Impossible to create new game with 'Rad new worlds update'

Based on this I tried a couple things and managed to generate a seed after modifying IdealLandingSite.yaml where I had to:

- Remove the `mincount: x` lines (not only for the oceanDeep biome)

- Comment out the poi_molten_gold_volcano and poi_molten_aluminium_volcano entries.

The game then managed to create a seed.

 

Very interesting. Could you attach that save so I could take a look? 

Here it is

Taudis.sav

A number of worlds failed to spawn in that save, there are only 6 when there should be 9. It's missing WaterMoonlet, MooMoonlet, and RegolithMoonlet. All the icons on the Starmap display the same terra icon too.

Seeing as @EricKlei has tried on fresh fedora and opensuse installs and had no failure, it seems to be either cruft from upgrading an existing install OR maybe a package that people don't have on a fresh install but which those experiencing the failure all have installed. I have a theory about it maybe being a shared library that overrides one of ours for example.

Open to hearing any theories, so far we are unable to reproduce on our end so we'll need one of you afflicted users to figure out what magic it is that breaks it.

6 hours ago, EricKlei said:

A number of worlds failed to spawn in that save, there are only 6 when there should be 9. It's missing WaterMoonlet, MooMoonlet, and RegolithMoonlet. All the icons on the Starmap display the same terra icon too.

That seed seems exactly like early, before moonlets + saptree worldgen. There are also some POI missing (Experiment 52B, temporal tear opener) and several other changes (temporal tear in the starmap, abyssalite overheating on Marshy asteroid, tungsten volcanos not in magma, niobium volcano that not always spawned) 

I don't know how but it seems exactly like one of my saves generated on January (+ a niobium volcano) Space Hut.sav

2 hours ago, nome said:

Seeing as @EricKlei has tried on fresh fedora and opensuse installs and had no failure, it seems to be either cruft from upgrading an existing install OR maybe a package that people don't have on a fresh install but which those experiencing the failure all have installed. I have a theory about it maybe being a shared library that overrides one of ours for example.

Open to hearing any theories, so far we are unable to reproduce on our end so we'll need one of you afflicted users to figure out what magic it is that breaks it.

Is there a way to make the seed generator and/or unity more verbose?

3 hours ago, nome said:

Seeing as @EricKlei has tried on fresh fedora and opensuse installs and had no failure, it seems to be either cruft from upgrading an existing install OR maybe a package that people don't have on a fresh install but which those experiencing the failure all have installed. I have a theory about it maybe being a shared library that overrides one of ours for example.

Open to hearing any theories, so far we are unable to reproduce on our end so we'll need one of you afflicted users to figure out what magic it is that breaks it.

I tried attaching gdb. Even if it can't print the backtrace due to the stripped executable, it catches a SIGSEGV

tested with:
 

gdb -p $(pidof OxygenNotIncluded)
(gdb) handle SIGPWR nostop noprint
(gdb) handle SIGXCPU nostop noprint
(gdb) cont
Continuing.                                                                                                                                                                                                       
[New Thread 0x7f9397357640 (LWP 12584)]                                                                                                                                                                           
[New Thread 0x7f9395356640 (LWP 12585)]                                                                                                                                                                           
[New Thread 0x7f9394b55640 (LWP 12586)]                                                                                                                                                                           
[Thread 0x7f9394b55640 (LWP 12586) exited]                                                                                                                                                    
[Thread 0x7f9395356640 (LWP 12585) exited]                                                                                                                                                    
                                                                                                                                                                                             
Thread 42 "OxygenNotInclud" received signal SIGSEGV, Segmentation fault.                                                                                                                      
[Switching to Thread 0x7f9397357640 (LWP 12584)]                                                                                                                                              
0x00000000420b5484 in ?? ()                                                                                                                                                                   
(gdb) bt                                                                                                                                                                                      
#0  0x00000000420b5484 in ?? ()                                                                                                                                                               
#1  0x00007f93a39c7e10 in ?? ()                                                                                                                                                               
#2  0x00007f93973564c0 in ?? ()                                                                                                                                                               
#3  0x00007f93ad1dffb8 in ?? ()                                                                                                                                                                                   
#4  0x0000000000000000 in ?? ()

(gdb) info registers                                                                                                                                                                                              
rax            0x7f93973563f0      140271873778672                                                                                                                                                                
rbx            0x7f93ad1aafc0      140272241127360                                                                                                                                                                
rcx            0x390               912                                                                                                                                                                            
rdx            0x10                16                                                                                                                                                                             
rsi            0x7f93ad587b90      140272245177232                                                                                                                                                                
rdi            0x7f93a14245f0      140272042395120                                                                                                                                            
rbp            0x0                 0x0                                                                                                                                                                            
rsp            0x7f9397356380      0x7f9397356380                                                                                                                                                                 
r8             0x0                 0                                                                                                                                                                              
r9             0x7f93a5514100      140272110485760                                                                                                                                            
r10            0x1a5e4b0           27649200                                                                                                                                                                       
r11            0x1c66a78           29780600                                                                                                                                                   
r12            0x7f9368b1ab40      140271093394240                                                                                                                                                                
r13            0x7f93a39c7e10      140272081862160                                                                                                                                                                
r14            0x0                 0                                                                                                                                                          
r15            0x7f93a07f06c0      140272029599424                                                                                                                                                                
rip            0x420b5484          0x420b5484                                                                                                                                                                     
eflags         0x10206             [ PF IF RF ]                                                                                                                                                                   
cs             0x33                51                                                                                                                                                                             
ss             0x2b                43                                                                                                                                                                             
ds             0x0                 0                                                                                                                                                                              
es             0x0                 0                                                                                                                                                                              
fs             0x0                 0                                                                                                                                                                              
gs             0x0                 0

 

 

I can reproduce the same.

Not an expert in asm, but for me this instruction looks like attempt to compare first field of some structure and getting nullptr dereference, since rbp contains 0x0:

cmpl $0x0,0x0($rbp)

 

Also will look into library overriding while running the game.

I've tried installing the latest Fedora and openSuse distros in a VM and played the game through Steam but could not reproduce the issue. We also know rolling back to an earlier version which did work now no longer works, which has me really puzzled since it should.

I'd love to attach a debugger and see the crash in person but since we can't reproduce this locally that's not happening. If anyone can figure steps to create a broken distro (be that "upgrade from distro X to Y", or "install package Z") I'd love to hear it, especially if it's "install package Z" as that'll probably be enough to answer what's going wrong right there.

Alright, not a reproduction, but a small workaround:

I started the game using Proton 6.3-8 and the world generation worked just fine.

Steps for enabling Proton:
ONI Preferences > Compatibility > [enable 'Force the use of a specific Steam Play compatibility tool'] > [select 'Proton 6.3-8']

On 12/1/2021 at 9:55 PM, nome said:

I'd love to attach a debugger and see the crash in person but since we can't reproduce this locally that's not happening. If anyone can figure steps to create a broken distro (be that "upgrade from distro X to Y", or "install package Z") I'd love to hear it, especially if it's "install package Z" as that'll probably be enough to answer what's going wrong right there.

If you can install an openSUSE Tumbleweed VM, I can give you the list of custom repositories I use, pack system configuration (there's nothing extraordinary) and all packages I have installed.

 

Do the linux builds have particular runtime dependencies?

Without surprise, the issue is still present after the release. Can we have a test build with more verbose logging? Was the assert cause in the code identified?

 

Just bought the game on Steam and I'm pretty sure I'm facing the same problem myself, also on Linux, where generating a new world always fails, even after trying +10 different seeds generated randomly by the game.

I'm running EndeavourOS (based on Arch), already verified game files integrity multiple times and reinstalled the game. Generating a world without the DLC loaded works. The only workaround I found was running the game with Proton, generating a new world, and then switching back to the regular Linux version and loading the save, which worked fine. 

From nome's comment, I guess that arch is not a supported distro either, but I see no reason why the game shouldn't run on all regular Linux distros.

Included Player.log below.

Player.log

13 hours ago, miguel1117 said:

From nome's comment, I guess that arch is not a supported distro either, but I see no reason why the game shouldn't run on all regular Linux distros.

It _should_ run on any reasonable linux distro, and we want it to, but for the purposes of support we only have a limited number of distros. Anything beyond that is best-effort, we can't have machines for nor test on every distro out there.

In this case we've been unable to reproduce the problem even by installing fedora so we're at a bit of a dead end. Until someone comes to us with either a simple repro (just install X to break it) or more info (e.g. "it seems to crash right after trying to access sound using ALSA but I'm using pulseaudio") we can't progress on this one. 

 

Can you tell us what is involved when generating seeds? How is the map randomness done?

I exclude newer glibc, or players with older fedora wouldn't get the issue

Using pipewire instead of pulseaudio ? No idea if it's present in older fedora.

I tried playing with LD_LIBRARY_PATH, MALLOC_CHECK_ without seeing anything particular.
I also tried running lsof every seconds while the game tries to generate a seed, but it also didn't reveal anything unusual. (except many UDP ports used even when disabling cloud saves and telemetry).
Moving the game from a brtfs to another partition type to check doesn't help,

My machine uses a X11 session, the NVidia binary blob, the tmp dir is a ramfs like, so nothing extravagant.

 

Does the game have command line parameters? (eg something to disable sound)

Edited December 21, 2021 by krop

One user reported a segfault (I think I remember it being a null dereference, which is very plausible). Can someone else confirm? If so, maybe a strace can find what the program is doing right before it segfaults and that might be the clue we need.

1 hour ago, nome said:

One user reported a segfault (I think I remember it being a null dereference, which is very plausible). Can someone else confirm? If so, maybe a strace can find what the program is doing right before it segfaults and that might be the clue we need.

No crash, only undefined symbols in libsteam_api.so
 

OxygenNotIncluded/OxygenNotIncluded_Data/Plugins/libsteam_api.so: error: symbol lookup error: undefined symbol: UnitySetGraphicsDevice (fatal)
[...] error: symbol lookup error: undefined symbol: UnityRenderEvent (fatal)
[...] error: symbol lookup error: undefined symbol: UnityGetAudioEffectDefinitions (fatal)
[...] error: symbol lookup error: undefined symbol: UnityPluginLoad (fatal)
[...] error: symbol lookup error: undefined symbol: UnityPluginUnload (fatal)

etc...

 

Other things I just tried:
* wiping most of ~/.local/share/Steam (except the game data and auth data), no change
* Disabling the DLC. Creating a new seed succeeds but logs unusual things:

Quote

[00:00:30.772] [1] [INFO] World Seeds: 89641100 for world SANDSTONE_DEFAULT [89641100/89641100/89641100/89641100]
Seeds are [89641100/89641100/89641100/89641100]
[00:00:40.242] [9] [WARNING] Room has no centerpoints. Terrain Cell [ shape: TallThin size: 2] [516 features/ocean/SaltCave (211.6, 250.0)]
THREAD - started 'SimThread' (139656598378048)
Unloading 2 Unused Serialized files (Serialized files now loaded: 0)

Quote

 

[00:00:55.280] [1] [INFO] Attempting to start a new game with current world gen
[00:00:55.842] [1] [INFO] Generating a world with the traits:
[00:00:55.888] [1] [INFO] Set world to [worlds/SandstoneDefault]
[00:00:55.930] [1] [INFO] Attempt success
THREAD - started 'SimThread' (139656606770752)
[00:00:59.115] [1] [INFO] 61.23052 Level Loaded.... backend
0(4) : error C0203: extension GL_AMD_vertex_shader_layer not supported in profile gp5vp

Note: Creation of internal variant of shader 'GUI/Text Shader' failed.
ShaderProgram is unsupported, but because jobified rendering is enabled the ShaderProgram can not be removed.

0(4) : error C0203: extension GL_AMD_vertex_shader_layer not supported in profile gp5vp

Note: Creation of internal variant of shader 'Hidden/InternalErrorShader' failed.
ShaderProgram is unsupported, but because jobified rendering is enabled the ShaderProgram can not be removed.

Shader warmup: 64 shaders 204 combinations 2.885s
[00:01:02.907] [1] [WARNING] No name tables provided to generate world name. Using GENERIC
[00:01:03.048] [1] [INFO] PopulatePOIsFromWorldGen
[00:01:03.051] [1] [INFO] -- GAME --
[00:01:03.093] [1] [WARNING] True Timelapse.OnNewDay but worldsToScreenshot is not empty
[00:01:03.233] [1] [INFO] Checking which steam branch we're on. Got: []
[00:01:03.627] [1] [INFO] Screen resolution updated, saving values to prefs: 1920x1009 @ 60, fullscreen: False
[00:01:03.639] [1] [WARNING] FMODUnity.EventNotFoundException: [FMOD] Event not found 'event:/expansion1/expansion1_FX/Radiation/Radiation_Geiger'
  at FMODUnity.RuntimeManager.CreateInstance (System.String path) [0x00015] in <0a23a27608924c5bae153e54c9475a79>:0
  at KFMOD.CreateInstance (System.String path) [0x00013] in <f655088ed32c40a5b1df667cf9f394a6>:0
[00:01:03.643] [1] [WARNING] FMODUnity.EventNotFoundException: [FMOD] Event not found 'event:/expansion1/expansion1_FX/Radiation/Radiation_Geiger'
  at FMODUnity.RuntimeManager.CreateInstance (System.String path) [0x00015] in <0a23a27608924c5bae153e54c9475a79>:0
  at KFMOD.CreateInstance (System.String path) [0x00013] in <f655088ed32c40a5b1df667cf9f394a6>:0
[00:01:03.643] [1] [WARNING] FMODUnity.EventNotFoundException: [FMOD] Event not found 'event:/expansion1/expansion1_FX/Radiation/Radiation_Geiger'
  at FMODUnity.RuntimeManager.CreateInstance (System.String path) [0x00015] in <0a23a27608924c5bae153e54c9475a79>:0
  at KFMOD.CreateInstance (System.String path) [0x00013] in <f655088ed32c40a5b1df667cf9f394a6>:0
[00:01:03.643] [1] [WARNING] FMODUnity.EventNotFoundException: [FMOD] Event not found 'event:/expansion1/expansion1_FX/Radiation/Radiation_Geiger'
  at FMODUnity.RuntimeManager.CreateInstance (System.String path) [0x00015] in <0a23a27608924c5bae153e54c9475a79>:0
  at KFMOD.CreateInstance (System.String path) [0x00013] in <f655088ed32c40a5b1df667cf9f394a6>:0

 

Quote

Unloading 3 Unused Serialized files (Serialized files now loaded: 0)
[00:01:30.796] [1] [WARNING] Failed to Unsubscribe event handler: Game+<>c.Void <.cctor>b__218_0(Game, System.Object)
Not subscribed to event
[00:01:30.796] [1] [WARNING] Failed to Unsubscribe event handler: Game+<>c.Void <.cctor>b__218_1(Game, System.Object)
Not subscribed to event
UnloadTime: 680.032707 ms

(No change after re-enabling the DLC after that)

Edited December 22, 2021 by krop

I faced the same problem and decided to do some testing.
I was checking the binary dependencies with ldd and seeing linux-vdso.so reminded me that I was using Zen Kernel. I switched back to Arch default kernel and it worked. I don't know if linux-vdso.so itself is the problem, changing the kernel is a big thing, so it might be something else, but i hope it helps someway.

 

I've been having the exact same problem on Arch with the same segfault, I have narrowed down the problem to pipewire / pipewire-pulse, this is certainly a result of a bug in fmod.

The quickest way to fix this is to switch back to just pulseaudio, on arch this can be done with:
sudo pacman -Syu pulseaudio
systemctl --user disable pipewire
systemctl --user enable pulseaudio
systemctl --user start pulseaudio

 

Edited January 2, 2022 by Chryseus

On 1/2/2022 at 11:45 AM, Chryseus said:

I've been having the exact same problem on Arch with the same segfault, I have narrowed down the problem to pipewire / pipewire-pulse, this is certainly a result of a bug in fmod.

The quickest way to fix this is to switch back to just pulseaudio, on arch this can be done with:
sudo pacman -Syu pulseaudio
systemctl --user disable pipewire
systemctl --user enable pulseaudio
systemctl --user start pulseaudio

 

Hi, just wanted to say that this fixed it for me. Using archlinux too. Thanks Chryseus.

On 1/2/2022 at 11:45 AM, Chryseus said:

I've been having the exact same problem on Arch with the same segfault, I have narrowed down the problem to pipewire / pipewire-pulse, this is certainly a result of a bug in fmod.

The quickest way to fix this is to switch back to just pulseaudio, on arch this can be done with:
sudo pacman -Syu pulseaudio
systemctl --user disable pipewire
systemctl --user enable pulseaudio
systemctl --user start pulseaudio

 

 

This shouldn't  be necessary then. Can you check if creating a fake pulseaudio symlink also works?

eg:

sudo ln -s /bin/true /usr/bin/pulseaudio

then launch steam

 

13 hours ago, krop said:

 

This shouldn't  be necessary then. Can you check if creating a fake pulseaudio symlink also works?

eg:

sudo ln -s /bin/true /usr/bin/pulseaudio

then launch steam

 

It does.