err: DxvkSubmissionQueue: Command submission failed: VK_ERROR_DEVICE_LOST

Once again, a driver crash.

Who do I have to report it to?

Can you post the link to the apitrace file you made?

Could be something wrong on our end, but I also wouldn't be surprised if it just replayed fine on other hardware.

https://mega.nz/file/491DGKwb#qIqIR4hA0NQueEWv8iqGj0M3AyqrQJ6BEsRyNiqCUvI

Ok, the game runs on Nvidia without DXVK, full of missing stuff and glitches... there's a lot of stuff ready to trace and tomorrow I'll be able to upload the apitrace.

I'm guessing this is Modern Warfare 2 campaign Remastered right? How do you get it to run with Wine? Crashes on startup for me and i haven't found why yet.

I'm guessing this is Modern Warfare 2 campaign Remastered right? How do you get it to run with Wine? Crashes on startup for me and i haven't found why yet.

Explain the crash please.

Start game from battlenet -> wine crash in game window.

Haven't looked at outputting to a log file yet. The wine crash window itself also crashes when i try to expand it lol

Start game from battlenet -> wine crash in game window.

Haven't looked at outputting to a log file yet. The wine crash window itself also crashes when i try to expand it lol

Probably it's something DRM related.

Ye figured as much.
How did you manage?

Edit: I'll try to figure something out.

Understood. I'll try to test some more.

Protip: you can compile and install apitrace then apitrace trace wine MW2CR.exe for example
It spares a lot of time and you just have to put the libraries on "built-in" without removing DXVK.

Problem: tracing makes the game crash, the problem is that without tracing you can play it without DXVK.

I have some things i can try. Including testing on Windows.

Will try again once i get my new setup later this week.

Understood. I'll try to test some more.

Protip: you can compile and install apitrace then apitrace trace wine MR2CR.exe for example It spares a lot of time and you just have to put the libraries on "built-in" without removing DXVK.

Problem: tracing makes the game crash, the problem is that without tracing you can play it without DXVK.

If you do that, you'll trace the OpenGL commands from wined3d and not the D3D11 commands, which makes it useless for debugging DXVK (or wined3d for that matter).

If you do that, you'll trace the OpenGL commands from wined3d and not the D3D11 commands, which makes it useless for debugging DXVK (or wined3d for that matter).

What about apitrace trace --API=D3D11 wine MW2CR.exe ?

Linux builds of apitrace don't support D3D, so that won't work either. And even for windows apitrace that would be wrong, it should be apitrace.exe --api=dxgi ... to trace D3D11.

Linux builds of apitrace don't support D3D, so that won't work either. And even for windows apitrace that would be wrong, it should be apitrace.exe --api=dxgi ... to trace D3D11.

Ok... my alternative was born because I can't apitrace this damn game, same with MW1R... the issue?
DXGIFactory issue.

Update: yeah, --api=dxgi, I wrote wrong stuff based on a single badly remembered event of yesterday at 2AM

@doitsujin any progress with the apitrace?

• With DXVK I can start everything, without it there's the 'DXGIFactory not implemented' issue.
• With DXVK it freeze, without DXVK and without any apitrace stuff the game works, full of glitches but it works.

I can't debug something that cannot be debugged using Apitrace without DXVK...

Trace replays fine without validation errors on AMD.

Yeah, but the latest frame is white?
The actual issue begins with DXVK and without DXVK you can play the game, but I can't trace something that gives DXGIFactory errors.

The problem is: of course it's fine, even on Nvidia, once you replay it, I guess?

I filed a bug on the nvidia forum

Trace replays fine without validation errors on AMD.

If so, does it mean that it's an nvidia issue for real?
Without DXVK, using the vulkan backend, the game crashes with an out-of-memory error, and at low-quality settings it crashes after a few seconds into the mission, which means that it could be a DXVK problem, since a few seconds is better than nothing, right?

If it's the same as the other remastered issue then it's not a Nvidia issue. But haven't checked it out yet so can't say for sure.

If it's the same as the other remastered issue then it's not a Nvidia issue. But haven't checked it out yet so can't say for sure.

Given how this was fixed in the latest driver without reason, I guess it is.

For those who don't know: there was this glitch where that piece of terrain would disappear if you didn't activate the precalculated spot-shadows.
Later on, the glitch appears again, in a different spot (it happened before so one place less, same glitch though).

I still have that on my R9 380. (well last time i tested at least which has been a while)
Haven't tested new 7900xtx

I still have that on my R9 380. (well last time i tested at least which has been a while) Haven't tested new 7900xtx

I must be dumb as hell... I've just realised that it's an AMD card.
By the way the latest trace has a glitch that if you try to replay it'll get skipped, but I guess it has been captured.
It must be.

Sorry for not specifying it was a AMD card.
I will try to check this issue and the other out here when i've got some time. Have my new setup now.

Sorry for not specifying it was a AMD card.

It's me who has to apologise, I assumed wrong when reading, I could've easily realised that there's no "R9" from Nvidia, they make something that starts with GTX and RTX and stuff...

I will try to check this issue and the other out here when i've got some time.

I still have to thank you to your precious contribution to helping people and this poor 22yo.
I'm trying to help in any way I can to this project, to the open-source community as a whole by spreading the word and stuff, it frustrates me that I can't help you of the devs by tracing something the right way, I thought of multiple ways to do it and it seems that this remastered and the other are a mess to debug.

I'll try as long as I have to, anyway.

When this issue happens, I'm getting:
NVRM: Xid (PCI:0000:01:00): 109, pid=15979, name=MW2CR.exe, Ch 00000056, errorString CTX SWITCH TIMEOUT, Info 0x348032
The "CTX SWITCH TIMEOUT" issue also happened with 'Resident Evil: Village' on vkd3d-proton, but seems to be fixed by https://github.com/HansKristian-Work/vkd3d-proton/commit/2a9c5fc522ca1f7b575054575e0127efdefab0b8
Maybe this issue is caused by something similar?

When this issue happens, I'm getting: NVRM: Xid (PCI:0000:01:00): 109, pid=15979, name=MW2CR.exe, Ch 00000056, errorString CTX SWITCH TIMEOUT, Info 0x348032 The "CTX SWITCH TIMEOUT" issue also happened with 'Resident Evil: Village' on vkd3d-proton, but seems to be fixed by HansKristian-Work/vkd3d-proton@2a9c5fc Maybe this issue is caused by something similar?

This game doesn't actually use DX12 as far as I know so that couldn't be something that this will fix, I guess.

There seems to be a race condition in the driver when submitting concurrently to different VkQueues.
This has nothing to do with DX12 and depends on how DXVK handles VkQueue submits (of which I have no idea).

To get this game working with an AMD GPU, is by faking a Nvidia GPU:
Put dxgi.customVendorId = 10de in your dxvk.conf.

Unfortunately, the game is also crashing with AMD when the 3rd mission is starting. My syslog shows:
Apr 28 20:02:18 zero-arch kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, but soft recovered
Apr 28 20:02:29 zero-arch kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, but soft recovered
Apr 28 20:02:39 zero-arch kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=2292085, emitted seq=2292087
Apr 28 20:02:39 zero-arch kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process MW2CR.exe pid 17583 thread MW2CR.exe pid 17583
Apr 28 20:02:39 zero-arch kernel: amdgpu 0000:03:00.0: amdgpu: GPU reset begin!

To get this game working with an AMD GPU, is by faking a Nvidia GPU: Put dxgi.customVendorId = 10de in your dxvk.conf.

That's already done by default

Not for Modern Warfare 2 Remastered no. Only the first remaster. If it is needed then we can add it

Not for Modern Warfare 2 Remastered no. Only the first remaster. If it is needed then we can add it

Well, the problem isn't solved but one problem out of two gets solved so...
The issue is that the main problem are those few levels that completely makes the game unplayable...

I can't remember if i ever looked in to this one specifically but i suspect it's similar or the same as the other Modern Warfare Remastered issues. But i have been unable to gather any kind of useful logs or traces or etc from those so far. So the devs need to look into it themselves.

I can't remember if i ever looked in to this one specifically but i suspect it's similar or the same as the other Modern Warfare Remastered issues. But i have been unable to gather any kind of useful logs or traces or etc from those so far. So the devs need to look into it themselves.

It's not just a similar issue, it's the same.
The only difference is that, years ago, Modern Warfare 1 Remastered would crash.
Now it doesn't crash and it just carries on with the glitches.

Modern Warfare 2 Remastered has the same problem, but it crashes, like it did years ago with Modern Warfare 1 Remastered.

Now it doesn't crash and it just carries on with the glitches.

It crashes for me. Or well causes a GPU hang. So something crashes i guess :stuck_out_tongue_closed_eyes:

I've went through the ordeal and installed Win11 Eval, then copied over Battlenet and placed DXVK dlls into the exe dir.
Result is, that the game crashes with DXVK under Win11 at the same point where it is crashing with wine.
I tried what's posted under https://github.com/doitsujin/dxvk/issues/252#issue-312221285, but there is nothing in the wine log and no radv-trace.txt is created at all.
Could anybody give me a hint, what diagnostic data could help with this issue?

I've played through the first MW Remastered (steamid: 393080) and noticed a similar crash at the bonus mission "Mile High Club", which starts after the main game is completed (about 3/4 into the mission, not right at the beginning).
Maybe the issues are related, since both games seem to use the same graphics engine?

Maybe the issues are related, since both games seem to use the same graphics engine?

Yes probably.
The devs just need to look at it. Not much i personally can do except test or gather info.

Hello, I have a similar issue with dxvk with this game on linux, crashes on the cliffhanger mission, it also freezes my steam deck in which I have to reset it

Here's a representative UMR dump from this hang on RX 7600 and Mesa 24.0.4 where it hangs in the same compute shader every time. I added some notes to the provided shader code with two of the places it timed out. I also double checked that it was the same hang on NV with RTX 3070 and 550.67.
MW2_UMR_dump_1.zip

This issue has been resolved in latest git-master, tested with AMD.
Seems like this https://github.com/doitsujin/dxvk/commit/14990dbb496c73d9310097159a8e2f737aea60ed did it?

This issue has been resolved in latest git-master, tested with AMD.

Seems like this https://github.com/doitsujin/dxvk/commit/14990dbb496c73d9310097159a8e2f737aea60ed did it?

Can confirm it no longer crashes on Steam Deck.

@gabriele2000 Are you able to confirm?

Well this is still open so I guess i will be able to confirm it in a bit..

@gabriele2000 Are you able to confirm?

I played through mw2 campaign remastered on my steam deck on steam os and all the crashes, at those specific three points are fixed

@gabriele2000 Are you able to confirm?

I played through mw2 campaign remastered on my steam deck on steam os and all the crashes, at those specific three points are fixed

Awesome!
Sorry for not replying, that mention has been lost, I'm sorry guys.
I think we can close it, since it has been fixed upstream months ago.

Thanks everyone!