Linux stuck/frozen after shutdown

Vernox · 2024-07-14 02:04:32

Hello, since I did a clean install of linux (kde reasons) back in march I am experiencing shutdown issues.

When trying to shutdown (either through "sudo shutdown" or kde's shutdown command), it doesn't cut the power.
The log shows the entire shutdown sequence succeeding (unmounting discs, killing services...), but at the end, it just stays powered-on (SysReq not working; CapsLock not working (mouse is not powered)).

Unfortunately, it's also like 2/3 of the time. Sometimes I change something in the bios and have like 3 consecutive days of no problems (maybe insanely unlucky), and then it happens again.
Additionally, it only happens when the system has been running for at least 8 hours. So just turning it on and off doesn't invoke anything.

I've already gone through a journalctl output with and one without the issue and there weren't any significant differences. (just discord being annoying and notification sounds)
The last few lines always look like:

Jun 27 04:02:23 pc systemd[1]: Shutting down.
Jun 27 04:02:24 pc systemd-shutdown[1]: Syncing filesystems and block devices.
Jun 27 04:02:24 pc systemd-shutdown[1]: Sending SIGTERM to remaining processes...
Jun 27 04:02:24 pc systemd-journald[299]: Received SIGTERM from PID 1 (systemd-shutdow).
Jun 27 04:02:24 pc systemd-journald[299]: Journal stopped
░░ Subject: The journal has been stopped
░░ Defined-By: systemd
░░ Support: https://lists.freedesktop.org/mailman/listinfo/systemd-devel
░░ 
░░ The system journal process has shut down and closed all currently
░░ active journal files.

I never reboot or put my system to sleep.

Other info:
Kernel: Linux 6.9.9-arch1-1
DE: KDE Plasma 6.1.2
Hardware: relatively old; ~7 years; nvidia gpu
Kernel parameters:

initrd=intel-ucode.img rw add_efi_memmap initrd=initramfs-%v.img nvidia nvidia-drm.modeset=1 nvidia_drm.fbdev=1 rootfstype=ext4

Kernel modules:

ahci nvidia nvidia_modeset nvidia_uvm nvidia_drm

Other forum posts with a similar problem often feature a bluetooth adapter that led to the same issue, but I don't have one.
Tomorrow I'll try running linux-lts, although I think I've already tried it. (EDIT: linux-lts also has the same problem)
I would really appreciate any hint / suggestion on how to fix this.

Last edited by Vernox (2024-07-15 12:52:38)

V1del · 2024-07-14 21:16:23

Somewhat of a known bug with nvidia drivers in version 550xx 555xx (especially if they are placed into the initramfs, maybe removing them there would already help...). You could try nvidia-535xx-dkms from the AUR which apparently has none of these issues, though you will regress on xwayland explicit sync support, should that be something you're currently using.

Vernox · 2024-07-15 00:45:35

V1del wrote:

Somewhat of a known bug with nvidia drivers in version 550xx 555xx (especially if they are placed into the initramfs, maybe removing them there would already help...). You could try nvidia-535xx-dkms from the AUR which apparently has none of these issues, though you will regress on xwayland explicit sync support, should that be something you're currently using.

What exactly do you mean by "removing them from the initramfs"?
Should I remove all the nvidia stuff from "MODULES=()" in "mkinitcpio.conf"?

I use X11, so regress in wayland wouldn't concern me.

Thank you so much for your help

Vernox · 2024-07-18 01:50:14

@V1del and anyone else who is possibly reading this

According to your suggestion, I have "MODULES=(ahci)" set and switched to nvidia-535xx-dkms, but that doesn't affect the error at all.

Maybe further information like this could be relevant:

GPU: nvidia 1080Ti
CPU: intel 7700k
Ram: 32GB
Boot Drive: Samsung 960 PRO
Boot Loader: refind

I'm really appreciate your help. This problem is a really big burden.

Last edited by Vernox (2024-07-18 01:59:51)

seth · 2024-07-18 06:34:07

Sometimes I change something in the bios and have like 3 consecutive days of no problems (maybe insanely unlucky), and then it happens again.
Additionally, it only happens when the system has been running for at least 8 hours. So just turning it on and off doesn't invoke anything.

Do you have a parallel windows installation?
Does it reliably work if you log out of KDE and shut down the system from SDDM?
Please post your complete system journal for a boot that failed to power off, eg.

sudo journalctl -b -1 | curl -F 'file=@-' 0x0.st

for the previous one.
The nvme could be a problem, https://wiki.archlinux.org/title/Solid_ … leshooting - disable APST and use "iommu=soft"

Vernox · 2024-07-19 01:47:24

seth wrote:

Do you have a parallel windows installation?

No. Arch Linux is the only bootable distro on my computer. (aside from VMs)

seth wrote:

Does it reliably work if you log out of KDE and shut down the system from SDDM?

Never tried it. Will try it. Good idea. EDIT: does not fix the problem

seth wrote:

Please post your complete system journal for a boot that failed to power off, eg.
sudo journalctl -b -1 | curl -F 'file=@-' 0x0.st
for the previous one.

https://0x0.st/X9yF.txt

seth wrote:

The nvme could be a problem, https://wiki.archlinux.org/title/Solid_ … leshooting - disable APST and use "iommu=soft"

Changes applied.

Posting this just to show that this issue is still active.
Results will come in the next few days (maybe a little longer).

Thank you very much for your assistance!

Last edited by Vernox (2024-07-19 02:35:49)

seth · 2024-07-19 07:00:08

plasma crashes out of the session, but eventually

Jul 18 03:41:49 pc. systemd[1]: systemd-poweroff.service: Deactivated successfully.
Jul 18 03:41:49 pc. systemd[1]: Finished System Power Off.
Jul 18 03:41:49 pc. systemd[1]: Reached target System Power Off.
Jul 18 03:41:49 pc. systemd[1]: Shutting down.
Jul 18 03:41:49 pc. systemd-shutdown[1]: Syncing filesystems and block devices.
Jul 18 03:41:49 pc. systemd-shutdown[1]: Sending SIGTERM to remaining processes...
Jul 18 03:41:49 pc. systemd-journald[281]: Received SIGTERM from PID 1 (systemd-shutdow).
Jul 18 03:41:49 pc. systemd-journald[281]: Journal stopped

so the system tries to power off

Jul 17 14:12:09 archlinux kernel: DMI: MSI MS-7A63/Z270 GAMING PRO CARBON (MS-7A63), BIOS 1.90 07/03/2018

Hardware: relatively old; ~7 years; nvidia gpu

I assume that's the latest available BIOS version for the board?

You're already running the 535xx nviida drivers, so that's unlikely the cause. Next random contender would probably be the nvme, try to disable APST and use the software IOMMU, https://wiki.archlinux.org/title/Solid_ … leshooting

Vernox · 2024-07-20 02:14:30

Hello

seth wrote:

Jul 17 14:12:09 archlinux kernel: DMI: MSI MS-7A63/Z270 GAMING PRO CARBON (MS-7A63), BIOS 1.90 07/03/2018
Hardware: relatively old; ~7 years; nvidia gpu
I assume that's the latest available BIOS version for the board?

That is correct.

seth wrote:

You're already running the 535xx nviida drivers, so that's unlikely the cause. Next random contender would probably be the nvme, try to disable APST and use the software IOMMU, https://wiki.archlinux.org/title/Solid_ … leshooting

I applied those changes, but that didn't resolve the issue, nor did it change anything in the systemd journal (messages containing "nvme").

What a bummer

seth · 2024-07-20 06:15:41

There're no nvme related *errors* in that journal anyway.
The system properly shuts down but the acpi message for the BIOS to cut power doesn't work™

What is there is this here

Jul 18 03:41:31 pc. systemd[1355]: Created slice Slice /app/dbus-:1.2-org.kde.LogoutPrompt.
Jul 18 03:41:31 pc. systemd[1355]: Started dbus-:1.2-org.kde.LogoutPrompt@0.service.
Jul 18 03:41:31 pc. systemd[1]: Starting PackageKit Daemon...
Jul 18 03:41:31 pc. PackageKit[1457319]: daemon start
Jul 18 03:41:31 pc. systemd[1]: Started PackageKit Daemon.
Jul 18 03:41:32 pc. PackageKit[1457319]: get-updates transaction /1_caeeebbe from uid 1000 finished with success after 198ms
Jul 18 03:41:34 pc. systemd[1355]: dbus-:1.2-org.kde.LogoutPrompt@0.service: Main process exited, code=exited, status=1/FAILURE
Jul 18 03:41:34 pc. systemd[1355]: dbus-:1.2-org.kde.LogoutPrompt@0.service: Failed with result 'exit-code'.
Jul 18 03:41:34 pc. systemd[1355]: Started kitty - Terminal emulator.
Jul 18 03:41:35 pc. kitty[1457361]: [0.267] The output buffer does not support sRGB color encoding, colors will be incorrect.
Jul 18 03:41:35 pc. systemd[1355]: Started kitty child process: 1457365 launched by: 1457361.
Jul 18 03:41:36 pc. sudo[1457428]: pam_systemd_home(sudo:account): New sd-bus connection (system-bus-pam-systemd-home-1457428) opened.
Jul 18 03:41:36 pc. systemd-logind[597]: The system will power off now!

"dbus-:1.2-org.kde.LogoutPrompt@0.service: Failed with result 'exit-code'" has shown up in several other threads, but typically prevented a complete shutdown.
Do you have a journal where you first log out of KDE (to get that out of the way) and then shut down the system?

Vernox · 2024-07-21 01:47:04

Hello

seth wrote:

There're no nvme related *errors* in that journal anyway.
The system properly shuts down but the acpi message for the BIOS to cut power doesn't work™

What is there is this here

Jul 18 03:41:31 pc. systemd[1355]: Created slice Slice /app/dbus-:1.2-org.kde.LogoutPrompt.
Jul 18 03:41:31 pc. systemd[1355]: Started dbus-:1.2-org.kde.LogoutPrompt@0.service.
Jul 18 03:41:31 pc. systemd[1]: Starting PackageKit Daemon...
Jul 18 03:41:31 pc. PackageKit[1457319]: daemon start
Jul 18 03:41:31 pc. systemd[1]: Started PackageKit Daemon.
Jul 18 03:41:32 pc. PackageKit[1457319]: get-updates transaction /1_caeeebbe from uid 1000 finished with success after 198ms
Jul 18 03:41:34 pc. systemd[1355]: dbus-:1.2-org.kde.LogoutPrompt@0.service: Main process exited, code=exited, status=1/FAILURE
Jul 18 03:41:34 pc. systemd[1355]: dbus-:1.2-org.kde.LogoutPrompt@0.service: Failed with result 'exit-code'.
Jul 18 03:41:34 pc. systemd[1355]: Started kitty - Terminal emulator.
Jul 18 03:41:35 pc. kitty[1457361]: [0.267] The output buffer does not support sRGB color encoding, colors will be incorrect.
Jul 18 03:41:35 pc. systemd[1355]: Started kitty child process: 1457365 launched by: 1457361.
Jul 18 03:41:36 pc. sudo[1457428]: pam_systemd_home(sudo:account): New sd-bus connection (system-bus-pam-systemd-home-1457428) opened.
Jul 18 03:41:36 pc. systemd-logind[597]: The system will power off now!

"dbus-:1.2-org.kde.LogoutPrompt@0.service: Failed with result 'exit-code'" has shown up in several other threads, but typically prevented a complete shutdown.
Do you have a journal where you first log out of KDE (to get that out of the way) and then shut down the system?

There you go: https://0x0.st/Xpiv.log

seth · 2024-07-21 07:53:38

Jul 21 03:19:14 pc. systemd[1347]: Created slice Slice /app/dbus-:1.2-org.kde.LogoutPrompt.
Jul 21 03:19:14 pc. systemd[1347]: Started dbus-:1.2-org.kde.LogoutPrompt@0.service.
Jul 21 03:19:18 pc. systemd[1347]: dbus-:1.2-org.kde.LogoutPrompt@0.service: Main process exited, code=exited, status=1/FAILURE
Jul 21 03:19:18 pc. systemd[1347]: dbus-:1.2-org.kde.LogoutPrompt@0.service: Failed with result 'exit-code'.
Jul 21 03:19:44 pc. systemd[1347]: Removed slice Slice /app/dbus-:1.2-org.kde.LogoutPrompt.

is still there and so is

Jul 21 03:19:44 pc. sddm[684]: Message received from greeter: PowerOff
Jul 21 03:19:44 pc. sddm-helper[26191]: Signal received: SIGTERM
Jul 21 03:19:44 pc. sddm-greeter-qt6[26206]: Signal received: SIGTERM
Jul 21 03:19:44 pc. systemd[1]: Stopping Session c2 of User sddm...
Jul 21 03:19:44 pc. sddm[684]: Error from greeter session: "Process crashed"
Jul 21 03:19:44 pc. sddm[684]: Auth: sddm-helper (--socket /tmp/sddm-auth-44ce81db-1ea6-4fe4-ba1d-f3f7d8741ad7 --id 4 --start /usr/bin/sddm-greeter-qt6 --socket /tmp/sddm-:0-ESzZsI --theme /usr/share/sddm/themes/breeze --user sddm --greeter) crashed (exit code 1)
Jul 21 03:19:44 pc. sddm[684]: Error from greeter session: "Process crashed"
Jul 21 03:19:44 pc. sddm[684]: Auth: sddm-helper exited with 1
Jul 21 03:19:44 pc. sddm[684]: Greeter stopped. SDDM::Auth::HELPER_AUTH_ERROR

(the greeter crashes on sigterm) but I don't believe this is the cause (I assume the problem exists even if you just boot the multi-user.target and shutdown from there, 2nd linke below)?

wrt cpupower-gui and the powersave profile, what if you don't engage in any measures to limit the CPU and maybe even restrict the c-states, https://wiki.archlinux.org/title/Intel_ … up_from_S3

Alternatively, try to lie to the ACPI,

acpi_osi=! acpi_osi="Windows 2022"

https://learn.microsoft.com/en-us/windo … inacpi-osi

Vernox · 2024-07-22 03:28:01

seth wrote:

I assume the problem exists even if you just boot the multi-user.target and shutdown from there, 2nd linke below?

I can try this if you want. But I doubt I can create the same amount of usage, as I currently do, while doing my day to day tasks.
So I don't believe it would be very conclusive (if it results in a successful a shutdown).
Just judging by how long the computer has to be running for the problem to even occur.

seth wrote:

wrt cpupower-gui and the powersave profile, what if you don't engage in any measures to limit the CPU and maybe even restrict the c-states, https://wiki.archlinux.org/title/Intel_ … up_from_S3

I deactivated all cpupower services and that didn't fix anything.
Whether the "intel_idle.max_cstate=1" affected anything, will reveal itself tomorrow.

seth wrote:

Alternatively, try to lie to the ACPI,
acpi_osi=! acpi_osi="Windows 2022"
https://learn.microsoft.com/en-us/windo … inacpi-osi

Configured. Results will come in about 24 hours.

PS:
And again, Thank you!
You're linking to wiki parts that I wouldn't have found in 100 years. <3

Last edited by Vernox (2024-07-22 03:29:57)

Vernox · 2024-07-24 02:01:09

@seth

So after failing to properly escape spaces in the kernel parameters two days in a row, I finally managed to boot like you recommended. But the same issue persists.

Last edited by Vernox (2024-07-24 02:11:27)

Vernox · 2024-08-05 00:50:32

Update:

Trying the nvidia driver "550.107.02", as suggested in this topic does not fix the issue.
But, there are also good news.

I went back to "nvidia-535xx" and found something very interesting.
When running "sudo shutdown now", the system hangs as usual.
But when I turn the system off using the kde interface, aka not violently kill everything, the shutdown process finally works.
This shouldn't be a fluke, as it has already worked 4 days in a row.

The acpi_osi and iommu kernel parameters did not affect the problem.

Last edited by Vernox (2024-08-06 21:37:33)

Arch Linux

#1 2024-07-14 02:04:32

Linux stuck/frozen after shutdown

#2 2024-07-14 21:16:23

Re: Linux stuck/frozen after shutdown

#3 2024-07-15 00:45:35

Re: Linux stuck/frozen after shutdown

#4 2024-07-18 01:50:14

Re: Linux stuck/frozen after shutdown

#5 2024-07-18 06:34:07

Re: Linux stuck/frozen after shutdown

#6 2024-07-19 01:47:24

Re: Linux stuck/frozen after shutdown

#7 2024-07-19 07:00:08

Re: Linux stuck/frozen after shutdown

#8 2024-07-20 02:14:30

Re: Linux stuck/frozen after shutdown

#9 2024-07-20 06:15:41

Re: Linux stuck/frozen after shutdown

#10 2024-07-21 01:47:04

Re: Linux stuck/frozen after shutdown

#11 2024-07-21 07:53:38

Re: Linux stuck/frozen after shutdown

#12 2024-07-22 03:28:01

Re: Linux stuck/frozen after shutdown

#13 2024-07-24 02:01:09

Re: Linux stuck/frozen after shutdown

#14 2024-08-05 00:50:32

Re: Linux stuck/frozen after shutdown

Board footer