Yu Watanabe [Fri, 17 Oct 2025 08:49:41 +0000 (17:49 +0900)]
coredump: enter mount namespace even when the crashed process is in the same PID namespace
Otherwise, we may not get stacktrace of the crashed process when it is
in running a mount namespace, especially when it is in a portable
service or service that uses RootImage=/RootDirectory=.
Yu Watanabe [Fri, 17 Oct 2025 00:30:33 +0000 (09:30 +0900)]
coredump-context: several cleanups
- Renames Context to CoredumpContext.
- Renames enum meta_argv_t to MetadataField, and use the correct type at
several pleaces, especially in loop.
- Parses argv (when invoked as kernel helper or with '--backtrace') or
iov (when invoked through socket) and store values in the context.
Previously we use the passed string as is, but let's make them in a
type safe way.
- Stores received or built iovw in the context. Also store input fd and
mountfs fd in the context.
- Introduces coredump_context_is_pid1() and _is_journald() helper
functions.
- Adds COREDUMP_PIDFDID= field when the kernel passed pidfd of the
crashed process.
Yu Watanabe [Wed, 8 Oct 2025 15:39:11 +0000 (00:39 +0900)]
coredump: shortcut to start backtrace mode
When running as backtrace mode, it is not necessary to use kmsg, not
necessary to disable coredump from the command, and unexpectedly
passed file descriptors can be silently ignored.
Yu Watanabe [Wed, 8 Oct 2025 05:30:25 +0000 (14:30 +0900)]
coredump-config: several modernizations
- introduce CoredumpConfig object that stores all configuration
parameters parsed from coredump.conf,
- use gperf to generate conf parser table,
- parse coredump.conf only when necessary: when systemd-coredump is
invoked as kernel helper or by socket actiavation.
Ronan Pigott [Sun, 26 Oct 2025 04:04:03 +0000 (21:04 -0700)]
zsh: add completion for dbus bus address
The DBUS_SESSION_BUS_ADDRESS and DBUS_SYSTEM_BUS_ADDRESS parameters have
an interesting syntax thats useful to complete. Let's include a
completion definition for these parameters.
core/exec-invoke: relax restriction for process name length
Previously, we limit the length of process name by 8.
This relax the restriction then at least process comm or
program_invocation_name contains the untrucated process name.
Yu Watanabe [Sat, 25 Oct 2025 06:34:44 +0000 (15:34 +0900)]
test: extend start limit interval
As the modified service requires about ~10 seconds for stopping, the
service never hit the start limit even if we tried to restart the
service more than 5 times.
This also checks that the service is actually triggered by dbus method
call.
Daniel Hast [Fri, 24 Oct 2025 22:47:59 +0000 (18:47 -0400)]
tree-wide: add basic validation of --background argument
Check whether the argument of the `--background` option of
`systemd-run`, `run0`, `systemd-nspawn`, `systemd-vmspawn`, and
`systemd-pty-forward` is either empty or looks like an ANSI color code,
and reject invalid values when parsing arguments.
We consider a string to look like an ANSI color code if it consists of
one or more sequences of ASCII digits separated by semicolons. This
permits every valid ANSI color code, and should reject anything that
results in garbled output.
discover-image: imply that hidden images are read-only
Marking a whole directory tree OS image as read-only is difficult
privilege-wise, because so far we rely on the FS_IMMUTABLE_FL which is
not accessible to unpriv clients.
One fundamental place where we currently rely on marking images
read-only is for keeping pristine copies of the originally downloaded
image around, which we place in "hidden" image directories. This is
probably the most relevant usecase for the read-only flag. And moreover,
the only usecase for the hidden images are these read-only pristine
copies.
Hence, let's make this work reasonably in the unpriv case, and simply
imply the read-only flag for hidden images. This is strictly speaking a
change in behaviour, but effectively it shouldn't be, because for nspawn
containers that are executed we insist on names that are hostname
compatible, and hidden names aren't (because they start with a dot).
rm-rf: make sure we can safely remove dirs we have no access to via rm_rf_at()
Previously, we'd first empty a dir, and then remove it. This works fine
as long as we have access to a dir. But in some cases (like for example
a foreign owned container tree) we might not have access to the dir, but
are still able to remove it (because it is empty, and in a dir we own).
Hence let's try that first. If it works, we do not need to enter the dir
(and thus fail).
Michal Sekletar [Fri, 24 Oct 2025 10:55:20 +0000 (12:55 +0200)]
coredump: handle ENOBUFS and EMSGSIZE the same way
Depending on the runtime configuration, e.g. sysctls
net.core.wmem_default= and net.core.rmem_default and on the actual
message size, sendmsg() can fail also with ENOBUFS. E.g. alloc_skb()
failure caused by net.core.[rw]mem_default=64MiB and huge fdinfo list
from process that has 90k opened FDs.
We should handle this case in the same way as EMSGSIZE and drop part of
the message.
Yu Watanabe [Mon, 20 Oct 2025 10:40:28 +0000 (19:40 +0900)]
core: increment start limit counter only when we can start the unit
Otherwise, e.g. requesting to start a unit that is under stopping may
enter the failed state.
This makes
- rename .can_start() -> .test_startable(), and make it allow to return
boolean and refuse to start units when it returns false,
- refuse earlier to start units that are in the deactivating state, so
several redundant conditions in .start() can be dropped,
- move checks for unit states mapped to UNIT_ACTIVATING from .start() to
.test_startable().
Frantisek Sumsal [Thu, 23 Oct 2025 08:28:07 +0000 (10:28 +0200)]
test: properly wait for the forked process
The process forked off by `systemd-notify --fork` is not a child of the
current shell, so using `wait` doesn't work. This then later causes a
race, when the test occasionally fails because it attempts to start a
new systemd-socket-activate instance before the old one is completely
gone:
[ 1488.947744] TEST-74-AUX-UTILS.sh[1938]: Child 1947 died with code 0
[ 1488.947952] TEST-74-AUX-UTILS.sh[1933]: + assert_eq hello hello
[ 1488.949716] TEST-74-AUX-UTILS.sh[1948]: + set +ex
[ 1488.950112] TEST-74-AUX-UTILS.sh[1950]: ++ cat /proc/1938/comm
[ 1488.945555] systemd[1]: Started systemd-networkd.service - Network Management.
[ 1488.950365] TEST-74-AUX-UTILS.sh[1933]: + assert_in systemd-socket systemd-socket-
[ 1488.950563] TEST-74-AUX-UTILS.sh[1951]: + set +ex
[ 1488.950766] TEST-74-AUX-UTILS.sh[1933]: + kill 1938
[ 1488.950766] TEST-74-AUX-UTILS.sh[1933]: + wait 1938
[ 1488.950766] TEST-74-AUX-UTILS.sh[1933]: .//usr/lib/systemd/tests/testdata/units/TEST-74-AUX-UTILS.socket-activate.sh: line 14: wait: pid 1938 is not a child of this shell
[ 1488.950766] TEST-74-AUX-UTILS.sh[1933]: + :
[ 1488.951486] TEST-74-AUX-UTILS.sh[1952]: ++ systemd-notify --fork -- systemd-socket-activate -l 1234 --now socat ACCEPT-FD:3 PIPE
[ 1488.952222] TEST-74-AUX-UTILS.sh[1953]: Failed to listen on [::]:1234: Address already in use
[ 1488.952222] TEST-74-AUX-UTILS.sh[1953]: Failed to open '1234': Address already in use
[ 1488.956831] TEST-74-AUX-UTILS.sh[1933]: + PID=1953
[ 1488.957078] TEST-74-AUX-UTILS.sh[102]: + echo 'Subtest /usr/lib/systemd/tests/testdata/units/TEST-74-AUX-UTILS.socket-activate.sh failed'
[ 1488.957078] TEST-74-AUX-UTILS.sh[102]: Subtest /usr/lib/systemd/tests/testdata/units/TEST-74-AUX-UTILS.socket-activate.sh failed
Yu Watanabe [Thu, 23 Oct 2025 00:35:03 +0000 (09:35 +0900)]
rereadpt: always update kernel partition tables from userspace in an incremental fashion (#39390)
Let's address #38672 comprehensively: let's avoid BLKRRPART as much as
we can, and always do careful userspace controlled, incremental updates
to the kernel partition tables.
This simply iterates through blkid's partition parsing, and turns it
into a BLKPG ioctls, adding, updating, removing partitions as necessary,
suppressing unnecessary changes. This has the major benefit that the
call becomes truly idempotent: if nothing changed then nothing is
removed/readed, like BLKRRPART is doing it.
This then ports over all code currently doing partition refreshing,
specifcially: udev, repart, and homed.
tree-wide: open block device locks in writable mode
udev's block device locking protocol has one pitfall not even the
example in the documentation got right so far (even though this is
explained in all detail above): udev's rescanning is only triggered when
an fd that is opened for writing is closed. This means that if a
separate locking fd is opened on a block device – one that is maintained
independently of the fd actually used for writing – it must be opened for
writing too, so that closing the lock definitely triggers a rescan. This
matters in cases where the lock fd is kept for longer than the fd used
for writing to disk. (Because otherwise udev might get the
IN_CLOSE_WRITE event, but when it tries to rescan will find the device
locked, and never retry because no IN_CLOSE_WRITE is triggred anymore.)
Let's fix that across the codebase, at 4 places:
1. in makefs (a lock fd is kept, and mkfs then invoked as child, which
uses a different fd, and the lock fd is closed only once the child
died)
2. in udevadm lock (embarassing!): which is intended to be used to wrap tools
that modify disk contents, very similar to the makefs case. The lock
is also kept until after the tool exited.
3. In storagetm: the kernel nvme-tcp layer writes to the device
directly, we just keep a lock fd.
blockdev-util: rename BlockDeviceLookupFlag to plural
This is a flags type and a flag function argument, let's name it in
plural, because it allows many flags combinations. Internally, the
implementation already used plural, but let's fix the prototypes too.
* 5650452e6b Install new files for upstream build
* 607afcd060 salsa: disable arm64/ppc64el again
* b1bb6d4849 systemd-tests: drop unused overrides
* b3790a36ca getty-static: add missing Documentation=
* 1cea27caba Backport patch to fix autopkgtest with new util-linux due to file move
* 2e74a7f969 Update changelog for 258.1-1 release
* 9250e242b9 Make /run/lock world writable by default
Luca Boccassi [Mon, 20 Oct 2025 23:37:44 +0000 (00:37 +0100)]
mountfsd: allow privileged users to mount bare unprotected filesystems
This is useful when we start to call mountfsd from root, for example
from the tests where we just use a simple squashfs/erofs.
Note that this requires the caller to be root, and it will be rejected
otherwise, as such images are classified as 'unprotected' and the
enforced policy does not accept them for unprivileged users.
Disable abort in log_assert in libsystemd/libudev (#39307)
See the second commit for details.
I think we might want to apply the same treatment to nss and pam
modules. Asserting in such "plugin code" seems iffy. But this PR doesn't
change those in any way.
Ryan Brue [Mon, 18 Aug 2025 17:12:26 +0000 (12:12 -0500)]
man: Clarify usage of /usr/share/factory/ in programs
As discussed in this thread:
https://github.com/redhat-performance/tuned/issues/798#issuecomment-3197697654
/usr/share/factory/ is not intended to be read from by programs,
but the wording in the FHS can be misread to think that programs
should be using /usr/share/factory/ as the vendor supplied configuration
directory rather than something like /usr/lib/foo/ or /usr/share/foo/.
This commit points developers to the UAPI configuration spec for how to
make their programs hermetic /usr/ compatible.
Do not use "critical assert_return" in libsystemd or libudev
Previously, when compiled in developer mode, a call into libsystemd with
invalid parameters would result in an abort. This means that it's effectively
impossible to install such libsystemd in a normal system, since various
third-party programs may now abort. A shared library should generally never
abort or exit the calling program.
In python-systemd, the test suite calls into libsystemd, to check if the proper
return values are received and propagated through the Python wrappers.
Obviously with libsystemd compiled from git, the test suite now fails
in a nasty way.
So rework the code to set assert_return_is_critical similarly to how we handle
mempool enablement: the function that returns true is declared as a week
symbol, and we "opt in" by linking a file that provides the function in
libsystemd-shared. Effectively, libsystemd and libudev always have
assert_return_is_critical==false, and our binaries and modules enable it
conditionally.
The function internally does caching which means that the result must
always be the same, the definition of a pure function. The compiler might
be able to optimize some repeated calls to the function.
Daniel Foster [Thu, 17 Jul 2025 23:59:14 +0000 (09:59 +1000)]
tree-wide: extend $LISTEN_FDS protocol with $LISTEN_PIDFDID
Although extremely unlikely, there is a race present in solely checking the
$LISTEN_PID environment variable, due to PID recycling. Fix that by introducing
$LISTEN_PIDFDID, which contains the 64-bit ID of a pidfd for the child process
that is not subject to recycling.
importd: downgrade log message about bound capability set dropping + netns
An unprivileged process cannot reduce its own capability bounding set,
hence, while it is nice to reduce the set, let's not log about it
loudly, in case we are invoked unpriv (which we explicity support these
days after all).
An unpriv process also cannot detach from its netns, hence also
downgrade the warning to a debug message.
importd: support unpacking tarballs to foreign UID range
When invoked unprivileged, let's use a transiently allocated userns, so
that we can properly untar UIDs/GIDs so that the trees appear owned by
the foreign UID/GID range.