For touchscreens we always send raw touch events to the compositor, and the
compositor or application toolkits do gesture recognition. This makes sense
because on a touchscreen which window / widget the touches are over is
important context to know to interpret gestures.
On touchpads however we never send raw events since a touchpad is an absolute
device which primary function is to send pointer motion delta-s, so we always
need to do processing (and a lot of it) on the raw events.
Moreover there is nothing underneath the finger which influences how to
interpret gestures, and there is a lot of touchpad and libinput configuration
specific context necessary for gesture recognition. E.g. is this a clickpad,
and if so are softbuttons or clickfinger used? What is the size of the
softbuttons? Is this a true multi-touch touchpad or a semi multi-touch touchpad
which only gives us a bounding box enclosing the fingers? Etc.
So for touchpads it is better to do gesture processing in libinput, this commit
adds an initial implementation of a Gesture event API which only supports swipe
gestures, other gestures will be added later following the same model wrt,
having clear start and stop events and the number of fingers involved being
fixed once a gesture sequence starts.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Jason Gerecke <jason.gerecke@wacom.com>
When the condition with continue was hit, syspath was still compared in the
loop condition, leading to crashes when strcmp()-ing a random string.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Drop the busy loop we had waiting for an event to appear and just call poll on
the libinput fd.
This actually makes the tests more correct, if we now time out where we didn't
before it means we're not setting the timers correctly.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
3s is too easy to hit on congested systems. 10s is overkill, but it's still
better to pass a test late than having to restart the whole test-suite again.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Set up a udev_monitor before each device creation and wait for the monitor to
notify us of the newly created device. This should take the place of the
various sleep loops we currently have sprinkled around the code and provide a
reliability when testing.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Hitting this limit too often on congested VMs, so let's increase the sleep
sleep (so the system can get done what needs to be done) and get the whole
timeout from 600ms to 2000ms.
Note: if we really hit 2000ms we may still fail on some tests with the check's
default 3 second timeout.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Quite similar to wacom tablets but the evdev protocol differs - no serials for
example and only two tools. The device has a wheel, but it's not apparently
part of the stylus like on the wacom tablets. And it has a bunch of keys.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
To quote Bryce Harrington from [1]:
"MIT has released software under several slightly different licenses,
including the old 'X11 License' or 'MIT License'. Some code under this
license was in fact included in X.org's Xserver in the past. However,
X.org now prefers the MIT Expat License as the standard (which,
confusingly, is also referred to as the 'MIT License'). See
http://cgit.freedesktop.org/xorg/xserver/tree/COPYING
When Wayland started, it was Kristian Høgsberg's intent to license it
compatibly with X.org. "I wanted Wayland to be usable (license-wise)
whereever X was usable." But, the text of the older X11 License was
taken for Wayland, rather than X11's current standard. This patch
corrects this by swapping in the intended text."
libinput is a fork of weston and thus inherited the original license intent
and the license boilerplate itself.
See this thread on wayland-devel here for a discussion:
http://lists.freedesktop.org/archives/wayland-devel/2015-May/022301.html
[1] http://lists.freedesktop.org/archives/wayland-devel/2015-June/022552.html
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Acked-by: Jonas Ådahl <jadahl@gmail.com>
We can't rely on the system having these files installed, at least not in the
latest version that we'd like.
Copy them over from the source directory into the /run/ and /etc/ directories
for each test and update udev and the hwdb. This ensures the tags we set in
the hwdb file are always set, regardless of the system configuration.
Note that the /run/udev/* files need to have a different filename to the ones
we ship to avoid getting overridden by local configuration.
systemd does not have support for /run/udev/hwdb.d [1]. So our hwdb.d file
is in /etc/udev/hwdb.d instead and marked them with a REMOVEME and a comment
that if that file is left after the tests, it should be removed by the user.
[1] https://github.com/systemd/systemd/issues/127
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Previous expansion had side-effects when litest_log was called in an if
condition without {}
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Added code to check for errors in getcwd() and system() that
were previously ignored and silently dropped.
Signed-off-by: Jon A. Cruz <jonc@osg.samsung.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
If asprintf fails for any reason, the contents of the pointer
are undefined. While some platforms set it to NULL, there is no
guarantee that all will.
This change adds a simple wrapper to ensure proper NULL results
on failure.
Signed-off-by: Jon A. Cruz <jonc@osg.samsung.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Added LIBINPUT_PRINTF attribute and the required declaration for it.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Wait after deleting a device so udev can catch up with everything and the
various hooks to make sure it's happy with any newly created devices after
this.
The sleep is in the delete path to also cover the tests where we manually
create uinput devices rather than using the litest hooks.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
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>
On slower machines, e.g. VMs, udev isn't fast enough to get the properties set
up by the time we're trying to get the device going. This fails when we try to
add the device with libinput_path_add_device().
We know that all litest devices will have ID_INPUT set, so check for that
before we continue.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Well, the patch looks like the HAVE_LIBUNWIND bit was moved down, which is the
same thing.
litest_log and litest_vlog are called from independent paths, we have a
compiler error otherwise.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Was added to other places, missing from here so some tests passed the filter
despite not being selected by it.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
The litest-selftest has its own main method and compiles litest.c with special
flags. Use that to ifdef out the litest.c main function, and inline the
litest_run/litest_parse_args functions so gcc doesn't complain about unused
functions.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Similar to the CK_RUN_CASE environment variable, but it does support
fnmatch()-style wildcards, e.g.
./test/test-touchpad --filter-device="synaptics*"
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Complementary to CK_RUN_SUITE and CK_RUN_CASE, this filters on actual test
function names with a simple fnmatch.
./test/test-touchpad --filter-test="*1fg_tap*"
Most of this patch is renaming litest_add_* to _litest_add_* so we can use the
macros to get at the function names.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
--verbose only works when running a specific test, sometimes we need make
check to be more verbose. Set this by default for make check, the log becomes
a lot more expressive that way
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
libunwind gives us a file and an address and usually a function name. Beyond
that, it's mostly guessing.
Fork off addr2line to resolve the addresses that libunwind gives us, if we
succeed we get a backtrace like this:
Backtrace:
0: litest_fail_comparison_int() (./test/litest.c:268)
1: disable_button_scrolling() (./test/pointer.c:115)
2: middlebutton_doubleclick() (./test/pointer.c:991)
3: /lib64/libcheck.so.0 (srunner_run+0x7f5) [0x7f6c12d8c025]
4: litest_run() (./test/litest.c:689)
5: main() (./test/pointer.c:1280)
6: /lib64/libc.so.6 (__libc_start_main+0xf0) [0x7f6c11a73790]
7: ./test/test-pointer (_start+0x29) [0x403d99]
8: ? (?+0x29) [0x29]
Note: I intentionally swapped function/file name in the output to make it
easier to spot which one is fully resolved and which one is the basic
libunwind output.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
The check unit test framework isn't particularly suited to having ck_assert*
calls in helper functions. A failed assertion in a helper function or the
litest framework merely gives us a the failed line in litest.c.
which doesn't tell us which test actually failed.
Add a backtracing facility with litest_backtrace(). And since this requires
wrapping all ck_assert macros with litest_assert() this patch ended up
replacing/duplicating a bunch of ck_assert_* bits. So rather than
ck_assert_int_eq() we now use litest_assert_int_eq(), etc. in the litest
framework itself.
The int comparison macros are more type-safe than ck_assert()'s macros which
just cast all the ints to intmax_t.
Backtrace is spewed to stderr, which is good enough for debugging. Example
backtrace:
FAILED COMPARISON: status != expected
Resolved to: 0 != 0
in disable_button_scrolling() (pointer.c:115)
Backtrace:
0: ./test/test-pointer (litest_fail_comparison_int+0xab) [0x40973b]
1: ./test/test-pointer (disable_button_scrolling+0x174) [0x40421b]
2: ./test/test-pointer (middlebutton_middleclick+0x40) [0x40829c]
3: /lib64/libcheck.so.0 (srunner_run+0x7f5) [0x7f0e8a277025]
4: ./test/test-pointer (litest_run+0x107) [0x40a42b]
5: ./test/test-pointer (main+0x2fa) [0x4090e7]
6: /lib64/libc.so.6 (__libc_start_main+0xf0) [0x7f0e88f5e790]
7: ./test/test-pointer (_start+0x29) [0x403ce9]
8: ? (?+0x29) [0x29]
litest_backtrace() itself is copied from xserver/os/backtrace.c which git
blame attributes to Marcin.
CC: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
