In an attempt to bring method into the madness, normalize the touchpad deltas
to those of a USB mouse with 400 dpi. This way the data we're dealing with in
the acceleration code is of a known quantity.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
The resolution-based scaling may result in deltas of 0. The accel code doesn't
handle 0 deltas too well. 0/0 deltas can't happen for EV_REL devices, so the
code just isn't designed for it.
Most notably, events with delta 0/0 have no direction. That messes up the
history-based velocity calculation which stops whenever the current vector's
direction changes from the one in the trackers.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This just moves a decimal point around, at the expense of making the approach
harder to understand. The only time the const acceleration matters is when
applied to the velocity but it only matters in relation to the threshold which
is a fixed number.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
velocity is in unit/ms, the threshold is in units/ms. Once we divide
velocity/threshold, we're not in units/ms anymore but have a unitless factor.
Use a separate variable to avoid confusion.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Instead of always going from an accel of 0.0 to 1.0 between a velocity of
0.0 and 1.0, make the lead-in length of the curve depend on the threshold
setting, using half of the threshold for the lead-in.
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>
There was an error in the value passed to the second calc_penumbral_gradient
call causing a jump in the acceleration curve. This commit fixes 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>
Cleanup the code a bit, and make sure accel is at least 1.0 .
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>
I doubt this does what we think it does. It doesn't soften the delta changes,
rather it introduces bumps in the smooth processing of the changes.
abs(delta) below 1.0 is untouched, and abs(delta) beyond 3 or 4 isn't
noticable much. But in the slow range around the 1/-1 mark there is a bump.
For example, if our last_delta is 1.0 and delta is 1.1, the "softened"
delta is set to 0.6. That is stored as last delta, so an input sequence of:
0.8, 0.9, 1.0, 1.1, 1.2, 1.0, 0.8, 1.1
results in "softened" deltas that don't match the input:
0.8, 0.9, 1.0, 0.6, 0.7, 1.0, 0.8, 0.6
A better approach at smoothing this out would be to calculate the softened as:
current = current ± diff(last, current) * 0.5
or even weighted towards the new delta
current = current ± diff(last, current) * 0.25
In tests, this makes little difference. Dropping this function altogether is
sufficient to make the pointer pointer behave slightly better at low speeds
though the increase is small enough to attribute to confirmation bias.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
We send two delta events. One may get eaten or softened by the accel code but
our expectation should be that both may get through, so the length of the
expected vector is √((2dx)² + (2dy)²). That is the maximum length we expect
though for deltas ranged [-1, 1].
Deltas above the threshold would fail this test but we can fix that when
needed.
Pointer acceleration is subject to timing changes. When running tests in
valgrind pointer accel timeouts and tracker resets may happen so we can't
guarantee a specific acceleration length.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
For better consistency with filter_dispatch(). And move the things around to keep
the consumable API together.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
The original intention of this state was to prevent an accidental move out of
the bottom software button to start moving the cursor. That ends up actually
preventing a number of normal moves that start low enough. Simply drop the
state.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
The current 20% is excessive. On the t440s, the button size amounts to ~14mm
from the bottom. On the x220 it amounts to ~9mm, leaving only 31mm as actual
touchpad.
Reduce it to 15% instead, which amounts to 10.5mm on the t440 and 6mm on the
x220. Cap the button height further by making buttons a maximum height of
10mm, anything larger than that is excessive anyway.
Smaller buttons should be acceptable since we can rely on the bottom edge to
be a haptic feedback and thus a good hit-target, somewhat simliar to how
screen edges are good hit-targets.
The top software buttons stay the same size, but prefer a physical size of 6mm
instead (which is 1mm below the button marker line on the T440s). If no y
resolution is available, fall back to the 8% which is 5.6mm on the T440s.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Now that we have run-time changes of the tap.enabled state move the check
to the IDLE state only. Otherwise the tap machine may hang if tapping is
disabled while a gesture is in progress.
Two basic tests are added to check for the tap default setting - which is now
"tap disabled by default", for two reasons:
* if you don't know that tapping is a thing (or enabled by default), you get
spurious button events that make the desktop feel buggy.
* if you do know what tapping is and you want it, you usually know where to
enable it, or at least you can search for it.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Provide an interface to enable/disable tapping, with a default mapping of
1/2/3 fingers mapping to L/R/M button events, respectively.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
timer.c: In function ‘libinput_timer_arm_timer_fd’:
timer.c:48: warning: missing initializer
timer.c:48: warning: (near initialization for ‘its.it_value.tv_nsec’)
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
The old touchpad accel code was clamping touchpad acceleration between
0.2 and 0.4, and on the test devices I have the constant_factor ended up
such that in practice the accel was almost always 0.2, so rather than having
a velocity based acceleration curve, in essence it was just always using an
acceleration of 0.2 .
This commit introduces actual velocity based acceleration based on the
recently added smooth simple acceleration code from filter.c .
Before feeding motion events to filter.c, they first get adjusted for touchpad
resolution. For touchpads where the driver does not provide resolution info,
scale based on the diagonal-size in units instead.
While at it rename tp_init_accel's dispatch parameter from touchpad to tp
to be consistent with all other functions.
Since the acceleration is also used for scrolling also adjust the scroll
start threshold for these changes.
Note that switching to the smooth simple accel code, as an added bonus gives
the tp code an accel profile with a threshold and a speed parameter, which
is exactly what is needed for the upcoming configuration interface support.
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>
Those three are the ones that matter for logging or device identification in
callers, so let's provide them.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
The current code triggers multi-finger tapping even if the finger released was
previously held on the touchpad for a while. For an event sequence of:
1. first finger down
2. first finger move past threshold/wait past timeout
3. second finger down
4. first finger up
The second finger initiates the two-finger tap state, but the button event is
sent when the first finger releases - despite that finger not meeting the
usual tap constraints. This sequence can happen whenever a user swaps fingers.
Add the finger state to the actual touchpoints and update them whenever the
constrains are broken. Then, discard button events if the respective touch
did not meet the conditions.
http://bugs.freedesktop.org/76760
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Once we get beyond the:
if (abs(diff) <= margin)
return center;
test, then diff is either > margin or < -margin, otherwise the test would
have triggered.
So the "return center + diff;" at the end will never be reached, and the
"else if (diff < -margin)" can be turned into a simple "else".
This commit does not just simplify tp_hysteresis, but (arguably more
important) also makes it clearer to the reader what it does.
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>
We have enough API breakage in this release that it's not worth keeping
this around. Every caller must be fixed for all the other stuff anyway,
so drop this too.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
We have enough API breakage in this release that it's not worth keeping
these around. Every caller must be fixed for all the other stuff anyway,
so drop this too.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
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>
And add an example xorg.conf.d .conf file for ignoring these devices under
xorg.
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>
There is no need to loop over the touch points twice.
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>
If the user puts down to fingers to scroll, then changes his mind and
lifts them, without having them moved past the initial scroll threshold in
either direction, then any movement which he has done will cause a spurious
scroll event when the second finger down is lifted first.
The problem is that t->is_pointer was not being set to false in this case,
since that is done in tp_post_twofinger_scroll after checking scroll.state
which never gets set in this scenario.
Instead of changing the order, simply completely remove scroll.state completely
it is a boolean, and everywhere we check for it we also check for the axis bits
in state.direction, so it is not necessary.
Also add a check to ensure there are no spurious motion events.
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>
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>
Inverted order of release from the other test
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
For those whose eyes struggle to focus on 5 zeros in a row, or those just sick
of forgetting one zero and wondering why things don't work.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This test relies on valgrind detecting the leak and use-after-free.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Instead of only allowing one owner keeping a libinput context alive,
make context reference counted, replacing libinput_destroy() with
libinput_unref() while adding another function libinput_ref().
Even though there might not be any current use cases, it doesn't mean we
should hard code this usage model in the API. The old behaviour can be
emulated by never calling libinput_ref() while replacing
libinput_destroy() with libinput_unref().
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
