I've been playing Descenders on Linux on Steam, and it's been running really well until now.

I can play for about a minute before the entire machine does a full reset. It just goes black and then a reboot.

Clearly there is a kernel problem, and I probably did an update when the problem started. This was on Fedora 31, and I upgraded to 32 a few days ago and even after the upgrade the problem remains.

I'd like some advice as to how to debug this. It's likely a driver issue, and the Radeon drivers are of course suspicious.

Any ideas?

Follow

@loke Anything in the /var/crash directory (or core dumps elsewhere)?

@modrobert That directory is empty. Then again, that wouldn't capture kernel crashes, would it?

@loke If you don't get any crash logs, and if this is a laptop. From experience, I would guess the motherboard temp is causing GFX (and CPU if combined package) BGA circuit to flex slightly against board (PCB) which over time eventually leads to solder cracks in the solder balls under the BGA. Common problem in laptops. If this is the case, and you have warranty, then manufacturer will replace the motherboard for you. If no warranty (usually the case), then you need to send the motherboard to be reballed by professional who repair. There are temporary fixes where they just reheat the board, but then it will only last like a month before the problem is back.

@modrobert That's not the issue here. This is a fairly new desktop machine, with a separate RX 5700 XT card.

It's somewhat weird that it started happening all of a sudden, and through several kernel updates (including an upgrade of Fedora from 31 to 32) the problem remains.

@modrobert Xorg. I tried it with Wayland too, and it was much more stable (I was able to play an entire evening). Then when I tried again yesterday it crashed, but it didn't kill the entire machine like it does with Xorg. Instead it hung the graphics card. I could still switch to a tty session, but trying to get another graphical session working (I tried to log in using Xorg) failed. GDM was still running, but the Xorg session hung in the middle of login.

@loke OK, are you using open source Radeon drivers (default?) or the AMD proprietary fglrx/Catalyst ones?

@modrobert I'm using the default in Fedora (part of the kernel package).

The proprietary drivers are only available for Redhat as far as I know.

@loke I've only had problems when moving from fglrx packages in Ubuntu to open source. Have to really delete a lot of packages with purge before installing open source Radeon, other than that it worked fine, performance is a lot better than it used to be in Radeon. Sorry, wish I could help more, still seems weird you don't get any crash dumps or logs from it. Steam sometimes put logs under /tmp when games crash there, but this seems more serious, hard crash. Anything in old 'dmesg' logs under /var/log?

@modrobert Perhaps I can get some more logs when I test it with wayland. The reason I don't get anything when it crashes with Xorg is that the entire machine is rebooted before any error is written.

@loke What about grub boot menu, do you have that in Fedora? Was thinking if you boot an older kernel (usually the last five are there when grub menu is activated).

@modrobert Good point. But I should have tried that the first time the problem happened. Since I have upgraded Fedora in between I don't think that's going to be possible.

Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.