roadrunner2
commented
7 years ago Interesting - I can confirm that behaviour exactly (I've been using libinput master too recently). However, I think what evemu-record is showing is correct: for a palm I see events start only when palm size threshold is reached (or more precisely: I see a few events, and then it stops as it recognizes that this is a palm, and doesn't start showing events until the pressure is increased to the point where the palm size threshold is reached), and on releasing they stop as soon as it goes under the threshold again. So I think the pointer-is-stuck is a bug in libinput.
peterychuang
This problem was first discovered when I was trying to make
libinputcapable of rejecting palm touches based on touch-sizes (see this branch and this thread ).When I release my palm very slowly from the touchpad, often times the machine fails to respond to the next finger input.
evemu-recordreveals that when a palm is released slowly, it sometimes stops registering anything in the middle of the slow release. IfABS_MT_TOUCH_MAJORis above the touch-size threshold the instanceevemu-recordis stuck, the cursor will fail to move on the next finger input even thoughevemu-recordwill register the said input. The cursor will only start moving again on the next touch.