Re: [PATCH v3] iio: imu: Add support for Kionix KMX61 sensor
From: Hartmut Knaack
Date: Tue Nov 25 2014 - 17:32:14 EST
Daniel Baluta schrieb am 25.11.2014 10:54:
> On Tue, Nov 25, 2014 at 2:45 AM, Hartmut Knaack <knaack.h@xxxxxx> wrote:
>> Daniel Baluta schrieb am 18.11.2014 17:47:
>>> Minimal implementation for KMX61 6-axis accelerometer/magnetometer. It exports
>>> raw accel/magn readings together with scale and sampling frequency.
>>>
>>> Datasheet will be available at:
>>> http://www.kionix.com/6-axis-accelerometer-magnetometer/kmx61
>>>
>> Finally managed to have a close look, and found a few issues.
>>> Signed-off-by: Daniel Baluta <daniel.baluta@xxxxxxxxx>
>>> ---
>>> Changes since v2:
>>> * mostly style fixes after comments received from Jonathan and Peter
>>> * http://marc.info/?l=linux-iio&m=141564019602400&w=2
>>> * implemented PM runtime and PM sleep resume/suspend hooks
>>> * introduced new parameter for kmx61_set_mode to help updating
>>> stby bits in kmx61_data.
>>> * use kmx6111021 instead of kmx61 for i2c_device_id.
>>>
>>> drivers/iio/imu/Kconfig | 9 +
>>> drivers/iio/imu/Makefile | 2 +
>>> drivers/iio/imu/kmx61.c | 766 +++++++++++++++++++++++++++++++++++++++++++++++
>>> 3 files changed, 777 insertions(+)
>>> create mode 100644 drivers/iio/imu/kmx61.c
>>>
>>> diff --git a/drivers/iio/imu/Kconfig b/drivers/iio/imu/Kconfig
>>> index 2b0e451..d675f43 100644
>>> --- a/drivers/iio/imu/Kconfig
>>> +++ b/drivers/iio/imu/Kconfig
>>> @@ -25,6 +25,15 @@ config ADIS16480
>>> Say yes here to build support for Analog Devices ADIS16375, ADIS16480,
>>> ADIS16485, ADIS16488 inertial sensors.
>>>
>>> +config KMX61
>>> + tristate "Kionix KMX61 6-axis accelerometer and magnetometer"
>>> + depends on I2C
>>> + help
>>> + Say Y here if you want to build a driver for Kionix KMX61 6-axis accelerometer
>>> + and magnetometer.
>>> + To compile this driver as module, choose M here: the module will be called
>>> + kmx61.
>>> +
>> This help text is more than 80 characters wide and may better be wrapped a bit differently.
>
> Good catch. checkpatch.pl didn't complain.
>
>>> source "drivers/iio/imu/inv_mpu6050/Kconfig"
>>>
>>> endmenu
>>> diff --git a/drivers/iio/imu/Makefile b/drivers/iio/imu/Makefile
>>> index 114d2c1..e1e6e3d 100644
>>> --- a/drivers/iio/imu/Makefile
>>> +++ b/drivers/iio/imu/Makefile
>>> @@ -14,3 +14,5 @@ adis_lib-$(CONFIG_IIO_ADIS_LIB_BUFFER) += adis_buffer.o
>>> obj-$(CONFIG_IIO_ADIS_LIB) += adis_lib.o
>>>
>>> obj-y += inv_mpu6050/
>>> +
>>> +obj-$(CONFIG_KMX61) += kmx61.o
>>> diff --git a/drivers/iio/imu/kmx61.c b/drivers/iio/imu/kmx61.c
>>> new file mode 100644
>>> index 0000000..f68b3ef
>>> --- /dev/null
>>> +++ b/drivers/iio/imu/kmx61.c
>>> @@ -0,0 +1,766 @@
>>> +/*
>>> + * KMX61 - Kionix 6-axis Accelerometer/Magnetometer
>>> + *
>>> + * Copyright (c) 2014, Intel Corporation.
>>> + *
>>> + * This file is subject to the terms and conditions of version 2 of
>>> + * the GNU General Public License. See the file COPYING in the main
>>> + * directory of this archive for more details.
>>> + *
>>> + * IIO driver for KMX61 (7-bit I2C slave address 0x0E or 0x0F).
>>> + *
>>> + * TODO: buffer, interrupt, thresholds, acpi, temperature sensor
>>> + *
>>> + */
>>> +
>>> +#include <linux/module.h>
>>> +#include <linux/i2c.h>
>>> +#include <linux/pm.h>
>>> +#include <linux/pm_runtime.h>
>>> +#include <linux/iio/iio.h>
>>> +#include <linux/iio/sysfs.h>
>>> +
>>> +#define KMX61_DRV_NAME "kmx61"
>>> +
>>> +#define KMX61_REG_WHO_AM_I 0x00
>>> +
>>> +/*
>>> + * three 16-bit accelerometer output registers for X/Y/Z axis
>>> + * we use only XOUT_L as a base register, all other addresses
>>> + * can be obtained by applying an offset and are provided here
>>> + * only for clarity.
>>> + */
>>> +#define KMX61_ACC_XOUT_L 0x0A
>>> +#define KMX61_ACC_XOUT_H 0x0B
>>> +#define KMX61_ACC_YOUT_L 0x0C
>>> +#define KMX61_ACC_YOUT_H 0x0D
>>> +#define KMX61_ACC_ZOUT_L 0x0E
>>> +#define KMX61_ACC_ZOUT_H 0x0F
>>> +
>>> +/*
>>> + * one 16-bit temperature output register
>>> + */
>>> +#define KMX61_TEMP_L 0x10
>>> +#define KMX61_TEMP_H 0x11
>>> +
>>> +/*
>>> + * three 16-bit magnetometer output registers for X/Y/Z axis
>>> + */
>>> +#define KMX61_MAG_XOUT_L 0x12
>>> +#define KMX61_MAG_XOUT_H 0x13
>>> +#define KMX61_MAG_YOUT_L 0x14
>>> +#define KMX61_MAG_YOUT_H 0x15
>>> +#define KMX61_MAG_ZOUT_L 0x16
>>> +#define KMX61_MAG_ZOUT_H 0x17
>>> +
>>> +#define KMX61_REG_ODCNTL 0x2C
>>> +#define KMX61_REG_STBY 0x29
>>> +#define KMX61_REG_CTRL1 0x2A
>>> +
>>> +#define KMX61_ACC_STBY_BIT BIT(0)
>>> +#define KMX61_MAG_STBY_BIT BIT(1)
>>> +#define KMX61_ACT_STBY_BIT BIT(7)
>>> +
>>> +#define KMX61_ALL_STBY (KMX61_ACC_STBY_BIT | KMX61_MAG_STBY_BIT)
>>> +
>>> +#define KMX61_REG_CTRL1_GSEL0_SHIFT 0
>>> +#define KMX61_REG_CTRL1_GSEL1_SHIFT 1
>>> +#define KMX61_REG_CTRL1_GSEL0_MASK 0x01
>>> +#define KMX61_REG_CTRL1_GSEL1_MASK 0x02
>> I don't see the benefit of separating GSEL0 and GSEL1.
>>> +
>>> +#define KMX61_REG_CTRL1_BIT_RES BIT(4)
>>> +
>>> +#define KMX61_ACC_ODR_SHIFT 0
>>> +#define KMX61_MAG_ODR_SHIFT 4
>>> +#define KMX61_ACC_ODR_MASK 0x0F
>>> +#define KMX61_MAG_ODR_MASK 0xF0
>>> +
>>> +#define KMX61_SLEEP_DELAY_MS 2000
>>> +
>>> +#define KMX61_CHIP_ID 0x12
>>> +
>>> +struct kmx61_data {
>>> + struct i2c_client *client;
>>> +
>>> + /* serialize access to non-atomic ops, e.g set_mode */
>>> + struct mutex lock;
>>> + u8 range;
>>> + u8 odr_bits;
>>> +
>>> + /* standby state */
>>> + u8 acc_stby;
>>> + u8 mag_stby;
>> Why u8 instead of bool?
>
> Hmm, correct. I was thinking bitwise, but you are right. Will fix.
>
>>> +
>>> + /* power state */
>>> + bool acc_ps;
>>> + bool mag_ps;
>>> +};
>>> +
>>> +enum kmx61_range {
>>> + KMX61_RANGE_2G,
>>> + KMX61_RANGE_4G,
>>> + KMX61_RANGE_8G,
>>> +};
>>> +
>>> +enum kmx61_scan {
>>> + KMX61_SCAN_ACC_X,
>>> + KMX61_SCAN_ACC_Y,
>>> + KMX61_SCAN_ACC_Z,
>>> + KMX61_SCAN_TEMP,
>>> + KMX61_SCAN_MAG_X,
>>> + KMX61_SCAN_MAG_Y,
>>> + KMX61_SCAN_MAG_Z,
>>> +};
>>> +
>>> +static const struct {
>>> + u16 uscale;
>>> + u8 gsel0;
>>> + u8 gsel1;
>>> +} kmx61_scale_table[] = {
>>> + {9582, 0, 0},
>>> + {19163, 1, 0},
>>> + {38326, 0, 1},
>> Why not consolidate gsel0 and gsel1 to gsel? Then this could become u16 kmx61_scale[] = {9582, 19163, 38326} with gsel as index.
>
> True. Will fix.
>
>>> +};
>>> +
>>> +/* KMX61 devices */
>>> +#define KMX61_ACC 0x01
>>> +#define KMX61_MAG 0x02
>>> +
>>> +static const struct {
>>> + int val;
>>> + int val2;
>>> + u8 odr_bits;
>>> +} kmx61_samp_freq_table[] = { {12, 500000, 0x00},
>>> + {25, 0, 0x01},
>>> + {50, 0, 0x02},
>>> + {100, 0, 0x03},
>>> + {200, 0, 0x04},
>>> + {400, 0, 0x05},
>>> + {800, 0, 0x06},
>>> + {1600, 0, 0x07},
>>> + {0, 781000, 0x08},
>>> + {1, 563000, 0x09},
>>> + {3, 125000, 0x0A},
>>> + {6, 250000, 0x0B} };
>>> +
>>> +static IIO_CONST_ATTR(accel_scale_available, "0.009582 0.019163 0.038326");
>>> +static IIO_CONST_ATTR(magn_scale_available, "0.001465");
>>> +static IIO_CONST_ATTR_SAMP_FREQ_AVAIL(
>>> + "0.781000 1.563000 3.125000 6.250000 12.500000 25 50 100 200 400 800");
>>> +
>>> +static struct attribute *kmx61_attributes[] = {
>>> + &iio_const_attr_accel_scale_available.dev_attr.attr,
>>> + &iio_const_attr_magn_scale_available.dev_attr.attr,
>>> + &iio_const_attr_sampling_frequency_available.dev_attr.attr,
>>> + NULL,
>>> +};
>>> +
>>> +static const struct attribute_group kmx61_attribute_group = {
>>> + .attrs = kmx61_attributes,
>>> +};
>>> +
>>> +#define KMX61_ACC_CHAN(_axis, _index) { \
>>> + .type = IIO_ACCEL, \
>>> + .modified = 1, \
>>> + .channel2 = IIO_MOD_ ## _axis, \
>>> + .info_mask_separate = BIT(IIO_CHAN_INFO_RAW), \
>>> + .info_mask_shared_by_type = BIT(IIO_CHAN_INFO_SCALE) | \
>>> + BIT(IIO_CHAN_INFO_SAMP_FREQ), \
>>> + .address = KMX61_ACC, \
>>> + .scan_index = _index, \
>>> + .scan_type = { \
>>> + .sign = 's', \
>>> + .realbits = 12, \
>>> + .storagebits = 16, \
>>> + .shift = 4, \
>>> + .endianness = IIO_LE, \
>>> + }, \
>>> +}
>>> +
>>> +#define KMX61_MAG_CHAN(_axis, _index) { \
>>> + .type = IIO_MAGN, \
>>> + .modified = 1, \
>>> + .channel2 = IIO_MOD_ ## _axis, \
>>> + .address = KMX61_MAG, \
>>> + .info_mask_separate = BIT(IIO_CHAN_INFO_RAW), \
>>> + .info_mask_shared_by_type = BIT(IIO_CHAN_INFO_SCALE) | \
>>> + BIT(IIO_CHAN_INFO_SAMP_FREQ), \
>>> + .scan_index = _index, \
>>> + .scan_type = { \
>>> + .sign = 's', \
>>> + .realbits = 14, \
>>> + .storagebits = 16, \
>>> + .shift = 2, \
>>> + .endianness = IIO_LE, \
>>> + }, \
>>> +}
>>> +
>>> +static const struct iio_chan_spec kmx61_channels[] = {
>>> + KMX61_ACC_CHAN(X, KMX61_SCAN_ACC_X),
>>> + KMX61_ACC_CHAN(Y, KMX61_SCAN_ACC_Y),
>>> + KMX61_ACC_CHAN(Z, KMX61_SCAN_ACC_Z),
>>> + KMX61_MAG_CHAN(X, KMX61_SCAN_MAG_X),
>>> + KMX61_MAG_CHAN(Y, KMX61_SCAN_MAG_Y),
>>> + KMX61_MAG_CHAN(Z, KMX61_SCAN_MAG_Z),
>>> +};
>>> +
>>> +static int kmx61_convert_freq_to_bit(int val, int val2)
>>> +{
>>> + int i;
>>> +
>>> + for (i = 0; i < ARRAY_SIZE(kmx61_samp_freq_table); i++)
>>> + if (val == kmx61_samp_freq_table[i].val &&
>>> + val2 == kmx61_samp_freq_table[i].val2)
>>> + return kmx61_samp_freq_table[i].odr_bits;
>>> + return -EINVAL;
>>> +}
>>> +/**
>> Like here, the first line of the following multiline comments have one asterisk too much.
>
> I think this is the convention, when adding doxygen style documentation:
> http://lxr.free-electrons.com/source/fs/eventpoll.c#L382
>
>>> + * kmx61_set_mode() - set KMX61 device operating mode
>>> + * @data - kmx61 device private data pointer
>>> + * @mode - bitmask, indicating operating mode for @device
>>> + * @device - bitmask, indicating device for which @mode needs to be set
>>> + * @update - update stby bits stored in device's private @data
>>> + *
>>> + * For each sensor (accelerometer/magnetometer) there are two operating modes
>>> + * STANDBY and OPERATION. Neither accel nor magn can be disabled independently
>>> + * if they are both enabled. Internal sensors state is saved in acc_stby and
>>> + * mag_stby members of driver's private @data.
>>> + */
>>> +static int kmx61_set_mode(struct kmx61_data *data, u8 mode, u8 device,
>>> + bool update)
>>> +{
>>> + int ret;
>>> + int acc_stby = -1, mag_stby = -1;
>>> +
>>> + ret = i2c_smbus_read_byte_data(data->client, KMX61_REG_STBY);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "Error reading reg_stby\n");
>>> + return ret;
>>> + }
>>> + if (device & KMX61_ACC) {
>>> + if (mode & KMX61_ACC_STBY_BIT) {
>>> + ret |= KMX61_ACC_STBY_BIT;
>>> + acc_stby = 1;
>>> + } else {
>>> + ret &= ~KMX61_ACC_STBY_BIT;
>>> + acc_stby = 0;
>>> + }
>>> + }
>>> +
>>> + if (device & KMX61_MAG) {
>>> + if (mode & KMX61_MAG_STBY_BIT) {
>>> + ret |= KMX61_MAG_STBY_BIT;
>>> + mag_stby = 1;
>>> + } else {
>>> + ret &= ~KMX61_MAG_STBY_BIT;
>>> + mag_stby = 0;
>>> + }
>>> + }
>>> +
>>> + ret = i2c_smbus_write_byte_data(data->client, KMX61_REG_STBY, ret);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "Error writing reg_stby\n");
>>> + return ret;
>>> + }
>>> +
>>> + if (acc_stby != -1 && update)
>>> + data->acc_stby = !!acc_stby;
>>> + if (mag_stby != -1 && update)
>>> + data->mag_stby = !!mag_stby;
>> Why is there a need for double negation?
>
> Correct. Now it doesn't make sense since data->acc_stby is u8.
>
> But, with your suggestion if I will use bool to represent data->acc_stby
> then I think the code should use double negation to make it clear
> that data->acc_stby is bool and acc_stby is int.
>
Well, if it makes you happy, then go ahead. My interpretation of a bool is, that an assignment of 0 becomes 0/false, and an assignment of anything else becomes 1/true. So, I wouldn't see a need here (especially since acc_stby/mag_stby were distinctly assigned either 0 or 1).
>>> +
>>> + return ret;
>> Since no error occured until this point, just return 0.
>
> Ok. Will fix.
>
>>> +}
>>> +
>>> +static int kmx61_get_mode(struct kmx61_data *data, u8 *mode, u8 device)
>>> +{
>>> + int ret;
>>> +
>>> + ret = i2c_smbus_read_byte_data(data->client, KMX61_REG_STBY);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "Error reading reg_stby\n");
>>> + return ret;
>>> + }
>>> + *mode = 0;
>>> +
>>> + if (device & KMX61_ACC) {
>>> + if (ret & KMX61_ACC_STBY_BIT)
>>> + *mode |= KMX61_ACC_STBY_BIT;
>>> + else
>>> + *mode &= ~KMX61_ACC_STBY_BIT;
>> Simply:
>> *mode |= ret & KMX61_ACC_STBY_BIT;
> Ok.
>
>>> + }
>>> +
>>> + if (device & KMX61_MAG) {
>>> + if (ret & KMX61_MAG_STBY_BIT)
>>> + *mode |= KMX61_MAG_STBY_BIT;
>>> + else
>>> + *mode &= ~KMX61_MAG_STBY_BIT;
>> Same here:
> Ok.
>
>> *mode |= ret & KMX61_MAG_STBY_BIT;
>>> + }
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int kmx61_set_odr(struct kmx61_data *data, int val, int val2, u8 device)
>>> +{
>>> + int ret;
>>> + u8 mode;
>>> + int lodr_bits, odr_bits;
>>> +
>>> + ret = kmx61_get_mode(data, &mode, KMX61_ACC | KMX61_MAG);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + lodr_bits = kmx61_convert_freq_to_bit(val, val2);
>>> + if (lodr_bits < 0)
>>> + return lodr_bits;
>>> +
>>> + /* To change ODR, accel and magn must be in STDBY */
>>> + ret = kmx61_set_mode(data, KMX61_ALL_STBY, KMX61_ACC | KMX61_MAG,
>>> + true);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + odr_bits = 0;
>>> + if (device & KMX61_ACC)
>>> + odr_bits |= lodr_bits;
>>> + if (device & KMX61_MAG)
>>> + odr_bits |= (lodr_bits << KMX61_MAG_ODR_SHIFT);
>> No need for parenthesis.
> ok.
>
>>> +
>>> + ret = i2c_smbus_write_byte_data(data->client, KMX61_REG_ODCNTL,
>>> + odr_bits);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + ret = kmx61_set_mode(data, mode, KMX61_ACC | KMX61_MAG, true);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + data->odr_bits = lodr_bits;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static
>>> +int kmx61_get_odr(struct kmx61_data *data, int *val, int *val2, u8 device)
>> Usually wrap parameters? And don't put code on line with {.
>
> Will fix.
>
>>> +{ int i;
>>> + u8 lodr_bits;
>>> +
>>> + if (device & KMX61_ACC)
>>> + lodr_bits = (data->odr_bits >> KMX61_ACC_ODR_SHIFT) &
>>> + KMX61_ACC_ODR_MASK;
>> KMX61_ACC_ODR_SHIFT is just a placebo and not used in kmx61_set_odr. Choose one way and stay consistent.
>
> placebo is good. will fix kmx61_set_odr :)
>
>>> + else if (device & KMX61_MAG)
>>> + lodr_bits = (data->odr_bits >> KMX61_MAG_ODR_SHIFT) &
>>> + KMX61_MAG_ODR_MASK;
>>> + else
>>> + return -EINVAL;
>>> +
>>> + for (i = 0; i < ARRAY_SIZE(kmx61_samp_freq_table); i++)
>>> + if (lodr_bits == kmx61_samp_freq_table[i].odr_bits) {
>>> + *val = kmx61_samp_freq_table[i].val;
>>> + *val2 = kmx61_samp_freq_table[i].val2;
>>> + return 0;
>> return IIO_VAL_INT_PLUS_MICRO allows you to faciliate _read_raw a bit to just return ret there.
>
> Hmm, If you don't mind I want to keep this as it is, just to
> be consistent with the rest of the functions. The convention
> used is to return 0 in case of success.
>
Up to you. Though kmx61_read_raw() would be an example of a function that doesn't follow this convention ;-) And there is no real benefit of checking an error code there, which is passed up anyway.
>>> + }
>>> + return -EINVAL;
>>> +}
>>> +
>>> +static int kmx61_set_range(struct kmx61_data *data, int range)
>>> +{
>>> + int ret;
>>> +
>>> + ret = i2c_smbus_read_byte_data(data->client, KMX61_REG_CTRL1);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "Error reading reg_ctrl1\n");
>>> + return ret;
>>> + }
>>> +
>>> + ret &= ~(KMX61_REG_CTRL1_GSEL0_MASK | KMX61_REG_CTRL1_GSEL1_MASK);
>>> + ret |= kmx61_scale_table[range].gsel0 << KMX61_REG_CTRL1_GSEL0_SHIFT;
>>> + ret |= kmx61_scale_table[range].gsel1 << KMX61_REG_CTRL1_GSEL1_SHIFT;
>>> +
>>> + ret = i2c_smbus_write_byte_data(data->client, KMX61_REG_CTRL1, ret);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "Error writing reg_ctrl1\n");
>>> + return ret;
>>> + }
>>> +
>>> + data->range = range;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int kmx61_set_scale(struct kmx61_data *data, int uscale)
>>> +{
>>> + int ret, i;
>>> + u8 mode;
>>> +
>>> + for (i = 0; i < ARRAY_SIZE(kmx61_scale_table); i++) {
>>> + if (kmx61_scale_table[i].uscale == uscale) {
>>> + ret = kmx61_get_mode(data, &mode,
>>> + KMX61_ACC | KMX61_MAG);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + ret = kmx61_set_mode(data, KMX61_ALL_STBY,
>>> + KMX61_ACC | KMX61_MAG, true);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + ret = kmx61_set_range(data, i);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + return kmx61_set_mode(data, mode,
>>> + KMX61_ACC | KMX61_MAG, true);
>>> + }
>>> + }
>>> + return -EINVAL;
>>> +}
>>> +
>>> +static int kmx61_chip_init(struct kmx61_data *data)
>>> +{
>>> + int ret;
>>> +
>>> + ret = i2c_smbus_read_byte_data(data->client, KMX61_REG_WHO_AM_I);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "Error reading who_am_i\n");
>>> + return ret;
>>> + }
>>> +
>>> + if (ret != KMX61_CHIP_ID) {
>>> + dev_err(&data->client->dev,
>>> + "Wrong chip id, got %x expected %x\n",
>>> + ret, KMX61_CHIP_ID);
>>> + return -EINVAL;
>>> + }
>>> +
>>> + /* set accel 12bit, 4g range */
>>> + ret = kmx61_set_range(data, KMX61_RANGE_4G);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + /* set acc/magn to OPERATION mode */
>>> + ret = kmx61_set_mode(data, 0, KMX61_ACC | KMX61_MAG, true);
>> Just return kmx61_set_mode(data, 0, KMX61_ACC | KMX61_MAG, true);
>
> Ok.
>
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + return 0;
>>> +}
>>> +/**
>>> + * kmx61_set_power_state() - set power state for kmx61 @device
>>> + * @data - kmx61 device private pointer
>>> + * @on - power state to be set for @device
>>> + * @device - bitmask indicating device for which @on state needs to be set
>>> + *
>>> + * Notice that when ACC power state needs to be set to ON and MAG is in
>>> + * OPERATION then we know that kmx61_runtime_resume was already called
>>> + * so we must set ACC OPERATION mode here. The same happens when MAG power
>>> + * state needs to be set to ON and ACC is in OPERATION.
>>> + */
>>> +static int kmx61_set_power_state(struct kmx61_data *data, bool on, u8 device)
>>> +{
>>> +#ifdef CONFIG_PM_RUNTIME
>>> + int ret;
>>> +
>>> + if (device & KMX61_ACC) {
>>> + if (on && !data->acc_ps && !data->mag_stby)
>>> + kmx61_set_mode(data, 0, KMX61_ACC, true);
>>> + data->acc_ps = on;
>>> + }
>>> + if (device & KMX61_MAG) {
>>> + if (on && !data->mag_ps && !data->acc_stby)
>>> + kmx61_set_mode(data, 0, KMX61_MAG, true);
>>> + data->mag_ps = on;
>>> + }
>>> +
>>> + if (on) {
>>> + ret = pm_runtime_get_sync(&data->client->dev);
>>> + } else {
>>> + pm_runtime_mark_last_busy(&data->client->dev);
>>> + ret = pm_runtime_put_autosuspend(&data->client->dev);
>>> + }
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev,
>>> + "Failed: kmx61_set_power_state for %d, ret %d\n",
>>> + on, ret);
>>> + return ret;
>>> + }
>>> +#endif
>>> + return 0;
>>> +}
>>> +
>>> +static int kmx61_read_measurement(struct kmx61_data *data, int base, int offset)
>> u8 base, u8 offset?
>
> Correct.
>
>>> +{
>>> + int ret;
>>> + u8 reg = base + offset * 2;
>>> +
>>> + ret = i2c_smbus_read_word_data(data->client, reg);
>>> + if (ret < 0) {
>>> + dev_err(&data->client->dev, "failed to read reg at %x\n", reg);
>>> + return ret;
>> This return is unnecessary.
>
> Ok.
>
>>> + }
>>> +
>>> + return ret;
>>> +}
>>> +
>>> +static int kmx61_read_raw(struct iio_dev *indio_dev,
>>> + struct iio_chan_spec const *chan, int *val,
>>> + int *val2, long mask)
>> Indentation of the parameters is a bit too high.
>
> Ok, will fix.
>
>>> +{
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + int ret;
>>> + u8 base_reg;
>>> +
>>> + switch (mask) {
>>> + case IIO_CHAN_INFO_RAW:
>>> + switch (chan->type) {
>>> + case IIO_ACCEL:
>>> + case IIO_MAGN:
>>> + base_reg = KMX61_ACC_XOUT_L;
>>> + break;
>>> + default:
>>> + return -EINVAL;
>>> + }
>>> + mutex_lock(&data->lock);
>>> +
>>> + kmx61_set_power_state(data, true, chan->address);
>>> + ret = kmx61_read_measurement(data, base_reg, chan->scan_index);
>>> + if (ret < 0) {
>>> + kmx61_set_power_state(data, false, chan->address);
>>> + mutex_unlock(&data->lock);
>>> + return ret;
>>> + }
>>> + *val = sign_extend32(ret >> chan->scan_type.shift,
>>> + chan->scan_type.realbits - 1);
>>> + kmx61_set_power_state(data, false, chan->address);
>>> +
>>> + mutex_unlock(&data->lock);
>>> + return IIO_VAL_INT;
>>> + case IIO_CHAN_INFO_SCALE:
>>> + switch (chan->type) {
>>> + case IIO_ACCEL:
>>> + *val = 0;
>>> + *val2 = kmx61_scale_table[data->range].uscale;
>>> + return IIO_VAL_INT_PLUS_MICRO;
>>> + case IIO_MAGN:
>>> + /* 14 bits res, 1465 microGauss per magn count */
>>> + *val = 0;
>>> + *val2 = 1465;
>>> + return IIO_VAL_INT_PLUS_MICRO;
>>> + default:
>>> + return -EINVAL;
>>> + }
>>> + case IIO_CHAN_INFO_SAMP_FREQ:
>>> + if (chan->type != IIO_ACCEL && chan->type != IIO_MAGN)
>>> + return -EINVAL;
>>> +
>>> + mutex_lock(&data->lock);
>>> + ret = kmx61_get_odr(data, val, val2, chan->address);
>>> + mutex_unlock(&data->lock);
>> Could be just return ret here, if kmx61_get_odr() returns IIO_VAL_INT_PLUS_MICRO on success and -EINVAL on error.
>
> As explained above, I prefer not to do this. More than that, I can't do ret here
> because the data->lock is held.
>
Well, I actually intended to have an unconditional "return ret;" right after the mutex_unlock. But up to you. I also see a good point to make it obvious here, that the return value is IIO_VAL_INT_PLUS_MICRO. So, up to you.
>>> + if (ret)
>>> + return -EINVAL;
>>> + return IIO_VAL_INT_PLUS_MICRO;
>>> + }
>>> + return -EINVAL;
>>> +}
>>> +
>>> +static int kmx61_write_raw(struct iio_dev *indio_dev,
>>> + struct iio_chan_spec const *chan, int val,
>>> + int val2, long mask)
>> Indentation of parameters is too high.
> Ok.
>>> +{
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + int ret;
>>> +
>>> + switch (mask) {
>>> + case IIO_CHAN_INFO_SAMP_FREQ:
>>> + if (chan->type != IIO_ACCEL && chan->type != IIO_MAGN)
>>> + return -EINVAL;
>>> +
>>> + mutex_lock(&data->lock);
>>> + ret = kmx61_set_odr(data, val, val2, chan->address);
>>> + mutex_unlock(&data->lock);
>>> + return ret;
>>> + case IIO_CHAN_INFO_SCALE:
>>> + switch (chan->type) {
>>> + case IIO_ACCEL:
>>> + if (val != 0)
>>> + return -EINVAL;
>>> + mutex_lock(&data->lock);
>>> + ret = kmx61_set_scale(data, val2);
>>> + mutex_unlock(&data->lock);
>>> + return ret;
>>> + default:
>>> + return -EINVAL;
>>> + }
>>> + return ret;
>> This won't be reached (if so, ret would not be initialized).
>>> + default:
>>> + return -EINVAL;
>>> + }
>>> + return ret;
>> Same here.
>
> I know. Would initializing ret with 0 will be enough?
Since the switch already catches all cases and returns, there is no need for return ret, here. But if you prefer to have a return at the end of the function, you could drop the default case, or leave it empty, or just put a break in there - and return -EINVAL at the end.
>
>>> +}
>>> +
>>> +static const struct iio_info kmx61_info = {
>>> + .driver_module = THIS_MODULE,
>>> + .read_raw = kmx61_read_raw,
>>> + .write_raw = kmx61_write_raw,
>>> + .attrs = &kmx61_attribute_group,
>>> +};
>>> +
>>> +static int kmx61_probe(struct i2c_client *client,
>>> + const struct i2c_device_id *id)
>>> +{
>>> + struct kmx61_data *data;
>>> + struct iio_dev *indio_dev;
>>> + int ret;
>>> + const char *name = NULL;
>>> +
>>> + indio_dev = devm_iio_device_alloc(&client->dev, sizeof(*data));
>>> + if (!indio_dev)
>>> + return -ENOMEM;
>>> +
>>> + data = iio_priv(indio_dev);
>>> + i2c_set_clientdata(client, indio_dev);
>>> + data->client = client;
>>> +
>>> + if (id)
>>> + name = id->name;
>>> +
>>> + indio_dev->dev.parent = &client->dev;
>>> + indio_dev->channels = kmx61_channels;
>>> + indio_dev->num_channels = ARRAY_SIZE(kmx61_channels);
>>> + indio_dev->name = name;
>>> + indio_dev->modes = INDIO_DIRECT_MODE;
>>> + indio_dev->info = &kmx61_info;
>>> +
>>> + mutex_init(&data->lock);
>>> +
>>> + ret = kmx61_chip_init(data);
>>> + if (ret < 0)
>>> + return ret;
>>> +
>>> + ret = iio_device_register(indio_dev);
>>> + if (ret < 0) {
>>> + dev_err(&client->dev, "Failed to register iio device\n");
>>> + goto err_iio_device_register;
>>> + }
>>> +
>>> + ret = pm_runtime_set_active(&client->dev);
>>> + if (ret < 0)
>>> + goto err_pm_runtime_set_active;
>>> +
>>> + pm_runtime_enable(&client->dev);
>>> + pm_runtime_set_autosuspend_delay(&client->dev, KMX61_SLEEP_DELAY_MS);
>>> + pm_runtime_use_autosuspend(&client->dev);
>>> +
>>> + return 0;
>>> +
>>> +err_pm_runtime_set_active:
>>> + iio_device_unregister(indio_dev);
>>> +err_iio_device_register:
>>> + kmx61_set_mode(data, KMX61_ALL_STBY, KMX61_ACC | KMX61_MAG, true);
>>> + return ret;
>>> +}
>>> +
>>> +static int kmx61_remove(struct i2c_client *client)
>>> +{
>>> + struct iio_dev *indio_dev = i2c_get_clientdata(client);
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + int ret;
>>> +
>>> + pm_runtime_disable(&client->dev);
>>> + pm_runtime_set_suspended(&client->dev);
>>> + pm_runtime_put_noidle(&client->dev);
>>> +
>>> + iio_device_unregister(indio_dev);
>>> +
>>> + mutex_lock(&data->lock);
>>> + ret = kmx61_set_mode(data, KMX61_ALL_STBY, KMX61_ACC | KMX61_MAG, true);
>>> + mutex_unlock(&data->lock);
>>> +
>>> + return ret;
>>> +}
>>> +
>>> +#ifdef CONFIG_PM_SLEEP
>>> +static int kmx61_suspend(struct device *dev)
>>> +{
>>> + struct iio_dev *indio_dev = i2c_get_clientdata(to_i2c_client(dev));
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + int ret;
>>> +
>>> + mutex_lock(&data->lock);
>>> + ret = kmx61_set_mode(data, KMX61_ALL_STBY, KMX61_ACC | KMX61_MAG,
>>> + false);
>>> + mutex_unlock(&data->lock);
>>> +
>>> + return ret;
>>> +}
>>> +
>>> +static int kmx61_resume(struct device *dev)
>>> +{
>>> + struct iio_dev *indio_dev = i2c_get_clientdata(to_i2c_client(dev));
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + u8 stby = 0;
>>> +
>>> + if (data->acc_stby)
>>> + stby |= KMX61_ACC_STBY_BIT;
>>> + if (data->mag_stby)
>>> + stby |= KMX61_MAG_STBY_BIT;
>>> +
>>> + return kmx61_set_mode(data, stby, KMX61_ACC | KMX61_MAG, true);
>>> +}
>>> +#endif
>>> +
>>> +#ifdef CONFIG_PM_RUNTIME
>>> +static int kmx61_runtime_suspend(struct device *dev)
>>> +{
>>> + struct iio_dev *indio_dev = i2c_get_clientdata(to_i2c_client(dev));
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + int ret;
>>> +
>>> + mutex_lock(&data->lock);
>>> + ret = kmx61_set_mode(data, KMX61_ALL_STBY, KMX61_ACC | KMX61_MAG, true);
>>> + mutex_unlock(&data->lock);
>>> +
>>> + return ret;
>>> +}
>>> +
>>> +static int kmx61_runtime_resume(struct device *dev)
>>> +{
>>> + struct iio_dev *indio_dev = i2c_get_clientdata(to_i2c_client(dev));
>>> + struct kmx61_data *data = iio_priv(indio_dev);
>>> + u8 stby = 0;
>>> +
>>> + if (!data->acc_ps)
>>> + stby |= KMX61_ACC_STBY_BIT;
>>> + if (!data->mag_ps)
>>> + stby |= KMX61_MAG_STBY_BIT;
>>> +
>>> + return kmx61_set_mode(data, stby, KMX61_ACC | KMX61_MAG, true);
>>> +}
>>> +#endif
>>> +
>>> +static const struct dev_pm_ops kmx61_pm_ops = {
>>> + SET_SYSTEM_SLEEP_PM_OPS(kmx61_suspend, kmx61_resume)
>>> + SET_RUNTIME_PM_OPS(kmx61_runtime_suspend, kmx61_runtime_resume, NULL)
>>> +};
>>> +
>>> +static const struct i2c_device_id kmx61_id[] = {
>>> + {"kmx611021", 0},
>>> + {}
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(i2c, kmx61_id);
>>> +
>>> +static struct i2c_driver kmx61_driver = {
>>> + .driver = {
>>> + .name = KMX61_DRV_NAME,
>>> + .pm = &kmx61_pm_ops,
>>> + },
>>> + .probe = kmx61_probe,
>>> + .remove = kmx61_remove,
>>> + .id_table = kmx61_id,
>>> +};
>>> +
>>> +module_i2c_driver(kmx61_driver);
>>> +
>>> +MODULE_AUTHOR("Daniel Baluta <daniel.baluta@xxxxxxxxx>");
>>> +MODULE_DESCRIPTION("KMX61 accelerometer/magnetometer driver");
>>> +MODULE_LICENSE("GPL v2");
>
> Hartmut, thanks a lot for your reviews. I will send a patch series
> with these cleanups together with changes for ACPI, buffer and trigger
> support.
Don't know if Jonathan would like to see it smashed together, already.
Take care
Hartmut
>
> Daniel.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-iio" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/