Add support for touchscreen on Colibri VF50

Hello,

The patchset adds support for 4 wire touchscreen on Toradex Colibri VF50 modules.

Thanks Dmitry and Stefan for your feedback.

Changes since v4:
1. Add missing header file for gpio consumer.
2. Purge unneeded gpio descriptor entry in struct
3. Do not cast from void * pointer in IRQ bh
4. Make sure to have the gpio ym inactive and pinmux in ADC state on close
5. Return error codes directly instead of assigning and returning
6. Fix leaking of IIO channels and install devm action earlier
7. Make sure that correct number of ADC channels are specified
8. Improve DT documentation
9. Also include DT pressure property in our evaluation board dts file

Changes since v3:
1. Add a #define for the average readings measurement.
2. Instead of configuring gpios every open, configure them on request.
3. Use a memory barrier and synchronize irq call to make sure IRQ is not running past close.
4. Drop inline for vf50_ts_get_gpiod
5. Convert "ret" to "error"
6. Use devm_add_action to add a function for releasing iio channels and drop remove completely.
7. Kill __set_bits()
8. Instead of requesting for pen irq as gpio and then irq, use the interrupts property for DT and let it be handled automatically. Anyways we never read it's state.
9. Just use IRQF_ONESHOT here, rely on the platform/OF to read and set up the trigger properly (via "interrupts" property).

Changes since v2:
1. Fix pin multiplexing for pins in idle state. Configuration of the pen detect pull up viz. PTA19__GPIO_9 resulted in generation of pen irq's on a continuous basis.
2. Fix pinmux of the ADC pins as per the recommended pinmux in TRM.
3. Use a threaded irq handler instead of a irq handler plus workqueue approach.
4. Use a low level trigger with oneshot flag specifier instead of the previous falling edge triggered irq's. This coupled with the fix in point 1 fixes the previous continuous spurious irq generation bug.
5. Change/fix the TS measurement logic to account for the fact that iio_channel_read_raw might actually return an error. To be more specific use break instead of continue and take care to close the FET's in case of channel read error.
6. Drop the first patch "Add io-channel-cells property for ADC node" as it has already been applied.
7. Move the iio channel get call again at the start. Having it in the end resulted in crashes sometimes when iio was not probed and the ts device got probed and opened earlier.

Changes since v1:
1. Fix/drop comments
2. Use an inline function for multiple gpiod_get calls in probe
3. Remove the pull up in the pinmux specified in DT for touchctrl_gpios
4. Add the io-channel-cells property before status property.
5. Add GPIOLIB as dependency in the Kconfig file

Version 4 of the patchset can be found here
https://lkml.org/lkml/2015/8/21/339

Version 3 of the patchset can be found here
https://lkml.org/lkml/2015/8/5/173

Version 2 of the patchset can be found here
https://www.mail-archive.com/linux-input at vger.kernel.org/msg18090.html

Version 1 of the patchset can be found here
https://lkml.org/lkml/2015/6/30/103

Regards,
Sanchayan.

Sanchayan Maity (3):
ARM: dts: vf500-colibri: Add device tree node for touchscreen support
input: Add DT binding documentation for Colibri VF50 touchscreen
touchscreen: colibri-vf50-ts: Add touchscreen support for Colibri VF50

 .../bindings/input/touchscreen/colibri-vf50-ts.txt |  36 ++
 arch/arm/boot/dts/vf500-colibri-eval-v3.dts        |   5 +
 arch/arm/boot/dts/vf500-colibri.dtsi               |  47 +++
 drivers/input/touchscreen/Kconfig                  |  12 +
 drivers/input/touchscreen/Makefile                 |   1 +
 drivers/input/touchscreen/colibri-vf50-ts.c        | 381 +++++++++++++++++++++
 6 files changed, 482 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/input/touchscreen/colibri-vf50-ts.txt
 create mode 100644 drivers/input/touchscreen/colibri-vf50-ts.c

2.5.0