Created attachment 187478 [details] journal log SUMMARY Since update to Plasma 6.5.4, sometimes my computer freezes completely (I can't even move the mouse cursor) immediately before the screen enters in power saving mode. The screen dims partially and then the computer freezes completely and I need to force reboot . The journal log says this when the computer freezes: kwin_wayland[1119]: atomic commit failed: Invalid argument I use a single screen with intel iGPU. I'm attaching the full journal log. STEPS TO REPRODUCE 1. set the screen to automatically turn off after timeout 2. wait until the timeout is reached 3. OBSERVED RESULT sometimes computer freezes immediately before the screen entering in power saving mode EXPECTED RESULT the observed result should not occur SOFTWARE/OS VERSIONS Operating System: Arch Linux KDE Plasma Version: 6.5.4 KDE Frameworks Version: 6.20.0 Qt Version: 6.10.1 Kernel Version: 6.17.9-arch1-1 (64-bit) Graphics Platform: Wayland Processors: 8 × Intel® Core™ i7-4790 CPU @ 3.60GHz Memory: 32 GiB of RAM (31,2 GiB usable) Graphics Processor: Intel® HD Graphics 4600
This is Bug 512511, which I thought was master-only. We'll investigate and backport the fix if needed. Thanks for reporting it! *** This bug has been marked as a duplicate of bug 512511 ***
Let's look at what exactly is happening first. The warnings look a bit different at least, there's no EBUSY here: > dez 09 21:55:19 Arch-PC kwin_wayland[1119]: Atomic modeset test failed! Invalid argument > dez 09 21:55:19 Arch-PC kwin_wayland[1119]: Setting dpms mode failed! Please get a drm debug log of the situation: https://invent.kde.org/plasma/kwin/-/wikis/Debugging/Debugging-DRM-issues
Created attachment 187497 [details] drm-debug log It just happened again. The computer froze completely when Plasma tried to turn the screen off after playing a video in youtube with Firefox 145. drm-debug log attached.
I'm having the exact same issue on Plasma 6.5.4, didn't occur on 6.5.3 and it doesn't happen anymore since I reverted my system back to 6.5.3 so it seems to be specific to this version. For me it's not linked to the system entering power saving mode or anything as I disabled it entirely but it happened when the screen locks itself automatically after some time. I would say it happened about half the time Plasma tried to lock the screen. I have the same "Atomic modeset test failed! Invalid argument" in my logs but not the "Setting dpms mode failed!". I'm also using Firefox to play YouTube video so there may be a link there.
This is the entire section from one of the failing atomic tests: [20772.926973] i915 0000:00:02.0: [drm:drm_ioctl] comm="kwin_wayland" pid=1082, dev=0xe201, auth=1, DRM_IOCTL_MODE_ATOMIC [20772.926984] i915 0000:00:02.0: [drm:drm_atomic_state_init] Allocated atomic state 000000001f7f4c60 [20772.926992] i915 0000:00:02.0: [drm:drm_atomic_get_plane_state] Added [PLANE:38:sprite A] 000000008b2a04a1 state to 000000001f7f4c60 [20772.926997] i915 0000:00:02.0: [drm:drm_atomic_set_fb_for_plane] Set [NOFB] for [PLANE:38:sprite A] state 000000008b2a04a1 [20772.927003] [drm:drm_mode_object_get] OBJ ID: 109 (3) [20772.927008] [drm:drm_mode_object_get] OBJ ID: 109 (4) [20772.927011] i915 0000:00:02.0: [drm:drm_atomic_get_plane_state] Added [PLANE:44:cursor A] 00000000af9d8963 state to 000000001f7f4c60 [20772.927017] [drm:drm_mode_object_get] OBJ ID: 115 (2) [20772.927020] [drm:drm_mode_object_get] OBJ ID: 116 (4) [20772.927023] [drm:drm_mode_object_get] OBJ ID: 116 (5) [20772.927026] [drm:drm_mode_object_get] OBJ ID: 116 (6) [20772.927029] i915 0000:00:02.0: [drm:drm_atomic_get_crtc_state] Added [CRTC:49:pipe A] 00000000c01439e5 state to 000000001f7f4c60 [20772.927034] i915 0000:00:02.0: [drm:drm_atomic_state_default_clear] Clearing atomic state 000000001f7f4c60 [20772.927039] [drm:drm_mode_object_put.part.0] OBJ ID: 115 (3) [20772.927043] [drm:drm_mode_object_put.part.0] OBJ ID: 116 (7) [20772.927046] [drm:drm_mode_object_put.part.0] OBJ ID: 116 (6) [20772.927048] [drm:drm_mode_object_put.part.0] OBJ ID: 116 (5) [20772.927052] [drm:drm_mode_object_put.part.0] OBJ ID: 109 (5) [20772.927055] [drm:drm_mode_object_put.part.0] OBJ ID: 109 (4) [20772.927059] i915 0000:00:02.0: [drm:__drm_atomic_state_free] Freeing atomic state 000000001f7f4c60 [20772.927064] i915 0000:00:02.0: [drm:drm_ioctl] comm="kwin_wayland", pid=1082, ret=-22 ...which is unfortunately completely useless, the driver doesn't log why it fails the commit. If it's new with 6.5.4 and happens pretty consistently, can you bisect?
Same error as soon as the screen dims, stuck at about 50% brightness. screen and keyborad frozen linux is not completely frozen because the file server is still working. These two lines are repeated 100 times : Dec. 10 11:14:15 bmax kwin_wayland[633]: atomic commit failed: Invalid argument ## THE FIRST AND INIQUE Dec. 10 11:14:28 bmax kwin_wayland[633]: Atomic modeset test failed! Invalid argument Dec. 10 11:14:28 bmax kwin_wayland[633]: Setting dpms mode failed! Dec. 10 11:14:28 bmax kwin_wayland[633]: Atomic modeset test failed! Invalid argument Dec. 10 11:14:28 bmax kwin_wayland[633]: Setting dpms mode failed!... Dec. 10 11:55:24 bmax kwin_wayland[633]: Setting dpms mode failed! Dec. 10 11:55:24 bmax kwin_wayland[633]: Atomic modeset test failed! Invalid argument archlinux, since new 6.5.4 wayland, pcu: intel M100 with Intel Alder Lake-N driver: i915, no gpu
Created attachment 187510 [details] systemctl
I have a similar problem : I use Arch Linux, Plasma 6.5.4-1, AMD Radeon RX7600 graphic card : - randomly when the screen enter to standby mode then I can not exit standby mode, I have to unplug the HDMI cable and plug back in order to have to exit the standby mode, when the bug occurs I can read this message on journalctl : Le bug est confirmé : https://bugs.kde.org/show_bug.cgi?id=512511 dans journalctl je peux voir le message quand le bug se produit : kwin_wayland[1662]: Atomic modeset test failed! Argument invalide
Same issue here using CachyOS (Arch) Prior to the upgrade to KDE 6.5.4 sleep worked flawlessly, but since the upgrade if I allow the OS to turn off the screen I am unable to bring the screen back up. This includes trying to bring up a TTY, so the only way to recover is a forced reset. Have disabled sleep and turn off screen till we get a solution to this. KDE Plasma 6.5.4 KDE Framework 6.20 Wayland Kernel 6.18 Intel Core i9-10850k @5.2GHz 32 gigs ram AMD RX 6900XT GPU
I have the same problem in CachyOS, although with this error message in PowerDevil: 11/12/25 17:58 org_kde_powerdevil Error detecting VCP version using VCP feature xDF: Error_Info[EIO in ddc_write_read_with_retry, causes: DDCRC_READ_ALL_ZERO, DDCRC_DDC_DATA(2), EIO] 11/12/25 17:58 org_kde_powerdevil [ 1535] Error detecting VCP version using VCP feature xDF: Error_Info[EIO in ddc_write_read_with_retry, causes: DDCRC_READ_ALL_ZERO, DDCRC_DDC_DATA(2), EIO] I will upload the log in the next post (Log_2025-12-11.log). Here are my system details: Operating System: CachyOS Linux KDE Plasma Version: 6.5.4 KDE Frameworks Version: 6.20.0 Qt Version: 6.10.1 Kernel Version: 6.18.0-5-cachyos (64-bit) Graphics Platform: Wayland Processors: 16 × AMD Ryzen 7 8845HS w/ Radeon 780M Graphics Memory: 64 GiB of RAM (59,7 GiB usable) Graphics Processor: AMD Radeon 780M Graphics Manufacturer: GMKtec Product Name: NucBox K8 Plus
Created attachment 187530 [details] CachyOS log of 2025-12-11.
*** Bug 513226 has been marked as a duplicate of this bug. ***
Just to clarify something about this problem, because I think I misunderstood it. The problem doesn't occur when it tries to enter sleep mode, but rather as soon as the screen automatically dims. As soon as the screen enters that state, the system stops responding (or at least the screen does). The keyboard remains active; I can toggle the Caps Lock key on and off, but the screen no longer reacts, and the system doesn't go into sleep mode by double-pressing the power button. The only option is to restart.
Confirmed, this is an issue with the point at which the screen tries to go into lock mode and dims first. The dim hangs half way, and then the display looks frozen until it's power cycled, at which point it comes back to a bright lock screen. Operating System: EndeavourOS KDE Plasma Version: 6.5.4 KDE Frameworks Version: 6.20.0 Qt Version: 6.10.1 Kernel Version: 6.17.9-arch1-1 (64-bit) Graphics Platform: Wayland Processors: 16 × AMD Ryzen 7 7800X3D 8-Core Processor Memory: 64 GiB of RAM (62.4 GiB usable) Graphics Processor: AMD Radeon RX 9070 XT Manufacturer: Gigabyte Technology Co., Ltd. Product Name: X670 GAMING X AX
I've verified that the problem also occurs when the screen automatically turns off (I tried disabling automatic screen dimming and leaving only automatic shutdown enabled, and the same issue persisted). Therefore, it seems to be happening in two situations: - When the screen automatically dims. - When the screen automatically turns off. Nate, please, when you find a solution to this, don't wait until KDE 6.5.5 to fix it. We can't keep working like this for a month until the new version is released. Please release a hotfix that includes the affected packages.
Wanted to add that several users are reporting that they have 'fixed' this issue by reverting (only) Kwin to the previous version.
Confirming that this issue happens with me too. Downgrading to Kwin 6.5.3 works (keeping other packages on 6.5.4) Issue is now widespread so hope that this will be taken up on priority. Using Arch Linux
The issue can also be triggered (sometimes) while the lock screen is active and the ESC key is pressed.
Seems like this doesn't occur every single time the screen dims, sometimes it freezes and sometimes it doesn't.
I just installed the new version KWin 6.5.4-2.1, which was recently released, and it seems to have solved the problem, at least in my case. I'm waiting for confirmation from other users.
*** Bug 511118 has been marked as a duplicate of this bug. ***
A possibly relevant merge request was started @ https://invent.kde.org/plasma/kwin/-/merge_requests/8560
*** Bug 513188 has been marked as a duplicate of this bug. ***
(In reply to Chema from comment #20) > I just installed the new version KWin 6.5.4-2.1, which was recently > released, and it seems to have solved the problem, at least in my case. I'm > waiting for confirmation from other users. I also just updated kwin (6.5.4-1 -> 6.5.4-2) on Arch, and the problem seems to be gone. Thanks!
A possibly relevant merge request was started @ https://invent.kde.org/plasma/kwin/-/merge_requests/8562
Git commit 30a7f52841dd1c449965d0c9d5cd4915697ab129 by Xaver Hugl. Committed on 12/12/2025 at 13:59. Pushed by zamundaaa into branch 'master'. core/sessions: don't take ownership of an fd that Qt will close QDBusUnixFileDescriptor closes the file descriptor in its destructor, if we want to take ownership of it, we need to duplicate the file descriptor. If we don't duplicate it, but directly take "ownership" of the file descriptor, the following sequence of events can happen: 1. ~QDBusUnixFileDescriptor closes file descriptor A 2. some new fd B is allocated, which happens to have the same handle as fd A 3. ~FileDescriptor closes fd A, but it has the same handle as fd, so it actually closes fd B! 4. whenever fd B is used, we run into problems To make such issues easier to debug in the future, this also adds debug logging for when closing a file descriptor fails. M +1 -1 src/core/session_consolekit.cpp M +1 -1 src/core/session_logind.cpp M +19 -3 src/utils/filedescriptor.cpp https://invent.kde.org/plasma/kwin/-/commit/30a7f52841dd1c449965d0c9d5cd4915697ab129
Git commit ef4504320de2c3a7c7aebcf083d75db361f802ae by Xaver Hugl. Committed on 12/12/2025 at 14:11. Pushed by zamundaaa into branch 'Plasma/6.5'. core/sessions: don't take ownership of an fd that Qt will close QDBusUnixFileDescriptor closes the file descriptor in its destructor, if we want to take ownership of it, we need to duplicate the file descriptor. If we don't duplicate it, but directly take "ownership" of the file descriptor, the following sequence of events can happen: 1. ~QDBusUnixFileDescriptor closes file descriptor A 2. some new fd B is allocated, which happens to have the same handle as fd A 3. ~FileDescriptor closes fd A, but it has the same handle as fd, so it actually closes fd B! 4. whenever fd B is used, we run into problems To make such issues easier to debug in the future, this also adds debug logging for when closing a file descriptor fails. (cherry picked from commit 30a7f52841dd1c449965d0c9d5cd4915697ab129) M +1 -1 src/core/session_consolekit.cpp M +1 -1 src/core/session_logind.cpp M +19 -3 src/utils/filedescriptor.cpp https://invent.kde.org/plasma/kwin/-/commit/ef4504320de2c3a7c7aebcf083d75db361f802ae
The commit will only be in Plasma 6.5.5, but I sent an email requesting distributions to backport it immediately as well.
(In reply to Chema from comment #20) > I just installed the new version KWin 6.5.4-2.1, which was recently > released, and it seems to have solved the problem, at least in my case. I'm > waiting for confirmation from other users. I just installed the updated package on Arch and switched back to Wayland. This appears to have resolved the issue.
Installed the updated kwin 6.5.4-2.1 and it does allow the PC to wake from sleep now. It does not however allow the video to recover if I set KDE to dim and then turn off the monitor. I have to unplug the HDMI to regain video (6900XT) when this happens. Till this is resolved I have disabled both dim and turn off monitor in KDE settings.
I've set "Turn off screen" to "Never" so far it's okay.
With kwin 6.5.4-3.1 everything seems to be working well again. Will let it sleep tonight with dim, and turn screen off enabled.