Legion Go 2: scroll wheel goes one way only.

[Patola]

package version: inputplumber 0.76.1-1.1 being used on CachyOS handheld.

The scroll wheel works but it goes only one way. Both scrolling up and scrolling down seemingly give the same result on applications and games, seemingly only the scroll down signal is sent either way.

I am not sure how to demonstrate that with text-based utilities and/or logs though. I can record games and applications if that is the case.

The best I could come up with was using libinput:

❯ sudo libinput debug-events
-event3   DEVICE_ADDED                 Video Bus                         seat0 default group1  cap:k
-event2   DEVICE_ADDED                   Legion Controller for Windows  Keyboard seat0 default group2  cap:k
-event4   DEVICE_ADDED                   Legion Controller for Windows  Mouse seat0 default group2  cap:p left scroll-nat scroll-button
-event5   DEVICE_ADDED                   Legion Controller for Windows  Touchpad seat0 default group2  cap:t  size 28x28mm ntouches 1 calib
-event10  DEVICE_ADDED                 Ideapad extra buttons             seat0 default group3  cap:k
-event8   DEVICE_ADDED                 GXTP6933:00 27C6:0133             seat0 default group4  cap:t  size 190x119mm ntouches 10 calib
-event0   DEVICE_ADDED                 Power Button                      seat0 default group5  cap:k
-event1   DEVICE_ADDED                 AT Translated Set 2 keyboard      seat0 default group6  cap:k
-event16  DEVICE_ADDED                 InputPlumber Mouse                seat0 default group7  cap:p left scroll-nat scroll-button
-event17  DEVICE_ADDED                 InputPlumber Keyboard             seat0 default group8  cap:k
-event18  DEVICE_ADDED                 BX17 (AVRCP)                      seat0 default group9  cap:k
-event16  POINTER_BUTTON               +0.000s  BTN_MIDDLE (274) pressed, seat count: 1
 event16  POINTER_BUTTON               +0.192s  BTN_MIDDLE (274) released, seat count: 0
 event16  POINTER_SCROLL_WHEEL         +1.128s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      2  +1.152s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      3  +1.464s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      4  +1.488s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      5  +2.256s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      6  +2.280s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      7  +2.712s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      8  +2.736s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL      9  +3.096s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     10  +3.120s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     11  +3.624s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     12  +3.648s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     13  +4.008s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     14  +4.032s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     15  +4.321s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     16  +4.344s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     17  +5.040s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     18  +5.065s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     19  +5.712s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     20  +5.736s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     21  +5.976s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     22  +6.000s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     23  +6.866s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     24  +6.890s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     25  +7.154s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     26  +7.178s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     27  +7.994s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     28  +8.018s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     29  +8.258s  vert 15.00/120.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     30  +8.282s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     31  +9.002s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     32  +9.026s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     33  +9.314s  vert 1905.00/15240.0* horiz 0.00/0.0 (wheel)
 event16  POINTER_SCROLL_WHEEL     34  +9.338s  vert 1920.00/15360.0* horiz 0.00/0.0 (wheel)

[Patola]

@pastaq would you be able to take a look?

[Patola]

@pastaq Ok, as the use of this mouse wheel is quite important for me, I forked the repository and used my Claude Code Pro plan to find the issue and fix it. After ~3h of iterations, it generated a quite short commit one one part of the code that affects the legion go/go 2 only, and I will be creating the pull request for the fix.

I have read the instructions for AI-generated code in the CONTRIBUTING.md and I will be following them.

[Patola]

Unfortunately it seems it needs additional testing from Legion Go 1 owners to make sure it doesn't break its input handling. Maybe the Legion Go S too.

[Patola]

BTW, the base code used for this patch is the most recent, which on Arch and CachyOS is on package inputplumber 0.77.2-1.

[Patola]

If someone wants to test, I've shared its build here: https://github.com/Patola/InputPlumber/releases/tag/0.77.3-legiongo2-mousewheel

[pastaq]

@Patola after discussing with ShadowApex again yesterday, we're going to be updating our policy for the time being on AI usage to not allow AI use for the time being, and we'll also be removing any existing AI assisted contributions and redeveloping them ourselves. This policy will be revisited after the legal questions about FOSS licensing and LLM generated code are answered. We're primarily concerned about copyleft contamination braking the GPL license.

I plan to work on go1/go2 issues later this week so this is an issue I'll be attempting to resolve. I the easiest answer may be to separate the drivers from go1 and go2, but the behavior difference may in fact be related to the older firmware being different.