protonscr

CONTROL Resonant: Xid 109 / VK_ERROR_DEVICE_LOST at a fixed location only when in-game HDR is enabled (NVIDIA)

vkd3dclosed appid 3669870
HansKristian-Work/vkd3d-proton#3330 · opened 2026-09-27 by daer68 · updated 2026-10-05 · 5 comments · github · game page · search this game
Ddaer68 2026-09-27 github

Possibly related to #3327, but the trigger here is specifically in-game HDR, and proton-cachyos-slr 20260703 (reported working there) does not avoid it.

Summary

With in-game HDR enabled, the game reliably hangs the GPU at one specific location. With HDR disabled, the same location plays fine. The hang is independent of graphics settings, compositor, Wine display driver and Proton build, and reproduces on official Proton Experimental (11.0-20260924) with gamescope HDR.

Steps to reproduce

  1. Proton Experimental, launch options: gamescope -f -W 2560 -H 1440 -r 280 --hdr-enabled -- env DXVK_HDR=1 %command%
    (or, on Proton builds with the Wayland driver: PROTON_ENABLE_WAYLAND=1 PROTON_ENABLE_HDR=1 %command% on an HDR-enabled Wayland session).
  2. Enable HDR in the game's display settings.
  3. Go to the West Incursion Zone, near the Parking Garage marker on the map (or load a save made there).
  4. The game freezes within about a minute of arriving / loading.
  5. Disable in-game HDR and repeat: the area plays through without a hang.

The hang also happens on the lowest graphics preset ("Handheld"), so it is not tied to specific quality settings or VRAM pressure.

Symptom

  • Picture freezes, audio continues, one game thread pegged, GPU utilization drops to ~12%. The process never recovers and has to be killed.
  • Kernel log:
    NVRM: Xid (PCI:0000:09:00): 109, pid=..., name=CONTROLResonant, channel 0x0000000e, errorString CTX SWITCH TIMEOUT, Info 0x8c019
    Seen with Info values 0x8c019, 0x8c01b, 0x8c020, 0x8c021, 0x8c02d, 0x1c024, 0x1c026, 0x1c02d across runs.
  • Proton log: many
    err:vkd3d-proton:d3d12_command_allocator_Reset: There are still 1 pending command lists awaiting execution from command allocator iface ...
    then
    warn:vkd3d-proton:d3d12_device_mark_as_removed: Device ... is lost (reason 0x887a0005, "VK_ERROR_DEVICE_LOST").

Test matrix (all at the same location)

Proton Presentation path In-game HDR Result
Proton Experimental 11.0-20260924 winex11 in gamescope 3.16.25 --hdr-enabled, DXVK_HDR=1 on hang
GE-Proton11-1 winewayland, Hyprland on (HDR10) hang
GE-Proton11-7 winewayland, Hyprland on (HDR10) hang
proton-cachyos-slr 11.0-20260703 winewayland, Hyprland on (HDR10) hang
proton-cachyos-slr 11.0-20260703 winewayland, Hyprland on (scRGB, m_bPreferScRGBHdr: true) hang
proton-cachyos-slr 11.0-20260703 winex11 in gamescope 3.16.25 --hdr-enabled (nested in Hyprland) on hang
proton-cachyos-slr 11.0-20260703 winewayland, KDE Plasma 6.7.5 on hang
proton-cachyos-slr 11.0-20260703 winewayland, Hyprland, "Handheld" preset on hang
proton-cachyos-slr 11.0-20260703 winex11 via XWayland (HDR unavailable) off no hang
proton-cachyos-slr 11.0-20260703 winewayland, Hyprland off no hang

Also did not help: VKD3D_CONFIG=no_upload_hvv, ray tracing fully off, closing other GPU applications.

System

  • GPU: NVIDIA GeForce RTX 3070 (GA104, 8 GB)
  • Driver: 615.71.09, open kernel modules
  • Distro / kernel: CachyOS, 7.2.8-1-cachyos
  • Compositors: Hyprland 0.56.2 and KDE Plasma 6.7.5 (Wayland, 10-bit, HDR on)
  • vkd3d-proton: build 7f0c30ad3c8f28d (Proton Experimental 11.0-20260924), build 3dfc6f07d0953b1 (proton-cachyos-slr 20260703); also hangs with the builds in GE-Proton11-1 and GE-Proton11-7
  • Game: DX12, DLSS SR on, frame generation off, ray tracing off

Workaround

Disable in-game HDR

Logs

Proton Experimental run (attached control-resonant-hdr-hang-experimental.log, launched with PROTON_LOG=1 VKD3D_CONFIG=fault, gamescope HDR). Kernel: Xid 109 ... Info 0x1c02d. Relevant part of the log:

err:winevulkan:signaller_worker wait timed out with non-empty poll list.
err:vkd3d-proton:vkd3d_wait_for_gpu_timeline_semaphore: Failed to wait for Vulkan timeline semaphore, vr 2.
warn:vkd3d-proton:d3d12_device_mark_as_removed: Device ... is lost (reason 0x887a0005, "VK_ERROR_DEVICE_LOST").
err:vkd3d-proton:d3d12_device_report_fault: Failed to query device fault info, vr 1.

VK_EXT_device_fault is exposed by the driver but vkGetDeviceFaultInfoEXT returns VK_NOT_READY, so no fault addresses/vendor info.

proton-cachyos-slr run (attached control-resonant-hdr-hang.log):
PROTON_LOG=1 log from a hanging run (proton-cachyos-slr 20260703, winewayland, in-game HDR on), launched with VKD3D_CONFIG=breadcrumbs. Kernel reported Xid 109 ... Info 0x1c026 at the moment of the hang; the log shows VK_ERROR_DEVICE_LOST but no breadcrumb dump was produced (NVIDIA, no AMD buffer markers). Happy to re-run with other debug options if you tell me which.

control-resonant-hdr-hang.log

control-resonant-hdr-hang-experimental.log

Hhelloabunai 2026-09-29 github

Same crash (potentially) here on my system. Different proton but seems to be only happening with in-game HDR on.

specs:

  • GPU: RTX 3090, driver 615.71.09 (open kernel modules)
  • CPU: Ryzen 9 5800X3D; kernel 7.2.7-arch1-1 (Arch)
  • Proton: GE-Proton11-7: vkd3d-proton vkd3d-1.1-5604-gaf89350cc, dxvk v3.1-19-g601930949, dxvk-nvapi v0.9.2-89-gfd105c0d
  • Game build: 25472515
  • output: wine wayland (PROTON_ENABLE_WAYLAND=1) on Hyprland 0.56.2, HDR via HDMI with forced EDID, captured by Sunshine

Interacting with the elevator in the Evacuation Site building (so a diff location) each time caused an instant crash. Gameplay up until this point (~4 hours from start of game) in the game worked fine with HDR on. After this point HDR seems to cause the crash. All RT/PT options are off in-game. After passing the elevator in SDR and switching HDR back on, it hung again a few minutes later at another spot (20:24 below in logs). So HDR seems to be the trigger for whatever reason.

from the kernel (all 7 hangs i tried; the first 5 are the elevator, latter two are after i toggled SDR to progress further in-game)

2026-09-29T19:54:29+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=1330309, name=CONTROLResonant, channel 0x00000029, errorString CTX SWITCH TIMEOUT, Info 0x4c02f
2026-09-29T19:56:32+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=1334484, name=CONTROLResonant, channel 0x00000029, errorString CTX SWITCH TIMEOUT, Info 0x2cc02f
2026-09-29T19:59:11+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=1337833, name=CONTROLResonant, channel 0x00000029, errorString CTX SWITCH TIMEOUT, Info 0xec02f
2026-09-29T20:04:32+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=8706, name=CONTROLResonant, channel 0x0000002b, errorString CTX SWITCH TIMEOUT, Info 0x2bc033
2026-09-29T20:07:52+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=13355, name=CONTROLResonant, channel 0x0000002b, errorString CTX SWITCH TIMEOUT, Info 0x20c033
2026-09-29T20:24:34+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=38103, name=CONTROLResonant, channel 0x0000002e, errorString CTX SWITCH TIMEOUT, Info 0x20c035
2026-09-29T20:34:33+03:00 NVRM: Xid (PCI:0000:07:00): 109, pid=50481, name=CONTROLResonant, channel 0x0000002f, errorString CTX SWITCH TIMEOUT, Info 0x4c037

Wine/vkd3d stderr (the same two lines come right before every Xid log above)

2026-09-29T20:24:32+03:00 err:winevulkan:signaller_worker wait timed out with non-empty poll list.
2026-09-29T20:24:34+03:00 err:winevulkan:signaller_worker error waiting for semaphores, vr -4.

Tried RenoDX HDR mod additionally but same deal. SDR for now.

Rrunar-work 2026-10-05 github

@daer68 @helloabunai Could you provide save files for these specific locations, or the nearest save spot with instructions on how to get there? You should be able to find them in your home directory, in .steam/steam/userdata/<id>/3669870/remote/

Ddaer68 2026-10-05 github

@runar-work
Here it is.

remote.zip

But i cannot replicate the bug now. Yesterday ive been tinkering with my setup, installed uwsm and hyprland-focused-booster. So maybe it was VRAM issue all along. Or maybe last game updates helped. Ive been running around this location for half an hour and game didn't hanged for me this time. Here is screenshot of parking garage around which game hanged before
Image
Save file should put you nearby.
But it was not limited by this location actually, game started hanging regularly in open world parts in hdr. In sdr performance was lacking but game never freezed for me.
Anyway - i ordered AMD RX9070 to replace my old RTX3070 and tomorrow will be the last day i can help with this issue on this config. What can i do to help?

Ddaer68 2026-10-05 github

Tried to replicate bug on KDE without fancy uwsm and vram-boost stuff. Also no hanges with hdr. I think as far as im concerned this issue could be closed
All that i updated in a few days was a game itself and nvidia kernel modules - Oct 4: linux-cachyos-nvidia-open 7.2.8-2 → 7.2.9-1.
Nothing other comes to mind.

Here is what Claude Code wrote about what could've been a reason for fix, dont know how much help can it be.

I couldn't find anything in the 7.2.9 update that directly touches NVIDIA or HDR. The one thing that plausibly fits your hang fix is the scheduler changes in CachyOS's patches, and that is a guess.

Upstream Linux 7.2.8 → 7.2.9 (changelog)

  • The fetched changelog showed no nouveau or NVIDIA commits.
  • The only GPU-related changes were AMD display (a DML frame-limit/UBSAN build fix) and Intel Xe (a workaround for IDLEDLY timer programming). Neither applies to your NVIDIA card.
  • The summary I got back lists no HDR, color-management or sync/timing changes. It was a summarized fetch, so it could have missed a generic fix.

CachyOS patches on top (cachyos-7.2.9-1 notes)

  • Scheduler (gaming-sched branch): EEVDF improvements, cgroup mode additions, and fixes for inter-class wakeup preemption and runqueue overflow. This is the most likely candidate for a hang fix, but I haven't confirmed it.

NVIDIA module

  • The driver is the same 615.71.09 as before. The modules were only recompiled for the new kernel, so there are no NVIDIA driver code changes in this update.

Likely causes of the HDR fix

  • I'm assuming the game is Control Resonant. A scheduler or wakeup fix could explain a hang during HDR rendering.
  • The Vulkan loader update to 1.4.363 on Oct 3 (22:31) is another candidate. It landed the evening before the kernel update, which makes it hard to tell the two apart.
  • The Proton Experimental and Hotfix updates on Oct 3 are a third candidate. -- it is not, i was using CachyOS Proton from oct 1 during this testing
BBlaster4385 2026-10-05 github

I'm still facing the same issue. The game freezes in a specific location with HDR, and I see the xid 109 in logs. I'm on the latest proton-cachyos, but I've had the same freeze with proton experimental and ge. Let me know if there's something specific I can provide to debug this.

Related threads

Proton versions

Launch options

Launch lines

Error codes