After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 752271 - 3d9cce79cf causing jumps in scrolling
3d9cce79cf causing jumps in scrolling
Status: RESOLVED FIXED
Product: gtk+
Classification: Platform
Component: Backend: X11
3.16.x
Other Linux
: Normal normal
: ---
Assigned To: gtk-bugs
gtk-bugs
Depends on:
Blocks:
 
 
Reported: 2015-07-11 17:36 UTC by adee.edu
Modified: 2015-07-16 09:47 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
xinput test-xi2 output (6.39 KB, text/plain)
2015-07-14 17:50 UTC, Carlos Garnacho
Details

Description adee.edu 2015-07-11 17:36:23 UTC
Commit 3d9cce79cf16bce335599b3b78b6bd0a04e37560 included in 3.16.5 causes big jumps in mouse scrolling, for the first scroll and sporadically afterwards.
This is weird because it doesn't happen at first right after startup. 
Also it happens with a Trust EVO Advanced mouse, but not with Logitech M705 (or maybe not as fast? did not try for long with that)

Reverting the aforementioned commit makes the problem go away.

To reproduce:
1. Fresh start of machine
2. Open Nautilus or Gedit with contents long enough for scrolling by default
3. scroll down -> big jump (requires multiple tries with Nautilus at least to trigger)
Comment 1 Carlos Garnacho 2015-07-14 17:50:21 UTC
Created attachment 307421 [details]
xinput test-xi2 output

From my investigation, I can only see strange behavior if the synaptics driver is in use, everything works as expected with the libinput x11 input driver.

The attachment contains xinput test-xi2 logs taken with the synaptics driver, the summary of the logs is:
- touchpad (device 14) scroll event, scroll valuators (2 and 3) have some high value
- regular mouse moves in between (device 10), we get SlaveSwitch, and a few motion events
- touchpad keeps scrolling, we get SlaveSwitch again, but scroll valuators made a noticeable jump

This rules out a GTK+ bug, we rather seem to be exposing the bug now. Our previous strategy of eating the first scroll event after SlaveSwitch helped as we forgot the previous value.

A short term fix is using the libinput X11 driver.
Comment 2 Carlos Garnacho 2015-07-14 17:53:12 UTC
*** Bug 752184 has been marked as a duplicate of this bug. ***
Comment 3 AnAkkk 2015-07-14 18:00:08 UTC
As I've said in the other bug report, I do use the libinput x11 driver, and the problem is even worse for me, it is randomly scrolling in the wrong direction.
Comment 4 Carlos Garnacho 2015-07-16 09:47:32 UTC
The affecting commit has been reverted in the end.