To run valgrind properly, we need a couple of arguments passed in so we check
for leaks and don't fail on bits of the stack we don't control. Add a
mesontest setup for this, the lot can now be run by
mesontest --setup=valgrind
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
No internal timeout we have takes longer than 2s, so we can abort if we don't
succeed. This gives us a better backtrace to figure out where we're hanging
than the SIGABRT that check will eventually send us.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Instead of just waiting for events, use a libinput_dispatch() and assume the
event is there when we want it.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This explains the heisenbugs when running the test suite. libevdev gives us
the syspath to the /sys/.../input123 node, not the one for the event node.
The device node path is created based on the sysfs tree, so there's a
window where the device node may not exist yet but we already returned the
device node path.
In litest, we're using a udev monitor to wait until the device is ready for
us, but the path interface only takes a device node path. So what happens is:
* libevdev gives us a syspath for the input node and a device path
* the monitor receives the input node udev device and matches the syspath
* we pass that up to the caller litest_add_device_with_overrides()
which opens the device node and adds it to libinput
* the path interface creates a udev device from the device node, which still
points to the old device node. Things fail because we don't have the device
we expect or it doesn't send events and eventually times out [1].
The errors triggered by this are either odd udev property mismatches or
timeouts because events are never processed.
This race is fixed by simply constructing the actual device node path we
expect from the udev device and waiting for the right device.
[1] We rely on the caller to notify us when to remove the device and thus
silently ignore ENODEV.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Leftovers from an earlier version where we had booleans and more function
nesting in the mix. Fix to return integers, and also rename the function name
accordingly.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
On unreliable tablets (Surface3), always force the lid switch to open when the
paired keyboard is removed. This way the lid can't be stuck in a closed state
when there's nothing attached that can actually trigger that state.
https://bugs.freedesktop.org/show_bug.cgi?id=101100
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
On unreliable LID switches, we might have the LID declared as closed
while it is actually not. We can not wait for the first switch event to setup
the keyboard listener: it will never occur.
https://bugs.freedesktop.org/show_bug.cgi?id=101099
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
What we do not want is libinput to believe the LID is closed while
it's not. But the internal notifier state need to be in sync with the evdev
node, or it's going to be a pain setting the keyboard listener.
But since we don't know if the state is reliable, we track the internal state
separately from the external state so that we can set up the keyboard listener
when the lid is closed, without having libinput actually send lid closed
events for unreliable devices.
https://bugs.freedesktop.org/show_bug.cgi?id=101099
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Only pair if the keyboard is either tagged as internal device.
This has another (unlikely) behaviour change: previously we would override the
paired keyboards with ones that look more accurate (e.g. a usb keyboard paired
before a serial would be unpaired and the serial keyboard takes its place).
Now we assume there can only be one internal keyboard, once we have it we
ignore all others. This shouldn't matter in real life provided the tagging is
correct.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
We have heuristics for detecting whether a keyboard is internal or external,
but in some cases (e.g. Surface 3) these heuristics fail. Add a udev property
that we can apply to these cases so we have something that's reliable.
This will likely eventually become ID_INPUT_KEYBOARD_INTEGRATION as shipped by
systemd, similar to the touchpad property.
https://bugs.freedesktop.org/show_bug.cgi?id=101101
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Having a litest_wait_for_event_of_type() in there causes us to silently
discard anything but the events we're looking for. This is risky, we want to
make sure that if we re-enable the lid that the key events arrive *after* the
lid open event, not before. So let's not paper over those.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
We were checking for the empty queue on the wrong context, the default context
got drained a few lines above.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This was motivated by the need to run the test runner from any directory, so
we need absolute paths to the files we copy.
Unfortunately, we can't get the absolute path from the object returned by
configure_file() and we can't feed that directly into join_paths() either.
So let' make it at least easier to handle: create a configure_file for all the
files we need (so they all end up in builddir/) and simply hardcode the name
for join_paths. Define the lot in config.h, no need to pass compiler flags
around.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This tool will eventually replace the different libinput tools we ship atm
with the various functionalities being commands to the single tool, rather
than multiple tools.
Right now, we still build both tools separately.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Include the source files directly, we'll need per-target compiler flags that
affect different tools differently in the future.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
commit 3925936 introduced changes to container_of, this is hopefully the
last part of it.
In the linux kernel, container_of() takes a type name, and not a
variable. Without this, in some cases it is needed to declare an unused
variable in order to call container_of().
example:
return container_of(dispatch, struct fallback_dispatch, base);
instead of:
struct fallback_dispatch *p;
return container_of(dispatch, p, base);
This introduce also list_first_entry(), a simple wrapper around
container_of() to retrieve the first element of a non empty list. It
allows to simplify list_for_each() and list_for_each_safe().
Signed-off-by: Gabriel Laskar <gabriel@lse.epita.fr>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
python installation does not always lives in /usr/bin, this allows to
use virtualenv for example.
Signed-off-by: Gabriel Laskar <gabriel@lse.epita.fr>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
gcc and clang supports offsetof (defined in stddef.h) as defined by C99
and POSIX.1-2001, use it in container_of.
Signed-off-by: Gabriel Laskar <gabriel@lse.epita.fr>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This is most likely the first error message a developer encounters when
running the test suite and the /run/udev/rules.d directory already exists.
Make it more meaningful than the current generic integer comparison failure.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This function used to be called inside a test run a long time ago but moved to
a pre-setup stage without switching to the more generic litest_abort_msg.
The only error message we got is "check_msg.c:80: No messaging setup".
https://github.com/libcheck/check/issues/18#issuecomment-301217615
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
We never installed the device groups file for the tests, effectively relying
on a system copy to be installed already.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This was originally left outside of the button areas in case users tap in
those zones, but we're getting false tap events in that zone.
On a 100mm touchpad, the edge zone is merely 5mm, it's acceptable to ignore
taps in that area even in the software button. We can revisit this if we see
tap detection failures in the future.
https://bugzilla.redhat.com/show_bug.cgi?id=1415796
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Because otherwise things go boom, but unless you passed -fshort-enums this
shouldn't happen anyway. And gcc's documentation says don't do that. So don't
do that, or we'll scream at you.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Tested-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
../src/libinput.c:56:17: warning: passing an object that undergoes default
argument promotion to 'va_start' has undefined behavior [-Wvarargs]
The enum's size is compiler-defined, so the enum gets promoted to whatever the
compiler chose. That promotion is undefined, so let's use an int here.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Tested-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Fixes a bunch of warnings of the kind
../src/evdev.h:378:32: warning: variable 'f' is uninitialized when used here [-Wuninitialized]
return container_of(dispatch, f, base);
Just typecasting NULL means we can ignore sample but for the type.
https://bugs.freedesktop.org/show_bug.cgi?id=100976
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Tested-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
clang supports __typeof__ which was the only real difference. Not sure any
other compilers matter (that don't support __typeof__)
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Armin Krezović <krezovic.armin@gmail.com>
Tested-by: Armin Krezović <krezovic.armin@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Running through mesontest also runs parse-hwdb through valgrind and
gives us a bunch of leaks that originate within Python somewhere - we don't
care about those.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Running valgrind through mesontest produces coredumps for a lot of tests
(unclear why, the core dump merely shows a call to abort). But even without
mesontest, creating a core dump for each failed test is a bad idea - if one
fails, most likely many others fail and the coredumps quickly fill up the file
system.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
If the event listener is added, then removed again on a lid switch on/off
event, the list is set to null. This can trigger two crashes:
* when the keyboard is removed first, the call to
libinput_device_remove_event_listener() dereferences the null pointer
* when the switch is removed first, the call to device_destroy will find a
remaining event listener and assert
https://bugzilla.redhat.com/show_bug.cgi?id=1440927
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Sequence triggered by the xorg driver, but basically: if the touchpad is
destroyed before the lid switch, the event listener wasn't removed and an
assertion was triggered.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
v2:
- meson 0.40 requirement
- add_project_arguments() instead of add_global_arguments()
- use cc.get_define('static_assert')
- use config.set10 and config.set_quoted instead of manual handling
- more use of join_paths
- use files() for model quirks hwdb check
- update options to all state 'true' as default instead of variations of
'true', 'enabled' and 'yes'
v3:
- drop -Wall -Wextra and -g, let meson set that with warning_level/debug build
- add meson files to EXTRA_DIST
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
