Bug 103594

I forgot to mention that my DE is GNOME (which is default for Ubuntu 17.10).

Please run sudo evemu-record and record one sequence that has bad cursor movement, *attach* the output here. 5-6 seconds will be enough, no need for anything longer.

Created attachment 135293 [details]
evemu-record log

Peter, thanks for your reaction.

I've attached my log. 

I'm not sure if that info will help, but I've noticed that jumps occur only on diagonal movements. If I move cursor strictly horizontally or striclty vertically (any direction) - cursor movement is perfect. But if I move both axes, here comes jumps.

I made three diagonal movements:
1. bottom left to top right
2. top left to bottom right
3. top right to bottom left

Cursor moved along screen edge 
1. upper edge
2. right edge
3. bottom edge

Jumps occured in each case.

If you need anything else, I'm at your service.

Hello,

I stumbled upon this bug; I noticed the same behaviour earlier this year.

You can find a short thread here: https://bbs.archlinux.org/viewtopic.php?pid=1731889#p1731889


I'm pretty sure it is no libinput-related issue but rather a kernel issue, since I was able to reproduce the issue only in Kernel 4.11, but not in 4.10.


At some point I got stuck, but I am pretty sure that the behaviour got intruced through some of the changes in this commit: http://lkml.iu.edu/hypermail/linux/kernel/1702.2/04627.html

I want to point out that the touchpad Synaptics TM3242-001 is in general not working well. In fact, I am using Peter Hutterer's "hack" (?) LIBINPUT_MODEL_HP_STREAM11_TOUCHPAD to enforce the option clickfinger on this touchpad, which it is obviously capable of, but is not being advertised to the kernel.

I would greatly appreciate any hints on where to submit this - is this a case for the kernel bug tracker?

FYI - I opened a kernel bug for this: https://bugzilla.kernel.org/show_bug.cgi?id=197885

(In reply to Max S from comment #5)

Are you sure it is not libinput, but kernel?

My kernel version is 
4.13.0-16-generic

I have described problems on Ubuntu 17.10 Wayland session. 
BUT on X.org session touchpad works perfect. Almost like my Macbook's touchpad. 

The difference between session is:
Wayland uses libinput
X.org uses synclient

Both sessions use same kernel.

The fixes for bug 98839 should make things a bit better, at least for small movements.

(In reply to Pavel from comment #7)

I have 2 Xiaomi 12.5 here. They already have kernel 4.14 running.

I am typing this now after switching to kernel 4.10.17-2-MANJARO
The problem does definitely not occur! This is reproducable to me only by switching kernels.

By the way, I use X.org.

(In reply to Max S from comment #9)

I don't think your bug is connected as 
   a) My kernel has higher version and 
   b) I don't have problems with X.org

Have you tried more modern kernel version? Maybe bugs you met are already fixed there.

(In reply to Peter Hutterer from comment #8)
> The fixes for bug 98839 should make things a bit better, at least for small
> movements.

Great. Thanks for update!

I'm new to bug reports and I have few questions. 

How do I know when those fixes will be available for end users like me?
How can I know version that will include the fix? 
Should I join mailing list of that bug to get emailed about that? Or I just wait patiently?

Thanks a lot.

you're tracking an upstream bug here, so once the bug is resolved here it means it's available in git. depending on how intrusive the fix is it may go into the next stable relase (atm the 1.9.x series) or you may have to wait for the next major release (atm 1.10). Then you have to wait for your distribution to ship updated packages.

Your distribution may decide to cherry-pick a patch independently and ship it, but that is up to them and you'd have to bring it to their attention in the respective distribution bugtracker.

You might also try out the patch in https://bugs.freedesktop.org/show_bug.cgi?id=101139 to see if that improves things for you. it feels drastically better for me.

Needinfo for too long, closing