Game: v3.1.0 GPU hang (VK_ERROR_DEVICE_LOST, NVIDIA Xid 109) when pulling the Master Sword (Temple of Time), Linux/gamescope, RTX 5060 Ti #2195
Labels
No labels
area/accessibility
area/audio
area/input
area/install
area/menu
area/renderer
area/saves
area/tooling
area/upscaling
beanstalk: approved
beanstalk: failed
beanstalk: needs approval
beanstalk: running
beanstalk: tested
ci: approved
good first issue
gpu/amd
gpu/igpu
gpu/intel
gpu/nvidia
help wanted
kind
bug
kind
chore
kind
docs
kind
feature
kind
performance
kind
question
needs measurements
platform/linux
platform/macos
platform/steam-deck
platform/windows
preview
priority
critical
priority
high
priority
low
regression
risk
high
risk
low
risk
medium
status
blocked
status
confirmed
status
duplicate
status
invalid
status
needs-info
status
wontfix
ci: approved
No milestone
No project
No assignees
2 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
MagicBeanDev/Longshot#2195
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Longshot version
v3.1.0
Channel
Release
What kind of problem
Crash, freeze or black screen
What happened
After updating from v3.0.0 to v3.1.0, the game freezes and closes with "the GPU device was lost (VK_ERROR_DEVICE_LOST)" three runs in a row. Each time the kernel logs an NVIDIA Xid 109 (CTX SWITCH TIMEOUT) for the longshot process, so the GPU hangs and the driver resets the channel.
Crash records (bug-reports/pending): reason "device lost", thread host, module_offset longshot+0x2272e470, backtrace longshot+0x2272e470 <- longshot+0x22715004.
Kernel:
NVRM: Xid (PCI:0000:01:00): 109, pid=30571, name=longshot, channel 0x00000021, errorString CTX SWITCH TIMEOUT, Info 0x7c01a
NVRM: Xid (PCI:0000:01:00): 109, pid=32106, name=longshot, channel 0x00000019, errorString CTX SWITCH TIMEOUT, Info 0x5c019
NVRM: Xid (PCI:0000:01:00): 109, pid=32474, name=longshot, channel 0x00000019, errorString CTX SWITCH TIMEOUT, Info 0x5c019
This PC had no Xid at all in November before this (rc.3 to v3.0.0 played fine, same saves, same driver). Nothing else changed between v3.0.0 and v3.1.0: same driver, kernel, gamescope and options file. The v3.1.0 build had 0 failures, and the title screen ran at 60 fps in a test start.
Where in the game
Temple of Time (child), pulling the Master Sword from the Pedestal of Time after Bottom of the Well (twice, same draw, ~45-55 s after loading the save in the Market); once when entering Hyrule Field (child) from the Market
How often
Every time
Graphics card and driver version
NVIDIA GeForce RTX 5060 Ti 16 GB, NVIDIA 620.70.06 (open kernel modules, linux-cachyos-deckify-nvidia-open)
CPU, RAM and operating system
Intel Core i7-14700K, 32 GB, CachyOS (Handheld Edition, kernel 7.3.4-1-cachyos-deckify), Steam game mode (gamescope 3.17.0), 2560x1440 60 Hz TV
Changes from the defaults
Built with the AppImage on the command line: --cli --update game --work work --jobs 8 --fetch-ngx-sdk --accept-nvidia-license (LONGSHOT_NGX_SDK set), so DLSS is built in.
Options: dlss=2 (DLAA, "NVIDIA DLAA after the game's bloom and tone map"), nvidia_reflex=1, frame_generation=0, hdr=0, vsync=0, frame_rate_cap=0, ultrawide=1, field_of_view=100, anisotropic_filtering=4, shadow_resolution=0, output_res=2048x1152.
No longshot_env.txt. Started from a Steam non-Steam shortcut with LD_PRELOAD="" %command% (so no Steam overlay). No ReShade, MangoHud or vkBasalt in the log.
longshot_log.txt
attached bellow
Screenshots or video
No response
Before you send it
Possibly related, for context:
Fixed in v3.1.1, thanks!
Same PC, driver, kernel, gamescope and options as in the report. After rebuilding with v3.1.1 (DLSS built in), I loaded the same save in the Market, walked to the Temple of Time and pulled the Master Sword (child, after Bottom of the Well) with no hang. Before, v3.1.0 hung there 2 of 2 times.
The v3.1.1 log has none of the signs from the v3.1.0 logs: no "simulated turf: ... -> 0 records" lines, no "a capture's wait for write-back ... timed out", no device lost, and no NVIDIA Xid in the kernel log.
It looks like #2174 (simulated-turf lists keep the game's size) was the fix: all three v3.1.0 logs showed the turf lists shrinking to 0 records and growing back from 0 right before each hang.