Currently there are various circular dependencies between headers
in core/. Let's get rid of these by making judicious use of forward
declarations and moving includes into implementation files instead of
having them in header files.
Getting rid of circular header includes simplifies the code and makes
various clang based tooling such as iwyu work much better on our code.
The most important change is getting rid of the manager.h include in
unit.h which is possible thanks to the previous commits. We also move
the OOMPolicy and StatusType enums to unit.h to remove the need for
other unit headers to include manager.h to get access to these enums.
inotify FD may take several milliseconds to close. We measured
daemon-reload
default: (0.427 ± 0.05) s
async: (0.323 ± 0.02) s
with 5 path units out of 422 units. I.e. ~1% of units cause ~25% of
delay, hence this fix seems like low-hanging fruit on the daemon-reload
critical path.
Particular inotify slowness pointed out by @fbuihuu.
When watching a given pathspec, systemd unconditionally installs
IN_ATTRIB watches to track the link count of the resolved file. This
way, we are notified if the watched path disappears, even if the
resolved file inode is not removed.
Similarly, systemd installs inotify watches on each parent directory, to
be notified when the specified path appears. However, for these watches
IN_ATTRIB is an unnecessary addition to the mask. In inotify, IN_ATTRIB
on a directory is emitted whenever the attributes of any child changes,
which, for many paths, has the potential to cause a high number of
spurious wakeups in systemd. Let's remove IN_ATTRIB from the mask when
installing watches on the parent directories of the specified path.
Previously, path units would remain in the running state while their
target unit is deactivating. This left a window of time where the target
unit is no longer operational (i.e. it is busy deactivating/cleaning
up/etc) but the path unit would continue to ignore inotify events. In
short: any inotify event that occurs while the target unit deactivates
would be completely lost.
With this commit, the path will go back into a waiting state when the
target unit starts deactivating. This means that any inotify event that
occurs while the target unit deactivates will queue a start job.
This is the equivalent of RequiresMountsFor=, but adds Wants= instead
of Requires=. It will be useful for example for the autogenerated
systemd-cryptsetup units.
Fixes https://github.com/systemd/systemd/issues/11646
This is supposed to be a help for compilers to apply optimizations on
functions where they can't determine whether they are const/pure on
their own. For static, local functions the compiler can do this on its
own easily however, hence the decoration with pure/const is just noise.
Let's drop it, and let the compiler to its thing better.
(Use it for exported functions, since compilers can't 'reach-over' into
other modules to determine if they are pure, except if LTO is used)
As it says on the tin, configures the unit to survive a soft reboot.
Currently all the following options have to be set by hand:
Conflicts=reboot.target kexec.target poweroff.target halt.target
Before=reboot.target kexec.target poweroff.target halt.target
After=sysinit.target basic.target
DefaultDependencies=no
IgnoreOnIsolate=yes
This is not very user friendly. If new default dependencies are added,
or new shutdown/reboot types, they also have to be added manually.
The new option is much simpler, easy to find, and does the right thing
by default.
Otherwise,
1. X.path triggered X.service, and the service has waiting start job,
2. systemctl stop X.service
3. the waiting start job is cancelled to install new stop job,
4. path_trigger_notify() is called, and may reinstall new start job,
5. the stop job cannot be installed, and triggeres assertion.
So, instead, let's add a defer event source, then enqueue the new start
job after the stop (or any other type) job finished.
Fixes https://github.com/systemd/systemd/issues/24577#issuecomment-1522628906.
-1 was used everywhere, but -EBADF or -EBADFD started being used in various
places. Let's make things consistent in the new style.
Note that there are two candidates:
EBADF 9 Bad file descriptor
EBADFD 77 File descriptor in bad state
Since we're initializating the fd, we're just assigning a value that means
"no fd yet", so it's just a bad file descriptor, and the first errno fits
better. If instead we had a valid file descriptor that became invalid because
of some operation or state change, the other errno would fit better.
In some places, initialization is dropped if unnecessary.
Often I end up debugging a problem on a system, and I
do e.g. `journalctl --grep=failed|error`. The use of the term
"failed" for condition checks adds a *lot* of unnecessary noise into
this.
Now, I know this regexp search isn't precise, but it has proven
to be useful to me.
I think "failed" is too strong of a term as a baseline, and also
just stands out to e.g. humans watching their servers boot or
whatever.
The term "met condition" is fairly widely used, e.g.
https://stackoverflow.com/questions/63751794/what-does-the-condition-is-met-exactly-mean-in-programming-languages
Use that instead.
When a service is triggered by a path unit, pass the
path unit name and the path that triggered it via env vars
to the spawned processes.
Note that this is best-effort, as there might be many triggers
at the same time, but we only get woken up by one.
This is a follow-up for f470cb6d13 which in
turn is a follow-up for a068aceafb.
The latter started to honour hidden files when deciding whether a
directory is empty. The former reverted to the old behaviour to fix
issue #23220.
It introduced a bug though: when a directory contains a larger number of
hidden entries the getdents64() buffer will not suffice to read them,
since we just allocate three entries for it (which is definitely enough
if we just ignore the . + .. entries, but not ig we ignore more).
I think it's a bit confusing that dir_is_empty() can return true even if
rmdir() on the dir would return ENOTEMPTY. Hence, let's rework the
function to make it optional whether hidden files are ignored or not.
After all, I looking at the users of this function I am pretty sure in
more cases we want to honour hidden files.
When conditions fail on a service unit, a path unit can cause
PID 1 to busy loop as it keeps trying to activate the service unit.
To avoid this from happening, add a trigger limit to the path unit,
identical to the trigger limit we have for socket units.
Initially, let's start with a high limit and not make it configurable.
If needed, we can add properties to configure the rate limit similar
to the ones we have for socket units.
Previously the mkdir_label() family of calls was implemented in
src/shared/mkdir-label.c but its functions partly declared ins
src/shared/label.h and partly in src/basic/mkdir.h (!!). That's weird
(and wrong).
Let's clean this up, and add a proper mkdir-label.h matching the .c
file.
Up until now the main reason why we didn't proceed with starting the
unit was exceed start limit burst. However, for unit types like mounts
the other reason could be effective ratelimit on /proc/self/mountinfo
event source. That means our mount unit state may not reflect current
kernel state. Hence, we need to attempt to re-run the start job again
after ratelimit on event source expires.
As we will be introducing another reason than start limit let's rename
the virtual function that implements the check.
Alternative to https://github.com/systemd/systemd/pull/20531.
Whenever a service triggered by another unit fails condition checks,
stop the triggering unit to prevent systemd busy looping trying to
start the triggered unit.
Fixes#17433. Currently, if any of the validations we do before we
check start rate limiting fail, we can still enter a busy loop as
no rate limiting gets applied. A common occurence of this scenario
is path units triggering a service that fails a condition check.
To fix the issue, we simply move up start rate limiting checks to
be the first thing we do when starting a unit. To achieve this,
we add a new method to the unit vtable and implement it for the
relevant unit types so that we can do the start rate limit checks
earlier on.
When watching paths that contain symlinks in some element we so far
always only watched the inode they are pointing to, not the symlink
inode itself. Let's fix that and always watch both. We do this by simply
installing the inotify watch once with and once without IN_DONT_FOLLOW.
For non-symlink inodes this just overrides the same watch twice (where
the second one replaces the first), which is has no effect effectively.
For symlinks it means we'll watch both source and destination.
Fixes: #17727
This mirrors the change done for systemd-resolved in
9793530228. Quoting that patch:
> We generally operate on the assumption that a source is "gone" as soon as we
> unref it. This is generally true because we have the only reference. But if
> something else holds the reference, our unref doesn't really stop the source
> and it could fire again.
In particular, we take temporary references from sd-event code, and when called
from an sd-event callback, we could temporarily see this elevated reference
count. This patch doesn't seem to change anything, but I think it's nicer to do
the same change as in other places and not rely on _unref() immediately
disabling the source.
We already do this for socket and automount units, do it for path units
too: if the triggered service keeps hitting the start limit, then fail
the triggering unit too, so that we don#t busy loop forever.
(Note that this leaves only timer units out in the cold for this kind of
protection, but it shouldn't matter there, as they are naturally
protected against busy loops: they are scheduled by time anyway).
Fixes: #16669
In 4c2ef32767 we enabled propagating
triggered unit state to the triggering unit for service units in more
load states, so that we don't accidentally stop tracking state
correctly.
Do the same for our other triggering unit states: automounts, paths, and
timers.
Also, make this an assertion rather than a simple test. After all it
should never happen that we get called for half-loaded units or units of
the wrong type. The load routines should already have made this
impossible.
Daan De Meyer
Michal Koutný
Luca Boccassi
Mike Yuan
Ronan Pigott
Adrian Vovk
Lennart Poettering
Yu Watanabe
Luca Boccassi
Zbigniew Jędrzejewski-Szmek
David Tardon
Colin Walters
Luca Boccassi
yangmingtai
Michal Sekletar