contrib: debian: add anacron and cron services

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

.travis.yml

@@ -5,9 +5,6 @@ language: c
 # Use docker for quicker builds, it now allows https://docs.travis-ci.com/user/apt/
 sudo: false
 
-# Use non-docker builds atm. to install libuEv and libite from source
-#sudo: true
-
 # Test build with both GCC and Clang (LLVM)
 compiler:
   - gcc
@@ -30,7 +27,7 @@ addons:
       name: "troglobit/finit"
       description: "finit | Fast & Extensible init for Linux"
     notification_email: troglobit@gmail.com
-    build_command_prepend: "./autogen.sh && ./configure --prefix=/tmp"
+    build_command_prepend: "./autogen.sh && ./configure --prefix= --disable-doc --disable-contrib --with-watchdog"
     build_command: "make"
     branch_pattern: dev
 
@@ -45,20 +42,20 @@ install:
 # Custom build script for the time being, no "make test" yet
 script:
   - ./autogen.sh
-  - ./configure --prefix=/tmp --enable-static
+  - ./configure --prefix= --enable-static
   - make clean
   - make V=1 -j5
-  - ./configure --prefix=/tmp --enable-x11-common-plugin --enable-fallback-shell --enable-watchdog --enable-logit
+  - ./configure --prefix= --enable-x11-common-plugin --with-watchdog
   - cat config.log
   - make clean
   - make V=1 -j5
-  - make install-strip
-  - tree /tmp
-  - ldd /tmp/sbin/finit
-  - size /tmp/sbin/finit
-  - size /tmp/sbin/initctl
-  - size /tmp/sbin/reboot
-  - LD_LIBRARY_PATH=/tmp/lib /tmp/sbin/initctl -h
+  - make install-strip DESTDIR=/tmp/finit
+  - tree /tmp/finit
+  - ldd /tmp/finit/sbin/finit
+  - size /tmp/finit/sbin/finit
+  - size /tmp/finit/sbin/initctl
+  - size /tmp/finit/sbin/reboot
+  - LD_LIBRARY_PATH=/tmp/lib /tmp/finit/sbin/initctl -h
   - unshare -V
   - nsenter -V
   - chroot --version

ChangeLog.md

@@ -4,12 +4,47 @@ Change Log
 All relevant changes are documented in this file.
 
 
-[4.0][UNRELEASED] - 2021-02-21
-------------------------------
+[4.1][UNRELEASED]
+-----------------
+
+Bug fix release.
+
+### Changes
+* Change behavior on SIGUSR1 to be compatible with sysvinit and systemd.
+  Previously SIGUSR1 caused Finit to halt, like BusyBox init.  This had
+  "interesting" side effects on Debian systems when coexisting with
+  sysvinit (upgrading/reinstalling causes scripts to `kill -USR1 1`)
+* Change how `contrib/debian/install.sh` sets up a Grub boot entry for
+  finit.  We now modify the $SUPPORTED_INITS variable in `10_linux`
+
+### Fixes
+* Stricter interface name validation in netlink plugin, modeled after
+  the kernel.  Suggested by Coverity Scan
+* The `contrib/*/install.sh` scripts failed to run from tarball
+* Fix #170: detect loss of default route when interfaces go down.  This
+  emulates the missing kernel netlink message to remove the condition
+  net/default/route to allow stopping dependent services
+* Fix #171: restore automatic mount of `/dev/shm`, `/dev/pts`, `/run`
+  and `/tmp`, unless mounted already by `/etc/fstab`.  This is what most
+  desktop systems expect PID 1 to do
+* Fix #173: netlink plugin runs out of socket buffer space;
+
+        finit[1]: nl_callback():recv(): No buffer space available
+
+  Fixed by adding support for resync with kernel on `ENOBUFS`.  See
+  netlink(7) for details.  As a spin-off the plugin now supports any
+  number of interfaces and routes on a system.  On resync, the following
+  message is now logged, as a warning:
+
+        finit[1]: nl_callback():busy system, resynchronizing with kernel.
+
+
+[4.0][] - 2021-04-26
+--------------------
 
 This release became v4.0, and not v3.2, because of incompatible changes
 to service conditions.  There are other significant changes as well, so
-make sure to read the whole changelog when upgrading.
+make sure to read the whole change log when upgrading.
 
 ### Changes
 * The stand-alone `reboot` tool has been replaced with a symlink to
@@ -18,31 +53,53 @@ make sure to read the whole changelog when upgrading.
   with a fallback to sending signals as per traditional SysV init.  The
   `-f` (force) flag remains, where `reboot(2)` is called directly
 * Introducing Finit progress 𝓜𝓸𝓭𝓮𝓻𝓷
+* The `inictl cond set|clear COND` have changed completely.  Constrained
+  to a flat `<usr/...>` namespace and automatically activated by a new
+  `usr.so` plugin that checks services for usr condition changes
 * Removed built-in inetd super server.  If you need this functionality,
   use an external inetd, like xinetd, instead.  A pull request for a
   stand-alone inetd, like watchdogd and getty, is most welcome!
-* Incompatible `configure` script changes, i.e., you must give proper
-  path arguments to the script, no more guessing just GNU defaults.
-  There are examples in the documentation and the `contrib/` section
-* Change service conditions from the non-obvious `<svc/path/to/foo>` to
-  `<pid/foo:id>`.  This to hopefully make it clear that one service's
-  'pid:!foo' pidfile is another service's `<pid/foo>` condition
-* Initial support for cgroups, every service now runs in its own cgroup
-  https://twitter.com/b0rk/status/1214341831049252870?s=20
+* Incompatible `configure` script changes, i.e., no guessing `--prefix`
+  and other paths.  Also, many options have been changed, renamed, or
+  flipped defaults, or even dropped altogether.  There are examples in
+  the documentation and the `contrib/` section
+* Service conditions change from the non-obvious `<svc/path/to/foo>` to
+  `<pid/foo:id>`.  Not only does this give simpler internal semantics,
+  it hopefully also makes it clear that one service's `pid:!foo` pidfile
+  is another service's `<pid/foo>` condition, issue #143
+* Initial support for cgroups v2:
+  - services runs in a cgroup named after their respective *.conf file
+  - top-level groups are; init, user, and system
+  - all top-level groups can be configured from finit *.conf files
+  - each service can tweak the cgroup settings
+  - Use `initctl [top|ps|cgroup]` commands to inspect runtime state
+  - https://twitter.com/b0rk/status/1214341831049252870?s=20
 * Major refactor of Finit's `main()` function to be able to start the
   event loop earlier.  This also facilitated factoring out functionality
   previously hard-coded in Finit, e.g., starting the bundled watchdogd,
   various distro packed udevd and other hotplugging tools
+* A proper rescue mode has been added.  It is started extremely early
+  and is protected with a bundled `suslogin`.  Exiting rescue mode now
+  brings up the system as a normal boot, as one expects
 * Support for `sysv` start/stop scripts as well as monitoring forking
   services, stared using `sysv` or `service` stanza
 * Support for custom `kill:DELAY`, default 3 sec.
 * Support for custom `halt:SIGNAL`, default SIGTERM
+* Support for `pre:script` and `post:script`, allows for setup and
+  teardown/cleanup before and after a service runs, issue #129
+* Support for `env:file` in `/etc/default/foo` or `/etc/conf.d/foo`, see
+  the contrib section for examples that utilize this feature.  Variables
+  expanded from env files, and the env files themselves, are tracked for
+  changes to see if a service .conf file is "dirty" and needs restart on
+  `initctl reload`
 * Support for tracking custom PID files, using `pid:!/path/to/foo.pid`,
   useful with new `sysv` or `service` which fork to background
 * Support starting run/task/services without absolute path, trust `$PATH`
 * Add support for `--disable-doc` and `--disable-contrib` to speed up
   builds and work around issue with massively parallel builds
-* Add support for `@console` also for external getty
+* Support for `@console` also for external getty
+* Support for `notty` option to built-in getty, for board bring-up
+* Support for `rescue` option to built-in getty, for rescue shells
 * Add `-b`, batch mode, for non-interactive use to `initctl`
 * Prefer udev to handle `/dev/` if mdev is also available
 * Redirect dbus daemon output to syslog
@@ -92,6 +149,8 @@ make sure to read the whole changelog when upgrading.
 * Always append `:ID` qualifier to conditions if set for a service
 * The IPC socket has moved from `/run/finit.sock` to `/run/finit/socket`
   officially only supported for use by the `initctl` tool
+* The IPC socket now uses `SOCK_SEQPACKET` instead of `SOCK_STREAM`.
+  Recommend using watchdogd v3.4, or later, which support this
 * Improved support for modern `/etc/network/interfaces`, which has
   include statements.  No more native `ifup` of individual interfaces,
   Finit now calls `ifup -a`, or `ifdown -a`, delegating all details to
@@ -109,16 +168,21 @@ make sure to read the whole changelog when upgrading.
   that do not have a watchdog
 * Fix #100: Early condition handling may not work if `/var/run` does
   not yet exist (symlink to `/run`).  Added compat layer for access
+* Fix #101: Built-in inetd removed
+* Fix #102: Start built-in watchdogd as a regular service
 * Fix #103: Register multiple getty if `@console` resolves to >1 TTY,
 * Fix #105: Only remove /etc/nologin when moving from runlevel 0, 1, 6
   Fixed by Jonas Johansson, Westermo
 * Fix #109: Support for PID files in sub-directories to `/var/run`
 * Handle rename of PID files, by Robert Andersson, Atlas Copco
+* Fix #110: automatic modprobe of RTC devices, built-in hwclock
 * Fix #120: Redirect `stdin` to `/dev/null` for services by default
 * Fix #122: Switch to `nanosleep()` to achieve "signal safe" sleep,
   fixed by Jacques de Laval, Westermo
 * Fix #124: Lingering processes in process group when session leader
   exits.  E.g., lingering `logit` processes when parent dies
+* Fix #127: Show all runparts scripts as they start, like rc.local, fixed
+  by Jacques de Laval, Westermo
 * Fix service name matching, e.g. for condition handling, may match with
   wrong service, by Jonas Holmberg, Westermo
 * Run all run-parts scripts using `/bin/sh -c foo` just like the standard
@@ -137,6 +201,8 @@ make sure to read the whole changelog when upgrading.
 * Provide service description for built-in watchdog daemon
 * Fix #138: Handle `SIGPWR` like `SIGSUR2`, i.e., power off the system
 * Drop the '%m' GNUism, for compat with older musl libc
+* Fix #139: call `tzset()` on `initctl reload` to activate system
+  timezone changes (for logging)
 
 
 [3.1][] - 2018-01-23
@@ -745,14 +811,14 @@ Major bug fix release.
 * Makefile fixes for installation, paths encoded wrong
 * Strip binaries + .so files, support for `$(CROSS)` toolchain strip
 * Default install is now to `/sbin/finit` and `/usr/`
-* Note change in `$PLUGIN_DIR` environemnt variable to `$plugindir`
+* Note change in `$PLUGIN_DIR` environment variable to `$plugindir`
 
 
 [1.2][] - 2012-09-27
 --------------------
 
 ### Changes
-* Update README with section on building and enviroment variables
+* Update README with section on building and environment variables
 
 ### Fixes
 * Fix installation paths encoded in finit binary
@@ -844,7 +910,7 @@ Major bug fix release.
 
 * Initial release
 
-[UNRELEASED]: https://github.com/troglobit/finit/compare/3.1...HEAD
+[UNRELEASED]: https://github.com/troglobit/finit/compare/4.0...HEAD
 [4.0]: https://github.com/troglobit/finit/compare/3.1...4.0
 [3.1]: https://github.com/troglobit/finit/compare/3.0...3.1
 [3.0]: https://github.com/troglobit/finit/compare/2.4...3.0
@@ -876,9 +942,3 @@ Major bug fix release.
 [0.2]: https://github.com/troglobit/finit/compare/0.1...0.2
 [libuEv]: https://github.com/troglobit/libuev
 [dea3ae8]: https://github.com/troglobit/finit/commit/dea3ae8
-
-<!--
-  -- Local Variables:
-  -- mode: markdown
-  -- End:
-  -->

Makefile.am

@@ -29,19 +29,3 @@ release: distcheck
 		printf "$$file.md5\t";    cat ../$$file.md5    | cut -f1 -d' ';	\
 		printf "$$file.sha256\t"; cat ../$$file.sha256 | cut -f1 -d' ';	\
 	done
-
-run:
-	srcdir=@srcdir@/test \
-	top_builddir=@top_builddir@ \
-	TENV_ROOT=$(tenv_root) \
-		@srcdir@/test/setup-root.sh
-
-	mkdir -p $(dir $(tenv_root)/@FINIT_CONF@)
-	echo 'tty [12345] @console linux noclear nologin' > $(tenv_root)/@FINIT_CONF@
-
-	-TENV_ROOT=$(tenv_root) \
-	 TESTENV_PATH=@sbindir@:/bin \
-	 FINIT_RCSD=@FINIT_RCSD@ \
-		test/tenv/start.sh finit
-
-run : tenv_root ?= @abs_top_builddir@/run-rootfs

README.md

@@ -26,6 +26,26 @@ Finit is a simple alternative to [SysV init][4] and [systemd][6].  It
 was reverse engineered from the [EeePC fastinit][] over ten years ago
 by [Claudio Matsuoka][] — "gaps filled with frog DNA …"
 
+![Alpine screenshot](img/finit4-screenshot.png)  
+*Figure 1: Screenshot showing Finit booting [Alpine Linux](https://www.alpinelinux.org/).*
+
+Features include:
+
+  * Runlevels
+  * One-shot tasks, daemons, or SysV init scripts
+  * Runparts and rc.local
+  * Process supervision
+  * Sourcing environment files
+  * Conditions for network/process/custom dependencies
+  * Pre/Post script actions
+  * Tooling to enable/disable services
+  * Built-in getty
+  * Built-in watchdog, with support for hand-over to [watchdogd](https://troglobit.com/watchdogd.html)
+  * Built-in support for Debian/BusyBox `/etc/network/interfaces`
+  * Cgroups v2, both configuration and monitoring in `initctl top`
+  * Plugin support for customization
+  * Proper rescue mode with bundled `sulogin` for protected maintenance shell
+
 Focus is on small and embedded systems, although Finit is fully usable
 on server and desktop systems as well.  For working examples, see the
 [contrib/](contrib/) section with tutorials for:
@@ -34,9 +54,6 @@ on server and desktop systems as well.  For working examples, see the
   * [Alpine Linux](contrib/alpine/), and
   * [Debian GNU/Linux](contrib/debian/)
 
-![Alpine screenshot](img/alpine-screenshot.png)  
-The screenshot shows [Alpine Linux](https://www.alpinelinux.org/).
-
 
 Example
 -------
@@ -58,10 +75,10 @@ runlevel 2
 log size=100k count=4
 
 # Services to be monitored and respawned as needed
-service [S12345] watchdogd -L -f                 -- System watchdog daemon
-service [S12345] syslogd -n -b 3 -D              -- System log daemon
-service [S12345] <pid/syslogd> klogd -n          -- Kernel log daemon
-service   [2345] lldpd -d -c -M1 -H0 -i          -- LLDP daemon (IEEE 802.1ab)
+service [S12345] env:-/etc/conf.d/watchdog watchdog $WATCHDOG_OPTS $WATCHDOG_DEV -- System watchdog daemon
+service [S12345] env:-/etc/conf.d/syslog syslogd -n $SYSLOGD_OPTS          -- System log daemon
+service [S12345] <pid/syslogd> env:-/etc/conf.d/klogd klogd -n $KLOGD_OPTS -- Kernel log daemon
+service   [2345] env:-/etc/conf.d/lldpd lldpd -d $LLDPD_OPTS               -- LLDP daemon (IEEE 802.1ab)
 
 # The BusyBox ntpd does not use syslog when running in the foreground
 # So we use this trick to redirect stdout/stderr to a log file.  The
@@ -76,24 +93,24 @@ service :80   [2345] merecat -n -p 80   /var/www -- Web server
 service :8080 [2345] merecat -n -p 8080 /var/www -- Old web server
 
 # Alternative method instead of below runparts, can also use /etc/rc.local
-#task [S] /etc/init.d/keyboard-setup start       -- Setting up preliminary keymap
-#task [S] /etc/init.d/acpid start                -- Starting ACPI Daemon
-#task [S] /etc/init.d/kbd start                  -- Preparing console
+#sysv [S] /etc/init.d/keyboard-setup       -- Setting up preliminary keymap
+#sysv [S] /etc/init.d/acpid                -- Starting ACPI Daemon
+#task [S] /etc/init.d/kbd                  -- Preparing console
 
 # Run start scripts from this directory
 # runparts /etc/start.d
 
 # Virtual consoles run BusyBox getty, keep kernel default speed
-tty [12345] /sbin/getty -L 0 /dev/tty1 linux
-tty [12345] /sbin/getty -L 0 /dev/tty2 linux
-tty [12345] /sbin/getty -L 0 /dev/tty3 linux
+tty [12345] /sbin/getty -L 0 /dev/tty1  linux nowait noclear
+tty [2345]  /sbin/getty -L 0 /dev/tty2  linux nowait noclear
+tty [2345]  /sbin/getty -L 0 /dev/tty3  linux nowait noclear
 
 # Use built-in getty for serial port and USB serial
-tty [12345] /dev/ttyAMA0 noclear nowait
-tty [12345] /dev/ttyUSB0 noclear
+#tty [12345] /dev/ttyAMA0 noclear nowait
+#tty [12345] /dev/ttyUSB0 noclear
 
 # Just give me a shell, I need to debug this embedded system!
-tty [12345] @console noclear nologin
+#tty [12345] console noclear nologin
 ```
 
 The `service` stanza, as well as `task`, `run` and others are described
@@ -117,7 +134,7 @@ Some components are optional: runlevel(s), condition(s) and description,
 making it easy to create simple start scripts and still possible for more
 advanced uses as well:
 
-    service sshd -D
+    service /usr/sbin/sshd -D
 
 Dependencies are handled using [conditions](doc/conditions.md).  One of
 the most common conditions is to wait for basic networking to become
@@ -232,8 +249,11 @@ For more information, see [doc/plugins.md](doc/plugins.md).
 By default, Finit monitors `/etc/finit.d/` and `/etc/finit.d/enabled/`
 registering any changes to `.conf` files.  To activate a change the user
 must call `initctl reload`, which reloads all modified files, stops any
-removed services, starts new ones, and restarts any modified ones, with
-SIGHUP if the process supports it.
+removed services, starts new ones, and restarts any modified ones.  If the
+command line arguments of a service have changed, the process will be
+terminated and then started again with the updated arguments. If the arguments
+have not been modified and the process supports SIGHUP, the process will
+receive a SIGHUP rather than being terminated and started.
 
 For some use-cases the extra step of calling `initctl reload` creates an
 unnecessary overhead, which can be removed at build-time using:
@@ -241,6 +261,46 @@ unnecessary overhead, which can be removed at build-time using:
     configure --enable-auto-reload
 
 
+**Cgroups**
+
+Finit supports cgroups v2 and comes with the following default groups
+in which services and user sessions are placed in:
+
+     /sys/fs/cgroup
+       |-- init/               # cpu.weight:100
+       |-- system/             # cpu.weight:9800
+       `-- user/               # cpu.weight:100
+
+Finit itself and its helper scripts and services are placed in the
+top-level leaf-node group `init/`, which also is _reserved_.
+
+All run/task/service/sysv processes are placed in their own sub-group
+in `system/`.  The name of each sub-group is taken from the respective
+`.conf` file from `/etc/finit.d`.
+
+All getty/tty processes are placed in their own sub-group in `user/`.
+The name of each sub-group is taken from the username.
+
+A fourth group also exists, the `root` group.  It is also _reserved_ and
+primarily intended for RT tasks.  If you have RT tasks they need to be
+declared as such in their service stanza like this:
+
+    service [...] <...> cgroup.root /path/to/foo args -- description
+
+or
+
+    cgroup.root
+    service [...] <...> /path/to/foo args -- description
+    service [...] <...> /path/to/bar args -- description
+
+See [doc/config.md](doc/config.md#cgroups) for more information, e.g.,
+how to configure per-group limits.
+
+The `initctl` tool has three commands to help debug and optimize the
+setup and monitoring of cgroups.  See the `ps`, `top`, and `cgroup`
+commands for details.
+
+
 Runparts & /etc/rc.local
 ------------------------
 
@@ -348,11 +408,18 @@ Finit also implements a modern API to query status, and start/stop
 services, called `initctl`.  Unlike `telinit` the `initctl` tool does
 not return until the given command has fully completed.
 
-```shell
-~ $ initctl -h
+```
+alpine:~# initctl help
 Usage: initctl [OPTIONS] [COMMAND]
 
 Options:
+  -b, --batch               Batch mode, no screen size probing
+  -c, --create              Create missing paths (and files) as needed
+  -f, --force               Ignore missing files and arguments, never prompt
+  -1, --once                Only one lap in commands like 'top'
+  -p, --plain               Use plain table headings, no ctrl chars
+  -q, --quiet               Silent, only return status of command
+  -t, --no-heading          Skip table headings
   -v, --verbose             Verbose output
   -h, --help                This help text
 
@@ -360,29 +427,39 @@ Commands:
   debug                     Toggle Finit (daemon) debug
   help                      This help text
   version                   Show Finit version
-  
-  list                      List all .conf in /etc/finit.d/
-  enable   <CONF>           Enable   .conf in /etc/finit.d/available/
-  disable  <CONF>           Disable  .conf in /etc/finit.d/[enabled/]
-  touch    <CONF>           Mark     .conf in /etc/finit.d/ for reload
-  reload                    Reload  *.conf in /etc/finit.d/ (activates changes)
-  
-  cond     show             Show condition status
+
+  ls | list                 List all .conf in /etc/finit.d
+  create   <CONF>           Create   .conf in /etc/finit.d/available
+  delete   <CONF>           Delete   .conf in /etc/finit.d/available
+  show     <CONF>           Show     .conf in /etc/finit.d/available
+  edit     <CONF>           Edit     .conf in /etc/finit.d/available
+  touch    <CONF>           Change   .conf in /etc/finit.d/available
+  enable   <CONF>           Enable   .conf in /etc/finit.d/available
+  disable  <CONF>           Disable  .conf in /etc/finit.d/enabled
+  reload                    Reload  *.conf in /etc/finit.d (activate changes)
+
+  cond     set   <COND>     Set (assert) user-defined condition     +usr/COND
+  cond     clear <COND>     Clear (deassert) user-defined condition -usr/COND
+  cond     status           Show condition status, default cond command
   cond     dump             Dump all conditions and their status
-  
+
   log      [NAME]           Show ten last Finit, or NAME, messages from syslog
-  start    <JOB|NAME>[:ID]  Start service by job# or name, with optional ID
-  stop     <JOB|NAME>[:ID]  Stop/Pause a running service by job# or name
-  restart  <JOB|NAME>[:ID]  Restart (stop/start) service by job# or name
-  status   <JOB|NAME>[:ID]  Show service status, by job# or name
-  status | show             Show status of services, default command
-  
+  start    <NAME>[:ID]      Start service by name, with optional ID
+  stop     <NAME>[:ID]      Stop/Pause a running service by name
+  reload   <NAME>[:ID]      Reload service by name (SIGHUP or restart)
+  restart  <NAME>[:ID]      Restart (stop/start) service by name
+  status   <NAME>[:ID]      Show service status, by name
+  status                    Show status of services, default command
+
+  cgroup                    List cgroup config overview
+  ps                        List processes based on cgroups
+  top                       Show top-like listing based on cgroups
+
   runlevel [0-9]            Show or set runlevel: 0 halt, 6 reboot
   reboot                    Reboot system
   halt                      Halt system
   poweroff                  Halt and power off system
-  
-  utmp     show             Raw dump of UTMP/WTMP db
+  suspend                   Suspend system
 ```
 
 For services *not* supporting `SIGHUP` the `<!>` notation in the .conf
@@ -390,22 +467,73 @@ file must be used to tell Finit to stop and start it on `reload` and
 `runlevel` changes.  If `<>` holds more [conditions](doc/conditions.md),
 these will also affect how a service is maintained.
 
-**Note:** even though it is possible to start services not belonging in
-the current runlevel these services will not be respawned automatically
-by Finit if they exit (crash).  Hence, if the runlevel is 2, the below
-Dropbear SSH service will not be restarted if it is killed or exits.
+> **Note:** even though it is possible to start services not belonging in
+> the current runlevel these services will not be respawned automatically
+> by Finit if they exit (crash).  Hence, if the runlevel is 2, the below
+> Dropbear SSH service will not be restarted if it is killed or exits.
 
-```shell
-~ $ initctl status -v
-1       running  476     [S12345]   /sbin/watchdog -T 16 -t 2 -F /dev/watchdog
-2       running  477     [S12345]   /sbin/syslogd -n -b 3 -D
-3       running  478     [S12345]   /sbin/klogd -n
-4:1       inetd  0       [2345]     internal time allow *:37,!eth0
-4:2       inetd  0       [2345]     internal time allow eth0:3737
-5:1       inetd  0       [2345]     /sbin/telnetd allow *:23 deny eth0,eth1
-5:2       inetd  0       [2345]     /sbin/telnetd allow eth0:2323,eth2:2323,eth1:2323
-6:1       inetd  0       [345]      /sbin/dropbear allow eth0:222
-6:2       inetd  0       [345]      /sbin/dropbear allow *:22 deny eth0
+The `status` command is the default, it displays a quick overview of all
+monitored run/task/services.  Here we call `initctl -p`, suitable for
+scripting and documentation:
+
+```
+alpine:~# initctl -p
+PID   IDENT     STATUS   RUNLEVELS    DESCRIPTION
+======================================================================
+1506  acpid     running  [--2345----] ACPI daemon
+1509  crond     running  [--2345----] Cron daemon
+1489  dropbear  running  [--2345----] Dropbear SSH daemon
+1511  klogd     running  [S12345----] Kernel log daemon
+1512  ntpd      running  [--2345----] NTP daemon
+1473  syslogd   running  [S12345----] Syslog daemon
+
+alpine:~# initctl -pv
+PID   IDENT     STATUS   RUNLEVELS    COMMAND
+======================================================================
+1506  acpid     running  [--2345----] acpid -f
+1509  crond     running  [--2345----] crond -f -S $CRON_OPTS
+1489  dropbear  running  [--2345----] dropbear -R -F $DROPBEAR_OPTS
+1511  klogd     running  [S12345----] klogd -n $KLOGD_OPTS
+1512  ntpd      running  [--2345----] ntpd -n $NTPD_OPTS
+1473  syslogd   running  [S12345----] syslogd -n
+```
+
+The environment variables to each of the services above are read from,
+in the case of Alpine Linux, `/etc/conf.d/`.  Other distributions may
+have other directories, e.g., Debian use `/etc/default/`.
+
+The `status` command takes an optional `NAME:ID` argument.  Here we
+check the status of `dropbear`, which only has one instance in this
+system:
+
+```
+alpine:~# initctl -p status dropbear
+     Status : running
+   Identity : dropbear
+Description : Dropbear SSH daemon
+     Origin : /etc/finit.d/enabled/dropbear.conf
+Environment : -/etc/conf.d/dropbear
+Condition(s):
+    Command : dropbear -R -F $DROPBEAR_OPTS
+   PID file : !/run/dropbear.pid
+        PID : 1485
+       User : root
+      Group : root
+     Uptime : 2 hour 46 min 56 sec
+  Runlevels : [--2345----]
+     Memory : 1.2M
+     CGroup : /system/dropbear cpu 0 [100, max] mem [--.--, max]
+              |- 1485 dropbear -R -F
+              |- 2634 dropbear -R -F
+              |- 2635 ash
+              `- 2652 initctl -p status dropbear
+
+Apr  8 12:19:49 alpine authpriv.info dropbear[1485]: Not backgrounding
+Apr  8 12:37:45 alpine authpriv.info dropbear[2300]: Child connection from 192.168.121.1:47834
+Apr  8 12:37:46 alpine authpriv.notice dropbear[2300]: Password auth succeeded for 'root' from 192.168.121.1:47834
+Apr  8 12:37:46 alpine authpriv.info dropbear[2300]: Exit (root) from <192.168.121.1:47834>: Disconnect received
+Apr  8 15:02:11 alpine authpriv.info dropbear[2634]: Child connection from 192.168.121.1:48576
+Apr  8 15:02:12 alpine authpriv.notice dropbear[2634]: Password auth succeeded for 'root' from 192.168.121.1:48576
 ```
 
 
@@ -473,6 +601,6 @@ and proposed extensions.
 [License]:          https://en.wikipedia.org/wiki/MIT_License
 [License Badge]:    https://img.shields.io/badge/License-MIT-teal.svg
 [Travis]:           https://travis-ci.org/troglobit/finit
-[Travis Status]:    https://travis-ci.org/troglobit/finit.png?branch=master
+[Travis Status]:    https://travis-ci.org/troglobit/finit.svg?branch=master
 [Coverity Scan]:    https://scan.coverity.com/projects/3545
 [Coverity Status]:  https://scan.coverity.com/projects/3545/badge.svg

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT(Finit, 4.0-rc1, https://github.com/troglobit/finit/issues, finit, https://troglobit.com/finit.html)
+AC_INIT(Finit, 4.1-rc1, https://github.com/troglobit/finit/issues, finit, https://troglobit.com/finit.html)
 AC_CONFIG_AUX_DIR(aux)
 AM_INIT_AUTOMAKE([1.11 foreign dist-xz subdir-objects])
 AM_SILENT_RULES([yes])
@@ -8,7 +8,7 @@ AC_CONFIG_HEADER([config.h])
 AC_CONFIG_FILES([Makefile
 	contrib/Makefile
 	contrib/alpine/Makefile contrib/alpine/finit.d/Makefile contrib/alpine/finit.d/available/Makefile
-	contrib/debian/Makefile contrib/debian/finit.d/Makefile contrib/debian/finit.d/available/Makefile contrib/debian/grub.d/Makefile
+	contrib/debian/Makefile contrib/debian/finit.d/Makefile contrib/debian/finit.d/available/Makefile
 	contrib/void/Makefile   contrib/void/finit.d/Makefile   contrib/void/finit.d/available/Makefile
 	doc/Makefile
 	plugins/Makefile
@@ -32,7 +32,7 @@ AC_PROG_MKDIR_P
 
 # Configuration.
 AC_HEADER_STDC
-AC_CHECK_HEADERS([fstaby.h termios.h])
+AC_CHECK_HEADERS([fstaby.h termios.h sys/ioctl.h])
 AC_CHECK_FUNCS([strstr getopt getfsenty])
 
 # Check for uint[8,16,32]_t
@@ -48,30 +48,13 @@ PKG_CHECK_MODULES([uev],  [libuev >= 2.2.0])
 PKG_CHECK_MODULES([lite], [libite >= 2.2.0])
 
 # Check for configured Finit features
-AC_ARG_ENABLE(emergency_shell,
-	AS_HELP_STRING([--enable-emergency-shell], [Experimental emergency fallback at boot.
-	                Use only for debugging very early boot problems with Finit]),,[
-	enable_emergency_shell=no])
-
-AC_ARG_ENABLE(fallback_shell,
-        AS_HELP_STRING([--enable-fallback-shell], [Start a shell if no TTYs can start]),,[
-	enable_fallback_shell=no])
-
 AC_ARG_ENABLE(auto_reload,
         AS_HELP_STRING([--enable-auto-reload], [Automatic reload on any .conf change]),,[
 	enable_auto_reload=no])
 
-AC_ARG_ENABLE(watchdog,
-        AS_HELP_STRING([--enable-watchdog], [Enable built-in watchdog, uses /dev/watchdog]),,[
-	enable_watchdog=no])
-
 AC_ARG_ENABLE(redirect,
-        AS_HELP_STRING([--enable-redirect], [Redirect service output to /dev/null, default: no]),,[
-	enable_redirect_output=no])
-
-AC_ARG_ENABLE(logit,
-     AS_HELP_STRING([--disable-logit], [Disable logit, used for log redirection, default: enable]),,[
-     enable_logit=yes])
+        AS_HELP_STRING([--disable-redirect], [Disable redirection of service output to /dev/null]),,[
+	enable_redirect=yes])
 
 AC_ARG_ENABLE(logrotate,
         AS_HELP_STRING([--disable-logrotate], [Disable built-in rotation of /var/log/wtmp, default enabled]),,[
@@ -128,31 +111,25 @@ AC_ARG_WITH(random-seed,
         AS_HELP_STRING([--with-random-seed=FILE], [Save a random seed for /dev/urandom across reboots, default /var/lib/misc/random-seed]),
 	[random_seed=$withval], [random_seed=yes])
 
+AC_ARG_WITH(watchdog,
+        AS_HELP_STRING([--with-watchdog=[DEV]], [Enable built-in watchdog, default: /dev/watchdog]),
+	[watchdog=$withval], [with_watchdog=no watchdog=])
+
 ### Enable features ###########################################################################
 
-# Create config.h from selected features and fallback defautls
-AS_IF([test "x$enable_emergency_shell" = "xyes"], [
-        AC_DEFINE(EMERGENCY_SHELL,  1, [Fork off init and let PID 1 wait and start /bin/sh if init crashes.])])
-
-AS_IF([test "x$enable_fallback_shell" = "xyes"], [
-        AC_DEFINE(FALLBACK_SHELL,  1, [Start /bin/sh on console if no TTYs.])])
-
+# Create config.h from selected features and fallback defaults
 AS_IF([test "x$enable_auto_reload" = "xyes"], [
         AC_DEFINE(AUTO_RELOAD, 1, [Finit automatically does `initctl reload` on any .conf change])])
 
-AS_IF([test "x$enable_watchdog" = "xyes"], [
-        AC_DEFINE(BUILTIN_WATCHDOG,  1, [Enable built-in watchdog, kicks on /dev/watchdog])])
-
 AS_IF([test "x$enable_redirect" = "xyes"], [
 	AC_DEFINE(REDIRECT_OUTPUT, 1, [Enable redirection of service output to /dev/null])])
 
 ### Disable features ###########################################################################
-AS_IF([test "x$enable_logit" != "xno"], [
-	AC_DEFINE(LOGIT_ENABLED,  1, [Enable logit tool, used for log redirection by Finit])])
-
 AS_IF([test "x$enable_logrotate" != "xno"], [
 	AC_DEFINE(LOGROTATE_ENABLED, 1, [Enable built-in rotation of /var/log/wtmp et al.])])
 
+AM_CONDITIONAL(LOGROTATE, [test "x$enable_logrotate" = "xyes"])
+
 ### With features ##############################################################################
 AS_IF([test "x$with_config" != "xno"], [
 	AS_IF([test "x$conf" = "xyes"], [
@@ -202,10 +179,16 @@ AS_IF([test "x$with_random_seed" != "xno"], [
 	AC_EXPAND_DIR(random_path, "$random_seed")
 	AC_DEFINE_UNQUOTED(RANDOMSEED, "$random_path", [Improve random at boot by seeding it with sth from before.])])
 
+AS_IF([test "x$with_watchdog" != "xno"], [
+	AS_IF([test "x$watchdog" = "xyes"], [
+		watchdog=/dev/watchdog])
+	with_watchdog=yes
+        AC_DEFINE_UNQUOTED(WDT_DEVNODE, "$watchdog", [Built-in watchdog device node])], [
+	watchdog=])
+
 # Control build with automake flags
 AM_CONDITIONAL(STATIC,    [test "x$enable_static" = "xyes"])
-AM_CONDITIONAL(WATCHDOGD, [test "x$enable_watchdog" = "xyes"])
-AM_CONDITIONAL(LOGIT,     [test "x$enable_logit" = "xyes"])
+AM_CONDITIONAL(WATCHDOGD, [test "x$with_watchdog" != "xno"])
 AM_CONDITIONAL(DOC,       [test "x$enable_doc" = "xyes"])
 AM_CONDITIONAL(CONTRIB,   [test "x$enable_contrib" = "xyes"])
 
@@ -224,9 +207,6 @@ AC_DEFINE_UNQUOTED(RESCUE_CONF, "$rescue_conf", [Finit rescue mode .conf])
 AC_EXPAND_DIR(sample_conf, "$libdir/finit/sample.conf")
 AC_DEFINE_UNQUOTED(SAMPLE_CONF, "$sample_conf", [Finit sameple.conf for initctl create])
 
-AC_EXPAND_DIR(path_stdpath, "$sbindir:$bindir:/usr/sbin:/usr/bin:/sbin:/bin")
-AC_DEFINE_UNQUOTED(_PATH_STDPATH, "$path_stdpath", [Default PATH])
-
 # Generate all files
 AC_OUTPUT
 
@@ -235,7 +215,7 @@ cat <<EOF
 ------------------ Summary ------------------
  $PACKAGE_NAME version $PACKAGE_VERSION
   Prefix................: $prefix
-  Exec prefix...........: `eval echo $exec_prefix`
+  Exec prefix...........: $eprefix
   Sysconfdir............: `eval echo $sysconfdir`
   Localstatedir.........: `eval echo $localstatedir`
   Finit config file.....: $conf_path
@@ -254,11 +234,9 @@ Behavior:
 Optional features:
   Install doc/..........: $enable_doc
   Install contrib/......: $enable_contrib
-  Built-in watchdogd....: $enable_watchdog
+  Built-in watchdogd....: $with_watchdog $watchdog
   Built-in logrotate....: $enable_logrotate
-  Scripting tool logit..: $enable_logit
-  Emergency shell.......: $enable_emergency_shell
-  Fallback shell........: $enable_fallback_shell
+  Redirect output.......: $enable_redirect
   Default hostname......: $hostname
   Default runlevel......: $runlevel
 

contrib/alpine/build.sh

@@ -17,7 +17,7 @@ PKG_CONFIG_LIBDIR=/usr/lib/pkgconfig:/usr/local/lib/pkgconfig ./configure	\
 		 --prefix=/usr			--exec-prefix=			\
 		 --sysconfdir=/etc		--localstatedir=/var		\
 		 --enable-dbus-plugin		--enable-x11-common-plugin	\
-		 --enable-alsa-utils-plugin	--enable-watchdog		\
+		 --enable-alsa-utils-plugin	--with-watchdog			\
 		 --with-heading="Alpine Linux 3.13"
 
 echo

contrib/alpine/finit.d/available/Makefile.am

@@ -1,5 +1,4 @@
 docsdir   := @docdir@/contrib/alpine/finit.d/available
 docs_DATA  = acpid.conf crond.conf dropbear.conf getty.conf keymap.conf	\
-             lxdm.conf modules.conf ntpd.conf klogd.conf syslog.conf	\
-	     watchdog.conf
+             lxdm.conf modules.conf ntpd.conf syslog.conf watchdog.conf
 EXTRA_DIST = $(docs_DATA)

contrib/alpine/finit.d/available/getty.conf

@@ -1,4 +1,5 @@
 # Consoles to start getty on when system is up
-tty [12345] /dev/tty1 noclear nowait
-tty [2345]  /dev/tty2 noclear nowait
-tty [2345]  /dev/tty3 noclear nowait
+tty [12345] /dev/ttyS0 noclear
+tty [12345] /dev/tty1  noclear nowait
+tty [2345]  /dev/tty2  noclear nowait
+tty [2345]  /dev/tty3  noclear nowait

contrib/alpine/finit.d/available/klogd.conf

@@ -1,2 +0,0 @@
-# Start klogd as soon as possible, after syslogd, should always run
-service [S12345] <pid/syslogd> env:-/etc/conf.d/klogd klogd -n $KLOGD_OPTS -- Kernel log daemon

contrib/alpine/finit.d/available/modules.conf

@@ -4,3 +4,4 @@
 #module evdev
 #module mousedev
 #module 8139cp
+module softdog

contrib/alpine/finit.d/available/syslog.conf

@@ -1,2 +1,4 @@
 # Start syslogd as soon as possible, should always run
-service [S12345] env:-/etc/conf.d/syslog syslogd -n $SYSLOGD_OPTS -- Syslog daemon
+# Start klogd as soon as possible, but after syslogd
+service [S12345] env:-/etc/conf.d/syslog syslogd -n $SYSLOGD_OPTS          -- Syslog daemon
+service [S12345] <pid/syslogd> env:-/etc/conf.d/klogd klogd -n $KLOGD_OPTS -- Kernel log daemon

contrib/alpine/install.sh

@@ -10,7 +10,7 @@ fi
 # Adjust base directory
 if [ -e alpine-release ]; then
     cd ../..
-elif [ ! -e autogen.sh ]; then
+elif [ ! -e configure.ac ]; then
     echo "*** Please run this script from the Finit base directory."
     exit 1
 fi

contrib/debian/Makefile.am

@@ -1,4 +1,4 @@
-SUBDIRS    = finit.d grub.d
+SUBDIRS    = finit.d
 docsdir   := @docdir@/contrib/debian
 docs_DATA  = README.md finit.conf os-release build.sh install.sh
 EXTRA_DIST = $(docs_DATA)

contrib/debian/README.md

@@ -20,29 +20,41 @@ tools pre-systemd work as intended.
 
     root@debian:~# apt install initscripts console-setup
 
+> **Note:** as of Debian 11 (Bullseye), libuev and libite are part of
+> the main section of Debian.  So just install the -dev packages :)
+
 The following script can then be used to configure, build, install and
 set up your system to run Finit:
 
     user@debian:~/finit$ contrib/debian/build.sh
 
 However, since `/sbin/init` already exists on your system, the script
-creates another entry in your GRUB config, in `/etc/grub.d/40_custom`,
-where `init=/sbin/finit` is added to your kernel line.  It is of course
-also possible to change the default init to Finit, if you do you can
-remove the custom Grub entry:
+creates another entry in your GRUB config by updating `$SUPPORTED_INITS`
+in `/etc/grub.d/10_inux` and run `update-grub`.  On reboot you will find
+a (finit) entry in the "Advanced options for Debian GNU/Linux" section.
+It is of course also possible to change the default init to Finit:
 
     user@debian:~/finit# cd /sbin
-    user@debian:/sbin# sudo rm init
+    user@debian:/sbin# sudo mv init oldinit
     user@debian:/sbin# sudo ln -s finit init
 
 Before rebooting, check the default [/etc/finit.conf](finit.conf) and
 `/etc/finit.d/*.conf` files.  The build + install script above provides
-a few sample `.conf` files. See `initctl list` after boot for a list of
-enabled and available services.
+a few sample `.conf` files.  See `initctl ls` after boot for a list of
+enabled and available services, you can then use `enable` and `disable`
+comands to `initctl`, followed by `reload` to activate your changes.
 
 You can also use a standard [/etc/rc.local](rc.local) for one-shot tasks
 and initialization like keyboard language etc.
 
+If you want to use the X Window system, `sudo apt install elogind`
+(Bullseye and later).  Followed by `initctl reload` (enabled by default)
+and logout/login again.  The elogind daemon ensures a regular non-root
+user can start and interact with an X session, otherwise keyboard and
+mouse won't work.  When you're happy, you can enable the lightdm.conf,
+change the default runlevel to 3, and presto! you have a desktop again.
+
+
 Have fun!  
  /Joachim ツ
 

contrib/debian/finit.d/Makefile.am

@@ -1,7 +1,9 @@
 SUBDIRS    = available
 docsdir   := @docdir@/contrib/debian/finit.d
-d_LINKS    = console-setup.conf getty.conf keyboard-setup.conf	\
-	     modules.conf rsyslogd.conf sshd.conf
+d_LINKS    = acpid.conf apparmor.conf avahi-demon.conf		\
+	     bluetooth.conf console-setup.conf elogind.conf 	\
+	     getty.conf keyboard-setup.conf modules.conf	\
+	     rsyslogd.conf sshd.conf
 
 install-exec-hook:
 	@$(MKDIR_P) $(DESTDIR)$(docsdir)/available

contrib/debian/finit.d/available/Makefile.am

@@ -1,5 +1,7 @@
 docsdir   := @docdir@/contrib/debian/finit.d/available
-docs_DATA  = acpid.conf console-setup.conf getty.conf keyboard-setup.conf \
-	     lightdm.conf modules.conf rsyslogd.conf sntpd.conf		  \
-	     sshd.conf syslogd.conf xdm.conf
+docs_DATA  = acpid.conf apparmor.conf avahi-daemon.conf bluetooth.conf	\
+	     console-setup.conf elogind.conf getty.conf			\
+	     keyboard-setup.conf lightdm.conf modules.conf sntpd.conf	\
+	     rsyslogd.conf sshd.conf syslogd.conf xdm.conf
 EXTRA_DIST = $(docs_DATA)
+

contrib/debian/finit.d/available/anacron.conf

@@ -0,0 +1 @@
+task [2345] <sys/pwr/ac,pid/syslogd> halt:usr1 env:-/etc/default/anacron anacron -d -q $ANACRON_ARGS -- Anacron jobs

contrib/debian/finit.d/available/apparmor.conf

@@ -0,0 +1,2 @@
+sysv [S] /etc/init.d/apparmor -- AppArmor initialization
+

contrib/debian/finit.d/available/avahi-daemon.conf

@@ -0,0 +1 @@
+service [2345] env:-/etc/default/avahi-daemon avahi-daemon -s $AVAHI_DAEMON_OPTS -- Avahi mDNS/DNS-SD Stack

contrib/debian/finit.d/available/bluetooth.conf

@@ -0,0 +1,3 @@
+# To ignore errors like "org.freedesktop.hostname1", update the env file with
+# "--noplugin=hostname", i.e. append to BLUETOOTHD_OPTS
+service [2345] log:null env:-/etc/default/bluetooth /usr/libexec/bluetooth/bluetoothd -n $BLUETOOTHD_OPTS -- Bluetooth service

contrib/debian/finit.d/available/cron.conf

@@ -0,0 +1 @@
+service [2345] <sys/pwr/ac,pid/syslogd> env:-/etc/default/cron cron -f $EXTRA_OPTS -- Background job processing daemon

contrib/debian/finit.d/available/elogind.conf

@@ -0,0 +1,4 @@
+# On some systems, bootstrap can be really quick and cause elogind
+# to not start before a user has managed to log in.  To prevent
+# this, add <pid/elogind> as condition to your TTYs in getty.conf
+service [2345] <pid/syslogd> /lib/elogind/elogind -- Login manager

contrib/debian/finit.d/available/getty.conf

@@ -1,4 +1,7 @@
 # Consoles to start getty on when system is up
 tty [12345] /dev/tty1 linux noclear nowait
-tty [2345]  /dev/tty2 linux noclear nowait
-tty [2345]  /dev/tty3 linux noclear nowait
+tty [2345]  /dev/tty2 linux nowait
+tty [2345]  /dev/tty3 linux nowait
+tty [2345]  /dev/tty4 linux nowait
+tty [2345]  /dev/tty5 linux nowait
+tty [2345]  /dev/tty6 linux nowait

contrib/debian/grub.d/40_custom

@@ -1,14 +0,0 @@
-#!/bin/sh
-exec tail -n +3 $0
-# This file provides an easy way to add custom menu entries.  Simply type the
-# menu entries you want to add after this comment.  Be careful not to change
-# the 'exec tail' line above.
-
-menuentry '$MENU with Finit' --class $TYPE --class gnu-linux --class gnu --class os --user {
-	insmod gzio
-	insmod part_msdos
-	insmod ext2
-	search --no-floppy --fs-uuid --set=root $BOOT
-	linux	/vmlinuz root=UUID=$ROOT ro quiet init=/sbin/finit
-	initrd	/initrd.img
-}

contrib/debian/grub.d/Makefile.am

@@ -1,3 +0,0 @@
-docsdir   := @docdir@/contrib/debian/grub.d
-docs_DATA  = 40_custom
-EXTRA_DIST = $(docs_DATA)

contrib/debian/install.sh

@@ -11,7 +11,7 @@ fi
 grep -q Debian os-release 2>/dev/null
 if [ $? -eq 0 ]; then
     cd ../..
-elif [ ! -e autogen.sh ]; then
+elif [ ! -e configure.ac ]; then
     echo "*** Please run this script from the Finit base directory."
     exit 1
 fi
@@ -23,7 +23,7 @@ echo "/sbin/finit           - PID 1"
 echo "/lib/finit/plugins/*  - All enabled Finit plugins"
 echo "/etc/finit.conf       - Finit configuration file"
 echo "/etc/finit.d/         - Finit services"
-echo "/etc/grub.d/40_custom - Add menu entry to the Grub boot loader"
+echo "/etc/grub.d/10_linux  - Add Finit to Grub's SUPPORTED_INITS"
 echo
 read -p "Do you want to continue (y/N)? " yorn
 echo
@@ -38,12 +38,18 @@ if [ "x$yorn" = "xy" -o "x$yorn" = "xY" ]; then
     cp -va finit.d /etc/
 
     echo "*** Setting up a GRUB boot entry ..."
-    MENU=`grep gnu-linux /boot/grub/grub.cfg |head -1 |sed "s/menuentry '\([^']*\).*/\1/"`
-    TYPE=`grep gnu-linux /boot/grub/grub.cfg |head -1 |sed "s/.*' --class \([a-z]*\).*/\1/"`
-    BOOT=`grep set=root /boot/grub/grub.cfg |head -1 |sed 's/.* \([a-f0-9]*-[a-f0-9]*-[a-f0-9]*-[a-f0-9]*-[a-f0-9]*\).*/\1/'`
-    ROOT=`grep root=UUID /boot/grub/grub.cfg |head -1 |sed 's/.*UUID=\([a-f0-9]*-[a-f0-9]*-[a-f0-9]*-[a-f0-9]*-[a-f0-9]*\).*/\1/'`
-    cat grub.d/40_custom | sed "s,\$BOOT,$BOOT,;s,\$ROOT,$ROOT,;s,\$TYPE,$TYPE,;s,\$MENU,$MENU," >/etc/grub.d/40_custom
-    update-grub
+    fn=/etc/grub.d/10_linux
+    if [ -e $fn ]; then
+	if `grep SUPPORTED_INITS $fn |head -1 |grep -q finit`; then
+	    echo "Already installed, done."
+	else
+	    echo "Adding Finit to list of SUPPORTED_INITS ..."
+	    sed -i 's/SUPPORTED_INITS="[^"]*/& finit:\/sbin\/finit/' $fn
+	    update-grub
+	fi
+    else
+	echo "Cannot find $fn, you'll have to set up your bootloader on your own."
+    fi
 
     echo
     echo "*** Done"

contrib/finit.conf

@@ -8,6 +8,18 @@ module evdev
 module loop
 module psmouse
 
+# Top-level cgroups and their default settings.  All groups mandatory
+# but more can be added, max 8 groups in total currently.  The cgroup
+# 'root' is also available, reserved for RT processes.  Settings are
+# as-is, only one shorthand 'mem.' exists, other than that it's the
+# cgroup v2 controller default names.
+cgroup init   cpu.weight:100
+cgroup user   cpu.weight:100
+cgroup system cpu.weight:9700 cpu.max:50000
+
+# Example extra cgroup
+cgroup maint  cpu.weight:100
+
 # Runlevel to start after bootstrap, runlevel 'S'
 runlevel 2
 
@@ -16,7 +28,7 @@ network service networking start
 
 # Max file size for each log file: 100 kiB, rotate max 4 copies:
 # log => log.1 => log.2.gz => log.3.gz => log.4.gz
-log size=100k count=4
+log size:100k count:4
 
 # Virtual consoles to start getty on
 tty /dev/tty1
@@ -30,9 +42,24 @@ tty /dev/tty3
 #run [2] /etc/init.d/networking start      -- Start networking
 
 # Services to be monitored and respawned as needed
+
+# klgod and syslogd are placed in the maint cgroup
+cgroup.maint
 service [2345] <pid/syslogd> klogd -n     -- Kernel logging server
 service [2345] syslogd -n                 -- Syslog server
-service [3]    gdm                        -- GNOME Display Manager
+
+# gdm in the system cgroup (default when read from it's own .conf file)
+cgroup.system
+service [3] cgroup:cpu.weight:250,mem.max:10M gdm -- GNOME Display Manager
+
+# Start SSH daemon with opts from /etc/default/ssh (if available), move
+# to user cgroup with cpu.weight:250.  Default for services is system
+# group.  Default weight in user group is 100 (above).  Max length of
+# cgroup argument is (currently) 63 chars.
+#
+# Notice alternative syntax where current cgroup is only set for this
+# particular service.
+service [2345] cgroup.user:cpu.weight:250,mem.max:10M env:-/etc/default/ssh /usr/sbin/sshd -D $SSHD_OPTS -- OpenSSH daemon
 
 # The BusyBox ntpd does not use syslog when running in the foreground
 # So we use this trick to redirect stdout/stderr to a log file.  The

contrib/void/build.sh

@@ -17,7 +17,7 @@ PKG_CONFIG_LIBDIR=/usr/lib/pkgconfig:/usr/local/lib/pkgconfig ./configure	\
 		 --prefix=/usr                  --exec-prefix=			\
 		 --sysconfdir=/etc              --localstatedir=/var		\
                  --enable-dbus-plugin           --enable-x11-common-plugin	\
-		 --enable-alsa-utils-plugin	--enable-watchdog		\
+		 --enable-alsa-utils-plugin	--with-watchdog			\
 		 --with-heading="Void Linux"
 
 echo

contrib/void/finit.d/available/Makefile.am

@@ -1,5 +1,5 @@
 docsdir   := @docdir@/contrib/void/finit.d/available
 docs_DATA  = acpid.conf crond.conf dhcpcd.conf dmeventd.conf getty.conf	\
-             keymap.conf klogd.conf lxdm.conf modules.conf ntpd.conf	\
-             sshd.conf syslogd.conf uuidd.conf
+             keymap.conf lxdm.conf modules.conf ntpd.conf		\
+             sshd.conf syslog.conf uuidd.conf
 EXTRA_DIST = $(docs_DATA)

contrib/void/finit.d/available/klogd.conf

@@ -1,2 +0,0 @@
-# Start klogd as soon as possible, should always run
-service [S12345] <pid/syslogd> klogd -n -- Kernel log daemon

contrib/void/finit.d/available/modules.conf

@@ -4,3 +4,4 @@
 #module evdev
 #module mousedev
 #module 8139cp
+module softdog

contrib/void/finit.d/available/syslog.conf

@@ -0,0 +1,4 @@
+# Start syslogd as soon as possible, should always run
+# Start klogd as soon as possible, but after syslogd
+service [S12345] syslogd -n             -- Syslog daemon
+service [S12345] <pid/syslogd> klogd -n -- Kernel log daemon

contrib/void/finit.d/available/syslogd.conf

@@ -1,2 +0,0 @@
-# Start syslogd as soon as possible, should always run
-service [S12345] syslogd -n -- Syslog daemon

contrib/void/install.sh

@@ -10,7 +10,7 @@ fi
 # Adjust base directory
 if [ -e xbps ]; then
     cd ../..
-elif [ ! -e autogen.sh ]; then
+elif [ ! -e configure.ac ]; then
     echo "*** Please run this script from the Finit base directory."
     exit 1
 fi

doc/TODO.md

@@ -19,9 +19,6 @@ Some of these are also registered in the issue tracker on GitHub.
 * Add new (complimentary) .svc file format to slowly migrate away
   from the very terse one-liner format currently used
 * Allow running as non-pid1 => read .conf and RCSD form cmdline
-* When a process dies, and Finit does not restart it, we should collect
-  a `siginfo_t` from the SIGCHLD signal and supply to the user in case
-  `initctl status <process>` is called to debug the issue.
 * Add `finit.conf` support for UPS notification (SIGPWR) to start a task
   using, e.g. <sys/power/{ok,fail,low}> conditions.  More info in sig.c
 * Add `finit.conf` support for ctrl-alt-delete (SIGINT) and kbrequest,
@@ -29,9 +26,18 @@ Some of these are also registered in the issue tracker on GitHub.
   e.g, <sys/key/ctrlaltdel> and <sys/key/signal> like SIGPWR handling.
 * Write man pages for finit and `finit.conf`, steal from the excellent
   `pimd` man pages ...
-* Add support for timed shutdown in Finit, including cancelled shutdown
+
+
+Initctl
+-------
+
+* Add support for timed shutdown in Finit, including cancelled shutdown:
+
+        shutdown -h 03:35 "foobar was here"
+
 * Add support for JSON output, or similar, from `initctl show`, e.g.
   `initctl show --json`
+* Add support for `-s` to redirect all output to syslog, for scripting
 
 
 Init

doc/bootstrap.md

@@ -9,9 +9,11 @@ Bootstrap
 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` and all `/etc/finit.d/*.conf` files
-10. Start built-in watchdog, if enabled
-11. Set hostname
+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
@@ -34,7 +36,7 @@ Bootstrap
 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, or a fallback shell on `/dev/console`
+27. Start all configured TTYs
 
 In (19) and (22) tasks and services defined in `/etc/finit.conf` and
 `/etc/sysctl.d/*.conf` are started.  Remember, all `service` and `task`

doc/build.md

@@ -153,19 +153,22 @@ kernel messages and add `debug` to enable Finit debug messages.
 append="init=/sbin/finit debug"
 ```
 
-To debug startup issues, in particular issues with getty/login, try
-`configure --enable-fallback-shell`.  When no TTYs are detected, and
-Finit is configured with this option, Finit will try to start a bare
-`/bin/sh` on the boot console.  Remember, this is only for debugging
-and would leave your production system potentially wide open.
+To debug startup issues, in particular issues with getty/login, add
+the following to your Finit .conf file:
+
+    tty [12345789] notty noclear
+
+The `notty` option ensures reusing the stdin/stdout set up by the
+kernel.  Remember, this is only for debugging and would leave your
+production system potentially wide open.
 
 There is also a rescue shell available, in case Finit crashes and the
 kernel usually reboots: `configure --enable-emergency-shell`.  However,
 the behavior of Finit is severely limited when this is enabled, so use
 it only for debugging start up issues when Finit crashes.
 
-**NOTE:** Neither of these two configure options should be enabled on
-  production systems since they can potentially give a user root access.
+**NOTE:** Neither of these options should be enabled on production
+         systems since they can potentially give a user root access.
 
 
 [1]:       ftp://troglobit.com/finit/finit-3.0.tar.xz

doc/cmdline.md

@@ -30,10 +30,12 @@ The `bool` setting is one of `on, off, true false, 1, 0`.
 * `quiet`: Suppress kernel logging to console, except for warnings and
   errors.
 
-* `rescue`: Start Finit rescue mode; no network, no `fsck` of file
-    systems in `/etc/fstab`, no `/etc/rc.local`, no `runparts`.  The
-    configuration is read from `/lib/finit/rescue.conf`, and can be
-    modified by the operator.
+* `rescue`: Start rescue/maintenance mode.  If your system comes with
+    the bundled `sulogin` program (Finit, or from util-linux/Busybox),
+    you will be given a root login to a maintenance shell.  However, if
+    `sulogin` is missing, the file `/lib/finit/rescue.conf` is read and
+    the system booted in a limited fallback mode.  See [config.md][]
+	for more information.
 
 * `single`, or `S`: Single user mode, runlevel 1, in Finit.  Useful to
     debug problems with the regular runlevel.  All services and TTYs in
@@ -42,5 +44,5 @@ The `bool` setting is one of `on, off, true false, 1, 0`.
 
 For more on kernel boot paramaters, see the man page [bootparam(7)][].
 
-
+[config.md][]:  config.md#rescue-mode
 [bootparam(7)]: https://www.man7.org/linux/man-pages/man7/bootparam.7.html

doc/conditions.md

@@ -45,7 +45,31 @@ being created.
 Triggering
 ----------
 
-Conditions can only be triggered by built-in plugins.
+Conditions are mainly triggered (asserted) by built-in plugins, e.g.,
+`netlink.so` and `pidfile.so`, see below for built-in conditions.  Finit
+also supports user-defined conditions, sometimes referred to as static
+or one-shot conditions.  They live in the `<usr/...` namespace and are
+constrained to a flat hierarchy without any sub-directories, unlike the
+pidfile plugin, for instance.
+
+User-defined conditions are controlled using the `initctl cond set` and
+`initctl cond clear` commands:
+
+    initctl cond set foo
+    initctl cond clear foo
+
+The purpose of user-defined conditions is to be able to start or stop
+services, or run/task jobs, on external site-dependent stimuli.
+Example:
+
+    service [2345] <usr/foo> alarm --arg foo -- Foo alarm
+
+For convenience, prefixing with `usr/` is allowed, but any other slashes
+or period characters are disallowed.  E.g., to trigger the `Foo alarm`,
+the same as above, can also be achieved like this:
+
+    initctl cond set usr/foo
+    initctl cond clear usr/foo
 
 Conditions retain their current state until the next reconfiguration or
 runlevel change.  At that point all set conditions transition into the
@@ -57,6 +81,8 @@ unchanged) state of it.
 > **Note:** For `pid/` conditions it is expected that services "touch"
 >           or recreate their PID file on `SIGHUP`.
 
+Static (one-shot) conditions, like `usr/`, never enter the `flux` state.
+
 
 Built-in Conditions
 -------------------

doc/config.md

@@ -2,7 +2,9 @@ Configuration
 =============
 
 * [Introduction](#introduction)
+* [Service Environment](#service-environment)
 * [Service Wrapper Scripts](#service-wrapper-scripts)
+* [Cgroups](#cgroups)
 * [Configuration File Syntax](#configuration-file-syntax)
   * [Hostname](#hostname)
   * [Kernel Modules](#kernel-modules)
@@ -19,9 +21,11 @@ Configuration
   * [TTYs and Consoles](#ttys-and-consoles)
   * [Non-privileged Services](#non-privileged-services)
   * [Redirecting Output](#redirecting-output)
+* [Rescue Mode](#rescue-mode)
 * [Limitations](#limitations)
   * [/etc/finit.conf](#etcfinitconf)
   * [/etc/finit.d](#etcfinitd)
+* [Watchdog](#watchdog)
 
 
 Introduction
@@ -35,8 +39,8 @@ Linux distributions -- each package can provide its own "script" file.
 - `/etc/finit.d/*.conf`: snippets, usually one service per file
 
 Not all configuration directives are available in `/etc/finit.d/*.conf`
-and some directives are only available at bootstrap, runlevel `S`, see
-the section [Limitations](#limitations) below for details.
+and some directives are only available at [bootstrap][], runlevel `S`,
+see the section [Limitations](#limitations) below for details.
 
 To add a new service, simply drop a `.conf` file in `/etc/finit.d` and
 run `initctl reload`.  (It is also possible to `SIGHUP` to PID 1, or
@@ -128,6 +132,67 @@ example employs a wrapper script in `/etc/start.d`.
 >           instead of sending SIGHUP at restart/reload events.
 
 
+Cgroups
+-------
+
+There are three major cgroup configuration directives:
+
+ 1. Global top-level group: init, system, user, or a custom group
+ 2. Selecting a top-level group for a set of run/task/services
+ 3. Per run/task/service limits
+
+Top-level group configuration.
+
+    # Top-level cgroups and their default settings.  All groups mandatory
+    # but more can be added, max 8 groups in total currently.  The cgroup
+    # 'root' is also available, reserved for RT processes.  Settings are
+    # as-is, only one shorthand 'mem.' exists, other than that it's the
+    # cgroup v2 controller default names.
+    cgroup init   cpu.weight:100
+    cgroup user   cpu.weight:100
+    cgroup system cpu.weight:9800
+
+Adding an extra cgroup `maint/` will require you to adjust the weight of
+the above three.  We leave `init/` and `user /` as-is reducing weight of
+`system/` to 9700.
+
+    cgroup system cpu.weight:9700
+
+    # Example extra cgroup 'maint'
+    cgroup maint  cpu.weight:100
+
+By default, the `system/` cgroup is selected for almost everything.  The
+`init/` cgroup is reserved for PID 1 itself and its closest relatives.
+The `user/` cgroup is for local TTY logins spawned by getty.
+
+To select a different top-level cgroup, e.g. `maint/`, one can either
+define it for a group of run/task/service directives in a `.conf` or per
+each stanza:
+
+    cgroup.maint
+    service [...] <...> /path/to/foo args -- description
+    service [...] <...> /path/to/bar args -- description
+
+or
+
+    service [...] <...> cgroup.maint /path/to/foo args -- description
+
+The latter form also allows per-stanza limits on the form:
+
+    service [...] <...> cgroup.maint:cpu.max:10000,mem.max:655360 /path/to/foo args -- description
+
+Notice the comma separation and the `mem.` exception to the rule: every
+cgroup setting maps directly to cgroup v2 syntax.  I.e., `cpu.max` maps
+to the file `/sys/fs/cgroup/maint/foo/cpu.max`.  There is not filtering,
+except for expanding the shorthand `mem.` to `memory.`, if the file is
+not available, either the cgroup controller is not available in your
+Linux kernel, or the name is misspelled.
+
+> Linux cgroups and details surrounding values are not explained in the
+> Finit documentation.  The Linux admin-guide cover this well:
+> <https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html>
+
+
 Configuration File Syntax
 -------------------------
 
@@ -140,7 +205,7 @@ 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
@@ -158,7 +223,7 @@ 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
@@ -171,7 +236,7 @@ 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][]).
 
 
 ### Resource Limits
@@ -205,13 +270,13 @@ stanzas can share the same rlimits if they are in the same .conf.
 
 **Syntax:** `runlevel <N>`
 
-Defines 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
+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.
 
 Default: 2
 
-> **Note:** only read and executed in runlevel S (bootstrap).
+> **Note:** only read and executed in runlevel S ([bootstrap][]).
 
 
 ### One-shot Commands (sequence)
@@ -333,6 +398,28 @@ process using a different signal, use the option `halt:SIGNAL`, e.g.,
 use the option `kill:SEC`, e.g., `kill:10` to wait 10 seconds before
 sending `SIGKILL`.
 
+Services support `pre:script` and `post:script` actions as well.  These
+run as the same `@USER:GROUP` as the service itself, with any `env:file`
+sourced.  The scripts must use an absolute path, but are executed from
+the `$HOME` of the given user.  The scripts are not called with any
+argument (currently), but both get the `SERVICE_IDENT=foo` environment
+variable set.  Here `foo` denotes the identity of the service, which if
+there are multiple services named `foo`, may be `foo:1`, or any unique
+identifier specified in the .conf file.  The `post:script` is called
+with an additional set of environment variables:
+
+ - `EXIT_CODE=[exited,signal]`: set to one of `exited` or `signal`
+ - `EXIT_STATUS=[num,SIGNAME]`: set to one of exit status code from
+   the program, if it exited normally, or the signal name (`HUP`,
+   `TERM`, etc.) if it exited due to signal
+
+The scripts have a default execution time of 3 seconds before they are
+SIGKILLed, this can be adjusted using the above `kill:SEC` syntax.
+
+> **Note:** the `pre:script` _must_ be idempotent, because a service
+>           can transition between READY and HALTED states any number
+>           of times before going to RUNNING.
+
 
 ### Run-parts Scripts
 
@@ -341,7 +428,7 @@ sending `SIGKILL`.
 Call [run-parts(8)][] on `DIR` to run start scripts.  All executable
 files, or scripts, in the directory are called, in alphabetic order.
 The scripts in this directory are executed at the very end of runlevel
-`S`, bootstrap.
+`S`, [bootstrap][].
 
 It can be beneficial to use `S01name`, `S02othername`, etc. if there
 is a dependency order between the scripts.  Symlinks to existing
@@ -351,7 +438,10 @@ Similar to the `/etc/rc.local` shell script, make sure that all your
 services and programs either terminate or start in the background or
 you will block Finit.
 
-> **Note:** only read and executed in runlevel S (bootstrap).
+> **Note:** only read and executed in runlevel S ([bootstrap][]).
+
+[run-parts(8)]: http://manpages.debian.org/cgi-bin/man.cgi?query=run-parts
+
 
 ### Including Finit Configs
 
@@ -377,8 +467,9 @@ size limit is reached.
 
 ### TTYs and Consoles
 
-**Syntax:** `tty [LVLS] <DEV> [BAUD] [noclear] [nowait] [nologin] [TERM]`  
-  `tty [LVLS] <CMD> <ARGS> [noclear] [nowait]`
+**Syntax:** `tty [LVLS] <COND> DEV [BAUD] [noclear] [nowait] [nologin] [TERM]`  
+  `tty [LVLS] <COND> CMD <ARGS> [noclear] [nowait]`  
+  `tty [LVLS] <COND> [notty] [rescue]`
 
 The first variant of this option uses the built-in getty on the given
 TTY device DEV, in the given runlevels.  The DEV may be the special
@@ -386,15 +477,29 @@ keyword `@console`, or `console`, useful on embedded systems.
 
 The default baud rate is 0, i.e., keep kernel default.
 
+> The `tty` stanza inherits runlevel, condition (and other feature)
+> parsing from the `service` stanza.  So TTYs can run in one or many
+> runlevels and depend on any condition supported by Finit.  This is
+> useful e.g. to depend on `<pid/elogind>` before starting a TTY.
+
 **Example:**
 
     tty [12345] /dev/ttyAMA0 115200 noclear vt220
 
 The second `tty` syntax variant is for using an external getty, like
 agetty or the BusyBox getty.
-    
-By default both variants *clear* the TTY and *wait* for the user to
-press enter before starting getty.
+
+The third variant is for board bringup and the `rescue` boot mode.  No
+device node is required in this variant, the same output that the kernel
+uses is reused for stdio.  If the `rescue` option is omitted, a shell is
+started (`nologin`, `noclear`, and `nowait` are implied), if the rescue
+option is set the bundled `/libexec/finit/sulogin` is started to present
+a bare-bones root login prompt.  If the root (uid:0, gid:0) user does
+not have a password set, no rescue is possible.  For more information,
+see the [Rescue Mode](#rescue-mode) section.
+
+By default, the first two syntax variants *clear* the TTY and *wait* for
+the user to press enter before starting getty.
 
 **Example:**
 
@@ -423,16 +528,23 @@ uses `ttyS0`, `ttyAMA0`, `ttyMXC0`, or anything else.  Finit figures
 it out by querying sysfs: `/sys/class/tty/console/active`.  The speed
 can be omitted to keep the kernel default.
 
+> Most systems get by fine by just using `console`, which will evaluate
+> to `/dev/console`.  If you have to use `@console` to get any output,
+> you may have some issue with your kernel config.
+
 **Example:**
 
     tty [12345] @console noclear vt220
 
-On really bare bones systems Finit offers a fallback shell, which
-should not be enabled on production systems since.  This because it
-may give a user root access without having to log in.  However, for
-bringup and system debugging it can come in handy:
+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
+device node:
 
-    configure --enable-fallback-shell
+    tty [12345789] notty
+
+This should of course not be enabled on production systems.  Because it
+may give a user root access without having to log in.  However, for
+board bringup and system debugging it can come in handy.
 
 One can also use the `service` stanza to start a stand-alone shell:
 
@@ -483,8 +595,58 @@ Log rotation is controlled using the global `log` setting.
 
     service log:prio:user.warn,tag:ntpd /sbin/ntpd pool.ntp.org -- NTP daemon
 
-Worth noting is that conditions is allowed for all these stanzas.  For a
-detailed description, see the [Conditions](conditions.md) document.
+
+Rescue Mode
+-----------
+
+Finit supports a rescue mode which is activated by the `rescue` option
+on the kernel command line.  See [cmdline docs](cmdline.md) for how to
+activate it.
+
+The rescue mode comes in two flavors; *traditional* and *fallback*.
+
+
+### Traditional
+
+This is what most users expect.  A very early maintenance login prompt,
+served by the bundled `/libexec/finit/sulogin` program, or the standard
+`sulogin` from util-linux or BusyBox is searched for in the UNIX default
+`$PATH`.  If a successful login is made, or the user exits (Ctrl-D), the
+rescue mode is ended and the system boots up normally.
+
+> **Note:** if the user (UID 0 and GID 0) does not have a password, or
+> __the account is locked__, the user is presented with a password-less
+> `"Press enter to enter maintenance mode."`, prompt which opens up a
+> root shell.
+
+
+### Fallback
+
+If no `sulogin` program is found, Finit tries to bring up as much of its
+own functionality as possible, yet limiting many aspects, meaning; no
+network, no `fsck` of file systems in `/etc/fstab`, no `/etc/rc.local`,
+no `runparts`, and most plugins are skipped (except those that provide
+functionality for the condition subsystem).
+
+Instead of reading `/etc/finit.conf` et al, system configuration is read
+from `/lib/finit/rescue.conf`, which can be freely modified by the
+system administrator.
+
+The bundled default `rescue.conf` contains nothing more than:
+
+    runlevel 1
+    tty [12345] rescue
+
+The `tty` has the `rescue` option set, which works similar to the board
+bring-up tty option `notty`.  The major difference being that `sulogin`
+is started to query for root/admin password.  If `sulogin` is not found,
+`rescue` behaves like `notty` and gives a plain root shell prompt.
+
+If Finit cannot find `/lib/finit/rescue.conf` it defaults to:
+
+    tty [12345] rescue
+
+There is no way to exit the *fallback* rescue mode.
 
 
 Limitations
@@ -496,15 +658,49 @@ from `/etc/finit.conf`, since it is the first `.conf` file Finit reads.
 
 Originally, `/etc/finit.conf` was the only way to set up a Finit system.
 Today it is mainly used for bootstrap settings like system hostname,
-network bringup and system shutdown.  These can now also be set in any
-`.conf` file in `/etc/finit.d`.
+early module loading for watchdogd, network bringup and system shutdown.
+These can now also be set in any `.conf` file in `/etc/finit.d`.
 
 There is, however, nothing preventing you from having all configuration
 settings in `/etc/finit.conf`.
 
 > **Note:** The `/etc/finit.d` directory was previously the default
->          Finit `runparts` directory.  Finit no longer has a default
->          `runparts`, make sure to update your setup, or the finit
->          configuration, accordingly.
+>          Finit [runparts](#run-parts-scripts) directory.  Finit no
+>          longer has a default `runparts`, make sure to update your
+>          setup, or the finit configuration, accordingly.
 
-[run-parts(8)]: http://manpages.debian.org/cgi-bin/man.cgi?query=run-parts
+
+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
+details.  This delegation is to ensure that the system is rebooted by a
+hardware watchdog timer -- on many embedded systems this is crucial to
+ensure all circuits on the board are properly reset for the next boot,
+in effect ensuring the system works the same after both a power-on and
+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
+`SIGPWR` is sent to advise watchdogd of the pending reboot.  Then, when
+the necessary steps of preparing the system for shutdown (umount etc.)
+are completed, Finit sends `SIGTERM` to watchdogd and puts itself in a
+10 sec timeout loop waiting for the WDT to reset the board.  If a reset
+is not done before the timeout, Finit falls back to`reboot(RB_AUTOBOOT)`
+which tells the kernel to do the reboot.
+
+An external watchdog service can also be used.  The more advanced cousin
+[watchdogd](https://github.com/troglobit/watchdogd/) is the recommended
+option here.  It can register itself with Finit using the same IPC as
+`initctl`.  If the bundled watchdogd is running a hand-over takes place,
+so it's safe to have both services installed on a system.  For the
+hand-over to work it requires that the WDT driver supports the safe exit
+functionality where `"V"` is written to the device before closing the
+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

doc/service.md

@@ -18,7 +18,7 @@ The current state depends on the two following conditions:
   
   A service may be stopped, or blocked, for several reasons:
 
-  - The user has manually stopped the service using `initctl stop JOB`
+  - The user has manually stopped the service using `initctl stop NAME`
   - The program exits immediately. I.e. keeps crashing (make sure to use
     the 'run this service in the foreground' command line option)
   - The binary is missing in the filesystem

doc/signals.md

@@ -12,28 +12,36 @@ have been adopted:
 Same effect as `finit q`, `init q`, or `initctl reload`, reloads all
 *.conf files in `/etc/finit.d/`
 
+This also restarts the API (initctl) socket, like SysV init and systemd
+does on USR1 with their FIFO/D-Bus.
+
+
 `SIGUSR1`
 ---------
 
-Calls shutdown hooks, including `HOOK_SHUTDOWN`, stopping all running
-processes, and unmounts all file systems.  Then tells kernel to halt.
-Most people these days want `SIGUSR2` though.
+Since Finit 4.1 this signal causes Finit to restart its API (initctl)
+socket, like SysV init and systemd does on USR1 with their FIFO/D-Bus.
+
+Finit <= 4.0 performed a system halt (like USR2 without power-off), but
+this caused compatibility problems with systemd and sysvinit on desktop
+systems.  Hence, since Finit 4.1 it is no longer possible to halt a
+system with a signal.
 
-SysV init and systemd use this to re-open their FIFO/D-Bus.
 
 `SIGUSR2`
 ---------
 
-Like SIGUSR1, but tell kernel to power-off the system, if ACPI or
-similar exists to actually do this.  If the kernel fails power-off,
-Finit falls back to halt.
- 
+Calls shutdown hooks, including `HOOK_SHUTDOWN`, stopping all running
+processes, and unmounts all file systems.  Then tells kernel to power
+off the system, if ACPI or similar exists to actually do this.  If the
+kernel fails power-off, Finit falls back to halt.
+
 SysV init N/A, systemd dumps its internal state to log.
 
 `SIGTERM`
 ---------
 
-Like `SIGUSR1`, but tell kernel to reboot the system when done.
+Like `SIGUSR2`, but tell kernel to reboot the system when done.
  
 SysV init N/A, systemd rexecutes itself.
 

plugins/Makefile.am

@@ -7,7 +7,7 @@ AM_CPPFLAGS        += $(lite_CFLAGS)
 
 if STATIC
 noinst_LTLIBRARIES  = libplug.la
-libplug_la_SOURCES  = bootmisc.c modprobe.c rtc.c pidfile.c procps.c tty.c urandom.c
+libplug_la_SOURCES  = bootmisc.c modprobe.c rtc.c pidfile.c procps.c sys.c tty.c urandom.c usr.c
 
 if BUILD_ALSA_UTILS_PLUGIN
 libplug_la_SOURCES += alsa-utils.c
@@ -34,7 +34,7 @@ libplug_la_SOURCES += resolvconf.c
 endif
 
 else
-pkglib_LTLIBRARIES  = bootmisc.la modprobe.la rtc.la pidfile.la procps.la tty.la urandom.la
+pkglib_LTLIBRARIES  = bootmisc.la modprobe.la rtc.la pidfile.la procps.la sys.la tty.la urandom.la usr.la
 
 if BUILD_ALSA_UTILS_PLUGIN
 pkglib_LTLIBRARIES += alsa-utils.la

plugins/alsa-utils.c

@@ -30,6 +30,11 @@
 
 static void save(void *arg)
 {
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	if (whichp(ALSACTL)) {
 		_d("Saving sound settings ...");
 		run_interactive(ALSACTL " -g store", "Saving sound settings");
@@ -38,6 +43,11 @@ static void save(void *arg)
 
 static void restore(void *arg)
 {
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	if (whichp(ALSACTL)) {
 		_d("Restoring sound settings ...");
 		run_interactive(ALSACTL " -g restore", "Restoring sound settings");

plugins/bootmisc.c

@@ -65,7 +65,7 @@ static int is_tmpfs(char *path)
 	return tmpfs;
 }
 
-static int do_clean(const char *fpath, const struct stat *sb, int tflag, struct FTW *ftw)
+static int bootclean(const char *fpath, const struct stat *sb, int tflag, struct FTW *ftw)
 {
 	if (ftw->level == 0)
 		return 1;
@@ -76,8 +76,14 @@ static int do_clean(const char *fpath, const struct stat *sb, int tflag, struct
 	return 0;
 }
 
-/* We can safely skip tmpfs, nothing to clean from previous boot there */
-static void bootclean(void)
+/*
+ * Cleanup stale files from previous boot, if any still linger on.
+ * Some systems, e.g. Alpine Linux, still have a persistent /run and
+ * /tmp, i.e. not tmpfs.
+ *
+ * We can safely skip tmpfs, nothing to clean there.
+ */
+static void clean(void *arg)
 {
 	char *dir[] = {
 		"/tmp/",
@@ -90,7 +96,7 @@ static void bootclean(void)
 		if (is_tmpfs(dir[i]))
 			continue;
 
-		nftw(dir[i], do_clean, 20, FTW_DEPTH);
+		nftw(dir[i], bootclean, 20, FTW_DEPTH);
 	}
 }
 
@@ -99,13 +105,11 @@ static void bootclean(void)
  */
 static void setup(void *arg)
 {
-	int uid, gid;
+	mode_t prev;
 
-	/* Cleanup stale files, if any still linger on. */
-	bootclean();
+	prev = umask(0);
 
 	_d("Setting up FHS structure in /var ...");
-	umask(0);
 	makedir("/var/cache",      0755);
 	makedir("/var/db",         0755); /* _PATH_VARDB on some systems */
 	makedir("/var/games",      0755);
@@ -115,15 +119,16 @@ static void setup(void *arg)
 	makedir("/var/lib/urandom",0755);
 	if (fisdir("/run")) {
 		_d("System with new /run tmpfs ...");
-		makedir("/run/lock",       1777);
+		if (!fisdir("/run/lock"))
+			makedir("/run/lock", 1777);
 		symlink("/run/lock", "/var/lock");
-		symlink("/dev/shm",  "/run/shm");
+		symlink("/dev/shm", "/run/shm");
 
 		/* compat only, should really be set up by OS/dist */
-		symlink("/run",      "/var/run");
+		symlink("/run", "/var/run");
 	} else {
-		makedir("/var/lock",       1777);
-		makedir("/var/run",        0755);
+		makedir("/var/lock", 1777);
+		makedir("/var/run", 0755);
 	}
 	makedir("/var/log",        0755);
 	makedir("/var/mail",       0755);
@@ -133,19 +138,21 @@ static void setup(void *arg)
 	makedir("/var/tmp",        0755);
 	makedir("/var/empty",      0755);
 
-	/*
-	 * If /etc/group or "utmp" group is missing,
-	 * default to "root"/"wheel" group
-	 */
-	_d("Setting up necessary UTMP files ...");
-	gid = getgroup("utmp");
-	if (gid < 0)
-		gid = 0;
-
 	/*
 	 * UTMP actually needs multiple db files
 	 */
 	if (has_utmp()) {
+		int gid;
+
+		/*
+		 * If /etc/group or "utmp" group is missing, default to
+		 * "root", or "wheel", group.
+		 */
+		_d("Setting up necessary UTMP files ...");
+		gid = getgroup("utmp");
+		if (gid < 0)
+			gid = 0;
+
 		create("/var/run/utmp",    0644, 0, gid); /* Currently logged in */
 		create("/var/log/wtmp",    0644, 0, gid); /* Login history       */
 		create("/var/log/btmp",    0600, 0, gid); /* Failed logins       */
@@ -165,39 +172,31 @@ static void setup(void *arg)
 	makedir("/var/run/network",0755); /* Needed by Debian/Ubuntu ifupdown */
 	makedir("/var/run/lldpd",  0755); /* Needed by lldpd */
 	makedir("/var/run/pluto",  0755); /* Needed by Openswan */
-	makedir("/var/run/quagga", 0755); /* Needed by Quagga */
-	makedir("/var/log/quagga", 0755); /* Needed by Quagga */
-	makedir("/var/run/frr",    0755); /* Needed by FRR */
-	makedir("/var/log/frr",    0755); /* Needed by FRR */
+	mksubsys("/var/run/quagga", 0755, "quagga", "quagga");
+	mksubsys("/var/log/quagga", 0755, "quagga", "quagga");
+	mksubsys("/var/run/frr",    0755, "frr", "frr");
+	mksubsys("/var/log/frr",    0755, "frr", "frr");
 	makedir("/var/run/sshd",  01755); /* OpenSSH  */
 
-	gid = getgroup("tty");
-	if (gid < 0)
-		gid = 0;
-	makefifo("/dev/xconsole",  0640); /* sysklogd */
-	chown("/dev/xconsole", 0, gid);
-
 	if (!fexist("/etc/mtab"))
 		symlink("../proc/self/mounts", "/etc/mtab");
 
 	/* Void Linux has a uuidd that runs as uuid:uuid and needs /run/uuid */
-	uid = getuser("uuidd", NULL);
-	if (uid >= 0) {
-		gid = getgroup("uuidd");
-		if (gid < 0)
-			gid = 0;
-		makedir("/var/run/uuidd", 0755);
-		chown("/var/run/uuidd", uid, gid);
-	}
+	mksubsys("/var/run/uuidd", 0755, "uuidd", "uuidd");
 
-	umask(022);
+	/* Debian has /run/sudo, ensure correct perms and SELinux label */
+	mksubsys("/var/run/sudo", 0711, "root", "root");
+	mksubsys("/var/run/sudo/ts", 0700, "root", "root");
+	if (whichp("restorecon"))
+		run("restorecon /var/run/sudo /var/run/sudo/ts");
+
+	umask(prev);
 }
 
 static plugin_t plugin = {
 	.name = __FILE__,
-	.hook[HOOK_BASEFS_UP] = {
-		.cb  = setup
-	},
+	.hook[HOOK_MOUNT_POST] = { .cb = clean },
+	.hook[HOOK_BASEFS_UP]  = { .cb = setup },
 	.depends = { "pidfile" },
 };
 

plugins/dbus.c

@@ -32,13 +32,21 @@
 #include "conf.h"
 
 #define DAEMON "dbus-daemon"
-#define ARGS   "--nofork --system"
+#define ARGS   "--nofork --system --syslog-only"
 #define DESC   "D-Bus message bus daemon"
 
+#define DAEMONUSER "messagebus"
+
 static void setup(void *arg)
 {
-	char *cmd;
 	char line[256];
+	mode_t prev;
+	char *cmd;
+
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
 
 	cmd = which(DAEMON);
 	if (!cmd) {
@@ -46,29 +54,28 @@ static void setup(void *arg)
 		return;
 	}
 
-	umask(0);
+	prev =umask(0);
 
-	makedir("/var/run/dbus", 0755);
-	makedir("/var/lock/subsys", 0755);
-	makedir("/var/lock/subsys/messagebus", 0755);
+	mksubsys("/var/run/dbus", 0755, DAEMONUSER, DAEMONUSER);
 	if (whichp("dbus-uuidgen"))
-		run("dbus-uuidgen --ensure");
+		run_interactive("dbus-uuidgen --ensure", "Creating machine UUID for D-Bus");
 
 	/* Clean up from any previous pre-bootstrap run */
-	erase("/var/run/dbus/pid");
+	remove("/var/run/dbus/pid");
 
 	/* Register service with Finit */
-	snprintf(line, sizeof(line), "[S12345] pid:!/run/dbus/pid log:prio:daemon.info,tag:dbus %s %s -- %s", cmd, ARGS, DESC);
+	snprintf(line, sizeof(line), "[S12345789] cgroup.system @%s:%s %s %s -- %s",
+		 DAEMONUSER, DAEMONUSER, cmd, ARGS, DESC);
 	if (service_register(SVC_TYPE_SERVICE, line, global_rlimit, NULL))
 		_pe("Failed registering %s", DAEMON);
 	free(cmd);
 
-	umask(022);
+	umask(prev);
 }
 
 static plugin_t plugin = {
 	.name = __FILE__,
-	.hook[HOOK_NETWORK_UP] = {
+	.hook[HOOK_BASEFS_UP] = {
 		.cb  = setup
 	},
 	.depends = { "bootmisc", },

plugins/hotplug.c

@@ -34,6 +34,11 @@ static void setup(void *arg)
 	char cmd[256];
 	char *path;
 
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	/*
 	 * Populate /dev and prepare for runtime events from kernel.
 	 * Prefer udev if mdev is also available on the system.
@@ -43,17 +48,17 @@ static void setup(void *arg)
 		path = which("/lib/systemd/systemd-udevd");
 	if (path) {
 		/* Register udevd as a monitored service */
-		snprintf(cmd, sizeof(cmd), "[S12345789] pid:udevd name:udevd log %s "
+		snprintf(cmd, sizeof(cmd), "[S12345789] cgroup.system pid:udevd name:udevd log %s "
 			 "-- Device event managing daemon", path);
 		if (service_register(SVC_TYPE_SERVICE, cmd, global_rlimit, NULL)) {
 			_pe("Failed registering %s", path);
 		} else {
-			snprintf(cmd, sizeof(cmd), ":1 [S] <pid/udevd> log "
+			snprintf(cmd, sizeof(cmd), "cgroup.init :1 [S] <pid/udevd> log "
 				 "udevadm trigger -c add -t devices "
 				 "-- Requesting device events");
 			service_register(SVC_TYPE_RUN, cmd, global_rlimit, NULL);
 
-			snprintf(cmd, sizeof(cmd), ":2 [S] <pid/udevd> log "
+			snprintf(cmd, sizeof(cmd), "cgroup.init :2 [S] <pid/udevd> log "
 				 "udevadm trigger -c add -t subsystems "
 				 "-- Requesting subsystem events");
 			service_register(SVC_TYPE_RUN, cmd, global_rlimit, NULL);
@@ -90,6 +95,7 @@ static plugin_t plugin = {
 	.hook[HOOK_BASEFS_UP] = {
 		.cb  = setup
 	},
+	.depends = { "bootmisc" },
 };
 
 PLUGIN_INIT(plugin_init)

plugins/modules-load.c

@@ -36,7 +36,7 @@
 #define MODULES_LOAD_PATH "/etc/modules-load.d"
 #endif
 #define SERVICE_LINE \
-	":%d name:modprobe.%s [2345] /sbin/modprobe %s %s -- Kernel module: %s"
+	"cgroup.init :%d name:modprobe.%s [2345] /sbin/modprobe %s %s -- Kernel module: %s"
 
 static void load(void *arg)
 {
@@ -71,8 +71,7 @@ static void load(void *arg)
 
 			mod = strtok_r(mod, " ", &args);
 
-			snprintf(cmd, sizeof(cmd), SERVICE_LINE,
-					 index, mod, mod, args, mod);
+			snprintf(cmd, sizeof(cmd), SERVICE_LINE, index, mod, mod, args, mod);
 			service_register(SVC_TYPE_TASK, cmd, global_rlimit, NULL);
 
 			index++;

plugins/netlink.c

@@ -17,6 +17,7 @@
  */
 
 #include <arpa/inet.h>
+#include <ctype.h>
 #include <errno.h>
 #include <net/if.h>		/* IFNAMSIZ */
 #include <sys/socket.h>
@@ -29,31 +30,35 @@
 #include "cond.h"
 #include "helpers.h"
 #include "plugin.h"
+#include "service.h"
 
-static int nlmsg_validate(struct nlmsghdr *nh, size_t len)
-{
-	if (!NLMSG_OK(nh, len))
-		return 1;
+#define  NL_BUFSZ	4096
 
-	if (nh->nlmsg_type == NLMSG_DONE) {
-		_d("Done with netlink messages.");
-		return 1;
-	}
+struct nl_request {
+	struct nlmsghdr nh;
+	union {
+		struct rtmsg     rtm;
+		struct ifinfomsg ifi;
+	};
+};
 
-	if (nh->nlmsg_type == NLMSG_ERROR) {
-		_d("Netlink reports error.");
-		return 1;
-	}
+static int   nl_defidx;
+static int   nl_ifdown;
+static char *nl_buf;
 
-	return 0;
-}
 
-static void nl_route(struct nlmsghdr *nlmsg)
+static void nl_route(struct nlmsghdr *nlmsg, ssize_t len)
 {
+	char daddr[INET_ADDRSTRLEN];
+	char gaddr[INET_ADDRSTRLEN];
+	struct in_addr ind, ing;
 	struct rtmsg *r;
 	struct rtattr *a;
+	int plen = 0;
+	int dst = 0;
+	int idx = 0;
+	int gw = 0;
 	int la;
-	int gw = 0, dst = 0, mask = 0, idx = 0;
 
 	if (nlmsg->nlmsg_len < NLMSG_LENGTH(sizeof(struct rtmsg))) {
 		_e("Packet too small or truncated!");
@@ -63,8 +68,14 @@ static void nl_route(struct nlmsghdr *nlmsg)
 	r  = NLMSG_DATA(nlmsg);
 	a  = RTM_RTA(r);
 	la = RTM_PAYLOAD(nlmsg);
+	if (la >= len) {
+		_e("Packet too large!");
+		return;
+	}
+
 	while (RTA_OK(a, la)) {
 		void *data = RTA_DATA(a);
+
 		switch (a->rta_type) {
 		case RTA_GATEWAY:
 			gw = *((int *)data);
@@ -73,8 +84,8 @@ static void nl_route(struct nlmsghdr *nlmsg)
 
 		case RTA_DST:
 			dst = *((int *)data);
-			mask = r->rtm_dst_len;
-			//_d("MASK: 0x%04x", mask);
+			plen = r->rtm_dst_len;
+			//_d("Prefix LEN: 0x%04x", plen);
 			break;
 
 		case RTA_OIF:
@@ -86,11 +97,20 @@ static void nl_route(struct nlmsghdr *nlmsg)
 		a = RTA_NEXT(a, la);
 	}
 
-	if ((!dst && !mask) && (gw || idx)) {
-		if (nlmsg->nlmsg_type == RTM_DELROUTE)
+	ind.s_addr = dst;
+	ing.s_addr = gw;
+	inet_ntop(AF_INET, &ind, daddr, sizeof(daddr));
+	inet_ntop(AF_INET, &ing, gaddr, sizeof(gaddr));
+	_d("Got gw %s dst/len %s/%d ifindex %d", gaddr, daddr, plen, idx);
+
+	if ((!dst && !plen) && (gw || idx)) {
+		if (nlmsg->nlmsg_type == RTM_DELROUTE) {
 			cond_clear("net/route/default");
-		else
+			nl_defidx = 0;
+		} else {
 			cond_set("net/route/default");
+			nl_defidx = idx;
+		}
 	}
 }
 
@@ -105,12 +125,45 @@ static void net_cond_set(char *ifname, char *cond, int set)
 		cond_clear(msg);
 }
 
-static void nl_link(struct nlmsghdr *nlmsg)
+static int validate_ifname(const char *ifname)
+{
+	if (!ifname || !ifname[0])
+		return 1;
+
+	if (strnlen(ifname, IFNAMSIZ) == IFNAMSIZ)
+		return 1;
+
+	if (!strcmp(ifname, ".") || !strcmp(ifname, ".."))
+		return 1;
+
+	while (*ifname) {
+		if (*ifname == '/' || *ifname == ':' || isspace(*ifname))
+			return 1;
+		ifname++;
+	}
+
+	return 0;
+}
+
+/*
+ * Check if this interface was associated with the default route
+ * previously, or if it's been removed.  If so, trigger a recheck
+ * of the system default route.
+ */
+static void nl_check_default(char *ifname)
+{
+	int idx = (int)if_nametoindex(ifname);
+
+	if ((nl_defidx > 0 && nl_defidx == idx) || (idx == 0 && errno == ENODEV))
+		nl_ifdown = 1;
+}
+
+static void nl_link(struct nlmsghdr *nlmsg, ssize_t len)
 {
-	int la;
 	char ifname[IFNAMSIZ + 1];
-	struct rtattr *a;
 	struct ifinfomsg *i;
+	struct rtattr *a;
+	int la;
 
 	if (nlmsg->nlmsg_len < NLMSG_LENGTH(sizeof(struct ifinfomsg))) {
 		_e("Packet too small or truncated!");
@@ -120,68 +173,228 @@ static void nl_link(struct nlmsghdr *nlmsg)
 	i  = NLMSG_DATA(nlmsg);
 	a  = (struct rtattr *)((char *)i + NLMSG_ALIGN(sizeof(struct ifinfomsg)));
 	la = NLMSG_PAYLOAD(nlmsg, sizeof(struct ifinfomsg));
+	if (la >= len) {
+		_e("Packet too large!");
+		return;
+	}
 
-	while (RTA_OK(a, la)) {
-		if (a->rta_type == IFLA_IFNAME) {
-			strlcpy(ifname, RTA_DATA(a), sizeof(ifname));
-			switch (nlmsg->nlmsg_type) {
-			case RTM_NEWLINK:
-				/*
-				 * New interface has appeared, or interface flags has changed.
-				 * Check ifi_flags here to see if the interface is UP/DOWN
-				 */
-				_d("%s: New link, flags 0x%x, change 0x%x", ifname, i->ifi_flags, i->ifi_change);
-				net_cond_set(ifname, "exist",   1);
-				net_cond_set(ifname, "up",      i->ifi_flags & IFF_UP);
-				net_cond_set(ifname, "running", i->ifi_flags & IFF_RUNNING);
-				break;
+	for (; RTA_OK(a, la); a = RTA_NEXT(a, la)) {
+		if (a->rta_type != IFLA_IFNAME)
+			continue;
 
-			case RTM_DELLINK:
-				/* NOTE: Interface has disappeared, not link down ... */
-				_d("%s: Delete link", ifname);
-				net_cond_set(ifname, "exist",   0);
-				net_cond_set(ifname, "up",      0);
-				net_cond_set(ifname, "running", 0);
-				break;
+		strlcpy(ifname, RTA_DATA(a), sizeof(ifname));
+		if (validate_ifname(ifname)) {
+			_d("Invalid interface name '%s', skipping ...", ifname);
+			continue;
+		}
 
-			case RTM_NEWADDR:
-				_d("%s: New Address", ifname);
-				break;
+		switch (nlmsg->nlmsg_type) {
+		case RTM_NEWLINK:
+			/*
+			 * New interface has appeared, or interface flags has changed.
+			 * Check ifi_flags here to see if the interface is UP/DOWN
+			 */
+			_d("%s: New link, flags 0x%x, change 0x%x", ifname, i->ifi_flags, i->ifi_change);
+			net_cond_set(ifname, "exist",   1);
+			net_cond_set(ifname, "up",      i->ifi_flags & IFF_UP);
+			net_cond_set(ifname, "running", i->ifi_flags & IFF_RUNNING);
+			if (!(i->ifi_flags & IFF_UP) || !(i->ifi_flags & IFF_RUNNING))
+				nl_check_default(ifname);
+			break;
 
-			case RTM_DELADDR:
-				_d("%s: Deconfig Address", ifname);
+		case RTM_DELLINK:
+			/* NOTE: Interface has disappeared, not link down ... */
+			_d("%s: Delete link", ifname);
+			net_cond_set(ifname, "exist",   0);
+			net_cond_set(ifname, "up",      0);
+			net_cond_set(ifname, "running", 0);
+			nl_check_default(ifname);
+			break;
+
+		case RTM_NEWADDR:
+			_d("%s: New Address", ifname);
+			break;
+
+		case RTM_DELADDR:
+			_d("%s: Deconfig Address", ifname);
+			break;
+
+		default:
+			_d("%s: Msg 0x%x", ifname, nlmsg->nlmsg_type);
+			break;
+		}
+	}
+}
+
+static int nl_parse(int sd)
+{
+	while (1) {
+		struct nlmsghdr *nh;
+		ssize_t len;
+
+		while ((len = recv(sd, nl_buf, NL_BUFSZ, 0)) < 0) {
+			switch (errno) {
+			case EAGAIN:	/* Nothing more right now. */
+				return 0;
+
+			case EINTR:	/* Signal */
+				continue;
+
+			case ENOBUFS:	/* netlink(7) */
 				break;
 
 			default:
-				_d("%s: Msg 0x%x", ifname, nlmsg->nlmsg_type);
+				_pe("recv()");
+				break;
+			}
+
+			return -1;
+		}
+
+		_d("recv %lld bytes", len);
+
+		for (nh = (struct nlmsghdr *)nl_buf; NLMSG_OK(nh, len); nh = NLMSG_NEXT(nh, len)) {
+			struct nlmsgerr *nle;
+
+			switch (nh->nlmsg_type) {
+			case NLMSG_DONE:
+				_d("Done with netlink messages.");
+				return 0;
+
+			case NLMSG_ERROR:
+				_d("Kernel netlink comm. error.");
+				nle = NLMSG_DATA(nh);
+				if (nle) {
+					errno = -nle->error;
+					_pe("Kernel netlink error %d", errno);
+				}
+				return -1;
+
+			case RTM_NEWROUTE:
+			case RTM_DELROUTE:
+				_d("Netlink route ...");
+				nl_route(nh, len);
+				break;
+
+			case RTM_NEWLINK:
+			case RTM_DELLINK:
+				_d("Netlink link ...");
+				nl_link(nh, len);
+				break;
+
+			default:
+				_w("unhandled netlink message, type %d", nh->nlmsg_type);
 				break;
 			}
 		}
-
-		a = RTA_NEXT(a, la);
 	}
 }
 
+static int nl_request(int sd, unsigned int seq, int type)
+{
+	struct nl_request *nlr = (struct nl_request *)nl_buf;
+
+	memset(nlr, 0, sizeof(struct nl_request));
+	nlr->nh.nlmsg_type  = type;
+	nlr->nh.nlmsg_flags = NLM_F_DUMP | NLM_F_REQUEST;
+	nlr->nh.nlmsg_seq   = seq;
+	nlr->nh.nlmsg_pid   = 1;
+
+	switch (type) {
+	case RTM_GETROUTE:
+		_d("RTM_GETROUTE");
+		nlr->rtm.rtm_family = AF_INET;
+		nlr->rtm.rtm_table  = RT_TABLE_MAIN;
+		nlr->nh.nlmsg_len   = NLMSG_LENGTH(sizeof(struct rtmsg));
+		break;
+
+	case RTM_GETLINK:
+		_d("RTM_GETLINK");
+		nlr->ifi.ifi_family = AF_UNSPEC;
+		nlr->ifi.ifi_change = 0xFFFFFFFF;
+		nlr->nh.nlmsg_len   = NLMSG_LENGTH(sizeof(struct ifinfomsg));
+		break;
+
+	default:
+		_w("Cannot resync, unhandled message type %d", type);
+		return -1;
+	}
+
+	if (send(sd, nlr, nlr->nh.nlmsg_len, 0) < 0)
+		return 1;
+
+	return nl_parse(sd);
+}
+
+static void nl_resync_routes(int sd, unsigned int seq)
+{
+	if (nl_request(sd, seq, RTM_GETROUTE))
+		_pe("Failed netlink route request");
+}
+
+static void nl_resync_ifaces(int sd, unsigned int seq)
+{
+	if (nl_request(sd, seq, RTM_GETLINK))
+		_pe("Failed netlink link request");
+}
+
+/*
+ * We've potentially lost netlink events, let's resync with kernel.
+ */
+static void nl_resync(int all)
+{
+	unsigned int seq = 0;
+	int sd;
+
+	sd = socket(AF_NETLINK, SOCK_DGRAM, NETLINK_ROUTE);
+	if (sd < 0) {
+		_pe("netlink socket");
+		return;
+	}
+
+	if (all) {
+		_d("============================ RESYNC =================================");
+		/* this doesn't update condtions, and thus does not stop services */
+		cond_deassert("net/");
+
+		nl_resync_ifaces(sd, seq++);
+		nl_resync_routes(sd, seq++);
+
+		/* delayed update after we've corrected things */
+		service_step_all(SVC_TYPE_ANY);
+		_d("=========================== RESYNCED ================================");
+	} else
+		nl_resync_routes(sd, seq++);
+
+	close(sd);
+}
+
 static void nl_callback(void *arg, int sd, int events)
 {
-	ssize_t len;
-	static char buf[4096];
-	struct nlmsghdr *nh;
-
-	memset(buf, 0, sizeof(buf));
-	len = recv(sd, buf, sizeof(buf), 0);
-	if (len < 0) {
-		if (errno != EINTR)	/* Signal */
-			_pe("recv()");
-		return;
+	if (nl_parse(sd) < 0) {
+		if (errno == ENOBUFS) {	/* netlink(7) */
+			_w("busy system, resynchronizing with kernel.");
+			nl_resync(1);
+			return;
+		}
 	}
 
-	for (nh = (struct nlmsghdr *)buf; !nlmsg_validate(nh, len); nh = NLMSG_NEXT(nh, len)) {
-		//_d("Well formed netlink message received. type %d ...", nh->nlmsg_type);
-		if (nh->nlmsg_type == RTM_NEWROUTE || nh->nlmsg_type == RTM_DELROUTE)
-			nl_route(nh);
-		else
-			nl_link(nh);
+	/*
+	 * Linux doesn't send route changes when interfaces go down, so
+	 * we need to check ourselves, e.g. for loss of default route.
+	 */
+	if (nl_ifdown) {
+		_d("interface down, checking default route.");
+		if (nl_defidx > 0) {
+			nl_defidx = 0;
+			nl_resync(0);
+			if (nl_defidx <= 0) {
+				cond_clear("net/route/default");
+				nl_defidx = 0;
+			}
+		}
+
+		nl_ifdown = 0;
 	}
 }
 
@@ -201,8 +414,8 @@ static plugin_t plugin = {
 
 PLUGIN_INIT(plugin_init)
 {
-	int sd;
 	struct sockaddr_nl sa;
+	int sd;
 
 	sd = socket(AF_NETLINK, SOCK_RAW | SOCK_NONBLOCK | SOCK_CLOEXEC, NETLINK_ROUTE);
 	if (sd < 0) {
@@ -212,7 +425,7 @@ PLUGIN_INIT(plugin_init)
 
 	memset(&sa, 0, sizeof(sa));
 	sa.nl_family = AF_NETLINK;
-	sa.nl_groups = RTMGRP_IPV4_ROUTE | RTMGRP_LINK; // | RTMGRP_NOTIFY | RTMGRP_IPV4_IFADDR;
+	sa.nl_groups = RTMGRP_IPV4_ROUTE | RTMGRP_LINK;
 	sa.nl_pid    = getpid();
 
 	if (bind(sd, (struct sockaddr *)&sa, sizeof(sa)) < 0) {
@@ -221,6 +434,13 @@ PLUGIN_INIT(plugin_init)
 		return;
 	}
 
+	nl_buf = malloc(NL_BUFSZ);
+	if (!nl_buf) {
+		_pe("malloc()");
+		close(sd);
+		return;
+	}
+
 	plugin.io.fd = sd;
 	plugin_register(&plugin);
 }

plugins/pidfile.c

@@ -58,7 +58,7 @@ static int pidfile_add_path(struct iwatch *iw, char *path)
 		}
 	}
 
-	return iwatch_add(iw, path, IN_ONLYDIR);
+	return iwatch_add(iw, path, IN_ONLYDIR | IN_CLOSE_WRITE);
 }
 
 static void pidfile_update_conds(char *dir, char *name, uint32_t mask)
@@ -68,11 +68,11 @@ static void pidfile_update_conds(char *dir, char *name, uint32_t mask)
 	svc_t *svc;
 
 	paste(fn, sizeof(fn), dir, name);
-	_d("path: %s, mask: %08x", fn, mask);
-
 	if (fnmatch("*\\.pid", fn, 0) && fnmatch("*/pid", fn, 0))
 		return;
 
+	_d("path: %s, mask: %08x", fn, mask);
+
 	svc = svc_find_by_pidfile(fn);
 	if (!svc) {
 		_d("No matching svc for %s", fn);
@@ -82,15 +82,25 @@ static void pidfile_update_conds(char *dir, char *name, uint32_t mask)
 	_d("Found svc %s for %s with pid %d", svc->name, fn, svc->pid);
 
 	mkcond(svc, cond, sizeof(cond));
-	if (mask & (IN_CREATE | IN_ATTRIB | IN_MODIFY | IN_MOVED_TO)) {
+	if (mask & (IN_CLOSE_WRITE | IN_ATTRIB | IN_MODIFY | IN_MOVED_TO)) {
 		svc_started(svc);
+		if (!svc_has_pidfile(svc))
+			pid_file_set(svc, fn, 1);
+
 		if (svc_is_forking(svc)) {
 			pid_t pid;
 
 			pid = pid_file_read(pid_file(svc));
-			_d("Forking service %s changed PID from %d to %d",
-			   svc->cmd, svc->pid, pid);
-			svc->pid = pid;
+			if (pid == -1) {
+				cond_clear(cond);
+				return;
+			}
+
+			if (pid != svc->pid) {
+				_d("Forking service %s changed PID from %d to %d",
+				   svc->cmd, svc->pid, pid);
+				svc->pid = pid;
+			}
 		}
 
 		cond_set(cond);
@@ -151,7 +161,8 @@ static void pidfile_callback(void *arg, int fd, int events)
 {
 	static char ev_buf[8 *(sizeof(struct inotify_event) + NAME_MAX + 1) + 1];
 	struct inotify_event *ev;
-	ssize_t sz, off;
+	ssize_t sz;
+	size_t off;
 
 	sz = read(fd, ev_buf, sizeof(ev_buf) - 1);
 	if (sz <= 0) {
@@ -160,15 +171,19 @@ static void pidfile_callback(void *arg, int fd, int events)
 	}
 	ev_buf[sz] = 0;
 
-	for (off = 0; off < sz; off += sizeof(*ev) + ev->len) {
+	for (off = 0; off < (size_t)sz; off += sizeof(*ev) + ev->len) {
 		struct iwatch_path *iwp;
 
+		if (off + sizeof(*ev) > (size_t)sz)
+			break;
+
 		ev = (struct inotify_event *)&ev_buf[off];
+		if (off + sizeof(*ev) + ev->len > (size_t)sz)
+			break;
+
 		if (!ev->mask)
 			continue;
 
-		_d("name %s, event: 0x%08x", ev->name, ev->mask);
-
 		/* Find base path for this event */
 		iwp = iwatch_find_by_wd(&iw_pidfile, ev->wd);
 		if (!iwp)
@@ -179,12 +194,7 @@ static void pidfile_callback(void *arg, int fd, int events)
 			continue;
 		}
 
-		if (ev->mask & IN_DELETE) {
-			_d("pidfile %s/%s removed ...", iwp->path, ev->name);
-			continue;
-		}
-
-		if (ev->mask & (IN_CREATE | IN_ATTRIB | IN_MODIFY | IN_MOVED_TO))
+		if (ev->mask & (IN_CLOSE_WRITE | IN_DELETE | IN_ATTRIB | IN_MODIFY | IN_MOVED_TO))
 			pidfile_update_conds(iwp->path, ev->name, ev->mask);
 	}
 }
@@ -226,8 +236,14 @@ static void pidfile_reconf(void *arg)
 
 static void pidfile_init(void *arg)
 {
+	char piddir[MAX_ARG_LEN];
 	char *path;
 
+	if (mkpath(pid_runpath(_PATH_CONDPID, piddir, sizeof(piddir)), 0755) && errno != EEXIST) {
+		_pe("Failed creating %s condition directory, %s", COND_PID, _PATH_CONDPID);
+		return;
+	}
+
 	/*
 	 * The bootmisc plugin is responsible for setting up /var/run.
 	 * and/or /run, with proper symlinks etc.  We depend on bootmisc
@@ -235,6 +251,7 @@ static void pidfile_init(void *arg)
 	 */
 	path = realpath(_PATH_VARRUN, NULL);
 	if (!path) {
+		_d("Failed querying realpath(%s): %s", _PATH_VARRUN, strerror(errno));
 		path = realpath("/run", NULL);
 		if (!path) {
 			_e("System does not have %s or /run, aborting.", _PATH_VARRUN);
@@ -244,6 +261,8 @@ static void pidfile_init(void *arg)
 
 	if (pidfile_add_path(&iw_pidfile, path))
 		iwatch_exit(&iw_pidfile);
+
+	free(path);
 }
 
 /*

plugins/procps.c

@@ -23,6 +23,9 @@
 
 #include <glob.h>
 #include <lite/lite.h>
+
+#include "finit.h"
+#include "helpers.h"
 #include "plugin.h"
 
 static void setup(void *arg)
@@ -31,6 +34,11 @@ static void setup(void *arg)
 	glob_t gl;
 	char buf[160];
 
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	glob("/run/sysctl.d/*.conf",           0, NULL, &gl);
 	glob("/etc/sysctl.d/*.conf",           GLOB_APPEND, NULL, &gl);
 	glob("/usr/local/lib/sysctl.d/*.conf", GLOB_APPEND, NULL, &gl);

plugins/rtc.c

@@ -84,6 +84,11 @@ static void rtc_save(void *arg)
 	int fd, rc = 0;
 	char tz[128];
 
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	fd = rtc_open();
 	if (fd < 0)
 		return;
@@ -121,6 +126,11 @@ static void rtc_restore(void *arg)
 	int fd, rc = 0;
 	char tz[128];
 
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	fd = rtc_open();
 	if (fd < 0)
 		return;

plugins/sys.c

@@ -0,0 +1,197 @@
+/* System condtion event monitor for the Finit condition engine
+ *
+ * Copyright (c) 2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include <dirent.h>
+#include <fnmatch.h>
+#include <glob.h>
+#include <limits.h>
+#include <paths.h>
+
+#include "finit.h"
+#include "cond.h"
+#include "pid.h"
+#include "plugin.h"
+#include "iwatch.h"
+
+static struct iwatch iw_sys;
+
+
+static int sys_add_path(struct iwatch *iw, char *path)
+{
+	return iwatch_add(iw, path, IN_ONLYDIR | IN_CLOSE_WRITE);
+}
+
+static void sys_update_conds(char *dir, char *name, uint32_t mask)
+{
+	char fn[PATH_MAX];
+	char *cond;
+
+	paste(fn, sizeof(fn), dir, name);
+	cond = strstr(fn, COND_BASE);
+	if (!cond)
+		return;
+
+	cond += strlen(COND_BASE) + 1;
+	_d("cond: %s set: %d", cond, mask & IN_CREATE ? 1 : 0);
+	cond_update(cond);
+}
+
+/* synthesize events in case of new run dirs */
+static void sys_scandir(struct iwatch *iw, char *dir, int len)
+{
+	glob_t gl;
+	size_t i;
+	char path[len + 6];
+	int rc;
+
+	snprintf(path, sizeof(path), "%s/*", dir);
+	rc = glob(path, GLOB_NOSORT, NULL, &gl);
+	if (rc && rc != GLOB_NOMATCH)
+		return;
+
+	for (i = 0; i < gl.gl_pathc; i++) {
+		_d("scan found %s", gl.gl_pathv[i]);
+		sys_update_conds(dir, gl.gl_pathv[i], IN_CREATE);
+	}
+	globfree(&gl);
+}
+
+/*
+ * create/remove sub-directory in monitored directory
+ */
+static void sys_handle_dir(struct iwatch *iw, char *dir, char *name, int mask)
+{
+	char path[strlen(dir) + strlen(name) + 2];
+	struct iwatch_path *iwp;
+
+	paste(path, sizeof(path), dir, name);
+	_d("path: %s", path);
+
+	iwp = iwatch_find_by_path(iw, path);
+
+	if (mask & IN_CREATE) {
+		if (!iwp) {
+			if (!sys_add_path(iw, path))
+				sys_scandir(iw, path, sizeof(path));
+		}
+	} else if (mask & IN_DELETE) {
+		if (iwp)
+			iwatch_del(&iw_sys, iwp);
+	}
+}
+
+static void sys_callback(void *arg, int fd, int events)
+{
+	static char ev_buf[8 *(sizeof(struct inotify_event) + NAME_MAX + 1) + 1];
+	struct inotify_event *ev;
+	ssize_t sz;
+	size_t off;
+
+	sz = read(fd, ev_buf, sizeof(ev_buf) - 1);
+	if (sz <= 0) {
+		_pe("invalid inotify event");
+		return;
+	}
+	ev_buf[sz] = 0;
+
+	for (off = 0; off < (size_t)sz; off += sizeof(*ev) + ev->len) {
+		struct iwatch_path *iwp;
+
+		if (off + sizeof(*ev) > (size_t)sz)
+			break;
+
+		ev = (struct inotify_event *)&ev_buf[off];
+		if (off + sizeof(*ev) + ev->len > (size_t)sz)
+			break;
+
+		_d("name %s, event: 0x%08x", ev->name, ev->mask);
+		if (!ev->mask)
+			continue;
+
+		/* Find base path for this event */
+		iwp = iwatch_find_by_wd(&iw_sys, ev->wd);
+		if (!iwp)
+			continue;
+
+		if (ev->mask & IN_ISDIR) {
+			sys_handle_dir(&iw_sys, iwp->path, ev->name, ev->mask);
+			continue;
+		}
+
+		if (ev->mask & (IN_CREATE | IN_DELETE))
+			sys_update_conds(iwp->path, ev->name, ev->mask);
+	}
+}
+
+static void sys_init(void *arg)
+{
+	char sysdir[MAX_ARG_LEN];
+	char *path;
+
+	if (mkpath(pid_runpath(_PATH_CONDSYS, sysdir, sizeof(sysdir)), 0755) && errno != EEXIST) {
+		_pe("Failed creating %s condition directory, %s", COND_SYS, _PATH_CONDSYS);
+		return;
+	}
+
+	path = realpath(_PATH_CONDSYS, NULL);
+	if (!path) {
+		_pe("Cannot figure out real path to %s, aborting", _PATH_CONDSYS);
+		return;
+	}
+
+	if (iwatch_add(&iw_sys, path, IN_ONLYDIR))
+		iwatch_exit(&iw_sys);
+}
+
+static plugin_t plugin = {
+	.name = __FILE__,
+	.hook[HOOK_BASEFS_UP]  = { .cb = sys_init },
+};
+
+PLUGIN_INIT(plugin_init)
+{
+	int fd;
+
+	fd = iwatch_init(&iw_sys);
+	if (fd < 0)
+		return;
+
+	plugin.io.fd = fd;
+	plugin.io.cb = sys_callback;
+	plugin.io.flags = PLUGIN_IO_READ;
+
+	plugin_register(&plugin);
+}
+
+PLUGIN_EXIT(plugin_exit)
+{
+	iwatch_exit(&iw_sys);
+	plugin_unregister(&plugin);
+}
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */

plugins/tty.c

@@ -32,13 +32,14 @@
 #include "finit.h"
 #include "helpers.h"
 #include "plugin.h"
+#include "service.h"
 #include "tty.h"
 
-static void watcher(void *arg, int fd, int events);
+static void tty_watcher(void *arg, int fd, int events);
 
 static plugin_t plugin = {
 	.io = {
-		.cb    = watcher,
+		.cb    = tty_watcher,
 		.flags = PLUGIN_IO_READ,
 	},
 };
@@ -53,33 +54,53 @@ static void setup(void)
 		_pe("Failed starting TTY watcher");
 }
 
-static void watcher(void *arg, int fd, int events)
+static void do_tty(char *tty, size_t len, int creat)
 {
-	char buf[EVENT_SIZE];
-	int len = 0;
+	char name[len + 6];
+	svc_t *svc;
 
-	while ((len = read(fd, buf, sizeof(buf)))) {
-		struct inotify_event *notified = (struct inotify_event *)buf;
-		char name[notified->len + 6];
-		struct tty *entry;
-
-		if (-1 == len) {
-			if (errno == EINVAL)
-				setup();
-			if (errno == EINTR)
-				continue;
-
-			break;	/* Likely EAGAIN */
+	snprintf(name, sizeof(name), "/dev/%s", tty);
+	svc = svc_find_by_tty(name);
+	if (svc) {
+		if (svc_is_blocked(svc) && creat) {
+			_d("%s: blocked, re-enabling", svc_ident(svc, NULL, 0));
+			svc_start(svc);
+		} else if (svc->pid) {
+			_d("%s: missing ...", svc_ident(svc, NULL, 0));
+			svc_missing(svc);
 		}
 
-		snprintf(name, sizeof(name), "/dev/%s", notified->name);
-		entry = tty_find(name);
-		if (entry && tty_enabled(entry)) {
-			if (notified->mask & IN_CREATE)
-				tty_start(entry);
-			else if (entry->pid)
-				tty_stop(entry);
-		}
+		service_step_all(SVC_TYPE_TTY);
+	}
+}
+
+static void tty_watcher(void *arg, int fd, int events)
+{
+	static char ev_buf[8 *(sizeof(struct inotify_event) + NAME_MAX + 1) + 1];
+	struct inotify_event *ev;
+	ssize_t sz;
+	size_t off;
+
+	sz = read(fd, ev_buf, sizeof(ev_buf) - 1);
+	if (sz <= 0) {
+		_pe("invalid inotify event");
+		return;
+	}
+	ev_buf[sz] = 0;
+
+	for (off = 0; off < (size_t)sz; off += sizeof(*ev) + ev->len) {
+		if (off + sizeof(*ev) > (size_t)sz)
+			break;
+
+		ev = (struct inotify_event *)&ev_buf[off];
+		if (off + sizeof(*ev) + ev->len > (size_t)sz)
+			break;
+
+		if (!ev->mask)
+			continue;
+
+		_d("tty %s, event: 0x%08x", ev->name, ev->mask);
+		do_tty(ev->name, ev->len, ev->mask & IN_CREATE);
 	}
 }
 

plugins/urandom.c

@@ -35,12 +35,18 @@
 static void setup(void *arg)
 {
 #ifdef RANDOMSEED
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
 	if (!fexist(RANDOMSEED)) {
 		int ret = 1;
+		mode_t prev;
 		FILE *fp;
 
 		print_desc("Bootstrapping random seed", NULL);
-		umask(077);
+		prev = umask(077);
 		fp = fopen(RANDOMSEED, "w");
 		if (fp) {
 			int iter = 128;
@@ -57,7 +63,7 @@ static void setup(void *arg)
 			ret = fclose(fp);
 		}
 		print_result(ret);
-		umask(0);
+		umask(prev);
 	}
 
 	print_desc("Initializing random number generator", NULL);
@@ -68,10 +74,19 @@ static void setup(void *arg)
 static void save(void *arg)
 {
 #ifdef RANDOMSEED
-	umask(077);
+	mode_t prev;
+
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
+	prev = umask(077);
+
 	print_desc("Saving random seed", NULL);
 	print_result(512 != copyfile("/dev/urandom", RANDOMSEED, 512, 0));
-	umask(0);
+
+	umask(prev);
 #endif
 }
 

plugins/usr.c

@@ -0,0 +1,133 @@
+/* User-defined condtion event monitor for the Finit condition engine
+ *
+ * Copyright (c) 2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include <fnmatch.h>
+#include <glob.h>
+#include <limits.h>
+#include <paths.h>
+
+#include "finit.h"
+#include "cond.h"
+#include "pid.h"
+#include "plugin.h"
+#include "iwatch.h"
+
+static struct iwatch iw_usr;
+
+
+static void usr_cond(char *name, uint32_t mask)
+{
+	char cond[MAX_COND_LEN] = COND_USR;
+
+	strlcat(cond, name, sizeof(cond));
+	cond_update(cond);
+}
+
+static void usr_callback(void *arg, int fd, int events)
+{
+	static char ev_buf[8 *(sizeof(struct inotify_event) + NAME_MAX + 1) + 1];
+	struct inotify_event *ev;
+	ssize_t sz;
+	size_t off;
+
+	sz = read(fd, ev_buf, sizeof(ev_buf) - 1);
+	if (sz <= 0) {
+		_pe("invalid inotify event");
+		return;
+	}
+	ev_buf[sz] = 0;
+
+	for (off = 0; off < (size_t)sz; off += sizeof(*ev) + ev->len) {
+		if (off + sizeof(*ev) > (size_t)sz)
+			break;
+
+		ev = (struct inotify_event *)&ev_buf[off];
+		if (off + sizeof(*ev) + ev->len > (size_t)sz)
+			break;
+
+		_d("name %s, event: 0x%08x", ev->name, ev->mask);
+		if (!ev->mask)
+			continue;
+
+		if (ev->mask & IN_ISDIR)
+			continue;	/* unsupported */
+
+		if (ev->mask & IN_DELETE)
+			sync();		/* dont ask */
+
+		usr_cond(ev->name, ev->mask);
+	}
+}
+
+static void usr_init(void *arg)
+{
+	char usrdir[MAX_ARG_LEN];
+	char *path;
+
+	if (mkpath(pid_runpath(_PATH_CONDUSR, usrdir, sizeof(usrdir)), 0755) && errno != EEXIST) {
+		_pe("Failed creating %s condition directory, %s", COND_USR, _PATH_CONDUSR);
+		return;
+	}
+
+	path = realpath(_PATH_CONDUSR, NULL);
+	if (!path) {
+		_pe("Cannot figure out real path to %s, aborting", _PATH_CONDUSR);
+		return;
+	}
+
+	if (iwatch_add(&iw_usr, path, IN_ONLYDIR))
+		iwatch_exit(&iw_usr);
+}
+
+static plugin_t plugin = {
+	.name = __FILE__,
+	.hook[HOOK_BASEFS_UP]  = { .cb = usr_init },
+};
+
+PLUGIN_INIT(plugin_init)
+{
+	int fd;
+
+	fd = iwatch_init(&iw_usr);
+	if (fd < 0)
+		return;
+
+	plugin.io.fd = fd;
+	plugin.io.cb = usr_callback;
+	plugin.io.flags = PLUGIN_IO_READ;
+
+	plugin_register(&plugin);
+}
+
+PLUGIN_EXIT(plugin_exit)
+{
+	iwatch_exit(&iw_usr);
+	plugin_unregister(&plugin);
+}
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */

plugins/x11-common.c

@@ -32,13 +32,19 @@
 
 static void setup(void *arg)
 {
+	const char *username = "nobody";
+	char line[LINE_SIZE];
+	mode_t prev;
 #ifdef PAM_CONSOLE
 	int fd;
 #endif
-	char line[LINE_SIZE];
-	const char *username = "nobody";
 
-	umask(0);
+	if (rescue) {
+		_d("Skipping %s plugin in rescue mode.", __FILE__);
+		return;
+	}
+
+	prev = umask(0);
 
 	makedir("/var/run/console", 01777);
 	snprintf(line, sizeof(line), "/var/run/console/%s", username);
@@ -58,7 +64,7 @@ static void setup(void *arg)
 	if (whichp("restorecon"))
 		run("restorecon /tmp/.ICE-unix /tmp/.X11-unix");
 
-	umask(022);
+	umask(prev);
 }
 
 static plugin_t plugin = {

src/.gitignore

@@ -1,8 +1,10 @@
 .deps/*
 finit
 initctl
+keventd
 logit
 reboot
+sulogin
 watchdogd
 Makefile
 Makefile.in

src/Makefile.am

@@ -1,6 +1,5 @@
-EXTRA_DIST         = cgreaper.sh rescue.conf sample.conf
-AM_CPPFLAGS        = -U_FORTIFY_SOURCE
-AM_CPPFLAGS       += -D_XOPEN_SOURCE=600 -D_BSD_SOURCE -D_GNU_SOURCE -D_DEFAULT_SOURCE
+EXTRA_DIST         = rescue.conf sample.conf
+AM_CPPFLAGS        = -D_XOPEN_SOURCE=600 -D_BSD_SOURCE -D_GNU_SOURCE -D_DEFAULT_SOURCE
 if STATIC
 AM_CPPFLAGS       += -DENABLE_STATIC
 AM_LDFLAGS         = -static-libtool-libs
@@ -8,21 +7,27 @@ else
 AM_LDFLAGS         = -export-dynamic
 endif
 
-sbin_PROGRAMS      = finit initctl
-pkglibexec_SCRIPTS = cgreaper.sh
+sbin_PROGRAMS        = finit initctl
+pkglibexec_PROGRAMS  = keventd logit sulogin
+if WATCHDOGD
+pkglibexec_PROGRAMS += watchdogd
+endif
 
 finit_pkglibdir      = $(pkglibdir)
 finit_pkglib_SCRIPTS = rescue.conf sample.conf
 
-if WATCHDOGD
-pkglibexec_PROGRAMS = watchdogd
-endif
+keventd_SOURCES      = keventd.c iwatch.c iwatch.h util.c util.h
+keventd_CFLAGS       = -W -Wall -Wextra -std=gnu99
+keventd_LDADD        = $(lite_LIBS)
 
-if LOGIT
-pkglibexec_PROGRAMS = logit
-logit_SOURCES       = logit.c
-logit_CFLAGS        = -W -Wall -Wextra -Wno-unused-parameter -std=gnu99
-endif
+sulogin_SOURCES      = sulogin.c
+sulogin_CFLAGS       = -W -Wall -Wextra -std=gnu99
+sulogin_LDADD        = -lcrypt
+
+logit_SOURCES        = logit.c logrotate.c
+logit_CFLAGS         = -W -Wall -Wextra -Wno-unused-parameter -std=gnu99
+logit_CFLAGS        += $(lite_CFLAGS)
+logit_LDADD          = $(lite_LIBS)
 
 finit_SOURCES      = api.c	cgroup.c	cgroup.h	\
 		     cond.c	cond-w.c	cond.h		\
@@ -43,7 +48,11 @@ finit_SOURCES      = api.c	cgroup.c	cgroup.h	\
 		     tty.c	tty.h				\
 		     util.c	util.h				\
 		     utmp-api.c	utmp-api.h
-pkginclude_HEADERS = cond.h finit.h helpers.h log.h plugin.h svc.h
+if LOGROTATE
+finit_SOURCES     += logrotate.c
+endif
+
+pkginclude_HEADERS = cgroup.h cond.h finit.h helpers.h log.h plugin.h svc.h
 
 finit_CFLAGS       = -W -Wall -Wextra -Wno-unused-parameter -std=gnu99
 finit_CFLAGS      += $(lite_CFLAGS) $(uev_CFLAGS)
@@ -54,20 +63,19 @@ else
 finit_LDADD       += -ldl
 endif
 
-initctl_SOURCES    = initctl.c client.c client.h 	\
-		     reboot.c serv.c serv.h svc.h	\
-		     cond.c cond.h util.c util.h
+initctl_SOURCES    = initctl.c initctl.h cgutil.c cgutil.h		\
+		     client.c client.h cond.c cond.h reboot.c		\
+		     serv.c serv.h svc.h util.c util.h
 initctl_CFLAGS     = -W -Wall -Wextra -Wno-unused-parameter -std=gnu99
-initctl_CFLAGS    += $(lite_CFLAGS)
-initctl_LDADD      = $(lite_LIBS)
+initctl_CFLAGS    += $(lite_CFLAGS) $(uev_CFLAGS)
+initctl_LDADD      = $(lite_LIBS) $(uev_LIBS)
 
-INIT_LNKS          = init
+INIT_LNKS          = init telinit
 REBOOT_LNKS        = reboot shutdown halt poweroff suspend
 
 if WATCHDOGD
 watchdogd_SOURCES  = watchdog.c watchdog.h
 watchdogd_CFLAGS   = -W -Wall -Wextra -Wno-unused-parameter -std=gnu99
-#log.c log.h
 endif
 
 # Hook in install to add finit and reboot symlink(s)

src/api.c

@@ -74,7 +74,24 @@ static int start(svc_t *svc)
 	return 0;
 }
 
+/*
+ * NOTE: this does not wait for svc to be stopped first, that is the
+ *       responsibility of initctl to do.  Otherwise we'd block PID 1,
+ *       or introduce some nasty race conditions.
+ */
 static int restart(svc_t *svc)
+{
+	if (!svc)
+		return 1;
+
+	svc_mark_dirty(svc);
+	svc_start(svc);
+	service_step(svc);
+
+	return 0;
+}
+
+static int reload(svc_t *svc)
 {
 	if (!svc)
 		return 1;
@@ -88,9 +105,10 @@ static int restart(svc_t *svc)
 	return 0;
 }
 
-static int do_start  (char *buf, size_t len) { return call(start,   buf, len); }
 static int do_stop   (char *buf, size_t len) { return call(stop,    buf, len); }
+static int do_start  (char *buf, size_t len) { return call(start,   buf, len); }
 static int do_restart(char *buf, size_t len) { return call(restart, buf, len); }
+static int do_reload (char *buf, size_t len) { return call(reload,  buf, len); }
 
 static char query_buf[368];
 static int missing(char *job, char *id)
@@ -110,7 +128,7 @@ static int missing(char *job, char *id)
 
 static int do_query(char *buf, size_t len)
 {
-	memset(query_buf, 0, sizeof(query_buf));
+	query_buf[0] = 0;
 	if (svc_parse_jobstr(buf, len, NULL, missing)) {
 		memcpy(buf, query_buf, len);
 		return 1;
@@ -237,7 +255,7 @@ ev_t ev_list[] = {
 
 static void send_svc(int sd, svc_t *svc)
 {
-	svc_t empty;
+	svc_t empty = { 0 };
 	size_t len;
 
 	if (!svc) {
@@ -252,10 +270,10 @@ static void send_svc(int sd, svc_t *svc)
 
 static void api_cb(uev_t *w, void *arg, int events)
 {
-	int sd, lvl;
-	svc_t *svc;
 	static svc_t *iter = NULL;
 	struct init_request rq;
+	int sd, lvl;
+	svc_t *svc;
 
 	sd = accept(w->fd, NULL, NULL);
 	if (sd < 0) {
@@ -327,16 +345,22 @@ static void api_cb(uev_t *w, void *arg, int events)
 			result = do_start(rq.data, sizeof(rq.data));
 			break;
 
+		case INIT_CMD_RESTART_SVC:
+			_d("restart %s", rq.data);
+			strterm(rq.data, sizeof(rq.data));
+			result = do_restart(rq.data, sizeof(rq.data));
+			break;
+
 		case INIT_CMD_STOP_SVC:
 			_d("stop %s", rq.data);
 			strterm(rq.data, sizeof(rq.data));
 			result = do_stop(rq.data, sizeof(rq.data));
 			break;
 
-		case INIT_CMD_RESTART_SVC:
-			_d("restart %s", rq.data);
+		case INIT_CMD_RELOAD_SVC:
+			_d("reload %s", rq.data);
 			strterm(rq.data, sizeof(rq.data));
-			result = do_restart(rq.data, sizeof(rq.data));
+			result = do_reload(rq.data, sizeof(rq.data));
 			break;
 
 		case INIT_CMD_GET_RUNLEVEL:
@@ -364,24 +388,30 @@ static void api_cb(uev_t *w, void *arg, int events)
 			}
 
 			_e("Request to hand-over wdog ... to PID %d", rq.runlevel);
-			if (!svc_find_by_pid(rq.runlevel)) {
+			svc = svc_find_by_pid(rq.runlevel);
+			if (!svc) {
 				logit(LOG_ERR, "Cannot find PID %d, not registered.", rq.runlevel);
 				break;
 			}
 
-			/* Disable and allow Finit to collect bundled watchdog */
-			if (wdog) {
-				logit(LOG_NOTICE, "Stopping and removing %s (PID:%d)", wdog->cmd, wdog->pid);
-				stop(wdog);
+			if (wdog && wdog != svc) {
+				char name[32];
+
+				svc_ident(svc, name, sizeof(name));
+				logit(LOG_NOTICE, "Handing over wdog ctrl from %s[%d] to %s[%d]",
+				      svc_ident(wdog, NULL, 0), wdog->pid, name, svc->pid);
+
 				if (wdog->protect) {
-					wdog->protect = 0;
-					wdog->runlevels = 0;
+					logit(LOG_NOTICE, "Stopping and deleting built-in watchdog.");
+					stop(wdog);
+					svc_del(wdog);
 				}
 			}
+			wdog = svc;
 			break;
 
 		case INIT_CMD_SVC_ITER:
-			_d("svc iter, first: %d", rq.runlevel);
+//			_d("svc iter, first: %d", rq.runlevel);
 			/*
 			 * XXX: This severly limits the number of
 			 * simultaneous client connections, but will
@@ -444,7 +474,7 @@ int api_init(uev_ctx_t *ctx)
 	};
 
 	_d("Setting up external API socket ...");
-	sd = socket(AF_UNIX, SOCK_STREAM | SOCK_CLOEXEC, 0);
+	sd = socket(AF_UNIX, SOCK_SEQPACKET | SOCK_CLOEXEC, 0);
 	if (-1 == sd) {
 		_pe("Failed starting external API socket");
 		return 1;

src/cgreaper.sh

@@ -1,11 +0,0 @@
-#!/bin/sh
-# Called by kernel when last child in cgroup exists
-
-BASE=/sys/fs/cgroup/finit
-PROC=`basename $1`
-DIR="$BASE$1"
-
-/libexec/finit/logit -p daemon.info -t finit "$PROC stopped, cleaning up control group $DIR"
-rmdir $DIR
-
-exit 0

src/cgroup.c

@@ -25,111 +25,461 @@
 #include <stdio.h>
 #include <string.h>
 #include <lite/lite.h>
+#include <lite/queue.h>
 #include <sys/mount.h>
+#include <sys/sysinfo.h>		/* get_nprocs_conf() */
 
+#include "cgroup.h"
+#include "finit.h"
+#include "iwatch.h"
 #include "log.h"
 #include "util.h"
 
-static int cg_init = 0;
+struct cg {
+	TAILQ_ENTRY(cg) link;
 
-/*
- * Called by Finit at early boot to mount initial cgroups
- */
-void cgroup_init(void)
+	char *name;		/* top-level group name */
+	char *cfg;		/* kernel settings */
+
+	int  active;		/* for mark & sweep */
+	int  protected;		/* for init/, user/, & system/ */
+};
+
+static TAILQ_HEAD(, cg) cgroups = TAILQ_HEAD_INITIALIZER(cgroups);
+
+static char controllers[256];
+
+static struct iwatch iw_cgroup;
+static uev_t cgw;
+
+
+static void cgset(const char *path, char *ctrl, char *prop)
 {
-	FILE *fp;
-	char buf[80];
-	int opts = MS_NODEV | MS_NOEXEC | MS_NOSUID;
+	char *val;
 
-	fp = fopen("/proc/cgroups", "r");
-	if (!fp) {
-		_d("No cgroup support");
+	_d("path %s, ctrl %s, prop %s", path ?: "NIL", ctrl ?: "NIL", prop ?: "NIL");
+	if (!path || !ctrl) {
+		_e("Missing path or controller, skipping!");
 		return;
 	}
 
-	if (mount("none", "/sys/fs/cgroup", "tmpfs", opts, NULL)) {
-		_d("Failed mounting cgroups");
-		goto fail;
+	if (!prop) {
+		prop = strchr(ctrl, '.');
+		if (!prop) {
+			_e("Invalid cgroup ctrl syntax: %s", ctrl);
+			return;
+		}
+
+		*prop++ = 0;
 	}
 
-	/* Skip first line, header */
-	fgets(buf, sizeof(buf), fp);
-	while (fgets(buf, sizeof(buf), fp)) {
-		char *cgroup;
-		char rc[80];
+	val = strchr(prop, ':');
+	if (!val) {
+		_e("Missing cgroup ctrl value, prop %s", prop);
+		return;
+	}
+	*val++ = 0;
 
-		cgroup = strtok(buf, "\t ");
-		if (!cgroup)
-			continue;
-
-		snprintf(rc, sizeof(rc), "/sys/fs/cgroup/%s", cgroup);
-		if (mkdir(rc, 0755) && EEXIST != errno)
-			continue;
-		if (mount("cgroup", rc, "cgroup", opts, cgroup))
-			_d("Failed mounting %s cgroup on %s", cgroup, rc);
+	/* disallow sneaky relative paths */
+	if (strstr(ctrl, "..") || strstr(prop, "..")) {
+		_e("Possible security violation; '..' not allowed in cgroup config!");
+		return;
 	}
 
-	/* Default cgroups for process monitoring */
-	if (mkdir("/sys/fs/cgroup/finit", 0755) && EEXIST != errno)
-		goto fail;
-	if (mount("none", "/sys/fs/cgroup/finit", "cgroup", opts, "none,name=finit")) {
-		_pe("Failed mounting Finit cgroup hierarchy");
-		goto fail;
-	}
-
-	/* Set up reaper and enable callbacks */
-	echo("/sys/fs/cgroup/finit/release_agent", 0, FINIT_LIBPATH_ "/cgreaper.sh");
-	echo("/sys/fs/cgroup/finit/notify_on_release", 0, "1");
-
-	/* Default groups, PID 1, services, and user/login processes */
-	if (mkdir("/sys/fs/cgroup/finit/init", 0755) && EEXIST != errno)
-		goto fail;
-	if (mkdir("/sys/fs/cgroup/finit/system", 0755) && EEXIST != errno)
-		goto fail;
-	if (mkdir("/sys/fs/cgroup/finit/user", 0755) && EEXIST != errno)
-		goto fail;
-
-	/* Move ourselves to init */
-	echo("/sys/fs/cgroup/finit/init/cgroup.procs", 0, "1");
-
-	/* We have signal, main screen turn on! */
-	cg_init = 1;
-fail:
-	fclose(fp);
+	_d("%s/%s.%s <= %s", path, ctrl, prop, val);
+	if (fnwrite(val, "%s/%s.%s", path, ctrl, prop))
+		_pe("Failed setting %s/%s.%s = %s", path, ctrl, prop, val);
 }
 
-static int move_pid(char *group, char *name, char *id, int pid)
+/*
+ * Settings for a cgroup are on the form: cpu.weight:1234,mem.max:4321,...
+ * Finit supports the short-form 'mem.', replacing it with 'memory.' when
+ * writing the setting to the file system.
+ */
+static void group_init(char *path, int leaf, const char *cfg)
+{
+	char *ptr, *s;
+
+	_d("path %s, leaf %d, cfg %s", path, leaf, cfg ?: "NIL");
+	if (!fisdir(path)) {
+		if (mkdir(path, 0755)) {
+			_pe("Failed creating cgroup %s", path);
+			return;
+		}
+
+		/* enable detected controllers on domain groups */
+		if (!leaf && fnwrite(controllers, "%s/cgroup.subtree_control", path))
+			_pe("Failed enabling %s for %s", controllers, path);
+	}
+
+	if (!cfg || !cfg[0])
+		return;
+
+	s = strdupa(cfg);
+	if (!s) {
+		_pe("Failed activating cgroup cfg for %s", path);
+		return;
+	}
+
+	_d("%s <=> %s", path, s);
+	ptr = strtok(s, ",");
+	while (ptr) {
+		_d("ptr: %s", ptr);
+		if (!strncmp("mem.", ptr, 4))
+			cgset(path, "memory", &ptr[4]);
+		else
+			cgset(path, ptr, NULL);
+
+		ptr = strtok(NULL, ",");
+	}
+}
+
+static int cgroup_leaf_init(char *group, char *name, int pid, const char *cfg)
 {
 	char path[256];
 
-	snprintf(path, sizeof(path), "/sys/fs/cgroup/%s/%s:%s", group, name, id);
-	if (mkdir(path, 0755) && errno != EEXIST)
-		return 1;
-
-	strlcat(path, "/cgroup.procs", sizeof(path));
-
-	return echo(path, 0, "%d", pid);
-}
-
-int cgroup_user(char *name)
-{
-	if (!cg_init)
-		return 0;
-
-	return move_pid("finit/user", name, "0", getpid());
-}
-
-int cgroup_service(char *nm, char *id, int pid)
-{
-	if (!cg_init)
-		return 0;
-
-	if (pid <= 0) {
+	_d("group %s, name %s, pid %d, cfg %s", group, name, pid, cfg ?: "NIL");
+	if (pid < 0 || pid == 1) {
 		errno = EINVAL;
 		return 1;
 	}
 
-	return move_pid("finit/system", nm, id, pid);
+	/* create and initialize new group */
+	snprintf(path, sizeof(path), "/sys/fs/cgroup/%s/%s", group, name);
+	group_init(path, 1, cfg);
+
+	/* move process to new group */
+	if (fnwrite(str("%d", pid), "%s/cgroup.procs", path))
+		_pe("Failed moving pid %d to group %s", pid, path);
+
+	strlcat(path, "/cgroup.events", sizeof(path));
+
+	return iwatch_add(&iw_cgroup, path, 0);
+}
+
+int cgroup_user(char *name, int pid)
+{
+	return cgroup_leaf_init("user", name, pid, NULL);
+}
+
+int cgroup_service(char *name, int pid, struct cgroup *cg)
+{
+	char *group = "system";
+
+	if (cg && cg->name[0]) {
+		char path[256];
+
+		if (!strcmp(cg->name, "root"))
+			return fnwrite(str("%d", pid), FINIT_CGPATH "/cgroup.procs");
+
+		if (!strcmp(cg->name, "init"))
+			return fnwrite(str("%d", pid), FINIT_CGPATH "/init/cgroup.procs");
+
+		snprintf(path, sizeof(path), "/sys/fs/cgroup/%s", cg->name);
+		if (fisdir(path))
+			group = cg->name;
+	}
+
+	return cgroup_leaf_init(group, name, pid, cg ? cg->cfg : NULL);
+}
+
+static void append_ctrl(char *ctrl)
+{
+	if (controllers[0])
+		strlcat(controllers, " ", sizeof(controllers));
+
+	strlcat(controllers, "+", sizeof(controllers));
+	strlcat(controllers, ctrl, sizeof(controllers));
+}
+
+static void cgroup_handle_event(char *event, uint32_t mask)
+{
+	char path[strlen(event) + 1];
+	char buf[80];
+	char *ptr;
+	FILE *fp;
+
+	_d("event: '%s', mask: %08x", event, mask);
+	if (!(mask & IN_MODIFY))
+		return;
+
+	fp = fopen(event, "r");
+	if (!fp) {
+		_d("Failed opening %s, skipping ...", event);
+		return;
+	}
+
+	while (fgets(buf, sizeof(buf), fp)) {
+		if (strncmp(buf, "populated", 9))
+			continue;
+
+		chomp(buf);
+		if (atoi(&buf[10]))
+			break;
+
+		strlcpy(path, event, sizeof(path));
+		ptr = strrchr(path, '/');
+		if (ptr) {
+			*ptr = 0;
+			if (!cgroup_del(path)) {
+				/*
+				 * try with parent, top-level group, we
+				 * may get events out-of-order *sigh*
+				 */
+				ptr = strrchr(path, '/');
+				if (!ptr)
+					break;
+				*ptr = 0;
+				cgroup_del(path);
+			}
+		}
+
+		break;
+	}
+
+	fclose(fp);
+}
+
+static void cgroup_events_cb(uev_t *w, void *arg, int events)
+{
+	static char ev_buf[8 *(sizeof(struct inotify_event) + NAME_MAX + 1) + 1];
+	struct inotify_event *ev;
+	ssize_t sz;
+	size_t off;
+
+	sz = read(w->fd, ev_buf, sizeof(ev_buf) - 1);
+	if (sz <= 0) {
+		_pe("invalid inotify event");
+		return;
+	}
+	ev_buf[sz] = 0;
+
+	for (off = 0; off < (size_t)sz; off += sizeof(*ev) + ev->len) {
+		struct iwatch_path *iwp;
+
+		if (off + sizeof(*ev) > (size_t)sz)
+			break;
+
+		ev = (struct inotify_event *)&ev_buf[off];
+		if (off + sizeof(*ev) + ev->len > (size_t)sz)
+			break;
+
+		if (!ev->mask)
+			continue;
+
+		/* Find base path for this event */
+		iwp = iwatch_find_by_wd(&iw_cgroup, ev->wd);
+		if (!iwp || !iwp->path)
+			continue;
+
+		cgroup_handle_event(iwp->path, ev->mask);
+	}
+
+#ifdef ENABLE_AUTO_RELOAD
+	if (conf_any_change())
+		service_reload_dynamic();
+#endif
+}
+
+static struct cg *cgroup_find(char *name)
+{
+	struct cg *cg;
+
+	TAILQ_FOREACH(cg, &cgroups, link) {
+		if (strcmp(cg->name, name))
+			continue;
+
+		return cg;
+	}
+
+	return NULL;
+}
+
+/*
+ * Marks all unprotected cgroups for deletion (during reload)
+ */
+void cgroup_mark_all(void)
+{
+	struct cg *cg;
+
+	TAILQ_FOREACH(cg, &cgroups, link) {
+		if (cg->protected)
+			continue;
+
+		cg->active = 0;
+	}
+}
+
+/*
+ * Remove (try to) all unused cgroups
+ */
+void cgroup_cleanup(void)
+{
+	struct cg *cg, *tmp;
+	char path[256];
+
+	TAILQ_FOREACH_SAFE(cg, &cgroups, link, tmp) {
+		if (cg->active)
+			continue;
+
+		snprintf(path, sizeof(path), FINIT_CGPATH "/%s", cg->name);
+		cgroup_del(path);
+	}
+}
+
+/*
+ * Add, or update, settings for top-level cgroup
+ */
+int cgroup_add(char *name, char *cfg, int protected)
+{
+	struct cg *cg;
+
+	if (!name)
+		return -1;
+	if (!cfg)
+		cfg = "";
+
+	cg = cgroup_find(name);
+	if (!cg) {
+		cg = malloc(sizeof(struct cg));
+		if (!cg) {
+			_pe("Failed allocating 'struct cg' for %s", name);
+			return -1;
+		}
+		cg->name = strdup(name);
+		if (!cg->name) {
+			_pe("Failed setting cgroup name %s", name);
+			free(cg);
+			return -1;
+		}
+		TAILQ_INSERT_TAIL(&cgroups, cg, link);
+	} else
+		free(cg->cfg);
+
+	cg->cfg = strdup(cfg);
+	if (!cg->cfg) {
+		_pe("Failed add/update of cgroup %s", name);
+		TAILQ_REMOVE(&cgroups, cg, link);
+		free(cg->name);
+		free(cg);
+		return -1;
+	}
+	cg->protected = protected;
+	cg->active = 1;
+
+	return 0;
+}
+
+/*
+ * Remove inactive top-level cgroup
+ */
+int cgroup_del(char *dir)
+{
+	struct cg *cg;
+	char path[256];
+
+	TAILQ_FOREACH(cg, &cgroups, link) {
+		snprintf(path, sizeof(path), FINIT_CGPATH "/%s", cg->name);
+		if (strcmp(path, dir))
+			continue;
+
+		if (cg->active)
+			return -1;
+
+		break;
+	}
+
+	if (rmdir(dir) && errno != ENOENT) {
+		_d("Failed removing %s: %s", dir, strerror(errno));
+		return -1;
+	}
+
+	if (cg) {
+		TAILQ_REMOVE(&cgroups, cg, link);
+		free(cg->name);
+		free(cg->cfg);
+		free(cg);
+	}
+
+	return 0;
+}
+
+/* the top-level init cgroup is a leaf, that's ensured in cgroup_init() */
+void cgroup_config(void)
+{
+	struct cg *cg;
+
+	TAILQ_FOREACH(cg, &cgroups, link) {
+		char path[256];
+		int leaf = 0;
+
+		if (!cg->active)
+			continue;
+		if (!strcmp(cg->name, "init"))
+			leaf = 1;	/* reserved */
+
+		snprintf(path, sizeof(path), "%s/%s", FINIT_CGPATH, cg->name);
+		group_init(path, leaf, cg->cfg);
+
+		strlcat(path, "/cgroup.events", sizeof(path));
+		iwatch_add(&iw_cgroup, path, 0);
+	}
+}
+
+/*
+ * Called by Finit at early boot to mount initial cgroups
+ */
+void cgroup_init(uev_ctx_t *ctx)
+{
+	int opts = MS_NODEV | MS_NOEXEC | MS_NOSUID;
+	char buf[80];
+	FILE *fp;
+	int fd;
+
+	if (mount("none", FINIT_CGPATH, "cgroup2", opts, NULL)) {
+		_pe("Failed mounting cgroup v2");
+		return;
+	}
+
+	/* Find available controllers */
+	fp = fopen(FINIT_CGPATH "/cgroup.controllers", "r");
+	if (!fp) {
+		_pe("Failed opening %s", FINIT_CGPATH "/cgroup.controllers");
+		return;
+	}
+
+	if (fgets(buf, sizeof(buf), fp)) {
+		char *cgroup;
+
+		cgroup = strtok(chomp(buf), "\t ");
+		while (cgroup) {
+			append_ctrl(cgroup);
+			cgroup = strtok(NULL, "\t ");
+		}
+	}
+
+	fclose(fp);
+
+	/* Enable all controllers */
+	if (fnwrite(controllers, FINIT_CGPATH "/cgroup.subtree_control"))
+		_pe("Failed enabling %s for %s", controllers, FINIT_CGPATH "/cgroup.subtree_control");
+
+	/* Default (protected) groups, PID 1, services, and user/login processes */
+	cgroup_add("init",   "cpu.weight:100",  1);
+	cgroup_add("system", "cpu.weight:9800", 1);
+	cgroup_add("user",   "cpu.weight:100",  1);
+	cgroup_config();
+
+	/* Move ourselves to init (best effort, otherwise run in 'root' group */
+	if (fnwrite("1", FINIT_CGPATH "/init/cgroup.procs"))
+		_pe("Failed moving PID 1 to cgroup ", FINIT_CGPATH "/init");
+
+	/* prepare cgroup.events watcher */
+	fd = iwatch_init(&iw_cgroup);
+	if (uev_io_init(ctx, &cgw, cgroup_events_cb, NULL, fd, UEV_READ)) {
+		_pe("Failed setting up cgroup.events watcher");
+		close(fd);
+	}
 }
 
 /**

src/cgroup.h

@@ -24,9 +24,23 @@
 #ifndef FINIT_CGROUP_H_
 #define FINIT_CGROUP_H_
 
-void cgroup_init   (void);
+#include <uev/uev.h>
 
-int cgroup_user    (char *name);
-int cgroup_service (char *cmd, char *id, int pid);
+struct cgroup {
+	char name[16];
+	char cfg[128];
+};
+
+void cgroup_mark_all(void);
+void cgroup_cleanup (void);
+
+int  cgroup_add     (char *name, char *cfg, int protected);
+int  cgroup_del     (char *dir);
+void cgroup_config  (void);
+
+void cgroup_init    (uev_ctx_t *ctx);
+
+int  cgroup_user    (char *name, int pid);
+int  cgroup_service (char *name, int pid, struct cgroup *cg);
 
 #endif /* FINIT_CGROUP_H_ */

src/cgutil.c

@@ -0,0 +1,654 @@
+/* New client tool, replaces old /dev/initctl API and telinit tool
+ *
+ * Copyright (c) 2019-2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "config.h"
+
+#include <err.h>
+#include <dirent.h>
+#include <stdio.h>
+#include <signal.h>
+#include <stdlib.h>
+#include <search.h>
+#include <inttypes.h>
+#include <lite/lite.h>
+#include <sys/sysinfo.h>		/* sysinfo() */
+
+#include "cgutil.h"
+#include "initctl.h"
+
+#define CDIM plain ? "" : "\e[2m"
+#define CRST plain ? "" : "\e[0m"
+
+#define NONE " "
+#define PIPE plain ? "| " : "│ "
+#define FORK plain ? "|-" : "├─"
+#define END  plain ? "`-" : "└─"
+
+uint64_t total_ram;			/* From sysinfo() */
+
+struct cg  dummy;			/* empty result "NULL"      */
+struct cg *list;
+
+
+char *pid_cmdline(int pid, char *buf, size_t len)
+{
+	size_t i, sz;
+	char *ptr;
+	FILE *fp;
+
+	fp = fopenf("r", "/proc/%d/cmdline", pid);
+	if (!fp) {
+		buf[0] = 0;
+		return buf;
+	}
+
+	sz = fread(buf, sizeof(buf[0]), len - 1, fp);
+	fclose(fp);
+	if (!sz)
+		return NULL;		/* kernel thread */
+
+	buf[sz] = 0;
+
+	ptr = strchr(buf, 0);
+	if (ptr && ptr != buf) {
+		ptr++;
+		sz -= ptr - buf;
+		memmove(buf, ptr, sz + 1);
+	}
+
+	for (i = 0; i < sz; i++) {
+		if (buf[i] == 0)
+			buf[i] = ' ';
+	}
+
+	return buf;
+}
+
+char *pid_comm(int pid, char *buf, size_t len)
+{
+	char *ptr = NULL;
+	FILE *fp;
+
+	fp = fopenf("r", "/proc/%d/comm", pid);
+	if (!fp)
+		return NULL;
+
+	if (fgets(buf, len, fp))
+		ptr = chomp(buf);
+	fclose(fp);
+
+	return ptr;
+}
+
+char *pid_cgroup(int pid, char *buf, size_t len)
+{
+	char *ptr = NULL;
+	FILE *fp;
+
+	fp = fopenf("r", "/proc/%d/cgroup", pid);
+	if (!fp)
+		return NULL;
+
+	if (fgets(buf, len, fp))
+		ptr = chomp(buf);
+	fclose(fp);
+
+	if (ptr)
+		ptr = strchr(buf, '/');
+
+	return ptr;
+}
+
+static char *cgroup_val(char *path, char *file, char *buf, size_t len)
+{
+	char *val = NULL;
+	FILE *fp;
+
+	fp = fopenf("r", "%s/%s", path, file);
+	if (fp) {
+		if (fgets(buf, len, fp)) {
+			val = chomp(buf);
+			len = strcspn(val, " \t");
+			val[len] = 0;
+		}
+
+		fclose(fp);
+	}
+
+	return val;
+}
+
+static uint64_t cgroup_uint64(char *path, char *file)
+{
+	uint64_t val = 0;
+	char buf[42];
+
+	if (cgroup_val(path, file, buf, sizeof(buf)))
+		val = strtoull(buf, NULL, 10);
+
+	return val;
+}
+
+static char *cgroup_memval(char *path, char *file, char *buf, size_t len)
+{
+	char data[42];
+
+	if (cgroup_val(path, file, data, sizeof(data))) {
+		if (!strcmp(data, "max"))
+			strlcpy(buf, data, len);
+		else
+			memsz(strtoull(data, NULL, 10), buf, len);
+	} else
+		buf[0] = 0;
+
+	return buf;
+}
+
+static uint64_t cgroup_memuse(struct cg *cg)
+{
+	char buf[42];
+	FILE *fp;
+
+	fp = fopenf("r", "%s/memory.stat", cg->cg_path);
+	if (fp) {
+		cg->cg_rss = 0;
+		cg->cg_vmlib = 0;
+
+		while (fgets(buf, sizeof(buf), fp)) {
+			chomp(buf);
+
+			if (!strncmp(buf, "anon", 4)) {
+				cg->cg_rss += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "slab", 4)) {
+				cg->cg_rss += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "kernel_stack", 12)) {
+				cg->cg_rss += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "pagetables", 10)) {
+				cg->cg_rss += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "percpu", 6)) {
+				cg->cg_rss += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "sock", 4)) {
+				cg->cg_rss += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "file", 4)) {
+				cg->cg_vmlib += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+			if (!strncmp(buf, "file_mapped", 11)) {
+				cg->cg_vmlib += strtoull(&buf[5], NULL, 10);
+				continue;
+			}
+		}
+		fclose(fp);
+	}
+
+	cg->cg_memshare = (float)(cg->cg_rss * 100 / total_ram);
+
+	return cg->cg_vmsize = cgroup_uint64(cg->cg_path, "memory.current");
+}
+
+uint64_t cgroup_memory(char *group)
+{
+	char path[256];
+
+	paste(path, sizeof(path), FINIT_CGPATH, group);
+
+	return cgroup_uint64(path, "memory.current");
+}
+
+static float cgroup_cpuload(struct cg *cg)
+{
+	char fn[256];
+	char buf[64];
+	FILE *fp;
+
+	snprintf(fn, sizeof(fn), "%s/cpu.stat", cg->cg_path);
+	fp = fopen(fn, "r");
+	if (!fp)
+		err(1, "Cannot open %s", fn);
+
+	while (fgets(buf, sizeof(buf), fp)) {
+		uint64_t curr;
+
+		chomp(buf);
+		if (strncmp(buf, "usage_usec", 10))
+			continue;
+
+		curr = strtoull(&buf[11], NULL, 10);
+		if (cg->cg_prev != 0) {
+			uint64_t diff = curr - cg->cg_prev;
+
+			/* this expects 1 sec poll interval */
+			cg->cg_load = (float)(diff / 1000000);
+			cg->cg_load *= 100.0;
+		}
+		cg->cg_prev = curr;
+		break;
+	}
+
+	fclose(fp);
+
+	return cg->cg_load;
+}
+
+static struct cg *append(char *path)
+{
+	struct cg *cg;
+	char fn[256];
+	ENTRY item;
+
+	snprintf(fn, sizeof(fn), "%s/cpu.stat", path);
+	if (access(fn, F_OK)) {
+		/* older kernels, 4.19, don't have summary cpu.stat in root */
+		if (strcmp(path, FINIT_CGPATH))
+			warn("not a cgroup path with cpu controller, %s", path);
+		return NULL;
+	}
+
+	cg = calloc(1, sizeof(struct cg));
+	if (!cg)
+		err(1, "failed allocating struct cg");
+
+	cg->cg_path = strdup(path);
+	if (list)
+		cg->cg_next = list;
+	list = cg;
+
+	item.key  = cg->cg_path;
+	item.data = cg;
+	if (!hsearch(item, ENTER))
+		err(1, "failed adding to hash table");
+
+	return cg;
+}
+
+static struct cg *find(char *path)
+{
+	ENTRY *ep, item = { path, NULL };
+
+	ep = hsearch(item, FIND);
+	if (ep)
+		return ep->data;
+
+	return append(path);
+}
+
+/* update stats */
+struct cg *cg_stats(char *path)
+{
+	struct cg *cg;
+
+	cg = find(path);
+	if (!cg)
+		return &dummy;
+
+	cgroup_cpuload(cg);
+	cgroup_memuse(cg);
+
+	return cg;
+}
+
+/* query config */
+struct cg *cg_conf(char *path)
+{
+	static struct cg cg;
+
+	cgroup_val(path, "memory.min", cg.cg_mem.min, sizeof(cg.cg_mem.min));
+	cgroup_memval(path, "memory.max", cg.cg_mem.max, sizeof(cg.cg_mem.max));
+	cgroup_val(path, "cpu.weight", cg.cg_cpu.weight, sizeof(cg.cg_cpu.weight));
+	cgroup_val(path, "cpu.max",    cg.cg_cpu.max, sizeof(cg.cg_cpu.max));
+	cgroup_val(path, "cpuset.cpus.effective", cg.cg_cpu.set, sizeof(cg.cg_cpu.set));
+	cg.cg_vmsize = cgroup_uint64(path, "memory.current");
+
+	return &cg;
+}
+
+static int cgroup_filter(const struct dirent *entry)
+{
+	/* Skip current dir ".", and prev dir "..", from list of files */
+	if ((1 == strlen(entry->d_name) && entry->d_name[0] == '.') ||
+	    (2 == strlen(entry->d_name) && !strcmp(entry->d_name, "..")))
+		return 0;
+
+	if (entry->d_name[0] == '.')
+		return 0;
+
+	if (entry->d_type != DT_DIR)
+		return 0;
+
+	return 1;
+}
+
+int cgroup_tree(char *path, char *pfx, int mode, int pos)
+{
+	struct dirent **namelist = NULL;
+	char s[32], r[32], l[32];
+	char row[ttcols + 9];		/* + control codes */
+	size_t rlen = sizeof(row) - 1;
+	size_t rplen = rlen - 9;
+	struct stat st;
+	struct cg *cg;
+	char buf[512];
+	int rc = 0;
+	FILE *fp;
+	int i, n;
+	int num;
+
+	if (pos >= ttrows)
+		return 0;
+
+	if (-1 == lstat(path, &st))
+		return 1;
+
+	if ((st.st_mode & S_IFMT) != S_IFDIR) {
+		errno = ENOTDIR;
+		return -1;
+	}
+
+	fp = fopenf("r", "%s/cgroup.procs", path);
+	if (!fp)
+		return -1;
+	num = 0;
+	while (fgets(buf, sizeof(buf), fp))
+		num++;
+
+	if (!pfx) {
+		pfx = "";
+		switch (mode) {
+		case 1:
+			cg = cg_stats(path);
+			snprintf(row, rplen, "\r %6.6s  %6.6s  %6.6s %5.1f %5.1f  %s",
+				 memsz(cg->cg_vmsize, s, sizeof(s)),
+				 memsz(cg->cg_rss,    r, sizeof(r)),
+				 memsz(cg->cg_vmlib,  l, sizeof(l)),
+				 cg->cg_memshare, cg->cg_load, path);
+			break;
+		case 2:
+			cg = cg_conf(path);
+			snprintf(row, rplen, "\r%6.6s [%-6.6s%6.6s] %6s [%-6.6s%6.6s] %s",
+				 memsz(cg->cg_vmsize, s, sizeof(s)),
+				 cg->cg_mem.min, cg->cg_mem.max, cg->cg_cpu.set,
+				 cg->cg_cpu.weight, cg->cg_cpu.max, path);
+			break;
+		default:
+			strlcpy(row, "\r", rplen);
+			strlcat(row, path, rplen);
+			break;
+		}
+
+		puts(row);
+	}
+
+	if (num > 0) {
+		rewind(fp);
+
+		i = 0;
+		while (fgets(buf, sizeof(buf), fp)) {
+			char comm[80] = { 0 };
+			pid_t pid;
+
+			pid = atoi(chomp(buf));
+			if (pid <= 0)
+				continue;
+
+			/* skip kernel threads for now (no cmdline) */
+			pid_comm(pid, comm, sizeof(comm));
+			if (pid_cmdline(pid, buf, sizeof(buf))) {
+				char proc[ttcols];
+				int len;
+
+				switch (mode) {
+				case 1:
+					snprintf(row, rplen, "\r%37s", " ");
+					break;
+				case 2:
+					snprintf(row, rplen, "\r --.-- [            ]        [            ] ");
+					break;
+				default:
+					strlcpy(row, "\r", rplen);
+					break;
+				}
+
+				strlcat(row, pfx, rplen);
+				strlcat(row, ++i == num ? END : FORK, rlen);
+
+				snprintf(proc, sizeof(proc), " %d %s %s", pid, comm, buf);
+
+				if (plain) {
+					strlcat(row, proc, rplen);
+				} else {
+					strlcat(row, CDIM, rlen);
+					strlcat(row, proc, rlen);
+
+					len = strlen(row) + strlen(CRST);
+					if (len > (int)rlen)
+						row[rlen - strlen(CRST)] = 0;
+
+					strlcat(row, CRST, sizeof(row));
+				}
+
+				puts(row);
+			}
+
+			if (mode == 1) {
+				pos += i;
+				if (pos >= ttrows)
+					break;
+			}
+		}
+	}
+
+	fclose(fp);
+
+	n = scandir(path, &namelist, cgroup_filter, alphasort);
+	if (n > 0) {
+		for (i = 0; i < n; i++) {
+			char *nm = namelist[i]->d_name;
+			char prefix[80];
+
+			snprintf(buf, sizeof(buf), "%s/%s", path, nm);
+			switch (mode) {
+			case 1:
+				cg = cg_stats(buf);
+				snprintf(row, rplen,
+					 "\r %6.6s  %6.6s  %6.6s %5.1f %5.1f  ",
+					 memsz(cg->cg_vmsize, s, sizeof(s)),
+					 memsz(cg->cg_rss,    r, sizeof(r)),
+					 memsz(cg->cg_vmlib,  l, sizeof(l)),
+					 cg->cg_memshare, cg->cg_load);
+				break;
+			case 2:
+				cg = cg_conf(buf);
+				snprintf(row, rplen, "\r%6.6s [%-6.6s%6.6s] %6.6s [%-6.6s%6.6s] ",
+					 memsz(cg->cg_vmsize, s, sizeof(s)),
+					 cg->cg_mem.min, cg->cg_mem.max,
+					 cg->cg_cpu.set, cg->cg_cpu.weight, cg->cg_cpu.max);
+				break;
+			default:
+				strlcpy(row, "\r", rplen);
+				break;
+			}
+
+			strlcat(row, pfx, rplen);
+			if (i + 1 == n) {
+				strlcat(row, END, rplen);
+				snprintf(prefix, sizeof(prefix), "%s   ", pfx);
+			} else {
+				strlcat(row, FORK, rplen);
+				snprintf(prefix, sizeof(prefix), "%s%s  ", pfx, PIPE);
+			}
+			strlcat(row, " ", rplen);
+
+			strlcat(row, nm,   rplen);
+			strlcat(row, "/ ", rplen);
+
+			puts(row);
+
+			rc += cgroup_tree(buf, prefix, mode, pos + i);
+
+			free(namelist[i]);
+		}
+
+		free(namelist);
+	}
+
+	return rc;
+}
+
+int show_cgps(char *arg)
+{
+	char path[512];
+
+	if (!arg)
+		arg = FINIT_CGPATH;
+	else if (arg[0] != '/') {
+		paste(path, sizeof(path), FINIT_CGPATH, arg);
+		arg = path;
+	}
+
+	return cgroup_tree(arg, NULL, 0, 0);
+}
+
+static void cgtop(uev_t *w, void *arg, int events)
+{
+	(void)w;
+	(void)events;
+
+	fputs("\e[2J\e[1;1H", stdout);
+	if (heading)
+		print_header(" VmSIZE     RSS   VmLIB  %%MEM  %%CPU  GROUP");
+	cgroup_tree(arg, NULL, 1, 0);
+}
+
+static void cleanup(void)
+{
+	ttcooked();
+	showcursor();
+	puts("");
+}
+
+static void leave(uev_t *w, void *arg, int events)
+{
+	(void)arg;
+	(void)events;
+
+	uev_exit(w->ctx);
+}
+
+static void key(uev_t *w, void *arg, int events)
+{
+	char ch;
+
+	(void)arg;
+	(void)events;
+
+	if (read(w->fd, &ch, sizeof(ch)) != -1) {
+		switch (ch) {
+		case 'q':
+			uev_exit(w->ctx);
+			break;
+
+		default:
+			dbg("Got char 0x%02x", ch);
+			break;
+		}
+	}
+}
+
+int show_cgtop(char *arg)
+{
+	struct sysinfo si = { 0 };
+        uev_t timer, input, sigint, sigterm, sigquit;
+	char path[512];
+        uev_ctx_t ctx;
+	int flags;
+
+	if (!arg)
+		arg = FINIT_CGPATH;
+	else if (arg[0] != '/') {
+		paste(path, sizeof(path), FINIT_CGPATH, arg);
+		arg = path;
+	}
+
+	if (!hcreate(ttrows + 25))
+		err(1, "failed creating hash table");
+
+	sysinfo(&si);
+	total_ram = si.totalram * si.mem_unit;
+
+        uev_init(&ctx);
+        uev_timer_init(&ctx, &timer, cgtop, arg, 1, ionce ? 0 : 1000);
+
+	if (!ionce && !plain) {
+		atexit(cleanup);
+		ttraw();
+		hidecursor();
+
+		flags = fcntl(STDIN_FILENO, F_GETFL);
+		if (flags != -1)
+			(void)fcntl(STDIN_FILENO, F_SETFL, flags | O_NONBLOCK);
+		(void)uev_io_init(&ctx, &input, key, NULL, STDIN_FILENO, UEV_READ);
+
+		(void)uev_signal_init(&ctx, &sigint, leave, NULL, SIGINT);
+		(void)uev_signal_init(&ctx, &sigterm, leave, NULL, SIGTERM);
+		(void)uev_signal_init(&ctx, &sigquit, leave, NULL, SIGQUIT);
+	}
+
+	return uev_run(&ctx, 0);
+}
+
+int show_cgroup(char *arg)
+{
+	char path[512];
+
+	if (!arg)
+		arg = FINIT_CGPATH;
+	else if (arg[0] != '/') {
+		paste(path, sizeof(path), FINIT_CGPATH, arg);
+		arg = path;
+	}
+
+	/* memory.current memory.min memory.max cpuset.cpus cpu.weight cpu.max */
+	if (heading)
+		print_header("   MEM [MIN      MAX]    CPU [WEIGHT   MAX] GROUP");
+
+	return cgroup_tree(arg, NULL, 2, 0);
+}
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */

src/cgutil.h

@@ -0,0 +1,77 @@
+/* Cgroup utility functions for initctl
+ *
+ * Copyright (c) 2019-2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef FINIT_CGUTIL_H_
+#define FINIT_CGUTIL_H_
+
+#include <stdint.h>
+#include <stdlib.h>
+
+struct cg {
+	struct cg *cg_next;
+
+	/* stats */
+	char      *cg_path;		/* path in /sys/fs/cgroup   */
+	uint64_t   cg_prev;		/* cpuacct.usage            */
+	uint64_t   cg_rss;		/* memory.stat              */
+	uint64_t   cg_vmlib;		/* memory.stat              */
+	uint64_t   cg_vmsize;		/* memory.current           */
+	float      cg_memshare;		/* cg_rss / total_ram * 100 */
+	float      cg_load;		/* curr - prev / 10000000.0 */
+
+	/* config */
+	struct {
+		char min[32];
+		char max[32];
+	} cg_mem;
+
+	struct {
+		char set[16];
+		char weight[32];
+		char max[32];
+	} cg_cpu;
+};
+
+char *pid_cmdline (int pid, char *buf, size_t len);
+char *pid_comm    (int pid, char *buf, size_t len);
+char *pid_cgroup  (int pid, char *buf, size_t len);
+
+uint64_t cgroup_memory(char *group);
+
+struct cg *cg_stats(char *path);
+struct cg *cg_conf (char *path);
+
+int   cgroup_tree  (char *path, char *pfx, int mode, int pos);
+
+int   show_cgroup (char *arg);
+int   show_cgtop  (char *arg);
+int   show_cgps   (char *arg);
+
+#endif /* FINIT_CGUTIL_H_ */
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */

src/client.c

@@ -21,7 +21,9 @@
  * THE SOFTWARE.
  */
 
+#include <err.h>
 #include <errno.h>
+#include <poll.h>
 #include <sys/socket.h>
 #include <sys/un.h>
 
@@ -36,19 +38,14 @@ int client_connect(void)
 		.sun_path   = INIT_SOCKET,
 	};
 
-	sd = socket(AF_UNIX, SOCK_STREAM, 0);
+	sd = socket(AF_UNIX, SOCK_SEQPACKET, 0);
 	if (-1 == sd)
-		goto error;
+		err(1, "Failed creating UNIX domain socket");
 
-	if (connect(sd, (struct sockaddr*)&sun, sizeof(sun)) == -1) {
-		close(sd);
-		goto error;
-	}
+	if (connect(sd, (struct sockaddr*)&sun, sizeof(sun)) == -1)
+		err(1, "Failed connecting to finit");
 
 	return sd;
-error:
-	perror("Failed connecting to finit");
-	return -1;
 }
 
 int client_disconnect(void)
@@ -68,27 +65,44 @@ int client_disconnect(void)
 
 int client_send(struct init_request *rq, ssize_t len)
 {
+	struct pollfd pfd = { 0 };
 	int sd, result = 255;
 
 	sd = client_connect();
 	if (-1 == sd)
 		return -1;
 
-	if (write(sd, rq, len) != len)
-		goto error;
+	pfd.fd     = sd;
+	pfd.events = POLLOUT;
+	if (poll(&pfd, 1, 2000) <= 0) {
+		warn("Timed out waiting for Finit, errno %d", errno);
+		goto exit;
+	}
 
-	if (read(sd, rq, len) != len)
-		goto error;
+	if (write(sd, rq, len) != len) {
+		warn("Failed communicating with Finit, errno %d", errno);
+		goto exit;
+	}
+
+	pfd.fd = sd;
+	pfd.events = POLLIN;
+	if (poll(&pfd, 1, 2000) <= 0) {
+		warn("Timed out waiting for reply from Finit, errno %d", errno);
+		goto exit;
+	}
+
+	if (read(sd, rq, len) != len) {
+		warn("Failed reading reply from Finit, errno %d", errno);
+		goto exit;
+	}
 
 	if (rq->cmd == INIT_CMD_NACK)
 		result = 1;
 	else
 		result = 0;
-	goto exit;
-error:
-	perror("Failed communicating with finit");
 exit:
 	client_disconnect();
+
 	return result;
 }
 

src/cond-w.c

@@ -99,13 +99,14 @@ static void cond_bump_reconf(void)
 	unsigned int rgen;
 
 	/*
-	 * If %COND_RECONF does not exist, cond_get_gen() returns 0
+	 * If %_PATH_RECONF does not exist, cond_get_gen() returns 0
 	 * meaning that rgen++ is always what we want.
 	 */
-	rgen = cond_get_gen(COND_RECONF);
+	rgen = cond_get_gen(_PATH_RECONF);
 	rgen++;
 
-	cond_set_gen(COND_RECONF, rgen);
+	if (cond_set_gen(_PATH_RECONF, rgen))
+		_pe("Failed setting %s to gen %d", _PATH_RECONF, rgen);
 }
 
 static int cond_checkpath(const char *path)
@@ -119,7 +120,7 @@ static int cond_checkpath(const char *path)
 		return 1;
 	}
 
-	if (makepath(dir) && errno != EEXIST) {
+	if (mkpath(dir, 0755) && errno != EEXIST) {
 		_pe("Failed creating dir '%s' for condition '%s'", dir, path);
 		return 1;
 	}
@@ -127,22 +128,22 @@ static int cond_checkpath(const char *path)
 	return 0;
 }
 
-int cond_set_path(const char *path, enum cond_state new)
+int cond_set_path(const char *path, enum cond_state next)
 {
-	enum cond_state old;
+	enum cond_state prev;
 	unsigned int rgen;
 
-	_d("%s", path);
+	_d("%s <= %d", path, next);
 
-	rgen = cond_get_gen(COND_RECONF);
+	rgen = cond_get_gen(_PATH_RECONF);
 	if (!rgen) {
 		_e("Unable to read configuration generation (%s)", path);
 		return -1;
 	}
 
-	old = cond_get_path(path);
+	prev = cond_get_path(path);
 
-	switch (new) {
+	switch (next) {
 	case COND_ON:
 		if (cond_checkpath(path))
 		    return 0;
@@ -159,30 +160,11 @@ int cond_set_path(const char *path, enum cond_state new)
 		return 0;
 	}
 
-	return new != old;
+	return next != prev;
 }
 
-/* Has condition in configuration and cond is allowed? */
-static int svc_has_cond(svc_t *svc)
-{
-	if (!svc->cond[0])
-		return 0;
-
-	switch (svc->type) {
-	case SVC_TYPE_SERVICE:
-	case SVC_TYPE_TASK:
-	case SVC_TYPE_RUN:
-	case SVC_TYPE_SYSV:
-		return 1;
-
-	default:
-		break;
-	}
-
-	return 0;
-}
-
-static void cond_update(const char *name)
+/* Should only be used by cond_set*(), cond_clear(), and usr/sys plugins! */
+void cond_update(const char *name)
 {
 	svc_t *svc, *iter = NULL;
 
@@ -196,13 +178,22 @@ static void cond_update(const char *name)
 	}
 }
 
-void cond_set(const char *name)
+int cond_set_noupdate(const char *name)
 {
 	_d("%s", name);
 	if (string_compare(name, "nop"))
-		return;
+		return 1;
 
 	if (!cond_set_path(cond_path(name), COND_ON))
+		return 1;
+
+	return 0;
+}
+
+void cond_set(const char *name)
+{
+	_d("%s", name);
+	if (cond_set_noupdate(name))
 		return;
 
 	cond_update(name);
@@ -221,17 +212,27 @@ void cond_set_oneshot(const char *name)
 	if (cond_checkpath(path))
 		return;
 
-	symlink(COND_RECONF, path);
+	if (symlink(_PATH_RECONF, path) && errno != EEXIST)
+		_pe("Failed creating onshot cond %s", name);
 	cond_update(name);
 }
 
+int cond_clear_noupdate(const char *name)
+{
+	_d("%s", name);
+	if (string_compare(name, "nop"))
+		return 1;
+
+	if (!cond_set_path(cond_path(name), COND_OFF))
+		return 1;
+
+	return 0;
+}
+
 void cond_clear(const char *name)
 {
 	_d("%s", name);
-	if (string_compare(name, "nop"))
-		return;
-
-	if (!cond_set_path(cond_path(name), COND_OFF))
+	if (cond_clear_noupdate(name))
 		return;
 
 	cond_update(name);
@@ -244,7 +245,7 @@ void cond_reload(void)
 	cond_bump_reconf();
 }
 
-static int reassert(const char *fpath, const struct stat *sb, int tflg, struct FTW *ftw)
+static int do_assert(const char *fpath, const struct stat *sb, int tflg, struct FTW *ftw, int set)
 {
 	char *nm;
 
@@ -254,35 +255,57 @@ static int reassert(const char *fpath, const struct stat *sb, int tflg, struct F
 	if (tflg != FTW_F)
 		return 0;
 
-	nm = strstr((char *)fpath, COND_DIR);
+	nm = strstr((char *)fpath, COND_BASE);
 	if (!nm) {
-		_e("Incorrect condtion path %s, cannot reassert", fpath);
+		_e("Incorrect condtion path %s, cannot %sassert", fpath, set ? "re" : "de");
 		return 1;
 	}
 
-	nm += sizeof(COND_DIR);
-	_d("Reasserting %s => %s", fpath, nm);
-	cond_set(nm);
+	nm += strlen(COND_BASE);
+	_d("%sasserting %s => %s", set ? "Re" : "De", fpath, nm);
+	if (set)
+		cond_set(nm);
+	else
+		cond_clear_noupdate(nm); /* important, see netlink plugin! */
 
 	return 0;
 }
 
+static int reassert(const char *fpath, const struct stat *sb, int tflg, struct FTW *ftw)
+{
+	return do_assert(fpath, sb, tflg, ftw, 1);
+}
+
+static int deassert(const char *fpath, const struct stat *sb, int tflg, struct FTW *ftw)
+{
+	return do_assert(fpath, sb, tflg, ftw, 0);
+}
+
 /*
  * Used only by netlink plugin atm.
  * type: is a one of pid/, net/, etc.
  */
-void cond_reassert(const char *type)
+void cond_reassert(const char *pat)
 {
-	_d("%s", type);
-	nftw(cond_path(type), reassert, 20, FTW_DEPTH);
+	_d("%s", pat);
+	nftw(cond_path(pat), reassert, 20, FTW_DEPTH);
+}
+
+/*
+ * Used only by netlink plugin atm.
+ */
+void cond_deassert(const char *pat)
+{
+	_d("%s", pat);
+	nftw(cond_path(pat), deassert, 20, FTW_DEPTH);
 }
 
 void cond_init(void)
 {
 	char path[MAX_ARG_LEN];
 
-	if (makepath(pid_runpath(COND_PATH, path, sizeof(path))) && errno != EEXIST) {
-		_pe("Failed creating condition base directory '%s'", COND_PATH);
+	if (mkpath(pid_runpath(_PATH_COND, path, sizeof(path)), 0755) && errno != EEXIST) {
+		_pe("Failed creating condition base directory '%s'", _PATH_COND);
 		return;
 	}
 

src/cond.c

@@ -45,15 +45,15 @@ const char *cond_path(const char *name)
 	static char path[256];
 	char tmp[MAX_ARG_LEN];
 
-	snprintf(tmp, sizeof(tmp), COND_PATH "/%s", name);
+	snprintf(tmp, sizeof(tmp), _PATH_COND "%s", name);
 
 	return pid_runpath(tmp, path, sizeof(path));
 }
 
 unsigned int cond_get_gen(const char *file)
 {
-	unsigned int gen;
 	char *ptr, path[256];
+	unsigned int gen;
 	FILE *fp;
 	int ret;
 
@@ -74,14 +74,14 @@ enum cond_state cond_get_path(const char *path)
 {
 	int cgen, rgen;
 
-	rgen = cond_get_gen(COND_RECONF);
-	if (!rgen)
-		return COND_OFF;
-
 	cgen = cond_get_gen(path);
 	if (!cgen)
 		return COND_OFF;
 
+	rgen = cond_get_gen(_PATH_RECONF);
+	if (!rgen)
+		return COND_OFF;
+
 	return (cgen == rgen) ? COND_ON : COND_FLUX;
 }
 

src/cond.h

@@ -4,9 +4,16 @@
 #include <paths.h>
 #include <svc.h>
 
-#define COND_DIR      "finit/cond"
-#define COND_PATH     _PATH_VARRUN COND_DIR
-#define COND_RECONF   COND_PATH "/reconf"
+#define COND_BASE      "finit/cond"
+#define COND_PID       "pid/"
+#define COND_SYS       "sys/"
+#define COND_USR       "usr/"
+
+#define _PATH_COND     _PATH_VARRUN COND_BASE "/"
+#define _PATH_CONDPID  _PATH_COND   COND_PID
+#define _PATH_CONDSYS  _PATH_COND   COND_SYS
+#define _PATH_CONDUSR  _PATH_COND   COND_USR
+#define _PATH_RECONF   _PATH_COND   "reconf"
 
 typedef enum cond_state {
 	COND_OFF = 0,
@@ -23,12 +30,19 @@ enum cond_state cond_get     (const char *name);
 enum cond_state cond_get_agg (const char *names);
 int             cond_affects (const char *name, const char *names);
 
+void cond_update      (const char *name);
 int  cond_set_path    (const char *path, enum cond_state new);
 void cond_set         (const char *name);
 void cond_set_oneshot (const char *name);
 void cond_clear       (const char *name);
 void cond_reload      (void);
+
+int cond_set_noupdate (const char *name);
+int cond_clear_noupdate(const char *name);
+
 void cond_reassert    (const char *pat);
+void cond_deassert    (const char *pat);
+
 void cond_init        (void);
 
 #endif	/* FINIT_COND_H_ */

src/conf.c

@@ -23,6 +23,7 @@
 
 #include "config.h"		/* Generated by configure script */
 
+#include <ctype.h>
 #include <dirent.h>
 #include <string.h>
 #include <sys/inotify.h>
@@ -47,8 +48,11 @@
 int logfile_size_max = 200000;	/* 200 kB */
 int logfile_count_max = 5;
 
+struct rlimit initial_rlimit[RLIMIT_NLIMITS];
 struct rlimit global_rlimit[RLIMIT_NLIMITS];
 
+char cgroup_current[16]; /* cgroup.NAME sets current cgroup for a set of services */
+
 struct conf_change {
 	TAILQ_ENTRY(conf_change) link;
 	char *name;
@@ -78,14 +82,14 @@ static void hide_args(int argc, char *argv[])
 	}
 }
 
-int get_bool(char *ptr, int default_value)
+int get_bool(char *arg, int default_value)
 {
-	if (!ptr)
+	if (!arg)
 		goto fallback;
 
-	if (string_compare(ptr, "true") || string_compare(ptr, "on") || string_compare(ptr, "1"))
+	if (string_compare(arg, "true") || string_compare(arg, "on") || string_compare(arg, "1"))
 		return 1;
-	if (string_compare(ptr, "false") || string_compare(ptr, "off") || string_compare(ptr, "0"))
+	if (string_compare(arg, "false") || string_compare(arg, "off") || string_compare(arg, "0"))
 		return 0;
 fallback:
 	return default_value;
@@ -97,19 +101,24 @@ fallback:
  */
 static void parse_finit_opts(char *opt)
 {
-	char *ptr;
+	char *arg;
 
-	ptr = strchr(opt, '=');
-	if (ptr)
-		*ptr++ = 0;
+	arg = strchr(opt, '=');
+	if (arg)
+		*arg++ = 0;
 
 	if (string_compare(opt, "debug")) {
-		debug = get_bool(ptr, 1);
+		debug = get_bool(arg, 1);
 		return;
 	}
 
 	if (string_compare(opt, "status_style")) {
-		if (string_compare(ptr, "old") || string_compare(ptr, "classic"))
+		if (!arg) {
+			_e("status_style option requires an argument, skipping.");
+			return;
+		}
+
+		if (string_compare(arg, "old") || string_compare(arg, "classic"))
 			show_progress(PROGRESS_CLASSIC);
 		else
 			show_progress(PROGRESS_MODERN);
@@ -117,7 +126,7 @@ static void parse_finit_opts(char *opt)
 	}
 
 	if (string_compare(opt, "show_status")) {
-		show_progress(get_bool(ptr, 1) ? PROGRESS_DEFAULT : PROGRESS_SILENT);
+		show_progress(get_bool(arg, 1) ? PROGRESS_DEFAULT : PROGRESS_SILENT);
 		return;
 	}
 }
@@ -346,7 +355,7 @@ char *lim2str(struct rlimit *rlim)
 	char tmp[25];
 	static char buf[42];
 
-	memset(buf, 0, sizeof(buf));
+	buf[0] = 0;
 	if (RLIM_INFINITY == rlim->rlim_cur)
 		snprintf(tmp, sizeof(tmp), "unlimited, ");
 	else
@@ -417,6 +426,29 @@ error:
 	logit(LOG_WARNING, "rlimit: parse error");
 }
 
+/* cgroup NAME ctrl.prop:value,ctrl.prop:value ... */
+static void conf_parse_cgroup(char *line)
+{
+	char config[strlen(line) + 1];
+	char *ptr, *name;
+
+	name = strtok(line, " \t");
+	if (!name)
+		return;
+
+	if (strstr(name, "..") || strchr(name, '/'))
+		return;		/* illegal */
+
+	config[0] = 0;
+	while ((ptr = strtok(NULL, " \t"))) {
+		if (config[0])
+			strlcat(config, ",", sizeof(config));
+		strlcat(config, ptr, sizeof(config));
+	}
+
+	cgroup_add(name, config, 0);
+}
+
 static void parse_static(char *line, int is_rcsd)
 {
 	char cmd[CMD_SIZE];
@@ -546,9 +578,21 @@ static void parse_dynamic(char *line, struct rlimit rlimit[], char *file)
 		return;
 	}
 
+	/* Read contrl group limits */
+	if (MATCH_CMD(line, "cgroup ", x)) {
+		conf_parse_cgroup(x);
+		return;
+	}
+
+	/* Set current cgroup for the following services/run/tasks */
+	if (MATCH_CMD(line, "cgroup.", x)) {
+		strlcpy(cgroup_current, x, sizeof(cgroup_current));
+		return;
+	}
+
 	/* Regular or serial TTYs to run getty */
 	if (MATCH_CMD(line, "tty ", x)) {
-		tty_register(strip_line(x), rlimit, file);
+		service_register(SVC_TYPE_TTY, strip_line(x), rlimit, file);
 		return;
 	}
 }
@@ -574,9 +618,11 @@ static int parse_conf(char *file, int is_rcsd)
 	if (!fp)
 		return 1;
 
-	/* Prepare default limits for each service in /etc/finit.d/ */
-	if (is_rcsd)
+	/* Prepare default limits and group for each service in /etc/finit.d/ */
+	if (is_rcsd) {
 		memcpy(rlimit, global_rlimit, sizeof(rlimit));
+		cgroup_current[0] = 0;
+	}
 
 	_d("*** Parsing %s", file);
 	while (!feof(fp)) {
@@ -616,25 +662,22 @@ int conf_reload(void)
 	   daylight, timezone, tzname[0], tzname[1]);
 
 	/* Mark and sweep */
+	cgroup_mark_all();
 	svc_mark_dynamic();
-	tty_mark();
 
 	/*
-	 * Get current global limits, which may be overridden from both
-	 * finit.conf, for Finit and its services like getty+watchdogd,
-	 * and *.conf in finit.d/, for each service(s) listed there.
+	 * Reset global rlimit to bootstrap values from conf_init().
 	 */
-	for (int i = 0; i < RLIMIT_NLIMITS; i++)
-		getrlimit(i, &global_rlimit[i]);
+	memcpy(global_rlimit, initial_rlimit, sizeof(global_rlimit));
 
 	if (rescue) {
 		int rc;
-		char line[80] = "tty [12345] @console noclear nologin";
+		char line[80] = "tty [12345] rescue";
 
 		/* If rescue.conf is missing, fall back to a root shell */
 		rc = parse_conf(RESCUE_CONF, 0);
 		if (rc)
-			tty_register(line, global_rlimit, NULL);
+			service_register(SVC_TYPE_TTY, line, global_rlimit, NULL);
 
 		print(rc, "Entering rescue mode");
 		goto done;
@@ -674,7 +717,7 @@ int conf_reload(void)
 
 		/* Check for dangling symlinks */
 		if (S_ISLNK(st.st_mode)) {
-			path = rp = realpath(path, NULL);
+			rp = realpath(path, NULL);
 			if (!rp) {
 				logit(LOG_WARNING, "Skipping %s, dangling symlink: %s", path, strerror(errno));
 				continue;
@@ -684,7 +727,7 @@ int conf_reload(void)
 		/* Check that file ends with '.conf' */
 		len = strlen(path);
 		if (len < 6 || strcmp(&path[len - 5], ".conf"))
-			_d("Skipping %s, not a valid .conf ... ", path);
+			_d("Skipping %s, not a Finit .conf file ... ", path);
 		else
 			parse_conf(path, 1);
 
@@ -694,7 +737,15 @@ int conf_reload(void)
 
 	globfree(&gl);
 
+	/* Mark any reverse deps as chenaged. */
+	service_update_rdeps();
+
+	/* Set up top-level cgroups */
+	cgroup_config();
 done:
+	/* Remove all unused top-level cgroups */
+	cgroup_cleanup();
+
 	/* Drop record of all .conf changes */
 	drop_changes();
 
@@ -787,20 +838,29 @@ int conf_any_change(void)
 
 int conf_changed(char *file)
 {
+	int rc = 0;
+	char *rp;
+
 	if (!file)
 		return 0;
 
-	if (conf_find(file))
-		return 1;
+	rp = realpath(file, NULL);
+	if (!rp)
+		return 0;
 
-	return 0;
+	if (conf_find(rp))
+		rc = 1;
+	free(rp);
+
+	return rc;
 }
 
 static void conf_cb(uev_t *w, void *arg, int events)
 {
 	static char ev_buf[8 *(sizeof(struct inotify_event) + NAME_MAX + 1) + 1];
 	struct inotify_event *ev;
-	ssize_t sz, off;
+	ssize_t sz;
+	size_t off;
 
 	sz = read(w->fd, ev_buf, sizeof(ev_buf) - 1);
 	if (sz <= 0) {
@@ -809,10 +869,16 @@ static void conf_cb(uev_t *w, void *arg, int events)
 	}
 	ev_buf[sz] = 0;
 
-	for (off = 0; off < sz; off += sizeof(*ev) + ev->len) {
+	for (off = 0; off < (size_t)sz; off += sizeof(*ev) + ev->len) {
 		struct iwatch_path *iwp;
 
+		if (off + sizeof(*ev) > (size_t)sz)
+			break;
+
 		ev = (struct inotify_event *)&ev_buf[off];
+		if (off + sizeof(*ev) + ev->len > (size_t)sz)
+			break;
+
 		if (!ev->mask)
 			continue;
 
@@ -868,7 +934,7 @@ int conf_monitor(void)
 }
 
 /*
- * Prepare .conf parser and load all .conf files
+ * Prepare .conf parser and load /etc/finit.conf for global settings
  */
 int conf_init(uev_ctx_t *ctx)
 {
@@ -882,8 +948,17 @@ int conf_init(uev_ctx_t *ctx)
          * finit.conf, for Finit and its services like getty+watchdogd,
          * and *.conf in finit.d/, for each service(s) listed there.
          */
-        for (int i = 0; i < RLIMIT_NLIMITS; i++)
-                getrlimit(i, &global_rlimit[i]);
+        for (int i = 0; i < RLIMIT_NLIMITS; i++) {
+                if (getrlimit(i, &initial_rlimit[i]))
+			logit(LOG_WARNING, "rlimit: Failed reading setting %s: %s",
+			      rlim2str(i), strerror(errno));
+	}
+
+	/* Initialize global rlimits, e.g. for built-in services */
+	memcpy(global_rlimit, initial_rlimit, sizeof(global_rlimit));
+
+	/* Read global rlimits and global cgroup setup from /etc/finit.conf */
+	parse_conf(FINIT_CONF, 0);
 
 	/* prepare /etc watcher */
 	fd = iwatch_init(&iw_conf);

src/conf.h

@@ -24,12 +24,14 @@
 #ifndef FINIT_CONF_H_
 #define FINIT_CONF_H_
 
+#include "cgroup.h"
 #include "svc.h"
 
 extern int logfile_size_max;
 extern int logfile_count_max;
 
 extern struct rlimit global_rlimit[];
+extern char cgroup_current[];
 
 int   str2rlim(char *str);
 char *rlim2str(int rlim);

src/exec.c

@@ -35,6 +35,7 @@
 #include <lite/lite.h>
 
 #include "finit.h"
+#include "cgroup.h"
 #include "conf.h"
 #include "helpers.h"
 #include "sig.h"
@@ -244,10 +245,6 @@ static void prepare_tty(char *tty, speed_t speed, char *procname, struct rlimit
 	char name[80];
 	int fd;
 
-	/* Detach from initial controlling TTY and become session leader */
-	vhangup();
-	setsid();
-
 	fd = open(tty, O_RDWR);
 	if (fd < 0) {
 		logit(LOG_ERR, "Failed opening %s: %s", tty, strerror(errno));
@@ -259,6 +256,11 @@ static void prepare_tty(char *tty, speed_t speed, char *procname, struct rlimit
 	dup2(fd, STDERR_FILENO);
 	close(fd);
 
+	/*
+	 * Become session leader and set controlling TTY
+	 * to enable Ctrl-C and job control in shell.
+	 */
+	setsid();
 	if (ioctl(STDIN_FILENO, TIOCSCTTY, 1) < 0)
 		logit(LOG_WARNING, "Failed TIOCSCTTY on %s: %s", tty, strerror(errno));
 
@@ -288,7 +290,6 @@ static void prepare_tty(char *tty, speed_t speed, char *procname, struct rlimit
 	sigemptyset(&sa.sa_mask);
 	sa.sa_flags   = SA_RESTART;
 	sa.sa_handler = SIG_IGN;
-	sigaction(SIGHUP,  &sa, NULL);
 	sigaction(SIGINT,  &sa, NULL);
 	sigaction(SIGQUIT, &sa, NULL);
 
@@ -315,7 +316,7 @@ static int activate_console(int noclear, int nowait)
 		return 1;
 
 	if (!noclear)
-		(void)write(STDERR_FILENO, "\e[r\e[H\e[J", 9);
+		dprint(STDERR_FILENO, "\e[r\e[H\e[J", 9);
 
 	/* Disable ECHO, XON/OFF while waiting for <CR> */
 	if (!tcgetattr(STDIN_FILENO, &orig)) {
@@ -339,16 +340,16 @@ static int activate_console(int noclear, int nowait)
 			continue;
 		}
 
-		(void)write(STDERR_FILENO, clr, strlen(clr));
-		(void)write(STDERR_FILENO, msg, strlen(msg));
+		dprint(STDERR_FILENO, clr, strlen(clr));
+		dprint(STDERR_FILENO, msg, strlen(msg));
 		while (read(STDIN_FILENO, &c, 1) == 1 && c != '\r')
 			continue;
 
 		if (fexist(SYNC_STOPPED))
 			continue;
 
-		(void)write(STDERR_FILENO, clr, strlen(clr));
-		(void)write(STDERR_FILENO, cup, strlen(cup));
+		dprint(STDERR_FILENO, clr, strlen(clr));
+		dprint(STDERR_FILENO, cup, strlen(cup));
 		ret = 1;
 		break;
 	}
@@ -381,62 +382,47 @@ static int activate_console(int noclear, int nowait)
  */
 pid_t run_getty(char *tty, char *baud, char *term, int noclear, int nowait, struct rlimit rlimit[])
 {
-	pid_t pid;
+	speed_t speed;
+	int rc = 1;
 
-	pid = fork();
-	if (!pid) {
-		speed_t speed;
-		int rc = 1;
-
-		speed = stty_parse_speed(baud);
+	speed = stty_parse_speed(baud);
+	prepare_tty(tty, speed, "tty", rlimit);
+	if (activate_console(noclear, nowait)) {
 		logit(LOG_INFO, "Starting built-in getty on %s, speed %u", tty, speed);
-		prepare_tty(tty, speed, "tty", rlimit);
-		if (activate_console(noclear, nowait))
-			rc = getty(tty, speed, term, NULL);
-
-		_exit(rc);
+		rc = getty(tty, speed, term, NULL);
 	}
 
-	return pid;
+	return rc;
 }
 
+/*
+ * Start external getty as defined by user in .conf file
+ */
 pid_t run_getty2(char *tty, char *cmd, char *args[], int noclear, int nowait, struct rlimit rlimit[])
 {
-	pid_t pid;
+	int rc = 1;
 
-	pid = fork();
-	if (!pid) {
-		int rc = 1;
-
-		/* Dunno speed, tell stty() to not mess with it */
-		logit(LOG_INFO, "Starting external getty on %s, speed %u", tty, B0);
-		prepare_tty(tty, B0, "getty", rlimit);
-		if (activate_console(noclear, nowait))
-			rc = execv(cmd, args);
-
-		vhangup();
-		_exit(rc);
+	/* Dunno speed, tell stty() to not mess with it */
+	prepare_tty(tty, B0, "getty", rlimit);
+	if (activate_console(noclear, nowait)) {
+		logit(LOG_INFO, "Starting external getty on %s", tty);
+		rc = execv(cmd, args);
 	}
 
-	return pid;
+	vhangup();
+
+	return rc;
 }
 
 pid_t run_sh(char *tty, int noclear, int nowait, struct rlimit rlimit[])
 {
-	pid_t pid;
+	int rc = 1;
 
-	pid = fork();
-	if (!pid) {
-		int rc = 1;
+	prepare_tty(tty, B0, "finitsh", rlimit);
+	if (activate_console(noclear, nowait))
+		rc = sh(tty);
 
-		prepare_tty(tty, B0, "finit-sh", rlimit);
-		if (activate_console(noclear, nowait))
-			rc = sh(tty);
-
-		_exit(rc);
-	}
-
-	return pid;
+	return rc;
 }
 
 int run_parts(char *dir, char *cmd)

src/finit.c

@@ -52,6 +52,7 @@
 #include "util.h"
 #include "utmp-api.h"
 #include "schedule.h"
+#include "watchdog.h"
 
 int   runlevel  = 0;		/* Bootstrap 'S' */
 int   cfglevel  = RUNLEVEL;	/* Fallback if no configured runlevel */
@@ -59,6 +60,7 @@ int   prevlevel = -1;
 int   debug     = 0;		/* debug mode from kernel cmdline */
 int   rescue    = 0;		/* rescue mode from kernel cmdline */
 int   single    = 0;		/* single user mode from kernel cmdline */
+int   bootstrap = 1;		/* set while bootrapping (for TTYs) */
 char *sdown     = NULL;
 char *network   = NULL;
 char *hostname  = NULL;
@@ -179,10 +181,74 @@ static void fs_remount_root(int fsckerr)
 	 * XXX: Untested, in the initramfs age we should
 	 *      probably use switch_root instead.
 	 */
-	rc = mount(SYSROOT, "/", NULL, MS_MOVE, NULL);
+	if (mount(SYSROOT, "/", NULL, MS_MOVE, NULL))
+		_pe("Failed %s / MS_MOVE");
 }
 #endif	/* SYSROOT */
 
+/*
+ * Opinionated file system setup.  Checks for critical mount points and
+ * mounts them as most users expect.  All file systems are checked with
+ * /proc/mounts before mounting.
+ *
+ * Embedded systems, and other people who want full control, can set up
+ * their system with /etc/fstab, which is handled before this function
+ * is called.  For systems like Debian/Ubuntu, who only have / and swap
+ * in their /etc/fstab, this function does all the magic necessary.
+ */
+static void fs_finalize(void)
+{
+	/*
+	 * Some systems rely on us to both create /dev/shm and, to mount
+	 * a tmpfs there.  Any system with dbus needs shared memory, so
+	 * mount it, unless its already mounted, but not if listed in
+	 * the /etc/fstab file already.
+	 */
+	if (!fismnt("/dev/shm")) {
+		makedir("/dev/shm", 0777);
+		mount("shm", "/dev/shm", "tmpfs", 0, "mode=0777");
+	}
+
+	/* Modern systems use /dev/pts */
+	if (!fismnt("/dev/pts")) {
+		char opts[32];
+		int mode;
+		int gid;
+
+		gid = getgroup("tty");
+		if (gid == -1)
+			gid = 0;
+
+		/* 0600 is default on Debian, use 0620 to get mesg y by default */
+		mode = 0620;
+		snprintf(opts, sizeof(opts), "gid=%d,mode=%d,ptmxmode=0666", gid, mode);
+
+		makedir("/dev/pts", 0755);
+		mount("devpts", "/dev/pts", "devpts", MS_NOSUID | MS_NOEXEC, opts);
+	}
+
+	/*
+	 * Modern systems use tmpfs for /run.  Fallback to /var/run if
+	 * /run doesn't exist is handled by the bootmisc plugin.  It
+	 * also sets up compat symlinks.
+	 *
+	 * The unconditional mount of /run/lock is for DoS prevention.
+	 * To override any of this behavior, add entries to /etc/fstab
+	 * for /run (and optionally /run/lock).
+	 */
+	if (fisdir("/run") && !fismnt("/run")) {
+		mount("tmpfs", "/run", "tmpfs", MS_NOSUID | MS_NODEV | MS_NOEXEC | MS_RELATIME, "mode=0755,size=10%");
+
+		/* This prevents user DoS of /run by filling /run/lock at the expense of another tmpfs, max 5MiB */
+		makedir("/run/lock", 1777);
+		mount("tmpfs", "/run/lock", "tmpfs", MS_NOSUID | MS_NODEV | MS_NOEXEC | MS_RELATIME, "mode=0777,size=5252880");
+	}
+
+	/* Modern systems use tmpfs for /tmp */
+	if (!fismnt("/tmp"))
+		mount("tmpfs", "/tmp", "tmpfs", MS_NOSUID | MS_NODEV, "mode=1777");
+}
+
 static void fs_mount(void)
 {
 	if (!rescue)
@@ -198,7 +264,9 @@ static void fs_mount(void)
 	plugin_run_hooks(HOOK_MOUNT_POST);
 
 	run("swapon -ea");
-	umask(0022);
+
+	_d("Finalize, ensure common file systems are available ...");
+	fs_finalize();
 }
 
 /*
@@ -206,6 +274,9 @@ static void fs_mount(void)
  * for early multi-console, and /sys for the cgroups support.  Any
  * occurrence of these file systems in /etc/fstab will replace these
  * mounts later in fs_mount()
+ *
+ * Ignore any mount errors with EBUSY, kernel likely alread mounted
+ * the filesystem for us automatically, e.g., CONFIG_DEVTMPFS_MOUNT.
  */
 static void fs_init(void)
 {
@@ -218,53 +289,19 @@ static void fs_init(void)
 	};
 	size_t i;
 
-	for (i = 0; i < NELEMS(fs); i++)
-		mount(fs[i].spec, fs[i].file, fs[i].type, 0, NULL);
-}
+	/* mask writable bit for g and o */
+	umask(022);
 
-/*
- * If everything goes south we can use this to give the operator an
- * emergency shell to debug the problem -- Finit should not crash!
- *
- * Note: Only use this for debugging a new Finit setup, don't use
- *       this in production since it gives a root shell to anyone
- *       if Finit crashes.
- *
- * This emergency shell steps in to prevent "Aieee, PID 1 crashed"
- * messages from the kernel, which usually results in a reboot, so
- * that the operator instead can debug the problem.
- */
-static void emergency_shell(void)
-{
-#ifdef EMERGENCY_SHELL
-	pid_t pid;
+	for (i = 0; i < NELEMS(fs); i++) {
+		int rc;
 
-	pid = fork();
-	if (pid) {
-		while (1) {
-			pid_t id;
-
-			/* Reap 'em (prevents Zombies) */
-			id = waitpid(-1, NULL, WNOHANG);
-			if (id == pid)
-				break;
-		}
-
-		fprintf(stderr, "\n=> Embarrassingly, Finit has crashed.  Check /dev/kmsg for details.\n");
-		fprintf(stderr,   "=> To debug, add 'debug' to the kernel command line.\n\n");
-
-		/*
-		 * Become session leader and set controlling TTY
-		 * to enable Ctrl-C and job control in shell.
-		 */
-		setsid();
-		ioctl(STDIN_FILENO, TIOCSCTTY, 1);
-
-		execl(_PATH_BSHELL, _PATH_BSHELL, NULL);
+		rc = mount(fs[i].spec, fs[i].file, fs[i].type, 0, NULL);
+		if (rc && errno != EBUSY)
+			_pe("Failed mounting %s on %s", fs[i].spec, fs[i].file);
 	}
-#endif /* EMERGENCY_SHELL */
 }
 
+
 /*
  * Handle bootstrap transition to configured runlevel, start TTYs
  *
@@ -276,20 +313,8 @@ static void emergency_shell(void)
  * finit.conf, or *.conf in finit.d/, run to completion before we
  * finalize the bootstrap process by calling this function.
  */
-static void finalize(void)
+static void finalize(void *unused)
 {
-	/*
-	 * Run startup scripts in the runparts directory, if any.
-	 */
-	if (runparts && fisdir(runparts) && !rescue)
-		run_parts(runparts, NULL);
-
-	/*
-	 * Start all tasks/services in the configured runlevel
-	 */
-	_d("Change to default runlevel, start all services ...");
-	service_runlevel(cfglevel);
-
 	/* Clean up bootstrap-only tasks/services that never started */
 	_d("Clean up all bootstrap-only tasks/services ...");
 	svc_prune_bootstrap();
@@ -311,9 +336,9 @@ static void finalize(void)
 	/* Disable progress output at normal runtime */
 	enable_progress(0);
 
-	/* Delayed start of TTYs at bootstrap */
-	_d("Launching all getty services ...");
-	tty_runlevel();
+	/* System bootrapped, launch TTYs et al */
+	bootstrap = 0;
+	service_step_all(SVC_TYPE_RESPAWN);
 }
 
 /*
@@ -331,12 +356,17 @@ static void crank_worker(void *unused)
 
 /*
  * Wait for system bootstrap to complete, all SVC_TYPE_RUNTASK must be
- * allowed to complete their work in [S], or timeout, before we call
- * finalize(), should not take more than 120 sec.
+ * allowed to complete their work in [S], or timeout, before we switch
+ * to the configured runlevel and call finalize(), should not take more
+ * than 120 sec.
  */
-static void final_worker(void *work)
+static void bootstrap_worker(void *work)
 {
 	static int cnt = 120;
+	static struct wq final = {
+		.cb = finalize,
+		.delay = 10
+	};
 
 	_d("Step all services ...");
 	service_step_all(SVC_TYPE_ANY);
@@ -352,12 +382,36 @@ static void final_worker(void *work)
 	else
 		_d("Timeout, resuming bootstrap.");
 
-	finalize();
+	_d("Starting runlevel change finalize ...");
+	schedule_work(&final);
+
+	/*
+	 * Run startup scripts in the runparts directory, if any.
+	 */
+	if (runparts && fisdir(runparts) && !rescue)
+		run_parts(runparts, NULL);
+
+	/*
+	 * Start all tasks/services in the configured runlevel
+	 */
+	_d("Change to default runlevel(%d), starting all services ...", cfglevel);
+	service_runlevel(cfglevel);
+}
+
+static int version(int rc)
+{
+	puts(PACKAGE_STRING);
+	printf("Bug report address: %-40s\n", PACKAGE_BUGREPORT);
+#ifdef PACKAGE_URL
+	printf("Project homepage: %s\n", PACKAGE_URL);
+#endif
+
+	return rc;
 }
 
 static int usage(int rc)
 {
-	printf("Usage: telinit [OPTIONS] [q | Q | 0-9]\n\n"
+	printf("Usage: %s [OPTIONS] [q | Q | 0-9]\n\n"
 	       "Options:\n"
 //	       "  -a       Ignored, compat SysV init\n"
 //	       "  -b       Ignored, compat SysV init\n"
@@ -375,12 +429,7 @@ static int usage(int rc)
 	       "  q, Q     Reload /etc/finit.conf and/or any *.conf in /etc/finit.d/\n"
 	       "           if modified, same as initctl reload or SIGHUP to PID 1\n"
 	       "  1, s, S  Enter system rescue mode, runlevel 1\n"
-	       "\n");
-
-	printf("Bug report address: %-40s\n", PACKAGE_BUGREPORT);
-#ifdef PACKAGE_URL
-	printf("Project homepage: %s\n", PACKAGE_URL);
-#endif
+	       "\n", prognm);
 
 	return rc;
 }
@@ -393,6 +442,7 @@ static int telinit(int argc, char *argv[])
 {
 	int c;
 
+	progname(argv[0]);
 	while ((c = getopt(argc, argv, "abe:h?st:vVz:")) != EOF) {
 		switch(c) {
 		case 'a': case 'b': case 'e': case 's': case 'z':
@@ -402,7 +452,7 @@ static int telinit(int argc, char *argv[])
 			break;
 
 		case 'v': case 'V':
-			return puts("v" VERSION) == EOF;
+			return version(0);
 
 		case 'h':
 		case '?':
@@ -414,13 +464,13 @@ static int telinit(int argc, char *argv[])
 		int req = (int)argv[optind][0];
 
 		if (isdigit(req))
-			return systemf("initctl runlevel %c", req);
+			return systemf("initctl -b runlevel %c", req);
 
 		if (req == 'q' || req == 'Q')
-			return systemf("initctl reload");
+			return systemf("initctl -b reload");
 
 		if (req == 's' || req == 'S')
-			return systemf("initctl runlevel %c", req);
+			return systemf("initctl -b runlevel %c", req);
 	}
 
 	/* XXX: add non-pid1 process monitor here
@@ -435,11 +485,12 @@ static int telinit(int argc, char *argv[])
 int main(int argc, char *argv[])
 {
 	struct wq crank = {
-		.cb = crank_worker
+		.cb = crank_worker,
+		.delay = 10
 	};
-	struct wq final = {
-		.cb = final_worker,
-		.delay = 1000
+	struct wq bootstrap = {
+		.cb = bootstrap_worker,
+		.delay = 100
 	};
 	uev_ctx_t loop;
 
@@ -470,33 +521,43 @@ int main(int argc, char *argv[])
 	ctx = &loop;
 
 	/*
-	 * Set PATH, SHELL, PWD, and umask early to something sane
+	 * Set PATH, SHELL, and PWD early to something sane
 	 */
 	setenv("PATH", _PATH_STDPATH, 1);
 	setenv("SHELL", _PATH_BSHELL, 1);
 
-	chdir("/");
-	umask(0);
-
-	/* Set up canvas */
-	if (!rescue && !debug)
-		screen_init();
-
-	/* Allow progress, if enabled */
-	enable_progress(1);
+	if (chdir("/"))
+		_pe("Failed cd /");
 
 	/*
 	 * In case of emergency.
 	 */
-	emergency_shell();
+	if (rescue) {
+		char *sulogin[] = {
+			_PATH_SULOGIN,
+			"sulogin",
+		};
+		size_t i;
 
-	/* Load plugins early, the first hook is in banner(), so we
-	 * need plugins loaded before calling it. */
+		for (i = 0; i < NELEMS(sulogin); i++) {
+			if (systemf(sulogin[i]))
+				continue;
+
+			rescue = 0;
+			break;
+		}
+	}
+
+	/*
+	 * Load plugins early, the first hook is in banner(), so we
+	 * need plugins loaded before calling it.
+	 */
 	plugin_init(&loop);
 
 	/*
 	 * Hello world.
 	 */
+	enable_progress(1);	/* Allow progress, if enabled */
 	banner();
 
 	/*
@@ -507,8 +568,7 @@ int main(int argc, char *argv[])
 	/*
 	 * Initialize default control groups, if available
 	 */
-	cgroup_init();
-
+	cgroup_init(&loop);
 
 	/* Check and mount filesystems. */
 	fs_mount();
@@ -526,6 +586,20 @@ int main(int argc, char *argv[])
 	 */
 	conf_init(&loop);
 
+	/*
+	 * Start built-in watchdogd as soon as possible, if enabled
+	 */
+	if (whichp(FINIT_LIBPATH_ "/watchdogd") && fexist(WDT_DEVNODE)) {
+		service_register(SVC_TYPE_SERVICE, "[123456789] cgroup.init name:watchdog :finit " FINIT_LIBPATH_ "/watchdogd -- Finit watchdog daemon", global_rlimit, NULL);
+		wdog = svc_find_by_nameid("watchdog", "finit");
+	}
+
+	/*
+	 * Start kernel event daemon as soon as possible, if enabled
+	 */
+	if (whichp(FINIT_LIBPATH_ "/keventd"))
+		service_register(SVC_TYPE_SERVICE, "[123456789] cgroup.init " FINIT_LIBPATH_ "/keventd -- Finit kernel event daemon", global_rlimit, NULL);
+
 	/* Base FS up, enable standard SysV init signals */
 	sig_setup(&loop);
 
@@ -540,13 +614,12 @@ int main(int argc, char *argv[])
 
 	_d("Starting initctl API responder ...");
 	api_init(&loop);
-	umask(022);
 
 	_d("Starting the big state machine ...");
 	schedule_work(&crank);
 
 	_d("Starting bootstrap finalize timer ...");
-	schedule_work(&final);
+	schedule_work(&bootstrap);
 
 	/*
 	 * Enter main loop to monitor /dev/initctl and services

src/finit.h

@@ -30,17 +30,28 @@
 #include <paths.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <sysexits.h>
 #include <sys/types.h>
 #include <sys/stat.h>
 #include <unistd.h>
 #include <lite/lite.h>
 #include <uev/uev.h>
 
-#ifndef _PATH_VARRUN
-#define _PATH_VARRUN    "/var/run/"
+/* just in case */
+#ifndef _PATH_STDPATH
+#define _PATH_STDPATH           "/usr/bin:/bin:/usr/sbin:/sbin"
 #endif
 
-#define LOGIT_PATH              FINIT_LIBPATH_ "/logit"
+#ifndef _PATH_VARRUN
+#define _PATH_VARRUN            "/var/run/"
+#endif
+
+#ifndef FINIT_CGPATH
+#define FINIT_CGPATH            "/sys/fs/cgroup"
+#endif
+
+#define _PATH_LOGIT             FINIT_LIBPATH_ "/logit"
+#define _PATH_SULOGIN           FINIT_LIBPATH_ "/sulogin"
 
 #define CMD_SIZE                256
 #define LINE_SIZE               1024
@@ -62,10 +73,10 @@
 /* Finit extensions over std SysV */
 #define INIT_CMD_DEBUG          8    /* Toggle Finit debug */
 #define INIT_CMD_RELOAD         9    /* Reload *.conf in /etc/finit.d/ */
-#define INIT_CMD_START_SVC      10
+#define INIT_CMD_START_SVC      10   /* START service */
 #define INIT_CMD_STOP_SVC       11
 #define INIT_CMD_RELOAD_SVC     12   /* SIGHUP service */
-#define INIT_CMD_RESTART_SVC    13   /* STOP + START service */
+#define INIT_CMD_RESTART_SVC    13   /* START service, clearing blocks */
 #define INIT_CMD_UNUSED2        14   /* Unused, was INIT_CMD_QUERY_INETD */
 #define INIT_CMD_UNUSED1        15   /* Unused, was INIT_CMD_EMIT */
 #define INIT_CMD_GET_RUNLEVEL   16
@@ -96,6 +107,7 @@ extern int    prevlevel;
 extern int    debug;
 extern int    rescue;
 extern int    single;
+extern int    bootstrap;
 extern char  *rcsd;
 extern char  *sdown;
 extern char  *network;

src/getty.c

@@ -41,9 +41,6 @@
 #define _PATH_LOGIN  "/bin/login"
 #endif
 
-#define print(s) (void)write(STDOUT_FILENO, s, strlen(s))
-
-
 /*
  * Read one character from stdin.
  */
@@ -54,7 +51,7 @@ static int readch(char *tty)
 
 	st = read(STDIN_FILENO, &ch1, 1);
 	if (st == 0) {
-		print("\n");
+		dprint(STDOUT_FILENO, "\n", 0);
 		_exit(0);
 	}
 
@@ -74,31 +71,32 @@ static void do_parse(char *line, struct utsname *uts, char *tty)
 	s0 = line;
 	for (s = line; *s != 0; s++) {
 		if (*s == '\\') {
-			(void)write(1, s0, s - s0);
+			if ((s - s0) > 0)
+				dprint(STDOUT_FILENO, s0, s - s0);
 			s0 = s + 2;
 			switch (*++s) {
 			case 'l':
-				print(tty);
+				dprint(STDOUT_FILENO, tty, 0);
 				break;
 			case 'm':
-				print(uts->machine);
+				dprint(STDOUT_FILENO, uts->machine, 0);
 				break;
 			case 'n':
-				print(uts->nodename);
+				dprint(STDOUT_FILENO, uts->nodename, 0);
 				break;
 #ifdef _GNU_SOURCE
 			case 'o':
-				print(uts->domainname);
+				dprint(STDOUT_FILENO, uts->domainname, 0);
 				break;
 #endif
 			case 'r':
-				print(uts->release);
+				dprint(STDOUT_FILENO, uts->release, 0);
 				break;
 			case 's':
-				print(uts->sysname);
+				dprint(STDOUT_FILENO, uts->sysname, 0);
 				break;
 			case 'v':
-				print(uts->version);
+				dprint(STDOUT_FILENO, uts->version, 0);
 				break;
 			case 0:
 				goto leave;
@@ -109,7 +107,8 @@ static void do_parse(char *line, struct utsname *uts, char *tty)
 	}
 
 leave:
-	(void)write(1, s0, s - s0);
+	if ((s - s0) > 0)
+		dprint(STDOUT_FILENO, s0, s - s0);
 }
 
 /*
@@ -159,7 +158,7 @@ static void do_getty(char *tty, char *name, size_t len)
 		while ((ch = readch(tty)) != '\n') {
 			if (ch == CTRL('U')) {
 				while (np > name) {
-					(void)write(1, "\b \b", 3);
+					dprint(STDOUT_FILENO, "\b \b", 3);
 					np--;
 				}
 				continue;
@@ -186,12 +185,12 @@ static int do_login(char *name)
 {
 	struct stat st;
 
-	cgroup_user(name);
+	cgroup_user(name, 0);
 	execl(_PATH_LOGIN, _PATH_LOGIN, name, NULL);
 
 	/*
-	 * Failed to exec login, should be impossible.  Try a starting a
-	 * fallback shell instead.
+	 * Failed to exec login, should not happen on normal systems.
+	 * Try a starting a rescue shell instead.
 	 *
 	 * Note: Add /etc/securetty handling.
 	 */
@@ -215,7 +214,7 @@ int getty(char *tty, speed_t speed, char *term, char *user)
 
 	/* Set up TTY, re-enabling ISIG et al. */
 	stty(STDIN_FILENO, speed);
-	(void)write(STDERR_FILENO, cln, strlen(cln));
+	dprint(STDERR_FILENO, cln, strlen(cln));
 
 	/* The getty process is responsible for the UTMP login record */
 	utmp_set_login(tty, NULL);

src/helpers.c

@@ -103,6 +103,9 @@ static void add_console(char *cons)
 	if (ptr)
 		*ptr = 0;
 
+	if (!strncmp(cons, "/dev/", 5))
+		cons += 5;
+
 	snprintf(path, sizeof(path), "/dev/%s", cons);
 
 	/* check for and skip duplicates */
@@ -121,7 +124,7 @@ static void add_console(char *cons)
 	num_cons++;
 }
 
-void console_init(void)
+static void find_active_consoles(void)
 {
 	char *cons, *ptr, *tok;
 	char buf[512];
@@ -131,7 +134,7 @@ void console_init(void)
 	if (!fp) {
 	fallback:
 		_d("Using fallback /dev/console");
-		add_console("/dev/console");
+		add_console("console");
 		return;
 	}
 
@@ -148,6 +151,12 @@ void console_init(void)
 	}
 }
 
+void console_init(void)
+{
+	find_active_consoles();
+	ttinit();
+}
+
 ssize_t cprintf(const char *fmt, ...)
 {
 	const size_t len = strlen(fmt) * 2;
@@ -161,7 +170,7 @@ ssize_t cprintf(const char *fmt, ...)
 	va_end(ap);
 
 	for (i = 0; i < num_cons; i++)
-		(void)write(fds[i], buf, size);
+		dprint(fds[i], buf, size);
 
 	return size;
 }
@@ -246,20 +255,20 @@ static char *pad(char *buf, size_t len, char *ch, size_t width)
 
 void print_banner(const char *heading)
 {
-	char buf[4 * SCREEN_WIDTH];
+	char buf[4 * ttcols];
 
 	if (progress_style == PROGRESS_SILENT)
 		return;
 
-	memset(buf, 0, sizeof(buf));
+	buf[0] = 0;
 	strlcat(buf, "\r\e[2K", sizeof(buf));
 
 	if (progress_style == PROGRESS_CLASSIC) {
 		strlcat(buf, "\e[1m", sizeof(buf));
 		strlcat(buf, heading, sizeof(buf));
-		pad(buf, sizeof(buf), "=", SCREEN_WIDTH + 10);
+		pad(buf, sizeof(buf), "=", ttcols + 10);
 	} else {
-		size_t wmax = 80 <= SCREEN_WIDTH ? 80 : SCREEN_WIDTH;
+		size_t wmax = 80 <= ttcols ? 80 : ttcols;
 
 		/* • • • Foo System ═══════════════════════════════════════════════ */
 		/* ⬤ ⬤ ⬤ Foo System ═══════════════════════════════════════════════ */
@@ -322,7 +331,7 @@ static char *status(int rc)
 
 void printv(const char *fmt, va_list ap)
 {
-	char buf[SCREEN_WIDTH];
+	char buf[ttcols];
 	size_t len;
 
 	if (!fmt || progress_style == PROGRESS_SILENT)
@@ -330,7 +339,7 @@ void printv(const char *fmt, va_list ap)
 
 	delline();
 
-	memset(buf, 0, sizeof(buf));
+	buf[0] = 0;
 	len = print_timestamp(buf, sizeof(buf));
 	vsnprintf(&buf[len], sizeof(buf) - len, fmt, ap);
 
@@ -404,6 +413,64 @@ int getgroup(char *group)
 #endif
 }
 
+/* get current user */
+int getcuser(char *buf, size_t len)
+{
+	char *ptr = "root";
+
+#ifndef ENABLE_STATIC
+	struct passwd *pw = getpwuid(getuid());
+
+	if (pw)
+		ptr = pw->pw_name;
+#endif
+	if (strlcpy(buf, ptr, len) > len)
+		return -1;
+
+	return 0;
+}
+
+/* get current group */
+int getcgroup(char *buf, size_t len)
+{
+	char *ptr = "root";
+
+#ifndef ENABLE_STATIC
+	struct group *gr = getgrgid(getgid());
+
+	if (gr)
+		ptr = gr->gr_name;
+#endif
+	if (strlcpy(buf, ptr, len) > len)
+		return -1;
+
+	return 0;
+}
+
+int mksubsys(const char *dir, mode_t mode, char *user, char *group)
+{
+	mode_t omask;
+	int uid, gid;
+	int rc = 0;
+
+	omask = umask(0);
+
+	uid = getuser(user, NULL);
+	if (uid >= 0) {
+		gid = getgroup(group);
+		if (gid < 0)
+			gid = 0;
+
+		rc = makedir(dir, mode);
+		if (chown(dir, uid, gid))
+			_pe("Failed chown(%s, %d, %d)", dir, uid, gid);
+	}
+
+	umask(omask);
+
+	return rc;
+}
+
 void set_hostname(char **hostname)
 {
 	FILE *fp;
@@ -426,14 +493,16 @@ void set_hostname(char **hostname)
 			return;
 		}
 
-		fgets(*hostname, st.st_size, fp);
-		chomp(*hostname);
+		if (fgets(*hostname, st.st_size, fp))
+			chomp(*hostname);
 		fclose(fp);
 	}
 
 done:
-	if (*hostname)
-		sethostname(*hostname, strlen(*hostname));
+	if (*hostname) {
+		if (sethostname(*hostname, strlen(*hostname)))
+			_pe("Failed sethostnaem(%s)", *hostname);
+	}
 }
 
 /*
@@ -471,6 +540,7 @@ void networking(int updown)
 				run("ifdown -a");
 			_exit(0);
 		}
+		cgroup_service("network", pid, NULL);
 
 		print(pid > 0 ? 0 : 1, "%s networking", updown ? "Starting" : "Stopping");
 	}
@@ -557,10 +627,12 @@ int ismnt(char *file, char *dir, char *mode)
 
 	while ((mnt = getmntent(fp))) {
 		if (!strcmp(mnt->mnt_dir, dir)) {
-			if (mode && hasopt(mnt->mnt_opts, mode))
-				found = 1;
-			else
+			if (mode) {
+				if (hasopt(mnt->mnt_opts, mode))
+					found = 1;
+			} else
 				found = 1;
+
 			break;
 		}
 	}

src/helpers.h

@@ -26,6 +26,7 @@
 #define FINIT_HELPERS_H_
 
 #include <mntent.h>
+#include <sched.h>		/* sched_yield() */
 #include <stdarg.h>
 #include <sys/resource.h>
 #include <sys/types.h>
@@ -70,7 +71,7 @@ void    show_progress   (pstyle_t style);
 int     getty           (char *tty, speed_t speed, char *term, char *user);
 int     sh              (char *tty);
 
-int     stty            (int fd, speed_t speed);
+void    stty            (int fd, speed_t speed);
 speed_t stty_parse_speed(char *baud);
 
 void    print_banner    (const char *heading);
@@ -81,6 +82,12 @@ int     print_result    (int fail);
 
 int     getuser         (char *username, char **home);
 int     getgroup        (char *group);
+
+int     getcuser        (char *buf, size_t len);
+int     getcgroup       (char *buf, size_t len);
+
+int     mksubsys        (const char *dir, mode_t mode, char *user, char *group);
+
 void    set_hostname    (char **hostname);
 void    networking      (int updown);
 
@@ -103,6 +110,24 @@ static inline int create(char *path, mode_t mode, uid_t uid, gid_t gid)
 	return 0;
 }
 
+static inline int dprint(int fd, const char *s, size_t len)
+{
+	size_t loop = 3;
+	int rc = -1;
+
+	if (!len)
+		len = strlen(s);
+
+	while (loop--) {
+		rc = write(fd, s, len);
+		if (rc == -1 && errno == EINTR)
+			continue;
+		break;
+	}
+
+	return rc;
+}
+
 int	ismnt		(char *file, char *dir, char *mode);
 int	fismnt		(char *dir);
 

src/initctl.h

@@ -0,0 +1,46 @@
+/* New client tool, replaces old /dev/initctl API and telinit tool
+ *
+ * Copyright (c) 2015-2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef FINIT_INITCTL_H_
+#define FINIT_INITCTL_H_
+
+#include "finit.h"
+#include "util.h"
+
+extern int icreate;			/* initctl -c */
+extern int iforce;			/* initctl -f */
+extern int ionce;			/* initctl -1 */
+extern int heading;			/* initctl -t */
+extern int verbose;			/* initctl -v */
+extern int plain;			/* initctl -p */
+
+extern void print_header(const char *fmt, ...);
+
+#endif /* FINIT_INITCTL_H_ */
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */

src/iwatch.c

@@ -81,10 +81,10 @@ int iwatch_add(struct iwatch *iw, char *file, uint32_t mask)
 		return -1;
 
 	if (!fexist(file)) {
-		_d("No such file or directory, skipping %s", file);
+		_d("skipping %s: no such file or directory", file);
 		return 0;
 	}
-	_d("Adding new watcher for path %s", file);
+	_d("adding new watcher for path %s", file);
 
 	path = strdup(file);
 	if (!path) {
@@ -94,7 +94,24 @@ int iwatch_add(struct iwatch *iw, char *file, uint32_t mask)
 
 	wd = inotify_add_watch(iw->fd, path, IWATCH_MASK | mask);
 	if (wd < 0) {
-		_pe("Failed adding watcher for %s", path);
+		switch (errno) {
+		case EEXIST:
+			break;
+
+		case EINVAL:
+			/*
+			 * some older kernels, notably 4.19, don't like
+			 * inotify on top-level cgroups in cgroup v2
+			 */
+			if (!strncmp(path, "/sys/fs/cgroup", 14))
+				break;	/* ignore error */
+			/* fallthrough */
+
+		default:
+			_pe("Failed adding watcher for %s", path);
+			break;
+		}
+
 		free(path);
 		return -1;
 	}

src/keventd.c

@@ -0,0 +1,278 @@
+/* Listens to kernel events like AC power status and manages sys/ conditions
+ *
+ * Copyright (c) 2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include <errno.h>
+#include <dirent.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <syslog.h>
+#include <unistd.h>
+
+#include <sys/poll.h>
+#include <sys/socket.h>
+#include <sys/types.h>
+
+#include <linux/types.h>
+#include <linux/netlink.h>
+
+#include "lite/lite.h"
+
+#include "cond.h"
+#include "pid.h"
+#include "util.h"
+
+#define _PATH_SYSFS_PWR  "/sys/class/power_supply"
+
+static int num_ac_online;
+static int num_ac;
+
+int debug;			/* debug in other modules as well */
+
+void logit(int prio, const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	if (!debug)
+		vsyslog(prio, fmt, ap);
+	else {
+		vfprintf(stderr, fmt, ap);
+		fprintf(stderr, "\n");
+	}
+	va_end(ap);
+}
+#define panic(fmt, args...) { logit(LOG_CRIT, fmt ":%s", ##args, strerror(errno)); exit(1); }
+#define warn(fmt, args...)  { logit(LOG_WARNING, fmt ":%s", ##args, strerror(errno)); }
+
+static void sys_cond(char *cond, int set)
+{
+	char oneshot[256];
+
+	snprintf(oneshot, sizeof(oneshot), "%s/%s", _PATH_CONDSYS, cond);
+	if (set) {
+		if (symlink(_PATH_RECONF, oneshot) && errno != EEXIST)
+			warn("failed asserting sys/%s", cond);
+	} else {
+		if (erase(oneshot) && errno != ENOENT)
+			warn("failed asserting sys/%s", cond);
+	}
+}
+
+static int fgetln(char *path, char *buf, size_t len)
+{
+	FILE *fp;
+
+	fp = fopen(path, "r");
+	if (!fp)
+		return -1;
+
+	if (!fgets(buf, len, fp)) {
+		fclose(fp);
+		return -1;
+	}
+
+	chomp(buf);
+	fclose(fp);
+
+	return 0;
+}
+
+static int check_online(char *online)
+{
+	int val;
+
+	if (!online)
+		return 0;
+
+	val = atoi(online);
+	logit(LOG_INFO, "AC %s", val ? "connected" : "disconnected");
+
+	return val;
+}
+
+static int is_ac(char *type)
+{
+	char *types[] = {
+		"Mains",
+		"USB",
+		"BrickID",
+		"Wireless",
+		NULL
+	};
+	int i;
+
+	for (i = 0; types[i]; i++) {
+		if (!strncmp(type, types[i], strlen(types[i])))
+			return 1;
+	}
+
+	return 0;
+}
+
+static void init(void)
+{
+	struct dirent **d = NULL;
+	char *cond_dirs[] = {
+		_PATH_CONDSYS,
+		_PATH_CONDSYS "/pwr",
+	};
+	char path[384];
+	int i, n;
+
+	for (i = 0; i < (int)NELEMS(cond_dirs); i++) {
+		if (mkpath(cond_dirs[i], 0755) && errno != EEXIST) {
+			_pe("Failed creating %s condition directory, %s", COND_SYS,
+			    cond_dirs[i]);
+			return;
+		}
+	}
+
+	n = scandir(_PATH_SYSFS_PWR, &d, NULL, alphasort);
+	for (i = 0; i < n; i++) {
+		char *nm = d[i]->d_name;
+ 		char buf[10];
+
+		snprintf(path, sizeof(path), "%s/%s/type", _PATH_SYSFS_PWR, nm);
+		if (!fgetln(path, buf, sizeof(buf)) && is_ac(buf)) {
+			num_ac++;
+
+			snprintf(path, sizeof(path), "%s/%s/online", _PATH_SYSFS_PWR, nm);
+			if (!fgetln(path, buf, sizeof(buf))) {
+				if (check_online(buf))
+					num_ac_online++;
+			}
+		}
+		free(d[i]);
+	}
+
+	if (n > 0)
+		free(d);
+
+	/* if any power_supply is online, or none can be found */
+	if (num_ac == 0 || num_ac_online > 0)
+		sys_cond("pwr/ac", 1);
+}
+
+/*
+ * Started by Finit as soon as possible when base filesystem is up,
+ * modules have been probed, or insmodded from /etc/finit.conf, so by
+ * now we should have /sys/class/power_supply/ available for probing.
+ * If none is found we assert /sys/pwr/ac condition anyway, this is what
+ * systemd does (ConditionACPower) and also makes most sense.
+ */
+int main(int argc, char *argv[])
+{
+	struct sockaddr_nl nls = { 0 };
+	struct pollfd pfd;
+	char buf[1024];
+
+	if (argc > 1) {
+		if (!strcmp(argv[1], "-d"))
+			debug = 1;
+	}
+
+	if (!debug) {
+		openlog("keventd", LOG_CONS | LOG_PID, LOG_DAEMON);
+		setlogmask(LOG_UPTO(LOG_NOTICE));
+	}
+
+	init();
+
+	pfd.events = POLLIN;
+	pfd.fd = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_KOBJECT_UEVENT);
+	if (pfd.fd == -1)
+		panic("failed creating netlink socket");
+
+	nls.nl_family = AF_NETLINK;
+	nls.nl_pid    = 0;
+	nls.nl_groups = -1;
+	if (bind(pfd.fd, (void *)&nls, sizeof(struct sockaddr_nl)))
+		panic("bind failed");
+
+	logit(LOG_DEBUG, "Waiting for events ...");
+	while (-1 != poll(&pfd, 1, -1)) {
+		char *path = NULL;
+		int ac = 0;
+		int i, len;
+
+		len = recv(pfd.fd, buf, sizeof(buf), MSG_DONTWAIT);
+		if (len == -1) {
+			switch (errno) {
+			case EINTR:
+				continue;
+			case ENOBUFS:
+				warn("lost events");
+				continue;
+			default:
+				panic("unhandled");
+				break;
+			}
+		}
+
+		/* skip libusb events, focus on kernel events/changes */
+		if (strncmp(buf, "change@", 7))
+			continue;
+
+		/* XXX: currently limited to monitoring this subsystem */
+		if (!strstr(buf, "power_supply"))
+			continue;
+
+		i = 0;
+		while (i < len) {
+			char *line = buf + i;
+
+			logit(LOG_DEBUG, "%s", line);
+			if (!strncmp(line, "DEVPATH=", 8)) {
+				path = &line[8];
+				logit(LOG_DEBUG, "Got path %s", path);
+			} else if (!strncmp(line, "POWER_SUPPLY_TYPE=", 18)) {
+				ac = is_ac(&line[18]);
+			} else if (!strncmp(line, "POWER_SUPPLY_ONLINE=", 20) && ac) {
+				if (check_online(&line[20])) {
+					if (!num_ac_online)
+						sys_cond("pwr/ac", 1);
+					num_ac_online++;
+				} else {
+					if (num_ac_online > 0)
+						num_ac_online--;
+					if (!num_ac_online)
+						sys_cond("pwr/ac", 0);
+				}
+			}
+
+			i += strlen(line) + 1;
+		}
+	}
+	close(pfd.fd);
+
+	return 0;
+}
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */

src/log.c

@@ -22,7 +22,6 @@
  */
 
 #include "config.h"
-#include <sched.h>		/* sched_yield() */
 #include <stdio.h>
 #include <stdarg.h>
 #include <lite/lite.h>
@@ -50,15 +49,13 @@ void log_exit(void)
 	 * Unless in debug mode at shutdown, Reinitialize screen,
 	 * terminal may have been resized at runtime
 	 */
-	if (debug)
-		screen_exit();
-	else
-		screen_init();
+	if (!debug)
+		ttinit();
 
 	enable_progress(1);
 }
 
-void log_open(void)
+static void log_open(void)
 {
 	int opts = LOG_PID;
 
@@ -79,10 +76,9 @@ void log_debug(void)
 
 	if (debug) {
 		loglevel = LOG_DEBUG;
-		screen_exit();
 	} else {
 		loglevel = LOG_NOTICE;
-		screen_init();
+		ttinit();
 	}
 	log_open();
 
@@ -103,11 +99,12 @@ void logit(int prio, const char *fmt, ...)
 	if (up || fexist("/dev/log")) {
 		if (!up)
 			log_open();
+
 		vsyslog(prio, fmt, ap);
 		goto done;
 	}
 
-	if (!debug && prio > loglevel)
+	if (prio > loglevel)
 		goto done;
 
 	fp = fopen("/dev/kmsg", "w");

src/logit.c

@@ -30,75 +30,12 @@
 #include <syslog.h>
 #include <unistd.h>
 #include <sys/stat.h>
+#include <lite/lite.h>
 
 static const char version_info[] = PACKAGE_NAME " v" PACKAGE_VERSION;
+extern int logrotate(char *file, int num, off_t sz);
 
 
-static int create(char *path, mode_t mode, uid_t uid, gid_t gid)
-{
-	return mknod(path, S_IFREG | mode, 0) || chown(path, uid, gid);
-}
-
-/*
- * This function triggers a log rotates of @file when size >= @sz bytes
- * At most @num old versions are kept and by default it starts gzipping
- * .2 and older log files.  If gzip is not available in $PATH then @num
- * files are kept uncompressed.
- */
-static int logrotate(char *file, int num, off_t sz)
-{
-	int cnt;
-	struct stat st;
-
-	if (stat(file, &st))
-		return 1;
-
-	if (sz > 0 && S_ISREG(st.st_mode) && st.st_size > sz) {
-		if (num > 0) {
-			size_t len = strlen(file) + 10 + 1;
-			char   ofile[len];
-			char   nfile[len];
-
-			/* First age zipped log files */
-			for (cnt = num; cnt > 2; cnt--) {
-				snprintf(ofile, len, "%s.%d.gz", file, cnt - 1);
-				snprintf(nfile, len, "%s.%d.gz", file, cnt);
-
-				/* May fail because ofile doesn't exist yet, ignore. */
-				(void)rename(ofile, nfile);
-			}
-
-			for (cnt = num; cnt > 0; cnt--) {
-				snprintf(ofile, len, "%s.%d", file, cnt - 1);
-				snprintf(nfile, len, "%s.%d", file, cnt);
-
-				/* May fail because ofile doesn't exist yet, ignore. */
-				(void)rename(ofile, nfile);
-
-				if (cnt == 2 && !access(nfile, F_OK)) {
-					size_t len = 5 + strlen(nfile) + 1;
-					char cmd[len];
-
-					snprintf(cmd, len, "gzip %s", nfile);
-					system(cmd);
-
-					remove(nfile);
-				}
-			}
-
-			if (rename(file, nfile))
-				(void)truncate(file, 0);
-			else
-				create(file, st.st_mode, st.st_uid, st.st_gid);
-		} else {
-			if (truncate(file, 0))
-				syslog(LOG_ERR | LOG_PERROR, "Failed truncating %s during logrotate: %s", file, strerror(errno));
-		}
-	}
-
-	return 0;
-}
-
 static int checksz(FILE *fp, off_t sz)
 {
 	struct stat st;

src/logrotate.c

@@ -0,0 +1,102 @@
+/*
+ * Copyright (c) 2018-2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include <lite/lite.h>
+#include <stdio.h>
+#include <syslog.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+
+static int recreate(char *path, mode_t mode, uid_t uid, gid_t gid)
+{
+	if (mknod(path, S_IFREG | mode, 0) || chown(path, uid, gid)) {
+		syslog(LOG_ERR, "Failed recreating %s: %s", path, strerror(errno));
+		return -1;
+	}
+
+	return 0;
+}
+
+/*
+ * This function triggers a log rotates of @file when size >= @sz bytes
+ * At most @num old versions are kept and by default it starts gzipping
+ * .2 and older log files.  If gzip is not available in $PATH then @num
+ * files are kept uncompressed.
+ */
+int logrotate(char *file, int num, off_t sz)
+{
+	int cnt;
+	struct stat st;
+
+	if (stat(file, &st))
+		return 1;
+
+	if (sz > 0 && S_ISREG(st.st_mode) && st.st_size > sz) {
+		if (num > 0) {
+			size_t len = strlen(file) + 10 + 1;
+			char   ofile[len];
+			char   nfile[len];
+
+			/* First age zipped log files */
+			for (cnt = num; cnt > 2; cnt--) {
+				snprintf(ofile, len, "%s.%d.gz", file, cnt - 1);
+				snprintf(nfile, len, "%s.%d.gz", file, cnt);
+
+				/* May fail because ofile doesn't exist yet, ignore. */
+				if (rename(ofile, nfile) && errno != ENOENT)
+					syslog(LOG_ERR, "Failed logrotate %s: %s",
+					       ofile, strerror(errno));
+			}
+
+			for (cnt = num; cnt > 0; cnt--) {
+				snprintf(ofile, len, "%s.%d", file, cnt - 1);
+				snprintf(nfile, len, "%s.%d", file, cnt);
+
+				/* May fail because ofile doesn't exist yet, ignore. */
+				if (rename(ofile, nfile) && errno != ENOENT) {
+					syslog(LOG_ERR, "Failed logrotate %s: %s",
+					       ofile, strerror(errno));
+					continue;
+				}
+
+				if (cnt == 2 && fexist(nfile)) {
+					if (systemf("gzip %s 2>/dev/null", nfile))
+						continue; /* no gzip, probably */
+
+					(void)remove(nfile);
+				}
+			}
+
+			if (rename(file, nfile))
+				goto fallback;
+			recreate(file, st.st_mode, st.st_uid, st.st_gid);
+		} else {
+		fallback:
+			if (truncate(file, 0))
+				syslog(LOG_ERR, "Failed truncating %s during logrotate: %s",
+				       file, strerror(errno));
+		}
+	}
+
+	return 0;
+}

src/pid.c

@@ -95,18 +95,15 @@ char *pid_get_name(pid_t pid, char *name, size_t len)
 
 char *pid_file(svc_t *svc)
 {
-	static char path[256];
-	char fn[MAX_ARG_LEN];
-
 	if (svc->pidfile[0]) {
 		if (svc->pidfile[0] == '!')
 			return &svc->pidfile[1];
 		return svc->pidfile;
 	}
 
-	snprintf(fn, sizeof(fn), "%s%s.pid", _PATH_VARRUN, basename(svc->cmd));
+	errno = ENOENT;
 
-	return pid_runpath(fn, path, sizeof(path));
+	return NULL;
 }
 
 pid_t pid_file_read(const char *fn)
@@ -143,12 +140,13 @@ int pid_file_create(svc_t *svc)
 	return fclose(fp);
 }
 
-static int pid_realpath(svc_t *svc, char *file)
+int pid_file_set(svc_t *svc, char *file, int not)
 {
-	int not = 0;
-
-	if (!file)
+	if (!file) {
 		file = pid_file(svc);
+		if (!file)
+			return -1;
+	}
 
 	if (file[0] == '!') {
 		not = 1;
@@ -202,7 +200,7 @@ int pid_file_parse(svc_t *svc, char *arg)
 
 		arg += 4;
 		if ((arg[0] == '!' && arg[1] == '/') || arg[0] == '/')
-			return pid_realpath(svc, arg);
+			return pid_file_set(svc, arg, 0);
 
 		if (arg[0] == '!') {
 			arg++;
@@ -213,12 +211,12 @@ int pid_file_parse(svc_t *svc, char *arg)
 		if (len > 4 && strcmp(&path[len - 4], ".pid"))
 			strlcat(path, ".pid", sizeof(path));
 
-		return pid_realpath(svc, path);
+		return pid_file_set(svc, path, 0);
 	}
 
 	/* 'pid' arg, no argument following */
 	if (!strcmp(arg, "pid"))
-		return pid_realpath(svc, NULL);
+		return pid_file_set(svc, NULL, 0);
 
 	return 1;
 }

src/pid.h

@@ -32,6 +32,7 @@ int   pid_alive       (pid_t pid);
 char *pid_get_name    (pid_t pid, char *name, size_t len);
 
 char *pid_file        (svc_t *svc);
+int   pid_file_set    (svc_t *svc, char *file, int not);
 pid_t pid_file_read   (const char *fn);
 int   pid_file_create (svc_t *svc);
 int   pid_file_parse  (svc_t *svc, char *arg);

src/plugin.c

@@ -196,7 +196,7 @@ void plugin_run_hook(hook_point_t no, void *arg)
 
 	PLUGIN_ITERATOR(p, tmp) {
 		if (p->hook[no].cb) {
-			_d("Calling %s hook n:o %d (arg: %p) ...", basename(p->name), no, arg);
+			_d("Calling %s hook n:o %d (arg: %p) ...", basename(p->name), no, arg ?: "NIL");
 			p->hook[no].cb(arg ? arg : p->hook[no].arg);
 		}
 	}

src/reboot.c

@@ -206,7 +206,7 @@ int reboot_main(int argc, char *argv[])
 
 	case CMD_HALT:
 		if (do_halt(NULL))
-			do_kill(SIGUSR1, msg);
+			do_kill(SIGUSR2, msg);
 		break;
 
 	case CMD_POWEROFF:

src/rescue.conf

@@ -1,3 +1,2 @@
 runlevel 1
-hostname rescue
-tty [12345] @console noclear nologin
+tty [12345] rescue

src/sample.conf

@@ -1,4 +1,12 @@
-# sample.conf: template run/task/service stanza for finit
+# sample.conf: template cgroup/run/task/service stanza for finit
+#
+# The top-level cgroups can be defined anywhere, but share the same
+# namespace.  It is up to the user to ensure groups are defined only
+# once, otherwise the last read .conf wins.  See below for assigning
+# run/task/services to cgroups.  The below example creates the cgroup
+# NAME with the following cpu and memory settings:
+#
+#   cgroup NAME cpu.weight:1000 mem.max:65
 #
 # The one-shot commands 'run' and 'task' are called only once per
 # runlevel.  Finit does not restart them when they exit.
@@ -25,6 +33,16 @@
 # |       |       `--------------------------------------------------------------- Optional conditions
 # |        `---------------------------------------------------------------------- Optional Runlevels
 #  `------------------------------------------------------------------------------ Monitored application
+#
+# Each stanza can also hold a 'cgroup' argument, or be prefixed with
+# cgroup.NAME to place all following stanzas in the same group.  In
+# the following example, foo and bar share the cgroup foo but baz
+# runs in its own cgroup baz:
+#
+#   cgroup.foo
+#   service cgroup:cpu.weight:250,mem.max:655350 foo args -- foo desc
+#   service cgroup:cpu.weight:150,mem.max:655350 bar args -- bar desc
+#   service cgroup.baz:cpu.weight:300 baz args -- baz desc
 
 # Debian GNU/Linux: start SSH daemon as soon as basic networking comes up
 #service [2345] <net/lo/up> env:-/etc/default/ssh /usr/sbin/sshd -D $SSHD_OPTS -- OpenSSH daemon
@@ -33,11 +51,11 @@
 #task [S] env:/etc/conf.d/loadkmap zcat $KEYMAP | loadkmap -- Loading keymap
 
 # Alpine Linux: start SSH daemon with $DROPBEAR_OPTS from /etc/conf.d
-#service [2345] env:-/etc/conf.d/dropbear dropbear -R -F $DROPBEAR_OPTS -- Dropbear SSH daemon
+#service [2345] cgroup.user env:-/etc/conf.d/dropbear dropbear -R -F $DROPBEAR_OPTS -- Dropbear SSH daemon
 
 # Handle PWR button to shutdown/reboot -- useful in Qemu (virt-manager)
 # Depends on syslogd having started. Redirect any output to log.
-#service [2345] <pid/syslogd> log acpid -f -- ACPI daemon
+#service [2345] <pid/syslogd> cgroup:mem.max:32000,cpu.max:1000 log acpid -f -- ACPI daemon
 
 # Start rsyslogd as soon as possible, should always run
 # Provides pid/syslogd condition

src/serv.c

@@ -30,10 +30,7 @@
 #include <stdlib.h>
 #include <lite/lite.h>
 
-#include "util.h"
-
-extern int icreate;			/* initctl -c */
-extern int iforce;			/* initctl -f */
+#include "initctl.h"
 
 
 static int calc_width(char *arr[], size_t len)
@@ -53,27 +50,66 @@ static int calc_width(char *arr[], size_t len)
 
 static void do_list(const char *path)
 {
- 	char buf[screen_cols];
+	static int once = 0;
 	int width, num, prev;
 	glob_t gl;
 	size_t i;
 
-	chdir(path);
+	if (chdir(path)) {
+		char *dir, *ptr;
+
+		if (!fexist(path))
+			return;
+		dir = strdupa(path);
+		if (!dir)
+			return;
+
+		ptr = strrchr(dir, '/');
+		if (ptr)
+			*ptr++ = 0;
+		else
+			ptr = dir;
+
+		if (heading)
+			print_header("%s%s ", once ? "\n" : "", dir);
+		puts(ptr);
+		puts("");
+		return;
+	}
+
 	if (glob("*.conf", 0, NULL, &gl))
 		return;
 
 	if (gl.gl_pathc <= 0)
 		goto done;
 
-	snprintf(buf, sizeof(buf), "%s ", path);
-	printheader(NULL, buf, 0);
+	if (plain) {
+		if (heading)
+			print_header("%s%s ", once ? "\n" : "", path);
+		once++;
+
+		for (i = 0; i < gl.gl_pathc; i++) {
+			if (!heading) {
+				char buf[512];
+
+				paste(buf, sizeof(buf), path, gl.gl_pathv[i]);
+				puts(buf);
+			} else
+				puts(gl.gl_pathv[i]);
+		}
+
+		goto done;
+	}
+
+	if (heading)
+		print_header("%s ", path);
 
 	width = calc_width(gl.gl_pathv, gl.gl_pathc);
 	if (width <= 0)
 		goto done;
 
-	num = (screen_cols - 2) / width;
-	if ((num - 1) * 2 + num * width > screen_cols)
+	num = (ttcols - 2) / width;
+	if ((num - 1) * 2 + num * width > ttcols)
 		num--;
 
 	prev = 0;
@@ -118,15 +154,19 @@ int serv_list(char *arg)
 	if (fisdir(FINIT_RCSD))
 		do_list(FINIT_RCSD);
 
+	if (fexist(FINIT_CONF))
+		do_list(FINIT_CONF);
+
 	return 0;
 }
 
 /*
- * Return path to configuration file for 'name', relative to FINIT_RCSD.
- * This may be any of the following, provided sysconfdir is /etc:
+ * Return path to configuration file for 'name'.  This may be any of the
+ * following, provided sysconfdir is /etc:
  *
- *   - /etc/finit.d/$name.conf
  *   - /etc/finit.d/available/$name.conf
+ *   - /etc/finit.d/$name.conf
+ *   - /etc/finit.conf
  *
  * The system *may* have a /etc/finit.d/available/ directory, or it may
  * just use a plain /etc/finit.d/ -- we do not set policy.
@@ -138,6 +178,11 @@ static char *conf(char *path, size_t len, char *name, int creat)
 {
 	char corr[40];
 
+	if (!name || !name[0] || !strcmp(name, "finit") || !strcmp(name, "finit.conf")) {
+		strlcpy(path, FINIT_CONF, len);
+		return path;
+	}
+
 	if (!strstr(name, ".conf")) {
 		snprintf(corr, sizeof(corr), "%s.conf", name);
 		name = corr;
@@ -155,6 +200,10 @@ static char *conf(char *path, size_t len, char *name, int creat)
 	} else
 		strlcat(path, name, len);
 
+	/* fall back to static service unless edit/create */
+	if (!creat && !fexist(path))
+		paste(path, len, FINIT_RCSD, name);
+
 	return path;
 }
 
@@ -208,7 +257,8 @@ int do_disable(char *arg, int check)
 
 	if (chdir(FINIT_RCSD))
 		err(1, "failed cd %s", FINIT_RCSD);
-	chdir("enabled");	   /* System *may* have enabled/ dir. */
+	if (chdir("enabled"))	   /* System *may* have enabled/ dir. */
+		dbg("Failed changing to %s/enabled/: %s", FINIT_RCSD, strerror(errno));
 
 	if (check && stat(arg, &st))
 		errx(1, "%s not (an) enabled (service).", arg);
@@ -250,6 +300,20 @@ int serv_touch(char *arg)
 	return 0;
 }
 
+int serv_show(char *arg)
+{
+	char path[256];
+	char *fn;
+
+	fn = conf(path, sizeof(path), arg, 0);
+	if (!fexist(fn)) {
+		warnx("Cannot find %s", arg);
+		return 1;
+	}
+
+	return systemf("cat %s", fn);
+}
+
 /*
  * Try to open an editor for the given file, if creat is given we
  * create a new file based on /lib/finit/sample.conf
@@ -279,7 +343,7 @@ static int do_edit(char *arg, int creat)
 	if (!fexist(fn)) {
 		if (!creat) {
 			warnx("Cannot find %s, use -c flag, create command, or select one of:", arg);
-			return serv_list("available");
+			return serv_list(NULL);
 		}
 
 #ifdef SAMPLE_CONF
@@ -301,8 +365,8 @@ static int do_edit(char *arg, int creat)
 int serv_edit(char *arg)
 {
 	if (!arg || !arg[0]) {
-		warnx("missing argument to edit, may be one of:");
-		return serv_list("available");
+		if (!yorn("Do you want to edit %s (y/N)? ", FINIT_CONF))
+			return 0;
 	}
 
 	return do_edit(arg, icreate);
@@ -346,8 +410,10 @@ int serv_delete(char *arg)
 	char buf[256];
 	char *fn;
 
-	if (!arg || !arg[0])
-		errx(1, "missing argument to delete");
+	if (!arg || !arg[0]) {
+		warnx("missing argument to delete, may be one of:");
+		return serv_list("available");
+	}
 
 	fn = conf(buf, sizeof(buf), arg, 0);
 	if (!fn)
@@ -358,7 +424,8 @@ int serv_delete(char *arg)
 
 	if (iforce || yorn("Remove file and symlink(s) to %s (y/N)? ", fn)) {
 		do_disable(arg, 0);
-		remove(fn);
+		if (remove(fn))
+			err(1, "Failed removing %s", fn);
 	}
 
 	return 0;

src/serv.h

@@ -28,6 +28,7 @@ int serv_list    (char *arg);
 int serv_enable  (char *arg);
 int serv_disable (char *arg);
 int serv_touch   (char *arg);
+int serv_show    (char *arg);
 int serv_edit    (char *arg);
 int serv_creat   (char *arg);
 int serv_delete  (char *arg);

src/service.c

@@ -25,7 +25,6 @@
 #include "config.h"		/* Generated by configure script */
 
 #include <ctype.h>		/* isblank() */
-#include <sched.h>		/* sched_yield() */
 #include <string.h>
 #include <sys/prctl.h>
 #include <sys/resource.h>
@@ -49,8 +48,6 @@
 #include "utmp-api.h"
 #include "schedule.h"
 
-#define RESPAWN_MAX    10	/* Prevent endless respawn of faulty services. */
-
 static struct wq work = {
 	.cb = service_worker,
 };
@@ -198,6 +195,8 @@ static int lredirect(svc_t *svc)
 		char *tag  = basename(svc->cmd);
 		char *prio = "daemon.info";
 
+		sched_yield();
+
 		fds = open(ptsname(fd), O_RDONLY);
 		close(fd);
 		if (fds == -1)
@@ -207,8 +206,8 @@ static int lredirect(svc_t *svc)
 		/* Reset signals */
 		sig_unblock();
 
-		if (!whichp(LOGIT_PATH)) {
-			logit(LOG_INFO, LOGIT_PATH " missing, using syslog for %s instead", svc->name);
+		if (!whichp(_PATH_LOGIT)) {
+			logit(LOG_INFO, _PATH_LOGIT " missing, using syslog for %s instead", svc->name);
 			fallback_logger(tag, prio);
 			_exit(0);
 		}
@@ -219,7 +218,7 @@ static int lredirect(svc_t *svc)
 			snprintf(sz, sizeof(sz), "%d", logfile_size_max);
 			snprintf(num, sizeof(num), "%d", logfile_count_max);
 
-			execlp(LOGIT_PATH, "logit", "-f", svc->log.file, "-n", sz, "-r", num, NULL);
+			execlp(_PATH_LOGIT, "logit", "-f", svc->log.file, "-n", sz, "-r", num, NULL);
 			_exit(1);
 		}
 
@@ -228,7 +227,7 @@ static int lredirect(svc_t *svc)
 		if (svc->log.prio[0])
 			prio = svc->log.prio;
 
-		execlp(LOGIT_PATH, "logit", "-t", tag, "-p", prio, NULL);
+		execlp(_PATH_LOGIT, "logit", "-t", tag, "-p", prio, NULL);
 		_exit(1);
 	}
 
@@ -262,37 +261,6 @@ static int redirect(svc_t *svc)
 	return 0;
 }
 
-/*
- * non-zero env, no checking if file exists or not
- */
-static char *svc_getenv(svc_t *svc)
-{
-	int v = 0;
-
-	if (!svc->env[0])
-		return NULL;
-
-	if (svc->env[0] == '-')
-		v = 1;
-
-	return &svc->env[v];
-}
-
-/*
- * Check if svc has env, if env file exists returns true
- * if it doesn't exist, but has '-', also true.  if svc
- * doesn't have env, also true.
- */
-static int check_env(svc_t *svc)
-{
-	char *env = svc_getenv(svc);
-
-	if (!env)
-		return 1;
-
-	return fexist(env);
-}
-
 /*
  * Source environment file, if it exists
  * Note: must be called from privsepped child
@@ -309,7 +277,7 @@ static void source_env(svc_t *svc)
 	if (!fn)
 		return;
 
-	/* Warning in service_start() after check_env() */
+	/* Warning in service_start() after svc_checkenv() */
 	fp = fopen(fn, "r");
 	if (!fp)
 		return;
@@ -384,6 +352,84 @@ static int is_norespawn(void)
 		fexist("/tmp/norespawn");
 }
 
+/* used for process group name, derived from originating filename,
+ * so to group multiple services, place them in the same .conf
+ */
+static char *group_name(svc_t *svc, char *buf, size_t len)
+{
+	char *ptr;
+
+	if (!svc->file[0])
+		return svc_ident(svc, buf, len);
+
+	ptr = strrchr(svc->file, '/');
+	if (ptr)
+		ptr++;
+	else
+		ptr = svc->file;
+
+	strlcpy(buf, ptr, len);
+	ptr = strstr(buf, ".conf");
+	if (ptr)
+		*ptr = 0;
+
+	return buf;
+}
+
+static pid_t service_fork(svc_t *svc)
+{
+	pid_t pid;
+
+	pid = fork();
+	if (pid == 0) {
+		char *home = NULL;
+#ifdef ENABLE_STATIC
+		int uid = 0; /* XXX: Fix better warning that dropprivs is disabled. */
+		int gid = 0;
+#else
+		int uid = getuser(svc->username, &home);
+		int gid = getgroup(svc->group);
+#endif
+
+		sched_yield();
+
+		/* Set configured limits */
+		for (int i = 0; i < RLIMIT_NLIMITS; i++) {
+			if (setrlimit(i, &svc->rlimit[i]) == -1)
+				logit(LOG_WARNING,
+				      "%s: rlimit: Failed setting %s",
+				      svc->cmd, rlim2str(i));
+		}
+
+		/* Set desired user+group */
+		if (gid >= 0) {
+			if (setgid(gid))
+				_pe("%s: failed setgid(%d)", svc->cmd, gid);
+		}
+
+		if (uid >= 0) {
+			if (setuid(uid))
+				_pe("%s: failed setuid(%d)", svc->cmd, uid);
+
+			/* Set default path for regular users */
+			if (uid > 0)
+				setenv("PATH", _PATH_DEFPATH, 1);
+			if (home) {
+				setenv("HOME", home, 1);
+				if (chdir(home)) {
+					if (chdir("/"))
+						_pe("%s: failed chdir(%s) and chdir(/)", svc->cmd, home);
+				}
+			}
+		}
+
+		/* Source any environment from env:/path/to/file */
+		source_env(svc);
+	}
+
+	return pid;
+}
+
 /**
  * service_start - Start service
  * @svc: Service to start
@@ -395,6 +441,7 @@ static int service_start(svc_t *svc)
 {
 	int result = 0, do_progress = 1;
 	sigset_t nmask, omask;
+	char grnam[80];
 	pid_t pid;
 	size_t i;
 
@@ -407,17 +454,28 @@ static int service_start(svc_t *svc)
 
 	/* Don't try and start service if it doesn't exist. */
 	if (!whichp(svc->cmd)) {
-		print(1, "%s: does not exist", svc->cmd);
+		logit(LOG_WARNING, "%s: missing or not in $PATH", svc->cmd);
 		svc_missing(svc);
 		return 1;
 	}
 
 	/* Unlike systemd we do not allow starting service if env is missing, unless - */
-	if (!check_env(svc)) {
-		print(1, "%s: environment file %s does not exist (yet)", svc->cmd, svc->env);
+	if (!svc_checkenv(svc)) {
+		logit(LOG_WARNING, "%s: missing env file %s", svc->cmd, svc->env);
+		svc_missing(svc);
 		return 1;
 	}
 
+	if (svc_is_tty(svc) && !svc->notty) {
+		char *dev = tty_canonicalize(svc->dev);
+
+		if (!dev || !tty_exists(dev)) {
+			_d("TTY %s missing or invalid, halting service.", svc->dev);
+			svc_missing(svc);
+			return 1;
+		}
+	}
+
 	if (svc_is_sysv(svc))
 		logit(LOG_CONSOLE | LOG_NOTICE, "Calling '%s start' ...", svc->cmd);
 
@@ -439,47 +497,10 @@ static int service_start(svc_t *svc)
 	sigaddset(&nmask, SIGCHLD);
 	sigprocmask(SIG_BLOCK, &nmask, &omask);
 
-	pid = fork();
-	cgroup_service(svc->name, svc->id, pid);
-
+	pid = service_fork(svc);
 	if (pid == 0) {
-		int status;
-		char *home = NULL;
-#ifdef ENABLE_STATIC
-		int uid = 0; /* XXX: Fix better warning that dropprivs is disabled. */
-		int gid = 0;
-#else
-		int uid = getuser(svc->username, &home);
-		int gid = getgroup(svc->group);
-#endif
 		char *args[MAX_NUM_SVC_ARGS + 1];
-
-		/* Set configured limits */
-		for (int i = 0; i < RLIMIT_NLIMITS; i++) {
-			if (setrlimit(i, &svc->rlimit[i]) == -1)
-				logit(LOG_WARNING,
-				      "%s: rlimit: Failed setting %s",
-				      svc->cmd, rlim2str(i));
-		}
-
-		/* Set desired user+group */
-		if (gid >= 0)
-			setgid(gid);
-
-		if (uid >= 0) {
-			setuid(uid);
-
-			/* Set default path for regular users */
-			if (uid > 0)
-				setenv("PATH", _PATH_DEFPATH, 1);
-			if (home) {
-				setenv("HOME", home, 1);
-				chdir(home);
-			}
-		}
-
-		/* Source any environment from env:/path/to/file */
-		source_env(svc);
+		int status;
 
 		if (!svc_is_sysv(svc)) {
 			wordexp_t we = { 0 };
@@ -543,7 +564,8 @@ static int service_start(svc_t *svc)
 		}
 		args[i] = NULL;
 
-		redirect(svc);
+		if (!svc_is_tty(svc))
+			redirect(svc);
 		sig_unblock();
 
 		/*
@@ -561,6 +583,8 @@ static int service_start(svc_t *svc)
 
 		if (svc_is_runtask(svc))
 			status = exec_runtask(args[0], &args[1]);
+		else if (svc_is_tty(svc))
+			status = tty_exec(svc);
 		else
 			status = execvp(args[0], &args[1]);
 
@@ -578,6 +602,11 @@ static int service_start(svc_t *svc)
 		_d("Starting %s %s", svc->cmd, buf);
 	}
 
+	if (svc_is_tty(svc))
+		cgroup_user("getty", pid);
+	else
+		cgroup_service(group_name(svc, grnam, sizeof(grnam)), pid, &svc->cgroup);
+
 	logit(LOG_CONSOLE | LOG_NOTICE, "Starting %s[%d]", svc_ident(svc, NULL, 0), pid);
 
 	svc->pid = pid;
@@ -646,12 +675,19 @@ static void service_cleanup(svc_t *svc)
 {
 	char *fn;
 
+	/* PID collected, cancel any pending SIGKILL */
+	service_timeout_cancel(svc);
+
 	fn = pid_file(svc);
-	if (remove(fn) && errno != ENOENT)
+	if (fn && remove(fn) && errno != ENOENT)
 		logit(LOG_CRIT, "Failed removing service %s pidfile %s",
 		      basename(svc->cmd), fn);
 
 	/* No longer running, update books. */
+	if (svc_is_tty(svc) && svc->pid > 1)
+		utmp_set_dead(svc->pid); /* Set DEAD_PROCESS UTMP entry */
+
+	svc->oldpid = svc->pid;
 	svc->start_time = svc->pid = 0;
 }
 
@@ -695,7 +731,7 @@ static int service_stop(svc_t *svc)
 		if (svc->pid > 1) {
 			/* Kill all children in the same proess group, e.g. logit */
 			rc = kill(-svc->pid, svc->sighalt);
-
+			_d("kill(-%d, %d) => rc %d", svc->pid, svc->sighalt, rc);
 			/* PID lost or forking process never really started */
 			if (rc == -1 && ESRCH == errno)
 				service_cleanup(svc);
@@ -862,6 +898,30 @@ static void parse_env(svc_t *svc, char *env)
 	strlcpy(svc->env, env, sizeof(svc->env));
 }
 
+static void parse_cgroup(svc_t *svc, char *cgroup)
+{
+	char *ptr = cgroup;
+
+	if (!cgroup)
+		return;
+
+	if (cgroup[0] == '.') {
+		ptr = strchr(cgroup, ':');
+		if (ptr)
+			*ptr++ = 0;
+		strlcpy(svc->cgroup.name, &cgroup[1], sizeof(svc->cgroup.name));
+		if (!ptr)
+			return;
+	}
+
+	if (strlen(ptr) >= sizeof(svc->cgroup)) {
+		_e("%s: cgroup settings too long (>%d chars)", svc->cmd, sizeof(svc->cgroup));
+		return;
+	}
+
+	strlcpy(svc->cgroup.cfg, ptr, sizeof(svc->cgroup.cfg));
+}
+
 static void parse_sighalt(svc_t *svc, char *arg)
 {
 	int signo;
@@ -888,6 +948,14 @@ static void parse_killdelay(svc_t *svc, char *delay)
 	svc->killdelay = (int)(sec * 1000);
 }
 
+static void parse_script(char *type, char *script, char *buf, size_t len)
+{
+	if (access(script, X_OK))
+		logit(LOG_WARNING, "%s: %s:%s is missing or not executable, skipping.", type, script);
+	else
+		strlcpy(buf, script, len);
+}
+
 /*
  * name:<name>
  */
@@ -903,13 +971,10 @@ static void parse_name(svc_t *svc, char *arg)
 	}
 
 	strlcpy(svc->name, name, sizeof(svc->name));
-
-	/* Warn if svc generates same condition as an existing service */
-	svc_validate(svc);
 }
 
-/**
- * parse_cmdline_args - Update the command line args in the svc struct
+/*
+ * Update the command line args in the svc struct
  *
  * strtok internal pointer must be positioned at first command line arg
  * when this function is called.
@@ -918,23 +983,29 @@ static void parse_name(svc_t *svc, char *arg)
  */
 static void parse_cmdline_args(svc_t *svc, char *cmd)
 {
+	int diff = 0;
 	char sep = 0;
 	char *arg;
 	int i = 0;
 
+	if (strcmp(svc->args[i], cmd))
+		diff++;
 	strlcpy(svc->args[i++], cmd, sizeof(svc->args[0]));
 
-	while (i < MAX_NUM_SVC_ARGS)
-		svc->args[i++][0] = 0;
-
 	/*
 	 * Copy supplied args. Stop at MAX_NUM_SVC_ARGS-1 to allow the args
 	 * array to be zero-terminated.
 	 */
-	for (i = 1; (arg = strtok(NULL, " ")) && i < (MAX_NUM_SVC_ARGS - 1);) {
+	while ((arg = strtok(NULL, " ")) && i < (MAX_NUM_SVC_ARGS - 1)) {
+		char prev[sizeof(svc->args[0])] = { 0 };
 		char ch = arg[0];
 		size_t len;
 
+		if (!sep) {
+			strlcpy(prev, svc->args[i], sizeof(prev));
+			svc->args[i][0] = 0;
+		}
+
 		/* XXX: ugly string arg re-concatenation, fixme */
 		if (ch == '"' || ch == '\'')
 			sep = ch;
@@ -951,6 +1022,13 @@ static void parse_cmdline_args(svc_t *svc, char *cmd)
 				continue;
 		}
 
+		/* replace any @console arg with the expanded device name */
+		if (svc_is_tty(svc) && tty_isatcon(svc->args[i]))
+			strlcpy(svc->args[i], svc->dev, sizeof(svc->args[i]));
+
+		if (strcmp(svc->args[i], prev))
+			diff++;
+
 		sep = 0;
 		i++;
 	}
@@ -959,15 +1037,22 @@ static void parse_cmdline_args(svc_t *svc, char *cmd)
 	 * Clear remaining args in case they were set earlier.
 	 * This also zero-terminates the args array.
 	 */
-	while (i < MAX_NUM_SVC_ARGS)
-		svc->args[i++][0] = 0;
-#if 0
-	for (i = 0; i < MAX_NUM_SVC_ARGS; i++) {
-		if (!svc->args[i][0])
-			break;
-		_d("%s ", svc->args[i]);
+	while (i < MAX_NUM_SVC_ARGS) {
+		if (svc->args[i++][0]) {
+			svc->args[i-1][0] = 0;
+			diff++;
+		}
 	}
-#endif
+
+	/*
+	 * Check also for changes to /etc/default/foo, because this
+	 * also constitutes changes to command line args.
+	 */
+	diff += conf_changed(svc_getenv(svc));
+
+	if (diff)
+		_d("Modified args for %s detected", cmd);
+	svc->args_dirty = (diff > 0);
 }
 
 
@@ -1031,7 +1116,10 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 	char *cmd, *desc, *runlevels = NULL, *cond = NULL;
 	char *username = NULL, *log = NULL, *pid = NULL;
 	char *name = NULL, *halt = NULL, *delay = NULL;
-	char *id = NULL, *env = NULL;
+	char *id = NULL, *env = NULL, *cgroup = NULL;
+	char *pre_script = NULL, *post_script = NULL;
+	struct tty tty = { 0 };
+	char *dev = NULL;
 	int levels = 0;
 	int manual = 0;
 	char *line;
@@ -1042,7 +1130,7 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 		return errno = EINVAL;
 	}
 
-	line = strdup(cfg);
+	line = strdupa(cfg);
 	if (!line)
 		return 1;
 
@@ -1064,15 +1152,17 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 		}
 	}
 
-	cmd = strtok(line, " ");
+	cmd = strtok(line, " \t");
 	if (!cmd) {
 	incomplete:
 		_e("Incomplete service '%s', cannot register", cfg);
-		free(line);
 		return errno = ENOENT;
 	}
 
 	while (cmd) {
+		if (type == SVC_TYPE_TTY && cmd[0] == '@')
+			break;		/* @console */
+
 		if (cmd[0] == '@')	/* @username[:group] */
 			username = &cmd[1];
 		else if (cmd[0] == '[')	/* [runlevels] */
@@ -1093,8 +1183,16 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 			halt = &cmd[5];
 		else if (!strncasecmp(cmd, "kill:", 5))
 			delay = &cmd[5];
+		else if (!strncasecmp(cmd, "pre:", 4))
+			pre_script = &cmd[4];
+		else if (!strncasecmp(cmd, "post:", 5))
+			post_script = &cmd[5];
 		else if (!strncasecmp(cmd, "env:", 4))
 			env = &cmd[4];
+		else if (!strncasecmp(cmd, "cgroup:", 7))
+			cgroup = &cmd[7]; /* only settings */
+		else if (!strncasecmp(cmd, "cgroup.", 7))
+			cgroup = &cmd[6]; /* with group */
 		else
 			break;
 
@@ -1107,25 +1205,66 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 	levels = conf_parse_runlevels(runlevels);
 	if (runlevel > 0 && !ISOTHER(levels, 0)) {
 		_d("Skipping %s, bootstrap is completed.", cmd);
-		free(line);
 		return 0;
 	}
 
 	if (!id)
 		id = "";
 
-	svc = svc_find(cmd, id);
+	if (type == SVC_TYPE_TTY) {
+		size_t i, len = 0;
+
+		if (tty_parse_args(cmd, &tty))
+			return errno;
+
+		if (tty.cmd)
+			len += strlen(tty.cmd);
+		else
+			len += 3;
+
+		len += tty.num + 1;
+		for (i = 0; i < tty.num; i++)
+			len += strlen(tty.args[i]) + 1;
+
+		line = alloca(len);
+		if (!line)
+			return errno;
+
+		snprintf(line, len, "%s", tty.cmd ?: "tty");
+		for (i = 0; i < tty.num; i++) {
+			strlcat(line, " ", len);
+			strlcat(line, tty.args[i], len);
+		}
+
+		cmd = strtok(line, " \t");
+		if (!cmd)
+			return errno;
+
+		if (tty_isatcon(tty.dev))
+			dev = tty_atcon();
+		else
+			dev = tty.dev;
+	next:
+		svc = svc_find_by_tty(dev);
+	} else
+		svc = svc_find(cmd, id);
+
 	if (!svc) {
 		_d("Creating new svc for %s id #%s type %d", cmd, id, type);
 		svc = svc_new(cmd, id, type);
 		if (!svc) {
 			_e("Out of memory, cannot register service %s", cmd);
-			free(line);
 			return errno = ENOMEM;
 		}
 
 		if (type == SVC_TYPE_SERVICE && manual)
 			svc_stop(svc);
+	} else {
+		/* update type, may have changed from service -> task */
+		svc->type = type;
+
+		/* e.g., if missing cmd or env before */
+		svc_unblock(svc);
 	}
 
 	/* Always clear svc PID file, for now.  See TODO */
@@ -1142,42 +1281,103 @@ int service_register(int type, char *cfg, struct rlimit rlimit[], char *file)
 			strlcpy(svc->group, ptr, sizeof(svc->group));
 		}
 		strlcpy(svc->username, username, sizeof(svc->username));
+	} else {
+		getcuser(svc->username, sizeof(svc->username));
+		getcgroup(svc->group, sizeof(svc->group));
 	}
 
-	parse_cmdline_args(svc, cmd);
-
 	svc->runlevels = levels;
-	_d("Service %s runlevel 0x%2x", svc->cmd, svc->runlevels);
+	_d("Service %s runlevel 0x%02x", svc->cmd, svc->runlevels);
 
 	conf_parse_cond(svc, cond);
 
-	parse_name(svc, name);
+	if (type == SVC_TYPE_TTY) {
+		char *ptr;
+
+		if (dev)
+			strlcpy(svc->dev, dev, sizeof(svc->dev));
+		if (tty.baud)
+			strlcpy(svc->baud, tty.baud, sizeof(svc->baud));
+		if (tty.term)
+			strlcpy(svc->term, tty.term, sizeof(svc->term));
+		svc->noclear = tty.noclear;
+		svc->nowait  = tty.nowait;
+		svc->nologin = tty.nologin;
+		svc->notty   = tty.notty;
+		svc->rescue  = tty.rescue;
+
+		/* TTYs cannot be redirected */
+		log = NULL;
+
+		/* Create name:id tuple for identiy, e.g., tty:S0 */
+		ptr = strrchr(svc->dev, '/');
+		if (ptr) {
+			ptr++;
+			if (!strncmp(ptr, "tty", 3))
+				ptr += 3;
+		} else
+			ptr = svc->dev;
+		if (!id || id[0] == 0)
+			id = ptr;
+		strlcpy(svc->name, "tty", sizeof(svc->name));
+		strlcpy(svc->id, id, sizeof(svc->id));
+	} else
+		parse_name(svc, name);
+
+	parse_cmdline_args(svc, cmd);
+
+	/*
+	 * Warn if svc generates same condition (based on name:id)
+	 * as an existing service.
+	 */
+	svc_validate(svc);
+
 	if (halt)
 		parse_sighalt(svc, halt);
 	if (delay)
 		parse_killdelay(svc, delay);
+	if (pre_script)
+		parse_script("pre", pre_script, svc->pre_script, sizeof(svc->pre_script));
+	if (post_script)
+		parse_script("post", post_script, svc->post_script, sizeof(svc->post_script));
 	if (log)
 		parse_log(svc, log);
 	if (desc)
 		strlcpy(svc->desc, desc, sizeof(svc->desc));
+	else if (type == SVC_TYPE_TTY)
+		snprintf(svc->desc, sizeof(svc->desc), "Getty on %s", svc->dev);
 	if (env)
 		parse_env(svc, env);
+	if (file)
+		strlcpy(svc->file, file, sizeof(svc->file));
 
 	/* Set configured limits */
 	memcpy(svc->rlimit, rlimit, sizeof(svc->rlimit));
 
+	/* Seed with currently active group, may be empty */
+	strlcpy(svc->cgroup.name, cgroup_current, sizeof(svc->cgroup.name));
+	if (cgroup)
+		parse_cgroup(svc, cgroup);
+
 	/* New, recently modified or unchanged ... used on reload. */
 	if ((file && conf_changed(file)) || conf_changed(svc_getenv(svc)))
 		svc_mark_dirty(svc);
 	else
 		svc_mark_clean(svc);
 
+	svc_enable(svc);
+
 	/* for finit native services only, e.g. plugins/hotplug.c */
 	if (!file)
 		svc->protect = 1;
 
-	/* Free duped line, from above */
-	free(line);
+	/* continue expanding any 'tty @console ...' */
+	if (tty_isatcon(tty.dev)) {
+		dev = tty_atcon();
+		if (dev)
+			goto next;
+	}
+
 	return 0;
 }
 
@@ -1205,36 +1405,47 @@ void service_monitor(pid_t lost, int status)
 	if (fexist(SYNC_SHUTDOWN) || lost <= 1)
 		return;
 
-	if (tty_respawn(lost))
-		return;
-
 	svc = svc_find_by_pid(lost);
 	if (!svc) {
 		_d("collected unknown PID %d", lost);
 		return;
 	}
 
-	_d("collected %s(%d), normal exit: %d, signaled: %d, exit code: %d",
-	   svc->cmd, lost, WIFEXITED(status), WIFSIGNALED(status), WEXITSTATUS(status));
-	svc->status = status;
+	switch (svc->state) {
+	case SVC_CLEANUP_STATE:
+	case SVC_SETUP_STATE:
+		_d("collected script %s(%d), normal exit: %d, signaled: %d, exit code: %d",
+		   svc->state == SVC_CLEANUP_STATE ? svc->post_script : svc->pre_script,
+		   lost, WIFEXITED(status), WIFSIGNALED(status), WEXITSTATUS(status));
+		kill(-svc->pid, SIGKILL);
+		goto done;
+
+	default:
+		_d("collected %s(%d), normal exit: %d, signaled: %d, exit code: %d",
+		   svc->cmd, lost, WIFEXITED(status), WIFSIGNALED(status), WEXITSTATUS(status));
+		svc->status = status;
+		break;
+	}
 
 	/* Forking sysv/services declare themselves with pid:!/path/to/pid.file  */
 	if (svc_is_starting(svc) && svc_is_forking(svc))
 		return;
 
+	/* Terminate any children in the same proess group, e.g. logit */
+	kill(-svc->pid, SIGKILL);
+
 	/* Try removing PID file (in case service does not clean up after itself) */
-	if (svc_is_daemon(svc)) {
+	if (svc_is_daemon(svc) || svc_is_tty(svc)) {
 		service_cleanup(svc);
 	} else if (svc_is_runtask(svc)) {
+		/* run/task should run at least once per runlevel */
 		if (WIFEXITED(status) && !WEXITSTATUS(status))
 			svc->started = 1;
 		else
 			svc->started = 0;
 	}
 
-	/* Terminate any children in the same proess group, e.g. logit */
-	kill(-svc->pid, SIGKILL);
-
+done:
 	/* No longer running, update books. */
 	svc->start_time = svc->pid = 0;
 
@@ -1247,6 +1458,112 @@ void service_monitor(pid_t lost, int status)
 	sm_step(&sm);
 }
 
+static void svc_mark_affected(char *cond)
+{
+	svc_t *svc, *iter = NULL;
+
+	for (svc = svc_iterator(&iter, 1); svc; svc = svc_iterator(&iter, 0)) {
+		if (!svc_has_cond(svc))
+			continue;
+
+		if (cond_affects(cond, svc->cond))
+			svc_mark_dirty(svc);
+	}
+}
+
+/*
+ * Called on conf_reload() to update service reverse dependencies.
+ * E.g., if ospfd depends on zebra and the zebra Finit conf has
+ * changed, we need to mark the ospfd Finit conf as changed too.
+ */
+void service_update_rdeps(void)
+{
+	svc_t *svc, *iter = NULL;
+
+	for (svc = svc_iterator(&iter, 1); svc; svc = svc_iterator(&iter, 0)) {
+		char cond[MAX_COND_LEN];
+
+		if (!svc_is_changed(svc))
+			continue;
+
+		svc_mark_affected(mkcond(svc, cond, sizeof(cond)));
+	}
+}
+
+static void service_kill_script(svc_t *svc)
+{
+	if (svc->pid <= 1)
+		return;
+
+	kill(-svc->pid, SIGKILL);
+}
+
+static void service_pre_script(svc_t *svc)
+{
+	svc->pid = service_fork(svc);
+	if (svc->pid < 0) {
+		_pe("Failed forking off %s pre-script %s", svc_ident(svc, NULL, 0), svc->pre_script);
+		return;
+	}
+
+	if (svc->pid == 0) {
+		char *argv[4] = {
+			"sh",
+			"-c",
+			svc->pre_script,
+			NULL
+		};
+
+		setenv("SERVICE_IDENT", svc_ident(svc, NULL, 0), 1);
+		execvp(_PATH_BSHELL, argv);
+		_exit(EX_OSERR);
+	}
+
+	/* Short hard-coded timeout to prevent locking up Finit */
+	service_timeout_after(svc, svc->killdelay, service_kill_script);
+}
+
+static void service_post_script(svc_t *svc)
+{
+	svc->pid = service_fork(svc);
+	if (svc->pid < 0) {
+		_pe("Failed forking off %s post-script %s", svc_ident(svc, NULL, 0), svc->post_script);
+		return;
+	}
+
+	if (svc->pid == 0) {
+		char *argv[4] = {
+			"sh",
+			"-c",
+			svc->post_script,
+			NULL
+		};
+		int rc, sig;
+
+		rc = WEXITSTATUS(svc->status);
+		sig = WTERMSIG(svc->status);
+
+		setenv("SERVICE_IDENT", svc_ident(svc, NULL, 0), 1);
+
+		if (WIFEXITED(svc->status)) {
+			char val[4];
+
+			setenv("EXIT_CODE", "exited", 1);
+			snprintf(val, sizeof(val), "%d", rc & 0xff);
+			setenv("EXIT_STATUS", val, 1);
+		} else if (WIFSIGNALED(svc->status)) {
+			setenv("EXIT_CODE", "signal", 1);
+			setenv("EXIT_STATUS", sig2str(sig), 1);
+		}
+
+		execvp(_PATH_BSHELL, argv);
+		_exit(EX_OSERR);
+	}
+
+	/* Short hard-coded timeout to prevent locking up Finit */
+	service_timeout_after(svc, svc->killdelay, service_kill_script);
+}
+
 static void service_retry(svc_t *svc)
 {
 	int timeout;
@@ -1256,12 +1573,13 @@ static void service_retry(svc_t *svc)
 
 	if (svc->state != SVC_HALTED_STATE ||
 	    svc->block != SVC_BLOCK_RESTARTING) {
-		_d("%s not crashing anymore", svc->cmd);
+		if (!svc_is_tty(svc))
+			_d("%s not crashing anymore", svc->cmd);
 		*restart_cnt = 0;
 		return;
 	}
 
-	if (*restart_cnt >= RESPAWN_MAX) {
+	if (*restart_cnt >= SVC_RESPAWN_MAX) {
 		logit(LOG_CONSOLE | LOG_WARNING, "Service %s keeps crashing, not restarting.",
 		      svc_ident(svc, NULL, 0));
 		svc_crashing(svc);
@@ -1274,12 +1592,12 @@ static void service_retry(svc_t *svc)
 
 	_d("%s crashed, trying to start it again, attempt %d", svc->cmd, *restart_cnt);
 	logit(LOG_CONSOLE | LOG_WARNING, "Service %s[%d] died, restarting (%d/%d)",
-	      svc_ident(svc, NULL, 0), svc->pid, *restart_cnt, RESPAWN_MAX);
+	      svc_ident(svc, NULL, 0), svc->oldpid, *restart_cnt, SVC_RESPAWN_MAX);
 	svc_unblock(svc);
 	service_step(svc);
 
 	/* Wait 2s for the first 5 respawns, then back off to 5s */
-	timeout = ((*restart_cnt) <= (RESPAWN_MAX / 2)) ? 2000 : 5000;
+	timeout = ((*restart_cnt) <= (SVC_RESPAWN_MAX / 2)) ? 2000 : 5000;
 	service_timeout_after(svc, timeout, service_retry);
 }
 
@@ -1290,7 +1608,7 @@ static void svc_set_state(svc_t *svc, svc_state_t new)
 	*state = new;
 
 	/* if PID isn't collected within SVC_TERM_TIMEOUT msec, kill it! */
-	if ((*state == SVC_STOPPING_STATE)) {
+	if (*state == SVC_STOPPING_STATE) {
 		_d("%s is stopping, wait %d sec before sending SIGKILL ...",
 		   svc->cmd, svc->killdelay / 1000);
 		service_timeout_cancel(svc);
@@ -1322,8 +1640,13 @@ restart:
 
 	switch (svc->state) {
 	case SVC_HALTED_STATE:
-		if (enabled)
-			svc_set_state(svc, SVC_READY_STATE);
+		if (enabled) {
+			if (svc_has_pre(svc)) {
+				svc_set_state(svc, SVC_SETUP_STATE);
+				service_pre_script(svc);
+			} else
+				svc_set_state(svc, SVC_READY_STATE);
+		}
 		break;
 
 	case SVC_DONE_STATE:
@@ -1333,13 +1656,20 @@ restart:
 
 	case SVC_STOPPING_STATE:
 		if (!svc->pid) {
-			/* PID was collected normally, no need to kill it */
-			_d("%s: Stopped normally, no need to send SIGKILL :)", svc->cmd);
+			char cond[MAX_COND_LEN];
+
+			_d("%s: stopped, cleaning up timers and conditions ...", svc->cmd);
 			service_timeout_cancel(svc);
+			cond_clear(mkcond(svc, cond, sizeof(cond)));
 
 			switch (svc->type) {
 			case SVC_TYPE_SERVICE:
-				svc_set_state(svc, SVC_HALTED_STATE);
+			case SVC_TYPE_TTY:
+				if (svc_has_post(svc)) {
+					svc_set_state(svc, SVC_CLEANUP_STATE);
+					service_post_script(svc);
+				} else
+					svc_set_state(svc, SVC_HALTED_STATE);
 				break;
 
 			case SVC_TYPE_TASK:
@@ -1355,6 +1685,16 @@ restart:
 		}
 		break;
 
+	case SVC_CLEANUP_STATE:
+		if (!svc->pid)
+			svc_set_state(svc, SVC_HALTED_STATE);
+		break;
+
+	case SVC_SETUP_STATE:
+		if (!svc->pid)
+			svc_set_state(svc, SVC_READY_STATE);
+		break;
+
 	case SVC_READY_STATE:
 		if (!enabled) {
 			svc_set_state(svc, SVC_HALTED_STATE);
@@ -1365,6 +1705,10 @@ restart:
 
 			err = service_start(svc);
 			if (err) {
+				if (svc_is_missing(svc)) {
+					svc_set_state(svc, SVC_HALTED_STATE);
+					break;
+				}
 				(*restart_cnt)++;
 				break;
 			}
@@ -1382,7 +1726,7 @@ restart:
 		}
 
 		if (!svc->pid) {
-			if (svc_is_daemon(svc)) {
+			if (svc_is_daemon(svc) || svc_is_tty(svc)) {
 				svc_restarting(svc);
 				svc_set_state(svc, SVC_HALTED_STATE);
 
@@ -1420,7 +1764,9 @@ restart:
 
 		case COND_ON:
 			if (svc_is_changed(svc)) {
-				if (svc->sighup) {
+				if (svc_nohup(svc))
+					service_stop(svc);
+				else {
 					/*
 					 * wait until all processes have been
 					 * stopped before continuing...
@@ -1428,8 +1774,7 @@ restart:
 					if (sm_is_in_teardown(&sm))
 						break;
 					service_restart(svc);
-				} else
-					service_stop(svc);
+				}
 
 				svc_mark_clean(svc);
 			}
@@ -1500,7 +1845,7 @@ void service_step_all(int types)
 
 void service_worker(void *unused)
 {
-	service_step_all(SVC_TYPE_SERVICE | SVC_TYPE_RUNTASK);
+	service_step_all(SVC_TYPE_RESPAWN | SVC_TYPE_RUNTASK);
 }
 
 /**

src/service.h

@@ -33,6 +33,7 @@ void      service_unregister     (svc_t *svc);
 
 void      service_runtask_clean  (void);
 void      service_reload_dynamic (void);
+void      service_update_rdeps   (void);
 
 int       service_step           (svc_t *svc);
 void      service_step_all       (int types);

src/sig.c

@@ -31,21 +31,24 @@
  *      does on USR1 with their FIFO/D-Bus.
  *
  * SIGUSR1
- *      Calls shutdown hooks, including HOOK_SHUTDOWN, stops all running
- *      processes, and unmounts all file systems.  Then tells kernel to
- *      halt.  Most people these days want SIGUSR2 though.
+ *      Restarts API (initctl) socket, like SysV init and systemd does
+ *      on USR1 with their FIFO/D-Bus.
  *
- *      (SysV init and systemd use this to re-open their FIFO/D-Bus.)
+ *      Finit <= 4.0 used to perform a system halt on USR1.  This caused
+ *      some quite nasty problems on systems with both systemd/sysvinit
+ *      installed alongside finit.  For compatibility reasons Finit 4.1
+ *      changed to partial SIGHUP (api_exit/init).
  *
  * SIGUSR2
- *      Like SIGUSR1, but tell kernel to power-off the system, if ACPI
- *      or similar exists to actually do this.  If the kernel fails
- *      power-off, Finit falls back to halt.
+ *      Calls shutdown hooks, including HOOK_SHUTDOWN, stops all running
+ *      processes, and unmounts all file systems.  Then tells kernel to
+ *      power-off the system, if ACPI or similar exists to actually do
+ *      this.  If the kernel fails power-off, Finit falls back to halt.
  *
  *      (SysV init N/A, systemd dumps its internal state to log.)
  *
  * SIGTERM
- *      Like SIGUSR1, but tell kernel to reboot the system when done.
+ *      Like SIGUSR2, but tell kernel to reboot the system when done.
  *
  *      (SysV init N/A, systemd rexecutes itself.)
  *
@@ -138,7 +141,9 @@ static struct sigmap {
 	/* Nonstandard signals found in all modern POSIX systems
 	   (including both BSD and Linux).  */
 	{ SIGWINCH,  "SIGWINCH"  },
+#ifdef SIGSTKFLT
 	{ SIGSTKFLT, "SIGSTKFLT" },
+#endif
 	{ SIGPWR,    "SIGPWR"    },
 
 	/* Archaic names for compatibility.  */
@@ -192,10 +197,10 @@ void do_kill(int signo)
 			if (fgets(file, sizeof(file), fp)) {
 				if (strstr(file, "gdbserver"))
 					_d("Skipping %s ...", file);
-				else if (file[0] != '@')
-					kill(pid, signo);
-				else
+				else if (file[0] == '@')
 					_d("Skipping %s ...", &file[1]);
+				else
+					kill(pid, signo);
 			}
 			fclose(fp);
 		}
@@ -214,11 +219,11 @@ void do_shutdown(shutop_t op)
 	utmp_set_halt();
 
 	/*
-	 * Tell all remaining non-monitored processes to exit, give them
-	 * some time to exit gracefully, 2 sec is customary.
+	 * Tell remaining non-monitored processes to exit, give them
+	 * time to exit gracefully, 2 sec was customary, we go for 1.
 	 */
 	do_kill(SIGTERM);
-	do_sleep(2);
+	do_sleep(1);
 	do_kill(SIGKILL);
 
 	/* Exit plugins and API gracefully */
@@ -265,7 +270,7 @@ void do_shutdown(shutop_t op)
 
 	/* Reboot via watchdog or kernel, or shutdown? */
 	if (op == SHUT_REBOOT) {
-		if (wdog) {
+		if (wdog && wdog->pid > 1) {
 			int timeout = 10;
 
 			/* Wait here until the WDT reboots, or timeout with fallback */
@@ -274,15 +279,15 @@ void do_shutdown(shutop_t op)
 				do_sleep(1);
 		}
 
-		_e("Rebooting ...");
+		_d("Rebooting ...");
 		reboot(RB_AUTOBOOT);
 	} else if (op == SHUT_OFF) {
-		_e("Powering down ...");
+		_d("Powering down ...");
 		reboot(RB_POWER_OFF);
 	}
 
 	/* Also fallback if any of the other two fails */
-	_e("Halting ...");
+	_d("Halting ...");
 	reboot(RB_HALT_SYSTEM);
 }
 
@@ -321,7 +326,7 @@ static void sigint_cb(uev_t *w, void *arg, int events)
 }
 
 /*
- * SIGUSR1: BusyBox style halt
+ * SIGUSR1: SysV init/systemd API socket restart
  */
 static void sigusr1_cb(uev_t *w, void *arg, int events)
 {
@@ -331,8 +336,9 @@ static void sigusr1_cb(uev_t *w, void *arg, int events)
 		return;
 	}
 
-	halt = SHUT_HALT;
-	service_runlevel(0);
+	/* Restart initctl API domain socket, similar to systemd/SysV init */
+	api_exit();
+	api_init(w->ctx);
 }
 
 /*
@@ -515,10 +521,14 @@ void sig_setup(uev_ctx_t *ctx)
 	erase(SYNC_SHUTDOWN);
 	erase(SYNC_STOPPED);
 
-	/* Standard SysV init calls ctrl-alt-delete handler */
+	/*
+	 * Standard SysV init calls ctrl-alt-delete handler
+	 * We need to disable kernel default so it sends us SIGINT
+	 */
+	reboot(RB_DISABLE_CAD);
 	uev_signal_init(ctx, &sigint_watcher, sigint_cb, NULL, SIGINT);
 
-	/* BusyBox init style signals for halt, power-off and reboot. */
+	/* BusyBox/SysV init style signals for halt, power-off and reboot. */
 	uev_signal_init(ctx, &sigusr1_watcher, sigusr1_cb, NULL, SIGUSR1);
 	uev_signal_init(ctx, &sigusr2_watcher, sigusr2_cb, NULL, SIGUSR2);
 	uev_signal_init(ctx, &sigpwr_watcher, sigusr2_cb, NULL, SIGPWR);

src/sm.c

@@ -156,9 +156,11 @@ restart:
 		/* Disable login in single-user mode as well as shutdown/reboot */
 		nologin();
 
-		/* Make sure to (re)load all *.conf in /etc/finit.d/ */
-		if (conf_any_change())
-			conf_reload();
+		if (runlevel != 0 && runlevel != 6) {
+			/* Make sure to (re)load all *.conf in /etc/finit.d/ */
+			if (conf_any_change())
+				conf_reload();
+		}
 
 		/* Reset once flag of runtasks */
 		service_runtask_clean();
@@ -204,10 +206,6 @@ restart:
 			break;
 		}
 
-		/* No TTYs run at bootstrap, they have a delayed start. */
-		if (prevlevel > 0)
-			tty_runlevel();
-
 		sm->state = SM_RUNNING_STATE;
 		break;
 
@@ -222,8 +220,7 @@ restart:
 		_d("Stopping services not allowed after reconf ...");
 		sm->in_teardown = 1;
 		cond_reload();
-		service_step_all(SVC_TYPE_SERVICE);
-		tty_reload(NULL);
+		service_step_all(SVC_TYPE_ANY);
 
 		sm->state = SM_RELOAD_WAIT_STATE;
 		break;
@@ -240,16 +237,17 @@ restart:
 		}
 
 		sm->in_teardown = 0;
-		/* Cleanup stale services */
-		svc_clean_dynamic(service_unregister);
 
 		_d("Starting services after reconf ...");
-		service_step_all(SVC_TYPE_SERVICE);
+		service_step_all(SVC_TYPE_ANY);
+
+		/* Cleanup stale services */
+		svc_clean_dynamic(service_unregister);
 
 		_d("Calling reconf hooks ...");
 		plugin_run_hooks(HOOK_SVC_RECONF);
 
-		service_step_all(SVC_TYPE_SERVICE);
+		service_step_all(SVC_TYPE_ANY);
 		_d("Reconfiguration done");
 
 		sm->state = SM_RUNNING_STATE;

src/stty.c

@@ -27,6 +27,8 @@
 #include <sys/ttydefaults.h>  /* Not included by default in musl libc */
 #include <termios.h>
 
+#include <helpers.h>
+
 /* B0 means; keep kernel default */
 speed_t stty_parse_speed(char *baud)
 {
@@ -61,10 +63,12 @@ speed_t stty_parse_speed(char *baud)
 		{ 1152000, B1152000 },
 		{ 1500000, B1500000 },
 		{ 2000000, B2000000 },
+#ifdef B4000000
 		{ 2500000, B2500000 },
 		{ 3000000, B3000000 },
 		{ 3500000, B3500000 },
 		{ 4000000, B4000000 },
+#endif
 	};
 
 	if (!baud || !baud[0])
@@ -121,10 +125,10 @@ void stty(int fd, speed_t speed)
 	tcsetattr(fd, TCSAFLUSH, &term);
 
 	/* Show cursor again, if it was hidden previously */
-	write(fd, "\033[?25h", 6);
+	dprint(fd, "\033[?25h", 6);
 
 	/* Enable line wrap, if disabled previously */
-	write(fd, "\033[7h", 4);
+	dprint(fd, "\033[7h", 4);
 }
 
 /**

src/sulogin.c

@@ -0,0 +1,213 @@
+/* poor man's sulogin, may be statically linked
+ *
+ * Copyright (c) 2021  Joachim Wiberg <troglobit@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include <crypt.h>
+#include <err.h>
+#include <paths.h>
+#include <pwd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <termios.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+
+/* getpwnam() cannot be used when statically linked */
+static int get_passwd(struct passwd *pw)
+{
+	char *ptr, *tmp;
+	char buf[256];
+	FILE *fp;
+
+	fp = fopen("/etc/passwd", "r");
+	if (!fp)
+		return 1;
+
+	while ((tmp = fgets(buf, sizeof(buf), fp))) {
+		if (!strstr(buf, ":0:0:"))
+			continue;	/* not uid 0 */
+
+		ptr = strsep(&tmp, ":");
+		if (!ptr)
+			break;
+		pw->pw_name = strdup(ptr);
+
+		ptr = strsep(&tmp, ":");
+		if (!ptr)
+			break;
+
+		/* x : password in /etc/shadow, see passwd(5) */
+		if (ptr[0] != 'x')
+			pw->pw_passwd = strdup(ptr);
+
+		break;			/* done, check it */
+	}
+	fclose(fp);
+
+	if (!pw->pw_name)
+		return 1;		/* no uid 0 here */
+	if (pw->pw_passwd)
+		return 0;		/* passwd in passwd */
+
+	fp = fopen("/etc/shadow", "r");
+	if (!fp) {
+		if (pw->pw_name)
+			free(pw->pw_name);
+		return 1;
+	}
+
+	while ((tmp = fgets(buf, sizeof(buf), fp))) {
+		ptr = strsep(&tmp, ":");
+		if (!ptr)
+			break;
+
+		if (!strcmp(tmp, pw->pw_name))
+			continue;
+
+		ptr = strsep(&tmp, ":");
+		if (!ptr)
+			break;
+
+		pw->pw_passwd = strdup(ptr);
+		break;
+	}
+	fclose(fp);
+
+	if (!pw->pw_passwd) {
+		if (pw->pw_name)
+			free(pw->pw_name);
+		return 1;		/* nopass => nologin */
+	}
+
+	return 0;			/* passwd in shadow */
+}
+
+static void freepw(struct passwd *pw)
+{
+	if (pw->pw_passwd) {
+		free(pw->pw_passwd);
+		pw->pw_passwd = NULL;
+	}
+	free(pw->pw_name);
+}
+
+static int sh(void)
+{
+	/*
+	 * Become session leader and set controlling TTY
+	 * to enable Ctrl-C and job control in shell.
+	 */
+	setsid();
+	ioctl(STDIN_FILENO, TIOCSCTTY, 1);
+	setenv("PS1", "# ", 1);
+
+	return execl(_PATH_BSHELL, "-sh", NULL);
+}
+
+int main(void)
+{
+	struct termios old, raw;
+	struct passwd pw = { 0 };
+	int rc = 1, ch;
+
+	if (get_passwd(&pw))
+		err(1, "reading credentials for UID 0");
+
+	switch (pw.pw_passwd[0]) {
+	case '*': case '!':	/* login disabled, allow */
+		printf("The %s account is locked, allowing entry.\n", pw.pw_name);
+		/* fallthrough */
+	case 0:			/* no password at all? */
+		printf("Press enter for maintenance (Ctrl-D to continue) ");
+		fflush(stdout);
+		ch = getchar();
+		if (ch == 4 || ch == EOF) { /* Ctrl-D */
+			rc = 0;
+			goto done;
+		}
+		goto nopass;
+
+	case '$': /* md5/shaN or similar, original DES not supported */
+		break;
+	default:
+		break;
+	}
+
+	if (tcgetattr(1, &old) < 0)
+		goto done;		/* failed reading tty */
+
+	raw = old;
+	raw.c_lflag &= ~ICANON;
+	raw.c_lflag &= ~ECHO;
+	raw.c_cc[VMIN] = 1;
+	raw.c_cc[VTIME] = 0;
+	if (tcsetattr(1, TCSANOW, &raw) < 0) {
+                tcsetattr(1, TCSANOW, &old);
+		goto done;		/* failed setting raw/pw mode */
+	}
+
+	while (1) {
+		char pwd[256];		/* reasonable max? */
+		size_t i = 0;
+		char *passwd;
+
+		printf("Give %s password for maintenance (Ctrl-D to continue): ", pw.pw_name);
+		fflush(stdout);
+		do {
+			ch = getchar();
+			if (ch == '\n' || ch == '\r')
+				break;
+			if (ch == 4 || ch == EOF) { /* Ctrl-D */
+				rc = 0;
+				goto exit;
+			}
+
+			pwd[i++] = ch;
+		} while (i + 1 < sizeof(pwd));
+		pwd[i] = 0;
+		puts("");
+
+		passwd = crypt(pwd, pw.pw_passwd);
+		if (passwd && !strcmp(passwd, pw.pw_passwd)) {
+			tcsetattr(1, TCSANOW, &old);	/* restore tty */
+		nopass:
+			freepw(&pw);
+
+			return sh();
+		}
+	}
+
+exit:
+	tcsetattr(1, TCSANOW, &old);	/* restore tty */
+	puts("");
+done:
+	freepw(&pw);
+	return rc;
+}
+
+/**
+ * Local Variables:
+ *  indent-tabs-mode: t
+ *  c-file-style: "linux"
+ * End:
+ */