do nvidia myself because upstream nvidia has been broken for days

Author: Synby

Containerfile

@@ -4,13 +4,16 @@ COPY build_files /build
 COPY system_files /files
 COPY cosign.pub /files/etc/pki/containers/zirc-addons.pub
 
-FROM ghcr.io/zirconium-dev/zirconium-nvidia:latest
+FROM ghcr.io/zirconium-dev/zirconium:latest
 
 RUN --mount=type=bind,from=ctx,source=/,target=/ctx \
     --mount=type=tmpfs,dst=/var \
     --mount=type=tmpfs,dst=/tmp \
     /ctx/build/01-packages.sh
 
+RUN --mount=type=bind,from=ctx,source=/,target=/ctx \
+    /ctx/build/02-nvidia.sh
+
 RUN --mount=type=bind,from=ctx,source=/,target=/ctx \
     /ctx/build/99-cleanup.sh
 

build_files/02-nvidia.sh

@@ -0,0 +1,70 @@
+#!/usr/bin/env bash
+
+set -xeuo pipefail
+uname -r
+KERNEL_VERSION="$(find "/usr/lib/modules" -maxdepth 1 -type d ! -path "/usr/lib/modules" -exec basename '{}' ';' | sort | tail -n 1)"
+
+dnf config-manager addrepo --from-repofile=https://negativo17.org/repos/fedora-nvidia.repo
+dnf config-manager setopt fedora-nvidia.enabled=0
+sed -i '/^enabled=/a\priority=90' /etc/yum.repos.d/fedora-nvidia.repo
+
+dnf -y install \
+    gcc-c++ \
+    kernel-devel-matched
+
+dnf -y install --enablerepo=fedora-nvidia akmod-nvidia
+mkdir -p /var/tmp # for akmods
+chmod 1777 /var/tmp
+    
+akmods --force --kernels "${KERNEL_VERSION}" --kmod "nvidia"
+cat /var/cache/akmods/nvidia/*.failed.log || true
+
+dnf -y install --enablerepo=fedora-nvidia \
+    nvidia-driver-cuda libnvidia-fbc libva-nvidia-driver nvidia-driver nvidia-modprobe nvidia-persistenced nvidia-settings
+
+dnf config-manager addrepo --from-repofile=https://nvidia.github.io/libnvidia-container/stable/rpm/nvidia-container-toolkit.repo
+dnf config-manager setopt nvidia-container-toolkit.enabled=0
+dnf config-manager setopt nvidia-container-toolkit.gpgcheck=1
+
+dnf -y install --enablerepo=nvidia-container-toolkit \
+    nvidia-container-toolkit
+
+curl --retry 3 -L https://raw.githubusercontent.com/NVIDIA/dgx-selinux/master/bin/RHEL9/nvidia-container.pp -o nvidia-container.pp
+semodule -i nvidia-container.pp
+rm -f nvidia-container.pp
+
+tee /usr/lib/modprobe.d/00-nouveau-blacklist.conf <<'EOF'
+blacklist nouveau
+options nouveau modeset=0
+EOF
+
+tee /usr/lib/bootc/kargs.d/00-nvidia.toml <<'EOF'
+kargs = ["rd.driver.blacklist=nouveau", "modprobe.blacklist=nouveau", "nvidia-drm.modeset=1"]
+EOF
+
+# Universal Blue specific Initramfs fixes
+mv /etc/modprobe.d/nvidia-modeset.conf /usr/lib/modprobe.d/nvidia-modeset.conf
+# we must force driver load to fix black screen on boot for nvidia desktops
+sed -i 's/omit_drivers/force_drivers/g' /usr/lib/dracut/dracut.conf.d/99-nvidia.conf
+# as we need forced load, also must pre-load intel/amd iGPU else chromium web browsers fail to use hardware acceleration
+sed -i 's/ nvidia / i915 amdgpu nvidia /g' /usr/lib/dracut/dracut.conf.d/99-nvidia.conf
+
+tee /usr/lib/systemd/system/nvctk-cdi.service <<'EOF'
+[Unit]
+Description=nvidia container toolkit CDI auto-generation
+ConditionFileIsExecutable=/usr/bin/nvidia-ctk
+After=local-fs.target
+
+[Service]
+Type=oneshot
+ExecStart=/usr/bin/nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml
+
+[Install]
+WantedBy=multi-user.target
+EOF
+
+systemctl enable nvctk-cdi.service
+systemctl disable akmods-keygen@akmods-keygen.service
+systemctl mask akmods-keygen@akmods-keygen.service
+systemctl disable akmods-keygen.target
+systemctl mask akmods-keygen.target