Disable Bash completion in 'make distcheck'

We cannot rely on the auto-detection of Bash completion dir during 'make
distcheck' because autotools does not use DESTDIR, only --prefix for the
install check, and pkg-config returns a system path.

Also, show detected path in configure summary for debug.

Signed-off-by: Joachim Wiberg <troglobit@gmail.com>

.github/workflows/release.yml

@@ -23,11 +23,16 @@ jobs:
         run: |
           ver=${GITHUB_REF#refs/tags/}
           echo "ver=${ver}" >> $GITHUB_OUTPUT
-          if echo $ver | grep -qE '[0-9.]+(-alpha|-beta|-rc)[0-9]*'; then
+          if echo $ver | grep -qE '^[0-9]+\.[0-9]+(\.[0-9]+)?(-alpha|-beta|-rc)[0-9]*$'; then
             echo "pre=true"  >> $GITHUB_OUTPUT
           else
             echo "pre=false" >> $GITHUB_OUTPUT
           fi
+          if echo $ver | grep -qE '^[0-9.]+\.[0-9.]+(\.[0-9]+)?$'; then
+            echo "latest=true"  >> $GITHUB_OUTPUT
+          else
+            echo "latest=false" >> $GITHUB_OUTPUT
+          fi
       - name: Creating Makefiles ...
         run: |
           ./autogen.sh
@@ -48,6 +53,7 @@ jobs:
       - uses: ncipollo/release-action@v1
         with:
           name: Finit v${{ github.ref_name }}
-          prerelease: ${{ needs.build.outputs.pre }}
+          prerelease: ${{ steps.build.outputs.pre }}
+          makeLatest: ${{ steps.build.outputs.latest }}
           bodyFile: "release.md"
           artifacts: "artifacts/*"

ChangeLog.md

@@ -4,7 +4,7 @@ Change Log
 All relevant changes are documented in this file.
 
 
-[4.5][] - 2023-10-03
+[4.5][] - 2023-10-30
 --------------------
 
 ### Changes
@@ -22,7 +22,7 @@ All relevant changes are documented in this file.
 - Improved logging on failure to `execvp()` a run/task/service, now
   with `errno`, e.g., "No such file or directory" when the command
   is missing from `$PATH`
-- Add support for Bash completion to `initctl`, issue #360
+- Add Bash completion support for `initctl`, configurable, issue #360
 - Handle absolute path to `initctl [enable|disable]`, not supported
 - Update `finit.conf(5)` man page with the recommended directory
   hierarchy in `/etc/finit.d/`
@@ -37,6 +37,24 @@ All relevant changes are documented in this file.
   a task only `if:<run/foo/failure>` (here the run task 'foo' failed)
 - Document new `if:`, `conflict:`, `nowarn` options for run/task/service
   first introduced in v4.4
+- Failure to open fstab should log to console, not just log to `/dev/kmsg`
+- Rename `/lib/finit/system/*.conf`, added numbered prefix to ensure
+  proper execution order, e.g., `udevd` should always run first
+- Plugins and bundled services: dbus, keventd, watchdogd, and runparts,
+  are now loaded *after* all services in `/lib/finit/system/`.  A new
+  runtime-only path (for inspection) in `/run/finit/system/` is used
+- Redirect `log*` output to console when `finit.debug` is enabled
+- Assert `<int/container>` condition if we detect running in container
+- Add support for mdev's netlink daemon mode, issue #367
+- Add support for mdevd in `10-hotplug.conf`, preferred over plain mdev
+- Disable modprobe plugin by default, udevd and mdev/mdevd loads modules
+- Update documentation for run/task shell limitations, issue #376
+- Update documentation regarding automount of `/run` and `/tmp`
+- Update plugin documentation, add section about limited tmpfiles.d(5) support
+- Skip registering service when `if:!name` matches a known service.
+  This allows conditional loading of alternative services, e.g. if udevd
+  is already loaded we do not need mdevd
+- Drop `doc/bootstrap.md`, inaccurate and confusing to users
 
 ### Fixes
 - Fix #227: believed to have been fixed in v4.3, the root cause was
@@ -52,9 +70,41 @@ All relevant changes are documented in this file.
 - Fix #366: document `fsck.*` kernel command line options and simplify
   the configure flags `--enable-fsckfix` and `--enable-fastboot` to
   only adjust the default values for the `fsck.*` options.
+- Fix #371: swap load order of `/lib/finit/system/*` vs `/etc/finit.d/*`
+  We must run `10-hotplug.conf` first to ensure devices and modules are
+  up and loaded before the user's run/task/services are called.  The order
+  at bootstrap is now saved in `/run/finit/conf.order` for inspection,
+  `/run/finit/exec.order` shows the start order of each process
+- Fix #372: lost `udevadm` calls due to overloading
+- Adjust final `udevadm settle` timeout: 5 -> 30 sec
+- Fixed dbus plugin, the function that located `<pidfile> ...` in the
+  `dbus/system.conf` caused spurious line breaks which led to the
+  service not being loaded properly
 - The `runparts` executor now skips backup files (`foo~`)
 - The `runparts` stanza now properly appends ` start` to scripts that
   start with `S[0-9]+`.  This has been broken for a very long time.
+- Fix #377: expand service `env:file` variables, allow constructs like:
+
+        RUNDIR=/var/run/somesvc
+        DAEMON_ARGS=--workdir $RUNDIR --other-args...
+- Fix #378: warn on console if run/task times out during bootstrap
+- Fix #378: add run/task support for `<!>` to allow transition from
+  bootstrap to multi-user runlevel even though task has not run yet.
+- Fix #382: do not clear `<service/foo/STATE>` conditions on reload.  
+  Introduced back in v4.3-rc2, 82cc10be8, the support for automatic
+  service conditions have had a weird and unintended behavior.  Any
+  change in state (see `doc/svc-machine.png`) caused Finit to clear
+  out *all* previously acquired service conditions.
+
+  However, when moving between RUNNING and PAUSED states, a service
+  should not have its conditions cleared.  The PAUSED state, seen
+  also by all conditions moving to FLUX, is only temporary while an
+  `initctl reload` is processed.  If a service has no changes to be
+  applied it will move back to RUNNING.
+
+  Also, we cannot clear the service conditions because other run/task
+  or services may depend on it and clearing them would cause Finit to
+  `SIGTERM` these processes (since they are no longer eligible to run).
 
 
 [4.4][] - 2023-05-15

Makefile.am

@@ -10,6 +10,10 @@ if DOC
 SUBDIRS           += doc
 endif
 
+if BASH
+bashdir            = $(BASH_DIR)
+dist_bash_DATA     = initctl
+endif
 # Compat hook
 install-dev:
 	@make -C src install-pkgincludeHEADERS
@@ -30,6 +34,7 @@ release: distcheck
 		printf "$$file.sha256\t"; cat ../$$file.sha256 | cut -f1 -d' ';	\
 	done
 
-DISTCHECK_CONFIGURE_FLAGS = --prefix=/usr --sysconfdir=/etc --localstatedir=/var \
+DISTCHECK_CONFIGURE_FLAGS = --prefix=/usr --sysconfdir=/etc --localstatedir=/var   \
-			    --enable-testserv-plugin --enable-x11-common-plugin  \
+			    --enable-testserv-plugin --enable-x11-common-plugin    \
-			    --with-watchdog --with-keventd --with-fstab=/etc/fstab
+			    --with-watchdog --with-keventd --with-fstab=/etc/fstab \
+			    --with-bash-completion-dir=no

README.md

@@ -3,7 +3,6 @@
 
 * [Introduction](#introduction)
 * [Features](#features)
-* [Bootstrap](doc/bootstrap.md#bootstrap)
 * [Runlevels](#runlevels)
 * [Syntax](doc/config.md#syntax)
   * [Runparts & /etc/rc.local](#runparts--etcrclocal)
@@ -391,8 +390,7 @@ Right after the runlevel change when all services have started properly,
 No configuration stanza in `/etc/finit.conf` is required for `rc.local`.
 If it exists and is an executable shell script Finit calls it at the very
 end of the boot, before calling the `HOOK_SYSTEM_UP`.  See more on hooks
-in [doc/plugins.md](doc/plugins.md#hooks), and about the system bootstrap
+in [doc/plugins.md](doc/plugins.md#hooks).
-in [doc/bootstrap.md](doc/bootstrap.md).
 
 
 ### Limitations

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([Finit], [4.5-rc4], [https://github.com/troglobit/finit/issues],
+AC_INIT([Finit], [4.5], [https://github.com/troglobit/finit/issues],
 	[finit], [https://troglobit.com/projects/finit/])
 AC_CONFIG_AUX_DIR(aux)
 AM_INIT_AUTOMAKE([1.11 foreign dist-xz subdir-objects])
@@ -15,10 +15,10 @@ AC_CONFIG_FILES([Makefile
 	man/Makefile
 	plugins/Makefile
 	src/Makefile
-	system/Makefile
+	system/Makefile system/10-hotplug.conf
-	tmpfiles.d/Makefile
 	test/test.env test/Makefile
-	test/lib/Makefile test/src/Makefile])
+	test/lib/Makefile test/src/Makefile
+	tmpfiles.d/Makefile])
 
 # Older versions of autoconf (<2.58) do not have AC_CONFIG_MACRO_DIR()
 #m4_include([m4/plugin.m4])
@@ -94,18 +94,22 @@ AC_ARG_ENABLE(all_plugins,
 AC_PLUGIN([alsa-utils],    [no],  [Save and restore ALSA sound settings using alsactl])
 AC_PLUGIN([dbus],          [no],  [Setup and start system message bus, D-Bus])
 AC_PLUGIN([modules-load],  [no],  [Scans /etc/modules-load.d for modules to load])
-AC_PLUGIN([modprobe],      [yes], [Coldplug modules using modalias magic])
+AC_PLUGIN([modprobe],      [no],  [Coldplug modules using modalias magic])
 AC_PLUGIN([resolvconf],    [no],  [Setup necessary files for resolvconf])
 AC_PLUGIN([x11-common],    [no],  [Console setup (for X)])
 AC_PLUGIN([netlink],       [yes], [Basic netlink plugin for IFUP/IFDN and GW events. Can be replaced with externally built plugin that links with libnl or similar.])
 AC_PLUGIN([hook-scripts],  [no],  [Trigger script execution from hook points])
-AC_PLUGIN([hotplug],       [yes], [Setup and start udev or mdev hotplug daemon])
+AC_PLUGIN([hotplug],       [yes], [Start udevd or mdev kernel event datamon])
 AC_PLUGIN([rtc],           [yes], [Save and restore RTC using hwclock])
 AC_PLUGIN([tty],           [yes], [Automatically activate new TTYs, e.g. USB-to-serial])
 AC_PLUGIN([urandom],       [yes], [Setup and save random seed at boot/shutdown])
 AC_PLUGIN([testserv],      [no],  [Test plugin to start test serv daemon])
 
 # Check for extra arguments or packages
+AC_ARG_WITH([bash-completion-dir],
+	AS_HELP_STRING([--with-bash-completion-dir[=PATH]], [Directory to install Bash completion, default: auto]),
+	[bash_dir=$withval], [bash_dir=yes])
+
 AC_ARG_WITH(fstab,
         AS_HELP_STRING([--with-fstab=FILE], [System default fstab, default $sysconfdir/fstab]),
 	[fstab=$withval], [fstab=yes])
@@ -196,6 +200,12 @@ AS_IF([test "x$enable_rescue" != "xno"], [
 AM_CONDITIONAL(LOGROTATE, [test "x$enable_logrotate" = "xyes"])
 
 ### With features ##############################################################################
+AS_IF([test "x$bash_dir" = "xyes"], [
+	PKG_CHECK_MODULES([BASH_COMPLETION], [bash-completion >= 2.0],
+		[BASH_DIR="$(pkg-config --variable=completionsdir bash-completion)"],
+		[BASH_DIR="$datadir/bash-completion/completions"])], [
+	BASH_DIR="$bash_dir"])
+
 AS_IF([test "x$with_fstab" != "xno"], [
 	AS_IF([test "x$fstab" = "xyes"], [fstab=$sysconfdir/fstab])
 	AC_EXPAND_DIR(fstab_path, "$fstab")
@@ -285,19 +295,26 @@ AS_IF([test "x$with_plugin_path" != "xno"], [
 	AC_DEFINE_UNQUOTED(EXTERNAL_PLUGIN_PATH, "$plugin_path", [Built-in default: /usr/lib/finit/plugins:/usr/local/lib/finit/plugins])])
 
 # Control build with automake flags
-AM_CONDITIONAL(STATIC,    [test "x$enable_static" = "xyes"])
+AM_CONDITIONAL(BASH,      [test "x$bash_dir"      != "xno"])
+AM_CONDITIONAL(STATIC,    [test "x$enable_static"  = "xyes"])
 AM_CONDITIONAL(KEVENTD,   [test "x$with_keventd"  != "xno"])
 AM_CONDITIONAL(SULOGIN,   [test "x$with_sulogin"  != "xno"])
 AM_CONDITIONAL(WATCHDOGD, [test "x$with_watchdog" != "xno"])
-AM_CONDITIONAL(DOC,       [test "x$enable_doc" = "xyes"])
+AM_CONDITIONAL(DOC,       [test "x$enable_doc"     = "xyes"])
 AM_CONDITIONAL(CONTRIB,   [test "x$enable_contrib" = "xyes"])
 AM_CONDITIONAL(TESTSERV,  [test "x$enable_testserv_plugin" = "xyes"])
 
+# Workaround for as-of-yet unreleased runstatedir support, planned for
+# autoconf 2.70, which some major distros have backported.
+AS_IF([test -z "$runstatedir"], runstatedir="$localstatedir/run")
+AC_SUBST(runstatedir)
+
 # Override default libdir, used for plugins and rescue.conf
 pkglibdir=$libdir/finit
 AC_SUBST(pkglibexecdir)
 AC_EXPAND_DIR(pkglibexecdir, "$libexecdir/finit")
 AC_DEFINE_UNQUOTED(FINIT_EXECPATH_, "$pkglibexecdir", [Finit libexec path])
+AC_DEFINE_UNQUOTED(FINIT_RUNPATH_, "$runstatedir/finit/system", [Finit runtime dynamic system services])
 
 AC_EXPAND_DIR(plugin_path, "$pkglibdir/plugins")
 AC_SUBST(plugin_path)
@@ -305,7 +322,7 @@ AC_DEFINE_UNQUOTED(PLUGIN_PATH, "$plugin_path", [Finit plugin path])
 
 AC_EXPAND_DIR(tmpfiles_path, "$libdir/tmpfiles.d")
 AC_SUBST(tmpfiles_path)
-AC_DEFINE_UNQUOTED(TMPFILES_PATH, "$tmpfiles_path", [Main tmpfiles.d/, override in /etc])
+AC_DEFINE_UNQUOTED(TMPFILES_PATH_, "$tmpfiles_path", [Main tmpfiles.d/, override in /etc])
 
 AC_EXPAND_DIR(finit_tmpfiles, "$pkglibdir/tmpfiles.d")
 AC_SUBST(finit_tmpfiles)
@@ -313,13 +330,16 @@ AC_DEFINE_UNQUOTED(FINIT_TMPFILES, "$finit_tmpfiles", [Finit tmpfiles.d/ .conf p
 
 AC_EXPAND_DIR(system_path, "$pkglibdir/system")
 AC_SUBST(system_path)
-AC_DEFINE_UNQUOTED(FINIT_SYSPATH, "$system_path", [Static services, previously generated at runtime by plugins])
+AC_DEFINE_UNQUOTED(FINIT_SYSPATH_, "$system_path", [Static services, previously generated at runtime by plugins])
 
 AC_EXPAND_DIR(rescue_conf, "$pkglibdir/rescue.conf")
 AC_DEFINE_UNQUOTED(RESCUE_CONF, "$rescue_conf", [Finit rescue mode .conf])
 AC_EXPAND_DIR(sample_conf, "$pkglibdir/sample.conf")
 AC_DEFINE_UNQUOTED(SAMPLE_CONF, "$sample_conf", [Finit sameple.conf for initctl create])
 
+AC_EXPAND_DIR(bash_completion_dir, "$BASH_DIR")
+AC_SUBST(BASH_DIR, "$bash_completion_dir")
+
 AS_IF([test "$heading" = ""],   [heading="<DEFAULT>"])
 AS_IF([test "$heading" = "no"], [heading="<NONE>"])
 
@@ -356,6 +376,7 @@ Behavior:
 Optional features:
   Install doc/..........: $enable_doc
   Install contrib/......: $enable_contrib
+  Bash completion.......: $bash_completion_dir
   Built-in keventd......: $with_keventd
   Built-in sulogin......: $with_sulogin $sulogin
   Built-in watchdogd....: $with_watchdog $watchdog

doc/bootstrap.md

@@ -1,59 +0,0 @@
-Bootstrap
-=========
-
-0. Get bearings, disable kernel logs, initialize TTY and print banner,
-   unless `HOOK_BANNER` is defined, then call that instead
-1. Prepare emergency shell, if enabled
-2. Set up initial signal handlers
-3. Set up default $PATH early
-4. Mount `/proc` and `/sys`
-5. Check kernel command line for `debug` to figure out log level
-6. Load all `.so` plugins
-7. Call `fsck` on file systems listed in `/etc/fstab`
-8. Populate `/dev` using either udev or mdev, depending on system type
-9. Parse `/etc/finit.conf`
-10. Start built-in watchdog, if enabled and `WDT_DEVNODE` exists.  This
-    means any WDT that requires a kernel module need to be either
-    compiled into the kernel, or insmod'ed in `/etc/finit.conf`
-11. Load all `/etc/finit.d/*.conf` files and set hostname
-12. Remount `/` read-write if `/` is listed in `/etc/fstab` without `ro`
-13. Call 1st level hooks, `HOOK_ROOTFS_UP`
-14. Mount all file systems listed in `/etc/fstab` and swap, if available.
-    On mount error `HOOK_MOUNT_ERROR` is called.  After mount, regardless
-    of error, `HOOK_MOUNT_POST` is called
-15. Enable SysV init signals
-16. Call 2nd level hooks, `HOOK_BASEFS_UP`
-17. Cleanup stale files from `/tmp/*` et al, handled by `bootmisc` plugin
-18. Load kernel params from `/etc/sysctl.d/*.conf`, `/etc/sysctl.conf`
-    et al. (Supports all locations that SysV init does.), handled by
-    `procps` plugin
-19. Start all 'S' runlevel tasks and services
-20. Bring up loopback interface and all `/etc/network/interfaces`, if
-    the `.conf` setting `network <SCRIPT>` is set, it is called instead
-21. Call 3rd level hooks, `HOOK_NETWORK_UP`
-22. If `runparts <DIR>` is set, [run-parts(8)][] is called on `<DIR>`
-23. Switch to the configured runlevel from `/etc/finit.conf`, default 2.
-    At every runlevel change all `*.conf` files in `/etc/finit.d/` are
-    (re)loaded and new services, tasks, and blocking run commands are
-    started.  Provided they are allowed in the new runlevel and all of
-    their conditions, if any, are set.
-24. Call 4th level hooks, `HOOK_SVC_UP`
-25. Call `/etc/rc.local`, if it exists and is an executable shell script
-26. Call 5th level (last) hooks, `HOOK_SYSTEM_UP`
-27. Start all configured TTYs
-
-In (19) and (23) tasks and services defined in `/etc/finit.conf` and
-`/etc/finit.d/*.conf` are started.
-
-Notice the seven hook points that are called at various point in the
-bootstrap process.  This is where plugins can extend the boot in any
-way they please.  There are other hook points available, for more on
-this, see [plugins.md](plugins.md).
-
-For instance, at `HOOK_BASEFS_UP` a plugin could read an XML file from a
-USB stick, convert/copy its contents to the system's `/etc/` directory,
-well before all 'S' runlevel tasks are started.  This could be used with
-system images that are created read-only and all configuration is stored
-on external media.
-
-[run-parts(8)]: http://manpages.debian.org/cgi-bin/man.cgi?query=run-parts

doc/conditions.md

@@ -27,6 +27,16 @@ specified separated by comma.  Multiple conditions are logically AND'ed
 during evaluation, i.e. all conditions must be satisfied in order for a
 service to run.
 
+A special syntax, using a leading `!` in run/task/service conditions,
+denote if a:
+
+ - service does not support `SIGHUP`
+ - run/task should not block runlevel changes (i.e., bootstrap)
+
+Finit guarantees by default that all run/tasks run (at least) once
+per runlevel.  For most tasks this is a good default, for example
+checking SSH host keys or loading keymap.  However, for conditions
+that are unlikely to happen it is not. (See example below.)
 
 ### Example
 
@@ -41,6 +51,19 @@ being created, i.e., the service's default readiness notification.
   asserted.  I.e., if the Zebra process above stops or restarts, netd
   will also stop or restart.
 
+Another example is `dropbear`, it does not support `SIGHUP`, but we can
+also see optional sourcing of arguments from an environment file:
+
+    service [2345789] <!> env:-/etc/default/dropbear dropbear -F -R $DROPBEAR_ARGS -- Dropbear SSH daemon
+
+Finally, the weird "block runlevel changes" example.  Here we see what
+happens when Finit receives `SIGPWR`, sent from a power daemon like
+[powstatd(8)][].  A condition is asserted and a user can set up their
+own task to act on it.  We do not want this task to block Finit from
+moving to the next runlevel after bootstrap, so we set `<!>`:
+
+    task [S0123456789] <!sys/pwr/fail> name:pwrfail initctl poweroff -- Power failure, shutting down
+
 
 Triggering
 ----------
@@ -260,3 +283,5 @@ Therefore, any plugin that supplies Finit with conditions must ensure
 that their state is updated after each reconfiguration.  This can be
 done by binding to the `HOOK_SVC_RECONF` hook.  For an example of how
 to do this, see `plugins/pidfile.c`.
+
+[powstatd(8)]: https://manpages.ubuntu.com/manpages/trusty/en/man8/powstatd.8.html

doc/config.md

@@ -52,21 +52,51 @@ PID file monitoring, or [conditions](conditions.md).
 
 ### Configuration Files
 
-Originally Finit was configured using a single file, `/etc/finit.conf`.
+Originally Finit was configured using a single file, `/etc/finit.conf`,
-Although still possible to use this single configuration file, today the
+and although still possible to use a single configuration file, today
-following familiar layout is recommended:
+the following layout is recommended:
 
-    /etc/
+    /
-      |- finit.d/
+    |- etc/
-      |   |- available/
+    |  |- finit.d/
-      |   |   `- my-service.conf
+    |  |   |- available/
-      :   |- enabled/
+    |  |   |  `- my-service.conf
-      :   |   `- my-service.conf -> ../available/my-service.conf
+    |  :   |- enabled/
-      :   :
+    |  :   |  `- my-service.conf -> ../available/my-service.conf
-      :   |- static-service.conf
+    |  :   :
-      :   `- another-static.conf
+    |  :   |- static-service.conf
-      :
+    |  :   `- another-static.conf
-      `- finit.conf
+    |  :
+    |  `- finit.conf
+    |- lib
+    |   `- finit/
+    |       `- system/
+    |           |- 10-hotplug.conf
+    |           `- ...
+    `- run/
+        `- finit/
+            `- system/
+                |- dbus.conf
+                |- keventd.conf
+                |- runparts.conf
+                |- watchdogd.conf
+                `- ...
+
+Configuration files in `/etc` are provided by the user, or projects like
+[finit-skel](https://github.com/troglobit/finit-skel) and extended by
+the user.
+
+The files in `/lib/finit/system/*.conf` are system critical services and
+setup provided by Finit, e.g. udev/mdev that must run very early at system
+bootstrap.  This system directory was introduced in Finit v4.4 to replace
+the hard-coded services provided by plugins before.  All .conf files in this
+directory be either replaced by a system administrator or overridden by a
+file with the same name in `/etc/finit.d/`.
+
+The files in `/run/finit/system/*.conf` are created by plugins and Finit
+bundled services like runparts, the [watchdog](#watchdog), and `keventd`
+if they are enabled.  Like `/lib/finit/system/*.conf`, these files can
+ be overridden by file with the same name in `/etc/finit.d/`.
 
 Services in the `available/` and `enabled/` sub-directories are called
 dynamic services, in contrast to static services -- the only difference
@@ -75,9 +105,27 @@ with the `enable` and `disable` commands.  An administrator can always
 create files and symlinks manually.
 
 At bootstrap, and `initctl reload`, all .conf files are read, starting
-with `finit.conf`, then, in alphabetical order, all `finit.d/*.conf`
+with `finit.conf`, then `/lib/finit/system/*.conf`, `finit.d/*.conf`,
-files, and finally (still in alphabetical order) all `enabled/*.conf`
+and finally all `finit.d/enabled/*.conf` files.  Each directory is a
-files.
+unique group, where files within each group are sorted alphabetically.
+
+**Example:**
+
+    /lib/finit/system/10-hotplug.conf
+    /lib/finit/system/90-testserv.conf
+    /run/finit/system/dbus.conf
+    /run/finit/system/runparts.conf
+    /etc/finit.d/10-abc.conf
+    /etc/finit.d/20-abc.conf
+    /etc/finit.d/enabled/1-aaa.conf
+    /etc/finit.d/enabled/1-abc.conf
+
+The resulting combined configuration is read line by line, each `run`,
+`task`, and `service` added to an ordered list that ensures they are
+started in the same order.  This is important because of the blocking
+properties of the `run` statement.  For an example on the relation of
+`service` and `run` statements, and dependency handling between them,
+see [Conditional Loading](#conditional-loading), below.
 
 > **Note:** the `finit.conf` and `finit.d/` names are only defaults.
 > They can be changed at compile-time with two `configure` options:
@@ -126,18 +174,22 @@ layout, as specified in the [FHS][]:
          |- spool/
          `- tmp/
 
-Finit starts (very early) by mounting `/dev`, `/proc/`, and `/sys`,
+Finit starts by mounting the critical file systems `/dev`, `/proc/`, and
-unless they (i.e., `/dev`) are already mounted.  When all plugins and
+`/sys`, unless they are already mounted.  When all plugins and other,
-other, core Finit functions, have been set up, all relevant filesystems
+core Finit functions, have been set up, all relevant filesystems (where
-(`PASS > 0`) are checked and mounted from the selected `fstab`, either
+`PASS > 0`) are checked and mounted from the selected `fstab`, either
 the default `/etc/fstab`, or any custom one selected from the command
 line, or at build time.
 
-Any file system not listed `fstab` are automatically mounted by Finit,
+To provide a smooth ride, file system not listed in the given `fstab`,
-as listed above, provided their respective mount point exists.
+e.g. `/tmp` and `/run`, are automatically mounted by Finit, as listed
+above, provided their respective mount point exists.
 
 With all filesystems mounted, Finit calls `swapon`.
 
+> **Tip:** to see what happens when all filesystems are mounted, have a
+> look at the [`bootmisc.so` plugin](plugins.md).
+
 At shutdown, and after having stopped all services and other lingering
 processes have been killed, filesystems are unmounted in the reverse
 order, and `swapoff` is called.
@@ -154,12 +206,13 @@ order, and `swapoff` is called.
  - `  0`: shutdown
 
 Runlevels are declared per service/run/task/sysv command.  Starting in
-runlevel S ([bootStrap](bootstrap.md)), usually only for tasks supposed
+runlevel S (bootStrap), usually only for tasks supposed to run once at
-to run once at boot, and services like `syslogd`, which you need to
+boot, and services like `syslogd`, which you need to start and run
-start and run throughout the whole time your system is up.  Before `S`
+throughout the whole time your system is up.
-is started, however, Finit performs a lot of housekeeping tasks like
+
-mounting all filesystems, calling `fsck` if needed, and making sure the
+Before `S` is started, however, Finit performs a lot of housekeeping
-everything is OK.
+tasks like mounting all filesystems, calling `fsck` if needed, and
+making sure the everything is OK.
 
     task [S] /lib/console-setup/console-setup.sh
     service [S12345] env:-/etc/default/rsyslog rsyslogd -n $RSYSLOGD_ARGS
@@ -472,7 +525,7 @@ following directives:
 
 
 > **Note:** not all directives are available in `/etc/finit.d/*.conf`
-> and some directives are only available at [bootstrap][], runlevel `S`,
+> and some directives are only available at bootstrap, runlevel `S`,
 > see [Limitations](#limitations) below for details.
 
 
@@ -485,7 +538,7 @@ the contents of that file is used.
 
 Deprecated.  We recommend using `/etc/hostname` instead.
 
-> **Note:** only read and executed in runlevel S ([bootstrap][]).
+> **Note:** only read and executed in runlevel S (bootstrap).
 
 
 ### Kernel Modules
@@ -503,7 +556,7 @@ BusyBox mdev tool, add to `/etc/mdev.conf`:
 
     $MODALIAS=.*  root:root       0660    @modprobe -b "$MODALIAS"
 
-> **Note:** only read and executed in runlevel S ([bootstrap][]).
+> **Note:** only read and executed in runlevel S (bootstrap).
 
 
 ### Networking
@@ -516,7 +569,7 @@ Deprecated.  We recommend using dedicated task/run stanzas per runlevel,
 or `/etc/network/interfaces` if you have a system with `ifupdown`, like
 Debian, Ubuntu, Linux Mint, or an embedded BusyBox system.
 
-> **Note:** only read and executed in runlevel S ([bootstrap][]).
+> **Note:** only read and executed in runlevel S (bootstrap).
 
 
 ### Alternate finit.d/
@@ -571,7 +624,7 @@ stanzas can share the same rlimits if they are in the same .conf.
 
 **Syntax:** `runlevel <N>`
 
-The system runlevel to go to after [bootstrap][] (S) has completed.  `N`
+The system runlevel to go to after bootstrap (S) has completed.  `N`
 is the runlevel number 0-9, where 6 is reserved for reboot and 0 for
 halt.
 
@@ -580,7 +633,7 @@ Finit disables networking in this mode.
 
 *Default:* 2
 
-> **Note:** only read and executed in runlevel S ([bootstrap][]).
+> **Note:** only read and executed in runlevel S (bootstrap).
 
 
 ### One-shot Commands (sequence)
@@ -620,11 +673,14 @@ also the `--quiet` and `--batch` options.
 
 One-shot like 'run', but starts in parallel with the next command.
   
-Both `run` and `task` commands are run in a shell, so pipes and
+Both `run` and `task` commands are run in a shell, so basic pipes and
-redirects can be freely used:
+redirects can be used:
 
     task [s] echo "foo" | cat >/tmp/bar
 
+Please note, `;`, `&&`, `||`, and similar are *not supported*.  Any
+non-trivial constructs are better placed in a separate shell script.
+
 
 ### SysV Init Scripts
 
@@ -860,7 +916,7 @@ if that is not found Finit automatically logs a warning.  The `nowarn`
 option disables this warning so that the second line can be evaluated,
 which also provides a service named `udevd`.
 
-    run nowarn if:udevd <pid/udevd> udevadm settle -t 0
+    run nowarn if:udevd <pid/udevd> :1 udevadm settle -t 0
 
 This line is only loaded if we know of a service named `udevd`.  Again,
 we do not warn if `udevadm` is not found, Execution will also stop here
@@ -915,9 +971,8 @@ previous stanza can also be written as:
 **Syntax:** `runparts [progress] [sysv] <DIR>`
 
 Call [run-parts(8)][] on `DIR` to run start scripts.  All executable
-files, or scripts, in the directory are called, in alphabetic order.
+files in the directory are called, in alphabetic order.  The scripts in
-The scripts in this directory are executed at the very end of runlevel
+this directory are executed at the very end of runlevel `S`.
-`S`, [bootstrap][].
 
 A common use-case for runparts scripts is to create and enable/disable
 services, which Finit will then apply when changing runlevel from S to
@@ -942,10 +997,9 @@ before continuing.  None of them can issue commands to start, stop, or
 restart other services.  Also, ensure all your services and programs
 either terminate or start in the background or you will block Finit.
 
-> **Note:** `runparts` scrips are only read and executed in runlevel S
+> **Note:** `runparts` scrips are only read and executed in runlevel S.
-> ([bootstrap][]).  See [hook scripts](plugins.md#hooks) for other ways
+> See [hook scripts](plugins.md#hooks) for other ways to run scripts at
-> to run scripts at certain points during the complete lifetime of the
+> certain points during the complete lifetime of the system.
-> system.
 
 **Recommendations:**
 
@@ -1058,8 +1112,8 @@ can be omitted to keep the kernel default.
     tty [12345] @console noclear vt220
 
 On really bare bones systems, or for board bringup, Finit can give you a
-shell prompt as soon as [bootstrap][] is done, without opening any
+shell prompt as soon as bootstrap is done, without opening any device
-device node:
+node:
 
     tty [12345789] notty
 
@@ -1294,12 +1348,11 @@ Watchdog
 When built `--with-watchdog` a separate service is built and installed
 in `/libexec/finit/watchdogd`.  If this exists at runtime, and the WDT
 device node exists, Finit will start it and treat it as the elected
-watchdog service to delegate its reboot to.  See [bootstrap][] for
+watchdog service to delegate its reboot to.  This delegation is to
-details.  This delegation is to ensure that the system is rebooted by a
+ensure that the system is rebooted by a hardware watchdog timer -- on
-hardware watchdog timer -- on many embedded systems this is crucial to
+many embedded systems this is crucial to ensure all circuits on the
-ensure all circuits on the board are properly reset for the next boot,
+board are properly reset for the next boot, in effect ensuring the
-in effect ensuring the system works the same after both a power-on and
+system works the same after both a power-on and reboot event.
-reboot event.
 
 The delegation is performed at the very last steps of system shutdown,
 if reboot has been selected and an elected watchdog is known, first a
@@ -1321,8 +1374,6 @@ device descriptor.  If the kernel driver has been built without this,
 the only option is to remove `/libexec/finit/watchdogd` or build without
 it at configure time.
 
-[bootstrap]: bootstrap.md
-
 
 keventd
 -------

doc/plugins.md

@@ -28,9 +28,16 @@ For your convenience a set of *optional* plugins are available:
   system and runs at `HOOK_BASEFS_UP`.  The `/var`, `/run`, and `/dev`
   file systems must be writable for this plugin to work.
 
-  Note: On an embedded system both `/var` and `/run` can be `tmpfs` RAM
+  This plugin is a wrapper for the [tmpfiles.d(5)][] implementation that
-  disks and `/dev` is usually a `devtmpfs`.  This must be defined in the
+  Finit has.  Capable but limited: no aging, attributes, or subvolumes.
-  `/etc/fstab` file and in the Linux kernel config.
+
+  By default, `/lib/finit/tmpfiles.d` carries all the default .conf
+  files distributed with Finit.  It is read first but can be overridden
+  by any of the standard tmpfiles.d directories, e.g. `/etc/tmpfiles.d`.
+
+  > **Note:** On an embedded system both `/var` and `/run` can be `tmpfs`
+  > RAM disks and `/dev` is usually a `devtmpfs`.  This must be defined
+  > in the `/etc/fstab` file and in the Linux kernel config.
 
 * *dbus.so*: Setup and start system message bus, D-Bus, at boot.
   _Optional plugin._
@@ -50,8 +57,11 @@ For your convenience a set of *optional* plugins are available:
   customized at build-time using the `--with-hook-scripts-path=PATH`
   argument to `configure`.
 
-* *hotplug.so*: Setup and start either udev or mdev hotplug daemon, if
+* *hotplug.so*: Replaced with `/lib/finit/system/10-hotplug.conf`, which
-  available.  Enabled by default.
+  checks for `udevd` and `mdev` at boot.  This file can be overridden by a
+  file in `/etc/finit.d/10-hotplug.conf`.
+
+  Enabled by default.
   
   > See the [Services](config.md#services) section in the configuration
   > guide for an example how to run `mdevd`, alternative to plain mdev.
@@ -217,3 +227,5 @@ Plugins like `tty.so` extend finit by acting on events, they are called
 I/O plugins and are called from the finit main loop when `poll()`
 detects an event.  See the source code for `plugins/*.c` for more help
 and ideas.
+
+[tmpfiles.d(5)]: https://www.freedesktop.org/software/systemd/man/tmpfiles.d.html

plugins/dbus.c

@@ -83,6 +83,7 @@ static char *dbus_pidfn(void)
 				break;
 			}
 			*pos = 0;
+			break;
 		}
 
 		fclose(fp);
@@ -142,7 +143,7 @@ static void setup(void *arg)
 
 	/* Generate machine id for dbus */
 	if (whichp("dbus-uuidgen"))
-		run_interactive("dbus-uuidgen --ensure", "Creating machine UUID for D-Bus");
+		run_interactive("dbus-uuidgen --ensure", "Verifying D-Bus machine UUID");
 
 	/*
 	 * Register service with Finit
@@ -156,9 +157,7 @@ static void setup(void *arg)
 		snprintf(line, sizeof(line), "[S123456789] cgroup.system name:dbus %s %s -- %s",
 			 cmd, DBUS_ARGS, DBUS_DESC);
 
-	if (service_register(SVC_TYPE_SERVICE, line, global_rlimit, NULL))
+	conf_save_service(SVC_TYPE_SERVICE, line, "dbus.conf");
-		err(1, "Failed registering %s", DBUS_DAEMON);
-
 	free(cmd);
 }
 

src/Makefile.am

@@ -19,6 +19,8 @@ if WATCHDOGD
 pkglibexec_PROGRAMS += watchdogd
 endif
 
+dist_pkglibexec_SCRIPTS = coldplug
+
 finit_pkglibdir      = $(pkglibdir)
 finit_pkglib_SCRIPTS = rescue.conf sample.conf
 

src/coldplug

@@ -0,0 +1,11 @@
+#!/bin/sh
+
+# With the daemon mode ('mdev -df') we no longer need the hotplug helper
+echo "" > /proc/sys/kernel/hotplug
+
+# Wait for 'mdev -df' to set up its event queue.
+sleep 1
+
+# Trigger 'add' event for all devices and subsystems to load all modules
+# and any firmware needed by the hardware.
+find /sys/devices -name uevent -print | while IFS='' read X; do echo add > $X; done

src/cond-w.c

@@ -222,8 +222,16 @@ static int do_delete(const char *fpath, const struct stat *sb, int tflag, struct
 static void cond_delpath(const char *path)
 {
 	nftw(path, do_delete, 20, FTW_DEPTH);
-	if (remove(path) && errno != ENOENT)
+	if (remove(path)) {
-		err(1, "Failed removing condition path %s", path);
+		switch (errno) {
+		case ENOENT:	/* Ignore */
+		case ENOTEMPTY:	/* Only at shutdown and for directories */
+			break;
+		default:
+			err(1, "Failed removing condition path %s", path);
+			break;
+		}
+	}
 }
 
 int cond_set_path(const char *path, enum cond_state next)

src/conf.c

@@ -414,6 +414,79 @@ void conf_saverc(void)
 	fclose(fp);
 }
 
+/*
+ * Called at bootstrap to log execution order (for debug)
+ */
+void conf_save_exec_order(svc_t *svc, char *cmdline, int result)
+{
+	const char *fn = _PATH_VARRUN "finit/exec.order";
+	const char *ststr = !result ? "[ OK ]" : "[FAIL]";
+	static char *prepared = NULL;
+	int first = !fexist(fn);
+	FILE *fp;
+
+	fp = fopen(fn, "a+");
+	if (!fp) {
+		err(1, "failed writing to %s", fn);
+		return;
+	}
+
+	if (first) {
+		fprintf(fp, "# Execution order of run/task/services at bootstrap\n");
+		fprintf(fp, "# ST    TYPE     COMMAND LINE (DESC)\n");
+	}
+
+	if (result == -1) {
+		int len;
+
+		len = snprintf(NULL, 0, "        %-7s  %-64s (%s)", svc_typestr(svc), cmdline, svc->desc);
+		if (len > 0) {
+			if (prepared)
+				free(prepared);
+			prepared = malloc(++len);
+			if (!prepared)
+				goto done;
+			snprintf(prepared, len, "        %-7s  %-64s (%s)", svc_typestr(svc), cmdline, svc->desc);
+		}
+	} else {
+		if (prepared) {
+			memcpy(prepared, ststr, strlen(ststr));
+			fprintf(fp, "%s\n", prepared);
+			free(prepared);
+			prepared = NULL;
+		} else {
+			fprintf(fp, "%s  %-7s  %-64s (%s)\n", ststr, svc_typestr(svc), cmdline, svc->desc);
+		}
+	}
+done:
+	fclose(fp);
+}
+
+/*
+ * Called by plugins and similar that dynamically generate system services
+ */
+void conf_save_service(int type, char *cfg, char *file)
+{
+	char fn[256];
+	svc_t foo;
+	FILE *fp;
+
+	mkpath(FINIT_RUNPATH_, 0755);
+	snprintf(fn, sizeof(fn), "%s/%s", FINIT_RUNPATH_, file);
+	if (fexist(fn))
+		warnx("File %s already exists, overwriting.", fn);
+	fp = fopen(fn, "w");
+	if (!fp) {
+		err(1, "Failed creating %s", fn);
+		return;
+	}
+
+	foo.type = type;
+	fprintf(fp, "# Generated by finit:%s()\n", __func__);
+	fprintf(fp, "%s %s\n", svc_typestr(&foo), cfg);
+	fclose(fp);
+}
+
 /*
  * Sets, and makes a note of, all KEY=VALUE lines in a given .conf line
  * from finit.conf, or other .conf file.  Note, PATH is always reset in
@@ -583,11 +656,22 @@ void conf_parse_cond(svc_t *svc, char *cond)
 		return;
 	}
 
-	/* First character must be '!' if SIGHUP is not supported. */
+	/*
+	 * First character must be '!' if:
+	 *   - service:  SIGHUP is not supported
+	 *   - run/task: Do not block bootstrap
+	 */
 	ptr = cond;
 	if (ptr[i] == '!') {
-		svc->sighup = 0;
 		ptr++;
+
+		if (svc_is_runtask(svc)) {
+			/* see service_runtask_clean() */
+			svc->sighup = 1;
+			svc->once = 1;
+		} else {
+			svc->sighup = 0;
+		}
 	}
 
 	while (ptr[i] != '>' && ptr[i] != 0)
@@ -1152,20 +1236,51 @@ int conf_reload(void)
 	 * but are now regular files that can be overridden by files in
 	 * /etc/finit.d -- similar to how tmfiles.d(5) work.  E.g., add
 	 * an override .conf, or an ignore by symlinking to /dev/null
+	 *
+	 * The .conf files (and run/task/service stanzas) are parsed and
+	 * started in order.  Each directory is sorted alphanumerically
+	 * and then the result is appended to the overall order:
+	 *
+	 *     /lib/finit/system/10-hotplug.conf
+	 *     /lib/finit/system/90-testserv.conf
+	 *     /run/finit/system/dbus.conf
+	 *     /run/finit/system/tty.conf
+	 *     /etc/finit.d/10-abc.conf
+	 *     /etc/finit.d/20-abc.conf
+	 *     /etc/finit.d/enabled/1-aaa.conf
+	 *     /etc/finit.d/enabled/1-abc.conf
+	 *     /etc/finit.d/enabled/2-aaa.conf
 	 */
-	glob_append(&gl, 0, "%s/*.conf", finit_rcsd);
+	glob_append(&gl, 0, "%s/*.conf", FINIT_SYSPATH_);
-	glob_append(&gl, 1, "%s/*.conf", FINIT_SYSPATH);
+	glob_append(&gl, 1, "%s/*.conf", FINIT_RUNPATH_);
+	glob_append(&gl, 1, "%s/*.conf", finit_rcsd);
 	glob_append(&gl, 1, "%s/enabled/*.conf", finit_rcsd);
 
+	if (bootstrap) {
+		const char *fn = _PATH_VARRUN "finit/conf.order";
+		FILE *fp;
+
+		fp = fopen(fn, "w");
+		if (!fp) {
+			err(1, "failed creating %s", fn);
+		} else {
+			fprintf(fp, "# Evaluation & execution order of .conf files\n");
+			for (i = 0; i < gl.gl_pathc; i++)
+				fprintf(fp, "%s\n", gl.gl_pathv[i]);
+			fclose(fp);
+		}
+	}
+
 	for (i = 0; i < gl.gl_pathc; i++) {
 		char *path = gl.gl_pathv[i];
 		char *rp = NULL;
 		struct stat st;
 		size_t j, len;
 
-		/* check for FINIT_SYSPATH overrides */
+		/* check for FINIT_SYSPATH_ or FINIT_RUNPATH_ overrides */
 		for (j = i + 1; j < gl.gl_pathc; j++) {
-			if (strncmp(path, FINIT_SYSPATH, strlen(FINIT_SYSPATH)))
+			if (strncmp(path, FINIT_SYSPATH_, strlen(FINIT_SYSPATH_)) &&
+			    strncmp(path, FINIT_RUNPATH_, strlen(FINIT_RUNPATH_)))
 				continue;
 			if (strcmp(basename(path), basename(gl.gl_pathv[j])))
 				continue;
@@ -1225,6 +1340,9 @@ done:
 	/* Drop record of all .conf changes */
 	drop_changes();
 
+	if (bootstrap)
+		wdog = svc_find("watchdog", "finit");
+
 	/* Override configured runlevel, user said 'S' on /proc/cmdline */
 	if (BOOTSTRAP && single)
 		cfglevel = 1;
@@ -1461,15 +1579,16 @@ int conf_init(uev_ctx_t *ctx)
 	 */
 #ifdef WDT_DEVNODE
 	if (whichp(FINIT_EXECPATH_ "/watchdogd") && fexist(WDT_DEVNODE)) {
-		service_register(SVC_TYPE_SERVICE, "[S0123456789] cgroup.init name:watchdog :finit " FINIT_EXECPATH_ "/watchdogd -- Finit watchdog daemon", global_rlimit, NULL);
+		conf_save_service(SVC_TYPE_SERVICE, "[S0123456789] cgroup.init name:watchdog :finit "
-		wdog = svc_find("watchdog", "finit");
+				  FINIT_EXECPATH_ "/watchdogd -- Finit watchdog daemon", "watchdogd.conf");
 	}
 #endif
 	/*
 	 * Start kernel event daemon as soon as possible, if enabled
 	 */
 	if (whichp(FINIT_EXECPATH_ "/keventd"))
-		service_register(SVC_TYPE_SERVICE, "[S0123456789] cgroup.init " FINIT_EXECPATH_ "/keventd -- Finit kernel event daemon", global_rlimit, NULL);
+		conf_save_service(SVC_TYPE_SERVICE, "[S0123456789] cgroup.init "
+				  FINIT_EXECPATH_ "/keventd -- Finit kernel event daemon", "keventd.conf");
 
 	dbg("Allow plugins to register early runlevel 1 run/task/services ...");
 	plugin_run_hooks(HOOK_SVC_PLUGIN);
@@ -1505,7 +1624,7 @@ int conf_init(uev_ctx_t *ctx)
 		snprintf(conf, sizeof(conf), "[S] <int/bootstrap> log:console %s %s %s"
 			 " -- Calling runparts %s in the background",
 			 _PATH_RUNPARTS, args, runparts, runparts);
-		service_register(SVC_TYPE_TASK, conf, global_rlimit, NULL);
+		conf_save_service(SVC_TYPE_TASK, conf, "runparts.conf");
 	}
 
 	return 0;

src/conf.h

@@ -44,6 +44,8 @@ int  conf_monitor         (void);
 
 void conf_reset_env       (void);
 void conf_saverc          (void);
+void conf_save_exec_order (svc_t *svc, char *cmdline, int result);
+void conf_save_service    (int type, char *cfg, char *file);
 void conf_parse_cmdline   (int argc, char *argv[]);
 int  conf_parse_runlevels (char *runlevels);
 void conf_parse_cond      (svc_t *svc, char *cond);

src/finit.c

@@ -204,7 +204,7 @@ static int fsck(int pass)
 
 	fp = setmntent(fstab, "r");
 	if (!fp) {
-		err(1, "Failed opening fstab: %s", fstab);
+		logit(LOG_CONSOLE | LOG_ALERT, "Failed opening fstab: %s", fstab);
 		sulogin(1);
 	}
 	dbg("Opened %s, pass %d", fstab, pass);
@@ -695,6 +695,10 @@ int main(int argc, char *argv[])
 	cond_set_oneshot(plugin_hook_str(HOOK_BANNER));
 	cond_set_oneshot(plugin_hook_str(HOOK_ROOTFS_UP));
 
+	/* Some bootstrap tasks may need to know if we're in a container. */
+	if (in_container())
+		cond_set("int/container");
+
 	/*
 	 * Initialize .conf system and load static /etc/finit.conf then
 	 * tell the world what we used.

src/service.c

@@ -330,14 +330,14 @@ static int lredirect(svc_t *svc)
 
 				snprintf(rot, sizeof(rot), "%d:%d", logfile_size_max, logfile_count_max);
 				snprintf(pid, sizeof(pid), "%d", svc_pid);
-				execlp("logger", "logger", "-f", svc->log.file, "-b", "-t", tag, "-p", prio, "-I", pid, "-r", rot, NULL);
+				execlp("logger", "logger", "-f", svc->log.file, "-b", "-t", tag, "-p", prio, debug ? "-s" : "", "-I", pid, "-r", rot, NULL);
 			} else {
 				char sz[20], num[3];
 
 				snprintf(sz, sizeof(sz), "%d", logfile_size_max);
 				snprintf(num, sizeof(num), "%d", logfile_count_max);
 
-				execlp(_PATH_LOGIT, "logit", "-f", svc->log.file, "-n", sz, "-r", num, NULL);
+				execlp(_PATH_LOGIT, "logit", "-f", svc->log.file, "-n", sz, "-r", num, debug ? "-s" : "", NULL);
 			}
 			_exit(1);
 		}
@@ -351,9 +351,9 @@ static int lredirect(svc_t *svc)
 			char pid[16];
 
 			snprintf(pid, sizeof(pid), "%d", svc_pid);
-			execlp("logger", "logger", "-t", tag, "-p", prio, "-I", pid, NULL);
+			execlp("logger", "logger", "-t", tag, "-p", prio, debug ? "-s" : "", "-I", pid, NULL);
 		} else {
-			execlp(_PATH_LOGIT, "logit", "-t", tag, "-p", prio, NULL);
+			execlp(_PATH_LOGIT, "logit", "-t", tag, "-p", prio, debug ? "-s" : "", NULL);
 		}
 		_exit(1);
 	}
@@ -413,6 +413,7 @@ static void source_env(svc_t *svc)
 	len = sizeof(buf);
 	while (fgets(line, len, fp)) {
 		char *key = chomp(line);
+		wordexp_t we = { 0 };
 		char *value, *end;
 
 		/* skip any leading whitespace */
@@ -466,7 +467,19 @@ static void source_env(svc_t *svc)
 				*end-- = 0;
 		}
 
-		setenv(key, value, 1);
+		switch (wordexp(value, &we, 0)) {
+		case 0:
+			setenv(key, we.we_wordv[0], 1);
+			wordfree(&we);
+			break;
+
+		case WRDE_NOSPACE:
+			wordfree(&we);
+			/* fallthrough */
+		default:
+			setenv(key, value, 1);
+			break;
+		}
 	}
 
 	fclose(fp);
@@ -629,6 +642,9 @@ static int service_start(svc_t *svc)
 	}
 
 	compose_cmdline(svc, cmdline, sizeof(cmdline));
+	if (bootstrap)
+		conf_save_exec_order(svc, cmdline, -1);
+
 	if (svc_is_sysv(svc))
 		logit(LOG_CONSOLE | LOG_NOTICE, "Calling '%s start' ...", cmdline);
 
@@ -838,6 +854,8 @@ fail:
 	sigprocmask(SIG_SETMASK, &omask, NULL);
 	if (do_progress && !run_block_pid)
 		print_result(result);
+	if (bootstrap && !run_block_pid)
+		conf_save_exec_order(svc, NULL, result);
 
 	return result;
 }
@@ -974,7 +992,11 @@ int service_stop(svc_t *svc)
 	if (!svc->desc[0])
 		do_progress = 0;
 
-	if (runlevel != 1 && do_progress)
+	/*
+	 * Skip run/tasks in progress, would otherwise print silly stuff
+	 * like: "Stopping Shutting down" ...
+	 */
+	if (runlevel != 1 && do_progress && (svc_is_daemon(svc) || svc_is_sysv(svc)))
 		print_desc("Stopping ", svc->desc);
 
 	if (!svc_is_sysv(svc)) {
@@ -1030,7 +1052,7 @@ int service_stop(svc_t *svc)
 		}
 	}
 
-	if (runlevel != 1 && do_progress)
+	if (runlevel != 1 && do_progress && (svc_is_daemon(svc) || svc_is_sysv(svc)))
 		print_result(rc);
 
 	return rc;
@@ -1396,7 +1418,11 @@ static void parse_cmdline_args(svc_t *svc, char *cmd, char **args)
  * before they are started.  Or restarted, or even SIGHUP'ed, when the
  * gateway changes or interfaces come and go.  The special case when a
  * service is declared with <!> means it does not support SIGHUP but
- * must be STOP/START'ed at system reconfiguration.
+ * must be STOP/START'ed at system reconfiguration.  For run/task a <!>
+ * means Finit can relax its promise to run at least once per runlevel.
+ * I.e., for run/task conditions that would otherwise block bootstrap:
+ *
+ *     task [S0123456789] <!sys/pwr/fail> name:pwrfail initctl poweroff -- Power failure, shutting down
  *
  * Conditions can for example be: pid/NAME:ID for process dependencies,
  * net/<IFNAME>/up or net/<IFNAME>/exists.  The condition handling is
@@ -1425,6 +1451,7 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 	char *id = NULL, *env = NULL, *cgroup = NULL;
 	char *pre_script = NULL, *post_script = NULL;
 	char *ready_script = NULL, *conflict = NULL;
+	char ident[MAX_IDENT_LEN];
 	char *ifstmt = NULL;
 	char *notify = NULL;
 	struct tty tty = { 0 };
@@ -1549,8 +1576,16 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 	}
 
 	name = parse_name(cmd, name);
-	if (!id)
+	strlcpy(ident, name, sizeof(ident));
+	if (!id) {
 		id = "";
+	} else {
+		strlcat(ident, ":", sizeof(ident));
+		strlcat(ident, id, sizeof(ident));
+	}
+
+	if (ifstmt && !svc_ifthen(1, ident, ifstmt))
+		return 0;
 
 	levels = conf_parse_runlevels(runlevels);
 	if (runlevel != INIT_LEVEL && !ISOTHER(levels, INIT_LEVEL)) {
@@ -1896,10 +1931,14 @@ done:
 	svc->start_time = svc->pid = 0;
 cont:
 	if (lost == run_block_pid) {
+		int result = WIFEXITED(status) ? WEXITSTATUS(status) : 1;
+
 		svc_mark_clean(svc); /* done, regardless of exit status */
 		run_block_pid = 0;
 		if (svc->desc[0])
-			print_result(WIFEXITED(status) ? WEXITSTATUS(status) : 1);
+			print_result(result);
+		if (bootstrap)
+			conf_save_exec_order(svc, NULL, result);
 	}
 
 	if (!service_step(svc)) {
@@ -2185,6 +2224,7 @@ static void service_retry(svc_t *svc)
 static void svc_set_state(svc_t *svc, svc_state_t new_state)
 {
 	svc_state_t *state = (svc_state_t *)&svc->state;
+	const svc_state_t old_state = svc->state;
 
 	/* if PID isn't collected within SVC_TERM_TIMEOUT msec, kill it! */
 	if (new_state == SVC_STOPPING_STATE) {
@@ -2205,7 +2245,7 @@ static void svc_set_state(svc_t *svc, svc_state_t new_state)
 		snprintf(failure, sizeof(failure), "%s/%s/failure", svc_typestr(svc), svc_ident(svc, NULL, 0));
 
 		/* create success/failure condition when entering SVC_DONE_STATE. */
-		if (*state == SVC_DONE_STATE) {
+		if (new_state == SVC_DONE_STATE) {
 			if (svc->started && !WEXITSTATUS(svc->status))
 				cond_set_oneshot(success);
 			else
@@ -2213,7 +2253,7 @@ static void svc_set_state(svc_t *svc, svc_state_t new_state)
 		}
 
 		/* clear all conditions when entering SVC_HALTED_STATE. */
-		if (*state == SVC_HALTED_STATE) {
+		if (new_state == SVC_HALTED_STATE) {
 			cond_clear(success);
 			cond_clear(failure);
 		}
@@ -2223,9 +2263,14 @@ static void svc_set_state(svc_t *svc, svc_state_t new_state)
 		char cond[MAX_COND_LEN];
 
 		snprintf(cond, sizeof(cond), "service/%s/", svc_ident(svc, NULL, 0));
-		cond_clear(cond);
 
-		switch (svc->state) {
+		if ((old_state == SVC_RUNNING_STATE && new_state == SVC_PAUSED_STATE) ||
+		    (old_state == SVC_PAUSED_STATE  && new_state == SVC_RUNNING_STATE))
+			; 	/* only paused during reload, don't clear conds. */
+		else
+			cond_clear(cond);
+
+		switch (new_state) {
 		case SVC_HALTED_STATE:
 		case SVC_RUNNING_STATE:
 			strlcat(cond, svc_status(svc), sizeof(cond));
@@ -2463,7 +2508,7 @@ restart:
 
 		case COND_ON:
 			if (svc_is_changed(svc)) {
-				if (svc_nohup(svc))
+				if (svc_nohup(svc) || !svc_is_daemon(svc))
 					service_stop(svc);
 				else {
 					/*
@@ -2564,7 +2609,12 @@ void service_runtask_clean(void)
 		if (!svc_is_runtask(svc))
 			continue;
 
-		svc->once = 0;
+		/* run/task declared with <!> */
+		if (svc->sighup)
+			svc->once = 1;
+		else
+			svc->once = 0;
+
 		if (svc->state == SVC_DONE_STATE)
 			svc_set_state(svc, SVC_HALTED_STATE);
 	}
@@ -2572,6 +2622,7 @@ void service_runtask_clean(void)
 
 /**
  * service_completed - Have run/task completed in current runlevel
+ * @svcp: On %FALSE return, this points to the first incomplete svc
  *
  * This function checks if all run/task have run once in the current
  * runlevel.  E.g., at bootstrap we must wait for these scripts or
@@ -2584,7 +2635,7 @@ void service_runtask_clean(void)
  * Returns:
  * %TRUE(1) or %FALSE(0)
  */
-int service_completed(void)
+int service_completed(svc_t **svcp)
 {
 	svc_t *svc, *iter = NULL;
 
@@ -2606,6 +2657,8 @@ int service_completed(void)
 
 		if (!svc->once) {
 			dbg("%s has not yet completed ...", svc_ident(svc, NULL, 0));
+			if (svcp)
+				*svcp = svc;
 			return 0;
 		}
 		dbg("%s has completed ...", svc_ident(svc, NULL, 0));

src/service.h

@@ -49,7 +49,7 @@ int       service_step           (svc_t *svc);
 void      service_step_all       (int types);
 void      service_worker         (void *unused);
 
-int       service_completed      (void);
+int       service_completed      (svc_t **svc);
 void      service_notify_reconf  (void);
 
 void      service_init           (uev_ctx_t *ctx);

src/sm.c

@@ -52,22 +52,30 @@ sm_t sm;
  */
 static void sm_check_bootstrap(void *work)
 {
-	static int cnt = 120 * 10;	/* We run with 100ms period */
+	static int timeout = 120;
-        int level = cfglevel;
+	int bootstrap_done;
+	int level = cfglevel;
+	svc_t *svc = NULL;
 
 	dbg("Step all services ...");
 	service_step_all(SVC_TYPE_ANY);
 
-	if (cnt-- > 0 && !service_completed()) {
+	bootstrap_done = service_completed(&svc);
-		dbg("Not all bootstrap run/tasks have completed yet ... %d", cnt);
+	if (timeout-- > 0 && !bootstrap_done) {
+		dbg("Not all bootstrap run/tasks have completed yet ... %d", timeout);
 		schedule_work(work);
 		return;
 	}
 
-	if (cnt > 0)
+	if (timeout > 0) {
 		dbg("All run/task have completed, resuming bootstrap.");
-	else
+	} else {
 		dbg("Timeout, resuming bootstrap.");
+		if (svc)
+			print(2, "Timeout waiting for %s to run, resuming bootstrap", svc_ident(svc, NULL, 0));
+		else
+			print(2, "Timeout waiting for unknown run/task, resuming bootstrap");
+	}
 
 	dbg("Flushing pending .conf file events ...");
 	conf_flush_events();

src/svc.c

@@ -710,6 +710,7 @@ int svc_conflicts(svc_t *svc)
  */
 int svc_ifthen(int is_conf, const char *ident, char *stmt)
 {
+	char stmts[MAX_IDENT_LEN];
 	int not = 0;
 	svc_t *svc;
 
@@ -724,39 +725,46 @@ int svc_ifthen(int is_conf, const char *ident, char *stmt)
 		if (is_conf)
 			return 1;
 
-		if (stmt[0] == '!') {
-			stmt++;
-			not++;
-		}
-
 		while (ptr[i] != '>' && ptr[i] != 0)
 			i++;
 		ptr[i] = 0;
 
-		cond = cond_get(stmt);
+		strlcpy(stmts, stmt, sizeof(stmts));
-		if (not && cond == COND_ON)
+		for (stmt = strtok(stmts, ","); stmt; stmt = strtok(NULL, ","), not = 0) {
-			return 0;
+			if (stmt[0] == '!') {
-		if (!not && cond == COND_OFF)
+				stmt++;
-			return 0;
+				not++;
+			}
+
+			cond = cond_get(stmt);
+			if (not && cond == COND_ON)
+				return 0;
+			if (!not && cond == COND_OFF)
+				return 0;
+		}
+
 		return 1;
 	}
 
 	if (!is_conf)
 		return 1;
 
-	if (stmt[0] == '!') {
+	strlcpy(stmts, stmt, sizeof(stmts));
-		stmt++;
+	for (stmt = strtok(stmts, ","); stmt; stmt = strtok(NULL, ","), not = 0) {
-		not++;
+		if (stmt[0] == '!') {
-	}
+			stmt++;
+			not++;
+		}
 
-	svc = svc_find_by_str(stmt);
+		svc = svc_find_by_str(stmt);
-	if (not && svc) {
+		if (not && svc) {
-		logit(LOG_NOTICE, "skipping %s, %s already loaded.", ident, svc_ident(svc, NULL, 0));
+			logit(LOG_NOTICE, "skipping %s, %s already loaded.", ident, svc_ident(svc, NULL, 0));
-		return 0;
+			return 0;
-	}
+		}
-	if (!not && !svc) {
+		if (!not && !svc) {
-		logit(LOG_NOTICE, "skipping %s, %s not available.", ident, stmt);
+			logit(LOG_NOTICE, "skipping %s, %s not available.", ident, stmt);
-		return 0;
+			return 0;
+		}
 	}
 
 	return 1;

src/tmpfiles.c

@@ -471,7 +471,7 @@ void tmpfilesd(void)
 	/* in priority order */
 	char *dir[] = {
 		FINIT_TMPFILES "/*.conf",
-		TMPFILES_PATH "/*.conf",
+		TMPFILES_PATH_ "/*.conf",
 		"/usr/lib/tmpfiles.d/*.conf",
 		"/run/tmpfiles.d/*.conf",
 		"/etc/tmpfiles.d/*.conf", /* local admin overrides */

system/.gitignore

@@ -1,2 +1,4 @@
 Makefile
 Makefile.in
+# generated
+10-hotplug.conf

system/10-hotplug.conf.in

@@ -0,0 +1,53 @@
+# System hotplug handled by udev, mdevd, or mdev (preferred order)
+#
+# Your system usually has either udevd, mdev, or mdevd.  They probe the system
+# for devices, loading modules and any firmware needed.  Since the bootstrap
+# often needs to be synchronized carefully, we suggest using your system log
+# daemon as the barrier for all other services.  E.g.:
+#
+#    service if:udevd <run/udevadm:5/success> [S0123456789] syslogd -F -- System log daemon
+#    service if:mdev  <run/coldplug/success>  [S0123456789] syslogd -F -- System log daemon
+#    service if:mdevd <run/coldplug/success>  [S0123456789] syslogd -F -- System log daemon
+#
+# This provdes a <pid/syslogd> condition that can act as a barrier for all
+# other services.  Notice the `if:` statement and the condition for starting
+# syslogd is only considered if either udevd or mdev (service) is loaded and
+# is guaranteed to run after each respective run stanza have completed.
+#
+# Override this file by copying it to /etc/finit.d/, using the same name, then
+# change the contents any way you like, it can even be empty.
+
+# Check for systemd-udevd and eudev, if we find both, we opt for the latter.
+service nowarn [S12345789] cgroup.system name:udevd pid:udevd log /lib/systemd/systemd-udevd -- Device event daemon (udev)
+service nowarn [S12345789] cgroup.system name:udevd pid:udevd log udevd -D -- Device event daemon (udev)
+
+# Wait for udevd to start, then trigger coldplug events and module loading.
+# The last 'settle' call waits for it to finalize processing all uevents.
+run nowarn if:udevd cgroup.init :1 [S] <pid/udevd> log udevadm settle -t 0                  --
+run nowarn if:udevd cgroup.init :2 [S] <pid/udevd> log udevadm control --reload             --
+run nowarn if:udevd cgroup.init :3 [S] <pid/udevd> log udevadm trigger -c add -t devices    -- Requesting device events
+run nowarn if:udevd cgroup.init :4 [S] <pid/udevd> log udevadm trigger -c add -t subsystems -- Requesting subsystem events
+run nowarn if:udevd cgroup.init :5 [S] <pid/udevd> log udevadm settle -t 30                 -- Waiting for udev to finish
+
+# If udevd is not available, check for mdevd.  Unlike plain 'mdev -df', the
+# mdevd-coldplug process exits when mdevd has completed.  So the condition
+# <run/coldplug/success> (same as for plain mdev below) can be relied on as a
+# barrier for services that, e.g., need to probe for switchcore ports or PHYs
+# that require firmware loading before they show up.
+service nowarn conflict:udevd,mdev cgroup.init if:!udevd \
+	[S12345789] notify:s6 mdevd -O 4 -D %n -- Device event daemon (mdevd)
+run     nowarn conflict:udevd,mdev cgroup.init name:coldplug <service/mdevd/ready> \
+	[S] mdevd-coldplug -- Cold plugging system
+
+# If neither udevd or mdevd are available, fall back to mdev.  The new daemon
+# mode calls `mdev -s` internally but does not create a PID file when that is
+# done, so synchronization with coldplug is fragile.
+service nowarn conflict:udevd,mdevd cgroup.init if:!udevd,!mdevd \
+	[S12345789] mdev -df -S -- Device event daemon (mdev)
+
+# Trigger a replay of all uevents from boot, before userspace was up and the
+# filesystem mounted.  Compared to udevd, we don't know when 'mdev -df' has
+# finished processing all events, i.e., depending on <run/coldplug/success> is
+# not an indicator of all devices having been (re)discovered and fully set up.
+run nowarn conflict:udevd,mdevd cgroup.init name:coldplug if:mdev <service/mdev/running> \
+	[S] @pkglibexecdir@/coldplug -- Cold plugging system

system/Makefile.am

@@ -1,10 +1,12 @@
+EXTRA_DIST        = 10-hotplug.conf.in
 systemdir         = $(system_path)
+system_DATA       =
 dist_system_DATA  =
 
 if BUILD_HOTPLUG_PLUGIN
-dist_system_DATA += hotplug.conf
+system_DATA      += 10-hotplug.conf
 endif
 
 if BUILD_TESTSERV_PLUGIN
-dist_system_DATA += testserv.conf
+dist_system_DATA += 90-testserv.conf
 endif

system/hotplug.conf

@@ -1,17 +0,0 @@
-# System hotplug can be handled by udev or mdev
-
-# Either we have systemd-udevd or eudev, on the rare occasion we find
-# both, we opt for the latter.  Remember, you can override this by a
-# custom hotplug.conf in /etc/finit.d/ or /etc/finit.d/enabled/.
-service nowarn [S12345789] cgroup.system name:udevd pid:udevd log /lib/systemd/systemd-udevd -- Device event managing daemon
-service nowarn [S12345789] cgroup.system name:udevd pid:udevd log udevd -- Device event managing daemon
-
-# Wait for udevd to start, then wait a couple of seconds for it to become ready
-run nowarn if:udevd cgroup.init    [S] <pid/udevd> udevadm settle -t 0 --
-run nowarn if:udevd cgroup.init    [S] <pid/udevd> log udevadm control --reload
-run nowarn if:udevd cgroup.init :1 [S] <pid/udevd> log udevadm trigger -c add -t devices    -- Requesting device events
-run nowarn if:udevd cgroup.init :2 [S] <pid/udevd> log udevadm trigger -c add -t subsystems -- Requesting subsystem events
-run nowarn if:udevd cgroup.init    [S] <pid/udevd> log udevadm settle -t 5 -- Waiting for udev to finish
-
-# If udevd is not available, try mdev
-run nowarn conflict:udevd [S] mdev -s -- Populating device tree

test/Makefile.am

@@ -32,6 +32,7 @@ EXTRA_DIST		+= crashing.sh
 EXTRA_DIST		+= depserv.sh
 EXTRA_DIST		+= devmon.sh
 EXTRA_DIST		+= failing-sysv.sh
+EXTRA_DIST		+= svc-env.sh
 EXTRA_DIST		+= global-envs.sh
 EXTRA_DIST		+= initctl-status-subset.sh
 EXTRA_DIST		+= notify.sh
@@ -53,6 +54,7 @@ EXTRA_DIST		+= start-stop-sysv.sh
 EXTRA_DIST		+= start-stop-serv.sh
 EXTRA_DIST		+= signal-service.sh
 EXTRA_DIST		+= testserv.sh
+EXTRA_DIST		+= unexpected-restart.sh
 
 AM_TESTS_ENVIRONMENT	 = SYSROOT='$(abs_builddir)/sysroot/';
 AM_TESTS_ENVIRONMENT	+= export SYSROOT;
@@ -75,6 +77,7 @@ TESTS			+= crashing.sh
 TESTS			+= depserv.sh
 TESTS			+= devmon.sh
 TESTS			+= failing-sysv.sh
+TESTS			+= svc-env.sh
 TESTS			+= global-envs.sh
 TESTS			+= initctl-status-subset.sh
 TESTS			+= notify.sh
@@ -98,6 +101,7 @@ TESTS			+= signal-service.sh
 if TESTSERV
 TESTS			+= testserv.sh
 endif
+TESTS			+= unexpected-restart.sh
 
 check-recursive: setup-chroot
 

test/lib/setup.sh

@@ -143,7 +143,15 @@ log()
 
 sep()
 {
-	printf "\e[2m――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――\e[0m\n"
+    heading=${1:-}
+    if [ -n "$heading" ]; then
+	num=$((72 - ${#heading} - 1))
+	printf "\e[1m%s " "$heading"
+	printf -- "―%.0s" $(seq 1 $num)
+	printf "\e[0m\n"
+    else
+	printf "\e[1m――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――\e[0m\n"
+    fi
 }
 
 say()

test/runparts.sh

@@ -37,5 +37,6 @@ while true; do
     sleep 1
 done
 
+sleep 1
 texec diff -u /usr/share/runparts/ref.log /tmp/runparts.log
 texec cmp /usr/share/runparts/ref.log /tmp/runparts.log || fail "runparts in wrong order"

test/setup-sysroot.sh

@@ -12,12 +12,12 @@ cp "$top_builddir/test/src/serv" "$SYSROOT/sbin/"
 FINITBIN="$(pwd)/$top_builddir/src/finit" DEST="$SYSROOT" make -f "$srcdir/lib/sysroot.mk"
 
 # Drop plugins we don't need in test, only causes confusing FAIL in logs.
-for plugin in tty.so urandom.so rtc.so modprobe.so hotplug.so; do
+for plugin in tty.so urandom.so rtc.so modprobe.so; do
     find "$SYSROOT" -name $plugin -delete
 done
 
 # Drop system .conf files we don't need in test, same as above
-for conf in hotplug.conf; do
+# shellcheck disable=SC2043
+for conf in 10-hotplug.conf; do
     find "$SYSROOT" -name $conf -delete
 done
-

test/skel/etc/env

@@ -3,3 +3,5 @@
 
 foo=bar
 baz=qux
+
+xyzzy=$foo

test/svc-env.sh

@@ -0,0 +1,34 @@
+#!/bin/sh
+# Ensure service environment variables are sourced and expanded properly.
+set -e
+
+TEST_DIR=$(dirname "$0")
+
+test_setup()
+{
+    say "Test start $(date)"
+}
+
+test_teardown()
+{
+    say "Test done $(date)"
+
+    say "Running test teardown."
+    run "rm -f $FINIT_CONF"
+}
+
+# shellcheck source=/dev/null
+. "$TEST_DIR/lib/setup.sh"
+
+sep "/etc/env:"
+run "cat /etc/env"
+run "echo 'service env:/etc/env serv -np -e xyzzy:bar -- serv checks xyzzy=bar' > $FINIT_CONF"
+
+say 'Reload Finit'
+run "initctl reload"
+
+retry 'assert_num_children 1 serv'
+
+say "Done, drop service from $FINIT_CONF ..."
+run "rm $FINIT_CONF"
+run "initctl reload"

test/sysvparts.sh

@@ -37,6 +37,7 @@ while true; do
     sleep 1
 done
 
+sleep 1
 texec cat /tmp/sysv.log
 texec diff -u /usr/share/runparts/sysv.log /tmp/sysv.log
 texec cmp /usr/share/runparts/sysv.log /tmp/sysv.log || fail "runparts sysv in wrong order"

test/unexpected-restart.sh

@@ -0,0 +1,73 @@
+#!/bin/sh
+# A service (B) going into flux, 'initctl reload' with no changes to
+# activate, should do not cause ripple effects removing any service
+# conditions, e.g., <service/B/running>.  During reload all services
+# are paused while Finit checks for any service3 changes to activate,
+# which may mean stopping or restarting existing services. 
+#
+# Issue #382
+set -e
+
+#DEBUG=true
+TEST_DIR=$(dirname "$0")
+
+test_setup()
+{
+    run "cat >> $FINIT_CONF" <<EOF
+service log:stdout name:A serv -np -i A
+service log:stdout notify:systemd <pid/A>           name:B serv -np -i B -N 0         -- B needs A
+service log:stdout notify:systemd <service/B/ready> name:C serv -np -i C -N 0         -- C needs B(service)
+service log:stdout notify:systemd <pid/B>           name:D serv -np -i D -N 0         -- D needs B(pid)
+task <service/C/ready,service/D/ready>              name:allup initctl cond set allup -- Everything is up
+EOF
+    say "Test start $(date)"
+}
+
+test_teardown()
+{
+    say "Test done $(date)"
+}
+
+pidof()
+{
+    texec initctl -j status "$1" | jq .pid
+}
+
+# shellcheck source=/dev/null
+. "$TEST_DIR/lib/setup.sh"
+
+sep "$FINIT_CONF"
+run "cat $FINIT_CONF"
+sep
+run "initctl reload"
+run "initctl status"
+sep
+
+say "waiting for primary startup to complete"
+retry 'assert_status allup "done"' 100 1
+assert_status C "running"
+oldpid=$(pidof C)
+assert_status D "running"
+doldpid=$(pidof D)
+
+sep "pre-reload status"
+run "initctl status"
+sep
+
+say "Reload Finit, who gets restarted?"
+#run "initctl debug"
+run "initctl reload"
+sleep 2
+
+assert_status A "running"
+assert_status B "running"
+assert_status C "running"
+assert_status D "running"
+
+newpid=$(pidof C)
+# shellcheck disable=SC2086
+assert "C was not restarted" $oldpid -eq $newpid
+
+dnewpid=$(pidof D)
+# shellcheck disable=SC2086
+assert "D was not restarted" $doldpid -eq $dnewpid

tmpfiles.d/finit.conf

@@ -11,4 +11,4 @@ d  /run/finit/cond	0755 - - -
 d  /run/finit/cond/pid	0755 - - -
 d  /run/finit/cond/sys	0755 - - -
 d  /run/finit/cond/usr	0755 - - -
-
+d  /run/finit/system	0755 - - -