Let's not leak details from src/shared and src/libsystemd into
src/basic, even though you can't actually do anything useful with
just forward declarations from src/shared.
The sd-forward.h header is put in src/libsystemd/sd-common as we
don't have a directory for shared internal headers for libsystemd
yet.
Let's also rename forward.h to basic-forward.h to keep things
self-explanatory.
bus_append_unit_property() and associated functions accept a long list of
properties. But the specific names are only available through code. But it is
useful to be able to know the specific list of properties that is supported, in
particular for shell completions. Thus, add a way to list the properties that
are supported by the code.
In the future we could also turn this into a test for the documentation. For
various reasons, the list of properties listed in the docs is a partially
overlapping set. E.g. for service type, the pull request
https://github.com/systemd/systemd/pull/37661 creates a list with 212 entries,
and this code generates 7 entries less and 184 more. I didn't check all the
differences, but in the few cases I did, the list generated here was actually
correctly supported by 'systemd-run -p'.
A smoke test is added.
This is not a trivial wrapper around user-created-buffer-based
syscall, so do not use _malloc suffix. Most of our functions
return an internally allocated buffer and this one's not special.
cg_get_keyed_attribute() is a more generic version of this,
and cg_is_empty_recursive() was the only user of this function,
which got converted in the previous commit.
This effectively renames cg_is_empty_recursive() to cg_is_empty().
Note that all existing code calls the former and not the latter,
hence with cgv1 support being dropped it's trivial to consult
cgroup.events directly for populated state everywhere.
Additionally, use more generic cg_get_keyed_attribute() helper
rather than cg_read_event().
- assert on supplied keys being unique
- Reject duplicate attributes with -EBADMSG
- Rename 'ret_values' to just 'values', given we don't
allocate the array
- Remove now unused cg_get_keyed_attribute_graceful()
Our baseline is v5.4 and cgroup v2 is enforced now,
which means CPU accounting is cheap everywhere without
requiring any controller, hence just remove the directive.
Let's return the size in a return parameter instead of the return value.
And if NULL is specified this tells us the caller doesn't care about the
size and expects a NUL terminated string. In that case look for an
embedded NUL byte, and refuse in that case.
This should lock things down a bit, as we'll systematically refuse
embedded NUL strings now when we expect strings.
Supersedes #35308 (cherry-picked one commit and replaced the rest)
(I left a few comments that's folded by GitHub. Please make sure to
check them too.)
When removing a cgroup, we always want to eliminate subcgroups
first, i.e. use cg_trim(). And cg_rmdir() (along with
CGROUP_REMOVE flag) is simply unused. Kill it.
Opening pidfds for non thread group leaders only works from 6.9 onwards with PIDFD_THREAD. On
older kernels or without PIDFD_THREAD pidfd_open() fails with EINVAL. Since we might read non
thread group leader IDs from cgroup.threads, we introduce and set CGROUP_NO_PIDFD to avoid
trying open pidfd's for them and instead use the pid as is.
We can only delegate actual controllers, not the BPF pseudo-controllers
we defined as there's imply no concept for that. Hence, when users set
Delegate=yes to do a wildcard delegation, only delegate the regular
controllers.
This means that we won't bother with BPF stuff for such units where it's
entirelly unnecessary.
This is just like cg_is_delegate() but operates on an fd instead of a
cgroup path.
Sooner or later we should access cgroupfs mostly via fds rather than
paths, but we aren't there yet. But let's at least get started.
This setting indicates that the given unit wants to receive coredumps
for processes that crash within the cgroup of this unit. This setting
requires that Delegate= is also true, and therefore is only available
where Delegate= is available.
This will be used by systemd-coredump to support forwarding coredumps to
containers.
Currently we spawn services by forking a child process, doing a bunch
of work, and then exec'ing the service executable.
There are some advantages to this approach:
- quick: we immediately have access to all the enourmous amount of
state simply by virtue of sharing the memory with the parent
- easy to refactor and add features
- part of the same binary, will never be out of sync
There are however significant drawbacks:
- doing work after fork and before exec is against glibc's supported
case for several APIs we call
- copy-on-write trap: anytime any memory is touched in either parent
or child, a copy of that page will be triggered
- memory footprint of the child process will be memory footprint of
PID1, but using the cgroup memory limits of the unit
The last issue is especially problematic on resource constrained
systems where hard memory caps are enforced and swap is not allowed.
As soon as PID1 is under load, with no page out due to no swap, and a
service with a low MemoryMax= tries to start, hilarity ensues.
Add a new systemd-executor binary, that is able to receive all the
required state via memfd, deserialize it, prepare the appropriate
data structures and call exec_child.
Use posix_spawn which uses CLONE_VM + CLONE_VFORK, to ensure there is
no copy-on-write (same address space will be used, and parent process
will be frozen, until exec).
The sd-executor binary is pinned by FD on startup, so that we can
guarantee there will be no incompatibilities during upgrades.
Daan De Meyer
Lennart Poettering
Mike Yuan
Zbigniew Jędrzejewski-Szmek
Jelle van der Waa
Jan Vaclav
Timo Rothenpieler
Yu Watanabe
Nick Rosbrook
Luca Boccassi