OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
audio-input high occasional DISPUTE

Get keyboard input back at the lock screen after resume when fcitx5 wedges

omarchyhyprlandwaylandfcitx5laptop
CORRECTED An auditor found the fix wrong or incomplete and rewrote it. · checked against Omarchy omarchy 4.0.4-1 on 2026-10-05
AUDIT NOTE

Re-checked on 4.0.4-1. /usr/lib/environment.d/10-omarchy-fcitx.conf (omarchy-settings 4.0.4-1) sets QT_IM_MODULE=fcitx, the lock field is a TextInput at /usr/share/omarchy/shell/plugins/lock/LockView.qml:132 with activeFocusOnPress at :143, the installed system-sleep hooks are force-igpu, keyboard-backlight and unmount-fuse (quattro has since added an unrelated fprintd-resume), and the graphical session is on tty1, so the TTY switch-back instruction holds. Issue #9325 read in full: the body, triage and the rek comment support the fcitx5 restart, the busctl check, the click test and the stale lock surface variant. The earlier correction (--no-block hook, warning against omarchy-restart-shell) still stands. One gap: issue #7461 has a timestamped incident (parnoldx, 2026-09-22) explicitly linking to #9325, in which D-Bus activation started a second fcitx5, the two displaced each other through the unit's restart loop, and the keyboard went dead because fcitx5 holds Hyprland's input-method keyboard grab. A resume hook that restarts the unit does not stop that route, so the cause and the diagnosis step now name it and point at the restart-loop drop-in. Not exercised: no suspend cycle and no sleep hook installed.

The cause above was rewritten on 2026-10-05 to match this note; the fix was corrected by the audit itself.
RISK

Do not run omarchy-restart-shell over ssh or from a TTY to recover: on a locked session it re-locks and leaves you with a fresh lock you still cannot release.

Symptom

After suspend and resume the Omarchy lock screen accepts no keystrokes at all. The "Enter Password" placeholder sits there and every key vanishes, which looks like a full system freeze, so people cold-boot. Hyprland is actually alive (hyprctl works over ssh) and the journal shows a clean resume. Clicking the password field does not help.

Cause

The lock screen's password field is a Quickshell (Qt) TextInput, and QT_IM_MODULE=fcitx from /usr/lib/environment.d/10-omarchy-fcitx.conf routes it through fcitx5, which also holds Hyprland's input-method keyboard grab. If fcitx5 comes back from sleep unable to serve its input method, it swallows the keys. Nothing in Omarchy restarts fcitx5 across sleep: the system-sleep hooks shipped in 4.0.4 are force-igpu, keyboard-backlight and unmount-fuse. A second route to the same dead keyboard is a second fcitx5 started by D-Bus activation or an autostart entry, which fights omarchy-fcitx5.service through its Restart=always loop until the grab wedges. If clicking the field does fix typing, it is the separate lock-focus bug, not this one.

Fix

**Recover now:** switch to a TTY (Ctrl+Alt+F3) or ssh in, log in as the same user, restart the managed unit, then switch back (Ctrl+Alt+F1 or F2):


systemctl --user restart omarchy-fcitx5.service

Do not run fcitx5 -r -d by hand. It takes fcitx5's D-Bus name away from the unit and omarchy-fcitx5.service then restart-loops every 2 seconds indefinitely.

**Confirm it is fcitx5** next time, before restarting anything:


busctl --user --timeout=3 introspect org.fcitx.Fcitx5 /controller   # a timeout means fcitx5 is wedged
systemctl --user show omarchy-fcitx5.service -p NRestarts           # climbing means two fcitx5 are fighting
journalctl --user -u omarchy-fcitx5 -b --no-pager | tail
omarchy-shell lock status

If NRestarts climbs every couple of seconds, a second fcitx5 is competing with the unit. Fix that first with the Restart=on-failure drop-in and the autostart check from the omarchy-fcitx5-service-restart-loop record, because a resume hook will not stop it.

If fcitx5 answers promptly and restarting it does not bring typing back, this is a different defect reported in the same issue: the lock surface is not re-created after the screen list changes on resume. Do not run omarchy-restart-shell from a TTY or ssh to recover from it, because that re-locks the session and you cannot release the new lock either. Type the password at the physical keyboard if any surface takes input, otherwise reboot.

**Restart it automatically after every resume** with a system-sleep hook. The reporter used a PrepareForSleep watcher. This hook is an equivalent built from systemd's documented interface and has not been exercised on hardware for this record. --no-block queues the restart and returns, so the job survives the end of the suspend service, unlike a backgrounded subshell. The heredoc expands $USER at creation time:


sudo tee /etc/systemd/system-sleep/restart-fcitx5 >/dev/null <<EOF
#!/bin/bash
if [[ \$1 == post ]]; then
  systemctl --user --machine=${USER}@ --no-block restart omarchy-fcitx5.service
fi
EOF
sudo chmod 755 /etc/systemd/system-sleep/restart-fcitx5

**If you type no CJK**, disabling fcitx5 removes it from the lock screen's key path entirely (you lose fcitx5-provided Compose sequences):


systemctl --user disable --now omarchy-fcitx5.service

Verify

Suspend, resume, and type at the lock screen without clicking. journalctl --user -u omarchy-fcitx5 -b --no-pager | tail shows a restart right after resume when the hook is installed.

Sources

lock-screen-no-keyboard-input-after-resume-fcitx5