You are not logged in.
Just updated all my system and KDE Plasma simply crashed each time it boots.
Reached to solve this by downgrading the linux firmware thanks to my friend gemini
Anyone else?
Offline
Great, But which firmware are you talking about? the recent AMD chaos?
What are you asking here? Validation from other fellow users? I doubt if that is supported here.
Last edited by 5hridhyan (2026-09-12 18:39:48)
Offline
Great, But which firmware are you talking about? the recent AMD chaos?
I take it, that's a well-known issue by now, then?
Something like https://bbs.archlinux.org/viewtopic.php?pid=2309521?
AFAICT, that thread originally focused on AMD iGPUs and freezes during boot, not login.
But to be fair, someone later wrote in that thread that dGPUs also seem to be affected.
I assume there's probably a few more threads that I overlooked. ![]()
Anyway, while I can't speak for thecatfromouterspace, I'd like to try and make this thread a little more helpful for anyone stumbling upon it. ![]()
First off, I'm on an AMD Radeon RX 7900 XTX and Kernel 7.2.4-arch1-2.
The symptoms are that the system freezes when trying to log into KDE/Wayland from SDDM.
Switching to a TTY instead works, however.
Downgrading all linux-firmware packages also "fixed" the problem for me.
That might have been a bit overkill, but I wanted them all to stay on the same version, since they seem to be versioned together, to avoid any potential mismatch-issues (if that is even possible).
EDIT (2026-09-13):
Seems like the actual issue was ddcutil 3.0.0-1 being broken.
Updating everything, including ddcutil to 3.0.1-1, fixed the issue on my end.
See my post below for more information (thanks to jbreizh for the hint!).
Checking journalctl listed several issues related to the amdgpu driver, stuff like this:
...
kernel: amdgpu 0000:03:00.0: [drm] *ERROR* flip_done timed out
...
kernel: amdgpu 0000:03:00.0: [drm] *ERROR* [PLANE:206:plane-3] commit wait timed out
...
kernel: amdgpu 0000:03:00.0: device lost from bus!
kernel: amdgpu 0000:03:00.0: SMU: bus error for message: DisallowGfxOff(41) response:0xFFFFFFFF
kernel: in params:00000000
kernel: amdgpu 0000:03:00.0: Failed to disable gfxoff!
kwin_wayland[1330]: Pageflip timed out! This is a bug in the amdgpu kernel driver
...... which does not look too nice to put it mildly.
So, after also taking a look at /var/log/pacman.log, I downgraded all linux-firmware* packages from 20260910-1 to 20260810-2, since they seemed the most suspicious from the last few updates I applied.
I used these steps to do so:
Downgrade linux-firmware from 20260910-1 to 20260810-2.
Uninstall linux-firmware-amd and linux-firmware-ti.
Downgrade the remaining linux-firmware-* packages (which were in my case):
linux-firmware-amdgpu
linux-firmware-atheros
linux-firmware-broadcom
linux-firmware-cirrus
linux-firmware-intel
linux-firmware-mediatek
linux-firmware-nvidia
linux-firmware-other
linux-firmware-radeon
linux-firmware-realtek
linux-firmware-whence
I had to downgrade linux-firmware before downgrading the rest, since the new version introduced the linux-firmware-amd and linux-firmware-ti packages.
Attempting to downgrade all at once leads to file owner issues with the packages, since some files seem to have been moved to the two new packages.
Last edited by Noirdragen (Yesterday 12:56:54)
Offline
Got the same issue. Yesterday my system downloaded the 7.2.4 kernel but I didn't reboot so it wasnt loaded. Today I turn it on and after about 20 minutes it starts rebooting by itself multiple times and sending my back to my desktop but frozen. After a hard shutdown I get the same issues of my WM crashing/freezing.
Switching to tty while in my display manager and downgrading the linux kernel to version 7.2.3 worked for me.
Last edited by Ribzee (Yesterday 11:00:29)
Offline
On my own amd system (ryzen 9 9950X + 9060 XT using startx / lxqt I haven't noticed any those issues (not even errors in logs ).
Could this have to do with using integrated gpus and/or booting to wayland ?
Can the next person with the issues please try appending systemd.unit=multi-user.target to their boot command to ensure they're not booting into wayland automatically ?
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
If i am allow to point this : https://bbs.archlinux.org/viewtopic.php … 8#p2309528
Now sorry for the me too : on my system (still AMD Ryzen 5 7600 and AMD Radeon RX 7900 GRE) i downgrade yesterday to linux-firmware-amdgpu 20260810-2 and plasma was logging. But today it fail again for no obvious reason. So i upgrade back and especially linux-firmware-amdgpu (20260810-2 -> 20260910-1) and ddcutil (3.0.0-1 -> 3.0.1-1) and plasma log again....
So in my case, it seems that it was ddcutil and not linux-firmware-amdgpu.
Sorry again.
Offline
Seems like the linux-firmware was a red herring and the "fix" was incidental for me, too, since I rebooted today to only land back on square one. ![]()
If i am allow to point this : https://bbs.archlinux.org/viewtopic.php … 8#p2309528
The GitHub issue in your linked post describes the symptoms on my system quite accurately, and I can confirm that updating everything again, including ddcutil to the newly released 3.0.1-1, fixed the issue for me.
Thanks for the link, I'll update my previous post to not confuse anyone!
Switching to tty while in my display manager and downgrading the linux kernel to version 7.2.3 worked for me.
Can you check if the kernel is also a false lead?
I've been on 7.2.4.arch1-2 for a few days without issues, and only linux-lts was updated yesterday.
Speaking of which, I ran into the same issue when trying to login with 6.18.51-1 when I encountered the issue again today.
Can the next person with the issues please try appending systemd.unit=multi-user.target to their boot command to ensure they're not booting into wayland automatically ?
Didn't need that, since SDDM still runs on X11, and I do not use auto-login.
Thanks for the tip though! ![]()
Last edited by Noirdragen (Yesterday 12:56:21)
Offline
`ddcutil` 3.0.1 seems to have patched this: https://github.com/rockowitz/ddcutil/bl … ANGELOG.md, it's now available in Arch repos.
> Examining /dev/i2c devices in parallel during initilaization caused system crashes on AMD Navi video devices. KDE Plasma failed to start. Reverted to serial examination by default
Can confirm Plasma no longer freezes post-login for me.
Last edited by karnage42 (Yesterday 15:51:20)
Offline
Could this have to do with using integrated gpus and/or booting to wayland ?
I don't think so, at least not in my case.
$ echo $XDG_SESSION_TYPE && echo $XDG_CURRENT_DESKTOP
wayland
Hyprland
$ pacman -Q linux-firmware-amdgpu
linux-firmware-amdgpu 20260910-1
$ lspci | grep -i vga
04:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Picasso/Raven 2 [Radeon Vega Series / Radeon Vega Mobile Series] (rev c2)
$ lscpu | grep "Model name"
Model name: AMD Ryzen 5 3500U with Radeon Vega Mobile Gfx
$ uname -r
7.2.4-arch1-2Offline
`ddcutil` 3.0.1 seems to have patched this: https://github.com/rockowitz/ddcutil/bl … ANGELOG.md, it's now available...
Looking back at my boot history, pacman logs and ddctuil package infos, I downloaded the 3.0.1 version during the same boot for when I downgraded my kernel which might explain why it worked. But I don't recall updating any packages during that time and the version was released before my issues happened so it doesn't make complete sense.
Offline