Message-ID: <4CAF2593.3020703@cam.ac.uk>
Date:	Fri, 08 Oct 2010 15:07:15 +0100
From:	Jonathan Cameron <jic23@....ac.uk>
To:	Samu Onkalo <samu.p.onkalo@...ia.com>
CC:	linux-i2c@...r.kernel.org, linux-kernel@...r.kernel.org,
	alan@...rguk.ukuu.org.uk
Subject: Re: [PATCHv2 5/5] Documentation: Short descriptions for bhsfh and
 apds990x drivers

On 10/08/10 14:42, Samu Onkalo wrote:
> Add short documentation for two ALS / proximity chip drivers.
> 
> bhsfh document update
Hi Samu,

Couple of clarity related questions below and some typos.

Otherwise pretty much fine.  I wonder if it would be useful to put
some of the platform data docs in the headers in kernel doc format
as well (or instead of here)?

I still personally prefer proximity rather than prox (and the isl29018
driver under IIO uses that and is in staging-next so will merge in the
next window).  Then again, I also prefer illuminance and I seem to have
lost that one for now ;)

Fix the typos and feel free to add,

Acked-by: Jonathan Cameron <jic23@....ac.uk>
> 
> Signed-off-by: Samu Onkalo <samu.p.onkalo@...ia.com>
> ---
>  Documentation/misc-devices/apds990x.txt |  152 +++++++++++++++++++++++++++++++
>  Documentation/misc-devices/bhsfh.txt    |  148 ++++++++++++++++++++++++++++++
>  2 files changed, 300 insertions(+), 0 deletions(-)
>  create mode 100644 Documentation/misc-devices/apds990x.txt
>  create mode 100644 Documentation/misc-devices/bhsfh.txt
> 
> diff --git a/Documentation/misc-devices/apds990x.txt b/Documentation/misc-devices/apds990x.txt
> new file mode 100644
> index 0000000..7b61495
> --- /dev/null
> +++ b/Documentation/misc-devices/apds990x.txt
> @@ -0,0 +1,152 @@
> +Kernel driver apds990x
> +======================
> +
> +Supported chips:
> +Avago APDS990X
> +
> +Data sheet:
> +Not freely available
> +
> +Author:
> +Samu Onkalo <samu.p.onkalo@...ia.com>
> +
> +Description
> +-----------
> +
> +APDS990x is a combined ambient light and proximity sensor. ALS and proximity
> +functionality are highly connected. ALS measurement path must be running
> +while the proximity functionality is enabled.
> +
> +ALS produces only raw measurement values. Further processing is needed
> +to get sensible LUX values. Vice versa, HW threshold control has nothing
> +to do with LUX values. Threshold detection triggs to raw conversion values.
Not quite clear what the above means.  Perhaps 
"Threshold detection is a comparison with the raw conversion value."?
> +Driver makes necessary conversions to both directions so that user handles
in both directions
> +only LUX values. ALS contains 4 different gain steps. Driver automatically
> +selects suitable gain step. After each measurement, reliability of the results
> +is estimated and new measurement is trigged if necessary.
> +
> +Platform data can provide tuned values to the conversion formulas if
> +values are known. Otherwise plain sensor default values are used.
> +
> +Proximity side is little bit simpler. There is no need for complex conversions.
> +It produces directly usable values.
> +
> +Driver controls chip operational state using pm_runtime framework.
> +Voltage regulators are controlled based on chip operational state.
> +
> +SYSFS
> +-----
> +
> +
> +chip_id
> +	RO - shows detected chip type and version
> +
> +power_state
> +	RW - enable / disable chip. Uses counting logic
> +	     1 enables the chip
> +	     0 disables the chip
> +lux0_input
> +	RO - measured LUX value
> +	     sysfs_notify called when threshold interrupt occurs
> +
> +lux0_sensor_range
> +	RO - lux0_input max value. Actually never reaches since sensor tends
> +	     to saturate much before that. Real max value varies depending
> +	     on the light spectrum etc.
(aside: That's why I'm anti range and favour 'scale' as the means of controlling
 this - fine here though and a nice explanation)
> +
> +
> +lux0_rate
> +	RW - measurement rate in Hz
> +
> +lux0_rate_avail
> +	RO - supported measurement rates
> +
> +lux0_calibscale
> +	RW - calibration value. Set to neutral value by default.
> +	     Output results are multiplied with calibscale / calibscale_default
> +	     value.
> +
> +lux0_calibscale_default
> +	RO - neutral calibration value
> +
> +lux0_thresh_above_value
> +	RW - HI level threshold value. All results above the value
> +	     trigs an interrupt. 65535 (i.e. sensor_range) disables the above
> +	     interrupt.
> +
> +lux0_thresh_below_value
> +	RW - LO level threshold value. All results below the value
> +	     trigs an interrupt. 0 disables the below interrupt.
> +
> +prox0_raw
> +	RO - measured proximity value
> +	     sysfs_notify called when threshold interrupt occurs
> +
> +prox0_sensor_range
> +	RO - prox0_raw max value (1023)
> +
> +prox0_raw_en
> +	RW - enable / disable proximity - uses counting logic
> +	     1 enables the proximity
> +	     0 disables the proximity
> +
> +prox0_reporting_mode
> +	RW - trigger / periodic. In "trigger" mode the driver tells two possible
> +	     values: 0 or prox0_sensor_range value. 0 means no proximity,
> +	     1023 means proximity. This causes minimal number of interrupts.
> +	     In "periodic" mode the driver reports all values above
> +	     prox0_thresh_above. This causes more interrupts, but it can give
> +	     _rough_ estimate about the distance.
> +
> +prox0_reporting_mode_avail
> +	RO - accepted values to prox0_reporting_mode (trigger, periodic)
> +
> +prox0_thresh_above_value
> +	RW - threshold level which trigs proximity events.
> +
> +Platform data
> +-------------
> +
> +
> +#define APDS_IRLED_CURR_12mA	0x3
> +#define APDS_IRLED_CURR_25mA	0x2
> +#define APDS_IRLED_CURR_50mA	0x1
> +#define APDS_IRLED_CURR_100mA	0x0
> +
> +/*
> + * Structure for tuning ALS calculation to match with environment.
> + * There depends on the material above the sensor and the sensor
> + * itself. If the GA is zero, driver will use uncovered sensor default values
> + * format: decimal value * APDS_PARAM_SCALE
> + */
> +#define APDS_PARAM_SCALE 4096
> +struct apds990x_chip_factors {
> +	int ga;		/* Glass attenuation */
> +	int cf1;	/* Clear channel factor 1 */
> +	int irf1;	/* Ir channel factor 1 */
> +	int cf2;	/* Clear channel factor 2 */
> +	int irf2;	/* Ir channel factor 2 */
> +	int df;		/* Device factor. Decimal number */
> +};
> +
> +struct apds990x_platform_data {
> +	struct apds990x_chip_factors cf;
> +	u8     pdrive;
> +	u8     ppcount;
> +	int    (*setup_resources)(void);
> +	int    (*release_resources)(void);
> +};
> +
> +chip factors are specific to for example dark plastic window
> +above the sensor. Driver uses plain (uncovered) sensor default values
> +if the platform data contains zero ga factor.
> +
> +"pdrive" is the IR led driver current
> +
> +"ppcount" is the number of the IR pulses used for proximity detection.
> +5 is for example quite good value.
> +
> +Proximity detection range and overall behaviour is heavily affected
> +by cover window, IR led driving strength and ppcount.

As an aside.  A lot of this stuff could also be usefully added to the
header in kernel doc format. That is typically where people go looking
for details on what platform data is needed rather than a documentation
file.
> +
> +setup / release resources handles interrupt line configurations.
> diff --git a/Documentation/misc-devices/bhsfh.txt b/Documentation/misc-devices/bhsfh.txt
> new file mode 100644
> index 0000000..84568fc
> --- /dev/null
> +++ b/Documentation/misc-devices/bhsfh.txt
> @@ -0,0 +1,148 @@
> +Kernel driver bhsfh
> +===================
> +
> +Supported chips:
> +ROHM BH1770GLC
> +OSRAM SFH7770
> +
> +Data sheet:
> +Not freely available
> +
> +Author:
> +Samu Onkalo <samu.p.onkalo@...ia.com>
> +
> +Description
> +-----------
> +BH1770GLC and SFH7770 (I'm calling them as bhsfh) are combined ambient
> +light and proximity sensors. ALS and proximity parts operates on their own,
> +but they shares common I2C interface and interrupt logic.
Out of curiosity, can they exist separately? E.g. would this driver work
(with minor changes) for a part that only contains one of the pair? 
> +
> +ALS produces 16 bit LUX values. The chip contains interrupt logic to produce
> +low and high threshold interrupts.
> +
> +Proximity part contains IR-led driver up to 3 IR leds. The chip measures
> +amount of reflected IR light and produces proximity result. Resolution is
> +8 bit. Driver supports only one channel. Driver uses ALS results to estimate
> +realibility of the proximity results. Thus ALS is always running while
typo realibility -> reliability?
> +proximity detection is needed.
> +
> +Driver uses threshold interrupts to avoid need for polling the values.
> +Proximity low interrupt doesn't exists in the chip. This is simulated
> +by using a delayed work.
I'm not clear why this is in driver?  Is it just to maintain the same interface
as for the other part?
> +
> +Chip state is controlled via runtime pm framework when enabled in config.
> +
> +Calibscale factor is used to hide differences between the chips. By default
> +value set to neutral state meaning factor of 1.00. To get proper values,
> +calibrated source of light is needed as a reference. Calibscale factor is set
> +so that measurement produces about the expected lux value.
> +
> +SYSFS
> +-----
> +
> +chip_id
> +	RO - shows detected chip type and version
> +
> +power_state
> +	RW - enable / disable chip. Uses counting logic
> +	     1 enables the chip
> +	     0 disables the chip
> +
> +lux0_input
> +	RO - measured LUX value
> +	     sysfs_notify called when threshold interrupt occurs
> +
> +lux0_sensor_range
> +	RO - lux0_input max value
> +
> +lux0_rate
> +	RW - measurement rate in Hz
> +
> +lux0_rate_avail
> +	RO - supported measurement rates
> +
> +lux0_thresh_above_value
> +	RW - HI level threshold value. All results above the value
> +	     trigs an interrupt. 65535 (i.e. sensor_range) disables the above
> +	     interrupt.
> +
> +lux0_thresh_below_value
> +	RW - LO level threshold value. All results below the value
> +	     trigs an interrupt. 0 disables the below interrupt.
> +
> +lux0_calibscale
> +	RW - calibration value. Set to neutral value by default.
> +	     Output results are multiplied with calibscale / calibscale_default
> +	     value.
> +
> +lux0_calibscale_default
> +	RO - neutral calibration value
> +
> +prox0_raw
> +	RO - measured proximity value
> +	     sysfs_notify called when threshold interrupt occurs
> +
> +prox0_sensor_range
> +	RO - prox0_raw max value
> +
> +prox0_raw_en
> +	RW - enable / disable proximity - uses counting logic
> +	     1 enables the proximity
> +	     0 disables the proximity
> +
> +prox0_thresh_above_count
> +	RW - number of proximity interrupts needed before triggering the event
> +
> +prox0_rate_above
> +	RW - Measurement rate (in Hz) when the level is above threshold
> +	     i.e. when proximity on has been reported.
> +
> +prox0_rate_below
> +	RW - Measurement rate (in Hz) when the level is below threshold
> +	     i.e. when proximity off has been reported.
> +
> +prox0_rate_avail
> +	RO - Supported proxity measurement rates in Hz
proxity -> proximity 
> +
> +
> +prox0_thresh_above0_value
> +	RW - threshold level which trigs proximity events.
> +	     Filtered by persistence filter (prox0_thresh_above_count)
> +
> +prox0_thresh_above1_value
> +	RW - threshold level which trigs event immediately
> +
> +
> +Platform data:
> +--------------
> +
> +struct bhsfh_platform_data {
> +#define BHSFH_LED_5mA	0
> +#define BHSFH_LED_10mA	1
> +#define BHSFH_LED_20mA	2
> +#define BHSFH_LED_50mA	3
> +#define BHSFH_LED_100mA 4
> +#define BHSFH_LED_150mA 5
> +#define BHSFH_LED_200mA 6
> +	__u8 led_def_curr;
> +#define BHFSH_NEUTRAL_GA 16384 /* 16384 / 16384 = 1 */
> +	__u32 glass_attenuation;
> +	int (*setup_resources)(void);
> +	int (*release_resources)(void);
> +};
> +
> +led_def_curr controls IR led driving current
> +glass_attenuation tells darkness of the covering window
> +setup / release resources are call back functions for controlling
> +interrupt line.
> +
> +Example:
> +static struct bhsfh_platform_data rm680_bhsfh_data = {
> +	.led_def_curr	   = BHSFH_LED_50mA,
> +	.glass_attenuation = (16384 * 385) / 100,
> +	.setup_resources   = bhsfh_setup,
> +	.release_resources = bhsfh_release,
> +};
> +
> +glass_attenuation: 385 in above formula means 3.85 in decimal.
> +light_above_sensors = light_above_cover_window / 3.85

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists