[Linux] Steam Games Don't Shut Down Properly Since Version 0.9

[BluesAdam]

Any Steam Game

Mods and mod versions used

OptiScaler_v0.9.0.0401 Fakenvapi 1.4

GPU

AMD RX 7600 8GB

OS

Fedora 43 Kinoite (Linux)

Used automated or manual install?

• Automated
• Manual

If on AMD/Intel and Automated, used DLSS inputs?

• Yes
• No

Also please mention in description which upscaler inputs are used (which upscaler you selected in game settings).

Did you check the Wiki and Compatibility List?

• Yes
• No

Please describe the issue and steps to reproduce it

I am using OptiScaler using Goverlay and if Opti is enabled, any Steam game fail to shut down properly and I have to force kill the application. This issue happened before and was fixed with the previous update. To reproduce the issue:

1. Install OptiScaler via Goverlay
2. Apply OptiScaler to a game via Goverlay's fgmod command
3. Start game
4. Quit game
5. Notice game doesn't shutdown and Steam shows "STOP" button

I am using game's native FSR 3.1 and swapping it for FSR 4.0.2 (Int8)

I have attached

• OptiScaler.log (set LogLevel=0 and LogToFile=true in OptiScaler.ini, zip it if too big)
• Screenshot of game folder (where you placed Opti)
• Screenshot of Opti overlay in-game (opens with shortcut, default Insert)

[TheRazerMD]

Where's the log? Also did you try Opti without Goverlay?

Edit: Also could you try reverting to FFX 2.1 SDK files?

https://github.com/GPUOpen-LibrariesAndSDKs/FidelityFX-SDK/releases/tag/v2.1.0

[BluesAdam]

Where's the log? Also did you try Opti without Goverlay?

Edit: Also could you try reverting to FFX 2.1 SDK files?

https://github.com/GPUOpen-LibrariesAndSDKs/FidelityFX-SDK/releases/tag/v2.1.0

Hey there. Here's the log. Will try the FFX 2.1 files asap and let you know. Shouldn't take long.

OptiScaler.log

[BluesAdam]

Tried using the files in the link you provided, which let me only use FSR 3.1.5 but the issue persists. I don't know how to use OptiScaler without Goverlay though, so can't help with that unfortunately.

[BluesAdam]

Where's the log? Also did you try Opti without Goverlay?

Edit: Also could you try reverting to FFX 2.1 SDK files?

https://github.com/GPUOpen-LibrariesAndSDKs/FidelityFX-SDK/releases/tag/v2.1.0

Hi,

Also tried manual installation and the issue persists. I also cannot swap to FSR 4.0.2 with manual installation for some reason but that's an issue for another day. So with or without Goverlay, the issue is there.

[TheRazerMD]

Could you elaborate if this happens on all games, or just specific ones (and which ones)? We need more info here.

[BluesAdam]

Could you elaborate if this happens on all games, or just specific ones (and which ones)? We need more info here.

So far tried Deep Rock Galactic and Warhammer 40K: Darktide and both exhibit the same issue. Darktide uses game's native FSR 3.1 and DRG uses DLSS as input. The closed post below might have some relevancy. I will try to find a game on Heroic and see if it happens on that platform as well.

Both games I installed Optiscaler manually. I will next try a game on Heroic and see if the issue happens there as well. Both Heroic Games Launcher and Steam are installed as flatpak by the way.

#512

[BluesAdam]

Tried Ghostwire Tokyo on Epic Store via Heroic Games Launcher. Again, did manual installation and there is no issue there. So the issue seems to be specific to Steam so far.

[al20ov]

Same here, games don't shut down properly, I have to click the blue "Stop" button on Steam.

Not sure if it's related or not but I recently came back to my gaming desktop (RX 7900 GRE) after not touching it for a couple months, reinstalled Void Linux over my previous Debian trixie install, steam, Cyberpunk 2077 and the latest optiscaler which happened to be 0.9, replaced the ffx dll with 4.0.2 int8 and I kept experiencing full on amdgpu freezes and crashes but only when exiting the inventory menu back into the game:

amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025cf83000 from client 10
amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00501031
amdgpu: Faulty UTCL2 client ID: TCP (0x8)
amdgpu: MORE_FAULTS: 0x1
amdgpu: WALKER_ERROR: 0x0
amdgpu: PERMISSION_FAULTS: 0x3
amdgpu: MAPPING_ERROR: 0x0
amdgpu: RW: 0x0
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025ce03000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025ce08000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025ce0b000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025cf83000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025cf81000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025cf80000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025cf81000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025cf82000 from client 10
amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32784)
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu:   in page starting at address 0x000080025ce1b000 from client 10
amdgpu: Dumping IP State
amdgpu: Dumping IP State Completed
amdgpu: [drm] AMDGPU device coredump file has been created
amdgpu: [drm] Check your /sys/class/drm/card0/device/devcoredump/data
amdgpu: ring gfx_0.0.0 timeout, signaled seq=225661, emitted seq=225663
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu: Starting gfx_0.0.0 ring reset
amdgpu: Ring gfx_0.0.0 reset succeeded
[drm] device wedged, but recovered through reset
amdgpu: Dumping IP State
amdgpu: Dumping IP State Completed
amdgpu: [drm] AMDGPU device coredump file has been created
amdgpu: [drm] Check your /sys/class/drm/card0/device/devcoredump/data
amdgpu: ring gfx_0.0.0 timeout, signaled seq=225662, emitted seq=225665
amdgpu:  Process GameThread pid 2642 thread vkd3d_queue pid 2756
amdgpu: Starting gfx_0.0.0 ring reset
amdgpu: Ring gfx_0.0.0 reset succeeded
[drm] device wedged, but recovered through reset
amdgpu: Dumping IP State
amdgpu: Dumping IP State Completed
amdgpu: [drm] AMDGPU device coredump file has been created
amdgpu: [drm] Check your /sys/class/drm/card0/device/devcoredump/data

I went down a huge rabbit hole thinking my RAM's timings were bad, tried multiple different amdgpu.x=y boot flags in GRUB, and it turns out it was OptiScaler all along (no offense, most of the time it's great). Then I simply disabled OptiScaler but kept fsr 4.0.2 int8 and it stopped crashing. Which is a shame since fsr 4.0.2 int8 looks better with DLSS inputs than it does with fsr 4 inputs in Cyberpunk 2077 imo.

I was not experiencing the same issue on 0.7.9 but once again I was on a completely different distro with an older kernel, older amdgpu, older radv, etc... so take that info with a grain of salt.

I have yet to try the old 0.7.9 on this new install though to see if the issue persists or if this is a regression.