NVIDIA GPU shows 7f? Check access before replacing drivers.

1. Find the GPU and the first error

Use the general 7f checks to find your GPU’s address. Inspect the graphics function and, if present, the adjacent HDMI audio function. If both fail at the same moment, include that observation in your report; it helps distinguish a shared device path from an application-only problem.

# Example GPU address: replace it with yours.
lspci -s 0000:01:00.0 -nnk
sudo lspci -s 0000:01:00.0 -vv
sudo journalctl -k -b --no-pager
nvidia-smi

nvidia-smi requires the NVIDIA utilities and a working driver connection. A failure from it alone does not distinguish an inaccessible card from a driver problem.

NVIDIA describes Xid 79 as the driver finding the GPU inaccessible through its PCI Express connection. Its catalog lists possible link, hardware and driver causes. Record earlier PCIe errors too; Xid 79 and header 7f are separate observations, not interchangeable codes. NVIDIA: Xid catalog and Xid 79.

2. Choose the branch that matches the evidence

The GPU is readable, but the NVIDIA module does not load

Check the installed driver against the running kernel and look for module build or signature failures. On Ubuntu, use Additional Drivers or the supported ubuntu-drivers workflow. Ubuntu recommends that tool for Secure Boot because it can select signed packages. Follow its key-enrollment instructions if your installation requires them. Ubuntu: supported NVIDIA driver installation.

If the evidence identifies a missing or unsuitable Ubuntu driver and you are ready for a package change, the documented automatic-selection command is sudo ubuntu-drivers install. Plan a reboot and follow any signing prompts. Do not start by purging every graphics package or disabling Secure Boot. A readable GPU with a rejected kernel module needs a module/signing fix; an unreadable PCIe device needs a bus-access investigation.

The failure starts after suspend or idle

Collect a working fresh-boot snapshot and a failed-resume snapshot. Record the laptop model, BIOS version, kernel, driver, AC/battery state and external display connections. The Linux PCI power-management documentation describes coordinated platform and driver resume handling; the useful question is which transition fails on this particular machine. Linux kernel: PCI power management.

The failure starts under GPU load

Record the workload and elapsed time. Follow the system maker’s diagnostics and power/cooling checks. If a desktop uses a riser, include it in the hardware description and ask the vendor about an approved direct-slot comparison. Perform physical work only with the system powered down according to its service instructions.

3. Recover, then retest

1. Save work and shut down affected GPU workloads.
2. Use an orderly host restart. If necessary, use the maker’s complete power-off procedure; a warm reboot and removal of device power are different tests.
3. Choose one supported kernel or driver comparison with a rollback option. Record the exact versions.
4. Repeat the original suspend, display or workload sequence. Confirm PCIe access and actual GPU use, not just a clean boot.

If the GPU is assigned to a VM, continue with the VFIO and Proxmox lifecycle guide. A guest driver reinstall is not a substitute for inspecting host state.

Questions GPU owners ask

Does broken HDMI prove this is an NVIDIA driver bug?

No. Display routing, module loading and hardware accessibility are distinct checks. Capture the GPU and audio function state instead of diagnosing from a blank display alone.

Should I apply an ASPM or D3 power workaround?

Only when the workaround is documented for your hardware and evidence fits. Global power changes can affect unrelated devices and battery life. Keep the original settings and verify the transition that originally failed.

What would make a useful bug report?

A minimal repeatable sequence, before/after device output, kernel log timing and supported-version comparison. Include whether an orderly restart recovers the device. Do not describe a temporary recovery as a permanent repair.