] = {
.channels = ti_ads7961_channels,
.num_channels = ARRAY_SIZE(ti_ads7961_channels),
},
};
/*
* ti_ads7950_update_scan_mode() setup the spi transfer buffer for the new
* scan mask
*/
static int ti_ads7950_update_scan_mode(struct iio_dev *indio_dev,
const unsigned long *active_scan_mask)
{
struct ti_ads7950_state *st = iio_priv(indio_dev);
int i, cmd, len;
len = 0;
for_each_set_bit(i, active_scan_mask, indio_dev->num_channels) {
cmd = TI_ADS7950_CR_WRITE | TI_ADS7950_CR_CHAN(i) | st->settings;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
st->tx_buf[len++] = cmd;
}
/* Data for the 1st channel is not returned until the 3rd transfer */
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
st->tx_buf[len++] = 0;
st->tx_buf[len++] = 0;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
st->ring_xfer.len = len * 2;
return 0;
}
static irqreturn_t ti_ads7950_trigger_handler(int irq, void *p)
{
struct iio_poll_func *pf = p;
struct iio_dev *indio_dev = pf->indio_dev;
struct ti_ads7950_state *st = iio_priv(indio_dev);
int ret;
mutex_lock(&st->slock);
ret = spi_sync(st->spi, &st->ring_msg);
if (ret < 0)
goto out;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
iio_push_to_buffers_with_timestamp(indio_dev, &st->rx_buf[2],
iio_get_time_ns(indio_dev));
out:
mutex_unlock(&st->slock);
iio_trigger_notify_done(indio_dev->trig);
return IRQ_HANDLED;
}
static int ti_ads7950_scan_direct(struct iio_dev *indio_dev, unsigned int ch)
{
struct ti_ads7950_state *st = iio_priv(indio_dev);
int ret, cmd;
mutex_lock(&st->slock);
cmd = TI_ADS7950_CR_WRITE | TI_ADS7950_CR_CHAN(ch) | st->settings;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
st->single_tx = cmd;
ret = spi_sync(st->spi, &st->scan_single_msg);
if (ret)
goto out;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
ret = st->single_rx;
out:
mutex_unlock(&st->slock);
return ret;
}
static int ti_ads7950_get_range(struct ti_ads7950_state *st)
{
int vref;
if (st->vref_mv) {
vref = st->vref_mv;
} else {
vref = regulator_get_voltage(st->reg);
if (vref < 0)
return vref;
vref /= 1000;
}
if (st->settings & TI_ADS7950_CR_RANGE_5V)
vref *= 2;
return vref;
}
static int ti_ads7950_read_raw(struct iio_dev *indio_dev,
struct iio_chan_spec const *chan,
int *val, int *val2, long m)
{
struct ti_ads7950_state *st = iio_priv(indio_dev);
int ret;
switch (m) {
case IIO_CHAN_INFO_RAW:
ret = ti_ads7950_scan_direct(indio_dev, chan->address);
if (ret < 0)
return ret;
if (chan->address != TI_ADS7950_EXTRACT(ret, 12, 4))
return -EIO;
*val = TI_ADS7950_EXTRACT(ret, chan->scan_type.shift,
chan->scan_type.realbits);
return IIO_VAL_INT;
case IIO_CHAN_INFO_SCALE:
ret = ti_ads7950_get_range(st);
if (ret < 0)
return ret;
*val = ret;
*val2 = (1 << chan->scan_type.realbits) - 1;
return IIO_VAL_FRACTIONAL;
}
return -EINVAL;
}
static const struct iio_info ti_ads7950_info = {
.read_raw = &ti_ads7950_read_raw,
.update_scan_mode = ti_ads7950_update_scan_mode,
};
static int ti_ads7950_probe(struct spi_device *spi)
{
struct ti_ads7950_state *st;
struct iio_dev *indio_dev;
const struct ti_ads7950_chip_info *info;
int ret;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
spi->bits_per_word = 16;
spi->mode |= SPI_CS_WORD;
ret = spi_setup(spi);
if (ret < 0) {
dev_err(&spi->dev, "Error in spi setup\n");
return ret;
}
indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st));
if (!indio_dev)
return -ENOMEM;
st = iio_priv(indio_dev);
spi_set_drvdata(spi, indio_dev);
st->spi = spi;
st->settings = TI_ADS7950_CR_MANUAL | TI_ADS7950_CR_RANGE_5V;
info = &ti_ads7950_chip_info[spi_get_device_id(spi)->driver_data];
indio_dev->name = spi_get_device_id(spi)->name;
indio_dev->dev.parent = &spi->dev;
indio_dev->modes = INDIO_DIRECT_MODE;
indio_dev->channels = info->channels;
indio_dev->num_channels = info->num_channels;
indio_dev->info = &ti_ads7950_info;
iio: adc: ti-ads7950: use SPI_CS_WORD to reduce CPU usage This changes how the SPI message for the triggered buffer is setup in the TI ADS7950 A/DC driver. By using the SPI_CS_WORD flag, we can read multiple samples in a single SPI transfer. If the SPI controller supports DMA transfers, we can see a significant reduction in CPU usage. For example, on an ARM9 system running at 456MHz reading just 4 channels at 100Hz: before this change, top shows the CPU usage of the IRQ thread of this driver to be ~7.7%. After this change, the CPU usage drops to ~3.8%. The use of big-endian for the raw data was cargo culted from another driver when this driver was originally written. It used an SPI word size of 8 bits and big-endian byte ordering to effectively emulate 16 bit words. Now, in order to inject a CS toggle between each word, we need to use the correct word size, otherwise we would get a CS toggle half way through each word 16-bit. The SPI subsystem uses CPU byte ordering for multi-byte words. So, the data we get back from the SPI is going to be CPU endian now no matter what. Converting that to big endian will just add overhead on little endian systems so we opt to change the raw data format from big endian to CPU endian. There is a small risk that this could break some lazy userspace programs that use the raw data without checking the data format. We can address this if/when it actually comes up. Signed-off-by: David Lechner <david@lechnology.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
2018-09-19 01:08:50 +08:00
/* build spi ring message */
spi_message_init(&st->ring_msg);
st->ring_xfer.tx_buf = &st->tx_buf[0];
st->ring_xfer.rx_buf = &st->rx_buf[0];
/* len will be set later */
st->ring_xfer.cs_change = true;
spi_message_add_tail(&st->ring_xfer, &st->ring_msg);
/*
* Setup default message. The sample is read at the end of the first
* transfer, then it takes one full cycle to convert the sample and one
* more cycle to send the value. The conversion process is driven by
* the SPI clock, which is why we have 3 transfers. The middle one is
* just dummy data sent while the chip is converting the sample that
* was read at the end of the first transfer.
*/
st->scan_single_xfer[0].tx_buf = &st->single_tx;
st->scan_single_xfer[0].len = 2;
st->scan_single_xfer[0].cs_change = 1;
st->scan_single_xfer[1].tx_buf = &st->single_tx;
st->scan_single_xfer[1].len = 2;
st->scan_single_xfer[1].cs_change = 1;
st->scan_single_xfer[2].rx_buf = &st->single_rx;
st->scan_single_xfer[2].len = 2;
spi_message_init_with_transfers(&st->scan_single_msg,
st->scan_single_xfer, 3);
/* Use hard coded value for reference voltage in ACPI case */
if (ACPI_COMPANION(&spi->dev))
st->vref_mv = TI_ADS7950_VA_MV_ACPI_DEFAULT;
mutex_init(&st->slock);
st->reg = devm_regulator_get(&spi->dev, "vref");
if (IS_ERR(st->reg)) {
dev_err(&spi->dev, "Failed get get regulator \"vref\"\n");
ret = PTR_ERR(st->reg);
goto error_destroy_mutex;
}
ret = regulator_enable(st->reg);
if (ret) {
dev_err(&spi->dev, "Failed to enable regulator \"vref\"\n");
goto error_destroy_mutex;
}
ret = iio_triggered_buffer_setup(indio_dev, NULL,
&ti_ads7950_trigger_handler, NULL);
if (ret) {
dev_err(&spi->dev, "Failed to setup triggered buffer\n");
goto error_disable_reg;
}
ret = iio_device_register(indio_dev);
if (ret) {
dev_err(&spi->dev, "Failed to register iio device\n");
goto error_cleanup_ring;
}
return 0;
error_cleanup_ring:
iio_triggered_buffer_cleanup(indio_dev);
error_disable_reg:
regulator_disable(st->reg);
error_destroy_mutex:
mutex_destroy(&st->slock);
return ret;
}
static int ti_ads7950_remove(struct spi_device *spi)
{
struct iio_dev *indio_dev = spi_get_drvdata(spi);
struct ti_ads7950_state *st = iio_priv(indio_dev);
iio_device_unregister(indio_dev);
iio_triggered_buffer_cleanup(indio_dev);
regulator_disable(st->reg);
mutex_destroy(&st->slock);
return 0;
}
static const struct spi_device_id ti_ads7950_id[] = {
{ "ads7950", TI_ADS7950 },
{ "ads7951", TI_ADS7951 },
{ "ads7952", TI_ADS7952 },
{ "ads7953", TI_ADS7953 },
{ "ads7954", TI_ADS7954 },
{ "ads7955", TI_ADS7955 },
{ "ads7956", TI_ADS7956 },
{ "ads7957", TI_ADS7957 },
{ "ads7958", TI_ADS7958 },
{ "ads7959", TI_ADS7959 },
{ "ads7960", TI_ADS7960 },
{ "ads7961", TI_ADS7961 },
{ }
};
MODULE_DEVICE_TABLE(spi, ti_ads7950_id);
static const struct of_device_id ads7950_of_table[] = {
{ .compatible = "ti,ads7950", .data = &ti_ads7950_chip_info[TI_ADS7950] },
{ .compatible = "ti,ads7951", .data = &ti_ads7950_chip_info[TI_ADS7951] },
{ .compatible = "ti,ads7952", .data = &ti_ads7950_chip_info[TI_ADS7952] },
{ .compatible = "ti,ads7953", .data = &ti_ads7950_chip_info[TI_ADS7953] },
{ .compatible = "ti,ads7954", .data = &ti_ads7950_chip_info[TI_ADS7954] },
{ .compatible = "ti,ads7955", .data = &ti_ads7950_chip_info[TI_ADS7955] },
{ .compatible = "ti,ads7956", .data = &ti_ads7950_chip_info[TI_ADS7956] },
{ .compatible = "ti,ads7957", .data = &ti_ads7950_chip_info[TI_ADS7957] },
{ .compatible = "ti,ads7958", .data = &ti_ads7950_chip_info[TI_ADS7958] },
{ .compatible = "ti,ads7959", .data = &ti_ads7950_chip_info[TI_ADS7959] },
{ .compatible = "ti,ads7960", .data = &ti_ads7950_chip_info[TI_ADS7960] },
{ .compatible = "ti,ads7961", .data = &ti_ads7950_chip_info[TI_ADS7961] },
{ },
};
MODULE_DEVICE_TABLE(of, ads7950_of_table);
static struct spi_driver ti_ads7950_driver = {
.driver = {
.name = "ads7950",
.of_match_table = ads7950_of_table,
},
.probe = ti_ads7950_probe,
.remove = ti_ads7950_remove,
.id_table = ti_ads7950_id,
};
module_spi_driver(ti_ads7950_driver);
MODULE_AUTHOR("David Lechner <david@lechnology.com>");
MODULE_DESCRIPTION("TI TI_ADS7950 ADC");
MODULE_LICENSE("GPL v2");