OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
omarchy-core medium occasional DISPUTE

Stop the lock screen re-blanking a slow-waking monitor before it shows

omarchyhyprlandquickshellomarchy-shellamdnvidiadesktopdisplayport
CORRECTED An auditor found the fix wrong or incomplete and rewrote it.
AUDIT NOTE

Checked the record against the shipped source on this workstation (omarchy 4.0.2-1, Hyprland 0.56.2), against the v4.0.1, v4.0.2, v4.0.3 and quattro copies of the file, and re-read issues #7749, #7399 and #7380 and PR #7643 in full today. The mechanism holds exactly as written. Confirmed on this machine, all read from the installed files. idleBlankTimer is interval: 5000, repeat: false at /usr/share/omarchy/shell/plugins/lock/Service.qml:415-417. runWake() at 167 to 170 starts wakeProcess and re-arms the timer when lockRequested. wakeProcess runs omarchy-system-wake (405 to 407) with no completion handler. omarchy-system-wake calls omarchy-brightness-display on, which dispatches hl.dsp.dpms({ action = "enable" }) and exits without waiting for a frame (bin/omarchy-brightness-display:70), and skips the dispatch entirely when every active monitor already reports dpmsStatus true (line 69). LockView.qml emits wakeRequested() on clicks, pointer motion, keystrokes and text changes (118, 119, 163 to 166, 177 to 178), which Service.qml:283 maps to runWake(). The tooling in the fix is real and works the way the record says. omarchy-plugin-clone builds the id as ${USER}.${source_id#omarchy.}, copies the entry points, stamps omarchy.clonedFrom, enables the clone and calls omarchy-shell shell rescanPlugins. resolveEnabledId (PluginRegistry.qml:146-157) routes calls made with the built-in id to the enabled clone. The shell watches ~/.config/omarchy/plugins with inotifywait -m -r (PluginRegistry.qml:636-651, and inotify-tools 4.25.9.0-1 is installed), so a save does reload. omarchy-plugin-remove restores the clonedFrom source.

The cited issues do support the claims. #7749's opening post names the ViewSonic VA3456-WQHD on a Radeon 890M at Omarchy 4.0.0-1, quotes the same timer, and reports the controlled clone test at 15000. The omarchybot collaborator comment confirms the source reading and says plainly that it was not reproduced on hardware. #7399's reporter is a 6K ASUS ProArt PA32QCV who saw no change at 30000, and the commenter orienw is the ASUS XG32UCWMG at 3840x2160@240 on an RTX 2070 who was fixed by 20000, so the record attributes both correctly. #7380 is the separate FALLBACK shell-crash defect the record points at.

Two things needed changing, neither of them the mechanism. First, the date and release framing: I fetched shell/plugins/lock/Service.qml at every tag and the local file is byte-identical to v4.0.2 and to v4.0.3 (tagged 2026-09-08, the newest tag), and interval: 5000 is still hardcoded even on quattro HEAD, where the only new timer is a 3000 ms monitorDpmsTimer that polls hyprctl monitors for video wallpapers and is not a readiness observer. PR #7643 reports state: open on 2026-09-11. The cause now says that, so the record cannot be read as describing a number that a release has already moved. Second, danger was empty and should not be. Enabling a clone of a non bar-widget plugin writes omarchy.lock into disabledPlugins in ~/.config/omarchy/shell.json (PluginRegistry.qml:523-525, and the lock manifest's kinds is ["service"]), and the lock IPC target lives inside the plugin (Service.qml:510-511), so a QML error in the clone leaves no lock service, and omarchy-system-lock sends omarchy-shell lock lock with output discarded and no exit check. The failure is a machine that silently stops locking, which is worth a danger on a fix that tells people to edit the lock screen's own source. I also added a listPlugins check to the fix and --yes to the remove command, since omarchy-plugin-remove refuses without a confirmation prompt outside a terminal.

Not exercised: I did not clone the plugin, did not edit any interval, did not lock this session and ran no DPMS or state-changing hyprctl command, because the audit is read-only here and Omarchy 4's lock surface cannot be released headlessly. The 15000 and 20000 values, the slow DisplayPort wake latency and the claim that a longer interval cures the symptom come from the two reporters in the issues rather than from anything I measured. Everything about the source, the plugin tooling and the clone's effect on shell.json is read from the installed files and from the upstream tags.

The cause above was rewritten on 2026-09-11 to match this note; the fix was corrected by the audit itself.
RISK

Enabling a cloned lock plugin turns the built-in one off, so a broken clone leaves the machine unable to lock. PluginRegistry.qml:523-525 writes omarchy.lock into disabledPlugins in ~/.config/omarchy/shell.json whenever the enabled clone declares a kind other than bar-widget, and the lock plugin's manifest.json declares "kinds": ["service"]. The lock IPC target lives inside the plugin itself (Service.qml:510-511), so a QML error in your copy means there is no lock service and no target for the omarchy-shell lock lock that omarchy-system-lock sends. That call discards its output and its exit status is not checked, so locking then fails silently and the session stays unlocked.

Change only the interval line, confirm the clone is loaded with omarchy-shell shell listPlugins before relying on it, and test with omarchy-system-lock while you are sitting at the machine. Do not test a lock-plugin clone over ssh on a headless or remote machine: Omarchy 4's lock surface is an ext-session-lock surface with no unlock() in its IPC, so the only ways back in are typing the password at the console or a reboot.

Symptom

On Omarchy 4 (Quattro) a locked display does not wake reliably. One key press or mouse movement turns the monitor on but the lock screen never appears, or the password prompt shows for a moment and the monitor goes black again. Pressing keys or moving the mouse repeatedly eventually gets the lock screen up. omarchy-shell stays alive and there are no FALLBACK output or fatal Wayland errors in the journal, which separates this from the shell crash in #7380. The same hardware woke on the first press on Omarchy 3.

Seen on DisplayPort monitors that take more than five seconds to complete a DPMS wake: a ViewSonic VA3456-WQHD on an AMD Radeon 890M (Omarchy 4.0.0-1) and an ASUS XG32UCWMG at 3840x2160@240 on an RTX 2070 (Omarchy 4.0.0-1, issue #7399). The shipped value is unchanged on 4.0.2-1.

Cause

The Quickshell lock plugin blanks the display on a fixed one-shot timer. In shell/plugins/lock/Service.qml idleBlankTimer has interval: 5000 with repeat: false (lines 415 to 417 on 4.0.2-1). runWake() (lines 167 to 170) starts omarchy-system-wake to turn DPMS on and, while lockRequested is true, immediately re-arms that timer. Nothing in the path observes whether the panel actually lit: wakeProcess has no completion handler (lines 405 to 407), and omarchy-brightness-display on returns once hyprctl has accepted the dispatch, not when the monitor shows a frame (bin/omarchy-brightness-display:70). A monitor whose DisplayPort wake handshake takes longer than five seconds is therefore blanked again before it becomes visible.

The budget is five seconds of inactivity on the lock surface rather than five seconds from the wake. LockView.qml emits wakeRequested() on pointer clicks and motion (lines 118 and 119) and on every keystroke and text change in the password field (lines 163 to 166 and 177 to 178), and the service maps that to runWake() at line 283, which re-arms the countdown. That is why repeated input eventually wins and why typing the password blind works.

A collaborator confirmed this reading against the source, and the reporter's controlled test (cloning the plugin and changing only the interval) confirmed it on hardware. A second user in #7399 fixed the same symptom the same way at 20000. The original reporter of #7399 saw no change at 30 seconds, so a second re-blank path exists on some hardware.

The interval is still a hardcoded literal in the newest release. shell/plugins/lock/Service.qml is byte-identical on this workstation (4.0.2-1) and at the v4.0.1, v4.0.2 and v4.0.3 tags, and interval: 5000 is still hardcoded on the unreleased quattro branch as well. PR #7643, which makes it configurable as idle.blank in shell.json, was still open and unmerged on 2026-09-11, so the clone below is the only way to change the number on any shipped release.

Fix

Clone the built-in lock plugin into your config and raise the interval. Cloning switches the shell to your copy and routes the lock IPC calls to it (PluginRegistry.qml resolveEnabledId, line 146), so nothing else changes:


omarchy plugin clone omarchy.lock

That creates ~/.config/omarchy/plugins/<username>.lock/ (for user dhh, dhh.lock), enables it and switches the shell to it. Edit Service.qml in that directory and change the one line:


    interval: 5000

to


    interval: 15000

15000 worked for the ViewSonic reporter, 20000 for the #7399 commenter. Change nothing else: the same file holds the PAM flows and the lock IPC target, and a QML error in it leaves you with no lock screen at all. See the danger below.

Saving a file under ~/.config/omarchy/plugins/ reloads plugin code automatically, because the shell runs inotifywait -m -r on that directory (services/PluginRegistry.qml:636-651). Check the clone came back before you trust it:


omarchy-shell shell listPlugins | jq -r '.[] | select(.id | endswith(".lock")) | "\(.id) enabled=\(.enabled)"'

If the reload did not happen, force it:


omarchy-shell shell rescanPlugins

The clone stops tracking updates to the built-in lock plugin, so remove it once a release makes the interval configurable or fixes the race:


omarchy plugin remove <username>.lock --yes

Removing an active clone switches the shell back to the built-in omarchy.lock. The --yes matters only outside an interactive terminal, where omarchy-plugin-remove refuses to continue without a confirmation it cannot prompt for.

If raising the interval changes nothing, this is not your cause. The #7399 reporter saw no difference at 30 seconds, and a lock screen that never appears alongside Got removal for monitor "FALLBACK" in the journal is #7380 instead.

Verify


omarchy-system-lock

Wait for the monitor to enter standby, then press Shift once. The monitor wakes and the lock screen stays visible without further input. omarchy plugin list shows <username>.lock enabled in place of omarchy.lock.

Sources

lock-screen-reblanks-slow-dpms-monitor