Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

Whoracle · 2026-05-13 14:05:44

So, I have let my game project lay dormant for half a year. Wanted to get back to it yesterday. In the meantime I switched from nvidia to nvidia-open and got myself a shiny new RTX 5070 TI. Last time everything worked was with nvidia and my RTX 3080 FE. Same Unreal version (5.7) worked back then.

Didn't know if I should put it here or in Games/Multimedia, but I guess this is more of a HW/Driver issue, so...

I can start Unreal Editor, load my project, but as soon as I try to open an Asset or do anything that spawns a modal Vulkan will crash out. Afterwards UE is unusable due to certain Windows/Modals not being drawn - even outside of the actual Project I'm developing, just inside the Project Picker. Only reboot helps. Tried disabling GSP, didn't make a difference.

Found this in dmesg:

[    2.329368] NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64  595.71.05  Release Build  (root@)  
[  365.969676] NVRM: GPU at PCI:0000:01:00: GPU-4e391288-897c-ba5d-6944-dfbea55ebcfa
[  365.969680] NVRM: GPU Board Serial Number: 0
[  365.969681] NVRM: Xid (PCI:0000:01:00): 31, pid=13705, name=UnrealEditor, channel 0x0000000c, intr 00000000. MMU Fault: ENGINE GRAPHICS GPC2 GPCCLIENT_PROP_0 faulted @ 0x0_00000000. Fault is of type FAULT_PDE ACCESS_TYPE_VIRT_READ

Already found a bug report that sounds kinda similar (https://github.com/NVIDIA/open-gpu-kern … ssues/1030), and found issues related to this going back to 2019, but since it worked half a year ago maybe there's a newer workaround/fix that I haven't discovered yet.

/etc/modprobe.d/nvidia.conf:

options nvidia_drm modeset=1 fbdev=1
options nvidia NVreg_EnableGpuFirmware=0

So, anybody got any ideas? Happy to provide further info as needed.

seth · 2026-05-13 15:19:56

"options nvidia_drm modeset=1 fbdev=1" are the default and "options nvidia NVreg_EnableGpuFirmware=0" is inert w/ nvidia-open (it critically depends on the GSP) and the blackwell chip isn't supported by https://aur.archlinux.org/packages?O=0&K=580xx - so that's not an option.

This is only the UE?

vulkaninfo --summary

and vkcube still work? Do https://aur.archlinux.org/packages?O=0&K=unigine ?
Do you still have the 3080 (to test nvidia-open on it)?

Random blame-game:
- is this a hybrid graphics system?
- GTK defaults to vulkan rendering, if you're globally exporting "GSK_RENDERER=gl" or "GSK_RENDERER=cairo" to prevent that, does that help anything?

Whoracle · 2026-05-13 15:47:30

seth wrote:

"options nvidia_drm modeset=1 fbdev=1" are the default and "options nvidia NVreg_EnableGpuFirmware=0" is inert w/ nvidia-open (it critically depends on the GSP) and the blackwell chip isn't supported by https://aur.archlinux.org/packages?O=0&K=580xx - so that's not an option.

Good to know, noted.

This is only the UE?

As far as I know, yes. I have one game that stopped working out of the blue after the swap, and that's a UE game. Reinstalled that, and that has a slightly different, but maybe related error in dmesg:

NVRM: Xid (PCI:0000:01:00): 109, pid=875276, name=GameThread, channel 0x00000016, errorString CTX SWITCH TIMEOUT, Info 0x2c01b

vulkaninfo --summary
and vkcube still work?

Yes. Along with any other games I played recently.

==========
VULKANINFO
==========

Vulkan Instance Version: 1.4.341


Instance Extensions: count = 25
-------------------------------
VK_EXT_acquire_drm_display             : extension revision 1
VK_EXT_acquire_xlib_display            : extension revision 1
VK_EXT_debug_report                    : extension revision 10
VK_EXT_debug_utils                     : extension revision 2
VK_EXT_direct_mode_display             : extension revision 1
VK_EXT_display_surface_counter         : extension revision 1
VK_EXT_surface_maintenance1            : extension revision 1
VK_EXT_swapchain_colorspace            : extension revision 5
VK_KHR_device_group_creation           : extension revision 1
VK_KHR_display                         : extension revision 23
VK_KHR_external_fence_capabilities     : extension revision 1
VK_KHR_external_memory_capabilities    : extension revision 1
VK_KHR_external_semaphore_capabilities : extension revision 1
VK_KHR_get_display_properties2         : extension revision 1
VK_KHR_get_physical_device_properties2 : extension revision 2
VK_KHR_get_surface_capabilities2       : extension revision 1
VK_KHR_portability_enumeration         : extension revision 1
VK_KHR_surface                         : extension revision 25
VK_KHR_surface_maintenance1            : extension revision 1
VK_KHR_surface_protected_capabilities  : extension revision 1
VK_KHR_wayland_surface                 : extension revision 6
VK_KHR_xcb_surface                     : extension revision 6
VK_KHR_xlib_surface                    : extension revision 6
VK_LUNARG_direct_driver_loading        : extension revision 1
VK_NV_display_stereo                   : extension revision 1

Instance Layers: count = 9
--------------------------
VK_LAYER_FROG_gamescope_wsi_x86_64 Gamescope WSI (XWayland Bypass) Layer (x86_64) 1.3.221  version 1
VK_LAYER_MANGOHUD_overlay_x86      Vulkan Hud Overlay                             1.3.0    version 1
VK_LAYER_MANGOHUD_overlay_x86_64   Vulkan Hud Overlay                             1.3.0    version 1
VK_LAYER_NV_optimus                NVIDIA Optimus layer                           1.4.329  version 1
VK_LAYER_NV_present                NVIDIA Presentation Layer                      1.4.329  version 1
VK_LAYER_VALVE_steam_fossilize_32  Steam Pipeline Caching Layer                   1.3.207  version 1
VK_LAYER_VALVE_steam_fossilize_64  Steam Pipeline Caching Layer                   1.3.207  version 1
VK_LAYER_VALVE_steam_overlay_32    Steam Overlay Layer                            1.3.207  version 1
VK_LAYER_VALVE_steam_overlay_64    Steam Overlay Layer                            1.3.207  version 1

Devices:
========
GPU0:
	apiVersion         = 1.4.329
	driverVersion      = 595.71.5.0
	vendorID           = 0x10de
	deviceID           = 0x2c05
	deviceType         = PHYSICAL_DEVICE_TYPE_DISCRETE_GPU
	deviceName         = NVIDIA GeForce RTX 5070 Ti
	driverID           = DRIVER_ID_NVIDIA_PROPRIETARY
	driverName         = NVIDIA
	driverInfo         = 595.71.05
	conformanceVersion = 1.4.3.3
	deviceUUID         = 4e391288-897c-ba5d-6944-dfbea55ebcfa
	driverUUID         = eec1eba6-4484-5d12-ab32-1ef922507ba0

Do https://aur.archlinux.org/packages?O=0&K=unigine ?

Superposition and Valley both work.

Do you still have the 3080 (to test nvidia-open on it)?

Not easily, will rip it out of the SOs PC tomorrow. Darn SFF cases make that a hassle.

Random blame-game:
- is this a hybrid graphics system?

Nope, B650 Board, Ryzen 7800X3D (iGPU disabled) and the full fat 5070ti

- GTK defaults to vulkan rendering, if you're globally exporting "GSK_RENDERER=gl" or "GSK_RENDERER=cairo" to prevent that, does that help anything?

Nope, neither changes anything here.

Edit: I'm also on an X11 session, no gamescope/xwayland/whatever voodoo involved. I have wayland installed for testing stuff, but it's dormant. WM is AwesomeWM if it helps.

Edit 2: Played around a little bit and found out that _seemingly_ it only crashes if I open "more complex" Assets in UE. A Widget Blueprint reproducibly makes it crash/Xid31, but a StringTable doesn't. Also also, I'm on Unreal 5.7.4 atm but thee same happened on 5.7.0 before I upgraded last night.

Edit 3: Tried Unreal 5.8 Preview - Same crash. INstalling 5.6 now to see if that makes it go away.

Edit 4: Tried the 3080, not reproducible with that. Also tried the 5070 on Windows, also not reproducible. So something specific to the 5070. Happens with UE5.6, 5.7 and 5.8 Preview-1. So for the time being it's dual boot again, until either Unreal or NVIDIA gets their act together or somebody comes up with a fix/workaround...

Last edited by Whoracle (2026-05-15 06:20:42)

seth · 2026-05-15 17:09:30

I'm also on an X11 session, no gamescope/xwayland/whatever voodoo involved.

Fwiw https://github.com/ValveSoftware/Proton/issues/2842 has a similar error and blames xwayland

No issue on X11 from my previous X11 experience.

VK_LAYER_FROG_gamescope_wsi_x86_64 Gamescope WSI (XWayland Bypass) Layer (x86_64) 1.3.221  version 1

What happens if you actually remove https://archlinux.org/packages/extra/x86_64/gamescope/ ?
Can you post the entire crash/report?

Whoracle · 2026-05-16 07:00:40

Uninstalled gamescope - The crash now takes slightly longer to appear (opening 3 or 4 assets as opposed to at the first one).

Full dmesg: https://paste.c-net.org/SheldonBethany
Full UE Log: https://paste.c-net.org/SeekersKinkle

Last few lines of dmesg (the relevant ones IMHO):

[Sat May 16 08:46:45 2026] vetha69e09b: entered promiscuous mode
[Sat May 16 08:46:45 2026] eth0: renamed from vethc239fb0
[Sat May 16 08:46:45 2026] br-d19c616e0797: port 1(vetha69e09b) entered blocking state
[Sat May 16 08:46:45 2026] br-d19c616e0797: port 1(vetha69e09b) entered forwarding state
[Sat May 16 08:46:54 2026] systemd-journald[436]: Time jumped backwards, rotating.
[Sat May 16 08:46:58 2026] process 'Common/Zen/Install/zenserver' started with executable stack
[Sat May 16 08:48:41 2026] NVRM: GPU at PCI:0000:01:00: GPU-4e391288-897c-ba5d-6944-dfbea55ebcfa
[Sat May 16 08:48:41 2026] NVRM: GPU Board Serial Number: 0
[Sat May 16 08:48:41 2026] NVRM: Xid (PCI:0000:01:00): 31, pid=11956, name=UnrealEditor, channel 0x0000000b, intr 00000000. MMU Fault: ENGINE GRAPHICS GPC2 GPCCLIENT_PROP_0 faulted @ 0x0_00002000. Fault is of type FAULT_PDE ACCESS_TYPE_VIRT_READ

This time, I didn't even get the usual (paraphrased) "VulkanRHI: Swapchain gone. Not even attempting to present." Error in UE - it just cuts off mid-logline, might be due to running tee ue.log. From the very next start, without teeing, here's the las few lines:

[2026.05.16-06.55.31:238][  1]LogAssetRegistry: AssetRegistryGather time 0.1377s: AssetDataDiscovery 0.0287s, AssetDataGather 0.0402s, StoreResults 0.0688s. Wall time 14.2470s.
	NumCachedDirectories 0. NumUncachedDirectories 2851. NumCachedFiles 16109. NumUncachedFiles 0.
	BackgroundTickInterruptions 3.
[2026.05.16-06.55.31:277][  1]LogAssetRegistry: Display: Starting OnFilesLoaded.Broadcast
[2026.05.16-06.55.31:299][  1]LogPlacementMode: Display: The Asset Registry is done with its initial scan, the list of placeable classes has been updated.
[2026.05.16-06.55.31:311][  1]LogCollectionManager: Fixed up redirectors for 0 collections in 0.000000 seconds (updated 0 objects)
[2026.05.16-06.55.31:311][  1]LogAssetRegistry: Display: Completed OnFilesLoaded.Broadcast: 0.035s
[2026.05.16-06.55.31:311][  1]LogSourceControl: Uncontrolled asset discovery started...
[2026.05.16-06.55.31:407][  1]LogSourceControl: Uncontrolled asset discovery finished in 0.095172 seconds (Found 5513 uncontrolled assets)
[2026.05.16-06.55.31:725][  2]LogEOSSDK: LogEOS: SDK Config Platform Update Request Successful, Time: 9.134498
[2026.05.16-06.55.31:726][  2]LogEOSSDK: LogEOSAnalytics: EOS SDK Analytics disabled for route [1].
[2026.05.16-06.55.31:726][  2]LogEOSSDK: LogEOS: Updating Product SDK Config, Time: 9.508634
[2026.05.16-06.55.31:802][  3]LogAssetRegistry: Display: Asset registry cache written as 130.2 MiB to ../../../../../home/anthrax/git/games/manifoldpath/Intermediate/CachedAssetRegistry_*.bin
[2026.05.16-06.55.32:072][  6]LogEOSSDK: LogEOSAnalytics: Start Session (User: ...)
[2026.05.16-06.55.32:223][ 24]LogEOSSDK: LogEOS: SDK Config Product Update Request Successful, Time: 9.998287
[2026.05.16-06.55.32:225][ 24]LogEOSSDK: LogEOS: SDK Config Data - Watermark: -1745642100
[2026.05.16-06.55.32:225][ 24]LogEOSSDK: LogEOS: ScheduleNextSDKConfigDataUpdate - Time: 9.998287, Update Interval: 343.873810
[2026.05.16-06.55.32:502][ 55]LogTemp: [ECABridge] Camera stopped moving - triggering describe_view
[77518:80175:0516/085534.842295:ERROR:cast_crl.cc(443)] CRL - Verification failed.
[2026.05.16-06.55.35:217][350]LogSlate: Took 0.000032 seconds to synchronously load lazily loaded font '../../../Engine/Content/Slate/Fonts/Roboto-Light.ttf' (167K)
[2026.05.16-06.55.36:974][447]LogTemp: [ECABridge] OnEditorSelectionChanged called! Object: SelectedObjects
[2026.05.16-06.55.37:131][449]LogAssetEditorSubsystem: Opening Asset editor for WidgetBlueprint /Game/ManifoldPath/UI/ingame/PauseMenu/UI_PauseMenu.UI_PauseMenu
[2026.05.16-06.55.37:132][449]LogTemp: [ECABridge] Actor spawned: WorldSettings (WorldSettings)
[2026.05.16-06.55.37:132][449]LogTemp: [ECABridge] Event queued: actor_spawned - Actor spawned: WorldSettings (WorldSettings)
[2026.05.16-06.55.37:133][449]LogTemp: [ECABridge] Actor spawned: WorldSettings (WorldSettings)
[2026.05.16-06.55.37:133][449]LogTemp: [ECABridge] Event queued: actor_spawned - Actor spawned: WorldSettings (WorldSettings)
[2026.05.16-06.55.37:134][449]LogSlate: Warning: The command 'UMGEditor.OpenAnimDrawer' has the same default chord as 'EditorViewport.Next' [Ctrl+Shift+Space Bar]
[2026.05.16-06.55.37:185][449]LogStreaming: Display: FlushAsyncLoading(363): 1 QueuedPackages, 0 AsyncPackages
[2026.05.16-06.55.37:187][449]LogTemp: [ECABridge] A Blueprint was compiled
[2026.05.16-06.55.37:187][449]LogTemp: [ECABridge] Event queued: blueprint_compiled - A Blueprint was compiled
[2026.05.16-06.55.37:192][449]LogTemp: [ECABridge] A Blueprint was compiled
[2026.05.16-06.55.37:192][449]LogTemp: [ECABridge] Event queued: blueprint_compiled - A Blueprint was compiled
[2026.05.16-06.55.37:300][449]LogVulkanRHI: AcquireNextImage() failed due to the outdated swapchain, not even attempting to present.

[ ~ ] Terminating server, no sponsors or connections active.                                                                          [ 8:55 ]
Listening cancelled, closing port...
Listening cancelled, closing port...
Received signal 15
Daemon is exiting without errors.

Only the last line looks relevant, and I have to killall -9 UnrealEditor afterwards, which then leads to the termination messages from the full ue.log above.

So, gamescope removal kinda helped, but not much. Loading heavy assets (a fuller UI BP for example) still crashes, but lighter BPs don't anymore (i.e. a single UI Button BP used to crash before removal, now doesn't). I could speculate about the nature of the error, but I don't have a real clue, just gut feelings, and either way I don't know what to do about it. Next up strace? Don't know if that'll yield much, tbh.

Edit: To make this clear: I don't get an in-editor crash report - UE just silently stops accepting input/freezes, with the "loading" modal overlaid in front, but transparent - UE is simply not rendering that surface due to Vulkan having a fit. Doesn't respond to CTRL+C in the terminal either in sane time (i.e. less than a minute), have to killall -9 UnrealEditor to quit.

Last edited by Whoracle (2026-05-16 07:03:00)

seth · 2026-05-16 07:33:35

Does any of iommu=soft, amd_iommu=off or amd_iommu=fullflush prevent this?

Whoracle · 2026-05-17 06:48:35

Nope. Tried all 3 one-by-one, assuming I don't want a combination of iommu and amd_iommu, right?

(Side note, although it doesn't _really_ seem related, but the Windows version of 5.7.4 has developed a MASSIVE memory leak. To the point of my 64 gigs being full after ~1h or less of having the engine open and being AFK and doing nothing. Not sure if it's REALLY Unreal or just my Windwos Install acting up, because I was too lazy to just let the Windows run without unreal open, but still, "funny" side fact...)

seth · 2026-05-17 07:43:43

I don't want a combination of iommu and amd_iommu

No, they're all mutually exclusive.

Does nvidia-smi and "free -h" indicate you're maybe running out of resources on linux as well?

Whoracle · 2026-05-17 08:08:20

Does nvidia-smi and "free -h" indicate you're maybe running out of resources on linux as well?

Nope, and I have traced the leak down to the abominable Codex App on Windows. That's what I get for trying agentic "coding", I guess. Even before that the behaviour was completely different: On Windows, my System RAM is filling up over time. On Linux, I get an instant crash after at max 2 or 3 open Assets (and sometimes even before I am in my Project - in the file Picker of the main UE interface) without anything other than AwesomeWM and Unreal open. Also can't be the ECABridge (UE MCP Bridge) Plugin, since I can reproduce the crashes with that disabled and even completely deleted.

And yes, I do realize this means I'm most likely SoL

Edit: Found something: I also have a Notebook with an 1650MaxQ that also exhibits the same behaviour on nvidia-open. Switching to nvidia-580xx-dkms now. Now I wonder why this wasn't an issue with the 3080FE, if it's indeed "more" driver related than GPU related.

Last edited by Whoracle (2026-05-18 09:52:00)

Arch Linux

#1 2026-05-13 14:05:44

Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#2 2026-05-13 15:19:56

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#3 2026-05-13 15:47:30

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#4 2026-05-15 17:09:30

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#5 2026-05-16 07:00:40

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#6 2026-05-16 07:33:35

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#7 2026-05-17 06:48:35

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#8 2026-05-17 07:43:43

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

#9 2026-05-17 08:08:20

Re: Xid 31 Vulkan Crash on 5070ti/nvidia-open/unreal engine 5.7

Board footer