Created attachment 136062 [details]
evemu-record of two-finger clickfinger when a palm is on the touchpad
Two-finger clickfinger is recognised as middle-click instead of right-click when a palm is rested on the touchpad, similar to bug103210 in some way.
One-finger and Three-finger clickfingers, however, behave correctly.
I've replayed the recording on my non-MacBook machine with "libinput debug-events --set-click-method=clickfinger", and it shows BTN_MIDDLE events. Hopefully we won't encounter the problem in bug103809 again.
looks to be an easier problem. tp_clickfinger_set_button() just takes nfingers_down which includes palm touches. Hence the mis-count.
Created attachment 136270 [details] [review]
Still need to write test cases for it but I think this one should be it?
Created attachment 136309 [details]
evemu-record of two-finger clickfinger when a palm is on the touchpad v.2
I've tested the patch, and it works for the most part. Though there are now miscount of fingers of a different kind.
In this new recording, I record a series of two-finger clicks on different places of the touchpad while my palm is on. Some of the two-finger clicks appear to be recognised as left clicks instead.
Created attachment 136310 [details]
evemu-record of three-finger clickfinger when a palm is on the touchpad
There are occasional miscounts with three-finger clicks as well. Sometimes they are recognised as right clicks, and more rarely as left clicks.
(In reply to Peter Y. Chuang from comment #3)
> In this new recording, I record a series of two-finger clicks on different
> places of the touchpad while my palm is on. Some of the two-finger clicks
> appear to be recognised as left clicks instead.
it appears that in all the left-click cases a second finger is also detected as palm, based on the touch size. so libinput thinks it's two palms + 1 finger, not one palm + 2 fingers. I suspect the same is true for the other attachement.
so, how are we going to proceed here? Judging by the last evemu-record it's a bit out of our hands now because if we can't rely on the touch size then we have general problems. But we seem to have gone down from always broken to occasionally broken, right?
Sorry for the late reply. I guess I will take it for now.
Closing based on comment #7
(In reply to Peter Hutterer from comment #8)
> Closing based on comment #7
Quick question: is the patch in git master?
oops, no, apparently not. sorry about that.
ok, this time:
Author: Peter Hutterer <>
Date: Tue Dec 19 14:54:53 2017 +1000
touchpad: ignore palm touches when handling clickfingers