Re: [PATCH] Added AMS/TAOS tsl2591 driver implementation

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Sat, 23 Jan 2021 23:32:36 +0000
Joe Sandom <joe.sandom@xxxxxxxxxxx> wrote:

> Driver implementation for AMS/TAOS tsl2591 ambient light sensor.
> 
> This driver supports configuration via device tree and sysfs.
> Supported channels for raw infrared light, raw visible light,
> raw combined light and combined lux value.
> The driver additionally supports iio events on lower and
> upper thresholds.
> 
> This is a very-high sensitivity light-to-digital converter that
> transforms light intensity into a digital signal.
> 
> Datasheet Available at: https://ams.com/tsl25911
> 
> Signed-off-by: Joe Sandom <joe.sandom@xxxxxxxxxxx>

Hi Joe,

Welcome to IIO :)

Some comments inline.  Mostly about getting the driver to fit into
standard userspace ABI.  Sometimes it takes a bit of work, but it
makes the driver useable without needing custom userspace code.

Thanks,

Jonathan

> ---
>  .../bindings/iio/light/tsl2591.yaml           |  100 ++

Filename for binding should be prefixed with the vendor prefix.
(we should fix the other tslxxxx parts that all have this wrong currently)

>  drivers/iio/light/Kconfig                     |   12 +
>  drivers/iio/light/Makefile                    |    1 +
>  drivers/iio/light/tsl2591.c                   | 1304 +++++++++++++++++
>  4 files changed, 1417 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/iio/light/tsl2591.yaml
>  create mode 100644 drivers/iio/light/tsl2591.c
> 
> diff --git a/Documentation/devicetree/bindings/iio/light/tsl2591.yaml b/Documentation/devicetree/bindings/iio/light/tsl2591.yaml
> new file mode 100644
> index 000000000000..1fb9cffff987
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/iio/light/tsl2591.yaml
> @@ -0,0 +1,100 @@
> +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/iio/light/tsl2591.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#

Bindings should be a separate patch in the series - reduces the searching that
binding reviewers need to do to find the bit they are interested in.
(obviously that happens to be easy here :)

> +
> +title: AMS/TAOS TSL2591 Ambient Light Sensor (ALS)
> +
> +maintainers:
> +  - Joe Sandom <joe.sandom@xxxxxxxxxxx>
> +
> +description: |
> +  Device driver for the TAOS TSL2591. This is a very-high sensitivity

This is a binding, not a driver, so don't mention the 'driver' bit in a binding
file.  I'd start at "This is"..

> +  light-to-digital converter that transforms light intensity into a digital
> +  signal.
> +
> +properties:
> +  compatible:
> +    const: amstaos,tsl2591
> +
> +  reg:
> +    maxItems: 1
> +
> +  interrupts:
> +    maxItems: 1
> +    description:
> +      Interrupt (INT:Pin 2) Active low. Should be set to IRQ_TYPE_EDGE_FALLING.
> +      interrupt is used to detect if lux value has reached below or above defined
> +      threshold values.
> +
> +  als-integration-time:
> +    $ref: /schemas/types.yaml#/definitions/uint32
> +    enum: [0x00, 0x01, 0x02, 0x03, 0x04, 0x05]
> +    default: 0x02
> +    description:
> +      Sensor integration time defines time taken for ADC to read samples.
> +      0x00=100ms, 0x01=200ms, 0x02=300ms, 0x03=400ms, 0x05=500ms, 0x06=600ms.
Please use the real world values for something like this and name it so that
is apparent.  Also, this isn't yet a generic binding so should be prefixed
with amstaos,

    amstaos,als-integration-time-msecs
      enum: [100, 200, 300, 400, 500, 600]
      default: 200
      description:
        etc.

The description doesn't define a typical integration time. So I'd drop it or
describe as something like:
     "Time over which incident light is aggregated by the sensor"
(not great either...)

> +      Additionally, configurable via sysfs.

Then why do we have it in devicetree?  If it makes sense to configure from
userspace then it's not a property of the system so probably doesn't want
to have a binding at all.

> +
> +
> +  als-gain:
> +    $ref: /schemas/types.yaml#/definitions/uint32
> +    enum: [0x00, 0x10, 0x20, 0x30]
> +    default: 0x10
> +    description:
> +      Sensor gain defines internal amplifiers for photodiode channels.
> +      0x00=Low, 0x10=Medium, 0x20=High, 0x30=Maximum.
> +      Additionally, configurable via sysfs.
As above, if it makes sense to have it in sysfs, probably not here.

Also figure 9 gives typical multiplier values, use those not names.
enum: [1, 25, 428, 9876]

> +
> +
> +  als-persist:
> +    $ref: /schemas/types.yaml#/definitions/uint32
> +    enum: [0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F]
> +    default: 0x01
> +    description:
> +      Sensor persist defines the number of consecutive cycles out of range before
> +      an interrupt is triggered.
> +      0x00=Every Cycle (test mode), 0x01=ANY, 0x02=2, 0x03=3, 0x04=5, 0x05=10, 0x06=15, 0x07=20,
> +      0x08=25, 0x09=30, 0x0A=35, 0x0B=40, 0x0C=45, 0x0D=50, 0x0E=55, 0x0F=60.

So this will need to map to _period in IIO and that will require taking sampling frequency into
account.  However, this definitely doesn't belong in DT as very much dependent on the
use case, not anything about the way the hardware is integrated with the device etc.

> +
> +
> +  als-lower-threshold:
> +    minimum: 100
> +    maximum: 65535
> +    description:
> +      Sensor lower threshold defines the lower raw threshold. If the photodiode channels
> +      read below this threshold for als-persist defined cycles, then an interrupt is triggered.
> +
> +  als-upper-threshold:
> +    minimum: 1500
> +    maximum: 65535
> +    description:
> +      Sensor upper threshold defines the upper raw threshold. If the photodiode channels
> +      read above this threshold for als-persist defined cycles, then an interrupt is triggered.

These are both things that need to be configure at run time, so not in DT.

> +
> +required:
> +  - compatible
> +  - reg
> +
> +additionalProperties: false
> +
> +examples:
> +  - |
> +    #include <dt-bindings/interrupt-controller/irq.h>
> +    i2c {
> +        #address-cells = <1>;
> +        #size-cells = <0>;
> +
> +        tsl2591@29 {
> +            compatible = "amstaos,tsl2591";
> +            reg = <0x29>;
> +            interrupts = <20 IRQ_TYPE_EDGE_FALLING>;
> +            als-integration-time = <0x03>;
> +            als-gain = <0x10>;
> +            als-persist = <0x04>;
> +            als-lower-threshold = <100>;
> +            als-upper-threshold = <1000>;
> +       };
> +    };
> +...
> diff --git a/drivers/iio/light/Kconfig b/drivers/iio/light/Kconfig
> index 33ad4dd0b5c7..8d2c75da3198 100644
> --- a/drivers/iio/light/Kconfig
> +++ b/drivers/iio/light/Kconfig
> @@ -519,6 +519,18 @@ config TSL4531
>  	  To compile this driver as a module, choose M here: the
>  	  module will be called tsl4531.
>  
> +config TSL2591
> +        tristate "TAOS TSL2591 ambient light sensor"
> +        depends on I2C
> +        help
> +          Select Y here for support of the TAOS TSL2591 ambient light sensor,
> +          featuring visible and infrared lux output. Access als data via
> +          iio and sysfs. Supports iio_events.
> +
> +          To compile this driver as a module, select M: the
> +          module will be called tsl2591.
> +
One line break only + numeric order.
> +
>  config US5182D
>  	tristate "UPISEMI light and proximity sensor"
>  	depends on I2C
> diff --git a/drivers/iio/light/Makefile b/drivers/iio/light/Makefile
> index ea376deaca54..6084ea6e118f 100644
> --- a/drivers/iio/light/Makefile
> +++ b/drivers/iio/light/Makefile
> @@ -50,6 +50,7 @@ obj-$(CONFIG_TCS3472)		+= tcs3472.o
>  obj-$(CONFIG_TSL2583)		+= tsl2583.o
>  obj-$(CONFIG_TSL2772)		+= tsl2772.o
>  obj-$(CONFIG_TSL4531)		+= tsl4531.o
> +obj-$(CONFIG_TSL2591)		+= tsl2591.o

Numeric order please.

>  obj-$(CONFIG_US5182D)		+= us5182d.o
>  obj-$(CONFIG_VCNL4000)		+= vcnl4000.o
>  obj-$(CONFIG_VCNL4035)		+= vcnl4035.o
> diff --git a/drivers/iio/light/tsl2591.c b/drivers/iio/light/tsl2591.c
> new file mode 100644
> index 000000000000..c3381dc51e8b
> --- /dev/null
> +++ b/drivers/iio/light/tsl2591.c
> @@ -0,0 +1,1304 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (C) 2020 Joe Sandom <joe.sandom@xxxxxxxxxxx>
> + *
> + * Datasheet Available at: https://ams.com/tsl25911
> + *
> + * @brief Device driver for the TAOS TSL2591. This is a very-high sensitivity
> + * light-to-digital converter that transforms light intensity into a digital
> + * signal.

This isn't kernel-doc, so the @brief isn't going to get picked up by anything.
I'd just drop that.

> + */
> +
> +#include <linux/kernel.h>
> +#include <linux/i2c.h>
> +#include <linux/module.h>
> +#include <linux/mutex.h>
> +#include <linux/pm_runtime.h>
> +#include <linux/debugfs.h>
> +#include <linux/delay.h>
> +#include <linux/interrupt.h>
Prefer alphabetical order for generic headers.  Fine to keep iio ones
clustered at the end in an IIO driver though.

> +#include <linux/iio/iio.h>
> +#include <linux/iio/sysfs.h>
> +#include <linux/iio/events.h>
> +
> +#define als_time_secs_to_ms(x) (((x) + 1) * 100)
> +#define als_time_ms_to_secs(x) (((x) / 100) - 1)
> +
> +/* TSL2591 Register Set */
> +#define TSL2591_ENABLE      0x00
> +#define TSL2591_CONTROL     0x01
> +#define TSL2591_AILTL       0x04
> +#define TSL2591_AILTH       0x05
> +#define TSL2591_AIHTL       0x06
> +#define TSL2591_AIHTH       0x07
> +#define TSL2591_NP_AILTL    0x08
> +#define TSL2591_NP_AILTH    0x09
> +#define TSL2591_NP_AIHTL    0x0A
> +#define TSL2591_NP_AIHTH    0x0B
> +#define TSL2591_PERSIST     0x0C
> +#define TSL2591_PACKAGE_ID  0x11
> +#define TSL2591_DEVICE_ID   0x12
> +#define TSL2591_STATUS      0x13
> +#define TSL2591_C0_DATAL    0x14
> +#define TSL2591_C0_DATAH    0x15
> +#define TSL2591_C1_DATAL    0x16
> +#define TSL2591_C1_DATAH    0x17
> +
> +/* TSL2591 Command Register Masks */
> +#define TSL2591_CMD_NOP             0xA0
> +#define TSL2591_CMD_SF_INTSET       0xE4
> +#define TSL2591_CMD_SF_CALS_I       0xE5
> +#define TSL2591_CMD_SF_CALS_NPI     0xE7
> +#define TSL2591_CMD_SF_CNP_ALSI     0xEA
> +
> +/* TSL2591 Enable Register Masks */
> +#define TSL2591_PWR_ON              0x01
> +#define TSL2591_PWR_OFF             0x00
> +#define TSL2591_ENABLE_ALS          0x02
> +#define TSL2591_ENABLE_ALS_INT      0x10
> +#define TSL2591_ENABLE_SLEEP_INT    0x40
> +#define TSL2591_ENABLE_NP_INT       0x80
> +
> +/* TSL2591 Control Register Masks */
> +#define TSL2591_CTRL_ALS_INTEGRATION_100MS  0x00
> +#define TSL2591_CTRL_ALS_INTEGRATION_200MS  0x01
> +#define TSL2591_CTRL_ALS_INTEGRATION_300MS  0x02
> +#define TSL2591_CTRL_ALS_INTEGRATION_400MS  0x03
> +#define TSL2591_CTRL_ALS_INTEGRATION_500MS  0x04
> +#define TSL2591_CTRL_ALS_INTEGRATION_600MS  0x05
> +#define TSL2591_CTRL_ALS_LOW_GAIN           0x00
> +#define TSL2591_CTRL_ALS_MED_GAIN           0x10
> +#define TSL2591_CTRL_ALS_HIGH_GAIN          0x20
> +#define TSL2591_CTRL_ALS_MAX_GAIN           0x30
> +#define TSL2591_CTRL_SYS_RESET              0x80
> +
> +/* TSL2591 Persist Register Masks */
> +#define TSL2591_PRST_ALS_INT_CYCLE_0        0x00
> +#define TSL2591_PRST_ALS_INT_CYCLE_ANY      0x01
> +#define TSL2591_PRST_ALS_INT_CYCLE_2        0x02
> +#define TSL2591_PRST_ALS_INT_CYCLE_3        0x03
> +#define TSL2591_PRST_ALS_INT_CYCLE_5        0x04
> +#define TSL2591_PRST_ALS_INT_CYCLE_10       0x05
> +#define TSL2591_PRST_ALS_INT_CYCLE_15       0x06
> +#define TSL2591_PRST_ALS_INT_CYCLE_20       0x07
> +#define TSL2591_PRST_ALS_INT_CYCLE_25       0x08
> +#define TSL2591_PRST_ALS_INT_CYCLE_30       0x09
> +#define TSL2591_PRST_ALS_INT_CYCLE_35       0x0A
> +#define TSL2591_PRST_ALS_INT_CYCLE_40       0x0B
> +#define TSL2591_PRST_ALS_INT_CYCLE_45       0x0C
> +#define TSL2591_PRST_ALS_INT_CYCLE_50       0x0D
> +#define TSL2591_PRST_ALS_INT_CYCLE_55       0x0E
> +#define TSL2591_PRST_ALS_INT_CYCLE_60       0x0F
> +
> +/* TSL2591 PID Register Mask */
> +#define TSL2591_PACKAGE_ID_MASK 0x30
> +
> +/* TSL2591 ID Register Mask */
> +#define TSL2591_DEVICE_ID_MASK  0xFF
> +
> +/* TSL2591 Status Register Masks */
> +#define TSL2591_STS_ALS_VALID   0x01
> +#define TSL2591_STS_ALS_INT     0x10
> +#define TSL2591_STS_NPERS_INT   0x20
> +#define TSL2591_STS_VAL_HIGH    0x01
> +
> +/* TSL2591 Constant Values */
> +#define TSL2591_PACKAGE_ID_VAL  0x00
> +#define TSL2591_DEVICE_ID_VAL   0x50
> +
> +/* Power off suspend delay time MS */
> +#define TSL2591_POWER_OFF_DELAY_MS	2000
> +
> +/* TSL2591 Default Values */
> +#define DEFAULT_ALS_INTEGRATION_TIME    TSL2591_CTRL_ALS_INTEGRATION_300MS
> +#define MAX_ALS_INTEGRATION_TIME_MS     600
> +#define DEFAULT_ALS_GAIN                TSL2591_CTRL_ALS_MED_GAIN
> +#define NUMBER_OF_DATA_CHANNELS         4
> +#define DEFAULT_ALS_PERSIST             TSL2591_PRST_ALS_INT_CYCLE_ANY
> +#define DEFAULT_ALS_LOWER_THRESHOLD	100
> +#define DEFAULT_ALS_UPPER_THRESHOLD	1500
> +
> +#define ALS_MIN_VALUE	0
> +#define ALS_MAX_VALUE	65535
> +#define ALS_STS_VALID_COUNT 10

Prefix everything.  These generic names may well clash with some standard
header definition sometime in the future + lack of prefix implies they
aren't local which can cause confusion.

> +
> +/* Literals */
> +#define TSL2591_CTRL_ALS_LOW_GAIN_LIT   "low"
> +#define TSL2591_CTRL_ALS_MED_GAIN_LIT   "med"
> +#define TSL2591_CTRL_ALS_HIGH_GAIN_LIT  "high"
> +#define TSL2591_CTRL_ALS_MAX_GAIN_LIT   "max"
> +
> +/* LUX Calculations */
> +/* AGAIN values from Adafruits TSL2591 Arduino library */
> +/* https://github.com/adafruit/Adafruit_TSL2591_Library */
> +#define TSL2591_CTRL_ALS_LOW_AGAIN   1
> +#define TSL2591_CTRL_ALS_MED_AGAIN   25
> +#define TSL2591_CTRL_ALS_HIGH_AGAIN  428
> +#define TSL2591_CTRL_ALS_MAX_AGAIN   9876
> +#define TSL2591_LUX_COEFFICIENT      408
> +
> +static const u8 tsl2591_data_channels[] = {
> +	TSL2591_C0_DATAL,
> +	TSL2591_C0_DATAH,
> +	TSL2591_C1_DATAL,
> +	TSL2591_C1_DATAH,
> +};
> +
> +struct tsl2591_als_readings {
> +	u16 als_ch0;
> +	u16 als_ch1;
> +	u16 als_visible;
> +	u16 als_lux_int;
> +	u16 als_lux_decimal;
> +};
> +
> +struct tsl2591_settings {
> +	u8 als_int_time;
> +	u8 als_gain;
> +	u8 als_persist;
> +	u16 als_lower_threshold;
> +	u16 als_upper_threshold;
> +};
> +
> +struct tsl2591_chip {
> +	/* Ambient light sensor mutex */
> +	struct mutex als_mutex;
> +	struct i2c_client *client;
> +	struct tsl2591_settings als_settings;
> +	struct tsl2591_als_readings als_readings;
> +};
> +
> +static char *tsl2591_gain_to_str(const u8 als_gain)
> +{
> +	char *gain_str;
> +
> +	switch (als_gain) {
> +	case TSL2591_CTRL_ALS_LOW_GAIN:
> +		gain_str = TSL2591_CTRL_ALS_LOW_GAIN_LIT;
> +		break;
> +	case TSL2591_CTRL_ALS_MED_GAIN:
> +		gain_str = TSL2591_CTRL_ALS_MED_GAIN_LIT;
> +		break;
> +	case TSL2591_CTRL_ALS_HIGH_GAIN:
> +		gain_str = TSL2591_CTRL_ALS_HIGH_GAIN_LIT;
> +		break;
> +	case TSL2591_CTRL_ALS_MAX_GAIN:
> +		gain_str = TSL2591_CTRL_ALS_MAX_GAIN_LIT;
> +		break;
> +	default:
> +		gain_str = "error";
> +		break;
> +	}
> +
> +	return gain_str;
> +}
> +
> +static int tsl2591_gain_to_again(const u8 als_gain)
> +{
> +	int a_gain;
> +
> +	switch (als_gain) {
> +	case TSL2591_CTRL_ALS_LOW_GAIN:
> +		a_gain = TSL2591_CTRL_ALS_LOW_AGAIN;
> +		break;
> +	case TSL2591_CTRL_ALS_MED_GAIN:
> +		a_gain = TSL2591_CTRL_ALS_MED_AGAIN;
> +		break;
> +	case TSL2591_CTRL_ALS_HIGH_GAIN:
> +		a_gain = TSL2591_CTRL_ALS_HIGH_AGAIN;
> +		break;
> +	case TSL2591_CTRL_ALS_MAX_GAIN:
> +		a_gain = TSL2591_CTRL_ALS_MAX_AGAIN;
> +		break;
> +	default:
> +		a_gain = -EINVAL;
> +		break;
> +	}
> +
> +	return a_gain;
> +}
> +
> +static int tsl2591_gain_from_str(const char *als_gain_str)

Numbers, they are on the datasheet so ignore the names it also gives
them as much less useful to anyone :)

> +{
> +	if (strstr(als_gain_str, TSL2591_CTRL_ALS_LOW_GAIN_LIT))
> +		return TSL2591_CTRL_ALS_LOW_GAIN;
> +	else if (strstr(als_gain_str, TSL2591_CTRL_ALS_MED_GAIN_LIT))
> +		return TSL2591_CTRL_ALS_MED_GAIN;
> +	else if (strstr(als_gain_str, TSL2591_CTRL_ALS_HIGH_GAIN_LIT))
> +		return TSL2591_CTRL_ALS_HIGH_GAIN;
> +	else if (strstr(als_gain_str, TSL2591_CTRL_ALS_MAX_GAIN_LIT))
> +		return TSL2591_CTRL_ALS_MAX_GAIN;
> +	else
> +		return -EINVAL;
> +}
> +
> +static int tsl2591_compatible_int_time(struct tsl2591_chip *chip,
> +				       u32 als_integration_time)
> +{
> +	switch (als_integration_time) {
> +	case TSL2591_CTRL_ALS_INTEGRATION_100MS:
> +	case TSL2591_CTRL_ALS_INTEGRATION_200MS:
> +	case TSL2591_CTRL_ALS_INTEGRATION_300MS:
> +	case TSL2591_CTRL_ALS_INTEGRATION_400MS:
> +	case TSL2591_CTRL_ALS_INTEGRATION_500MS:
> +	case TSL2591_CTRL_ALS_INTEGRATION_600MS:
> +		chip->als_settings.als_int_time = als_integration_time;
> +		return 0;
> +	default:
> +		break;
> +	}
> +
> +	return -EINVAL;
> +}
> +
> +static int tsl2591_compatible_gain(struct tsl2591_chip *chip, u32 als_gain)
> +{
> +	switch (als_gain) {
> +	case TSL2591_CTRL_ALS_LOW_GAIN:
> +	case TSL2591_CTRL_ALS_MED_GAIN:
> +	case TSL2591_CTRL_ALS_HIGH_GAIN:
> +	case TSL2591_CTRL_ALS_MAX_GAIN:
> +		chip->als_settings.als_gain = als_gain;
> +		return 0;
> +	default:
> +		break;
> +	}
> +
> +	return -EINVAL;
> +}
> +
> +static int tsl2591_compatible_als_persist(struct tsl2591_chip *chip,
> +					  u32 als_persist)
> +{
> +	switch (als_persist) {
> +	case TSL2591_PRST_ALS_INT_CYCLE_0:

If you want to support this, it needs to be done as a dataready
signal and wrapped up in a trigger + buffered output etc.  For a light
sensor I'd not bother (they are too slow).  So I'd reject this as
a value.

> +	case TSL2591_PRST_ALS_INT_CYCLE_ANY:

To get this into standard ABI you'll need to deal with this in seconds
rather than cycles (its _period btw)

> +	case TSL2591_PRST_ALS_INT_CYCLE_2:
> +	case TSL2591_PRST_ALS_INT_CYCLE_3:
> +	case TSL2591_PRST_ALS_INT_CYCLE_5:
> +	case TSL2591_PRST_ALS_INT_CYCLE_10:
> +	case TSL2591_PRST_ALS_INT_CYCLE_15:
> +	case TSL2591_PRST_ALS_INT_CYCLE_20:
> +	case TSL2591_PRST_ALS_INT_CYCLE_25:
> +	case TSL2591_PRST_ALS_INT_CYCLE_30:
> +	case TSL2591_PRST_ALS_INT_CYCLE_35:
> +	case TSL2591_PRST_ALS_INT_CYCLE_40:
> +	case TSL2591_PRST_ALS_INT_CYCLE_45:
> +	case TSL2591_PRST_ALS_INT_CYCLE_50:
> +	case TSL2591_PRST_ALS_INT_CYCLE_55:
> +	case TSL2591_PRST_ALS_INT_CYCLE_60:
> +		chip->als_settings.als_persist = als_persist;
> +		return 0;
> +	default:
> +		break;
> +	}
> +
> +	return -EINVAL;
> +}
> +
> +static int tsl2591_wait_adc_complete(struct tsl2591_chip *chip)
> +{
> +	struct i2c_client *client = chip->client;
> +	struct tsl2591_settings settings = chip->als_settings;
> +	int delay = als_time_secs_to_ms(settings.als_int_time);
> +	int ret;
> +	int i;
> +
> +	if (!delay) {
> +		delay = MAX_ALS_INTEGRATION_TIME_MS;
> +		dev_warn(&chip->client->dev,
> +			 "Failed to get int time, setting default delay: %d\n",
> +			delay)

That is rather worrying if you can get here. I'm guessing you can't in which
case if you want to be paranoid just return an error if we do get here rather than
trying to muddle on.

> +	}
> +
> +	/*
> +	 * Sleep for als integration time to allow enough time
> +	 * for an ADC read cycle to complete. Check status after
> +	 * delay for als valid
> +	 */
> +	msleep(delay);
> +
> +	/* Check for status als valid flag for up to 100ms */
> +	for (i = 0; i < ALS_STS_VALID_COUNT; ++i) {
> +		ret = i2c_smbus_read_byte_data(client, TSL2591_CMD_NOP |
> +			TSL2591_STATUS);
> +
> +		if (ret < 0) {
> +			dev_err(&client->dev, "%s:failed to read register\n", __func__);
> +			return -EINVAL;
> +		}
> +
> +		if ((ret & TSL2591_STS_ALS_VALID) == TSL2591_STS_VAL_HIGH)
> +			break;
> +
> +		if (i == (ALS_STS_VALID_COUNT - 1))
> +			return -ENODATA;
> +
> +		usleep_range(9000, 10000);
> +	}
> +
> +	return 0;
> +}
> +
> +/**
> + * tsl2591_get_lux_data() - Reads raw channel data and calculates lux
> + *
> + * Formula for lux calculation;
> + * Derived from Adafruit's TSL2591 library
> + * Link: https://github.com/adafruit/Adafruit_TSL2591_Library
> + * Counts Per Lux (CPL) = (ATIME_ms * AGAIN) / LUX DF
> + * lux = ((C0DATA - C1DATA) * (1 - (C1DATA / C0DATA))) / CPL
> + *
> + * Scale values to get more representative value of lux i.e.
> + * lux = ((C0DATA - C1DATA) * (1000 - ((C1DATA * 1000) / C0DATA))) / CPL
> + *
> + * Channel 0 = IR + Visible
> + * Channel 1 = IR only
> + * Visible = Channel 0 - Channel 1
> + *
> + */
> +static int tsl2591_get_lux_data(struct iio_dev *indio_dev)
> +{
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct tsl2591_settings *settings = &chip->als_settings;
> +	struct i2c_client *client = chip->client;
> +	int i;
> +	int ret;
> +	u8 channel_data[NUMBER_OF_DATA_CHANNELS];
> +
> +	int counts_per_lux;
> +	int lux;
> +
> +	ret = tsl2591_wait_adc_complete(chip);
> +	if (ret < 0) {
> +		dev_warn(&client->dev, "No data available. Err: %d\n", ret);
> +		return -ENODATA;
> +	}
> +
> +	for (i = 0; i < NUMBER_OF_DATA_CHANNELS; ++i) {
> +		int reg = TSL2591_CMD_NOP | tsl2591_data_channels[i];
> +
> +		ret = i2c_smbus_read_byte_data(client, TSL2591_CMD_NOP | reg);
> +
> +		if (ret < 0) {
> +			dev_err(&client->dev,
> +				"%s: failed to read register %#04x\n",
> +				__func__, reg);
> +			return -EINVAL;
> +		}
> +		channel_data[i] = ret;
> +	}
> +
> +	chip->als_readings.als_ch0 =
> +		le16_to_cpup((const __le16 *)&channel_data[0]);
> +	chip->als_readings.als_ch1 =
> +		le16_to_cpup((const __le16 *)&channel_data[2]);
> +
> +	dev_dbg(&client->dev, "both: %d\n", chip->als_readings.als_ch0);
> +	dev_dbg(&client->dev, "ir: %d\n", chip->als_readings.als_ch1);
> +
> +	/* Visible light is channel 0 - channel 1 i.e. both - IR = visible */
> +	chip->als_readings.als_visible =
> +		chip->als_readings.als_ch0 - chip->als_readings.als_ch1;

I'm not sure I can see that this is very useful to calculate or expose.
It doesn't really have much meaning.

> +
> +	/* Calculate counts per lux value */
> +	counts_per_lux = (als_time_secs_to_ms(settings->als_int_time) *
> +		tsl2591_gain_to_again(settings->als_gain)) /
> +		TSL2591_LUX_COEFFICIENT;
> +
> +	dev_dbg(&client->dev, "Counts Per Lux (CPL): %d\n", counts_per_lux);
> +
> +	/* Calculate lux value */
> +	lux = ((chip->als_readings.als_ch0 - chip->als_readings.als_ch1) *
> +	(1000 - ((chip->als_readings.als_ch1 * 1000) /
> +	chip->als_readings.als_ch0))) / counts_per_lux;
> +
> +	/* Divide by 1000 to get real lux value before scaling */
> +	chip->als_readings.als_lux_int = lux / 1000;
> +
> +	/* Get the decimal part of lux reading */
> +	chip->als_readings.als_lux_decimal =
> +		(lux - (chip->als_readings.als_lux_int * 1000));
> +
> +	dev_dbg(&client->dev, "Raw lux calculation: %d\n", lux);
> +
> +	dev_dbg(&client->dev, "Lux Value: %d.%d\n",
> +		chip->als_readings.als_lux_int,
> +		chip->als_readings.als_lux_decimal);
> +
> +	return ret;
> +}
> +
> +static int tsl2591_als_calibrate(struct tsl2591_chip *chip)
This appears to be setting the integration time and gain. I'm not
following why it would be called _calibrate()?
> +{
> +	struct i2c_client *client = chip->client;
> +	struct tsl2591_settings als_settings = chip->als_settings;
> +	int ret;
> +
> +	u8 configuration;
> +
> +	dev_dbg(&client->dev, "Setting configuration - als_int_time: %#04x\n",
> +		als_settings.als_int_time);
> +	dev_dbg(&client->dev, "Setting configuration - als_gain: %#04x\n",
> +		als_settings.als_gain);
These are readable from userspace, so this is overly noisy. I'd drop the
dev_dbg.

It's normal to have lots of this during development and only retain those
things where there isn't an easy way to read them after development is done.

> +
> +	configuration = als_settings.als_int_time | als_settings.als_gain;
> +
> +	ret = i2c_smbus_write_byte_data(client,
> +					TSL2591_CMD_NOP | TSL2591_CONTROL, configuration);
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set configuration to %#04x\n", __func__,
> +			configuration);
> +
> +	return ret;
> +}
> +
> +static int tsl2591_clear_als_irq(struct tsl2591_chip *chip)
> +{
> +	struct i2c_client *client = chip->client;
> +	int ret;
> +
> +	ret = i2c_smbus_write_byte(client, TSL2591_CMD_SF_CALS_NPI);
> +	if (ret < 0)
> +		dev_err(&client->dev, "failed to clear als irq\n");

No point in having a wrapper for this. Just do it inline as that
will be easier to review.

> +
> +	return ret;
> +}
> +
> +static int tsl2591_als_thresholds(struct tsl2591_chip *chip)
> +{
> +	struct i2c_client *client = chip->client;
> +	struct tsl2591_settings als_settings = chip->als_settings;
> +	int ret;
> +
> +	u8 als_lower_l;
> +	u8 als_lower_h;
> +	u8 als_upper_l;
> +	u8 als_upper_h;
> +
> +	if (als_settings.als_lower_threshold >= als_settings.als_upper_threshold) {
> +		dev_warn(&client->dev, "lower threshold higher than upper\n");
> +		return -EINVAL;
> +	}
> +
> +	if (als_settings.als_upper_threshold > ALS_MAX_VALUE) {
> +		dev_warn(&client->dev, "upper threshold higher than max\n");
> +		return -EINVAL;
> +	}
> +
> +	als_lower_l = (als_settings.als_lower_threshold & 0x00FF);
> +	als_lower_h = ((als_settings.als_lower_threshold >> 8) & 0x00FF);
> +	als_upper_l = (als_settings.als_upper_threshold & 0x00FF);
> +	als_upper_h = ((als_settings.als_upper_threshold >> 8) & 0x00FF);
> +
> +	dev_dbg(&client->dev, "Setting configuration - als lower l: %#04x\n",
> +		als_lower_l);
> +	dev_dbg(&client->dev, "Setting configuration - als lower h: %#04x\n",
> +		als_lower_h);
> +	dev_dbg(&client->dev, "Setting configuration - als upper l: %#04x\n",
> +		als_upper_l);
> +	dev_dbg(&client->dev, "Setting configuration - als upper h: %#04x\n",
> +		als_upper_h);
> +
> +	ret = i2c_smbus_write_byte_data(client, TSL2591_CMD_NOP |
> +		TSL2591_PERSIST, als_settings.als_persist);
> +
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set als persist any\n", __func__);
> +
> +	ret = i2c_smbus_write_byte_data(client, TSL2591_CMD_NOP |
> +		TSL2591_AILTL, als_lower_l);
> +
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set als lower threshold\n", __func__);
> +
> +	ret = i2c_smbus_write_byte_data(client, TSL2591_CMD_NOP |
> +		TSL2591_AILTH, als_lower_h);
> +
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set als lower threshold\n", __func__);
> +
> +	ret = i2c_smbus_write_byte_data(client, TSL2591_CMD_NOP |
> +		TSL2591_AIHTL, als_upper_l);
> +
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set als upper threshold\n", __func__);
> +
> +	ret = i2c_smbus_write_byte_data(client, TSL2591_CMD_NOP |
> +		TSL2591_AIHTH, als_upper_h);
> +
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set als upper threshold\n", __func__);
> +
> +	return ret;
> +}
> +
> +static int tsl2591_set_power_state(struct tsl2591_chip *chip, u8 state)
> +{
> +	struct i2c_client *client = chip->client;
> +	int ret;
> +
> +	ret = i2c_smbus_write_byte_data(client, TSL2591_CMD_NOP |
> +					TSL2591_ENABLE, state);
> +	if (ret < 0)
> +		dev_err(&client->dev,
> +			"%s: failed to set the power state to %#04x\n", __func__,
> +			state);
> +
> +	return ret;
> +}
> +
> +static int tsl2591_set_pm_runtime_busy(struct tsl2591_chip *chip, bool busy)
> +{
> +	struct i2c_client *client = chip->client;
> +	int ret;
> +
> +	if (!client->irq) {
Given the two cases are totally different code, I'd have

tsl2591_set_pm_runtime_busy()
tsl2591_set_pm_runtime_not_busy()

+ as below, the condition shouldn't be the presence of the irq, but rather if
events have been enabled or not.

The runtime pm framework uses reference counts for all of this, so you
should be able to get away with just elevating them with a pm_runtime_get_sync()
in the event enable path.  Then this lot becomes unconditional (it won't do
anything if events are enabled)

Once you've done that, then put this code inline rather than hiding it
in wrapper. Much easier to follow that way.

> +		if (busy) {
> +			ret = pm_runtime_get_sync(&chip->client->dev);
> +			if (ret < 0)
> +				pm_runtime_put_noidle(&chip->client->dev);
> +		} else {
> +			pm_runtime_mark_last_busy(&chip->client->dev);
> +			ret = pm_runtime_put_autosuspend(&chip->client->dev);
> +		}
> +
> +		return ret;
> +	}
> +
> +	dev_dbg(&client->dev, "irq enabled, skipping pm runtime\n");
> +
> +	return 0;
> +}
> +
> +static ssize_t in_illuminance_integration_time_show(struct device *dev,
> +						    struct device_attribute *attr,
> +						    char *buf)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	int als_int_time = als_time_secs_to_ms(chip->als_settings.als_int_time);
> +
> +	mutex_lock(&chip->als_mutex);
> +	ret = sprintf(buf, "%d\n", als_int_time);
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static ssize_t in_illuminance_integration_time_store(struct device *dev,
> +						     struct device_attribute *attr,
> +						     const char *buf, size_t len)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct i2c_client *client = chip->client;
> +
> +	u32 int_time;
> +	int value;
> +
> +	if (kstrtoint(buf, 0, &value) || !value)
> +		return -EINVAL;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	int_time = als_time_ms_to_secs(value);
> +
> +	if (tsl2591_compatible_int_time(chip, int_time))
> +		goto calibrate_error;
> +
> +	if (tsl2591_als_calibrate(chip))
> +		goto calibrate_error;
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return len;
> +
> +calibrate_error:
> +	dev_err(&client->dev, "Failed to calibrate sensor\n");
> +	mutex_unlock(&chip->als_mutex);
> +	return -EINVAL;
> +}
> +
> +static ssize_t in_illuminance_gain_show(struct device *dev,
> +					struct device_attribute *attr,
> +					char *buf)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	char *gain = tsl2591_gain_to_str(chip->als_settings.als_gain);
> +
> +	mutex_lock(&chip->als_mutex);
> +	ret = sprintf(buf, "%s\n", gain);
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static ssize_t in_illuminance_gain_store(struct device *dev,
> +					 struct device_attribute *attr,
> +					 const char *buf, size_t len)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct i2c_client *client = chip->client;
> +
> +	int gain;
> +
> +	if (!buf)
> +		return -EINVAL;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	gain = tsl2591_gain_from_str(buf);
> +	if (gain == -EINVAL)
> +		goto calibrate_error;
> +
> +	chip->als_settings.als_gain = gain;
> +
> +	if (tsl2591_als_calibrate(chip))
> +		goto calibrate_error;
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return len;
> +
> +calibrate_error:
> +	dev_err(&client->dev, "Failed to calibrate sensor\n");
> +	mutex_unlock(&chip->als_mutex);
> +	return -EINVAL;
> +}
> +
> +static ssize_t in_illuminance_lower_threshold_show(struct device *dev,
> +						   struct device_attribute *attr,
> +						   char *buf)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	mutex_lock(&chip->als_mutex);
> +	ret = sprintf(buf, "%d\n", chip->als_settings.als_lower_threshold);
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static ssize_t in_illuminance_lower_threshold_store(struct device *dev,
> +						    struct device_attribute *attr,
> +						    const char *buf, size_t len)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct i2c_client *client = chip->client;
> +
> +	int value;
> +
> +	if (kstrtoint(buf, 0, &value) || !value)
> +		return -EINVAL;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	if (value >= ALS_MAX_VALUE ||
> +	    value >= chip->als_settings.als_upper_threshold)
> +		goto calibrate_error;
> +	else
> +		chip->als_settings.als_lower_threshold = value;
> +
> +	if (tsl2591_als_thresholds(chip))
> +		goto calibrate_error;
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return len;
> +
> +calibrate_error:
> +	dev_err(&client->dev, "Failed to calibrate sensor\n");
> +	mutex_unlock(&chip->als_mutex);
> +	return -EINVAL;
> +}
> +
> +static ssize_t in_illuminance_upper_threshold_show(struct device *dev,
> +						   struct device_attribute *attr,
> +						   char *buf)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	mutex_lock(&chip->als_mutex);
> +	ret = sprintf(buf, "%d\n", chip->als_settings.als_upper_threshold);
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static ssize_t in_illuminance_upper_threshold_store(struct device *dev,
> +						    struct device_attribute *attr,
> +						    const char *buf, size_t len)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct i2c_client *client = chip->client;
> +
> +	int value;
> +
> +	if (kstrtoint(buf, 0, &value) || !value)
> +		return -EINVAL;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	if (value > ALS_MAX_VALUE ||
> +	    value <= chip->als_settings.als_lower_threshold)
> +		goto calibrate_error;
> +	else
> +		chip->als_settings.als_upper_threshold = value;
> +
> +	if (tsl2591_als_thresholds(chip))
> +		goto calibrate_error;
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return len;
> +
> +calibrate_error:
> +	dev_err(&client->dev, "Failed to calibrate sensor\n");
> +	mutex_unlock(&chip->als_mutex);
> +	return -EINVAL;
> +}
> +
> +static ssize_t in_illuminance_persist_filter_show(struct device *dev,
> +						  struct device_attribute *attr,
> +						  char *buf)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	mutex_lock(&chip->als_mutex);
> +	ret = sprintf(buf, "%d\n", chip->als_settings.als_persist);
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static ssize_t in_illuminance_persist_filter_store(struct device *dev,
> +						   struct device_attribute *attr,
> +						   const char *buf, size_t len)
> +{
> +	struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct i2c_client *client = chip->client;
> +
> +	int value;
> +
> +	if (kstrtoint(buf, 0, &value) || !value)
> +		return -EINVAL;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	if (tsl2591_compatible_als_persist(chip, value))
> +		goto calibrate_error;
> +
> +	if (tsl2591_als_thresholds(chip))
> +		goto calibrate_error;
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return len;
> +
> +calibrate_error:
> +	dev_err(&client->dev, "Failed to calibrate sensor\n");
> +	mutex_unlock(&chip->als_mutex);
> +	return -EINVAL;
> +}
> +
> +static IIO_CONST_ATTR(in_illuminance_integration_time_available_ms,
> +				"100 200 300 400 500 600");
> +static IIO_CONST_ATTR(in_illuminance_gain_available,
> +				"low med high max");
> +static IIO_CONST_ATTR(in_illuminance_persist_filter_available,
> +				"0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15");
> +static IIO_DEVICE_ATTR_RW(in_illuminance_integration_time, 0);
> +static IIO_DEVICE_ATTR_RW(in_illuminance_gain, 0);
> +static IIO_DEVICE_ATTR_RW(in_illuminance_lower_threshold, 0);
> +static IIO_DEVICE_ATTR_RW(in_illuminance_upper_threshold, 0);
> +static IIO_DEVICE_ATTR_RW(in_illuminance_persist_filter, 0);
> +
> +static struct attribute *sysfs_attrs_ctrl[] = {
> +	&iio_const_attr_in_illuminance_integration_time_available_ms.dev_attr.attr,

These need to map to standard ABI.  Note that if you do need new ABI then
you also need to document it in Documentation/ABI/testing/sysfs-bus-iio*
but here, I'm not seeing anything unusual.

They all map to either info_mask elements or to parameters of the
events.


> +	&iio_const_attr_in_illuminance_gain_available.dev_attr.attr,
> +	&iio_const_attr_in_illuminance_persist_filter_available.dev_attr.attr,
> +	&iio_dev_attr_in_illuminance_integration_time.dev_attr.attr,
> +	&iio_dev_attr_in_illuminance_gain.dev_attr.attr,
> +	&iio_dev_attr_in_illuminance_lower_threshold.dev_attr.attr,
> +	&iio_dev_attr_in_illuminance_upper_threshold.dev_attr.attr,
> +	&iio_dev_attr_in_illuminance_persist_filter.dev_attr.attr,
> +	NULL
> +};
> +
> +static const struct attribute_group tsl2591_attribute_group = {
> +	.attrs = sysfs_attrs_ctrl,
> +};
> +
> +static const struct iio_event_spec tsl2591_events[] = {
> +	{
> +		.type = IIO_EV_TYPE_THRESH,
> +		.dir = IIO_EV_DIR_RISING,
> +		.mask_separate = BIT(IIO_EV_INFO_VALUE) |
> +				BIT(IIO_EV_INFO_ENABLE),
> +	}, {
> +		.type = IIO_EV_TYPE_THRESH,
> +		.dir = IIO_EV_DIR_FALLING,
> +		.mask_separate = BIT(IIO_EV_INFO_VALUE) |
> +				BIT(IIO_EV_INFO_ENABLE),
> +	},
> +};
> +
> +static const struct iio_chan_spec tsl2591_channels[] = {
> +	{
> +		.type = IIO_INTENSITY,
> +		.modified = 1,
> +		.channel2 = IIO_MOD_LIGHT_IR,
> +		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW),
> +	},
> +	{
> +		.type = IIO_INTENSITY,
> +		.modified = 1,
> +		.channel2 = IIO_MOD_LIGHT_BOTH,
> +		.event_spec = tsl2591_events,
> +		.num_event_specs = ARRAY_SIZE(tsl2591_events),
> +		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW),
> +	},
> +	{
> +		.type = IIO_INTENSITY,
> +		.info_mask_separate = BIT(IIO_CHAN_INFO_PROCESSED),

I'm not following what the intended interface is here for these last two
channels.  Please describe it.  The next channel is kind of standard
for thee devices (bunch of maths to get illuminance from 2 diodes
with different responses).  This channel however is less normal...

> +	},
> +	{
> +		.type = IIO_LIGHT,
> +		.info_mask_separate = BIT(IIO_CHAN_INFO_PROCESSED),
> +	},
> +};
> +
> +static int tsl2591_read_raw(struct iio_dev *indio_dev,
> +			    struct iio_chan_spec const *chan,
> +			    int *val, int *val2, long mask)
> +{
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	struct i2c_client *client = chip->client;
> +	int ret, pm_ret;
> +
> +	dev_dbg(&client->dev, "Reading from sensor");
> +
> +	ret = tsl2591_set_pm_runtime_busy(chip, true);
> +
For blocks where you do something then check for error it's more common
in kernel code to keep them together (i.e. no blank line).

e.g.
	ret = tsl2591_set_pm_runtime_busy(chip, true);
	if (ret < 0)
		return ret;

> +	if (ret < 0)
> +		return ret;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	ret = -EINVAL;
> +	switch (mask) {
> +	case IIO_CHAN_INFO_RAW:
> +		if (chan->type == IIO_INTENSITY) {
> +			ret = tsl2591_get_lux_data(indio_dev);

Interesting function name.  lux is the units of illuminance so if
you are calling lux_data I'd expect this to be a light channel.
Rename the function so it's not missleading.

> +			if (ret < 0)
> +				break;
> +
> +			if (chan->channel2 == IIO_MOD_LIGHT_BOTH)
> +				*val = chip->als_readings.als_ch0;
> +			else if (chan->channel2 == IIO_MOD_LIGHT_IR)
> +				*val = chip->als_readings.als_ch1;
> +			else
> +				*val = 0;

Why?  this should be an error case so I'd expect ret = -EINVAL;
No need to set *val to anything if you return an error.

> +
> +			ret = IIO_VAL_INT;
> +		}
> +		break;
> +	case IIO_CHAN_INFO_PROCESSED:
> +		if (chan->type == IIO_LIGHT) {
> +			ret = tsl2591_get_lux_data(indio_dev);
> +			if (ret < 0)
> +				break;
> +			*val = chip->als_readings.als_lux_int;
> +			*val2 = (chip->als_readings.als_lux_decimal * 1000);
> +			ret = IIO_VAL_INT_PLUS_MICRO;
> +		} else if (chan->type == IIO_INTENSITY) {
> +			ret = tsl2591_get_lux_data(indio_dev);
> +			if (ret < 0)
> +				break;
> +
> +			*val = chip->als_readings.als_visible;
> +			ret = IIO_VAL_INT;
> +		}
> +		break;
> +	}
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	pm_ret = tsl2591_set_pm_runtime_busy(chip, false);
> +	if (pm_ret < 0)
> +		return pm_ret;
> +
> +	return ret;
> +}
> +
> +static const struct iio_info tsl2591_info = {
> +	.attrs = &tsl2591_attribute_group,
> +	.read_raw = tsl2591_read_raw,
> +};
> +
> +static int __maybe_unused tsl2591_suspend(struct device *dev)
> +{
> +	struct iio_dev *indio_dev = i2c_get_clientdata(to_i2c_client(dev));
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	dev_dbg(dev, "PM Suspending\n");
> +	ret = tsl2591_set_power_state(chip, TSL2591_PWR_OFF);
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static int __maybe_unused tsl2591_resume(struct device *dev)
> +{
> +	struct iio_dev *indio_dev = i2c_get_clientdata(to_i2c_client(dev));

There was a move a while ago to clear out this bouncing back and forwards.
dev_get_drvdata(dev) is the same thing.

> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +	int ret;
> +
> +	mutex_lock(&chip->als_mutex);
> +
> +	dev_dbg(dev, "PM Resuming\n");
> +	ret = tsl2591_set_power_state(chip, TSL2591_PWR_ON |
> +				TSL2591_ENABLE_ALS |
> +				TSL2591_ENABLE_ALS_INT);
> +
> +	mutex_unlock(&chip->als_mutex);
> +
> +	return ret;
> +}
> +
> +static const struct dev_pm_ops tsl2591_pm_ops = {
> +	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
> +				pm_runtime_force_resume)
> +	SET_RUNTIME_PM_OPS(tsl2591_suspend, tsl2591_resume, NULL)
> +};
> +
> +static irqreturn_t tsl2591_irq_handler(int irq, void *private)
> +{
> +	struct iio_dev *dev_info = private;
> +	struct tsl2591_chip *chip = iio_priv(dev_info);
> +
> +	s64 timestamp = iio_get_time_ns(dev_info);

Do this inline as local variable doesn't add anything.

> +
> +	iio_push_event(dev_info,
> +		       IIO_UNMOD_EVENT_CODE(IIO_LIGHT, 0,
> +					    IIO_EV_TYPE_THRESH,
> +					IIO_EV_DIR_EITHER),
> +		       timestamp);
> +
> +	tsl2591_clear_als_irq(chip);
> +
> +	return IRQ_HANDLED;
> +}
> +
> +#ifdef CONFIG_OF

As mentioned below, this is definitely not a good idea as presence
of OF support doesn't mean there is a device tree available.

> +static int tsl2591_probe_of(struct tsl2591_chip *chip)
> +{
> +	struct device *dev = &chip->client->dev;
> +	struct device_node *np = dev->of_node;
Please use the generic firmware functions for this, not the of_ specific
ones.  

fwnode and device_property etc.

That enables various other types of firmware and costs nothing in
driver complexity etc.

> +	int ret;
> +
> +	u32 als_integration_time;
> +	u32 als_gain;
> +	u32 als_persist;
> +	u32 als_lower_threshold;
> +	u32 als_upper_threshold;

I've commented on the suitability of these above in reviewing the
dt binding.

> +
> +	if (!np)
> +		return -ENODEV;
> +
> +	dev_dbg(dev, "Probing device tree\n");
> +
> +	ret = of_property_read_u32(np, "als-integration-time",
> +				   &als_integration_time);

device_property_read_u32() etc

> +
> +	if (ret) {
> +		dev_warn(dev,
> +			 "Setting default als-integration-time: %d\n",
> +			 DEFAULT_ALS_INTEGRATION_TIME);

No need to warn, it's the documented default.
Do warn on invalid values though as you do below.

> +		chip->als_settings.als_int_time = DEFAULT_ALS_INTEGRATION_TIME;
> +	} else {
> +		if (tsl2591_compatible_int_time(chip, als_integration_time)) {
> +			dev_warn(dev, "Setting default als-integration-time\n");
> +			chip->als_settings.als_int_time =
> +				DEFAULT_ALS_INTEGRATION_TIME;
> +		}
> +		dev_info(dev, "als-integration-time = %#04x\n",
> +			 chip->als_settings.als_int_time);
> +	}
> +
> +	ret = of_property_read_u32(np, "als-gain", &als_gain);
> +
> +	if (ret) {
> +		dev_warn(dev, "Setting default als-gain: %#04x\n",
> +			 DEFAULT_ALS_GAIN);
> +		chip->als_settings.als_int_time = DEFAULT_ALS_GAIN;
> +	} else {
> +		if (tsl2591_compatible_gain(chip, als_gain)) {
> +			dev_warn(dev, "Setting default als-gain\n");
> +			chip->als_settings.als_gain = DEFAULT_ALS_GAIN;
> +		}
> +		dev_info(dev, "als-gain = %#04x\n",
> +			 chip->als_settings.als_gain);
> +	}
> +
> +	ret = of_property_read_u32(np, "als-persist", &als_persist);
> +
> +	if (ret) {
> +		dev_warn(dev,
> +			 "Setting default als-persist: %#04x\n",
> +			DEFAULT_ALS_PERSIST);
> +		chip->als_settings.als_persist = DEFAULT_ALS_PERSIST;
> +	} else {
> +		if (tsl2591_compatible_als_persist(chip, als_persist)) {
> +			dev_warn(dev, "Setting default als-persist\n");
> +			chip->als_settings.als_persist = DEFAULT_ALS_PERSIST;
> +		}
> +		dev_info(dev, "als-persist = %#04x\n",
> +			 chip->als_settings.als_persist);
> +	}
> +
> +	ret = of_property_read_u32(np, "als-lower-threshold",
> +				   &als_lower_threshold);
> +
> +	if (ret) {
> +		dev_warn(dev,
> +			 "Setting default als-lower-threshold: %d\n",
> +			DEFAULT_ALS_LOWER_THRESHOLD);
> +		chip->als_settings.als_lower_threshold =
> +			DEFAULT_ALS_LOWER_THRESHOLD;
> +	} else {
> +		if (als_lower_threshold >= ALS_MAX_VALUE) {
> +			dev_warn(dev, "Setting default als-lower-threshold\n");
> +			chip->als_settings.als_lower_threshold =
> +				DEFAULT_ALS_LOWER_THRESHOLD;
> +		} else {
> +			chip->als_settings.als_lower_threshold =
> +				als_lower_threshold;
> +		}
> +		dev_info(dev, "als-lower-threshold = %d\n",
> +			 chip->als_settings.als_lower_threshold);
> +	}
> +
> +	ret = of_property_read_u32(np, "als-upper-threshold",
> +				   &als_upper_threshold);
> +
> +	if (ret) {
> +		dev_warn(dev,
> +			 "Setting default als-upper-threshold: %d\n",
> +			DEFAULT_ALS_UPPER_THRESHOLD);
> +		chip->als_settings.als_upper_threshold =
> +			DEFAULT_ALS_UPPER_THRESHOLD;
> +	} else {
> +		if (als_upper_threshold > ALS_MAX_VALUE) {
> +			dev_warn(dev, "Setting default als-upper-threshold\n");
> +			chip->als_settings.als_upper_threshold =
> +				DEFAULT_ALS_UPPER_THRESHOLD;
> +		} else {
> +			chip->als_settings.als_upper_threshold =
> +				als_upper_threshold;
> +		}
> +		dev_info(dev, "als-upper-threshold = %d\n",
> +			 chip->als_settings.als_upper_threshold);
> +	}
> +
> +	return 0;
> +}
> +#else
> +static int tsl2591_default_config(struct tsl2591_chip *chip)
> +{
> +	struct device *dev = &chip->client->dev;
> +
> +	dev_dbg(dev, "Loading default configuration\n");
> +
> +	chip->als_settings.als_int_time = DEFAULT_ALS_INTEGRATION_TIME;
> +	chip->als_settings.als_int_time = DEFAULT_ALS_GAIN;
> +	chip->als_settings.als_persist = DEFAULT_ALS_PERSIST;
> +	chip->als_settings.als_lower_threshold = DEFAULT_ALS_LOWER_THRESHOLD;
> +	chip->als_settings.als_upper_threshold = DEFAULT_ALS_UPPER_THRESHOLD;
> +
> +	dev_dbg(dev, "als-integration-time = %d\n",
> +		chip->als_settings.als_int_time);
> +	dev_dbg(dev, "als-gain = %d\n", chip->als_settings.als_gain);
> +	dev_dbg(dev, "als-persist = %d\n", chip->als_settings.als_persist);
> +	dev_dbg(dev, "als-lower-threshold = %d\n",
> +		chip->als_settings.als_lower_threshold);
> +	dev_dbg(dev, "als-upper-threshold = %d\n",
> +		chip->als_settings.als_upper_threshold);
> +
> +	return 0;
> +}
> +#endif
> +
> +static int tsl2591_probe(struct i2c_client *client)
> +{
> +	int ret = 0;
I think this is set in all paths where it is used, so no need to initialize here.

> +	struct tsl2591_chip *chip;
> +	struct iio_dev *indio_dev;
> +
> +	dev_info(&client->dev, "Start probing device.\n");
Noise, please clean this sort of print out it tells us nothing useful and kernel logs
are busy enough as it is!

> +
> +	if (!i2c_check_functionality(client->adapter,
> +				     I2C_FUNC_SMBUS_BYTE_DATA)) {
> +		dev_err(&client->dev,
> +			"%s: i2c smbus byte data functionality is not supported\n", __func__);

Don't print __func__.  IIRC that can be done via dynamic debug tricks anyway if a user
wants it.

> +		return -EOPNOTSUPP;
> +	}
> +
> +	indio_dev = devm_iio_device_alloc(&client->dev, sizeof(*chip));
> +	if (!indio_dev)
> +		return -ENOMEM;
> +
> +	chip = iio_priv(indio_dev);
> +	chip->client = client;
> +	i2c_set_clientdata(client, indio_dev);
> +
> +#ifdef CONFIG_OF

Definitely don't want to see this.   Just because a kernel is built with
OF doesn't mean it's using it on a given platform.

Nothing wrong with putting in documented defaults, but do it when the
various read functions fail, not in a separate function.

> +	if (tsl2591_probe_of(chip)) {
> +		dev_err(&client->dev, "No platform data\n");
> +		return -ENODEV;
> +	}
> +#else
> +	if (tsl2591_default_config(chip)) {
> +		dev_err(&client->dev, "Failed to load default config\n");
> +		return -EINVAL;
> +	}
> +#endif
> +
> +	if (client->irq) {
> +		dev_dbg(&client->dev, "Registering interrupt\n");

I'd drop dev_dbg() that is just related to flow. It can be useful if values
are being returned etc, but there are lots of other ways to know things like
this are going to have happened so they are just noise and tend to bitrot.

> +		ret = devm_request_threaded_irq(&client->dev, client->irq,
> +						NULL, tsl2591_irq_handler,
> +				IRQF_TRIGGER_FALLING | IRQF_ONESHOT,
> +				"tsl2591_irq", indio_dev);
> +		if (ret) {
> +			dev_err(&client->dev, "irq request error %d\n", -ret);
> +			return -EINVAL;
> +		}
> +	}
> +
> +	mutex_init(&chip->als_mutex);
> +
> +	ret = i2c_smbus_read_byte_data(client,
> +				       TSL2591_CMD_NOP | TSL2591_DEVICE_ID);
> +
> +	if (ret < 0) {
> +		dev_err(&client->dev,
> +			"%s: failed to read the device ID register\n",
> +			__func__);
> +		return ret;
> +	}
> +
> +	if ((ret & TSL2591_DEVICE_ID_MASK) != TSL2591_DEVICE_ID_VAL) {

FIELD_GET can make this sort of code a little more readable.

> +		dev_err(&client->dev, "%s: received an unknown device ID %#04x\n",
> +			__func__, ret);
> +		return -EINVAL;
> +	}
> +
> +	indio_dev->info = &tsl2591_info;
> +	indio_dev->channels = tsl2591_channels;
> +	indio_dev->num_channels = ARRAY_SIZE(tsl2591_channels);
> +	indio_dev->modes = INDIO_DIRECT_MODE;
> +	indio_dev->name = chip->client->name;
> +
> +	pm_runtime_enable(&client->dev);
> +	pm_runtime_set_autosuspend_delay(&client->dev,
> +					 TSL2591_POWER_OFF_DELAY_MS);
> +	pm_runtime_use_autosuspend(&client->dev);
> +
> +	/*
> +	 * Power device on from probe to allow als cycles to
> +	 * generate interrupts when intensity is out of bounds.
> +	 * When interrupts are not enabled, device only needs to be
> +	 * powered on when reading
Don't have events from probe.  No one will be listening for them
unless they have requested them so it makes little sense.  Default
to disabled.

As you note, for this device you'll need to disable runtime pm once
they are enabled, but don't do it before then.

> +	 */
> +	if (client->irq) {
> +		tsl2591_resume(&client->dev);
> +		pm_runtime_set_active(&client->dev);
> +	}
> +
> +	ret = devm_iio_device_register(&client->dev, indio_dev);
> +
> +	if (ret) {
> +		dev_err(&client->dev, "%s: iio registration failed\n",
> +			__func__);
> +		return ret;
> +	}
> +

All the following 'after' you've exposed the interfaces to userspace?
Anything like this needs to happen before userspace interfaces and inevitable
race conditions are exposed.

> +	if (tsl2591_als_calibrate(chip)) {
> +		dev_err(&client->dev, "Failed to calibrate sensor\n");
> +		return -EINVAL;
> +	}
> +	if (tsl2591_clear_als_irq(chip)) {
> +		dev_err(&client->dev, "Failed to clear irq\n");
> +		return -EINVAL;
> +	}
> +	if (tsl2591_als_thresholds(chip)) {
> +		dev_err(&client->dev, "Failed to set als thresholds\n");
> +		return -EINVAL;
> +	}
> +
> +	dev_info(&client->dev, "Probe complete - tsl2591 found.\n");
> +
> +	return 0;
> +}
> +
> +static int tsl2591_remove(struct i2c_client *client)
> +{
> +	struct iio_dev *indio_dev = i2c_get_clientdata(client);
> +	struct tsl2591_chip *chip = iio_priv(indio_dev);
> +
> +	dev_dbg(&client->dev, "Removing device.\n");

This is useful during driver development but by the time of upstreaming
it's noise, even as a debug message.

> +
> +	pm_runtime_disable(&client->dev);
> +	pm_runtime_set_suspended(&client->dev);
> +	pm_runtime_put_noidle(&client->dev);
> +
> +	return tsl2591_set_power_state(chip, TSL2591_PWR_OFF);
This just turned the chip off before removing the userspace interfaces. Not a good
idea. 

Basic rule of thumb is that you should be able to read remove in reverse order
and see everything done in probe.  When managed interfaces (devm_ etc) then
they occur after the remove function is done.  That means you can't mix
and match.  The moment you hit something in probe that isn't managed you can't
use devm_* after that point.  It might be safe, but it makes review harder
so isn't worth it.

There is devm_add_action_or_reset() which lets you put additional functions
into the automatically managed path.  That can simplify things a lot if
you only have one or two unmanaged calls to deal with.

> +}
> +
> +static const struct of_device_id tsl2591_of_match[] = {
> +	{ .compatible = "amstaos,tsl2591", },
> +	{},
> +};
> +MODULE_DEVICE_TABLE(of, tsl2591_of_match);
> +
> +static struct i2c_driver tsl2591_driver = {
> +	.driver = {
> +		.name = "tsl2591",
> +		.pm = &tsl2591_pm_ops,
> +		.of_match_table = of_match_ptr(tsl2591_of_match),

Don't use of_match_ptr.  It stops the driver being used with
the magic of ACPI device type that maps to device tree
(odd but it exists :)  See PRP0001


> +	},
> +	.probe_new = tsl2591_probe,
> +	.remove = tsl2591_remove,
> +};
> +module_i2c_driver(tsl2591_driver);
> +
> +MODULE_AUTHOR("Joe Sandom <joe.sandom@xxxxxxxxxxx>");
> +MODULE_DESCRIPTION("TAOS tsl2591 ambient light sensor driver");
> +MODULE_LICENSE("GPL");





[Index of Archives]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Input]     [Linux Kernel]     [Linux SCSI]     [X.org]

  Powered by Linux