actually bind /dev/kvm into the container, and test it initializes

DeviceAllow= only grants cgroup permission to open a device node - it
doesn't create one, and nspawn can't mknod a device inside a user
namespace. The node needs an explicit bind from the host as well, so
both are required together.

Also add a test that actually boots qemu-system-x86_64 -enable-kvm
(no disk, killed by timeout) and checks for "Permission denied" /
"Could not access KVM" in its output, rather than just checking the
node opens - a wrong-gid or unmapped node still opens but fails the
ioctls qemu needs.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Dzy7tEKLCFFXD96LmR8we3

configuration.nix

@@ -4,26 +4,12 @@
   lib,
   hostInfo,
   customPkgs,
-  flakeRev,
   ...
 }:
 let
   inherit (hostInfo) paths;
 in
 {
-  imports = [ ./modules/autodeploy.nix ];
-
-  services.server-autodeploy = {
-    enable = true;
-    giteaUrl = "https://git.floweytf.com/";
-    giteaRepo = "server/server-config";
-    tokenFile = "${hostInfo.paths.secrets}/gitea-autodeploy-token";
-    allowedSigners = [
-      "ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIPaGpTX3EzfezVc2nl3MgsKVamYYP2h2VJaRny8pZkvi flowey@arch"
-    ];
-    stateDir = hostInfo.paths.config;
-  };
-
   assertions =
     let
       isMounted =
@@ -61,14 +47,6 @@ in
 
   time.timeZone = "UTC";
   system.stateVersion = "26.05";
-  system.nixos.label =
-    let
-      labelFile = ./.nixos-label;
-    in
-    if builtins.pathExists labelFile then
-      lib.removeSuffix "\n" (builtins.readFile labelFile)
-    else
-      flakeRev;
 
   nix = {
     settings = {

dsl.nix

@@ -222,13 +222,15 @@ rec {
     : Bind mounts, keyed by mount point.
 
     `resources.devices`
-    : Host device nodes to expose, as paths, added to the unit's
-      `DeviceAllow=`. nspawn populates a handful of privileged devices
-      (`/dev/kvm`, `/dev/net/tun`, ...) into the container's own `/dev` once
-      the cgroup allows it, and silently skips ones the host doesn't have -
-      no `resources.fs` bind mount needed. Nodes keep their host ownership,
-      so a consumer inside needs `isolation.privateUsers = "identity"` for
-      the owning gid to mean anything.
+    : Host device nodes to expose, as paths. Bound into the container at the
+      same path and added to the unit's `DeviceAllow=` - the cgroup
+      permission alone doesn't get you a working node (nothing can `mknod` a
+      device inside a user namespace), and the bind alone leaves the cgroup
+      denying `open()`, so both are needed. The bind is unconditional, so
+      the device must actually exist on whatever host runs the container.
+      Nodes keep their host ownership, so a consumer inside needs
+      `isolation.privateUsers = "identity"` for the owning gid to mean
+      anything.
 
     `isolation.nixDaemon`
     : Whether the Nix daemon and CLI are available inside the container.
@@ -295,6 +297,10 @@ rec {
         + lib.optionalString useIdmap ":idmap"
       ) allBindMounts;
       bindMountHostPaths = map (bm: bm.hostPath) (lib.attrValues allBindMounts);
+      # DeviceAllow= alone only grants cgroup permission - nspawn can't
+      # mknod a device node inside a user namespace, so the node still
+      # needs an explicit bind from the host.
+      deviceFlags = map (d: "--bind=${d}") devices;
       inherit (deriveIdentity ifPrefix.nspawn name)
         iface
         mac
@@ -381,7 +387,7 @@ rec {
       nixosContainer = {
         inherit autoStart privateUsers;
         bindMounts = { };
-        extraFlags = bindMountExtraFlags;
+        extraFlags = bindMountExtraFlags ++ deviceFlags;
         privateNetwork = true;
         extraVeths = linkVeths;
         config =

flake.nix

@@ -11,7 +11,6 @@
 
   outputs =
     {
-      self,
       nixpkgs,
       mtce,
       microvm,
@@ -22,8 +21,6 @@
       inherit (nixpkgs) lib;
       pkgs = import nixpkgs { inherit system; };
 
-      flakeRev = self.shortRev or self.dirtyShortRev or "unknown";
-
       customPkgs = {
         hopper-rs = pkgs.callPackage ./packages/hopper-rs.nix { };
         sshpiper = pkgs.callPackage ./packages/sshpiper.nix { };
@@ -32,7 +29,6 @@
         server-rebuild = pkgs.callPackage ./packages/server-rebuild.nix { };
         server-status = pkgs.callPackage ./packages/server-status.nix { };
         server-kexec = pkgs.callPackage ./packages/server-kexec.nix { };
-        server-autodeploy = pkgs.callPackage ./packages/server-autodeploy.nix { };
       };
 
       ci = import ./ci.nix { inherit lib pkgs; };
@@ -50,13 +46,12 @@
           server = "/persist/@containers";
           data = "/persist/@containers/data";
           secrets = "/persist/@secrets/";
-          config = "/persist/@config/";
         };
       };
 
       specialArgsFor = hostName: {
         hostInfo = hostInfoFor hostName;
-        inherit inputs customPkgs flakeRev;
+        inherit inputs customPkgs;
       };
 
       hostDefFor =
@@ -152,7 +147,6 @@
           server-rebuild
           server-status
           server-kexec
-          server-autodeploy
           ;
       };
     };

modules/autodeploy.nix

@@ -1,84 +0,0 @@
-{
-  config,
-  lib,
-  pkgs,
-  customPkgs,
-  ...
-}:
-let
-  cfg = config.services.server-autodeploy;
-in
-{
-  options.services.server-autodeploy = {
-    enable = lib.mkEnableOption "automatic deployment from signed git tags";
-
-    interval = lib.mkOption {
-      type = lib.types.str;
-      default = "15min";
-    };
-
-    branch = lib.mkOption {
-      type = lib.types.str;
-      default = "master";
-    };
-
-    giteaUrl = lib.mkOption {
-      type = lib.types.str;
-    };
-
-    giteaRepo = lib.mkOption {
-      type = lib.types.str;
-    };
-
-    tokenFile = lib.mkOption {
-      type = lib.types.str;
-    };
-
-    allowedSigners = lib.mkOption {
-      type = lib.types.listOf lib.types.str;
-      default = [ ];
-    };
-
-    stateDir = lib.mkOption {
-      type = lib.types.str;
-    };
-  };
-
-  config = lib.mkIf cfg.enable {
-    systemd.services.server-autodeploy = {
-      description = "Deploy latest signed tag";
-      after = [ "network-online.target" ];
-      wants = [ "network-online.target" ];
-
-      environment = {
-        GITEA_URL = cfg.giteaUrl;
-        GITEA_REPO = cfg.giteaRepo;
-        TOKEN_FILE = cfg.tokenFile;
-        ALLOWED_SIGNERS = builtins.toFile "signers" (lib.join "\n" cfg.allowedSigners);
-        DEPLOY_BRANCH = cfg.branch;
-        STATE_DIR = cfg.stateDir;
-        REPO_DIR = "/persist/@configuration";
-      };
-
-      serviceConfig = {
-        Type = "oneshot";
-        ExecStart = lib.getExe customPkgs.server-autodeploy;
-        TimeoutStartSec = "30min";
-      };
-
-      path = [
-        config.nix.package
-        pkgs.nixos-rebuild
-      ];
-    };
-
-    systemd.timers.server-autodeploy = {
-      wantedBy = [ "timers.target" ];
-      timerConfig = {
-        OnBootSec = "2min";
-        OnUnitActiveSec = cfg.interval;
-        RandomizedDelaySec = "2min";
-      };
-    };
-  };
-}

packages/server-autodeploy.nix

@@ -1,14 +0,0 @@
-{
-  writeShellApplication,
-  git,
-  curl,
-  jq,
-}:
-import ./mk-script.nix { inherit writeShellApplication; } {
-  name = "server-autodeploy";
-  runtimeInputs = [
-    git
-    curl
-    jq
-  ];
-}

scripts/server-autodeploy.sh

@@ -1,58 +0,0 @@
-REPO_DIR="${REPO_DIR:-/persist/@configuration}"
-STATE_DIR="${STATE_DIR:-/persist/@autodeploy}"
-BRANCH="origin/${DEPLOY_BRANCH:-master}"
-
-mkdir -p "$STATE_DIR"
-exec 9>"$STATE_DIR/lock"
-flock -n 9 || { echo "another deploy running"; exit 0; }
-
-report_status() {
-  [[ -f "$TOKEN_FILE" ]] || return 0
-  curl -sf -X POST \
-    -H "Authorization: token $(<"$TOKEN_FILE")" \
-    -H "Content-Type: application/json" \
-    -d "$(jq -n --arg s "$1" --arg d "$2" --arg c "deploy/$(hostname)" \
-      '{state:$s,description:$d,context:$c}')" \
-    "$GITEA_URL/api/v1/repos/$GITEA_REPO/statuses/$3" >/dev/null
-}
-
-cd "$REPO_DIR"
-
-export GIT_CONFIG_COUNT=2
-export GIT_CONFIG_KEY_0=gpg.format
-export GIT_CONFIG_VALUE_0=ssh
-export GIT_CONFIG_KEY_1=gpg.ssh.allowedSignersFile
-export GIT_CONFIG_VALUE_1="$ALLOWED_SIGNERS"
-
-git fetch origin --tags --prune-tags
-
-latest=$(git describe --abbrev=0 "$BRANCH" 2>/dev/null) || {
-  echo "no tags on $BRANCH"
-  exit 0
-}
-
-current=$(cat "$STATE_DIR/tag" 2>/dev/null) || true
-[[ "$latest" != "$current" ]] || { echo "already at $latest"; exit 0; }
-
-echo "new tag: $latest (current: ${current:-none})"
-
-git tag -v "$latest" || {
-  sha=$(git rev-list -1 "$latest")
-  report_status "failure" "tag signature verification failed" "$sha"
-  exit 1
-}
-
-sha=$(git rev-list -1 "$latest")
-report_status "pending" "deploying $latest" "$sha"
-
-# build from remote, pinned to the verified rev - no local checkout needed
-flake_url="git+${GITEA_URL}/${GITEA_REPO}?ref=refs/tags/${latest}&rev=${sha}#$(hostname)"
-
-if nixos-rebuild switch --flake "$flake_url"; then
-  echo "$latest" > "$STATE_DIR/tag"
-  report_status "success" "deployed $latest" "$sha"
-  echo "deployed $latest"
-else
-  report_status "failure" "nixos-rebuild failed" "$sha"
-  exit 1
-fi

tests/server-n0/integration.py

@@ -333,9 +333,28 @@ with subtest("(functional): published ports follow the container's lifetime"):
 with subtest("(functional): declared devices are usable in-container"):
     for name, devices in DEVICES.items():
         for dev in devices:
+            # opened, not stat'd: without DeviceAllow= the node is still visible
             server.succeed(ct_cmd(name, f"exec 3<> {dev}"))
 
 
+with subtest("(functional): /dev/kvm actually initializes, not just opens"):
+    # a wrong-gid or unmapped node still opens but fails the ioctls qemu
+    # needs, so boot with no disk (never completes) and kill it - the
+    # failure mode this catches is qemu refusing to start at all.
+    for name, devices in DEVICES.items():
+        if "/dev/kvm" not in devices:
+            continue
+        out = server.succeed(
+            ct_cmd(
+                name,
+                f"{TIMEOUT} 3 qemu-system-x86_64 -enable-kvm -m 256 -nographic -no-reboot "
+                ">/tmp/qemu.out 2>&1; cat /tmp/qemu.out",
+            )
+        )
+        assert "Permission denied" not in out, f"{name}: qemu couldn't use /dev/kvm:\n{out}"
+        assert "Could not access KVM" not in out, f"{name}: qemu couldn't use /dev/kvm:\n{out}"
+
+
 with subtest("(isolation): devices reach only the container that declared them"):
     for name in enterable:
         for dev in {d for ds in DEVICES.values() for d in ds} - set(DEVICES.get(name, [])):