summary refs log tree commit diff
path: root/nixos/modules/virtualisation
diff options
context:
space:
mode:
Diffstat (limited to 'nixos/modules/virtualisation')
-rw-r--r--nixos/modules/virtualisation/azure-agent.nix18
-rw-r--r--nixos/modules/virtualisation/azure-image.nix46
-rw-r--r--nixos/modules/virtualisation/google-compute-config.nix4
-rw-r--r--nixos/modules/virtualisation/google-compute-image.nix13
-rw-r--r--nixos/modules/virtualisation/incus.nix236
-rw-r--r--nixos/modules/virtualisation/lxc-container.nix41
-rw-r--r--nixos/modules/virtualisation/lxd.nix4
-rw-r--r--nixos/modules/virtualisation/nixos-containers.nix15
-rw-r--r--nixos/modules/virtualisation/oci-common.nix2
-rw-r--r--nixos/modules/virtualisation/oci-containers.nix59
-rw-r--r--nixos/modules/virtualisation/qemu-vm.nix85
-rw-r--r--nixos/modules/virtualisation/vagrant-guest.nix1
-rw-r--r--nixos/modules/virtualisation/virtualbox-host.nix2
13 files changed, 441 insertions, 85 deletions
diff --git a/nixos/modules/virtualisation/azure-agent.nix b/nixos/modules/virtualisation/azure-agent.nix
index 6e6021cf80f..e712fac17a4 100644
--- a/nixos/modules/virtualisation/azure-agent.nix
+++ b/nixos/modules/virtualisation/azure-agent.nix
@@ -61,7 +61,7 @@ in
 
         # Which provisioning agent to use. Supported values are "auto" (default), "waagent",
         # "cloud-init", or "disabled".
-        Provisioning.Agent=disabled
+        Provisioning.Agent=auto
 
         # Password authentication for root account will be unavailable.
         Provisioning.DeleteRootPassword=n
@@ -241,7 +241,16 @@ in
       after = [ "network-online.target" "sshd.service" ];
       wants = [ "network-online.target" ];
 
-      path = [ pkgs.e2fsprogs pkgs.bash ];
+      path = [
+        pkgs.e2fsprogs
+        pkgs.bash
+
+        # waagent's Microsoft.OSTCExtensions.VMAccessForLinux needs Python 3
+        pkgs.python39
+
+        # waagent's Microsoft.CPlat.Core.RunCommandLinux needs lsof
+        pkgs.lsof
+      ];
       description = "Windows Azure Agent Service";
       unitConfig.ConditionPathExists = "/etc/waagent.conf";
       serviceConfig = {
@@ -250,5 +259,10 @@ in
       };
     };
 
+    # waagent will generate files under /etc/sudoers.d during provisioning
+    security.sudo.extraConfig = ''
+      #includedir /etc/sudoers.d
+    '';
+
   };
 }
diff --git a/nixos/modules/virtualisation/azure-image.nix b/nixos/modules/virtualisation/azure-image.nix
index 17cfd393830..d909680cca1 100644
--- a/nixos/modules/virtualisation/azure-image.nix
+++ b/nixos/modules/virtualisation/azure-image.nix
@@ -16,6 +16,13 @@ in
         Size of disk image. Unit is MB.
       '';
     };
+    virtualisation.azureImage.contents = mkOption {
+      type = with types; listOf attrs;
+      default = [ ];
+      description = lib.mdDoc ''
+        Extra contents to add to the image.
+      '';
+    };
   };
   config = {
     system.build.azureImage = import ../../lib/make-disk-image.nix {
@@ -26,46 +33,9 @@ in
       '';
       configFile = ./azure-config-user.nix;
       format = "raw";
-      inherit (cfg) diskSize;
+      inherit (cfg) diskSize contents;
       inherit config lib pkgs;
     };
 
-    # Azure metadata is available as a CD-ROM drive.
-    fileSystems."/metadata".device = "/dev/sr0";
-
-    systemd.services.fetch-ssh-keys = {
-      description = "Fetch host keys and authorized_keys for root user";
-
-      wantedBy = [ "sshd.service" "waagent.service" ];
-      before = [ "sshd.service" "waagent.service" ];
-
-      path  = [ pkgs.coreutils ];
-      script =
-        ''
-          eval "$(cat /metadata/CustomData.bin)"
-          if ! [ -z "$ssh_host_ecdsa_key" ]; then
-            echo "downloaded ssh_host_ecdsa_key"
-            echo "$ssh_host_ecdsa_key" > /etc/ssh/ssh_host_ed25519_key
-            chmod 600 /etc/ssh/ssh_host_ed25519_key
-          fi
-
-          if ! [ -z "$ssh_host_ecdsa_key_pub" ]; then
-            echo "downloaded ssh_host_ecdsa_key_pub"
-            echo "$ssh_host_ecdsa_key_pub" > /etc/ssh/ssh_host_ed25519_key.pub
-            chmod 644 /etc/ssh/ssh_host_ed25519_key.pub
-          fi
-
-          if ! [ -z "$ssh_root_auth_key" ]; then
-            echo "downloaded ssh_root_auth_key"
-            mkdir -m 0700 -p /root/.ssh
-            echo "$ssh_root_auth_key" > /root/.ssh/authorized_keys
-            chmod 600 /root/.ssh/authorized_keys
-          fi
-        '';
-      serviceConfig.Type = "oneshot";
-      serviceConfig.RemainAfterExit = true;
-      serviceConfig.StandardError = "journal+console";
-      serviceConfig.StandardOutput = "journal+console";
-    };
   };
 }
diff --git a/nixos/modules/virtualisation/google-compute-config.nix b/nixos/modules/virtualisation/google-compute-config.nix
index 3c503f027d7..887af26949f 100644
--- a/nixos/modules/virtualisation/google-compute-config.nix
+++ b/nixos/modules/virtualisation/google-compute-config.nix
@@ -84,6 +84,10 @@ in
     { groups = [ "google-sudoers" ]; commands = [ { command = "ALL"; options = [ "NOPASSWD" ]; } ]; }
   ];
 
+  security.sudo-rs.extraRules = mkIf config.users.mutableUsers [
+    { groups = [ "google-sudoers" ]; commands = [ { command = "ALL"; options = [ "NOPASSWD" ]; } ]; }
+  ];
+
   users.groups.google-sudoers = mkIf config.users.mutableUsers { };
 
   boot.extraModprobeConfig = readFile "${pkgs.google-guest-configs}/etc/modprobe.d/gce-blacklist.conf";
diff --git a/nixos/modules/virtualisation/google-compute-image.nix b/nixos/modules/virtualisation/google-compute-image.nix
index 197ebb18b9a..e4a18fd81d7 100644
--- a/nixos/modules/virtualisation/google-compute-image.nix
+++ b/nixos/modules/virtualisation/google-compute-image.nix
@@ -44,10 +44,22 @@ in
         GZIP compression level of the resulting disk image (1-9).
       '';
     };
+    virtualisation.googleComputeImage.efi = mkEnableOption "EFI booting";
   };
 
   #### implementation
   config = {
+    boot.initrd.availableKernelModules = [ "nvme" ];
+    boot.loader.grub = mkIf cfg.efi {
+      device = mkForce "nodev";
+      efiSupport = true;
+      efiInstallAsRemovable = true;
+    };
+
+    fileSystems."/boot" = mkIf cfg.efi {
+      device = "/dev/disk/by-label/ESP";
+      fsType = "vfat";
+    };
 
     system.build.googleComputeImage = import ../../lib/make-disk-image.nix {
       name = "google-compute-image";
@@ -62,6 +74,7 @@ in
       '';
       format = "raw";
       configFile = if cfg.configFile == null then defaultConfigFile else cfg.configFile;
+      partitionTableType = if cfg.efi then "efi" else "legacy";
       inherit (cfg) diskSize;
       inherit config lib pkgs;
     };
diff --git a/nixos/modules/virtualisation/incus.nix b/nixos/modules/virtualisation/incus.nix
new file mode 100644
index 00000000000..3a4f0d7157a
--- /dev/null
+++ b/nixos/modules/virtualisation/incus.nix
@@ -0,0 +1,236 @@
+{ config, lib, pkgs, ... }:
+
+let
+  cfg = config.virtualisation.incus;
+  preseedFormat = pkgs.formats.yaml { };
+in
+{
+  meta.maintainers = [ lib.maintainers.adamcstephens ];
+
+  options = {
+    virtualisation.incus = {
+      enable = lib.mkEnableOption (lib.mdDoc ''
+        incusd, a daemon that manages containers and virtual machines.
+
+        Users in the "incus-admin" group can interact with
+        the daemon (e.g. to start or stop containers) using the
+        {command}`incus` command line tool, among others.
+      '');
+
+      package = lib.mkPackageOptionMD pkgs "incus" { };
+
+      lxcPackage = lib.mkPackageOptionMD pkgs "lxc" { };
+
+      preseed = lib.mkOption {
+        type = lib.types.nullOr (
+          lib.types.submodule { freeformType = preseedFormat.type; }
+        );
+
+        default = null;
+
+        description = lib.mdDoc ''
+          Configuration for Incus preseed, see
+          <https://linuxcontainers.org/incus/docs/main/howto/initialize/#non-interactive-configuration>
+          for supported values.
+
+          Changes to this will be re-applied to Incus which will overwrite existing entities or create missing ones,
+          but entities will *not* be removed by preseed.
+        '';
+
+        example = {
+          networks = [
+            {
+              name = "incusbr0";
+              type = "bridge";
+              config = {
+                "ipv4.address" = "10.0.100.1/24";
+                "ipv4.nat" = "true";
+              };
+            }
+          ];
+          profiles = [
+            {
+              name = "default";
+              devices = {
+                eth0 = {
+                  name = "eth0";
+                  network = "incusbr0";
+                  type = "nic";
+                };
+                root = {
+                  path = "/";
+                  pool = "default";
+                  size = "35GiB";
+                  type = "disk";
+                };
+              };
+            }
+          ];
+          storage_pools = [
+            {
+              name = "default";
+              driver = "dir";
+              config = {
+                source = "/var/lib/incus/storage-pools/default";
+              };
+            }
+          ];
+        };
+      };
+
+      socketActivation = lib.mkEnableOption (
+        lib.mdDoc ''
+          socket-activation for starting incus.service. Enabling this option
+          will stop incus.service from starting automatically on boot.
+        ''
+      );
+
+      startTimeout = lib.mkOption {
+        type = lib.types.ints.unsigned;
+        default = 600;
+        apply = toString;
+        description = lib.mdDoc ''
+          Time to wait (in seconds) for incusd to become ready to process requests.
+          If incusd does not reply within the configured time, `incus.service` will be
+          considered failed and systemd will attempt to restart it.
+        '';
+      };
+    };
+  };
+
+  config = lib.mkIf cfg.enable {
+    # https://github.com/lxc/incus/blob/f145309929f849b9951658ad2ba3b8f10cbe69d1/doc/reference/server_settings.md
+    boot.kernel.sysctl = {
+      "fs.aio-max-nr" = lib.mkDefault 524288;
+      "fs.inotify.max_queued_events" = lib.mkDefault 1048576;
+      "fs.inotify.max_user_instances" = lib.mkOverride 1050 1048576; # override in case conflict nixos/modules/services/x11/xserver.nix
+      "fs.inotify.max_user_watches" = lib.mkOverride 1050 1048576; # override in case conflict nixos/modules/services/x11/xserver.nix
+      "kernel.dmesg_restrict" = lib.mkDefault 1;
+      "kernel.keys.maxbytes" = lib.mkDefault 2000000;
+      "kernel.keys.maxkeys" = lib.mkDefault 2000;
+      "net.core.bpf_jit_limit" = lib.mkDefault 1000000000;
+      "net.ipv4.neigh.default.gc_thresh3" = lib.mkDefault 8192;
+      "net.ipv6.neigh.default.gc_thresh3" = lib.mkDefault 8192;
+      # vm.max_map_count is set higher in nixos/modules/config/sysctl.nix
+    };
+
+    boot.kernelModules = [
+      "veth"
+      "xt_comment"
+      "xt_CHECKSUM"
+      "xt_MASQUERADE"
+      "vhost_vsock"
+    ] ++ lib.optionals (!config.networking.nftables.enable) [ "iptable_mangle" ];
+
+    environment.systemPackages = [ cfg.package ];
+
+    # Note: the following options are also declared in virtualisation.lxc, but
+    # the latter can't be simply enabled to reuse the formers, because it
+    # does a bunch of unrelated things.
+    systemd.tmpfiles.rules = [ "d /var/lib/lxc/rootfs 0755 root root -" ];
+
+    security.apparmor = {
+      packages = [ cfg.lxcPackage ];
+      policies = {
+        "bin.lxc-start".profile = ''
+          include ${cfg.lxcPackage}/etc/apparmor.d/usr.bin.lxc-start
+        '';
+        "lxc-containers".profile = ''
+          include ${cfg.lxcPackage}/etc/apparmor.d/lxc-containers
+        '';
+      };
+    };
+
+    systemd.services.incus = {
+      description = "Incus Container and Virtual Machine Management Daemon";
+
+      wantedBy = lib.mkIf (!cfg.socketActivation) [ "multi-user.target" ];
+      after = [
+        "network-online.target"
+        "lxcfs.service"
+      ] ++ (lib.optional cfg.socketActivation "incus.socket");
+      requires = [
+        "lxcfs.service"
+      ] ++ (lib.optional cfg.socketActivation "incus.socket");
+      wants = [
+        "network-online.target"
+      ];
+
+      path = lib.mkIf config.boot.zfs.enabled [ config.boot.zfs.package ];
+
+      environment = {
+        # Override Path to the LXC template configuration directory
+        INCUS_LXC_TEMPLATE_CONFIG = "${pkgs.lxcfs}/share/lxc/config";
+      };
+
+      serviceConfig = {
+        ExecStart = "${cfg.package}/bin/incusd --group incus-admin";
+        ExecStartPost = "${cfg.package}/bin/incusd waitready --timeout=${cfg.startTimeout}";
+        ExecStop = "${cfg.package}/bin/incus admin shutdown";
+
+        KillMode = "process"; # when stopping, leave the containers alone
+        Delegate = "yes";
+        LimitMEMLOCK = "infinity";
+        LimitNOFILE = "1048576";
+        LimitNPROC = "infinity";
+        TasksMax = "infinity";
+
+        Restart = "on-failure";
+        TimeoutStartSec = "${cfg.startTimeout}s";
+        TimeoutStopSec = "30s";
+      };
+    };
+
+    systemd.sockets.incus = lib.mkIf cfg.socketActivation {
+      description = "Incus UNIX socket";
+      wantedBy = [ "sockets.target" ];
+
+      socketConfig = {
+        ListenStream = "/var/lib/incus/unix.socket";
+        SocketMode = "0660";
+        SocketGroup = "incus-admin";
+        Service = "incus.service";
+      };
+    };
+
+    systemd.services.incus-preseed = lib.mkIf (cfg.preseed != null) {
+      description = "Incus initialization with preseed file";
+
+      wantedBy = ["incus.service"];
+      after = ["incus.service"];
+      bindsTo = ["incus.service"];
+      partOf = ["incus.service"];
+
+      script = ''
+        ${cfg.package}/bin/incus admin init --preseed <${
+          preseedFormat.generate "incus-preseed.yaml" cfg.preseed
+        }
+      '';
+
+      serviceConfig = {
+        Type = "oneshot";
+        RemainAfterExit = true;
+      };
+    };
+
+    users.groups.incus-admin = { };
+
+    users.users.root = {
+      # match documented default ranges https://linuxcontainers.org/incus/docs/main/userns-idmap/#allowed-ranges
+      subUidRanges = [
+        {
+          startUid = 1000000;
+          count = 1000000000;
+        }
+      ];
+      subGidRanges = [
+        {
+          startGid = 1000000;
+          count = 1000000000;
+        }
+      ];
+    };
+
+    virtualisation.lxc.lxcfs.enable = true;
+  };
+}
diff --git a/nixos/modules/virtualisation/lxc-container.nix b/nixos/modules/virtualisation/lxc-container.nix
index 9402d3bf37d..61d7c4cb73f 100644
--- a/nixos/modules/virtualisation/lxc-container.nix
+++ b/nixos/modules/virtualisation/lxc-container.nix
@@ -9,15 +9,16 @@ in {
 
   options = {
     virtualisation.lxc = {
-      privilegedContainer = lib.mkOption {
-        type = lib.types.bool;
-        default = false;
-        description = lib.mdDoc ''
-          Whether this LXC container will be running as a privileged container or not. If set to `true` then
-          additional configuration will be applied to the `systemd` instance running within the container as
-          recommended by [distrobuilder](https://linuxcontainers.org/distrobuilder/introduction/).
-        '';
-      };
+      nestedContainer = lib.mkEnableOption (lib.mdDoc ''
+        Whether this container is configured as a nested container. On LXD containers this is recommended
+        for all containers and is enabled with `security.nesting = true`.
+      '');
+
+      privilegedContainer = lib.mkEnableOption (lib.mdDoc ''
+        Whether this LXC container will be running as a privileged container or not. If set to `true` then
+        additional configuration will be applied to the `systemd` instance running within the container as
+        recommended by [distrobuilder](https://linuxcontainers.org/distrobuilder/introduction/).
+      '');
     };
   };
 
@@ -36,7 +37,6 @@ in {
         ${config.nix.package.out}/bin/nix-env -p /nix/var/nix/profiles/system --set /run/current-system
       '';
 
-    # TODO: build rootfs as squashfs for faster unpack
     system.build.tarball = pkgs.callPackage ../../lib/make-system-tarball.nix {
       extraArgs = "--owner=0";
 
@@ -63,11 +63,30 @@ in {
       extraCommands = "mkdir -p proc sys dev";
     };
 
+    system.build.squashfs = pkgs.callPackage ../../lib/make-squashfs.nix {
+      fileName = "nixos-lxc-image-${pkgs.stdenv.hostPlatform.system}";
+
+      noStrip = true; # keep directory structure
+      comp = "zstd -Xcompression-level 6";
+
+      storeContents = [config.system.build.toplevel];
+
+      pseudoFiles = [
+        "/sbin d 0755 0 0"
+        "/sbin/init s 0555 0 0 ${config.system.build.toplevel}/init"
+        "/dev d 0755 0 0"
+        "/proc d 0555 0 0"
+        "/sys d 0555 0 0"
+      ];
+    };
+
     system.build.installBootLoader = pkgs.writeScript "install-lxd-sbin-init.sh" ''
       #!${pkgs.runtimeShell}
-      ln -fs "$1/init" /sbin/init
+      ${pkgs.coreutils}/bin/ln -fs "$1/init" /sbin/init
     '';
 
+    systemd.additionalUpstreamSystemUnits = lib.mkIf cfg.nestedContainer ["systemd-udev-trigger.service"];
+
     # Add the overrides from lxd distrobuilder
     # https://github.com/lxc/distrobuilder/blob/05978d0d5a72718154f1525c7d043e090ba7c3e0/distrobuilder/main.go#L630
     systemd.packages = [
diff --git a/nixos/modules/virtualisation/lxd.nix b/nixos/modules/virtualisation/lxd.nix
index e30fbebb662..6f628c4a6e3 100644
--- a/nixos/modules/virtualisation/lxd.nix
+++ b/nixos/modules/virtualisation/lxd.nix
@@ -145,9 +145,7 @@ in {
       };
 
       ui = {
-        enable = lib.mkEnableOption (lib.mdDoc ''
-          Enables the (experimental) LXD UI.
-        '');
+        enable = lib.mkEnableOption (lib.mdDoc "(experimental) LXD UI");
 
         package = lib.mkPackageOption pkgs.lxd-unwrapped "ui" { };
       };
diff --git a/nixos/modules/virtualisation/nixos-containers.nix b/nixos/modules/virtualisation/nixos-containers.nix
index 5df9942dbc0..6fdb177b968 100644
--- a/nixos/modules/virtualisation/nixos-containers.nix
+++ b/nixos/modules/virtualisation/nixos-containers.nix
@@ -649,6 +649,15 @@ in
               '';
             };
 
+            restartIfChanged = mkOption {
+              type = types.bool;
+              default = true;
+              description = lib.mdDoc ''
+                Whether the container should be restarted during a NixOS
+                configuration switch if its definition has changed.
+              '';
+            };
+
             timeoutStartSec = mkOption {
               type = types.str;
               default = "1min";
@@ -745,7 +754,7 @@ in
                   { services.postgresql.enable = true;
                     services.postgresql.package = pkgs.postgresql_14;
 
-                    system.stateVersion = "21.05";
+                    system.stateVersion = "${lib.trivial.release}";
                   };
               };
           }
@@ -826,7 +835,7 @@ in
                 containerConfig.path
                 config.environment.etc."${configurationDirectoryName}/${name}.conf".source
               ];
-              restartIfChanged = true;
+              restartIfChanged = containerConfig.restartIfChanged;
             }
           )
       )) config.containers)
@@ -897,4 +906,6 @@ in
       "tun"
     ];
   });
+
+  meta.buildDocsInSandbox = false;
 }
diff --git a/nixos/modules/virtualisation/oci-common.nix b/nixos/modules/virtualisation/oci-common.nix
index ac9405e3ecf..a620df06315 100644
--- a/nixos/modules/virtualisation/oci-common.nix
+++ b/nixos/modules/virtualisation/oci-common.nix
@@ -56,5 +56,5 @@ in
 
   # Otherwise the instance may not have a working network-online.target,
   # making the fetch-ssh-keys.service fail
-  networking.useNetworkd = true;
+  networking.useNetworkd = lib.mkDefault true;
 }
diff --git a/nixos/modules/virtualisation/oci-containers.nix b/nixos/modules/virtualisation/oci-containers.nix
index a9f4ab77f86..a4a40346f09 100644
--- a/nixos/modules/virtualisation/oci-containers.nix
+++ b/nixos/modules/virtualisation/oci-containers.nix
@@ -66,6 +66,17 @@ let
           '';
         };
 
+        labels = mkOption {
+          type = with types; attrsOf str;
+          default = {};
+          description = lib.mdDoc "Labels to attach to the container at runtime.";
+          example = literalExpression ''
+            {
+              "traefik.https.routers.example.rule" = "Host(`example.container`)";
+            }
+          '';
+        };
+
         entrypoint = mkOption {
           type = with types; nullOr str;
           description = lib.mdDoc "Override the default entrypoint of the image.";
@@ -203,6 +214,13 @@ let
           '';
         };
 
+        hostname = mkOption {
+          type = with types; nullOr str;
+          default = null;
+          description = lib.mdDoc "The hostname of the container.";
+          example = "hello-world";
+        };
+
         extraOptions = mkOption {
           type = with types; listOf str;
           default = [];
@@ -228,6 +246,25 @@ let
   mkService = name: container: let
     dependsOn = map (x: "${cfg.backend}-${x}.service") container.dependsOn;
     escapedName = escapeShellArg name;
+    preStartScript = pkgs.writeShellApplication {
+      name = "pre-start";
+      runtimeInputs = [ ];
+      text = ''
+        ${cfg.backend} rm -f ${name} || true
+        ${optionalString (isValidLogin container.login) ''
+          ${cfg.backend} login \
+          ${container.login.registry} \
+          --username ${container.login.username} \
+          --password-stdin < ${container.login.passwordFile}
+        ''}
+        ${optionalString (container.imageFile != null) ''
+          ${cfg.backend} load -i ${container.imageFile}
+        ''}
+        ${optionalString (cfg.backend == "podman") ''
+          rm -f /run/podman-${escapedName}.ctr-id
+        ''}
+      '';
+    };
   in {
     wantedBy = [] ++ optional (container.autoStart) "multi-user.target";
     after = lib.optionals (cfg.backend == "docker") [ "docker.service" "docker.socket" ]
@@ -242,23 +279,6 @@ let
       else if cfg.backend == "podman" then [ config.virtualisation.podman.package ]
       else throw "Unhandled backend: ${cfg.backend}";
 
-    preStart = ''
-      ${cfg.backend} rm -f ${name} || true
-      ${optionalString (isValidLogin container.login) ''
-        cat ${container.login.passwordFile} | \
-          ${cfg.backend} login \
-            ${container.login.registry} \
-            --username ${container.login.username} \
-            --password-stdin
-        ''}
-      ${optionalString (container.imageFile != null) ''
-        ${cfg.backend} load -i ${container.imageFile}
-        ''}
-      ${optionalString (cfg.backend == "podman") ''
-        rm -f /run/podman-${escapedName}.ctr-id
-        ''}
-      '';
-
     script = concatStringsSep " \\\n  " ([
       "exec ${cfg.backend} run"
       "--rm"
@@ -266,6 +286,8 @@ let
       "--log-driver=${container.log-driver}"
     ] ++ optional (container.entrypoint != null)
       "--entrypoint=${escapeShellArg container.entrypoint}"
+      ++ optional (container.hostname != null)
+      "--hostname=${escapeShellArg container.hostname}"
       ++ lib.optionals (cfg.backend == "podman") [
         "--cidfile=/run/podman-${escapedName}.ctr-id"
         "--cgroups=no-conmon"
@@ -277,6 +299,7 @@ let
       ++ map (p: "-p ${escapeShellArg p}") container.ports
       ++ optional (container.user != null) "-u ${escapeShellArg container.user}"
       ++ map (v: "-v ${escapeShellArg v}") container.volumes
+      ++ (mapAttrsToList (k: v: "-l ${escapeShellArg k}=${escapeShellArg v}") container.labels)
       ++ optional (container.workdir != null) "-w ${escapeShellArg container.workdir}"
       ++ map escapeShellArg container.extraOptions
       ++ [container.image]
@@ -306,7 +329,7 @@ let
       ###
       # ExecReload = ...;
       ###
-
+      ExecStartPre = [ "${preStartScript}/bin/pre-start" ];
       TimeoutStartSec = 0;
       TimeoutStopSec = 120;
       Restart = "always";
diff --git a/nixos/modules/virtualisation/qemu-vm.nix b/nixos/modules/virtualisation/qemu-vm.nix
index 74c3e1ecd03..6f275baf60d 100644
--- a/nixos/modules/virtualisation/qemu-vm.nix
+++ b/nixos/modules/virtualisation/qemu-vm.nix
@@ -198,6 +198,39 @@ let
         fi
       ''}
 
+      ${lib.optionalString cfg.tpm.enable ''
+        NIX_SWTPM_DIR=$(readlink -f "''${NIX_SWTPM_DIR:-${config.system.name}-swtpm}")
+        mkdir -p "$NIX_SWTPM_DIR"
+        ${lib.getExe cfg.tpm.package} \
+          socket \
+          --tpmstate dir="$NIX_SWTPM_DIR" \
+          --ctrl type=unixio,path="$NIX_SWTPM_DIR"/socket,terminate \
+          --pid file="$NIX_SWTPM_DIR"/pid --daemon \
+          --tpm2 \
+          --log file="$NIX_SWTPM_DIR"/stdout,level=6
+
+        # Enable `fdflags` builtin in Bash
+        # We will need it to perform surgical modification of the file descriptor
+        # passed in the coprocess to remove `FD_CLOEXEC`, i.e. close the file descriptor
+        # on exec.
+        # If let alone, it will trigger the coprocess to read EOF when QEMU is `exec`
+        # at the end of this script. To work around that, we will just clear
+        # the `FD_CLOEXEC` bits as a first step.
+        enable -f ${hostPkgs.bash}/lib/bash/fdflags fdflags
+        # leave a dangling subprocess because the swtpm ctrl socket has
+        # "terminate" when the last connection disconnects, it stops swtpm.
+        # When qemu stops, or if the main shell process ends, the coproc will
+        # get signaled by virtue of the pipe between main and coproc ending.
+        # Which in turns triggers a socat connect-disconnect to swtpm which
+        # will stop it.
+        coproc waitingswtpm {
+          read || :
+          echo "" | ${lib.getExe hostPkgs.socat} STDIO UNIX-CONNECT:"$NIX_SWTPM_DIR"/socket
+        }
+        # Clear `FD_CLOEXEC` on the coprocess' file descriptor stdin.
+        fdflags -s-cloexec ''${waitingswtpm[1]}
+      ''}
+
       cd "$TMPDIR"
 
       ${lib.optionalString (cfg.emptyDiskImages != []) "idx=0"}
@@ -267,6 +300,7 @@ let
   };
 
   storeImage = import ../../lib/make-disk-image.nix {
+    name = "nix-store-image";
     inherit pkgs config lib;
     additionalPaths = [ regInfo ];
     format = "qcow2";
@@ -656,8 +690,8 @@ in
       package =
         mkOption {
           type = types.package;
-          default = hostPkgs.qemu_kvm;
-          defaultText = literalExpression "config.virtualisation.host.pkgs.qemu_kvm";
+          default = if hostPkgs.stdenv.hostPlatform.qemuArch == pkgs.stdenv.hostPlatform.qemuArch then hostPkgs.qemu_kvm else hostPkgs.qemu;
+          defaultText = literalExpression "if hostPkgs.stdenv.hostPlatform.qemuArch == pkgs.stdenv.hostPlatform.qemuArch then config.virtualisation.host.pkgs.qemu_kvm else config.virtualisation.host.pkgs.qemu";
           example = literalExpression "pkgs.qemu_test";
           description = lib.mdDoc "QEMU package to use.";
         };
@@ -862,6 +896,32 @@ in
       };
     };
 
+    virtualisation.tpm = {
+      enable = mkEnableOption "a TPM device in the virtual machine with a driver, using swtpm.";
+
+      package = mkPackageOptionMD cfg.host.pkgs "swtpm" { };
+
+      deviceModel = mkOption {
+        type = types.str;
+        default = ({
+          "i686-linux" = "tpm-tis";
+          "x86_64-linux" = "tpm-tis";
+          "ppc64-linux" = "tpm-spapr";
+          "armv7-linux" = "tpm-tis-device";
+          "aarch64-linux" = "tpm-tis-device";
+        }.${pkgs.hostPlatform.system} or (throw "Unsupported system for TPM2 emulation in QEMU"));
+        defaultText = ''
+          Based on the guest platform Linux system:
+
+          - `tpm-tis` for (i686, x86_64)
+          - `tpm-spapr` for ppc64
+          - `tpm-tis-device` for (armv7, aarch64)
+        '';
+        example = "tpm-tis-device";
+        description = lib.mdDoc "QEMU device model for the TPM, uses the appropriate default based on th guest platform system and the package passed.";
+      };
+    };
+
     virtualisation.useDefaultFilesystems =
       mkOption {
         type = types.bool;
@@ -937,7 +997,7 @@ in
               virtualisation.memorySize is above 2047, but qemu is only able to allocate 2047MB RAM on 32bit max.
             '';
           }
-          { assertion = cfg.directBoot.initrd != options.virtualisation.directBoot.initrd.default -> cfg.directBoot.enable;
+          { assertion = cfg.directBoot.enable || cfg.directBoot.initrd == options.virtualisation.directBoot.initrd.default;
             message =
               ''
                 You changed the default of `virtualisation.directBoot.initrd` but you are not
@@ -1027,7 +1087,8 @@ in
 
     boot.initrd.availableKernelModules =
       optional cfg.writableStore "overlay"
-      ++ optional (cfg.qemu.diskInterface == "scsi") "sym53c8xx";
+      ++ optional (cfg.qemu.diskInterface == "scsi") "sym53c8xx"
+      ++ optional (cfg.tpm.enable) "tpm_tis";
 
     virtualisation.additionalPaths = [ config.system.build.toplevel ];
 
@@ -1098,6 +1159,11 @@ in
       (mkIf (!cfg.graphics) [
         "-nographic"
       ])
+      (mkIf (cfg.tpm.enable) [
+        "-chardev socket,id=chrtpm,path=\"$NIX_SWTPM_DIR\"/socket"
+        "-tpmdev emulator,id=tpm_dev_0,chardev=chrtpm"
+        "-device ${cfg.tpm.deviceModel},tpmdev=tpm_dev_0"
+      ])
     ];
 
     virtualisation.qemu.drives = mkMerge [
@@ -1121,11 +1187,12 @@ in
       }) cfg.emptyDiskImages)
     ];
 
-    # Use mkVMOverride to enable building test VMs (e.g. via `nixos-rebuild
-    # build-vm`) of a system configuration, where the regular value for the
-    # `fileSystems' attribute should be disregarded (since those filesystems
-    # don't necessarily exist in the VM).
-    fileSystems = mkVMOverride cfg.fileSystems;
+    # By default, use mkVMOverride to enable building test VMs (e.g. via
+    # `nixos-rebuild build-vm`) of a system configuration, where the regular
+    # value for the `fileSystems' attribute should be disregarded (since those
+    # filesystems don't necessarily exist in the VM). You can disable this
+    # override by setting `virtualisation.fileSystems = lib.mkForce { };`.
+    fileSystems = lib.mkIf (cfg.fileSystems != { }) (mkVMOverride cfg.fileSystems);
 
     virtualisation.fileSystems = let
       mkSharedDir = tag: share:
diff --git a/nixos/modules/virtualisation/vagrant-guest.nix b/nixos/modules/virtualisation/vagrant-guest.nix
index 263b1ebca08..2fad376086e 100644
--- a/nixos/modules/virtualisation/vagrant-guest.nix
+++ b/nixos/modules/virtualisation/vagrant-guest.nix
@@ -55,4 +55,5 @@ in
   };
 
   security.sudo.wheelNeedsPassword = false;
+  security.sudo-rs.wheelNeedsPassword = false;
 }
diff --git a/nixos/modules/virtualisation/virtualbox-host.nix b/nixos/modules/virtualisation/virtualbox-host.nix
index b1565a09682..9741ea090f7 100644
--- a/nixos/modules/virtualisation/virtualbox-host.nix
+++ b/nixos/modules/virtualisation/virtualbox-host.nix
@@ -91,7 +91,7 @@ in
   };
 
   config = mkIf cfg.enable (mkMerge [{
-    warnings = mkIf (config.nixpkgs.config.virtualbox.enableExtensionPack or false)
+    warnings = mkIf (pkgs.config.virtualbox.enableExtensionPack or false)
       ["'nixpkgs.virtualbox.enableExtensionPack' has no effect, please use 'virtualisation.virtualbox.host.enableExtensionPack'"];
     boot.kernelModules = [ "vboxdrv" "vboxnetadp" "vboxnetflt" ];
     boot.extraModulePackages = [ kernelModules ];