From 4edae1724d14cef30ce2e5617e28f5d4b0135724 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rafael=20Fern=C3=A1ndez=20L=C3=B3pez?=
 <ereslibre@ereslibre.es>
Date: Wed, 15 Oct 2025 16:34:18 +0200
Subject: [PATCH 1/2] nvidia-container-toolkit: wait for all device nodes to be
 present

---
 .../nvidia-container-toolkit/default.nix      | 47 ++++++++++++++++++-
 1 file changed, 46 insertions(+), 1 deletion(-)

diff --git a/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix b/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix
index 35a32b9ddcdb..7361b039a1ac 100644
--- a/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix
+++ b/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix
@@ -188,6 +188,10 @@
         ])
       ];
 
+      services.udev.extraRules = ''
+        KERNEL=="nvidia", RUN+="${pkgs.systemd}/bin/systemctl restart nvidia-container-toolkit-cdi-generator.service'"
+      '';
+
       virtualisation = {
         containers.containersConf.settings = {
           engine = {
@@ -289,11 +293,52 @@
 
       systemd.services.nvidia-container-toolkit-cdi-generator = {
         description = "Container Device Interface (CDI) for Nvidia generator";
+        requiredBy = lib.mkMerge [
+          (lib.mkIf config.virtualisation.docker.enable [ "docker.service" ])
+          (lib.mkIf config.virtualisation.podman.enable [ "podman.service" ])
+        ];
         wantedBy = [ "multi-user.target" ];
-        after = [ "systemd-udev-settle.service" ];
         serviceConfig = {
           RuntimeDirectory = "cdi";
           RemainAfterExit = true;
+          ExecStartPre = pkgs.writeShellScript "wait-for-nvidia-devices" ''
+            set -eu
+
+            gpus_dir="/proc/driver/nvidia/gpus"
+            max_wait_seconds=60
+
+            if [ ! -d "$gpus_dir" ]; then
+              echo "wait-for-nvidia-devices: $gpus_dir does not exist; nothing to wait for."
+              exit 0
+            fi
+
+            gpu_count=$(find "$gpus_dir" -mindepth 1 -maxdepth 1 -type d 2>/dev/null | wc -l | tr -d ' ')
+
+            if [ "$gpu_count" -eq 0 ]; then
+              echo "wait-for-nvidia-devices: no GPU entries found in $gpus_dir; nothing to wait for."
+              exit 0
+            fi
+
+            echo "wait-for-nvidia-devices: expecting $gpu_count /dev/nvidiaN device node(s)."
+
+            elapsed=0
+            while true; do
+              dev_count=$(find /dev -mindepth 1 -maxdepth 1 -type c -regex '.*/nvidia[0-9]+' 2>/dev/null | wc -l | tr -d ' ')
+
+              if [ "$dev_count" -eq "$gpu_count" ]; then
+                echo "wait-for-nvidia-devices: found $dev_count matching device node(s)."
+                exit 0
+              fi
+
+              if [ "$elapsed" -ge "$max_wait_seconds" ]; then
+                echo "wait-for-nvidia-devices: timed out after $max_wait_seconds seconds; expected $gpu_count node(s) but found $dev_count." >&2
+                exit 1
+              fi
+
+              sleep 1
+              elapsed=$((elapsed + 1))
+            done
+          '';
           ExecStart =
             let
               script = pkgs.callPackage ./cdi-generate.nix {

From f246b4444454b43048fed8655ad9651b16dda51a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rafael=20Fern=C3=A1ndez=20L=C3=B3pez?=
 <ereslibre@ereslibre.es>
Date: Tue, 4 Nov 2025 18:29:23 +0100
Subject: [PATCH 2/2] chore: only pull systemctl binary

---
 .../services/hardware/nvidia-container-toolkit/default.nix      | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix b/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix
index 7361b039a1ac..9e715762fe9a 100644
--- a/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix
+++ b/nixos/modules/services/hardware/nvidia-container-toolkit/default.nix
@@ -189,7 +189,7 @@
       ];
 
       services.udev.extraRules = ''
-        KERNEL=="nvidia", RUN+="${pkgs.systemd}/bin/systemctl restart nvidia-container-toolkit-cdi-generator.service'"
+        KERNEL=="nvidia", RUN+="${lib.getExe' config.systemd.package "systemctl"} restart nvidia-container-toolkit-cdi-generator.service'"
       '';
 
       virtualisation = {