sky_line/Slipstream

Fork 1

[viewer] [crash] Crash: graphics driver crash on intense load (?) #1063

New issue

Open

opened 2026-10-25 18:55:32 +00:00 by slipstream-reports · 1 comment

slipstream-reports commented

2026-10-25 18:55:32 +00:00

Collaborator

Source: in-viewer report (Help > Report Bug)

Reporter (avatar name as given; not verified):

silverfoot

Filed after a crash. Previous session ended with:

Previous session lost the graphics device (the graphics driver s

System

Version: 1.5.0.1
Channel: Slipstream Beta
OS: Linux 7.2.11-arch1-1 #1 SMP PREEMPT_DYNAMIC Fri, 16 Oct 2026 00:28:58 +0000 x86_64
CPU: AMD Ryzen 7 3700X 8-Core Processor (4426.67 MHz)
Memory: 31998 MB
GPU: AMD Radeon RX 6700 XT (RADV NAVI22)
GPU driver: radv Mesa 26.2.5-arch1.1
Vulkan: 1.4.354; ray tracing: enabled
Window: 1920x1008
Display: swapchain 1920x1008 BGRA8/sRGB fifo x4 (30 MB); monitors 2, window on 1920x1080@60Hz scale 1.00, HDR on
VRAM: n/a (no Vulkan world renderer)
Host memory: host-visible 38 MB / 23 allocs [cull 0/8 rt-aux 0/4 rt-mask 1/2 staging 33/2 readback 0/1 part 2/4 ui 2/2] device-only 107 MB | /s: alloc hv 0.0 (0 MB) dev 0.0 (0 MB), free hv 0.0 (0 MB) dev 0.0 (0 MB
Pipeline compiles: async 318/237 0.0s +10MB t15 c15 ob0: mMirrorRigAlphaPipe/Frag 0.0s 0M c15; mCullPipe (Cull) 0.0s 0M c15; mRtFwdPipe (DeferRtFwd) (DeferRtFwd) 0.0s 0M c15; mPdnMergeCorePipe (DeferShadowMergeCore) (De
Settings:
  RenderDeferredSSAO = false
  RenderDepthOfField = 0
  RenderFPSLimit = 60
  RenderFSAAType = 0
  RenderFarClip = 64
  RenderHiDPI = 0
  RenderMaxTextureResolution = 512
  RenderMaxVideoMemoryMB = 0
  RenderMirrors = 0
  RenderQualityPerformance = 0
  RenderReflectionProbeCount = 1
  RenderReflectionProbeDetail = 0
  RenderReflectionProbeLevel = 0
  RenderReflectionProbeResolution = 128
  RenderScreenSpaceReflections = 0
  RenderShadowDetail = 0
  RenderTransparentWater = false
  RenderVKAOStandard = 1
  RenderVKAlphaShadingRate = 0
  RenderVKDlssRayReconstruction = 2
  RenderVKFramePacing = 1
  RenderVKGodRays = 0
  RenderVKNoVsyncPresentMode = 0
  RenderVKOpaqueShadingRate = 0
  RenderVKPipelineCompileBudgetMB = 0
  RenderVKPipelineCompileThreads = 0
  RenderVKProjectorShadows = 1
  RenderVKRayTracing = 1
  RenderVKRenderScale = 1
  RenderVKShadowResolution = 512
  RenderVKSunShadows = 1
  RenderVKTAA = 0
  RenderVKTextureCompression = 4
  RenderVKUpscaler = 1
  RenderVSyncEnable = 1
  RenderVolumeLODFactor = 1.125
  SlipstreamRenderTonemapType = 1
  UIScaleFactor = 1
  VKDebugForceIntelCompileCap = 0

Description

i was playing helldivers with slipstream on the lowest settings
i was also playing video on a cytube stream

Steps to reproduce

play helldivers on max graphics
play slipstream on lowest graphics
run cytube

Log excerpt

Full scrubbed excerpt: slipstream-log-excerpt.txt

Last 40 lines

[vktest]    4 compute  q_defer_light    wait gfx 134568 -> signal comp 134568
[vktest]    5 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    6 graphics                  wait comp 134568 (+ 0) -> signal gfx 134569
[vktest]    7 compute  q_volumetrics    wait gfx 134569 -> signal comp 134569
[vktest]    8 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    9 graphics                  wait comp 134569 (+ 0) -> signal gfx 0
[vktest] async frame (registry frame 51264, slot 1): 10 segments in submission order
[vktest]    0 graphics                  wait comp 134569 (+ 0) -> signal gfx 134570
[vktest]    1 compute  q_rt_build       wait gfx 134570 -> signal comp 134570
[vktest]    2 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    3 graphics                  wait comp 134570 (+ 0) -> signal gfx 134571
[vktest]    4 compute  q_defer_light    wait gfx 134571 -> signal comp 134571
[vktest]    5 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    6 graphics                  wait comp 134571 (+ 0) -> signal gfx 134572
[vktest]    7 compute  q_volumetrics    wait gfx 134572 -> signal comp 134572
[vktest]    8 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    9 graphics                  wait comp 134572 (+ 0) -> signal gfx 0
[vktest] async frame (registry frame 51265, slot 0): 10 segments in submission order
[vktest]    0 graphics                  wait comp 134572 (+ 0) -> signal gfx 134573
[vktest]    1 compute  q_rt_build       wait gfx 134573 -> signal comp 134573
[vktest]    2 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    3 graphics                  wait comp 134573 (+ 0) -> signal gfx 134574
[vktest]    4 compute  q_defer_light    wait gfx 134574 -> signal comp 134574
[vktest]    5 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    6 graphics                  wait comp 134574 (+ 0) -> signal gfx 134575
[vktest]    7 compute  q_volumetrics    wait gfx 134575 -> signal comp 134575
[vktest]    8 graphics                  wait comp 0 (+ 0) -> signal gfx 0
[vktest]    9 graphics                  wait comp 134575 (+ 0) -> signal gfx 0
[vktest] AS build inputs: 7285 checked over registry frames 51262..51266 (7285 recorded on the async queue): 0 outside a live buffer / AS, below 1 MB, or running past their buffer's end
[vktest] RT scene audit (#874): registry frame 51266, 2 readback slot(s); CPU tables checked
[vktest]   slot 1: RT frame 44500 (registry frame 51266, viewer frame 44500) TLAS REFIT, 131072 records (cap 163840), 819 rig entries -- NOT executed (no GPU stamp: recorded, never ran or still in flight)
[vktest]   slot 0: RT frame 44499 (registry frame 51265, viewer frame 44499) TLAS REFIT, 131072 records (cap 163840), 819 rig entries -- NOT executed (no GPU stamp: recorded, never ran or still in flight)
[vktest]   CPU tables now (registry frame 51266, RT frame 44500; the GPU's copy lags by the dirty-range upload):
[vktest]   CPU slot-addr: 34037 rows, 17351 set: ok 17351; owner mismatch 0
[vktest]   CPU omm-addr: 0 rows, 0 set: -; owner mismatch 0
[vktest]   CPU rig-group: 12 rows, 12 set: ok 12; owner mismatch 0
[vktest]   CPU rig-entry: 819 rows, 12 set: ok 12; owner mismatch 0
[vktest]   dummy BLAS 0xffff800152df6600: live AS 'blas-dummy' id=0 [0xffff800152df6600, +512) created reg frame 6767
[vktest] RT read guard (#874 / #889): on; session 0 skipped read(s) in 0 frame(s) (harvested)
[vktest]   guard slot 0: RT frame 44499 (registry frame 51265, viewer frame 44499, 13707 bindless slots): 0 skipped read(s)

_{Filed by the Slipstream report relay. The reporter's identity is not verified; system details are as sent by the reporter.}

**Source:** in-viewer report (Help > Report Bug) **Reporter** (avatar name as given; not verified): ~~~text silverfoot ~~~ **Filed after a crash.** Previous session ended with: ~~~text Previous session lost the graphics device (the graphics driver s ~~~ ### System ~~~text Version: 1.5.0.1 Channel: Slipstream Beta OS: Linux 7.2.11-arch1-1 #1 SMP PREEMPT_DYNAMIC Fri, 16 Oct 2026 00:28:58 +0000 x86_64 CPU: AMD Ryzen 7 3700X 8-Core Processor (4426.67 MHz) Memory: 31998 MB GPU: AMD Radeon RX 6700 XT (RADV NAVI22) GPU driver: radv Mesa 26.2.5-arch1.1 Vulkan: 1.4.354; ray tracing: enabled Window: 1920x1008 Display: swapchain 1920x1008 BGRA8/sRGB fifo x4 (30 MB); monitors 2, window on 1920x1080@60Hz scale 1.00, HDR on VRAM: n/a (no Vulkan world renderer) Host memory: host-visible 38 MB / 23 allocs [cull 0/8 rt-aux 0/4 rt-mask 1/2 staging 33/2 readback 0/1 part 2/4 ui 2/2] device-only 107 MB | /s: alloc hv 0.0 (0 MB) dev 0.0 (0 MB), free hv 0.0 (0 MB) dev 0.0 (0 MB Pipeline compiles: async 318/237 0.0s +10MB t15 c15 ob0: mMirrorRigAlphaPipe/Frag 0.0s 0M c15; mCullPipe (Cull) 0.0s 0M c15; mRtFwdPipe (DeferRtFwd) (DeferRtFwd) 0.0s 0M c15; mPdnMergeCorePipe (DeferShadowMergeCore) (De Settings: RenderDeferredSSAO = false RenderDepthOfField = 0 RenderFPSLimit = 60 RenderFSAAType = 0 RenderFarClip = 64 RenderHiDPI = 0 RenderMaxTextureResolution = 512 RenderMaxVideoMemoryMB = 0 RenderMirrors = 0 RenderQualityPerformance = 0 RenderReflectionProbeCount = 1 RenderReflectionProbeDetail = 0 RenderReflectionProbeLevel = 0 RenderReflectionProbeResolution = 128 RenderScreenSpaceReflections = 0 RenderShadowDetail = 0 RenderTransparentWater = false RenderVKAOStandard = 1 RenderVKAlphaShadingRate = 0 RenderVKDlssRayReconstruction = 2 RenderVKFramePacing = 1 RenderVKGodRays = 0 RenderVKNoVsyncPresentMode = 0 RenderVKOpaqueShadingRate = 0 RenderVKPipelineCompileBudgetMB = 0 RenderVKPipelineCompileThreads = 0 RenderVKProjectorShadows = 1 RenderVKRayTracing = 1 RenderVKRenderScale = 1 RenderVKShadowResolution = 512 RenderVKSunShadows = 1 RenderVKTAA = 0 RenderVKTextureCompression = 4 RenderVKUpscaler = 1 RenderVSyncEnable = 1 RenderVolumeLODFactor = 1.125 SlipstreamRenderTonemapType = 1 UIScaleFactor = 1 VKDebugForceIntelCompileCap = 0 ~~~ ### Description ~~~text i was playing helldivers with slipstream on the lowest settings i was also playing video on a cytube stream ~~~ ### Steps to reproduce ~~~text play helldivers on max graphics play slipstream on lowest graphics run cytube ~~~ ### Log excerpt Full scrubbed excerpt: [slipstream-log-excerpt.txt](https://git.slipstreamviewer.org/attachments/675ef7c4-7ff0-4798-b02b-49c547e663cd) <details><summary>Last 40 lines</summary> ~~~log [vktest] 4 compute q_defer_light wait gfx 134568 -> signal comp 134568 [vktest] 5 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 6 graphics wait comp 134568 (+ 0) -> signal gfx 134569 [vktest] 7 compute q_volumetrics wait gfx 134569 -> signal comp 134569 [vktest] 8 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 9 graphics wait comp 134569 (+ 0) -> signal gfx 0 [vktest] async frame (registry frame 51264, slot 1): 10 segments in submission order [vktest] 0 graphics wait comp 134569 (+ 0) -> signal gfx 134570 [vktest] 1 compute q_rt_build wait gfx 134570 -> signal comp 134570 [vktest] 2 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 3 graphics wait comp 134570 (+ 0) -> signal gfx 134571 [vktest] 4 compute q_defer_light wait gfx 134571 -> signal comp 134571 [vktest] 5 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 6 graphics wait comp 134571 (+ 0) -> signal gfx 134572 [vktest] 7 compute q_volumetrics wait gfx 134572 -> signal comp 134572 [vktest] 8 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 9 graphics wait comp 134572 (+ 0) -> signal gfx 0 [vktest] async frame (registry frame 51265, slot 0): 10 segments in submission order [vktest] 0 graphics wait comp 134572 (+ 0) -> signal gfx 134573 [vktest] 1 compute q_rt_build wait gfx 134573 -> signal comp 134573 [vktest] 2 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 3 graphics wait comp 134573 (+ 0) -> signal gfx 134574 [vktest] 4 compute q_defer_light wait gfx 134574 -> signal comp 134574 [vktest] 5 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 6 graphics wait comp 134574 (+ 0) -> signal gfx 134575 [vktest] 7 compute q_volumetrics wait gfx 134575 -> signal comp 134575 [vktest] 8 graphics wait comp 0 (+ 0) -> signal gfx 0 [vktest] 9 graphics wait comp 134575 (+ 0) -> signal gfx 0 [vktest] AS build inputs: 7285 checked over registry frames 51262..51266 (7285 recorded on the async queue): 0 outside a live buffer / AS, below 1 MB, or running past their buffer's end [vktest] RT scene audit (#874): registry frame 51266, 2 readback slot(s); CPU tables checked [vktest] slot 1: RT frame 44500 (registry frame 51266, viewer frame 44500) TLAS REFIT, 131072 records (cap 163840), 819 rig entries -- NOT executed (no GPU stamp: recorded, never ran or still in flight) [vktest] slot 0: RT frame 44499 (registry frame 51265, viewer frame 44499) TLAS REFIT, 131072 records (cap 163840), 819 rig entries -- NOT executed (no GPU stamp: recorded, never ran or still in flight) [vktest] CPU tables now (registry frame 51266, RT frame 44500; the GPU's copy lags by the dirty-range upload): [vktest] CPU slot-addr: 34037 rows, 17351 set: ok 17351; owner mismatch 0 [vktest] CPU omm-addr: 0 rows, 0 set: -; owner mismatch 0 [vktest] CPU rig-group: 12 rows, 12 set: ok 12; owner mismatch 0 [vktest] CPU rig-entry: 819 rows, 12 set: ok 12; owner mismatch 0 [vktest] dummy BLAS 0xffff800152df6600: live AS 'blas-dummy' id=0 [0xffff800152df6600, +512) created reg frame 6767 [vktest] RT read guard (#874 / #889): on; session 0 skipped read(s) in 0 frame(s) (harvested) [vktest] guard slot 0: RT frame 44499 (registry frame 51265, viewer frame 44499, 13707 bindless slots): 0 skipped read(s) ~~~ </details> --- <sub>Filed by the Slipstream report relay. The reporter's identity is not verified; system details are as sent by the reporter.</sub>

slipstream-log-excerpt.txt

192 KiB

slipstream-reports added the

from-viewer

triage

labels

2026-10-25 18:55:32 +00:00

slipstream-reports added this to the Slipstream project

2026-10-25 18:55:33 +00:00

sky_line referenced this issue

2026-10-25 19:41:35 +00:00

Preference: hardware acceleration for web pages and media (CEF GPU on/off) #1068

sky_line referenced this issue

2026-10-25 19:41:35 +00:00

Survive a graphics device loss: rebuild the Vulkan renderer instead of logging out #1069

sky_line commented

2026-10-25 19:46:22 +00:00

Owner

Thanks for the kernel log. What happened, in order:

A GPU page fault in the web/media plugin's GPU process (dullahan_host, Chromium/CEF rendering web or video content through Mesa's OpenGL driver): a permission fault in the command processor. That fits the video stream that was playing.
Two seconds later the GPU locked up; the kernel's ring reset failed, so amdgpu did a full GPU reset ("VRAM is lost due to GPU reset").
A full reset invalidates every program's GPU state, so the viewer's Vulkan renderer reported device lost at that moment (the desktop compositor was hit as well).

So the viewer's renderer was collateral here, not the trigger. Follow-ups:

#1068: a Preferences > Advanced option to turn hardware acceleration for web pages and media off, for anyone who hits this again (default stays on).
#1069: survive a device loss by rebuilding the renderer in-session instead of logging out.

If it happens again without any video/web content open, please file a new report with the kernel log from that boot.

Thanks for the kernel log. What happened, in order: 1. A GPU page fault in the **web/media plugin's GPU process** (`dullahan_host`, Chromium/CEF rendering web or video content through Mesa's OpenGL driver): a permission fault in the command processor. That fits the video stream that was playing. 2. Two seconds later the GPU locked up; the kernel's ring reset failed, so amdgpu did a **full GPU reset** ("VRAM is lost due to GPU reset"). 3. A full reset invalidates every program's GPU state, so the viewer's Vulkan renderer reported device lost at that moment (the desktop compositor was hit as well). So the viewer's renderer was collateral here, not the trigger. Follow-ups: - #1068: a Preferences > Advanced option to turn hardware acceleration for web pages and media off, for anyone who hits this again (default stays on). - #1069: survive a device loss by rebuilding the renderer in-session instead of logging out. If it happens again without any video/web content open, please file a new report with the kernel log from that boot.

No milestone

No project

No assignees

2 participants

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

sky_line/Slipstream#1063

No description provided.

Rows
Columns