Re: [PATCHv5][ 1/4] Input: tsc2007: Add device tree support.

[Thierry Reding <thierry.reding@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 3514 bytes --]

On Thu, Oct 24, 2013 at 08:51:50AM +0200, Lothar Waßmann wrote:
> Hi,
> 
> Rob Herring wrote:
> > On 10/23/2013 07:10 AM, Denis Carikli wrote:
> [...]
> > > diff --git
> a/Documentation/devicetree/bindings/input/touchscreen/tsc2007.txt
> b/Documentation/devicetree/bindings/input/touchscreen/tsc2007.txt
> > > new file mode 100644
> > > index 0000000..fadd3f6
> > > --- /dev/null
> > > +++ b/Documentation/devicetree/bindings/input/touchscreen/tsc2007.txt
> > > @@ -0,0 +1,44 @@
> > > +* Texas Instruments tsc2007 touchscreen controller
> > > +
> > > +Required properties:
> > > +- compatible: must be "ti,tsc2007".
> > > +- reg: I2C address of the chip.
> > > +- pinctrl-0: Should specify pin control groups used for this controller
> > > +  (see pinctrl bindings[0]).
> > > +- pinctrl-names: Should contain only one value - "default"
> > > +  (see pinctrl bindings[0]).
> > 
> > I'm confused why an i2c slave needs pinctl binding?
> > 
> for the pendetect GPIO.

Shouldn't that be done transparently to users of the GPIO API? I was
under the impression that gpio_request() would set everything up (or
return an error if unable to do so) so that the GPIO can be used,
including any required pinmuxing.

> [...]
> > > diff --git a/drivers/input/touchscreen/tsc2007.c b/drivers/input/touchscreen/tsc2007.c
> > > index 0b67ba4..0625fe1 100644
> > > --- a/drivers/input/touchscreen/tsc2007.c
> > > +++ b/drivers/input/touchscreen/tsc2007.c
> > > @@ -26,6 +26,9 @@
> > >  #include <linux/interrupt.h>
> > >  #include <linux/i2c.h>
> > >  #include <linux/i2c/tsc2007.h>
> > > +#include <linux/of_device.h>
> > > +#include <linux/of.h>
> > > +#include <linux/of_gpio.h>
> > >  
> > >  #define TSC2007_MEASURE_TEMP0		(0x0 << 4)
> > >  #define TSC2007_MEASURE_AUX		(0x2 << 4)
> > > @@ -74,7 +77,10 @@ struct tsc2007 {
> > >  	u16			max_rt;
> > >  	unsigned long		poll_delay;
> > >  	unsigned long		poll_period;
> > > +	int			fuzzy;
> > > +	char			of;
> > >  
> > > +	unsigned		gpio;
> > >  	int			irq;
> > >  
> > >  	wait_queue_head_t	wait;
> [...]
> > > @@ -273,34 +295,65 @@ static void tsc2007_close(struct input_dev *input_dev)
> > >  	tsc2007_stop(ts);
> > >  }
> > >  
> > > -static int tsc2007_probe(struct i2c_client *client,
> > > -				   const struct i2c_device_id *id)
> > > +#ifdef CONFIG_OF
> > > +static int tsc2007_probe_dt(struct i2c_client *client, struct tsc2007 *ts,
> > > +			    struct device_node *np)
> > >  {
> > > -	struct tsc2007 *ts;
> > > -	struct tsc2007_platform_data *pdata = client->dev.platform_data;
> > > -	struct input_dev *input_dev;
> > > -	int err;
> > > -
> > > -	if (!pdata) {
> > > -		dev_err(&client->dev, "platform data is required!\n");
> > > +	int err = 0;
> > > +	u32 val32;
> > > +	u64 val64;
> > > +
> > > +	if (!of_property_read_u32(np, "max-rt", &val32))
> > > +		ts->max_rt = val32;
> > > +	else
> > > +		ts->max_rt = MAX_12BIT;
> > 
> > These functions don't overwrite the value if the property isn't present.
> > So you can set the values to the defaults and just pass the variable
> > (i.e. ts->max_rt) to of_property_read_u32 directly.
> > 
> Not quite. Since max_rt is an u16 you can't pass it to
> of_property_read_u32(). And using of_property_read_u16() requires the
> abominable DT notation: "/bits/ 16 <value>;"

In that case perhaps you need to check that whatever u32 value you read
from DT is actually in the expected range before assigning to u16?

Thierry

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]