If we add a drop-in for init.scope (e.g.: to set some memory limit),
it will be loaded long after the cgroup has already been realized.
Do it again when creating the special unit.
This reverts commit e59678b2cf.
We also modify the repart integration tests to make them pass with the
changes in this commit. In short, we have to make sure every file is
owned by the user executing repart. We use tee instead of cat since it
makes that easier. This also has the benefit of improving debugability
as seeing the config file contents on stdout makes it easier to know
which test is failing.
Fix incorrection assumption about the Debian patch being buggy and
actually making TEST-73-LOCALE fail on Debian.
```
# localectl set-locale LANG=C.UTF-8
# cat /etc/default/locale
LANG=C.UTF-8
```
Suggested at https://github.com/systemd/systemd/pull/25805#discussion_r1054871210
> we now write multiple files and we cannot possibly guarantee that
> either both updates worked or neither. unix fs apis simply have
> no concept that would allow us to implement that. hence I think
> we should update what we can, log about what we cannot, but return
> success as long as our in-memory state was updated.
Similary, this also renames context_free_x11() -> context_clear_x11(),
to make it clear that these functions do not free Context.
Also, this makes context_clear() support to be called more than once.
- add missing assertions,
- rename arguments for storing results,
- always initialize results on success,
- use _cleanup_ attribute at more places, etc.
We would set .path in all cases except INSTALL_CHANGE_AUXILIARY_FAILED, where
we would just just .source. This special case is just not worth it, because
we can't easily assert that .path is set. Let's remove this special case to
help the compiler know that .path is actually set.
Avoids a warning with gcc-13.0.1-0.1.fc38.x86_64.
Whether we use a newline at the beginning of <programlisting> or not doesn't
seem to have any effect. So let's use the newline in multi-line examples; it is
easier to get the indentation right this way.
Let's make sure the timestamp file's mtime is guaranteed to change for
each boot, so that it is a useful indicator of time. Or in other words
this gurantees that systemd-timesyncd.service acts as a new kind of
milestone: that time definitely progressed on this boot even the machine
died abnormally imediately after.
It's useful being able to easily detect if a disk-based clock bump was
done, let's make it a structure message, the same way as acquiring an
NTP fix already is.
Also, set the clock to 1 µs further than the timestamp from the disk,
after all we know that that timestamp was current when it was written,
hence it can't be the right one right now anymore.
If output fields are specified, let's store this in a Set right-away,
instead of converting between strv and Set again and again for each
line.
This is not only faster, but also simpler and shorter.
Zbigniew Jędrzejewski-Szmek
Luca Boccassi
Lennart Poettering
Daan De Meyer
Jan Janssen
Michael Biebl
Yu Watanabe
Steve Ramage
DaPigGuy
David Tardon
13r0ck