You are not logged in.
Just putting out a quick warning in case anyone else hits this after a routine system upgrade.
After updating my kernel from 7.2.14.arch1-1 to 7.2.15.arch1-1, my built-in laptop keyboard developed severe chattering/double-typing (making entering passwords quite an adventure!).
Interestingly:
- The issue was completely absent in the TTY.
- It was heavy and persistent inside my Wayland graphical environment using labwc.
Downgrading back to 7.2.14.arch1-1 immediately resolved the issue, with no chattering present since.
Hardware & System Details:
• Keyboard: laptop, Dell 16 Pro Plus (built-in keyboard)
• Display Server / DE: Wayland / labwc
• Broken kernel: 7.2.15.arch1-1
• Working Kernel: 7.2.14.arch1-1
If you are experiencing similar input behaviour on this kernel version, rolling back or holding the kernel package temporarily might save you some frustration.
dif
Last edited by dif (Yesterday 17:44:21)
Offline
double-typing … completely absent in the TTY … heavy and persistent inside my Wayland
Are other compositors or X11 affected?
Offline
My Wayland compositor (labwc) and Xwayland were completely unaffected.
The chattering keyboard turned out to be a kernel input regression (likely in i8042/atkbd on my Dell laptop)—downgrading the kernel fixed it immediately, so I'm holding off on that kernel version for now.
By the way:
$ pacman -Qs xorg|grep '^[^ ]'
local/xorg-fonts-encodings 1.1.0-2 (xorg-fonts xorg)
local/xorg-mkfontscale 1.2.4-2 (xorg-apps xorg)
local/xorg-server-common 21.1.26-1 (xorg)
local/xorg-setxkbmap 1.3.5-1 (xorg-apps xorg)
local/xorg-xdpyinfo 1.4.0-1 (xorg-apps xorg)
local/xorg-xkbcomp 1.5.0-1 (xorg-apps xorg)
local/xorg-xprop 1.2.8-1 (xorg-apps xorg)
local/xorg-xwayland 24.1.15-1 (xorg)
local/xorgproto 2026.1-1And that's it.
Perhaps I should use an i8042.* kernel boot parameter. Since I run a complex EFISTUB setup, I didn't feel like modifying boot entries when the problem first occurred, but now I'm considering testing it seriously.
Offline
Your OP says it exclusively affected labwc but notably not the console, so the idea was to check other environments (or inspect evtest and libinput debug-events) to see whether your compositor (config) takes input from two devices ("doubled") reflecting the same hardware (where the second one shows up w/ the updated kernel)
If it was strictly the kernel you'd expect similar behavior on the console
Offline
You are right.
Checked my baseline on 7.2.14 with libinput list-devices and /proc/bus/input/devices—right now AT Translated Set 2 keyboard (/dev/input/event3) is the only device mapping regular typing keys (the other keyboard-tagged nodes are just Power/Sleep buttons, Dell WMI hotkeys, Dell Privacy Driver, PMF-TA, etc.).
When I test the newer kernel next, I'll check sudo libinput debug-events --show-keycodes first to see whether a second virtual device is duplicating keystrokes in Wayland or if event3 itself is firing multiple times."
But first I have to sort out my EFISTUB entries, so I don't get stranded when I boot in linux 7.2.15.
Offline
Update: You were right that it isn't isolated to Wayland/labwc—in fact, the behavior is much worse than simple key chattering.
After setting up linux-lts as a fallback EFISTUB entry and upgrading back to 7.2.15.arch1-1, the system now locks up completely at the early boot LUKS prompt. Every time I start typing my passphrase, the machine hangs mid-input before I can even press Enter. Because it freezes in initramfs, I can't get into the system to run libinput debug-events or evtest on 7.2.15.
Booted into linux-lts to recover and downgraded back to 7.2.14.arch1-2, where everything works normally again. It looks like a kernel-level input/interrupt regression that freezes the system under keyboard input—I must have just got lucky getting past LUKS the first time around.
Holding linux at 7.2.14.arch1-2 for now and updating the thread title to reflect the early-boot freeze.
Offline
Nothing in https://github.com/gregkh/linux/commits … t/keyboard or https://github.com/gregkh/linux/commits … rivers/i2c
How did you arrive at the original situation resp. do you still have a journal of that boot (and a 7.2.14 boot for comparism)?
Offline