Hello, Libinput 1.10.x branch introduced a regression regarding three-finger click on my Elantech touchpad. Sporadically, any middle click attempt is ineffective. When that happens, the following errors get logged: (EE) event1 - ETPS/2 Elantech Touchpad: libinput bug: invalid tap event TAP_EVENT_RELEASE in state TAP_STATE_TAPPED (EE) event1 - ETPS/2 Elantech Touchpad: libinput bug: invalid tap event TAP_EVENT_RELEASE in state TAP_STATE_MULTITAP Do you have an idea about what goes wrong?
I'll need an evemu-record that triggers that bug please, thanks. You're on 1.10.3?
(In reply to Peter Hutterer from comment #1) > I'll need an evemu-record that triggers that bug please, thanks. Ok, will do as soon as possible. > You're on 1.10.3? Indeed.
Created attachment 138261 [details] evemu-record capture Hi Peter, I finally found the time to record my touchpad’s behavior. You’ll notice that after each middle-click attempt I slightly move the cursor; that’s because it tends to be “stuck” for a few milliseconds afterward. Cheers, Vincent
what's the output of udevadm info /sys/class/input/eventXX for your event node? I can't reproduce the bug here, not even with 1.10.0 so there may be a device-specific udev property that doesn't get applied on the test host.
Created attachment 138262 [details] udevadm_output
can you reproduce the bug by replaying that event sequence? Still doesn't work for me here, and that's after matching all the udev properties. This doesn't even get into the multitap state, afaict all touches are detected as palms.
Yup, it seems I’m able to reproduce it, not all the time though. Now that you mention palm events, I just saw that a different error message has been logged yesterday: (EE) event3 - ETPS/2 Elantech Touchpad: libinput bug: invalid tap event TAP_EVENT_PALM in state TAP_STATE_TAPPED
I wonder if that is similar to bug 105160 where the state changes happen within the same EV_SYN frame and thus confuse the state machine. If so, that's going to need *a lot* of tests...
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/libinput/libinput/issues/15.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.