Xid109 CTX SWITCH TIMEOUT Driver Crashes In Many Applications

Graphics / Linux Linux
,
42

I thought I had solved the issue using NTSYNC a couple of months ago, but it’s happening all over again and it seems even worse than before. I’m throwing the towel here… I’ve tried it on Windows and it’s failing there too. I’m totally giving up on nVidia. It’s been MORE THAN AN YEAR I’ve been dealing with this issue. I’m building a desktop with an AMD GPU and I hope I never have to deal with nVidia ever again.

43

This is a persistent issue in Unreal Engine on my Arch installation. I am running Hyprland and have attempted to use both XWayland and native Wayland to run the engine. As soon as I move within a viewport that uses the Lit option set, the entire program freezes and becomes unresponsive. The only message I get is the same CTX SWITCH TIMEOUT error. The driver version is 580.95.05-1. I have compiled Unreal Engine 5.6 from source using the 5.6 branch at the commit cdda65cecacd9a1278020925c357bf4cc0b17e8c.

Here is an example of the error from journalctl:
NVRM: Xid (PCI:0000:0b:00): 109, pid=18707, name=UnrealEditor-Li, channel 0x00000021, errorString CTX SWITCH TIMEOUT, Info 0x4c026

44

Same with 580.105.08, ntsync, Proton GE 10.25.

nvidia-bug-report.log.gz (688.0 KB)

45

Hi @madoxav

Could you please help to share reliable repro steps and nvidia bug report from repro state.

46

I don’t get what you need or i didn’t understood your answer.

In the attached bug report (in my previous post Xid109 CTX SWITCH TIMEOUT Driver Crashes In Many Applications - #44 by madoxav )you can see the Xid109 error freezing the game for two differents games, Duckov (Steam) and SW:Outlaws (Ubisoft play).

The crash is random but usually appears on the ~15 first minutes of gameplay. In outlwas, it mostly happens when loading a new zone. In duckov, at no specific moment.

I don’t know what else I could provide.

47

I can reproduce this very consistently in the Unreal Engine 5.6.1 editor. Open a project in the editor, right click and move the camera around extremely fast in the viewport and it should freeze and spike the GPU to 100%. This is in the Linux version of UE5, not running via Wine or Proton.

nvidia-bug-report.log.gz (836.2 KB)

48

So I upgraded to 590.44.01, and this issue has gone away completely in UE 5.6’s editor.

49

Hello Madoxav, I am setting up a local repro environment meanwhile can you please check this behavior on latest 580.126.09 driver or 590.48.01 driver please?

50

Hello,

After ~30mins of playthroug on Outlaws I wan’t able to reproduce on 580.119.02 (I had the crash on 580.105.08). I’ll try again in the next days and let you know if it happens again.

51 
Xid (PCI:0000:01:00): 109, pid=31348, name=GameThread, channel 0x0000001d, errorString CTX SWITCH TIMEOUT, Info 0x328032

I had this issue in 575.51.02 , not in 575.57.08 and back again starting since 580.x.x until the current 590.48.01 (included) on my Blackwell RTX 5080. Not sure if this helps, but Proton logs err:winevulkan:signaller_worker wait timed out with non-empty poll list. together with the Xid 109 error.

I can consistently reproduce the issue on Black Myth: Wukong Benchmark Tool (Steam Game ID 3132990) when I set the option Visual Effect Quality to either Very High or Cinematic. Any other option can be set between Low or Cinematic, it wouldn’t make any impact. If I set every option to Cinematic but Visual Effect Quality to either Low, Medium or High, I will not get the CTX SWITCH TIMEOUT. Note this error happens in all games and 3D applications, but it’s with the Wukong game I was able to narrow-down how to precisely trigger the issue.

I have zero issues on Windows, which unfortunately I am forced to fall back to due to this specific issue. This is a list of all settings combinations I tried but did not make any difference on avoiding the issue:

  • Different Proton versions (9.x, 10.x, Experimental, GE 10-xx)
  • KDE Wayland, KDE X11, Gamescope on DRM
  • Switching between Full Screen, lower game resolution, toggle VRR on and off
  • Switched between NT-Sync and FSync
  • CachyOS on Linux 6.19-rc5, or Bazzite on Linux 6.17 on a clean installation.
  • Disabled EXPO on UEFI BIOS
  • Toggled ReBAR and SR-IOV on UEFI BIOS
  • Switch between PCIe Gen5 and PCIe Gen4

I hope this will help the NVIDIA Developers to reproduce the issue more easily.

1 Like
52

Thanks for the update, Please let me know if you able to reproduce it with 580.119.02, Thanks!

53

Can you please attach nvidia-bug-report here?

54

nvidia-bug-report.log.gz (716.7 KB)

please find the nvidia-bug-report attached with the reproducible steps I described in my post above.

55

Thank you, let me have a look and do the needful.

56

I am having this issue in 590.48.01
[Sat Feb 14 11:30:26 2026] NVRM: Xid (PCI:0000:01:00): 109, pid=142590, name=hitman3.exe, channel 0x0000000f, errorString CTX SWITCH TIMEOUT, Info 0x7c00e
nvidia-bug-report.log.gz (730.8 KB)

57

I also have the same issues, not in games but in… web browser! RTX 5080, nvidia open driver 590.48.01.

nvidia bug report and journalctl output attached. Crash happened at 14:53.

lut 19 14:53:25 PC kernel: NVRM: Xid (PCI:0000:01:00): 154, GPU recovery action changed from 0x0 (None) to 0x1 (GPU Reset Required)
lut 19 14:53:35 PC kernel: NVRM: Xid (PCI:0000:01:00): 109, pid=3688, name=msedge, channel 0x00000015, errorString CTX SWITCH TIMEOUT, Info 0x274027
lut 19 14:53:37 PC kernel: NVRM: _kgspLogXid119: ********************************* GSP Timeout **********************************
lut 19 14:53:37 PC kernel: NVRM: _kgspLogXid119: Note: Please also check logs above.
lut 19 14:53:37 PC kernel: NVRM: Xid (PCI:0000:01:00): 119, Timeout after 6s of waiting for RPC response from GPU0 GSP! Expected function 76 (GSP_RM_CONTROL) sequence 2472296 (0x2080a61d 0xc).

nvidia-bug-report.log.gz (630.6 KB)

journalctl_crashlog.txt (284.3 KB)

58

Silent Hill 2 Remake with 590.xx.

59

For the record, here is a new instance of the bug on The Witcher 3, with kernel 595:
595 release feedback & discussion - #107 by Thrim (it didn’t happen to me on 590)

60

Hi. I have been getting pretty much the same error recently. Currently I am using the 590.48.01 driver, and I am attaching my last bug report for exactly that driver. But I have also tried the latest 580 and 535 drivers before and I was getting the same error as well. I am on Ubuntu 24.04.4, have an RTX 2060 (laptop) and this error keeps happening for me when playing Diablo 2 Resurrected, the Reign of the Warlock DLC. But I was getting the same error previously before the DLC as well. I play the game via Lutris with the following configuration:

Wine version: wine-11.2-staging-tkg-amd64
DXVK version: v2.6.2
VKD3D version: vkd3d-proton-3.0b
DXVK NVAPI version: v0.9.0

And I usually see those two lines in the kernel.log:

2026-03-08T23:10:01.240645+01:00 legion5i kernel: NVRM: GPU at PCI:0000:01:00: GPU-ede4171a-33b2-58f4-0e27-d978740a12b1
2026-03-08T23:10:01.240661+01:00 legion5i kernel: NVRM: Xid (PCI:0000:01:00): 109, pid=13176, name=Main, channel 0x00000029, errorString CTX SWITCH TIMEOUT, Info 0x5c02e

But sometimes I also get the other line that says:

GPU recovery action changed from 0x0 (None) to 0x1 (GPU Reset Required)

And finally the error usually happens after a while of playing, I would say about 20, 30 minutes maybe. Sometimes it happens almost immediately and other times it does not happen even after a few hours of playing. But more often than not it does happen.

nvidia-bug-report.log.gz (902.1 KB)