Potential Staging Clean-up: accel:adis16201.
Jonathan Cameron
maintainerHi All
I was looking over what we have left in staging for any low hanging fruit and came across the adis16201 driver.
I was wondering if anyone would like to clean it up?
This would make a good opportuntity for someone who is fairly new to kernel development to make a useful contribution.
To that end I've done a review as if it had been submitted as a new driver.
If you want to take this on, please reply to say so in order to avoid several people looking at this one driver. Obviously if things change and you then can't get to it reasonably quickly then 'hand it back' on the mailing list.
I'm happy to do this for a few more drivers if there are enough interested people so do ping me if someone has claimed this one and you would like a similar starting point.
Ideally we would find someone who actually has hardware, but for the simple stuff in this one I think we are unlikely to break anything that can't be picked up in review. Hence I'm not totally bothered about it if no one comes forward!
Alison, Daniel. Feel free to add this to the outreachy task list if no one else grabs it before the process starts again (just let us know if you are doing that).
Thanks,
Jonathan
Quoted 6 lines
/*
* ADIS16201 Dual-Axis Digital Inclinometer and Accelerometer *
* Copyright 2010 Analog Devices Inc.
*
* Licensed under the GPL-2 or later.
Might as well convert to SPDX format for the license if we are working on the driver anyway.
Quoted 10 lines
*/
#include <
linux/delay.h>
#include <linux/mutex.h>
#include <linux/device.h>
#include <linux/kernel.h>
#include <linux/spi/spi.h>
#include <linux/slab.h>
#include <linux/sysfs.h>
#include <linux/module.h>
Slight preference for alphabetical order in headers.
However, it is useful to keep the iio headers separate as they are here.
Quoted 7 lines
#include <
linux/iio/iio.h>
#include <linux/iio/sysfs.h>
#include <linux/iio/buffer.h>
#include <linux/iio/imu/adis.h>#define
ADIS16201_STARTUP_DELAY220 /* ms */
Why not just make this ADIS16201_STARTUP_DELAY_MS as then it will be obvious what the units are throughout the driver and remove the need for the comment.
Quoted 3 lines
/* Flash memory write count */
#defineADIS16201_FLASH_CNT0x00
To my mind there is far more commenting going on here than is actually useful. If the name of the register doesn't make it reasonably clear what it is used for then it is not a good choice of name.
It is made more complex by the fact these are the datasheet names so we need to keep 'close' to those names for ease of comparisom.
So unless something odd is going on I would not expect to see any comments in here at all.
Formatting wise, should be using tabs rather than spaces before the numbers.
Also, we have a mixture of defines for register addresses and fields with in registers. That is fine, but I would prefer the nice clear _REG postfix on the registers defines to make it more obvious which is which.
e.g. What I'd expect here is something like:
#define ADIS16201_FLASH_CNT_REG 0x00
#define ADIS16201_SUPPLY_OUT_REG 0x02
#define ADIS16201_XACCL_OUT_REG 0x04
etc.
Quoted 3 lines
/* Output, power supply */
#defineADIS16201_SUPPLY_OUT0x02
The out naming is awkward given it doesn't mean the device is supplying the power, but rather that this is the reading the device is outputting. Still can't do that much about this without breaking away from datasheet naming.
Quoted 76 lines
/* Output, x-axis accelerometer */
#defineADIS16201_XACCL_OUT0x04/* Output, y-axis accelerometer */
#defineADIS16201_YACCL_OUT0x06/* Output, auxiliary ADC input */
#defineADIS16201_AUX_ADC0x08/* Output, temperature */
#defineADIS16201_TEMP_OUT0x0A/* Output, x-axis inclination */
#defineADIS16201_XINCL_OUT0x0C/* Output, y-axis inclination */
#defineADIS16201_YINCL_OUT0x0E/* Calibration, x-axis acceleration offset */
#defineADIS16201_XACCL_OFFS0x10/* Calibration, y-axis acceleration offset */
#defineADIS16201_YACCL_OFFS0x12/* x-axis acceleration scale factor */
#defineADIS16201_XACCL_SCALE0x14/* y-axis acceleration scale factor */
#defineADIS16201_YACCL_SCALE0x16/* Calibration, x-axis inclination offset */
#defineADIS16201_XINCL_OFFS0x18/* Calibration, y-axis inclination offset */
#defineADIS16201_YINCL_OFFS0x1A/* x-axis inclination scale factor */
#defineADIS16201_XINCL_SCALE0x1C/* y-axis inclination scale factor */
#defineADIS16201_YINCL_SCALE0x1E/* Alarm 1 amplitude threshold */
#defineADIS16201_ALM_MAG10x20/* Alarm 2 amplitude threshold */
#defineADIS16201_ALM_MAG20x22/* Alarm 1, sample period */
#defineADIS16201_ALM_SMPL10x24/* Alarm 2, sample period */
#defineADIS16201_ALM_SMPL20x26/* Alarm control */
#defineADIS16201_ALM_CTRL0x28/* Auxiliary DAC data */
#defineADIS16201_AUX_DAC0x30/* General-purpose digital input/output control */
#defineADIS16201_GPIO_CTRL0x32/* Miscellaneous control */
#defineADIS16201_MSC_CTRL0x34/* Internal sample period (rate) control */
#defineADIS16201_SMPL_PRD0x36/* Operation, filter configuration */
#defineADIS16201_AVG_CNT0x38/* Operation, sleep mode control */
#defineADIS16201_SLP_CNT0x3A
Quoted 8 lines
/* Diagnostics, system status register */
#defineADIS16201_DIAG_STAT0x3C/* Operation, system command register */
#defineADIS16201_GLOB_CMD0x3E/*
MSC_CTRL*/
A really nice convention that has come up in recent drivers and makes all these register / register field relationships obvious is to use some subtle indenting of the fields and put them next to the register address. An example is:
#define ADIS16201_MSC_CTRL_REG 0x34
#define ADIS16201_MSC_CTRL_SELF_TEST_EN BIT(8) #define ADIS16201_MSC_CTRL_DATA_RDY_EN BIT(2) #define ADIS16201_MSC_CTRL_ACTIVE_HIGH BIT(1) #define ADIS16201_MSC_CTRL_DATA_RDY_DIO1 BIT(0)
Again the naming of the fields is sufficient and the documentation doesn't add anything so I would drop it.
Quoted 49 lines
/* Self-test enable */
#defineADIS16201_MSC_CTRL_SELF_TEST_ENBIT(8)/* Data-ready enable: 1 = enabled, 0 = disabled */
#defineADIS16201_MSC_CTRL_DATA_RDY_ENBIT(2)/* Data-ready polarity: 1 = active high, 0 = active low */
#defineADIS16201_MSC_CTRL_ACTIVE_HIGHBIT(1)/* Data-ready line selection: 1 = DIO1, 0 = DIO0 */
#defineADIS16201_MSC_CTRL_DATA_RDY_DIO1BIT(0)/*
DIAG_STAT*//* Alarm 2 status: 1 = alarm active, 0 = alarm inactive */
#defineADIS16201_DIAG_STAT_ALARM2BIT(9)/* Alarm 1 status: 1 = alarm active, 0 = alarm inactive */
#defineADIS16201_DIAG_STAT_ALARM1BIT(8)/* SPI communications failure */
#defineADIS16201_DIAG_STAT_SPI_FAIL_BIT3/* Flash update failure */
#defineADIS16201_DIAG_STAT_FLASH_UPT_BIT2/* Power supply above 3.625 V */
#defineADIS16201_DIAG_STAT_POWER_HIGH_BIT1/* Power supply below 3.15 V */
#defineADIS16201_DIAG_STAT_POWER_LOW_BIT0/*
GLOB_CMD*/#define
ADIS16201_GLOB_CMD_SW_RESETBIT(7)
#defineADIS16201_GLOB_CMD_FACTORY_CALBIT(1)#define
ADIS16201_ERROR_ACTIVEBIT(14)enum adis16201_scan { ADIS16201_SCAN_ACC_X, ADIS16201_SCAN_ACC_Y, ADIS16201_SCAN_INCLI_X, ADIS16201_SCAN_INCLI_Y, ADIS16201_SCAN_SUPPLY, ADIS16201_SCAN_AUX_ADC, ADIS16201_SCAN_TEMP, };
Quoted 1 line
static const u8 adis16201_addresses[] = {
It is rather odd that this is only used for the offsets, not
any of the other channel specific registers (scales, or the reading itself).
Quoted 10 lines
[ADIS16201_SCAN_ACC_X] = ADIS16201_XACCL_OFFS, [ADIS16201_SCAN_ACC_Y] = ADIS16201_YACCL_OFFS, [ADIS16201_SCAN_INCLI_X] = ADIS16201_XINCL_OFFS, [ADIS16201_SCAN_INCLI_Y] = ADIS16201_YINCL_OFFS, };static int adis16201_read_raw(struct
iio_dev*indio_dev,
structiio_chan_specconst *chan, int *val, int *val2, long mask)
Quoted 11 lines
{
struct adis *st = iio_priv(indio_dev);
int ret;
int bits;
u8 addr;
s16 val16;
switch (mask) {
case IIO_CHAN_INFO_RAW:
return adis_single_conversion(indio_dev, chan,
ADIS16201_ERROR_ACTIVE, val);Indenting could I think make the parameters align with the opening bracket.
Quoted 6 lines
case IIO_CHAN_INFO_SCALE:
switch (chan->type) {
case IIO_VOLTAGE:
if (chan->channel == 0) {
*val = 1;
*val2 = 220000; /* 1.22 mV */Comments don't add a huge amount given they kind of just state the value before them.
Quoted 1 line
} else {Given there are only two relevant channels I would like to see them move explicitly matched using a switch statement.
The default case would be an error return -EINVAL;
Quoted 11 lines
*val = 0; *val2 = 610000; /* 0.610 mV */ } return IIO_VAL_INT_PLUS_MICRO; case IIO_TEMP: *val = -470; /* 0.47 C */ *val2 = 0; return IIO_VAL_INT_PLUS_MICRO; case IIO_ACCEL: *val = 0; *val2 = IIO_G_TO_M_S_2(462400); /* 0.4624 mg */
Not sure that comment adds anything..
Quoted 4 lines
return IIO_VAL_INT_PLUS_NANO; case IIO_INCLI: *val = 0; *val2 = 100000; /* 0.1 degree */
Or this one.
Quoted 16 lines
return IIO_VAL_INT_PLUS_MICRO;
default:
return -EINVAL;
}
break;
case IIO_CHAN_INFO_OFFSET:
*val = 25000 / -470 - 1278; /* 25 C = 1278 */
return IIO_VAL_INT;
case IIO_CHAN_INFO_CALIBBIAS:
switch (chan->type) {
case IIO_ACCEL:
bits = 12;
break;
case IIO_INCLI:
bits = 9;
break;Quoted 8 lines
default: return -EINVAL; } addr = adis16201_addresses[chan->scan_index]; ret = adis_read_reg_16(st, addr, &val16); if (ret) return ret; val16 &= (1 << bits) - 1;
Given the following sign extend, I don't think this does anything useful.
Quoted 1 line
val16 = (s16)(val16 << (16 - bits)) >> (16 - bits);
This looks like a 16bit sign extend to me. We have functions to handle this nicely. See how it is done in other driverss
Quoted 1 line
*val = val16;
blank line before return always makes them more obvious and hence the code slightly more readable. Do this throughout (including 2 lines below this)
Quoted 21 lines
return IIO_VAL_INT; } return -EINVAL; }static int adis16201_write_raw(struct
iio_dev*indio_dev,
structiio_chan_specconst *chan, int val,
int val2,
long mask){ struct adis *st = iio_priv(indio_dev); int bits; s16 val16; u8 addr; switch (mask) { case IIO_CHAN_INFO_CALIBBIAS: switch (chan->type) { case IIO_ACCEL: bits = 12;
Quoted 8 lines
break; case IIO_INCLI: bits = 9; break; default: return -EINVAL; } val16 = val & ((1 << bits) - 1);
Given the above bits value is used only for this, why not create a mask in the switch statement instead of carrying the number of bits around.
Then you can use GENMASK to make it nice and obvious what is going on.
Quoted 51 lines
addr = adis16201_addresses[chan->scan_index]; return adis_write_reg_16(st, addr, val16); } return -EINVAL; }static const struct iio_chan_spec adis16201_channels[] = { ADIS_SUPPLY_CHAN(ADIS16201_SUPPLY_OUT, ADIS16201_SCAN_SUPPLY, 0, 12), ADIS_TEMP_CHAN(ADIS16201_TEMP_OUT, ADIS16201_SCAN_TEMP, 0, 12), ADIS_ACCEL_CHAN(X, ADIS16201_XACCL_OUT, ADIS16201_SCAN_ACC_X, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_ACCEL_CHAN(Y, ADIS16201_YACCL_OUT, ADIS16201_SCAN_ACC_Y, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_AUX_ADC_CHAN(ADIS16201_AUX_ADC, ADIS16201_SCAN_AUX_ADC, 0, 12), ADIS_INCLI_CHAN(X, ADIS16201_XINCL_OUT, ADIS16201_SCAN_INCLI_X, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_INCLI_CHAN(X, ADIS16201_YINCL_OUT, ADIS16201_SCAN_INCLI_Y, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), IIO_CHAN_SOFT_TIMESTAMP(7) };static const struct iio_info adis16201_info = { .read_raw = adis16201_read_raw, .write_raw = adis16201_write_raw, .update_scan_mode = adis_update_scan_mode, };static const char * const adis16201_status_error_msgs[] = { [ADIS16201_DIAG_STAT_SPI_FAIL_BIT] = "SPI failure", [ADIS16201_DIAG_STAT_FLASH_UPT_BIT] = "Flash update failed", [ADIS16201_DIAG_STAT_POWER_HIGH_BIT] = "Power supply above 3.625V", [ADIS16201_DIAG_STAT_POWER_LOW_BIT] = "Power supply below 3.15V", };static const struct adis_data adis16201_data = { .read_delay = 20, .msc_ctrl_reg = ADIS16201_MSC_CTRL, .glob_cmd_reg = ADIS16201_GLOB_CMD, .diag_stat_reg = ADIS16201_DIAG_STAT, .self_test_mask = ADIS16201_MSC_CTRL_SELF_TEST_EN, .self_test_no_autoclear = true, .startup_delay = ADIS16201_STARTUP_DELAY, .status_error_msgs = adis16201_status_error_msgs, .status_error_mask = BIT(ADIS16201_DIAG_STAT_SPI_FAIL_BIT) | BIT(ADIS16201_DIAG_STAT_FLASH_UPT_BIT) | BIT(ADIS16201_DIAG_STAT_POWER_HIGH_BIT) | BIT(ADIS16201_DIAG_STAT_POWER_LOW_BIT), };
Quoted 5 lines
static int
adis16201_probe(struct spi_device *spi){ int ret; struct adis *st; struct iio_dev *indio_dev;
There is a convention (when nothing else is guiding ordering) of doing things in reverse Christmas tree.
That means longest line first and shortest last. Very minor point though.
Quoted 2 lines
/* setup the industrialio driver allocated elements */
Comment doesn't really add anything so drop it.
Quoted 6 lines
indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st)); if (!indio_dev) return -ENOMEM; st = iio_priv(indio_dev); /* this is only used for removal purposes */
Again, comment isn't really that useful...
Quoted 6 lines
spi_set_drvdata(spi, indio_dev); indio_dev->name = spi->dev.driver->name; indio_dev->dev.parent = &spi->dev; indio_dev->info = &adis16201_info;
Quoted 7 lines
indio_dev->channels = adis16201_channels; indio_dev->num_channels = ARRAY_SIZE(adis16201_channels); indio_dev->modes = INDIO_DIRECT_MODE; ret = adis_init(st, indio_dev, spi, &adis16201_data); if (ret) return ret;
A blank line here would help readability by separating these two unconnected operations.
Quoted 4 lines
ret = adis_setup_buffer_and_trigger(st, indio_dev, NULL); if (ret) return ret;
Quoted 8 lines
/* Get the device into a sane initial state */ ret = adis_initial_startup(st); if (ret) goto error_cleanup_buffer_trigger; ret = iio_device_register(indio_dev); if (ret < 0) goto error_cleanup_buffer_trigger;
Blank line here.
Quoted 4 lines
return 0;
error_cleanup_buffer_trigger:adis_cleanup_buffer_and_trigger(st, indio_dev);
blank line here.
Quoted 14 lines
return ret; }static int
adis16201_remove(struct spi_device *spi){ struct iio_dev *indio_dev = spi_get_drvdata(spi); struct adis *st = iio_priv(indio_dev); iio_device_unregister(indio_dev); adis_cleanup_buffer_and_trigger(st, indio_dev); return 0; }
Quoted 8 lines
static struct spi_driver adis16201_driver = { .driver = { .name = "adis16201", }, .probe = adis16201_probe, .remove = adis16201_remove, };
module_spi_driver(adis16201_driver);
Probably want to add devicetree bindings but not strictly necessary to move out of staging...
Quoted 5 lines
MODULE_AUTHOR("Barry Song <[email protected]>");MODULE_DESCRIPTION("Analog Devices ADIS16201 Dual-Axis Digital Inclinometer and Accelerometer");MODULE_LICENSE("GPL v2");MODULE_ALIAS("spi:adis16201");
Hi Jonathan,
Quoted 22 lines · Jonathan Cameron
Hi All
I was looking over what we have left in staging for any low hanging fruit and came across the adis16201 driver.
I was wondering if anyone would like to clean it up?
This would make a good opportuntity for someone who is fairly new to kernel development to make a useful contribution.
To that end I've done a review as if it had been submitted as a new driver.
If you want to take this on, please reply to say so in order to avoid several people looking at this one driver. Obviously if things change and you then can't get to it reasonably quickly then 'hand it back' on the mailing list.
I'm happy to do this for a few more drivers if there are enough interested people so do ping me if someone has claimed this one and you would like a similar starting point.
Quoted 8 lines
Ideally we would find someone who actually has hardware, but for the simple stuff in this one I think we are unlikely to break anything that can't be picked up in review. Hence I'm not totally bothered about it if no one comes forward!
Alison, Daniel. Feel free to add this to the outreachy task list if no one else grabs it before the process starts again (just let us know if you are doing that).
I would love to clean this up as I have just started learning about IIO subsytem.
Do you want incremental patches everyday or all work done in a patchset while taking time ?
Though I'm not new to kernel development but yes I am new to IIO community.
Signature · 2 lines
Thanks Himanshu Jha
Quoted 22 lines · Jonathan Cameron
Hi All
I was looking over what we have left in staging for any low hanging fruit and came across the adis16201 driver.
I was wondering if anyone would like to clean it up?
This would make a good opportuntity for someone who is fairly new to kernel development to make a useful contribution.
To that end I've done a review as if it had been submitted as a new driver.
If you want to take this on, please reply to say so in order to avoid several people looking at this one driver. Obviously if things change and you then can't get to it reasonably quickly then 'hand it back' on the mailing list.
I'm happy to do this for a few more drivers if there are enough interested people so do ping me if someone has claimed this one and you would like a similar starting point.
Quoted 8 lines
Ideally we would find someone who actually has hardware, but for the simple stuff in this one I think we are unlikely to break anything that can't be picked up in review. Hence I'm not totally bothered about it if no one comes forward!
Alison, Daniel. Feel free to add this to the outreachy task list if no one else grabs it before the process starts again (just let us know if you are doing that).
Hi Jonathan,
Will keep this in mind for next round of Outreachy that starts in March.
I also submitted a proposal for GSoC:
https://wiki.linuxfoundation.org/gsoc/2018-gsoc-iio-driver
Let's see if this is accepted and we can have people working for the cleanups as a part of their application process.
thanks,
Daniel.
Quoted 37 lines · Himanshu Jha
Hi Jonathan,
Quoted 3 lines · Jonathan Cameron
Hi All
=20
I was looking over what we have left in staging for any low hangingfruit
Quoted 4 lines
and came across the adis16201 driver=2E
=20
I was wondering if anyone would like to clean it up? =20Quoted 8 lines
This would make a good opportuntity for someone who is fairly new to kernel development to make a useful contribution=2E
=20
To that end I've done a review as if it had been submitted as a new driver=2E
=20
If you want to take this on, please reply to say so in order to avoid several people looking at this one driver=2E Obviously if thingschange
Quoted 5 lines
and you then can't get to it reasonably quickly then 'hand it back' on the mailing list=2E
=20
I'm happy to do this for a few more drivers if there are enough interested people so do ping me if someone has claimed this one andyou
Quoted 3 lines
would like a similar starting point=2E
=20
Ideally we would find someone who actually has hardware, but for thesimple
Quoted 1 line
stuff in this one I think we are unlikely to break anything that
can't be
Quoted 1 line
picked up in review=2E Hence I'm not totally bothered about it if no
one
Quoted 3 lines
comes forward!
=20
Alison, Daniel=2E Feel free to add this to the outreachy task list if
Quoted 7 lines
no
Quoted 1 line
one else grabs it before the process starts again (just let us know
if you are
Quoted 1 line
doing that)=2E
I would love to clean this up as I have just started learning about IIO subsytem=2E
Cool
Quoted 3 lines
Do you want incremental patches everyday or all work done in a patchset while taking time ?
I am fine with either way, but tend to take a little while to get to revie= wing so might be easier to=20
go for a series=2E
Quoted 3 lines
Though I'm not new to kernel development but yes I am new to IIO community=2E
Great=2E
Jonathan
--=20
Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E
Hi jonathan,
I would like to clean adis 16203.
Thanks,
Nijam Haider
Quoted 555 lines · On Sun, 4 Feb 2018 at 20:56 Jonathan Cameron <[email protected]> wrote:
Hi All
I was looking over what we have left in staging for any low hanging fruit and came across the adis16201 driver.
I was wondering if anyone would like to clean it up?
This would make a good opportuntity for someone who is fairly new to kernel development to make a useful contribution.
To that end I've done a review as if it had been submitted as a new driver.
If you want to take this on, please reply to say so in order to avoid several people looking at this one driver. Obviously if things change and you then can't get to it reasonably quickly then 'hand it back' on the mailing list.
I'm happy to do this for a few more drivers if there are enough interested people so do ping me if someone has claimed this one and you would like a similar starting point.
Ideally we would find someone who actually has hardware, but for the simple stuff in this one I think we are unlikely to break anything that can't be picked up in review. Hence I'm not totally bothered about it if no one comes forward!
Alison, Daniel. Feel free to add this to the outreachy task list if no one else grabs it before the process starts again (just let us know if you are
doing that).Thanks,
Jonathan
Quoted 6 lines
/*
* ADIS16201 Dual-Axis Digital Inclinometer and Accelerometer *
* Copyright 2010 Analog Devices Inc.
*
* Licensed under the GPL-2 or later.Might as well convert to SPDX format for the license if we are working on the driver anyway.
Quoted 10 lines
*/
#include <
linux/delay.h>
#include <linux/mutex.h>
#include <linux/device.h>
#include <linux/kernel.h>
#include <linux/spi/spi.h>
#include <linux/slab.h>
#include <linux/sysfs.h>
#include <linux/module.h>Slight preference for alphabetical order in headers.
However, it is useful to keep the iio headers separate as they are here.
Quoted 7 lines
#include <
linux/iio/iio.h>
#include <linux/iio/sysfs.h>
#include <linux/iio/buffer.h>
#include <linux/iio/imu/adis.h>#define
ADIS16201_STARTUP_DELAY220 /* ms */Why not just make this
ADIS16201_STARTUP_DELAY_MSas then it will be obvious what the units are throughout the driver and remove the need for the comment.Quoted 3 lines
/* Flash memory write count */
#defineADIS16201_FLASH_CNT0x00To my mind there is far more commenting going on here than is actually useful. If the name of the register doesn't make it reasonably clear what it is used for then it is not a good choice of name.
It is made more complex by the fact these are the datasheet names so we need to keep 'close' to those names for ease of comparisom.
So unless something odd is going on I would not expect to see any comments in here at all.
Formatting wise, should be using tabs rather than spaces before the numbers.
Also, we have a mixture of defines for register addresses and fields with in registers. That is fine, but I would prefer the nice clear
_REGpostfix on the registers defines to make it more obvious which is which.e.g. What I'd expect here is something like:
#define ADIS16201_FLASH_CNT_REG 0x00 #define ADIS16201_SUPPLY_OUT_REG 0x02 #define ADIS16201_XACCL_OUT_REG 0x04etc.
Quoted 3 lines
/* Output, power supply */
#defineADIS16201_SUPPLY_OUT0x02The out naming is awkward given it doesn't mean the device is supplying the power, but rather that this is the reading the device is outputting. Still can't do that much about this without breaking away from datasheet naming.
Quoted 76 lines
/* Output, x-axis accelerometer */
#defineADIS16201_XACCL_OUT0x04/* Output, y-axis accelerometer */
#defineADIS16201_YACCL_OUT0x06/* Output, auxiliary ADC input */
#defineADIS16201_AUX_ADC0x08/* Output, temperature */
#defineADIS16201_TEMP_OUT0x0A/* Output, x-axis inclination */
#defineADIS16201_XINCL_OUT0x0C/* Output, y-axis inclination */
#defineADIS16201_YINCL_OUT0x0E/* Calibration, x-axis acceleration offset */
#defineADIS16201_XACCL_OFFS0x10/* Calibration, y-axis acceleration offset */
#defineADIS16201_YACCL_OFFS0x12/* x-axis acceleration scale factor */
#defineADIS16201_XACCL_SCALE0x14/* y-axis acceleration scale factor */
#defineADIS16201_YACCL_SCALE0x16/* Calibration, x-axis inclination offset */
#defineADIS16201_XINCL_OFFS0x18/* Calibration, y-axis inclination offset */
#defineADIS16201_YINCL_OFFS0x1A/* x-axis inclination scale factor */
#defineADIS16201_XINCL_SCALE0x1C/* y-axis inclination scale factor */
#defineADIS16201_YINCL_SCALE0x1E/* Alarm 1 amplitude threshold */
#defineADIS16201_ALM_MAG10x20/* Alarm 2 amplitude threshold */
#defineADIS16201_ALM_MAG20x22/* Alarm 1, sample period */
#defineADIS16201_ALM_SMPL10x24/* Alarm 2, sample period */
#defineADIS16201_ALM_SMPL20x26/* Alarm control */
#defineADIS16201_ALM_CTRL0x28/* Auxiliary DAC data */
#defineADIS16201_AUX_DAC0x30/* General-purpose digital input/output control */
#defineADIS16201_GPIO_CTRL0x32/* Miscellaneous control */
#defineADIS16201_MSC_CTRL0x34/* Internal sample period (rate) control */
#defineADIS16201_SMPL_PRD0x36/* Operation, filter configuration */
#defineADIS16201_AVG_CNT0x38/* Operation, sleep mode control */
#defineADIS16201_SLP_CNT0x3AQuoted 8 lines
/* Diagnostics, system status register */
#defineADIS16201_DIAG_STAT0x3C/* Operation, system command register */
#defineADIS16201_GLOB_CMD0x3E/*
MSC_CTRL*/A really nice convention that has come up in recent drivers and makes all these register / register field relationships obvious is to use some subtle indenting of the fields and put them next to the register address. An example is:
#define ADIS16201_MSC_CTRL_REG 0x34 #define ADIS16201_MSC_CTRL_SELF_TEST_EN BIT(8) #define ADIS16201_MSC_CTRL_DATA_RDY_EN BIT(2) #define ADIS16201_MSC_CTRL_ACTIVE_HIGH BIT(1) #define ADIS16201_MSC_CTRL_DATA_RDY_DIO1 BIT(0)Again the naming of the fields is sufficient and the documentation doesn't add anything so I would drop it.
Quoted 49 lines
/* Self-test enable */
#defineADIS16201_MSC_CTRL_SELF_TEST_ENBIT(8)/* Data-ready enable: 1 = enabled, 0 = disabled */
#defineADIS16201_MSC_CTRL_DATA_RDY_ENBIT(2)/* Data-ready polarity: 1 = active high, 0 = active low */
#defineADIS16201_MSC_CTRL_ACTIVE_HIGHBIT(1)/* Data-ready line selection: 1 = DIO1, 0 = DIO0 */
#defineADIS16201_MSC_CTRL_DATA_RDY_DIO1BIT(0)/*
DIAG_STAT*//* Alarm 2 status: 1 = alarm active, 0 = alarm inactive */
#defineADIS16201_DIAG_STAT_ALARM2BIT(9)/* Alarm 1 status: 1 = alarm active, 0 = alarm inactive */
#defineADIS16201_DIAG_STAT_ALARM1BIT(8)/* SPI communications failure */
#defineADIS16201_DIAG_STAT_SPI_FAIL_BIT3/* Flash update failure */
#defineADIS16201_DIAG_STAT_FLASH_UPT_BIT2/* Power supply above 3.625 V */
#defineADIS16201_DIAG_STAT_POWER_HIGH_BIT1/* Power supply below 3.15 V */
#defineADIS16201_DIAG_STAT_POWER_LOW_BIT0/*
GLOB_CMD*/#define
ADIS16201_GLOB_CMD_SW_RESETBIT(7)
#defineADIS16201_GLOB_CMD_FACTORY_CALBIT(1)#define
ADIS16201_ERROR_ACTIVEBIT(14)enum adis16201_scan { ADIS16201_SCAN_ACC_X, ADIS16201_SCAN_ACC_Y, ADIS16201_SCAN_INCLI_X, ADIS16201_SCAN_INCLI_Y, ADIS16201_SCAN_SUPPLY, ADIS16201_SCAN_AUX_ADC, ADIS16201_SCAN_TEMP, };Quoted 1 line
static const u8 adis16201_addresses[] = {
It is rather odd that this is only used for the offsets, not any of the other channel specific registers (scales, or the reading itself).
Quoted 10 lines
[ADIS16201_SCAN_ACC_X] = ADIS16201_XACCL_OFFS, [ADIS16201_SCAN_ACC_Y] = ADIS16201_YACCL_OFFS, [ADIS16201_SCAN_INCLI_X] = ADIS16201_XINCL_OFFS, [ADIS16201_SCAN_INCLI_Y] = ADIS16201_YINCL_OFFS, };static int adis16201_read_raw(struct
iio_dev*indio_dev,
structiio_chan_specconst *chan, int *val, int *val2, long mask)Quoted 11 lines
{ struct adis *st = iio_priv(indio_dev); int ret; int bits; u8 addr; s16 val16; switch (mask) { case IIO_CHAN_INFO_RAW: return adis_single_conversion(indio_dev, chan, ADIS16201_ERROR_ACTIVE, val);Indenting could I think make the parameters align with the opening bracket.
Quoted 6 lines
case IIO_CHAN_INFO_SCALE: switch (chan->type) { case IIO_VOLTAGE: if (chan->channel == 0) { *val = 1; *val2 = 220000; /* 1.22 mV */Comments don't add a huge amount given they kind of just state the value before them.
Quoted 1 line
} else {Given there are only two relevant channels I would like to see them move explicitly matched using a switch statement.
The default case would be an error return
-EINVAL;Quoted 11 lines
*val = 0; *val2 = 610000; /* 0.610 mV */ } return IIO_VAL_INT_PLUS_MICRO; case IIO_TEMP: *val = -470; /* 0.47 C */ *val2 = 0; return IIO_VAL_INT_PLUS_MICRO; case IIO_ACCEL: *val = 0; *val2 = IIO_G_TO_M_S_2(462400); /* 0.4624 mg */Not sure that comment adds anything..
Quoted 4 lines
return IIO_VAL_INT_PLUS_NANO; case IIO_INCLI: *val = 0; *val2 = 100000; /* 0.1 degree */Or this one.
Quoted 16 lines
return IIO_VAL_INT_PLUS_MICRO; default: return -EINVAL; } break; case IIO_CHAN_INFO_OFFSET: *val = 25000 / -470 - 1278; /* 25 C = 1278 */ return IIO_VAL_INT; case IIO_CHAN_INFO_CALIBBIAS: switch (chan->type) { case IIO_ACCEL: bits = 12; break; case IIO_INCLI: bits = 9; break;Quoted 8 lines
default: return -EINVAL; } addr = adis16201_addresses[chan->scan_index]; ret = adis_read_reg_16(st, addr, &val16); if (ret) return ret; val16 &= (1 << bits) - 1;Given the following sign extend, I don't think this does anything useful.
Quoted 1 line
val16 = (s16)(val16 << (16 - bits)) >> (16 - bits);This looks like a 16bit sign extend to me. We have functions to handle this nicely. See how it is done in other driverss
Quoted 1 line
*val = val16;blank line before return always makes them more obvious and hence the code slightly more readable. Do this throughout (including 2 lines below this)
Quoted 21 lines
return IIO_VAL_INT; } return -EINVAL; }static int adis16201_write_raw(struct
iio_dev*indio_dev,
structiio_chan_specconst *chan, int val,
int val2,
long mask){ struct adis *st = iio_priv(indio_dev); int bits; s16 val16; u8 addr; switch (mask) { case IIO_CHAN_INFO_CALIBBIAS: switch (chan->type) { case IIO_ACCEL: bits = 12;Quoted 8 lines
break; case IIO_INCLI: bits = 9; break; default: return -EINVAL; } val16 = val & ((1 << bits) - 1);Given the above bits value is used only for this, why not create a mask in the switch statement instead of carrying the number of bits around.
Then you can use GENMASK to make it nice and obvious what is going on.
Quoted 8 lines
addr = adis16201_addresses[chan->scan_index]; return adis_write_reg_16(st, addr, val16); } return -EINVAL; }static const struct iio_chan_spec adis16201_channels[] = { ADIS_SUPPLY_CHAN(ADIS16201_SUPPLY_OUT, ADIS16201_SCAN_SUPPLY, 0,12),
Quoted 6 lines
ADIS_TEMP_CHAN(ADIS16201_TEMP_OUT, ADIS16201_SCAN_TEMP, 0, 12), ADIS_ACCEL_CHAN(X, ADIS16201_XACCL_OUT, ADIS16201_SCAN_ACC_X, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_ACCEL_CHAN(Y, ADIS16201_YACCL_OUT, ADIS16201_SCAN_ACC_Y, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_AUX_ADC_CHAN(ADIS16201_AUX_ADC, ADIS16201_SCAN_AUX_ADC, 0,12),
Quoted 37 lines
ADIS_INCLI_CHAN(X, ADIS16201_XINCL_OUT, ADIS16201_SCAN_INCLI_X, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_INCLI_CHAN(X, ADIS16201_YINCL_OUT, ADIS16201_SCAN_INCLI_Y, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), IIO_CHAN_SOFT_TIMESTAMP(7) };static const struct iio_info adis16201_info = { .read_raw = adis16201_read_raw, .write_raw = adis16201_write_raw, .update_scan_mode = adis_update_scan_mode, };static const char * const adis16201_status_error_msgs[] = { [ADIS16201_DIAG_STAT_SPI_FAIL_BIT] = "SPI failure", [ADIS16201_DIAG_STAT_FLASH_UPT_BIT] = "Flash update failed", [ADIS16201_DIAG_STAT_POWER_HIGH_BIT] = "Power supply above 3.625V", [ADIS16201_DIAG_STAT_POWER_LOW_BIT] = "Power supply below 3.15V", };static const struct adis_data adis16201_data = { .read_delay = 20, .msc_ctrl_reg = ADIS16201_MSC_CTRL, .glob_cmd_reg = ADIS16201_GLOB_CMD, .diag_stat_reg = ADIS16201_DIAG_STAT, .self_test_mask = ADIS16201_MSC_CTRL_SELF_TEST_EN, .self_test_no_autoclear = true, .startup_delay = ADIS16201_STARTUP_DELAY, .status_error_msgs = adis16201_status_error_msgs, .status_error_mask = BIT(ADIS16201_DIAG_STAT_SPI_FAIL_BIT) | BIT(ADIS16201_DIAG_STAT_FLASH_UPT_BIT) | BIT(ADIS16201_DIAG_STAT_POWER_HIGH_BIT) | BIT(ADIS16201_DIAG_STAT_POWER_LOW_BIT), };Quoted 5 lines
static int
adis16201_probe(struct spi_device *spi){ int ret; struct adis *st; struct iio_dev *indio_dev;There is a convention (when nothing else is guiding ordering) of doing things in reverse Christmas tree.
That means longest line first and shortest last. Very minor point though.
Quoted 2 lines
/* setup the industrialio driver allocated elements */Comment doesn't really add anything so drop it.
Quoted 6 lines
indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st)); if (!indio_dev) return -ENOMEM; st = iio_priv(indio_dev); /* this is only used for removal purposes */Again, comment isn't really that useful...
Quoted 6 lines
spi_set_drvdata(spi, indio_dev); indio_dev->name = spi->dev.driver->name; indio_dev->dev.parent = &spi->dev; indio_dev->info = &adis16201_info;Quoted 7 lines
indio_dev->channels = adis16201_channels; indio_dev->num_channels = ARRAY_SIZE(adis16201_channels); indio_dev->modes = INDIO_DIRECT_MODE; ret = adis_init(st, indio_dev, spi, &adis16201_data); if (ret) return ret;A blank line here would help readability by separating these two unconnected operations.
Quoted 4 lines
ret = adis_setup_buffer_and_trigger(st, indio_dev, NULL); if (ret) return ret;Quoted 8 lines
/* Get the device into a sane initial state */ ret = adis_initial_startup(st); if (ret) goto error_cleanup_buffer_trigger; ret = iio_device_register(indio_dev); if (ret < 0) goto error_cleanup_buffer_trigger;Blank line here.
Quoted 4 lines
return 0;
error_cleanup_buffer_trigger:adis_cleanup_buffer_and_trigger(st, indio_dev);blank line here.
Quoted 14 lines
return ret; }static int
adis16201_remove(struct spi_device *spi){ struct iio_dev *indio_dev = spi_get_drvdata(spi); struct adis *st = iio_priv(indio_dev); iio_device_unregister(indio_dev); adis_cleanup_buffer_and_trigger(st, indio_dev); return 0; }Quoted 8 lines
static struct spi_driver adis16201_driver = { .driver = { .name = "adis16201", }, .probe = adis16201_probe, .remove = adis16201_remove, };
module_spi_driver(adis16201_driver);Probably want to add devicetree bindings but not strictly necessary to move out of staging...
Quoted 3 lines
MODULE_AUTHOR("Barry Song <[email protected]>"); MODULE_DESCRIPTION("Analog Devices ADIS16201 Dual-Axis DigitalInclinometer and Accelerometer");
Quoted 2 lines
MODULE_LICENSE("GPL v2");MODULE_ALIAS("spi:adis16201");--
To unsubscribe from this list: send the line "unsubscribe linux-iio" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Quoted 3 lines · nizam haider
Hi jonathan,
I would like to clean adis 16203.
Cool. Anyone else, shout if you want a suggestion of what to tackle.
If not I will do some more of these Todo reviews as time allows.
Jonathan
Quoted 588 lines
Thanks,
Nijam HaiderQuoted 3 lines · On Sun, 4 Feb 2018 at 20:56 Jonathan Cameron <[email protected]> wrote:
Hi All
I was looking over what we have left in staging for any low hanging
fruit
Quoted 4 lines
and came across the adis16201 driver.
I was wondering if anyone would like to clean it up?
Quoted 8 lines
This would make a good opportuntity for someone who is fairly new to kernel development to make a useful contribution.
To that end I've done a review as if it had been submitted as a new driver.
If you want to take this on, please reply to say so in order to avoid several people looking at this one driver. Obviously if things
change
Quoted 5 lines
and you then can't get to it reasonably quickly then 'hand it back' on the mailing list.
I'm happy to do this for a few more drivers if there are enough interested people so do ping me if someone has claimed this one and
you
Quoted 3 lines
would like a similar starting point.
Ideally we would find someone who actually has hardware, but for the
simple
Quoted 1 line
stuff in this one I think we are unlikely to break anything that
can't be
Quoted 1 line
picked up in review. Hence I'm not totally bothered about it if no
one
Quoted 3 lines
comes forward!
Alison, Daniel. Feel free to add this to the outreachy task list if
no
Quoted 1 line
one else grabs it before the process starts again (just let us know
if you
Quoted 46 lines
are
doing that).Thanks,
Jonathan
Quoted 6 lines
/*
* ADIS16201 Dual-Axis Digital Inclinometer and Accelerometer *
* Copyright 2010 Analog Devices Inc.
*
* Licensed under the GPL-2 or later.Might as well convert to SPDX format for the license if we are working on the driver anyway.
Quoted 10 lines
*/
#include <
linux/delay.h>
#include <linux/mutex.h>
#include <linux/device.h>
#include <linux/kernel.h>
#include <linux/spi/spi.h>
#include <linux/slab.h>
#include <linux/sysfs.h>
#include <linux/module.h>Slight preference for alphabetical order in headers.
However, it is useful to keep the iio headers separate as they are here.
Quoted 7 lines
#include <
linux/iio/iio.h>
#include <linux/iio/sysfs.h>
#include <linux/iio/buffer.h>
#include <linux/iio/imu/adis.h>#define
ADIS16201_STARTUP_DELAY220 /* ms */Why not just make this
ADIS16201_STARTUP_DELAY_MSas then it will be obvious what the units are throughout the driver and remove the need for the comment.Quoted 1 line
Quoted 4 lines
Quoted 2 lines
/* Flash memory write count */
#defineADIS16201_FLASH_CNT0x00To my mind there is far more commenting going on here than is
actually
Quoted 23 lines
useful. If the name of the register doesn't make it reasonably clear what it is used for then it is not a good choice of name.
It is made more complex by the fact these are the datasheet names so we need to keep 'close' to those names for ease of comparisom.
So unless something odd is going on I would not expect to see any comments in here at all.
Formatting wise, should be using tabs rather than spaces before the numbers.
Also, we have a mixture of defines for register addresses and fields with in registers. That is fine, but I would prefer the nice clear
_REGpostfix on the registers defines to make it more obvious which is which.e.g. What I'd expect here is something like:
#define ADIS16201_FLASH_CNT_REG 0x00 #define ADIS16201_SUPPLY_OUT_REG 0x02 #define ADIS16201_XACCL_OUT_REG 0x04Quoted 7 lines
etc.
Quoted 3 lines
/* Output, power supply */
#defineADIS16201_SUPPLY_OUT0x02The out naming is awkward given it doesn't mean the device is
supplying the
Quoted 1 line
power, but rather that this is the reading the device is outputting.
Still
Quoted 1 line
can't do that much about this without breaking away from datasheet
naming.
Quoted 171 lines
Quoted 77 lines
/* Output, x-axis accelerometer */
#defineADIS16201_XACCL_OUT0x04/* Output, y-axis accelerometer */
#defineADIS16201_YACCL_OUT0x06/* Output, auxiliary ADC input */
#defineADIS16201_AUX_ADC0x08/* Output, temperature */
#defineADIS16201_TEMP_OUT0x0A/* Output, x-axis inclination */
#defineADIS16201_XINCL_OUT0x0C/* Output, y-axis inclination */
#defineADIS16201_YINCL_OUT0x0E/* Calibration, x-axis acceleration offset */
#defineADIS16201_XACCL_OFFS0x10/* Calibration, y-axis acceleration offset */
#defineADIS16201_YACCL_OFFS0x12/* x-axis acceleration scale factor */
#defineADIS16201_XACCL_SCALE0x14/* y-axis acceleration scale factor */
#defineADIS16201_YACCL_SCALE0x16/* Calibration, x-axis inclination offset */
#defineADIS16201_XINCL_OFFS0x18/* Calibration, y-axis inclination offset */
#defineADIS16201_YINCL_OFFS0x1A/* x-axis inclination scale factor */
#defineADIS16201_XINCL_SCALE0x1C/* y-axis inclination scale factor */
#defineADIS16201_YINCL_SCALE0x1E/* Alarm 1 amplitude threshold */
#defineADIS16201_ALM_MAG10x20/* Alarm 2 amplitude threshold */
#defineADIS16201_ALM_MAG20x22/* Alarm 1, sample period */
#defineADIS16201_ALM_SMPL10x24/* Alarm 2, sample period */
#defineADIS16201_ALM_SMPL20x26/* Alarm control */
#defineADIS16201_ALM_CTRL0x28/* Auxiliary DAC data */
#defineADIS16201_AUX_DAC0x30/* General-purpose digital input/output control */
#defineADIS16201_GPIO_CTRL0x32/* Miscellaneous control */
#defineADIS16201_MSC_CTRL0x34/* Internal sample period (rate) control */
#defineADIS16201_SMPL_PRD0x36/* Operation, filter configuration */
#defineADIS16201_AVG_CNT0x38/* Operation, sleep mode control */
#defineADIS16201_SLP_CNT0x3AQuoted 8 lines
/* Diagnostics, system status register */
#defineADIS16201_DIAG_STAT0x3C/* Operation, system command register */
#defineADIS16201_GLOB_CMD0x3E/*
MSC_CTRL*/A really nice convention that has come up in recent drivers and makes all these register / register field relationships obvious is to use some subtle indenting of the fields and put them next to the register address. An example is:
#define ADIS16201_MSC_CTRL_REG 0x34 #define ADIS16201_MSC_CTRL_SELF_TEST_EN BIT(8) #define ADIS16201_MSC_CTRL_DATA_RDY_EN BIT(2) #define ADIS16201_MSC_CTRL_ACTIVE_HIGH BIT(1) #define ADIS16201_MSC_CTRL_DATA_RDY_DIO1 BIT(0)Again the naming of the fields is sufficient and the documentation doesn't add anything so I would drop it.
Quoted 49 lines
/* Self-test enable */
#defineADIS16201_MSC_CTRL_SELF_TEST_ENBIT(8)/* Data-ready enable: 1 = enabled, 0 = disabled */
#defineADIS16201_MSC_CTRL_DATA_RDY_ENBIT(2)/* Data-ready polarity: 1 = active high, 0 = active low */
#defineADIS16201_MSC_CTRL_ACTIVE_HIGHBIT(1)/* Data-ready line selection: 1 = DIO1, 0 = DIO0 */
#defineADIS16201_MSC_CTRL_DATA_RDY_DIO1BIT(0)/*
DIAG_STAT*//* Alarm 2 status: 1 = alarm active, 0 = alarm inactive */
#defineADIS16201_DIAG_STAT_ALARM2BIT(9)/* Alarm 1 status: 1 = alarm active, 0 = alarm inactive */
#defineADIS16201_DIAG_STAT_ALARM1BIT(8)/* SPI communications failure */
#defineADIS16201_DIAG_STAT_SPI_FAIL_BIT3/* Flash update failure */
#defineADIS16201_DIAG_STAT_FLASH_UPT_BIT2/* Power supply above 3.625 V */
#defineADIS16201_DIAG_STAT_POWER_HIGH_BIT1/* Power supply below 3.15 V */
#defineADIS16201_DIAG_STAT_POWER_LOW_BIT0/*
GLOB_CMD*/#define
ADIS16201_GLOB_CMD_SW_RESETBIT(7)
#defineADIS16201_GLOB_CMD_FACTORY_CALBIT(1)#define
ADIS16201_ERROR_ACTIVEBIT(14)enum adis16201_scan { ADIS16201_SCAN_ACC_X, ADIS16201_SCAN_ACC_Y, ADIS16201_SCAN_INCLI_X, ADIS16201_SCAN_INCLI_Y, ADIS16201_SCAN_SUPPLY, ADIS16201_SCAN_AUX_ADC, ADIS16201_SCAN_TEMP, };Quoted 1 line
static const u8 adis16201_addresses[] = {
It is rather odd that this is only used for the offsets, not any of the other channel specific registers (scales, or the reading itself).
Quoted 17 lines
[ADIS16201_SCAN_ACC_X] = ADIS16201_XACCL_OFFS, [ADIS16201_SCAN_ACC_Y] = ADIS16201_YACCL_OFFS, [ADIS16201_SCAN_INCLI_X] = ADIS16201_XINCL_OFFS, [ADIS16201_SCAN_INCLI_Y] = ADIS16201_YINCL_OFFS, };static int adis16201_read_raw(struct
iio_dev*indio_dev,
structiio_chan_specconst *chan, int *val, int *val2, long mask){ struct adis *st = iio_priv(indio_dev); int ret; int bits; u8 addr; s16 val16;Quoted 6 lines
Quoted 4 lines
switch (mask) { case IIO_CHAN_INFO_RAW: return adis_single_conversion(indio_dev, chan, ADIS16201_ERROR_ACTIVE, val);Indenting could I think make the parameters align with the opening
bracket.
Quoted 11 lines
Quoted 7 lines
case IIO_CHAN_INFO_SCALE: switch (chan->type) { case IIO_VOLTAGE: if (chan->channel == 0) { *val = 1; *val2 = 220000; /* 1.22 mV */Comments don't add a huge amount given they kind of just state the value before them.
Quoted 18 lines
Quoted 1 line
} else {Given there are only two relevant channels I would like to see them move explicitly matched using a switch statement.
The default case would be an error return
-EINVAL;Quoted 11 lines
*val = 0; *val2 = 610000; /* 0.610 mV */ } return IIO_VAL_INT_PLUS_MICRO; case IIO_TEMP: *val = -470; /* 0.47 C */ *val2 = 0; return IIO_VAL_INT_PLUS_MICRO; case IIO_ACCEL: *val = 0; *val2 = IIO_G_TO_M_S_2(462400); /* 0.4624 mg*/
Quoted 35 lines
Not sure that comment adds anything..
Quoted 4 lines
return IIO_VAL_INT_PLUS_NANO; case IIO_INCLI: *val = 0; *val2 = 100000; /* 0.1 degree */Or this one.
Quoted 16 lines
return IIO_VAL_INT_PLUS_MICRO; default: return -EINVAL; } break; case IIO_CHAN_INFO_OFFSET: *val = 25000 / -470 - 1278; /* 25 C = 1278 */ return IIO_VAL_INT; case IIO_CHAN_INFO_CALIBBIAS: switch (chan->type) { case IIO_ACCEL: bits = 12; break; case IIO_INCLI: bits = 9; break;Quoted 8 lines
default: return -EINVAL; } addr = adis16201_addresses[chan->scan_index]; ret = adis_read_reg_16(st, addr, &val16); if (ret) return ret; val16 &= (1 << bits) - 1;Quoted 1 line
Given the following sign extend, I don't think this does anything
useful.
Quoted 42 lines
Quoted 2 lines
val16 = (s16)(val16 << (16 - bits)) >> (16 - bits);This looks like a 16bit sign extend to me. We have functions to handle this nicely. See how it is done in other driverss
Quoted 1 line
*val = val16;blank line before return always makes them more obvious and hence the code slightly more readable. Do this throughout (including 2 lines below this)
Quoted 21 lines
return IIO_VAL_INT; } return -EINVAL; }static int adis16201_write_raw(struct
iio_dev*indio_dev,
structiio_chan_specconst *chan, int val,
int val2,
long mask){ struct adis *st = iio_priv(indio_dev); int bits; s16 val16; u8 addr; switch (mask) { case IIO_CHAN_INFO_CALIBBIAS: switch (chan->type) { case IIO_ACCEL: bits = 12;Quoted 8 lines
break; case IIO_INCLI: bits = 9; break; default: return -EINVAL; } val16 = val & ((1 << bits) - 1);Quoted 15 lines
Given the above bits value is used only for this, why not create a mask in the switch statement instead of carrying the number of bits around.
Then you can use GENMASK to make it nice and obvious what is going on.
Quoted 8 lines
addr = adis16201_addresses[chan->scan_index]; return adis_write_reg_16(st, addr, val16); } return -EINVAL; }static const struct iio_chan_spec adis16201_channels[] = { ADIS_SUPPLY_CHAN(ADIS16201_SUPPLY_OUT, ADIS16201_SCAN_SUPPLY,0,
Quoted 2 lines
12),
Quoted 1 line
ADIS_TEMP_CHAN(ADIS16201_TEMP_OUT, ADIS16201_SCAN_TEMP, 0,12),
Quoted 5 lines
Quoted 5 lines
ADIS_ACCEL_CHAN(X, ADIS16201_XACCL_OUT, ADIS16201_SCAN_ACC_X, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_ACCEL_CHAN(Y, ADIS16201_YACCL_OUT, ADIS16201_SCAN_ACC_Y, BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_AUX_ADC_CHAN(ADIS16201_AUX_ADC, ADIS16201_SCAN_AUX_ADC,0,
Quoted 2 lines
12),
Quoted 1 line
ADIS_INCLI_CHAN(X, ADIS16201_XINCL_OUT,
ADIS16201_SCAN_INCLI_X,Quoted 2 lines
Quoted 2 lines
BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), ADIS_INCLI_CHAN(X, ADIS16201_YINCL_OUT,
ADIS16201_SCAN_INCLI_Y,Quoted 10 lines
Quoted 10 lines
BIT(IIO_CHAN_INFO_CALIBBIAS), 0, 14), IIO_CHAN_SOFT_TIMESTAMP(7) };static const struct iio_info adis16201_info = { .read_raw = adis16201_read_raw, .write_raw = adis16201_write_raw, .update_scan_mode = adis_update_scan_mode, };Quoted 4 lines
Quoted 4 lines
static const char * const adis16201_status_error_msgs[] = { [ADIS16201_DIAG_STAT_SPI_FAIL_BIT] = "SPI failure", [ADIS16201_DIAG_STAT_FLASH_UPT_BIT] = "Flash update failed", [ADIS16201_DIAG_STAT_POWER_HIGH_BIT] = "Power supply above3.625V",
Quoted 1 line
Quoted 1 line
[
ADIS16201_DIAG_STAT_POWER_LOW_BIT] = "Power supply below3.15V",
Quoted 19 lines
Quoted 19 lines
};
static const struct adis_data adis16201_data = { .read_delay = 20, .msc_ctrl_reg = ADIS16201_MSC_CTRL, .glob_cmd_reg = ADIS16201_GLOB_CMD, .diag_stat_reg = ADIS16201_DIAG_STAT, .self_test_mask = ADIS16201_MSC_CTRL_SELF_TEST_EN, .self_test_no_autoclear = true, .startup_delay = ADIS16201_STARTUP_DELAY, .status_error_msgs = adis16201_status_error_msgs, .status_error_mask = BIT(ADIS16201_DIAG_STAT_SPI_FAIL_BIT) | BIT(ADIS16201_DIAG_STAT_FLASH_UPT_BIT) | BIT(ADIS16201_DIAG_STAT_POWER_HIGH_BIT) | BIT(ADIS16201_DIAG_STAT_POWER_LOW_BIT), };Quoted 6 lines
Quoted 5 lines
static int
adis16201_probe(struct spi_device *spi){ int ret; struct adis *st; struct iio_dev *indio_dev;There is a convention (when nothing else is guiding ordering) of
doing
Quoted 3 lines
things in reverse Christmas tree.
That means longest line first and shortest last. Very minor point
though.
Quoted 77 lines
Quoted 3 lines
/* setup the industrialio driver allocated elements */Comment doesn't really add anything so drop it.
Quoted 6 lines
indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st)); if (!indio_dev) return -ENOMEM; st = iio_priv(indio_dev); /* this is only used for removal purposes */Again, comment isn't really that useful...
Quoted 6 lines
spi_set_drvdata(spi, indio_dev); indio_dev->name = spi->dev.driver->name; indio_dev->dev.parent = &spi->dev; indio_dev->info = &adis16201_info;Quoted 7 lines
indio_dev->channels = adis16201_channels; indio_dev->num_channels = ARRAY_SIZE(adis16201_channels); indio_dev->modes = INDIO_DIRECT_MODE; ret = adis_init(st, indio_dev, spi, &adis16201_data); if (ret) return ret;A blank line here would help readability by separating these two unconnected operations.
Quoted 4 lines
ret = adis_setup_buffer_and_trigger(st, indio_dev, NULL); if (ret) return ret;Quoted 8 lines
/* Get the device into a sane initial state */ ret = adis_initial_startup(st); if (ret) goto error_cleanup_buffer_trigger; ret = iio_device_register(indio_dev); if (ret < 0) goto error_cleanup_buffer_trigger;Blank line here.
Quoted 4 lines
return 0;
error_cleanup_buffer_trigger:adis_cleanup_buffer_and_trigger(st, indio_dev);blank line here.
Quoted 14 lines
return ret; }static int
adis16201_remove(struct spi_device *spi){ struct iio_dev *indio_dev = spi_get_drvdata(spi); struct adis *st = iio_priv(indio_dev); iio_device_unregister(indio_dev); adis_cleanup_buffer_and_trigger(st, indio_dev); return 0; }Quoted 8 lines
static struct spi_driver adis16201_driver = { .driver = { .name = "adis16201", }, .probe = adis16201_probe, .remove = adis16201_remove, };
module_spi_driver(adis16201_driver);Quoted 1 line
Probably want to add devicetree bindings but not strictly necessary
to
Quoted 10 lines
move out of staging...
Quoted 3 lines
MODULE_AUTHOR("Barry Song <[email protected]>"); MODULE_DESCRIPTION("Analog Devices ADIS16201 Dual-Axis DigitalInclinometer and Accelerometer");
Quoted 2 lines
MODULE_LICENSE("GPL v2");MODULE_ALIAS("spi:adis16201");--
To unsubscribe from this list: send the line "unsubscribe linux-iio"in
Quoted 3 lines
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Signature · 1 line
Sent from my Android device with K-9 Mail. Please excuse my brevity.