[SOLVED] Freezing in graphical DEs when playing media

SamuraiGelato · 2026-06-12 23:53:10

Hello Arch community,

I am getting freezes/DE crashes where the system becomes unresponsive. Audio and video might keep playing for a moment or so, but shortly after the freeze, the screen will go black. Sometimes my cursor is still responsive, but if I attempt to switch to another tty by CTRL+ALT+F6 then the system will be entirely frozen. I have the magic sysrq keys enabled, so at that point I usually do an ALT + SYSRQ+r,e,i,s,u,b

Alas, I have not been successful in collecting logs from these crashes. I would appreciate guidance on resolving this issue. I have gathered the relevant information below in a way that I hope is useful. I am trying to determine if it is a filesystem corruption, hardware/kernel issue, or something else.

Timeline / wall of text

I had a previous desktop computer on the Fedora KDE spin (and then briefly CachyOS) which this current machine has replaced. It had occasional stability issues and, at the time, I chalked most of it up to the Nvidia 1080ti card. The only hardware carried over to the new system is my monitor, keyboard, mouse, and usb/wireless headset.

On the new desktop machine, mistakes were made, and I got some pretty bad file corruption. I think this was due to an unsafe shutdown after a freeze during a package install. It was bad enough that I reformatted the drive, and started over. On the second install, I made sure to enable the magic sysrq keys in the kernel parameters.

Things went well until I was trying to build a container with Podman to run a Mindwtr sync service. I was getting crashes which, at the time, I thought were related to podman. I eventually gave up on Podman and built the container and ran it in Docker. Shortly afterwards, when I joined some friends for a movie night via Discord, I ran into another freeze. I was up using the AUR Vesktop package instead of the official Discord client, and it worked great, until everything froze about 30 minutes into the movie.

I don't remember exactly what I did afterwards, but the problem got worse. GNOME was freezing immediately after login. I tried restoring to some older btrfs snapshots, but it did not help. I was able to switch to tty from gdm instead of starting a gnome session. and disable the Docker service, but that did not fix the problem. I then uninstalled all of my GNOME packages, and related packages, and installed KDE Plasma. Since then, the system has been stable unless I have some media playing. e.g.after 10 to 60 minutes with youtube playing in Firefox, it will freeze as described above.

I took an embarrassingly long time typing this up, all without any chill beats to relax/chill to. And it has not crashed during this time. It really seems correlated with playing videos or perhaps having my wireless headset powered on and connected.

Troubleshooting attempts

Ran memtest86 and an nvme test from the motherboard's UEFI settings. No errors detected.

I booted to my live archiso USB and ran

# btrfs check --readonly /dev/nvme0n1

It did find some issues. I will have to boot to that environment again to get those specific error logs. btrfs scrub, on the other hand, detected no issues. (I will edit this post shortly once I get that output saved)
Edit: Darn, after this last boot to get the btrfs check output, KDE Plasma is also crashing right after login, just like GNOME was.

I wanted to get journal logs from when these freezes happen, to see if it is indeed a filesystem issue, or if it is something else. So, I tried a couple of AUR kdump solutions from the wiki. kdumpst didn't work, so I tried simple-kdump, and I was able to generate crash logs from a test crash via

# sync; echo c > /proc/sysrq-trigger

Unfortunately, today when it froze, I did not get a kdump log. After it was frozen at the black screen for a while, I attempted ALT+SYSRQ+r,s,c but It did not automatically reboot into the crashkernel, or if it did, something went wrong, as all I had was an unresponsive system, so I held down the power button to power it off and then booted it back up.

System Info

inxi -ez

Kernel Version: 7.0.11-arch1-1 (64-bit)
Graphics Platform: Wayland
Processors: 12 × AMD Ryzen 5 7500X3D 6-Core Processor
Memory: 16 GiB of RAM (14.7 GiB usable)
Graphics Processor: AMD Radeon RX 9070 XT

simple-kdump

$ cat /proc/cmdline
BOOT_IMAGE=/vmlinuz-linux root=UUID=acc44320-1a27-4adb-a8ec-80c1071e995f rw rootflags=subvol=@ crashkernel=512M loglevel=3 sysrq_always_enabled=1 quiet splash

$ cat /etc/conf.d/simple-kdump.conf 
# Kernel and initramfs for the kexec environment.
# Recommended to use the linux or linux-lts kernel with 'default' preset.
KERNEL=/boot/vmlinuz-linux
INITRAMFS=/boot/initramfs-linux.img

# No crashkernel= option for the kexec kernel cmdline.
# Just regular boot options, the extra needed ones will be added by
# simple-kdump-setup service automatically.
BOOT_OPTIONS=root=UUID=acc44320-1a27-4adb-a8ec-80c1071e995f rw rootflags=subvol=@ loglevel=5 video=2560x1440@59.94e

Logs

journalctl -b -1 -n 400 -W
journalctl -b -1 --dmesg -n 400 -W
cat /var/log/pacman.log

EDIT2:

I tried to pipe the output from btrfs check --readonly /dev/nvme0n1p2 to a file, but it didn't give all the output

Opening filesystem to check...
Checking filesystem on /dev/nvme0n1p2
UUID: acc44320-1a27-4adb-a8ec-80c1071e995f
found 205445181440 bytes used, error(s) found
total csum bytes: 188301852
total tree bytes: 3488694272
total fs tree bytes: 3162406912
total extent tree bytes: 124125184
btree space waste bytes: 506990402
file data blocks allocated: 490399457280
 referenced 349471985664

I feel like my fstab may be relevant too:

# Static information about the filesystems.
# See fstab(5) for details.

# <file system> <dir> <type> <options> <dump> <pass>
# /dev/nvme0n1p2 LABEL=ArchLinuxFS
UUID=af67c079-2f5a-49f5-b4b0-309cf5357017	/         	btrfs     	rw,relatime,ssd,discard=async,space_cache=v2,subvol=/@	0 0

# /dev/nvme0n1p1
UUID=1DD0-54C5      				/boot     	vfat      	rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,utf8,errors=remount-ro	0 2

UUID=af67c079-2f5a-49f5-b4b0-309cf5357017	/home         	btrfs     	rw,relatime,ssd,discard=async,space_cache=v2,subvol=/@home	0 0

UUID=af67c079-2f5a-49f5-b4b0-309cf5357017	/swap        	btrfs     	rw,relatime,ssd,discard=async,space_cache=v2,subvol=/@swap	0 0


/swap/swapfile					none      	swap      	defaults 	0 0

I am starting to think maybe btrfs was a bad choice for me. But it works so nicely on my laptop. I have Arch on there and it has been a good time. I am at least able to get to a tty, so I am willing to try to find out what went wrong instead of wiping and starting over again. If you are reading all my nonsense here, let me know if you have any constructive thoughts

Last edited by SamuraiGelato (2026-06-17 15:27:05)

seth · 2026-06-13 07:31:45

Trigger the problem, see whether you can still ssh into the system, sysrq+reisub and then post your complete system journal for the boot:

sudo journalctl -b -1 | curl -s -H "Accept: application/json, */*" --upload-file - 'https://paste.c-net.org/'

(No random filters)

From the last journal post

Jun 12 11:10:02 kernel: amdgpu 0000:03:00.0: VM memory stats for proc kscreenlocker_g(5763) task kscreenloc:cs0(5744) is non-zero when fini
Jun 12 11:20:19 kernel: amdgpu 0000:03:00.0: VM memory stats for proc (0) task (0) is non-zero when fini
Jun 12 11:20:19 kernel: amdgpu 0000:03:00.0: VM memory stats for proc firefox(4847) task firefox:cs0(4785) is non-zero when fini

but missing the context it's not clear whether this is indicating anything (or eg. artifact of the reboot)

SamuraiGelato · 2026-06-13 16:34:11

I haven't been able to ssh in once the issue problem occurs, but here is what I got while attempting this over an active ssh connection and triggering the problem:

~]# journalctl -b -1 | curl -s -H "Accept: application/json, */*" --upload-file - 'https://paste.c-net.org/'
-bash: sed: command not found
-bash: curl: command not found
-bash: /usr/bin/journalctl: Input/output error
-bash: sed: command not found
[root@pspec-arch-26 ~]# dmesg
-bash: sed: command not found
-bash: dmesg: command not found
-bash: sed: command not found
[root@pspec-arch-26 ~]# pwd
-bash: sed: command not found
/root
-bash: sed: command not found

(I had done sudo -i)

Seems like the ssh connection is still up, but the shell is borked.

Here are a couple of literal screenshots of the console when I do the sysrq
https://ibb.co/7xzb0CDF
https://ibb.co/DPdNh7R4

As far as those journals, the one thing I would note is that the unfiltered log ends about an hour after the last dmesg log.

I was looking into how I might otherwise try to get these logs:

My motherboard has a JCOM1 connector on it. I could in theory get serial console going, but I would need to purchase the appropriate port to connect to it. Or maybe I could try setting up the systemd-journal-upload.service to forward to my laptop.

Last edited by SamuraiGelato (2026-06-13 16:38:16)

seth · 2026-06-13 18:57:58

Why does sed show up?
But it sounds you're losing the disk…

=> https://wiki.archlinux.org/title/SMART and keep "dmesg -W" running while waiting for the incident (via ssh or locally visible terminal)

**Might** be caused by APST, https://wiki.archlinux.org/title/Solid_ … leshooting
"nvme_core.default_ps_max_latency_us=0 pcie_aspm=off iommu=soft", https://wiki.archlinux.org/title/Kernel_parameters

SamuraiGelato · 2026-06-16 15:18:22

Ok good, running "dsmesg -W" I can at least get these logs out of the ssh terminal.

https://paste.c-net.org/RelativeWreath

looks like the very first thing once I finally could capture logs it it recommending that kernel parameter workaround.

[  806.963157] sysrq: Changing Loglevel
[  806.963165] sysrq: Loglevel set to 7
[  936.483510] nvme nvme0: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0x10
[  936.483516] nvme nvme0: Does your device have a faulty power saving mode enabled?
[  936.483518] nvme nvme0: Try "nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off" and report a bug

I haven't tried that kernel parameter yet, but I did update the firmware for my nvme drive, a WD Blue SN580, from 281040WD to 281070WD. Unforunately that didn't seem to make a difference, so I will try the kernel parameter, and report back.

Last edited by SamuraiGelato (2026-06-16 15:32:02)

SamuraiGelato · 2026-06-16 16:14:29

Good news! Disabling "ASPM Control for CPU PCIe" in the UEFI/BIOS firmware (and rebooting) settings seems to have cleared the problem!

I will keep using it for a bit while listening to some music on Youtube to see if it remains stable.

Setting the kernel parameters alone didn't work:

[roscoe@pspec-arch-26 ~]$ cat /proc/cmdline
BOOT_IMAGE=/vmlinuz-linux root=UUID=acc44320-1a27-4adb-a8ec-80c1071e995f rw rootflags=subvol=@ nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off crashkernel=512M loglevel=3 sysrq_always_enabled=1 quiet splash
[roscoe@pspec-arch-26 ~]$ sudo dmesg -W
[  244.848549] nvme nvme0: using unchecked data buffer
[  244.852423] block nvme0n1: No UUID available providing old NGUID
[  246.493131] warning: `kdeconnectd' uses wireless extensions which will stop working for Wi-Fi 7 hardware; use nl80211
[  298.015506] nvme nvme0: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0x10
[  298.015514] nvme nvme0: Does your device have a faulty power saving mode enabled?
[  298.015517] nvme nvme0: Try "nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off" and report a bug

full dmesg output: https://paste.c-net.org/GermanManagers

Should this be reported as an instance of this type of bug as mentioned in that last quoted dmesg line?

seth · 2026-06-16 20:19:00

Setting the kernel parameters alone didn't work:

Are they still required?
What does a general journal currently look like (nvme failure apart, more the overall ASPM situation)
Please post your complete system journal for the boot:

sudo journalctl -b | curl -s -H "Accept: application/json, */*" --upload-file - 'https://paste.c-net.org/'

Should this be reported as an instance of this type of bug as mentioned in that last quoted dmesg line?

Not if it's entirely driven by the UEFI setting, in that case you should rather look for a UEFI update.

SamuraiGelato · 2026-06-16 22:36:40

Here you go:
https://paste.c-net.org/RevolveBrava
I removed those kernel parameters, and things still seem to be working alright.

seth · 2026-06-16 22:48:51

Jun 16 18:30:08 archlinux kernel: acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI EDR HPX-Type3]

Looks like ASPM now got enabled (by the UEFI setting?)

Jun 16 18:30:08 archlinux kernel: pci 0000:06:00.0: broken device, retraining non-functional downstream link at 2.5GT/s
Jun 16 18:30:08 archlinux kernel: pci 0000:06:00.0: retraining failed
Jun 16 18:30:08 archlinux kernel: pci 0000:06:00.0: PME# supported from D0 D3hot D3cold
Jun 16 18:30:08 archlinux kernel: pci 0000:06:01.0: [1022:43f5] type 01 class 0x060400 PCIe Switch Downstream Port
Jun 16 18:30:08 archlinux kernel: pci 0000:06:01.0: PCI bridge to [bus 08]
Jun 16 18:30:08 archlinux kernel: pci 0000:06:01.0: enabling Extended Tags
Jun 16 18:30:08 archlinux kernel: pci 0000:06:01.0: broken device, retraining non-functional downstream link at 2.5GT/s
Jun 16 18:30:08 archlinux kernel: pci 0000:06:01.0: retraining failed
Jun 16 18:30:08 archlinux kernel: pci 0000:06:01.0: PME# supported from D0 D3hot D3cold
Jun 16 18:30:08 archlinux kernel: pci 0000:06:02.0: [1022:43f5] type 01 class 0x060400 PCIe Switch Downstream Port
Jun 16 18:30:08 archlinux kernel: pci 0000:06:02.0: PCI bridge to [bus 09]
Jun 16 18:30:08 archlinux kernel: pci 0000:06:02.0: enabling Extended Tags
Jun 16 18:30:08 archlinux kernel: pci 0000:06:02.0: broken device, retraining non-functional downstream link at 2.5GT/s
Jun 16 18:30:08 archlinux kernel: pci 0000:06:02.0: retraining failed

But the nvme is on 0000:04:00.0

Jun 16 18:30:08 archlinux kernel: nvme 0000:04:00.0: platform quirk: setting simple suspend

When looking at an older journal from before the UEFI change, are there any (other) aspm related messages?

SamuraiGelato · 2026-06-16 23:19:14

Output of "sudo journalctl -b -30 | grep -i --context 20 "aspm""
https://paste.c-net.org/KubelikAppeal
Seems like that same line, not sure what exactly it means

same thing if I just go back 6 boots
https://paste.c-net.org/RatchetMeteor

Output of "sudo journalctl -b -30 | grep -i "kernel: pci""
https://paste.c-net.org/ShamblesBreast

Looking into firmware updates, fwupd doesn't seem to have an update. But I expect I can get one from MSI.

I suppose I could have gone into the pre-installed Windows on this machine before installing Arch and applied firmware updates through the supported OS, but..

seth · 2026-06-17 06:49:16

Jun 12 20:11:06 archlinux kernel: acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug SHPCHotplug PME AER PCIeCapability LTR DPC]

is gone.

Does adding this entire list to the kernel parameters allow you to flip back the UEFI setting?
"nvme_core.default_ps_max_latency_us=0 pcie_aspm=off iommu=soft pcie_ports=compat pcie_port_pm=off"

SamuraiGelato · 2026-06-17 07:22:35

I just finished updating the UEFI firmware, and I removed all those related kernel parameters. Here is the journal from this boot: https://paste.c-net.org/TonaneGlance

Here is the firmware page:
https://us.msi.com/Motherboard/PRO-B850 … 6E/support

One of the recent updates included the note:

Improved PCIe-based devices compatibility under certain CPU or system combinations.

I seem to be up and running. Will confirm with further testing when I have some more time. So far so good.

SamuraiGelato · 2026-06-17 15:31:45

I believe this to be solved. Thank you for your help, Seth.

Arch Linux

#1 2026-06-12 23:53:10