Weird freezing on Wolfenstein: The New Order. Only on desktop, not Steam Deck. Is Valve interested in a bug report?
I'm on Kubuntu 24.04, rocking a build that was pretty darn high end in 2021 with an AMD 6800 XT, and of course, Wolfenstein: The New Order was already old news by then. Proton does miracles, but this game freezes my entire machine. The last time I saw something like this happen was with Monster Hunter World in 2018, on a much older version of Proton. I can reliably get the game to freeze my machine in the opening level of The New Order, even across multiple versions of Proton, even with the renderapi launch parameter that should switch it back to OpenGL. Of course, even if I report this to Steam support, they'll tell me that they only support Steam Deck and not bespoke Linux desktops, and the game works fine on my Steam Deck, but would they be interested in some logs and a bug reported against the GitHub project? This is assuming no one here has an easy fix, of course. But if not, how would I get the logs? I wouldn't know what I'm looking at in those logs, personally. I'm also not sure if they'll write out correctly. Because it freezes the entire machine, I end up having to hard shut down the computer by the power button, and once or twice during my experiments, it failed to mount my game SSD (a separate drive from where my OS is installed) at boot, and I had to set up the automatic mount in the partition manager again. So assuming that doesn't impact the ability to write out the logs, I can collect them with some instructions, if you kind strangers in the know wouldn't mind providing them, please. And if Valve is interested in looking at them.
To be fair, I can't find evidence of anyone on the internet experiencing this same freeze, so it's probably more specific to me than AMD in general. When I saw freezes like this in Monster Hunter, years ago, I was on Nvidia.
Have you ever tested your RAM? Like a proper memtest overnight?
If I ever suspect hardware, I also stress test the cpu with prime95.
I could give it a shot. Will it test VRAM too? Last I ran memtest was years ago. It needs a boot device like a USB, right?
No. It only tests system RAM.
Yes. https://memtest.org
Thanks. I'll give it a go. I don't think I'm convinced it's a hardware issue, since that error says something about permissions and faulty IDs, but what do I know? Couldn't hurt to check.
I’m not so sure it is either, but since you’re on kubuntu you don’t easily have access to newer mesa/kernel to test software.
Troubleshooting is not about finding the issue right away, it’s about eliminating by going through all of it and trying it to narrow it down.
Have you tried a distro with a bit more recent packages? If it also crashes then we know it’s more likely something with your hardware.
I'm not in a position to test on more than just these two machines/distros. Once upon a time, I tried switching to Fedora, but some of the behaviors were not to my liking, so I went back to Kubuntu.
I’ll end it here.
If you are curious I’d check out the following link and see what version of mesa you are running vs the latest. Maybe you can find more info or other experiences from users on that specific version.
https://linuxcapable.com/how-to-upgrade-mesa-drivers-on-ubuntu-linux/
If you want to explore a newer kernel: https://ubuntuhandbook.org/index.php/2023/08/install-latest-kernel-new-repository/amp/
Finally, I’m curious if you boot using USB into a live distro like Nobara, if you can manage to test the game. Or even actually install Nobara to a spare USB drive drive and test the game.
Maybe someone else can chime in, but that’s about all I can think of. If you want to find a root cause, you have to run through and find what’s not the issue.
I understand the nature of troubleshooting, but I don't think testing a 45 GB game is feasible off of a live distro, and any way to test it on my hardware outside of logs is a whole lot of work to get one game working; I don't have a spare drive around either. I just figured these errors would mean something to someone who could take action on it to make a better Proton for everyone. I haven't checked the kernel yet, but my version of mesa matches the latest stable version in your link.
EDIT: I did find this link that sounds like it's a mesa bug. I'm on the same major version but a different minor version.
FYI
Since AMD GFX works out of the box and steam is preinstalled in some live distro’s… simply mounting the hard drive with the game and then adding it to steam will work. Allows you to add the game ( steamlibrary) without having to download. Just a thought.
Thanks, I'll give it a try, probably over the weekend with Nobara.
EDIT: In case anyone finds this later, I just tested this with Nobara 40 and had a similar result. Nobara was on kernel 6.10 compared to my 6.8. The game ran fine from beginning to end on SteamOS with a custom kernel branched off of 6.1.52. I'm still operating under the assumption that this is a mesa bug. I don't have another machine to test this with and rule out hardware issues, but this is the information I collected, and besides, it's the only game exhibiting problems for me at this point in time.
EDIT to that EDIT: The New Order didn't run perfectly from beginning to end, I just remembered. It crashed back to the SteamOS menu once and hard froze much like my desktop once, but even that freeze was gracefully caught well enough that I could still use the Steam menu to force quit the program, unlike my experience on desktop.
I would recommend against installing ppas in general.
I think that Kubuntu/ubuntu Noble is on a pretty recent stable kernel as is. There would be something even fresher in the HWE track, dunno if that exists for noble yet. The DXVK version is up to Proton (so Proton-GE would be slightly fresher).
The Mesa version, I'm not sure where that comes from. You have an OS installed copy, you have a flatpak/snap version, but aiui Steam Runtime and/or Proton also likes to bring its own version.
Better gpu crash handling is a todo on Linux.
https://ubuntu.com/kernel/lifecycle https://www.phoronix.com/news/AMDGPU-Per-Ring-Resets
If it's an amdgpu bug try reporting it to the amd linux gitlab