Otherwise poll() keeps on triggering on our fd until udev gets around to
notifying the server that the device is in some state of deadness. Meanwhile,
the input thread is busy filling the log up with "Read error" messages. Great
way to test file systems and their capacity to handle out of space scenarios
but otherwise a bit pointless. Those FS developers should write their own
tests instead of relying on the synaptics driver!
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Adam Jackson <ajax@redhat.com>
hw.x and the motion history are integers so our deltas are always integers.
It's a bit pointless to split them into the fractional and integral part.
obsolete since defc1d008e or so
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
If opening the fd fails we still need to fail the device. This is particularly
the case when a device disappears before we can open it - the current code
wouldn't exit but instead switch to auto-probe touchpad devices on the system.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
When ignoring modifiers, ensure the touchpad is enabled once a modifier
key is pressed disregarding any previous key press that caused the
touchpad to be disabled.
Signed-off-by: Anton Lindqvist <anton.lindqvist@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Bump up the synaptics driver to 70, so it get's preferred over libinput, which
was dropped down to 60. The synaptics driver is more of a leaf package
than libinput (which covers a multitude of device types) and can be removed by
default. When specifically installed by the user, the synaptics driver should
override the system default.
Similar to what was done for wacom configuration file.
https://bugzilla.suse.com/show_bug.cgi?id=979554
Signed-off-by: Stefan Dirsch <sndirsch@suse.de>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This was a bad idea. No-one seems to use this and it gives us little benefits.
To even get this feature a number of other features need to be turned off
(like two-finger scrolling and tapping). Many of these are enabled by default,
if they are disabled a client stack with full touchpad gesture support could
in theory support true touchpad gestures. This has never happened.
Drop the touch events from the synaptics driver. This allows us to switch the
touchpad fully over to look like a relative device, thus also removing the
bug that changes the touchpad speed whenever a monitor is added/removed in.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
An MT device without X/Y is not a touchpad. And neither are fake MT devices.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This came up as a kernel bug, but it's valid to create uinput devices with a
min == max range for x/y. Technically valid, but effectively useless, so catch
it, complain and hobble on along.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This reverts commit 064445364b.
The Lenovo *50 series, including the X1 Carbon 3rd always require multiple
kernel patches to enable the touchpad buttons. This patch in synaptics only
addresses the re-routing of the top buttons.
The final iteration of the kernel patches also route the trackpoint buttons
through the trackpoint device, rendering this patch unnecessary. These patches
are queued for 4.0.
See kernel patch series up to commit cdd9dc195916ef5644cfac079094c3c1d1616e4c
Author: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Date: Sun Mar 8 22:35:41 2015 -0700
Input: synaptics - re-route tracksticks buttons on the Lenovo 2015 series
Currently in Dmitry's for-linus branch.
Distributions running older kernels or the kernel stable series which has
partial backports of the above patch series are encouraged to leave the
0644453 commit in and undo this revert.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Some applications ignore the second tap of double taps because of the
lack of a delay between the button down and button up events.
Prevent this by replacing the transition from TS_2B to TS_START with a
transition from TS_2B to TS_SINGLETAP that emits only a button down
event. The button up event will be emitted when transitioning from
TS_SINGLETAP to TS_START.
In addition, decrease the default value of MaxDoubleTapTime from 180 ms
to 100 ms in order to make double taps faster.
Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This device has the trackpoint buttons wired up to the touchpad to send BTN_0,
BTN_1 and BTN_2 for left, right, middle. This conflicts with previous
touchpads that used those event codes for dedicated scroll buttons.
Add an option HasTrackpointButtons that can be set via a xorg.conf.d
snippets. This option is not intended as a user-set option, rather
we expect distributions to ship some conglomerate of udev/hal rules with
xorg.conf snippets that take effect.
If the option is set, we look at the three affected buttons at the beginning
of HandleState and send button events immediately for them. The HW state is
reset to neutral and other processing continues. This saves us from having to
synchronize these buttons with software buttons (also present on this device),
tapping, etc.
Since the buttons are physically different and (mentally) associated with the
trackpoint device we also don't need to worry about having finger motion event
correctly synced up with the button presses - it's acceptable to send the
presses before the motion events.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Tested-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
The palm detection relies on both the width and the pressure.
a897147be0 ("Use ABS_MT events for the palm detection when supported")
assumed that all the touch devices can report both ABS_MT_TOUCH_MAJOR
and ABS_MT_PRESSURE, but this is not necessarily true. This assumption
could hence break the palm detection when at least one of the mentioned
events is not declared but both ABS_TOOL_WIDTH and ABS_PRESSURE are
reported.
Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Use ABS_MT_TOUCH_MAJOR and ABS_MT_PRESSURE instead of ABS_TOOL_WIDTH
and ABS_PRESSURE when supported so that the pressure and the width of
all the fingers is taken into account for the palm detection.
This also fixes the palm detection for those touchpads for which the
kernel only sends ABS_MT_TOUCH_MAJOR and not ABS_TOOL_WIDTH.
Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Touchpads are limited by a fixed sampling rate (usually 80Hz). Some finger
changes may happen too fast for this sampling rate, resulting in two distinct
event sequences:
* finger 1 up and finger 2 down in the same EV_SYN frame. Synaptics sees one
finger down before and after and the changed coordinates
* finger 1 up and finger 2 down _between_ two EV_SYN frames. Synaptics sees one
touchpoint move from f1 position to f2 position.
That move causes a large cursor jump. The former could be solved (with
difficulty) by adding fake EV_SYN handling after releasing touchpoints but
that won't fix the latter case.
So as a solution for now limit the finger movement to 20mm per event.
Tests on a T440 and an x220 showed that this is just above what a reasonable
finger movement would trigger. If a movement is greater than that limit, reset
it to 0/0.
On devices without resolution, use 0.25 of the touchpad's diagonal instead.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Default resolution is 1, don't allow setting 0 to avoid divisions by 0 or
just general weirdness.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
open_slots holds the slot index, resetting it to 0 is a bad idea. And make
sure that we do reset after DEVICE_INIT. We already do so on DEVICE_CLOSE, but
after the first DEVICE_ON the data could still be random.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
xf86-input-synaptics-1.8.0/src/synaptics.c:498: var_compare_op: Comparing
"end_str" to null implies that "end_str" might be null.
end_str can't be null, so no need for this check and no need to get Coverity
all confused.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Default on evdev devices is CLOCK_REALTIME. If that clock falls behind the
server's CLOCK_MONOTONIC, motion after a clickpad click may be delayed by the
difference in the clocks.
In detail:
When the timer func is triggered, GetTimeInMillis() which is CLOCK_MONOTONIC,
is stored as hwState->millis. The eventcomm backend uses struct
input_event time (CLOCK_REALTIME).
When we read events from the device, if the evdev time is less than the server
time, the fix for (#48777) sets the current event time to hwState->millis.
Until the evdev time overtakes that stored time, all events have the
hwState->millis time.
If during that time a clickpad triggers a physical click,
clickpad_click_millis is set to hwState->millis + the ignore-motion timeout.
Thus, all motion is ignored until the event time overtakes that stored
time.
The whole issue is further enhanced by us unconditionally setting the timer
func if we get any events, which is a separate issue anyway.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
In file included from /usr/include/string.h:634:0,
from /usr/include/xorg/os.h:53,
from /usr/include/xorg/misc.h:115,
from /usr/include/xorg/xf86str.h:37,
from /usr/include/xorg/xf86Xinput.h:54,
from synproto.h:36,
from synproto.c:24:
/usr/include/xorg/os.h:579:1: error: expected identifier or '(' before '__extension__'
strndup(const char *str, size_t n);
See http://lists.freedesktop.org/archives/xorg-devel/2014-July/043070.html
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
When two fingers are used, the coordinates of only one of them is taken into
account. This can lead to sudden variations of the absolute coordinates when
two-fingers taps are performed if the finger considered changes.
Take into account coordinates variations to prevent unwanted taps only if
the number of fingers doesn't change.
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Per-device logging functions don't interfere with other drivers if they also
use libevdev, so use those instead the global log handler if available. If not
available, drop libevdev logging, I don't want to maintain the ifdef mess and
the logging doesn't give us _that_ much benefit.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
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 does not seem to need this, presumably the
haptic feedback of the little dots sticking out from the surface make hitting
the button easier. Its size is increased to 15% anyway, for simplicity and
because a sample set of 3 is too small to be definitive about this.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
When we required a grab on the device, this was a shortcut so we didn't have
to query the device only to realise we can't read events off it anyway. Now
that we don't actually grab the device by default, this is unnecessary.
Something else may have a temporary grab on the device during init, in which
case we just continue as usual and read events if and when they become
available.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
