[PR #9] [MERGED] Add a script to automatically install the Nvidia Xorg driver if possible #106

New issue

Closed

opened 2026-10-15 12:49:08 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-15 12:49:08 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/games-on-whales/gow/pull/9
Author: @zb140
Created: 6/23/2021
Status: ✅ Merged
Merged: 6/24/2021
Merged by: @ABeltramo

Base: master ← Head: auto-x-driver

📝 Commits (2)

2d17cef Add a script to automatically install the Nvidia Xorg driver if possible
7d20803 Refactor the nvidia driver script to allow caching, and to clarify how it works a bit.

📊 Changes

6 files changed (+225 additions, -13 deletions)

View changed files

📝 .dockerignore (+2 -1)
📝 README.md (+77 -11)
➕ configs/xorg-nvidia.conf (+11 -0)
➕ ensure-nvidia-xorg-driver.sh (+126 -0)
📝 gpu.Dockerfile (+2 -0)
📝 startup-gpu.sh (+7 -1)

📄 Description

This is one possible way to do this. What do you think?

If you launch the container in the nvidia-docker runtime, the startup process will try to ensure you have an appropriate Xorg driver installed. It will look for one in /usr/lib/x86_64-linux-gnu/nvidia/xorg (which is where the Ubuntu package installs the driver), and it will look in /nvidia/xorg (where you can use Docker to map in your host's driver, or another specific version).

If it doesn't find one in one of those locations, it tries to identify an Ubuntu package with the same exact version as the host's driver. If it can find such a package, it installs it. If it can't, it prints a failure message. It will still attempt to launch Xorg in this case, but it warns that it may fail.

If you want to skip all this and just assume that the driver will be there, you can set the environment variable NVIDIA_SKIP_DRIVER_CHECK to 1.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/games-on-whales/gow/pull/9 **Author:** [@zb140](https://github.com/zb140) **Created:** 6/23/2021 **Status:** ✅ Merged **Merged:** 6/24/2021 **Merged by:** [@ABeltramo](https://github.com/ABeltramo) **Base:** `master` ← **Head:** `auto-x-driver` --- ### 📝 Commits (2) - [`2d17cef`](https://github.com/games-on-whales/gow/commit/2d17cef6907d77c6d1062c3064c37740f68d5614) Add a script to automatically install the Nvidia Xorg driver if possible - [`7d20803`](https://github.com/games-on-whales/gow/commit/7d208035b7af7e527ac433fa8d0f2ecc84430f99) Refactor the nvidia driver script to allow caching, and to clarify how it works a bit. ### 📊 Changes **6 files changed** (+225 additions, -13 deletions) <details> <summary>View changed files</summary> 📝 `.dockerignore` (+2 -1) 📝 `README.md` (+77 -11) ➕ `configs/xorg-nvidia.conf` (+11 -0) ➕ `ensure-nvidia-xorg-driver.sh` (+126 -0) 📝 `gpu.Dockerfile` (+2 -0) 📝 `startup-gpu.sh` (+7 -1) </details> ### 📄 Description This is one possible way to do this. What do you think? If you launch the container in the `nvidia-docker` runtime, the startup process will try to ensure you have an appropriate Xorg driver installed. It will look for one in `/usr/lib/x86_64-linux-gnu/nvidia/xorg` (which is where the Ubuntu package installs the driver), and it will look in `/nvidia/xorg` (where you can use Docker to map in your host's driver, or another specific version). If it doesn't find one in one of those locations, it tries to identify an Ubuntu package with the same exact version as the host's driver. If it can find such a package, it installs it. If it can't, it prints a failure message. It will still _attempt_ to launch Xorg in this case, but it warns that it may fail. If you want to skip all this and just assume that the driver will be there, you can set the environment variable `NVIDIA_SKIP_DRIVER_CHECK` to `1`. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-15 12:49:08 +00:00