protonscr

Shader compile related crashing?

dxvkclosed
doitsujin/dxvk#5681 · opened 2026-06-04 by BlueGoliath · updated 2026-06-05 · 8 comments · github
BBlueGoliath 2026-06-04 github

Hi,

Just wondering if any of the developers here are aware of any shader related game crashing bugs either in DXVK or Proton/Wine. I've been getting a lot of crashes and might be related to shader compilation.

Endfied will crash with error 4 general protection fault randomly, especially in the AIC areas.

The Finals randomly crashes when doing shader compilation and/or attempting to join a game.

Other Proton games will sometimes just randomly crash.

Thinking that this might somehow be a hardware issue, I did OCCT and memtester passes. Everything comes back green. They even happen after a fresh reboot and no overclocks applied.

I'm not sure if this is even a DXVK issue, since all of this started after Arch system updates, but I'm not sure where else to ask. I've also been seeing this in dmesg lately:

NVRM: dmaAllocMapping_GM107: can't alloc VA space for mapping.
NVRM: dmaAllocMapping_GM107: can't alloc VA space for mapping.
NVRM: nvAssertOkFailedNoLog: Assertion failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from pReuseMappingDb->pMapCb(pReuseMappingDb->pGlobalCtx, pAllocCtx, range, cachingFlags, &token, _reusemappingdbAddMappingCallback) @ mapping_reuse.c:273

It seems like they might be related as once this shows up, other graphical applications start crashing at the same time.

(I've had sound stop working in The Finals too, but I have no idea if that's related to any of this.)

Again, not even sure if any of this caused by DXVK but with everyone vibecoding things these days it's hard to pin down who to blame. Thanks for any insight.

Ddanielfikko 2026-06-04 github

Not a developer here, but to save the maintainers some time, check if the issue happens with the option d3d9.useDxbcSpirv = False (See: https://github.com/doitsujin/dxvk/issues/5680#issuecomment-4620078179). Also, give the names of the games and, if possible, an apitrace.

Ddoitsujin maintainer 2026-06-04 github

isn't The Finals a dx12 game?

Either way, missing a bunch of info here (proton log, which gpu), but the error message indicates that you're running out of memory (or address space in 32-bit games) in some way.

BBlueGoliath 2026-06-04 github

isn't The Finals a dx12 game?

Either way, missing a bunch of info here (proton log, which gpu), but the error message indicates that you're running out of memory (or address space in 32-bit games) in some way.

It's a 5060 TI 16GB on the latest 610 driver, but this happened on 595 too starting a few weeks ago. All games are 64-bit. Endfield is Vulkan while The Finals is DX12. Not sure about the others.

By proton logs do you mean an API trace? IIRC those aren't helpful if the crash isn't quickly ran into.

BBlueGoliath 2026-06-04 github

Not a developer here, but to save the maintainers some time, check if the issue happens with the option d3d9.useDxbcSpirv = False (See: [#5680 (comment)](https://github.com/doitsujin/dxvk/issues/5680#issuecomment-4620078179)). Also, give the names of the games and, if possible, an apitrace.

Names were provided and DX9 specific options aren't going to do anything here unless they somehow impact DX11/DX12.

Mmbriar 2026-06-04 github

It's definitely not a DXVK issue if it happens with native vulkan or dx12 games where DXVK isn't used at all.

BBlueGoliath 2026-06-04 github

It's definitely not a DXVK issue if it happens with native vulkan or dx12 games where DXVK isn't used at all.

I just checked and there is a game that is DX11. No idea if it's caused by the same issue because most crashes don't print anything to dmesg.

Again, wasn't sure if it was DXVK, especially since it happened after a system update. I don't think it's the Nvidia driver either because 595 was working fine before too. Not sure who else to ask about it.

It's really strange how some games(e.g. GTA 5) work fine.

Ppchome 2026-06-05 github

Driver bug (likely).
https://forums.developer.nvidia.com/t/bug-report-wayland-only-dmaallocmapping-gm107-va-mapping-failures-does-not-reproduce-on-x11/353598
https://docs.nvidia.com/deploy/xid-errors/analyzing-xid-catalog.html - look for XID errors in dmesg and use table for more info

tl;dr:
580.119.02 - last known OK
580.119.09 - first known KO

BBlueGoliath 2026-06-05 github

Driver bug (likely). https://forums.developer.nvidia.com/t/bug-report-wayland-only-dmaallocmapping-gm107-va-mapping-failures-does-not-reproduce-on-x11/353598 https://docs.nvidia.com/deploy/xid-errors/analyzing-xid-catalog.html - look for XID errors in dmesg and use table for more info

tl;dr: 580.119.02 - last known OK 580.119.09 - first known KO

Yeah looks like the VA issue is the cause:

[91418.679901] NVRM: dmaAllocMapping_GM107: can't alloc VA space for mapping.
[91418.679911] NVRM: nvAssertOkFailedNoLog: Assertion failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from pReuseMappingDb->pMapCb(pReuseMappingDb->pGlobalCtx, pAllocCtx, range, cachingFlags, &token, _reusemappingdbAddMappingCallback) @ mapping_reuse.c:273
[91418.679922] NVRM: dmaAllocMapping_GM107: can't alloc VA space for mapping.
[94246.104420] traps: wineserver[907730] general protection fault ip:55a949e4238a sp:7fffdf5fdac0 error:0 in wineserver[2e38a,55a949e29000+71000]

Guess I'll go bark up a different tree.

Upstream links