Currently, if a unit file is enabled from outside of the search path,
and that unit has an alias, then the symlink ends up pointing outside of
the search path too. For example:
$ cat /tmp/a.service
[Service]
ExecStart=sleep infinity
[Install]
Alias=b.service
WantedBy=multi-user.target
$ systemctl enable /tmp/a.service
Created symlink /etc/systemd/system/a.service → /tmp/a.service.
Created symlink /etc/systemd/system/b.service → /tmp/a.service.
Created symlink /etc/systemd/system/multi-user.target.wants/a.service → /tmp/a.service.
This then means the alias is treated as a separate unit:
$ systemctl start a.service
$ sudo systemctl status a
● a.service
Loaded: loaded (/etc/systemd/system/a.service; enabled; preset: enabled)
Active: active (running) since Fri 2024-03-15 15:17:49 EDT; 9s ago
Main PID: 769593 (sleep)
Tasks: 1 (limit: 18898)
Memory: 220.0K
CPU: 5ms
CGroup: /system.slice/a.service
└─769593 sleep infinity
Mar 15 15:17:49 six systemd[1]: Started a.service.
$ sudo systemctl status b
○ b.service
Loaded: loaded (/etc/systemd/system/b.service; alias)
Active: inactive (dead)
To fix this, make sure the alias uses a target that is inside the search
path. Since the unit file itself is outside of the search path, a
symlink inside the search path will have been created already. Hence,
just point the alias symlink to that recently created symlink.
For unit instances install_info_discover() returns path to the template,
which then generates confusing errors when passed to
do_unit_file_enable():
~# build/systemctl --root=/tmp/systemctl-test.N9ysbz reenable templ1@two.service
Unit name: templ1@two.service; p: /etc/systemd/system/templ1@.service
Removed "/tmp/systemctl-test.N9ysbz/etc/systemd/system/services.target.wants/templ1@two.service".
Failed to reenable templ1@.service, destination unit services.target is a non-template unit.
This can also be seen with a different reproducer using getty@.service
and a simple bind mount to / - there's no error this time, but it tries
to create a symlink for the default instance (from DefaultInstance=tty1),
which is also incorrect:
~# SYSTEMD_LOG_LEVEL=debug systemctl --root /mnt/bindroot/ reenable getty@test.service
Symlink /mnt/bindroot/etc/systemd/system/getty.target.wants/getty@tty1.service → /usr/lib/systemd/system/getty@.service already exists
Follow-up to: 29a7c59abbResolves: #24740
To make it consistent with other env vars, e.g. $SYSTEMD_ESP_PATH or
$SYSTEMD_XBOOTLDR_PATH.
This is useful when the root is specified by a file descriptor, instead
of a path.
This reverts commit d6c9411072.
I still think this is something that needs to be done, but we're hitting some
unexplained failures, e.g. https://github.com/systemd/systemd/issues/22920.
So let's revert this for now, so -rc2 can be released, with a plan to return
to this after a release.
Closes#22920.
On arm, we'd fail with:
target@v:5.16.8-200.fc35.armv7hl+lpae.socket: not a valid unit name "target@v:5.16.8-200.fc35.armv7hl+lpae.socket": Invalid argument
Inspired by 7910ec3bcd.
'! true' passes, because it's a conditional expression.
But '( ! true )' fails, because '( … )' creates a subshell, i.e. a separate
program, and '! true' becomes the return value of that program, and the whole
thing apparently is not a conditional expression for the outer shell.
This is shorter, so let's just do this.
We would only accept "identical" links, but having e.g. a symlink
/usr/lib/systemd/system/foo-alias.service → /usr/lib/systemd/system/foo.service
when we're trying to create /usr/lib/systemd/system/foo-alias.service →
./foo.service is OK. This fixes an issue found in ubuntuautopkg package
installation, where we'd fail when enabling systemd-resolved.service, because
the existing alias was absolute, and (with the recent patches) we were trying
to create a relative one.
A test is added.
(For .wants/.requires symlinks we were already doing OK. A test is also
added, to verify.)
When we have a symlink that goes outside of our search path, we should just
ignore the target file name. But we were verifying it, and rejecting in
the case where a symlink was created manually.
This is a fairly noticable change, but I think it needs to be done.
So far we'd create an absolute symlink to the target unit file:
.wants/foo.service → /usr/lib/systemd/system/foo.service
or
alias.service → /etc/systemd/system/aliased.service.
This works reasonably well, except in one case: where the unit file
is linked. When we look at a file link, the name of the physical file
isn't used, and we only take the account the symlink source name.
(In fact, the destination filename may not even be a well-formed unit name,
so we couldn't use it, even if we wanted to.) But this means that if
a file is linked, and specifies aliases, we'd create absolute links for
those aliases, and systemd would consider each "alias" to be a separate
unit. This isn't checked by the tests here, because we don't have a running
systemd instance, but it is easy enough to check manually.
The most reasonable way to fix this is to create relative links to the
unit file:
.wants/foo.service → ../foo.service
alias.service → aliased.service.
I opted to use no prefix for aliases, both normal and 'default.target',
and to add "../" for .wants/ and .requires/. Note that the link that is
created doesn't necessarily point to the file. E.g. if we're enabling
a file under /usr/lib/systemd/system, and create a symlink in /etc/systemd/system,
it'll still be "../foo.service", not "../../usr/lib/systemd/system/foo.service".
For our unit loading logic this doesn't matter, and figuring out a path
that actually leads somewhere would be more work. Since the user is allowed
to move the unit file, or add a new unit file in a different location, and
we don't actually follow the symlink, I think it's OK to create a dangling
symlink. The prefix of "../" is useful to give a hint that the link points
to files that are conceptually "one level up" in the directory hierarchy.
With the relative symlinks, systemd knows that those are aliases.
The tests are adjusted to use the new forms. There were a few tests that
weren't really testing something useful: 'test -e x' fails if 'x' is a
a dangling symlink. Absolute links in the chroot would be dangling, even
though the target existed in the expected path, but become non-dangling
when made relative and the test fails.
This should be described in NEWS, but I'm not adding that here, because
it'd likely result in conflicts.
I was considering deduplicating the list of target units in
WantedBy/RequiredBy. But to do this meaningfully, we'd need to do alias
expansion first, i.e. after the initial parsing is done. This seems to be
more trouble than it would be worth.
Instead, I added tests that we're doing the right thing and creating symlinks
as expected. For duplicate links, we create the link, and on the second time we
see that the link is already there, so the output is correct.
We'd create aliases and other symlinks first, and only then try to create
the main link. Since that can fail, let's do things in opposite order, and
abort immediately if we can't link the file itself.
We don't need to talk about Alias=. The approach of using Alias= to enable
units is still supported, but hasn't been advertised as the way to do thing
for many years. Using it as an explanation is just confusing.
Also, the description of templated units did not take DefaultInstance=
into account. It is updated and extended.
We had a check that was done in unit_file_resolve_symlink(). Let's move
the check to unit_validate_alias_symlink_or_warn(), which makes it available
to the code in install.c.
With this, unit_file_resolve_symlink() behaves almost the same. The warning
about "suspicious symlink" is done a bit later. I think this should be OK.
So far we'd issue a warning (before this series, just in the logs on the server
side, and before this commit, on stderr on the caller's side), but return
success. It seems that successfull return was introduced by mistake in
aa0f357fd8 (my fault :( ), which was supposed to
be a refactoring without a functional change. I think it's better to fail,
because if enablement fails, the user will most likely want to diagnose the
issue.
Note that we still do partial enablement, as far as that is possible. So if
e.g. we have [Install] Alias=foo.service foobar, we'll create the symlink
'foo.service', but not 'foobar', since that's not a valid unit name. We'll
print info about the action taken, and about 'foobar' being invalid, and return
failure.
We would resolve those specifiers to the calling user/group. This is mostly OK
when done in the manager, because the manager generally operates as root
in system mode, and a non-root in user mode. It would still be wrong if
called with --test though. But in systemctl, this would be generally wrong,
since we can call 'systemctl --system' as a normal user, either for testing
or even for actual operation with '--root=…'.
When operating in --global mode, %u/%U/%g/%G should return an error.
The information whether we're operating in system mode, user mode, or global
mode is passed as the data pointer to specifier_group_name(), specifier_user_name(),
specifier_group_id(), specifier_user_id(). We can't use userdata, because
it's already used for other things.
ENOENT is easily confused with the file that we're working on not being
present, e.g. when the file contains %o or something else that requires
os-release to be present. Let's use -EUNATCH instead to reduce that chances of
confusion if the context of the error is lost.
And once we have pinpointed the reason, let's provide a proper error message:
+ build/systemctl --root=/tmp/systemctl-test.TO7Mcb enable some-some-link6@.socket
/tmp/systemctl-test.TO7Mcb/etc/systemd/system/some-some-link6@.socket: Failed to resolve alias "target@A:%A.socket": Protocol driver not attached
Failed to enable unit, cannot resolve specifiers in "target@A:%A.socket".
The test for the variable is added in test-systemctl-enable because there we
can do it almost for free, and the variable is most likely to be used with
'systemctl enable --root' anyway.
This test has overlap with test-install-root, but it tests things at a
different level, so I think it's useful to add. It immediately shows various
bugs which will be fixed in later patches.
