The order of endpoints is well defined on official Xbox pads, but
we have found at least one 3rd-party pad that doesn't follow the
standard ("Titanfall 2 Xbox One controller" 0e6f:0165).

Fortunately, we get lucky with this specific pad because it uses
endpoint addresses that differ only by direction. We know that
there are other pads out where this is not true, so let's go
ahead and fix this.

Signed-off-by: Cameron Gutman <aicommander@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Replace first goto with simple returns as we really are just returning
one error code.

Signed-off-by: Pavel Rojtberg <rojtberg@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

xbox one was the only device that has a *_process_buttons routine.

Signed-off-by: Pavel Rojtberg <rojtberg@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Signed-off-by: Daniel Tobias <dan.g.tob@gmail.com>
Signed-off-by: Pavel Rojtberg <rojtberg@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

msleep (1~20) may not do what the caller intends, and will often sleep
longer.  (~20 ms actual sleep for any value given in the 1~20ms range) This
is not the desired behaviour for many cases like device resume time, device
suspend time, device enable time, etc.  Thus, change msleep to usleep_range
for precise wakeups.

Signed-off-by: Aniroop Mathur <a.mathur@samsung.com>
Acked by: Albert Zhang <xu.zhang@bosch-sensortec.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

We rely on the result of scanning PNP bus for keyboard and mouse devices to
decide whether we should continue initialization of i8042 driver or not, so
the probes can not be asynchronous.

Also, it is not a good idea to unbind either of the PNP devices while i8042
is running, so let's disable bind/unbind for the PNP drivers through sysfs
and rely on i8042 cleaning after itself properly. User can still unbind and
rebind entire i8042 and that will take care of PNP drivers as well.

Reviewed-by: Luis R. Rodriguez <mgrof@kernel.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Merge updated Synaptics RMI4 support, including support for SMBus
controllers and flashing firmware.

Make the avarage-samples property a general touchscreen property
rather than imx6ul device specific.

Signed-off-by: Guy Shapiro <guy.shapiro@mobi-wize.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Update driver drv260x to use generic device properties so that it can be
used on non-DT systems. We also remove platform data as generic device
properties work on static board code as well.

Signed-off-by: Jingkui Wang <jkwang@google.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Let's introduce a temporary for "client->dev" is probe() as we use
it quite a few times and "dev" is shorter.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

We were assigning I2C bus controller instead of client as parent device.
Besides being logically wrong, it messed up with devm handling of input
device. As a result we were leaving input device and event node behind
after rmmod-ing the driver, which lead to a kernel oops if one were to
access the event node later.

Let's remove the assignment and rely on devm_input_allocate_device() to
set it up properly for us.

Signed-off-by: Jingkui Wang <jkwang@google.com>
Fixes: 7132fe4f ("Input: drv260x - add TI drv260x haptics driver")
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Port firmware update code from Samsung Galaxy S7 driver into
mainline framework.

This patch has been tested on Synaptics S7813.

Signed-off-by: Nick Dyer <nick@shmanahar.org>
Tested-by: Chris Healy <cphealy@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

We were accidentally initializing haptics->rated_voltage twice, and did not
initialize overdrive voltage.

Acked-by: Dan Murphy <dmurphy@ti.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Reviewed-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

As this define check if huge, this makes easier to read the code.

Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Semantics of NR_IRQS is different on machines with SPARSE_IRQ option
disabled or enabled, in the latter case IRQs are allocated starting
at least from the value specified by NR_IRQS and going upwards, so
the check of (irq >= NR_IRQ) to decide about an error code returned by
platform_get_irq() is completely invalid, don't attempt to overrule
irq subsystem in the driver.

The change fixes lpc32xx_keys driver initialization on boot:

  lpc32xx_keys 40050000.key: failed to get platform irq
  lpc32xx_keys: probe of 40050000.key failed with error -22

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Acked-by: Sylvain Lemieux <slemieux.tyco@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The debug would only ever output zero for the clip information.

Signed-off-by: Nick Dyer <nick@shmanahar.org>
Reviewed-by: Andrew Duggan <aduggan@synaptics.com>
Tested-by: Chris Healy <cphealy@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Now that we have a proper API to set the attention data, there is
no point in keeping it in the transport driver.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Reviewed-by: Andrew Duggan <aduggan@synaptics.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The HID implementation of RMI4 provides the data during
the interrupt (in the input report). We need to provide
a way for this transport driver to provide the attention
data while calling an IRQ.

We use a fifo in rmi_core to not lose any incoming event.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Reviewed-by: Andrew Duggan <aduggan@synaptics.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

First check if there are data available passed by the transport device.
If data available use these data. If there are no data available
try to read the rmi block if dsata are passed this way.

This is the way the other rmi function handlers will do this.

This patch is needed on HID devices because the firmware reads F03 data
registers and adds them to the HID attention report. Reading those
registers from the driver after the firmware read them will result in
invalid data.

Reviewed-by: Andrew Duggan <aduggan@synaptics.com>
Signed-off-by: Dennis Wassenberg <dennis.wassenberg@secunet.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

This adds basic functionality for PS/2 passthrough on Synaptics
Touchpads using RMI4 through smbus.

Reviewed-by: Andrew Duggan <aduggan@synaptics.com>
Signed-off-by: Lyude Paul <thatslyude@gmail.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The code uses of_property_read_u32 and expects positive values.  However,
the values are stored in signed int variables.  Additionally, the registers
values are also stored in signed variables without a good reason
(readl/writel expect u32).

The only time this caused a real bug was in the new average-samples
property, in which the property is numerically compared and implicitly
expected to be positive.

I believe it's better to change all the properties and registers to u32,
for consistency and warnings reduction.

Signed-off-by: Guy Shapiro <guy.shapiro@mobi-wize.com>
Reported-by: Bjørn Forsman <bjorn.forsman@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Add mask of each function bits of REG_ADC_CFG, and clear these
function bits first, otherwise use '|=' operation may get the
wrong setting which depends on the original value of REG_ADC_CFG.

Signed-off-by: Haibo Chen <haibo.chen@nxp.com>
Reviewed-by: Guy Shapiro <guy.shapiro@mobi-wize.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

If struct rmi_device_platform_data contains pointers to other struct,
it gets difficult to allocate a fixed size struct and copy it over between
drivers.

Change the pointers into a struct and change the code in rmi4 accordingly.

Reviewed-by: Andrew Duggan <aduggan@synaptics.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

those functions should not be used outside of rmi_core.ko.
There is no point in exporting them to the world.

It looks like rmi_read_pdt_entry() should be static too.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

This partially reverts commit 29fd0ec2 ("Input: synaptics-rmi4 -
add support for F34 device reflash")

irq_mutex should be used only to protect data->current_irq_mask, not
preventing incoming input to be processed while the upgrade of the
firmware is happening. We can simply disable the irqs when we don't
want them to interfere with the upgrade process.

Tested on S7300 and S7800 (with F34 v7 patch added)

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Nick Dyer <nick@shmanahar.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Using regmap_update_bits(..., mask, 1) with 'mask' following (1 << k)
and k greater than 0 is wrong. Indeed, _regmap_update_bits will perform
(mask & 1), which results in 0 if LSB of mask is 0. Thus the call
regmap_update_bits(..., mask, 1) is in reality equivalent to
regmap_update_bits(..., mask, 0).

In such a case, the correct use is regmap_update_bits(..., mask, mask).

This driver is performing such a mistake with the DRV2667_STANDBY mask,
which equals (1 << 6). Fix the driver to make it consistent with the
API, and fix the alignment problem at the same time. Please note that
this change is untested, as I do not have this piece of hardware.
Testers are welcome!

Signed-off-by: Florian Vaussard <florian.vaussard@heig-vd.ch>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Using regmap_update_bits(..., mask, 1) with 'mask' following (1 << k)
and k greater than 0 is wrong. Indeed, _regmap_update_bits will perform
(mask & 1), which results in 0 if LSB of mask is 0. Thus the call
regmap_update_bits(..., mask, 1) is in reality equivalent to
regmap_update_bits(..., mask, 0).

In such a case, the correct use is regmap_update_bits(..., mask, mask).

This driver is performing such a mistake with the DRV2665_STANDBY mask,
which equals BIT(6). Fix the driver to make it consistent with the API,
and fix the alignment problem at the same time. Please note that this
change is untested, as I do not have this piece of hardware. Testers
are welcome!

Signed-off-by: Florian Vaussard <florian.vaussard@heig-vd.ch>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Set the .enabled boolean and trigger an event processing when enabling
for edge-triggered systems.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

These fields are not used anywhere, there is no point in carrying them.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Typos...

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The return value of alps_get_pkt_id_ss4_v2() should really be "enum
SS4_PACKET_ID", not "unsigned char".  Correct this.

Also, most of the Alps SS5 (SS4 v2) packet byte parsing code is implemented
using macros, but there are a few places where bytes are directly
manipulated in alps.c.  For consistency, migrate the rest of these to
macros.

Reviewed-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Paul Donohue <linux-kernel@PaulSD.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

For consistency and clarity, the input_report_*() functions should be
called by alps_process_packet_ss4_v2() instead of by alps_decode_ss4_v2().

Reviewed-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Paul Donohue <linux-kernel@PaulSD.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The current Alps SS5 (SS4 v2) code generates bogus TouchPad events when
TrackStick packets are processed.

This causes the xorg synaptics driver to print
"unable to find touch point 0" and
"BUG: triggered 'if (priv->num_active_touches > priv->num_slots)'"
messages.  It also causes unexpected TouchPad button release and re-click
event sequences if the TrackStick is moved while holding a TouchPad
button.

This commit corrects the problem by adjusting alps_process_packet_ss4_v2()
so that it only sends TrackStick reports when processing TrackStick
packets.

Reviewed-by: Pali Rohár <pali.rohar@gmail.com>
Signed-off-by: Paul Donohue <linux-kernel@PaulSD.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The i.MX6UL internal touchscreen controller contains an option to
average upon samples. This feature reduces noise from the produced
touch locations.

This patch adds sample averaging support to the imx6ul_tsc device
driver.

Signed-off-by: Guy Shapiro <guy.shapiro@mobi-wize.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Unlike previous Xbox pads, the Xbox One pad doesn't have "sticky" rumble
packets. The duration is encoded into the command and expiration is handled
by the pad firmware.

ff-memless needs pseudo-sticky behavior for rumble effects to behave
properly for long duration effects. We already specify the maximum rumble
on duration in the command packets, but it's still only good for about 2.5
seconds of rumble. This is easily reproducible running fftest's sine
vibration test.

It turns out there's a repeat count encoded in the rumble command. We can
abuse that to get the pseudo-sticky behavior needed for rumble to behave as
expected for effects with long duration.

By my math, this change should allow a single ff_effect to rumble for 10
minutes straight, which should be more than enough for most needs.

Signed-off-by: Cameron Gutman <aicommander@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

This is the new gamepad that ships with the Xbox One S which
includes Bluetooth functionality.

Signed-off-by: Cameron Gutman <aicommander@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

The PNP0C40 device of the Surface 3 doesn't have any GPIO attached to it.
Instead of trying to access the GPIO, request the count beforehand and
bail out if it is null or if an error is returned.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

gpio_keys will later use gpio_is_valid(). To match the actual
behavior, we should use it here too.

Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>