The x and y absolute axis may have different resolutions, meaning 1 unit
long motion delta on one axis is not physically as long as 1 unit motion
delta on the other axis.
In order to make these anisotropic input motion deltas output as isotropic
motion deltas, apply scaling to one of the axis making it have the same
dimension as the other, before passing it through the motion filter
which assumes all deltas are isotropic.
https://bugs.freedesktop.org/show_bug.cgi?id=79056
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This patch reimplements the simple smooth pointer acceleration profile
from X.org xserver. The algorithm is identical to the classic profile
with a non-zero pointer acceleration threshold.
When support for changable parameters is in place, to get a pointer
acceleration the same as the default classic profile of X.org a
polynomial acceleration profile should be used for when the threshold
parameter is zero.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Pointer acceleration filters may absorb the first event, so queue two,
just in case.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
A test cannot exactly predict the resulting motion event from a given
evdev event series without having to reimplement the acceleration
algorithm. To still be able to test that sane relative motion events are
produced, check that the length and direction of the resulting motion
event vectors are close to the same as the expected vectors.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
On a button click / tap the scrolling event handler no longer gets called,
ensure that any in progress scrolling is stopped.
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>
Put the actual scroll event posting in the straight path.
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>
Setting tp->scroll.direction = 0 before checking tp->scroll.direction
to see if we need to send stop scroll events for vert / horz scrolling does
not really work well.
Also we need to check direction with an axis mask, not the axis number.
While at it also add a tp_stop_scroll_events() helper function 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>
When doing 2 finger scrolling we don't want any spurious movement events after
scrolling. touchpad_2fg_no_motion tests for this, but it lifts touch 0
(which is the pointer as it came down first) first, so it only catches the
case where touch 1 suddenly gets promoted to being the pointer.
However if touch 1 is lifted first, then touch 0 is still the pointer and
will cause spurious movement events. Swap the 2 litest_touch_up calls to
catch this (and make the test fail), and add code to clear the is_pointer
flag on all touched when doing 2 finger scrolling to fix it again.
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>
Alter the so far hard coded parameters a bit to make the pointer
acceleration profile accelerate the pointer movement a bit more than
before.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
We store timestamps in ms since system boot (CLOCK_MONOTONIC). This will wrap
after circa 50 days.
I've considered making our code wrapping safe, but that won't work. We also
use our internal timestamps to program timer-fds for timeouts. And we store
ms in a single integer but the kernel uses 2 integers, one for seconds and
one for usec/nanosec. So at 32 bits our ms containing integer will wrap
in 50 days, while the kernels seconds storing integer lasts a lot longer.
So when we wrap our ms timestamps, we will be programming the timer-fds
with a seconds value in the past.
So change all our internal timestamps to uint64_t to avoid the wrapping
when programming the timer-fds. Note that we move from 64-bit timestamps to
32-bit timestamps when calling the foo_notify_bar functions from
libinput-private.h. Having 64 bit timestamps has no use past this point,
since the wayland input protocol uses 32 bit timestamps (and clients will
have to deal with wrapping).
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
This is no longer needed now that we take the button area and pinned fingers
into account.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
We already check for old != current everywhere, so this is not needed;
And in tp_post_softbutton_buttons we want to do delay button down reporting if
we don't have touch info yet in which case this check actually gets in the way.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
We should never get any non left button events on clickpads, but if we
do these might confuse our state, so complain about it and ignore these.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
And warn if INPUT_PROP_BUTTONPAD mismatches right/middle buttons presence.
Also fix the bcm5974 to properly advertise INPUT_PROP_BUTTONPAD.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
We don't want touches in the button area to cause the pointer to move. So
instead of making a touch the pointer when it moves to TOUCH_BEGIN, wait
with making it the pointer until its buttons state moves to BUTTON_STATE_AREA.
Note that a touch in the main area of the touchpad will move to
BUTTON_STATE_AREA immediately. If software-buttons are not enabled, any finger
is in the BUTTON_STATE_AREA.
While at it also refactor the is_pointer setting in general, removing
code duplicition wrt checking that another touch is not already
the pointer on unpinning a finger, and add safeguards that unpinning
does not make a finger which is not in button state BUTTON_STATE_AREA the
pointer, nor that the button code makes a pinned finger the pointer.
All these sanity checks are combined into a new tp_button_active function,
since they should be taken into account for 2 finger scrolling, etc. too.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Almost all non Apple touchpads have visible markings for software button areas,
so limit clickfinger behavior to Apple clickpads, and implement software button
areas for others.
This is a slightly fancier implementation than the simplest model and ported
over from libtouchpad. It implements a state machine for the software buttons
with left and right buttons currently implemented. Buttons are oriented
left-to-right, in a horizontal bar. No random button placement allowed.
In general, the procedure is:
- if a finger sets down in the left button area, a click is a left click
- if a finger sets down in the right button area, a click is a right click
- if a finger leaves the button area, a click is a left click
- if a finger starts outside the button area, a click is a left click
Two timeouts are used to handle buttons more smoothly:
- if a finger sets down in a button area but "immediately" moves over
to a different area, that area takes effect on a click.
- if a finger leaves a button area and "immediately" clicks or moves back into
the area, the button still takes effect on a click.
- if a finger changes between areas and stays there for a timeout, that area
takes effect on a click.
Note the button area states are named BOTTOM_foo to make it easier to later
add support for a top button area such as can be found on the Thinkpad [2-5]40
series.
Co-authored-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
To avoid having a button left press and a button right release if the number
of fingers changes.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This is about to become more complicated with the support for software button
areas. Move it to a separate file to have it logically grouped together.
No functional changes.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
No real effect, just for safety
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
On clickpads, clicking the pad usually causes some motion events. To avoid
erroneous movements, lock all fingers into position on the click and don't
allow for motion events until a finger moves past a given threshold
(currently 2% of the touchpad diagonal).
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
A single-touch touchpad that provides BTN_TOOL_TRIPLETAP has 3 touches, etc.
There aren't a lot of these out there, but some touchpads don't have slots but
do provide two- or three-finger detection.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Check the value returned by calloc.
Signed-off-by: Carlos Olmedo Escobar <carlos.olmedo.e@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
We depend on device creation on the host system, having the tests run in
parallel runs a risk of random failure.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
When building on a system with an older kernel, some KEY_ macros might
be missing. To be able to build on such system, define them if they are
missing.
It is probably better to keep our own copy of input.h somewhere in our
tree, and include that one instead of the system one, but that can be
added later.
Signed-off-by: Jonas Ådahl <jadahl@opera.com>
Compositors will need to keep provide virtual devices of supported
generic device types (pointer, keyboard, touch etc). Events from each
device capable of a certain device type abstraction should be combined
as if it was only one device.
For key and button events this means counting presses of every key or
button. With this patch, libinput provides two new API for doing just
this; libinput_event_pointer_get_seat_button_count() and
libinput_event_keyboard_get_seat_key_count().
With these functions, a compositor can sort out what key or button events
that should be ignored for a virtual device. This could for example
look like:
event = libinput_get_event(libinput);
switch (libinput_event_get_type(event)) {
...
case LIBINPUT_EVENT_POINTER_BUTTON:
device = libinput_event_get_device(event);
seat = libinput_event_get_seat(device);
pevent = libinput_event_get_pointer_event(event);
if (libinput_event_pointer_get_button_state(pevent) &&
libinput_event_pointer_get_seat_button_count(pevent) == 1)
notify_pointer_button_press(seat);
else if (libinput_event_pointer_get_button_state(pevent) &&
libinput_event_pointer_get_seat_button_count(pevent) == 0)
notify_pointer_button_release(seat);
break;
...
}
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
libinput currently handles 16 per device touch points. Test that we
behave as expected when a device has an even higher number of active
touch points.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
When the kernel sends multiple touch down or touch up for the same slot
in a row, ignore any such subsequent event ensuring libinput always
produces 1 x touch down -> [n x touch motion] -> 1 x touch up event
series.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Don't have a hard coded slot array size; instead allocate the array
needed according to the abs info reported by either libmtdev or libevdev.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
