The abs-fuzz and abs-flat properties are documented as optional.  When
these are absent, fwnode_property_read_u32() will leave the input
unchanged, meaning that an axis either picks up the value for the
previous axis or an uninitialized value.

Explicitly set these values to zero when they are unspecified to match
the documented behaviour in the device tree bindings.
Signed-off-by: John Keeping <jkeeping@inmusicbrands.com>
Link: https://lore.kernel.org/r/20240731093310.3696919-1-jkeeping@inmusicbrands.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The kernel reports that the touchpad for this device can support a
different bus.

With SMBus enabled the touchpad movement is smoother and three-finger
gestures are recognized.
Signed-off-by: Jonathan Denose <jdenose@google.com>
Link: https://lore.kernel.org/r/20240719180612.1.Ib652dd808c274076f32cd7fc6c1160d2cf71753b@changeidSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

In case the touch controller is not connected, this message keeps scrolling
on the console indefinitelly. Ratelimit it to avoid filling kernel logs.

"
ads7846 spi2.1: spi_sync --> -22
"
Signed-off-by: Marek Vasut <marex@denx.de>
Link: https://lore.kernel.org/r/20240708211913.171243-1-marex@denx.deSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Prepare input updates for 6.11 merge window.

The locking rules in the driver came from era when sysfs attributes
could live past the point of time when device would be unbound from
the driver, and so used module-global semaphore (potentially shared
between multiple yealink devices). Thankfully these times are long
gone and attributes will not be accessible once they are removed.

Simplify the logic by moving to per-device mutex, stop checking if
there is driver data instance attached to the interface, and use
guard notation to acquire the mutex.
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/20240710234855.311366-2-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Instead of manually creating driver-specific device attributes
set struct usb_driver->dev_groups pointer to have the driver core
do it.
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/20240710234855.311366-1-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Instead of manually creating driver-specific device attributes
set struct usb_driver->dev_groups pointer to have the driver core
do it.
Reviewed-by: Ville Syrjälä <syrjala@sci.fi>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/Zo8gaF_lKPAfcye1@google.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Instead of manually creating driver-specific device attributes
set struct driver->dev_groups pointer to have the driver core
do it.

This also fixes issue with the attribute not being deleted on driver
unbind.
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/Zo9nofWJ1xg9MgKs@google.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Instead of manually creating driver-specific device attributes,
set struct driver->dev_groups pointer to have the driver core
do it.
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/Zo9kSFeGOZB9b3rq@google.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Add support for EXC81W32 controllers.

Tested with firmware reported as type "PCAP81X32 Series",
model "Orion_0183_1019", fw_version "8001280G".
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Link: https://lore.kernel.org/r/20240710-input-exc3000-exc81w32-v3-2-4272183628b4@pengutronix.deSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Add compatible for EXC81W32 touchscreen controllers.
They use the same protocol and have the same resolution as EXC80H84.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20240710-input-exc3000-exc81w32-v3-1-4272183628b4@pengutronix.deSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Do not use kernel-doc style for comment describing contents of the
source file, as it trips the script:

  scripts/kernel-doc -none   drivers/input/misc/twl4030-pwrbutton.c
drivers/input/misc/twl4030-pwrbutton.c:2: info: Scanning doc for function twl4030
drivers/input/misc/twl4030-pwrbutton.c:33: warning: expecting prototype for twl4030(). Prototype was for PWR_PWRON_IRQ() instead
1 warnings

Also remove file name from the same comment - it it not the best idea
to have it as they tend to get stale when sources get moved or renamed.
Reported-by: Mirsad Todorovac <mtodorovac69@gmail.com>
Tested-by: Mirsad Todorovac <mtodorovac69@gmail.com>
Link: https://lore.kernel.org/r/Zo3QE00GqCrA3M9b@google.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The HX83100A is a bit of an outlier in the Himax HX831xxx series of
touch controllers as it requires reading touch events through the AHB
interface of the MCU rather than providing a dedicated FIFO address like
the other chips do.
This patch implements the specific read function and introduces the
HX83100A chip with an appropriate i2c ID and DT compatible string.

The HX83100A doesn't have a straightforward way to do chip
identification, which is why it is not implemented in this patch.

Tested on: Lenovo ThinkSmart View (CD-18781Y) / Innolux P080DDD-AB2 LCM
Signed-off-by: Felix Kaechele <felix@kaechele.ca>
Tested-by: Paul Gale <paul@siliconpixel.com>
Link: https://lore.kernel.org/r/20240620145019.156187-6-felix@kaechele.caSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

In preparation for HX83100A support allow defining separate functions
for specific chip operations.
Signed-off-by: Felix Kaechele <felix@kaechele.ca>
Link: https://lore.kernel.org/r/20240620145019.156187-5-felix@kaechele.caSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Implement reading from the MCU in a more universal fashion. This allows
properly handling reads of more than 4 bytes using the AHB FIFO
implemented in the chip.
Signed-off-by: Felix Kaechele <felix@kaechele.ca>
Link: https://lore.kernel.org/r/20240620145019.156187-4-felix@kaechele.caSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Himax uses an AHB-style bus to communicate with different parts of the
display driver and touch controller system.
Use more descriptive names for the register and address defines.
The names were taken from a driver submission for the similar HX83102J
chip.
Signed-off-by: Felix Kaechele <felix@kaechele.ca>
Link: https://lore.kernel.org/all/TY0PR06MB561105A3386E9D76F429110D9E0F2@TY0PR06MB5611.apcprd06.prod.outlook.com/
Link: https://lore.kernel.org/r/20240620145019.156187-3-felix@kaechele.caSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Add a compatible string for the Himax HX83100A touch controller.

The HX83100A presents touch events on the internal bus rather than
offering a dedicated event register like the other chips in this family
do.
Signed-off-by: Felix Kaechele <felix@kaechele.ca>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20240620145019.156187-2-felix@kaechele.caSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Now that the input_dev->vals array is always there we can be assured
that input_pass_values() is always called with a non-0 number of
events. Remove the check.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-8-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Preallocate memory for holding event values (input_dev->vals) so that
there is no need to check if it was allocated or not in the event
processing code.

The amount of memory will be adjusted after input device has been fully
set up upon registering device with the input core.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-7-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

In preparation to have dev->vals memory pre-allocated rearrange
code in input_alloc_device() so that it allows handling multiple
points of failure.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-6-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Streamline event handling code by providing batch implementations for
filtering and event processing and using them in place of the main
event handler, as needed, instead of having complex branching logic
in the middle of the event processing code.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-5-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

In preparation to consolidating filtering and event processing in the
input core change events() method to return number of events processed
by it.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-4-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Input core expects input handlers to be either filters, or regular
handlers, but not both. Additionally, for regular handlers it does
not make sense to define both single event method and batch method.

Refuse registering handler if it defines more than one method.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-3-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Input core favors ->events() (batch) method over ->event() method
if the former is defined, so there is no point in defining evdev_event()
as it is never called. Remove it.
Reviewed-by: Jeff LaBundy <jeff@labundy.com>
Reviewed-by: Benjamin Tissoires <bentiss@kernel.org>
Link: https://lore.kernel.org/r/20240703213756.3375978-2-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Imagis IST3038 is another variant of Imagis IST3038 IC, which has
a different register interface from IST3038C (possibly firmware defined).

Unlike IST3038C/IST3032C, IST3038 has different registers for commands,
which means IST3038 doesn't use protocol B.
Similar to IST3032C and maybe the other variants, IST3038 has touch keys
support, which provides KEY_APPSELECT and KEY_BACK.

Add support for IST3038 with touch keys.
Signed-off-by: Raymond Hackley <raymondhackley@protonmail.com>
Link: https://lore.kernel.org/r/20240613025631.5425-4-raymondhackley@protonmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Imagis IST3038 is a variant of Imagis touchscreen IC. Document it in
imagis,ist3038c bindings.
Signed-off-by: Raymond Hackley <raymondhackley@protonmail.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20240613025631.5425-3-raymondhackley@protonmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

protocol_b is a property, which tells Imagis panel to use a different
format for coordinates.

IST30XXC series is known for using protocol B, while the other series
aren't. Note this could be confusing, unlike the model name implies.

Adjust the usage of protocol_b to avoid confusion.
Signed-off-by: Raymond Hackley <raymondhackley@protonmail.com>
Link: https://lore.kernel.org/r/20240613025631.5425-2-raymondhackley@protonmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Convert TI-NSPIRE Keypad controller bindings to DT schema.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20240612150711.26706-1-afd@ti.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

`mx25_tcq_regconfig` is not modified and can be declared as const to
move its data to a read-only section.
Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
Link: https://lore.kernel.org/r/20240705-input-const-regmap_config-v1-2-f712a4494883@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

`qt1050_regmap_config` is not modified and can be declared as const to
move its data to a read-only section.
Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
Link: https://lore.kernel.org/r/20240705-input-const-regmap_config-v1-1-f712a4494883@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

When dealing with byte data use u8 instead of unsigned char or int.
Stop layering error handling in cypress_ps2_sendbyte() and simply
pass on error code from ps2_sendbyte().

Additionally use u8 instead of unisgned char throughout the code.

Link: https://lore.kernel.org/r/20240628224728.2180126-5-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Do not override errors reported by lower layers with generic "-1",
but propagate them to the callers. Change the checks for errors to be
in the form of "if (error)" to maintain consistency.

Link: https://lore.kernel.org/r/20240628224728.2180126-4-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Report -ETIMEDOUT error code from cypress_ps2_read_cmd_status() when
device does not send enough data within the allotted time in response
to a command.

Link: https://lore.kernel.org/r/20240628224728.2180126-3-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Stop layering error handling in cypress_ps2_sendbyte() and simply
pass on error code from ps2_sendbyte() and use it in the callers.

This fixes mishandling of error condition in
cypress_ps2_read_cmd_status() which expects errors to be negative.
Reported-by: Igor Artemiev <Igor.A.Artemiev@mcst.ru>
Link: https://lore.kernel.org/r/20240628224728.2180126-2-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Casting an integer field containing desired rate to a pointer to bytes
works on little endian architectures where the driver is used, but not
a good practice. Use a temporary of proper type instead.

Link: https://lore.kernel.org/r/20240628224728.2180126-1-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Make sure interrupts are not left disabled when we fail to suspend the
touch controller.

Fixes: 6696777c ("Input: add driver for Elan I2C/SMbus touchpad")
Link: https://lore.kernel.org/r/ZmKiiL-1wzKrhqBj@google.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

This makes the code more compact and error handling more robust.

Link: https://lore.kernel.org/r/20240609234757.610273-3-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Start using __free() attributes to simplify the code and error handling.

Link: https://lore.kernel.org/r/20240609234757.610273-2-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Allocating a contiguous buffer of 64K may fail if memory is sufficiently
fragmented, and may cause OOM kill of an unrelated process. However we
do not need to have contiguous memory. We also do not need to zero
out the buffer since it will be overwritten with firmware data.

Switch to using kvmalloc() instead of kzalloc().

Link: https://lore.kernel.org/r/20240609234757.610273-1-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Start using __free() and guard() primitives to simplify the code
and error handling.

Link: https://lore.kernel.org/r/20240609235134.614592-3-dmitry.torokhov@gmail.comSigned-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>