In commit 9a9466b6a9 ("evdev: discard any frame with EV_SYN SYN_REPORT 1")
all frames with a SYN_REPORT 1 were discarded on the assumption of those
being key repeat frames. Unfortunately the kernel uses the same sequence
to simply mark *any* injected/emulated event, regardless of the cause. Key repeat
events are merely the most numerous ones but as shown in commit
7140f13d82 ("evdev: track KEY_SYSRQ frames and pass them even as repeat frames")
Alt+PrintScreen is also an emulated event.
Issue #1165 details another case: keyboards with n-key rollover can
exceed the kernel-internal event buffer, typically 8 events for devices
without EV_REL/EV_ABS. Those events will be broken up by the kernel into
multiple frames - once nevents == buffer_size the current state is
flushed as SYN_REPORT 1 frame. Then, if any more events are pending
those are flushed as SYN_REPORT 0 frame. In the case of exactly 8
events, the second frame is never present, so we cannot easily detect if
another one is coming.
Issue #1145 only affects us in the touchpad code, the rest of the
backends seem to (so far) be fine. So let's move the discarding of
SYN_REPORT 1 to the touchpad backend and leave the rest of the code
as-is.
This effectively
Reverts: 7140f13d82 ("evdev: track KEY_SYSRQ frames and pass them even as repeat frames")
Reverts: 9a9466b6a9 ("evdev: discard any frame with EV_SYN SYN_REPORT 1")
Closes#1165
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1282>
When the kernel inserts a repeat frame it does that with EV_KEY code
value of 2 and the frame itself is a SYN_REPORT with value 1. Nothing in
libinput wants those repeat values, so let's discard them here before
anything tries to process them.
This inserted frame causes bugs on touchpads with EV_REP (rare enough)
because while the key event itself is dropped, the timestamp of the
frame still causes the next real frame's delta time to shorten,
resulting in wrong acceleration values.
Closes#1149
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1255>
This device was added before high-res scroll events existing in the
kernel and it's used in a test to verify that a device that has
ABS_MT_POSITION_X but not _Y doesn't get automatically ignored.
Said test (device_quirks_no_abs_mt_y) uses a wheel event to verify that
we do get events from this device.
Since then we've long had kernels that support hi-res scrolling and the
kernel takes care of those events for us. So let's update the device
description and the events we send to include the high-resolution
events. That doesn't change the validity of the test but stops it from
becoming a false positive.
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1251>
It's too much effort fighting clang-format for these snippets which
all don't really do much anyway but are important to be read easily.
Let's categorically disable all formatting in the test collections and
move on.
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1246>
touchpad_move_after_touch contains support for up to 5 fingers but was only run
with 2..4 fingers.
touchpad_multitap and several others using the same parameter definition were
long ago tested with 3..7 taps. In 8f92b091 this was reduced to 3..4 for CI
performance, while the commit message indicates 3..5 were intended.
The common theme is that the upper bound of `struct range` is interpreted
as exclusive, while some uses assumed it would be inclusive.
There are relatively recent helper functions range_init_inclusive and
range_init_exclusive (since 817dc423) to avoid this trap. Use them on the
remaining two ranged tests.
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1142>
Slightly less efficient but easier to read and it's not possible to
accidentally provide the wrong length. Plus it handles null pointers
correctly so get to skip the checks (which weren't needed for strneq()
either, but still).
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1121>
This replaces check. The code is a copy of pwtest which I wrote years
ago for pipewire but adjusted for us here the last few days.
There are a few advantages over check:
- Ability to SKIP tests or mark them as NOT_APPLICABLE, the latter
of which is used for early checks if a device doesn't meet
requirements.
- it captures stdout/stderr separately
- colors!
- YAML output format makes it a lot easier to read the results and
eventually parse them for e.g. "restart failed tests"
Less abstraction: we set up the tests, pass them to the runner and run
them with the given number of forks. This is an improvement over before
where we forked into N test suites which each called check which then
forked again. Since we're now keeping track of those processes
ourselves we can also write tests that are expected to fail with
signals.
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1067>
This is the first step in switching away from the check framework.
Our litest macros already do almost exactly the same anyway so most of
this is a simple sed with a few compiler fixes where things mismatch
(nonnull -> notnull) and (_tol -> _epsilon).
This now generates a whole bunch of integer mismatch warnings: check
casts everything to intmax_t whereas we use typeof, so lots of warnings
especially for enums.
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1059>
Wraps libinput_dispatch() with a location which will make things a bit
easier to track. Output (in --verbose) is something like:
gestures_swipe_3fg_unaccel_fn():1346 - dispatching
Which makes it easier to associate the various calls to libinput
dispatch with the other output from libinput.
This patch switches all uses of libinput_dispatch() in test cases over
but not the litest functions that may call dispatch too. Remains to be
seen if that is necessary.
Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1048>
Some touchpads, notably those on the Dell XPS 15 9500, are prone to registering
touchpad clicks when the case is sufficiently flexed. Ignore these by
disregarding any clicks that are registered without touchpad touch.
Signed-off-by: Rob Glossop <robgssp@gmail.com>
This switches the quirk from AttrEventCodeEnable/Disable to just
AttrEventCode with a +/- prefix for each entry.
This switches the quirk from AttrInputPropEnable/Disable to just
AttrInputProp with a +/- prefix for each entry.
Previously, both event codes and input props would only apply the
last-matching section entry for a device. Furthermore, an earlier Disable entry
would take precedence over a later Enable entry. For example, a set of
sections with these lines *should* enable left, right and middle:
[first]
AttrEventCodeEnable=BTN_LEFT;BTN_RIGHT;BTN_MIDDLE
[second]
AttrEventCodeDisable=BTN_RIGHT
[third]
AttrEventCodeEnable=BTN_LEFT;BTN_RIGHT;
Alas: the first line was effectively ignored (quirks only returned the
last-matching one, i.e. the one from "third"). And due to implementation
details in evdev.c, the Disable attribute was processed after Enable,
i.e. the device was enabled for left + right and then disabled for
right. As a result, the device only had BTN_LEFT enabled.
Fix this by changing the attribute to carry both enable/disable
information and merging the commands together.
Internally, all quirks matching a device are simply ref'd into an array
in the struct quirks. The applied value is simply the last entry in the
array corresponding to our quirk.
For AttrEventCode and AttrInputProp instead do this:
- switch them to a tuple with the code as first entry and a boolean
enable/disable as second entry
- if the struct quirk already has an entry for either, append the more
recent one to the existing entry (instead of creating a new entry in
the array). This way we have all entries that match and in-order of
precedence - i.e. we can process them left-to-right to end up
with the right state.
Fixes: https://gitlab.freedesktop.org/libinput/libinput/-/issues/821
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
All cases we have in our code base have an otherwise unused variable to
loop through the array. Let's auto-declare this as part of the loop.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Starting with kernel v5.0 two new axes are available for high-resolution wheel
scrolling: REL_WHEEL_HI_RES and REL_HWHEEL_HI_RES. Both axes send data in
fractions of 120 where each multiple of 120 amounts to one logical scroll
event. Fractions of 120 indicate a wheel movement less than one detent.
This commit adds a new API for scroll events. Three new event types that encode
the axis source in the event type name and a new API to get a normalized-to-120
value that also used by Windows and the kernel (each multiple of 120 represents
a logical scroll click).
This addresses a main shortcoming with the existing API - it was unreliable to
calculate the click angle based on the axis value+discrete events and thus any
caller using the axis value alone would be left with some ambiguity. With the
v120 API it's now possible to (usually) calculate the click angle, but more
importantly it provides the simplest hw-independent way of scrolling by a
click or a fraction of a click.
A new event type is required, the only way to integrate the v120 value
otherwise was to start sending events with a discrete value of 0. This
would break existing xf86-input-libinput (divide by zero, fixed in 0.28.2) and
weston (general confusion). mutter, kwin are unaffected.
With the new API, the old POINTER_AXIS event are deprecated - callers should use
the new API where available and discard any POINTER_AXIS events.
Notable: REL_WHEEL/REL_HWHEEL are emulated by the kernel but there's no
guarantee that they'll come every accumulated 120 values, e.g. Logitech mice
often send events that don't add up to 120 per detent.
We use the kernel's wheel click emulation instead of doing our own.
libinput guarantees high-resolution events even on pre-5.0 kernels.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: José Expósito <jose.exposito89@gmail.com>
The group names are forced by check (they are called suite names there) but
for our test suite they provide very little benefit. Much easier to just
use the filename a test is in as group name.
This removes the pure substring match for --filter-group, it's now fnmatch
only. group names are short enough that the typing isn't an issue and we don't
want to run tests twice (e.g. 'pad' is also in 'touchpad').
This patch caused #574 until it got fixed in d838e3a3a4
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
The latter requires libevdev 1.10 but since that'll take a while to filter
into our various CI systems, let's make it conditional.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Currently unused, but let's get this in because we may need this very soon for
broken tablets.
Enabling EV_ABS axes requires an absinfo struct - we default to a simple 0-1
axis range for those as the most generic option. Anything more custom will
need more custom treatment when we need it.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
SW_MAX changed and the device_capability_nocaps_ignored test will fail on
older kernels. Change that test to use some other unhandled-by-libinput switch
code instead.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Symmetrical to litest_create_context(), this allows us to store special data
in that context that we have access to during the tests.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This is merely the simple support that we use in the fallback backend as
well. It doesn't interact with touch arbitration directly but it'll be
good enough for the default use-case.
Fixes#476
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This device looks similar to a MT device on the kernel side, but it's not a
MT device and it's not quite a tablet either. It uses slots to track up to 4
totems off the same device and the only hint that it's not a MT device is that
it sends ABS_MT_TOOL_TYPE / MT_TOOL_DIAL.
udev thinks it's a touchscreen and a tablet but we currently init those
devices as touchscreen (because all wacom tablet touch devices are udev
tablets+tochscreens). So we need a quirk to hook onto this device.
And we use a completely separate dispatch implementation, because adding the
behavior to the tablet interface requires so many exceptions that it's easier
to just add a separate dispatch interface.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This leaves a bug open, on a Lenovo T440 generation touchpad with top software
buttons, the button will not be leased correctly. This is caused by
device->is_suspended=true by the time we try to clear the state and the
button events thus getting filtered.
This used to affect all touchpads, this patch just moves it so it only affects
the T440-like devices now.
Fixes#233
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
When we disable the touch device, any existing touches should be cancelled,
not just released.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This forces events for every ~10ms now. If we want a slower movement, we need
more steps - just like a real touchpad does it.
Cocinelle spatch files were variants of:
@@
expression A, B, C, D, E, F, G, H, I, J, K;
@@
- litest_touch_move_two_touches(A, B, C, D, E, F, G, H, I)
+ litest_touch_move_two_touches(A, B, C, D, E, F, G, H)
The only test that needed a real fix was touchpad_no_palm_detect_2fg_scroll,
it used 12ms before, now it's using 10ms so on the bcm5974 touchpad the second
finger was a speed-thumb. Increasing the events and thus slowing down the
pointer means it's a normal finger and the test succeeds again.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This is via a simple search & replace. Later auditing is needed to switch
clicks that should not be debounced (e.g. touchpads) back to a non-debounced
version.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
