This is the first patch in a series to add support for
log:/path/to/logfile,priority:facility.level,tag:ident
In this patch support for 'priority:' and 'tag:', log file support
coming later.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The SVC_RECONF, SVC_LOST, SVC_START, and RUNLEVEL_CHANGE hooks are not
one-shot, they are also not regular conditions since there exist no
mechanism to reset them from flux.
One idea was to turn them into actions, but the lost + start hooks need
to be called multiple times per trigger, e.g. `initctl reload`, which
turned out to be non-trivial to implement right now.
Therefore, for (at least) the Finit v3.1 release these conditions have
been disabled ("nop") and ignored by Finit. Only actual C-style plugins
will be called.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch cleans up the implmentation, adds documentation and also
makes sure to adjust the prefix path of a pid:/prefix/path argument.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The pidfile plugin just listens to changes in files in /run or /var/run,
so when trying to locate an svc_t from a file change we need to compare
using the proper prefix path.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
A service may now have a custom PID file, possibly because it doesn't
really create one itself. This needs to be taken into account also in
the pidfile plugin.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for creating, and removing, a PID file for
programs (services) that for some reason does not create one. With
Finit this is very useful since it allows us to use our condition
support even for "broken" daemons.
To enable this for a service, use the following syntax:
service pid[:[/path/to/]file[.pid]] /path/to/daemon
I.e., `pid` is a required argument if you want to enable the feature,
but the rest is optional:
pid --> /run/`basename $CMD`.pid
pid:file --> /run/file.pid
pid:file.dip --> /run/file.dip
pid:/mnt/k --> /mnt/k.pid
pid:/mnt/k.foo --> /mnt/k.foo
There is an additional feature with this patch. For daemons that run
with a different identiy one usually wants to create the PID file with
that identity, e.g.:
service inadyn -I dyn -- Dyn.com DDNS client
service inadyn -I freedns -- FreeDNS DDNS client
Here inadyn will look for /etc/dyn.conf and create /run/dyn.pid, and
/etc/freedns.conf and create /run/freedns.pid, respetively. While Finit
will in both cases assume the PID file is /run/inadyn.pid ... err fail!
To get out of this mess the `pid` argument can be used to tell Finit
what the real PID file the daemon creates is called:
service pid:!dyn inadyn -I dyn -- Dyn.com DDNS client
service pid:!freedns inadyn -I freedns -- FreeDNS DDNS client
The same composition rules listed above applies also to the `!` variant.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The service_step() function may call svc_del() on an svc from the
service_monitor(), so we need a way to tell the latter that it no
longer can dereference the pointer.
This regression is limited to v3.1, due to the introduction of the
new non-shared-memory store of svc_t.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Per service rlimit is new in v3.1. When an inetd service connection is
created we do a selective copy from the parent inetd. We also need to
copy the parent's rlimits, otherwise the child will be killed as soon
as the kernel starts it.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch increases the default wait, 2 --> 5 sec, while pending for
Finit + watchdogd controlled reboot/shutdown. We don't want to hand
back the prompt when we're not really ready yet.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch builds on top of the recently introduced "silent by default"
and support for kernel cmdline `splash` to disable silent operation.
At shutdown/reboot `log_exit()` is called, the intention is to re-enable
progress output so the user can see what's going on. So, we need to re-
set `silent` to the same setting as log_init() started with, hence the
new `muffler()`.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Automatic redirection of output was introduced in v3.0 and for the
upcoming v3.1 a configure option was added to disable it.
By popular demand this patch flips the default. One now has to
explicitly enable this feature to get redirection.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for restarting the API socket on SIGHUP, similar
to how SysV init and systemd use SIGUSR1 to restart their FIFO or D-Bus.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
An interface may come up with IFF_RUNNING already set, but we would
only detect IFF_UP. This patch simplfies the code a lot by pivoting
the qualifier from using ifi_change to ifi_flags, which is the actual
current status of an interface.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This is a quite intrusive change. In addition to the already reserved
runlevels 0 and 6, halt and reboot, respectively, this patch reserves
runlevel 1 as *no networking*, like the traditional SysV init did. One
of the reasons for this change is the new 'single' user mode, introduced
earlier. Most users when entering 'S' or 'single' on the kernel cmdline
expect the single user mode (runlevel 1) to be without networking.
This leaves 2, 3, 4, 5, and 7, 8, 9 as fully user-configurable runlevels
with networking.
Of course, networking can still be enabled in runlevel 1 using ifconfig
or ifupdown, or change runlevel.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If hostname is missing from /lib/finit/rescue.conf the system will
default to the configured default hostname (noname).
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a rescue (revovery) mode to Finit. Simply add `rescue`
to the kernel cmdline at boot and the following steps at bootstrap will
be skipped:
- Load of services/run/task/etc in /etc/finit.conf + /etc/finit.d/*.conf
- fsck of filesystems in /etc/fstab
- Remount of / to read-write
- Mount of all filesystems in /etc/fstab
- swapon
- A few plugin hooks:
- HOOK_ROOTFS_UP
- HOOK_MOUNT_ERROR
- HOOK_BASEFS_UP
- `runparts DIR`
- /etc/rc.local
Instead of loading /etc/finit.conf and /etc/finit.d/*.conf Finit loads
the file /lib/finit/rescue.conf, which can be overridden by the operator
if needed. If this file is removed (or lost), Finit falls back to start
a simple root shell, without any login.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The /lib/finit/plugins path is as of this patch not longer possible to
change. Use --prefix or --libdir instead to change the leading path.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Finit v3.1 adds support for reloading /etc/finit.conf, document what
actually is read at runtime, compared to at bootstrap, and clarify
which stanzas are for what part of the boot process.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The difference between `runparts DIR` and `/etc/rc.local` is that one
should run just before the first runlevel change (to the configured
runlevel) at the end of bootstrap, whereas the other should run just
before launcing the TTYs, i.e. after all tasks in the configured
runlevel have been started.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch is an optimization. Now that we monitor for any .conf file
changes (as soon as the event loop starts) we can skip automatic reload
of all .conf files and services if no .conf file has changed. Unless
the user issues an `initctl reload`, `init q`, or sends us SIGHUP.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
As reported in #92, systems with an udevd very early register a service
using global_rlimit[]. The change was introduced in b68c5c1 but did
not take the ordering problem into consideration, global_rlimit[] was
left uninitialized, which likely led to severely limited udevd that
was continously crashing.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Validate function argument
- Skip¹ shutdown() when closing a TCP socket, not needed
___
¹ also, shutdown() segfaults a late shutdown/reboot
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
On shutdown/reboot, send SIGPR as a heads-up to watchdogd that the
system is going down. All file systems are about to be unmounted.
It's just about the last process left running in the system, so we
give it two seconds to complete its work before we continue.
When everything has been unmounted and we're ready to call reboot(),
we send SIGTERM to the registered watchdog, to give ti the chance to
let the WDT reset the system.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When Finit enters shutdown/reboot we need to advise any watchdogd that
we are going down *before* we unmount any file systems. This to let
the user save any state, e.g. reboot counters, to disk. Right before
we call reboot() in Finit we send SIGTERM to the watchdogd to give it
the chance to use the WDT to reboot the system.
The WDT cricuitry on embedded systems is often connected to the RESETn
logic of all relevant board compoents, whereas reboot() only reboots the
SoC with the CPU. Hence, a WDT reboot is more efficient and more like a
regular power cycle, which users of embedded systems often want.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch makes it possible to control the default redirection of
(misbehaving) services output to /dev/null. With this disabled,
a service may write to its stdout/stderr and mess up the console.
By default all service/run/tasks stdout/stderr is redirected.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a `nologin` flag to the `tty` configuration directive,
making it possible to set up really simple embedded systems with no
login.
Example, put the following line in `/etc/finit.d/getty.conf` to get
a "Please press Enter to ..." prompt before being presented with a
root shell prompt.
tty [12345] @console noclear nologin
Needless to say, this is not very secure, but can be really useful
for developer setups, during board bringup, or similar.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In f3669c7 the difference between static and dynamic services was
removed. This led to a regression in handling internally created
services for, e.g., dbus-daemon and udevd, mentioned in issue #90.
This patch introduces a "protected" flag for svc_t to prevent the
mark-and-sweep handling of regularly loaded services.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The new condition triggered hooks requrie /var/run/finit/cond to
be set up before calling the first user-configurable hooks.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Add newline to cond generation ID, like PID files, easier when
debugging. Does not affect fscanf() in cond_get_gen()
- Update copyright years
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Until now Finit has used mtime to determine if a file has been changed,
or simply touched to request reload, when `initctl reload` is called.
Using mtime for this purpose is a monumentally bad idea since time can
be changed at any point: a user may adjust the time, NTP continously
tunes the clock, and time stamps are stored as the walltime. So if we
compare timestamps against the last time we (booted or) did reload, we
would miss .conf file changes.
This patch replaces the mtime mistake with an inotify watcher for the
following files: /etc/finit.d/*.conf, /etc/finit.d/available/*.conf,
and /etc/finit.conf. All file changes between (bootstraps and) calls
to `initctl reload` are tracked, like the mtime backend it replaces.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
TIL, using mtimes for tracking event orderings is a monumentally bad
idea (queue the nodding UNIX-beards). Mtimes are in wallclock time which
is not necessarily monotonically increasing. A user may adjust the time,
an NTP daemon will continously tune the clock and so on.
Instead, store an explicit generation number in each condition file,
which will be monotonically increased by finit on each reconf.
The following configuration directives, previously only allowed
in /etc/finit.conf, are now also only allowed in runlevel S, i.e.
at bootstrap:
- host
- mknod
- network
- runparts
- runlevel
This change is in preparation for allowing reload of finit.conf
as well as /etc/finit.d/*.conf, which we already support.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This is a prelude to a major behavioral change. The difference between
a static and dynamic service will be removed in Finit v3.1. This means
all .conf files, including /etc/finit.conf, will be re-read on reload.
However, not all configuration directives will be re-read, some will
only be applicable at bootstrap, i.e. runlevel S.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch moves the networking bit to the latter part of bootstrap,
called finalize(), in favor of starting the event loop earlier.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This change removes a long standing kernel requirement and architectural
mishap in the design of Finit v1. It is no longer necessary to include
CONFIG_SYSVIPC in the kernel to use Finit and its command line tool
initctl. It is also no longer possible to connect to the internals of
PID 1 using shmat(), the initctl tool has already been updated to use
the domain socket API.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This change is one of the required intermediate steps to remove the
shared memory store for all svc_t, this in turn to avoid any external
programs manipulating the internal memory of PID 1.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
For ease of use, allow short forms of commands, like:
initctl h
initctl c d
initctl l
Also, default to show conditions if no argument to cond is given.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We need the name client.c for the client side of api.c, so let's rename
the client to its traditional name.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Previously UDP packets were never dropped and if a request came
from a not valid interface inetd got stuck in an inifinite
loop looking at the same packet every time and not dropping it.
Signed-off-by: Lennart Eriksson <lennart.eriksson@westermo.se>
This patch fixes the problem with run/task jobs with conditions, run at
runlevel S (bootstrap), not being given enough time to actually start
and run even once.
The new svc counter 'once' is now inspected before transitioning away
from runlevel S to the configured runlevel. A global (currently hard
coded) timeout of 10 sec is given to all run/tasks.
As a spin-off, this patch starts up the event loop much earlier than
before. Meaning the initial runlevel change at boot is now done in
a finalize() callback when all run/tasks have completed, or when the
global timeout has expired. Starting the event loop earlier gives us
the possibility to rely on the pidfile plugin to provide our run/tasks
with the conditions necessary to start.
TODO: What remains is to add similar handling to all runlevel
transitions, not just the initial one at boot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch implements support for calling run/task jobs on the built-in
`HOOK_*` points in Finit. To map each HOOK point to a condition a new
classification was made, as shown in plugin.h and in the example below.
task <hook/mount/root> /sbin/attachubi.sh # run in parallel with other tasks/services
run <hook/net/up> /sbin/dosomething.sh # run in sequence with other commands
Please note, only rudimentary tested.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Make svc state transition to RUNNING *before* service_start(), because
in that function we may call service_step() as well, which would start
the svc again (since it was still in READY).
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Only run HOOK_SVC_START on successful start, and *after* having printed
the result of the started service, otherwise any print-outs from plugins
or tasks we start may overwrite the started service's progress.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Don't really see the point of this code anymore, was possibly an issue
before we refactored the logging subsystem. Comment out for now, and
remove later.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If you have lots of service/run/tasks using the same cmd it's impossible
to tell them apart in debug mode. This patch logs description as well
as cmd to improve on this situation a bit.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a 'once' counter to the svc_t for run/task jobs. This
counter is incremented when Finit calls service_start() and is reset on
runlevel changes.
Also, at the end of a runlevel each run/task have their state reset to
SVC_HALTED_STATE so they can be launched again for the new runlevel, if
so declared.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Systems running OverlayFS, or similar, on /etc may have new/modified
.conf files in /etc/finit.d. This patch makes sure to reload them
after `mount -a` has run so any configured bootstrap tasks and the
new hook callback scripts may run as expected.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a rather nasty bug in the SIGTERM+SIGKILL logic
employed by Finit when stopping tasks, e.g. at a runlevel change.
When a process terminates, after Finit having sent it SIGTERM, and the
SIGKILL timer expires *before* the timer has been cancelled, then the
timer callback would cause kill(0, SIGKILL).
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch extends the existing rlimit implementation to support setting
limits per service, run/task, inetd, and tty.
Use rlimit in /etc/finit.conf to change the global setting, which is
then inherited to each /etc/finit.d/*.conf. For each .conf file the
rlimit is reinitialized to the global finit.conf settings.
Also, add `unlimited` keyword, to replace the now deprecated `infinity`
keyword. The latter is however kept, for compatibility with previous
releases, for the foreseeable future.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch not only greatly simplifies the registration of the built-in
watchdog daemon, it also fixes a nasty race condition with an external
watchdog daemon. The INIT_CMD_WDOG_HELLO could reach finit in reversed
order, leading to the built-in daemon shutting down, but finit killing
the just registered external daemon.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Most Linux distributions, embedded or otherwise, do not need this
and only confuses the user of an OS with Finit.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Most Linux distributions, embedded or otherwise, do not need this
and only confuses the user of an OS with Finit.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Some systems use /var/db while others use /var/lib/misc, we need to
support both and cannot rely on _PATH_VARDB, since it too might not
be set properly.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When alibuEv event watcher callback has `UEV_ERROR` in the events mask
we must handle the error. Unfortunately, in most cases there is not
much we can do but ignore it or log the error.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If the speed argument to the `tty` stanza is invalid we should not
default to some other random speed. Instead, leave the line speed
as-is, like we do when starting an external getty.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch relocates the speed parser so that exec.c can set the TTY
speed properly as soon as possible. For external getty we make sure
to not touch the TTY speed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch makes the Finit built-in getty behave a bit more like other
getty implementations. We now not only disable ISIG but also VEOF and
VINTR when prearing the TTY for the getty (built-in or external).
Like before, it is up to the getty process to leave the TTY in a
usable state to /bin/login, so we also move the stty() call to
just before calling /bin/login in getty.c
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch moves the stty() function from getty.c to a separate file so
it can be called from exec.c as well. The fix to #84 is simply to call
stty() in prepare_tty(), which is shared by run_getty() and run_getty2()
and called before "Please press Enter ...".
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Time comparision should be >= not just >. May seem a bit silly to
nitpick like this since we're comparing nano seconds, but on systems
with no high-res timers this happens more often than you would like.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When performing an `initctl reload` with one (unchanged) service
depending on, e.g. `net/iface/lo`, its condition will not be set
to ON by the pidfile plugin unless the netlink plugin hook runs
first.
Example:
service <net/iface/lo> /sbin/dropbear ...
Which provides the <svc/sbin/dropbear> condition, will not be
set by pidfile.so during `initctl reload` because dropbear is
still SIGSTP:ed waiting for <net/iface/lo>.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
All this code did was trigger prctl() to set process name to "console",
which you could use as a simple means of figuring out if a process was
started from the system console.
There are other ways to do this, which Finit should not be involved in.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Before we call screen_init(), which in turn calls libite:initscr(),
which in turn does a lot of tcgetattr() and termios to STDOUT to figure
out the size, we need to check that we're not backgrounded, or don't
actually have a TTY. Problem reported by @wkz
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for a new TTY device keyword `@console` which
auto-detects the system console by reading the SysFS file:
/sys/class/tty/console/active
Embedded systems with one of /dev/ttyAMA0, /tty/ttyMXC0 or the original
/dev/ttyS0, as their active serial console can now instead use a single
line to /etc/finit.d/getty.conf to start the built-in getty on the same
TTY as the kernel already runs on:
tty [12345789] @console 115200 noclear
With no TTY defined Finit will either do nothing (default), or start a
fallback shell if enabled in the configure script.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Disable INTR, QUIT, SUSP, and DSUSP while handing over to getty. It is
up to the getty process to allow ISIG again.
This patch is due to the differences in control line setup between ARM,
x86/x86_64, and most notably PPC. PPC in particular seem to have ISIG
set by default from the kernel.
Without this patch pressing Ctrl-C before getty is started will lead to
Finit receiving SIGINT and defaulting to a reboot, as introduced in the
BusyBox init signal adoption in eeb59eb.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In GCC 7 the -Wall infers the new -Wformat-truncation which finds
problems with string truncation in functions like snprintf().
This patch suppresses the warning by allowing only a subset of the
parent inetd service description to be copied to the child connection.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In GCC 7 the -Wall infers the new -Wformat-truncation which finds
problems with string truncation in functions like snprintf().
This patch fixes a possibly problem in path name creation which,
combined with the scandir() function, might cause loss of .conf
files, failure to call run-parts properly, and kill services at
shutdown.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In GCC 7 the -Wall infers the new -Wformat-truncation which finds
problems with string truncation in functions like snprintf().
This patch fixes a problem in condition parsing for very long
PID filenames which might be truncated in internal buffers
causing a mismatch in Finit condition tracking.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In GCC 7 the -Wextra infers the new -Wimplicit-fallthrough which can be
suppressed by adding a simple comment
/* fallthrough */
Which we had, but was too verbose (and misspelled) for GCC to notice.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In commit 34a1524 the start of udevd was refactored to allow it to be a
properly monitored service caused systems using eudev to malfunction.
This patch reverts part of that commit and adds a workaround found when
trying to reproduce issue #82 on Void Linux:
- Start a temporary udevd in the background
- Call udevadm to trigger bringup of all subsystems and devices
- Wait for everything to settle
- Tell temporary udevd to exit (new)
In the world of `mdev` this song and dance act is not required.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Finit has long since supported an @user:group syntax to services, tasks,
and run stanzas. Setting UID was implemented at that time, but setting
GID never was. This patch fixes that oversight.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes an invalid use of strpbrk() causing all cond
sub-commands, except initctl cond show, to fail.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a '-' for each runlevel a service/task is not part of.
The resulting listings are easier to read and also makes it clearer to
the user what happens when changing runlevels.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch, similar to the work done in the d-bus plugin, converts the
udevd service from a non-monitored run_interactive() task that forks
itself into the background, to a proper Finit monitored service.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch prevents output of stderr/stdout from services to leak to the
console. Use the 'log' keyword to service/run/task stanza to redirect
output to syslog.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch changes how the D-Bus plugin starts dbus-daemon, from being
started in the background to letting Finit start and monitor it.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
For some reason PPC is much more sensitive than ARM when it comes to
termios changes. This patch fixes a regression in Ctrl-C handling
after login introduced in 907595d when blocking of Ctrl-C was added
to prevent breaking the login-prompt.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch removes the following very annoying warning for TTY devices
not yet available, i.e. plugged-in to the system.
Loading configuration ............................................................................. [ OK ]
[ 11.097414] finit[1]:tty_register():Incomplete or non-existing TTY device given, cannot register.
[ 11.101457] finit[1]:tty_register():Incomplete or non-existing TTY device given, cannot register.
Mounting filesystems .............................................................................. [ OK ]
Finit will not start a getty (built-in or external) if the TTY does not
exist, but having the configuration enabled makes it possible to do a
simple `initctl reload` to activate a plugged in TTY.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In case the kernel is incapable of handling a reboot,
power-off, or halt request we want to be able to let
the user perform some basic diagnose of the system.
The existing debug messages are good enough, and use
the correct grammar form, to be enabled always.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When collecting a bootstrap task we need to allow its state machine to
complete before cleaning them out.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
During the runlevel change from S --> 2 the state machine will pause,
waiting to collect all S-only tasks/services, before continuing.
When an S-only task/service is collected we therefore need to call the
state machine again so that it can reassert all conditions properly.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We should not let PID 1 fail on systems where Linux is built without
CONFIG_SYSVIPC, or the libc does not support shmget()/shmat().
This workaround implements a fallback to using calloc() for PID 1,
which means the initctl tool will not work properly. A better fix
would be to refactor the shm implementation to instead have initctl
call Finit using api.c for each svc_t. This has been added to the
TODO file as future work.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Building finit with a uclibc-ng-toolchain fails with:
```
../src/helpers.h:53:43: error: unknown type name ‘va_list’
void printv (const char *fmt, va_list ap);
```
- Detail limitations in separate section
- Focus on syntax early in document
- Overhaul and simplify introduction
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Enable LXDM using:
initctl enable lxdm
initctl reload
The latter command reloads all /etc/finit.d/*.conf files and starts lxdm
in runlevels 2-5.
Remember to first edit /etc/lxdm/lxdm.conf, necessary to get a session.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch enables support for bootstrap services in /etc/finit.d/
Please note, however, that this does not mean that all configuration
directives are now allowed in /etc/finit.d/*.conf -- the intention is
only to simplify distribution setups.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Restore original file layout somewhat, it's better if we clear the space
at the end of the file for easy `cat BOOTSTRAP.conf >>/etc/finit.conf`
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Finit now supports pipes and redirects in run/task stanzas. Hence, we
can reduce rc.local to an example template.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
A run stanza blocks operation by calling waitpid(), so it will never be
caught by the service_monitor() function. Hence, we need to clean up
any bootsrap calls after waitpid(), or at least perform a manual state
change.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch extends the capabilities of task and run .conf stanzas by
enabling pipes and redirects.
Example:
task [s] echo "foo" | cat >/tmp/bar
run [s] echo "$HOME" >/tmp/secret
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Show progress when booting
- Use noclear + nowait on TTY's
- Update `loadkmap` args
Works on Alpine Linux 3.4 .. 3.6
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch aligns the two tty configuration stanzas, thus clear and wait
are now the default for both.
tty [LVLS] <DEV> [BAUD] [noclear] [nowait] [TERM]
tty [LVLS] <CMD> <ARGS> [noclear] [nowait]
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch makes it possible to disable the "Please press Enter" prompt
when using an external getty. Simply append `nowait` to the .conf entry
and Finit will spawn the getty immediately.
A future patch will add support for `noclear` as well to external getty,
as well as a change in default (=> `wait`) for the built-in getty.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
It is strongly recommended to instead use the service stanza to start
XDM/GDM, or use startx after a regular login.
Note: the `user NAME` setting was always reserved for use with startx
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The emergency shell is just a Finit debug mode. This patch adresses a
concern from GitHub issue #76, with zombie processes in this mode.
Emergency shell could be further improved, e.g. informing the real Finit
process of each collected child (zombie) to be restarted if needed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In commit eec32e2 support for reloading TTY configurations was added.
This patch is a follow-up to that with support for cleaning up and
removing allocated TTYs.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
With the new scheme using /etc/finit.d/available/, and symlinks from
there to the enabled directory, we need to check mtime of .conf files
using stat(), not lstat() which we need to check if .conf is a symlink.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch disables console heading and bootup progress messages by
default. Use --enable-progress to get the old behavior back, or try
the new HOOK_BANNER and HOOK_SVC_UP to run a custom progress.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When the TTY fallback mechanism is enabled tty_fallback() prints this
annoying debug message whenever *any* is process is collected by PID 1:
"Not starting fallback (%d) shell, lost pid %d", fallback, lost"
This patch removes it, and a two other useless debug messages.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch moves the plugin loader much earlier, making it possible to
provide a custom banner/logo in a plugin.
To facilitate this move we need to also move banner later and parsing
of the kernel cmdline earlier, right after mouting /proc.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch changes the traditional look of the Finit boot sequence.
Now we stick to the screen width probed by screen_init(), or max
132 chars.
Also, some minor cleanup of the currently unused print_timestamp()
method used by printv() when showing [ OK ] or [FAIL] of services
at boot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch introduces the screen_width() method from the SMCRoute
project. The code is hereby also available under MIT/X11.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Since at least a44daf2 both run() and run_interactive() can access all
system tools in _PATH_STDPATH. A system that uses non-standard paths,
e.g. /system/bin and /apps/bin, should update _PATH_STDPATH.
For reference, see issue #69 and PRs #70 and #72.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch implemements a new feature of libite v1.9.1, whichp(), which
instead of fexist() that requires an absolute path, searches $PATH.
For reference, see issue #69 and PRs #70 and #72.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Ways to log:
- Set debug, using kernel cmdline, or edit the sources
- Change loglevel severity, default blocks > LOG_NOTICE
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The generic term for a non-running service is blocked, but a service may
be blocked for several different reasons. When a user stops a service
with `initctl stop` the intuitive expected output from `initctl show` is
"stopped" not "blocked".
When the user later restarts the service with `initctl start` it may of
course be blocked due to wrong runlevel or a missing condition. As can
be expected from the service's configuration.
The rest of the patch refactors the internal API names a bit to reflect
this change.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch is a follow-up to 1f1c2e0 which introduced the new semantics
for the very confusing quiet and silent modes.
The initial idea was to loosely follow the kernel cmdline 'quiet' option
and as a working concept this was good. The resulting code, however,
became unnecessarily complex and the configure options impossible to
explain.
With this patch the quiet mode, and thus the --disable-quiet configure
option, is removed. Finit now defaults to a sane behavior: services are
shown only at startup. The --enable-silent option works like it did
before, silencing everything.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch is the first in a series of changes to clean up the internal
logging APIs of Finit:
1. Convert log macros to use logit() for everything (this patch)
2. Set correct syslog() log level based on debug flag
3. Do closelog()/openlog() when toggling debug flag to control LOG_PERROR
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for enabling and disabling services in
/etc/finit.d/ by adding another subdirectory:
/etc/finit.d/available
Enabled service/run/task stanzas are by convention symlinked
to the parent dir. The initctl tool has three new convenience
commands to facilitate operating with this:
initctl list
initctl enable <CONF>
initctl disable <CONF>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
getprotobyname() returns p_proto in host endian, so it should not be
converted with ntoh. (Note: getservbyname() on the other hand does
return s_port in network endian)
Signed-off-by: Petrus Hellgren <petrus.hellgren@westermo.se>
Check first ifup exists before using it since
this is really something only exits on some
distros an in different ways. Debian/Ubunu has
an package for this while others have it in the
initscripts package(s) , meaning the file can be
around even without the prog.
Signed-off-by: crazy <crazy@frugalware.org>
This should allow us to run 'prog' without
absolute paths. Eg:
run("/bin/foo"); -> run("foo");
However we still need absolute PATH for
checking since fexist() need full path right now.
Signed-off-by: crazy <crazy@frugalware.org>
That should never be needed , however I saw
already systems with broken paths.h so better
have something around in case something breaks.
Signed-off-by: crazy <crazy@frugalware.org>
Also, update the service help text in doc/config.md to include info on
both runlevels and explain the condition in the listed example.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch to the logrotate() function extends the log rotattion of a
file with a gzip of .2 and older log files. If gzip is not available
in $PATH then @num files are left uncompressed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a simple log-rotate function to the UTMP, WTMP, BTMP,
and lastlog files. Enabled by default, disable with:
configure --disable-logrotate
Even though it can rotate /run/utmp it is unlikely this file will grow
that much during a system's uptime.
The hard-coded defaults are to keep up to two (2) previous versions
of the files and rotate on file size >= 100k.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
For issue #64 we need to have /proc/mounts, or /etc/mtab but that's
usually symlinked to /proc/mounts.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Remove verbose of initctl cond, and replace it with 'initctl cond dump'
Show all conditions if they are in flux/on even if there are no other
services depending on this condition.
Signed-off-by: Mattias Walström <lazzer@gmail.com>
We have repeatedly seen use-cases when sub-second resolution for
updatating the mtime of files have been needed. Yes, all those
cases have been Autobots.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch improves the error handling of initctl so that it returns
non-zero if a service does not exist.
$ initctl <start|stop|restart> SVC
$ echo $?
1
In verbose mode an error message is also returned:
$ initctl -v <start|stop|restart> SVC
No such service SVC
$ echo $?
1
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch breaks out the JOB:ID parser from api.c, placing it in a new
function called svc_parse_jobstr().
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We want to emulate the behavior of the reboot() syscall on success.
Primarily for a coherency/symmetry, but also to prevent printing
a new shell prompt in the middle of system shutdown messages.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When finit completes quicker than one second it will
not find if any configurations has been changed.
Signed-off-by: Mattias Walström <mattias.walstrom@westermo.se>
If a service depend on the service last in the list the application
will be kept in flux state forever.
Signed-off-by: Mattias Walström <mattias.walstrom@westermo.se>
This patch changes the confusing logic in all #ifndef INETD_DISABLED
statements to instead be #ifdef INETD_ENABLED, which is quite a bit
more readable.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This (initial) patch for issue #56 adds a very basic validation of the
device before attempting to start a getty on it.
TODO: Verify a USB/serial device can be plugged in at any point at
runtime, either automatically start getty or by `initctl reload`
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch (untested) patch calls `service_reload_dynamic()` after both
run-parts (if directory exists) and after `/etc/rc.local` has run. The
intention is to reload any services modified/added/removed by a user
script so the system is properly up to date.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch allows all services to go from READY -> RUNNING, this to
prevent started inetd connections from being treated as "stubborn"
and inadvertently killing them off after a few seconds.
The same actually applies to run/task applications.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes issue #51 by reverting b4b63d3, which is not needed
anymore since service callbacks have been removed from Finit3.
The `sig_unblock()` function, called for each forked off child, should
now be balanced with caught signals.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We can safely skip tmpfs, nothing to clean from previous boot there.
In fact, if we clean it we may spoil things for udev, which saves
pre-boot files in /run/udev.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch removes the long standing Watchdog daemon from the TODO list.
We now have a bare-bones watchdog service built-in to Finit, with
support for handing over to an external, and possibly more advanced,
using the initctl API.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a simple but effective watchdog daemon with support for
handover to an external daemon, should one register using the API.
Disabled by default, enable by:
configure --enable-watchdog
The watchdog forks to the background, opening /dev/watchdog setting a
three second timeout, and kicking every second. At shutdown/reboot the
daemon will receive SIGPWR from Finit to exit and cause a WDT reboot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
To prevent returning the prompt too soon, and thereby messing up the
output from Finit, wait a bit after successful delivery of signal to
reboot/halt/poweroff.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds initial support for registering a watchdog daemon with
Finit. Needed for situations when you want the system to actually be
rebooted by the watchdog timer (WDT) circuitry rather than the kernel.
The initctl API has been extended with a INIT_CMD_WDOG_HELLO command
that is used to both register and "ping" Finit periodically. As long as
Finit responds (with an ACK) the watchdog daemon knows Finit is alive.
At shutdown Finit stops all services (except kernel threads and daemons
marked with @ in the first character of their process name) and, if a
watchdog has registered, SIGPWR is sent to it when Finit has finished
unmounting etc., just prior to kernel reboot/halt/poweroff.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Now that we change runlevel at shutdown/reboot, we also need to restore
terse mode before calling the shutdown hooks, like before.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch replaces kill(-1, signo) with a more fine grained
do_kill(signo) that traverses each /proc/PID, skipping kernel
threads and special [0][0] == '@' processes.
The reason for this patch is not only future compliance with
freedesktop.org standards, but also a way of not killing any
critical processes at shutdown, e.g. mdadm or watchdogd.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The shutdown hook is supposed to run right before services start
receiving SIGTERM. Change due to previous refactor.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch moves both the definition and the call to bootclean() from
finit.c to the bootmisc plugin. We want to call bootclean() *after*
any /run and /var file systems have been mounted.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- New requirements section about what Finit expects from the system
- Remove commented out `udevadm settle` code, now in README
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
On systems with systemd-udev/udevadm the trigger addition of devices at
boot wrecks havoc of /dev/console. This patch forces Finit to go silent
by default on such systems. Far from ideal, but since the actual root
cause for this messup is unknown this will have to do for now.
Also, add a settle timeout of 120 sec to the udev device tree population.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch factors out the fgetint() fallback code for parsing
/etc/services and /etc/protocols. Useful on systems with a
malfunctioning C library.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Without trigger, kernel modules won't be loaded automatically. Also
devices are not managed. For example, i915 was not loaded after boot
and my network card was not renamed from wlan0 to wlp3s0, which means
it was not handled properly by udev.
Signed-off-by: Wang He <xw897002528@gmail.com>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a long-standing issue with how Finit handles shutdown,
reboot and poweroff -- and even runlevel changes to runlevel 0 and 6.
Prior to this patch sig.c:do_shutdown() was called immediately on any
of the above events, disregarding any services that a user might have
wanted to run as part of the shutdown/reboot.
With this change the handling is unified in all frontends to instead
do a runlevel change followed by the do_shutdown() action.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The /etc/nologin file must be set as soon as the runlevel change to
0/1/6 comes in, not when done with runlevel change and services have
been started, if any.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
As a last ditch effort to understand why an embedded system doesn't
boot, the operator should be able to see as much logs as possible
when enabling --debug mode on the kernel command line.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds an `--enable-fallback-shell` option to confiugre.
When enabled Finit will start `/bin/sh` on the console, but only
if there are no active TTYs running in the current runlevel.
With this it will be easier to do board bringup with Finit, but it also
enables using the fallback shell in a rescue mode (runlevel) where no
TTYs are allowed to run.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Also changes the default to set $HOME to / instead of nothing. This can
of course be improved immensely. Pull requests always welcome ;-)
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch removes the now heavily outdated (and non-working) support
for callback plugins. I.e. plugins that attach themselves to services
listed in /etc/finit.conf and can be queried to determine if a service
should be started or stopped.
The Finit3 support for adding/removing services at runtime with plain
text *.conf files in /etc/finit.d and `initctl reload`, this old, and
rather cumbersome service callback feature, is not longer supported.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Add support for registering callbacks to be called after a given
timeout.
Use this to implement forceful termination of services that refuse to
terminate upon receiving a SIGTERM. If the service PID is not collected
within 3 seconds, send a SIGKILL to it.
Also, rework restarting of crashing services. When a service crashes for
the first time, try to restart it immediately. If it crashes again, wait
a few seconds before trying again. Just as before, if it continues to
crash, block the service.
This patch prevents double critical/error messages from being printed on
the console when there is no syslog daemon running yet.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Delay setup of initctl plugin to after the base filesystems are up and
after the bootmisc plugin has run.
This patch factors out the I/O initialization of plugins so it does not
have to run when plugins are loaded. Instead, plugins can now call this
API to register an I/O callback when they are ready.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
In UTMP a terminating zero in string fields is optional, so we need to
take that into account when preparing entrys.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch makes Finit require libuEv v1.4.2, a minor bugfix update, and
libite (-lite) v1.8.2, which is quite a bit more than just a few bug
fixes. The most relevant fixes are the removal of implicit logging for
touch(), makedir(), makefifo(), and erase() inline functions. The
release also removed implicit logging to stderr on error, as well as a
a few build fixes for running on musl libc.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a very special debug mode to Finit. With this feature
enabled the actual duties of PID 1 is forked off early to a child. If
that child ever dies, usually due to a SIGSEGV, this wakes up the real
PID 1 which lauches a debug shell on the console.
Remember to disable 'quiet' and enable '--debug' on the kernel command
line to get all the debug logs -- which these days goes into the kernel
log in /dev/kmsg -- simply `cat /dev/kmsg` and use Ctrl-C to get the
shell prompt back.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Very early, before syslogd has started, we have traditionally been
forced to use stdout/stderr for debug, warning and error messages.
With this patch we can now rely on /dev/kmsg to preserve all logs
from early Finit problems. When syslogd/klogd starts up they read
/dev/kmsg and copy logs to /var/log/* files.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Refactor getent() for improved separation and readability
- Warn and exit if /etc/protocols or /etc/services is missing
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Adjust debug levels slightly and add _w() macro for LOG_WARNING,
useful when having /dev/kmsg as backend. The latter outputs any
log <= LOG_ERR to console as well.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a build problem when configured without inetd support.
Signed-off-by: Markus Svilans <msvilans@aeonyx.ca>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Add support for SysV init setenv to activate poweroff
- Update help text considerably to be more helpful
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The UTMP db tracks currently logged in users and WTMP is an audit trail
of all user logins and logouts, as well as runlevel changes and reboot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The new logit() function takes care to print to stderr if syslogd has
not yet started. The _d(), _e(), and _pe() macros have an additional
requirement: to always print on stderr.
After the boot we also want to see what happned during, so this patch
simply adds syslog() support to _d(), _e(), and _pe().
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch removes internal logging macros and adds support for logging
to stderr until a syslogd is up and running, i.e. /dev/log appears.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch removes accounting for all processes to instead implement the
traditional SysV getty accounting. (Telnet/SSH et al maintain their own
accounding for UTMP.)
TODO: WTMP support
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes the /etc/grub.d/40_custom generation for the Debian
sample so that it properly extracts the UUID For /boot and /
Note: the maintenance free symlinks /vmlinuz and /initrd.img may not
work properly with /boot on a separate partition or disk. If only
the symlinks where in /boot as well ...
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Instead of ugly #ifdefs we now probe for mdev or udevd at runtime. Take
care to check for udev-finalize and location of udevd, it differs a lot
between distros.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds initial support for handling MD RAID arrays. Adding
a pass at the very end of the shutdown sequence to signal any external
RAID controller to clear its state before we proceed to halt/reboot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for a more fine grained approach to what
filesystems are unmounted at shutdown/reboot. This so we later
can add support for `mdadm --wait-clean --scan`, which relies on
/proc, /sys, and /dev being there before we call reboot().
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Also, this code will be refactored soon to be a part of the finit
binary, like telinit, and then we will be called after unmount to
do a forced halt/reboot/poweroff.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch is a complete rewrite of the Finit reboot command. Not only
does it bring it up to speed with the new signal model, but it also adds
new features like poweroff, halt, and suspend -- including a forced mode
that goes for reboot(2) immediately.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Still TODO, replace `run_interactive()` start of dbus-daemon with a
proper service. Possibly only service_register() is needed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes the ordering of UTMP events. The first event that is
expected in /var/run/utmp is reboot, next is the runlevel change to
the configured runlevel, e.g. from S --> 2.
Let the bootmisc plugin handle this, it knows when /var/run/utmp has
been created and is writable. It is not the eact boot time, but close
enough for most usees.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for initializing /var/run/utmp with the initial
reboot as well as a shutdown record in the UTMP db.
Also, the run level change now properly records ~ and ~~ as line and id,
respectively, and the kernel release and time of each UTMP record is now
set properly.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Minor coding style cleanup, code simplification, and change service
'crashing' status text to 'crashed' to fit column width.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Simple wrappers for the signals Finit responds to. Will need complete
rewrite when support for delayed reboot/halt is added.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The condition flux state does not work like that, so keeping
this sub-command is only confusing to the user.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
These scripts help a user to configure, build, and install Finit
correctly on a Debian 8 (Jessie) server install.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The fs.fs_type is only for *BSD, we hard-code it to "rw"
in the meantime. Should be safe in most cases.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The debug mode cannot be used in run_interactive() that early, so we
must disable it during the complete fsck run.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for the following /etc/fstab API's on systems
that do not have them, e.g. musl libc on Alpine Linux.
- setfsent()
- getfsent(), and
- endfsent()
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Run fsck much earlier, as soon as we can in fact.
The first 13 characters of the UUID, label or device name should
be sufficient to identify the device, without violating max width
of 78 chars.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for running fsck on all file systems listed in
/etc/fstab with a passno > 0. The fsck runs sequentially, starting with
the root fs (passno 1).
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Disabled until we decide on config to disable/enable. It is quite
annoying to not be able to see the boot messages with it on ...
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Plugins without declared name registered with full path + ext
=> Strip path and extension
- Plugins registered with __NAME__ => strip extension
- Do *not* call dlclose() from destructor, must be called by Finit
Also, some minor cleanup and code simplifcation.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Fix compat INIT_CMD_ defines so Finit clients can talk to SysV init
- Add setenv support, only for INIT_HALT={POWERDOWN,HALT} to runlevel 6
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adopts the BusyBox init signals for handling halt, reboot,
and poweroff. The SIGINT and SIGPWR signals still behave the same as
before, even though they likely will become configurable in a future
release.
Note: the disabling of Ctrl-Alt-Delete (SIGINT) is also removed!
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
That's maybe a bit unfair. This patch implements a brief version of
what is really needed to safely unmount all file systems on a modern
Linux distribution. Yes, it's this insane. (Who am I kidding, it's
worse! Just have a look at the ancient scrolls hidden in your /etc!)
Here is where we use the new plugin_exit() and api_exit() API's from an
earlier patch. Making sure to exit and close any kernel descriptors.
Then follows a bizarre spell cast on leaking descriptors, rebooting in
a child, remounting / as read-only, and a whole slew of other R rated
stuff to prevent mount/umount from returning 32 ...
Stay safe people, use protection always!
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
FIFOs, PID files, and other non-devices belong in /var/run (or /run
these days), not /dev -- as per FHS. Compat with Debian SysV init.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The TTY handling in Finit has always been a bit of a patch work. This
commit is an attempt at a first cleanup and simplification.
Included is a brand new built-in getty which should work fine on both
virtual and serial consoles. It is however completely untested on the
latter serial consoles.
Still TODO: add support for /etc/securetty, add support for actually
setting TERM when calling /bin/login.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Embedded Linux systems usually employ BusyBox mdev. This patch assumes
that all non-embedded systems employ udev. Adding support for cleaning
up the udev db after the initramfs, or mounting /dev if that not done.
It also runs the udev-finish script at the end.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Look for 'auto' *and* 'allow-hotplug' in /etc/network/interfaces when
bringing up Debian/BusyBox interfaces at boot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If the user specifies no 'tty /dev/ttyN' in finit.conf this code
took upon itself to add a backdoor to the system. Let's remove
it and rely on the operator deciding what is best practise.
With this patch no getty, or fallback shell, will be started if
no tty line in finit.conf is available.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds prototype support for the Debian/Ubuntu/Busybox network
configuration file /etc/netork/interfaces. If this file exists Finit
will call `ifup` for each (non-commented-out) interface marked `auto`.
With this implementation we block boot until all ifup commands have
returned, which is less than ideal. A better approach would be to ifup
only loopback and create deferred tasklets to be run in parallel with
the rest of the boot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a long standing problem with TTY's. The original code
tried to workaround a limitation in the console of some embedded systems
that was detrimental on actual TTYs.
As an added benefit, the console specific "Press Enter" prompt for the
console port is now available on all TTYs -- meaning no getty needs to
be started and kept needlessly in the background until it is needed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We never get any NEWLINK message for lo since it as an interface exists
very early. So we add this workaround when lo comes up so services can
depend on exists for lo as well as for other interfaces.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The introduction of conditions to Finit only focused on
services (monitored daemons) and not parallell one-shot
tasks or blocking one-shot runs.
This patch enables condition support for 'task' and 'run'
stanzas in both finit.conf and finit.d
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
By default the urandom plugin will use /var/lib/misc/random-seed across
reboots to seed /dev/urandom at boot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Do not tag log messages with service description, but use process name.
Possibly add :ID later to denote instance if multiple instances exist.
Also, add priority daemon.info log level.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If user has no start script fall back to set up at least loopback. This
fixes (at least) a problem on Alpine Linux where `ifup -a` returns error
if `lo` is already up.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Modern systems today use a RAM disk for /run and /run/lock, with
symlinks to the FHS locations in /var/run and /var/lock. See the
following link for details http://lwn.net/Articles/436012/
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds protection against a common inetd attack vector wherein
the reply port to UDP inetd services is forged to an internal inetd
service port.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch replaces the old 'initctl emit <EV>' idea with a more
full blown condition generator:
initctl cond <set|clear|flux> <COND>
This aligns better with terminology used elsewhere in the new Finit
service state machine and dependency handling already documented.
The 'flux' sub-command is mostly intended for simulation/test of
Finit itself and should perhaps not be documented.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This is a first effort at cleaning up and consolidating all information
about the new service state machine and how conditions control this
state machine.
Also, clarified and added sections for all documents regarding how
services and conditions relate.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Some vendors provide their own Netlink plugin to Finit. This patch
makes it possible to disable the default netlink plugin from being
built. However, even before this patch it was possible to overload
the plugin. This just discards it from being built.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The AC_PLUGIN() could not properly handle if a plugin is to be enabled
by default. This patch adds support for outputting the correct help
text.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Libraries and system headers should be included with <> instead of "".
This makes a great difference for the automatic dependency tracking.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Prune standard configure flags and focus on the most important, whatever
they may be. Mention --disable-inetd and describe how --enable-static
works.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes regressions in --enable-static introduced with the
conversion to the GNU configure and build system.
The default link flag to libtool is set to -static-libtool-libs, meaning
not a fully static build -- the C library will still be used as an .so
which is usually what people want -- whereas libite (-lite) and libuEv
are linked statically.
$ ldd /sbin/finit /sbin/initctl
/sbin/finit:
linux-vdso.so.1 => (0x00007ffffde62000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f6979ad8000)
/lib64/ld-linux-x86-64.so.2 (0x000055d893e96000)
/sbin/initctl:
linux-vdso.so.1 => (0x00007ffed0dd6000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007feb3a29c000)
/lib64/ld-linux-x86-64.so.2 (0x00005556ebd2b000)
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Now that we have a proper GNU configure script there is no such thing as
lightweight, and the respected environment variables are listed at the
end of `./configure --help`
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch corrects a long standing issue with Finit. System initial
bootstrap runlevel must be set to 'S' (0), not the configured fallback
runlevel 'RUNLEVEL', which likely is '2'.
Also, minor cleanup, reduce code duplication.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The new state machine handling is not necessary to export to external
plugins, so this patch reduces the exposure of the global variable sm
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Setting O_NONBLOCK on an inetd server socket may fail, we
must check for this or the event loop may stall and block
Finit completely.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Make sure to check return value of remove(), it may actually signal
something useful. Log useful errors.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Must check return value from open(), found by Coverity Scan
- Minor refactor (and cleanup) to reduce code complexity
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Must initialize init_request buffer to zero before communicating with
Finit. Found by Coverity Scan.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If no console is configured, then the following code gets executed:
int conf_parse_config(void)
{
...
char *fallback = FALLBACK_SHELL;
...
tty_register(fallback);
...
}
int tty_register(char *line)
{
...
cmd = strtok(line, " ");
...
}
FALLBACK_SHELL is constant character array and strtok tries to modify
the first argument, which causes a crash.
Finit <3 did not use GNU configure and build system. This patch adds an
install-dev rule that is used by some to install required headers for
external plugin developers.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Add support for changing the initial hard and soft resource limits for
finit and any processes it launches.
See /etc/finit.conf section in README.md for more information.
When a service changes type from inetd to a regular daemon we must make
sure to first deregister the currently running inetd service properly.
Example: a system boots up with httpd.conf as an inetd service, so Finit
opens a socket 0.0.0.0:80 pending connections to start `httpd -i`.
After a while a user changes httpd.conf to be a regular service and
calls `initctl reload`. Without this patch the socket will remain
open and prevent httpd from opening and binding to the same port.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a problem with lingering inet sockets when inetd
services are removed. When removing an inetd service we must stop
the watcher and close the socket.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We cannot clear the svc_t memory because other applications may have
connected to the same shared memory. An application may be inspecting
the current list of services (initctl) and pulling the rug from under
them could cause some really unpredicatble results.
This reverts commit 06d1e90bac.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Both the original MIT Athena inetd and the GNU inetutils inetd only keep
a backlog of 10 connections in listen() for TCP sockets. This patch
adjusts the Finit hard coded default accordingly.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When starting an inetd service in verbose mode we print a progress
message to the console appended with "Connection". This patch simply
changes that to append with "connection" in lower case.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes several issues related to the problem of launching
UDP inetd services:
- Make sure to disable O_NONBLOCK on the socket before passing it to
the inetd service, that's what is expected. This goes for both UDP
and TCP (accepted) connections -- there is no guarantee even for TCP
sockets that they do *not* inherit the parent socket's flags.
- Mark the UDP inetd parent's SVC as busy and stop its watcher before
starting the child task to service the connection. The watcher must
be stopped or multiple child tasks may be spawned!
- Only copy the most relevant parts to the UDP task. In particular
must *not* copy the libuEv watcher data to the task!
- With the child task done and the parent inetd service unblocked we
must restore O_NONBLOCK to the socket before restarting the watcher
When a UDP inetd service connection is handled we create a child task
to service the request, meanwhile we block the parent service. There
was a problem with blocking the parent since we called inetd_stop(),
which both did a shutdown() and close() of the socket ... this patch
introduces separate handling of blocked (busy) inetd services in the
inetd_stop() function and also make sure to handle restarting the
parent watcher in inetd_start() when the child task is done.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Both the new initctl API and the new pidfile watcher plugin failed to
set CLOEXEC on their sockets. This caused forked-off and exec()'d
children to inherit all these descriptors.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Patrick Stewart added the optional `log` keyword to service, run, task
and inetd stanzas in 889bce1. It redirects stdout/stderr to syslog
using the logger tool. This addition to the README is mainly just a
place holder so it's not forgotten before the upcoming README refactor.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Adds an extra 'log' parameter to services that routes STDOUT and STDERR to syslog.
Uses a PTY rather than a pipe, as this prevents log messages hanging around in a buffer indefinitely.
In service_start() the SIGCHLD handler is reenabled before calling complete(), so the service could be reaped by the SIGCHLD handler leaving service_start() waiting forever
Also remove code that duplicates sig_unblock()
The alsa-utils, dbus, lost (PID), resolvconf, and x11-common plugins are
rarely used in the main use-case for Finit (embedded head-less systems).
This patch disables them by default and makes them possible to enable by
configure --enable-$PLUGIN switches.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Slightly more readable code and fix issue with --enable-$PLUGIN not
biting properly in the generated Makefile.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch replaces the verbose and kernel-quiet runtime options with
a (hopefully) more useful quiet and silent modes of operation.
By default Finit progressively shows all services it starts at boot, but
when boot has completed and services are started/restarted Finit prints
nothing. If `--disable-quiet` is given all started/restared/stoppping
messages are shown even after boot.
However, if `--enable-silent` is given Finit is completely silent even
at boot, no progress is shown, until the first service is started, which
on most systems is login.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Remove confusing `--with-aix-soname` option to configure. Replace with
hard-coded `aix-soname=both` => shared + static archives are build as to
be expected.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch prevents too early start of services depending on other
services. E.g, if service B depends on A then we must wait for A to
signal that it is ready before we allow B to start -- in a Finit based
system A does this by creating, or touching (updating mtime), its PID
file. Services (like A) that support SIGHUP should call utime(), or
utimensat(), on the PID file at the end of their SIGHUP handler.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
From now on the user is required to have libuEv and libite installed
before building and installing finit. This is for maintenance reasons.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for the runtime `HOOK_SVC_LOST` and the (very
noisy) *example* plugin `lost.so`.
Customer request.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
To distinguish between variables and functions in comments Finit use ()
at the end of function names.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch builds Finit dynamically, including both plugins and GIT
submodules for libite and libuev. This causes libite.so and libuev.so
to be installed in a public path, which is not good if the user has
other versions of the same libraries already installed.
The correct fix is to remove the GIT submodules for now and require
users of Finit to build and install libite and libuev separately.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Flush any existing filter rules when a configuration file is
updated. Otherwise filters are simply appended to the old
configuration which can cause finit to accept connections on
interfaces that are blocked in the new configuration.
Same as for libite, libuev has also changed to autotools, so
call its autogen.sh and configure script in cascade from our
own. Aslo add safeguards to clean/distclean targets so we
do not attempt to enter the libuev subdirectory if it's not
configured.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The new libite uses autotools to build, and since we use it as a GIT
submodule we have to call its autogen.sh before we call the configure
script, from within the Finit configure script. We may have to move
to use autotools even for Finit ...
Also, make sure to only call clean/distclean on libite if the local
version is selected at configure time.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When a service's condition transitions to `flux`, put it in the
waiting state, even if it does not support SIGHUP. That way, if the
condition returns to `on` we can simply SIGCONT it. If it goes to
`off` it will still be stop/started as before.
When reloading dynamic services, inetd services deleted marker was not
being cleaned. This caused finit to stop and start all inetd services
at every other reload.
This patch introduces a new service type called SVC_TYPE_INETD_CONN,
which represents a running inetd connection. These are handled in much
the same way as tasks. I.e. they are considered one-time jobs, with
the difference that they are removed upon completion.
When cleaning up removed services, the assumption was made that if a
service was not 'removed', it was 'clean'. In reality, it could also
be 'updated', which was then overwritten, causing services not being
restarted. Don't do that.
As a future precaution, make the dirty field const, only allowing
updates from accessor methods.
During documentation of the state machine, some theoretical problems
where discovered that could have lead finit to spawn a new instance of
a daemon before the previous one had been collected.
Now a service will always go through the STOPPING state when leaving
RUNNING. This ensures that the PID has been collected before any calls
to service_start.
...documenting your work is, apparently, not a bad idea. :)
Old event system has been replaced with a more generic condition
concept. The idea is that finit plugins may provide arbitrary
conditions that services may specify as dependencies that they require
to run.
In order to accomodate this, the service management has been
redesigned to use a state machine.
Parser would erroneously interpret an event specification containing
slashes as an inet service/proto specification.
E.g. "<net/gw>" was interpreted as port "<net" using protocol "gw>".
Stopped services are now (again) listed as "stopped" and halted
services, due to a runlevel change, are listed as "halted". Several
users complained that this change was just not intuitive. I agreee.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
By popular request, this patch adds support for querying status of a
single service. The syntax is similar to that of stop/start/reload.
Both `initctl show ...` and `initctl status ...` is supported.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When no services were stopped we must ensure to start/SIGHUP any new or
SIGHUP:able services as the last pass of the `initctl reload` cycle.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a hang when performing a system reconfiguration using
`initctl reload`, `SIGHUP` or running `(f)init q`, which caused Finit
to wait for SIGHUP:able services to stop.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch reenables verbose mode at shutdown/reboot, if it was disabled
at boot. This allows the user to see the output from the `urandom.so`
and `hwclock.so` plugins at reboot.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The urandom.so plugin uncovered that `print_desc()` does not honor the
`verbose` flag. At runtime `print_desc()` et al should no be displaying
anyting when we finit operates in terse mode.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Leverage new svc_t states in initctl status/show output. Also, ensure
different svc_t types have correct start state.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
new postponed stop handling tried to reload already stopped and removed
services. We should of course skip already stopped services and halt
those that have been removed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch reverts the change in behavior introduced in Finit v2.2,
commit 924bf041, where `initctl restart JOB` changed from 'stop + start'
to `SIGHUP` for services that support `SIGHUP`.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When `service_stop_done()` is called with a `NULL` argument and no
dynamic services have been stopped, this regression is triggered.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch makes it possible to see debug traces when running the client
program initctl, using `--debug`. Please note, this does not affect the
debug command to the daemon.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for declaring *why* a service is being stopped
or reloaded. This is later used when all stopped services have been
collected by the `service_monitor()` to issue start or postponed SIGHUP.
As an added benefit we can now also see why a service is not running.
If it has been paused by a user, halted when moving to another runlevel,
waiting for a condition (event), or similar.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The `initctl reload <JOB>` command did not work properly. This
was due to the service not being properly initialized to default
settings when reading the Finit configuration file.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch fixes a race condition when doing an ordered RELOAD and STOP
event sequence. For services that shall be stopped it is imperative
that we wait for them to be colleced by the `service_monitor()`, so that
we know they have really been stopped, before calling HOOK_SVC_RECONF.
The latter is a promise to external services that all services have been
indeed stopped.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The TTY plugin might fail in read() with EINVAL if the inotify event
buffer is too small. Fix: set the receive event buffer to the size
recommended in inotify(7). Also add a handler for EINVAL to close and
reinitialize the inotify descriptor so we do not accidentally overrun
the CPU with any unhandled errors.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When building a static Finit the plugins do not get a unique name
automatically from the file. This led to only the first plugin being
loaded, since its name was 'unknown' it was registered as such and all
other plugins conflicted, since 'unknown' was already ... known and
loaded.
This patch gives all plugins a default name, __FILE__, making it
possible to use all of them when building a static Finit.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
For embedded systems with no initial random seed, usually stored in
/var/lib/urandom/urandom-seed on desktops, servers and laptops, we now
provide an initial seed file using a built-in PRNG.
Also, let the configure script default to *enable* the random seeding
at boot, granted the urandom plugin is enabled. We should encourage
users to properly seed their random, for security reasons.
Moreover, the plugin did not properly regenerate the random seed on
reboot, or shutdown, this patch also rectifies that problem.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Upgrade libite (LITE) for new UNIX file parser API to improve static
builds of Finit.
- Re-enable bootmisc plugin for static builds by adding built-in support
for reading GID from /etc/group.
- Allow inetd support for static builds by adding built-in support for
reading service and protocol from /etc/services and /etc/protocols.
- Do not try to install/uninstall any plugins for static builds.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
service.c:
Don't declare find_inetd_svc() if inetd is disabled.
api.c:
Disable inetd API
initctl.c:
Disable presentation of inetd services
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
After forking off a new child, all signals that are blocked by finit
should be unblocked before exec-ing the final binary. Also, finit now
blocks SIGHUP, so add it to the set of signals that should be
unblocked.
We should not use -Os as default optimization, it is sometimes unstable
in cross compilers. Instead, default to -O2, which is the more tried
and tested path.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When integrating finit as a component in another system, that build
system may set VERSION as an environment variable for that OS -- we do
not want that version to be set as the version of finit.
Also, bump version to 2.0.1-dev for upcoming patch release cycle.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Check return code from `mount -na`, if there where any errors give
plugins a chance to do something about it.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds a simple `service <script> <stop|start|reload|restart>`
script for for SysV init compatibility. Very common with users of the
Ubuntu upstart init as well.
Also, fixed `make uninstall` for plugins.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Even though libite and libuEv can be disabled, using the --with-libnnn
option to configure, we let make descend into those submodules for the
clean and distclean targets. This way we can rely on make dev to work.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit makes it possible to build Finit with an external libuEv.
Without this configure script option the libuev submodule is used.
With an external libuEv we make sure to set the SONAME to libuev.so.1
to prevent any future runtime ABI issues.
Fixes issue #4.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit makes it possible to build Finit with an external libite.
Without this configure script option the libite submodule is used.
With an external libite we make sure to set the SONAME to libite.so.2
to prevent runtime ABI issues with older libite versions.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds support for event based services to Finit. Along with
the previously added support for multiple instances of the same process,
the feature scope for Finit v2.0 has been reached!
As of now a service can be declared in /etc/finit.conf or /etc/finit.d/
like this:
service :1 [2345] <!IFUP:eth0,GW> /sbin/dropbear -R -F -p 22 -- SSH daemon
Here the first instance `:1` of dropbear is declared to run in runlevels
2-5, but only if eth0 `IFUP:eth0` is up and a gateway `GW` is set. When
the configuration changes, a new gateway is set, or somehow a new `IFUP`
event for eth0 is received, then dropbear is not SIGHUP'ed, but instead
stop-started `<!>`. The latter trick applies to all services, even
those that do not define any events.
Currently inetd services are not supported for event based starting, but
it could easily be added. Likely scenario is to deny incoming requests
on, e.g., eth0 if there is no gateway.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The event API, including Netlink events to services, looks OK now
to build the remaining two features on:
- Netlink plugin for GW and IFUP IFDN events
- Generic event support for services
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Previously `HOOK_SVC_RECONF` was only called on `initctl reload`,
now we also call it on `initctl emit "STOP"`. This seems like a
good idea atm, hopefully it will remain a good idea.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This part of the README must be completely overhauled before the
release, so much of the contents is basically just TODO or design
ideas atm.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The tempfile() replacement in Finit leaked temporary files in /tmp.
The mkstemp() based implementation does not remove temporary files
when they are closed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When Finit receives SIGHUP or a reload command all services listed in
/etc/finit.d/*.conf are first stopped and then started again. I.e.,
removed/changed services are stopped and then new/changed services are
started again. In between we call HOOK_SVC_RECONF. It should always be
called, even though no services were removed/changed/added. This to
allow plugins connected to that hook may need to run as intended.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch makes it possible to communicate process name or JOB[:ID]
from initctl to Finit. When the process name translates into a service
with multiple instances Finit applies the requested command to all
instances. For example, two Dropbear SSH services running on two
different ports, calling `initctl stop dropbear` stops both services.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Fix side effect with run stanza children not being collected if finit is
not run in verbose mode. Regression introduced in release cycle.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Replace `remove()` with our own `erase()` which checks the return value
and warns on actual real failure.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The new iterators `svc_inetd_iterator()` and `svc_dynamic_iterator()`
used unsafe constructs that could cause them, at least the latter, to
dereference a `NULL` pointer.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We must check the return value from `accept()`, it may fail. Also, we
should probably retry the syscall in some cases ...
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When receiving a start/stop/reload/restart request from initctl we must
make sure to validate and NUL terminate the job ID. The `read()` API
does not NUL terminate strings for us and we cannot rely on `initctl` or
any other external API user to do so for us.
This patch improves error and boundary checking.
- Use memcpy() + hard termination instead.
- Fix use of strchr() on possibly non-NUL terminated string.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch changes the syntax for custom inetd services and adds support
for deny filters. The new syntax is:
inetd service/proto[@iface,!iface,...] </path/to/cmd | internal[.service]>
This means the second column now defines what@from and the third to/what
process. For internal services on a custom port the internal.service
syntax must be specified, so Finit can properly bind the inetd service
to the correct plugin. Here follows a few examples:
inetd time/udp wait [2345] internal -- UNIX rdate service
inetd time/tcp nowait [2345] internal -- UNIX rdate service
inetd 3737/tcp nowait [2345] internal.time -- UNIX rdate service
inetd telnet/tcp@*,!eth1,!eth0, nowait [2345] /sbin/telnetd -i -F -- Telnet service
inetd 2323/tcp@eth1,eth2,eth0 nowait [2345] /sbin/telnetd -i -F -- Telnet service
inetd 222/tcp@eth0 nowait [2345] /sbin/dropbear -i -R -F -- SSH service
inetd ssh/tcp@*,!eth0 nowait [2345] /sbin/dropbear -i -R -F -- SSH service
Access to telnet on port `2323` is only possible from interfaces `eth0`,
`eth1` and `eth2`. The standard telnet port (`23`) is available from
all other interfaces, but also `eth2`. The `*` notation used in the ssh
stanza means *any* interface, however, here `eth0` is not allowed.
NOTE: This patch breaks syntax compatibility with Finit v1.12!
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Reduce size of `helpers.c`, for linking against new `initctl` tool,
by moving out functions to `pid.c` and `exec.c`
- Add `AF_UNIX` API to Finit, to complement old `/dev/initctl` FIFO
- Let old FIFO API be used by init/telinit: `init <q | 1-9>`
- Move all advanced initctl code from `client.c` to `initctl.c`, yes
its a bit confusing to call the *new* tool the same as the old FIFO
but this is more in line with what, e.g Upstart does.
- Move all advanced server side code from `plugins/initctl.c` to `api.c`
- Update TODO with upcoming inetd syntax change and dynamic events.
- Temporarily fix display of inetd services from `initctl status -v`
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch further extends the capabilities of the client with the
ability to control the running state of services:
finit <stop|start|reload|restart> JOB
Since services can now exist with multiple instances the concept of
"jobs" is introduced in this patch. A job is a numeric identifier of
the form `NUM[:NUM]`, the latter `:NUM` is the new instance syntax
introduced earlier. See the output of `finit status` for the JOB id.
Also, with the introduction of multiple instances we broke support for
multiple related inetd services. This became obvious when the shiny new
`finit status` was tested ... hence, this patch also contains fixes to
the inetd support.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Turns out that #ID syntax for multiple instances, introduced in cb07556,
was very cumbersome to reference from the shell command line in the new
Finit client. This patch changes the syntax from #ID to :ID, which also
means that listing services/jobs in the shell will look like JOB:ID.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Further refactoring of the client code. Supported commands:
- debug: Toggle PID 1 debug flag
- version: Show Finit version or custom header string
- reload: Reload config files in /etc/finit.d/, same as SIGHUP
Also supported is the legacy 'init q' command, to reload config.
The 'status' command now supports an optiona '-v', either as an
argument to init itself, or as argument to the command. When it
is given 'status' lists verbose information about all services.
Currently only the service arguments are listed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch refactors svc.c into two files: svc.c now as a low-level
svc_t API and service.c for the more advanced rest.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
These two new hooks run when SIGHUP is received and when changing
system runlevel. The hooks are called when all services have been
stopped, just before starting up new/modified services. Making it
the perfect hook for reconfiguring the system/hardware before new
services or modified services are started up again.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When receiving SIGCHLD we must reap *all* children. We only receive one
signal, but multiple processes may have exited.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for starting and supervising multiple
services (run, task, or service) using an extra #ID number.
service #1 [2345] /sbin/httpd -f -h /http -p 80 -- Web server
service #2 [2345] /sbin/httpd -f -h /http -p 8080 -- Old web server
Also included in this patch is a fix for endless respawn of faulty
services. For instance, a buggy daemon that crashes repeatedly will now
be stopped after 10 respawns.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for adding and removing services from Finit at
runtime. Configuration file stanzas for service, run and task may now
be written to files in /etc/finit.d/*.conf, using the exact same syntax
as before in /etc/finit.conf. When a file is added, removed or modified
the user may simply SIGHUP Finit (PID 1) to activate the changes.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When a user changes runlevel, or wants to reload the config, we want the
command to return only when the operation has completed. For instance,
rolling in a new system configuration in runlevel 1 requires that we are
sure we've actually entered that runlevel before rolling it in.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for a configure script and with it
support for disabling inetd support. See the output from
./configure --help
* Add configure script
* Sprinkle #ifdef fairy dust on svc.c when inetd support is disabled
* Use GCC built-in autodep calculator (-MMD -MP)
* Fix name space issue with include files, use relative paths
* Disable username/group name to uid/gid functions in static builds
* Bug out (error) if a user tries to build the bootmisc plugin static
Possibly fixes GitHub issue #5 and issue #6.
NOTE: The static build has not received any testing at all!
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adresses issue #5. However, there still remains a few issues
to work around before Finit can be built completely standalone.
$ make clean all STATIC=1
LINK finit
helpers.o: In function `getgroup':
helpers.c:(.text+0x10a7): warning: Using 'getgrnam' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
helpers.o: In function `getuser':
helpers.c:(.text+0x108e): warning: Using 'getpwnam' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
inetd.o: In function `getent':
inetd.c:(.text+0x127): warning: Using 'getprotobyname' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
inetd.c:(.text+0xce): warning: Using 'getservbyname' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
The problem is the NSS plugins and similar issues that prevent static
linking when using the above functions. See issue #5 for more on this.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Without this patch binding to sockets when restarting them fails. This
was found when quickly changing between runlevels where a service runs,
not runs, then back to a runlevel where is shall run.
Also, refactor previous setsockopt() call to use new common macro.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
There seems to have been a regression from earlier when this worked.
I'm sure it worked ... or am I? This patch makes the whole thing look
a lot more sane, so I wonder ... might have worked in my unit tests on
x86, but not when I cross-compile to ARM.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When changing runlevels netstat still listed closed sockets as being
open and listening. With this patch all sockets are properly shut down
before we close them.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
With O_CLOEXEC we prevent forked+exec'ed children from accessing the
initctl fifo via that socket.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Stop event watcher before running callback, the plugin may manipulate
the descriptor and cause the kernel into a EPOLLHUP frenzy.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If no tty is configured in finit.conf, only use FALLBACK_SHELL (usually
/bin/sh on UNIX) when no console is configured either.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Fix `restart_any_lost_procs()`, called when finit recovers from being
stopped for a while. This function should not restart task/run/inetd
services when called -- must check for type `SVC_CMD_SERVICE` when
iterating over all registered `svc_t`, just like a regular lost pid.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Fix double close of TCP socket introduced in GIT ded4b39,
replaced with closing all open stdio descriptors, which
shouldn't really be needed.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Fixes, or rather works around the Coverity warning of possible NULL
pointer dereference. (Not an error.)
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds support for custom port and interface filtering to inetd
services, in a very condensed syntax:
inetd ssh@eth0:222/tcp nowait [2345] /usr/sbin/sshd -i
inetd ssh@eth1:22/tcp nowait [2345] /usr/sbin/sshd -i
In this example eth0 and eth1 are allowed inbound interfaces for SSH
connections, on port 222 and 22, respectively. Attempting to connect
from any other interface is denied. Also, if the system is not in
runlevel 2, 3, 4, or 5, ports 222 and 22 will not even be opened.
If eth0 is your upstream interface you may want to avoid using the
default port. To run ssh on port 222, and all others on port 22:
inetd ssh@eth0:222/tcp nowait [2345] /usr/sbin/sshd -i
inetd ssh/tcp nowait [2345] /usr/sbin/sshd -i
This actually adds a deny rule for eth0 on ssh/tcp, implicitly. You can
even list the services in the reverse order with the same result:
inetd ssh/tcp nowait [2345] /usr/sbin/sshd -i
inetd ssh@eth0:222/tcp nowait [2345] /usr/sbin/sshd -i
There is no specific deny syntax available yet, see the TODO file for
more details on how this can be implemented.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Initial support for unloading plugins. Currently we only save dlopen()
handle. The plugin_unregister() function itself is never called atm.
This should at the very least make Coverity happy.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch is slightly sub-optimal, since it will causes us to load all
availble plugins, not just those referenced by finit.conf. However, in
the short term perspective we need to reference internal inted services
provided by plugins from finit.conf, so the dependency order is clear.
In the midterm perspective we want to add support for SIGHUP to reload
finit.conf, but not plugins. So again, this is an OK patch.
One fix would be to allow loading of all plugins, parse finit.conf and
then unload all unused plugins before continuing.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch adds initial support (only SOCK_STREAM verified!) for
figuring out the inbound interface for inetd service connections.
The intention is to use this later on for a very simple tcpwrappers
replacement, e.g: `deny telnet eth0 eth3`
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch traps segfaults caused by 3rd party service callbacks in a
separate process context. Effectively preventing a single programming
mistake from taking down the entire system. Previously it was up to the
callback coder to write error free code so that PID 1 did not crash.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This reverts commit 074686b520. Which
turned out to not work so well after all. For instance, launching TTYs
in a background process completely blocked inetd services from even
starting up listening sockets ... proper fork seems to work fine though.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
We don't want to run out of scan tokens for every README update. This
also means the introduction of a new workflow. All new development must
be staged to the dev branch before being merged to master.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Bump version for coming 1.11 release
- Update `make dist` to use Kentzo's `git-archive-all` from
https://github.com/Kentzo/git-archive-all
- Rename NEWS --> CHANGELOG and update for coming release
- Update CHANGELOG with info on the libuEv migration
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Make sure to close dup()'ed descriptors when we're done with them.
Thanks to Coverity Scan for finding this nasty one!
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
If not standard finit.conf, then we want to show just the base name.
E.g., loading first /etc/finit.conf then /etc/finit.d/services.conf
Loading configuration ............. vs
Loading services configuration ....
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Try to do better by the user, by default, if no tty is specified in the
finit.conf. This way the user will at least have a way to interact with
the system and repair it.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Simplify: reduce indentation depth
- Always check for error condition first, cleaner code
- Make sure buffer is always freed, use same exit everyhere
- Use readable, but still short, local variable names
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The cppcheck tool raised an alert to the missing close() for
stdin when a console is opened for a getty. This patch fixes
that warning and also makes the code slighlty more readable.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Add support for symlinks to copyfile()
- Fix memory leak in copy_filep()
- New function, fsendfile(), copy data between FILE streams
Also, minor whitespace and coding style cleanup.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Revert crazy BANNER macro idea from 5ba7bc1. It is simply too difficult to
set -DBANNER='"Some distributions specific text"' in a parent Makefile
without entering Shell/Make escape hell ... instead, downstream will
have to patch the banner() function.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
On some platforms GCC does not zero the heap. This can lead to
serious problems, e.g., tty->pid may be non-zero while never
having been started.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Standard init implementations save current and previous runlevels
in utmp, now finit does too. Also, the initial runlevel at boot
has changed from 10 to the more obvious 0.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Add support for runlevels to tty command and support for reading baud
rate and TERM variable from finit.conf.
The tty-watcher thread has been converted to a standard I/O plugin.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Most systems need an early bootstrap phase for one-shot tasks and
necessary one-time probing. This phase usually runs well before any
networking is setup. This commit adds support for that 'S' runlevel,
and also extends the number of possible total runlevels to nine.
For the one-shot tasks that run in bootstrap two new finit.conf cmds
have been added:
- task: For one-shot tasks that can be started in parallell
- run: For one-shot tasks that must run in sequence
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit is a refactor of how arguments to finit.conf file settings
are parsed. Not all settings are run directly, but only need the arg
for future use. The code should be a lot easier to understand now.
Also, added support for specifying the desired runlevel in finit.conf
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
chomp():
Removed first \n instead of last \n
set_hostname():
If /etc/hostname exists we need to realloc the size of
the current hostname variable. Also, only try to set
the hostname if !NULL
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The ROOTDIR variable was only set in the top-level Makefile,
so when building from a subdirectory the output was slightly
garbled. Moved ROOTDIR declaration to common.mk.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Remove indentation of module loading and device node creation
- Clarify that we load kernel modules
- Clarify that we load Finit plugins
- Remove cls()
- Remove verbose condition for echo()
- Make banner a separate function with optional distribution banner
- Add a few neat ANSI escape sequences for cls(), clreol(), delline(), etc.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
More inspiration and courage from the launchd Google Talk video!
Fascinating stuff about replacing init/inetd/cron/watchdogd that
I've been thinking about for ages!
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds a line: "Loading kernel modules ..." before increasing
indentation level of module loading messages.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds support for -d, --debug option to init, which can
be used to toggle the finit debug mode at runtime.
Signed-off-by: Mattias Walström <lazzer@gmail.com>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Like most other init implementations do, when entering runlevel 1, finit
creates the file /etc/nologin to prevent users from logging in during
single-user mode, this file is then removed upon leaving runlevel 1.
Also, bugfix how FIFO I/O plugins are handled in finit. A FIFO will
need to be reopened by the server side when the client closes the pipe.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds support for specifying TTY and console to be used
for starting a getty to login to the system
Add the following to your finit.conf to use this feature:
tty ttyS0
tty ttyUSB0
console ttyS0
Signed-off-by: Mattias Walström <lazzer@gmail.com>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds a statistics counter to the finit svc_t process struct
which is incremented every time a process is restarted by the service
monitor.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This patch reorders the shutdown/reboot handler slightly and fixes
a bug in sleep(). It seems the sleep() syscall was interrupted by
receiving lots of SIGCHLD. Unsurprisingly since we just sent all
processes SIGTERM, but unfortunate since that was too short time
to deliver SIGTERM to watchdogd.
We now call sleep() again, as long as it returns a non-zero value,
until at most two (2) seconds have passed. During this time we
basically pend for watchdogd to do its job and let the WDT reboot
the system. If the two second timeout passes we send SIGKILL and
forcibly reboot(RB_AUTOBOOT) the system.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
When using the simple rsync() or copyfile() functions of libite some
error codes should not be interpreted as error. I.e., if a file or
directory already exists when attempting to create it.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The ROOTDIR variable is used for pretty formatting during compilation.
Some users might want to have this display the build from the top of
their tree instead of the finit build tree.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This also adds support for monitor /dev for TTYs appering after boot,
USB-to-serial converters for example.
Some limitations: Baudrate (and other settings) is hardcoded for now,
should be read from config file.
This is a bugfix to 785c21d, which introduced the new runparts
directive in finit.conf
- Do not attempt to free() a static string, use same approach as
username and hostname.
- Do not overwrite hostname, set the correct variable if runparts
directive is found in finit.conf
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Many callbacks check their svc->pid to determine SVC_RELOAD or SVC_START,
they rely on it to be updated. When a service is lost we must therefore
clear its recorded PID before asking the callback what to do.
Also, swap comparison in debug code for dumping service arguments ...
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The run_parts() implementation used variable arguments in a bad way,
resulting in it not working properly. Replaced va_*() code with a
simple cmd argument that's passed to each script called.
Also, the opendir()/readdir() code has been replaced with scandir()
using regular alphasort(). This simplifies the code somewhat as well.
Otherwise, minor cleanups.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Sometimes people type in the wrong path to a service directive in
finit.conf, or the executable isn't installed. This commit makes
sure to check if the command exists before attempting to start it.
This also means that we now require service commands to start with
/ to indicate an absolute path. We used to search the PATH for a
suitable executable, using execvp(), but that's rather insecure as
well, so we now use only execv().
Also, added support for adding "start" or "stop" to entries found
in /etc/finit.d starting with SNNname, or KNNname, where NN is a
number. This makes it possible to simply import sysvinit style
start scripts from /etc/rcS.d and /etc/rcN.d/.
The alsa-utils plugin also had some minor fixes, added -g option
to ignore non-existing soundcards and fixed cut-paste bug on the
_d() debug messages.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds support for the "runparts <DIR>" directive to
finit.conf. Enabling the user to override the built-in default,
which by default is /etc/finit.d.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Fix from field testing. Need to skip "." and ".." directory entries
when loading plugins from /lib/finit/plugins. Also, trigger an error
when a required plugin fails to load. We need to print to console
since syslogd has not started yet.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This has been something I wanted to avoid, but then I stumbled so hard
on it myself I simply had to add it. With this commit finit actually
comes of age -- it's now possible to do anything a modern init can --
only faster! :-)
Included in this commit is the OpenBSD version of sys/queue.h, which
unline all versions of the same file on Linux has the _FOREACH_SAFE
macros for iterating over a linked list.
The plugins are now in a TAILQ, loaded in alphabetic order and, unless
the dependency resolver says otherwise, added last in the queue. So,
naming your plugins according to standard sysvinit rules: S01, S02,
etc. takes care of the most basic dependency resolution. When that is
not sufficient, simply add a line ".depends = { "other_plugin",
"another_plugin" }," to your plugin_t definition. Be careful though
for circular dependencies, we don't check for those.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This was missing from the 1.4 release due to it being hidden in one
of my private plugins. Sorry!
Also, add another log macro. Warnings are quite useful things to have.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Restructure start by calling the plugin loader earlier. This makes it
possible to run hooks even earlier. This also adds a new hook:
HOOK_ROOTFS_UP which runs as soon as the system root (/) is available.
Another very notable change is the slow refactor of helpers.c into a finit
helper library called libite.so (-lite). This is a light-weight
Also, lots of misc. fixes all over the place. Signal handling still really
needs to be looked into once more!
Makefile:
Add dependency handling for parallell build
Add libite library
ChangeLog:
Update with new 1.4 release. There is a lot of fixes!
libite/:
Collection of utility functions and C library extensions for
finit & its plugins
reboot.c:
New reboot tool. Simply sends SIGINT to PID 1
finit.h:
Cleanup old LISTEN_INITCTL define, code is in an optional plugin now
finit.c:
Let mdev/udev run *after* cls()
Add useful debug statement for /proc/cmdline
Earlier plugin load => new early hook point
Show OK/FAIL for total plugin load, not for each plugin => speedup
helpers.h:
Rename syslog macros to prevent common name clashes (DEBUG/ERROR)
Add __FILE__ to _e(), _d() and _pe() macros
helpers.c:
In ifconfig() the addr and mask arguments are now optional
The copyfile() has been moved to libite.so
Temporarily remove const directive from some functions to build
Use DFLSIG() macro in run_getty() to simplify code
plugin.h:
Add new HOOK_ROOTFS_UP hook
plugin.c:
Fix init_plugins() to actually register the plugin's I/O fd
Skip verbose listing of all plugins in plugin_load_all()
svc.c:
svc_start(): Dito
plugins/Makefile:
Inherit CFLAGS and LDFLAGS, but make sure to filter out -rdynamic
from LDFLAGS, it's used by the daemon, only
plugins/initctl.c:
New includes after helpers.h refactor
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- Use RTLD_GLOBAL flag to tell dynamic loader to load dependent .so
files as well. Also lets other plugins use globally scoped symbols.
- Check for plugin hooks as well before dismissing plugin as unused.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit removes a lot of unnecessary overhead for handling external
service plugins. Instead of traversing the plugin array every time a
service callback should run we set the plugin callback in the svc_t when
the plugin is registered.
Also, make sure to check for norespawn/SIGSTOP in svc_reload() as well.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Improve upon API names, both internal and external, to make it clearer
where methods are located and clarify responsibility.
Add new Makefile target "dev" to create untagged development snapshots.
Bump version for upcoming release.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Updates to README include changed environment variables and a
clarification of why there is no default /etc/finit.conf after make
install.
Also, the /etc/finit.d directory is now created, as expected, by
make install.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
- 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/
- Use the 'plugindir' environemnt variable instead of 'PLUGIN_DIR'
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Break up install/uninstall targets in top Makefile into separate
install-exec, install-data, and install-dev for packagers and
distributions wishing to call Makefile with different DESTDIR's
depending on building -bin or -devel package.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Minor cleanup of Makefile and added support for installing necessary
header files, for developing external plugins, to system include path.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Move utility macros from finit.h to helpers.h and continue refactoring
and streamlining of public plugin API.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
This commit adds support for calling plugin hooks when all services
have been started properly. Useful for, e.g., a 3rd party service
monitor to trigger further events on.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
The installation path for plugins was encoded wrong in the finit binary
if *not* unset, i.e., "prefix=".
Updated the README with information on all build related environment
variables. Much of it is perhaps only needed for packagers and
distributions.
New version bump, v1.2 already.
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
2012-09-27 09:33:55 +02:00
191 changed files with 17341 additions and 2691 deletions
AC_PLUGIN([netlink], [yes], [Basic netlink plugin for IFUP/IFDN and GW events. Can be replaced with externally built plugin that links with libnl or similar.])
# Check for extra arguments or packages
AC_ARG_WITH(heading,
AS_HELP_STRING([--with-heading=GREET], [Display GREET instead of Finit version at boot]),
# Alternative boot method for other distributions than Debian 6.0
# Place this file in a /etc/finit.d/ directory. At boot finit will
# run all files, using its built-in run_parts(), in that directory.
#
# Here we chain load Debian /etc/rcS.d and /etc/rc2.d directories
# in sequence using the standard run-parts tool.
run-parts /etc/rcS.d --arg=start
run-parts /etc/rc2.d --arg=start
# You can also place other commands in this services file, akin to
# the old finit /usr/sbin/services.sh (not supported anymore).
#service acpid start
#service networking start
#service ssh start
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Joachim Nilsson
