platform/upstream/libinput.git
9 years agofilter: adjust acceleration curve depending on speed
Peter Hutterer [Fri, 19 Sep 2014 04:25:22 +0000 (14:25 +1000)]
filter: adjust acceleration curve depending on speed

The acceleration curve consists of four parts, in ascii-art like this:
        _____________
       /
  ____/
 /
/

where the x axis is the speed, y is the acceleration factor.
The first plateau is at the acceleration factor 1 (i.e. unaccelerated
movement), the second plateau is at the max acceleration factor. The threshold
in the code defines where and how long the plateau is.

This patch adjusts the curve based on a [-1, 1] range. For anything below 0,
the plateau is longer (i.e. accel kicks in at a higher speed), the second
incline is flatter (i.e. accel kicks in slower) and the max accel factor is
lower (i.e. maximum speed is slower). For anything above 0, the inverse is
true, acceleration kicks in earlier, harder and is faster in general. So the
default/min/max curves overlaid look something like this:
      ________ max
     | _______ default
    | /  _____ min
  _|_/_/
 /
/

Note that there's a limit to what ascii art can do...

Note that there are additional tweaks we can introduce later, such as
decreaseing the unaccelerated speed of the device (i.e. lowering the first
plateau).

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agofilter: add a configurable speed interface
Peter Hutterer [Thu, 3 Jul 2014 05:53:56 +0000 (15:53 +1000)]
filter: add a configurable speed interface

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agofilter: move the threshold/accel into the filter struct
Peter Hutterer [Thu, 3 Jul 2014 05:32:40 +0000 (15:32 +1000)]
filter: move the threshold/accel into the filter struct

No functional changes, prep work for the config interface.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agofilter: add a filter-private.h header file
Peter Hutterer [Thu, 3 Jul 2014 23:29:11 +0000 (09:29 +1000)]
filter: add a filter-private.h header file

To keep the implementation of a filter separate from the users of a filter.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoReplace pointer acceleration with a much simpler linear one
Peter Hutterer [Fri, 19 Sep 2014 01:10:17 +0000 (11:10 +1000)]
Replace pointer acceleration with a much simpler linear one

We ran a userstudy, evaluating three different accel methods. Detailed results are
available at:
http://www.who-t.net/publications/hutterer2014_libinput_ptraccel_study.pdf

We found that there was little difference between the method we had in
libinput 0.6 and this three-line function. Users didn't really notice a
difference, but measured data suggests that it has slight advantages in some
use-cases.

The method proposed here is the one labeled "linear" in the paper, its profile
looks roughly like this:

        _____________
       /
  ____/
 /
/

where the x axis is the speed, y is the acceleration factor.
The first plateau is at the acceleration factor 1 (i.e. unaccelerated
movement), the second plateau is at the max acceleration factor. The threshold
in the code defines where and how long the plateau is.

Differences to the previous accel function:
- both inclines are linear rather than curved
- the second incline is less steep than the current method

From a maintainer's point-of-view, this function is significantly easier to
understand and manipulate than the previous one.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoDrop two semicolons after a macro definition
Peter Hutterer [Wed, 17 Sep 2014 05:11:41 +0000 (15:11 +1000)]
Drop two semicolons after a macro definition

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: drop unused declaration evdev_proces_event
Peter Hutterer [Sun, 21 Sep 2014 23:30:38 +0000 (09:30 +1000)]
evdev: drop unused declaration evdev_proces_event

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: Add trackpoint middlebutton scrolling tests
Hans de Goede [Thu, 18 Sep 2014 09:15:27 +0000 (11:15 +0200)]
test: Add trackpoint middlebutton scrolling tests

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agolitest: Add litest_assert_scroll() helper function
Hans de Goede [Wed, 17 Sep 2014 13:35:31 +0000 (15:35 +0200)]
litest: Add litest_assert_scroll() helper function

Make check_2fg_scroll functionality available outside of touchpad.c ,
no functional changes.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Enlarge topbutton area a bit while the touchpad is disabled
Hans de Goede [Wed, 17 Sep 2014 13:35:30 +0000 (15:35 +0200)]
touchpad: Enlarge topbutton area a bit while the touchpad is disabled

Make it easier to hit the topbutton area when the touchpad is disabled,
normally we don't want to make the topbutton area too big, so as to not
interfere with normal touchpad operation, but when disabled we have no such
worries.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agocosmetic: Remove prototype for nonexistent create_linear_acceleration_filter()
Derek Foreman [Thu, 18 Sep 2014 22:02:39 +0000 (17:02 -0500)]
cosmetic: Remove prototype for nonexistent create_linear_acceleration_filter()

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: Test the topbuttons of a disabled touchpad still send events through the trackpoint
Peter Hutterer [Tue, 16 Sep 2014 14:22:42 +0000 (16:22 +0200)]
test: Test the topbuttons of a disabled touchpad still send events through the trackpoint

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: When disabling a TOPBUTTONPAD, leave the top buttons enabled
Hans de Goede [Tue, 16 Sep 2014 14:22:41 +0000 (16:22 +0200)]
touchpad: When disabling a TOPBUTTONPAD, leave the top buttons enabled

On a TOPBUTTONPAD, we can't disable the touchpad altogether - the trackstick
relies on the touchpad's top software buttons.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Put state unrolling code in a tp_clear_state() helper function
Hans de Goede [Tue, 16 Sep 2014 14:22:40 +0000 (16:22 +0200)]
touchpad: Put state unrolling code in a tp_clear_state() helper function

For touchpads with top softbuttons we don't want to fully disable the touchpad
on suspend, as we want to keep the top softbuttons working for the trackpoint.

So in the suspended state some of the touchpad sub-statemachines will keep
running (e.g. buttons) where others (e.g. tap) will not. This means that
we will need to clear the touchpad state on resume too, to avoid things
being in an inconsistent state after resume.

This commit factors out the state clearing code into a helper functions, so
that the same code can be used on resume.

No functional changes.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Route top softbuttons through the trackstick if we've one
Hans de Goede [Tue, 16 Sep 2014 14:22:39 +0000 (16:22 +0200)]
touchpad: Route top softbuttons through the trackstick if we've one

The touchpad top softbuttons such as found on the Lenove T440 are intended for
use with the trackstick. Route their events through the trackstick, so that
they can be used for e.g. middle button scrolling with the trackstick.

Note that sending top button events to a disabled trackpoint makes no sense
(and will mess up internal state). Likely a user with a disabled trackpoint
will still expect the top buttons to work, so rather than not sending events
in that case, simply treat a suspendeded trackpoint as not being there, and
send the events directly from the touchpad device.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Keep track of associated trackpoint device
Peter Hutterer [Tue, 16 Sep 2014 14:22:38 +0000 (16:22 +0200)]
touchpad: Keep track of associated trackpoint device

The top soft buttons are intended for use with a trackpoint, and to e.g.
make middle button scrolling work correctly, we must post the events for
these "buttons" through the trackpoint device.

This commit is a preparation patch for this, it adds a link to the
trackpoint to the touchpad, but does not yet do anything with it.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: Add an internal device suspend / resume notification system
Hans de Goede [Tue, 16 Sep 2014 14:22:37 +0000 (16:22 +0200)]
evdev: Add an internal device suspend / resume notification system

We have the ability for a device to form a link to another device through the
device_added / device_removed callbacks. A device having such a link to
another device may also want to know when that other device is disabled /
enabled (suspended / resumed). So add a notification mechanism for this too.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: Add middle button scrolling for trackpoints
Hans de Goede [Tue, 16 Sep 2014 14:22:36 +0000 (16:22 +0200)]
evdev: Add middle button scrolling for trackpoints

Most trackpoint users want to be able to scroll using the trackpoint with
the middle button pressed, add support for this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: Move generic scroll code from evdev-mt-touchpad.c to evdev.c
Hans de Goede [Tue, 16 Sep 2014 14:22:35 +0000 (16:22 +0200)]
evdev: Move generic scroll code from evdev-mt-touchpad.c to evdev.c

So that it can be used for middle button trackpoint scrolling too.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: move softbutton initialization to separate function
Peter Hutterer [Thu, 4 Sep 2014 02:55:02 +0000 (12:55 +1000)]
touchpad: move softbutton initialization to separate function

No functional changes

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: implement conditional disabling
Peter Hutterer [Wed, 3 Sep 2014 05:45:57 +0000 (15:45 +1000)]
touchpad: implement conditional disabling

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: add internal tagging system
Peter Hutterer [Wed, 3 Sep 2014 05:50:28 +0000 (15:50 +1000)]
evdev: add internal tagging system

For conditional touchpad disabling we need two pieces of knowledge: is the
device an internal touchpad and is another device an external mouse-like
device. For that use-case it's enough to tag any device that's on USB and
Bluetooth with pointer capabilities as external mouse. A more complex can be
done when needed.

The tag function is part of the dispatch interface (to save on udev code) and
called before the caller is notified about the new device, i.e. the device is
fully configured by the time it needs to be tagged, and other devices can rely
on the tags being assigned by the time they get notified about the new device.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: add a internal device notification mechanism
Peter Hutterer [Wed, 3 Sep 2014 04:34:52 +0000 (14:34 +1000)]
evdev: add a internal device notification mechanism

When a device is added or removed, notify all internal devices about the
device change. This allows all devices to configure themselves depending on
other devices in the system. Prime use-case here is an internal touchpad that
wants to know if an external mouse is connected.

On device added, notification goes both ways: existing devices are notified
about the new device, and the new device is notified about existing devices.
On device removed, notification only goes one way.

In both cases, the internal notification is complete before the event is sent
to the caller.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: test for release events on device suspend
Peter Hutterer [Wed, 20 Aug 2014 00:32:29 +0000 (10:32 +1000)]
test: test for release events on device suspend

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: hook up sendevents configuration
Peter Hutterer [Mon, 1 Sep 2014 07:17:18 +0000 (17:17 +1000)]
touchpad: hook up sendevents configuration

We may be in the middle of a software button click or a tap, so make sure we
go back to the device-neutral state by unwinding.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: split handling the state into a separate function
Peter Hutterer [Mon, 1 Sep 2014 07:05:49 +0000 (17:05 +1000)]
touchpad: split handling the state into a separate function

No functional changes

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: drop the button count when releasing keys on remove
Peter Hutterer [Wed, 20 Aug 2014 01:11:28 +0000 (11:11 +1000)]
evdev: drop the button count when releasing keys on remove

We previously called this function only before device removal, so failing to
update the button state didn't matter. To make this function generic for the
device suspend/resume, we need to keep track of the button/key count properly.

If we have a key down multiple times on suspend though, log a bug.
The dispatch should release the keys before we even get here (functionality
added in a subsequent patch).

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add test for device suspend/resume
Peter Hutterer [Thu, 30 Jan 2014 05:40:35 +0000 (15:40 +1000)]
test: add test for device suspend/resume

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: don't resume a removed device
Peter Hutterer [Thu, 30 Jan 2014 06:33:24 +0000 (16:33 +1000)]
evdev: don't resume a removed device

A device may disappear and a new device may re-appear with the same device
node while the original device is suspended. Prevent a device resume to open
the wrong device.

In a path context, a changing syspath is the only indicator we get of the
device changing.
In a udev context, if the device was removed and libinput_dispatch() was
called, we can short-cut the syspath comparison by setting it to NULL.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: hook up a generic enable/disable interface for devices
Peter Hutterer [Fri, 22 Aug 2014 03:40:40 +0000 (13:40 +1000)]
evdev: hook up a generic enable/disable interface for devices

The evdev fallback dispatch supports enabling and disabling devices. That's
fairly easy to support since we don't (yet) have extra event generation within
the fallback backend. Thus, we can simply close the fd and re-open it again
later.

Touchpads are currently excluded here, they generate extra events on tapping,
scrolling, and software buttons and need a more complex implementation.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: add helper function evdev_need_mtdev
Peter Hutterer [Mon, 15 Sep 2014 04:59:07 +0000 (14:59 +1000)]
evdev: add helper function evdev_need_mtdev

Prep work for upcoming patches

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoAdd a config interface for enabling/disabling event generation from a device
Peter Hutterer [Thu, 30 Jan 2014 06:18:55 +0000 (16:18 +1000)]
Add a config interface for enabling/disabling event generation from a device

Rather than adding a config interface to disable a device merely allow a
caller to toggle the "send events" mode on the device. If off, the device
won't send events (though further events may be received depending on the
current state of the device).
Default is enabled, i.e. the device sends events.

A special mode is added to the obvious enable/disable: disable the device when
an external mouse is connected. Once set, the device will be enabled when no
mouse is present and stop sending events otherwise. This isn't hooked up to
anything yet though.

Built into the config API is the default option of "enabled". Any device
supports this, for the obvious reason. Disabling or conditionally disabling is
left to the implementation.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: prevent double-suspending a device
Peter Hutterer [Thu, 30 Jan 2014 05:55:37 +0000 (15:55 +1000)]
evdev: prevent double-suspending a device

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: factor out closing a device into evdev_suspend()
Peter Hutterer [Thu, 30 Jan 2014 05:34:00 +0000 (15:34 +1000)]
evdev: factor out closing a device into evdev_suspend()

No functional changes, just prep work for an upcoming patch

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: prefix the hw key/button bitmask with 'hw'
Peter Hutterer [Fri, 22 Aug 2014 00:41:20 +0000 (10:41 +1000)]
evdev: prefix the hw key/button bitmask with 'hw'

This bitmask reflects the hw state, prefix it accordingly.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: fix tap-and-drag handling for timeouts
Peter Hutterer [Wed, 3 Sep 2014 01:16:20 +0000 (11:16 +1000)]
touchpad: fix tap-and-drag handling for timeouts

Doing a tap-and-drag gesture but just holding the finger instead of moving
should trigger a timeout and still switchin into tap-and-drag.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: wait for events in litest_assert_button_events
Peter Hutterer [Wed, 3 Sep 2014 00:57:25 +0000 (10:57 +1000)]
test: wait for events in litest_assert_button_events

Takes the onus of waiting from the caller.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: move assert_button_event to litest proper
Peter Hutterer [Wed, 3 Sep 2014 00:53:00 +0000 (10:53 +1000)]
test: move assert_button_event to litest proper

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add litest_add_device()
Peter Hutterer [Fri, 29 Aug 2014 04:08:46 +0000 (14:08 +1000)]
test: add litest_add_device()

For adding a litest device to an existing context.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: avoid erroneous devices to be passed into the test suites
Peter Hutterer [Mon, 4 Aug 2014 02:49:59 +0000 (12:49 +1000)]
test: avoid erroneous devices to be passed into the test suites

The litest features overlap with the litest device specifiers, so it's easy to
pass in LITEST_MOUSE where LITEST_POINTER should be passed in, and vice versa.
Lacking proper type checking the best we can do here is simply move the
devices into the negative range and check for that.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoAdd a helper function for clock_gettime
Peter Hutterer [Mon, 1 Sep 2014 06:47:28 +0000 (16:47 +1000)]
Add a helper function for clock_gettime

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: print strerror() if uinput device creation fails
Peter Hutterer [Wed, 17 Sep 2014 00:30:45 +0000 (10:30 +1000)]
test: print strerror() if uinput device creation fails

The most common error running the test suite is not running as root, but the
error message is hard to interpret. Make it more explicit when it failed,
printing the strerror of the errno.

Note that libevdev 1.3 is needed to get EACCES instead of EBADF
http://cgit.freedesktop.org/libevdev/commit/?id=debe9b030c8069cdf78307888ef3b65830b25122
A workaround is put in place for now until libevdev 1.3 is commonplace.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: fix a jumping touch movement
Peter Hutterer [Tue, 16 Sep 2014 04:30:44 +0000 (14:30 +1000)]
test: fix a jumping touch movement

touch_move_to() should usually continue from the touch_down() location

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoconfigure.ac: libinput 0.6 0.6.0
Jonas Ådahl [Thu, 11 Sep 2014 20:32:51 +0000 (22:32 +0200)]
configure.ac: libinput 0.6

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
9 years agobuild: symbol ck_assert_ptr_ne requires check-0.9.10
Jan Engelhardt [Tue, 9 Sep 2014 23:32:24 +0000 (01:32 +0200)]
build: symbol ck_assert_ptr_ne requires check-0.9.10

openSUSE 12.3 ships with check-0.9.9 and subsequently fails to build
the tests. Change the call to look for check >= 0.9.10 where that
symbol is available.

Signed-off-by: Jan Engelhardt <jengelh@inai.de>
9 years agoDocument the static udev configuration options we support
Peter Hutterer [Tue, 9 Sep 2014 23:31:08 +0000 (09:31 +1000)]
Document the static udev configuration options we support

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoDocument LIBINPUT_CALIBRATION_MATRIX properly
Peter Hutterer [Tue, 9 Sep 2014 23:00:09 +0000 (09:00 +1000)]
Document LIBINPUT_CALIBRATION_MATRIX properly

Make this part of our API proper and outline the 4 most common examples.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agolitest-alps-semi-mt: Fix compiler warnings
Hans de Goede [Tue, 2 Sep 2014 14:34:48 +0000 (16:34 +0200)]
litest-alps-semi-mt: Fix compiler warnings

This fixes the following (false positive) compiler warnings:

litest-alps-semi-mt.c: In function 'alps_touch_move':
litest-alps-semi-mt.c:163:3: warning: 'b' may be used uninitialized in this function [-Wmaybe-uninitialized]
   send_abs_mt_xy(d, r, b);
   ^
litest-alps-semi-mt.c:163:3: warning: 'r' may be used uninitialized in this function [-Wmaybe-uninitialized]
litest-alps-semi-mt.c: In function 'alps_touch_down':
litest-alps-semi-mt.c:127:3: warning: 'b' may be used uninitialized in this function [-Wmaybe-uninitialized]
   send_abs_mt_xy(d, r, b);
   ^
litest-alps-semi-mt.c:127:3: warning: 'r' may be used uninitialized in this function [-Wmaybe-uninitialized]

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Only break out of tap FSM for clickpad button presses
Jonas Ådahl [Sun, 27 Jul 2014 21:28:31 +0000 (23:28 +0200)]
touchpad: Only break out of tap FSM for clickpad button presses

It should be possible to initiate a drag by tapping-drag, but continue
it by pressing a physical button continuing to drag by subsequent finger
motions.

As the generic evdev layer helps us ignore multiple button presses we
can have the tap machine run completely separate from and uneffected by
regular physical button presses, making the tap FSM much simpler than
adding new states for handling button presse life times from outside
of the tap state machine.

A touchpad test is updated to test click while tapping instead of tap
FSM break out. The updated test is re-added but only for clickpads only.

The tap FSM svg is updated to say "clickpad button press" instead of
"phys button press".

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: use the evironment variable for check's verbosity
Peter Hutterer [Mon, 1 Sep 2014 02:40:59 +0000 (12:40 +1000)]
test: use the evironment variable for check's verbosity

Allows to set CK_VERBOSITY to be set to "silent", "minimal", "normal", or
"verbose". Falls back to CK_NORMAL if unset.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: fix infinite loop in litest_wait_for_event_of_type()
Peter Hutterer [Mon, 1 Sep 2014 02:39:38 +0000 (12:39 +1000)]
test: fix infinite loop in litest_wait_for_event_of_type()

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: load the LIBINPUT_CALIBRATION_MATRIX as default matrix
Peter Hutterer [Tue, 26 Aug 2014 03:44:03 +0000 (13:44 +1000)]
evdev: load the LIBINPUT_CALIBRATION_MATRIX as default matrix

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoChange calibration into a configuration option
Peter Hutterer [Tue, 26 Aug 2014 01:41:19 +0000 (11:41 +1000)]
Change calibration into a configuration option

New configuration API:
libinput_device_config_calibration_has_matrix()
libinput_device_config_calibration_set_matrix()
libinput_device_config_calibration_get_matrix()
libinput_device_config_calibration_get_default_matrix()

Deprecates libinput_device_calibrate().

For coordinate transformation, we're using a precalculated matrix. Thus, to
support ..._get_matrix() we need to store the original user-specified matrix
separately, in an unmangled state.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add tests for touch calibration
Peter Hutterer [Wed, 20 Aug 2014 07:16:06 +0000 (17:16 +1000)]
test: add tests for touch calibration

Basic tests for rotation, translation and scaling events.

Note that tests need to be added separately for single-touch and touch
devices, this is a restriction of the litest framework.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoudev: drop WL_CALIBRATION, replace with LIBINPUT_CALIBRATION_MATRIX
Peter Hutterer [Tue, 26 Aug 2014 04:02:44 +0000 (14:02 +1000)]
udev: drop WL_CALIBRATION, replace with LIBINPUT_CALIBRATION_MATRIX

WL_CALIBRATION is a weston-specific property that required the translation
component of the matrix to be in pixels. libinput can't provide calibration
based on unknown outputs, so drop support for this property.

The basic functionality is maintained, renaming just makes the change in
behavior more explicit.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: switch to a normalized transformation matrix
Peter Hutterer [Tue, 26 Aug 2014 00:34:05 +0000 (10:34 +1000)]
evdev: switch to a normalized transformation matrix

The big change here is the requirement to have the translation component in a
device-normalized coordinate space. Without that, we cannot reliably rotate as
the coordinate space is effectively unknown and may differ between the axes.
This affects any rotation matrix or translation matrix, pure scale matrices
were working just fine since they're unit-less.

Requiring the matrix in device-normalized space makes it possible for libinput
to rotate or otherwise handle the matrix independent of the screen resolution.
The rotation matrix is documented in a bit more detail to make it easier for
users to figure it out.

This changes the definition of the WL_CALIBRATION property (which is currently
broken).

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoutil: add a couple of 3x3 matrix helper functions
Peter Hutterer [Mon, 25 Aug 2014 23:37:29 +0000 (09:37 +1000)]
util: add a couple of 3x3 matrix helper functions

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: apply calibration to multitouch values as well
Peter Hutterer [Wed, 20 Aug 2014 08:23:43 +0000 (18:23 +1000)]
evdev: apply calibration to multitouch values as well

We apply calibration to single-touch and absolute devices, but we might as
well do so for multitouch events.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: constify evdev_device_calibrate
Peter Hutterer [Tue, 26 Aug 2014 02:57:41 +0000 (12:57 +1000)]
evdev: constify evdev_device_calibrate

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoudev: use evdev_device_calibrate() instead of manually writing the matrix
Peter Hutterer [Tue, 26 Aug 2014 00:33:39 +0000 (10:33 +1000)]
udev: use evdev_device_calibrate() instead of manually writing the matrix

We have a wrapper, use it.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add a generic single-touch device
Peter Hutterer [Thu, 21 Aug 2014 01:08:29 +0000 (11:08 +1000)]
test: add a generic single-touch device

With a non-zero absmin for both axes and different ranges for x/y, just to
detect those errors.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add helpers to wait for specific events
Peter Hutterer [Wed, 20 Aug 2014 07:15:50 +0000 (17:15 +1000)]
test: add helpers to wait for specific events

litest_wait_for_event() returns if any event is available.
litest_wait_for_event_of_type(... type, type, type, -1) returns if any of the
given event types is availble. All other events are discarded.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoFix a doxygen reference
Peter Hutterer [Tue, 26 Aug 2014 01:33:59 +0000 (11:33 +1000)]
Fix a doxygen reference

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: silence Coverity warnings about uninitialized use
Peter Hutterer [Fri, 22 Aug 2014 05:15:17 +0000 (15:15 +1000)]
touchpad: silence Coverity warnings about uninitialized use

container_of() accesses tp for offset calculation. Which is fine, but
Coverity doesn't know that.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoSwap conditions for ARRAY_FOR_EACH()
Peter Hutterer [Fri, 22 Aug 2014 05:08:57 +0000 (15:08 +1000)]
Swap conditions for ARRAY_FOR_EACH()

The current conditions result in _elem being assigned _arr[i] before the
condition is checked. This is fine since we then break from the loop and don't
access it anyway, but it makes Coverity unhappy.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: mark a intentional switch case fallthrough as such
Peter Hutterer [Fri, 22 Aug 2014 04:58:48 +0000 (14:58 +1000)]
touchpad: mark a intentional switch case fallthrough as such

Both motion and timeout expiry transition into the TOUCH_2_HOLD state, but
only for motion do we need to cancel the current timeout.

Found by Coverity.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoDon't close the fd if libinput_add_fd() fails
Peter Hutterer [Fri, 22 Aug 2014 04:54:06 +0000 (14:54 +1000)]
Don't close the fd if libinput_add_fd() fails

Let the caller decide what to do with the fd.

In the current code the caller can't know if the fd was closed on error since
we return NULL on malloc failure as well as on epoll_ctl() failure. In the
latter case the fd was closed, not in the former. The caller had to close
the fd anyway (and all three callers do), so drop closing the fd from this
function.

Found by Coverity.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: plug memory leak on libevdev_new_from_fd failure
Peter Hutterer [Fri, 22 Aug 2014 04:48:42 +0000 (14:48 +1000)]
evdev: plug memory leak on libevdev_new_from_fd failure

Found by coverity.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoDisable static libraries
Peter Hutterer [Mon, 18 Aug 2014 21:59:10 +0000 (07:59 +1000)]
Disable static libraries

Our static library leaks symbols like crazy, some of which are likely
conflicts with users of this library (log_msg, open_restricted, ...).

Disale static linking by default so we don't have to spend time debugging
this.

Related to:
https://bugs.freedesktop.org/show_bug.cgi?id=82292
https://bugs.freedesktop.org/show_bug.cgi?id=82785

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoUse -no-install instead of -static for local noinst linking
Peter Hutterer [Mon, 18 Aug 2014 21:57:25 +0000 (07:57 +1000)]
Use -no-install instead of -static for local noinst linking

The goal of -static was to avoid the libtool wrappers for easier debugging.
The -no-install flag does exactly that, without requiring static linking.

Related to https://bugs.freedesktop.org/show_bug.cgi?id=82292

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: Release still pressed keys/buttons when removing device
Jonas Ådahl [Sun, 27 Jul 2014 14:02:46 +0000 (16:02 +0200)]
evdev: Release still pressed keys/buttons when removing device

When removing a device, its not guaranteed that all button or key
presses have been released, resulting in an invalid seat wide button
count.

Note that kernel devices normally will send release events when being
unplugged, but this won't happen when removing a device from the path
backend.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: Keep track of button/key press count per device
Jonas Ådahl [Sun, 27 Jul 2014 13:54:49 +0000 (15:54 +0200)]
evdev: Keep track of button/key press count per device

Keep track of the number of times a given button or key is pressed on a
device. For regular mouse devices or keyboard devices, such a count will
never exceed 1, but counting button presses could help when button
presses with the same code can originate from different sources. One could
for example implement overlapping tap-drags with button presses by
having them deal with their own life-time independently, sorting out
when the user should receive button presses or not depending on the
pressed count.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: Ignore key/button release events if key was never pressed
Jonas Ådahl [Sun, 27 Jul 2014 13:43:59 +0000 (15:43 +0200)]
evdev: Ignore key/button release events if key was never pressed

The kernel may send a 'release' event without ever having sent a key
'pressed' event in case the key was pressed before libinput was
initiated. Ignore these events so that we always guarantee a release
event always comes after a pressed event for any given key or button.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
9 years agoevdev: Use helper for separating buttons from keys
Jonas Ådahl [Sat, 12 Jul 2014 22:25:36 +0000 (00:25 +0200)]
evdev: Use helper for separating buttons from keys

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoevdev: Let dispatch instances set their own capabilities
Jonas Ådahl [Sat, 26 Jul 2014 19:34:47 +0000 (21:34 +0200)]
evdev: Let dispatch instances set their own capabilities

It's up to a evdev device backend to configure seat capabilities it
supports. Even though it may be possible for a touchpad to have extra
keys, there is currently no support for sending keyboard events from the
touchpad driver, and if that would be implemented, it'd be a detail of
the touchpad driver, not the generic evdev device part.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: Don't fail when events are enabled multiple times
Jonas Ådahl [Mon, 14 Jul 2014 21:34:25 +0000 (23:34 +0200)]
test: Don't fail when events are enabled multiple times

When overriding events of a test device, if one would enable an event
that was already enabled by default for the overridden device, an assert
checking if the event was already enabled would fail and cause the test
to fail.

Since the merging of the default and overriding event lists is implemented
by simply concatinating them letting libevdev deal with ignoring
superfluous event enabling, remove the assert to allow the implementation
to work.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: Remove test device from context when deleting
Jonas Ådahl [Sun, 13 Jul 2014 22:04:14 +0000 (00:04 +0200)]
test: Remove test device from context when deleting

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: Use only one test device for some udev and path tests
Jonas Ådahl [Sun, 13 Jul 2014 22:01:10 +0000 (00:01 +0200)]
test: Use only one test device for some udev and path tests

Some tests in test/path.c and test/udev.c are not dependent on
device behaviour but rather managing of device lifetime etc. Run those
tests only once with only one device, resulting more or less the same
code coverage but shorter run time.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: increase top software button area to 15%
Peter Hutterer [Wed, 6 Aug 2014 02:22:38 +0000 (12:22 +1000)]
touchpad: increase top software button area to 15%

We had reports that the top software button area is hard to hit for those
using the trackpoint and clicking the buttons with their thumb.

Analysis of event recordings (3 different people) for left, right and middle
clicks shows that there is a significant amount of events up to about 10mm
(with outliers up to 12mm) from the top of the touchpad. That maps to 15%.

Interestingly, the middle button is not affected by this, presumably the
haptic feedback of the little dots sticking out from the surface make hitting
the button easier.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add a semi-mt Alps test device devel upstream/0.5.0+5+gd1cc842
Peter Hutterer [Thu, 24 Jul 2014 03:18:56 +0000 (13:18 +1000)]
test: add a semi-mt Alps test device

Provides the bounding box only, with slot 0 always being the upper/left, slot
1 being the lower-right touch. This needs to use the touch_down etc. litest
interfaces, which are now widened to double (leftover from 489630f58) and a
device-specific private pointer in the litest device.

New device feature for litest: LITEST_SEMI_MT

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoevdev: don't return a width/height if we faked the resolution
Peter Hutterer [Thu, 24 Jul 2014 06:15:43 +0000 (16:15 +1000)]
evdev: don't return a width/height if we faked the resolution

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: reset motion history when nfingers changes on semi-mt pads
Hans de Goede [Mon, 21 Jul 2014 13:25:47 +0000 (15:25 +0200)]
touchpad: reset motion history when nfingers changes on semi-mt pads

On semi-mt touchpads the reported position of the first finger down may
jump when the pad switches from st to mt mode. When this happens a large
delta gets seen on the first finger at the same time the second fingers
is first seen down, causing a spurious 2 finger scroll event.

Reset the motion history when nfingers changes on semi-mt pads to avoid this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoDocument that the delta from pointer events is accelerated
Peter Hutterer [Wed, 23 Jul 2014 08:03:15 +0000 (18:03 +1000)]
Document that the delta from pointer events is accelerated

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotest: auto-update for BTN_TOOL_* when using litest_touch_ functions
Peter Hutterer [Mon, 21 Jul 2014 02:30:40 +0000 (12:30 +1000)]
test: auto-update for BTN_TOOL_* when using litest_touch_ functions

Set BTN_TOUCH, BTN_TOOL_DOUBLETAP automatically depending on the number of
fingers down.

This emulates real event sequences a bit better than the current approach,
though it's not a 100% correct emulation:
1) On real devices, BTN_* are usually sent last before the SYN_REPORT - here
   they are sent first to slot in with the custom, device-specific event
   sequence. We should only ever look at the complete sequence anyway, so this
   shouldn't matter.
2) On real devices, the switch from BTN_TOOL_DOUBLETAP to TRIPLETAP and vice
   versa is not always toggled within the same SYN_REPORT
3) On synaptics devices, BTN_TOUCH is released in the frame where
   BTN_TOOL_DOUBLETAP is set. It is then immediately set again in the next
   frame.  With the current litest framework this is hard to integrate, so we
   just leave BTN_TOUCH set the whole time, which is what MT devices do if
   they don't have BTN_TOOL_DOUBLETAP.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoconfigure.ac: libinput 0.5 0.5.0
Jonas Ådahl [Tue, 22 Jul 2014 19:37:02 +0000 (21:37 +0200)]
configure.ac: libinput 0.5

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
9 years agotest/path: Avoid creating ignored test devices
Jonas Ådahl [Wed, 16 Jul 2014 19:14:51 +0000 (21:14 +0200)]
test/path: Avoid creating ignored test devices

Some tests doesn't use or doesn't need to use the test device
automatically created when adding a test case for certain types of
devices. For these tests, to shorten test run time, don't create the
test devices that would be ignored.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agostyle fix: Remove duplicate empty lines
Peter Hutterer [Mon, 21 Jul 2014 23:00:27 +0000 (09:00 +1000)]
style fix: Remove duplicate empty lines

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotools: always enable tapping in the event-gui program
Peter Hutterer [Mon, 21 Jul 2014 00:58:24 +0000 (10:58 +1000)]
tools: always enable tapping in the event-gui program

This is a debugging tool, so the features to debug should be enabled by
default.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agoUse an enum to enable/disable tapping configuration
Peter Hutterer [Mon, 21 Jul 2014 01:07:25 +0000 (11:07 +1000)]
Use an enum to enable/disable tapping configuration

More expressive in the caller and less ambiguous about return values (is it 1?
is it non-zero? can it be negative?)

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: add test for 3-finger tapping
Peter Hutterer [Mon, 21 Jul 2014 02:29:19 +0000 (12:29 +1000)]
test: add test for 3-finger tapping

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: make tp_estimate_delta return fractions
Hans de Goede [Fri, 18 Jul 2014 09:06:39 +0000 (11:06 +0200)]
touchpad: make tp_estimate_delta return fractions

Force a cast of the input arguments to a double before the divide, rather
than after the divide.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Create fake touches for BTN_TOOL_FOO on multi-touch pads too
Hans de Goede [Fri, 18 Jul 2014 09:06:38 +0000 (11:06 +0200)]
touchpad: Create fake touches for BTN_TOOL_FOO on multi-touch pads too

Multi-touch pads may track less touches then they can report fingers being
present through BTN_TOOL_FOO. So create fake touches for fingers reported
by BTN_TOOL_FOO on multi-touch pads too (when necessary).

This fixes e.g. 3 finger tap not working on the T440s.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Don't process fake touches if they are not dirty
Hans de Goede [Fri, 18 Jul 2014 09:06:37 +0000 (11:06 +0200)]
touchpad: Don't process fake touches if they are not dirty

Don't process fake touches, e.g. re-adding the same position to the motion
history when they are not dirty. This could trigger for example on a button
press.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: Protect tp_begin_touch and tp_end_touch against being called twice
Hans de Goede [Fri, 18 Jul 2014 09:06:36 +0000 (11:06 +0200)]
touchpad: Protect tp_begin_touch and tp_end_touch against being called twice

They were already protected against being called twice between SYN_REPORT, but not
for being called twice before a SYN_REPORT arrives.

This allows simplifying tp_process_fake_touch a bit. Note while at it also
also flip the if condition in tp_process_fake_touch so that the if is
true when the finger is down, and remove the bogus t->state = TOUCH_NONE.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: always enable palm detection on apple touchpads
Peter Hutterer [Mon, 21 Jul 2014 04:20:35 +0000 (14:20 +1000)]
touchpad: always enable palm detection on apple touchpads

They don't set resolution so we can't calculate the size but we know they're
big enough to need palm detection.

And fix the descriptor for the bcm5974. For some reason this was advertising
synaptics coordinates. Fix it to represent (one of) the apple touchpads.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotest: touchpads are too small for palm if we can't get the dimensions
Peter Hutterer [Mon, 21 Jul 2014 04:35:42 +0000 (14:35 +1000)]
test: touchpads are too small for palm if we can't get the dimensions

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: always init the left/right palm edge to INT_MIN/MAX
Peter Hutterer [Mon, 21 Jul 2014 01:27:21 +0000 (11:27 +1000)]
touchpad: always init the left/right palm edge to INT_MIN/MAX

A touchpad without resolution support had the values set to 0, disabling
pointer movement.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agoFix two doxygen errors
Peter Hutterer [Mon, 21 Jul 2014 01:15:43 +0000 (11:15 +1000)]
Fix two doxygen errors

Add the config status enum to the config doxygen group, and remove a
superfluous argument for an @ingroup command.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
9 years agotouchpad: don't init edge palm detection on touchpads less than 8cm across
Peter Hutterer [Tue, 15 Jul 2014 04:01:00 +0000 (14:01 +1000)]
touchpad: don't init edge palm detection on touchpads less than 8cm across

On small touchpads, a touch that intends to go across the width of the
touchpad is likely to start in the edge zone. Likewise, on those touchpads the
chances of a palm event happening on the edge is small.

A minimum width of 8cm determined by an elaborate process of completely
unscientific guesswork: the x220 is roughly 7.5cm across and doesn't suffer
much from edge events, the T440s is 10cm across and definitely suffers from
it. So the trigger width likely somewhere in between which makes 8cm about as
valid as any other guess.

Note that this disables palm detection for resolution-less touchpads too - if
we don't know how big the touchpad is we can't know if palm detection on the
edges is necessary.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
9 years agotouchpad: require a <45 degree movement for a palm to become a touch
Peter Hutterer [Mon, 14 Jul 2014 06:38:46 +0000 (16:38 +1000)]
touchpad: require a <45 degree movement for a palm to become a touch

Any legitimate finger movement that starts in the palm area is expected to
move out of the palm area at an angle roughly orthogonal to the edge of the
touchpad. Check for the direction of the movement vector, and if it is within
the accepted cardinal/ordinal directions then proceed.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>