Commit 94b7836456 ("filter: support accelerating high-resolution
scroll wheel events") introduced a regression where high-res scroll
wheel events were incorrectly normalized by DPI. Mice with non-default
DPI (e.g., Logitech G502 at 2400 DPI) had their scroll wheel speed
reduced by the DPI ratio (1000/2400), resulting in 2-3x slower
scrolling.
The "noop" filter functions were actually performing DPI normalization
or applying a constant acceleration factor, which is appropriate for
button scrolling but incorrect for scroll wheels that have their own
units.
Add a filter_scroll_type enum (CONTINOUS, WHEEL, FINGER to match the
public events) passed through the filter_scroll interface. Update all
filter implementations to skip acceleration and normalization for wheel
events while maintaining existing behavior for button scrolling and
touchpad scrolling.
The custom acceleration profile continues to accelerate high-res wheel
events as designed.
Fixes: 94b7836456 ("filter: support accelerating high-resolution scroll wheel events")
Closes: #1212
Signed-off-by: Yinon Burgansky <yinonburgansky@gmail.com>
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1363>
Adds a dedicated scroll movement type to the custom acceleration profile.
Supported by physical mouse and touchpad.
Other profiles remain the same by using the same unaccelerated filter for the scroll filter.
Signed-off-by: Yinon Burgansky <51504-Yinon@users.noreply.gitlab.freedesktop.org>
The filter vs const filter is supposed to be for accelerated vs
non-accelerated motion (e.g. pointer motion vs scrolling) - in both
cases the returned value is supposed to be in the same coordinate
system, just once with an extra accel factor applied.
This was broken in the flat and low-dpi profiles: in both of those the
accelerated filter does *not* normalize, it merely applies the fixed/adaptive factor.
The constant filter normalized however. The result was that on e.g. a
5000dpi mouse the constant motion was 5 times slower than the
accelerated motion, even with a factor of 1.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This is a leftover from when some of the filter code was shared between
pointer acceleration methods (pre v1.11 or so). Now these functions are
duplicated across files, so both the names and what they do isn't
necessarily reflective anymore.
Let's drop one layer of indirection to make the code a bit easier to
understand.
No functional changes.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
libinput applies averaging to the velocity of most pointer devices. Averaging
the velocity makes the motion look smooth and may be of benefit to bad input
devices. For good devices, however, it comes at the unfortunate price of
decreased accuaracy.
This change turns velocity averaging off by default (sets ntrackers to 2 instead
of 16) and allows for it to be turned back on via a quirk, for bad devices which
require it.
Yeah, it's duplication. But this way it's also separation and we can't
accidentally use the wrong struct.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Plenty of duplication there from the normal filter.c, but that also makes it
less likely to break if we adjust the other one.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
