Some PIDs can remain in the watched list even though their processes have
exited since a long time. It can easily happen if the main process of a forking
service manages to spawn a child before the control process exits for example.
However when a pid is about to be mapped to a unit by calling unit_watch_pid(),
the caller usually knows if the pid should belong to this unit exclusively: if
we just forked() off a child, then we can be sure that its PID is otherwise
unused. In this case we take this opportunity to remove any stalled PIDs from
the watched process list.
If we learnt about a PID in any other form (for example via PID file, via
searching, MAINPID= and so on), then we can't assume anything.
The function so far always returned -ECANCELLED, which is ignored in all
cases the function is invoked, except one: in unit_test_start_limit()
where -ECANCELLED is returned when the start limit is hit, which is part
of unit_start()'s protocol of return values.
Since the emergency_action() logic should be relatively generic and is
used in many places, let's drop the return value from it, since it's
constant anyway, and in alll cases useless. Instead, let's return it in
unit_test_start_limit(), where it's part of the protocol.
No change in behaviour.
Let's add a safety precaution: if the start condition checks for a unit
are tested too often and fail each time, let's rate limit this too.
This should add extra safety in case people define .path, .timer or
.automount units that trigger a service that as a conditoin that always
fails.
Just some renaming, no change in behaviour.
Background: I'd like to add more functions unit_test_xyz() that test
various things, hence let's streamline the naming a bit.
unit_require_mounts_for may be passed path arguments that contain "."
components like for user's home directories where "." is sometimes used
to specify some form of anchor point.
This change stops considering such path as an error and removes the "."
components instead.
Closes: #11910
KillMode=mixed and control group are used to indicate that all
process should be killed off. SendSIGKILL is used for services
that require a clean shutdown. These are typically database
service where a SigKilled process would result in a lengthy
recovery and who's shutdown or startup time is quite variable
(so Timeout settings aren't of use).
Here we take these two factors and refuse to start a service if
there are existing processes within a control group. Databases,
while generally having some protection against multiple instances
running, lets not stress the rigor of these. Also ExecStartPre
parts of the service aren't as rigoriously written to protect
against against multiple use.
closes#8630
Fixes: #11499
Let's return -EAGAIN so that on state change, unit_process_job tries to
add our job to run_queue again so that all the reloads that coalesced
into the installed reload (which itself merged into a running one)
inititally atleast runs *once*. This should ensure service picks up all
config changes reliably.
See the issue being fixed for a detailed explanation.
It would be very wrong if any of the specfier printf calls modified
any of the objects or data being printed. Let's mark all arguments as const
(primarily to make it easier for the reader to see where modifications cannot
occur).
Previously, we'd immediately propagate unit state changes into any jobs
pending for them, always. With this we only do this if the manager is
out of the "reload" state. This fixes the problem #8803 tried to
address, by simply not completing jobs until after the reload (and thus
reestablishment of the dbus connection) is complete.
Note that there's no need to later on explicitly catch up with the
missed job state changes (i.e. there's no need to call
unit_process_job() later one explicitly). That's because for jobs in
JOB_WAITING state on deserialization all jobs are requeued into the run
queue anyway, and thus checked again if they can complete now. And for
JOB_RUNNING jobs unit_catchup() phase is going to trigger missed out
state changes *after* the reload complete anyway (after all that's what
distinguishes from unit_coldplug()).
Replaces: #8803
Let's add a helper call unit_deserialize_job() for this purpose, and
let's move registration in the global jobs hash table into
job_install_deserialized() so that it it is done after all superficial
checks are done, and before transitioning into installed states, so that
rollback code is not necessary anymore.
This splits out a bunch of functions from fileio.c that have to do with
temporary files. Simply to make the header files a bit shorter, and to
group things more nicely.
No code changes, just some rearranging of source files.
Previously, we'd enqueue a unit to the dbus queue whenever the state
changed, after we processed the state change fully. This commit to the
beginning of the state change. This has the benefit that when the state
change causes a job to complete the unit is already in the dbus queue,
and thus we get the guarantee that any unit change can be sent out to
clients before the job change.
Let's inform the clients about assert/condition property changes as they
happen, it's basically for free because assert/condition property
changes generally coincide with other unit state changes (after all
these checks are done on unit_start())
add new "systemd-run-generator" for running arbitrary commands from the kernel command line as system services using the "systemd.run=" kernel command line switch
This adds SuccessActionExitStatus= and FailureActionExitStatus= that may
be used to configure the exit status to propagate in when
SuccessAction=exit or FailureAction=exit is used.
When not specified let's also propagate the exit status of the main
process we fork off for the unit.
Let's highlight the unit description string in the status updates, to
separate them a bit more the english sentence they are part of, and thus
make the different casing less surprising.
This way we can corectly ensure that when a unit that requires some
controller goes away, we propagate the removal of it all the way up, so
that the controller is turned off in all the parents too.
Previously we tried to be smart: when a new unit appeared and it only
added controllers to the cgroup mask we'd update the cached members mask
in all parents by ORing in the controller flags in their cached values.
Unfortunately this was quite broken, as we missed some conditions when
this cache had to be reset (for example, when a unit got unloaded),
moreover the optimization doesn't work when a controller is removed
anyway (as in that case there's no other way for the parent to iterate
though all children if any other, remaining child unit still needs it).
Hence, let's simplify the logic substantially: instead of updating the
cache on the right events (which we didn't get right), let's simply
invalidate the cache, and generate it lazily when we encounter it later.
This should actually result in better behaviour as we don't have to
calculate the new members mask for a whole subtree whever we have the
suspicion something changed, but can delay it to the point where we
actually need the members mask.
This allows us to simplify things quite a bit, which is good, since
validating this cache for correctness is hard enough.
Fixes: #9512
Ideally we'd even propagate this all the way to the client, by having a
separate JobType enum value for this. But it's hard to add this without
breaking compat, hence for now let's at least internally propagate this
case differently from the case "already on it".
This is then used to call job_finish_and_invalidate() slightly
differently, with the already= parameter false, as in the failed
condition case no message was likely produced so far.
This call is only used by job.c and very specific to job handling.
Moreover the very similar logic of job_emit_status_message() is already
in job.c.
Hence, let's clean this up, and move both sets of functions to job.c,
and rename them a bit so that they express precisely what they do:
1. unit_status_emit_starting_stopping_reloading() →
job_emit_begin_status_message()
2. job_emit_status_message() → job_emit_done_status_message()
The first call is after all what we call when we begin with the
execution of a job, and the second call what we call when we are done
wiht it.
Just some moving and renaming, not other changes, and hence no change in
behaviour.
Add unit name in StartLimitAction=, FailureAction= and SuccessAction=
emergency_action() reason messages, so that the problematic unit is
easily visible, for example:
"unit dbus.service failed"
This splits the "environment" field of Manager into two:
transient_environment and client_environment. The former is generated
from configuration file, kernel cmdline, environment generators. The
latter is the one the user can control with "systemctl set-environment"
and similar.
Both sets are merged transparently whenever needed. Separating the two
sets has the benefit that we can safely flush out the former while
keeping the latter during daemon reload cycles, so that env var settings
from env generators or configuration files do not accumulate, but
dynamic API changes are kept around.
Note that this change is not entirely transparent to users: if the user
first uses "set-environment" to override a transient variable, and then
uses "unset-environment" to unset it again things will revert to the
original transient variable now, while previously the variable was fully
removed. This change in behaviour should not matter too much though I
figure.
Fixes: #9972
Let's be more careful with what we serialize: let's ensure we never
serialize strings that are longer than LONG_LINE_MAX, so that we know we
can read them back with read_line(…, LONG_LINE_MAX, …) safely.
In order to implement this all serialization functions are move to
serialize.[ch], and internally will do line size checks. We'd rather
skip a serialization line (with a loud warning) than write an overly
long line out. Of course, this is just a second level protection, after
all the data we serialize shouldn't be this long in the first place.
While we are at it also clean up logging: while serializing make sure to
always log about errors immediately. Also, (void)ify all calls we don't
expect errors in (or catch errors as part of the general
fflush_and_check() at the end.
This should be much better than fgets(), as we can read substantially
longer lines and overly long lines result in proper errors.
Fixes a vulnerability discovered by Jann Horn at Google.
CVE-2018-15686
LP: #1796402https://bugzilla.redhat.com/show_bug.cgi?id=1639071
Don't add an implicit RequiresMountsFor depenency for the
WorkingDirectory of a unit if the "-" character was used to
indicate that "a missing working directory is not considered fatal"
(see systemd.exec(5)). Otherwise systemd might fail the unit
because of missing dependencies.
Add LogRateLimitIntervalSec= and LogRateLimitBurst= options for
services. If provided, these values get passed to the journald
client context, and those values are used in the rate limiting
function in the journal over the the journald.conf values.
Part of #10230
For example in a container we'd log:
Oct 17 17:01:10 rawhide systemd[1]: Started Power-Off.
Oct 17 17:01:10 rawhide systemd[1]: Forcibly powering off: unit succeeded
Oct 17 17:01:10 rawhide systemd[1]: Reached target Power-Off.
Oct 17 17:01:10 rawhide systemd[1]: Shutting down.
and on the console we'd write (in red)
[ !! ] Forcibly powering off: unit succeeded
This is not useful in any way, and the fact that we're calling an "emergency action"
is an internal implementation detail. Let's log about c-a-d and the watchdog actions
only.
The setting is now only looked at when considering an action for a job timeout
or unit start limit. It is ignored for ctrl-alt-del, SuccessAction, SuccessFailure.
v2: turn the parameter into a flag field
v3: rename Options to Flags
Let's make them typesafe, and let's add a nice macro helper for checking
if we are in a test run, which should make testing for this much easier
to read for most cases.
Cgroup v2 provides the eBPF-based device controller, which isn't currently
supported by systemd. This commit aims to provide such support.
There are no user-visible changes, just the device policy and whitelist
start working if cgroup v2 is used.
This shouldn't change control flow, with one exception: we won't send
notifications for boot progress to plymouth anymore during reload, which
is something we really shouldn't.
Allows configuring the watchdog signal (with a default of SIGABRT).
This allows an alternative to SIGABRT when coredumps are not desirable.
Appropriate references to SIGABRT or aborting were renamed to reflect
more liberal watchdog signals.
Closes#8658
Previously, we'd act immediately on StopWhenUnneeded= when a unit state
changes. With this rework we'll maintain a queue instead: whenever
there's the chance that StopWhenUneeded= might have an effect we enqueue
the unit, and process it later when we have nothing better to do.
This should make the implementation a bit more reliable, as the unit notify event
cannot immediately enqueue tons of side-effect jobs that might
contradict each other, but we do so only in a strictly ordered fashion,
from the main event loop.
This slightly changes the check when to consider a unit "unneeded".
Previously, we'd assume that a unit in "deactivating" state could also
be cleaned up. With this new logic we'll only consider units unneeded
that are fully up and have no job queued. This means that whenever
there's something pending for a unit we won't clean it up.
RootImage= may require the following settings
```
DeviceAllow=/dev/loop-control rw
DeviceAllow=block-loop rwm
DeviceAllow=block-blkext rwm
```
This adds the following settings implicitly when RootImage= is
specified.
Fixes#9737.
Usecase is to allow changing the final kill from SIGKILL to SIGQUIT which
should create a core dump useful for debugging why the service didn't stop
with the SIGTERM
These lines are generally out-of-date, incomplete and unnecessary. With
SPDX and git repository much more accurate and fine grained information
about licensing and authorship is available, hence let's drop the
per-file copyright notice. Of course, removing copyright lines of others
is problematic, hence this commit only removes my own lines and leaves
all others untouched. It might be nicer if sooner or later those could
go away too, making git the only and accurate source of authorship
information.
This part of the copyright blurb stems from the GPL use recommendations:
https://www.gnu.org/licenses/gpl-howto.en.html
The concept appears to originate in times where version control was per
file, instead of per tree, and was a way to glue the files together.
Ultimately, we nowadays don't live in that world anymore, and this
information is entirely useless anyway, as people are very welcome to
copy these files into any projects they like, and they shouldn't have to
change bits that are part of our copyright header for that.
hence, let's just get rid of this old cruft, and shorten our codebase a
bit.
Since bb28e68477 parsing failures of
certain unit file settings will result in load failures of units. This
introduces a new load state "bad-setting" that is entered in precisely
this case.
With this addition error messages on bad settings should be a lot more
explicit, as we don't have to show some generic "errno" error in that
case, but can explicitly say that a bad setting is at fault.
Internally this unit load state is entered as soon as any configuration
loader call returns ENOEXEC. Hence: config parser calls should return
ENOEXEC now for such essential unit file settings. Turns out, they
generally already do.
Fixes: #9107
This is very similar to the existing unit method coldplug() but is
called a bit later. The idea is that that coldplug() restores the unit
state from before any prior reload/restart, i.e. puts the deserialized
state in effect. The catchup() call is then called a bit later, to
catch up with the system state for which we missed notifications while
we were reloading. This is only really useful for mount, swap and device
mount points were we should be careful to generate all missing unit
state change events (i.e. call unit_notify() appropriately) for
everything that happened while we were reloading.
This reworks how systemd tracks processes on cgroupv1 systems where
cgroup notification is not reliable. Previously, whenever we had reason
to believe that new processes showed up or got removed we'd scan the
cgroup of the scope or service unit for new processes, and would tidy up
the list of PIDs previously watched. This scanning is relatively slow,
and does not scale well. With this change behaviour is changed: instead
of scanning for new/removed processes right away we do this work in a
per-unit deferred event loop job. This event source is scheduled at a
very low priority, so that it is executed when we have time but does not
starve other event sources. This has two benefits: this expensive work is
coalesced, if events happen in quick succession, and we won't delay
SIGCHLD handling for too long.
This patch basically replaces all direct invocation of
unit_watch_all_pids() in scope.c and service.c with invocations of the
new unit_enqueue_rewatch_pids() call which just enqueues a request of
watching/tidying up the PID sets (with one exception: in
scope_enter_signal() and service_enter_signal() we'll still do
unit_watch_all_pids() synchronously first, since we really want to know
all processes we are about to kill so that we can track them properly.
Moreover, all direct invocations of unit_tidy_watch_pids() and
unit_synthesize_cgroup_empty_event() are removed too, when the
unit_enqueue_rewatch_pids() call is invoked, as the queued job will run
those operations too.
All of this is done on cgroupsv1 systems only, and is disabled on
cgroupsv2 systems as cgroup-empty notifications are reliable there, and
we do not need SIGCHLD events to track processes there.
Fixes: #9138
This way we don't need to repeat the argument twice.
I didn't replace all instances. I think it's better to leave out:
- asserts
- comparisons like x & y == x, which are mathematically equivalent, but
here we aren't checking if flags are set, but if the argument fits in the
flags.
The function is similar to path_kill_slashes() but also removes
initial './', trailing '/.', and '/./' in the path.
When the second argument of path_simplify() is false, then it
behaves as the same as path_kill_slashes(). Hence, this also
replaces path_kill_slashes() with path_simplify().
This adds a flags parameter to unit_notify() which can be used to pass
additional notification information to the function. We the make the old
reload_failure boolean parameter one of these flags, and then add a new
flag that let's unit_notify() if we are configured to restart the
service.
Note that this adjusts behaviour of systemd to match what the docs say.
Fixes: #8398
When I see "test", I have to think three times what the return value
means. With "below" this is immediately clear. ratelimit_below(&limit)
sounds almost like English and is imho immediately obvious.
(I also considered ratelimit_ok, but this strongly implies that being under the
limit is somehow better. Most of the times this is true, but then we use the
ratelimit to detect triple-c-a-d, and "ok" doesn't fit so well there.)
C.f. a1bcaa07.
Scope units are populated from PIDs specified by the bus client. We do
that when a scope is started. We really shouldn't allow scopes to be
started multiple times, as the PIDs then might be heavily out of date.
Moreover, clients should have the guarantee that any scope they allocate
has a clear runtime cycle which is not repetitive.
Double newlines (i.e. one empty lines) are great to structure code. But
let's avoid triple newlines (i.e. two empty lines), quadruple newlines,
quintuple newlines, …, that's just spurious whitespace.
It's an easy way to drop 121 lines of code, and keeps the coding style
of our sources a bit tigther.
Files which are installed as-is (any .service and other unit files, .conf
files, .policy files, etc), are left as is. My assumption is that SPDX
identifiers are not yet that well known, so it's better to retain the
extended header to avoid any doubt.
I also kept any copyright lines. We can probably remove them, but it'd nice to
obtain explicit acks from all involved authors before doing that.
Currently we add target dependencies while we are loading units. This
can create ordering loops even if configuration doesn't contain any
loop. Take for example following configuration,
$ systemctl get-default
multi-user.target
$ cat /etc/systemd/system/test.service
[Unit]
After=default.target
[Service]
ExecStart=/bin/true
[Install]
WantedBy=multi-user.target
If we encounter such unit file early during manager start-up (e.g. load
queue is dispatched while enumerating devices due to SYSTEMD_WANTS in
udev rules) we would add stub unit default.target and we order it Before
test.service. At the same time we add implicit Before to
multi-user.target. Later we merge two units and we create ordering cycle
in the process.
To fix the issue we will now never add any target dependencies until we
loaded all the unit files and resolved all the aliases.
This macro will read a pointer of any type, return it, and set the
pointer to NULL. This is useful as an explicit concept of passing
ownership of a memory area between pointers.
This takes inspiration from Rust:
https://doc.rust-lang.org/std/option/enum.Option.html#method.take
and was suggested by Alan Jenkins (@sourcejedi).
It drops ~160 lines of code from our codebase, which makes me like it.
Also, I think it clarifies passing of ownership, and thus helps
readability a bit (at least for the initiated who know the new macro)
Let's better check this inside of the call than before it, so that we
never issue this while reloading, even should these calls be called due
to other reasons than just the unit notify.
This makes sure the reload state is unset a bit earlier in
manager_reload() so that we can safely call this function from there and
they do the right thing.
Follow-up for e63ebf71ed.
path_is_normalized() will reject paths longer than 4095 bytes, so it's better
to not create a stack variable of unbounded size, but instead do the check first
and only then do that allocation.
Also use _cleanup_ to make things a bit shorter.
https://oss-fuzz.com/v2/issue/5424177403133952/7000
manager_recheck_journal() and manager_recheck_dbus() would be called to early
while we were deserialiazing units, before the systemd-journald.service and
dbus.service have been deserialized. In effect we'd disable logging to the
journald and close the bus connection. The first is not very noticable, it
mostly means that logs emitted during deserialization are lost. The second is
more noticeable, because manager_recheck_dbus() would call bus_done_api() and
bus_done_system() and close dbus connections. Logging and bus connection would
then be restored later after the respective units have been deserialized.
This is easily reproduced by calling:
$ sudo gdbus call --system --dest org.freedesktop.systemd1 --object-path /org/freedesktop/systemd1 --method "org.freedesktop.systemd1.Manager.Reload"
which works fine before 8559b3b75c, and then starts failing with:
Error: GDBus.Error:org.freedesktop.DBus.Error.NoReply: Remote peer disconnected
None of this should happen, and we should delay changing state until after
deserialization is complete when reloading. manager_reload() already included
the calls to manager_recheck_journal() and manager_recheck_dbus(), so the
connection state will be updated after deserialization during reloading is done.
Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1554578.
So, the kernel's management of cgroup/BPF programs is a bit misdesigned:
if you attach a BPF program to a cgroup and close the fd for it it will
stay pinned to the cgroup with no chance of ever removing it again (or
otherwise getting ahold of it again), because the fd is used for
selecting which BPF program to detach. The only way to get rid of the
program again is to destroy the cgroup itself.
This is particularly bad for root the cgroup (and in fact any other
cgroup that we cannot realistically remove during runtime, such as
/system.slice, /init.scope or /system.slice/dbus.service) as getting rid
of the program only works by rebooting the system.
To counter this let's closely keep track to which cgroup a BPF program
is attached and let's implicitly detach the BPF program when we are
about to close the BPF fd.
This hence changes the bpf_program_cgroup_attach() function to track
where we attached the program and changes bpf_program_cgroup_detach() to
use this information. Moreover bpf_program_unref() will now implicitly
call bpf_program_cgroup_detach().
In order to simplify things, bpf_program_cgroup_attach() will now
implicitly invoke bpf_program_load_kernel() when necessary, simplifying
the caller's side.
Finally, this adds proper reference counting to BPF programs. This
is useful for working with two BPF programs in parallel: the BPF program
we are preparing for installation and the BPF program we so far
installed, shortening the window when we detach the old one and reattach
the new one.
Before this change all unit types would default to "private" in the
system service manager and "inherit" to in the user service manager.
With this change this is slightly altered: non-service units of the
system service manager are now run with KeyringMode=shared. This appears
to be the more appropriate choice as isolation is not as desirable for
mount tools, which regularly consume key material. After all mounts are
a shared resource themselves as they appear system-wide hence it makes a
lot of sense to share their key material too.
Fixes: #8159
When various references to the unit were dropped during cleanup in unit_free(),
add_to_gc_queue() could be called on this unit. If the unit was previously in
the gc queue (at the time when unit_free() was called on it), this wouldn't
matter, because it'd have in_gc_queue still set even though it was already
removed from the queue. But if it wasn't set, then the unit could be added to
the queue. Then after unit_free() would deallocate the unit, we would be left
with a dangling pointer in gc_queue.
A unit could be added to the gc queue in two places called from unit_free():
in the job_install calls, and in unit_ref_unset(). The first was OK, because
it was above the LIST_REMOVE(gc_queue,...) call, but the second was not, because
it was after that. Move the all LIST_REMOVE() calls down.
We would free stuff like the names of the unit first, and then recurse
into other structures to remove the unit from there. Technically this
was OK, since the code did not access the name, but this makes debugging
harder. And if any log messages are added in any of those functions, they
are likely to access u->id and such other basic information about the unit.
So let's move the removal of this "basic" information towards the end
of unit_free().
A .socket will reference a .service unit, by registering a UnitRef with the
.service unit. If this .service unit has the .socket unit listed in Wants or
Sockets or such, a cycle will be created. We would not free this cycle
properly, because we treated any unit with non-empty refs as uncollectable. To
solve this issue, treats refs with UnitRef in u->refs_by_target similarly to
the refs in u->dependencies, and check if the "other" unit is known to be
needed. If it is not needed, do not treat the reference from it as preventing
the unit we are looking at from being freed.
No functional change.
The source unit manages the reference. It allocates the UnitRef structure and
registers it in the target unit, and then the reference must be destroyed
before the source unit is destroyed. Thus, is should be OK to include the
pointer to the source unit, it should be live as long as the reference exists.
v2:
- rename refs to refs_by_target
"check" is unclear: what is true, what is false? Let's rename to "can_gc" and
revert the return value ("positive" values are easier to grok).
v2:
- rename from unit_can_gc to unit_may_gc
This adds a new bus call to service and scope units called
AttachProcesses() that moves arbitrary processes into the cgroup of the
unit. The primary user for this new API is systemd itself: the systemd
--user instance uses this call of the systemd --system instance to
migrate processes if itself gets the request to migrate processes and
the kernel refuses this due to access restrictions.
The primary use-case of this is to make "systemd-run --scope --user …"
invoked from user session scopes work correctly on pure cgroupsv2
environments. There, the kernel refuses to migrate processes between two
unprivileged-owned cgroups unless the requestor as well as the ownership
of the closest parent cgroup all match. This however is not the case
between the session-XYZ.scope unit of a login session and the
user@ABC.service of the systemd --user instance.
The new logic always tries to move the processes on its own, but if
that doesn't work when being the user manager, then the system manager
is asked to do it instead.
The new operation is relatively restrictive: it will only allow to move
the processes like this if the caller is root, or the UID of the target
unit, caller and process all match. Note that this means that
unprivileged users cannot attach processes to scope units, as those do
not have "owning" users (i.e. they have now User= field).
Fixes: #3388
This removes the current bus_init() call, as it had multiple problems:
it munged handling of the three bus connections we care about (private,
"api" and system) into one, even though the conditions when which was
ready are very different. It also added redundant logging, as the
individual calls it called all logged on their own anyway.
The three calls bus_init_api(), bus_init_private() and bus_init_system()
are now made public. A new call manager_dbus_is_running() is added that
works much like manager_journal_is_running() and is a lot more careful
when checking whether dbus is around. Optionally it checks the unit's
deserialized_state rather than state, in order to accomodate for cases
where we cant to connect to the bus before deserializing the
"subscribed" list, before coldplugging the units.
manager_recheck_dbus() is added, that works a lot like
manager_recheck_journal() and is invoked in unit_notify(), i.e. when
units change state.
All in all this should make handling a bit more alike to journal
handling, and it also fixes one major bug: when running in user mode
we'll now connect to the system bus early on, without conditionalizing
this in anyway.
Let's simplify things a bit: we so far called both functions every
single time, let's just merge one into the other, so that we have fewer
functions to call.
Currently we allowed delegation for alluntis with cgroup backing
except for slices. Let's make this a bit more strict for now, and only
allow this in service and scope units.
Let's also add a generic accessor unit_cgroup_delegate() for checking
whether a unit has delegation turned on that checks the new bool first.
Also, when doing transient units, let's explcitly refuse turning on
delegation for unit types that don#t support it. This is mostly
cosmetical as we wouldn't act on the delegation request anyway, but
certainly helpful for debugging.
Previous code was using the basename(id->fragment_path) which returned
incorrect result if the unit was an instance.
For example, assuming that no instances of "template" have been created so far:
$ systemctl enable template@1
Created symlink from /etc/systemd/system/multi-user.target.wants/template@1.service to /usr/lib/systemd/system/template@.service.
$ systemctl is-enabled template@3.service
disabled
$ systemctl status template@3.service
● template@3.service - openQA Worker #3
Loaded: loaded (/usr/lib/systemd/system/template@.service; enabled; vendor preset: disabled)
[...]
Here the unit file states reported by "status" and "is-enabled" were different.
Before this, each ExecRuntime object is owned by a unit. However,
it may be shared with other units which enable JoinsNamespaceOf=.
Thus, by the serialization/deserialization process, its sharing
information, more specifically, reference counter is lost, and
causes issue #7790.
This makes ExecRuntime objects be managed by manager, and changes
the serialization/deserialization process.
Fixes#7790.
Let's add a per-unit boolean that tells us whether our unit is currently
counted or not. This way it's unlikely we get out of sync again and
things are generally more robust.
This also allows us to remove the counting logic specific to service
units (which was in fact mostly a copy from the generic implementation),
in favour of fully generic code.
Replaces: #7824
This call determines whether a specific unit currently needs access to
the console. It's a fancy wrapper around
exec_context_may_touch_console() ultimately, however for service units
we'll explicitly exclude the SERVICE_EXITED state from when we report
true.
Previously, we'd maintain two hashmaps keyed by PIDs, pointing to Unit
interested in SIGCHLD events for them. This scheme allowed a specific
PID to be watched by exactly 0, 1 or 2 units.
With this rework this is replaced by a single hashmap which is primarily
keyed by the PID and points to a Unit interested in it. However, it
optionally also keyed by the negated PID, in which case it points to a
NULL terminated array of additional Unit objects also interested. This
scheme means arbitrary numbers of Units may now watch the same PID.
Runtime and memory behaviour should not be impact by this change, as for
the common case (i.e. each PID only watched by a single unit) behaviour
stays the same, but for the uncommon case (a PID watched by more than
one unit) we only pay with a single additional memory allocation for the
array.
Why this all? Primarily, because allowing exactly two units to watch a
specific PID is not sufficient for some niche cases, as processes can
belong to more than one unit these days:
1. sd_notify() with MAINPID= can be used to attach a process from a
different cgroup to multiple units.
2. Similar, the PIDFile= setting in unit files can be used for similar
setups,
3. By creating a scope unit a main process of a service may join a
different unit, too.
4. On cgroupsv1 we frequently end up watching all processes remaining in
a scope, and if a process opens lots of scopes one after the other it
might thus end up being watch by many of them.
This patch hence removes the 2-unit-per-PID limit. It also makes a
couple of other changes, some of them quite relevant:
- manager_get_unit_by_pid() (and the bus call wrapping it) when there's
ambiguity will prefer returning the Unit the process belongs to based on
cgroup membership, and only check the watch-pids hashmap if that
fails. This change in logic is probably more in line with what people
expect and makes things more stable as each process can belong to
exactly one cgroup only.
- Every SIGCHLD event is now dispatched to all units interested in its
PID. Previously, there was some magic conditionalization: the SIGCHLD
would only be dispatched to the unit if it was only interested in a
single PID only, or the PID belonged to the control or main PID or we
didn't dispatch a signle SIGCHLD to the unit in the current event loop
iteration yet. These rules were quite arbitrary and also redundant as
the the per-unit handlers would filter the PIDs anyway a second time.
With this change we'll hence relax the rules: all we do now is
dispatch every SIGCHLD event exactly once to each unit interested in
it, and it's up to the unit to then use or ignore this. We use a
generation counter in the unit to ensure that we only invoke the unit
handler once for each event, protecting us from confusion if a unit is
both associated with a specific PID through cgroup membership and
through the "watch_pids" logic. It also protects us from being
confused if the "watch_pids" hashmap is altered while we are
dispatching to it (which is a very likely case).
- sd_notify() message dispatching has been reworked to be very similar
to SIGCHLD handling now. A generation counter is used for dispatching
as well.
This also adds a new test that validates that "watch_pid" registration
and unregstration works correctly.
It was forbidden to create mount units for a symlink. But the reason is
that the mount unit needs to know the real path that will appear in
/proc/self/mountinfo. The kernel dereferences *all* the symlinks in the
path at mount time (I checked this with `mount -c` running under `strace`).
This will have no effect on most systems. As recommended by docs, most
systems use /etc/fstab, as opposed to native mount unit files.
fstab-generator dereferences symlinks for backwards compatibility.
A relatively minor issue regarding Time Of Check / Time Of Use also exists
here. I can't see how to get rid of it entirely. If we pass an absolute
path to mount, the racing process can replace it with a symlink. If we
chdir() to the mount point and pass ".", the racing process can move the
directory. The latter might potentially be nicer, except that it breaks
WorkingDirectory=.
I'm not saying the race is relevant to security - I just want to consider
how bad the effect is. Currently, it can make the mount unit active (and
hence the job return success), despite there never being a matching entry
in /proc/self/mountinfo. This wart will be removed in the next commit;
i.e. it will make the mount unit fail instead.
Let's remove a number of synchronization points from our service
startups: let's drop synchronous match installation, and let's opt for
asynchronous instead.
Also, let's use sd_bus_match_signal() instead of sd_bus_add_match()
where we can.
This adds a new safe_fork() wrapper around fork() and makes use of it
everywhere. The new wrapper does a couple of things we previously did
manually and separately in a safer, more correct and automatic way:
1. Optionally resets signal handlers/mask in the child
2. Sets a name on all processes we fork off right after forking off (and
the patch assigns useful names for all processes we fork off now,
following a systematic naming scheme: always enclosed in () – in order
to indicate that these are not proper, exec()ed processes, but only
forked off children, and if the process is long-running with only our
own code, without execve()'ing something else, it gets am "sd-" prefix.)
3. Optionally closes all file descriptors in the child
4. Optionally sets a PR_SET_DEATHSIG to SIGTERM in the child, in a safe
way so that the parent dying before this happens being handled
safely.
5. Optionally reopens the logs
6. Optionally connects stdin/stdout/stderr to /dev/null
7. Debug logs about the forked off processes.
An auto-restarted unit B may depend on unit A with StopWhenUnneeded=yes.
If A stops before B's restart timeout expires, it'll be started again as part
of B's dependent jobs. However, if stopping takes longer than the timeout, B's
running stop job collides start job which also cancels B's start job. Result is
that neither A or B are active.
Currently, when a service with automatic restarting fails, it transitions
through following states:
1) SERVICE_FAILED or SERVICE_DEAD to indicate the failure,
2) SERVICE_AUTO_RESTART while restart timer is running.
The StopWhenUnneeded= check takes place in service_enter_dead between the two
state mentioned above. We temporarily store the auto restart flag to query it
during the check. Because we don't return control to the main event loop, this
new service unit flag needn't be serialized.
This patch prevents the pathologic situation when the service with Restart=
won't restart automatically. As a side effect it also avoid restarting the
dependency unit with StopWhenUnneeded=yes.
Fixes: #7377
This changes the unit search path logic to never drop the transient and
control directories from the unit search path. This is necessary as we
add new entries to both during runtime, due to the "systemctl
set-property" and transient unit logic.
Previously, the "transient" directory was created during early boot to
deal with this, but the "control" directories were not covered like
that. Creating the control directories early at boot is not possible
however, as /etc might be read-only then, and we do define a persistent
control directory. Hence, let's create these dirs on-demand when we need
them, and make sure the search path clean-up logic never drops them from
the search path even if they are initially missing.
(Also, always create these paths properly labelled)
This majorly refactors the transient unit file and drop-in writing
logic, so that we properly C-escape and specifier-escape (% → %%)
everything we write out, so that when we read it back again, specifiers
are parsed that aren't supposed to be parsed.
This renames unit_write_drop_in() and friends by unit_write_setting().
The name change is supposed to clarify that the functions are not only
used to write drop-in files, but also transient unit files.
The previous "mode" parameter to this function is replaced by a more
generic "flags", which knows additional flags for implicit C-style and
specifier escaping before writing things out. This can cover most
properties where either form of escaping is defined. For the cases where
this isn't sufficient, we add helpers unit_escape_setting() and
unit_concat_strv() for escaping individual strings or strvs properly.
While we are at it, we also prettify generation of transient unit files:
we try to reduce the number of section headers written out: previously
we'd write the right section header our for each setting. With this
change we do so only if the setting lives in a different section than
the one before.
(This should also be considered preparation for when we add proper APIs
to systemd to write normal, persistant unit files through the bus API)
Now that we don't kill control processes anymore, let's at least warn
about any processes left-over in the unit cgroup at the moment of
starting the unit.
Let's move the cgroup empty check for all unit types into the generic
unit_check_gc() call, out of the per-unit-type _check_gc() type. This
not only allows us to share some code, but also hooks up mount and
socket units with this kind of check, for free, as it was missing there
previously.
Before this patch, the bpf cgroup realization state was implicitly set
to "NO", meaning that the bpf configuration was realized but was turned
off. That means invalidation requests for the bpf stuff (which we issue
in blanket fashion when doing a daemon reload) would actually later
result in a us re-realizing the unit, under the assumption it was
already realized once, even though in reality it never was realized
before.
This had the effect that after each daemon-reload we'd end up realizing
*all* defined units, even the unloaded ones, populating cgroupfs with
lots of unneeded empty cgroups.
With this fix we properly set the realiazation state to "INVALIDATED",
i.e. indicating the bpf stuff was never set up for the unit, and hence
when we try to invalidate it later we won't do anything.
This introduces a new function unit_prepare_exec() that encapsulates a
number of calls we do in preparation for spawning off some processes in
all our unit types that do so.
This allows us to neatly unify a bit of code between unit types and
shorten our code.
SuccessAction= is similar to FailureAction= but declares what to do on
success of a unit, rather than on failure. This is useful for running
commands in qemu/nspawn images, that shall power down on completion. We
frequently see "ExecStopPost=/usr/bin/systemctl poweroff" or so in unit
files like this. Offer a simple, more declarative alternative for this.
While we are at it, hook up failure action with unit_dump() and
transient units too.
Already, path_is_safe() refused paths container the "." dir. Doing that
isn't strictly necessary to be "safe" by most definitions of the word.
But it is necessary in order to consider a path "normalized". Hence,
"path_is_safe()" is slightly misleading a name, but
"path_is_normalize()" is more descriptive, hence let's rename things
accordingly.
No functional changes.
Right now, the option only takes one of two possible values "inactive"
or "inactive-or-failed", the former being the default, and exposing same
behaviour as the status quo ante. If set to "inactive-or-failed" units
may be collected by the GC logic when in the "failed" state too.
This logic should be a nicer alternative to using the "-" modifier for
ExecStart= and friends, as the exit data is collected and logged about
and only removed when the GC comes along. This should be useful in
particular for per-connection socket-activated services, as well as
"systemd-run" command lines that shall leave no artifacts in the
system.
I was thinking about whether to expose this as a boolean, but opted for
an enum instead, as I have the suspicion other tweaks like this might be
a added later on, in which case we extend this setting instead of having
to add yet another one.
Also, let's add some documentation for the GC logic.
When preparing for a restart we quickly go through the DEAD/INACTIVE
service state before entering AUTO_RESTART. When doing this, we need to
make sure we don't destroy the FD store. Previously this was done by
checking the failure state of the unit, and keeping the FD store around
when the unit failed, under the assumption that the restart logic will
then get into action.
This is not entirely correct howver, as there might be failure states
that will no result in restarts.
With this commit we slightly alter the logic: a ref counter for the fd
store is added, that is increased right before we handle the restart
logic, and decreased again right-after.
This should ensure that the fdstore lives exactly as long as it needs.
Follow-up for f0bfbfac43.
