stornic56/debianito-post-install
1
0
Fork 0
mirror of https://github.com/stornic56/debianito-post-install.git synced 2026-09-14 06:02:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

nvidia & zram fixes

Browse Source 
- Fixed critical NVIDIA driver bug on Debian 13 (Trixie) where repo configuration installed v550 instead of selected v595. Implemented version-specific enabling: cuda-keyring for Trixie, extrepo for Bookworm. Added explicit apt update after repo enable and candidate verification before pinning checks to abort cleanly if expected driver unavailable.
- Rewrote `_install_nvidia_standard()` with auto-detected architecture via `_get_nvidia_arch_family()`. Turing/Ampere/Ada/Blackwell → nvidia-open-kernel-dkms (open module); Maxwell/Pascal/Volta/unknown/empty → nvidia-kernel-dkms (closed, safe fallback). Removed all `nvidia-detect` references across modules and docs.
- Fixed `_install_nvidia_cuda_repo()` to use correct CUDA repository package names: v590/v595 → nvidia-open metapackage (includes nvidia-kernel-open-dkms); v550 and below → nvidia-driver. Added dynamic warning message based on installed module type.
- Fixed uninitialized variable bug in `modules/gaming/tools.sh` causing script abortion under `set -u`. Initialized all local variables (`deb_url`, `sha256`, `json`, `deb_suffix`) with default values and added defensive initialization for `$USER` environment variable.
- Fixed ZRAM configuration persistence bug where old config persisted after changes. Added device reset sequence (swapoff + modprobe -r) before writing new config, post-config verification check, and changed default recommendation from 50% to 25% for systems with >16GB RAM.
- Added `_warn_nvidia_gnome_wayland()` function in `gpu.sh` to warn about Debian bug #1109409 (GDM3 + NVIDIA + Wayland black screen issue). Warning displays only when gdm3 present, GNOME Shell 4.x detected, and NVIDIA driver installed.
- Standardized menu headers across all modules using centralized SCROLL_HINT variable, replacing hardcoded strings with uniform Whiptail instructions. Updated documentation for cuda-keyring (Trixie) and ZRAM configuration flow diagrams.
This commit is contained in:
stornic56
2026-08-05 21:13:32 -05:00
committed by GitHub
parent 0ec9639030
commit 59c93ba459
6 changed files with 168 additions and 127 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/gpu.md
+2 -2
View File
@@ -175,7 +175,7 @@ The NVIDIA driver installation process is inherently complex due to proprietary
#### **Blackwell Architecture & CUDA v590**
- **Detection**: `_helpers.sh` function `is_nvidia_blackwell()` identifies GPUs via PCI IDs `10de:24xx`, `0x2900–0x29BF`, and `0x2B80–0x31FF`.
- **Reason for v590**: Debian 13 (Trixie) stable drivers only support up to v550, which lacks Blackwell (GB20x) architecture. The NVIDIA CUDA repository provides production branch v590 with unified driver packages. Specifically, the goal is to install the latest version of the nvidia-driver from the 590 branch, which would be [590.48.01](https://download.nvidia.com/XFree86/Linux-x86_64/590.48.01/README/supportedchips.html).
- **Extrepo Mechanism**: Enables `nvidia-cuda` repository via `extrepo`, creates APT pinning in `/etc/apt/preferences.d/block-nvidia` to lock to version `590.*`, and installs `nvidia-driver-pinning-590`, `nvidia-driver`, and `firmware-nvidia-gsp`.
- **CUDA Repository Enablement**: On Debian 12 (Bookworm) the `nvidia-cuda` repository is enabled via `extrepo`. On Debian 13 (Trixie) the official `cuda-keyring` package is used instead (`wget` + `dpkg -i`), since extrepo cannot configure the repo correctly on Trixie. After enabling, the script always runs an explicit `apt update` and verifies the APT candidate matches the requested branch (`595.*`) before writing APT pinning in `/etc/apt/preferences.d/block-nvidia` and installing `nvidia-driver-pinning-<ver>`, `nvidia-driver`, and `firmware-nvidia-gsp`. If the candidate does not match, the installation aborts cleanly.
#### **Installation Flow**
```bash
@@ -212,7 +212,7 @@ Depending on your GPU generation and your Debian ecosystem, you must select the
* **Blackwell (RTX 5000)**: Not officially supported by standard Debian drivers yet. The script provides a path to the Enterprise Repo for users who need this hardware to function immediately.
#### **Kepler Interception in Bookworm:**
When Kepler is detected on Bookworm, `nvidia.sh` bypasses `nvidia-detect` (which might recommend v535) and forces installation of `nvidia-tesla-470-driver`:
When Kepler is detected on Bookworm, `nvidia.sh` forces installation of `nvidia-tesla-470-driver` (the modern v535/v550 branches do not support Kepler):
```bash
if [ "$is_kepler" = "true" ] && [ "$DEBIAN_CODENAME" = "bookworm" ]; then
nv_pkg="nvidia-tesla-470-driver"