Merge branch 'frcc-kvm' of https://git.floweytf.com/server/server-config into frcc-kvm

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

.github/workflows/diff-pr.yml

@@ -17,18 +17,10 @@ jobs:
     steps:
       - uses: actions/checkout@v4
         with:
-          ref: ${{ github.event.pull_request.head.sha }}
           fetch-depth: 0
 
-      - name: merge base branch
-        run: |
-          git config user.email "ci@localhost"
-          git config user.name "CI"
-          git fetch origin ${{ github.event.pull_request.base.ref }}
-          git merge --no-edit origin/${{ github.event.pull_request.base.ref }}
-
       - name: worktree base for diff
-        run: git worktree add "$RUNNER_TEMP/base" "origin/${{ github.event.pull_request.base.ref }}"
+        run: git worktree add "$RUNNER_TEMP/base" "${{ github.event.pull_request.base.sha }}"
 
       - name: build closures and render diff
         run: |

.github/workflows/flake-update.yml

@@ -5,19 +5,21 @@ on:
     - cron: "17 4 * * *"
   workflow_dispatch:
 
+permissions:
+  contents: write
+  pull-requests: write
+
 jobs:
   update:
     runs-on: nix
     steps:
       - uses: actions/checkout@v4
-        with:
-          persist-credentials: false
 
       - run: nix flake update -L
 
       - name: open/update PR
         env:
-          GH_TOKEN: ${{ secrets.BOT_TOKEN }}
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           API: ${{ github.api_url }}
           REPO: ${{ github.repository }}
         run: |

.github/workflows/vulnix.yml

@@ -5,6 +5,10 @@ on:
     - cron: "0 6 * * 1"
   workflow_dispatch:
 
+permissions:
+  contents: read
+  issues: write
+
 jobs:
   scan:
     runs-on: nix

ci.nix

@@ -1,17 +1,10 @@
 { lib, pkgs }:
-let
-  # vulnix's bundled nix-2.34 broke its own --experimental-features detection
-  nixConf = pkgs.writeText "ci-nix.conf" "experimental-features = nix-command flakes";
-in
 # tools must track ci.py's subprocess calls; nix stays ambient (uses the
 # runner's own nix, not a pinned one).
 pkgs.writers.writePython3Bin "ci" {
   flakeIgnore = [ "E501" ];
   libraries = [ pkgs.python3Packages.subprocess-tee ];
   makeWrapperArgs = [
-    "--set-default"
-    "NIX_USER_CONF_FILES"
-    (toString nixConf)
     "--prefix"
     "PATH"
     ":"

ci.py

@@ -8,7 +8,6 @@ import re
 import subprocess
 import subprocess_tee
 import sys
-import urllib.error
 import urllib.request
 from collections import Counter
 from pathlib import Path
@@ -19,18 +18,13 @@ def run(*args, **kw):
 
 
 def nix(*args):
-    # tee to stderr - stdout leaks into GITHUB_OUTPUT $() captures
-    return subprocess_tee.run(("nix", "--extra-experimental-features", "nix-command flakes", "-L", *args), check=True, text=True, stdout=sys.stderr).stdout
+    return subprocess_tee.run(("nix", "--extra-experimental-features", "nix-command flakes", "-L", *args), check=True, text=True).stdout
 
 
 def git(*args, check=True):
     return run("git", *args, check=check, capture_output=True)
 
 
-class GiteaError(Exception):
-    pass
-
-
 def gitea(method, path, body=None):
     url = f'{os.environ["API"]}/repos/{os.environ["REPO"]}{path}'
     data = json.dumps(body).encode() if body is not None else None
@@ -38,20 +32,13 @@ def gitea(method, path, body=None):
     req.add_header("Authorization", f'token {os.environ["GH_TOKEN"]}')
     if data is not None:
         req.add_header("Content-Type", "application/json")
-    try:
-        with urllib.request.urlopen(req) as resp:
-            raw = resp.read()
-    except urllib.error.HTTPError as e:
-        try:
-            detail = json.loads(e.read()).get("message", e.reason)
-        except (json.JSONDecodeError, UnicodeDecodeError):
-            detail = e.reason
-        raise GiteaError(f"{method} {path}: {e.code} {detail}") from None
+    with urllib.request.urlopen(req) as resp:
+        raw = resp.read()
     return json.loads(raw) if raw else None
 
 
 def find_open_issue(label):
-    issues = gitea("GET", f"/issues?state=open&type=issues&labels={label}")
+    issues = gitea("GET", f"/issues?state=open&labels={label}")
     return issues[0] if issues else None
 
 
@@ -248,7 +235,7 @@ def format_vulnix(vulns):
     for v in vulns:
         for cve in v["affected_by"]:
             score = v.get("cvssv3_basescore", {}).get(cve)
-            desc = v.get("description", {}).get(cve, "").replace("\n", " ").replace("|", "∣")
+            desc = v.get("description", {}).get(cve, "")
             if len(desc) > 120:
                 desc = desc[:117] + "..."
             rows.append((v["pname"], v["version"], cve, score, desc))
@@ -267,7 +254,7 @@ def format_vulnix(vulns):
 
 def vulnix(args):
     paths = list(build_hosts(args.ref).values())
-    result = run("vulnix", "-j", *paths, capture_output=True)
+    result = run("vulnix", "-j", "-C", *paths, capture_output=True)
     if result.returncode not in (0, 2):
         print(result.stderr, file=sys.stderr)
         sys.exit(result.returncode)
@@ -475,24 +462,16 @@ def open_pr(args):
 
     host = os.environ["API"].removesuffix("/api/v1")
     auth = base64.b64encode(f'token:{os.environ["GH_TOKEN"]}'.encode()).decode()
-    push = git("-c", f"http.extraHeader=Authorization: Basic {auth}",
-               "push", "-f", f'{host}/{os.environ["REPO"]}.git', f"HEAD:{args.branch}",
-               check=False)
-    if push.returncode != 0:
-        raise RuntimeError(f"git push failed:\n{push.stderr}")
+    git("-c", f"http.extraHeader=Authorization: Basic {auth}",
+        "push", "-f", f'{host}/{os.environ["REPO"]}.git', f"HEAD:{args.branch}")
 
-    pulls = gitea("GET", f"/pulls?state=open&base_branch={args.base}")
-    existing = next((pr for pr in pulls if pr["head"]["ref"] == args.branch), None)
-    if existing:
-        print(f"PR #{existing['number']} already open, pushed update")
+    owner = os.environ["REPO"].split("/")[0]
+    if gitea("GET", f"/pulls?state=open&head={owner}:{args.branch}"):
         return
     labels = [label_id(args.label)] if args.label else []
-    pr = gitea("POST", "/pulls",
-               {"title": args.title, "head": args.branch, "base": args.base,
-                "body": Path(args.body_file).read_text(), "labels": labels})
-    if not pr or "number" not in pr:
-        raise GiteaError(f"POST /pulls: unexpected response: {json.dumps(pr)}")
-    print(f"opened PR #{pr['number']}")
+    gitea("POST", "/pulls",
+          {"title": args.title, "head": args.branch, "base": args.base,
+           "body": Path(args.body_file).read_text(), "labels": labels})
 
 
 def main():

configuration.nix

@@ -4,27 +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 = [
-      "flowey@arch namespaces=\"git\" ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIPaGpTX3EzfezVc2nl3MgsKVamYYP2h2VJaRny8pZkvi"
-    ];
-    stateDir = hostInfo.paths.config;
-    repoOwner = "server";
-  };
-
   assertions =
     let
       isMounted =
@@ -62,14 +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 = {

containers/server-n0/mail.nix

@@ -146,8 +146,6 @@ let
 in
 mkContainer {
   inherit name;
-  includedInTests = false; # TODO: fixme, clearly mail is flakey
-
   net.access = [ { to = "@internet"; } ];
   svc.autoStart = true;
   net.forward.tcp = [

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

@@ -1,96 +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 = [ ];
-    };
-    
-    repoOwner = lib.mkOption {
-      type = lib.types.str;
-      description = "User that owns the configuration checkout";
-    };
-
-    stateDir = lib.mkOption {
-      type = lib.types.str;
-    };
-
-    hostname = lib.mkOption {
-      type = lib.types.str;
-      default = config.networking.hostName;
-    };
-  };
-
-  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";
-        REPO_OWNER = cfg.repoOwner;
-        HOSTNAME = cfg.hostname;
-      };
-
-      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

@@ -1,18 +0,0 @@
-{
-  writeShellApplication,
-  git,
-  curl,
-  jq,
-  util-linux,
-  openssh,
-}:
-import ./mk-script.nix { inherit writeShellApplication; } {
-  name = "server-autodeploy";
-  runtimeInputs = [
-    git
-    curl
-    jq
-    util-linux
-    openssh
-  ];
-}

scripts/server-autodeploy.sh

@@ -1,62 +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>"$STATE_DIR/lock"
-flock -n 9 || { echo "another deploy running"; exit 0; }
-
-runas() { runuser -u "$REPO_OWNER" -- "$@"; }
-
-report_status() {
-  echo "updating status to: $1 - $2"
-  [[ -f "$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="$ALLOWED_SIGNERS"
-
-runas git fetch origin --tags --prune-tags
-
-latest=$(runas 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: ${current:-none})"
-
-runas git tag -v "$latest" || {
-  sha=$(runas git rev-list -1 "$latest")
-  report_status "failure" "tag signature verification failed" "$sha"
-  exit 1
-}
-
-sha=$(runas git rev-list -1 "$latest")
-report_status "pending" "deploying $latest" "$sha"
-
-runas git checkout "$latest"
-runas tee .nixos-label <<< "$latest" >/dev/null
-runas git add .nixos-label
-
-if nixos-rebuild switch --flake ".#$HOSTNAME"; then
-  echo "$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

@@ -153,8 +153,6 @@ with subtest("(functional): web services respond"):
     for name, port in [("devenv", 8080), ("maven", 8080), ("git", 3000)]:
         server.succeed(net_cmd(name, f"{CURL} -sf --max-time 5 -o /dev/null http://{ADDR[name]}:{port}/ >&2"))
 
-# TODO: re-enable mail
-"""
 with subtest("(functional): mail accepts connections"):
     smtp_probe = net_cmd("mail", f"{SH} -c 'exec 3<>/dev/tcp/{ADDR['mail']}/25 && head -1 <&3'")
     server.wait_until_succeeds(f"{smtp_probe} | grep -q '^220'", timeout=60)
@@ -169,7 +167,6 @@ with subtest("(functional): mail accepts connections"):
         return greeting.strip().startswith("* OK")
 
     retry(imap_ready, timeout_seconds=60)
-"""
 
 with subtest("(functional): nginx proxies to backend containers"):
     server.succeed(
@@ -200,7 +197,7 @@ with subtest("(isolation): undeclared container edges are refused"):
         ("devenv", "git", 3000),
         ("git", "devenv", 8080),
         ("maven", "git", 3000),
-        # ("nginx", "mail", 25),
+        ("nginx", "mail", 25),
         ("ssh-botm", "ssh-admin", 22),
         ("overseer", "git", 3000),
     ]:
@@ -253,7 +250,7 @@ with subtest("(isolation): renaming the interface changes nothing"):
     rename(iface, "evil0")
     try:
         server.succeed(ct_connect("nginx", ADDR["git"], 3000))
-        # server.fail(ct_connect("nginx", ADDR["mail"], 25, timeout=2))
+        server.fail(ct_connect("nginx", ADDR["mail"], 25, timeout=2))
     finally:
         rename("evil0", iface)