RTX 5080 (GB203): Random Xid 79 with zero precursor — any load level, instant atomic GPU death, open kernel module 595.71.05

[luciaLebrun]

NVIDIA Open GPU Kernel Modules Version

kmod-nvidia-595.71.05-1.fc44.x86_64

Please confirm this issue does not happen with the proprietary driver (of the same version). This issue tracker is only for bugs specific to the open kernel driver.

• I confirm that this does not happen with the proprietary driver package.

Operating System and Version

Bazzite

Kernel Release

Linux bazzite 6.19.14

Please confirm you are running a stable release kernel (e.g. not a -rc). We do not accept bug reports for unreleased kernels.

• I am running on a stable kernel release.

Hardware: GPU

NVIDIA GeForce RTX 5080

Describe the bug

Symptom

Random Xid 79 (GPU has fallen off the bus) at completely unpredictable intervals — minutes to 9+ hours after boot. Reproducible under:

• Heavy Vulkan/Proton gaming (Diablo IV)
• Light/indie gaming (Hyper Light Drifter)
• Desktop idle with only Spotify running
• Locked screen with no active GPU workload

Crash presents as black screen + audio artifacts. Wayland compositor collapses. Hard reboot required.

Key log — Xid 79 with zero precursor

  May 18 19:02:39 kernel: NVRM: GPU at PCI:0000:2b:00: GPU-0b11d014-1685-04bd-26ed-5fc0bc1c44ae
  May 18 19:02:39 kernel: NVRM: Xid (PCI:0000:2b:00): 79, GPU has fallen off the bus.
  May 18 19:02:39 kernel: NVRM: GPU 0000:2b:00.0: GPU has fallen off the bus.
  May 18 19:02:39 kernel: NVRM: krcRcAndNotifyAllChannels_IMPL: RC all channels for critical error 79.
  May 18 19:02:39 kernel: NVRM: _threadNodeCheckTimeout: API_GPU_ATTACHED_SANITY_CHECK failed!
  [x35 repetitions]
  May 18 19:02:47 kernel: NVRM: GPU0 _issueRpcAndWait: rpcSendMessage failed with status 0x0000000f for fn 10 sequence 137124
  [cascade of GspRmFree failures with status=0x0000000f]
  May 18 19:02:47 kernel: nvidia-modeset: ERROR: GPU:0: Failed detecting connected display devices

Zero kernel events in the 10 seconds before 19:02:39. No AER errors anywhere in the boot. The GPU disappears atomically with no
software-visible trigger.

In a different boot the same crash was preceded only by 9+ hours of clean operation (last NVIDIA log entry was Steam GPU detection at
session start).

What has been ruled out

Fix attempted	Result
NVreg_EnableGpuFirmware=0	Ignored — open module on Blackwell enforces GSP
pcie_aspm=off kernel arg	No effect
PCIe Gen 3 forced in BIOS	No effect
GPU power limit 300W (nvidia-smi -pl 300)	No effect
nvidia-persistenced + min clock lock (500 MHz floor)	No effect
PCIe AER hardware errors	None present — physical link never drops

Additional observations

• nvidia-powerd exits at boot with ERROR! Running on an unsupported system (PCI device Id: 0x2c02) — RTX 5080 is absent from its
  supported device list, leaving Dynamic Boost entirely unmanaged
• No Xid other than 79 has ever appeared
• The failure mode matches issue RTX 5090 (GB202): GSP heartbeat timeout -> Xid 109/8 under Vulkan load via Proton (595.58.03, 590.48.01) #1080 (GB202, GSP silent death under Proton/Vulkan) but extends to near-zero GPU load, suggesting the
  GSP firmware on GB20x can die spontaneously regardless of command queue pressure
• status=0x0000000f (NV_ERR_GPU_NOT_FULL_POWER) in all post-crash GspRmFree calls

The crash also occurs on Windows 11 with the official NVIDIA Windows driver — identical black screen + hard reboot required. The issue is absent on an RTX 2070 in the same system (same slot, same PSU, same everything), confirming this is specific to the RTX 5080 (GB203/Blackwell) and not an environment issue.

To Reproduce

1. Use the system normally or leave it idle
2. After an unpredictable interval (minutes to 9+ hours), Xid 79 fires with no preceding warning — display goes black, hard reboot required

Notes:

• There is no reliable on-demand trigger — the crash occurs under any load level including near-zero (locked screen, Spotify in background)
• Crash does not reproduce on RTX 2070 in the same system (same slot, PSU, and motherboard), confirming the issue is specific to GB203/Blackwell silicon or this GPU's GSP firmware
• Zero kernel events precede the Xid 79 — the GPU vanishes atomically with no AER errors, no GSP heartbeat timeout log, and no PCIe link drop

Bug Incidence

Always

nvidia-bug-report.log.gz

nvidia-bug-report.log.gz

More Info

No response

[HackZers7]

I can confirm — I am hitting the same issue on NVIDIA GeForce RTX 5090.

The crash is fully unpredictable, but I was able to reproduce it on both driver 580.159.04 and 595.71.05.

I have verified the PCIe slot seating and power connectors — everything is fine on the hardware side.

System Information

• GPU: NVIDIA GeForce RTX 5090
• Driver: 580.159.04 and 595.71.05 (both affected, NVIDIA Open Kernel Module)
• Kernel: 6.17.0-29-generic

Symptoms

GPU suddenly falls off the PCIe bus during normal operation. Recovery requires a full reboot (Node Reboot Required). No prior warnings, no thermal issues, no power events on the rail.

Relevant Kernel Log

NVRM: GPU at PCI:0000:06:00: GPU-0b6b7808-c002-f593-ac86-cd8e57757b94
NVRM: GPU Board Serial Number: 0
NVRM: Xid (PCI:0000:06:00): 79, GPU has fallen off the bus.
NVRM: GPU 0000:06:00.0: GPU has fallen off the bus.
NVRM: GPU 0000:06:00.0: GPU serial number is 0.
NVRM: kgspRcAndNotifyAllChannels_IMPL: RC all channels for critical error 79.
NVRM: _threadNodeCheckTimeout: API_GPU_ATTACHED_SANITY_CHECK failed!
...
NVRM: _issueRpcAndWait: rpcSendMessage failed with status 0x0000000f for fn 78 sequence 7869!
NVRM: nvCheckOkFailedNoLog: Check failed: GPU lost from the bus [NV_ERR_GPU_IS_LOST] (0x0000000F) returned from nvdEngineDumpCallbackHelper(pGp>
...
NVRM: RmLogGpuCrash: RmLogGpuCrash: failed to save GPU crash data
NVRM: _kgspLogRpcSanityCheckFailure: GPU1 sanity check failed 0xf waiting for RPC response from GSP. Expected function 76 (GSP_RM_CONTROL) sequ>
NVRM: GPU1 GSP RPC buffer contains function 78 (DUMP_PROTOBUF_COMPONENT) sequence 7893 and data 0x0000000000000000 0x0000000000000000.
NVRM: GPU1 RPC history (CPU -> GSP):
NVRM:     entry function                     sequence data0              data1              ts_start           ts_end             duration acti>
NVRM:      0    76   GSP_RM_CONTROL              7868 0x000000002080a0d1 0x00000000000007e8 0x00065279a0990a4b 0x0000000000000000          y
NVRM:     -1    76   GSP_RM_CONTROL              7867 0x000000002080a0d1 0x00000000000007e8 0x00065279a089be78 0x00065279a089bffe    390us  
  ...

NVRM: GPU1 RPC event history (CPU <- GSP):
NVRM:     entry function                     sequence data0              data1              ts_start           ts_end             duration duri>
NVRM:      0    4124 GSP_LOCKDOWN_NOTICE            0 0x0000000000000000 0x0000000000000000 0x000652798bd0fed2 0x000652798bd0fed2           
NVRM:     -1    4124 GSP_LOCKDOWN_NOTICE            0 0x0000000000000001 0x0000000000000000 0x000652798bd0fdf0 0x000652798bd0fdf0           
NVRM:     -2    4124 GSP_LOCKDOWN_NOTICE            0 0x0000000000000000 0x0000000000000000 0x000652798bd0ef07 0x000652798bd0ef07           
  ...

NVRM: Xid (PCI:0000:06:00): 154, GPU recovery action changed from 0x0 (None) to 0x2 (Node Reboot Required)
NVRM: nvAssertFailedNoLog: Assertion failed: (status == NV_OK) || (status == NV_ERR_GPU_IN_FULLCHIP_RESET) @ rs_client.c:844