When path_find_first_component() returns the last component, the iterator
must be an empty string. The fact is heavily used in chaseat(). Let's
explicitly test it.
Previously, struct stat may not be correctly synced with the currently
opened fd, e.g. when a path contains symlink which points to an absolute
path.
This also rename variables for struct stat, to make them consistent with
the corresponding fd.
My original assumption in 61961e693d was wrong, since we do reset
/dev/console during reexec (see make_console_stdio()), so the test was
quite racy. Let's replace it with something, hopefully, more reliable.
Follow-up to 61961e693d.
After manually editing /etc/locale.gen, calling localectl set-locale
sometimes fails. When it fails, the systemd journal shows:
systemd-localed: free() / invalid pointer.
It turned out that it only fails if some of the uncommented lines in
/etc/locale.gen have leading spaces, as in:
* C.UTF-8 <= OK
* en_US.UTF-8 <= OK
* fr_FR.UTF-8 <= NOK
After parsing a line from /etc/locale.gen, we use strstrip() to obtain
the "trimmed" line (without leading or trailing spaces).
However, we store the result of strstrip() in the original pointer
containing the untrimmed line. This pointer is later passed to free
(this is done automatically using _cleanup_free_).
This is a problem because if any leading space is present, the pointer
will essentially be shifted from its original value. This will result in
an invalid free upon cleanup.
The same issue is present in the locale_gen_locale_supported function.
Fixed by storing the result of strstrip() in a different pointer.
The documentation suggests that the mount point (i.e. the directory
where the device is mounted) is removed when the device vanishes.
However only stopping of the automount unit is implemented.
So adapt the documentation to reality.
Before this commit, if `original_path` is given,
it will always be used to overwrite `path`.
After this commit, it's controlled by the newly-added
switch `overwrite_with_origin`.
This also migrates the configuration to the new format that was
just merged in mkosi. Specifically, we make use of the new [Match]
sections to only include specific config snippets per distro.
When adding a sysext image to the system and manuall merging it, a
later "systemctl (re)start systemd-sysext" won't work because "merge"
refuses to work when something is merged already. Another problem with
"merge" at start plus "unmerge" at stop is that a service restart can't
make use of the new MOVE_MOUNT_BENEATH in the future even which would
only be available in "refresh". It also prepares us for setting up the
merged overlay for the sysroot from the initrd already, which also
would lead to the mentioned start problem of the service (One
optimization could be to skip the loading but only if we are sure that
all images were loaded and weren't modified since - this assumption is
hard because early services could want to inject a sysext, too).
Use "refresh" on service start to fix the problem that the service
can't start as soon as a manual merge was done. Also add a reload
action that allows to issue "systemctl reload systemd-sysext" and it
will make use of MOVE_MOUNT_BENEATH once we implement this in
systemd-sysext refresh (and it's available from the kernel).
The /usr/lib/extensions/ location for systemd-sysext images is not
supported anymore. In https://github.com/systemd/systemd/pull/26013
systemd-confext images got introduced and we can list its path under
/usr instead.
With certain fstabs we may propagate ENXIO from the $SYSTEMD_SYSFS_CHECK
check all the way up, making fstab-generator exit with a non-zero EC and
without any helpful message, which is really confusing.
Daan De Meyer
Yu Watanabe
Frantisek Sumsal
Olivier Gayot
Uwe Kleine-König
Dmitrii Fomchenkov
Zbigniew Jędrzejewski-Szmek
Ludwig Nussel
Mike Yuan
Cristian Rodríguez
Sorah Fukumori
Kai Lueke
Luca Boccassi
maanyagoenka