sky_line/Slipstream

Fork 1

Hard crash when alt camming through the sim #99

New issue

Closed

opened 2026-09-27 09:04:28 +00:00 by schizo-bat · 1 comment

schizo-bat commented

2026-09-27 09:04:28 +00:00

Environment

Alchemy_VK_654c07053f_x86_64

Description

While alt camming through the sim the viewer crashed:

radv/amdgpu: The CS has been cancelled because the context is lost. This context is guilty of a hard recovery.
radv: GPUVM fault detected at address 0x8000c3c28000.
GCVM_L2_PROTECTION_FAULT_STATUS: 0x401430
	 CLIENT_ID: (SQC (data)) 0xa
	 MORE_FAULTS: 0
	 WALKER_ERROR: 0
	 PERMISSION_FAULTS: 3
	 MAPPING_ERROR: 0
	 RW: 0
[vktest] VK_CHECK failed: vkQueueSubmit2(mImpl->mDevice.queue(), 1, &submit, mImpl->mSwapchain.inFlightFence()) = VK_ERROR_DEVICE_LOST (-4)
  at /home/bryan/ci/badalchestorm/indra/newview/vkgfx/llvulkancontext.cpp:637
[vktest] DEVICE_LOST fault: A GPUVM fault has been detected
[vktest]   address type=1 addr=0xffff8000c3c28000 precision=4096
[vktest]     no live/recently-freed buffer or AS contains it; neighbours:
[vktest]     above buffer '-' usage=0x20022 [0xffff800100a70000, 0xffff800100a7f0b0) size=61616 created frame 1275 by do-not-directly-run-alchemy-bin+0x265723f (now 128550)
[vktest]     (registry: 117 live buffers, 0 AS, 12522 recent frees, frame 128550)

*** SIGABRT backtrace ***
bin/do-not-directly-run-alchemy-bin(+0x310d82b)[0x5e7e4304c82b]
/lib/x86_64-linux-gnu/libc.so.6(+0x45330)[0x766a92c45330]
/lib/x86_64-linux-gnu/libc.so.6(pthread_kill+0x11c)[0x766a92c9ec0c]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0x1e)[0x766a92c4527e]
/lib/x86_64-linux-gnu/libc.so.6(abort+0xdf)[0x766a92c288ff]
bin/do-not-directly-run-alchemy-bin(+0x82dcf1)[0x5e7e4076ccf1]
bin/do-not-directly-run-alchemy-bin(+0xdff56e)[0x5e7e40d3e56e]
bin/do-not-directly-run-alchemy-bin(+0x20c0295)[0x5e7e41fff295]
bin/do-not-directly-run-alchemy-bin(+0xe09264)[0x5e7e40d48264]
bin/do-not-directly-run-alchemy-bin(+0xe0ad05)[0x5e7e40d49d05]
bin/do-not-directly-run-alchemy-bin(+0x9e2f49)[0x5e7e40921f49]
/lib/x86_64-linux-gnu/libc.so.6(+0x2a1ca)[0x766a92c2a1ca]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0x8b)[0x766a92c2a28b]
bin/do-not-directly-run-alchemy-bin(+0xa048e5)[0x5e7e409438e5]
*** end backtrace ***

Alchemy and console logs attached

Reproduction steps

Steps:

Move camera around the sim using Object View and panning
Expected: Viewer doesn't crash
Actual: Viewer crashed

### Environment Alchemy_VK_654c07053f_x86_64 ### Description While alt camming through the sim the viewer crashed: ``` radv/amdgpu: The CS has been cancelled because the context is lost. This context is guilty of a hard recovery. radv: GPUVM fault detected at address 0x8000c3c28000. GCVM_L2_PROTECTION_FAULT_STATUS: 0x401430 CLIENT_ID: (SQC (data)) 0xa MORE_FAULTS: 0 WALKER_ERROR: 0 PERMISSION_FAULTS: 3 MAPPING_ERROR: 0 RW: 0 [vktest] VK_CHECK failed: vkQueueSubmit2(mImpl->mDevice.queue(), 1, &submit, mImpl->mSwapchain.inFlightFence()) = VK_ERROR_DEVICE_LOST (-4) at /home/bryan/ci/badalchestorm/indra/newview/vkgfx/llvulkancontext.cpp:637 [vktest] DEVICE_LOST fault: A GPUVM fault has been detected [vktest] address type=1 addr=0xffff8000c3c28000 precision=4096 [vktest] no live/recently-freed buffer or AS contains it; neighbours: [vktest] above buffer '-' usage=0x20022 [0xffff800100a70000, 0xffff800100a7f0b0) size=61616 created frame 1275 by do-not-directly-run-alchemy-bin+0x265723f (now 128550) [vktest] (registry: 117 live buffers, 0 AS, 12522 recent frees, frame 128550) *** SIGABRT backtrace *** bin/do-not-directly-run-alchemy-bin(+0x310d82b)[0x5e7e4304c82b] /lib/x86_64-linux-gnu/libc.so.6(+0x45330)[0x766a92c45330] /lib/x86_64-linux-gnu/libc.so.6(pthread_kill+0x11c)[0x766a92c9ec0c] /lib/x86_64-linux-gnu/libc.so.6(gsignal+0x1e)[0x766a92c4527e] /lib/x86_64-linux-gnu/libc.so.6(abort+0xdf)[0x766a92c288ff] bin/do-not-directly-run-alchemy-bin(+0x82dcf1)[0x5e7e4076ccf1] bin/do-not-directly-run-alchemy-bin(+0xdff56e)[0x5e7e40d3e56e] bin/do-not-directly-run-alchemy-bin(+0x20c0295)[0x5e7e41fff295] bin/do-not-directly-run-alchemy-bin(+0xe09264)[0x5e7e40d48264] bin/do-not-directly-run-alchemy-bin(+0xe0ad05)[0x5e7e40d49d05] bin/do-not-directly-run-alchemy-bin(+0x9e2f49)[0x5e7e40921f49] /lib/x86_64-linux-gnu/libc.so.6(+0x2a1ca)[0x766a92c2a1ca] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0x8b)[0x766a92c2a28b] bin/do-not-directly-run-alchemy-bin(+0xa048e5)[0x5e7e409438e5] *** end backtrace *** ``` Alchemy and console logs attached ### Reproduction steps Steps: 1. Move camera around the sim using Object View and panning Expected: Viewer doesn't crash Actual: Viewer crashed

console.log

87 KiB

Alchemy.log

26 MiB

schizo-bat added the

bug

label

2026-09-27 09:04:28 +00:00

schizo-bat changed title from ~~Hard crash when alt cammon through the sim~~ to Hard crash when alt camming through the sim

2026-09-27 09:05:51 +00:00

sky_line referenced this issue

2026-09-27 16:02:13 +00:00

Windows: abrupt exit with no dump or log error (fail-fast?) — tracking #107

sky_line commented

2026-09-29 01:53:57 +00:00

Owner

Thanks for the report, and sorry it sat for a while. This crash was on an old test build (Alchemy_VK_654c07053f), from before 0.11, and the renderer has changed a lot since.

The log shows the GPU reset during rendering (VK_ERROR_DEVICE_LOST, a GPU memory fault on RADV). 0.12.0 Beta 3 also handles this much better if it does happen:

you get a message explaining that the graphics driver reset, instead of the viewer vanishing;
the log records which render pass the GPU was in when it crashed;
a crash report is written.

Could you try the same thing (alt-camming through a sim) on the latest release?
https://git.slipstreamviewer.org/backspace119/Slipstream/releases

If it crashes again, please open a new issue with Slipstream.log from that session (in ~/.slipstreamviewer/logs/) and link it here. The new log will tell us exactly where to look. Closing this one for now since it's from a very old build.

Thanks for the report, and sorry it sat for a while. This crash was on an old test build (`Alchemy_VK_654c07053f`), from before 0.11, and the renderer has changed a lot since. The log shows the GPU reset during rendering (`VK_ERROR_DEVICE_LOST`, a GPU memory fault on RADV). 0.12.0 Beta 3 also handles this much better if it does happen: - you get a message explaining that the graphics driver reset, instead of the viewer vanishing; - the log records which render pass the GPU was in when it crashed; - a crash report is written. Could you try the same thing (alt-camming through a sim) on the latest release? https://git.slipstreamviewer.org/backspace119/Slipstream/releases If it crashes again, please open a new issue with `Slipstream.log` from that session (in `~/.slipstreamviewer/logs/`) and link it here. The new log will tell us exactly where to look. Closing this one for now since it's from a very old build.

sky_line closed this issue

2026-09-29 01:53:57 +00:00

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#99

No description provided.

Rows
Columns