I am posting this here as the Linux support staff have either not seen my mails or are otherwise preoccupied (7+ days, no response).
I’ve recently purchased an Ultra 5 (S1) version of the Framework 13 but had a bad time on Ubuntu 24.04.3 (kernel 6.14.0-27):
Graphical glitches.
Random Freezing and locking-up, requiring hard reboot.
Freezing on Framework logo during shutdown.
I resolved all this by installing Fedora 42 (kernel: 6.15.10-200) and applying latest updates.
It’s been over 6+ days and no issues whatsoever and finally starting to like the laptop.
So seems to be a kernel bug?
Yet, I don’t see anyone else reporting this in these forums.
For context, the very first OS I installed was Bazzite (KDE) and experienced the occasional hard freeze with that too, despite being on a newer kernel. However, it was much more usable than Ubuntu 24.04.3 and probably had a different root cause.
I’ve not seen the symptoms you describe using ubuntu 24.03 with kde either on 6.14.0-27 or now on 6.14.0-28 with the Intel Ultra 5 125H (matte display at 60 Hz).
What display are you using?
EDIT: I assume your using wayland (with kde) now on Fedora and were using X on Ubuntu. I’m not sure what Bazzite uses, but it would be interesting to know. I’m planning on trying Fedora soon.
I’m definitely on Wayland now, while on Fedora and very stable. The display is the same matte 60Hz.
But I would have thought it was Wayland even in Ubuntu since that’s the default?
The only possible difference is that I had used Kingston Fury RAM, 5600. Whilst this is approved, it may not have been extensively tested, from a Framework perspective. Indeed, Framework’s own documentation mentions this.
Yet, it’s highly stable on Fedora 42!
One other change is that I’ve stopped using the Logitech universal receiver, opting instead to use Bluetooth. But I doubt that would have been the issue.
I believe kubuntu 24.04 lts defualts to X (it was for me). It’s possible to run wayland on 24.04 lts but I have not tried yet.
I thought about ram as well, but I suspect that would be difficult to track down. If your happy on fedora 42/wayland and its stable then maybe not worry about it.
If the display ever happens to freeze again, see if you can get to the console (ctl-alt-f3 for me) or set up sshd and allow remote access. If the issue is just related to X or wayland, you may be able login and get a better idea of what has failed. If its a total system lockup (no console or remote ssh access possible), then I would suspect kernel/hardware.
I’ve had a similar problem with Kubuntu 24.04. I’ve had the FW13 for about a month, it’s frozen twice. I switched from X to wayland after the first lockup, for better scaling support when docking/undocking. Wayland itself doesn’t seem to have fixed it.
I haven’t enabled ssh but I have a zabbix server so I installed the agent to see if I could at least collect more information. I’ve left the laptop connected using wired NIC, WiFi disabled. When it froze the second time I got an alert from zabbix that the agent stopped responding at 5:40 AM. No unusual performance spikes before dropping offline. If zabbix couldn’t see it I doubt ssh would work but I could try a bit more (ping, ARP, ssh if it’s online) next time. Ctrl+alt+f3 etc didn’t work, power charging light did go off and on when unplugging, single press power didn’t work. Had to hold the power button to cycle it. The fan was spinning fairly fast and it was kind of hot. Zabbix didn’t show any increased CPU load prior to dropping offline, but it feels like there was possibly a high CPU after it dropped offline while frozen.
Syslog/journald didn’t show anything suspicious, some crone job ran 5 minutes earlier. Uptime was about a week stable sitting on my desk so it seemed quite random. I’m on the AMD 7060u, also using Kingston Fury RAM. So not exactly the same configuration but close.
Edit: kernel 6.14.0-29 on Kubuntu 24.04.3 to be more specific
Another fw13 intel core ultra user (see the link below the op’s first post) also experienced similar symptoms and has tried dropping back to a 6.8 kernel. No idea yet if that is a kernel/hardware issue or just a display server freeze. You don’t mention what kernel(s) your using and also your amd not intel…
That said, given your symptoms and inability to get to a tty console, I suspect you will not get too much further without some effort.
EDIT: See if you have anything in /var/lib/systemd/pstore/. Directories there may contain dmesg.txt files with some debugging info. I just tested with sudo echo c > /proc/sysrq-trigger (WARNING running this command will immediately trigger a kernel panic and you will lose any unsaved work), rebooted and found that ubuntu is saving some logs upon a kernel panic. Using ramoops described below may not be necessary (or it is enabled).
If its a kernel panic (and it still may not be), you could try to use a kernel with ramoops/pstore enabled. In the event of a kernel panic (crash) ramoops has a chance to save debug data to ram (in /sys/fs/pstore) which can be examined on reboot provided the system is not powered off before/during rebooting. I think you will have to change bios options so you can restart your framework (via the power button) without turning it off.
A quick inspection indicates pstore is enabled by default with kubuntu 24.04.3 (kernel 6.14.0-29) I dont see ramoops in dmesg so I’m gussing its not enabled in the normal ubuntu kernels. You could look to see if ubuntu provides a perbuilt debugging kernel with ramoops (or some other similar option) or build your own kernel with ramoops enabled.
I’ve used pstore/ramoops on an arm based system. Its not hard to setup/use and it is a lifesaver; however, keep in mind that if the kernel crashes badly enough, it may not be able to log anything. At that point your stuck building your own kernel, disabling kernel modules/features and “instrumenting” the kernel with debug statements until you can find a clue.