Linux Port – Feedback, Known Issues, and FAQ

I cannot, my entire system freezes.

 

so it doesn't recover afterwards?

 

Nope. My system freezes, displays disconnect then reconnect, and the system stays frozen. Sometimes you can hear audio, but nothing is responsive. Can't even open tty.

 

really weird, does it happen on any map?

 

Hello,
I'm experiencing the same issue as Gecked.
My configuration is:

• CPU: AMD Ryzen 9 7950X
• GPU: AMD Radeon RX 7900XTX
• OS: Manjaro
• Kernel: 6.4.6
• Mesa: 23.0.4
  
• Desktop: Plasma 5.27.6

The computer totally freeze, I have to use the reset button.
Game works fine with Proton.

 

As far as I can tell, yes.

 

I've been getting random crashes where my entire system freezes and the displays turn off when playing the game, sometimes it recovers and sometimes I have to restart the PC.

https://pastebin.com/VYdRQnp7

 

Same here. The game crashes my GPU driver.
I'm on AMD, now running an RX 6700 XT

Got this when it crashed:

Aug 19 13:40:35 lolcat kernel: snd_hda_intel 0000:0a:00.1: CORB reset timeout#2, CORBRP = 65535
Aug 19 13:40:35 lolcat kernel: amdgpu 0000:0a:00.0: amdgpu: GPU reset end with ret = -62
Aug 19 13:40:35 lolcat kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* GPU Recovery Failed: -62
Aug 19 13:40:42 lolcat kernel: [drm:amdgpu_dm_atomic_check [amdgpu]] *ERROR* [CRTC:80:crtc-1] hw_done or flip_done timed out
Aug 19 13:40:45 lolcat kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring sdma0 timeout, signaled seq=17037, emitted seq=17039
Aug 19 13:40:45 lolcat kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process pid 0 thread pid 0

 

I wonder if this really is a game bug or a Mesa bug... But yeah I can confirm weird crashes with BeamNG on the 5700XT too.

 

I'm also getting GPU hangs (system freezes/black screen etc.) on RADV on my RX 6700 XT.

Kernel messages like "ring <something> timeout" are normal indicators for GPU hangs and appear regardless of whether they're caused by application or driver bugs.

I've collected some logs with the Vulkan Validation Layers active, and the logs contain quite a few errors which I suspect could cause GPU hangs like this. I've attached the log to this message.

 

Linux is not really top priority, is it? I feel like a dev peeks into this post once every few months. We don't know what is known, and what is not, what is being worked on and what is not.

This uncertainty is a bit frustrating, even more so considering a major crash problem is completely preventing any Gameplay with actually decent performance right now. A while ago I was able to almost perfectly enjoy the linux build, and for weeks (months) this crash has been preventing that.

Don't get me wrong, I really appreciate the Linux build, but it's seemingly always dormant in some way or another, making it often unplayable for a long time.

I would love to see some activity, and if it's just a little progress update once a week or something like that.

 

I just found this on the arch wiki: https://wiki.archlinux.org/title/Vulkan#AMDGPU_-_Hangs_when_playing_DirectX_Vulkan_games

I guess I'll give AMDVLK a try. This might be the source of the vulkan issues.

Edit: If "hangs" really just refers to hangs and not crashes, then maybe the section right below is of relevance here. It lists out a few threads of people experiencing driver crashes with AMDVLK, but not mesa? Interesting.

 

It seems that the mesa devs have fixed the crashing issue, because I couldn't get the game to crash my driver, however, both with amdvlk and radv there are ~5 second long game hangs every about two seconds.

Maybe I just didn't test for long enough to experience the crash. These lagspikes prevented me from playing any longer than I absolutely had to for this test

 

Just updated to 0.30, started getting this issue (seizure warning?)

The map in the video is automation test track, but it happens on every map. The in game console and the stdout both spam the same error:

Code:

{timestamp}|E|_updateDynamicDescriptorSet| Missing VK resource

The same thing does not happen if I launch the windows version (through proton) with the vulkan renderer. It only happens on the linux bin. I'm running Arch, NVidia GPU with proprietary drivers, all up to date

P.S. Thanks for all the work that goes into the linux port!

 

I get the same errors, but then (probably just before switching off the loading screen) it freezes my whole OS, then everything blackscreens for about a second, then 2/4 Screens return, but with a still image.
Eruption (Keyboard RGB Control software still works, but not even Caps Lock seems to work (LED doesn't come on).*

looking into journalctl -kb -1 shows this a bunch of times

Code:

Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:3 pasid:32819, for process BeamNG.drive.x6 pid 41280 thread BeamNG.drive.x6 pid 41334)
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:   in page starting at address 0x0000000b00020000 from client 0x1b (UTCL2)
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00301431
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:          Faulty UTCL2 client ID: SQC (data) (0xa)
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:          MORE_FAULTS: 0x1
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:          WALKER_ERROR: 0x0
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:          PERMISSION_FAULTS: 0x3
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:          MAPPING_ERROR: 0x0
Sep 20 19:17:25 JoLi-HauptPC-archlinux kernel: amdgpu 0000:03:00.0: amdgpu:          RW: 0x0

also, after the GPU reset, this line appears ~150 times in one second, then not anymore. after that the sysrq stuff comes and then I hard rebooted.

Code:

Sep 20 19:17:37 JoLi-HauptPC-archlinux kernel: [drm] Skip scheduling IBs!

*) I tried rebooting with REISUB (sysrq), but didn't seem to work. looking into the logs reveals that it did the emergency sync, but didn't reboot from that...

 

Im also getting a crash when i try to load any map on 0.30.

Code:

43.22695|D|GELua.clientStartMission| starting levelPath: /levels/west_coast_usa/info.json
 43.22700|D|GELua.gameplay_missions_clustering.| Clustering cleared.
 43.23364|I|GELua.core_extendedTriggers.extendedTriggers| Added 3 BeamNGTriggers to triggers data
 43.23365|I|GELua.core_extendedTriggers.extendedTriggers| Added 0 site zones to triggers data
 43.23396|I|GELua.core_paths.core_paths.loadPath| 1 Camera Path(s) loaded.
 43.60270|I|GELua.levelLoading| *** Loaded everything in 21.681 s
 43.62660|D|GELua.serverConnection| Everything should be loaded setting worldReadyState to 1
 53.75083|D|GELua.core_camera.| Running cameras order:
 53.75085|D|GELua.core_camera.|  #1: order=0.200, name=transition
 53.75086|D|GELua.core_camera.|  #2: order=0.500, name=trackir
 53.75086|D|GELua.core_camera.|  #3: order=0.600, name=fallback
 53.75087|D|GELua.core_camera.|  #4: order=1.000, name=gameengine
 53.75594|I|GELua.freeroam_facilities.| Loaded facilities on level west_coast_usa (1 garages, 6 gasStations, 4 dealerships)
 53.76317|D|GELua.gamestate| Checking material finished loading
 53.76318|D|GELua.core_gamestate.gamestate| exiting : worldReadyState
 53.76319|D|GELua.core_gamestate.gamestate| show main menu (false)
 53.76320|D|GELua.core_gamestate.gamestate| exiting loading screen to menu
 54.17755|E|_updateDynamicDescriptorSet| Missing VK resource
....(many many lines of that error)
54.27414|D|_mapMaterial| [Material::mapMaterial] - Cannot map unnamed Material -
54.46507|E|_updateDynamicDescriptorSet| Missing VK resource
....(many many lines of that error)
58.14174|E|compileHlslGlsl| /shaders/common/downsamplerCS.hlsl: hlsl.hlsl:33:12: error: no matching function for call to 'reduce_avg'
    return reduce_avg(samples);
           ^~~~~~~~~~
hlsl.hlsl:28:9: note: candidate function not viable: no known conversion from 'type2x2' to 'float4 [4]' for 1st argument
type1x1 reduce_avg(type1x1 samples[4]) {
        ^

On my system this is only a soft crash and I can always terminate the program without needing a reboot

System specs:
OS: Manjaro kernel 6.1.49-1
CPU: Ryzen 7 3700X
GPU: RTX 3070Ti (Nvidia Drivers)
RAM: 32 gb

 

I can't launch any map now, when loading the whole system goes blackscreen and crashes to display manager.

https://pastebin.com/KEcRsjfS

Tried deleting the .local Beamng directory and that didn't work.