System is left unusable if there are no enabled monitors

[alba4k]

Say I have a single monitor and choose to disable it with hyprctl keyword monitor eDP-1,disable.

I am left with an anusable system and need to reboot with reisub/power button, I can't even switch to another tty anymore.

[fiffeek]

I wanted to create a similar post but seems this already exists, so I'll piggy back on it with some more details and how to reproduce -- also I believe in some cases it's still "working" but in most it doesn't. I got to the point where I can reproduce a crash for this (only on start up though, when applied to a running system it just swallows the errors, or it's a red herring but anyway, my findings below).

---

Hey!

I think there is a bug when hyprland fails to get to an interactible state after all the monitors are disabled on startup (this works however, after the startup) -- wanted to know if this is desired, and if so why.

Imagine a scenario when as the user I have one of my two desired "monitor profiles" in ~/.config/hypr/monitors.conf (and I source them in ~/.config/hypr/hyprland.conf).).
The profiles are:

1. Two monitors, a laptop screen and the external screen; the laptop screen is disabled since it's unneeded, e.g.:

# disable a laptop monitor
monitor=desc:BOE NE135A1M-NY1,disable
# configure the external monitor
monitor=desc:LG Electronics LG SDQHD 301NTBKDU037,[email protected],-1800x0,1.60000000,transform,3,vrr,0

2. Only the laptop monitor:

# configure the laptop monitor
monitor=eDP-1,[email protected],0x0,2.0,vrr,1

How I switch between these is somewhat unimportant (can be manual, symlinks, etc.). And essentially the "profiles" are just settings for where the user'd use the settings at, nothing super special about these but it's the easiest to illustrate on a working example.

Now, let's say the contents of monitors.conf is:

monitor=desc:BOE NE135A1M-NY1,disable
monitor=desc:LG Electronics LG SDQHD 301NTBKDU037,[email protected],-1800x0,1.60000000,transform,3,vrr,0

then I shutdown the computer, remove the HDMI cable for the external monitor, and boot again. Then run Hyprland again to get to the session -- the screen is blank, that is somewhat to be "expected", since the output is disabled. What surprises me more, though is even after changing the contents to (with the blank screen, we will get to how shortly):

monitor=eDP-1,[email protected],0x0,2.0,vrr,1

Hyprland seems to be stuck and the monitor is not interactive anymore. The same is reported in this reddit post. When hyprland gets to this state the keybinds do not seem to work (e.g. exiting the session), same for exec-once and other directives.

This does work when I start up with a "working state" (e.g. a laptop screen configured, only one screen, then disable it, then reenable it).

Here are steps to most easily reproduce, assuming you have one display attached (eDP-1 in my case):
~/hyprtest/hypr.conf:

# the main config file, only the first line matters
source=~/hyprtest/hypr-disabled.conf
exec-once=bash -c 'echo "hello" > /tmp/hyprrun.log'
debug:disable_logs = false
$mainMod = SUPER
bind = $mainMod, X, exit,

~/hyprtest/hypr-enabled.conf:

monitor=eDP-1,[email protected],0x0,2.0,vrr,1

~/hyprtest/hypr-disabled.conf:

monitor=eDP-1,disable

~/hyprtest/switch-monitors.sh:

#!/bin/bash

(sleep 10 && sed -i 's/disabled/enabled/g' ~/hyprtest/hypr.conf) &

~/hyprtest/save-logs.sh:

#!/bin/bash

(sleep 15 && cp -r /run/user/1000/hypr/ ~/hyprtest/logs/) &

Then run with (not under a nested session):

~/hyprtest/save-logs.sh && ~/hyprtest/switch-monitors.sh && HYPRLAND_TRACE=1 AQ_TRACE=1 Hyprland --config ~/hyprtest/hypr.conf

Explanation:

• you can see a black screen instantly
• after 10s a monitor switch is issued by changing source= line (can see that in logs too, below)
• after 15s save all logs from ephemeral /run/user/ to a dir (just in case the crash does not happen; the desktop is not responsive so a reboot might be required)
• the above two run in the bg independently to hyprland command
• a crash might happen

For me sometimes a crash happens, sometimes everything is just frozen (can't even grab the keyboard input to kernel to switch to another tty).

Here are the logs (stripped to only errors):

❯ grep "ERR" 967c3c7404d4fa00234e29c70df3e263386d2597_1763494120_547214715/hyprland.log
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] drm: getCurrentCRTC: No CRTC 0
[ERR] [AQ] Wayland backend cannot start: wl_display_connect failed (is a wayland compositor running?)
[ERR] [AQ] Requested backend (wayland) could not start, enabling fallbacks
[ERR] [AQ] Implementation wayland failed, erasing.
[ERR] Hyprcursor failed loading theme "", falling back to Xcursor.
[ERR] [AQ] drm: No format for output
[ERR] [AQ] drm: No format for output
[ERR] [AQ] drm: No format for output
[ERR] Couldn't commit output eDP-1 in ensureVRR -> true
[ERR] CConfigWatcher: got an event for wd 2 which we don't have?!

same but stripped to only eDP-1 mentions:

❯ grep "eDP-1" 967c3c7404d4fa00234e29c70df3e263386d2597_1763494120_547214715/hyprland.log
[LOG] [AQ] drm: Connector gets name eDP-1
[LOG] [AQ] drm: Skipping connector eDP-1, has crtc 80 and is connected
[LOG] [AQ] drm: slot 0 crtc 80 taken by eDP-1, skipping
[LOG] [AQ] drm: Connector eDP-1 connected
[LOG] [AQ] drm: Connecting connector eDP-1, CRTC ID 80
[LOG] [AQ] drm: Description BOE NE135A1M-NY1  (eDP-1)
[LOG] [AQ] drm: connector eDP-1 crtc is capable of vrr: props.vrr_capable -> 97, crtc->props.vrr_enabled -> 24
[LOG] [AQ] drm: connector eDP-1 crtc supports CTM
[LOG] [AQ] drm: connector eDP-1 crtc supports HDR (7)
[LOG] [AQ] drm: connector eDP-1 crtc supports Colorspace (96)
[LOG] [AQ] drm: onReady: connector 94 has output name eDP-1
[LOG] New aquamarine output with name eDP-1
[LOG] New output with name eDP-1
[LOG] [AQ] drm: Disabling output eDP-1
[LOG] Applying monitor rule for eDP-1
[LOG] Added new monitor with name eDP-1 at [-1, -1] with size [0, 0], pointer 5561566c3700
[LOG] Pending output eDP-1 does not accept VRR.
[ERR] Couldn't commit output eDP-1 in ensureVRR -> true
[LOG] checking if we have seen this monitor before: eDP-1
[LOG] Monitor eDP-1 was not on any workspace
[LOG] [IWaylandProtocol] Registered global [WLOutput (eDP-1)]
[LOG] arrangeMonitors: eDP-1 explicit [0, 0]
[LOG] arrangeMonitors: eDP-1 xwayland [1920, 0]
[LOG] Monitor eDP-1 is disabled but is requested to be enabled
[TRACE] Monitor eDP-1 requested modes:
[LOG] Monitor eDP-1: requested [email protected], using preferred mode [email protected]
[LOG] output eDP-1 succeeded basic test on format DRM_FORMAT_XRGB8888
[LOG] [AQ] drm: Modesetting eDP-1 with [email protected]
[LOG] Monitor eDP-1 -> destroyed all render data
[LOG] Monitor eDP-1 data dump: res [email protected], scale 2.00, transform 0, pos 0x0, 10b 0
[LOG] arrangeMonitors: eDP-1 explicit [0, 0]
[LOG] arrangeMonitors: eDP-1 xwayland [1920, 0]
[TRACE] [pointer] monitor: eDP-1, size: [Vector2D: x: 24, y: 24], hw buf: [Vector2D: x: 24, y: 24], scale: 1.00, monscale: 2.00, xbox: [Vector2D: x: 48, y: 48]
[TRACE] [pointer] hw transformed hotspot for eDP-1: [Vector2D: x: 6, y: 2]
[TRACE] [pointer] monitor: eDP-1, size: [Vector2D: x: 24, y: 24], hw buf: [Vector2D: x: 24, y: 24], scale: 1.00, monscale: 2.00, xbox: [Vector2D: x: 48, y: 48]
[TRACE] [pointer] hw transformed hotspot for eDP-1: [Vector2D: x: 6, y: 2]
[TRACE] [pointer] monitor: eDP-1, size: [Vector2D: x: 24, y: 24], hw buf: [Vector2D: x: 24, y: 24], scale: 1.00, monscale: 2.00, xbox: [Vector2D: x: 48, y: 48]
[TRACE] [pointer] hw transformed hotspot for eDP-1: [Vector2D: x: 6, y: 2]

same but stripped to FALLBACK mentions:

❯ grep "FALLBACK" 967c3c7404d4fa00234e29c70df3e263386d2597_1763494120_547214715/hyprland.log
[LOG] New output with name FALLBACK
[WARN] No rule found for FALLBACK, trying to use the first.
[LOG] Applying monitor rule for FALLBACK
[TRACE] Monitor FALLBACK requested modes:
[LOG] Monitor FALLBACK: requested preferred, using preferred mode [email protected]
[LOG] output FALLBACK succeeded basic test on format DRM_FORMAT_XRGB8888
[LOG] Monitor FALLBACK -> destroyed all render data
[LOG] Monitor FALLBACK data dump: res [email protected], scale 1.00, transform 0, pos -1x-1, 10b 0
[LOG] Added new monitor with name FALLBACK at [-1, -1] with size [1920, 1080], pointer 556157769090
[LOG] checking if we have seen this monitor before: FALLBACK
[LOG] Monitor FALLBACK was not on any workspace
[LOG] arrangeMonitors: FALLBACK auto [-1, -1]
[LOG] arrangeMonitors: FALLBACK xwayland [0, 0]
[TRACE] Output FALLBACK rejected hardware cursors, falling back to sw
[LOG] arrangeMonitors: FALLBACK auto [0, 0]
[LOG] arrangeMonitors: FALLBACK xwayland [0, 0]
[TRACE] Output FALLBACK rejected hardware cursors, falling back to sw
[LOG] arrangeMonitors: FALLBACK auto [0, 0]
[LOG] arrangeMonitors: FALLBACK xwayland [0, 0]
[LOG] arrangeMonitors: FALLBACK auto [0, 0]
[LOG] arrangeMonitors: FALLBACK xwayland [0, 0]
[LOG] onDisconnect called for FALLBACK
[LOG] Monitor FALLBACK -> destroyed all render data
[LOG] Disconnecting Monitor FALLBACK was on workspace 1
[LOG] Removed monitor FALLBACK!
[TRACE] onCursorMoved for output FALLBACK: cursor entered the output, but no front buffer, forcing recalc
[TRACE] onCursorMoved for output FALLBACK: cursor entered the output, but no front buffer, forcing recalc

(I noticed a fallback is added when no other output is specified)

Here's the crash report:

❯ cat $XDG_CACHE_HOME/hyprland/hyprlandCrashReport10326.txt
--------------------------------------------
   Hyprland Crash Report
--------------------------------------------
Maybe you should try dusting your PC in the meantime?

Hyprland received signal 11(SEGV)
Version: 967c3c7404d4fa00234e29c70df3e263386d2597
Tag: v0.52.1
Date: Sun Nov 9 23:31:36 2025
Flags:

System info:
        System name: Linux
        Node name: bloodhound
        Release: 6.17.8-arch1-1
        Version: #1 SMP PREEMPT_DYNAMIC Fri, 14 Nov 2025 06:54:20 +0000

GPU:
        c1:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Phoenix1 [1002:15bf] (rev c4) (prog-if 00 [VGA controller])


os-release:
        NAME="Arch Linux"
        PRETTY_NAME="Arch Linux"
        ID=arch
        BUILD_ID=rolling
        ANSI_COLOR="38;2;23;147;209"
        HOME_URL="https://archlinux.org/"
        DOCUMENTATION_URL="https://wiki.archlinux.org/"
        SUPPORT_URL="https://bbs.archlinux.org/"
        BUG_REPORT_URL="https://gitlab.archlinux.org/groups/archlinux/-/issues"
        PRIVACY_POLICY_URL="https://terms.archlinux.org/docs/privacy-policy/"
        LOGO=archlinux-logo

Libraries:
Hyprgraphics: built against 0.3.0, system has 0.3.0
Hyprutils: built against 0.10.2, system has 0.10.2
Hyprcursor: built against 0.1.13, system has 0.1.13
Hyprlang: built against 0.6.3, system has 0.6.3
Aquamarine: built against 0.9.5, system has 0.9.5

Backtrace:
        # | Hyprland(_Z12getBacktracev+0x61) [0x5561474981c1]
                getBacktrace()
                ??:?
        #1 | Hyprland(_ZN14NCrashReporter18createAndSaveCrashEi+0xcd6) [0x5561473c4066]
                NCrashReporter::createAndSaveCrash(int)
                ??:?
        #2 | Hyprland(+0x2b0a02) [0x556147325a02]
                _start
                ??:?
        #3 | /usr/lib/libc.so.6(+0x3e4d0) [0x7fa35683e4d0]
                ??
                ??:0
        #4 | Hyprland(+0x1d9006) [0x55614724e006]
                std::__throw_bad_any_cast()
                ??:?
        #5 | Hyprland(_ZN22CMonitorFrameScheduler7onFrameEv+0x207) [0x5561474ca4b7]
                CMonitorFrameScheduler::onFrame()
                ??:?
        #6 | /usr/lib/libhyprutils.so.9(_ZN9Hyprutils6Signal15CSignalListener12emitInternalEPv+0x2d) [0x7fa3577356ad]
                ??
                ??:0
        #7 | /usr/lib/libhyprutils.so.9(_ZN9Hyprutils6Signal11CSignalBase12emitInternalEPv+0x1aa) [0x7fa35773591a]
                ??
                ??:0
        #8 | /usr/lib/libaquamarine.so.8(_ZN10Aquamarine8CBackend12dispatchIdleEv+0x17e) [0x7fa3578025de]
                ??
                ??:0
        #9 | Hyprland(+0x5336da) [0x5561475a86da]
                Cursor::CShapeOverrideController::unsetOverride(Cursor::eCursorShapeOverrideGroup)
                ??:?
        #1 | /usr/lib/libwayland-server.so.0(wl_event_loop_dispatch+0x1d2) [0x7fa35760a182]
                ??
                ??:0
        #11 | /usr/lib/libwayland-server.so.0(wl_display_run+0x37) [0x7fa35760c297]
                ??
                ??:0
        #12 | Hyprland(_ZN17CEventLoopManager9enterLoopEv+0x291) [0x5561475ab301]
                CEventLoopManager::enterLoop()
                ??:?
        #13 | Hyprland(main+0x1234) [0x55614729c414]
                main
                ??:?
        #14 | /usr/lib/libc.so.6(+0x27635) [0x7fa356827635]
                ??
                ??:0
        #15 | /usr/lib/libc.so.6(__libc_start_main+0x89) [0x7fa3568276e9]
                ??
                ??:0
        #16 | Hyprland(_start+0x25) [0x556147324ea5]
                _start
                ??:?


Log tail:
[TRACE] [AQ] drm: new buffer 165
[LOG] [AQ] drm: Cursor buffer imported into KMS with id 165
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 3, needsFrame true, isPageFlipPending true, frameEventScheduled false
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 3, needsFrame true, isPageFlipPending true, frameEventScheduled false
[ERR] CConfigWatcher: got an event for wd 2 which we don't have?!
[TRACE] [AQ] drm: pf event seq 0 sec 0 usec 0 crtc 80
[WARN] Attempted to render frame on inactive session!
[LOG] Leaving unsafe state
[LOG] [hookSystem] New hook event registered: preMonitorRemoved
[LOG] onDisconnect called for FALLBACK
[LOG] Monitor FALLBACK -> destroyed all render data
[LOG] Disconnecting Monitor FALLBACK was on workspace 1
[LOG] Removed monitor FALLBACK!
[TRACE] onCursorMoved for output FALLBACK: cursor entered the output, but no front buffer, forcing recalc
[TRACE] [AQ] atomic moveCursor
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 4, needsFrame true, isPageFlipPending false, frameEventScheduled false
[TRACE] Not updating hw cursors: disabled / dpms off display
[TRACE] [AQ] atomic moveCursor
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 4, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [pointer] monitor: eDP-1, size: [Vector2D: x: 24, y: 24], hw buf: [Vector2D: x: 24, y: 24], scale: 1.00, monscale: 2.00, xbox: [Vector2D: x: 48, y: 48]
[TRACE] [pointer] hw transformed hotspot for eDP-1: [Vector2D: x: 6, y: 2]
[TRACE] [AQ] drm: CDRMFB: buffer has drmfb attachment with fb 556157a9a970
[LOG] [AQ] drm: Cursor buffer imported into KMS with id 165
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 3, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 3, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 6, needsFrame true, isPageFlipPending false, frameEventScheduled true
[LOG] moveWorkspaceToMonitor: Moving 1 to monitor 0
[LOG] moveWorkspaceToMonitor: Plugging gap with new 3
[LOG] moveWorkspaceToMonitor: Plugging gap with existing 3
[LOG] [hookSystem] New hook event registered: workspace
[LOG] moveWorkspaceToMonitor: SWITCHINGISACTIVE, active 2 -> 1
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 6, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 6, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] onCursorMoved for output FALLBACK: cursor entered the output, but no front buffer, forcing recalc
[TRACE] [AQ] atomic moveCursor
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 4, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] Not updating hw cursors: disabled / dpms off display
[TRACE] [AQ] atomic moveCursor
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 4, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [pointer] monitor: eDP-1, size: [Vector2D: x: 24, y: 24], hw buf: [Vector2D: x: 24, y: 24], scale: 1.00, monscale: 2.00, xbox: [Vector2D: x: 48, y: 48]
[TRACE] [pointer] hw transformed hotspot for eDP-1: [Vector2D: x: 6, y: 2]
[TRACE] [AQ] drm: CDRMFB: buffer has drmfb attachment with fb 556157a9a970
[LOG] [AQ] drm: Cursor buffer imported into KMS with id 165
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 3, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 3, needsFrame true, isPageFlipPending false, frameEventScheduled true
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 6, needsFrame true, isPageFlipPending false, frameEventScheduled true
[LOG] Destroying workspace ID 2
[LOG] [hookSystem] New hook event registered: destroyWorkspace
[LOG] [hookSystem] New hook event registered: moveWorkspace
[LOG] Destroying workspace ID 3
[TRACE] CMonitorState::ensureBufferPresent: Ignoring, monitor is not enabled
[LOG] [CLinuxDMABufV1Protocol] Resetting format table
[TRACE] [AQ] CDRMOutput::scheduleFrame: reason 5, needsFrame true, isPageFlipPending false, frameEventScheduled true

The above is fine if we start up with a monitor, then disable it, then reenable it, which tells me there is some hidden assumption being made or some invariant that does not hold.

Would appreciate some input whether this is desired -- this connects to a monitor switching tool I'm developing and likely a workaround is needed there since the assumption was that this'd work correctly.

[maxjo020418]

I also had the same issue and was with @fiffeek to confirm,

commenting just in case because you can recover by editing the kernel cmdline on the bootloader(limine, grub) by adding systemd.unit=multi-user.target at the end of the cmdline argument (ephemeral)

after rebooting that should drop you into the cli

[fiffeek]

I thought about this for a little bit, and I think the easiest solution would be to just not let the user disable the last enabled monitor/output (when parsing config and/or when issuing hyprctl commands), that would guarantee this magic invariant is held always (so no errors on the user side), and generally IMO this is what the user wants -- no one really should rely on the behavior of disabling all the displays?

If the above does not seem coherent or sound -- just do not let the user disable all the monitors (throw a warn/err or whatever and surface at the top like other config errors) on start up (since when running this can be recovered).

Above are just prevention method, I haven't yet got to read the code for this but I suspect for anyone not that familiar with the codebase this will be a nightmare to debug (even reproducing this takes lotta time, I can't get a nested hyprland session to reproduce this at all since disabling the only output [WAYLAND-1] on start-up will result in completely loosing access to the screen).