NVIDIA dGPU disappeared from PCI/ePSA after s2idle resume, Xid 79 on Dell XPS 15 9500

Graphics / Linux Linux
, , ,
1

I am looking for advice on whether there is any remaining recovery path for a missing NVIDIA dGPU, or whether this is probably motherboard-level failure.

System:

  • Dell XPS 15 9500
  • BIOS 1.40.0, dated 2025-11-26
  • Fedora Linux 44
  • NVIDIA dGPU expected: GTX 1650 Ti / Max-Q family
  • NVIDIA PCI ID before failure: 10de:1f95
  • Intel iGPU still works: 8086:9bc4

The GPU was working before Apr 23, 2026. The laptop then went through a Linux s2idle suspend/resume. On resume, the logs showed NVIDIA ACPI/power-event errors, then around 90 seconds later:

NVRM: Xid (PCI:0000:01:00): 79, pid=6990, name=chrome, GPU has fallen off the bus.
NVRM: GPU 0000:01:00.0: GPU has fallen off the bus.
NVRM: Xid (PCI:0000:01:00): 154, GPU recovery action changed from 0x0 (None) to 0x2 (Node Reboot Required)

Before this happened, Linux saw the GPU like this:

pci 0000:00:01.0: [8086:1901] PCIe Root Port
pci 0000:00:01.0: PCI bridge to [bus 01]
pci 0000:01:00.0: [10de:1f95] NVIDIA PCIe Endpoint
ACPI: video: Video Device [PEGP]

After powering off and booting again, the NVIDIA GPU was gone. More importantly, the CPU PEG root port 0000:00:01.0 was also gone. Bus 01 is now assigned to an NVMe SSD instead:

pci 0000:00:1b.0: PCI bridge to [bus 01]
pci 0000:01:00.0: [1179:0116] NVMe PCIe Endpoint
ACPI: video: Video Device [GFX0]

Current Linux state:

  • No NVIDIA device in lspci
  • No 10de device
  • No /dev/nvidia*
  • Only Intel graphics appears in /dev/dri
  • NVIDIA packages are installed
  • Secure Boot is disabled
  • This does not look like a normal driver issue because the PCI device itself is missing

I also ran Dell preboot/ePSA diagnostics and checked System Information. The VIDEO section only showed Intel graphics:

Vendor Id: 8086
Device Id: 9BC4
Product Name: Intel(R) Graphics Controller

I did not see NVIDIA / 10DE anywhere in the ePSA System Info page. I also tried poweroff/reboot and a Dell flea-power/CMOS-style reset, but the GPU did not come back.

One extra oddity: ePSA repeatedly reports this thermal error:

Thermal: The (SKIN) reading (128C) exceeds the thermal limit

I do not know whether that is related, but it makes me wonder if the embedded controller or a board sensor is involved.

My current understanding is that Linux suspend/resume likely triggered the failure, but the dGPU is now missing below the OS. Since Dell ePSA also does not show it, I am worried the issue is motherboard/dGPU/power-rail related.

Questions:

  1. Has anyone seen an XPS 15 9500 lose its NVIDIA dGPU completely after Linux suspend/resume?
  2. If the GPU is missing from both Linux lspci and Dell ePSA System Info, is there any realistic software/firmware fix left?
  3. Could the SKIN sensor reading 128C cause Dell firmware/EC to suppress the NVIDIA GPU?
  4. Is there any deeper EC/RTC/NVRAM reset procedure for this model?
  5. Would BIOS reflashing/downgrading be worth trying, or is this likely a board replacement situation?

I do not know where to go from here. I am trying to avoid replacing the motherboard if there is still a plausible firmware/EC recovery step, but I also do not want to waste time reinstalling Linux or NVIDIA drivers if the GPU is not even being enumerated.