1 systemd System and Service Manager
5 Announcements of Future Feature Removals and Incompatible Changes:
7 * Support for automatic flushing of the nscd user/group database caches
8 will be dropped in a future release.
10 * Support for cgroup v1 ('legacy' and 'hybrid' hierarchies) is now
11 considered obsolete and systemd by default will refuse to boot under
12 it. To forcibly reenable cgroup v1 support,
13 SYSTEMD_CGROUP_ENABLE_LEGACY_FORCE=1 must be set on kernel command
14 line. The meson option 'default-hierarchy=' is also deprecated, i.e.
15 only cgroup v2 ('unified' hierarchy) can be selected as build-time
18 * Previously, systemd-networkd did not explicitly remove any bridge
19 VLAN IDs assigned on bridge master and ports. Since version 256, if a
20 .network file for an interface has at least one valid setting in the
21 [BridgeVLAN] section, then all assigned VLAN IDs on the interface
22 that are not configured in the .network file are removed.
24 * systemd-gpt-auto-generator will stop generating units for ESP or
25 XBOOTLDR partitions if it finds mount entries for or below the /boot/
26 or /efi/ hierarchies in /etc/fstab. This is to prevent the generator
27 from interfering with systems where the ESP is explicitly configured
28 to be mounted at some path, for example /boot/efi/ (this type of
29 setup is obsolete, but still commonly found).
31 * The behavior of systemd-sleep and systemd-homed has been updated to
32 freeze user sessions when entering the various sleep modes or when
33 locking a homed-managed home area. This is known to cause issues with
34 the proprietary NVIDIA drivers. Packagers of the NVIDIA proprietary
35 drivers may want to add drop-in configuration files that set
36 SYSTEMD_SLEEP_FREEZE_USER_SESSION=false for systemd-suspend.service
37 and related services, and SYSTEMD_HOME_LOCK_FREEZE_SESSION=false for
38 systemd-homed.service.
40 * systemd-tmpfiles and systemd-sysusers, when given a relative
41 configuration file path (with at least one directory separator '/'),
42 will open the file directly, instead of searching for the given
43 partial path in the standard locations. The old mode wasn't useful
44 because tmpfiles.d/ and sysusers.d/ configuration has a flat
45 structure with no subdirectories under the standard locations and
46 this change makes it easier to work with local files with those
49 * systemd-tmpfiles now properly applies nested configuration to 'R' and
50 'D' stanzas. For example, with the combination of 'R /foo' and 'x
51 /foo/bar', /foo/bar will now be excluded from removal.
53 General Changes and New Features:
55 * Various programs will now attempt to load the main configuration file
56 from locations below /usr/lib/, /usr/local/lib/, and /run/, not just
57 below /etc/. For example, systemd-logind will look for
58 /etc/systemd/logind.conf, /run/systemd/logind.conf,
59 /usr/local/lib/systemd/logind.conf, and /usr/lib/systemd/logind.conf,
60 and use the first file that is found. This means that the search
61 logic for the main config file and for drop-ins is now the same.
63 Similarly, kernel-install will look for the config files in
64 /usr/lib/kernel/ and the other search locations, and now also
67 systemd-udevd now supports drop-ins for udev.conf.
69 * A new 'systemd-vpick' binary has been added. It implements the new
70 vpick protocol, where a "*.v/" directory may contain multiple files
71 which have versions (following the UAPI version format specification)
72 embedded in the file name. The files are ordered by version and
73 the newest one is selected.
75 systemd-nspawn --image=/--directory=, systemd-dissect, systemd-portabled,
76 and the RootDirectory=, RootImage=, ExtensionImages=, and
77 ExtensionDirectories= settings for units now support the vpick
78 protocol and allow the latest version to be selected automatically if
79 a "*.v/" directory is specified as the source.
81 * Encrypted service credentials can now be made accessible to
82 unprivileged users. systemd-creds gained new options --user/--uid=
83 for encrypting/decrypting a credential for a specific user.
85 * New command-line tool 'importctl' to download, import, and export
86 disk images via systemd-importd is added with the following verbs:
87 pull-tar, pull-raw, import-tar, import-raw, import-fs, export-tar,
88 export-raw, list-transfers, and cancel-transfer. This functionality
89 was previously available in "machinectl", where it was used
90 exclusively for machine images. The new "importctl" generalizes this
91 for sysext, confext, and portable service images.
95 * New system manager setting ProtectSystem= has been added. It is
96 analogous to the unit setting, but applies to the whole system. It is
97 enabled by default in the initrd.
99 * New unit setting WantsMountsFor= has been added. It is analogous to
100 RequiresMountsFor=, but creates a Wants= dependency instead of
101 Requires=. This new logic is now used in various places where mounts
102 were added as dependencies for other settings (WorkingDirectory=-…,
103 PrivateTmp=yes, cryptsetup lines with 'nofail').
105 * New unit setting MemoryZSwapWriteback= can be used to control the new
106 memory.zswap.writeback cgroup knob added in kernel 6.8.
108 * The manager gained a org.freedesktop.systemd1.StartAuxiliaryScope()
109 D-Bus method to devolve some processes from a service into a new
110 scope. This new scope will remain running, even when the original
111 service unit is restarted or stopped. This allows a service unit to
112 split out some worker processes which need to continue running.
113 Control group properties of the new scope are copied from the
114 originating unit, so various limits are retained.
116 * Units now expose properties EffectiveMemoryMax=,
117 EffectiveMemoryHigh=, and EffectiveTasksMax=, which report the
118 most stringent limit systemd is aware of for the given unit.
120 * A new unit file specifier %D expands to $XDG_DATA_HOME (for user
121 services) or /usr/share/ (for system services).
123 * AllowedCPUs= now supports specifier expansion.
125 * What= setting in .mount and .swap units now accepts fstab-style
126 identifiers, for example UUID=… or LABEL=….
128 * RestrictNetworkInterfaces= now supports alternative network interface
131 * PAMName= now implies SetLoginEnvironment=yes.
133 * systemd.firstboot=no can be used on the kernel command-line to
134 disable interactive queries, but allow other first boot configuration
135 to happen based on credentials.
137 * The system's hostname can be configured via the systemd.hostname
140 * The systemd binary will no longer chainload sysvinit's "telinit"
141 binary when called under the init/telinit name on a system that isn't
142 booted with systemd. This previously has been supported to make sure
143 a distribution that has both init systems installed can reasonably
144 switch from one to the other via a simple reboot. Distributions
145 apparently have lost interest in this, and the functionality has not
146 been supported on the primary distribution this was still intended
147 for for a long time, and hence has been removed now.
149 * A new concept called "capsules" has been introduced. "Capsules" wrap
150 additional per-user service managers, whose users are transient and
151 are only defined as long as the service manager is running. (This is
152 implemented via DynamicUser=1), allowing a user manager to be used to
153 manager a group of processes without needing to create an actual user
154 account. These service managers run with home directories of
155 /var/lib/capsules/<capsule-name> and can contain regular services and
156 other units. A capsule is started via a simple "systemctl start
157 capsule@<name>.service". See the capsule@.service(5) man page for
160 Various systemd tools (including, and most importantly, systemctl and
161 systemd-run) have been updated to interact with capsules via the new
162 "--capsule="/"-C" switch.
164 * .socket units gained a new setting PassFileDescriptorsToExec=, taking
165 a boolean value. If set to true the file descriptors the socket unit
166 encapsulates are passed to the ExecStartPost=, ExecStopPre=,
167 ExecStopPost= using the usual $LISTEN_FDS interface. This may be used
168 for doing additional initializations on the sockets once they are
169 allocated. (For example, to install an additional eBPF program on
172 * The .socket setting MaxConnectionsPerSource= (which so far put a
173 limit on concurrent connections per IP in Accept=yes socket units),
174 now also has an effect on AF_UNIX sockets: it will put a limit on the
175 number of simultaneous connections from the same source UID (as
176 determined via SO_PEERCRED). This is useful for implementing IPC
177 services in a simple Accept=yes mode.
179 * The service manager will now maintain a counter of soft reboot cycles
180 the system went through. It may be queried via the D-Bus APIs.
182 * systemd's execution logic now supports the new pidfd_spawn() API
183 introduced by glibc 2.39, which allows us to invoke a subprocess in a
184 target cgroup and get a pidfd back in a single operation.
186 * systemd/PID 1 will now send an additional sd_notify() message to its
187 supervising VMM or container manager reporting the selected hostname
188 ("X_SYSTEMD_HOSTNAME=") and machine ID ("X_SYSTEMD_MACHINE_ID=") at
189 boot. Moreover, the service manager will send additional sd_notify()
190 messages ("X_SYSTEMD_UNIT_ACTIVE=") whenever a target unit is
191 reached. This can be used by VMMs/container managers to schedule
192 access to the system precisely. For example, the moment a system
193 reports "ssh-access.target" being reached a VMM/container manager
194 knows it can now connect to the system via SSH. Finally, a new
195 sd_notify() message ("X_SYSTEMD_SIGNALS_LEVEL=2") is sent the moment
196 PID 1 has successfully completed installation of its various UNIX
197 process signal handlers (i.e. the moment where SIGRTMIN+4 sent to
198 PID 1 will start to have the effect of shutting down the system
201 * New D-Bus properties ExecMainHandoverTimestamp and
202 ExecMainHandoverTimestampMonotonic are now published by services of
203 type exec, dbus, notify, and notify-reload.
204 This timestamp is taken as the very last operation before executing
205 a service's binary, which allows users to accurately track when
206 execution control of the process is handed over from systemd to the
211 * systemd-journald can now forward journal entries to a socket
212 (AF_INET, AF_INET6, AF_UNIX, or AF_VSOCK). The socket can be
213 specified in journald.conf via a new option ForwardAddress= or via
214 the 'journald.forward_address' credential. Log records are sent in
215 the Journal Export Format. A related setting MaxLevelSocket= has been
216 added to control the maximum log levels for the messages sent to this
219 * systemd-vmspawn gained a new --forward-journal= option to forward the
220 virtual machine's journal entries to the host. This is done over a
221 AF_VSOCK socket, i.e. it does not require networking in the guest.
223 * journalctl gained option '-i' as a shortcut for --file=.
225 * journalctl gained a new -T/--exclude-identifier= option to filter
226 out certain syslog identifiers.
228 * journalctl gained a new --list-namespaces option.
230 * systemd-journal-remote now also accepts AF_VSOCK and AF_UNIX sockets
231 (so it can be used to receive entries forwarded by systemd-journald).
233 * systemd-journal-gatewayd allows restricting the time range of
234 retrieved entries with a new "realtime=[<since>]:[<until>]" URL
237 * systemd-cat gained a new option --namespace= to specify the target
238 journal namespace to which the output shall be connected.
240 * systemd-bsod gained a new option --tty= to specify the output TTY
244 * /dev/ now contains symlinks that combine by-path and by-{label,uuid}
247 /dev/disk/by-path/<path>/by-<label|uuid|…>/<label|uuid|…>
249 This allows distinguishing partitions with identical contents on
250 multiple storage devices. This is useful, for example, when copying
251 raw disk contents between devices.
253 * systemd-udevd now creates persistent /dev/media/by-path/ symlinks for
254 media controllers. For example, the uvcvideo driver may create
255 /dev/media0 which will be linked as
256 /dev/media/by-path/pci-0000:04:00.3-usb-0:1:1.0-media-controller.
258 * A new unit systemd-udev-load-credentials.service has been added
259 to pick up udev.conf drop-ins and udev rules from credentials.
261 * An allowlist/denylist may be specified to filter which sysfs
262 attributes are used when crafting network interface names. Those
263 lists are stored as hwdb entries
264 ID_NET_NAME_ALLOW_<sysfsattr>=0|1
266 ID_NET_NAME_ALLOW=0|1.
268 The goal is to avoid unexpected changes to interface names when the
269 kernel is updated and new sysfs attributes become visible.
271 * A new unit tpm2.target has been added to provide a synchronization
272 point for units which expect the TPM hardware to be available. A new
273 generator "systemd-tpm2-generator" has been added that will insert
274 this target whenever it detects that the firmware has initialized a
275 TPM, but Linux hasn't loaded a driver for it yet.
277 * systemd-backlight now properly supports numbered devices which the
278 kernel creates to avoid collisions in the leds subsystem.
280 * systemd-hwdb update operation can be disabled with a new environment
281 variable SYSTEMD_HWDB_UPDATE_BYPASS=1.
285 * systemd-hostnamed now exposes the machine ID and boot ID via
286 D-Bus. It also exposes the hosts AF_VSOCK CID, if available.
288 * systemd-hostnamed now provides a basic Varlink interface.
290 * systemd-hostnamed exports the full data in os-release(5) and
291 machine-info(5) via D-Bus and Varlink.
293 * hostnamectl now shows the system's product UUID and hardware serial
298 * systemd-networkd now provides a basic Varlink interface.
300 * systemd-networkd's ARP proxy support gained a new option to configure
301 a private VLAN variant of the proxy ARP supported by the kernel under
302 the name IPv4ProxyARPPrivateVLAN=.
304 * systemd-networkd now exports the NamespaceId and NamespaceNSID
305 properties via D-Bus and Varlink. (which expose the inode and NSID of
306 the network namespace the networkd instance manages)
308 * systemd-networkd now supports IPv6RetransmissionTimeSec= and
309 UseRetransmissionTime= settings in .network files to configure
310 retransmission time for IPv6 neighbor solicitation messages.
312 * networkctl gained new verbs 'mask' and 'unmask' for masking networkd
313 configuration files such as .network files.
315 * 'networkctl edit --runtime' allows editing volatile configuration
316 under /run/systemd/network/.
318 * The implementation behind TTLPropagate= network setting has been
319 removed and the setting is now ignored.
321 * systemd-network-generator will now pick up .netdev/.link/.network/
322 networkd.conf configuration from system credentials.
324 * systemd-networkd will now pick up wireguard secrets from
327 * systemd-networkd's Varlink API now supports enumerating LLDP peers.
329 * .link files now support new Property=, ImportProperty=,
330 UnsetProperty= fields for setting udev properties on a link.
332 * The various .link files that systemd ships for interfaces that are
333 supposed to be managed by systemd-networkd only now carry a
334 ID_NET_MANAGED_BY=io.systemd.Network udev property ensuring that
335 other network management solutions honouring this udev property do
336 not come into conflict with networkd, trying to manage these
339 * .link files now support a new ReceivePacketSteeringCPUMask= setting
340 for configuring which CPUs to steer incoming packets to.
344 * systemd-nspawn now provides a /run/systemd/nspawn/unix-export/
345 directory where the container payload can expose AF_UNIX sockets to
346 allow them them to be accessed from outside.
348 * systemd-nspawn will tint the terminal background for containers in a
349 blueish color. This can be controller with the new --background=
352 * systemd-nspawn gained support for the 'owneridmap' option for --bind=
353 mounts to map the target directory owner from inside the container to
354 the owner of the directory bound from the host filesystem.
356 * systemd-nspawn now supports moving Wi-Fi network devices into a
357 container, just like other network interfaces.
361 * systemd-resolved now reads RFC 8914 EDE error codes provided by
362 upstream DNS services.
364 * systemd-resolved and resolvectl now support RFC 9460 SVCB and HTTPS
365 records, as well as RFC 2915 NAPTR records.
367 * resolvectl gained a new option --relax-single-label= to allow
368 querying single-label hostnames via unicast DNS on a per-query basis.
370 * systemd-resolved's Varlink IPC interface now supports resolving
371 DNS-SD services as well as an API for resolving raw DNS RRs.
373 * systemd-resolved's .dnssd DNS_SD service description files now
374 support DNS-SD "subtypes" via the new SubType= setting.
376 * systemd-resolved's configuration may now be reloaded without
377 restarting the service. (i.e. "systemctl reload systemd-resolved" is
382 * An sshd config drop-in to allow ssh keys acquired via userdbctl (for
383 example expose by homed accounts) to be used for authorization of
384 incoming SSH connections.
386 * A small new unit generator "systemd-ssh-generator" has been added. It
387 checks if the sshd binary is installed. If so, it binds it via
388 per-connection socket activation to various sockets depending on the
391 • If the system is run in a VM providing AF_VSOCK support, it
392 automatically binds sshd to AF_VSOCK port 22.
394 • If the system is invoked as a full-OS container and the container
395 manager pre-mounts a directory /run/host/unix-export/, it will
396 bind sshd to an AF_UNIX socket /run/host/unix-export/ssh. The
397 idea is the container manager bind mounts the directory to an
398 appropriate place on the host as well, so that the AF_UNIX socket
399 may be used to easily connect from the host to the container.
401 • sshd is also bound to an AF_UNIX socket
402 /run/ssh-unix-local/socket, which may be to use ssh/sftp in a
403 "sudo"-like fashion to access resources of other local users.
405 • Via the kernel command line option "systemd.ssh_listen=" and the
406 system credential "ssh.listen" sshd may be bound to additional,
407 explicitly configured options, including AF_INET/AF_INET6 ports.
409 In particular the first two mechanisms should make dealing with local
410 VMs and full OS containers a lot easier, as SSH connections will
411 *just* *work* from the host – even if no networking is available
414 systemd-ssh-generator optionally generates a per-connection
415 socket activation service file wrapping sshd. This is only done if
416 the distribution does not provide one on its own under the name
417 "sshd@.service". The generated unit only works correctly if the SSH
418 privilege separation ("privsep") directory exists. Unfortunately
419 distributions vary wildly where they place this directory. An
420 incomprehensive list:
422 • /usr/share/empty.sshd/ (new fedora)
425 • /run/sshd/ (debian/ubuntu?)
427 If the SSH privsep directory is placed below /var/ or /run/ care
428 needs to be taken that the directory is created automatically at boot
429 if needed, since these directories possibly or always come up
430 empty. This can be done via a tmpfiles.d/ drop-in. You may use the
431 "sshdprivsepdir" meson option provided by systemd to configure the
432 directory, in case you want systemd to create the directory as needed
433 automatically, if your distribution does not cover this natively.
435 Recommendations to distributions, in order to make things just work:
437 • Please provide a per-connection SSH service file under the name
440 • Please move the SSH privsep dir into /usr/ (so that it is truly
441 immutable on image-based operating systems, is strictly under
442 package manager control, and never requires recreation if the
443 system boots up with an empty /run/ or /var/).
445 • As an extension of this: please consider following Fedora's lead
446 here, and use /usr/share/empty.sshd/ to minimize needless
447 differences between distributions.
449 • If your distribution insists on placing the directory in /var/ or
450 /run/ then please at least provide a tmpfiles.d/ drop-in to
451 recreate it automatically at boot, so that the sshd binary just
452 works, regardless in which context it is called.
454 * A small tool "systemd-ssh-proxy" has been added, which is supposed to
455 act as counterpart to "systemd-ssh-generator". It's a small plug-in
456 for the SSH client (via ProxyCommand/ProxyUseFdpass) to allow it to
457 connect to AF_VSOCK or AF_UNIX sockets. Example: "ssh vsock/4711"
458 connects to a local VM with cid 4711, or "ssh
459 unix/run/ssh-unix-local/socket" to connect to the local host via the
460 AF_UNIX socket /run/ssh-unix-local/socket.
462 systemd-boot and systemd-stub and Related Tools:
464 * TPM 1.2 PCR measurement support has been removed from systemd-stub.
465 TPM 1.2 is obsolete and – due to the (by today's standards) weak
466 cryptographic algorithms it only supports – does not actually provide
467 the security benefits it's supposed to provide. Given that the rest
468 of systemd's codebase never supported TPM 1.2, the support has now
469 been removed from systemd-stub as well.
471 * systemd-stub will now measure its payload via the new EFI
472 Confidential Computing APIs (CC), in addition to the pre-existing
475 * confexts are loaded by systemd-stub from the ESP as well.
477 * kernel-install gained support for --root= for the 'list' verb.
479 * bootctl now provides a basic Varlink interface and can be run as a
480 daemon via a template unit.
482 * systemd-measure gained new options --certificate=, --private-key=,
483 and --private-key-source= to allow using OpenSSL's "engines" or
484 "providers" as the signing mechanism to use when creating signed
485 TPM2 PCR measurement values.
487 * ukify gained support for signing of PCR signatures via OpenSSL's
488 engines and providers.
490 * ukify now supports zboot kernels.
492 * systemd-boot now supports passing additional kernel command line
493 switches to invoked kernels via an SMBIOS Type #11 string
494 "io.systemd.boot.kernel-cmdline-extra". This is similar to the
495 pre-existing support for this in systemd-stub, but also applies to
496 Type #1 Boot Loader Specification Entries.
498 * systemd-boot's automatic SecureBoot enrollment support gained support
499 for enrolling "dbx" too (Previously, only db/KEK/PK enrollment was
500 supported). It also now supports UEFI "Custom" mode.
502 * The pcrlock policy is saved in an unencrypted credential file
503 "pcrlock.<entry-token>.cred" under XBOOTLDR/ESP in the
504 /loader/credentials/ directory. It will be picked up at boot by
505 systemd-stub and passed to the initrd, where it can be used to unlock
506 the root file system.
508 * systemd-pcrlock gained an --entry-token= option to configure the
511 * systemd-pcrlock now provides a basic Varlink interface and can be run
512 as a daemon via a template unit.
514 * systemd-pcrlock's TPM nvindex access policy has been modified, this
515 means that previous pcrlock policies stored in nvindexes are
516 invalidated. They must be removed (systemd-pcrlock remove-policy) and
517 recreated (systemd-pcrlock make-policy). For the time being
518 systemd-pcrlock remains an experimental feature, but it is expected
519 to become stable in the next release, i.e. v257.
521 * systemd-pcrlock's --recovery-pin= switch now takes three values:
522 "hide", "show", "query". If "show" is selected the automatically
523 generated recovery PIN is shown to the user. If "query" is selected
524 then the PIN is queried from the user.
528 * systemd-run is now a multi-call binary. When invoked as 'run0', it
529 provides as interface similar to 'sudo', with all arguments starting
530 at the first non-option parameter being treated the command to invoke
531 as root. Unlike 'sudo' and similar tools, it does not make use of
532 setuid binaries or other privilege escalation methods, but instead
533 runs the specified command as a transient unit, which is started by
534 the system service manager, so privileges are dropped, rather than
535 gained, thus implementing a much more robust and safe security
536 model. As usual, authorization is managed via Polkit.
538 * systemd-run/run0 will now tint the terminal background on supported
539 terminals: in a reddish tone when invoking a root service, in a
540 yellowish tone otherwise. This may be controlled and turned off via
541 the new --background= switch.
543 * systemd-run gained a new option '--ignore-failure' to suppress
548 * 'systemctl edit --stdin' allows creation of unit files and drop-ins
549 with contents supplied via standard input. This is useful when creating
550 configuration programmatically; the tool takes care of figuring out
551 the file name, creating any directories, and reloading the manager
554 * 'systemctl disable --now' and 'systemctl mask --now' now work
555 correctly with template units.
557 * 'systemd-analyze architectures' lists known CPU architectures.
559 * 'systemd-analyze --json=…' is supported for 'architectures',
560 'capability', 'exit-status'.
562 * 'systemd-tmpfiles --purge' will purge (remove) all files and
563 directories created via tmpfiles.d configuration.
565 * systemd-id128 gained new options --no-pager, --no-legend, and
568 * hostnamectl gained '-j' as shortcut for '--json=pretty' or
571 * loginctl now supports -j/--json=.
573 * resolvectl now supports -j/--json= for --type=.
575 * systemd-tmpfiles gained a new option --dry-run to print what would be
576 done without actually taking action.
578 * varlinkctl gained a new --collect switch to collect all responses of
579 a method call that supports multiple replies and turns it into a
582 * systemd-dissect gained a new --make-archive option to generate an
583 archive file (tar.gz and similar) from a disk image.
587 * systemd-vmspawn gained a new --firmware= option to configure or list
588 firmware definitions for Qemu, a new --tpm= option to enable or
589 disable the use of a software TPM, a new --linux= option to specify a
590 kernel binary for direct kernel boot, a new --initrd= option to
591 specify an initrd for direct kernel boot, a new -D/--directory option
592 to use a plain directory as the root file system, a new
593 --private-users option similar to the one in systemd-nspawn, new
594 options --bind= and --bind-ro= to bind part of the host's file system
595 hierarchy into the guest, a new --extra-drive= option to attach
596 additional storage, and -n/--network-tap/--network-user-mode to
597 configure networking.
599 * A new systemd-vmspawn@.service can be used to launch systemd-vmspawn
602 * systemd-vmspawn gained the new --console= and --background= switches
603 that control how to interact with the VM. As before, by default an
604 interactive terminal interface is provided, but now with a background
605 tinted with a greenish hue.
607 * systemd-vmspawn can now register its VMs with systemd-machined,
608 controlled via the --register= switch.
610 * machinectl's start command (and related) can now invoke images either
611 as containers via `systemd-nspawn` (switch is --runner=nspawn, the
612 default) or as VMs via `systemd-vmspawn` (switch is --runner=vmspawn,
615 * systemd-vmspawn now supports two switches --pass-ssh-key= and
616 --ssh-key-type= to optionally set up transient SSH keys to pass to the
617 invoked VMs in order to be able to SSH into them once booted.
619 * systemd-vmspawn will now enable various "HyperV enlightenments" and
620 the "VM Generation ID" on the VMs.
622 * A new environment variable $SYSTEMD_VMSPAWN_QEMU_EXTRA may carry
623 additional qemu command line options to pass to qemu.
627 * systemd-repart gained new options --generate-fstab= and
628 --generate-crypttab= to write out fstab and crypttab files matching the
629 generated partitions.
631 * systemd-repart gained a new option --private-key-source= to allow
632 using OpenSSL's "engines" or "providers" as the signing mechanism to
633 use when creating verity signature partitions.
635 * systemd-repart gained a new DefaultSubvolume= setting in repart.d/
636 drop-ins that allow configuring the default btrfs subvolume for newly
637 formatted btrfs file systems.
641 * libsystemd gained new call sd_bus_creds_new_from_pidfd() to get a
642 credentials object for a pidfd and sd_bus_creds_get_pidfd_dup() to
643 retrieve the pidfd from a credentials object.
645 * sd-bus' credentials logic will now also acquire peer's UNIX group
646 lists and peer's pidfd if supported and requested.
648 * RPM macro %_kernel_install_dir has been added with the path
649 to the directory for kernel-install plugins.
651 * The liblz4, libzstd, liblzma, libkmod, libgcrypt dependencies have
652 been changed from regular shared library dependencies into dlopen()
655 * The sd-journal API gained a new call
656 sd_journal_stream_fd_with_namespace() which is just like
657 sd_journal_stream_fd() but creates a log stream targeted at a
658 specific log namespace.
660 * The sd-id128 API gained a new API call
661 sd_id128_get_invocation_app_specific() for acquiring an app-specific
662 ID that is derived from the service invocation ID.
664 systemd-cryptsetup/systemd-cryptenroll:
666 * systemd-cryptenroll can now enroll directly with a PKCS11 public key
667 (instead of a certificate).
669 * systemd-cryptsetup/systemd-cryptenroll now may lock a disk against a
670 PKCS#11 provided EC key (before it only supported RSA).
672 * systemd-cryptsetup gained support for crypttab option
673 link-volume-key= to link the volume key into the kernel keyring when
674 the volume is opened.
676 * systemd-cryptenroll will no longer enable Dictionary Attack
677 Protection (i.e. turn on NO_DA) for TPM enrollments that do not
678 involve a PIN. DA should not be necessary in that case (since key
679 entropy is high enough to make this unnecessary), but risks
680 accidental lock-out in case of unexpected PCR changes.
682 * systemd-cryptenroll now supports enrolling a new slot while unlocking
683 the old slot via TPM2 (previously unlocking only worked via password
688 * The remaining documentation that was on
689 https://freedesktop.org/wiki/Software/systemd/ has been moved to
692 * A new text describing the VM integration interfaces of systemd has
695 https://systemd.io/VM_INTERFACE
697 * The sd_notify() man page has gained an example with C code that shows
698 how to implement the interface in C without involving libsystemd.
700 systemd-homed, systemd-logind, systemd-userdbd:
702 * systemd-homed now supports unlocking of home directories when logging
703 in via SSH. Previously home directories needed to be unlocked before
704 an SSH login is attempted.
706 * JSON User Records have been extended with a separate public storage
707 area called "User Record Blob Directories". This is intended to store
708 the user's background image, avatar picture, and other similar items
709 which are too large to fit into the User Record itself.
711 systemd-homed, userdbctl, and homectl gained support for blob
712 directories. homectl gained --avatar= and --login-background= to
713 control two specific items of the blob directories.
715 * A new "additionalLanguages" field has been added to JSON user records
716 (as supported by systemd-homed and systemd-userdbd), which is closely
717 related to the pre-existing "preferredLanguage", and allows
718 specifying multiple additional languages for the user account. It is
719 used to initialize the $LANGUAGES environment variable when used.
721 * A new pair of "preferredSessionType" and "preferredSessionLauncher"
722 fields have been added to JSON user records, that may be used to
723 control which kind of desktop session to preferable activate on
726 * homectl gained a new verb 'firstboot', and a new
727 systemd-homed-firstboot.service unit uses this verb to create users
728 in a first boot environment, either from system credentials or by
729 querying interactively.
731 * systemd-logind now supports a new "background-light" session class
732 which does not pull in the user@.service unit. This is intended in
733 particular for lighter weight per-user cron jobs which do require any
734 per-user service manager to be around.
736 * The per-user service manager will now be tracked as a distinct "manager"
737 session type among logind sessions of each user.
739 * homectl now supports an --offline mode, by which certain account
740 properties can be changed without unlocking the home directory.
742 * systemd-logind gained a new
743 org.freedesktop.login1.Manager.ListSessionsEx() method that provides
744 additional metadata compared to ListSessions(). loginctl makes use of
745 this to list additional fields in list-sessions.
747 * systemd-logind gained a new org.freedesktop.login1.Manager.Sleep()
748 method that automatically redirects to SuspendThenHibernate(),
749 Suspend(), HybridSleep(), or Hibernate(), depending on what is
750 supported and configured, a new configuration setting SleepOperation=,
751 and an accompanying helper method
752 org.freedesktop.login1.Manager.CanSleep() and property
753 org.freedesktop.login1.Manager.SleepOperation.
755 'systemctl sleep' calls the new method to automatically put the
756 machine to sleep in the most appropriate way.
758 Credential Management:
760 * systemd-creds now provides a Varlink IPC API for encrypting and
761 decrypting credentials.
763 * systemd-creds' "tpm2-absent" key selection has been renamed to
764 "null", since that's what it actually does: "encrypt" and "sign"
765 with a fixed null key. --with-key=null should only be used in very
766 specific cases, as it provides zero integrity or confidentiality
767 protections. (i.e. it's only safe to use as fallback in environments
768 lacking both a TPM and access to the root fs to use the host
769 encryption key, or when integrity is provided some other way.)
771 * systemd-creds gained a new switch --allow-null. If specified, the
772 "decrypt" verb will decode encrypted credentials that use the "null"
773 key (by default this is refused, since using the "null" key defeats
774 the authenticated encryption normally done).
778 * The sleep.conf configuration file gained a new MemorySleepMode=
779 setting for configuring the sleep mode in more detail.
781 * A tiny new service systemd-hibernate-clear.service has been added
782 which clears hibernation information from the HibernateLocation EFI
783 variable, in case the resume device is gone. Normally, this variable
784 is supposed to be cleaned up by the code that initiates the resume
785 from hibernation image. But when the device is missing and that code
786 doesn't run, this service will now do the necessary work, ensuring
787 that no outdated hibernation image information remains on subsequent
790 Unprivileged User Namespaces & Mounts:
792 * A small new service systemd-nsresourced.service has been added. It
793 provides a Varlink IPC API that assigns a free, transiently allocated
794 64K UID/GID range to an uninitialized user namespace a client
795 provides. It may be used to implement unprivileged container managers
796 and other programs that need dynamic user ID ranges. It also provides
797 interfaces to then delegate mount file descriptors, control groups
798 and network interfaces to user namespaces set up this way.
800 * A small new service systemd-mountfsd.service has been added. It
801 provides a Varlink IPC API for mounting DDI images, and returning a set
802 of mount file descriptors for it. If a user namespace fd is provided
803 as input, then the mounts are registered with the user namespace. To
804 ensure trust in the image it must provide Verity information (or
805 alternatively interactive polkit authentication is required).
807 * The systemd-dissect tool now can access DDIs fully unprivileged by
808 using systemd-nsresourced/systemd-mountfsd.
810 * If the service manager runs unprivileged (i.e. systemd --user) it now
811 supports RootImage= for accessing DDI images, also implemented via
812 the systemd-nsresourced/systemd-mountfsd.
814 * systemd-nspawn may now operate without privileges, if a suitable DDI
815 is provided via --image=, again implemented via
816 systemd-nsresourced/systemd-mountfsd.
820 * timedatectl and machinectl gained option '-P', an alias for
821 '--value --property=…'.
823 * Various tools that pretty-print config files will now highlight
824 configuration directives.
826 * varlinkctl gained support for the "ssh:" transport. This requires
827 OpenSSH 9.4 or newer.
829 * systemd-sysext gained support for enabling system extensions in
830 mutable fashion, where a writeable upperdir is stored under
831 /var/lib/extensions.mutable/, and a new --mutable= option to
832 configure this behaviour. An "ephemeral" mode is not also supported
833 where the mutable layer is configured to be a tmpfs that is
834 automatically released when the system extensions are reattached.
836 * Coredumps are now retained for two weeks by default (instead of three
839 * portablectl --copy= parameter gained a new 'mixed' argument, that will
840 result in resources owned by the OS (e.g.: portable profiles) to be linked
841 but resources owned by the portable image (e.g.: the unit files and the
842 images themselves) to be copied.
844 * systemd will now register MIME types for various of its file types
845 (e.g. journal files, DDIs, encrypted credentials …) via the XDG
846 shared-mime-info infrastructure. (Files of these types will thus be
847 recognized as their own thing in desktop file managers such as GNOME
850 * systemd-dissect will now show the detected sector size of a given DDI
851 in its default output.
853 * systemd-portabled now generates recognizable structured log messages
854 whenever a portable service is attached or detached.
856 * Verity signature checking in userspace (i.e. checking against
857 /etc/verity.d/ keys) when activating DDIs can now be turned on/off
858 via a kernel command line option systemd.allow_userspace_verity= and
859 an environment variable SYSTEMD_ALLOW_USERSPACE_VERITY=.
861 * ext4/xfs file system quota handling has been reworked, so that
862 quotacheck and quotaon are now invoked as per-file-system templated
863 services (as opposed to single system-wide singletons), similar in
864 style to the fsck, growfs, pcrfs logic. This means file systems with
865 quota enabled can now be reasonably enabled at runtime of the system,
868 * "systemd-analyze dot" will now also show BindsTo= dependencies.
870 * systemd-debug-generator gained the ability add in arbitrary units
871 based on them being passed in via system credentials.
873 * A new kernel command-line option systemd.default_debug_tty= can be
874 used to specify the TTY for the debug shell, independently of
875 enabling or disabling it.
877 Contributions from: A S Alam, AKHIL KUMAR, Abraham Samuel Adekunle,
878 Adrian Vovk, Adrian Wannenmacher, Alan Liang, Alberto Planas,
879 Alexander Zavyalov, Anders Jonsson, Andika Triwidada, Andres Beltran,
880 Andrew Sayers, Antonio Alvarez Feijoo, Arthur Zamarin, Artur Pak,
881 AtariDreams, Benjamin Franzke, Bernhard M. Wiedemann, Black-Hole1,
882 Bryan Jacobs, Burak Gerz, Carlos Garnacho, Chandra Pratap,
883 Chris Simons, Christian Wesselhoeft, Clayton Craft, Colin Geniet,
884 Colin Walters, Costa Tsaousis, Cristian Rodríguez, Daan De Meyer,
885 Damien Challet, Dan Streetman, David Tardon, David Venhoek,
886 Diego Viola, Dionna Amalie Glaze, Dmitry Konishchev,
887 Edson Juliano Drosdeck, Eisuke Kawashima, Eli Schwartz,
888 Emanuele Giuseppe Esposito, Eric Daigle, Evgeny Vereshchagin,
889 Felix Riemann, Fernando Fernandez Mancera, Florian Schmaus, Franck Bui,
890 Frantisek Sumsal, Friedrich Altheide, Gabríel Arthúr Pétursson,
891 Gaël Donval, Georges Basile Stavracas Neto, Gerd Hoffmann, GNOME
892 Foundation, Guilhem Lettron, Göran Uddeborg, Hans de Goede, Harald
893 Brinkmann, Heinrich Schuchardt, Henry Li, Holger Assmann, Ivan Kruglov,
894 Ivan Shapovalov, Jakub Sitnicki, James Muir, Jan Engelhardt, Jan Macku,
895 Jeff King, JmbFountain, Joakim Nohlgård, Julius Alexandre,
896 Jörg Behrmann, Keian, Kirk, Kristian Klausen, Krzesimir Nowak,
897 Lars Ellenberg, Lennart Poettering, Luca Boccassi, Ludwig Nussel,
898 Lukáš Nykrýn, Luxiter, Maanya Goenka, Mariano Giménez,
899 Markus Merklinger, Martin Ivicic, Martin Trigaux, Martin Wilck,
900 Matt Layher, Matt Muggeridge, Matteo Croce, Matthias Lisin,
901 Max Gautier, Max Staudt, Michael Biebl, Michal Koutný, Michal Sekletár,
902 Mike Gilbert, Mike Yuan, Mikko Ylinen, MkfsSion, MrSmör,
903 Nandakumar Raghavan, Nick Cao, Nick Rosbrook, Ole Peder Brandtzæg,
904 Ondrej Kozina, Oğuz Ersen, Pablo Méndez Hernández, Pierre GRASSER,
905 Piotr Drąg, QuonXF, Rafaël Kooi, Raito Bezarius, Reid Wahl,
906 Renjaya Raga Zenta, Richard Maw, Roland Hieber, Ronan Pigott, Rose,
907 Ross Burton, Sam Leonard, Samuel BF, Sergei Zhmylev, Sergey A, Shulhan,
908 SidhuRupinder, Sludge, Stuart Hayhurst, Susant Sahani,
909 Takashi Sakamoto, Temuri Doghonadze, Thilo Fromm, Thomas Blume,
910 TobiPeterG, Tomáš Pecka, Topi Miettinen, Tycho Andersen, Unique-Usman,
911 Usman Akinyemi, Vasiliy Kovalev, Vasiliy Stelmachenok,
912 Vishal Chillara Srinivas, Vitaly Kuznetsov, Vito Caputo,
913 Vladimir Stoiakin, Werner Sembach, Will Springer, Winterhuman,
914 Xiaotian Wu, Yu Watanabe, Yuri Chornoivan, Zbigniew Jędrzejewski-Szmek,
915 Zmyeir, aslepykh, chenjiayi, cpackham-atlnz, cunshunxia, djantti,
916 hfavisado, hulkoba, ksaleem, medusalix, mille-feuille, mkubiak, mooo,
917 msizanoen, networkException, nl6720, r-vdp, runiq, samuelvw01,
918 sharad3001, sushmbha, wangyuhang, zzywysm, İ. Ensar Gülşen,
919 Štěpán Němec, 我超厉害, 김인수
921 — Happy Place, 2024-04-XX
925 Announcements of Future Feature Removals and Incompatible Changes:
927 * Support for split-usr (/usr/ mounted separately during late boot,
928 instead of being mounted by the initrd before switching to the rootfs)
929 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
930 /usr/lib/, …) has been removed. For more details, see:
931 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
933 * We intend to remove cgroup v1 support from a systemd release after
934 the end of 2023. If you run services that make explicit use of
935 cgroup v1 features (i.e. the "legacy hierarchy" with separate
936 hierarchies for each controller), please implement compatibility with
937 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
938 Most of Linux userspace has been ported over already.
940 * Support for System V service scripts is now deprecated and will be
941 removed in a future release. Please make sure to update your software
942 *now* to include a native systemd unit file instead of a legacy
943 System V script to retain compatibility with future systemd releases.
945 * Support for the SystemdOptions EFI variable is deprecated.
946 'bootctl systemd-efi-options' will emit a warning when used. It seems
947 that this feature is little-used and it is better to use alternative
948 approaches like credentials and confexts. The plan is to drop support
949 altogether at a later point, but this might be revisited based on
952 * systemd-run's switch --expand-environment= which currently is disabled
953 by default when combined with --scope, will be changed in a future
954 release to be enabled by default.
956 * "systemctl switch-root" is now restricted to initrd transitions only.
958 Transitions between real systems should be done with
959 "systemctl soft-reboot" instead.
961 * The "ip=off" and "ip=none" kernel command line options interpreted by
962 systemd-network-generator will now result in IPv6RA + link-local
963 addressing being disabled, too. Previously DHCP was turned off, but
964 IPv6RA and IPv6 link-local addressing was left enabled.
966 * The NAMING_BRIDGE_MULTIFUNCTION_SLOT naming scheme has been deprecated
969 * SuspendMode=, HibernateState= and HybridSleepState= in the [Sleep]
970 section of systemd-sleep.conf are now deprecated and have no effect.
971 They did not (and could not) take any value other than the respective
972 default. HybridSleepMode= is also deprecated, and will now always use
973 the 'suspend' disk mode.
977 * The way services are spawned has been overhauled. Previously, a
978 process was forked that shared all of the manager's memory (via
979 copy-on-write) while doing all the required setup (e.g.: mount
980 namespaces, CGroup configuration, etc.) before exec'ing the target
981 executable. This was problematic for various reasons: several glibc
982 APIs were called that are not supposed to be used after a fork but
983 before an exec, copy-on-write meant that if either process (the
984 manager or the child) touched a memory page a copy was triggered, and
985 also the memory footprint of the child process was that of the
986 manager, but with the memory limits of the service. From this version
987 onward, the new process is spawned using CLONE_VM and CLONE_VFORK
988 semantics via posix_spawn(3), and it immediately execs a new internal
989 binary, systemd-executor, that receives the configuration to apply
990 via memfd, and sets up the process before exec'ing the target
991 executable. The systemd-executor binary is pinned by file descriptor
992 by each manager instance (system and users), and the reference is
993 updated on daemon-reexec - it is thus important to reexec all running
994 manager instances when the systemd-executor and/or libsystemd*
995 libraries are updated on the filesystem.
997 * Most of the internal process tracking is being changed to use PIDFDs
998 instead of PIDs when the kernel supports it, to improve robustness
1001 * A new option SurviveFinalKillSignal= can be used to configure the
1002 unit to be skipped in the final SIGTERM/SIGKILL spree on shutdown.
1003 This is part of the required configuration to let a unit's processes
1004 survive a soft-reboot operation.
1006 * System extension images (sysext) can now set
1007 EXTENSION_RELOAD_MANAGER=1 in their extension-release files to
1008 automatically reload the service manager (PID 1) when
1009 merging/refreshing/unmerging on boot. Generally, while this can be
1010 used to ship services in system extension images it's recommended to
1011 do that via portable services instead.
1013 * The ExtensionImages= and ExtensionDirectories= options now support
1014 confexts images/directories.
1016 * A new option NFTSet= provides a method for integrating dynamic cgroup
1017 IDs into firewall rules with NFT sets. The benefit of using this
1018 setting is to be able to use control group as a selector in firewall
1019 rules easily and this in turn allows more fine grained filtering.
1020 Also, NFT rules for cgroup matching use numeric cgroup IDs, which
1021 change every time a service is restarted, making them hard to use in
1022 systemd environment.
1024 * A new option CoredumpReceive= can be set for service and scope units,
1025 together with Delegate=yes, to make systemd-coredump on the host
1026 forward core files from processes crashing inside the delegated
1027 CGroup subtree to systemd-coredump running in the container. This new
1028 option is by default used by systemd-nspawn containers that use the
1031 * A new ConditionSecurity=measured-uki option is now available, to ensure
1032 a unit can only run when the system has been booted from a measured UKI.
1034 * MemoryAvailable= now considers physical memory if there are no CGroup
1035 memory limits set anywhere in the tree.
1037 * The $USER environment variable is now always set for services, while
1038 previously it was only set if User= was specified. A new option
1039 SetLoginEnvironment= is now supported to determine whether to also set
1040 $HOME, $LOGNAME, and $SHELL.
1042 * Socket units now support a new pair of
1043 PollLimitBurst=/PollLimitInterval= options to configure a limit on
1044 how often polling events on the file descriptors backing this unit
1045 will be considered within a time window.
1047 * Scope units can now be created using PIDFDs instead of PIDs to select
1048 the processes they should include.
1050 * Sending SIGRTMIN+18 with 0x500 as sigqueue() value will now cause the
1051 manager to dump the list of currently pending jobs.
1053 * If the kernel supports MOVE_MOUNT_BENEATH, the systemctl and
1054 machinectl bind and mount-image verbs will now cause the new mount to
1055 replace the old mount (if any), instead of overmounting it.
1057 * Units now have MemoryPeak, MemorySwapPeak, MemorySwapCurrent and
1058 MemoryZSwapCurrent properties, which respectively contain the values
1059 of the cgroup v2's memory.peak, memory.swap.peak, memory.swap.current
1060 and memory.zswap.current properties. This information is also shown in
1061 "systemctl status" output, if available.
1063 TPM2 Support + Disk Encryption & Authentication:
1065 * systemd-cryptenroll now allows specifying a PCR bank and explicit hash
1066 value in the --tpm2-pcrs= option.
1068 * systemd-cryptenroll now allows specifying a TPM2 key handle (nv
1069 index) to be used instead of the default SRK via the new
1070 --tpm2-seal-key-handle= option.
1072 * systemd-cryptenroll now allows TPM2 enrollment using only a TPM2
1073 public key (in TPM2B_PUBLIC format) – without access to the TPM2
1074 device itself – which enables offline sealing of LUKS images for a
1075 specific TPM2 chip, as long as the SRK public key is known. Pass the
1076 public to the tool via the new --tpm2-device-key= switch.
1078 * systemd-cryptsetup is now installed in /usr/bin/ and is no longer an
1079 internal-only executable.
1081 * The TPM2 Storage Root Key will now be set up, if not already present,
1082 by a new systemd-tpm2-setup.service early boot service. The SRK will
1083 be stored in PEM format and TPM2_PUBLIC format (the latter is useful
1084 for systemd-cryptenroll --tpm2-device-key=, as mentioned above) for
1085 easier access. A new "srk" verb has been added to systemd-analyze to
1086 allow extracting it on demand if it is already set up.
1088 * The internal systemd-pcrphase executable has been renamed to
1091 * The systemd-pcrextend tool gained a new --pcr= switch to override
1092 which PCR to measure into.
1094 * systemd-pcrextend now exposes a Varlink interface at
1095 io.systemd.PCRExtend that can be used to do measurements and event
1098 * TPM measurements are now also written to an event log at
1099 /run/log/systemd/tpm2-measure.log, using a derivative of the TCG
1100 Canonical Event Log format. Previously we'd only log them to the
1101 journal, where they however were subject to rotation and similar.
1103 * A new component "systemd-pcrlock" has been added that allows managing
1104 local TPM2 PCR policies for PCRs 0-7 and similar, which are hard to
1105 predict by the OS vendor because of the inherently local nature of
1106 what measurements they contain, such as firmware versions of the
1107 system and extension cards and suchlike. pcrlock can predict PCR
1108 measurements ahead of time based on various inputs, such as the local
1109 TPM2 event log, GPT partition tables, PE binaries, UKI kernels, and
1110 various other things. It can then pre-calculate a TPM2 policy from
1111 this, which it stores in an TPM2 NV index. TPM2 objects (such as disk
1112 encryption keys) can be locked against this NV index, so that they
1113 are locked against a specific combination of system firmware and
1114 state. Alternatives for each component are supported to allowlist
1115 multiple kernel versions or boot loader version simultaneously
1116 without losing access to the disk encryption keys. The tool can also
1117 be used to analyze and validate the local TPM2 event log.
1118 systemd-cryptsetup, systemd-cryptenroll, systemd-repart have all been
1119 updated to support such policies. There's currently no support for
1120 locking the system's root disk against a pcrlock policy, this will be
1121 added soon. Moreover, it is currently not possible to combine a
1122 pcrlock policy with a signed PCR policy. This component is
1123 experimental and its public interface is subject to change.
1125 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
1127 * bootctl will now show whether the system was booted from a UKI in its
1130 * systemd-boot and systemd-stub now use different project keys in their
1131 respective SBAT sections, so that they can be revoked individually if
1134 * systemd-boot will no longer load unverified Devicetree blobs when UEFI
1135 SecureBoot is enabled. For more details see:
1136 https://github.com/systemd/systemd/security/advisories/GHSA-6m6p-rjcq-334c
1138 * systemd-boot gained new hotkeys to reboot and power off the system
1139 from the boot menu ("B" and "O"). If the "auto-poweroff" and
1140 "auto-reboot" options in loader.conf are set these entries are also
1141 shown as menu items (which is useful on devices lacking a regular
1144 * systemd-boot gained a new configuration value "menu-disabled" for the
1145 set-timeout option, to allow completely disabling the boot menu,
1146 including the hotkey.
1148 * systemd-boot will now measure the content of loader.conf in TPM2
1151 * systemd-stub will now concatenate the content of all kernel
1152 command-line addons before measuring them in TPM2 PCR 12, in a single
1153 measurement, instead of measuring them individually.
1155 * systemd-stub will now measure and load Devicetree Blob addons, which
1156 are searched and loaded following the same model as the existing
1157 kernel command-line addons.
1159 * systemd-stub will now ignore unauthenticated kernel command line options
1160 passed from systemd-boot when running inside Confidential VMs with UEFI
1163 * systemd-stub will now load a Devicetree blob even if the firmware did
1164 not load any beforehand (e.g.: for ACPI systems).
1166 * ukify is no longer considered experimental, and now ships in /usr/bin/.
1168 * ukify gained a new verb inspect to describe the sections of a UKI and
1169 print the contents of the well-known sections.
1171 * ukify gained a new verb genkey to generate a set of key pairs for
1172 signing UKIs and their PCR data.
1174 * The 90-loaderentry kernel-install hook now supports installing device
1177 * kernel-install now supports the --json=, --root=, --image=, and
1178 --image-policy= options for the inspect verb.
1180 * kernel-install now supports new list and add-all verbs. The former
1181 lists all installed kernel images (if those are available in
1182 /usr/lib/modules/). The latter will install all the kernels it can
1187 * A new option --copy-from= has been added that synthesizes partition
1188 definitions from the given image, which are then applied by the
1189 systemd-repart algorithm.
1191 * A new option --copy-source= has been added, which can be used to specify
1192 a directory to which CopyFiles= is considered relative to.
1194 * New --make-ddi=confext, --make-ddi=sysext, and --make-ddi=portable
1195 options have been added to make it easier to generate these types of
1196 DDIs, without having to provide repart.d definitions for them.
1198 * The dm-verity salt and UUID will now be derived from the specified
1201 * New VerityDataBlockSizeBytes= and VerityHashBlockSizeBytes= can now be
1202 configured in repart.d/ configuration files.
1204 * A new Subvolumes= setting is now supported in repart.d/ configuration
1205 files, to indicate which directories in the target partition should be
1208 * A new --tpm2-device-key= option can be used to lock a disk against a
1209 specific TPM2 public key. This matches the same switch the
1210 systemd-cryptenroll tool now supports (see above).
1214 * The journalctl --lines= parameter now accepts +N to show the oldest N
1215 entries instead of the newest.
1217 * journald now ensures that sealing happens once per epoch, and sets a
1218 new compatibility flag to distinguish old journal files that were
1219 created before this change, for backward compatibility.
1223 * udev will now create symlinks to loopback block devices in the
1224 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
1225 string field selected during allocation. The systemd-dissect tool and
1226 the util-linux losetup command now supports a complementing new switch
1227 --loop-ref= for selecting the string. This means a loopback block
1228 device may now be allocated under a caller-chosen reference and can
1229 subsequently be referenced without first having to look up the block
1230 device name the caller ended up with.
1232 * udev also creates symlinks to loopback block devices in the
1233 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
1234 of the inode attached to the loopback block device. This means that
1235 attaching a file to a loopback device will implicitly make a handle
1236 available to be found via that file's inode information.
1238 * udevadm info gained support for JSON output via a new --json= flag, and
1239 for filtering output using the same mechanism that udevadm trigger
1242 * The predictable network interface naming logic is extended to include
1243 the SR-IOV-R "representor" information in network interface names.
1244 This feature was intended for v254, but even though the code was
1245 merged, the part that actually enabled the feature was forgotten.
1246 It is now enabled by default and is part of the new "v255" naming
1249 * A new hwdb/rules file has been added that sets the
1250 ID_NET_AUTO_LINK_LOCAL_ONLY=1 udev property on all network interfaces
1251 that should usually only be configured with link-local addressing
1252 (IPv4LL + IPv6LL), i.e. for PC-to-PC cables ("laplink") or
1253 Thunderbolt networking. systemd-networkd and NetworkManager (soon)
1254 will make use of this information to apply an appropriate network
1255 configuration by default.
1257 * The ID_NET_DRIVER property on network interfaces is now set
1258 relatively early in the udev rule set so that other rules may rely on
1259 its use. This is implemented in a new "net-driver" udev built-in.
1263 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
1264 is now dropped, as it never worked, hence it should not be used by
1267 * The 'prefixstable' ipv6 address generation mode now considers the SSID
1268 when generating stable addresses, so that a different stable address
1269 is used when roaming between wireless networks. If you already use
1270 'prefixstable' addresses with wireless networks, the stable address
1271 will be changed by the update.
1273 * The DHCPv4 client gained a RapidCommit option, true by default, which
1274 enables RFC4039 Rapid Commit behavior to obtain a lease in a
1275 simplified 2-message exchange instead of the typical 4-message
1276 exchange, if also supported by the DHCP server.
1278 * The DHCPv4 client gained new InitialCongestionWindow= and
1279 InitialAdvertisedReceiveWindow= options for route configurations.
1281 * The DHCPv4 client gained a new RequestAddress= option that allows
1282 to send a preferred IP address in the initial DHCPDISCOVER message.
1284 * The DHCPv4 server and client gained support for IPv6-only mode
1287 * The SendHostname= and Hostname= options are now available for the
1288 DHCPv6 client, independently of the DHCPv4= option, so that these
1289 configuration values can be set independently for each client.
1291 * The DHCPv4 and DHCPv6 client state can now be queried via D-Bus,
1292 including lease information.
1294 * The DHCPv6 client can now be configured to use a custom DUID type.
1296 * .network files gained a new IPv4ReversePathFilter= setting in the
1297 [Network] section, to control sysctl's rp_filter setting.
1299 * .network files gaiend a new HopLimit= setting in the [Route] section,
1300 to configure a per-route hop limit.
1302 * .network files gained a new TCPRetransmissionTimeoutSec= setting in
1303 the [Route] section, to configure a per-route TCP retransmission
1306 * A new directive NFTSet= provides a method for integrating network
1307 configuration into firewall rules with NFT sets. The benefit of using
1308 this setting is that static network configuration or dynamically
1309 obtained network addresses can be used in firewall rules with the
1310 indirection of NFT set types.
1312 * The [IPv6AcceptRA] section supports the following new options:
1313 UsePREF64=, UseHopLimit=, UseICMP6RateLimit=, and NFTSet=.
1315 * The [IPv6SendRA] section supports the following new options:
1316 RetransmitSec=, HopLimit=, HomeAgent=, HomeAgentLifetimeSec=, and
1317 HomeAgentPreference=.
1319 * A new [IPv6PREF64Prefix] set of options, containing Prefix= and
1320 LifetimeSec=, has been introduced to append pref64 options in router
1321 advertisements (RFC8781).
1323 * The network generator now configures the interfaces with only
1324 link-local addressing if "ip=link-local" is specified on the kernel
1327 * The prefix of the configuration files generated by the network
1328 generator from the kernel command line is now prefixed with '70-',
1329 to make them have higher precedence over the default configuration
1332 * Added a new -Ddefault-network=BOOL meson option, that causes more
1333 .network files to be installed as enabled by default. These configuration
1334 files will which match generic setups, e.g. 89-ethernet.network matches
1335 all Ethernet interfaces and enables both DHCPv4 and DHCPv6 clients.
1337 * If a ID_NET_MANAGED_BY= udev property is set on a network device and
1338 it is any other string than "io.systemd.Network" then networkd will
1339 not manage this device. This may be used to allow multiple network
1340 management services to run in parallel and assign ownership of
1341 specific devices explicitly. NetworkManager will soon implement a
1346 * systemctl is-failed now checks the system state if no unit is
1349 * systemctl will now automatically soft-reboot if a new root file system
1350 is found under /run/nextroot/ when a reboot operation is invoked.
1354 * Wall messages now work even when utmp support is disabled, using
1355 systemd-logind to query the necessary information.
1357 * systemd-logind now sends a new PrepareForShutdownWithMetadata D-Bus
1358 signal before shutdown/reboot/soft-reboot that includes additional
1359 information compared to the PrepareForShutdown signal. Currently the
1360 additional information is the type of operation that is about to be
1363 Hibernation & Suspend:
1365 * The kernel and OS versions will no longer be checked on resume from
1368 * Hibernation into swap files backed by btrfs are now
1369 supported. (Previously this was supported only for other file
1374 * A new systemd-vmspawn tool has been added, that aims to provide for VMs
1375 the same interfaces and functionality that systemd-nspawn provides for
1376 containers. For now it supports QEMU as a backend, and exposes some of
1377 its options to the user. This component is experimental and its public
1378 interface is subject to change.
1380 * "systemd-analyze plot" has gained tooltips on each unit name with
1381 related-unit information in its svg output, such as Before=,
1382 Requires=, and similar properties.
1384 * A new varlinkctl tool has been added to allow interfacing with
1385 Varlink services, and introspection has been added to all such
1386 services. This component is experimental and its public interface is
1389 * systemd-sysext and systemd-confext now expose a Varlink service
1390 at io.systemd.sysext.
1392 * portable services now accept confexts as extensions.
1394 * systemd-sysupdate now accepts directories in the MatchPattern= option.
1396 * systemd-run will now output the invocation ID of the launched
1397 transient unit and its peak memory usage.
1399 * systemd-analyze, systemd-tmpfiles, systemd-sysusers, systemd-sysctl,
1400 and systemd-binfmt gained a new --tldr option that can be used instead
1401 of --cat-config to suppress uninteresting configuration lines, such as
1402 comments and whitespace.
1404 * resolvectl gained a new "show-server-state" command that shows
1405 current statistics of the resolver. This is backed by a new
1406 DumpStatistics() Varlink method provided by systemd-resolved.
1408 * systemd-timesyncd will now emit a D-Bus signal when the LinkNTPServers
1411 * vconsole now supports KEYMAP=@kernel for preserving the kernel keymap
1414 * seccomp now supports the LoongArch64 architecture.
1416 * seccomp may now be enabled for services running as a non-root User=
1417 without NoNewPrivileges=yes.
1419 * systemd-id128 now supports a new -P option to show only values. The
1420 combination of -P and --app options is also supported.
1422 * A new pam_systemd_loadkey.so PAM module is now available, which will
1423 automatically fetch the passphrase used by cryptsetup to unlock the
1424 root file system and set it as the PAM authtok. This enables, among
1425 other things, configuring auto-unlock of the GNOME Keyring / KDE
1426 Wallet when autologin is configured.
1428 * Many meson options now use the 'feature' type, which means they
1429 take enabled/disabled/auto as values.
1431 * A new meson option -Dconfigfiledir= can be used to change where
1432 configuration files with default values are installed to.
1434 * Options and verbs in man pages are now tagged with the version they
1435 were first introduced in.
1437 * A new component "systemd-storagetm" has been added, which exposes all
1438 local block devices as NVMe-TCP devices, fully automatically. It's
1439 hooked into a new target unit storage-target-mode.target that is
1440 suppsoed to be booted into via
1441 rd.systemd.unit=storage-target-mode.target on the kernel command
1442 line. This is intended to be used for installers and debugging to
1443 quickly get access to the local disk. It's inspired by MacOS "target
1444 disk mode". This component is experimental and its public interface is
1447 * A new component "systemd-bsod" has been added, which can show logged
1448 error messages full screen, if they have a log level of LOG_EMERG log
1449 level. This component is experimental and its public interface is
1452 * The systemd-dissect tool's --with command will now set the
1453 $SYSTEMD_DISSECT_DEVICE environment variable to the block device it
1454 operates on for the invoked process.
1456 * The systemd-mount tool gained a new --tmpfs switch for mounting a new
1457 'tmpfs' instance. This is useful since it does so via .mount units
1458 and thus can be executed remotely or in containers.
1460 * The various tools in systemd that take "verbs" (such as systemctl,
1461 loginctl, machinectl, …) now will suggest a close verb name in case
1462 the user specified an unrecognized one.
1464 * libsystemd now exports a new function sd_id128_get_app_specific()
1465 that generates "app-specific" 128bit IDs from any ID. It's similar to
1466 sd_id128_get_machine_app_specific() and
1467 sd_id128_get_boot_app_specific() but takes the ID to base calculation
1468 on as input. This new functionality is also exposed in the
1469 "systemd-id128" tool where you can now combine --app= with `show`.
1471 * All tools that parse timestamps now can also parse RFC3339 style
1472 timestamps that include the "T" and Z" characters.
1474 * New documentation has been added:
1476 https://systemd.io/FILE_DESCRIPTOR_STORE
1477 https://systemd.io/TPM2_PCR_MEASUREMENTS
1478 https://systemd.io/MOUNT_REQUIREMENTS
1480 * The codebase now recognizes the suffix .confext.raw and .sysext.raw
1481 as alternative to the .raw suffix generally accepted for DDIs. It is
1482 recommended to name configuration extensions and system extensions
1483 with such suffixes, to indicate their purpose in the name.
1485 * The sd-device API gained a new function
1486 sd_device_enumerator_add_match_property_required() which allows
1487 configuring matches on properties that are strictly required. This is
1488 different from the existing sd_device_enumerator_add_match_property()
1489 matches of which one one needs to apply.
1491 * The MAC address the veth side of an nspawn container shall get
1492 assigned may now be controlled via the $SYSTEMD_NSPAWN_NETWORK_MAC
1493 environment variable.
1495 * The libiptc dependency is now implemented via dlopen(), so that tools
1496 such as networkd and nspawn no longer have a hard dependency on the
1497 shared library when compiled with support for libiptc.
1499 * New rpm macros have been added: %systemd_user_daemon_reexec does
1500 daemon-reexec for all user managers, and %systemd_postun_with_reload
1501 and %systemd_user_postun_with_reload do a reload for system and user
1504 * coredumpctl now propagates SIGTERM to the debugger process.
1506 Contributions from: 김인수, Abderrahim Kitouni, Adam Goldman,
1507 Adam Williamson, Alexandre Peixoto Ferreira, Alex Hudspith,
1508 Alvin Alvarado, André Paiusco, Antonio Alvarez Feijoo,
1509 Anton Lundin, Arian van Putten, Arseny Maslennikov, Arthur Shau,
1510 Balázs Úr, beh_10257, Benjamin Peterson, Bertrand Jacquin,
1511 Brian Norris, Charles Lee, Cheng-Chia Tseng, Chris Patterson,
1512 Christian Hergert, Christian Hesse, Christian Kirbach,
1513 Clayton Craft, commondservice, cunshunxia, Curtis Klein, cvlc12,
1514 Daan De Meyer, Daniele Medri, Daniel P. Berrangé, Daniel Rusek,
1515 Daniel Thompson, Dan Nicholson, Dan Streetman, David Rheinsberg,
1516 David Santamaría Rogado, David Tardon, dependabot[bot],
1517 Diego Viola, Dmitry V. Levin, Emanuele Giuseppe Esposito,
1518 Emil Renner Berthing, Emil Velikov, Etienne Dechamps, Fabian Vogt,
1519 felixdoerre, Felix Dörre, Florian Schmaus, Franck Bui,
1520 Frantisek Sumsal, G2-Games, Gioele Barabucci, Hugo Carvalho,
1521 huyubiao, Iago López Galeiras, IllusionMan1212, Jade Lovelace,
1522 janana, Jan Janssen, Jan Kuparinen, Jan Macku, Jeremy Fleischman,
1523 Jin Liu, jjimbo137, Joerg Behrmann, Johannes Segitz, Jordan Rome,
1524 Jordan Williams, Julien Malka, Juno Computers, Khem Raj, khm,
1525 Kingbom Dou, Kiran Vemula, Krzesimir Nowak, Laszlo Gombos,
1526 Lennart Poettering, linuxlion, Luca Boccassi, Lucas Adriano Salles,
1527 Lukas, Lukáš Nykrýn, Maanya Goenka, Maarten, Malte Poll,
1528 Marc Pervaz Boocha, Martin Beneš, Martin Joerg, Martin Wilck,
1529 Mathieu Tortuyaux, Matthias Schiffer, Maxim Mikityanskiy,
1530 Max Kellermann, Michael A Cassaniti, Michael Biebl, Michael Kuhn,
1531 Michael Vasseur, Michal Koutný, Michal Sekletár, Mike Yuan,
1532 Milton D. Miller II, mordner, msizanoen, NAHO, Nandakumar Raghavan,
1533 Neil Wilson, Nick Rosbrook, Nils K, NRK, Oğuz Ersen,
1534 Omojola Joshua, onenowy, Paul Meyer, Paymon MARANDI, pelaufer,
1535 Peter Hutterer, PhylLu, Pierre GRASSER, Piotr Drąg, Priit Laes,
1536 Rahil Bhimjiani, Raito Bezarius, Raul Cheleguini, Reto Schneider,
1537 Richard Maw, Robby Red, RoepLuke, Roland Hieber, Roland Singer,
1538 Ronan Pigott, Sam James, Sam Leonard, Sergey A, Susant Sahani,
1539 Sven Joachim, Tad Fisher, Takashi Sakamoto, Thorsten Kukuk, Tj,
1540 Tomasz Świątek, Topi Miettinen, Valentin David,
1541 Valentin Lefebvre, Victor Westerhuis, Vincent Haupert,
1542 Vishal Chillara Srinivas, Vito Caputo, Warren, Weblate,
1543 Xiaotian Wu, xinpeng wang, Yaron Shahrabani, Yo-Jung Lin,
1544 Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zeroskyx,
1545 Дамјан Георгиевски, наб
1547 — Edinburgh, 2023-12-06
1551 Announcements of Future Feature Removals and Incompatible Changes:
1553 * The next release (v255) will remove support for split-usr (/usr/
1554 mounted separately during late boot, instead of being mounted by the
1555 initrd before switching to the rootfs) and unmerged-usr (parallel
1556 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
1558 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1560 * We intend to remove cgroup v1 support from a systemd release after
1561 the end of 2023. If you run services that make explicit use of
1562 cgroup v1 features (i.e. the "legacy hierarchy" with separate
1563 hierarchies for each controller), please implement compatibility with
1564 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
1565 Most of Linux userspace has been ported over already.
1567 * Support for System V service scripts is now deprecated and will be
1568 removed in a future release. Please make sure to update your software
1569 *now* to include a native systemd unit file instead of a legacy
1570 System V script to retain compatibility with future systemd releases.
1572 * Support for the SystemdOptions EFI variable is deprecated.
1573 'bootctl systemd-efi-options' will emit a warning when used. It seems
1574 that this feature is little-used and it is better to use alternative
1575 approaches like credentials and confexts. The plan is to drop support
1576 altogether at a later point, but this might be revisited based on
1579 * EnvironmentFile= now treats the line following a comment line
1580 trailing with escape as a non comment line. For details, see:
1581 https://github.com/systemd/systemd/issues/27975
1583 * PrivateNetwork=yes and NetworkNamespacePath= now imply
1584 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
1586 * Behaviour of sandboxing options for the per-user service manager
1587 units has changed. They now imply PrivateUsers=yes, which means user
1588 namespaces will be implicitly enabled when a sandboxing option is
1589 enabled in a user unit. Enabling user namespaces has the drawback
1590 that system users will no longer be visible (and processes/files will
1591 appear as owned by 'nobody') in the user unit.
1593 By definition a sandboxed user unit should run with reduced
1594 privileges, so impact should be small. This will remove a great
1595 source of confusion that has been reported by users over the years,
1596 due to how these options require an extra setting to be manually
1597 enabled when used in the per-user service manager, which is not
1598 needed in the system service manager. For more details, see:
1599 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
1601 * systemd-run's switch --expand-environment= which currently is disabled
1602 by default when combined with --scope, will be changed in a future
1603 release to be enabled by default.
1605 Security Relevant Changes:
1607 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
1608 process capability to invoked session processes of regular users on
1609 local seats (as well as to systemd --user), unless configured
1610 otherwise via data from JSON user records, or via the PAM module's
1611 parameter list. This is useful in order allow desktop tools such as
1612 GNOME's Alarm Clock application to set a timer for
1613 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
1614 per-user service unit file may thus use AmbientCapability= to pass
1615 the capability to invoked processes. Note that this capability is
1616 relatively narrow in focus (in particular compared to other process
1617 capabilities such as CAP_SYS_ADMIN) and we already — by default —
1618 permit more impactful operations such as system suspend to local
1623 * Memory limits that apply while the unit is activating are now
1624 supported. Previously IO and CPU settings were already supported via
1625 StartupCPUWeight= and similar. The same logic has been added for the
1626 various manager and unit memory settings (DefaultStartupMemoryLow=,
1627 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
1628 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
1630 * The service manager gained support for enqueuing POSIX signals to
1631 services that carry an additional integer value, exposing the
1632 sigqueue() system call. This is accessible via new D-Bus calls
1633 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
1634 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
1635 via the new --kill-value= option.
1637 * systemctl gained a new "list-paths" verb, which shows all currently
1638 active .path units, similarly to how "systemctl list-timers" shows
1639 active timers, and "systemctl list-sockets" shows active sockets.
1641 * systemctl gained a new --when= switch which is honoured by the various
1642 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
1643 scheduling these operations by time, similar in fashion to how this
1644 has been supported by SysV shutdown.
1646 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
1647 supports the new PR_SET_MDWE prctl() call, it is used instead of the
1648 seccomp()-based system call filter to achieve the same effect.
1650 * A new set of kernel command line options is now understood:
1651 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
1652 systemd.tty.columns.<name>= allow configuring the TTY type and
1653 dimensions for the tty specified via <name>. When systemd invokes a
1654 service on a tty (via TTYName=) it will look for these and configure
1655 the TTY accordingly. This is particularly useful in VM environments
1656 to propagate host terminal settings into the appropriate TTYs of the
1659 * A new RootEphemeral= setting is now understood in service units. It
1660 takes a boolean argument. If enabled for services that use RootImage=
1661 or RootDirectory= an ephemeral copy of the disk image or directory
1662 tree is made when the service is started. It is removed automatically
1663 when the service is stopped. That ephemeral copy is made using
1664 btrfs/xfs reflinks or btrfs snapshots, if available.
1666 * The service activation logic gained new settings RestartSteps= and
1667 RestartMaxDelaySec= which allow exponentially-growing restart
1668 intervals for Restart=.
1670 * The service activation logic gained a new setting RestartMode= which
1671 can be set to 'direct' to skip the inactive/failed states when
1672 restarting, so that dependent units are not notified until the service
1673 converges to a final (successful or failed) state. For example, this
1674 means that OnSuccess=/OnFailure= units will not be triggered until the
1675 service state has converged.
1677 * PID 1 will now automatically load the virtio_console kernel module
1678 during early initialization if running in a suitable VM. This is done
1679 so that early-boot logging can be written to the console if available.
1681 * Similarly, virtio-vsock support is loaded early in suitable VM
1682 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
1683 to the VMM if configured, thus loading this early is beneficial.
1685 * A new verb "fdstore" has been added to systemd-analyze to show the
1686 current contents of the file descriptor store of a unit. This is
1687 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
1688 the service manager.
1690 * The service manager will now set a new $FDSTORE environment variable
1691 when invoking processes for services that have the file descriptor
1694 * A new service option FileDescriptorStorePreserve= has been added that
1695 allows tuning the lifecycle of the per-service file descriptor store.
1696 If set to "yes", the entries in the fd store are retained even after
1697 the service has been fully stopped.
1699 * The "systemctl clean" command may now be used to clear the fdstore of
1702 * Unit *.preset files gained a new directive "ignore", in addition to
1703 the existing "enable" and "disable". As the name suggests, matching
1704 units are left unchanged, i.e. neither enabled nor disabled.
1706 * Service units gained a new setting DelegateSubgroup=. It takes the
1707 name of a sub-cgroup to place any processes the service manager forks
1708 off in. Previously, the service manager would place all service
1709 processes directly in the top-level cgroup it created for the
1710 service. This usually meant that main process in a service with
1711 delegation enabled would first have to create a subgroup and move
1712 itself down into it, in order to not conflict with the "no processes
1713 in inner cgroups" rule of cgroup v2. With this option, this step is
1714 now handled by PID 1.
1716 * The service manager will now look for .upholds/ directories,
1717 similarly to the existing support for .wants/ and .requires/
1718 directories. Symlinks in this directory result in Upholds=
1721 The [Install] section of unit files gained support for a new
1722 UpheldBy= directive to generate .upholds/ symlinks automatically when
1725 * The service manager now supports a new kernel command line option
1726 systemd.default_device_timeout_sec=, which may be used to override
1727 the default timeout for .device units.
1729 * A new "soft-reboot" mechanism has been added to the service manager.
1730 A "soft reboot" is similar to a regular reboot, except that it
1731 affects userspace only: the service manager shuts down any running
1732 services and other units, then optionally switches into a new root
1733 file system (mounted to /run/nextroot/), and then passes control to a
1734 systemd instance in the new file system which then starts the system
1735 up again. The kernel is not rebooted and neither is the hardware,
1736 firmware or boot loader. This provides a fast, lightweight mechanism
1737 to quickly reset or update userspace, without the latency that a full
1738 system reset involves. Moreover, open file descriptors may be passed
1739 across the soft reboot into the new system where they will be passed
1740 back to the originating services. This allows pinning resources
1741 across the reboot, thus minimizing grey-out time further. This new
1742 reboot mechanism is accessible via the new "systemctl soft-reboot"
1745 * Services using RootDirectory= or RootImage= will now have read-only
1746 access to a copy of the host's os-release file under
1747 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
1748 This was already the case for Portable Services, and the feature has
1749 now been extended to all services that do not run off the host's
1752 * A new service setting MemoryKSM= has been added to enable kernel
1753 same-page merging individually for services.
1755 * A new service setting ImportCredentials= has been added that augments
1756 LoadCredential= and LoadCredentialEncrypted= and searches for
1757 credentials to import from the system, and supports globbing.
1759 * A new job mode "restart-dependencies" has been added to the service
1760 manager (exposed via systemctl --job-mode=). It is only valid when
1761 used with "start" jobs, and has the effect that the "start" job will
1762 be propagated as "restart" jobs to currently running units that have
1763 a BindsTo= or Requires= dependency on the started unit.
1765 * A new verb "whoami" has been added to "systemctl" which determines as
1766 part of which unit the command is being invoked. It writes the unit
1767 name to standard output. If one or more PIDs are specified reports
1768 the unit names the processes referenced by the PIDs belong to.
1770 * The system and service credential logic has been improved: there's
1771 now a clearly defined place where system provisioning tools running
1772 in the initrd can place credentials that will be imported into the
1773 system's set of credentials during the initrd → host transition: the
1774 /run/credentials/@initrd/ directory. Once the credentials placed
1775 there are imported into the system credential set they are deleted
1776 from this directory, and the directory itself is deleted afterwards
1779 * A new kernel command line option systemd.set_credential_binary= has
1780 been added, that is similar to the pre-existing
1781 systemd.set_credential= but accepts arbitrary binary credential data,
1782 encoded in Base64. Note that the kernel command line is not a
1783 recommend way to transfer credentials into a system, since it is
1784 world-readable from userspace.
1786 * The default machine ID to use may now be configured via the
1787 system.machine_id system credential. It will only be used if no
1788 machine ID was set yet on the host.
1790 * On Linux kernel 6.4 and newer system and service credentials will now
1791 be placed in a tmpfs instance that has the "noswap" mount option
1792 set. Previously, a "ramfs" instance was used. By switching to tmpfs
1793 ACL support and overall size limits can now be enforced, without
1794 compromising on security, as the memory is never paged out either
1797 * The service manager now can detect when it is running in a
1798 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
1799 accepted by ConditionSecurity= for units that want to conditionalize
1800 themselves on this. systemd-detect-virt gained new 'cvm' and
1801 '--list-cvm' switches to respectively perform the detection or list
1802 all known flavours of confidential VM, depending on the vendor. The
1803 manager will publish a 'ConfidentialVirtualization' D-Bus property,
1804 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
1805 variable for unit generators. Finally, udev rules can match on a new
1806 'cvm' key that will be set when in a confidential VM.
1807 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
1808 strings and QEMU's fw_cfg protocol will not be used to import
1809 credentials and kernel command line parameters by the system manager,
1810 systemd-boot and systemd-stub, because the hypervisor is considered
1811 untrusted in this particular setting.
1815 * The sd-journal API gained a new call sd_journal_get_seqnum() to
1816 retrieve the current log record's sequence number and sequence number
1817 ID, which allows applications to order records the same way as
1818 journal does internally. The sequence number is now also exported in
1819 the JSON and "export" output of the journal.
1821 * journalctl gained a new switch --truncate-newline. If specified
1822 multi-line log records will be truncated at the first newline,
1823 i.e. only the first line of each log message will be shown.
1825 * systemd-journal-upload gained support for --namespace=, similar to
1826 the switch of the same name of journalctl.
1830 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
1831 may be used to exclude certain files from the effect of CopyFiles=.
1833 * systemd-repart's Verity support now implements the Minimize= setting
1834 to minimize the size of the resulting partition.
1836 * systemd-repart gained a new --offline= switch, which may be used to
1837 control whether images shall be built "online" or "offline",
1838 i.e. whether to make use of kernel facilities such as loopback block
1839 devices and device mapper or not.
1841 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
1842 partition with some files, it will now default to VFAT rather than
1845 * systemd-repart gained a new --architecture= switch. If specified, the
1846 per-architecture GPT partition types (i.e. the root and /usr/
1847 partitions) configured in the partition drop-in files are
1848 automatically adjusted to match the specified CPU architecture, in
1849 order to simplify cross-architecture DDI building.
1851 * systemd-repart will now default to a minimum size of 300MB for XFS
1852 filesystems if no size parameter is specified. This matches what the
1853 XFS tools (xfsprogs) can support.
1855 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
1857 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
1858 Instead, pyelftools is now needed, and it will be used to perform the
1859 ELF -> PE relocations at build time.
1861 * bootctl gained a new switch --print-root-device/-R that prints the
1862 block device the root file system is backed by. If specified twice,
1863 it returns the whole disk block device (as opposed to partition block
1864 device) the root file system is on. It's useful for invocations such
1865 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
1868 * systemd-stub will now look for the SMBIOS Type 1 field
1869 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
1870 kernel command line it invokes. This is useful for VMMs such as qemu
1871 to pass additional kernel command lines into the system even when
1872 booting via full UEFI. The contents of the field are measured into
1875 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
1876 value "auto". With this value, a kernel will be automatically
1877 analyzed, and if it qualifies as UKI, it will be installed as if the
1878 setting was to set to "uki", otherwise as "bls".
1880 * systemd-stub can now optionally load UEFI PE "add-on" images that may
1881 contain additional kernel command line information. These "add-ons"
1882 superficially look like a regular UEFI executable, and are expected
1883 to be signed via SecureBoot/shim. However, they do not actually
1884 contain code, but instead a subset of the PE sections that UKIs
1885 support. They are supposed to provide a way to extend UKIs with
1886 additional resources in a secure and authenticated way. Currently,
1887 only the .cmdline PE section may be used in add-ons, in which case
1888 any specified string is appended to the command line embedded into
1889 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
1890 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
1891 the future we expect other sections to be made extensible like this as
1894 * ukify has been updated to allow building these UEFI PE "add-on"
1895 images, using the new 'addon<EFI-ARCH>.efi.stub'.
1897 * ukify now accepts SBAT information to place in the .sbat PE section
1898 of UKIs and addons. If a UKI is built the SBAT information from the
1899 inner kernel is merged with any SBAT information associated with
1900 systemd-stub and the SBAT data specified on the ukify command line.
1902 * The kernel-install script has been rewritten in C, and reuses much of
1903 the infrastructure of existing tools such as bootctl. It also gained
1904 --esp-path= and --boot-path= options to override the path to the ESP,
1905 and the $BOOT partition. Options --make-entry-directory= and
1906 --entry-token= have been added as well, similar to bootctl's options
1909 * A new kernel-install plugin 60-ukify has been added which will
1910 combine kernel/initrd locally into a UKI and optionally sign them
1911 with a local key. This may be used to switch to UKI mode even on
1912 systems where a local kernel or initrd is used. (Typically UKIs are
1913 built and signed by the vendor.)
1915 * The ukify tool now supports "pesign" in addition to the pre-existing
1916 "sbsign" for signing UKIs.
1918 * systemd-measure and systemd-stub now look for the .uname PE section
1919 that should contain the kernel's "uname -r" string.
1921 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
1922 "offline", i.e. without access to a TPM (physical or
1925 Memory Pressure & Control:
1927 * The sd-event API gained new calls sd_event_add_memory_pressure(),
1928 sd_event_source_set_memory_pressure_type(),
1929 sd_event_source_set_memory_pressure_period() to create and configure
1930 an event source that is called whenever the OS signals memory
1931 pressure. Another call sd_event_trim_memory() is provided that
1932 compacts the process' memory use by releasing allocated but unused
1933 malloc() memory back to the kernel. Services can also provide their
1934 own custom callback to do memory trimming. This should improve system
1935 behaviour under memory pressure, as on Linux traditionally provided
1936 no mechanism to return process memory back to the kernel if the
1937 kernel was under memory pressure. This makes use of the kernel's PSI
1938 interface. Most long-running services in systemd have been hooked up
1939 with this, and in particular systems with low memory should benefit
1942 * Service units gained new settings MemoryPressureWatch= and
1943 MemoryPressureThresholdSec= to configure the PSI memory pressure
1944 logic individually. If these options are used, the
1945 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
1946 variables will be set for the invoked processes to inform them about
1947 the requested memory pressure behaviour. (This is used by the
1948 aforementioned sd-events API additions, if set.)
1950 * systemd-analyze gained a new "malloc" verb that shows the output
1951 generated by glibc's malloc_info() on services that support it. Right
1952 now, only the service manager has been updated accordingly. This
1953 call requires privileges.
1955 User & Session Management:
1957 * The sd-login API gained a new call sd_session_get_username() to
1958 return the user name of the owner of a login session. It also gained
1959 a new call sd_session_get_start_time() to retrieve the time the login
1960 session started. A new call sd_session_get_leader() has been added to
1961 return the PID of the "leader" process of a session. A new call
1962 sd_uid_get_login_time() returns the time since the specified user has
1963 most recently been continuously logged in with at least one session.
1965 * JSON user records gained a new set of fields capabilityAmbientSet and
1966 capabilityBoundingSet which contain a list of POSIX capabilities to
1967 set for the logged in users in the ambient and bounding sets,
1968 respectively. homectl gained the ability to configure these two sets
1969 for users via --capability-bounding-set=/--capability-ambient-set=.
1971 * pam_systemd learnt two new module options
1972 default-capability-bounding-set= and default-capability-ambient-set=,
1973 which configure the default bounding sets for users as they are
1974 logging in, if the JSON user record doesn't specify this explicitly
1975 (see above). The built-in default for the ambient set now contains
1976 the CAP_WAKE_ALARM, thus allowing regular users who may log in
1977 locally to resume from a system suspend via a timer.
1979 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
1980 call to update the TTY of a session after it has been allocated. This
1981 is useful for SSH sessions which are typically allocated first, and
1982 for which a TTY is added later.
1984 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
1985 combines the various other sd_pid_notify() flavours into one: takes a
1986 format string, an overriding PID, and a set of file descriptors to
1987 send. It also gained a new call sd_pid_notify_barrier() call which is
1988 equivalent to sd_notify_barrier() but allows the originating PID to
1991 * "loginctl list-users" and "loginctl list-sessions" will now show the
1992 state of each logged in user/session in their tabular output. It will
1993 also show the current idle state of sessions.
1997 * systemd-dissect will now show the intended CPU architecture of an
2000 * systemd-dissect will now install itself as mount helper for the "ddi"
2001 pseudo-file system type. This means you may now mount DDIs directly
2002 via /bin/mount or /etc/fstab, making full use of embedded Verity
2003 information and all other DDI features.
2005 Example: mount -t ddi myimage.raw /some/where
2007 * The systemd-dissect tool gained the new switches --attach/--detach to
2008 attach/detach a DDI to a loopback block device without mounting it.
2009 It will automatically derive the right sector size from the image
2010 and set up Verity and similar, but not mount the file systems in it.
2012 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
2013 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
2014 implied. Given that these file systems are typically untrusted, this
2015 should make mounting them automatically have less of a security
2018 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
2019 systemd-tmpfiles, …) now understand a new switch --image-policy= which
2020 takes a string encoding image dissection policy. With this mechanism
2021 automatic discovery and use of specific partition types and the
2022 cryptographic requirements on the partitions (Verity, LUKS, …) can be
2023 restricted, permitting better control of the exposed attack surfaces
2024 when mounting disk images. systemd-gpt-auto-generator will honour such
2025 an image policy too, configurable via the systemd.image_policy= kernel
2026 command line option. Unit files gained the RootImagePolicy=,
2027 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
2028 disk images a service runs off.
2030 * systemd-analyze gained a new verb "image-policy" to validate and
2031 parse image policy strings.
2033 * systemd-dissect gained support for a new --validate switch to
2034 superficially validate DDI structure, and check whether a specific
2035 image policy allows the DDI.
2037 * systemd-dissect gained support for a new --mtree-hash switch to
2038 optionally disable calculating mtree hashes, which can be slow on
2041 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
2042 are now able to operate on directories too, other than images.
2046 * networkd's GENEVE support as gained a new .network option
2047 InheritInnerProtocol=.
2049 * The [Tunnel] section in .netdev files has gained a new setting
2050 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
2052 * A new global IPv6PrivacyExtensions= setting has been added that
2053 selects the default value of the per-network setting of the same
2056 * The predictable network interface naming logic was extended to
2057 include SR-IOV-R "representor" information in network interface
2058 names. Unfortunately, this feature was not enabled by default and can
2059 only be enabled at compilation time by setting
2060 -Ddefault-net-naming-scheme=v254.
2062 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
2063 the RFC8910 captive portal option.
2067 * udevadm gained the new "verify" verb for validating udev rules files
2070 * udev gained a new tool "iocost" that can be used to configure QoS IO
2071 cost data based on hwdb information onto suitable block devices. Also
2072 see https://github.com/iocost-benchmark/iocost-benchmarks.
2074 TPM2 Support + Disk Encryption & Authentication:
2076 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
2077 ("Storage Root Key") as first step in the TPM2, and then use that
2078 for binding FDE to, if TPM2 support is used. This matches
2079 recommendations of TCG (see
2080 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
2082 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
2083 understand textual identifiers for these PCRs.
2085 * systemd-veritysetup + /etc/veritytab gained support for a series of
2086 new options: hash-offset=, superblock=, format=, data-block-size=,
2087 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
2088 fec-offset=, fec-roots= to configure various aspects of a Verity
2091 * systemd-cryptsetup + /etc/crypttab gained support for a new
2092 veracrypt-pim= option for setting the Personal Iteration Multiplier
2093 of veracrypt volumes.
2095 * systemd-integritysetup + /etc/integritytab gained support for a new
2096 mode= setting for controlling the dm-integrity mode (journal, bitmap,
2097 direct) for the volume.
2099 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
2100 registers, their symbolic names and current values.
2104 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
2105 access right is specified this is equivalent to "x" but only if the
2106 inode in question already has the executable bit set for at least
2107 some user/group. Otherwise the "x" bit will be turned off.
2109 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
2110 with C+ will result in a "merge" copy, i.e. all files of the source
2111 tree are copied into the target tree, even if that tree already
2112 exists, resulting in a combined tree of files already present in the
2113 target tree and those copied in.
2115 * systemd-tmpfiles gained a new --graceful switch. If specified lines
2116 with unknown users/groups will silently be skipped.
2120 * systemd-notify gained two new options --fd= and --fdname= for sending
2121 arbitrary file descriptors to the service manager (while specifying an
2122 explicit name for it).
2124 * systemd-notify gained a new --exec switch, which makes it execute the
2125 specified command line after sending the requested messages. This is
2126 useful for sending out READY=1 first, and then continuing invocation
2127 without changing process ID, so that the tool can be nicely used
2128 within an ExecStart= line of a unit file that uses Type=notify.
2130 sd-event + sd-bus APIs:
2132 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
2133 which may be used to explicitly end a rate-limit state an event
2134 source might be in, resetting all rate limiting counters.
2136 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
2137 sockets, it will now encode the "description" set via
2138 sd_bus_set_description() into the source socket address. It will also
2139 look for this information when accepting a connection. This is useful
2140 to track individual D-Bus connections on a D-Bus broker for debug
2145 * systemd-resolved gained a new resolved.conf setting
2146 StateRetentionSec= which may be used to retain cached DNS records
2147 even after their nominal TTL, and use them in case upstream DNS
2148 servers cannot be reached. This can be used to make name resolution
2149 more resilient in case of network problems.
2151 * resolvectl gained a new verb "show-cache" to show the current cache
2152 contents of systemd-resolved. This verb communicates with the
2153 systemd-resolved daemon and requires privileges.
2157 * Meson >= 0.60.0 is now required to build systemd.
2159 * The default keymap to apply may now be chosen at build-time via the
2160 new -Ddefault-keymap= meson option.
2162 * Most of systemd's long-running services now have a generic handler of
2163 the SIGRTMIN+18 signal handler which executes various operations
2164 depending on the sigqueue() parameter sent along. For example, values
2165 0x100…0x107 allow changing the maximum log level of such
2166 services. 0x200…0x203 allow changing the log target of such
2167 services. 0x300 make the services trim their memory similarly to the
2168 automatic PSI-triggered action, see above. 0x301 make the services
2169 output their malloc_info() data to the logs.
2171 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
2172 files, inspired by systemctl's verbs of the same name which edit unit
2173 files. Similarly, networkctl gained the same verbs for editing
2174 .network, .netdev, .link files.
2176 * A new syscall filter group "@sandbox" has been added that contains
2177 syscalls for sandboxing system calls such as those for seccomp and
2180 * New documentation has been added:
2182 https://systemd.io/COREDUMP
2183 https://systemd.io/MEMORY_PRESSURE
2186 * systemd-firstboot gained a new --reset option. If specified, the
2187 settings in /etc/ it knows how to initialize are reset.
2189 * systemd-sysext is now a multi-call binary and is also installed under
2190 the systemd-confext alias name (via a symlink). When invoked that way
2191 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
2192 powerful, atomic, secure configuration management of sorts, that
2193 locally can merge configuration from multiple confext configuration
2194 images into a single immutable tree.
2196 * The --network-macvlan=, --network-ipvlan=, --network-interface=
2197 switches of systemd-nspawn may now optionally take the intended
2198 network interface inside the container.
2200 * All our programs will now send an sd_notify() message with their exit
2201 status in the EXIT_STATUS= field when exiting, using the usual
2202 protocol, including PID 1. This is useful for VMMs and container
2203 managers to collect an exit status from a system as it shuts down, as
2204 set via "systemctl exit …". This is particularly useful in test cases
2205 and similar, as invocations via a VM can now nicely propagate an exit
2206 status to the host, similar to local processes.
2208 * systemd-run gained a new switch --expand-environment=no to disable
2209 server-side environment variable expansion in specified command
2210 lines. Expansion defaults to enabled for all execution types except
2211 --scope, where it defaults to off (and prints a warning) for backward
2212 compatibility reasons. --scope will be flipped to enabled by default
2213 too in a future release. If you are using --scope and passing a '$'
2214 character in the payload you should start explicitly using
2215 --expand-environment=yes/no according to the use case.
2217 * The systemd-system-update-generator has been updated to also look for
2218 the special flag file /etc/system-update in addition to the existing
2219 support for /system-update to decide whether to enter system update
2222 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
2223 mount options by default.
2225 * systemd-fstab-generator now understands two new kernel command line
2226 options systemd.mount-extra= and systemd.swap-extra=, which configure
2227 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
2228 will be ran on these block devices, like it already happens for
2229 'root='. It also now supports the new fstab.extra and
2230 fstab.extra.initrd credentials that may contain additional /etc/fstab
2231 lines to apply at boot.
2233 * systemd-getty-generator now understands two new credentials
2234 getty.ttys.container and getty.ttys.serial. These credentials may
2235 contain a list of TTY devices – one per line – to instantiate
2236 container-getty@.service and serial-getty@.service on.
2238 * The getty/serial-getty/container-getty units now import the 'agetty.*'
2239 and 'login.*' credentials, which are consumed by the 'login' and
2240 'agetty' programs starting from util-linux v2.40.
2242 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
2243 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
2244 which case the Path= setting is taken relative to the ESP or XBOOTLDR
2245 partitions, rather than the system's root directory /. The relevant
2246 directories are automatically discovered.
2248 * The systemd-ac-power tool gained a new switch --low, which reports
2249 whether the battery charge is considered "low", similar to how the
2250 s2h suspend logic checks this state to decide whether to enter system
2251 suspend or hibernation.
2253 * The /etc/os-release file can now have two new optional fields
2254 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
2257 * When the system hibernates, information about the device and offset
2258 used is now written to a non-volatile EFI variable. On next boot the
2259 system will attempt to resume from the location indicated in this EFI
2260 variable. This should make hibernation a lot more robust, while
2261 requiring no manual configuration of the resume location.
2263 * The $XDG_STATE_HOME environment variable (added in more recent
2264 versions of the XDG basedir specification) is now honoured to
2265 implement the StateDirectory= setting in user services.
2267 * A new component "systemd-battery-check" has been added. It may run
2268 during early boot (usually in the initrd), and checks the battery
2269 charge level of the system. In case the charge level is very low the
2270 user is notified (graphically via Plymouth – if available – as well
2271 as in text form on the console), and the system is turned off after a
2272 10s delay. The feature can be disabled by passing
2273 systemd.battery_check=0 through the kernel command line.
2275 * The 'passwdqc' library is now supported as an alternative to the
2276 'pwquality' library and can be selected at build time.
2278 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
2279 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
2280 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
2281 Arian van Putten, Arthur Shau, A S Alam,
2282 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
2283 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
2284 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
2285 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
2286 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
2287 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
2288 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
2289 David Schroeder, David Tardon, dependabot[bot],
2290 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
2291 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
2292 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
2293 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
2294 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
2295 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
2296 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
2297 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
2298 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
2299 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
2300 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
2301 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
2302 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
2303 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
2304 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
2305 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
2306 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
2307 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
2308 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
2309 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
2310 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
2311 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
2312 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
2313 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
2314 saikat0511, Samanta Navarro, Sam James, Sam Morris,
2315 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
2316 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
2317 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
2318 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
2319 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
2320 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
2321 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
2322 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
2323 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
2324 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
2325 Дамјан Георгиевски, наб
2327 — Edinburgh, 2023-07-28
2331 Announcements of Future Feature Removals and Incompatible Changes:
2333 * We intend to remove cgroup v1 support from systemd release after the
2334 end of 2023. If you run services that make explicit use of cgroup v1
2335 features (i.e. the "legacy hierarchy" with separate hierarchies for
2336 each controller), please implement compatibility with cgroup v2 (i.e.
2337 the "unified hierarchy") sooner rather than later. Most of Linux
2338 userspace has been ported over already.
2340 * We intend to remove support for split-usr (/usr mounted separately
2341 during boot) and unmerged-usr (parallel directories /bin and
2342 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
2343 half of 2023, in the first release that falls into that time window.
2344 For more details, see:
2345 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
2347 * We intend to change behaviour w.r.t. units of the per-user service
2348 manager and sandboxing options, so that they work without having to
2349 manually enable PrivateUsers= as well, which is not required for
2350 system units. To make this work, we will implicitly enable user
2351 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
2352 user unit. The drawback is that system users will no longer be visible
2353 (and appear as 'nobody') to the user unit when a sandboxing option is
2354 enabled. By definition a sandboxed user unit should run with reduced
2355 privileges, so impact should be small. This will remove a great source
2356 of confusion that has been reported by users over the years, due to
2357 how these options require an extra setting to be manually enabled when
2358 used in the per-user service manager, as opposed as to the system
2359 service manager. We plan to enable this change in the next release
2360 later this year. For more details, see:
2361 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
2363 Deprecations and incompatible changes:
2365 * systemctl will now warn when invoked without /proc/ mounted
2366 (e.g. when invoked after chroot() into an directory tree without the
2367 API mount points like /proc/ being set up.) Operation in such an
2368 environment is not fully supported.
2370 * The return value of 'systemctl is-active|is-enabled|is-failed' for
2371 unknown units is changed: previously 1 or 3 were returned, but now 4
2372 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
2374 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
2375 systemd-hwdb (added in 2014) should be used instead.
2377 * 'bootctl --json' now outputs a single JSON array, instead of a stream
2378 of newline-separated JSON objects.
2380 * Udev rules in 60-evdev.rules have been changed to load hwdb
2381 properties for all modalias patterns. Previously only the first
2382 matching pattern was used. This could change what properties are
2383 assigned if the user has more and less specific patterns that could
2384 match the same device, but it is expected that the change will have
2385 no effect for most users.
2387 * systemd-networkd-wait-online exits successfully when all interfaces
2388 are ready or unmanaged. Previously, if neither '--any' nor
2389 '--interface=' options were used, at least one interface had to be in
2390 configured state. This change allows the case where systemd-networkd
2391 is enabled, but no interfaces are configured, to be handled
2392 gracefully. It may occur in particular when a different network
2393 manager is also enabled and used.
2395 * Some compatibility helpers were dropped: EmergencyAction= in the user
2396 manager, as well as measuring kernel command line into PCR 8 in
2397 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
2400 * The '-Dupdate-helper-user-timeout=' build-time option has been
2401 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
2402 integer as parameter instead of a string.
2404 * The DDI image dissection logic (which backs RootImage= in service
2405 unit files, the --image= switch in various tools such as
2406 systemd-nspawn, as well as systemd-dissect) will now only mount file
2407 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
2408 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
2409 variable. These file systems are fairly well supported and maintained
2410 in current kernels, while others are usually more niche, exotic or
2411 legacy and thus typically do not receive the same level of security
2414 * The default per-link multicast DNS mode is changed to "yes"
2415 (that was previously "no"). As the default global multicast DNS mode
2416 has been "yes" (but can be changed by the build option), now the
2417 multicast DNS is enabled on all links by default. You can disable the
2418 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
2419 or on an interface by calling "resolvectl mdns INTERFACE no".
2423 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
2424 (UKIs) has been added. This replaces functionality provided by
2425 'dracut --uefi' and extends it with automatic calculation of PE file
2426 offsets, insertion of signed PCR policies generated by
2427 systemd-measure, support for initrd concatenation, signing of the
2428 embedded Linux image and the combined image with sbsign, and
2429 heuristics to autodetect the kernel uname and verify the splash
2432 Changes in systemd and units:
2434 * A new service type Type=notify-reload is defined. When such a unit is
2435 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
2436 service process. The manager will then wait until it receives a
2437 "RELOADING=1" followed by a "READY=1" notification from the unit as
2438 response (via sd_notify()). Otherwise, this type is the same as
2439 Type=notify. A new setting ReloadSignal= may be used to change the
2440 signal to send from the default of SIGHUP.
2442 user@.service, systemd-networkd.service, systemd-udevd.service, and
2443 systemd-logind have been updated to this type.
2445 * Initrd environments which are not on a pure memory file system (e.g.
2446 overlayfs combination as opposed to tmpfs) are now supported. With
2447 this change, during the initrd → host transition ("switch root")
2448 systemd will erase all files of the initrd only when the initrd is
2449 backed by a memory file system such as tmpfs.
2451 * New per-unit MemoryZSwapMax= option has been added to configure
2452 memory.zswap.max cgroup properties (the maximum amount of zswap
2455 * A new LogFilterPatterns= option has been added for units. It may be
2456 used to specify accept/deny regular expressions for log messages
2457 generated by the unit, that shall be enforced by systemd-journald.
2458 Rejected messages are neither stored in the journal nor forwarded.
2459 This option may be used to suppress noisy or uninteresting messages
2462 * The manager has a new
2463 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
2464 query process ownership via a PIDFD, which is more resilient against
2465 PID recycling issues.
2467 * Scope units now support OOMPolicy=. Login session scopes default to
2468 OOMPolicy=continue, allowing login scopes to survive the OOM killer
2469 terminating some processes in the scope.
2471 * systemd-fstab-generator now supports x-systemd.makefs option for
2472 /sysroot/ (in the initrd).
2474 * The maximum rate at which daemon reloads are executed can now be
2475 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
2476 options. (Or the equivalent on the kernel command line:
2477 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
2478 addition, systemd now logs the originating unit and PID when a reload
2479 request is received over D-Bus.
2481 * When enabling a swap device systemd will now reinitialize the device
2482 when the page size of the swap space does not match the page size of
2483 the running kernel. Note that this requires the 'swapon' utility to
2484 provide the '--fixpgsz' option, as implemented by util-linux, and it
2485 is not supported by busybox at the time of writing.
2487 * systemd now executes generator programs in a mount namespace
2488 "sandbox" with most of the file system read-only and write access
2489 restricted to the output directories, and with a temporary /tmp/
2490 mount provided. This provides a safeguard against programming errors
2491 in the generators, but also fixes here-docs in shells, which
2492 previously didn't work in early boot when /tmp/ wasn't available
2493 yet. (This feature has no security implications, because the code is
2494 still privileged and can trivially exit the sandbox.)
2496 * The system manager will now parse a new "vmm.notify_socket"
2497 system credential, which may be supplied to a VM via SMBIOS. If
2498 found, the manager will send a "READY=1" notification on the
2499 specified socket after boot is complete. This allows readiness
2500 notification to be sent from a VM guest to the VM host over a VSOCK
2503 * The sample PAM configuration file for systemd-user@.service now
2504 includes a call to pam_namespace. This puts children of user@.service
2505 in the expected namespace. (Many distributions replace their file
2506 with something custom, so this change has limited effect.)
2508 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
2509 can be used to override the mount units burst late limit for
2510 parsing '/proc/self/mountinfo', which was introduced in v249.
2513 * Drop-ins for init.scope changing control group resource limits are
2514 now applied, while they were previously ignored.
2516 * New build-time configuration options '-Ddefault-timeout-sec=' and
2517 '-Ddefault-user-timeout-sec=' have been added, to let distributions
2518 choose the default timeout for starting/stopping/aborting system and
2519 user units respectively.
2521 * Service units gained a new setting OpenFile= which may be used to
2522 open arbitrary files in the file system (or connect to arbitrary
2523 AF_UNIX sockets in the file system), and pass the open file
2524 descriptor to the invoked process via the usual file descriptor
2525 passing protocol. This is useful to give unprivileged services access
2526 to select files which have restrictive access modes that would
2527 normally not allow this. It's also useful in case RootDirectory= or
2528 RootImage= is used to allow access to files from the host environment
2529 (which is after all not visible from the service if these two options
2534 * The new net naming scheme "v253" has been introduced. In the new
2535 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
2536 a PCI bus. This extends the coverage of predictable interface names
2537 in some embedded systems.
2539 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
2540 a more informative path on some embedded systems.
2542 * Partition block devices will now also get symlinks in
2543 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
2544 block device nodes via the kernel's "diskseq" value. Previously those
2545 symlinks were only created for the main block device.
2547 * A new operator '-=' is supported for SYMLINK variables. This allows
2548 symlinks to be unconfigured even if an earlier rule added them.
2550 * 'udevadm --trigger --settle' now also works for network devices
2551 that are being renamed.
2553 Changes in sd-boot, bootctl, and the Boot Loader Specification:
2555 * systemd-boot now passes its random seed directly to the kernel's RNG
2556 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
2557 means the RNG gets seeded very early in boot before userspace has
2560 * systemd-boot will pass a disk-backed random seed – even when secure
2561 boot is enabled – if it can additionally get a random seed from EFI
2562 itself (via EFI's RNG protocol), or a prior seed in
2563 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
2565 * systemd-boot-system-token.service was renamed to
2566 systemd-boot-random-seed.service and extended to always save a random
2567 seed to ESP on every boot when a compatible boot loader is used. This
2568 allows a refreshed random seed to be used in the boot loader.
2570 * systemd-boot handles various seed inputs using a domain- and
2571 field-separated hashing scheme.
2573 * systemd-boot's 'random-seed-mode' option has been removed. A system
2574 token is now always required to be present for random seeds to be
2577 * systemd-boot now supports being loaded from other locations than the
2578 ESP, for example for direct kernel boot under QEMU or when embedded
2581 * systemd-boot now parses SMBIOS information to detect
2582 virtualization. This information is used to skip some warnings which
2583 are not useful in a VM and to conditionalize other aspects of
2586 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
2587 Secure Boot automated certificate enrollment from the ESP only if it
2588 is considered 'safe' to do so. At the moment 'safe' means running in
2591 * systemd-stub now processes random seeds in the same way as
2592 systemd-boot already does, in case a unified kernel image is being
2593 used from a different bootloader than systemd-boot, or without any
2596 * bootctl will now generate a system token on all EFI systems, even
2597 virtualized ones, and is activated in the case that the system token
2598 is missing from either sd-boot and sd-stub booted systems.
2600 * bootctl now implements two new verbs: 'kernel-identify' prints the
2601 type of a kernel image file, and 'kernel-inspect' provides
2602 information about the embedded command line and kernel version of
2605 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
2606 as for kernel-install.
2608 * The JSON output of "bootctl list" will now contain two more fields:
2609 isDefault and isSelected are boolean fields set to true on the
2610 default and currently booted boot menu entries.
2612 * bootctl gained a new verb "unlink" for removing a boot loader entry
2613 type #1 file from disk in a safe and robust way.
2615 * bootctl also gained a new verb "cleanup" that automatically removes
2616 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
2617 is not referenced anymore by any installed Type #1 boot loader
2618 specification entry. This is particularly useful in environments where
2619 a large number of entries reference the same or partly the same
2620 resources (for example, for snapshot-based setups).
2622 Changes in kernel-install:
2624 * A new "installation layout" can be configured as layout=uki. With
2625 this setting, a Boot Loader Specification Type#1 entry will not be
2626 created. Instead, a new kernel-install plugin 90-uki-copy.install
2627 will copy any .efi files from the staging area into the boot
2628 partition. A plugin to generate the UKI .efi file must be provided
2631 Changes in systemctl:
2633 * 'systemctl reboot' has dropped support for accepting a positional
2634 argument as the argument to the reboot(2) syscall. Please use the
2635 --reboot-argument= option instead.
2637 * 'systemctl disable' will now warn when called on units without
2638 install information. A new --no-warn option has been added that
2639 silences this warning.
2641 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
2642 of the drop-in to edit. (Previously, 'override.conf' was always
2645 * 'systemctl list-dependencies' now respects --type= and --state=.
2647 * 'systemctl kexec' now supports XEN VMM environments.
2649 * 'systemctl edit' will now tell the invoked editor to jump into the
2650 first line with actual unit file data, skipping over synthesized
2653 Changes in systemd-networkd and related tools:
2655 * The [DHCPv4] section in .network file gained new SocketPriority=
2656 setting that assigns the Linux socket priority used by the DHCPv4 raw
2657 socket. This may be used in conjunction with the
2658 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
2659 desired ethernet 802.1Q frame priority for DHCPv4 initial
2660 packets. This cannot be achieved with netfilter mangle tables because
2661 of the raw socket bypass.
2663 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
2664 new QuickAck= boolean setting that enables the TCP quick ACK mode for
2665 the routes configured by the acquired DHCPv4 lease or received router
2666 advertisements (RAs).
2668 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
2669 routes) now accepts three values, for high, medium, and low preference
2670 of the router (which can be set with the RouterPreference=) setting.
2672 * systemd-networkd-wait-online now supports matching via alternative
2675 * The [DHCPv6] section in .network file gained new SendRelease=
2676 setting which enables the DHCPv6 client to send release when
2677 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
2678 It is enabled by default.
2680 * If the Address= setting in [Network] or [Address] sections in .network
2681 specified without its prefix length, then now systemd-networkd assumes
2682 /32 for IPv4 or /128 for IPv6 addresses.
2684 * networkctl shows network and link file dropins in status output.
2686 Changes in systemd-dissect:
2688 * systemd-dissect gained a new option --list, to print the paths of
2689 all files and directories in a DDI.
2691 * systemd-dissect gained a new option --mtree, to generate a file
2692 manifest compatible with BSD mtree(5) of a DDI
2694 * systemd-dissect gained a new option --with, to execute a command with
2695 the specified DDI temporarily mounted and used as working
2696 directory. This is for example useful to convert a DDI to "tar"
2697 simply by running it within a "systemd-dissect --with" invocation.
2699 * systemd-dissect gained a new option --discover, to search for
2700 Discoverable Disk Images (DDIs) in well-known directories of the
2701 system. This will list machine, portable service and system extension
2704 * systemd-dissect now understands 2nd stage initrd images stored as a
2705 Discoverable Disk Image (DDI).
2707 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
2708 disk UUID stored in the GPT header) among the other data it can show.
2710 * systemd-dissect gained a new --in-memory switch to operate on an
2711 in-memory copy of the specified DDI file. This is useful to access a
2712 DDI with write access without persisting any changes. It's also
2713 useful for accessing a DDI without keeping the originating file
2716 * The DDI dissection logic will now automatically detect the intended
2717 sector size of disk images stored in files, based on the GPT
2718 partition table arrangement. Loopback block devices for such DDIs
2719 will then be configured automatically for the right sector size. This
2720 is useful to make dealing with modern 4K sector size DDIs fully
2721 automatic. The systemd-dissect tool will now show the detected sector
2722 size among the other DDI information in its output.
2724 Changes in systemd-repart:
2726 * systemd-repart gained new options --include-partitions= and
2727 --exclude-partitions= to filter operation on partitions by type UUID.
2728 This allows systemd-repart to be used to build images in which the
2729 type of one partition is set based on the contents of another
2730 partition (for example when the boot partition shall include a verity
2731 hash of the root partition).
2733 * systemd-repart also gained a --defer-partitions= option that is
2734 similar to --exclude-partitions=, but the size of the partition is
2735 still taken into account when sizing partitions, but without
2738 * systemd-repart gained a new --sector-size= option to specify what
2739 sector size should be used when an image is created.
2741 * systemd-repart now supports generating erofs file systems via
2742 CopyFiles= (a read-only file system similar to squashfs).
2744 * The Minimize= option was extended to accept "best" (which means the
2745 most minimal image possible, but may require multiple attempts) and
2746 "guess" (which means a reasonably small image).
2748 * The systemd-growfs binary now comes with a regular unit file template
2749 systemd-growfs@.service which can be instantiated directly for any
2750 desired file system. (Previously, the unit was generated dynamically
2751 by various generators, but no regular unit file template was
2754 Changes in journal tools:
2756 * Various systemd tools will append extra fields to log messages when
2757 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
2758 this includes information about D-Bus messages when sd-bus is used,
2759 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
2760 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
2761 Details of what is logged and when are subject to change.
2763 * The systemd-journald-audit.socket can now be disabled via the usual
2764 "systemctl disable" mechanism to stop collection of audit
2765 messages. Please note that it is not enabled statically anymore and
2766 must be handled by the preset/enablement logic in package
2767 installation scripts.
2769 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
2770 be used to curtail disk use by systemd-journal-remote. This is
2771 similar to the options supported by systemd-journald.
2773 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
2776 * When enrolling new keys systemd-cryptenroll now supports unlocking
2777 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
2778 password was strictly required to be specified.
2780 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
2781 (except for tokens with user verification, UV) to identify tokens
2782 before authentication. Multiple FIDO2 tokens can now be enrolled at
2783 the same time, and systemd-cryptsetup will automatically select one
2784 that corresponds to one of the available LUKS key slots.
2786 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
2787 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
2788 bank and number into which the volume key should be measured. This is
2789 automatically enabled for the encrypted root volume discovered and
2790 activated by systemd-gpt-auto-generator.
2792 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
2793 "noexec,nosuid,nodev".
2795 * systemd-gpt-auto-generator will now honour the rootfstype= and
2796 rootflags= kernel command line switches for root file systems it
2797 discovers, to match behaviour in case an explicit root fs is
2798 specified via root=.
2800 * systemd-pcrphase gained new options --machine-id and --file-system=
2801 to measure the machine-id and mount point information into PCR 15.
2802 New service unit files systemd-pcrmachine.service and
2803 systemd-pcrfs@.service have been added that invoke the tool with
2804 these switches during early boot.
2806 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
2807 with a success exit code even if no TPM device is detected.
2809 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
2810 making it harder to brute-force.
2812 Changes in other tools:
2814 * systemd-homed gained support for luksPbkdfForceIterations (the
2815 intended number of iterations for the PBKDF operation on LUKS).
2817 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
2818 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
2819 may now be used to specify additional arguments for mkfs when
2820 systemd-homed formats a file system.
2822 * systemd-hostnamed now exports the contents of
2823 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
2824 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
2825 unprivileged code to access those values.
2827 systemd-hostnamed also exports the SUPPORT_END= field from
2828 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
2829 this to show the status of the installed system.
2831 * systemd-measure gained an --append= option to sign multiple phase
2832 paths with different signing keys. This allows secrets to be
2833 accessible only in certain parts of the boot sequence. Note that
2834 'ukify' provides similar functionality in a more accessible form.
2836 * systemd-timesyncd will now write a structured log message with
2837 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
2838 on a on-disk timestamp, similarly to what it did when reaching
2839 synchronization via NTP.
2841 * systemd-timesyncd will now update the on-disk timestamp file on each
2842 boot at least once, making it more likely that the system time
2843 increases in subsequent boots.
2845 * systemd-vconsole-setup gained support for system/service credentials:
2846 vconsole.keymap/vconsole.keymap_toggle and
2847 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
2848 the similarly-named options in vconsole.conf.
2850 * systemd-localed will now save the XKB keyboard configuration to
2851 /etc/vconsole.conf, and also read it from there with a higher
2852 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
2853 file. Previously, this information was stored in the former file in
2854 converted form, and only in latter file in the original form. Tools
2855 which want to access keyboard configuration can now do so from a
2858 * systemd-resolved gained support for configuring the nameservers and
2859 search domains via kernel command line (nameserver=, domain=) and
2860 credentials (network.dns, network.search_domains).
2862 * systemd-resolved will now synthesize host names for the DNS stub
2863 addresses it supports. Specifically when "_localdnsstub" is resolved,
2864 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
2865 127.0.0.54 is returned.
2867 * systemd-notify will now send a "RELOADING=1" notification when called
2868 with --reloading, and "STOPPING=1" when called with --stopping. This
2869 can be used to implement notifications from units where it's easier
2870 to call a program than to use the sd-daemon library.
2872 * systemd-analyze's 'plot' command can now output its information in
2873 JSON, controlled via the --json= switch. Also, new --table, and
2874 --no-legend options have been added.
2876 * 'machinectl enable' will now automatically enable machines.target
2877 unit in addition to adding the machine unit to the target.
2879 Similarly, 'machinectl start|stop' gained a --now option to enable or
2880 disable the machine unit when starting or stopping it.
2882 * systemd-sysusers will now create /etc/ if it is missing.
2884 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
2885 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
2886 added to provide the new initial value for the new automated battery
2887 estimation functionality. If 'HibernateDelaySec=' is set to any value,
2888 the automated estimate (and thus the automated hibernation on low
2889 battery to avoid data loss) functionality will be disabled.
2891 * Default tmpfiles.d/ configuration will now automatically create
2892 credentials storage directory '/etc/credstore/' with the appropriate,
2893 secure permissions. If '/run/credstore/' exists, its permissions will
2894 be fixed too in case they are not correct.
2896 Changes in libsystemd and shared code:
2898 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
2899 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
2901 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
2902 128-bit ID in files such as /etc/machine-id has an invalid
2903 format. They also accept NULL as output parameter in more places,
2904 which is useful when the caller only wants to validate the inputs and
2905 does not need the output value.
2907 * sd-login gained new functions sd_pidfd_get_session(),
2908 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
2909 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
2910 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
2911 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
2912 but accept a PIDFD instead of a PID.
2914 * sd-path (and systemd-path) now export four new paths:
2915 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
2916 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
2917 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
2918 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
2920 * sd_notify() now supports AF_VSOCK as transport for notification
2921 messages (in addition to the existing AF_UNIX support). This is
2922 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
2924 * Detection of chroot() environments now works if /proc/ is not
2925 mounted. This affects systemd-detect-virt --chroot, but also means
2926 that systemd tools will silently skip various operations in such an
2929 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
2930 virtualization is now detected.
2932 Changes in the build system:
2934 * Standalone variants of systemd-repart and systemd-shutdown may now be
2935 built (if -Dstandalone=true).
2937 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
2938 example, allow scripts to conditionalize execution on AC power
2941 * The libp11kit library is now loaded through dlopen(3).
2943 Changes in the documentation:
2945 * Specifications that are not closely tied to systemd have moved to
2946 https://uapi-group.org/specifications/: the Boot Loader Specification
2947 and the Discoverable Partitions Specification.
2949 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
2950 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
2951 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
2952 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
2953 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
2954 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
2955 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
2956 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
2957 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
2958 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
2959 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
2960 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
2961 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
2962 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
2963 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
2964 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
2965 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
2966 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
2967 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
2968 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
2969 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
2970 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
2971 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
2972 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
2973 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
2974 reuben olinsky, Richard E. van der Luit, Richard Phibel,
2975 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
2976 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
2977 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
2978 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
2979 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
2980 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
2981 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
2982 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
2985 — Warsaw, 2023-02-15
2989 Announcements of Future Feature Removals:
2991 * We intend to remove cgroup v1 support from systemd release after the
2992 end of 2023. If you run services that make explicit use of cgroup v1
2993 features (i.e. the "legacy hierarchy" with separate hierarchies for
2994 each controller), please implement compatibility with cgroup v2 (i.e.
2995 the "unified hierarchy") sooner rather than later. Most of Linux
2996 userspace has been ported over already.
2998 * We intend to remove support for split-usr (/usr mounted separately
2999 during boot) and unmerged-usr (parallel directories /bin and
3000 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
3001 half of 2023, in the first release that falls into that time window.
3002 For more details, see:
3003 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
3005 Compatibility Breaks:
3007 * ConditionKernelVersion= checks that use the '=' or '!=' operators
3008 will now do simple string comparisons (instead of version comparisons
3009 à la stverscmp()). Version comparisons are still done for the
3010 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
3011 specified, a shell-style glob match is now done. This creates a minor
3012 incompatibility compared to older systemd versions when the '*', '?',
3013 '[', ']' characters are used, as these will now match as shell globs
3014 instead of literally. Given that kernel version strings typically do
3015 not include these characters we expect little breakage through this
3018 * The service manager will now read the SELinux label used for SELinux
3019 access checks from the unit file at the time it loads the file.
3020 Previously, the label would be read at the moment of the access
3021 check, which was problematic since at that time the unit file might
3022 already have been updated or removed.
3026 * systemd-measure is a new tool for calculating and signing expected
3027 TPM2 PCR values for a given unified kernel image (UKI) booted via
3028 sd-stub. The public key used for the signature and the signed
3029 expected PCR information can be embedded inside the UKI. This
3030 information can be extracted from the UKI by external tools and code
3031 in the image itself and is made available to userspace in the booted
3034 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
3035 updated to make use of this information if available in the booted
3036 kernel: when locking an encrypted volume/credential to the TPM
3037 systemd-cryptenroll/systemd-creds will use the public key to bind the
3038 volume/credential to any kernel that carries PCR information signed
3039 by the same key pair. When unlocking such volumes/credentials
3040 systemd-cryptsetup/systemd-creds will use the signature embedded in
3041 the booted UKI to gain access.
3043 Binding TPM-based disk encryption to public keys/signatures of PCR
3044 values — instead of literal PCR values — addresses the inherent
3045 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
3046 disks remain accessible even if the UKI is updated, without any TPM
3047 specific preparation during the OS update — as long as each UKI
3048 carries the necessary PCR signature information.
3050 Net effect: if you boot a properly prepared kernel, TPM-bound disk
3051 encryption now defaults to be locked to kernels which carry PCR
3052 signatures from the same key pair. Example: if a hypothetical distro
3053 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
3054 by default – bound to only FooOS kernels, and encrypted volumes bound
3055 to the TPM cannot be unlocked on kernels from other sources. (But do
3056 note this behaviour requires preparation/enabling in the UKI, and of
3057 course users can always enroll non-TPM ways to unlock the volume.)
3059 * systemd-pcrphase is a new tool that is invoked at six places during
3060 system runtime, and measures additional words into TPM2 PCR 11, to
3061 mark milestones of the boot process. This allows binding access to
3062 specific TPM2-encrypted secrets to specific phases of the boot
3063 process. (Example: LUKS2 disk encryption key only accessible in the
3064 initrd, but not later.)
3066 Changes in systemd itself, i.e. the manager and units
3068 * The cpu controller is delegated to user manager units by default, and
3069 CPUWeight= settings are applied to the top-level user slice units
3070 (app.slice, background.slice, session.slice). This provides a degree
3071 of resource isolation between different user services competing for
3074 * Systemd can optionally do a full preset in the "first boot" condition
3075 (instead of just enable-only). This behaviour is controlled by the
3076 compile-time option -Dfirst-boot-full-preset. Right now it defaults
3077 to 'false', but the plan is to switch it to 'true' for the subsequent
3080 * Drop-ins are now allowed for transient units too.
3082 * Systemd will set the taint flag 'support-ended' if it detects that
3083 the OS image is past its end-of-support date. This date is declared
3084 in a new /etc/os-release field SUPPORT_END= described below.
3086 * Two new settings ConditionCredential= and AssertCredential= can be
3087 used to skip or fail units if a certain system credential is not
3090 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
3092 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
3093 specify the SMACK security label to use when not specified in a unit
3096 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
3097 specify the default timeout when waiting for device units to
3100 * C.UTF-8 is used as the default locale if nothing else has been
3103 * [Condition|Assert]Firmware= have been extended to support certain
3104 SMBIOS fields. For example
3106 ConditionFirmware=smbios-field(board_name = "Custom Board")
3108 conditionalizes the unit to run only when
3109 /sys/class/dmi/id/board_name contains "Custom Board" (without the
3112 * ConditionFirstBoot= now correctly evaluates as true only during the
3113 boot phase of the first boot. A unit executed later, after booting
3114 has completed, will no longer evaluate this condition as true.
3116 * Socket units will now create sockets in the SELinuxContext= of the
3117 associated service unit, if any.
3119 * Boot phase transitions (start initrd → exit initrd → boot complete →
3120 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
3121 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
3122 unsealed only in the initrd.
3124 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
3125 also be provided to ExecStartPre= processes.
3127 * Various units are now correctly ordered against
3128 initrd-switch-root.target where previously a conflict without
3129 ordering was configured. A stop job for those units would be queued,
3130 but without the ordering it could be executed only after
3131 initrd-switch-root.service, leading to units not being restarted in
3132 the host system as expected.
3134 * In order to fully support the IPMI watchdog driver, which has not yet
3135 been ported to the new common watchdog device interface,
3136 /dev/watchdog0 will be tried first and systemd will silently fallback
3137 to /dev/watchdog if it is not found.
3139 * New watchdog-related D-Bus properties are now published by systemd:
3140 WatchdogDevice, WatchdogLastPingTimestamp,
3141 WatchdogLastPingTimestampMonotonic.
3143 * At shutdown, API virtual files systems (proc, sys, etc.) will be
3146 * At shutdown, systemd will now log about processes blocking unmounting
3149 * A new meson build option 'clock-valid-range-usec-max' was added to
3150 allow disabling system time correction if RTC returns a timestamp far
3153 * Propagated restart jobs will no longer be discarded while a unit is
3156 * PID 1 will now import system credentials from SMBIOS Type 11 fields
3157 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
3158 simple, fast and generic path for supplying credentials to a VM,
3159 without involving external tools such as cloud-init/ignition.
3161 * The CPUWeight= setting of unit files now accepts a new special value
3162 "idle", which configures "idle" level scheduling for the unit.
3164 * Service processes that are activated due to a .timer or .path unit
3165 triggering will now receive information about this via environment
3166 variables. Note that this is information is lossy, as activation
3167 might be coalesced and only one of the activating triggers will be
3168 reported. This is hence more suited for debugging or tracing rather
3169 than for behaviour decisions.
3171 * The riscv_flush_icache(2) system call has been added to the list of
3172 system calls allowed by default when SystemCallFilter= is used.
3174 * The selinux context derived from the target executable, instead of
3175 'init_t' used for the manager itself, is now used when creating
3176 listening sockets for units that specify SELinuxContextFromNet=yes.
3178 Changes in sd-boot, bootctl, and the Boot Loader Specification:
3180 * The Boot Loader Specification has been cleaned up and clarified.
3181 Various corner cases in version string comparisons have been fixed
3182 (e.g. comparisons for empty strings). Boot counting is now part of
3183 the main specification.
3185 * New PCRs measurements are performed during boot: PCR 11 for the
3186 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
3187 took place this is now reported to userspace via the new
3188 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
3190 * As before, systemd-stub will measure kernel parameters and system
3191 credentials into PCR 12. It will now report this fact via the
3192 StubPcrKernelParameters EFI variable to userspace.
3194 * The UEFI monotonic boot counter is now included in the updated random
3195 seed file maintained by sd-boot, providing some additional entropy.
3197 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
3198 of arranging the image manually and jumping to the kernel entry
3199 point. sd-stub also installs a temporary UEFI SecurityOverride to
3200 allow the (unsigned) nested image to be booted. This is safe because
3201 the outer (signed) stub+kernel binary must have been verified before
3202 the stub was executed.
3204 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
3205 is now supported by sd-boot.
3207 * bootctl gained a bunch of new options: --all-architectures to install
3208 binaries for all supported EFI architectures, --root= and --image=
3209 options to operate on a directory or disk image, and
3210 --install-source= to specify the source for binaries to install,
3211 --efi-boot-option-description= to control the name of the boot entry.
3213 * The sd-boot stub exports a StubFeatures flag, which is used by
3214 bootctl to show features supported by the stub that was used to boot.
3216 * The PE section offsets that are used by tools that assemble unified
3217 kernel images have historically been hard-coded. This may lead to
3218 overlapping PE sections which may break on boot. The UKI will now try
3219 to detect and warn about this.
3221 Any tools that assemble UKIs must update to calculate these offsets
3222 dynamically. Future sd-stub versions may use offsets that will not
3223 work with the currently used set of hard-coded offsets!
3225 * sd-stub now accepts (and passes to the initrd and then to the full
3226 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
3227 signatures of expected PCR values, to allow sealing secrets via the
3228 TPM2 against pre-calculated PCR measurements.
3230 Changes in the hardware database:
3232 * 'systemd-hwdb query' now supports the --root= option.
3234 Changes in systemctl:
3236 * systemctl now supports --state= and --type= options for the 'show'
3239 * systemctl gained a new verb 'list-automounts' to list automount
3242 * systemctl gained support for a new --image= switch to be able to
3243 operate on the specified disk image (similar to the existing --root=
3244 which operates relative to some directory).
3246 Changes in systemd-networkd:
3248 * networkd can set Linux NetLabel labels for integration with the
3249 network control in security modules via a new NetLabel= option.
3251 * The RapidCommit= is (re-)introduced to enable faster configuration
3252 via DHCPv6 (RFC 3315).
3254 * networkd gained a new option TCPCongestionControlAlgorithm= that
3255 allows setting a per-route TCP algorithm.
3257 * networkd gained a new option KeepFileDescriptor= to allow keeping a
3258 reference (file descriptor) open on TUN/TAP interfaces, which is
3259 useful to avoid link flaps while the underlying service providing the
3260 interface is being serviced.
3262 * RouteTable= now also accepts route table names.
3264 Changes in systemd-nspawn:
3266 * The --bind= and --overlay= options now support relative paths.
3268 * The --bind= option now supports a 'rootidmap' value, which will
3269 use id-mapped mounts to map the root user inside the container to the
3270 owner of the mounted directory on the host.
3272 Changes in systemd-resolved:
3274 * systemd-resolved now persists DNSOverTLS in its state file too. This
3275 fixes a problem when used in combination with NetworkManager, which
3276 sends the setting only once, causing it to be lost if resolved was
3277 restarted at any point.
3279 * systemd-resolved now exposes a Varlink socket at
3280 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
3281 root. Processed DNS requests in a JSON format will be published to
3282 any clients connected to this socket.
3284 resolvectl gained a 'monitor' verb to make use of this.
3286 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
3287 instead of returning SERVFAIL, as per RFC:
3288 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
3290 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
3291 is still supported.)
3293 Changes in libsystemd and other libraries:
3295 * libsystemd now exports sd_bus_error_setfv() (a convenience function
3296 for setting bus errors), sd_id128_string_equal (a convenience
3297 function for 128-bit ID string comparisons), and
3298 sd_bus_message_read_strv_extend() (a function to incrementally read
3301 * libsystemd now exports sd_device_get_child_first()/_next() as a
3302 high-level interface for enumerating child devices. It also supports
3303 sd_device_new_child() for opening a child device given a device
3306 * libsystemd now exports sd_device_monitor_set()/get_description()
3307 which allow setting a custom description that will be used in log
3308 messages by sd_device_monitor*.
3310 * Private shared libraries (libsystemd-shared-nnn.so,
3311 libsystemd-core-nnn.so) are now installed into arch-specific
3312 directories to allow multi-arch installs.
3314 * A new sd-gpt.h header is now published, listing GUIDs from the
3315 Discoverable Partitions specification. For more details see:
3316 https://systemd.io/DISCOVERABLE_PARTITIONS/
3318 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
3319 database given an explicit path to the file.
3321 * The signal number argument to sd_event_add_signal() now can now be
3322 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
3323 be automatically invoked to block the specified signal. This is
3324 useful to simplify invocations as the caller doesn't have to do this
3327 * A new convenience call sd_event_set_signal_exit() has been added to
3328 sd-event to set up signal handling so that the event loop
3329 automatically terminates cleanly on SIGTERM/SIGINT.
3331 Changes in other components:
3333 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
3334 can now be provided via the credential mechanism.
3336 * systemd-analyze gained a new verb 'compare-versions' that implements
3337 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
3338 'dpkg --compare-versions').
3340 * 'systemd-analyze dump' is extended to accept glob patterns for unit
3341 names to limit the output to matching units.
3343 * tmpfiles.d/ lines can read file contents to write from a credential.
3344 The new modifier char '^' is used to specify that the argument is a
3345 credential name. This mechanism is used to automatically populate
3346 /etc/motd, /etc/issue, and /etc/hosts from credentials.
3348 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
3349 an inode if the specification is prefixed with ':' and the inode
3352 * Default tmpfiles.d/ configuration now carries a line to automatically
3353 use an 'ssh.authorized_keys.root' credential if provided to set up
3354 the SSH authorized_keys file for the root user.
3356 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
3359 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
3360 in base64. This is useful to write arbitrary binary data into files.
3362 * The pkgconfig and rpm macros files now export the directory for user
3363 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
3365 * Detection of Apple Virtualization and detection of Parallels and
3366 KubeVirt virtualization on non-x86 archs have been added.
3368 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
3369 user when their system will become unsupported.
3371 * When performing suspend-then-hibernate, the system will estimate the
3372 discharge rate and use that to set the delay until hibernation and
3373 hibernate immediately instead of suspending when running from a
3374 battery and the capacity is below 5%.
3376 * systemd-sysctl gained a --strict option to fail when a sysctl
3377 setting is unknown to the kernel.
3379 * machinectl supports --force for the 'copy-to' and 'copy-from'
3382 * coredumpctl gained the --root and --image options to look for journal
3383 files under the specified root directory, image, or block device.
3385 * 'journalctl -o' and similar commands now implement a new output mode
3386 "short-delta". It is similar to "short-monotonic", but also shows the
3387 time delta between subsequent messages.
3389 * journalctl now respects the --quiet flag when verifying consistency
3392 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
3393 will indicate whether a message was logged in the 'initrd' phase or
3394 in the 'system' phase of the boot process.
3396 * Journal files gained a new compatibility flag
3397 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
3398 to the storage format that allow reducing size on disk. As with other
3399 compatibility flags, older journalctl versions will not be able to
3400 read journal files using this new format. The environment variable
3401 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
3402 disable this functionality. It is enabled by default.
3404 * systemd-run's --working-directory= switch now works when used in
3405 combination with --scope.
3407 * portablectl gained a --force flag to skip certain sanity checks. This
3408 is implemented using new flags accepted by systemd-portabled for the
3409 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
3410 flag now means that the attach/detach checks whether the units are
3411 already present and running will be skipped. Similarly,
3412 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
3413 image name matches the name declared inside of the image will be
3414 skipped. Callers must be sure to do those checks themselves if
3417 * systemd-portabled will now use the original filename to check
3418 extension-release.NAME for correctness, in case it is passed a
3421 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
3424 * sysext's extension-release files now support '_any' as a special
3425 value for the ID= field, to allow distribution-independent extensions
3426 (e.g.: fully statically compiled binaries, scripts). It also gained
3427 support for a new ARCHITECTURE= field that may be used to explicitly
3428 restrict an image to hosts of a specific architecture.
3430 * systemd-repart now supports creating squashfs partitions. This
3431 requires mksquashfs from squashfs-tools.
3433 * systemd-repart gained a --split flag to also generate split
3434 artifacts, i.e. a separate file for each partition. This is useful in
3435 conjunction with systemd-sysupdate or other tools, or to generate
3436 split dm-verity artifacts.
3438 * systemd-repart is now able to generate dm-verity partitions, including
3441 * systemd-repart can now set a partition UUID to zero, allowing it to
3442 be filled in later, such as when using verity partitions.
3444 * systemd-repart now supports drop-ins for its configuration files.
3446 * Package metadata logged by systemd-coredump in the system journal is
3449 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
3451 * systemd-oomd now automatically links against libatomic, if available.
3453 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
3456 * scope units now also provide oom-kill status.
3458 * systemd-pstore will now try to load only the efi_pstore kernel module
3459 before running, ensuring that pstore can be used.
3461 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
3462 session after a preconfigure timeout.
3464 * systemd-homed will now wait up to 30 seconds for workers to terminate,
3465 rather than indefinitely.
3467 * homectl gained a new '--luks-sector-size=' flag that allows users to
3468 select the preferred LUKS sector size. Must be a power of 2 between 512
3469 and 4096. systemd-userdbd records gained a corresponding field.
3471 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
3472 variable when generating the 'sp_lstchg' field, to ensure an image
3473 build can be reproducible.
3475 * 'udevadm wait' will now listen to kernel uevents too when called with
3478 * When naming network devices udev will now consult the Devicetree
3479 "alias" fields for the device.
3481 * systemd-udev will now create infiniband/by-path and
3482 infiniband/by-ibdev links for Infiniband verbs devices.
3484 * systemd-udev-trigger.service will now also prioritize input devices.
3486 * ConditionACPower= and systemd-ac-power will now assume the system is
3487 running on AC power if no battery can be found.
3489 * All features and tools using the TPM2 will now communicate with it
3490 using a bind key. Beforehand, the tpm2 support used encrypted sessions
3491 by creating a primary key that was used to encrypt traffic. This
3492 creates a problem as the key created for encrypting the traffic could
3493 be faked by an active interposer on the bus. In cases when a pin is
3494 used, a bind key will be used. The pin is used as the auth value for
3495 the seal key, aka the disk encryption key, and that auth value will be
3496 used in the session establishment. An attacker would need the pin
3497 value to create the secure session and thus an active interposer
3498 without the pin cannot interpose on TPM2 traffic.
3500 * systemd-growfs no longer requires udev to run.
3502 * systemd-backlight now will better support systems with multiple
3505 * systemd-cryptsetup's keyfile-timeout= option now also works when a
3506 device is used as a keyfile.
3508 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
3509 unlocking key from a key file (instead of prompting the user). Note
3510 that this is the key for unlocking the volume in order to be able to
3511 enroll a new key, but it is not the key that is enrolled.
3513 * systemd-dissect gained a new --umount switch that will safely and
3514 synchronously unmount all partitions of an image previously mounted
3515 with 'systemd-dissect --mount'.
3517 * When using gcrypt, all systemd tools and services will now configure
3518 it to prefer the OS random number generator if present.
3520 * All example code shipped with documentation has been relicensed from CC0
3523 * Unit tests will no longer fail when running on a system without
3526 Experimental features:
3528 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
3529 and bpftool >= 7.0).
3531 * sd-boot can automatically enroll SecureBoot keys from files found on
3532 the ESP. This enrollment can be either automatic ('force' mode) or
3533 controlled by the user ('manual' mode). It is sufficient to place the
3534 SecureBoot keys in the right place in the ESP and they will be picked
3535 up by sd-boot and shown in the boot menu.
3537 * The mkosi config in systemd gained support for automatically
3538 compiling a kernel with the configuration appropriate for testing
3539 systemd. This may be useful when developing or testing systemd in
3540 tandem with the kernel.
3542 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
3543 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
3544 Alexander Graf, Alexander Shopov, Alexander Wilson,
3545 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
3546 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
3547 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
3548 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
3549 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
3550 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
3551 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
3552 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
3553 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
3554 David Jaša, David Rheinsberg, David Seifert, David Tardon,
3555 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
3556 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
3557 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
3558 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
3559 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
3560 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
3561 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
3562 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
3563 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
3564 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
3565 JeroenHD, jiangchuangang, João Loureiro,
3566 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
3567 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
3568 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
3569 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
3570 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
3571 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
3572 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
3573 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
3574 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
3575 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
3576 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
3577 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
3578 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
3579 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
3580 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
3581 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
3582 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
3583 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
3584 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
3585 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
3586 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
3587 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
3588 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
3589 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
3591 – The Great Beyond, 2022-10-31 👻
3595 Backwards-incompatible changes:
3597 * The minimum kernel version required has been bumped from 3.13 to 4.15,
3598 and CLOCK_BOOTTIME is now assumed to always exist.
3600 * C11 with GNU extensions (aka "gnu11") is now used to build our
3601 components. Public API headers are still restricted to ISO C89.
3603 * In v250, a systemd-networkd feature that automatically configures
3604 routes to addresses specified in AllowedIPs= was added and enabled by
3605 default. However, this causes network connectivity issues in many
3606 existing setups. Hence, it has been disabled by default since
3607 systemd-stable 250.3. The feature can still be used by explicitly
3608 configuring RouteTable= setting in .netdev files.
3610 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
3611 when a Condition*= check does not succeed, restoring the JobRemoved
3612 signal to the behaviour it had before v250.
3614 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
3615 GetImageMetadataWithExtensions() have been fixed to provide an extra
3616 return parameter, containing the actual extension release metadata.
3617 The current implementation was judged to be broken and unusable, and
3618 thus the usual procedure of adding a new set of methods was skipped,
3619 and backward compatibility broken instead on the assumption that
3620 nobody can be affected given the current state of this interface.
3622 * All kernels supported by systemd mix bytes returned by RDRAND (or
3623 similar) into the entropy pool at early boot. This means that on
3624 those systems, even if /dev/urandom is not yet initialized, it still
3625 returns bytes that are of at least RDRAND quality. For that reason,
3626 we no longer have reason to invoke RDRAND from systemd itself, which
3627 has historically been a source of bugs. Furthermore, kernels ≥5.6
3628 provide the getrandom(GRND_INSECURE) interface for returning random
3629 bytes before the entropy pool is initialized without warning into
3630 kmsg, which is what we attempt to use if available. systemd's direct
3631 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
3632 running an older kernel may experience kmsg warnings that were not
3633 seen with 250. For newer kernels, non-x86 systems, or older x86
3634 systems, there should be no visible changes.
3636 * sd-boot will now measure the kernel command line into TPM PCR 12
3637 rather than PCR 8. This improves usefulness of the measurements on
3638 systems where sd-boot is chainloaded from Grub. Grub measures all
3639 commands its executes into PCR 8, which makes it very hard to use
3640 reasonably, hence separate ourselves from that and use PCR 12
3641 instead, which is what certain Ubuntu editions already do. To retain
3642 compatibility with systems running older systemd systems a new meson
3643 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
3644 If enabled, the measurement is done twice: into the new-style PCR 12
3645 *and* the old-style PCR 8. It's strongly advised to migrate all users
3646 to PCR 12 for this purpose in the long run, as we intend to remove
3647 this compatibility feature in two years' time.
3649 * busctl capture now writes output in the newer pcapng format instead
3652 * A udev rule that imported hwdb matches for USB devices with lowercase
3653 hexadecimal vendor/product ID digits was added in systemd 250. This
3654 has been reverted, since uppercase hexadecimal digits are supposed to
3655 be used, and we already had a rule with the appropriate match.
3657 Users might need to adjust their local hwdb entries.
3659 * arch_prctl(2) has been moved to the @default set in the syscall filters
3660 (as exposed via the SystemCallFilter= setting in service unit files).
3661 It is apparently used by the linker now.
3663 * The tmpfiles entries that create the /run/systemd/netif directory and
3664 its subdirectories were moved from tmpfiles.d/systemd.conf to
3665 tmpfiles.d/systemd-network.conf.
3667 Users might need to adjust their files that override tmpfiles.d/systemd.conf
3668 to account for this change.
3670 * The requirement for Portable Services images to contain a well-formed
3671 os-release file (i.e.: contain at least an ID field) is now enforced.
3672 This applies to base images and extensions, and also to systemd-sysext.
3674 Changes in the Boot Loader Specification, kernel-install and sd-boot:
3676 * kernel-install's and bootctl's Boot Loader Specification Type #1
3677 entry generation logic has been reworked. The user may now pick
3678 explicitly by which "token" string to name the installation's boot
3679 entries, via the new /etc/kernel/entry-token file or the new
3680 --entry-token= switch to bootctl. By default — as before — the
3681 entries are named after the local machine ID. However, in "golden
3682 image" environments, where the machine ID shall be initialized on
3683 first boot (as opposed to at installation time before first boot) the
3684 machine ID will not be available at build time. In this case the
3685 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
3686 file) may be used to override the "token" for the entries, for
3687 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
3688 make the OS images independent of any machine ID, and ensure that the
3689 images will not carry any identifiable information before first boot,
3690 but on the other hand means that multiple parallel installations of
3691 the very same image on the same disk cannot be supported.
3693 Summary: if you are building golden images that shall acquire
3694 identity information exclusively on first boot, make sure to both
3695 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
3696 value of the IMAGE_ID= or ID= field of /etc/os-release or another
3697 suitable identifier before deploying the image.
3699 * The Boot Loader Specification has been extended with
3700 /loader/entries.srel file located in the EFI System Partition (ESP)
3701 that disambiguates the format of the entries in the /loader/entries/
3702 directory (in order to discern them from incompatible uses of this
3703 directory by other projects). For entries that follow the
3704 Specification, the string "type1" is stored in this file.
3706 bootctl will now write this file automatically when installing the
3707 systemd-boot boot loader.
3709 * kernel-install supports a new initrd_generator= setting in
3710 /etc/kernel/install.conf, that is exported as
3711 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
3712 allows choosing different initrd generators.
3714 * kernel-install will now create a "staging area" (an initially-empty
3715 directory to gather files for a Boot Loader Specification Type #1
3716 entry). The path to this directory is exported as
3717 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
3718 drop files there instead of writing them directly to the final
3719 location. kernel-install will move them when all files have been
3720 prepared successfully.
3722 * New option sort-key= has been added to the Boot Loader Specification
3723 to override the sorting order of the entries in the boot menu. It is
3724 read by sd-boot and bootctl, and will be written by kernel-install,
3725 with the default value of IMAGE_ID= or ID= fields from
3726 os-release. Together, this means that on multiboot installations,
3727 entries should be grouped and sorted in a predictable way.
3729 * The sort order of boot entries has been updated: entries which have
3730 the new field sort-key= are sorted by it first, and all entries
3731 without it are ordered later. After that, entries are sorted by
3732 version so that newest entries are towards the beginning of the list.
3734 * The kernel-install tool gained a new 'inspect' verb which shows the
3735 paths and other settings used.
3737 * sd-boot can now optionally beep when the menu is shown and menu
3738 entries are selected, which can be useful on machines without a
3739 working display. (Controllable via a loader.conf setting.)
3741 * The --make-machine-id-directory= switch to bootctl has been replaced
3742 by --make-entry-directory=, given that the entry directory is not
3743 necessarily named after the machine ID, but after some other suitable
3744 ID as selected via --entry-token= described above. The old name of
3745 the option is still understood to maximize compatibility.
3747 * 'bootctl list' gained support for a new --json= switch to output boot
3748 menu entries in JSON format.
3750 * 'bootctl is-installed' now supports the --graceful, and various verbs
3751 omit output with the new option --quiet.
3753 Changes in systemd-homed:
3755 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
3756 of activated home directories it manages (if the kernel and selected
3757 file systems support it). So far it mapped three UID ranges: the
3758 range from 0…60000, the user's own UID, and the range 60514…65534,
3759 leaving everything else unmapped (in other words, the 16-bit UID range
3760 is mapped almost fully, with the exception of the UID subrange used
3761 for systemd-homed users, with one exception: the user's own UID).
3762 Unmapped UIDs may not be used for file ownership in the home
3763 directory — any chown() attempts with them will fail. With this
3764 release a fourth range is added to these mappings:
3765 524288…1879048191. This range is the UID range intended for container
3768 https://systemd.io/UIDS-GIDS
3770 This range may be used for container managers that place container OS
3771 trees in the home directory (which is a questionable approach, for
3772 quota, permission, SUID handling and network file system
3773 compatibility reasons, but nonetheless apparently commonplace). Note
3774 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
3775 UID assignments from the range are not managed or mapped by
3776 `systemd-homed`, and must be managed with other mechanisms, in the
3777 context of the local system.
3779 Typically, a better approach to user namespacing in relevant
3780 container managers would be to leave container OS trees on disk at
3781 UID offset 0, but then map them to a dynamically allocated runtime
3782 UID range via another UID mount map at container invocation
3783 time. That way user namespace UID ranges become strictly a runtime
3784 concept, and do not leak into persistent file systems, persistent
3785 user databases or persistent configuration, thus greatly simplifying
3786 handling, and improving compatibility with home directories intended
3787 to be portable like the ones managed by systemd-homed.
3789 Changes in shared libraries:
3791 * A new libsystemd-core-<version>.so private shared library is
3792 installed under /usr/lib/systemd/system, mirroring the existing
3793 libsystemd-shared-<version>.so library. This allows the total
3794 installation size to be reduced by binary code reuse.
3796 * The <version> tag used in the name of libsystemd-shared.so and
3797 libsystemd-core.so can be configured via the meson option
3798 'shared-lib-tag'. Distributions may build subsequent versions of the
3799 systemd package with unique tags (e.g. the full package version),
3800 thus allowing multiple installations of those shared libraries to be
3801 available at the same time. This is intended to fix an issue where
3802 programs that link to those libraries would fail to execute because
3803 they were installed earlier or later than the appropriate version of
3806 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
3807 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
3808 format instead of as a simple series of hex characters.
3810 * The sd-device API gained two new calls sd_device_new_from_devname()
3811 and sd_device_new_from_path() which permit allocating an sd_device
3812 object from a device node name or file system path.
3814 * sd-device also gained a new call sd_device_open() which will open the
3815 device node associated with a device for which an sd_device object
3816 has been allocated. The call is supposed to address races around
3817 device nodes being removed/recycled due to hotplug events, or media
3818 change events: the call checks internally whether the major/minor of
3819 the device node and the "diskseq" (in case of block devices) match
3820 with the metadata loaded in the sd_device object, thus ensuring that
3821 the device once opened really matches the provided sd_device object.
3823 Changes in PID1, systemctl, and systemd-oomd:
3825 * A new set of service monitor environment variables will be passed to
3826 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
3827 handler unit as OnFailure=/OnSuccess=. The variables are:
3828 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
3829 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
3830 handler needs to watch multiple units, use a templated handler.
3832 * A new ExtensionDirectories= setting in service unit files allows
3833 system extensions to be loaded from a directory. (It is similar to
3834 ExtensionImages=, but takes paths to directories, instead of
3837 'portablectl attach --extension=' now also accepts directory paths.
3839 * The user.delegate and user.invocation_id extended attributes on
3840 cgroups are used in addition to trusted.delegate and
3841 trusted.invocation_id. The latter pair requires privileges to set,
3842 but the former doesn't and can be also set by the unprivileged user
3845 (Only supported on kernels ≥5.6.)
3847 * Units that were killed by systemd-oomd will now have a service result
3848 of 'oom-kill'. The number of times a service was killed is tallied
3849 in the 'user.oomd_ooms' extended attribute.
3851 The OOMPolicy= unit file setting is now also honoured by
3854 * In unit files the new %y/%Y specifiers can be used to refer to
3855 normalized unit file path, which is particularly useful for symlinked
3858 The new %q specifier resolves to the pretty hostname
3859 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
3861 The new %d specifier resolves to the credentials directory of a
3862 service (same as $CREDENTIALS_DIRECTORY).
3864 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
3865 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
3866 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
3867 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
3868 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
3869 MountFlags= service settings now also work in unprivileged user
3870 services, i.e. those run by the user's --user service manager, as long
3871 as user namespaces are enabled on the system.
3873 * Services with Restart=always and a failing ExecCondition= will no
3874 longer be restarted, to bring ExecCondition= behaviour in line with
3875 Condition*= settings.
3877 * LoadCredential= now accepts a directory as the argument; all files
3878 from the directory will be loaded as credentials.
3880 * A new D-Bus property ControlGroupId is now exposed on service units,
3881 that encapsulates the service's numeric cgroup ID that newer kernels
3882 assign to each cgroup.
3884 * PID 1 gained support for configuring the "pre-timeout" of watchdog
3885 devices and the associated governor, via the new
3886 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
3887 options in /etc/systemd/system.conf.
3889 * systemctl's --timestamp= option gained a new choice "unix", to show
3890 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
3892 * A new "taint" flag named "old-kernel" is introduced which is set when
3893 the kernel systemd runs on is older then the current baseline version
3894 (see above). The flag is shown in "systemctl status" output.
3896 * Two additional taint flags "short-uid-range" and "short-gid-range"
3897 have been added as well, which are set when systemd notices it is run
3898 within a userns namespace that does not define the full 0…65535 UID
3901 * A new "unmerged-usr" taint flag has been added that is set whenever
3902 running on systems where /bin/ + /sbin/ are *not* symlinks to their
3903 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
3906 * Generators invoked by PID 1 will now have a couple of useful
3907 environment variables set describing the execution context a
3908 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
3909 system service manager, or from the per-user service
3910 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
3911 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
3912 systemd considers the current boot to be a "first"
3913 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
3914 detected and which type of hypervisor/container
3915 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
3916 kernel is built for.
3918 * PID 1 will now automatically pick up system credentials from qemu's
3919 fw_cfg interface, thus allowing passing arbitrary data into VM
3920 systems similar to how this is already supported for passing them
3921 into `systemd-nspawn` containers. Credentials may now also be passed
3922 in via the new kernel command line option `systemd.set_credential=`
3923 (note that kernel command line options are world-readable during
3924 runtime, and only useful for credentials that require no
3925 confidentiality). The credentials that can be passed to unified
3926 kernels that use the `systemd-stub` UEFI stub are now similarly
3927 picked up automatically. Automatic importing of system credentials
3928 this way can be turned off via the new
3929 `systemd.import_credentials=no` kernel command line option.
3931 * LoadCredential= will now automatically look for credentials in the
3932 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
3933 the argument is not an absolute path. Similarly,
3934 LoadCredentialEncrypted= will check the same directories plus
3935 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
3936 /usr/lib/credstore.encrypted/. The idea is to use those directories
3937 as the system-wide location for credentials that services should pick
3940 * System and service credentials are described in great detail in a new
3943 https://systemd.io/CREDENTIALS
3945 Changes in systemd-journald:
3947 * The journal JSON export format has been added to listed of stable
3948 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
3950 * journalctl --list-boots now supports JSON output and the --reverse option.
3952 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
3953 updated, BUILDING_IMAGES is new:
3955 https://systemd.io/JOURNAL_EXPORT_FORMATS
3956 https://systemd.io/BUILDING_IMAGES
3960 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
3961 calculators, etc.), the other AV production devices (DJ tables,
3962 keypads, etc.) that should accessible to the seat owner user by
3965 * udevadm trigger gained a new --prioritized-subsystem= option to
3966 process certain subsystems (and all their parent devices) earlier.
3968 systemd-udev-trigger.service now uses this new option to trigger
3969 block and TPM devices first, hopefully making the boot a bit faster.
3971 * udevadm trigger now implements --type=all, --initialized-match,
3972 --initialized-nomatch to trigger both subsystems and devices, only
3973 already-initialized devices, and only devices which haven't been
3974 initialized yet, respectively.
3976 * udevadm gained a new "wait" command for safely waiting for a specific
3977 device to show up in the udev device database. This is useful in
3978 scripts that asynchronously allocate a block device (e.g. through
3979 repartitioning, or allocating a loopback device or similar) and need
3980 to synchronize on the creation to complete.
3982 * udevadm gained a new "lock" command for locking one or more block
3983 devices while formatting it or writing a partition table to it. It is
3984 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
3985 usable in scripts dealing with block devices.
3987 * udevadm info will show a couple of additional device fields in its
3988 output, and will not apply a limited set of coloring to line types.
3990 * udevadm info --tree will now show a tree of objects (i.e. devices and
3991 suchlike) in the /sys/ hierarchy.
3993 * Block devices will now get a new set of device symlinks in
3994 /dev/disk/by-diskseq/<nr>, which may be used to reference block
3995 device nodes via the kernel's "diskseq" value. Note that this does
3996 not guarantee that opening a device by a symlink like this will
3997 guarantee that the opened device actually matches the specified
3998 diskseq value. To be safe against races, the actual diskseq value of
3999 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
4000 the one in the symlink path.
4002 * .link files gained support for setting MDI/MID-X on a link.
4004 * .link files gained support for [Match] Firmware= setting to match on
4005 the device firmware description string. By mistake, it was previously
4006 only supported in .network files.
4008 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
4009 and [SR-IOV] section to configure SR-IOV virtual functions.
4011 Changes in systemd-networkd:
4013 * The default scope for unicast routes configured through [Route]
4014 section is changed to "link", to make the behavior consistent with
4015 "ip route" command. The manual configuration of [Route] Scope= is
4018 * A new unit systemd-networkd-wait-online@<interface>.service has been
4019 added that can be used to wait for a specific network interface to be
4022 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
4023 that configures the eponymous kernel attribute on the bridge.
4025 * .netdev files now can be used to create virtual WLAN devices, and
4026 configure various settings on them, via the [WLAN] section.
4028 * .link/.network files gained support for [Match] Kind= setting to match
4029 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
4031 This value is also shown by 'networkctl status'.
4033 * The Local= setting in .netdev files for various virtual network
4034 devices gained support for specifying, in addition to the network
4035 address, the name of a local interface which must have the specified
4038 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
4039 files, to configure tunnels in external mode (a.k.a. collect metadata
4042 * [Network] L2TP= setting was removed. Please use interface specifier in
4043 Local= setting in .netdev files of corresponding L2TP interface.
4045 * New [DHCPServer] BootServerName=, BootServerAddress=, and
4046 BootFilename= settings can be used to configure the server address,
4047 server name, and file name sent in the DHCP packet (e.g. to configure
4050 Changes in systemd-resolved:
4052 * systemd-resolved is started earlier (in sysinit.target), so it
4053 available earlier and will also be started in the initrd if installed
4056 Changes in disk encryption:
4058 * systemd-cryptenroll can now control whether to require the user to
4059 enter a PIN when using TPM-based unlocking of a volume via the new
4060 --tpm2-with-pin= option.
4062 Option tpm2-pin= can be used in /etc/crypttab.
4064 * When unlocking devices via TPM, TPM2 parameter encryption is now
4065 used, to ensure that communication between CPU and discrete TPM chips
4066 cannot be eavesdropped to acquire disk encryption keys.
4068 * A new switch --fido2-credential-algorithm= has been added to
4069 systemd-cryptenroll allowing selection of the credential algorithm to
4070 use when binding encryption to FIDO2 tokens.
4072 Changes in systemd-hostnamed:
4074 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
4075 to override the values gleaned from the hwdb.
4077 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
4078 /sys/class/dmi/id) to override the chassis that is reported by
4081 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
4082 for reading the hardware serial number, as reportd by DMI. It also
4083 exposes a new method D-Bus property FirmwareVersion that encode the
4084 firmware version of the system.
4086 Changes in other components:
4088 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
4089 handling with the values that were configured during systemd build
4090 (if /etc/locale.conf has not been created through some other
4091 mechanism). This means that /etc/locale.conf should always have
4092 reasonable contents and we avoid a potential mismatch in defaults.
4094 * The userdbctl tool will now show UID range information as part of the
4095 list of known users.
4097 * A new build-time configuration setting default-user-shell= can be
4098 used to set the default shell for user records and nspawn shell
4099 invocations (instead of the default /bin/bash).
4101 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
4102 information dynamically at runtime via IPC.
4104 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
4105 whether a functioning TPM2 infrastructure is available, i.e. if
4106 firmware, kernel driver and systemd all have TPM2 support enabled and
4109 * The systemd-creds tool gained support for generating encrypted
4110 credentials that are using an empty encryption key. While this
4111 provides no integrity nor confidentiality it's useful to implement
4112 codeflows that work the same on TPM-ful and TPM2-less systems. The
4113 service manager will only accept credentials "encrypted" that way if
4114 a TPM2 device cannot be detected, to ensure that credentials
4115 "encrypted" like that cannot be used to trick TPM2 systems.
4117 * When deciding whether to colorize output, all systemd programs now
4118 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
4121 * Meson's new install_tag feature is now in use for several components,
4122 allowing to build and install select binaries only: pam, nss, devel
4123 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
4124 $ meson build systemd-boot
4125 $ meson install --tags systemd-boot --no-rebuild
4126 https://mesonbuild.com/Installing.html#installation-tags
4128 * A new build configuration option has been added, to allow selecting the
4129 default compression algorithm used by systemd-journald and systemd-coredump.
4130 This allows to build-in support for decompressing all supported formats,
4131 but choose a specific one for compression. E.g.:
4132 $ meson -Ddefault-compression=xz
4134 Experimental features:
4136 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
4137 loader.conf that implements booting Microsoft Windows from the
4138 sd-boot in a way that first reboots the system, to reset the TPM
4139 PCRs. This improves compatibility with BitLocker's TPM use, as the
4140 PCRs will only record the Windows boot process, and not sd-boot
4141 itself, thus retaining the PCR measurements not involving sd-boot.
4142 Note that this feature is experimental for now, and is likely going
4143 to be generalized and renamed in a future release, without retaining
4144 compatibility with the current implementation.
4146 * A new systemd-sysupdate component has been added that automatically
4147 discovers, downloads, and installs A/B-style updates for the host
4148 installation itself, or container images, portable service images,
4149 and other assets. See the new systemd-sysupdate man page for updates.
4151 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
4152 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
4153 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
4154 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
4155 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
4156 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
4157 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
4158 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
4159 David, David Bond, Davide Cavalca, David Tardon, davijosw,
4160 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
4161 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
4162 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
4163 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
4164 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
4165 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
4166 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
4167 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
4168 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
4169 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
4170 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
4171 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
4172 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
4173 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
4174 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
4175 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
4176 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
4177 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
4178 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
4179 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
4180 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
4181 Simon Ellmann, Sonali Srivastava, Stefan Seering,
4182 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
4183 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
4184 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
4185 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
4186 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
4187 Zbigniew Jędrzejewski-Szmek, наб
4189 — Edinburgh, 2022-05-21
4193 * Support for encrypted and authenticated credentials has been added.
4194 This extends the credential logic introduced with v247 to support
4195 non-interactive symmetric encryption and authentication, based on a
4196 key that is stored on the /var/ file system or in the TPM2 chip (if
4197 available), or the combination of both (by default if a TPM2 chip
4198 exists the combination is used, otherwise the /var/ key only). The
4199 credentials are automatically decrypted at the moment a service is
4200 started, and are made accessible to the service itself in unencrypted
4201 form. A new tool 'systemd-creds' encrypts credentials for this
4202 purpose, and two new service file settings LoadCredentialEncrypted=
4203 and SetCredentialEncrypted= configure such credentials.
4205 This feature is useful to store sensitive material such as SSL
4206 certificates, passwords and similar securely at rest and only decrypt
4207 them when needed, and in a way that is tied to the local OS
4208 installation or hardware.
4210 * systemd-gpt-auto-generator can now automatically set up discoverable
4211 LUKS2 encrypted swap partitions.
4213 * The GPT Discoverable Partitions Specification has been substantially
4214 extended with support for root and /usr/ partitions for the majority
4215 of architectures systemd supports. This includes platforms that do
4216 not natively support UEFI, because even though GPT is specified under
4217 UEFI umbrella, it is useful on other systems too. Specifically,
4218 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
4219 Portable Services use the concept without requiring UEFI.
4221 * The GPT Discoverable Partitions Specifications has been extended with
4222 a new set of partitions that may carry PKCS#7 signatures for Verity
4223 partitions, encoded in a simple JSON format. This implements a simple
4224 mechanism for building disk images that are fully authenticated and
4225 can be tested against a set of cryptographic certificates. This is
4226 now implemented for the various systemd tools that can operate with
4227 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
4228 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
4229 The PKCS#7 signatures are passed to the kernel (where they are
4230 checked against certificates from the kernel keyring), or can be
4231 verified against certificates provided in userspace (via a simple
4232 drop-in file mechanism).
4234 * systemd-dissect's inspection logic will now report for which uses a
4235 disk image is intended. Specifically, it will display whether an
4236 image is suitable for booting on UEFI or in a container (using
4237 systemd-nspawn's --image= switch), whether it can be used as portable
4238 service, or attached as system extension.
4240 * The system-extension.d/ drop-in files now support a new field
4241 SYSEXT_SCOPE= that may encode which purpose a system extension image
4242 is for: one of "initrd", "system" or "portable". This is useful to
4243 make images more self-descriptive, and to ensure system extensions
4244 cannot be attached in the wrong contexts.
4246 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
4247 be used in portable service images to indicate which unit prefixes
4250 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
4251 now is able to decode images for non-native architectures as well.
4252 This allows systemd-nspawn to boot images of non-native architectures
4253 if the corresponding user mode emulator is installed and
4254 systemd-binfmtd is running.
4256 * systemd-logind gained new settings HandlePowerKeyLongPress=,
4257 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
4258 HandleHibernateKeyLongPress= which may be used to configure actions
4259 when the relevant keys are pressed for more than 5s. This is useful
4260 on devices that only have hardware for a subset of these keys. By
4261 default, if the reboot key is pressed long the poweroff operation is
4262 now triggered, and when the suspend key is pressed long the hibernate
4263 operation is triggered. Long pressing the other two keys currently
4264 does not trigger any operation by default.
4266 * When showing unit status updates on the console during boot and
4267 shutdown, and a service is slow to start so that the cylon animation
4268 is shown, the most recent sd_notify() STATUS= text is now shown as
4269 well. Services may use this to make the boot/shutdown output easier
4270 to understand, and to indicate what precisely a service that is slow
4271 to start or stop is waiting for. In particular, the per-user service
4272 manager instance now reports what it is doing and which service it is
4273 waiting for this way to the system service manager.
4275 * The service manager will now re-execute on reception of the
4276 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
4277 only when running as PID 1. There was no signal to request this when
4278 running as per-user service manager, i.e. as any other PID than 1.
4279 SIGRTMIN+25 works for both system and user managers.
4281 * The hardware watchdog logic in PID 1 gained support for operating
4282 with the default timeout configured in the hardware, instead of
4283 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
4284 request this behavior.
4286 * A new kernel command line option systemd.watchdog_sec= is now
4287 understood which may be used to override the hardware watchdog
4288 time-out for the boot.
4290 * A new setting DefaultOOMScoreAdjust= is now supported in
4291 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
4292 to set the default process OOM score adjustment value for processes
4293 started by the service manager. For per-user service managers this
4294 now defaults to 100, but for per-system service managers is left as
4295 is. This means that by default now services forked off the user
4296 service manager are more likely to be killed by the OOM killer than
4297 system services or the managers themselves.
4299 * A new per-service setting RestrictFileSystems= as been added that
4300 restricts the file systems a service has access to by their type.
4301 This is based on the new BPF LSM of the Linux kernel. It provides an
4302 effective way to make certain API file systems unavailable to
4303 services (and thus minimizing attack surface). A new command
4304 "systemd-analyze filesystems" has been added that lists all known
4305 file system types (and how they are grouped together under useful
4308 * Services now support a new setting RestrictNetworkInterfaces= for
4309 restricting access to specific network interfaces.
4311 * Service unit files gained new settings StartupAllowedCPUs= and
4312 StartupAllowedMemoryNodes=. These are similar to their counterparts
4313 without the "Startup" prefix and apply during the boot process
4314 only. This is useful to improve boot-time behavior of the system and
4315 assign resources differently during boot than during regular
4316 runtime. This is similar to the preexisting StartupCPUWeight=
4319 * Related to this: the various StartupXYZ= settings
4320 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
4321 during shutdown. The settings not prefixed with "Startup" hence apply
4322 during regular runtime, and those that are prefixed like that apply
4323 during boot and shutdown.
4325 * A new per-unit set of conditions/asserts
4326 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
4327 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
4328 pressure is above the configured threshold, using the kernel PSI
4329 feature. For more details see systemd.unit(5) and
4330 https://docs.kernel.org/accounting/psi.html
4332 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
4333 ProtectKernelLogs=yes can now be used.
4335 * The default maximum numbers of inodes have been raised from 64k to 1M
4336 for /dev/, and from 400k to 1M for /tmp/.
4338 * The per-user service manager learnt support for communicating with
4339 systemd-oomd to acquire OOM kill information.
4341 * A new service setting ExecSearchPath= has been added that allows
4342 changing the search path for executables for services. It affects
4343 where we look for the binaries specified in ExecStart= and similar,
4344 and the specified directories are also added the $PATH environment
4345 variable passed to invoked processes.
4347 * A new setting RuntimeRandomizedExtraSec= has been added for service
4348 and scope units that allows extending the runtime time-out as
4349 configured by RuntimeMaxSec= with a randomized amount.
4351 * The syntax of the service unit settings RuntimeDirectory=,
4352 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
4353 if the specified value is now suffixed with a colon, followed by
4354 another filename, the latter will be created as symbolic link to the
4355 specified directory. This allows creating these service directories
4356 together with alias symlinks to make them available under multiple
4359 * Service unit files gained two new settings TTYRows=/TTYColumns= for
4360 configuring rows/columns of the TTY device passed to
4361 stdin/stdout/stderr of the service. This is useful to propagate TTY
4362 dimensions to a virtual machine.
4364 * A new service unit file setting ExitType= has been added that
4365 specifies when to assume a service has exited. By default systemd
4366 only watches the main process of a service. By setting
4367 ExitType=cgroup it can be told to wait for the last process in a
4370 * Automount unit files gained a new setting ExtraOptions= that can be
4371 used to configure additional mount options to pass to the kernel when
4372 mounting the autofs instance.
4374 * "Urlification" (generation of ESC sequences that generate clickable
4375 hyperlinks in modern terminals) may now be turned off altogether
4378 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
4379 settings that default to 200 and 2 s respectively. The ratelimit
4380 ensures that a path unit cannot cause PID1 to busy-loop when it is
4381 trying to trigger a service that is skipped because of a Condition*=
4382 not being satisfied. This matches the configuration and behaviour of
4385 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
4386 as a plug-in for cryptsetup. This means the plain cryptsetup command
4387 may now be used to unlock volumes set up this way.
4389 * The TPM2 logic in cryptsetup will now automatically detect systems
4390 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
4391 updates the SHA1 banks. In such a case PCR policies will be
4392 automatically bound to the latter, not the former. This makes the PCR
4393 policies reliable, but of course do not provide the same level of
4394 trust as SHA256 banks.
4396 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
4397 RSA primary keys in addition to ECC, improving compatibility with
4398 TPM2 chips that do not support ECC. RSA keys are much slower to use
4399 than ECC, and hence are only used if ECC is not available.
4401 * /etc/crypttab gained support for a new token-timeout= setting for
4402 encrypted volumes that allows configuration of the maximum time to
4403 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
4404 the logic will query the user for a regular passphrase/recovery key
4407 * Support for activating dm-integrity volumes at boot via a new file
4408 /etc/integritytab and the tool systemd-integritysetup have been
4409 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
4410 with dm-integrity instead of dm-crypt/dm-verity.
4412 * The systemd-veritysetup-generator now understands a new usrhash=
4413 kernel command line option for specifying the Verity root hash for
4414 the partition backing the /usr/ file system. A matching set of
4415 systemd.verity_usr_* kernel command line options has been added as
4416 well. These all work similar to the corresponding options for the
4419 * The sd-device API gained a new API call sd_device_get_diskseq() to
4420 return the DISKSEQ property of a device structure. The "disk
4421 sequence" concept is a new feature recently introduced to the Linux
4422 kernel that allows detecting reuse cycles of block devices, i.e. can
4423 be used to recognize when loopback block devices are reused for a
4424 different purpose or CD-ROM drives get their media changed.
4426 * A new unit systemd-boot-update.service has been added. If enabled
4427 (the default) and the sd-boot loader is detected to be installed, it
4428 is automatically updated to the newest version when out of date. This
4429 is useful to ensure the boot loader remains up-to-date, and updates
4430 automatically propagate from the OS tree in /usr/.
4432 * sd-boot will now build with SBAT by default in order to facilitate
4433 working with recent versions of Shim that require it to be present.
4435 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
4436 This is used to robustly generate boot entry titles for Windows.
4438 * A new generic target unit factory-reset.target has been added. It is
4439 hooked into systemd-logind similar in fashion to
4440 reboot/poweroff/suspend/hibernate, and is supposed to be used to
4441 initiate a factory reset operation. What precisely this operation
4442 entails is up for the implementer to decide, the primary goal of the
4443 new unit is provide a framework where to plug in the implementation
4444 and how to trigger it.
4446 * A new meson build-time option 'clock-valid-range-usec-max' has been
4447 added which takes a time in µs and defaults to 15 years. If the RTC
4448 time is noticed to be more than the specified time ahead of the
4449 built-in epoch of systemd (which by default is the release timestamp
4450 of systemd) it is assumed that the RTC is not working correctly, and
4451 the RTC is reset to the epoch. (It already is reset to the epoch when
4452 noticed to be before it.) This should increase the chance that time
4453 doesn't accidentally jump too far ahead due to faulty hardware or
4456 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
4457 which may be used to automatically save the current system time to
4458 disk in regular intervals. This is useful to maintain a roughly
4459 monotonic clock even without RTC hardware and with some robustness
4460 against abnormal system shutdown.
4462 * systemd-analyze verify gained support for a pair of new --image= +
4463 --root= switches for verifying units below a specific root
4464 directory/image instead of on the host.
4466 * systemd-analyze verify gained support for verifying unit files under
4467 an explicitly specified unit name, independently of what the filename
4470 * systemd-analyze verify gained a new switch --recursive-errors= which
4471 controls whether to only fail on errors found in the specified units
4472 or recursively any dependent units.
4474 * systemd-analyze security now supports a new --offline mode for
4475 analyzing unit files stored on disk instead of loaded units. It may
4476 be combined with --root=/--image to analyze unit files under a root
4477 directory or disk image. It also learnt a new --threshold= parameter
4478 for specifying an exposure level threshold: if the exposure level
4479 exceeds the specified value the call will fail. It also gained a new
4480 --security-policy= switch for configuring security policies to
4481 enforce on the units. A policy is a JSON file that lists which tests
4482 shall be weighted how much to determine the overall exposure
4483 level. Altogether these new features are useful for fully automatic
4484 analysis and enforcement of security policies on unit files.
4486 * systemd-analyze security gain a new --json= switch for JSON output.
4488 * systemd-analyze learnt a new --quiet switch for reducing
4489 non-essential output. It's honored by the "dot", "syscall-filter",
4490 "filesystems" commands.
4492 * systemd-analyze security gained a --profile= option that can be used
4493 to take into account a portable profile when analyzing portable
4494 services, since a lot of the security-related settings are enabled
4497 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
4498 files, binaries and executables and prints metadata information,
4499 including the build-id and other info described on:
4500 https://systemd.io/COREDUMP_PACKAGE_METADATA/
4502 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
4503 section, for automatically propagating DNS settings from other
4506 * The static lease DHCP server logic in systemd-networkd may now serve
4507 IP addresses outside of the configured IP pool range for the server.
4509 * CAN support in systemd-networkd gained four new settings Loopback=,
4510 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
4511 control modes. It gained a number of further settings for tweaking
4514 * The [CAN] section in .network file gained new TimeQuantaNSec=,
4515 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
4516 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
4517 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
4518 DataSyncJumpWidth= settings to control bit-timing processed by the
4521 * DHCPv4 client support in systemd-networkd learnt a new Label= option
4522 for configuring the address label to apply to configure IPv4
4525 * The [IPv6AcceptRA] section of .network files gained support for a new
4526 UseMTU= setting that may be used to control whether to apply the
4527 announced MTU settings to the local interface.
4529 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
4530 setting to control whether the DHCPv4 client request and process the
4533 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
4534 [DHCPPrefixDelegation], as now the prefix delegation is also supported
4535 with DHCPv4 protocol by enabling the Use6RD= setting.
4537 * The [DHCPPrefixDelegation] section in .network file gained a new
4538 setting UplinkInterface= to specify the upstream interface.
4540 * The [DHCPv6] section in .network file gained a new setting
4541 UseDelegatedPrefix= to control whether the delegated prefixes will be
4542 propagated to the downstream interfaces.
4544 * The [IPv6AcceptRA] section of .network files now understands two new
4545 settings UseGateway=/UseRoutePrefix= for explicitly configuring
4546 whether to use the relevant fields from the IPv6 Router Advertisement
4549 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
4550 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
4551 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
4552 [IPv6AcceptRA] section to control when the DHCPv6 client is started
4553 and how the delegated prefixes are handled by the DHCPv6 client.
4555 * The IPv6Token= section in the [Network] section is deprecated, and
4556 the [IPv6AcceptRA] section gained the Token= setting for its
4557 replacement. The [IPv6Prefix] section also gained the Token= setting.
4558 The Token= setting gained 'eui64' mode to explicitly configure an
4559 address with the EUI64 algorithm based on the interface MAC address.
4560 The 'prefixstable' mode can now optionally take a secret key. The
4561 Token= setting in the [DHCPPrefixDelegation] section now supports all
4562 algorithms supported by the same settings in the other sections.
4564 * The [RoutingPolicyRule] section of .network file gained a new
4565 SuppressInterfaceGroup= setting.
4567 * The IgnoreCarrierLoss= setting in the [Network] section of .network
4568 files now allows a duration to be specified, controlling how long to
4569 wait before reacting to carrier loss.
4571 * The [DHCPServer] section of .network file gained a new Router=
4572 setting to specify the router address.
4574 * The [CAKE] section of .network files gained various new settings
4575 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
4576 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
4577 and UseRawPacketSize= for configuring CAKE.
4579 * systemd-networkd now ships with new default .network files:
4580 80-container-vb.network which matches host-side network bridge device
4581 created by systemd-nspawn's --network-bridge or --network-zone
4582 switch, and 80-6rd-tunnel.network which matches automatically created
4583 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
4585 * systemd-networkd's handling of Endpoint= resolution for WireGuard
4586 interfaces has been improved.
4588 * systemd-networkd will now automatically configure routes to addresses
4589 specified in AllowedIPs=. This feature can be controlled via
4590 RouteTable= and RouteMetric= settings in [WireGuard] or
4591 [WireGuardPeer] sections.
4593 * systemd-networkd will now once again automatically generate persistent
4594 MAC addresses for batadv and bridge interfaces. Users can disable this
4595 by using MACAddress=none in .netdev files.
4597 * systemd-networkd and systemd-udevd now support IP over InfiniBand
4598 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
4599 systemd.netdev files gained the [IPoIB] section.
4601 * systemd-networkd and systemd-udevd now support net.ifname_policy=
4602 option on the kernel command-line. This is implemented through the
4603 systemd-network-generator service that automatically generates
4604 appropriate .link, .network, and .netdev files.
4606 * The various systemd-udevd "ethtool" buffer settings now understand
4607 the special value "max" to configure the buffers to the maximum the
4610 * systemd-udevd's .link files may now configure a large variety of
4611 NIC coalescing settings, plus more hardware offload settings.
4613 * .link files gained a new WakeOnLanPassword= setting in the [Link]
4614 section that allows to specify a WoL "SecureOn" password on hardware
4617 * systemd-nspawn's --setenv= switch now supports an additional syntax:
4618 if only a variable name is specified (i.e. without being suffixed by
4619 a '=' character and a value) the current value of the environment
4620 variable is propagated to the container. e.g. --setenv=FOO will
4621 lookup the current value of $FOO in the environment, and pass it down
4622 to the container. Similar behavior has been added to homectl's,
4623 machinectl's and systemd-run's --setenv= switch.
4625 * systemd-nspawn gained a new switch --suppress-sync= which may be used
4626 to optionally suppress the effect of the sync()/fsync()/fdatasync()
4627 system calls for the container payload. This is useful for build
4628 system environments where safety against abnormal system shutdown is
4629 not essential as all build artifacts can be regenerated any time, but
4630 the performance win is beneficial.
4632 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
4633 same value that PID 1 uses for most forked off processes.
4635 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
4636 uidmap/nouidmap options as last parameter. If "uidmap" is used the
4637 bind mounts are created with UID mapping taking place that ensures
4638 the host's file ownerships are mapped 1:1 to container file
4639 ownerships, even if user namespacing is used. This way
4640 files/directories bound into containers will no longer show up as
4641 owned by the nobody user as they typically did if no special care was
4642 taken to shift them manually.
4644 * When discovering Windows installations sd-boot will now attempt to
4645 show the Windows version.
4647 * The color scheme to use in sd-boot may now be configured at
4650 * sd-boot gained the ability to change screen resolution during
4651 boot-time, by hitting the "r" key. This will cycle through available
4652 resolutions and save the last selection.
4654 * sd-boot learnt a new hotkey "f". When pressed the system will enter
4655 firmware setup. This is useful in environments where it is difficult
4656 to hit the right keys early enough to enter the firmware, and works
4657 on any firmware regardless which key it natively uses.
4659 * sd-boot gained support for automatically booting into the menu item
4660 selected on the last boot (using the "@saved" identifier for menu
4663 * sd-boot gained support for automatically loading all EFI drivers
4664 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
4665 Partition (ESP). These drivers are loaded before the menu entries are
4666 loaded. This is useful e.g. to load additional file system drivers
4667 for the XBOOTLDR partition.
4669 * systemd-boot will now paint the input cursor on its own instead of
4670 relying on the firmware to do so, increasing compatibility with broken
4671 firmware that doesn't make the cursor reasonably visible.
4673 * sd-boot now embeds a .osrel PE section like we expect from Boot
4674 Loader Specification Type #2 Unified Kernels. This means sd-boot
4675 itself may be used in place of a Type #2 Unified Kernel. This is
4676 useful for debugging purposes as it allows chain-loading one a
4677 (development) sd-boot instance from another.
4679 * sd-boot now supports a new "devicetree" field in Boot Loader
4680 Specification Type #1 entries: if configured the specified device
4681 tree file is installed before the kernel is invoked. This is useful
4682 for installing/applying new devicetree files without updating the
4685 * Similarly, sd-stub now can read devicetree data from a PE section
4686 ".dtb" and apply it before invoking the kernel.
4688 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
4689 gained the ability to pick up credentials and sysext files, wrap them
4690 in a cpio archive, and pass as an additional initrd to the invoked
4691 Linux kernel, in effect placing those files in the /.extra/ directory
4692 of the initrd environment. This is useful to implement trusted initrd
4693 environments which are fully authenticated but still can be extended
4694 (via sysexts) and parameterized (via encrypted/authenticated
4695 credentials, see above).
4697 Credentials can be located next to the kernel image file (credentials
4698 specific to a single boot entry), or in one of the shared directories
4699 (credentials applicable to multiple boot entries).
4701 * sd-stub now comes with a full man page, that explains its feature set
4702 and how to combine a kernel image, an initrd and the stub to build a
4703 complete EFI unified kernel image, implementing Boot Loader
4704 Specification Type #2.
4706 * sd-stub may now provide the initrd to the executed kernel via the
4707 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
4708 non-x86 architectures.
4710 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
4711 may be used to set the boot menu time-out of the boot loader (for all
4712 or just the subsequent boot).
4714 * bootctl and kernel-install will now read variables
4715 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
4716 /etc/kernel/install.conf. When set, it specifies the layout to use
4717 for installation directories on the boot partition, so that tools
4718 don't need to guess it based on the already-existing directories. The
4719 only value that is defined natively is "bls", corresponding to the
4721 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
4722 kernel-install that implement a different layout can declare other
4723 values for this variable.
4725 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
4726 assumption that if the user installed sd-boot to the ESP, they intend
4727 to use the entry layout understood by sd-boot. It'll also write
4728 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
4729 (and it wasn't specified in the config file yet). Similarly,
4730 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
4731 wasn't specified in the config file yet). Effectively, those changes
4732 mean that the machine-id used for boot loader entry installation is
4733 "frozen" upon first use and becomes independent of the actual
4736 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
4737 images created for distribution ("golden images") are built with no
4738 machine-id, so that a unique machine-id can be created on the first
4739 boot. But those images may contain boot loader entries with the
4740 machine-id used during build included in paths. Using a "frozen"
4741 value allows unambiguously identifying entries that match the
4742 specific installation, while still permitting parallel installations
4745 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
4746 kernel-install to guess the installation layout. This fixes the
4747 problem where a (possibly empty) directory in the boot partition is
4748 created from a different layout causing kernel-install plugins to
4749 assume the wrong layout. A particular example of how this may happen
4750 is the grub2 package in Fedora which includes directories under /boot
4751 directly in its file list. Various other packages pull in grub2 as a
4752 dependency, so it may be installed even if unused, breaking
4753 installations that use the bls layout.
4755 * bootctl and systemd-bless-boot can now be linked statically.
4757 * systemd-sysext now optionally doesn't insist on extension-release.d/
4758 files being placed in the image under the image's file name. If the
4759 file system xattr user.extension-release.strict is set on the
4760 extension release file, it is accepted regardless of its name. This
4761 relaxes security restrictions a bit, as system extension may be
4762 attached under a wrong name this way.
4764 * udevadm's test-builtin command learnt a new --action= switch for
4765 testing the built-in with the specified action (in place of the
4768 * udevadm info gained new switches --property=/--value for showing only
4769 specific udev properties/values instead of all.
4771 * A new hwdb database has been added that contains matches for various
4772 types of signal analyzers (protocol analyzers, logic analyzers,
4773 oscilloscopes, multimeters, bench power supplies, etc.) that should
4774 be accessible to regular users.
4776 * A new hwdb database entry has been added that carries information
4777 about types of cameras (regular or infrared), and in which direction
4778 they point (front or back).
4780 * A new rule to allow console users access to rfkill by default has been
4783 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
4784 now also owned by the system group "sgx".
4786 * A new build-time meson option "extra-net-naming-schemes=" has been
4787 added to define additional naming schemes for udev's network
4788 interface naming logic. This is useful for enterprise distributions
4789 and similar which want to pin the schemes of certain distribution
4790 releases under a specific name and previously had to patch the
4791 sources to introduce new named schemes.
4793 * The predictable naming logic for network interfaces has been extended
4794 to generate stable names from Xen netfront device information.
4796 * hostnamed's chassis property can now be sourced from chassis-type
4797 field encoded in devicetree (in addition to the existing DMI
4800 * systemd-cgls now optionally displays cgroup IDs and extended
4801 attributes for each cgroup. (Controllable via the new --xattr= +
4802 --cgroup-id= switches.)
4804 * coredumpctl gained a new --all switch for operating on all
4805 Journal files instead of just the local ones.
4807 * systemd-coredump will now use libdw/libelf via dlopen() rather than
4808 directly linking, allowing users to easily opt-out of backtrace/metadata
4809 analysis of core files, and reduce image sizes when this is not needed.
4811 * systemd-coredump will now analyze core files with libdw/libelf in a
4812 forked, sandboxed process.
4814 * systemd-homed will now try to unmount an activate home area in
4815 regular intervals once the user logged out fully. Previously this was
4816 attempted exactly once but if the home directory was busy for some
4817 reason it was not tried again.
4819 * systemd-homed's LUKS2 home area backend will now create a BSD file
4820 system lock on the image file while the home area is active
4821 (i.e. mounted). If a home area is found to be locked, logins are
4822 politely refused. This should improve behavior when using home areas
4823 images that are accessible via the network from multiple clients, and
4824 reduce the chance of accidental file system corruption in that case.
4826 * Optionally, systemd-homed will now drop the kernel buffer cache once
4827 a user has fully logged out, configurable via the new --drop-caches=
4830 * systemd-homed now makes use of UID mapped mounts for the home areas.
4831 If the kernel and used file system support it, files are now
4832 internally owned by the "nobody" user (i.e. the user typically used
4833 for indicating "this ownership is not mapped"), and dynamically
4834 mapped to the UID used locally on the system via the UID mapping
4835 mount logic of recent kernels. This makes migrating home areas
4836 between different systems cheaper because recursively chown()ing file
4837 system trees is no longer necessary.
4839 * systemd-homed's CIFS backend now optionally supports CIFS service
4840 names with a directory suffix, in order to place home directories in
4841 a subdirectory of a CIFS share, instead of the top-level directory.
4843 * systemd-homed's CIFS backend gained support for specifying additional
4844 mount options in the JSON user record (cifsExtraMountOptions field,
4845 and --cifs-extra-mount-options= homectl switch). This is for example
4846 useful for configuring mount options such as "noserverino" that some
4847 SMB3 services require (use that to run a homed home directory from a
4848 FritzBox SMB3 share this way).
4850 * systemd-homed will now default to btrfs' zstd compression for home
4851 areas. This is inspired by Fedora's recent decision to switch to zstd
4854 * Additional mount options to use when mounting the file system of
4855 LUKS2 volumes in systemd-homed has been added. Via the
4856 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
4857 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
4858 systemd-homed or via the luksExtraMountOptions user record JSON
4859 property. (Exposed via homectl --luks-extra-mount-options)
4861 * homectl's resize command now takes the special size specifications
4862 "min" and "max" to shrink/grow the home area to the minimum/maximum
4863 size possible, taking disk usage/space constraints and file system
4864 limitations into account. Resizing is now generally graceful: the
4865 logic will try to get as close to the specified size as possible, but
4866 not consider it a failure if the request couldn't be fulfilled
4869 * systemd-homed gained the ability to automatically shrink home areas
4870 on logout to their minimal size and grow them again on next
4871 login. This ensures that while inactive, a home area only takes up
4872 the minimal space necessary, but once activated, it provides
4873 sufficient space for the user's needs. This behavior is only
4874 supported if btrfs is used as file system inside the home area
4875 (because only for btrfs online growing/shrinking is implemented in
4876 the kernel). This behavior is now enabled by default, but may be
4877 controlled via the new --auto-resize-mode= setting of homectl.
4879 * systemd-homed gained support for automatically re-balancing free disk
4880 space among active home areas, in case the LUKS2 backends are used,
4881 and no explicit disk size was requested. This way disk space is
4882 automatically managed and home areas resized in regular intervals and
4883 manual resizing when disk space becomes scarce should not be
4884 necessary anymore. This behavior is only supported if btrfs is used
4885 within the home areas (as only then online shrinking and growing is
4886 supported), and may be configured via the new rebalanceWeight JSON
4887 user record field (as exposed via the new --rebalance-weight= homectl
4888 setting). Re-balancing is mostly automatic, but can also be requested
4889 explicitly via "homectl rebalance", which is synchronous, and thus
4890 may be used to wait until the rebalance run is complete.
4892 * userdbctl gained a --json= switch for configured the JSON formatting
4893 to use when outputting user or group records.
4895 * userdbctl gained a new --multiplexer= switch for explicitly
4896 configuring whether to use the systemd-userdbd server side user
4897 record resolution logic.
4899 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
4900 for chaining up another command to execute after completing the
4901 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
4902 configuration of a single command to invoke, this maybe used to
4903 invoke multiple: first userdbctl's own implementation, and then any
4904 other also configured in the command line.
4906 * The sd-event API gained a new function sd_event_add_inotify_fd() that
4907 is similar to sd_event_add_inotify() but accepts a file descriptor
4908 instead of a path in the file system for referencing the inode to
4911 * The sd-event API gained a new function
4912 sd_event_source_set_ratelimit_expire_callback() that may be used to
4913 define a callback function that is called whenever an event source
4914 leaves the rate limiting phase.
4916 * New documentation has been added explaining which steps are necessary
4917 to port systemd to a new architecture:
4919 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
4921 * The x-systemd.makefs option in /etc/fstab now explicitly supports
4922 ext2, ext3, and f2fs file systems.
4924 * Mount units and units generated from /etc/fstab entries with 'noauto'
4925 are now ordered the same as other units. Effectively, they will be
4926 started earlier (if something actually pulled them in) and stopped
4927 later, similarly to normal mount units that are part of
4928 fs-local.target. This change should be invisible to users, but
4929 should prevent those units from being stopped too early during
4932 * The systemd-getty-generator now honors a new kernel command line
4933 argument systemd.getty_auto= and a new environment variable
4934 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
4935 example useful to turn off gettys inside of containers or similar
4938 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
4939 (in addition to 127.0.0.53, as before). If DNS requests are sent to
4940 this address they are propagated in "bypass" mode only, i.e. are
4941 almost not processed locally, but mostly forwarded as-is to the
4942 current upstream DNS servers. This provides a stable DNS server
4943 address that proxies all requests dynamically to the right upstream
4944 DNS servers even if these dynamically change. This stub does not do
4945 mDNS/LLMNR resolution. However, it will translate look-ups to
4946 DNS-over-TLS if necessary. This new stub is particularly useful in
4947 container/VM environments, or for tethering setups: use DNAT to
4948 redirect traffic to any IP address to this stub.
4950 * systemd-importd now honors new environment variables
4951 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
4952 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
4953 generation, btrfs quota setup and disk synchronization.
4955 * systemd-importd and systemd-resolved can now be optionally built with
4956 OpenSSL instead of libgcrypt.
4958 * systemd-repart no longer requires OpenSSL.
4960 * systemd-sysusers will no longer create the redundant 'nobody' group
4961 by default, as the 'nobody' user is already created with an
4962 appropriate primary group.
4964 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
4966 * systemctl show-environment gained support for --output=json.
4968 * pam_systemd will now first try to use the X11 abstract socket, and
4969 fallback to the socket file in /tmp/.X11-unix/ only if that does not
4972 * systemd-journald will no longer go back to volatile storage
4973 regardless of configuration when its unit is restarted.
4975 * Initial support for the LoongArch architecture has been added (system
4976 call lists, GPT partition table UUIDs, etc).
4978 * systemd-journald's own logging messages are now also logged to the
4979 journal itself when systemd-journald logs to /dev/kmsg.
4981 * systemd-journald now re-enables COW for archived journal files on
4982 filesystems that support COW. One benefit of this change is that
4983 archived journal files will now get compressed on btrfs filesystems
4984 that have compression enabled.
4986 * systemd-journald now deduplicates fields in a single log message
4987 before adding it to the journal. In archived journal files, it will
4988 also punch holes for unused parts and truncate the file as
4989 appropriate, leading to reductions in disk usage.
4991 * journalctl --verify was extended with more informative error
4994 * More of sd-journal's functions are now resistant against journal file
4997 * The shutdown command learnt a new option --show, to display the
5000 * A LICENSES/ directory is now included in the git tree. It contains a
5001 README.md file that explains the licenses used by source files in
5002 this repository. It also contains the text of all applicable
5003 licenses as they appear on spdx.org.
5005 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
5006 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
5007 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
5008 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
5009 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
5010 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
5011 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
5012 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
5013 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
5014 Christian Brauner, Christian Göttsche, Christian Wehrli,
5015 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
5016 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
5017 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
5018 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
5019 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
5020 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
5021 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
5022 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
5023 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
5024 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
5025 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
5026 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
5027 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
5028 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
5029 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
5030 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
5031 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
5032 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
5033 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
5034 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
5035 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
5036 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
5037 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
5038 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
5039 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
5040 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
5041 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
5042 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
5043 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
5044 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
5045 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
5046 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
5047 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
5048 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
5049 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
5050 Дамјан Георгиевски, наб
5052 — Warsaw, 2021-12-23
5056 * When operating on disk images via the --image= switch of various
5057 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
5058 no 'root=' parameter on the kernel command line, and multiple
5059 suitable root or /usr/ partitions exist in the image, then a simple
5060 comparison inspired by strverscmp() is done on the GPT partition
5061 label, and the newest partition is picked. This permits a simple and
5062 generic whole-file-system A/B update logic where new operating system
5063 versions are dropped into partitions whose label is then updated with
5064 a matching version identifier.
5066 * systemd-sysusers now supports querying the passwords to set for the
5067 users it creates via the "credentials" logic introduced in v247: the
5068 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
5069 credentials are consulted for the password to use (either in UNIX
5070 hashed form, or literally). By default these credentials are inherited
5071 down from PID1 (which in turn imports it from a container manager if
5072 there is one). This permits easy configuration of user passwords
5073 during first boot. Example:
5075 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
5077 Note that systemd-sysusers operates in purely additive mode: it
5078 executes no operation if the declared users already exist, and hence
5079 doesn't set any passwords as effect of the command line above if the
5080 specified root user exists already in the image. (Note that
5081 --volatile=yes ensures it doesn't, though.)
5083 * systemd-firstboot now also supports querying various system
5084 parameters via the credential subsystems. Thus, as above this may be
5085 used to initialize important system parameters on first boot of
5086 previously unprovisioned images (i.e. images with a mostly empty
5089 * PID 1 may now show both the unit name and the unit description
5090 strings in its status output during boot. This may be configured with
5091 StatusUnitFormat=combined in system.conf or
5092 systemd.status-unit-format=combined on the kernel command line.
5094 * The systemd-machine-id-setup tool now supports a --image= switch for
5095 provisioning a machine ID file into an OS disk image, similar to how
5096 --root= operates on an OS file tree. This matches the existing switch
5097 of the same name for systemd-tmpfiles, systemd-firstboot, and
5098 systemd-sysusers tools.
5100 * Similarly, systemd-repart gained support for the --image= switch too.
5101 In combination with the existing --size= option, this makes the tool
5102 particularly useful for easily growing disk images in a single
5103 invocation, following the declarative rules included in the image
5106 * systemd-repart's partition configuration files gained support for a
5107 new switch MakeDirectories= which may be used to create arbitrary
5108 directories inside file systems that are created, before registering
5109 them in the partition table. This is useful in particular for root
5110 partitions to create mount point directories for other partitions
5111 included in the image. For example, a disk image that contains a
5112 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
5113 create /home/ and /var/ as empty directories in the root file system
5114 on its creation, so that the resulting image can be mounted
5115 immediately, even in read-only mode.
5117 * systemd-repart's CopyBlocks= setting gained support for the special
5118 value "auto". If used, a suitable matching partition on the booted OS
5119 is found as source to copy blocks from. This is useful when
5120 implementing replicating installers, that are booted from one medium
5121 and then stream their own root partition onto the target medium.
5123 * systemd-repart's partition configuration files gained support for a
5124 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
5125 GPT partition flags for the created partitions: this is useful for
5126 marking newly created partitions as read-only, or as not being
5127 subject for automatic mounting from creation on.
5129 * The /etc/os-release file has been extended with two new (optional)
5130 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
5131 information for OS images that are updated comprehensively and
5132 atomically as one image. Two new specifiers %M, %A now resolve to
5133 these two fields in the various configuration options that resolve
5136 * portablectl gained a new switch --extension= for enabling portable
5137 service images with extensions that follow the extension image
5138 concept introduced with v248, and thus allows layering multiple
5139 images when setting up the root filesystem of the service.
5141 * systemd-coredump will now extract ELF build-id information from
5142 processes dumping core and include it in the coredump report.
5143 Moreover, it will look for ELF .note.package sections with
5144 distribution packaging meta-information about the crashing process.
5145 This is useful to directly embed the rpm or deb (or any other)
5146 package name and version in ELF files, making it easy to match
5147 coredump reports with the specific package for which the software was
5148 compiled. This is particularly useful on environments with ELF files
5149 from multiple vendors, different distributions and versions, as is
5150 common today in our containerized and sand-boxed world. For further
5153 https://systemd.io/COREDUMP_PACKAGE_METADATA
5155 * A new udev hardware database has been added for FireWire devices
5158 * The "net_id" built-in of udev has been updated with three
5159 backwards-incompatible changes:
5161 - PCI hotplug slot names on s390 systems are now parsed as
5162 hexadecimal numbers. They were incorrectly parsed as decimal
5163 previously, or ignored if the name was not a valid decimal
5166 - PCI onboard indices up to 65535 are allowed. Previously, numbers
5167 above 16383 were rejected. This primarily impacts s390 systems,
5168 where values up to 65535 are used.
5170 - Invalid characters in interface names are replaced with "_".
5172 The new version of the net naming scheme is "v249". The previous
5173 scheme can be selected via the "net.naming_scheme=v247" kernel
5174 command line parameter.
5176 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
5177 NULL bus object, for which they will return false. Or in other words,
5178 an unallocated bus connection is neither ready nor open.
5180 * The sd-device API acquired a new API function
5181 sd_device_get_usec_initialized() that returns the monotonic time when
5182 the udev device first appeared in the database.
5184 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
5185 sd_device_get_trigger_uuid(). The former is similar to
5186 sd_device_trigger() but returns a randomly generated UUID that is
5187 associated with the synthetic uevent generated by the call. This UUID
5188 may be read from the sd_device object a monitor eventually receives,
5189 via the sd_device_get_trigger_uuid(). This interface requires kernel
5190 4.13 or above to work, and allows tracking a synthetic uevent through
5191 the entire device management stack. The "udevadm trigger --settle"
5192 logic has been updated to make use of this concept if available to
5193 wait precisely for the uevents it generates. "udevadm trigger" also
5194 gained a new parameter --uuid that prints the UUID for each generated
5197 * sd-device also gained new APIs sd_device_new_from_ifname() and
5198 sd_device_new_from_ifindex() for allocating an sd-device object for
5199 the specified network interface. The former accepts an interface name
5200 (either a primary or an alternative name), the latter an interface
5203 * The native Journal protocol has been documented. Clients may talk
5204 this as alternative to the classic BSD syslog protocol for locally
5205 delivering log records to the Journal. The protocol has been stable
5206 for a long time and in fact been implemented already in a variety
5207 of alternative client libraries. This documentation makes the support
5210 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
5212 * A new BPFProgram= setting has been added to service files. It may be
5213 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
5214 file, or a bind mount or symlink to one. This may be used to upload
5215 and manage BPF programs externally and then hook arbitrary systemd
5218 * The "home.arpa" domain that has been officially declared as the
5219 choice for domain for local home networks per RFC 8375 has been added
5220 to the default NTA list of resolved, since DNSSEC is generally not
5221 available on private domains.
5223 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
5225 * A new ManageForeignRoutingPolicyRules= setting has been added to
5226 .network files which may be used to exclude foreign-created routing
5227 policy rules from systemd-networkd management.
5229 * systemd-network-wait-online gained two new switches -4 and -6 that
5230 may be used to tweak whether to wait for only IPv4 or only IPv6
5233 * .network files gained a new RequiredFamilyForOnline= setting to
5234 fine-tune whether to require an IPv4 or IPv6 address in order to
5235 consider an interface "online".
5237 * networkctl will now show an over-all "online" state in the per-link
5240 * In .network files a new OutgoingInterface= setting has been added to
5241 specify the output interface in bridge FDB setups.
5243 * In .network files the Multipath group ID may now be configured for
5244 [NextHop] entries, via the new Group= setting.
5246 * The DHCP server logic configured in .network files gained a new
5247 setting RelayTarget= that turns the server into a DHCP server relay.
5248 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
5249 to further tweak the DHCP relay behaviour.
5251 * The DHCP server logic also gained a new ServerAddress= setting in
5252 .network files that explicitly specifies the server IP address to
5253 use. If not specified, the address is determined automatically, as
5256 * The DHCP server logic in systemd-networkd gained support for static
5257 DHCP leases, configurable via the [DHCPServerStaticLease]
5258 section. This allows explicitly mapping specific MAC addresses to
5259 fixed IP addresses and vice versa.
5261 * The RestrictAddressFamilies= setting in service files now supports a
5262 new special value "none". If specified sockets of all address
5263 families will be made unavailable to services configured that way.
5265 * systemd-fstab-generator and systemd-repart have been updated to
5266 support booting from disks that carry only a /usr/ partition but no
5267 root partition yet, and where systemd-repart can add it in on the
5268 first boot. This is useful for implementing systems that ship with a
5269 single /usr/ file system, and whose root file system shall be set up
5270 and formatted on a LUKS-encrypted volume whose key is generated
5271 locally (and possibly enrolled in the TPM) during the first boot.
5273 * The [Address] section of .network files now accepts a new
5274 RouteMetric= setting that configures the routing metric to use for
5275 the prefix route created as effect of the address configuration.
5276 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
5277 gained matching settings for their prefix routes. (The option of the
5278 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
5279 it conceptually belongs there; the old option is still understood for
5282 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
5285 * A new udev property ID_NET_DHCP_BROADCAST on network interface
5286 devices is now honoured by systemd-networkd, controlling whether to
5287 issue DHCP offers via broadcasting. This is used to ensure that s390
5288 layer 3 network interfaces work out-of-the-box with systemd-networkd.
5290 * nss-myhostname and systemd-resolved will now synthesize address
5291 records for a new special hostname "_outbound". The name will always
5292 resolve to the local IP addresses most likely used for outbound
5293 connections towards the default routes. On multi-homed hosts this is
5294 useful to have a stable handle referring to "the" local IP address
5295 that matters most, to the point where this is defined.
5297 * The Discoverable Partition Specification has been updated with a new
5298 GPT partition flag "grow-file-system" defined for its partition
5299 types. Whenever partitions with this flag set are automatically
5300 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
5301 of systemd-nspawn or other tools; and as opposed to explicit mounting
5302 via /etc/fstab), the file system within the partition is
5303 automatically grown to the full size of the partition. If the file
5304 system size already matches the partition size this flag has no
5305 effect. Previously, this functionality has been available via the
5306 explicit x-systemd.growfs mount option, and this new flag extends
5307 this to automatically discovered mounts. A new GrowFileSystem=
5308 setting has been added to systemd-repart drop-in files that allows
5309 configuring this partition flag. This new flag defaults to on for
5310 partitions automatically created by systemd-repart, except if they
5311 are marked read-only. See the specification for further details:
5313 https://systemd.io/DISCOVERABLE_PARTITIONS
5315 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
5316 section. If enabled (which is the default), and an NTP server address
5317 is acquired through a DHCP lease on this interface an explicit route
5318 to this address is created on this interface to ensure that NTP
5319 traffic to the NTP server acquired on an interface is also routed
5320 through that interface. The pre-existing RoutesToDNS= setting that
5321 implements the same for DNS servers is now enabled by default.
5323 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
5324 been added that may be used to restrict the network interfaces
5325 sockets created by the service may be bound to. This is implemented
5328 * A new ConditionFirmware= setting has been added to unit files to
5329 conditionalize on certain firmware features. At the moment it may
5330 check whether running on a UEFI system, a device.tree system, or if
5331 the system is compatible with some specified device-tree feature.
5333 * A new ConditionOSRelease= setting has been added to unit files to
5334 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
5335 operators may be used to check if some field has some specific value
5336 or do an alphanumerical comparison. Equality comparisons are useful
5337 for fields like ID, but relative comparisons for fields like
5338 VERSION_ID or IMAGE_VERSION.
5340 * hostnamed gained a new Describe() D-Bus method that returns a JSON
5341 serialization of the host data it exposes. This is exposed via
5342 "hostnamectl --json=" to acquire a host identity description in JSON.
5343 It's our intention to add a similar features to most services and
5344 objects systemd manages, in order to simplify integration with
5345 program code that can consume JSON.
5347 * Similarly, networkd gained a Describe() method on its Manager and
5348 Link bus objects. This is exposed via "networkctl --json=".
5350 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
5351 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
5352 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
5353 that is used both to get the value (when no argument is given), and
5354 to set the value (when an argument is specified). The old names
5355 continue to be supported for compatibility.
5357 * systemd-detect-virt and ConditionVirtualization= are now able to
5358 correctly identify Amazon EC2 environments.
5360 * The LogLevelMax= setting of unit files now applies not only to log
5361 messages generated *by* the service, but also to log messages
5362 generated *about* the service by PID 1. To suppress logs concerning a
5363 specific service comprehensively, set this option to a high log
5366 * bootctl gained support for a new --make-machine-id-directory= switch
5367 that allows precise control on whether to create the top-level
5368 per-machine directory in the boot partition that typically contains
5369 Type 1 boot loader entries.
5371 * During build SBAT data to include in the systemd-boot EFI PE binaries
5372 may be specified now.
5374 * /etc/crypttab learnt a new option "headless". If specified any
5375 requests to query the user interactively for passwords or PINs will
5376 be skipped. This is useful on systems that are headless, i.e. where
5377 an interactive user is generally not present.
5379 * /etc/crypttab also learnt a new option "password-echo=" that allows
5380 configuring whether the encryption password prompt shall echo the
5381 typed password and if so, do so literally or via asterisks. (The
5382 default is the same behaviour as before: provide echo feedback via
5385 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
5386 systemd-homed has been updated to allow explicit configuration of the
5387 "user presence" and "user verification" checks, as well as whether a
5388 PIN is required for authentication, via the new switches
5389 --fido2-with-user-presence=, --fido2-with-user-verification=,
5390 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
5391 features are available, and may be enabled or disabled depends on the
5394 * systemd-nspawn's --private-user= switch now accepts the special value
5395 "identity" which configures a user namespacing environment with an
5396 identity mapping of 65535 UIDs. This means the container UID 0 is
5397 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
5398 this doesn't appear to be useful, however it does reduce the attack
5399 surface a bit, since the resulting container will possess process
5400 capabilities only within its namespace and not on the host.
5402 * systemd-nspawn's --private-user-chown switch has been replaced by a
5403 more generic --private-user-ownership= switch that accepts one of
5404 three values: "chown" is equivalent to the old --private-user-chown,
5405 and "off" is equivalent to the absence of the old switch. The value
5406 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
5407 of files and directories of the underlying image to the chosen UID
5408 range for the container. "auto" is equivalent to "map" if UID mapping
5409 mount are supported, otherwise it is equivalent to "chown". The short
5410 -U switch systemd-nspawn now implies --private-user-ownership=auto
5411 instead of the old --private-user-chown. Effectively this means: if
5412 the backing file system supports UID mapping mounts the feature is
5413 now used by default if -U is used. Generally, it's a good idea to use
5414 UID mapping mounts instead of recursive chown()ing, since it allows
5415 running containers off immutable images (since no modifications of
5416 the images need to take place), and share images between multiple
5417 instances. Moreover, the recursive chown()ing operation is slow and
5418 can be avoided. Conceptually it's also a good thing if transient UID
5419 range uses do not leak into persistent file ownership anymore. TLDR:
5420 finally, the last major drawback of user namespacing has been
5421 removed, and -U should always be used (unless you use btrfs, where
5422 UID mapped mounts do not exist; or your container actually needs
5423 privileges on the host).
5425 * nss-systemd now synthesizes user and group shadow records in addition
5426 to the main user and group records. Thus, hashed passwords managed by
5427 systemd-homed are now accessible via the shadow database.
5429 * The userdb logic (and thus nss-systemd, and so on) now read
5430 additional user/group definitions in JSON format from the drop-in
5431 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
5432 /usr/lib/userdb/. This is a simple and powerful mechanism for making
5433 additional users available to the system, with full integration into
5434 NSS including the shadow databases. Since the full JSON user/group
5435 record format is supported this may also be used to define users with
5436 resource management settings and other runtime settings that
5437 pam_systemd and systemd-logind enforce at login.
5439 * The userdbctl tool gained two new switches --with-dropin= and
5440 --with-varlink= which can be used to fine-tune the sources used for
5441 user database lookups.
5443 * systemd-nspawn gained a new switch --bind-user= for binding a host
5444 user account into the container. This does three things: the user's
5445 home directory is bind mounted from the host into the container,
5446 below the /run/userdb/home/ hierarchy. A free UID is picked in the
5447 container, and a user namespacing UID mapping to the host user's UID
5448 installed. And finally, a minimal JSON user and group record (along
5449 with its hashed password) is dropped into /run/host/userdb/. These
5450 records are picked up automatically by the userdb drop-in logic
5451 describe above, and allow the user to login with the same password as
5452 on the host. Effectively this means: if host and container run new
5453 enough systemd versions making a host user available to the container
5454 is trivially simple.
5456 * systemd-journal-gatewayd now supports the switches --user, --system,
5457 --merge, --file= that are equivalent to the same switches of
5458 journalctl, and permit exposing only the specified subset of the
5461 * The OnFailure= dependency between units is now augmented with a
5462 implicit reverse dependency OnFailureOf= (this new dependency cannot
5463 be configured directly it's only created as effect of an OnFailure=
5464 dependency in the reverse order — it's visible in "systemctl show"
5465 however). Similar, Slice= now has an reverse dependency SliceOf=,
5466 that is also not configurable directly, but useful to determine all
5467 units that are members of a slice.
5469 * A pair of new dependency types between units PropagatesStopTo= +
5470 StopPropagatedFrom= has been added, that allows propagation of unit
5471 stop events between two units. It operates similar to the existing
5472 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
5474 * A new dependency type OnSuccess= has been added (plus the reverse
5475 dependency OnSuccessOf=, which cannot be configured directly, but
5476 exists only as effect of the reverse OnSuccess=). It is similar to
5477 OnFailure=, but triggers in the opposite case: when a service exits
5478 cleanly. This allows "chaining up" of services where one or more
5479 services are started once another service has successfully completed.
5481 * A new dependency type Upholds= has been added (plus the reverse
5482 dependency UpheldBy=, which cannot be configured directly, but exists
5483 only as effect of Upholds=). This dependency type is a stronger form
5484 of Wants=: if a unit has an UpHolds= dependency on some other unit
5485 and the former is active then the latter is started whenever it is
5486 found inactive (and no job is queued for it). This is an alternative
5487 to Restart= inside service units, but less configurable, and the
5488 request to uphold a unit is not encoded in the unit itself but in
5489 another unit that intends to uphold it.
5491 * The systemd-ask-password tool now also supports reading passwords
5492 from the credentials subsystem, via the new --credential= switch.
5494 * The systemd-ask-password tool learnt a new switch --emoji= which may
5495 be used to explicit control whether the lock and key emoji (🔐) is
5496 shown in the password prompt on suitable TTYs.
5498 * The --echo switch of systemd-ask-password now optionally takes a
5499 parameter that controls character echo. It may either show asterisks
5500 (default, as before), turn echo off entirely, or echo the typed
5501 characters literally.
5503 * The systemd-ask-password tool also gained a new -n switch for
5504 suppressing output of a trailing newline character when writing the
5505 acquired password to standard output, similar to /bin/echo's -n
5508 * New documentation has been added that describes the organization of
5509 the systemd source code tree:
5511 https://systemd.io/ARCHITECTURE
5513 * Units using ConditionNeedsUpdate= will no longer be activated in
5516 * It is now possible to list a template unit in the WantedBy= or
5517 RequiredBy= settings of the [Install] section of another template
5518 unit, which will be instantiated using the same instance name.
5520 * A new MemoryAvailable property is available for units. If the unit,
5521 or the slices it is part of, have a memory limit set via MemoryMax=/
5522 MemoryHigh=, MemoryAvailable will indicate how much more memory the
5523 unit can claim before hitting the limits.
5525 * systemd-coredump will now try to stay below the cgroup memory limit
5526 placed on itself or one of the slices it runs under, if the storage
5527 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
5528 since files written on such filesystems count toward the cgroup memory
5529 limit. If there is not enough available memory in such cases to store
5530 the core file uncompressed, systemd-coredump will skip to compressed
5531 storage directly (if enabled) and it will avoid analyzing the core file
5532 to print backtrace and metadata in the journal.
5534 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
5535 of a path matches the configured expectations, and remove it if not.
5537 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
5538 specify which of the several available filesystem timestamps (access
5539 time, birth time, change time, modification time) to look at when
5540 deciding whether a path has aged enough to be cleaned.
5542 * A new IPv6StableSecretAddress= setting has been added to .network
5543 files, which takes an IPv6 address to use as secret for IPv6 address
5546 * The [DHCPServer] logic in .network files gained support for a new
5547 UplinkInterface= setting that permits configuration of the uplink
5548 interface name to propagate DHCP lease information from.
5550 * The WakeOnLan= setting in .link files now accepts a list of flags
5551 instead of a single one, to configure multiple wake-on-LAN policies.
5553 * User-space defined tracepoints (USDT) have been added to udev at
5554 strategic locations. This is useful for tracing udev behaviour and
5555 performance with bpftrace and similar tools.
5557 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
5558 setting a network timeout time.
5560 * If a system service is running in a new mount namespace (RootDirectory=
5561 and friends), all file systems will be mounted with MS_NOSUID by
5562 default, unless the system is running with SELinux enabled.
5564 * When enumerating time zones the timedatectl tool will now consult the
5565 'tzdata.zi' file shipped by the IANA time zone database package, in
5566 addition to 'zone1970.tab', as before. This makes sure time zone
5567 aliases are now correctly supported. Some distributions so far did
5568 not install this additional file, most do however. If you
5569 distribution does not install it yet, it might make sense to change
5572 * Intel HID rfkill event is no longer masked, since it's the only
5573 source of rfkill event on newer HP laptops. To have both backward and
5574 forward compatibility, userspace daemon needs to debounce duplicated
5575 events in a short time window.
5577 Contributions from: Aakash Singh, adrian5, Albert Brox,
5578 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
5579 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
5580 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
5581 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
5582 Christian Hesse, Daniel Schaefer, Dan Streetman,
5583 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
5584 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
5585 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
5586 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
5587 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
5588 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
5589 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
5590 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
5591 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
5592 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
5593 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
5594 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
5595 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
5596 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
5597 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
5598 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
5599 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
5600 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
5601 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
5602 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
5603 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
5604 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
5605 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
5606 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
5607 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
5609 — Edinburgh, 2021-07-07
5613 * A concept of system extension images is introduced. Such images may
5614 be used to extend the /usr/ and /opt/ directory hierarchies at
5615 runtime with additional files (even if the file system is read-only).
5616 When a system extension image is activated, its /usr/ and /opt/
5617 hierarchies and os-release information are combined via overlayfs
5618 with the file system hierarchy of the host OS.
5620 A new systemd-sysext tool can be used to merge, unmerge, list, and
5621 refresh system extension hierarchies. See
5622 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
5624 The systemd-sysext.service automatically merges installed system
5625 extensions during boot (before basic.target, but not in very early
5626 boot, since various file systems have to be mounted first).
5628 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
5629 supported system extension level.
5631 * A new ExtensionImages= unit setting can be used to apply the same
5632 system extension image concept from systemd-sysext to the namespaced
5633 file hierarchy of specific services, following the same rules and
5636 * Support for a new special "root=tmpfs" kernel command-line option has
5637 been added. When specified, a tmpfs is mounted on /, and mount.usr=
5638 should be used to point to the operating system implementation.
5640 * A new configuration file /etc/veritytab may be used to configure
5641 dm-verity integrity protection for block devices. Each line is in the
5642 format "volume-name data-device hash-device roothash options",
5643 similar to /etc/crypttab.
5645 * A new kernel command-line option systemd.verity.root_options= may be
5646 used to configure dm-verity behaviour for the root device.
5648 * The key file specified in /etc/crypttab (the third field) may now
5649 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
5650 acquired by connecting to that socket and reading from it. This
5651 allows the implementation of a service to provide key information
5652 dynamically, at the moment when it is needed.
5654 * When the hostname is set explicitly to "localhost", systemd-hostnamed
5655 will respect this. Previously such a setting would be mostly silently
5656 ignored. The goal is to honour configuration as specified by the
5659 * The fallback hostname that will be used by the system manager and
5660 systemd-hostnamed can now be configured in two new ways: by setting
5661 DEFAULT_HOSTNAME= in os-release(5), or by setting
5662 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
5663 also be configured during compilation. The environment variable is
5664 intended for testing and local overrides, the os-release(5) field is
5665 intended to allow customization by different variants of a
5666 distribution that share the same compiled packages.
5668 * The environment block of the manager itself may be configured through
5669 a new ManagerEnvironment= setting in system.conf or user.conf. This
5670 complements existing ways to set the environment block (the kernel
5671 command line for the system manager, the inherited environment and
5672 user@.service unit file settings for the user manager).
5674 * systemd-hostnamed now exports the default hostname and the source of
5675 the configured hostname ("static", "transient", or "default") as
5678 * systemd-hostnamed now exports the "HardwareVendor" and
5679 "HardwareModel" D-Bus properties, which are supposed to contain a
5680 pair of cleaned up, human readable strings describing the system's
5681 vendor and model. It's typically sourced from the firmware's DMI
5682 tables, but may be augmented from a new hwdb database. hostnamectl
5683 shows this in the status output.
5685 * Support has been added to systemd-cryptsetup for extracting the
5686 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
5687 metadata header. This allows the information how to open the
5688 encrypted device to be embedded directly in the device and obviates
5689 the need for configuration in an external file.
5691 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
5692 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
5693 pre-existing support for PKCS#11 security tokens).
5695 * systemd-repart may enroll encrypted partitions using TPM2
5696 hardware. This may be useful for example to create an encrypted /var
5697 partition bound to the machine on first boot.
5699 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
5700 and PKCS#11 security tokens to LUKS volumes, list and destroy
5703 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
5705 It also supports enrolling "recovery keys" and regular passphrases.
5707 * The libfido2 dependency is now based on dlopen(), so that the library
5708 is used at runtime when installed, but is not a hard runtime
5711 * systemd-cryptsetup gained support for two new options in
5712 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
5713 request synchronous processing of encryption/decryption IO.
5715 * The manager may be configured at compile time to use the fexecve()
5716 instead of the execve() system call when spawning processes. Using
5717 fexecve() closes a window between checking the security context of an
5718 executable and spawning it, but unfortunately the kernel displays
5719 stale information in the process' "comm" field, which impacts ps
5722 * The configuration option -Dcompat-gateway-hostname has been dropped.
5723 "_gateway" is now the only supported name.
5725 * The ConditionSecurity=tpm2 unit file setting may be used to check if
5726 the system has at least one TPM2 (tpmrm class) device.
5728 * A new ConditionCPUFeature= has been added that may be used to
5729 conditionalize units based on CPU features. For example,
5730 ConditionCPUFeature=rdrand will condition a unit so that it is only
5731 run when the system CPU supports the RDRAND opcode.
5733 * The existing ConditionControlGroupController= setting has been
5734 extended with two new values "v1" and "v2". "v2" means that the
5735 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
5736 hierarchy or the hybrid hierarchy are used.
5738 * A new PrivateIPC= setting on a unit file allows executed processes to
5739 be moved into a private IPC namespace, with separate System V IPC
5740 identifiers and POSIX message queues.
5742 A new IPCNamespacePath= allows the unit to be joined to an existing
5745 * The tables of system calls in seccomp filters are now automatically
5746 generated from kernel lists exported on
5747 https://fedora.juszkiewicz.com.pl/syscalls.html.
5749 The following architectures should now have complete lists:
5750 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
5751 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
5753 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
5754 on /run/ if it is not already a mount point. A writable /run/ has
5755 always been a requirement for a functioning system, but this was not
5756 guaranteed when using a read-only image.
5758 Users can always specify BindPaths= or InaccessiblePaths= as
5759 overrides, and they will take precedence. If the host's root mount
5760 point is used, there is no change in behaviour.
5762 * New bind mounts and file system image mounts may be injected into the
5763 mount namespace of a service (without restarting it). This is exposed
5764 respectively as 'systemctl bind <unit> <path>…' and
5765 'systemctl mount-image <unit> <image>…'.
5767 * The StandardOutput= and StandardError= settings can now specify files
5768 to be truncated for output (as "truncate:<path>").
5770 * The ExecPaths= and NoExecPaths= settings may be used to specify
5771 noexec for parts of the file system.
5773 * sd-bus has a new function sd_bus_open_user_machine() to open a
5774 connection to the session bus of a specific user in a local container
5775 or on the local host. This is exposed in the existing -M switch to
5776 systemctl and similar tools:
5778 systemctl --user -M lennart@foobar start foo
5780 This will connect to the user bus of a user "lennart" in container
5781 "foobar". If no container name is specified, the specified user on
5782 the host itself is connected to
5784 systemctl --user -M lennart@ start quux
5786 * sd-bus also gained a convenience function sd_bus_message_send() to
5787 simplify invocations of sd_bus_send(), taking only a single
5788 parameter: the message to send.
5790 * sd-event allows rate limits to be set on event sources, for dealing
5791 with high-priority event sources that might starve out others. See
5792 the new man page sd_event_source_set_ratelimit(3) for details.
5794 * systemd.link files gained a [Link] Promiscuous= switch, which allows
5795 the device to be raised in promiscuous mode.
5797 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
5798 number of TX and RX queues to be configured.
5800 New [Link] TransmitQueueLength= setting allows the size of the TX
5801 queue to be configured.
5803 New [Link] GenericSegmentOffloadMaxBytes= and
5804 GenericSegmentOffloadMaxSegments= allow capping the packet size and
5805 the number of segments accepted in Generic Segment Offload.
5807 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
5808 wireless routing protocol that operates on ISO/OSI Layer 2 only and
5809 uses ethernet frames to route/bridge packets. This encompasses a new
5810 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
5811 new settings in .netdev files, and a new BatmanAdvanced= setting in
5814 * systemd.network files gained a [Network] RouteTable= configuration
5815 switch to select the routing policy table.
5817 systemd.network files gained a [RoutingPolicyRule] Type=
5818 configuration switch (one of "blackhole, "unreachable", "prohibit").
5820 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
5821 RouteAllowList= settings to ignore/accept route advertisements from
5822 routers matching specified prefixes. The DenyList= setting has been
5823 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
5826 systemd.network files gained a [DHCPv6] UseAddress= setting to
5827 optionally ignore the address provided in the lease.
5829 systemd.network files gained a [DHCPv6PrefixDelegation]
5830 ManageTemporaryAddress= switch.
5832 systemd.network files gained a new ActivationPolicy= setting which
5833 allows configuring how the UP state of an interface shall be managed,
5834 i.e. whether the interface is always upped, always downed, or may be
5835 upped/downed by the user using "ip link set dev".
5837 * The default for the Broadcast= setting in .network files has slightly
5838 changed: the broadcast address will not be configured for wireguard
5841 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
5842 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
5843 configuration options for VLAN packet handling.
5845 * udev rules may now set log_level= option. This allows debug logs to
5846 be enabled for select events, e.g. just for a specific subsystem or
5847 even a single device.
5849 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
5850 DATA_PREPARED_ID properties for block devices with ISO9660 file
5853 * udev now exports decoded DMI information about installed memory slots
5854 as device properties under the /sys/class/dmi/id/ pseudo device.
5856 * /dev/ is not mounted noexec anymore. This didn't provide any
5857 significant security benefits and would conflict with the executable
5858 mappings used with /dev/sgx device nodes. The previous behaviour can
5859 be restored for individual services with NoExecPaths=/dev (or by allow-
5860 listing and excluding /dev from ExecPaths=).
5862 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
5863 and /dev/vhost-net are owned by the kvm group.
5865 * The hardware database has been extended with a list of fingerprint
5866 readers that correctly support USB auto-suspend using data from
5869 * systemd-resolved can now answer DNSSEC questions through the stub
5870 resolver interface in a way that allows local clients to do DNSSEC
5871 validation themselves. For a question with DO+CD set, it'll proxy the
5872 DNS query and respond with a mostly unmodified packet received from
5873 the upstream server.
5875 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
5876 resolved.conf. If true the service will provide caching even for DNS
5877 lookups made to an upstream DNS server on the 127.0.0.1/::1
5878 addresses. By default (and when the option is false) systemd-resolved
5879 will not cache such lookups, in order to avoid duplicate local
5880 caching, under the assumption the local upstream server caches
5883 * systemd-resolved now implements RFC5001 NSID in its local DNS
5884 stub. This may be used by local clients to determine whether they are
5885 talking to the DNS resolver stub or a different DNS server.
5887 * When resolving host names and other records resolvectl will now
5888 report where the data was acquired from (i.e. the local cache, the
5889 network, locally synthesized, …) and whether the network traffic it
5890 effected was encrypted or not. Moreover the tool acquired a number of
5891 new options --cache=, --synthesize=, --network=, --zone=,
5892 --trust-anchor=, --validate= that take booleans and may be used to
5893 tweak a lookup, i.e. whether it may be answered from cached
5894 information, locally synthesized information, information acquired
5895 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
5896 anchor, and whether DNSSEC validation shall be executed for the
5899 * systemd-nspawn gained a new --ambient-capability= setting
5900 (AmbientCapability= in .nspawn files) to configure ambient
5901 capabilities passed to the container payload.
5903 * systemd-nspawn gained the ability to configure the firewall using the
5904 nftables subsystem (in addition to the existing iptables
5905 support). Similarly, systemd-networkd's IPMasquerade= option now
5906 supports nftables as back-end, too. In both cases NAT on IPv6 is now
5907 supported too, in addition to IPv4 (the iptables back-end still is
5910 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
5911 retains its meaning, but has been deprecated. Please switch to either
5912 "ivp4" or "both" (if covering IPv6 is desired).
5914 * systemd-importd will now download .verity and .roothash.p7s files
5915 along with the machine image (as exposed via machinectl pull-raw).
5917 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
5918 setting to configure the time a unit's cgroup needs to exceed memory
5919 pressure limits before action will be taken, and a new
5920 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
5923 systemd-oomd is now considered fully supported (the usual
5924 backwards-compatibility promises apply). Swap is not required for
5925 operation, but it is still recommended.
5927 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
5928 configures the retry delay when trying to contact servers.
5930 * systemd-stdio-bridge gained --system/--user options to connect to the
5931 system bus (previous default) or the user session bus.
5933 * systemd-localed may now call locale-gen to generate missing locales
5934 on-demand (UTF-8-only). This improves integration with Debian-based
5935 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
5937 * systemctl --check-inhibitors=true may now be used to obey inhibitors
5938 even when invoked non-interactively. The old --ignore-inhibitors
5939 switch is now deprecated and replaced by --check-inhibitors=false.
5941 * systemctl import-environment will now emit a warning when called
5942 without any arguments (i.e. to import the full environment block of
5943 the called program). This command will usually be invoked from a
5944 shell, which means that it'll inherit a bunch of variables which are
5945 specific to that shell, and usually to the TTY the shell is connected
5946 to, and don't have any meaning in the global context of the system or
5947 user service manager. Instead, only specific variables should be
5948 imported into the manager environment block.
5950 Similarly, programs which update the manager environment block by
5951 directly calling the D-Bus API of the manager, should also push
5952 specific variables, and not the full inherited environment.
5954 * systemctl's status output now shows unit state with a more careful
5955 choice of Unicode characters: units in maintenance show a "○" symbol
5956 instead of the usual "●", failed units show "×", and services being
5959 * coredumpctl gained a --debugger-arguments= switch to pass arguments
5960 to the debugger. It also gained support for showing coredump info in
5961 a simple JSON format.
5963 * systemctl/loginctl/machinectl's --signal= option now accept a special
5964 value "list", which may be used to show a brief table with known
5965 process signals and their numbers.
5967 * networkctl now shows the link activation policy in status.
5969 * Various tools gained --pager/--no-pager/--json= switches to
5970 enable/disable the pager and provide JSON output.
5972 * Various tools now accept two new values for the SYSTEMD_COLORS
5973 environment variable: "16" and "256", to configure how many terminal
5974 colors are used in output.
5976 * less 568 or newer is now required for the auto-paging logic of the
5977 various tools. Hyperlink ANSI sequences in terminal output are now
5978 used even if a pager is used, and older versions of less are not able
5979 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
5980 disable this output again.
5982 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
5983 builds, non-merged-usr builds) are now officially deprecated. A
5984 warning is emitted during build. Support is slated to be removed in
5985 about a year (when the Debian Bookworm release development starts).
5987 * Systems with the legacy cgroup v1 hierarchy are now marked as
5988 "tainted", to make it clearer that using the legacy hierarchy is not
5991 * systemd-localed will now refuse to configure a keymap which is not
5992 installed in the file system. This is intended as a bug fix, but
5993 could break cases where systemd-localed was used to configure the
5994 keymap in advanced of it being installed. It is necessary to install
5995 the keymap file first.
5997 * The main git development branch has been renamed to 'main'.
5999 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
6000 for partitions, as in the vast majority of cases they contain none
6001 and are used internally by the bootloader (eg: uboot).
6003 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
6004 spawned processes to the PID of the process itself. This may be used
6005 by programs for detecting whether they were forked off by the service
6006 manager itself or are a process forked off further down the tree.
6008 * The sd-device API gained four new calls: sd_device_get_action() to
6009 determine the uevent add/remove/change/… action the device object has
6010 been seen for, sd_device_get_seqno() to determine the uevent sequence
6011 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
6012 object from stat(2) data of a device node, and sd_device_trigger() to
6013 write to the 'uevent' attribute of a device.
6015 * For most tools the --no-legend= switch has been replaced by
6016 --legend=no and --legend=yes, to force whether tables are shown with
6019 * Units acquired a new property "Markers" that takes a list of zero,
6020 one or two of the following strings: "needs-reload" and
6021 "needs-restart". These markers may be set via "systemctl
6022 set-property". Once a marker is set, "systemctl reload-or-restart
6023 --marked" may be invoked to execute the operation the units are
6024 marked for. This is useful for package managers that want to mark
6025 units for restart/reload while updating, but effect the actual
6026 operations at a later step at once.
6028 * The sd_bus_message_read_strv() API call of sd-bus may now also be
6029 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
6032 * bootctl will now report whether the UEFI firmware used a TPM2 device
6033 and measured the boot process into it.
6035 * systemd-tmpfiles learnt support for a new environment variable
6036 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
6037 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
6038 even if the root fs of the system is not itself a btrfs volume.
6040 * systemd-detect-virt/ConditionVirtualization= will now explicitly
6041 detect Docker/Podman environments where possible. Moreover, they
6042 should be able to generically detect any container manager as long as
6043 it assigns the container a cgroup.
6045 * portablectl gained a new "reattach" verb for detaching/reattaching a
6046 portable service image, useful for updating images on-the-fly.
6048 * Intel SGX enclave device nodes (which expose a security feature of
6049 newer Intel CPUs) will now be owned by a new system group "sgx".
6051 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
6052 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
6053 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
6054 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
6055 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
6056 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
6057 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
6058 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
6059 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
6060 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
6061 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
6062 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
6063 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
6064 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
6065 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
6066 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
6067 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
6068 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
6069 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
6070 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
6071 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
6072 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
6073 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
6074 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
6075 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
6076 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
6077 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
6078 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
6079 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
6080 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
6081 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
6082 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
6083 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
6084 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
6085 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
6086 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
6087 Zmicer Turok, Дамјан Георгиевски
6089 — Berlin, 2021-03-30
6093 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
6094 "bind" and "unbind" to the Linux device model. When this kernel
6095 change was made, systemd-udevd was only minimally updated to handle
6096 and propagate these new event types. The introduction of these new
6097 uevents (which are typically generated for USB devices and devices
6098 needing a firmware upload before being functional) resulted in a
6099 number of issues which we so far didn't address. We hoped the kernel
6100 maintainers would themselves address these issues in some form, but
6101 that did not happen. To handle them properly, many (if not most) udev
6102 rules files shipped in various packages need updating, and so do many
6103 programs that monitor or enumerate devices with libudev or sd-device,
6104 or otherwise process uevents. Please note that this incompatibility
6105 is not fault of systemd or udev, but caused by an incompatible kernel
6106 change that happened back in Linux 4.14, but is becoming more and
6107 more visible as the new uevents are generated by more kernel drivers.
6109 To minimize issues resulting from this kernel change (but not avoid
6110 them entirely) starting with systemd-udevd 247 the udev "tags"
6111 concept (which is a concept for marking and filtering devices during
6112 enumeration and monitoring) has been reworked: udev tags are now
6113 "sticky", meaning that once a tag is assigned to a device it will not
6114 be removed from the device again until the device itself is removed
6115 (i.e. unplugged). This makes sure that any application monitoring
6116 devices that match a specific tag is guaranteed to both see uevents
6117 where the device starts being relevant, and those where it stops
6118 being relevant (the latter now regularly happening due to the new
6119 "unbind" uevent type). The udev tags concept is hence now a concept
6120 tied to a *device* instead of a device *event* — unlike for example
6121 udev properties whose lifecycle (as before) is generally tied to a
6122 device event, meaning that the previously determined properties are
6123 forgotten whenever a new uevent is processed.
6125 With the newly redefined udev tags concept, sometimes it's necessary
6126 to determine which tags are the ones applied by the most recent
6127 uevent/database update, in order to discern them from those
6128 originating from earlier uevents/database updates of the same
6129 device. To accommodate for this a new automatic property CURRENT_TAGS
6130 has been added that works similar to the existing TAGS property but
6131 only lists tags set by the most recent uevent/database
6132 update. Similarly, the libudev/sd-device API has been updated with
6133 new functions to enumerate these 'current' tags, in addition to the
6134 existing APIs that now enumerate the 'sticky' ones.
6136 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
6137 essential that all udev rules files and applications are updated to
6138 handle the new events. Specifically:
6140 • All rule files that currently use a header guard similar to
6141 ACTION!="add|change",GOTO="xyz_end" should be updated to use
6142 ACTION=="remove",GOTO="xyz_end" instead, so that the
6143 properties/tags they add are also applied whenever "bind" (or
6144 "unbind") is seen. (This is most important for all physical device
6145 types — those for which "bind" and "unbind" are currently
6146 generated, for all other device types this change is still
6147 recommended but not as important — but certainly prepares for
6148 future kernel uevent type additions).
6150 • Similarly, all code monitoring devices that contains an 'if' branch
6151 discerning the "add" + "change" uevent actions from all other
6152 uevents actions (i.e. considering devices only relevant after "add"
6153 or "change", and irrelevant on all other events) should be reworked
6154 to instead negatively check for "remove" only (i.e. considering
6155 devices relevant after all event types, except for "remove", which
6156 invalidates the device). Note that this also means that devices
6157 should be considered relevant on "unbind", even though conceptually
6158 this — in some form — invalidates the device. Since the precise
6159 effect of "unbind" is not generically defined, devices should be
6160 considered relevant even after "unbind", however I/O errors
6161 accessing the device should then be handled gracefully.
6163 • Any code that uses device tags for deciding whether a device is
6164 relevant or not most likely needs to be updated to use the new
6165 udev_device_has_current_tag() API (or sd_device_has_current_tag()
6166 in case sd-device is used), to check whether the tag is set at the
6167 moment an uevent is seen (as opposed to the existing
6168 udev_device_has_tag() API which checks if the tag ever existed on
6169 the device, following the API concept redefinition explained
6172 We are very sorry for this breakage and the requirement to update
6173 packages using these interfaces. We'd again like to underline that
6174 this is not caused by systemd/udev changes, but result of a kernel
6177 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
6178 packages have not retriggered devices once the udev package (or any
6179 auxiliary package installing additional udev rules) is updated. We
6180 intend to work with major distributions to change this, so that
6181 "udevadm trigger -c change" is issued on such upgrades, ensuring that
6182 the updated ruleset is applied to the devices already discovered, so
6183 that (asynchronously) after the upgrade completed the udev database
6184 is consistent with the updated rule set. This means udev rules must
6185 be ready to be retriggered with a "change" action any time, and
6186 result in correct and complete udev database entries. While the
6187 majority of udev rule files known to us currently get this right,
6188 some don't. Specifically, there are udev rules files included in
6189 various packages that only set udev properties on the "add" action,
6190 but do not handle the "change" action. If a device matching those
6191 rules is retriggered with the "change" action (as is intended here)
6192 it would suddenly lose the relevant properties. This always has been
6193 problematic, but as soon as all udev devices are triggered on relevant
6194 package upgrades this will become particularly so. It is strongly
6195 recommended to fix offending rules so that they can handle a "change"
6196 action at any time, and acquire all necessary udev properties even
6197 then. Or in other words: the header guard mentioned above
6198 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
6199 this, as it makes sure rules are rerun on "change" correctly, and
6200 accumulate the correct and complete set of udev properties. udev rule
6201 definitions that cannot handle "change" events being triggered at
6202 arbitrary times should be considered buggy.
6204 * The MountAPIVFS= service file setting now defaults to on if
6205 RootImage= and RootDirectory= are used, which means that with those
6206 two settings /proc/, /sys/ and /dev/ are automatically properly set
6207 up for services. Previous behaviour may be restored by explicitly
6208 setting MountAPIVFS=off.
6210 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
6211 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
6212 latter it takes precedence over the former, similar to how most of
6213 systemd's own configuration is handled. Given that PAM stack
6214 definitions are primarily put together by OS vendors/distributions
6215 (though possibly overridden by users), this systemd release moves its
6216 own PAM stack configuration for the "systemd-user" PAM service (i.e.
6217 for the PAM session invoked by the per-user user@.service instance)
6218 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
6219 packages' vendor versions of their PAM stack definitions from
6220 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
6221 desired the location to which systemd installs its PAM stack
6222 configuration may be changed via the -Dpamconfdir Meson option.
6224 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
6225 libpwquality and libcryptsetup have been changed to be based on
6226 dlopen(): instead of regular dynamic library dependencies declared in
6227 the binary ELF headers, these libraries are now loaded on demand
6228 only, if they are available. If the libraries cannot be found the
6229 relevant operations will fail gracefully, or a suitable fallback
6230 logic is chosen. This is supposed to be useful for general purpose
6231 distributions, as it allows minimizing the list of dependencies the
6232 systemd packages pull in, permitting building of more minimal OS
6233 images, while still making use of these "weak" dependencies should
6234 they be installed. Since many package managers automatically
6235 synthesize package dependencies from ELF shared library dependencies,
6236 some additional manual packaging work has to be done now to replace
6237 those (slightly downgraded from "required" to "recommended" or
6238 whatever is conceptually suitable for the package manager). Note that
6239 this change does not alter build-time behaviour: as before the
6240 build-time dependencies have to be installed during build, even if
6241 they now are optional during runtime.
6243 * sd-event.h gained a new call sd_event_add_time_relative() for
6244 installing timers relative to the current time. This is mostly a
6245 convenience wrapper around the pre-existing sd_event_add_time() call
6246 which installs absolute timers.
6248 * sd-event event sources may now be placed in a new "exit-on-failure"
6249 mode, which may be controlled via the new
6250 sd_event_source_get_exit_on_failure() and
6251 sd_event_source_set_exit_on_failure() functions. If enabled, any
6252 failure returned by the event source handler functions will result in
6253 exiting the event loop (unlike the default behaviour of just
6254 disabling the event source but continuing with the event loop). This
6255 feature is useful to set for all event sources that define "primary"
6256 program behaviour (where failure should be fatal) in contrast to
6257 "auxiliary" behaviour (where failure should remain local).
6259 * Most event source types sd-event supports now accept a NULL handler
6260 function, in which case the event loop is exited once the event
6261 source is to be dispatched, using the userdata pointer — converted to
6262 a signed integer — as exit code of the event loop. Previously this
6263 was supported for IO and signal event sources already. Exit event
6264 sources still do not support this (simply because it makes little
6265 sense there, as the event loop is already exiting when they are
6268 * A new per-unit setting RootImageOptions= has been added which allows
6269 tweaking the mount options for any file system mounted as effect of
6270 the RootImage= setting.
6272 * Another new per-unit setting MountImages= has been added, that allows
6273 mounting additional disk images into the file system tree accessible
6276 * Timer units gained a new FixedRandomDelay= boolean setting. If
6277 enabled, the random delay configured with RandomizedDelaySec= is
6278 selected in a way that is stable on a given system (though still
6279 different for different units).
6281 * Socket units gained a new setting Timestamping= that takes "us", "ns"
6282 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
6285 * systemd-repart now generates JSON output when requested with the new
6288 * systemd-machined's OpenMachineShell() bus call will now pass
6289 additional policy metadata data fields to the PolicyKit
6290 authentication request.
6292 * systemd-tmpfiles gained a new -E switch, which is equivalent to
6293 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
6294 --exclude=/sys. It's particularly useful in combination with --root=,
6295 when operating on OS trees that do not have any of these four runtime
6296 directories mounted, as this means no files below these subtrees are
6297 created or modified, since those mount points should probably remain
6300 * systemd-tmpfiles gained a new --image= switch which is like --root=,
6301 but takes a disk image instead of a directory as argument. The
6302 specified disk image is mounted inside a temporary mount namespace
6303 and the tmpfiles.d/ drop-ins stored in the image are executed and
6304 applied to the image. systemd-sysusers similarly gained a new
6305 --image= switch, that allows the sysusers.d/ drop-ins stored in the
6306 image to be applied onto the image.
6308 * Similarly, the journalctl command also gained an --image= switch,
6309 which is a quick one-step solution to look at the log data included
6312 * journalctl's --output=cat option (which outputs the log content
6313 without any metadata, just the pure text messages) will now make use
6314 of terminal colors when run on a suitable terminal, similarly to the
6317 * JSON group records now support a "description" string that may be
6318 used to add a human-readable textual description to such groups. This
6319 is supposed to match the user's GECOS field which traditionally
6320 didn't have a counterpart for group records.
6322 * The "systemd-dissect" tool that may be used to inspect OS disk images
6323 and that was previously installed to /usr/lib/systemd/ has now been
6324 moved to /usr/bin/, reflecting its updated status of an officially
6325 supported tool with a stable interface. It gained support for a new
6326 --mkdir switch which when combined with --mount has the effect of
6327 creating the directory to mount the image to if it is missing
6328 first. It also gained two new commands --copy-from and --copy-to for
6329 copying files and directories in and out of an OS image without the
6330 need to manually mount it. It also acquired support for a new option
6331 --json= to generate JSON output when inspecting an OS image.
6333 * The cgroup2 file system is now mounted with the
6334 "memory_recursiveprot" mount option, supported since kernel 5.7. This
6335 means that the MemoryLow= and MemoryMin= unit file settings now apply
6336 recursively to whole subtrees.
6338 * systemd-homed now defaults to using the btrfs file system — if
6339 available — when creating home directories in LUKS volumes. This may
6340 be changed with the DefaultFileSystemType= setting in homed.conf.
6341 It's now the default file system in various major distributions and
6342 has the major benefit for homed that it can be grown and shrunk while
6343 mounted, unlike the other contenders ext4 and xfs, which can both be
6344 grown online, but not shrunk (in fact xfs is the technically most
6345 limited option here, as it cannot be shrunk at all).
6347 * JSON user records managed by systemd-homed gained support for
6348 "recovery keys". These are basically secondary passphrases that can
6349 unlock user accounts/home directories. They are computer-generated
6350 rather than user-chosen, and typically have greater entropy.
6351 homectl's --recovery-key= option may be used to add a recovery key to
6352 a user account. The generated recovery key is displayed as a QR code,
6353 so that it can be scanned to be kept in a safe place. This feature is
6354 particularly useful in combination with systemd-homed's support for
6355 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
6356 security tokens are lost. Recovery keys may be entered wherever the
6357 system asks for a password.
6359 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
6360 home directory which indicates that a home directory has not been
6361 deactivated cleanly when offline. This flag is useful to identify
6362 home directories for which the offline discard logic did not run when
6363 offlining, and where it would be a good idea to log in again to catch
6366 * systemctl gained a new parameter --timestamp= which may be used to
6367 change the style in which timestamps are output, i.e. whether to show
6368 them in local timezone or UTC, or whether to show µs granularity.
6370 * Alibaba's "pouch" container manager is now detected by
6371 systemd-detect-virt, ConditionVirtualization= and similar
6372 constructs. Similar, they now also recognize IBM PowerVM machine
6375 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
6376 place to use for propagating external mounts into the
6377 container. Similarly /run/host/notify is now used as the socket path
6378 for container payloads to communicate with the container manager
6379 using sd_notify(). The container manager now uses the
6380 /run/host/inaccessible/ directory to place "inaccessible" file nodes
6381 of all relevant types which may be used by the container payload as
6382 bind mount source to over-mount inodes to make them inaccessible.
6383 /run/host/container-manager will now be initialized with the same
6384 string as the $container environment variable passed to the
6385 container's PID 1. /run/host/container-uuid will be initialized with
6386 the same string as $container_uuid. This means the /run/host/
6387 hierarchy is now the primary way to make host resources available to
6388 the container. The Container Interface documents these new files and
6391 https://systemd.io/CONTAINER_INTERFACE
6393 * Support for the "ConditionNull=" unit file condition has been
6394 deprecated and undocumented for 6 years. systemd started to warn
6395 about its use 1.5 years ago. It has now been removed entirely.
6397 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
6398 a sd_bus_error struct and a list of error names, and checks if the
6399 error matches one of these names. It's a convenience wrapper that is
6400 useful in cases where multiple errors shall be handled the same way.
6402 * A new system call filter list "@known" has been added, that contains
6403 all system calls known at the time systemd was built.
6405 * Behaviour of system call filter allow lists has changed slightly:
6406 system calls that are contained in @known will result in EPERM by
6407 default, while those not contained in it result in ENOSYS. This
6408 should improve compatibility because known system calls will thus be
6409 communicated as prohibited, while unknown (and thus newer ones) will
6410 be communicated as not implemented, which hopefully has the greatest
6411 chance of triggering the right fallback code paths in client
6414 * "systemd-analyze syscall-filter" will now show two separate sections
6415 at the bottom of the output: system calls known during systemd build
6416 time but not included in any of the filter groups shown above, and
6417 system calls defined on the local kernel but known during systemd
6420 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
6421 systemd-nspawn all system call filter violations will be logged by
6422 the kernel (audit). This is useful for tracking down system calls
6423 invoked by container payloads that are prohibited by the container's
6424 system call filter policy.
6426 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
6427 systemd-nspawn (and other programs that use seccomp) all seccomp
6428 filtering is turned off.
6430 * Two new unit file settings ProtectProc= and ProcSubset= have been
6431 added that expose the hidepid= and subset= mount options of procfs.
6432 All processes of the unit will only see processes in /proc that are
6433 are owned by the unit's user. This is an important new sandboxing
6434 option that is recommended to be set on all system services. All
6435 long-running system services that are included in systemd itself set
6436 this option now. This option is only supported on kernel 5.8 and
6437 above, since the hidepid= option supported on older kernels was not a
6438 per-mount option but actually applied to the whole PID namespace.
6440 * Socket units gained a new boolean setting FlushPending=. If enabled
6441 all pending socket data/connections are flushed whenever the socket
6442 unit enters the "listening" state, i.e. after the associated service
6445 * The unit file setting NUMAMask= gained a new "all" value: when used,
6446 all existing NUMA nodes are added to the NUMA mask.
6448 * A new "credentials" logic has been added to system services. This is
6449 a simple mechanism to pass privileged data to services in a safe and
6450 secure way. It's supposed to be used to pass per-service secret data
6451 such as passwords or cryptographic keys but also associated less
6452 private information such as user names, certificates, and similar to
6453 system services. Each credential is identified by a short user-chosen
6454 name and may contain arbitrary binary data. Two new unit file
6455 settings have been added: SetCredential= and LoadCredential=. The
6456 former allows setting a credential to a literal string, the latter
6457 sets a credential to the contents of a file (or data read from a
6458 user-chosen AF_UNIX stream socket). Credentials are passed to the
6459 service via a special credentials directory, one file for each
6460 credential. The path to the credentials directory is passed in a new
6461 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
6462 are passed in the file system they may be easily referenced in
6463 ExecStart= command lines too, thus no explicit support for the
6464 credentials logic in daemons is required (though ideally daemons
6465 would look for the bits they need in $CREDENTIALS_DIRECTORY
6466 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
6467 backed by unswappable memory if privileges allow it, immutable if
6468 privileges allow it, is accessible only to the service's UID, and is
6469 automatically destroyed when the service stops.
6471 * systemd-nspawn supports the same credentials logic. It can both
6472 consume credentials passed to it via the aforementioned
6473 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
6474 to its payload. The service manager/PID 1 has been updated to match
6475 this: it can also accept credentials from the container manager that
6476 invokes it (in fact: any process that invokes it), and passes them on
6477 to its services. Thus, credentials can be propagated recursively down
6478 the tree: from a system's service manager to a systemd-nspawn
6479 service, to the service manager that runs as container payload and to
6480 the service it runs below. Credentials may also be added on the
6481 systemd-nspawn command line, using new --set-credential= and
6482 --load-credential= command line switches that match the
6483 aforementioned service settings.
6485 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
6486 the partition drop-ins which may be used to format/LUKS
6487 encrypt/populate any created partitions. The partitions are
6488 encrypted/formatted/populated before they are registered in the
6489 partition table, so that they appear atomically: either the
6490 partitions do not exist yet or they exist fully encrypted, formatted,
6491 and populated — there is no time window where they are
6492 "half-initialized". Thus the system is robust to abrupt shutdown: if
6493 the tool is terminated half-way during its operations on next boot it
6494 will start from the beginning.
6496 * systemd-repart's --size= operation gained a new "auto" value. If
6497 specified, and operating on a loopback file it is automatically sized
6498 to the minimal size the size constraints permit. This is useful to
6499 use "systemd-repart" as an image builder for minimally sized images.
6501 * systemd-resolved now gained a third IPC interface for requesting name
6502 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
6503 interface is now supported. The nss-resolve NSS module has been
6504 modified to use this new interface instead of D-Bus. Using Varlink
6505 has a major benefit over D-Bus: it works without a broker service,
6506 and thus already during earliest boot, before the dbus daemon has
6507 been started. This means name resolution via systemd-resolved now
6508 works at the same time systemd-networkd operates: from earliest boot
6509 on, including in the initrd.
6511 * systemd-resolved gained support for a new DNSStubListenerExtra=
6512 configuration file setting which may be used to specify additional IP
6513 addresses the built-in DNS stub shall listen on, in addition to the
6514 main one on 127.0.0.53:53.
6516 * Name lookups issued via systemd-resolved's D-Bus and Varlink
6517 interfaces (and thus also via glibc NSS if nss-resolve is used) will
6518 now honour a trailing dot in the hostname: if specified the search
6519 path logic is turned off. Thus "resolvectl query foo." is now
6520 equivalent to "resolvectl query --search=off foo.".
6522 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
6523 exposes how /etc/resolv.conf is currently managed: by resolved (and
6524 in which mode if so) or another subsystem. "resolvctl" will display
6525 this property in its status output.
6527 * The resolv.conf snippets systemd-resolved provides will now set "."
6528 as the search domain if no other search domain is known. This turns
6529 off the derivation of an implicit search domain by nss-dns for the
6530 hostname, when the hostname is set to an FQDN. This change is done to
6531 make nss-dns using resolv.conf provided by systemd-resolved behave
6532 more similarly to nss-resolve.
6534 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
6535 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
6536 "birth" time (btime) of a file in addition to the atime, mtime, and
6539 * systemd-analyze gained a new verb "capability" that lists all known
6540 capabilities by the systemd build and by the kernel.
6542 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
6543 advance the system clock to it at boot if it is noticed to be before
6544 that time. Previously, PID 1 would only advance the time to an epoch
6545 time that is set during build-time. With this new file OS builders
6546 can change this epoch timestamp on individual OS images without
6547 having to rebuild systemd.
6549 * systemd-logind will now listen to the KEY_RESTART key from the Linux
6550 input layer and reboot the system if it is pressed, similarly to how
6551 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
6552 was originally defined in the Multimedia context (to restart playback
6553 of a song or film), but is now primarily used in various embedded
6554 devices for "Reboot" buttons. Accordingly, systemd-logind will now
6555 honour it as such. This may configured in more detail via the new
6556 HandleRebootKey= and RebootKeyIgnoreInhibited=.
6558 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
6559 copying OS trees, for example in "systemd-nspawn --ephemeral",
6560 "systemd-nspawn --template=", "machinectl clone" and similar. This is
6561 useful when operating with OSTree images, which use hardlinks heavily
6562 throughout, and where such copies previously resulting in "exploding"
6565 * systemd-nspawn's --console= setting gained support for a new
6566 "autopipe" value, which is identical to "interactive" when invoked on
6567 a TTY, and "pipe" otherwise.
6569 * systemd-networkd's .network files gained support for explicitly
6570 configuring the multicast membership entries of bridge devices in the
6571 [BridgeMDB] section. It also gained support for the PIE queuing
6572 discipline in the [FlowQueuePIE] sections.
6574 * systemd-networkd's .netdev files may now be used to create "BareUDP"
6575 tunnels, configured in the new [BareUDP] setting.
6577 * systemd-networkd's Gateway= setting in .network files now accepts the
6578 special values "_dhcp4" and "_ipv6ra" to configure additional,
6579 locally defined, explicit routes to the gateway acquired via DHCP or
6580 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
6581 but still accepted for backwards compatibility.
6583 * systemd-networkd's [IPv6PrefixDelegation] section and
6584 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
6585 IPv6SendRA= (the old names are still accepted for backwards
6588 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
6589 boolean setting in [Network] section. If enabled, the delegated prefix
6590 gained by another link will be configured, and an address within the
6591 prefix will be assigned.
6593 * systemd-networkd's .network files gained the Announce= boolean setting
6594 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
6595 prefix will be announced through IPv6 router advertisement (IPv6 RA).
6596 The setting is enabled by default.
6598 * VXLAN tunnels may now be marked as independent of any underlying
6599 network interface via the new Independent= boolean setting.
6601 * systemctl gained support for two new verbs: "service-log-level" and
6602 "service-log-target" may be used on services that implement the
6603 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
6604 adjust the log level and target. All of systemd's long-running
6605 services support this now, but ideally all system services would
6606 implement this interface to make the system more uniformly
6609 * The SystemCallErrorNumber= unit file setting now accepts the new
6610 "kill" and "log" actions, in addition to arbitrary error number
6611 specifications as before. If "kill" the processes are killed on the
6612 event, if "log" the offending system call is audit logged.
6614 * A new SystemCallLog= unit file setting has been added that accepts a
6615 list of system calls that shall be logged about (audit).
6617 * The OS image dissection logic (as used by RootImage= in unit files or
6618 systemd-nspawn's --image= switch) has gained support for identifying
6619 and mounting explicit /usr/ partitions, which are now defined in the
6620 discoverable partition specification. This should be useful for
6621 environments where the root file system is
6622 generated/formatted/populated dynamically on first boot and combined
6623 with an immutable /usr/ tree that is supplied by the vendor.
6625 * In the final phase of shutdown, within the systemd-shutdown binary
6626 we'll now try to detach MD devices (i.e software RAID) in addition to
6627 loopback block devices and DM devices as before. This is supposed to
6628 be a safety net only, in order to increase robustness if things go
6629 wrong. Storage subsystems are expected to properly detach their
6630 storage volumes during regular shutdown already (or in case of
6631 storage backing the root file system: in the initrd hook we return to
6634 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
6635 will now log the thread ID in their log output. This is useful when
6636 working with heavily threaded programs.
6638 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
6639 not use the RDRAND CPU instruction. This is useful in environments
6640 such as replay debuggers where non-deterministic behaviour is not
6643 * The autopaging logic in systemd's various tools (such as systemctl)
6644 has been updated to turn on "secure" mode in "less"
6645 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
6646 detected. This disables invoking external programs from the pager,
6647 via the pipe logic. This behaviour may be overridden via the new
6648 $SYSTEMD_PAGERSECURE environment variable.
6650 * Units which have resource limits (.service, .mount, .swap, .slice,
6651 .socket, and .slice) gained new configuration settings
6652 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
6653 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
6654 limits and optional action taken by systemd-oomd.
6656 * A new service systemd-oomd has been added. It monitors resource
6657 contention for selected parts of the unit hierarchy using the PSI
6658 information reported by the kernel, and kills processes when memory
6659 or swap pressure is above configured limits. This service is only
6660 enabled by default in developer mode (see below) and should be
6661 considered a preview in this release. Behaviour details and option
6662 names are subject to change without the usual backwards-compatibility
6665 * A new helper oomctl has been added to introspect systemd-oomd state.
6666 It is only enabled by default in developer mode and should be
6667 considered a preview without the usual backwards-compatibility
6670 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
6671 enabled, systemd reads the system UID boundaries from /etc/login.defs
6672 at runtime, instead of using the built-in values selected during
6673 build. This is an option to improve compatibility for upgrades from
6674 old systems. It's strongly recommended not to make use of this
6675 functionality on new systems (or even enable it during build), as it
6676 makes something runtime-configurable that is mostly an implementation
6677 detail of the OS, and permits avoidable differences in deployments
6678 that create all kinds of problems in the long run.
6680 * New meson option '-Dmode=developer|release' has been added. When
6681 'developer', additional checks and features are enabled that are
6682 relevant during upstream development, e.g. verification that
6683 semi-automatically-generated documentation has been properly updated
6684 following API changes. Those checks are considered hints for
6685 developers and are not actionable in downstream builds. In addition,
6686 extra features that are not ready for general consumption may be
6687 enabled in developer mode. It is thus recommended to set
6688 '-Dmode=release' in end-user and distro builds.
6690 * systemd-cryptsetup gained support for processing detached LUKS
6691 headers specified on the kernel command line via the header=
6692 parameter of the luks.options= kernel command line option. The same
6693 device/path syntax as for key files is supported for header files
6696 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
6697 slot index data for devices that are connected through a PCI bridge
6698 where the _SUN index is associated with the bridge instead of the
6699 network device itself. Previously this would create ambiguous device
6700 naming if multiple network interfaces were connected to the same PCI
6701 bridge. Since this is a naming scheme incompatibility on systems that
6702 possess hardware like this it has been introduced as new naming
6703 scheme "v247". The previous scheme can be selected via the
6704 "net.naming_scheme=v245" kernel command line parameter.
6706 * ConditionFirstBoot= semantics have been modified to be safe towards
6707 abnormal system power-off during first boot. Specifically, the
6708 "systemd-machine-id-commit.service" service now acts as boot
6709 milestone indicating when the first boot process is sufficiently
6710 complete in order to not consider the next following boot also a
6711 first boot. If the system is reset before this unit is reached the
6712 first time, the next boot will still be considered a first boot; once
6713 it has been reached, no further boots will be considered a first
6714 boot. The "first-boot-complete.target" unit now acts as official hook
6715 point to order against this. If a service shall be run on every boot
6716 until the first boot fully succeeds it may thus be ordered before
6717 this target unit (and pull it in) and carry ConditionFirstBoot=
6720 * bootctl's set-default and set-oneshot commands now accept the three
6721 special strings "@default", "@oneshot", "@current" in place of a boot
6722 entry id. These strings are resolved to the current default and
6723 oneshot boot loader entry, as well as the currently booted one. Thus
6724 a command "bootctl set-default @current" may be used to make the
6725 currently boot menu item the new default for all subsequent boots.
6727 * "systemctl edit" has been updated to show the original effective unit
6728 contents in commented form in the text editor.
6730 * Units in user mode are now segregated into three new slices:
6731 session.slice (units that form the core of graphical session),
6732 app.slice ("normal" user applications), and background.slice
6733 (low-priority tasks). Unless otherwise configured, user units are
6734 placed in app.slice. The plan is to add resource limits and
6735 protections for the different slices in the future.
6737 * New GPT partition types for RISCV32/64 for the root and /usr
6738 partitions, and their associated Verity partitions have been defined,
6739 and are now understood by systemd-gpt-auto-generator, and the OS
6740 image dissection logic.
6742 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
6743 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
6744 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
6745 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
6746 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
6747 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
6748 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
6749 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
6750 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
6751 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
6752 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
6753 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
6754 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
6755 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
6756 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
6757 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
6758 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
6759 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
6760 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
6761 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
6762 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
6763 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
6764 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
6765 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
6766 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
6767 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
6768 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
6769 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
6770 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
6771 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
6772 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
6773 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
6774 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
6776 – Warsaw, 2020-11-26
6780 * The service manager gained basic support for cgroup v2 freezer. Units
6781 can now be suspended or resumed either using new systemctl verbs,
6782 freeze and thaw respectively, or via D-Bus.
6784 * PID 1 may now automatically load pre-compiled AppArmor policies from
6785 /etc/apparmor/earlypolicy during early boot.
6787 * The CPUAffinity= setting in service unit files now supports a new
6788 special value "numa" that causes the CPU affinity masked to be set
6789 based on the NUMA mask.
6791 * systemd will now log about all left-over processes remaining in a
6792 unit when the unit is stopped. It will now warn about services using
6793 KillMode=none, as this is generally an unsafe thing to make use of.
6795 * Two new unit file settings
6796 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
6797 added. They may be used to check whether a specific file system path
6798 resides on a block device that is encrypted on the block level
6799 (i.e. using dm-crypt/LUKS).
6801 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
6802 has been added that may be used for simple environment checks. This
6803 is particularly useful when passing in environment variables from a
6804 container manager (or from PAM in case of the systemd --user
6807 * .service unit files now accept a new setting CoredumpFilter= which
6808 allows configuration of the memory sections coredumps of the
6809 service's processes shall include.
6811 * .mount units gained a new ReadWriteOnly= boolean option. If set
6812 it will not be attempted to mount a file system read-only if mounting
6813 in read-write mode doesn't succeed. An option x-systemd.rw-only is
6814 available in /etc/fstab to control the same.
6816 * .socket units gained a new boolean setting PassPacketInfo=. If
6817 enabled, the kernel will attach additional per-packet metadata to all
6818 packets read from the socket, as an ancillary message. This controls
6819 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
6820 depending on socket type.
6822 * .service units gained a new setting RootHash= which may be used to
6823 specify the root hash for verity enabled disk images which are
6824 specified in RootImage=. RootVerity= may be used to specify a path to
6825 the Verity data matching a RootImage= file system. (The latter is
6826 only useful for images that do not contain the Verity data embedded
6827 into the same image that carries a GPT partition table following the
6828 Discoverable Partition Specification). Similarly, systemd-nspawn
6829 gained a new switch --verity-data= that takes a path to a file with
6830 the verity data of the disk image supplied in --image=, if the image
6831 doesn't contain the verity data itself.
6833 * .service units gained a new setting RootHashSignature= which takes
6834 either a base64 encoded PKCS#7 signature of the root hash specified
6835 with RootHash=, or a path to a file to read the signature from. This
6836 allows validation of the root hash against public keys available in
6837 the kernel keyring, and is only supported on recent kernels
6838 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
6839 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
6840 this mechanism has also been added to systemd-veritysetup.
6842 * .service unit files gained two new options
6843 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
6844 tune behaviour if a start or stop timeout is hit, i.e. whether to
6845 terminate the service with SIGTERM, SIGABRT or SIGKILL.
6847 * Most options in systemd that accept hexadecimal values prefixed with
6848 0x in additional to the usual decimal notation now also support octal
6849 notation when the 0o prefix is used and binary notation if the 0b
6852 * Various command line parameters and configuration file settings that
6853 configure key or certificate files now optionally take paths to
6854 AF_UNIX sockets in the file system. If configured that way a stream
6855 connection is made to the socket and the required data read from
6856 it. This is a simple and natural extension to the existing regular
6857 file logic, and permits other software to provide keys or
6858 certificates via simple IPC services, for example when unencrypted
6859 storage on disk is not desired. Specifically, systemd-networkd's
6860 Wireguard and MACSEC key file settings as well as
6861 systemd-journal-gatewayd's and systemd-journal-remote's PEM
6862 key/certificate parameters support this now.
6864 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
6865 configuration files that support specifier expansion learnt six new
6866 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
6867 resolve to the various ID fields from /etc/os-release, %l resolves to
6868 the "short" hostname of the system, i.e. the hostname configured in
6869 the kernel truncated at the first dot.
6871 * Support for the .include syntax in unit files has been removed. The
6872 concept has been obsolete for 6 years and we started warning about
6873 its pending removal 2 years ago (also see NEWS file below). It's
6876 * StandardError= and StandardOutput= in unit files no longer support
6877 the "syslog" and "syslog-console" switches. They were long removed
6878 from the documentation, but will now result in warnings when used,
6879 and be converted to "journal" and "journal+console" automatically.
6881 * If the service setting User= is set to the "nobody" user, a warning
6882 message is now written to the logs (but the value is nonetheless
6883 accepted). Setting User=nobody is unsafe, since the primary purpose
6884 of the "nobody" user is to own all files whose owner cannot be mapped
6885 locally. It's in particular used by the NFS subsystem and in user
6886 namespacing. By running a service under this user's UID it might get
6887 read and even write access to all these otherwise unmappable files,
6888 which is quite likely a major security problem.
6890 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
6891 and others) now have a size and inode limits applied (50% of RAM for
6892 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
6893 that the implicit kernel default is 50% too, so there is no change
6894 in the size limit for /tmp and /dev/shm.
6896 * nss-mymachines lost support for resolution of users and groups, and
6897 now only does resolution of hostnames. This functionality is now
6898 provided by nss-systemd. Thus, the 'mymachines' entry should be
6899 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
6900 (and 'systemd' added if it is not already there).
6902 * A new kernel command line option systemd.hostname= has been added
6903 that allows controlling the hostname that is initialized early during
6906 * A kernel command line option "udev.blockdev_read_only" has been
6907 added. If specified all hardware block devices that show up are
6908 immediately marked as read-only by udev. This option is useful for
6909 making sure that a specific boot under no circumstances modifies data
6910 on disk. Use "blockdev --setrw" to undo the effect of this, per
6913 * A new boolean kernel command line option systemd.swap= has been
6914 added, which may be used to turn off automatic activation of swap
6915 devices listed in /etc/fstab.
6917 * New kernel command line options systemd.condition_needs_update= and
6918 systemd.condition_first_boot= have been added, which override the
6919 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
6922 * A new kernel command line option systemd.clock_usec= has been added
6923 that allows setting the system clock to the specified time in µs
6924 since Jan 1st, 1970 early during boot. This is in particular useful
6925 in order to make test cases more reliable.
6927 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
6928 systemd-coredump to save core files for suid processes. When saving
6929 the core file, systemd-coredump will use the effective uid and gid of
6930 the process that faulted.
6932 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
6933 now automatically set to "Y" at boot, in order to enable pstore
6934 generation for collection with systemd-pstore.
6936 * We provide a set of udev rules to enable auto-suspend on PCI and USB
6937 devices that were tested to correctly support it. Previously, this
6938 was distributed as a set of udev rules, but has now been replaced by
6939 by a set of hwdb entries (and a much shorter udev rule to take action
6940 if the device modalias matches one of the new hwdb entries).
6942 As before, entries are periodically imported from the database
6943 maintained by the ChromiumOS project. If you have a device that
6944 supports auto-suspend correctly and where it should be enabled by
6945 default, please submit a patch that adds it to the database (see
6946 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
6948 * systemd-udevd gained the new configuration option timeout_signal= as well
6949 as a corresponding kernel command line option udev.timeout_signal=.
6950 The option can be used to configure the UNIX signal that the main
6951 daemon sends to the worker processes on timeout. Setting the signal
6952 to SIGABRT is useful for debugging.
6954 * .link files managed by systemd-udevd gained options RxFlowControl=,
6955 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
6956 order to configure various flow control parameters. They also gained
6957 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
6958 frame ring buffer sizes.
6960 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
6961 enabled systemd-networkd manages all routes configured by other tools.
6963 * .network files managed by systemd-networkd gained a new section
6964 [SR-IOV], in order to configure SR-IOV capable network devices.
6966 * systemd-networkd's [IPv6Prefix] section in .network files gained a
6967 new boolean setting Assign=. If enabled an address from the prefix is
6968 automatically assigned to the interface.
6970 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
6971 controls delegated prefixes assigned by DHCPv6 client. The section
6972 has three settings: SubnetID=, Assign=, and Token=. The setting
6973 SubnetID= allows explicit configuration of the preferred subnet that
6974 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
6975 Assign= is enabled (which is the default) an address from any acquired
6976 delegated prefix is automatically chosen and assigned to the
6977 interface. The setting Token= specifies an optional address generation
6980 * systemd-networkd's [Network] section gained a new setting
6981 IPv4AcceptLocal=. If enabled the interface accepts packets with local
6984 * systemd-networkd gained support for configuring the HTB queuing
6985 discipline in the [HierarchyTokenBucket] and
6986 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
6987 be configured in the [PFIFO] section, "GRED" in
6988 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
6989 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
6990 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
6991 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
6992 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
6993 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
6995 * systemd-networkd gained support for a new Termination= setting in the
6996 [CAN] section for configuring the termination resistor. It also
6997 gained a new ListenOnly= setting for controlling whether to only
6998 listen on CAN interfaces, without interfering with traffic otherwise
6999 (which is useful for debugging/monitoring CAN network
7000 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
7001 been added to configure various CAN-FD aspects.
7003 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
7004 When enabled, DHCPv6 will be attempted right-away without requiring an
7005 Router Advertisement packet suggesting it first (i.e. without the 'M'
7006 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
7007 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
7008 the RA packets suggest it.
7010 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
7011 which may be used to turn off use of the gateway information provided
7012 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
7013 used to configure how to process leases that lack a lifetime option.
7015 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
7016 setting SendVendorOption= allowing configuration of additional vendor
7017 options to send in the DHCP requests/responses. The [DHCPv6] section
7018 gained a new SendOption= setting for sending arbitrary DHCP
7019 options. RequestOptions= has been added to request arbitrary options
7020 from the server. UserClass= has been added to set the DHCP user class
7023 * systemd-networkd's [DHCPServer] section gained a new set of options
7024 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
7025 information about these three protocols in the DHCP lease. It also
7026 gained support for including "MUD" URLs ("Manufacturer Usage
7027 Description"). Support for "MUD" URLs was also added to the LLDP
7028 stack, configurable in the [LLDP] section in .network files.
7030 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
7031 mode. Also, the sections now support a new setting SourceMACAddress=.
7033 * systemd-networkd's .netdev files now support a new setting
7034 VLANProtocol= in the [Bridge] section that allows configuration of
7035 the VLAN protocol to use.
7037 * systemd-networkd supports a new Group= setting in the [Link] section
7038 of the .network files, to control the link group.
7040 * systemd-networkd's [Network] section gained a new
7041 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
7042 link local address is generated.
7044 * A new default .network file is now shipped that matches TUN/TAP
7045 devices that begin with "vt-" in their name. Such interfaces will
7046 have IP routing onto the host links set up automatically. This is
7047 supposed to be used by VM managers to trivially acquire a network
7048 interface which is fully set up for host communication, simply by
7049 carefully picking an interface name to use.
7051 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
7052 which sets the route priority for routes specified by the DHCP server.
7054 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
7055 which configures the vendor class information sent to DHCP server.
7057 * The BlackList= settings in .network files' [DHCPv4] and
7058 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
7059 are still understood to provide compatibility.
7061 * networkctl gained the new "forcerenew" command for forcing all DHCP
7062 server clients to renew their lease. The interface "status" output
7063 will now show numerous additional fields of information about an
7064 interface. There are new "up" and "down" commands to bring specific
7065 interfaces up or down.
7067 * systemd-resolved's DNS= configuration option now optionally accepts a
7068 port number (after ":") and a host name (after "#"). When the host
7069 name is specified, the DNS-over-TLS certificate is validated to match
7070 the specified hostname. Additionally, in case of IPv6 addresses, an
7071 interface may be specified (after "%").
7073 * systemd-resolved may be configured to forward single-label DNS names.
7074 This is not standard-conformant, but may make sense in setups where
7075 public DNS servers are not used.
7077 * systemd-resolved's DNS-over-TLS support gained SNI validation.
7079 * systemd-nspawn's --resolv-conf= switch gained a number of new
7080 supported values. Specifically, options starting with "replace-" are
7081 like those prefixed "copy-" but replace any existing resolv.conf
7082 file. And options ending in "-uplink" and "-stub" can now be used to
7083 propagate other flavours of resolv.conf into the container (as
7084 defined by systemd-resolved).
7086 * The various programs included in systemd can now optionally output
7087 their log messages on stderr prefixed with a timestamp, controlled by
7088 the $SYSTEMD_LOG_TIME environment variable.
7090 * systemctl gained a new "-P" switch that is a shortcut for "--value
7093 * "systemctl list-units" and "systemctl list-machines" no longer hide
7094 their first output column with --no-legend. To hide the first column,
7097 * "systemctl reboot" takes the option "--reboot-argument=".
7098 The optional positional argument to "systemctl reboot" is now
7099 being deprecated in favor of this option.
7101 * systemd-run gained a new switch --slice-inherit. If specified the
7102 unit it generates is placed in the same slice as the systemd-run
7105 * systemd-journald gained support for zstd compression of large fields
7106 in journal files. The hash tables in journal files have been hardened
7107 against hash collisions. This is an incompatible change and means
7108 that journal files created with new systemd versions are not readable
7109 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
7110 environment variable for systemd-journald.service is set to 0 this
7111 new hardening functionality may be turned off, so that generated
7112 journal files remain compatible with older journalctl
7115 * journalctl will now include a clickable link in the default output for
7116 each log message for which a URL with further documentation is
7117 known. This is only supported on terminal emulators that support
7118 clickable hyperlinks, and is turned off if a pager is used (since
7119 "less" still doesn't support hyperlinks,
7120 unfortunately). Documentation URLs may be included in log messages
7121 either by including a DOCUMENTATION= journal field in it, or by
7122 associating a journal message catalog entry with the log message's
7123 MESSAGE_ID, which then carries a "Documentation:" tag.
7125 * journald.conf gained a new boolean setting Audit= that may be used to
7126 control whether systemd-journald will enable audit during
7129 * when systemd-journald's log stream is broken up into multiple lines
7130 because the PID of the sender changed this is indicated in the
7131 generated log records via the _LINE_BREAK=pid-change field.
7133 * journalctl's "-o cat" output mode will now show one or more journal
7134 fields specified with --output-fields= instead of unconditionally
7135 MESSAGE=. This is useful to retrieve a very specific set of fields
7136 without any decoration.
7138 * The sd-journal.h API gained two new functions:
7139 sd_journal_enumerate_available_unique() and
7140 sd_journal_enumerate_available_data() that operate like their
7141 counterparts that lack the _available_ in the name, but skip items
7142 that cannot be read and processed by the local implementation
7143 (i.e. are compressed in an unsupported format or such),
7145 * coredumpctl gained a new --file= switch, matching the same one in
7146 journalctl: a specific journal file may be specified to read the
7149 * coredumps collected by systemd-coredump may now be compressed using
7152 * systemd-binfmt gained a new switch --unregister for unregistering all
7153 registered entries at once. This is now invoked automatically at
7154 shutdown, so that binary formats registered with the "F" flag will
7155 not block clean file system unmounting.
7157 * systemd-notify's --pid= switch gained new values: "parent", "self",
7158 "auto" for controlling which PID to send to the service manager: the
7159 systemd-notify process' PID, or the one of the process invoking it.
7161 * systemd-logind's Session bus object learnt a new method call
7162 SetType() for temporarily updating the session type of an already
7163 allocated session. This is useful for upgrading tty sessions to
7164 graphical ones once a compositor is invoked.
7166 * systemd-socket-proxy gained a new switch --exit-idle-time= for
7167 configuring an exit-on-idle time.
7169 * systemd-repart's --empty= setting gained a new value "create". If
7170 specified a new empty regular disk image file is created under the
7171 specified name. Its size may be specified with the new --size=
7172 option. The latter is also supported without the "create" mode, in
7173 order to grow existing disk image files to the specified size. These
7174 two new options are useful when creating or manipulating disk images
7175 instead of operating on actual block devices.
7177 * systemd-repart drop-ins now support a new UUID= setting to control
7178 the UUID to assign to a newly created partition.
7180 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
7183 * systemd-repart's Label= setting now support the usual, simple
7184 specifier expansion.
7186 * systemd-homed's LUKS backend gained the ability to discard empty file
7187 system blocks automatically when the user logs out. This is enabled
7188 by default to ensure that home directories take minimal space when
7189 logged out but get full size guarantees when logged in. This may be
7190 controlled with the new --luks-offline-discard= switch to homectl.
7192 * If systemd-homed detects that /home/ is encrypted as a whole it will
7193 now default to the directory or subvolume backends instead of the
7194 LUKS backend, in order to avoid double encryption. The default
7195 storage and file system may now be configured explicitly, too, via
7196 the new /etc/systemd/homed.conf configuration file.
7198 * systemd-homed now supports unlocking home directories with FIDO2
7199 security tokens that support the 'hmac-secret' extension, in addition
7200 to the existing support for PKCS#11 security token unlocking
7201 support. Note that many recent hardware security tokens support both
7202 interfaces. The FIDO2 support is accessible via homectl's
7203 --fido2-device= option.
7205 * homectl's --pkcs11-uri= setting now accepts two special parameters:
7206 if "auto" is specified and only one suitable PKCS#11 security token
7207 is plugged in, its URL is automatically determined and enrolled for
7208 unlocking the home directory. If "list" is specified a brief table of
7209 suitable PKCS#11 security tokens is shown. Similar, the new
7210 --fido2-device= option also supports these two special values, for
7211 automatically selecting and listing suitable FIDO2 devices.
7213 * The /etc/crypttab tmp option now optionally takes an argument
7214 selecting the file system to use. Moreover, the default is now
7215 changed from ext2 to ext4.
7217 * There's a new /etc/crypttab option "keyfile-erase". If specified the
7218 key file listed in the same line is removed after use, regardless if
7219 volume activation was successful or not. This is useful if the key
7220 file is only acquired transiently at runtime and shall be erased
7221 before the system continues to boot.
7223 * There's also a new /etc/crypttab option "try-empty-password". If
7224 specified, before asking the user for a password it is attempted to
7225 unlock the volume with an empty password. This is useful for
7226 installing encrypted images whose password shall be set on first boot
7227 instead of at installation time.
7229 * systemd-cryptsetup will now attempt to load the keys to unlock
7230 volumes with automatically from files in
7231 /etc/cryptsetup-keys.d/<volume>.key and
7232 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
7234 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
7235 /etc/crypttab, during boot.
7237 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
7238 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
7241 * A new generator systemd-xdg-autostart-generator has been added. It
7242 generates systemd unit files from XDG autostart .desktop files, and
7243 may be used to let the systemd user instance manage services that are
7244 started automatically as part of the desktop session.
7246 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
7247 to query and change the firmware's 'Reboot Into Firmware Interface'
7250 * systemd-firstboot gained a new switch --kernel-command-line= that may
7251 be used to initialize the /etc/kernel/cmdline file of the image. It
7252 also gained a new switch --root-password-hashed= which is like
7253 --root-password= but accepts a pre-hashed UNIX password as
7254 argument. The new option --delete-root-password may be used to unset
7255 any password for the root user (dangerous!). The --root-shell= switch
7256 may be used to control the shell to use for the root account. A new
7257 --force option may be used to override any already set settings with
7258 the parameters specified on the command line (by default, the tool
7259 will not override what has already been set before, i.e. is purely
7262 * systemd-firstboot gained support for a new --image= switch, which is
7263 similar to --root= but accepts the path to a disk image file, on
7264 which it then operates.
7266 * A new sd-path.h API has been added to libsystemd. It provides a
7267 simple API for retrieving various search paths and primary
7268 directories for various resources.
7270 * A new call sd_notify_barrier() has been added to the sd-daemon.h
7271 API. The call will block until all previously sent sd_notify()
7272 messages have been processed by the service manager. This is useful
7273 to remove races caused by a process already having disappeared at the
7274 time a notification message is processed by the service manager,
7275 making correct attribution impossible. The systemd-notify tool will
7276 now make use of this call implicitly, but this can be turned off again
7277 via the new --no-block switch.
7279 * When sending a file descriptor (fd) to the service manager to keep
7280 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
7281 may be specified. If passed the service manager will refrain from
7282 poll()ing on the file descriptor. Traditionally (and when the
7283 parameter is not specified), the service manager will poll it for
7284 POLLHUP or POLLERR events, and immediately close the fds in that
7287 * The service manager (PID1) gained a new D-Bus method call
7288 SetShowStatus() which may be used to control whether it shall show
7289 boot-time status output on the console. This method has a similar
7290 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
7292 * The sd-bus API gained a number of convenience functions that take
7293 va_list arguments rather than "...". For example, there's now
7294 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
7295 it easier to build wrappers that accept variadic arguments and want
7296 to pass a ready va_list structure to sd-bus.
7298 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
7299 flag which alters how the userdata pointer to pass to the callbacks
7300 is determined. When the flag is set, the offset field is converted
7301 as-is into a pointer, without adding it to the object pointer the
7302 vtable is associated with.
7304 * sd-bus now exposes four new functions:
7305 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
7306 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
7307 validate strings to check if they qualify as various D-Bus concepts.
7309 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
7310 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
7311 that simplify adding argument names to D-Bus methods and signals.
7313 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
7315 * Various D-Bus APIs of systemd daemons now have man pages that
7316 document the methods, signals and properties.
7318 * The expectations on user/group name syntax are now documented in
7319 detail; documentation on how classic home directories may be
7320 converted into home directories managed by homed has been added;
7321 documentation regarding integration of homed/userdb functionality in
7322 desktops has been added:
7324 https://systemd.io/USER_NAMES
7325 https://systemd.io/CONVERTING_TO_HOMED
7326 https://systemd.io/USERDB_AND_DESKTOPS
7328 * Documentation for the on-disk Journal file format has been updated
7329 and has now moved to:
7331 https://systemd.io/JOURNAL_FILE_FORMAT
7333 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
7334 has been extended by a set of environment variables that expose
7335 select fields from the host's os-release file to the container
7336 payload. Similarly, host's os-release files can be mounted into the
7337 container underneath /run/host. Together, those mechanisms provide a
7338 standardized way to expose information about the host to the
7339 container payload. Both interfaces are implemented in systemd-nspawn.
7341 * All D-Bus services shipped in systemd now implement the generic
7342 LogControl1 D-Bus API which allows clients to change log level +
7343 target of the service during runtime.
7345 * Only relevant for developers: the mkosi.default symlink has been
7346 dropped from version control. Please create a symlink to one of the
7347 distribution-specific defaults in .mkosi/ based on your preference.
7349 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
7350 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
7351 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
7352 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
7353 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
7354 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
7355 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
7356 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
7357 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
7358 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
7359 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
7360 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
7361 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
7362 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
7363 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
7364 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
7365 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
7366 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
7367 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
7368 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
7369 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
7370 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
7371 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
7372 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
7373 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
7374 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
7375 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
7376 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
7377 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
7378 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
7379 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
7380 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
7381 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
7382 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
7383 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
7384 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
7385 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
7386 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
7388 – Warsaw, 2020-07-30
7392 * A new tool "systemd-repart" has been added, that operates as an
7393 idempotent declarative repartitioner for GPT partition tables.
7394 Specifically, a set of partitions that must or may exist can be
7395 configured via drop-in files, and during every boot the partition
7396 table on disk is compared with these files, creating missing
7397 partitions or growing existing ones based on configurable relative
7398 and absolute size constraints. The tool is strictly incremental,
7399 i.e. does not delete, shrink or move partitions, but only adds and
7400 grows them. The primary use-case is OS images that ship in minimized
7401 form, that on first boot are grown to the size of the underlying
7402 block device or augmented with additional partitions. For example,
7403 the root partition could be extended to cover the whole disk, or a
7404 swap or /home partitions could be added on first boot. It can also be
7405 used for systems that use an A/B update scheme but ship images with
7406 just the A partition, with B added on first boot. The tool is
7407 primarily intended to be run in the initrd, shortly before
7408 transitioning into the host OS, but can also be run after the
7409 transition took place. It automatically discovers the disk backing
7410 the root file system, and should hence not require any additional
7411 configuration besides the partition definition drop-ins. If no
7412 configuration drop-ins are present, no action is taken.
7414 * A new component "userdb" has been added, along with a small daemon
7415 "systemd-userdbd.service" and a client tool "userdbctl". The framework
7416 allows defining rich user and group records in a JSON format,
7417 extending on the classic "struct passwd" and "struct group"
7418 structures. Various components in systemd have been updated to
7419 process records in this format, including systemd-logind and
7420 pam-systemd. The user records are intended to be extensible, and
7421 allow setting various resource management, security and runtime
7422 parameters that shall be applied to processes and sessions of the
7423 user as they log in. This facility is intended to allow associating
7424 such metadata directly with user/group records so that they can be
7425 produced, extended and consumed in unified form. We hope that
7426 eventually frameworks such as sssd will generate records this way, so
7427 that for the first time resource management and various other
7428 per-user settings can be configured in LDAP directories and then
7429 provided to systemd (specifically to systemd-logind and pam-system)
7430 to apply on login. For further details see:
7432 https://systemd.io/USER_RECORD
7433 https://systemd.io/GROUP_RECORD
7434 https://systemd.io/USER_GROUP_API
7436 * A small new service systemd-homed.service has been added, that may be
7437 used to securely manage home directories with built-in encryption.
7438 The complete user record data is unified with the home directory,
7439 thus making home directories naturally migratable. Its primary
7440 back-end is based on LUKS volumes, but fscrypt, plain directories,
7441 and other storage schemes are also supported. This solves a couple of
7442 problems we saw with traditional ways to manage home directories, in
7443 particular when it comes to encryption. For further discussion of
7444 this, see the video of Lennart's talk at AllSystemsGo! 2019:
7446 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
7448 For further details about the format and expectations on home
7449 directories this new daemon makes, see:
7451 https://systemd.io/HOME_DIRECTORY
7453 * systemd-journald is now multi-instantiable. In addition to the main
7454 instance systemd-journald.service there's now a template unit
7455 systemd-journald@.service, with each instance defining a new named
7456 log 'namespace' (whose name is specified via the instance part of the
7457 unit name). A new unit file setting LogNamespace= has been added,
7458 taking such a namespace name, that assigns services to the specified
7459 log namespaces. As each log namespace is serviced by its own
7460 independent journal daemon, this functionality may be used to improve
7461 performance and increase isolation of applications, at the price of
7462 losing global message ordering. Each instance of journald has a
7463 separate set of configuration files, with possibly different disk
7464 usage limitations and other settings.
7466 journalctl now takes a new option --namespace= to show logs from a
7467 specific log namespace. The sd-journal.h API gained
7468 sd_journal_open_namespace() for opening the log stream of a specific
7469 log namespace. systemd-journald also gained the ability to exit on
7470 idle, which is useful in the context of log namespaces, as this means
7471 log daemons for log namespaces can be activated automatically on
7472 demand and will stop automatically when no longer used, minimizing
7475 * When systemd-tmpfiles copies a file tree using the 'C' line type it
7476 will now label every copied file according to the SELinux database.
7478 * When systemd/PID 1 detects it is used in the initrd it will now boot
7479 into initrd.target rather than default.target by default. This should
7480 make it simpler to build initrds with systemd as for many cases the
7481 only difference between a host OS image and an initrd image now is
7482 the presence of the /etc/initrd-release file.
7484 * A new kernel command line option systemd.cpu_affinity= is now
7485 understood. It's equivalent to the CPUAffinity= option in
7486 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
7487 itself and the default for all other processes.
7489 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
7490 equivalent), the SELinux database is now reloaded, ensuring that
7491 sockets and other file system objects are generated taking the new
7492 database into account.
7494 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
7495 "quiet" has been changed to imply that instead of
7496 "systemd.show-status=auto". In this mode, only messages about errors
7497 and significant delays in boot are shown on the console.
7499 * The sd-event.h API gained native support for the new Linux "pidfd"
7500 concept. This permits watching processes using file descriptors
7501 instead of PID numbers, which fixes a number of races and makes
7502 process supervision more robust and efficient. All of systemd's
7503 components will now use pidfds if the kernel supports it for process
7504 watching, with the exception of PID 1 itself, unfortunately. We hope
7505 to move PID 1 to exclusively using pidfds too eventually, but this
7506 requires some more kernel work first. (Background: PID 1 watches
7507 processes using waitid() with the P_ALL flag, and that does not play
7508 together nicely with pidfds yet.)
7510 * Closely related to this, the sd-event.h API gained two new calls
7511 sd_event_source_send_child_signal() (for sending a signal to a
7512 watched process) and sd_event_source_get_child_process_own() (for
7513 marking a process so that it is killed automatically whenever the
7514 event source watching it is freed).
7516 * systemd-networkd gained support for configuring Token Bucket Filter
7517 (TBF) parameters in its qdisc configuration support. Similarly,
7518 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
7519 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
7521 * systemd-networkd gained support for Intermediate Functional Block
7522 (IFB) network devices.
7524 * systemd-networkd gained support for configuring multi-path IP routes,
7525 using the new MultiPathRoute= setting in the [Route] section.
7527 * systemd-networkd's DHCPv4 client has been updated to support a new
7528 SendDecline= option. If enabled, duplicate address detection is done
7529 after a DHCP offer is received from the server. If a conflict is
7530 detected, the address is declined. The DHCPv4 client also gained
7531 support for a new RouteMTUBytes= setting that allows to configure the
7532 MTU size to be used for routes generated from DHCPv4 leases.
7534 * The PrefixRoute= setting in systemd-networkd's [Address] section of
7535 .network files has been deprecated, and replaced by AddPrefixRoute=,
7536 with its sense inverted.
7538 * The Gateway= setting of [Route] sections of .network files gained
7539 support for a special new value "_dhcp". If set, the configured
7540 static route uses the gateway host configured via DHCP.
7542 * New User= and SuppressPrefixLength= settings have been implemented
7543 for the [RoutingPolicyRule] section of .network files to configure
7544 source routing based on UID ranges and prefix length, respectively.
7546 * The Type= match property of .link files has been generalized to
7547 always match the device type shown by 'networkctl status', even for
7548 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
7551 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
7552 D-Bus message object as "sensitive". Those objects are erased from
7553 memory when they are freed. This concept is intended to be used for
7554 messages that contain security sensitive data. A new flag
7555 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
7556 in sd-bus vtables, causing any incoming and outgoing messages of
7557 those methods to be implicitly marked as "sensitive".
7559 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
7560 contents of a message (or parts thereof) to standard output for
7563 * systemd-sysusers gained support for creating users with the primary
7564 group named differently than the user.
7566 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
7567 gained support for growing XFS partitions. Previously it supported
7568 only ext4 and btrfs partitions.
7570 * The support for /etc/crypttab gained a new x-initrd.attach option. If
7571 set, the specified encrypted volume is unlocked already in the
7572 initrd. This concept corresponds to the x-initrd.mount option in
7575 * systemd-cryptsetup gained native support for unlocking encrypted
7576 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
7577 encryption of volumes to YubiKeys. This is exposed in the new
7578 pkcs11-uri= option in /etc/crypttab.
7580 * The /etc/fstab support in systemd now supports two new mount options
7581 x-systemd.{required,wanted}-by=, for explicitly configuring the units
7582 that the specified mount shall be pulled in by, in place of
7583 the usual local-fs.target/remote-fs.target.
7585 * The https://systemd.io/ web site has been relaunched, directly
7586 populated with most of the documentation included in the systemd
7587 repository. systemd also acquired a new logo, thanks to Tobias
7590 * systemd-udevd gained support for managing "alternative" network
7591 interface names, as supported by new Linux kernels. For the first
7592 time this permits assigning multiple (and longer!) names to a network
7593 interface. systemd-udevd will now by default assign the names
7594 generated via all supported naming schemes to each interface. This
7595 may be further tweaked with .link files and the AlternativeName= and
7596 AlternativeNamesPolicy= settings. Other components of systemd have
7597 been updated to support the new alternative names wherever
7598 appropriate. For example, systemd-nspawn will now generate
7599 alternative interface names for the host-facing side of container
7600 veth links based on the full container name without truncation.
7602 * systemd-nspawn interface naming logic has been updated in another way
7603 too: if the main interface name (i.e. as opposed to new-style
7604 "alternative" names) based on the container name is truncated, a
7605 simple hashing scheme is used to give different interface names to
7606 multiple containers whose names all begin with the same prefix. Since
7607 this changes the primary interface names pointing to containers if
7608 truncation happens, the old scheme may still be requested by
7609 selecting an older naming scheme, via the net.naming_scheme= kernel
7610 command line option.
7612 * PrivateUsers= in service files now works in services run by the
7613 systemd --user per-user instance of the service manager.
7615 * A new per-service sandboxing option ProtectClock= has been added that
7616 locks down write access to the system clock. It takes away device
7617 node access to /dev/rtc as well as the system calls that set the
7618 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
7619 Note that this option does not affect access to auxiliary services
7620 that allow changing the clock, for example access to
7623 * The systemd-id128 tool gained a new "show" verb for listing or
7624 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
7625 GPT partition table types.
7627 * The Discoverable Partitions Specification has been updated to support
7628 /var and /var/tmp partition discovery. Support for this has been
7629 added to systemd-gpt-auto-generator. For details see:
7631 https://systemd.io/DISCOVERABLE_PARTITIONS
7633 * "systemctl list-unit-files" has been updated to show a new column
7634 with the suggested enablement state based on the vendor preset files
7635 for the respective units.
7637 * "systemctl" gained a new option "--with-dependencies". If specified
7638 commands such as "systemctl status" or "systemctl cat" will now show
7639 all specified units along with all units they depend on.
7641 * networkctl gained support for showing per-interface logs in its
7644 * systemd-networkd-wait-online gained support for specifying the maximum
7645 operational state to wait for, and to wait for interfaces to
7648 * The [Match] section of .link and .network files now supports a new
7649 option PermanentMACAddress= which may be used to check against the
7650 permanent MAC address of a network device even if a randomized MAC
7653 * The [TrafficControlQueueingDiscipline] section in .network files has
7654 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
7655 dropped from the individual setting names.
7657 * Any .link and .network files that have an empty [Match] section (this
7658 also includes empty and commented-out files) will now be
7659 rejected. systemd-udev and systemd-networkd started warning about
7660 such files in version 243.
7662 * systemd-logind will now validate access to the operation of changing
7663 the virtual terminal via a polkit action. By default, only users
7664 with at least one session on a local VT are granted permission.
7666 * When systemd sets up PAM sessions that invoked service processes
7667 shall run in, the pam_setcred() API is now invoked, thus permitting
7668 PAM modules to set additional credentials for the processes.
7670 * portablectl attach/detach verbs now accept --now and --enable options
7671 to combine attachment with enablement and invocation, or detachment
7672 with stopping and disablement.
7674 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
7675 fixed, which in turn exposed bugs in unit configuration of services
7676 which have Type=oneshot and should only run once, but do not have
7677 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
7678 service may be started again after exiting successfully, for example
7679 as a dependency in another transaction. Affected services included
7680 some internal systemd services (most notably
7681 systemd-vconsole-setup.service, which was updated to have
7682 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
7683 plymouth has been suitably updated or patched before upgrading to
7684 this systemd release. See
7685 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
7686 additional discussion.
7688 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
7689 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
7690 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
7691 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
7692 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
7693 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
7694 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
7695 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
7696 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
7697 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
7698 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
7699 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
7700 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
7701 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
7702 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
7703 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
7704 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
7705 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
7706 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
7707 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
7708 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
7709 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
7710 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
7711 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
7714 – Warsaw, 2020-03-06
7718 * Support for the cpuset cgroups v2 controller has been added.
7719 Processes may be restricted to specific CPUs using the new
7720 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
7721 AllowedMemoryNodes= setting.
7723 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
7724 now be configured using a new RestartKillSignal= setting. This
7725 allows units which signals to request termination to implement
7726 different behaviour when stopping in preparation for a restart.
7728 * "systemctl clean" may now be used also for socket, mount, and swap
7731 * systemd will also read configuration options from the EFI variable
7732 SystemdOptions. This may be used to configure systemd behaviour when
7733 modifying the kernel command line is inconvenient, but configuration
7734 on disk is read too late, for example for the options related to
7735 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
7736 set the EFI variable.
7738 * systemd will now disable printk ratelimits in early boot. This should
7739 allow us to capture more logs from the early boot phase where normal
7740 storage is not available and the kernel ring buffer is used for
7741 logging. Configuration on the kernel command line has higher priority
7742 and overrides the systemd setting.
7744 systemd programs which log to /dev/kmsg directly use internal
7745 ratelimits to prevent runaway logging. (Normally this is only used
7746 during early boot, so in practice this change has very little
7749 * Unit files now support top level dropin directories of the form
7750 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
7751 that affects all corresponding unit files.
7753 * systemctl gained support for 'stop --job-mode=triggering' which will
7754 stop the specified unit and any units which could trigger it.
7756 * Unit status display now includes units triggering and triggered by
7757 the unit being shown.
7759 * The RuntimeMaxSec= setting is now supported by scopes, not just
7760 .service units. This is particularly useful for PAM sessions which
7761 create a scope unit for the user login. systemd.runtime_max_sec=
7762 setting may used with the pam_systemd module to limit the duration
7763 of the PAM session, for example for time-limited logins.
7765 * A new @pkey system call group is now defined to make it easier to
7766 allow-list memory protection syscalls for containers and services
7767 which need to use them.
7769 * systemd-udevd: removed the 30s timeout for killing stale workers on
7770 exit. systemd-udevd now waits for workers to finish. The hard-coded
7771 exit timeout of 30s was too short for some large installations, where
7772 driver initialization could be prematurely interrupted during initrd
7773 processing if the root file system had been mounted and init was
7774 preparing to switch root. If udevd is run without systemd and workers
7775 are hanging while udevd receives an exit signal, udevd will now exit
7776 when udev.event_timeout is reached for the last hanging worker. With
7777 systemd, the exit timeout can additionally be configured using
7778 TimeoutStopSec= in systemd-udevd.service.
7780 * udev now provides a program (fido_id) that identifies FIDO CTAP1
7781 ("U2F")/CTAP2 security tokens based on the usage declared in their
7782 report and descriptor and outputs suitable environment variables.
7783 This replaces the externally maintained allow lists of all known
7784 security tokens that were used previously.
7786 * Automatically generated autosuspend udev rules for allow-listed
7787 devices have been imported from the Chromium OS project. This should
7788 improve power saving with many more devices.
7790 * udev gained a new "CONST{key}=value" setting that allows matching
7791 against system-wide constants without forking a helper binary.
7792 Currently "arch" and "virt" keys are supported.
7794 * udev now opens CDROMs in non-exclusive mode when querying their
7795 capabilities. This should fix issues where other programs trying to
7796 use the CDROM cannot gain access to it, but carries a risk of
7797 interfering with programs writing to the disk, if they did not open
7798 the device in exclusive mode as they should.
7800 * systemd-networkd does not create a default route for IPv4 link local
7801 addressing anymore. The creation of the route was unexpected and was
7802 breaking routing in various cases, but people who rely on it being
7803 created implicitly will need to adjust. Such a route may be requested
7804 with DefaultRouteOnDevice=yes.
7806 Similarly, systemd-networkd will not assign a link-local IPv6 address
7807 when IPv6 link-local routing is not enabled.
7809 * Receive and transmit buffers may now be configured on links with
7810 the new RxBufferSize= and TxBufferSize= settings.
7812 * systemd-networkd may now advertise additional IPv6 routes. A new
7813 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
7816 * systemd-networkd may now configure "next hop" routes using the
7817 [NextHop] section and Gateway= and Id= settings.
7819 * systemd-networkd will now retain DHCP config on restarts by default
7820 (but this may be overridden using the KeepConfiguration= setting).
7821 The default for SendRelease= has been changed to true.
7823 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
7824 received from the server.
7826 The client will use the received SIP server list if UseSIP=yes is
7829 The client may be configured to request specific options from the
7830 server using a new RequestOptions= setting.
7832 The client may be configured to send arbitrary options to the server
7833 using a new SendOption= setting.
7835 A new IPServiceType= setting has been added to configure the "IP
7836 service type" value used by the client.
7838 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
7839 request prefix hints in the DHCPv6 solicitation.
7841 * The DHCPv4 server may be configured to send arbitrary options using
7842 a new SendOption= setting.
7844 * The DHCPv4 server may now be configured to emit SIP server list using
7845 the new EmitSIP= and SIP= settings.
7847 * systemd-networkd and networkctl may now renew DHCP leases on demand.
7848 networkctl has a new 'networkctl renew' verb.
7850 * systemd-networkd may now reconfigure links on demand. networkctl
7851 gained two new verbs: "reload" will reload the configuration, and
7852 "reconfigure DEVICE…" will reconfigure one or more devices.
7854 * .network files may now match on SSID and BSSID of a wireless network,
7855 i.e. the access point name and hardware address using the new SSID=
7856 and BSSID= options. networkctl will display the current SSID and
7857 BSSID for wireless links.
7859 .network files may also match on the wireless network type using the
7860 new WLANInterfaceType= option.
7862 * systemd-networkd now includes default configuration that enables
7863 link-local addressing when connected to an ad-hoc wireless network.
7865 * systemd-networkd may configure the Traffic Control queueing
7866 disciplines in the kernel using the new
7867 [TrafficControlQueueingDiscipline] section and Parent=,
7868 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
7869 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
7870 NetworkEmulatorDuplicateRate= settings.
7872 * systemd-tmpfiles gained a new w+ setting to append to files.
7874 * systemd-analyze dump will now report when the memory configuration in
7875 the kernel does not match what systemd has configured (usually,
7876 because some external program has modified the kernel configuration
7879 * systemd-analyze gained a new --base-time= switch instructs the
7880 'calendar' verb to resolve times relative to that timestamp instead
7881 of the present time.
7883 * journalctl --update-catalog now produces deterministic output (making
7884 reproducible image builds easier).
7886 * A new devicetree-overlay setting is now documented in the Boot Loader
7889 * The default value of the WatchdogSec= setting used in systemd
7890 services (the ones bundled with the project itself) may be set at
7891 configuration time using the -Dservice-watchdog= setting. If set to
7892 empty, the watchdogs will be disabled.
7894 * systemd-resolved validates IP addresses in certificates now when GnuTLS
7897 * libcryptsetup >= 2.0.1 is now required.
7899 * A configuration option -Duser-path= may be used to override the $PATH
7900 used by the user service manager. The default is again to use the same
7901 path as the system manager.
7903 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
7904 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
7907 * Service units gained a new sandboxing option ProtectKernelLogs= which
7908 makes sure the program cannot get direct access to the kernel log
7909 buffer anymore, i.e. the syslog() system call (not to be confused
7910 with the API of the same name in libc, which is not affected), the
7911 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
7912 inaccessible to the service. It's recommended to enable this setting
7913 for all services that should not be able to read from or write to the
7914 kernel log buffer, which are probably almost all.
7916 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
7917 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
7918 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
7919 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
7920 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
7921 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
7922 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
7923 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
7924 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
7925 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
7926 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
7927 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
7928 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
7929 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
7930 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
7931 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
7932 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
7933 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
7934 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
7935 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
7936 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
7937 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
7938 Zach Smith, Zbigniew Jędrzejewski-Szmek
7940 – Warsaw, 2019-11-29
7944 * This release enables unprivileged programs (i.e. requiring neither
7945 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
7946 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
7947 kernel for the whole UNIX group range, i.e. all processes. This
7948 change should be reasonably safe, as the kernel support for it was
7949 specifically implemented to allow safe access to ICMP Echo for
7950 processes lacking any privileges. If this is not desirable, it can be
7951 disabled again by setting the parameter to "1 0".
7953 * Previously, filters defined with SystemCallFilter= would have the
7954 effect that any calling of an offending system call would terminate
7955 the calling thread. This behaviour never made much sense, since
7956 killing individual threads of unsuspecting processes is likely to
7957 create more problems than it solves. With this release the default
7958 action changed from killing the thread to killing the whole
7959 process. For this to work correctly both a kernel version (>= 4.14)
7960 and a libseccomp version (>= 2.4.0) supporting this new seccomp
7961 action is required. If an older kernel or libseccomp is used the old
7962 behaviour continues to be used. This change does not affect any
7963 services that have no system call filters defined, or that use
7964 SystemCallErrorNumber= (and thus see EPERM or another error instead
7965 of being killed when calling an offending system call). Note that
7966 systemd documentation always claimed that the whole process is
7967 killed. With this change behaviour is thus adjusted to match the
7970 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
7971 4194304 by default, i.e. the full 22bit range the kernel allows, up
7972 from the old 16-bit range. This should improve security and
7973 robustness, as PID collisions are made less likely (though certainly
7974 still possible). There are rumours this might create compatibility
7975 problems, though at this moment no practical ones are known to
7976 us. Downstream distributions are hence advised to undo this change in
7977 their builds if they are concerned about maximum compatibility, but
7978 for everybody else we recommend leaving the value bumped. Besides
7979 improving security and robustness this should also simplify things as
7980 the maximum number of allowed concurrent tasks was previously bounded
7981 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
7982 only a single knob is left ("kernel.threads-max"). There have been
7983 concerns that usability is affected by this change because larger PID
7984 numbers are harder to type, but we believe the change from 5 digits
7985 to 7 digits doesn't hamper usability.
7987 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
7988 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
7989 hierarchically set default memory protection values for a particular
7990 subtree of the unit hierarchy.
7992 * Memory protection directives can now take a value of zero, allowing
7993 explicit opting out of a default value propagated by an ancestor.
7995 * systemd now defaults to the "unified" cgroup hierarchy setup during
7996 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
7997 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
7998 change reflects the fact that cgroupsv2 support has matured
7999 substantially in both systemd and in the kernel, and is clearly the
8000 way forward. Downstream production distributions might want to
8001 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
8002 their builds as unfortunately the popular container managers have not
8003 caught up with the kernel API changes.
8005 * Man pages are not built by default anymore (html pages were already
8006 disabled by default), to make development builds quicker. When
8007 building systemd for a full installation with documentation, meson
8008 should be called with -Dman=true and/or -Dhtml=true as appropriate.
8009 The default was changed based on the assumption that quick one-off or
8010 repeated development builds are much more common than full optimized
8011 builds for installation, and people need to pass various other
8012 options to when doing "proper" builds anyway, so the gain from making
8013 development builds quicker is bigger than the one time disruption for
8016 Two scripts are created in the *build* directory to generate and
8017 preview man and html pages on demand, e.g.:
8019 build/man/man systemctl
8020 build/man/html systemd.index
8022 * libidn2 is used by default if both libidn2 and libidn are installed.
8023 Please use -Dlibidn=true if libidn is preferred.
8025 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
8026 big-endian machines. Before, bytes were written and read in native
8027 machine order as exposed by the native libc __cpu_mask interface.
8028 Now, little-endian order is always used (CPUs 0–7 are described by
8029 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
8030 This change fixes D-Bus calls that cross endianness boundary.
8032 The presentation format used for CPUAffinity= by "systemctl show" and
8033 "systemd-analyze dump" is changed to present CPU indices instead of
8034 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
8035 shown as CPUAffinity=03000000000000000000000000000… (on
8036 little-endian) or CPUAffinity=00000000000000300000000000000… (on
8037 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
8038 input format. The maximum integer that will be printed in the new
8039 format is 8191 (four digits), while the old format always used a very
8040 long number (with the length varying by architecture), so they can be
8041 unambiguously distinguished.
8043 * /usr/sbin/halt.local is no longer supported. Implementation in
8044 distributions was inconsistent and it seems this functionality was
8047 To replace this functionality, users should:
8048 - either define a new unit and make it a dependency of final.target
8049 (systemctl add-wants final.target my-halt-local.service)
8050 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
8051 and ensure that it accepts "halt", "poweroff", "reboot", and
8052 "kexec" as an argument, see the description in systemd-shutdown(8).
8054 * When a [Match] section in .link or .network file is empty (contains
8055 no match patterns), a warning will be emitted. Please add any "match
8056 all" pattern instead, e.g. OriginalName=* or Name=* in case all
8057 interfaces should really be matched.
8059 * A new setting NUMAPolicy= may be used to set process memory
8060 allocation policy. This setting can be specified in
8061 /etc/systemd/system.conf and hence will set the default policy for
8062 PID1. The default policy can be overridden on a per-service
8063 basis. The related setting NUMAMask= is used to specify NUMA node
8064 mask that should be associated with the selected policy.
8066 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
8067 generates when processes it manages are reaching their memory limits,
8068 and will place their units in a special state, and optionally kill or
8069 stop the whole unit.
8071 * The service manager will now expose bus properties for the IO
8072 resources used by units. This information is also shown in "systemctl
8073 status" now (for services that have IOAccounting=yes set). Moreover,
8074 the IO accounting data is included in the resource log message
8075 generated whenever a unit stops.
8077 * Units may now configure an explicit timeout to wait for when killed
8078 with SIGABRT, for example when a service watchdog is hit. Previously,
8079 the regular TimeoutStopSec= timeout was applied in this case too —
8080 now a separate timeout may be set using TimeoutAbortSec=.
8082 * Services may now send a special WATCHDOG=trigger message with
8083 sd_notify() to trigger an immediate "watchdog missed" event, and thus
8084 trigger service termination. This is useful both for testing watchdog
8085 handling, but also for defining error paths in services, that shall
8086 be handled the same way as watchdog events.
8088 * There are two new per-unit settings IPIngressFilterPath= and
8089 IPEgressFilterPath= which allow configuration of a BPF program
8090 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
8091 to apply to the IP packet ingress/egress path of all processes of a
8092 unit. This is useful to allow running systemd services with BPF
8093 programs set up externally.
8095 * systemctl gained a new "clean" verb for removing the state, cache,
8096 runtime or logs directories of a service while it is terminated. The
8097 new verb may also be used to remove the state maintained on disk for
8098 timer units that have Persistent= configured.
8100 * During the last phase of shutdown systemd will now automatically
8101 increase the log level configured in the "kernel.printk" sysctl so
8102 that any relevant loggable events happening during late shutdown are
8103 made visible. Previously, loggable events happening so late during
8104 shutdown were generally lost if the "kernel.printk" sysctl was set to
8105 high thresholds, as regular logging daemons are terminated at that
8106 time and thus nothing is written to disk.
8108 * If processes terminated during the last phase of shutdown do not exit
8109 quickly systemd will now show their names after a short time, to make
8110 debugging easier. After a longer timeout they are forcibly killed,
8113 * journalctl (and the other tools that display logs) will now highlight
8114 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
8115 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
8116 are now shown in blue color, to separate them visually from regular
8117 logs. References to configuration files are now turned into clickable
8118 links on terminals that support that.
8120 * systemd-journald will now stop logging to /var/log/journal during
8121 shutdown when /var/ is on a separate mount, so that it can be
8122 unmounted safely during shutdown.
8124 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
8126 * systemd-resolved "Cache=" configuration option in resolved.conf has
8127 been extended to also accept the 'no-negative' value. Previously,
8128 only a boolean option was allowed (yes/no), having yes as the
8129 default. If this option is set to 'no-negative', negative answers are
8130 not cached while the old cache heuristics are used positive answers.
8131 The default remains unchanged.
8133 * The predictable naming scheme for network devices now supports
8134 generating predictable names for "netdevsim" devices.
8136 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
8139 Those two changes form a new net.naming_scheme= entry. Distributions
8140 which want to preserve naming stability may want to set the
8141 -Ddefault-net-naming-scheme= configuration option.
8143 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
8144 interfaces natively.
8146 * systemd-networkd's bridge FDB support now allows configuration of a
8147 destination address for each entry (Destination=), as well as the
8148 VXLAN VNI (VNI=), as well as an option to declare what an entry is
8149 associated with (AssociatedWith=).
8151 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
8152 option for configuring the maximum number of DHCP lease requests. It
8153 also learnt a new BlackList= option for deny-listing DHCP servers (a
8154 similar setting has also been added to the IPv6 RA client), as well
8155 as a SendRelease= option for configuring whether to send a DHCP
8156 RELEASE message when terminating.
8158 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
8159 separately in the [DHCPv4] and [DHCPv6] sections.
8161 * systemd-networkd's DHCP support will now optionally create an
8162 implicit host route to the DNS server specified in the DHCP lease, in
8163 addition to the routes listed explicitly in the lease. This should
8164 ensure that in multi-homed systems DNS traffic leaves the systems on
8165 the interface that acquired the DNS server information even if other
8166 routes such as default routes exist. This behaviour may be turned on
8167 with the new RoutesToDNS= option.
8169 * systemd-networkd's VXLAN support gained a new option
8170 GenericProtocolExtension= for enabling VXLAN Generic Protocol
8171 Extension support, as well as IPDoNotFragment= for setting the IP
8172 "Don't fragment" bit on outgoing packets. A similar option has been
8173 added to the GENEVE support.
8175 * In systemd-networkd's [Route] section you may now configure
8176 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
8177 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
8178 propagation. The Type= setting now supports local, broadcast,
8179 anycast, multicast, any, xresolve routes, too.
8181 * systemd-networkd's [Network] section learnt a new option
8182 DefaultRouteOnDevice= for automatically configuring a default route
8183 onto the network device.
8185 * systemd-networkd's bridging support gained two new options ProxyARP=
8186 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
8187 MulticastRouter= for configuring multicast routing behaviour. A new
8188 option MulticastIGMPVersion= may be used to change bridge's multicast
8189 Internet Group Management Protocol (IGMP) version.
8191 * systemd-networkd's FooOverUDP support gained the ability to configure
8192 local and peer IP addresses via Local= and Peer=. A new option
8193 PeerPort= may be used to configure the peer's IP port.
8195 * systemd-networkd's TUN support gained a new setting VnetHeader= for
8196 tweaking Generic Segment Offload support.
8198 * The address family for policy rules may be specified using the new
8199 Family= option in the [RoutingPolicyRule] section.
8201 * networkctl gained a new "delete" command for removing virtual network
8202 devices, as well as a new "--stats" switch for showing device
8205 * networkd.conf gained a new setting SpeedMeter= and
8206 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
8207 measured speed may be shown by 'networkctl status'.
8209 * "networkctl status" now displays MTU and queue lengths, and more
8210 detailed information about VXLAN and bridge devices.
8212 * systemd-networkd's .network and .link files gained a new Property=
8213 setting in the [Match] section, to match against devices with
8214 specific udev properties.
8216 * systemd-networkd's tunnel support gained a new option
8217 AssignToLoopback= for selecting whether to use the loopback device
8218 "lo" as underlying device.
8220 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
8221 been renamed to LinkLayerAddress=, and it now allows configuration of
8224 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
8225 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
8226 IPv6 configuration (static or DHCPv6) was found for a given
8227 interface. It will not touch the sysctl otherwise.
8229 * The order of entries is $PATH used by the user manager instance was
8230 changed to put bin/ entries before the corresponding sbin/ entries.
8231 It is recommended to not rely on this order, and only ever have one
8232 binary with a given name in the system paths under /usr.
8234 * A new tool systemd-network-generator has been added that may generate
8235 .network, .netdev and .link files from IP configuration specified on
8236 the kernel command line in the format used by Dracut.
8238 * The CriticalConnection= setting in .network files is now deprecated,
8239 and replaced by a new KeepConfiguration= setting which allows more
8240 detailed configuration of the IP configuration to keep in place.
8242 * systemd-analyze gained a few new verbs:
8244 - "systemd-analyze timestamp" parses and converts timestamps. This is
8245 similar to the existing "systemd-analyze calendar" command which
8246 does the same for recurring calendar events.
8248 - "systemd-analyze timespan" parses and converts timespans (i.e.
8249 durations as opposed to points in time).
8251 - "systemd-analyze condition" will parse and test ConditionXYZ=
8254 - "systemd-analyze exit-status" will parse and convert exit status
8255 codes to their names and back.
8257 - "systemd-analyze unit-files" will print a list of all unit
8258 file paths and unit aliases.
8260 * SuccessExitStatus=, RestartPreventExitStatus=, and
8261 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
8262 is equivalent to "65"). Those exit status name mappings may be
8263 displayed with the systemd-analyze exit-status verb describe above.
8265 * systemd-logind now exposes a per-session SetBrightness() bus call,
8266 which may be used to securely change the brightness of a kernel
8267 brightness device, if it belongs to the session's seat. By using this
8268 call unprivileged clients can make changes to "backlight" and "leds"
8269 devices securely with strict requirements on session membership.
8270 Desktop environments may use this to generically make brightness
8271 changes to such devices without shipping private SUID binaries or
8272 udev rules for that purpose.
8274 * "udevadm info" gained a --wait-for-initialization switch to wait for
8275 a device to be initialized.
8277 * systemd-hibernate-resume-generator will now look for resumeflags= on
8278 the kernel command line, which is similar to rootflags= and may be
8279 used to configure device timeout for the hibernation device.
8281 * sd-event learnt a new API call sd_event_source_disable_unref() for
8282 disabling and unref'ing an event source in a single function. A
8283 related call sd_event_source_disable_unrefp() has been added for use
8284 with gcc's cleanup extension.
8286 * The sd-id128.h public API gained a new definition
8287 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
8290 * "busctl introspect" gained a new switch --xml-interface for dumping
8291 XML introspection data unmodified.
8293 * PID 1 may now show the unit name instead of the unit description
8294 string in its status output during boot. This may be configured in
8295 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
8296 kernel command line option systemd.status_unit_format=.
8298 * PID 1 now understands a new option KExecWatchdogSec= in
8299 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
8300 Previously watchdog functionality was only available for regular
8301 reboots. The new setting defaults to off, because we don't know in
8302 the general case if the watchdog will be reset after kexec (some
8303 drivers do reset it, but not all), and the new userspace might not be
8304 configured to handle the watchdog.
8306 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
8307 RebootWatchdogSec= to more clearly communicate what it is about. The
8308 old name is still accepted for compatibility.
8310 * The systemd.debug_shell kernel command line option now optionally
8311 takes a tty name to spawn the debug shell on, which allows a
8312 different tty to be selected than the built-in default.
8314 * Service units gained a new ExecCondition= setting which will run
8315 before ExecStartPre= and either continue execution of the unit (for
8316 clean exit codes), stop execution without marking the unit failed
8317 (for exit codes 1 through 254), or stop execution and fail the unit
8318 (for exit code 255 or abnormal termination).
8320 * A new service systemd-pstore.service has been added that pulls data
8321 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
8324 * timedatectl gained new verbs for configuring per-interface NTP
8325 service configuration for systemd-timesyncd.
8327 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
8328 2019. (You can set non-UTF-8 locales though, if you know their name.)
8330 * If variable assignments in sysctl.d/ files are prefixed with "-" any
8331 failures to apply them are now ignored.
8333 * systemd-random-seed.service now optionally credits entropy when
8334 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
8335 true for the service to enable this behaviour, but please consult the
8336 documentation first, since this comes with a couple of caveats.
8338 * systemd-random-seed.service is now a synchronization point for full
8339 initialization of the kernel's entropy pool. Services that require
8340 /dev/urandom to be correctly initialized should be ordered after this
8343 * The systemd-boot boot loader has been updated to optionally maintain
8344 a random seed file in the EFI System Partition (ESP). During the boot
8345 phase, this random seed is read and updated with a new seed
8346 cryptographically derived from it. Another derived seed is passed to
8347 the OS. The latter seed is then credited to the kernel's entropy pool
8348 very early during userspace initialization (from PID 1). This allows
8349 systems to boot up with a fully initialized kernel entropy pool from
8350 earliest boot on, and thus entirely removes all entropy pool
8351 initialization delays from systems using systemd-boot. Special care
8352 is taken to ensure different seeds are derived on system images
8353 replicated to multiple systems. "bootctl status" will show whether
8354 a seed was received from the boot loader.
8356 * bootctl gained two new verbs:
8358 - "bootctl random-seed" will generate the file in ESP and an EFI
8359 variable to allow a random seed to be passed to the OS as described
8362 - "bootctl is-installed" checks whether systemd-boot is currently
8365 * bootctl will warn if it detects that boot entries are misconfigured
8366 (for example if the kernel image was removed without purging the
8369 * A new document has been added describing systemd's use and support
8370 for the kernel's entropy pool subsystem:
8372 https://systemd.io/RANDOM_SEEDS
8374 * When the system is hibernated the swap device to write the
8375 hibernation image to is now automatically picked from all available
8376 swap devices, preferring the swap device with the highest configured
8377 priority over all others, and picking the device with the most free
8378 space if there are multiple devices with the highest priority.
8380 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
8381 option that permits selecting the timeout how long to wait for a
8382 device with an encryption key before asking for the password.
8384 * IOWeight= has learnt to properly set the IO weight when using the
8385 BFQ scheduler officially found in kernels 5.0+.
8387 * A new mailing list has been created for reporting of security issues:
8388 systemd-security@redhat.com. For mode details, see
8389 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
8391 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
8392 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
8393 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
8394 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
8395 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
8396 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
8397 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
8398 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
8399 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
8400 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
8401 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
8402 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
8403 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
8404 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
8405 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
8406 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
8407 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
8408 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
8409 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
8410 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
8411 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
8412 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
8413 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
8414 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
8415 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
8416 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
8417 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
8418 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
8419 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
8420 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
8422 – Camerino, 2019-09-03
8426 * In .link files, MACAddressPolicy=persistent (the default) is changed
8427 to cover more devices. For devices like bridges, tun, tap, bond, and
8428 similar interfaces that do not have other identifying information,
8429 the interface name is used as the basis for persistent seed for MAC
8430 and IPv4LL addresses. The way that devices that were handled
8431 previously is not changed, and this change is about covering more
8432 devices then previously by the "persistent" policy.
8434 MACAddressPolicy=random may be used to force randomized MACs and
8435 IPv4LL addresses for a device if desired.
8437 Hint: the log output from udev (at debug level) was enhanced to
8438 clarify what policy is followed and which attributes are used.
8439 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
8440 may be used to view this.
8442 Hint: if a bridge interface is created without any slaves, and gains
8443 a slave later, then now the bridge does not inherit slave's MAC.
8444 To inherit slave's MAC, for example, create the following file:
8446 # /etc/systemd/network/98-bridge-inherit-mac.link
8451 MACAddressPolicy=none
8454 * The .device units generated by systemd-fstab-generator and other
8455 generators do not automatically pull in the corresponding .mount unit
8456 as a Wants= dependency. This means that simply plugging in the device
8457 will not cause the mount unit to be started automatically. But please
8458 note that the mount unit may be started for other reasons, in
8459 particular if it is part of local-fs.target, and any unit which
8460 (transitively) depends on local-fs.target is started.
8462 * networkctl list/status/lldp now accept globbing wildcards for network
8463 interface names to match against all existing interfaces.
8465 * The $PIDFILE environment variable is set to point the absolute path
8466 configured with PIDFile= for processes of that service.
8468 * The fallback DNS server list was augmented with Cloudflare public DNS
8469 servers. Use `-Ddns-servers=` to set a different fallback.
8471 * A new special target usb-gadget.target will be started automatically
8472 when a USB Device Controller is detected (which means that the system
8473 is a USB peripheral).
8475 * A new unit setting CPUQuotaPeriodSec= assigns the time period
8476 relatively to which the CPU time quota specified by CPUQuota= is
8479 * A new unit setting ProtectHostname= may be used to prevent services
8480 from modifying hostname information (even if they otherwise would
8481 have privileges to do so).
8483 * A new unit setting NetworkNamespacePath= may be used to specify a
8484 namespace for service or socket units through a path referring to a
8485 Linux network namespace pseudo-file.
8487 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
8488 have an effect on .socket units: when used the listening socket is
8489 created within the configured network namespace instead of the host
8492 * ExecStart= command lines in unit files may now be prefixed with ':'
8493 in which case environment variable substitution is
8494 disabled. (Supported for the other ExecXYZ= settings, too.)
8496 * .timer units gained two new boolean settings OnClockChange= and
8497 OnTimezoneChange= which may be used to also trigger a unit when the
8498 system clock is changed or the local timezone is
8499 modified. systemd-run has been updated to make these options easily
8500 accessible from the command line for transient timers.
8502 * Two new conditions for units have been added: ConditionMemory= may be
8503 used to conditionalize a unit based on installed system
8504 RAM. ConditionCPUs= may be used to conditionalize a unit based on
8505 installed CPU cores.
8507 * The @default system call filter group understood by SystemCallFilter=
8508 has been updated to include the new rseq() system call introduced in
8511 * A new time-set.target has been added that indicates that the system
8512 time has been set from a local source (possibly imprecise). The
8513 existing time-sync.target is stronger and indicates that the time has
8514 been synchronized with a precise external source. Services where
8515 approximate time is sufficient should use the new target.
8517 * "systemctl start" (and related commands) learnt a new
8518 --show-transaction option. If specified brief information about all
8519 jobs queued because of the requested operation is shown.
8521 * systemd-networkd recognizes a new operation state 'enslaved', used
8522 (instead of 'degraded' or 'carrier') for interfaces which form a
8523 bridge, bond, or similar, and an new 'degraded-carrier' operational
8524 state used for the bond or bridge master interface when one of the
8525 enslaved devices is not operational.
8527 * .network files learnt the new IgnoreCarrierLoss= option for leaving
8528 networks configured even if the carrier is lost.
8530 * The RequiredForOnline= setting in .network files may now specify a
8531 minimum operational state required for the interface to be considered
8532 "online" by systemd-networkd-wait-online. Related to this
8533 systemd-networkd-wait-online gained a new option --operational-state=
8534 to configure the same, and its --interface= option was updated to
8535 optionally also take an operational state specific for an interface.
8537 * systemd-networkd-wait-online gained a new setting --any for waiting
8538 for only one of the requested interfaces instead of all of them.
8540 * systemd-networkd now implements L2TP tunnels.
8542 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
8543 may be used to cause autonomous and onlink prefixes received in IPv6
8544 Router Advertisements to be ignored.
8546 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
8547 file settings may be used to tweak bridge behaviour.
8549 * The new TripleSampling= option in .network files may be used to
8550 configure CAN triple sampling.
8552 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
8553 used to point to private or preshared key for a WireGuard interface.
8555 * /etc/crypttab now supports the same-cpu-crypt and
8556 submit-from-crypt-cpus options to tweak encryption work scheduling
8559 * systemd-tmpfiles will now take a BSD file lock before operating on a
8560 contents of directory. This may be used to temporarily exclude
8561 directories from aging by taking the same lock (useful for example
8562 when extracting a tarball into /tmp or /var/tmp as a privileged user,
8563 which might create files with really old timestamps, which
8564 nevertheless should not be deleted). For further details, see:
8566 https://systemd.io/TEMPORARY_DIRECTORIES
8568 * systemd-tmpfiles' h line type gained support for the
8569 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
8570 controlling project quota inheritance.
8572 * sd-boot and bootctl now implement support for an Extended Boot Loader
8573 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
8574 addition to the ESP partition mounted to /efi or /boot/efi.
8575 Configuration file fragments, kernels, initrds and other EFI images
8576 to boot will be loaded from both the ESP and XBOOTLDR partitions.
8577 The XBOOTLDR partition was previously described by the Boot Loader
8578 Specification, but implementation was missing in sd-boot. Support for
8579 this concept allows using the sd-boot boot loader in more
8580 conservative scenarios where the boot loader itself is placed in the
8581 ESP but the kernels to boot (and their metadata) in a separate
8584 * A system may now be booted with systemd.volatile=overlay on the
8585 kernel command line, which causes the root file system to be set up
8586 an overlayfs mount combining the root-only root directory with a
8587 writable tmpfs. In this setup, the underlying root device is not
8588 modified, and any changes are lost at reboot.
8590 * Similar, systemd-nspawn can now boot containers with a volatile
8591 overlayfs root with the new --volatile=overlay switch.
8593 * systemd-nspawn can now consume OCI runtime bundles using a new
8594 --oci-bundle= option. This implementation is fully usable, with most
8595 features in the specification implemented, but since this a lot of
8596 new code and functionality, this feature should most likely not
8597 be used in production yet.
8599 * systemd-nspawn now supports various options described by the OCI
8600 runtime specification on the command-line and in .nspawn files:
8601 --inaccessible=/Inaccessible= may be used to mask parts of the file
8602 system tree, --console=/--pipe may be used to configure how standard
8603 input, output, and error are set up.
8605 * busctl learned the `emit` verb to generate D-Bus signals.
8607 * systemd-analyze cat-config may be used to gather and display
8608 configuration spread over multiple files, for example system and user
8609 presets, tmpfiles.d, sysusers.d, udev rules, etc.
8611 * systemd-analyze calendar now takes an optional new parameter
8612 --iterations= which may be used to show a maximum number of iterations
8613 the specified expression will elapse next.
8615 * The sd-bus C API gained support for naming method parameters in the
8618 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
8619 the reboot() system call expects.
8621 * journalctl learnt a new --cursor-file= option that points to a file
8622 from which a cursor should be loaded in the beginning and to which
8623 the updated cursor should be stored at the end.
8625 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
8626 detected by systemd-detect-virt (and may also be used in
8627 ConditionVirtualization=).
8629 * The behaviour of systemd-logind may now be modified with environment
8630 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
8631 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
8632 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
8633 skip the relevant operation completely (when set to false), or to
8634 create a flag file in /run/systemd (when set to true), instead of
8635 actually commencing the real operation when requested. The presence
8636 of /run/systemd/reboot-to-firmware-setup,
8637 /run/systemd/reboot-to-boot-loader-menu, and
8638 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
8639 boot loader implementations to replace some steps logind performs
8640 during reboot with their own operations.
8642 * systemctl can be used to request a reboot into the boot loader menu
8643 or a specific boot loader entry with the new --boot-load-menu= and
8644 --boot-loader-entry= options to a reboot command. (This requires a
8645 boot loader that supports this, for example sd-boot.)
8647 * kernel-install will no longer unconditionally create the output
8648 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
8649 snippets, but will do only if the machine-specific parent directory
8650 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
8651 to create this parent directory during sd-boot installation.
8653 This makes it easier to use kernel-install with plugins which support
8654 a different layout of the bootloader partitions (for example grub2).
8656 * During package installation (with `ninja install`), we would create
8657 symlinks for getty@tty1.service, systemd-networkd.service,
8658 systemd-networkd.socket, systemd-resolved.service,
8659 remote-cryptsetup.target, remote-fs.target,
8660 systemd-networkd-wait-online.service, and systemd-timesyncd.service
8661 in /etc, as if `systemctl enable` was called for those units, to make
8662 the system usable immediately after installation. Now this is not
8663 done anymore, and instead calling `systemctl preset-all` is
8664 recommended after the first installation of systemd.
8666 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
8667 is built on seccomp. When turned on creation of SUID/SGID files is
8670 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
8671 implied if DynamicUser= is turned on for a service. This hardens
8672 these services, so that they neither can benefit from nor create
8673 SUID/SGID executables. This is a minor compatibility breakage, given
8674 that when DynamicUser= was first introduced SUID/SGID behaviour was
8675 unaffected. However, the security benefit of these two options is
8676 substantial, and the setting is still relatively new, hence we opted
8677 to make it mandatory for services with dynamic users.
8679 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
8680 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
8681 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
8682 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
8683 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
8684 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
8685 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
8686 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
8687 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
8688 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
8689 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
8690 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
8691 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
8692 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
8693 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
8694 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
8695 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
8696 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8698 — Warsaw, 2019-04-11
8702 * The default locale can now be configured at compile time. Otherwise,
8703 a suitable default will be selected automatically (one of C.UTF-8,
8704 en_US.UTF-8, and C).
8706 * The version string shown by systemd and other tools now includes the
8707 git commit hash when built from git. An override may be specified
8708 during compilation, which is intended to be used by distributions to
8709 include the package release information.
8711 * systemd-cat can now filter standard input and standard error streams
8712 for different syslog priorities using the new --stderr-priority=
8715 * systemd-journald and systemd-journal-remote reject entries which
8716 contain too many fields (CVE-2018-16865) and set limits on the
8717 process' command line length (CVE-2018-16864).
8719 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
8722 * A new network device NamePolicy "keep" is implemented for link files,
8723 and used by default in 99-default.link (the fallback configuration
8724 provided by systemd). With this policy, if the network device name
8725 was already set by userspace, the device will not be renamed again.
8726 This matches the naming scheme that was implemented before
8727 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
8728 is also enabled by default, even if not specified. Effectively, this
8729 means that if naming-scheme >= 240 is specified, network devices will
8730 be renamed according to the configuration, even if they have been
8731 renamed already, if "keep" is not specified as the naming policy in
8732 the .link file. The 99-default.link file provided by systemd includes
8733 "keep" for backwards compatibility, but it is recommended for user
8734 installed .link files to *not* include it.
8736 The "kernel" policy, which keeps kernel names declared to be
8737 "persistent", now works again as documented.
8739 * kernel-install script now optionally takes the paths to one or more
8740 initrd files, and passes them to all plugins.
8742 * The mincore() system call has been dropped from the @system-service
8743 system call filter group, as it is pretty exotic and may potentially
8744 used for side-channel attacks.
8746 * -fPIE is dropped from compiler and linker options. Please specify
8747 -Db_pie=true option to meson to build position-independent
8748 executables. Note that the meson option is supported since meson-0.49.
8750 * The fs.protected_regular and fs.protected_fifos sysctls, which were
8751 added in Linux 4.19 to make some data spoofing attacks harder, are
8752 now enabled by default. While this will hopefully improve the
8753 security of most installations, it is technically a backwards
8754 incompatible change; to disable these sysctls again, place the
8755 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
8757 fs.protected_regular = 0
8758 fs.protected_fifos = 0
8760 Note that the similar hardlink and symlink protection has been
8761 enabled since v199, and may be disabled likewise.
8763 * The files read from the EnvironmentFile= setting in unit files now
8764 parse backslashes inside quotes literally, matching the behaviour of
8767 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
8768 now automatically become NOPs when run in a chroot() environment.
8770 * The tmpfiles.d/ "C" line type will now copy directory trees not only
8771 when the destination is so far missing, but also if it already exists
8772 as a directory and is empty. This is useful to cater for systems
8773 where directory trees are put together from multiple separate mount
8774 points but otherwise empty.
8776 * A new function sd_bus_close_unref() (and the associated
8777 sd_bus_close_unrefp()) has been added to libsystemd, that combines
8778 sd_bus_close() and sd_bus_unref() in one.
8780 * udevadm control learnt a new option for --ping for testing whether a
8781 systemd-udevd instance is running and reacting.
8783 * udevadm trigger learnt a new option for --wait-daemon for waiting
8784 systemd-udevd daemon to be initialized.
8786 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
8787 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
8788 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
8789 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
8790 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
8791 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
8792 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
8793 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
8794 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
8795 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
8796 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
8797 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
8798 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
8799 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
8800 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
8801 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
8802 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
8804 — Berlin, 2019-02-14
8808 * NoNewPrivileges=yes has been set for all long-running services
8809 implemented by systemd. Previously, this was problematic due to
8810 SELinux (as this would also prohibit the transition from PID1's label
8811 to the service's label). This restriction has since been lifted, but
8812 an SELinux policy update is required.
8813 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
8815 * DynamicUser=yes is dropped from systemd-networkd.service,
8816 systemd-resolved.service and systemd-timesyncd.service, which was
8817 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
8818 and since v236 for systemd-timesyncd.service. The users and groups
8819 systemd-network, systemd-resolve and systemd-timesync are created
8820 by systemd-sysusers again. Distributors or system administrators
8821 may need to create these users and groups if they not exist (or need
8822 to re-enable DynamicUser= for those units) while upgrading systemd.
8823 Also, the clock file for systemd-timesyncd may need to move from
8824 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
8826 * When unit files are loaded from disk, previously systemd would
8827 sometimes (depending on the unit loading order) load units from the
8828 target path of symlinks in .wants/ or .requires/ directories of other
8829 units. This meant that unit could be loaded from different paths
8830 depending on whether the unit was requested explicitly or as a
8831 dependency of another unit, not honouring the priority of directories
8832 in search path. It also meant that it was possible to successfully
8833 load and start units which are not found in the unit search path, as
8834 long as they were requested as a dependency and linked to from
8835 .wants/ or .requires/. The target paths of those symlinks are not
8836 used for loading units anymore and the unit file must be found in
8839 * A new service type has been added: Type=exec. It's very similar to
8840 Type=simple but ensures the service manager will wait for both fork()
8841 and execve() of the main service binary to complete before proceeding
8842 with follow-up units. This is primarily useful so that the manager
8843 propagates any errors in the preparation phase of service execution
8844 back to the job that requested the unit to be started. For example,
8845 consider a service that has ExecStart= set to a file system binary
8846 that doesn't exist. With Type=simple starting the unit would be
8847 considered instantly successful, as only fork() has to complete
8848 successfully and the manager does not wait for execve(), and hence
8849 its failure is seen "too late". With the new Type=exec service type
8850 starting the unit will fail, as the manager will wait for the
8851 execve() and notice its failure, which is then propagated back to the
8854 NOTE: with the next release 241 of systemd we intend to change the
8855 systemd-run tool to default to Type=exec for transient services
8856 started by it. This should be mostly safe, but in specific corner
8857 cases might result in problems, as the systemd-run tool will then
8858 block on NSS calls (such as user name look-ups due to User=) done
8859 between the fork() and execve(), which under specific circumstances
8860 might cause problems. It is recommended to specify "-p Type=simple"
8861 explicitly in the few cases where this applies. For regular,
8862 non-transient services (i.e. those defined with unit files on disk)
8863 we will continue to default to Type=simple.
8865 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
8866 userspace processes is set to 1024 (soft) and 4096
8867 (hard). Previously, systemd passed this on unmodified to all
8868 processes it forked off. With this systemd release the hard limit
8869 systemd passes on is increased to 512K, overriding the kernel's
8870 defaults and substantially increasing the number of simultaneous file
8871 descriptors unprivileged userspace processes can allocate. Note that
8872 the soft limit remains at 1024 for compatibility reasons: the
8873 traditional UNIX select() call cannot deal with file descriptors >=
8874 1024 and increasing the soft limit globally might thus result in
8875 programs unexpectedly allocating a high file descriptor and thus
8876 failing abnormally when attempting to use it with select() (of
8877 course, programs shouldn't use select() anymore, and prefer
8878 poll()/epoll, but the call unfortunately remains undeservedly popular
8879 at this time). This change reflects the fact that file descriptor
8880 handling in the Linux kernel has been optimized in more recent
8881 kernels and allocating large numbers of them should be much cheaper
8882 both in memory and in performance than it used to be. Programs that
8883 want to take benefit of the increased limit have to "opt-in" into
8884 high file descriptors explicitly by raising their soft limit. Of
8885 course, when they do that they must acknowledge that they cannot use
8886 select() anymore (and neither can any shared library they use — or
8887 any shared library used by any shared library they use and so on).
8888 Which default hard limit is most appropriate is of course hard to
8889 decide. However, given reports that ~300K file descriptors are used
8890 in real-life applications we believe 512K is sufficiently high as new
8891 default for now. Note that there are also reports that using very
8892 high hard limits (e.g. 1G) is problematic: some software allocates
8893 large arrays with one element for each potential file descriptor
8894 (Java, …) — a high hard limit thus triggers excessively large memory
8895 allocations in these applications. Hopefully, the new default of 512K
8896 is a good middle ground: higher than what real-life applications
8897 currently need, and low enough for avoid triggering excessively large
8898 allocations in problematic software. (And yes, somebody should fix
8901 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
8902 to the highest possible values, as separate accounting of file
8903 descriptors is no longer necessary, as memcg tracks them correctly as
8904 part of the memory accounting anyway. Thus, from the four limits on
8905 file descriptors currently enforced (fs.file-max, fs.nr_open,
8906 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
8907 and keep only the latter two. A set of build-time options
8908 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
8909 has been added to revert this change in behaviour, which might be
8910 an option for systems that turn off memcg in the kernel.
8912 * When no /etc/locale.conf file exists (and hence no locale settings
8913 are in place), systemd will now use the "C.UTF-8" locale by default,
8914 and set LANG= to it. This locale is supported by various
8915 distributions including Fedora, with clear indications that upstream
8916 glibc is going to make it available too. This locale enables UTF-8
8917 mode by default, which appears appropriate for 2018.
8919 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
8920 default. This effectively switches the RFC3704 Reverse Path filtering
8921 from Strict mode to Loose mode. This is more appropriate for hosts
8922 that have multiple links with routes to the same networks (e.g.
8923 a client with a Wi-Fi and Ethernet both connected to the internet).
8925 Consult the kernel documentation for details on this sysctl:
8926 https://docs.kernel.org/networking/ip-sysctl.html
8928 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
8931 * CPUAccounting=yes no longer enables the CPU controller when using
8932 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
8933 statistics are now provided independently from the CPU controller.
8935 * Support for disabling a particular cgroup controller within a sub-tree
8936 has been added through the DisableControllers= directive.
8938 * cgroup_no_v1=all on the kernel command line now also implies
8939 using the unified cgroup hierarchy, unless one explicitly passes
8940 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
8942 * The new "MemoryMin=" unit file property may now be used to set the
8943 memory usage protection limit of processes invoked by the unit. This
8944 controls the cgroup v2 memory.min attribute. Similarly, the new
8945 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
8946 cgroup v2 io.latency cgroup property for configuring per-service I/O
8949 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
8950 to the cgroup v1 "devices" cgroup controller.
8952 * systemd-escape now is able to combine --unescape with --template. It
8953 also learnt a new option --instance for extracting and unescaping the
8954 instance part of a unit name.
8956 * sd-bus now provides the sd_bus_message_readv() which is similar to
8957 sd_bus_message_read() but takes a va_list object. The pair
8958 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
8959 has been added for configuring the default method call timeout to
8960 use. sd_bus_error_move() may be used to efficiently move the contents
8961 from one sd_bus_error structure to another, invalidating the
8962 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
8963 be used to control whether a bus connection object is automatically
8964 flushed when an sd-event loop is exited.
8966 * When processing classic BSD syslog log messages, journald will now
8967 save the original time-stamp string supplied in the new
8968 SYSLOG_TIMESTAMP= journal field. This permits consumers to
8969 reconstruct the original BSD syslog message more correctly.
8971 * StandardOutput=/StandardError= in service files gained support for
8972 new "append:…" parameters, for connecting STDOUT/STDERR of a service
8973 to a file, and appending to it.
8975 * The signal to use as last step of killing of unit processes is now
8976 configurable. Previously it was hard-coded to SIGKILL, which may now
8977 be overridden with the new KillSignal= setting. Note that this is the
8978 signal used when regular termination (i.e. SIGTERM) does not suffice.
8979 Similarly, the signal used when aborting a program in case of a
8980 watchdog timeout may now be configured too (WatchdogSignal=).
8982 * The XDG_SESSION_DESKTOP environment variable may now be configured in
8983 the pam_systemd argument line, using the new desktop= switch. This is
8984 useful to initialize it properly from a display manager without
8985 having to touch C code.
8987 * Most configuration options that previously accepted percentage values
8988 now also accept permille values with the '‰' suffix (instead of '%').
8990 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
8993 * systemd-resolved's configuration file resolved.conf gained a new
8994 option ReadEtcHosts= which may be used to turn off processing and
8995 honoring /etc/hosts entries.
8997 * The "--wait" switch may now be passed to "systemctl
8998 is-system-running", in which case the tool will synchronously wait
8999 until the system finished start-up.
9001 * hostnamed gained a new bus call to determine the DMI product UUID.
9003 * On x86-64 systemd will now prefer using the RDRAND processor
9004 instruction over /dev/urandom whenever it requires randomness that
9005 neither has to be crypto-grade nor should be reproducible. This
9006 should substantially reduce the amount of entropy systemd requests
9007 from the kernel during initialization on such systems, though not
9008 reduce it to zero. (Why not zero? systemd still needs to allocate
9009 UUIDs and such uniquely, which require high-quality randomness.)
9011 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
9012 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
9013 for forcing the "Other Information" bit in IPv6 RA messages. The
9014 bonding logic gained four new options AdActorSystemPriority=,
9015 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
9016 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
9017 shuffling of flows. The tunnel logic gained a new
9018 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
9019 Deployment. The policy rule logic gained four new options IPProtocol=,
9020 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
9021 support for the MulticastToUnicast= option. networkd also gained
9022 support for configuring static IPv4 ARP or IPv6 neighbor entries.
9024 * .preset files (as read by 'systemctl preset') may now be used to
9025 instantiate services.
9027 * /etc/crypttab now understands the sector-size= option to configure
9028 the sector size for an encrypted partition.
9030 * Key material for encrypted disks may now be placed on a formatted
9031 medium, and referenced from /etc/crypttab by the UUID of the file
9032 system, followed by "=" suffixed by the path to the key file.
9034 * The "collect" udev component has been removed without replacement, as
9035 it is neither used nor maintained.
9037 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
9038 LogsDirectory=, ConfigurationDirectory= settings are used in a
9039 service the executed processes will now receive a set of environment
9040 variables containing the full paths of these directories.
9041 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
9042 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
9043 are used. Note that these options may be used multiple times per
9044 service in which case the resulting paths will be concatenated and
9045 separated by colons.
9047 * Predictable interface naming has been extended to cover InfiniBand
9048 NICs. They will be exposed with an "ib" prefix.
9050 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
9051 which case the respective line failing is ignored.
9053 * .link files may now be used to configure the equivalent to the
9054 "ethtool advertise" commands.
9056 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
9057 alternative to libudev.h. Previously, the latter was just an internal
9058 wrapper around the former, but now these two APIs are exposed
9061 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
9062 which calculates an app-specific boot ID similar to how
9063 sd_id128_get_machine_app_specific() generates an app-specific machine
9066 * A new tool systemd-id128 has been added that can be used to determine
9067 and generate various 128-bit IDs.
9069 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
9072 * systemd-hibernate-resume-generator will now honor the "noresume"
9073 kernel command line option, in which case it will bypass resuming
9074 from any hibernated image.
9076 * The systemd-sleep.conf configuration file gained new options
9077 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
9078 AllowHybridSleep= for prohibiting specific sleep modes even if the
9079 kernel exports them.
9081 * portablectl is now officially supported and has thus moved to
9084 * bootctl learnt the two new commands "set-default" and "set-oneshot"
9085 for setting the default boot loader item to boot to (either
9086 persistently or only for the next boot). This is currently only
9087 compatible with sd-boot, but may be implemented on other boot loaders
9088 too, that follow the boot loader interface. The updated interface is
9089 now documented here:
9091 https://systemd.io/BOOT_LOADER_INTERFACE
9093 * A new kernel command line option systemd.early_core_pattern= is now
9094 understood which may be used to influence the core_pattern PID 1
9095 installs during early boot.
9097 * busctl learnt two new options -j and --json= for outputting method
9098 call replies, properties and monitoring output in JSON.
9100 * journalctl's JSON output now supports simple ANSI coloring as well as
9101 a new "json-seq" mode for generating RFC7464 output.
9103 * Unit files now support the %g/%G specifiers that resolve to the UNIX
9104 group/GID of the service manager runs as, similar to the existing
9105 %u/%U specifiers that resolve to the UNIX user/UID.
9107 * systemd-logind learnt a new global configuration option
9108 UserStopDelaySec= that may be set in logind.conf. It specifies how
9109 long the systemd --user instance shall remain started after a user
9110 logs out. This is useful to speed up repetitive re-connections of the
9111 same user, as it means the user's service manager doesn't have to be
9112 stopped/restarted on each iteration, but can be reused between
9113 subsequent options. This setting defaults to 10s. systemd-logind also
9114 exports two new properties on its Manager D-Bus objects indicating
9115 whether the system's lid is currently closed, and whether the system
9118 * systemd gained support for a generic boot counting logic, which
9119 generically permits automatic reverting to older boot loader entries
9120 if newer updated ones don't work. The boot loader side is implemented
9121 in sd-boot, but is kept open for other boot loaders too. For details
9124 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
9126 * The SuccessAction=/FailureAction= unit file settings now learnt two
9127 new parameters: "exit" and "exit-force", which result in immediate
9128 exiting of the service manager, and are only useful in systemd --user
9129 and container environments.
9131 * Unit files gained support for a pair of options
9132 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
9133 exit status to use as service manager exit status when
9134 SuccessAction=/FailureAction= is set to exit or exit-force.
9136 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
9137 options may now be used to configure the log rate limiting applied by
9138 journald per-service.
9140 * systemd-analyze gained a new verb "timespan" for parsing and
9141 normalizing time span values (i.e. strings like "5min 7s 8us").
9143 * systemd-analyze also gained a new verb "security" for analyzing the
9144 security and sand-boxing settings of services in order to determine an
9145 "exposure level" for them, indicating whether a service would benefit
9146 from more sand-boxing options turned on for them.
9148 * "systemd-analyze syscall-filter" will now also show system calls
9149 supported by the local kernel but not included in any of the defined
9152 * .nspawn files now understand the Ephemeral= setting, matching the
9153 --ephemeral command line switch.
9155 * sd-event gained the new APIs sd_event_source_get_floating() and
9156 sd_event_source_set_floating() for controlling whether a specific
9157 event source is "floating", i.e. destroyed along with the even loop
9160 * Unit objects on D-Bus gained a new "Refs" property that lists all
9161 clients that currently have a reference on the unit (to ensure it is
9164 * The JoinControllers= option in system.conf is no longer supported, as
9165 it didn't work correctly, is hard to support properly, is legacy (as
9166 the concept only exists on cgroup v1) and apparently wasn't used.
9168 * Journal messages that are generated whenever a unit enters the failed
9169 state are now tagged with a unique MESSAGE_ID. Similarly, messages
9170 generated whenever a service process exits are now made recognizable,
9171 too. A tagged message is also emitted whenever a unit enters the
9172 "dead" state on success.
9174 * systemd-run gained a new switch --working-directory= for configuring
9175 the working directory of the service to start. A shortcut -d is
9176 equivalent, setting the working directory of the service to the
9177 current working directory of the invoking program. The new --shell
9178 (or just -S) option has been added for invoking the $SHELL of the
9179 caller as a service, and implies --pty --same-dir --wait --collect
9180 --service-type=exec. Or in other words, "systemd-run -S" is now the
9181 quickest way to quickly get an interactive in a fully clean and
9182 well-defined system service context.
9184 * machinectl gained a new verb "import-fs" for importing an OS tree
9185 from a directory. Moreover, when a directory or tarball is imported
9186 and single top-level directory found with the OS itself below the OS
9187 tree is automatically mangled and moved one level up.
9189 * systemd-importd will no longer set up an implicit btrfs loop-back
9190 file system on /var/lib/machines. If one is already set up, it will
9191 continue to be used.
9193 * A new generator "systemd-run-generator" has been added. It will
9194 synthesize a unit from one or more program command lines included in
9195 the kernel command line. This is very useful in container managers
9198 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
9200 This will run "systemd-nspawn" on an image, invoke the specified
9201 command line and immediately shut down the container again, returning
9202 the command line's exit code.
9204 * The block device locking logic is now documented:
9206 https://systemd.io/BLOCK_DEVICE_LOCKING
9208 * loginctl and machinectl now optionally output the various tables in
9209 JSON using the --output= switch. It is our intention to add similar
9210 support to systemctl and all other commands.
9212 * udevadm's query and trigger verb now optionally take a .device unit
9215 * systemd-udevd's network naming logic now understands a new
9216 net.naming_scheme= kernel command line switch, which may be used to
9217 pick a specific version of the naming scheme. This helps stabilizing
9218 interface names even as systemd/udev are updated and the naming logic
9221 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
9222 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
9223 initialize one to all 0xFF.
9225 * After loading the SELinux policy systemd will now recursively relabel
9226 all files and directories listed in
9227 /run/systemd/relabel-extra.d/*.relabel (which should be simple
9228 newline separated lists of paths) in addition to the ones it already
9229 implicitly relabels in /run, /dev and /sys. After the relabelling is
9230 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
9231 removed. This is useful to permit initrds (i.e. code running before
9232 the SELinux policy is in effect) to generate files in the host
9233 filesystem safely and ensure that the correct label is applied during
9234 the transition to the host OS.
9236 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
9237 mknod() handling in user namespaces. Previously mknod() would always
9238 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
9239 but device nodes generated that way cannot be opened, and attempts to
9240 open them result in EPERM. This breaks the "graceful fallback" logic
9241 in systemd's PrivateDevices= sand-boxing option. This option is
9242 implemented defensively, so that when systemd detects it runs in a
9243 restricted environment (such as a user namespace, or an environment
9244 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
9245 where device nodes cannot be created the effect of PrivateDevices= is
9246 bypassed (following the logic that 2nd-level sand-boxing is not
9247 essential if the system systemd runs in is itself already sand-boxed
9248 as a whole). This logic breaks with 4.18 in container managers where
9249 user namespacing is used: suddenly PrivateDevices= succeeds setting
9250 up a private /dev/ file system containing devices nodes — but when
9251 these are opened they don't work.
9253 At this point it is recommended that container managers utilizing
9254 user namespaces that intend to run systemd in the payload explicitly
9255 block mknod() with seccomp or similar, so that the graceful fallback
9258 We are very sorry for the breakage and the requirement to change
9259 container configurations for newer kernels. It's purely caused by an
9260 incompatible kernel change. The relevant kernel developers have been
9261 notified about this userspace breakage quickly, but they chose to
9264 * PermissionsStartOnly= setting is deprecated (but is still supported
9265 for backwards compatibility). The same functionality is provided by
9266 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
9269 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
9270 pam_systemd anymore.
9272 * The naming scheme for network devices was changed to always rename
9273 devices, even if they were already renamed by userspace. The "kernel"
9274 policy was changed to only apply as a fallback, if no other naming
9277 * The requirements to build systemd is bumped to meson-0.46 and
9280 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
9281 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
9282 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
9283 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
9284 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
9285 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
9286 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
9287 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
9288 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
9289 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
9290 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
9291 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
9292 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
9293 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
9294 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
9295 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
9296 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
9297 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
9298 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
9299 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
9300 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
9301 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
9302 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
9303 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
9304 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
9305 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
9306 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
9307 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
9308 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
9309 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
9310 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
9311 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
9312 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
9313 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
9314 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
9315 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
9316 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
9317 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
9318 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
9319 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
9320 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
9321 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
9322 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
9323 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
9324 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
9326 — Warsaw, 2018-12-21
9330 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
9331 builtin will name network interfaces differently than in previous
9332 versions for virtual network interfaces created with SR-IOV and NPAR
9333 and for devices where the PCI network controller device does not have
9334 a slot number associated.
9336 SR-IOV virtual devices are now named based on the name of the parent
9337 interface, with a suffix of "v<N>", where <N> is the virtual device
9338 number. Previously those virtual devices were named as if completely
9341 The ninth and later NPAR virtual devices will be named following the
9342 scheme used for the first eight NPAR partitions. Previously those
9343 devices were not renamed and the kernel default (eth<n>) was used.
9345 "net_id" will also generate names for PCI devices where the PCI
9346 network controller device does not have an associated slot number
9347 itself, but one of its parents does. Previously those devices were
9348 not renamed and the kernel default (eth<n>) was used.
9350 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
9351 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
9352 the unit. So, it is expected that the default behavior of
9353 systemd-logind is not changed. However, if distribution packagers or
9354 administrators disabled or modified IPAddressDeny= setting by a
9355 drop-in config file, then it may be necessary to update the file to
9356 re-enable AF_INET and AF_INET6 to support network user name services,
9359 * When the RestrictNamespaces= unit property is specified multiple
9360 times, then the specified types are merged now. Previously, only the
9361 last assignment was used. So, if distribution packagers or
9362 administrators modified the setting by a drop-in config file, then it
9363 may be necessary to update the file.
9365 * When OnFailure= is used in combination with Restart= on a service
9366 unit, then the specified units will no longer be triggered on
9367 failures that result in restarting. Previously, the specified units
9368 would be activated each time the unit failed, even when the unit was
9369 going to be restarted automatically. This behaviour contradicted the
9370 documentation. With this release the code is adjusted to match the
9373 * systemd-tmpfiles will now print a notice whenever it encounters
9374 tmpfiles.d/ lines referencing the /var/run/ directory. It will
9375 recommend reworking them to use the /run/ directory instead (for
9376 which /var/run/ is simply a symlinked compatibility alias). This way
9377 systemd-tmpfiles can properly detect line conflicts and merge lines
9378 referencing the same file by two paths, without having to access
9381 * systemctl disable/unmask/preset/preset-all cannot be used with
9382 --runtime. Previously this was allowed, but resulted in unintuitive
9383 behaviour that wasn't useful. systemctl disable/unmask will now undo
9384 both runtime and persistent enablement/masking, i.e. it will remove
9385 any relevant symlinks both in /run and /etc.
9387 * Note that all long-running system services shipped with systemd will
9388 now default to a system call allow list (rather than a deny list, as
9389 before). In particular, systemd-udevd will now enforce one too. For
9390 most cases this should be safe, however downstream distributions
9391 which disabled sandboxing of systemd-udevd (specifically the
9392 MountFlags= setting), might want to disable this security feature
9393 too, as the default allow-listing will prohibit all mount, swap,
9394 reboot and clock changing operations from udev rules.
9396 * sd-boot acquired new loader configuration settings to optionally turn
9397 off Windows and MacOS boot partition discovery as well as
9398 reboot-into-firmware menu items. It is also able to pick a better
9399 screen resolution for HiDPI systems, and now provides loader
9400 configuration settings to change the resolution explicitly.
9402 * systemd-resolved now supports DNS-over-TLS. It's still
9403 turned off by default, use DNSOverTLS=opportunistic to turn it on in
9404 resolved.conf. We intend to make this the default as soon as couple
9405 of additional techniques for optimizing the initial latency caused by
9406 establishing a TLS/TCP connection are implemented.
9408 * systemd-resolved.service and systemd-networkd.service now set
9409 DynamicUser=yes. The users systemd-resolve and systemd-network are
9410 not created by systemd-sysusers anymore.
9412 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
9413 that embed a network facing module into any process using getpwuid()
9414 or related call: the dynamic allocation of the user ID for
9415 systemd-resolved.service means the service manager has to check NSS
9416 if the user name is already taken when forking off the service. Since
9417 the user in the common case won't be defined in /etc/passwd the
9418 lookup is likely to trigger nss-ldap which in turn might use NSS to
9419 ask systemd-resolved for hostname lookups. This will hence result in
9420 a deadlock: a user name lookup in order to start
9421 systemd-resolved.service will result in a hostname lookup for which
9422 systemd-resolved.service needs to be started already. There are
9423 multiple ways to work around this problem: pre-allocate the
9424 "systemd-resolve" user on such systems, so that nss-ldap won't be
9425 triggered; or use a different NSS package that doesn't do networking
9426 in-process but provides a local asynchronous name cache; or configure
9427 the NSS package to avoid lookups for UIDs in the range `pkg-config
9428 systemd --variable=dynamicuidmin` … `pkg-config systemd
9429 --variable=dynamicuidmax`, so that it does not consider itself
9430 authoritative for the same UID range systemd allocates dynamic users
9433 * The systemd-resolve tool has been renamed to resolvectl (it also
9434 remains available under the old name, for compatibility), and its
9435 interface is now verb-based, similar in style to the other <xyz>ctl
9436 tools, such as systemctl or loginctl.
9438 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
9439 compatibility. It may be symlinked under the 'resolvconf' name, in
9440 which case it will take arguments and input compatible with the
9441 Debian and FreeBSD resolvconf tool.
9443 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
9444 where the system initially suspends, and after a timeout resumes and
9447 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
9448 set the client will only send a DUID as client identifier. (EDIT: the
9449 option was broken, and was dropped in v255.)
9451 * The nss-systemd glibc NSS module will now enumerate dynamic users and
9452 groups in effect. Previously, it could resolve UIDs/GIDs to user
9453 names/groups and vice versa, but did not support enumeration.
9455 * journald's Compress= configuration setting now optionally accepts a
9456 byte threshold value. All journal objects larger than this threshold
9457 will be compressed, smaller ones will not. Previously this threshold
9458 was not configurable and set to 512.
9460 * A new system.conf setting NoNewPrivileges= is now available which may
9461 be used to turn off acquisition of new privileges system-wide
9462 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
9463 for all its children). Note that turning this option on means setuid
9464 binaries and file system capabilities lose their special powers.
9465 While turning on this option is a big step towards a more secure
9466 system, doing so is likely to break numerous pre-existing UNIX tools,
9467 in particular su and sudo.
9469 * A new service systemd-time-sync-wait.service has been added. If
9470 enabled it will delay the time-sync.target unit at boot until time
9471 synchronization has been received from the network. This
9472 functionality is useful on systems lacking a local RTC or where it is
9473 acceptable that the boot process shall be delayed by external network
9476 * When hibernating, systemd will now inform the kernel of the image
9477 write offset, on kernels new enough to support this. This means swap
9478 files should work for hibernation now.
9480 * When loading unit files, systemd will now look for drop-in unit files
9481 extensions in additional places. Previously, for a unit file name
9482 "foo-bar-baz.service" it would look for dropin files in
9483 "foo-bar-baz.service.d/*.conf". Now, it will also look in
9484 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
9485 service name truncated after all inner dashes. This scheme allows
9486 writing drop-ins easily that apply to a whole set of unit files at
9487 once. It's particularly useful for mount and slice units (as their
9488 naming is prefix based), but is also useful for service and other
9489 units, for packages that install multiple unit files at once,
9490 following a strict naming regime of beginning the unit file name with
9491 the package's name. Two new specifiers are now supported in unit
9492 files to match this: %j and %J are replaced by the part of the unit
9493 name following the last dash.
9495 * Unit files and other configuration files that support specifier
9496 expansion now understand another three new specifiers: %T and %V will
9497 resolve to /tmp and /var/tmp respectively, or whatever temporary
9498 directory has been set for the calling user. %E will expand to either
9499 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
9501 * The ExecStart= lines of unit files are no longer required to
9502 reference absolute paths. If non-absolute paths are specified the
9503 specified binary name is searched within the service manager's
9504 built-in $PATH, which may be queried with 'systemd-path
9505 search-binaries-default'. It's generally recommended to continue to
9506 use absolute paths for all binaries specified in unit files.
9508 * Units gained a new load state "bad-setting", which is used when a
9509 unit file was loaded, but contained fatal errors which prevent it
9510 from being started (for example, a service unit has been defined
9511 lacking both ExecStart= and ExecStop= lines).
9513 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
9514 support alternative debuggers, for example lldb. The old name
9515 continues to be available however, for compatibility reasons. Use the
9516 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
9517 to pick an alternative debugger instead of the default gdb.
9519 * systemctl and the other tools will now output escape sequences that
9520 generate proper clickable hyperlinks in various terminal emulators
9521 where useful (for example, in the "systemctl status" output you can
9522 now click on the unit file name to quickly open it in the
9523 editor/viewer of your choice). Note that not all terminal emulators
9524 support this functionality yet, but many do. Unfortunately, the
9525 "less" pager doesn't support this yet, hence this functionality is
9526 currently automatically turned off when a pager is started (which
9527 happens quite often due to auto-paging). We hope to remove this
9528 limitation as soon as "less" learns these escape sequences. This new
9529 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
9530 environment variable. For details on these escape sequences see:
9531 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
9533 * networkd's .network files now support a new IPv6MTUBytes= option for
9534 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
9535 option in the [Route] section to configure the MTU to use for
9536 specific routes. It also gained support for configuration of the DHCP
9537 "UserClass" option through the new UserClass= setting. It gained
9538 three new options in the new [CAN] section for configuring CAN
9539 networks. The MULTICAST and ALLMULTI interface flags may now be
9540 controlled explicitly with the new Multicast= and AllMulticast=
9543 * networkd will now automatically make use of the kernel's route
9544 expiration feature, if it is available.
9546 * udevd's .link files now support setting the number of receive and
9547 transmit channels, using the RxChannels=, TxChannels=,
9548 OtherChannels=, CombinedChannels= settings.
9550 * Support for UDPSegmentationOffload= has been removed, given its
9551 limited support in hardware, and waning software support.
9553 * networkd's .netdev files now support creating "netdevsim" interfaces.
9555 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
9556 to query the unit belonging to a specific kernel control group.
9558 * systemd-analyze gained a new verb "cat-config", which may be used to
9559 dump the contents of any configuration file, with all its matching
9560 drop-in files added in, and honouring the usual search and masking
9561 logic applied to systemd configuration files. For example use
9562 "systemd-analyze cat-config systemd/system.conf" to get the complete
9563 system configuration file of systemd how it would be loaded by PID 1
9564 itself. Similar to this, various tools such as systemd-tmpfiles or
9565 systemd-sysusers, gained a new option "--cat-config", which does the
9566 corresponding operation for their own configuration settings. For
9567 example, "systemd-tmpfiles --cat-config" will now output the full
9568 list of tmpfiles.d/ lines in place.
9570 * timedatectl gained three new verbs: "show" shows bus properties of
9571 systemd-timedated, "timesync-status" shows the current NTP
9572 synchronization state of systemd-timesyncd, and "show-timesync"
9573 shows bus properties of systemd-timesyncd.
9575 * systemd-timesyncd gained a bus interface on which it exposes details
9578 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
9579 understood by systemd-timedated. It takes a colon-separated list of
9580 unit names of NTP client services. The list is used by
9581 "timedatectl set-ntp".
9583 * systemd-nspawn gained a new --rlimit= switch for setting initial
9584 resource limits for the container payload. There's a new switch
9585 --hostname= to explicitly override the container's hostname. A new
9586 --no-new-privileges= switch may be used to control the
9587 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
9588 --oom-score-adjust= switch controls the OOM scoring adjustment value
9589 for the payload. The new --cpu-affinity= switch controls the CPU
9590 affinity of the container payload. The new --resolv-conf= switch
9591 allows more detailed control of /etc/resolv.conf handling of the
9592 container. Similarly, the new --timezone= switch allows more detailed
9593 control of /etc/localtime handling of the container.
9595 * systemd-detect-virt gained a new --list switch, which will print a
9596 list of all currently known VM and container environments.
9598 * Support for "Portable Services" has been added, see
9599 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
9600 experimental, but this is expected to change soon. Reflecting this
9601 experimental state, the "portablectl" binary is not installed into
9602 /usr/bin yet. The binary has to be called with the full path
9603 /usr/lib/systemd/portablectl instead.
9605 * journalctl's and systemctl's -o switch now knows a new log output
9606 mode "with-unit". The output it generates is very similar to the
9607 regular "short" mode, but displays the unit name instead of the
9608 syslog tag for each log line. Also, the date is shown with timezone
9609 information. This mode is probably more useful than the classic
9610 "short" output mode for most purposes, except where pixel-perfect
9611 compatibility with classic /var/log/messages formatting is required.
9613 * A new --dump-bus-properties switch has been added to the systemd
9614 binary, which may be used to dump all supported D-Bus properties.
9615 (Options which are still supported, but are deprecated, are *not*
9618 * sd-bus gained a set of new calls:
9619 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
9620 enable/disable the "floating" state of a bus slot object,
9621 i.e. whether the slot object pins the bus it is allocated for into
9622 memory or if the bus slot object gets disconnected when the bus goes
9623 away. sd_bus_open_with_description(),
9624 sd_bus_open_user_with_description(),
9625 sd_bus_open_system_with_description() may be used to allocate bus
9626 objects and set their description string already during allocation.
9628 * sd-event gained support for watching inotify events from the event
9629 loop, in an efficient way, sharing inotify handles between multiple
9630 users. For this a new function sd_event_add_inotify() has been added.
9632 * sd-event and sd-bus gained support for calling special user-supplied
9633 destructor functions for userdata pointers associated with
9634 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
9635 functions sd_bus_slot_set_destroy_callback,
9636 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
9637 sd_bus_track_get_destroy_callback,
9638 sd_event_source_set_destroy_callback,
9639 sd_event_source_get_destroy_callback have been added.
9641 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
9643 * PID 1 will now automatically reschedule .timer units whenever the
9644 local timezone changes. (They previously got rescheduled
9645 automatically when the system clock changed.)
9647 * New documentation has been added to document cgroups delegation,
9648 portable services and the various code quality tools we have set up:
9650 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
9651 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
9652 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
9654 * The Boot Loader Specification has been added to the source tree.
9656 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
9658 While moving it into our source tree we have updated it and further
9659 changes are now accepted through the usual github PR workflow.
9661 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
9662 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
9663 earlier PAM modules. The data in these fields is used to initialize
9664 the session scope's resource properties. Thus external PAM modules
9665 may now configure per-session limits, for example sourced from
9666 external user databases.
9668 * socket units with Accept=yes will now maintain a "refused" counter in
9669 addition to the existing "accepted" counter, counting connections
9670 refused due to the enforced limits.
9672 * The "systemd-path search-binaries-default" command may now be use to
9673 query the default, built-in $PATH PID 1 will pass to the services it
9676 * A new unit file setting PrivateMounts= has been added. It's a boolean
9677 option. If enabled the unit's processes are invoked in their own file
9678 system namespace. Note that this behaviour is also implied if any
9679 other file system namespacing options (such as PrivateTmp=,
9680 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
9681 primarily useful for services that do not use any of the other file
9682 system namespacing options. One such service is systemd-udevd.service
9683 where this is now used by default.
9685 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
9686 when the system is booted in UEFI "secure mode".
9688 * A new unit "system-update-pre.target" is added, which defines an
9689 optional synchronization point for offline system updates, as
9690 implemented by the pre-existing "system-update.target" unit. It
9691 allows ordering services before the service that executes the actual
9692 update process in a generic way.
9694 * Systemd now emits warnings whenever .include syntax is used.
9696 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
9697 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
9698 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
9699 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
9700 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
9701 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
9702 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
9703 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
9704 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
9705 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
9706 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
9707 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
9708 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
9709 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
9710 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
9711 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
9712 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
9713 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
9714 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
9715 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
9716 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
9717 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
9718 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
9719 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
9720 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
9721 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
9722 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
9723 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
9724 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9726 — Berlin, 2018-06-22
9730 * The MemoryAccounting= unit property now defaults to on. After
9731 discussions with the upstream control group maintainers we learnt
9732 that the negative impact of cgroup memory accounting on current
9733 kernels is finally relatively minimal, so that it should be safe to
9734 enable this by default without affecting system performance. Besides
9735 memory accounting only task accounting is turned on by default, all
9736 other forms of resource accounting (CPU, IO, IP) remain off for now,
9737 because it's not clear yet that their impact is small enough to move
9738 from opt-in to opt-out. We recommend downstreams to leave memory
9739 accounting on by default if kernel 4.14 or higher is primarily
9740 used. On very resource constrained systems or when support for old
9741 kernels is a necessity, -Dmemory-accounting-default=false can be used
9742 to revert this change.
9744 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
9745 %udev_rules_update) and the journal catalog (%journal_catalog_update)
9746 from the upgrade scriptlets of individual packages now do nothing.
9747 Transfiletriggers have been added which will perform those updates
9748 once at the end of the transaction.
9750 Similar transfiletriggers have been added to execute any sysctl.d
9751 and binfmt.d rules. Thus, it should be unnecessary to provide any
9752 scriptlets to execute this configuration from package installation
9755 * systemd-sysusers gained a mode where the configuration to execute is
9756 specified on the command line, but this configuration is not executed
9757 directly, but instead it is merged with the configuration on disk,
9758 and the result is executed. This is useful for package installation
9759 scripts which want to create the user before installing any files on
9760 disk (in case some of those files are owned by that user), while
9761 still allowing local admin overrides.
9763 This functionality is exposed to rpm scriptlets through a new
9764 %sysusers_create_package macro. Old %sysusers_create and
9765 %sysusers_create_inline macros are deprecated.
9767 A transfiletrigger for sysusers.d configuration is now installed,
9768 which means that it should be unnecessary to call systemd-sysusers from
9769 package installation scripts, unless the package installs any files
9770 owned by those newly-created users, in which case
9771 %sysusers_create_package should be used.
9773 * Analogous change has been done for systemd-tmpfiles: it gained a mode
9774 where the command-line configuration is merged with the configuration
9775 on disk. This is exposed as the new %tmpfiles_create_package macro,
9776 and %tmpfiles_create is deprecated. A transfiletrigger is installed
9777 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
9778 from package installation scripts.
9780 * sysusers.d configuration for a user may now also specify the group
9781 number, in addition to the user number ("u username 123:456"), or
9782 without the user number ("u username -:456").
9784 * Configution items for systemd-sysusers can now be specified as
9785 positional arguments when the new --inline switch is used.
9787 * The login shell of users created through sysusers.d may now be
9788 specified (previously, it was always /bin/sh for root and
9789 /sbin/nologin for other users).
9791 * systemd-analyze gained a new --global switch to look at global user
9792 configuration. It also gained a unit-paths verb to list the unit load
9793 paths that are compiled into systemd (which can be used with
9794 --systemd, --user, or --global).
9796 * udevadm trigger gained a new --settle/-w option to wait for any
9797 triggered events to finish (but just those, and not any other events
9798 which are triggered meanwhile).
9800 * The action that systemd-logind takes when the lid is closed and the
9801 machine is connected to external power can now be configured using
9802 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
9803 was determined by HandleLidSwitch=, and, for backwards compatibility,
9804 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
9806 * journalctl will periodically call sd_journal_process() to make it
9807 resilient against inotify queue overruns when journal files are
9808 rotated very quickly.
9810 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
9811 sd_bus_get_n_queued_write — may be used to check the number of
9812 pending bus messages.
9814 * systemd gained a new
9815 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
9816 which can be used to migrate foreign processes to scope and service
9817 units. The primary user for this new API is systemd itself: the
9818 systemd --user instance uses this call of the systemd --system
9819 instance to migrate processes if it itself gets the request to
9820 migrate processes and the kernel refuses this due to access
9821 restrictions. Thanks to this "systemd-run --scope --user …" works
9822 again in pure cgroup v2 environments when invoked from the user
9825 * A new TemporaryFileSystem= setting can be used to mask out part of
9826 the real file system tree with tmpfs mounts. This may be combined
9827 with BindPaths= and BindReadOnlyPaths= to hide files or directories
9828 not relevant to the unit, while still allowing some paths lower in
9829 the tree to be accessed.
9831 ProtectHome=tmpfs may now be used to hide user home and runtime
9832 directories from units, in a way that is mostly equivalent to
9833 "TemporaryFileSystem=/home /run/user /root".
9835 * Non-service units are now started with KeyringMode=shared by default.
9836 This means that mount and swapon and other mount tools have access
9837 to keys in the main keyring.
9839 * /sys/fs/bpf is now mounted automatically.
9841 * QNX virtualization is now detected by systemd-detect-virt and may
9842 be used in ConditionVirtualization=.
9844 * IPAccounting= may now be enabled also for slice units.
9846 * A new -Dsplit-bin= build configuration switch may be used to specify
9847 whether bin and sbin directories are merged, or if they should be
9848 included separately in $PATH and various listings of executable
9849 directories. The build configuration scripts will try to autodetect
9850 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
9851 system, but distributions are encouraged to configure this
9854 * A new -Dok-color= build configuration switch may be used to change
9855 the colour of "OK" status messages.
9857 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
9858 PrivateNetwork=yes was buggy in previous versions of systemd. This
9859 means that after the upgrade and daemon-reexec, any such units must
9862 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
9863 will not exclude read-only files owned by root from cleanup.
9865 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
9866 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
9867 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
9868 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
9869 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
9870 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
9871 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
9872 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
9873 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
9874 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
9875 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
9876 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
9877 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
9878 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
9879 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
9880 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
9882 — Warsaw, 2018-03-05
9886 * Some keyboards come with a zoom see-saw or rocker which until now got
9887 mapped to the Linux "zoomin/out" keys in hwdb. However, these
9888 keycodes are not recognized by any major desktop. They now produce
9889 Up/Down key events so that they can be used for scrolling.
9891 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
9892 slightly: previously, if an argument was specified for lines of this
9893 type (i.e. the right-most column was set) this string was appended to
9894 existing files each time systemd-tmpfiles was run. This behaviour was
9895 different from what the documentation said, and not particularly
9896 useful, as repeated systemd-tmpfiles invocations would not be
9897 idempotent and grow such files without bounds. With this release
9898 behaviour has been altered to match what the documentation says:
9899 lines of this type only have an effect if the indicated files don't
9900 exist yet, and only then the argument string is written to the file.
9902 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
9903 systemd-tmpfiles behaviour: previously, read-only files owned by root
9904 were always excluded from the file "aging" algorithm (i.e. the
9905 automatic clean-up of directories like /tmp based on
9906 atime/mtime/ctime). We intend to drop this restriction, and age files
9907 by default even when owned by root and read-only. This behaviour was
9908 inherited from older tools, but there have been requests to remove
9909 it, and it's not obvious why this restriction was made in the first
9910 place. Please speak up now, if you are aware of software that requires
9911 this behaviour, otherwise we'll remove the restriction in v238.
9913 * A new environment variable $SYSTEMD_OFFLINE is now understood by
9914 systemctl. It takes a boolean argument. If on, systemctl assumes it
9915 operates on an "offline" OS tree, and will not attempt to talk to the
9916 service manager. Previously, this mode was implicitly enabled if a
9917 chroot() environment was detected, and this new environment variable
9918 now provides explicit control.
9920 * .path and .socket units may now be created transiently, too.
9921 Previously only service, mount, automount and timer units were
9922 supported as transient units. The systemd-run tool has been updated
9923 to expose this new functionality, you may hence use it now to bind
9924 arbitrary commands to path or socket activation on-the-fly from the
9925 command line. Moreover, almost all properties are now exposed for the
9926 unit types that already supported transient operation.
9928 * The systemd-mount command gained support for a new --owner= parameter
9929 which takes a user name, which is then resolved and included in uid=
9930 and gid= mount options string of the file system to mount.
9932 * A new unit condition ConditionControlGroupController= has been added
9933 that checks whether a specific cgroup controller is available.
9935 * Unit files, udev's .link files, and systemd-networkd's .netdev and
9936 .network files all gained support for a new condition
9937 ConditionKernelVersion= for checking against specific kernel
9940 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
9941 support for configuring device flags in the Flags= setting. In the
9942 same files, the [Tunnel] section gained support for configuring
9943 AllowLocalRemote=. The [Route] section in .network files gained
9944 support for configuring InitialCongestionWindow=,
9945 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
9946 understands RapidCommit=.
9948 * systemd-networkd's DHCPv6 support gained support for Prefix
9951 * sd-bus gained support for a new "watch-bind" feature. When this
9952 feature is enabled, an sd_bus connection may be set up to connect to
9953 an AF_UNIX socket in the file system as soon as it is created. This
9954 functionality is useful for writing early-boot services that
9955 automatically connect to the system bus as soon as it is started,
9956 without ugly time-based polling. systemd-networkd and
9957 systemd-resolved have been updated to make use of this
9958 functionality. busctl exposes this functionality in a new
9959 --watch-bind= command line switch.
9961 * sd-bus will now optionally synthesize a local "Connected" signal as
9962 soon as a D-Bus connection is set up fully. This message mirrors the
9963 already existing "Disconnected" signal which is synthesized when the
9964 connection is terminated. This signal is generally useful but
9965 particularly handy in combination with the "watch-bind" feature
9966 described above. Synthesizing of this message has to be requested
9967 explicitly through the new API call sd_bus_set_connected_signal(). In
9968 addition a new call sd_bus_is_ready() has been added that checks
9969 whether a connection is fully set up (i.e. between the "Connected" and
9970 "Disconnected" signals).
9972 * sd-bus gained two new calls sd_bus_request_name_async() and
9973 sd_bus_release_name_async() for asynchronously registering bus
9974 names. Similar, there is now sd_bus_add_match_async() for installing
9975 a signal match asynchronously. All of systemd's own services have
9976 been updated to make use of these calls. Doing these operations
9977 asynchronously has two benefits: it reduces the risk of deadlocks in
9978 case of cyclic dependencies between bus services, and it speeds up
9979 service initialization since synchronization points for bus
9980 round-trips are removed.
9982 * sd-bus gained two new calls sd_bus_match_signal() and
9983 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
9984 and sd_bus_add_match_async() but instead of taking a D-Bus match
9985 string take match fields as normal function parameters.
9987 * sd-bus gained two new calls sd_bus_set_sender() and
9988 sd_bus_message_set_sender() for setting the sender name of outgoing
9989 messages (either for all outgoing messages or for just one specific
9990 one). These calls are only useful in direct connections as on
9991 brokered connections the broker fills in the sender anyway,
9992 overwriting whatever the client filled in.
9994 * sd-event gained a new pseudo-handle that may be specified on all API
9995 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
9996 used this refers to the default event loop object of the calling
9997 thread. Note however that this does not implicitly allocate one —
9998 which has to be done prior by using sd_event_default(). Similarly
9999 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
10000 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
10001 to the default bus of the specified type of the calling thread. Here
10002 too this does not implicitly allocate bus connection objects, this
10003 has to be done prior with sd_bus_default() and friends.
10005 * sd-event gained a new call pair
10006 sd_event_source_{get|set}_io_fd_own(). This may be used to request
10007 automatic closure of the file descriptor an IO event source watches
10008 when the event source is destroyed.
10010 * systemd-networkd gained support for natively configuring WireGuard
10013 * In previous versions systemd synthesized user records both for the
10014 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
10015 internally. In order to simplify distribution-wide renames of the
10016 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
10017 new transitional flag file has been added: if
10018 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
10019 user and group record within the systemd codebase is disabled.
10021 * systemd-notify gained a new --uid= option for selecting the source
10022 user/UID to use for notification messages sent to the service
10025 * journalctl gained a new --grep= option to list only entries in which
10026 the message matches a certain pattern. By default matching is case
10027 insensitive if the pattern is lowercase, and case sensitive
10028 otherwise. Option --case-sensitive=yes|no can be used to override
10029 this an specify case sensitivity or case insensitivity.
10031 * There's now a "systemd-analyze service-watchdogs" command for printing
10032 the current state of the service runtime watchdog, and optionally
10033 enabling or disabling the per-service watchdogs system-wide if given a
10034 boolean argument (i.e. the concept you configure in WatchdogSec=), for
10035 debugging purposes. There's also a kernel command line option
10036 systemd.service_watchdogs= for controlling the same.
10038 * Two new "log-level" and "log-target" options for systemd-analyze were
10039 added that merge the now deprecated get-log-level, set-log-level and
10040 get-log-target, set-log-target pairs. The deprecated options are still
10041 understood for backwards compatibility. The two new options print the
10042 current value when no arguments are given, and set them when a
10043 level/target is given as an argument.
10045 * sysusers.d's "u" lines now optionally accept both a UID and a GID
10046 specification, separated by a ":" character, in order to create users
10047 where UID and GID do not match.
10049 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
10050 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
10051 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
10052 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
10053 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
10054 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
10055 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
10056 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
10057 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
10058 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
10059 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
10060 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
10061 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
10062 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
10063 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
10064 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
10065 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
10066 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
10067 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
10068 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
10075 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
10076 in v235 has been extended to also set the dummy.ko module option
10077 numdummies=0, preventing the kernel from automatically creating
10078 dummy0. All dummy interfaces must now be explicitly created.
10080 * Unknown '%' specifiers in configuration files are now rejected. This
10081 applies to units and tmpfiles.d configuration. Any percent characters
10082 that are followed by a letter or digit that are not supposed to be
10083 interpreted as the beginning of a specifier should be escaped by
10084 doubling ("%%"). (So "size=5%" is still accepted, as well as
10085 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
10086 valid specifiers today.)
10088 * systemd-resolved now maintains a new dynamic
10089 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
10090 recommended to make /etc/resolv.conf a symlink to it. This file
10091 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
10092 includes dynamically acquired search domains, achieving more correct
10093 DNS resolution by software that bypasses local DNS APIs such as NSS.
10095 * The "uaccess" udev tag has been dropped from /dev/kvm and
10096 /dev/dri/renderD*. These devices now have the 0666 permissions by
10097 default (but this may be changed at build-time). /dev/dri/renderD*
10098 will now be owned by the "render" group along with /dev/kfd.
10100 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
10101 systemd-journal-gatewayd.service and
10102 systemd-journal-upload.service. This means "nss-systemd" must be
10103 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
10104 services are resolved properly.
10106 * In /etc/fstab two new mount options are now understood:
10107 x-systemd.makefs and x-systemd.growfs. The former has the effect that
10108 the configured file system is formatted before it is mounted, the
10109 latter that the file system is resized to the full block device size
10110 after it is mounted (i.e. if the file system is smaller than the
10111 partition it resides on, it's grown). This is similar to the fsck
10112 logic in /etc/fstab, and pulls in systemd-makefs@.service and
10113 systemd-growfs@.service as necessary, similar to
10114 systemd-fsck@.service. Resizing is currently only supported on ext4
10117 * In systemd-networkd, the IPv6 RA logic now optionally may announce
10118 DNS server and domain information.
10120 * Support for the LUKS2 on-disk format for encrypted partitions has
10121 been added. This requires libcryptsetup2 during compilation and
10124 * The systemd --user instance will now signal "readiness" when its
10125 basic.target unit has been reached, instead of when the run queue ran
10126 empty for the first time.
10128 * Tmpfiles.d with user configuration are now also supported.
10129 systemd-tmpfiles gained a new --user switch, and snippets placed in
10130 ~/.config/user-tmpfiles.d/ and corresponding directories will be
10131 executed by systemd-tmpfiles --user running in the new
10132 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
10133 running in the user session.
10135 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
10136 %S resolves to the top-level state directory (/var/lib for the system
10137 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
10138 top-level cache directory (/var/cache for the system instance,
10139 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
10140 logs directory (/var/log for the system instance,
10141 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
10142 existing %t specifier, that resolves to the top-level runtime
10143 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
10146 * journalctl learnt a new parameter --output-fields= for limiting the
10147 set of journal fields to output in verbose and JSON output modes.
10149 * systemd-timesyncd's configuration file gained a new option
10150 RootDistanceMaxSec= for setting the maximum root distance of servers
10151 it'll use, as well as the new options PollIntervalMinSec= and
10152 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
10154 * bootctl gained a new command "list" for listing all available boot
10155 menu items on systems that follow the boot loader specification.
10157 * systemctl gained a new --dry-run switch that shows what would be done
10158 instead of doing it, and is currently supported by the shutdown and
10161 * ConditionSecurity= can now detect the TOMOYO security module.
10163 * Unit file [Install] sections are now also respected in unit drop-in
10164 files. This is intended to be used by drop-ins under /usr/lib/.
10166 * systemd-firstboot may now also set the initial keyboard mapping.
10168 * Udev "changed" events for devices which are exposed as systemd
10169 .device units are now propagated to units specified in
10170 ReloadPropagatedFrom= as reload requests.
10172 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
10173 unit template name (i.e. a name in the form of 'foobar@.service',
10174 without the instance component between the '@' and - the '.'), then
10175 the escaped sysfs path of the device is automatically used as the
10178 * SystemCallFilter= in unit files has been extended so that an "errno"
10179 can be specified individually for each system call. Example:
10180 SystemCallFilter=~uname:EILSEQ.
10182 * The cgroup delegation logic has been substantially updated. Delegate=
10183 now optionally takes a list of controllers (instead of a boolean, as
10184 before), which lists the controllers to delegate at least.
10186 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
10188 * A new LogLevelMax= setting configures the maximum log level any
10189 process of the service may log at (i.e. anything with a lesser
10190 priority than what is specified is automatically dropped). A new
10191 LogExtraFields= setting allows configuration of additional journal
10192 fields to attach to all log records generated by any of the unit's
10195 * New StandardInputData= and StandardInputText= settings along with the
10196 new option StandardInput=data may be used to configure textual or
10197 binary data that shall be passed to the executed service process via
10198 standard input, encoded in-line in the unit file.
10200 * StandardInput=, StandardOutput= and StandardError= may now be used to
10201 connect stdin/stdout/stderr of executed processes directly with a
10202 file or AF_UNIX socket in the file system, using the new "file:" option.
10204 * A new unit file option CollectMode= has been added, that allows
10205 tweaking the garbage collection logic for units. It may be used to
10206 tell systemd to garbage collect units that have failed automatically
10207 (normally it only GCs units that exited successfully). systemd-run
10208 and systemd-mount expose this new functionality with a new -G option.
10210 * "machinectl bind" may now be used to bind mount non-directories
10211 (i.e. regularfiles, devices, fifos, sockets).
10213 * systemd-analyze gained a new verb "calendar" for validating and
10214 testing calendar time specifications to use for OnCalendar= in timer
10215 units. Besides validating the expression it will calculate the next
10216 time the specified expression would elapse.
10218 * In addition to the pre-existing FailureAction= unit file setting
10219 there's now SuccessAction=, for configuring a shutdown action to
10220 execute when a unit completes successfully. This is useful in
10221 particular inside containers that shall terminate after some workload
10222 has been completed. Also, both options are now supported for all unit
10223 types, not just services.
10225 * networkds's IP rule support gained two new options
10226 IncomingInterface= and OutgoingInterface= for configuring the incoming
10227 and outgoing interfaces of configured rules. systemd-networkd also
10228 gained support for "vxcan" network devices.
10230 * networkd gained a new setting RequiredForOnline=, taking a
10231 boolean. If set, systemd-wait-online will take it into consideration
10232 when determining that the system is up, otherwise it will ignore the
10233 interface for this purpose.
10235 * The sd_notify() protocol gained support for a new operation: with
10236 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
10237 store again, ahead of POLLHUP or POLLERR when they are removed
10240 * A new document doc/UIDS-GIDS.md has been added to the source tree,
10241 that documents the UID/GID range and assignment assumptions and
10242 requirements of systemd.
10244 * The watchdog device PID 1 will ping may now be configured through the
10245 WatchdogDevice= configuration file setting, or by setting the
10246 systemd.watchdog_service= kernel command line option.
10248 * systemd-resolved's gained support for registering DNS-SD services on
10249 the local network using MulticastDNS. Services may either be
10250 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
10251 the same dir below /run, /usr/lib), or through its D-Bus API.
10253 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
10254 extend the effective start, runtime, and stop time. The service must
10255 continue to send EXTEND_TIMEOUT_USEC within the period specified to
10256 prevent the service manager from making the service as timedout.
10258 * systemd-resolved's DNSSEC support gained support for RFC 8080
10259 (Ed25519 keys and signatures).
10261 * The systemd-resolve command line tool gained a new set of options
10262 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
10263 --set-nta= and --revert to configure per-interface DNS configuration
10264 dynamically during runtime. It's useful for pushing DNS information
10265 into systemd-resolved from DNS hook scripts that various interface
10266 managing software supports (such as pppd).
10268 * systemd-nspawn gained a new --network-namespace-path= command line
10269 option, which may be used to make a container join an existing
10270 network namespace, by specifying a path to a "netns" file.
10272 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
10273 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
10274 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
10275 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
10276 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
10277 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
10278 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
10279 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
10280 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
10281 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
10282 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
10283 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
10284 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
10285 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
10286 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
10287 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
10288 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
10289 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
10290 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
10291 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
10292 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
10293 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
10294 Jędrzejewski-Szmek, Zeal Jagannatha
10296 — Berlin, 2017-12-14
10300 * INCOMPATIBILITY: systemd-logind.service and other long-running
10301 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
10302 communication with the outside. This generally improves security of
10303 the system, and is in almost all cases a safe and good choice, as
10304 these services do not and should not provide any network-facing
10305 functionality. However, systemd-logind uses the glibc NSS API to
10306 query the user database. This creates problems on systems where NSS
10307 is set up to directly consult network services for user database
10308 lookups. In particular, this creates incompatibilities with the
10309 "nss-nis" module, which attempts to directly contact the NIS/YP
10310 network servers it is configured for, and will now consistently
10311 fail. In such cases, it is possible to turn off IP sandboxing for
10312 systemd-logind.service (set IPAddressDeny= in its [Service] section
10313 to the empty string, via a .d/ unit file drop-in). Downstream
10314 distributions might want to update their nss-nis packaging to include
10315 such a drop-in snippet, accordingly, to hide this incompatibility
10316 from the user. Another option is to make use of glibc's nscd service
10317 to proxy such network requests through a privilege-separated, minimal
10318 local caching daemon, or to switch to more modern technologies such
10319 sssd, whose NSS hook-ups generally do not involve direct network
10320 access. In general, we think it's definitely time to question the
10321 implementation choices of nss-nis, i.e. whether it's a good idea
10322 today to embed a network-facing loadable module into all local
10323 processes that need to query the user database, including the most
10324 trivial and benign ones, such as "ls". For more details about
10325 IPAddressDeny= see below.
10327 * A new modprobe.d drop-in is now shipped by default that sets the
10328 bonding module option max_bonds=0. This overrides the kernel default,
10329 to avoid conflicts and ambiguity as to whether or not bond0 should be
10330 managed by systemd-networkd or not. This resolves multiple issues
10331 with bond0 properties not being applied, when bond0 is configured
10332 with systemd-networkd. Distributors may choose to not package this,
10333 however in that case users will be prevented from correctly managing
10334 bond0 interface using systemd-networkd.
10336 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
10337 which print the logging level and target of the system manager. They
10338 complement the existing "set-log-level" and "set-log-target" verbs
10339 used to change those values.
10341 * journald.conf gained a new boolean setting ReadKMsg= which defaults
10342 to on. If turned off kernel log messages will not be read by
10343 systemd-journald or included in the logs. It also gained a new
10344 setting LineMax= for configuring the maximum line length in
10345 STDOUT/STDERR log streams. The new default for this value is 48K, up
10346 from the previous hardcoded 2048.
10348 * A new unit setting RuntimeDirectoryPreserve= has been added, which
10349 allows more detailed control of what to do with a runtime directory
10350 configured with RuntimeDirectory= (i.e. a directory below /run or
10351 $XDG_RUNTIME_DIR) after a unit is stopped.
10353 * The RuntimeDirectory= setting for units gained support for creating
10354 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
10355 one top-level directory.
10357 * Units gained new options StateDirectory=, CacheDirectory=,
10358 LogsDirectory= and ConfigurationDirectory= which are closely related
10359 to RuntimeDirectory= but manage per-service directories below
10360 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
10361 possible to write unit files which when activated automatically gain
10362 properly owned service specific directories in these locations, thus
10363 making unit files self-contained and increasing compatibility with
10364 stateless systems and factory reset where /etc or /var are
10365 unpopulated at boot. Matching these new settings there's also
10366 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
10367 ConfigurationDirectoryMode= for configuring the access mode of these
10368 directories. These settings are particularly useful in combination
10369 with DynamicUser=yes as they provide secure, properly-owned,
10370 writable, and stateful locations for storage, excluded from the
10371 sandbox that such services live in otherwise.
10373 * Automake support has been removed from this release. systemd is now
10376 * systemd-journald will now aggressively cache client metadata during
10377 runtime, speeding up log write performance under pressure. This comes
10378 at a small price though: as much of the metadata is read
10379 asynchronously from /proc/ (and isn't implicitly attached to log
10380 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
10381 metadata stored alongside a log entry might be slightly
10382 out-of-date. Previously it could only be slightly newer than the log
10383 message. The time window is small however, and given that the kernel
10384 is unlikely to be improved anytime soon in this regard, this appears
10387 * nss-myhostname/systemd-resolved will now by default synthesize an
10388 A/AAAA resource record for the "_gateway" hostname, pointing to the
10389 current default IP gateway. Previously it did that for the "gateway"
10390 name, hampering adoption, as some distributions wanted to leave that
10391 hostname open for local use. The old behaviour may still be
10392 requested at build time.
10394 * systemd-networkd's [Address] section in .network files gained a new
10395 Scope= setting for configuring the IP address scope. The [Network]
10396 section gained a new boolean setting ConfigureWithoutCarrier= that
10397 tells systemd-networkd to ignore link sensing when configuring the
10398 device. The [DHCP] section gained a new Anonymize= boolean option for
10399 turning on a number of options suggested in RFC 7844. A new
10400 [RoutingPolicyRule] section has been added for configuring the IP
10401 routing policy. The [Route] section has gained support for a new
10402 Type= setting which permits configuring
10403 blackhole/unreachable/prohibit routes.
10405 * The [VRF] section in .netdev files gained a new Table= setting for
10406 configuring the routing table to use. The [Tunnel] section gained a
10407 new Independent= boolean field for configuring tunnels independent of
10408 an underlying network interface. The [Bridge] section gained a new
10409 GroupForwardMask= option for configuration of propagation of link
10410 local frames between bridge ports.
10412 * The WakeOnLan= setting in .link files gained support for a number of
10413 new modes. A new TCP6SegmentationOffload= setting has been added for
10414 configuring TCP/IPv6 hardware segmentation offload.
10416 * The IPv6 RA sender implementation may now optionally send out RDNSS
10417 and RDNSSL records to supply DNS configuration to peers.
10419 * systemd-nspawn gained support for a new --system-call-filter= command
10420 line option for adding and removing entries in the default system
10421 call filter it applies. Moreover systemd-nspawn has been changed to
10422 implement a system call allow list instead of a deny list.
10424 * systemd-run gained support for a new --pipe command line option. If
10425 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
10426 are directly passed on to the activated transient service
10427 executable. This allows invoking arbitrary processes as systemd
10428 services (for example to take benefit of dependency management,
10429 accounting management, resource management or log management that is
10430 done automatically for services) — while still allowing them to be
10431 integrated in a classic UNIX shell pipeline.
10433 * When a service sends RELOAD=1 via sd_notify() and reload propagation
10434 using ReloadPropagationTo= is configured, a reload is now propagated
10435 to configured units. (Previously this was only done on explicitly
10436 requested reloads, using "systemctl reload" or an equivalent
10439 * For each service unit a restart counter is now kept: it is increased
10440 each time the service is restarted due to Restart=, and may be
10441 queried using "systemctl show -p NRestarts …".
10443 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
10444 @signal and @timer have been added, for usage with SystemCallFilter=
10445 in unit files and the new --system-call-filter= command line option
10446 of systemd-nspawn (see above).
10448 * ExecStart= lines in unit files gained two new modifiers: when a
10449 command line is prefixed with "!" the command will be executed as
10450 configured, except for the credentials applied by
10451 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
10452 "+", but does still apply namespacing options unlike "+". There's
10453 also "!!" now, which is mostly identical, but becomes a NOP on
10454 systems that support ambient capabilities. This is useful to write
10455 unit files that work with ambient capabilities where possible but
10456 automatically fall back to traditional privilege dropping mechanisms
10457 on systems where this is not supported.
10459 * ListenNetlink= settings in socket units now support RDMA netlink
10462 * A new unit file setting LockPersonality= has been added which permits
10463 locking down the chosen execution domain ("personality") of a service
10466 * A new special target "getty-pre.target" has been added, which is
10467 ordered before all text logins, and may be used to order services
10468 before textual logins acquire access to the console.
10470 * systemd will now attempt to load the virtio-rng.ko kernel module very
10471 early on if a VM environment supporting this is detected. This should
10472 improve entropy during early boot in virtualized environments.
10474 * A _netdev option is now supported in /etc/crypttab that operates in a
10475 similar way as the same option in /etc/fstab: it permits configuring
10476 encrypted devices that need to be ordered after the network is up.
10477 Following this logic, two new special targets
10478 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
10479 added that are to cryptsetup.target what remote-fs.target and
10480 remote-fs-pre.target are to local-fs.target.
10482 * Service units gained a new UnsetEnvironment= setting which permits
10483 unsetting specific environment variables for services that are
10484 normally passed to it (for example in order to mask out locale
10485 settings for specific services that can't deal with it).
10487 * Units acquired a new boolean option IPAccounting=. When turned on, IP
10488 traffic accounting (packet count as well as byte count) is done for
10489 the service, and shown as part of "systemctl status" or "systemd-run
10492 * Service units acquired two new options IPAddressAllow= and
10493 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
10494 for configuring a simple IP access control list for all sockets of
10495 the unit. These options are available also on .slice and .socket
10496 units, permitting flexible access list configuration for individual
10497 services as well as groups of services (as defined by a slice unit),
10498 including system-wide. Note that IP ACLs configured this way are
10499 enforced on every single IPv4 and IPv6 socket created by any process
10500 of the service unit, and apply to ingress as well as egress traffic.
10502 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
10503 structured log message is generated each time the unit is stopped,
10504 containing information about the consumed resources of this
10507 * A new setting KeyringMode= has been added to unit files, which may be
10508 used to control how the kernel keyring is set up for executed
10511 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
10512 "systemctl kexec" and "systemctl exit" are now always asynchronous in
10513 behaviour (that is: these commands return immediately after the
10514 operation was enqueued instead of waiting for the operation to
10515 complete). Previously, "systemctl poweroff" and "systemctl reboot"
10516 were asynchronous on systems using systemd-logind (i.e. almost
10517 always, and like they were on sysvinit), and the other three commands
10518 were unconditionally synchronous. With this release this is cleaned
10519 up, and callers will see the same asynchronous behaviour on all
10520 systems for all five operations.
10522 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
10525 * .timer units now accept calendar specifications in other timezones
10526 than UTC or the local timezone.
10528 * The tmpfiles snippet var.conf has been changed to create
10529 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
10530 the "utmp" group already, and it appears to be generally understood
10531 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
10532 databases. Previously this was implemented correctly for all these
10533 databases excepts btmp, which has been opened up like this now
10534 too. Note that while the other databases are world-readable
10535 (i.e. 0644), btmp is not and remains more restrictive.
10537 * The systemd-resolve tool gained a new --reset-server-features
10538 switch. When invoked like this systemd-resolved will forget
10539 everything it learnt about the features supported by the configured
10540 upstream DNS servers, and restarts the feature probing logic on the
10541 next resolver look-up for them at the highest feature level
10544 * The status dump systemd-resolved sends to the logs upon receiving
10545 SIGUSR1 now also includes information about all DNS servers it is
10546 configured to use, and the features levels it probed for them.
10548 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
10549 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
10550 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
10551 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
10552 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
10553 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
10554 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
10555 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
10556 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
10557 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
10558 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
10559 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
10560 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
10561 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
10562 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
10563 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
10564 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
10565 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
10566 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
10567 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
10569 — Berlin, 2017-10-06
10573 * Meson is now supported as build system in addition to Automake. It is
10574 our plan to remove Automake in one of our next releases, so that
10575 Meson becomes our exclusive build system. Hence, please start using
10576 the Meson build system in your downstream packaging. There's plenty
10577 of documentation around how to use Meson, the extremely brief
10580 ./autogen.sh && ./configure && make && sudo make install
10584 meson build && ninja -C build && sudo ninja -C build install
10586 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
10587 which permits configuring a timeout on the time a job is
10588 running. This is particularly useful for setting timeouts on jobs for
10591 * Unit files gained two new options ConditionUser= and ConditionGroup=
10592 for conditionalizing units based on the identity of the user/group
10593 running a systemd user instance.
10595 * systemd-networkd now understands a new FlowLabel= setting in the
10596 [VXLAN] section of .network files, as well as a Priority= in
10597 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
10598 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
10599 gained support for configuration of GENEVE links, and IPv6 address
10600 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
10602 * .link files now understand a new Port= setting.
10604 * systemd-networkd's DHCP support gained support for DHCP option 119
10605 (domain search list).
10607 * systemd-networkd gained support for serving IPv6 address ranges using
10608 the Router Advertisement protocol. The new .network configuration
10609 section [IPv6Prefix] may be used to configure the ranges to
10610 serve. This is implemented based on a new, minimal, native server
10611 implementation of RA.
10613 * journalctl's --output= switch gained support for a new parameter
10614 "short-iso-precise" for a mode where timestamps are shown as precise
10617 * systemd-udevd's "net_id" builtin may now generate stable network
10618 interface names from IBM PowerVM VIO devices as well as ACPI platform
10621 * MulticastDNS support in systemd-resolved may now be explicitly
10622 enabled/disabled using the new MulticastDNS= configuration file
10625 * systemd-resolved may now optionally use libidn2 instead of the libidn
10626 for processing internationalized domain names. Support for libidn2
10627 should be considered experimental and should not be enabled by
10630 * "machinectl pull-tar" and related call may now do verification of
10631 downloaded images using SUSE-style .sha256 checksum files in addition
10632 to the already existing support for validating using Ubuntu-style
10635 * sd-bus gained support for a new sd_bus_message_appendv() call which
10636 is va_list equivalent of sd_bus_message_append().
10638 * sd-boot gained support for validating images using SHIM/MOK.
10640 * The SMACK code learnt support for "onlycap".
10642 * systemd-mount --umount is now much smarter in figuring out how to
10643 properly unmount a device given its mount or device path.
10645 * The code to call libnss_dns as a fallback from libnss_resolve when
10646 the communication with systemd-resolved fails was removed. This
10647 fallback was redundant and interfered with the [!UNAVAIL=return]
10648 suffix. See nss-resolve(8) for the recommended configuration.
10650 * systemd-logind may now be restarted without losing state. It stores
10651 the file descriptors for devices it manages in the system manager
10652 using the FDSTORE= mechanism. Please note that further changes in
10653 other components may be required to make use of this (for example
10654 Xorg has code to listen for stops of systemd-logind and terminate
10655 itself when logind is stopped or restarted, in order to avoid using
10656 stale file descriptors for graphical devices, which is now
10657 counterproductive and must be reverted in order for restarts of
10658 systemd-logind to be safe. See
10659 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
10661 * All kernel-install plugins are called with the environment variable
10662 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
10663 /etc/machine-id. If the machine ID could not be determined,
10664 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
10665 anything in the entry directory (passed as the second argument) if
10666 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
10667 temporary directory is passed as the entry directory and removed
10668 after all the plugins exit.
10670 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
10671 will now use its value as the machine ID instead of the machine ID
10672 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
10673 /etc/machine-info and no machine ID is set in /etc/machine-id,
10674 kernel-install will try to store the current machine ID there as
10675 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
10676 will generate a new UUID, store it in /etc/machine-info as
10677 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
10679 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
10680 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
10681 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
10682 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
10683 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
10684 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
10685 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
10686 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
10687 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
10688 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
10689 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
10690 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
10691 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
10692 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
10693 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
10694 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
10695 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
10696 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
10697 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
10698 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
10699 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
10700 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
10701 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
10702 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
10703 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
10704 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
10705 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
10706 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
10707 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
10710 — Berlin, 2017-07-12
10714 * The "hybrid" control group mode has been modified to improve
10715 compatibility with "legacy" cgroups-v1 setups. Specifically, the
10716 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
10717 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
10718 cgroups-v1 hierarchy), the only externally visible change being that
10719 the cgroups-v2 hierarchy is also mounted, to
10720 /sys/fs/cgroup/unified. This should provide a large degree of
10721 compatibility with "legacy" cgroups-v1, while taking benefit of the
10722 better management capabilities of cgroups-v2.
10724 * The default control group setup mode may be selected both a boot-time
10725 via a set of kernel command line parameters (specifically:
10726 systemd.unified_cgroup_hierarchy= and
10727 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
10728 default selected on the configure command line
10729 (--with-default-hierarchy=). The upstream default is "hybrid"
10730 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
10731 this will change in a future systemd version to be "unified" (pure
10732 cgroups-v2 mode). The third option for the compile time option is
10733 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
10734 distributions to default to "hybrid" mode for release distributions,
10735 starting with v233. We recommend "unified" for development
10736 distributions (specifically: distributions such as Fedora's rawhide)
10737 as that's where things are headed in the long run. Use "legacy" for
10738 greatest stability and compatibility only.
10740 * Note one current limitation of "unified" and "hybrid" control group
10741 setup modes: the kernel currently does not permit the systemd --user
10742 instance (i.e. unprivileged code) to migrate processes between two
10743 disconnected cgroup subtrees, even if both are managed and owned by
10744 the user. This effectively means "systemd-run --user --scope" doesn't
10745 work when invoked from outside of any "systemd --user" service or
10746 scope. Specifically, it is not supported from session scopes. We are
10747 working on fixing this in a future systemd version. (See #3388 for
10748 further details about this.)
10750 * DBus policy files are now installed into /usr rather than /etc. Make
10751 sure your system has dbus >= 1.9.18 running before upgrading to this
10752 version, or override the install path with --with-dbuspolicydir= .
10754 * All python scripts shipped with systemd (specifically: the various
10755 tests written in Python) now require Python 3.
10757 * systemd unit tests can now run standalone (without the source or
10758 build directories), and can be installed into /usr/lib/systemd/tests/
10759 with 'make install-tests'.
10761 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
10762 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
10765 * Support for the %c, %r, %R specifiers in unit files has been
10766 removed. Specifiers are not supposed to be dependent on configuration
10767 in the unit file itself (so that they resolve the same regardless
10768 where used in the unit files), but these specifiers were influenced
10769 by the Slice= option.
10771 * The shell invoked by debug-shell.service now defaults to /bin/sh in
10772 all cases. If distributions want to use a different shell for this
10773 purpose (for example Fedora's /sbin/sushell) they need to specify
10774 this explicitly at configure time using --with-debug-shell=.
10776 * The confirmation spawn prompt has been reworked to offer the
10779 (c)ontinue, proceed without asking anymore
10780 (D)ump, show the state of the unit
10781 (f)ail, don't execute the command and pretend it failed
10783 (i)nfo, show a short summary of the unit
10784 (j)obs, show jobs that are in progress
10785 (s)kip, don't execute the command and pretend it succeeded
10786 (y)es, execute the command
10788 The 'n' choice for the confirmation spawn prompt has been removed,
10789 because its meaning was confusing.
10791 The prompt may now also be redirected to an alternative console by
10792 specifying the console as parameter to systemd.confirm_spawn=.
10794 * Services of Type=notify require a READY=1 notification to be sent
10795 during startup. If no such message is sent, the service now fails,
10796 even if the main process exited with a successful exit code.
10798 * Services that fail to start up correctly now always have their
10799 ExecStopPost= commands executed. Previously, they'd enter "failed"
10800 state directly, without executing these commands.
10802 * The option MulticastDNS= of network configuration files has acquired
10803 an actual implementation. With MulticastDNS=yes a host can resolve
10804 names of remote hosts and reply to mDNS A and AAAA requests.
10806 * When units are about to be started an additional check is now done to
10807 ensure that all dependencies of type BindsTo= (when used in
10808 combination with After=) have been started.
10810 * systemd-analyze gained a new verb "syscall-filter" which shows which
10811 system call groups are defined for the SystemCallFilter= unit file
10812 setting, and which system calls they contain.
10814 * A new system call filter group "@filesystem" has been added,
10815 consisting of various file system related system calls. Group
10816 "@reboot" has been added, covering reboot, kexec and shutdown related
10817 calls. Finally, group "@swap" has been added covering swap
10818 configuration related calls.
10820 * A new unit file option RestrictNamespaces= has been added that may be
10821 used to restrict access to the various process namespace types the
10822 Linux kernel provides. Specifically, it may be used to take away the
10823 right for a service unit to create additional file system, network,
10824 user, and other namespaces. This sandboxing option is particularly
10825 relevant due to the high amount of recently discovered namespacing
10826 related vulnerabilities in the kernel.
10828 * systemd-udev's .link files gained support for a new AutoNegotiation=
10829 setting for configuring Ethernet auto-negotiation.
10831 * systemd-networkd's .network files gained support for a new
10832 ListenPort= setting in the [DHCP] section to explicitly configure the
10833 UDP client port the DHCP client shall listen on.
10835 * .network files gained a new Unmanaged= boolean setting for explicitly
10836 excluding one or more interfaces from management by systemd-networkd.
10838 * The systemd-networkd ProxyARP= option has been renamed to
10839 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
10840 renamed to ReduceARPProxy=. The old names continue to be available
10843 * systemd-networkd gained support for configuring IPv6 Proxy NDP
10844 addresses via the new IPv6ProxyNDPAddress= .network file setting.
10846 * systemd-networkd's bonding device support gained support for two new
10847 configuration options ActiveSlave= and PrimarySlave=.
10849 * The various options in the [Match] section of .network files gained
10850 support for negative matching.
10852 * New systemd-specific mount options are now understood in /etc/fstab:
10854 x-systemd.mount-timeout= may be used to configure the maximum
10855 permitted runtime of the mount command.
10857 x-systemd.device-bound may be set to bind a mount point to its
10858 backing device unit, in order to automatically remove a mount point
10859 if its backing device is unplugged. This option may also be
10860 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
10861 on the block device, which is now automatically set for all CDROM
10862 drives, so that mounted CDs are automatically unmounted when they are
10863 removed from the drive.
10865 x-systemd.after= and x-systemd.before= may be used to explicitly
10866 order a mount after or before another unit or mount point.
10868 * Enqueued start jobs for device units are now automatically garbage
10869 collected if there are no jobs waiting for them anymore.
10871 * systemctl list-jobs gained two new switches: with --after, for every
10872 queued job the jobs it's waiting for are shown; with --before the
10873 jobs which it's blocking are shown.
10875 * systemd-nspawn gained support for ephemeral boots from disk images
10876 (or in other words: --ephemeral and --image= may now be
10877 combined). Moreover, ephemeral boots are now supported for normal
10878 directories, even if the backing file system is not btrfs. Of course,
10879 if the file system does not support file system snapshots or
10880 reflinks, the initial copy operation will be relatively expensive, but
10881 this should still be suitable for many use cases.
10883 * Calendar time specifications in .timer units now support
10884 specifications relative to the end of a month by using "~" instead of
10885 "-" as separator between month and day. For example, "*-02~03" means
10886 "the third last day in February". In addition a new syntax for
10887 repeated events has been added using the "/" character. For example,
10888 "9..17/2:00" means "every two hours from 9am to 5pm".
10890 * systemd-socket-proxyd gained a new parameter --connections-max= for
10891 configuring the maximum number of concurrent connections.
10893 * sd-id128 gained a new API for generating unique IDs for the host in a
10894 way that does not leak the machine ID. Specifically,
10895 sd_id128_get_machine_app_specific() derives an ID based on the
10896 machine ID in a well-defined, non-reversible, stable way. This is
10897 useful whenever an identifier for the host is needed but where the
10898 identifier shall not be useful to identify the system beyond the
10899 scope of the application itself. (Internally this uses HMAC-SHA256 as
10900 keyed hash function using the machine ID as input.)
10902 * NotifyAccess= gained a new supported value "exec". When set
10903 notifications are accepted from all processes systemd itself invoked,
10904 including all control processes.
10906 * .nspawn files gained support for defining overlay mounts using the
10907 Overlay= and OverlayReadOnly= options. Previously this functionality
10908 was only available on the systemd-nspawn command line.
10910 * systemd-nspawn's --bind= and --overlay= options gained support for
10911 bind/overlay mounts whose source lies within the container tree by
10912 prefixing the source path with "+".
10914 * systemd-nspawn's --bind= and --overlay= options gained support for
10915 automatically allocating a temporary source directory in /var/tmp
10916 that is removed when the container dies. Specifically, if the source
10917 directory is specified as empty string this mechanism is selected. An
10918 example usage is --overlay=+/var::/var, which creates an overlay
10919 mount based on the original /var contained in the image, overlaid
10920 with a temporary directory in the host's /var/tmp. This way changes
10921 to /var are automatically flushed when the container shuts down.
10923 * systemd-nspawn --image= option does now permit raw file system block
10924 devices (in addition to images containing partition tables, as
10927 * The disk image dissection logic in systemd-nspawn gained support for
10928 automatically setting up LUKS encrypted as well as Verity protected
10929 partitions. When a container is booted from an encrypted image the
10930 passphrase is queried at start-up time. When a container with Verity
10931 data is started, the root hash is search in a ".roothash" file
10932 accompanying the disk image (alternatively, pass the root hash via
10933 the new --root-hash= command line option).
10935 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
10936 be used to dissect disk images the same way as systemd-nspawn does
10937 it, following the Bootable Partition Specification. It may even be
10938 used to mount disk images with complex partition setups (including
10939 LUKS and Verity partitions) to a local host directory, in order to
10940 inspect them. This tool is not considered public API (yet), and is
10941 thus not installed into /usr/bin. Please do not rely on its
10942 existence, since it might go away or be changed in later systemd
10945 * A new generator "systemd-verity-generator" has been added, similar in
10946 style to "systemd-cryptsetup-generator", permitting automatic setup of
10947 Verity root partitions when systemd boots up. In order to make use of
10948 this your partition setup should follow the Discoverable Partitions
10949 Specification, and the GPT partition ID of the root file system
10950 partition should be identical to the upper 128-bit of the Verity root
10951 hash. The GPT partition ID of the Verity partition protecting it
10952 should be the lower 128-bit of the Verity root hash. If the partition
10953 image follows this model it is sufficient to specify a single
10954 "roothash=" kernel command line argument to both configure which root
10955 image and verity partition to use as well as the root hash for
10956 it. Note that systemd-nspawn's Verity support follows the same
10957 semantics, meaning that disk images with proper Verity data in place
10958 may be booted in containers with systemd-nspawn as well as on
10959 physical systems via the verity generator. Also note that the "mkosi"
10960 tool available at https://github.com/systemd/mkosi has been updated
10961 to generate Verity protected disk images following this scheme. In
10962 fact, it has been updated to generate disk images that optionally
10963 implement a complete UEFI SecureBoot trust chain, involving a signed
10964 kernel and initrd image that incorporates such a root hash as well as
10965 a Verity-enabled root partition.
10967 * The hardware database (hwdb) udev supports has been updated to carry
10968 accelerometer quirks.
10970 * All system services are now run with a fresh kernel keyring set up
10971 for them. The invocation ID is stored by default in it, thus
10972 providing a safe, non-overridable way to determine the invocation
10973 ID of each service.
10975 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
10976 options for bind mounting arbitrary paths in a service-specific
10977 way. When these options are used, arbitrary host or service files and
10978 directories may be mounted to arbitrary locations in the service's
10981 * Documentation has been added that lists all of systemd's low-level
10982 environment variables:
10984 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
10986 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
10987 whether a specific socket file descriptor matches a specified socket
10990 * systemd-firstboot has been updated to check for the
10991 systemd.firstboot= kernel command line option. It accepts a boolean
10992 and when set to false the first boot questions are skipped.
10994 * systemd-fstab-generator has been updated to check for the
10995 systemd.volatile= kernel command line option, which either takes an
10996 optional boolean parameter or the special value "state". If used the
10997 system may be booted in a "volatile" boot mode. Specifically,
10998 "systemd.volatile" is used, the root directory will be mounted as
10999 tmpfs, and only /usr is mounted from the actual root file system. If
11000 "systemd.volatile=state" is used, the root directory will be mounted
11001 as usual, but /var is mounted as tmpfs. This concept provides similar
11002 functionality as systemd-nspawn's --volatile= option, but provides it
11003 on physical boots. Use this option for implementing stateless
11004 systems, or testing systems with all state and/or configuration reset
11005 to the defaults. (Note though that many distributions are not
11006 prepared to boot up without a populated /etc or /var, though.)
11008 * systemd-gpt-auto-generator gained support for LUKS encrypted root
11009 partitions. Previously it only supported LUKS encrypted partitions
11010 for all other uses, except for the root partition itself.
11012 * Socket units gained support for listening on AF_VSOCK sockets for
11013 communication in virtualized QEMU environments.
11015 * The "configure" script gained a new option --with-fallback-hostname=
11016 for specifying the fallback hostname to use if none is configured in
11017 /etc/hostname. For example, by specifying
11018 --with-fallback-hostname=fedora it is possible to default to a
11019 hostname of "fedora" on pristine installations.
11021 * systemd-cgls gained support for a new --unit= switch for listing only
11022 the control groups of a specific unit. Similar --user-unit= has been
11023 added for listing only the control groups of a specific user unit.
11025 * systemd-mount gained a new --umount switch for unmounting a mount or
11026 automount point (and all mount/automount points below it).
11028 * systemd will now refuse full configuration reloads (via systemctl
11029 daemon-reload and related calls) unless at least 16MiB of free space
11030 are available in /run. This is a safety precaution in order to ensure
11031 that generators can safely operate after the reload completed.
11033 * A new unit file option RootImage= has been added, which has a similar
11034 effect as RootDirectory= but mounts the service's root directory from
11035 a disk image instead of plain directory. This logic reuses the same
11036 image dissection and mount logic that systemd-nspawn already uses,
11037 and hence supports any disk images systemd-nspawn supports, including
11038 those following the Discoverable Partition Specification, as well as
11039 Verity enabled images. This option enables systemd to run system
11040 services directly off disk images acting as resource bundles,
11041 possibly even including full integrity data.
11043 * A new MountAPIVFS= unit file option has been added, taking a boolean
11044 argument. If enabled /proc, /sys and /dev (collectively called the
11045 "API VFS") will be mounted for the service. This is only relevant if
11046 RootDirectory= or RootImage= is used for the service, as these mounts
11047 are of course in place in the host mount namespace anyway.
11049 * systemd-nspawn gained support for a new --pivot-root= switch. If
11050 specified the root directory within the container image is pivoted to
11051 the specified mount point, while the original root disk is moved to a
11052 different place. This option enables booting of ostree images
11053 directly with systemd-nspawn.
11055 * The systemd build scripts will no longer complain if the NTP server
11056 addresses are not changed from the defaults. Google now supports
11057 these NTP servers officially. We still recommend downstreams to
11058 properly register an NTP pool with the NTP pool project though.
11060 * coredumpctl gained a new "--reverse" option for printing the list
11061 of coredumps in reverse order.
11063 * coredumpctl will now show additional information about truncated and
11064 inaccessible coredumps, as well as coredumps that are still being
11065 processed. It also gained a new --quiet switch for suppressing
11066 additional informational message in its output.
11068 * coredumpctl gained support for only showing coredumps newer and/or
11069 older than specific timestamps, using the new --since= and --until=
11070 options, reminiscent of journalctl's options by the same name.
11072 * The systemd-coredump logic has been improved so that it may be reused
11073 to collect backtraces in non-compiled languages, for example in
11074 scripting languages such as Python.
11076 * machinectl will now show the UID shift of local containers, if user
11077 namespacing is enabled for them.
11079 * systemd will now optionally run "environment generator" binaries at
11080 configuration load time. They may be used to add environment
11081 variables to the environment block passed to services invoked. One
11082 user environment generator is shipped by default that sets up
11083 environment variables based on files dropped into /etc/environment.d
11084 and ~/.config/environment.d/.
11086 * systemd-resolved now includes the new, recently published 2017 DNSSEC
11089 * hostnamed has been updated to report a new chassis type of
11090 "convertible" to cover "foldable" laptops that can both act as a
11091 tablet and as a laptop, such as various Lenovo Yoga devices.
11093 Contributions from: Adrián López, Alexander Galanin, Alexander
11094 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
11095 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
11096 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
11097 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
11098 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
11099 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
11100 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
11101 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
11102 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
11103 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
11104 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
11105 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
11106 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
11107 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
11108 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
11109 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
11110 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
11111 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
11112 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
11113 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
11114 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
11115 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
11116 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
11117 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
11118 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
11119 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
11120 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
11123 — Berlin, 2017-03-01
11127 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
11128 RestrictAddressFamilies= enabled. These sandboxing options should
11129 generally be compatible with the various external udev call-out
11130 binaries we are aware of, however there may be exceptions, in
11131 particular when exotic languages for these call-outs are used. In
11132 this case, consider turning off these settings locally.
11134 * The new RemoveIPC= option can be used to remove IPC objects owned by
11135 the user or group of a service when that service exits.
11137 * The new ProtectKernelModules= option can be used to disable explicit
11138 load and unload operations of kernel modules by a service. In
11139 addition access to /usr/lib/modules is removed if this option is set.
11141 * ProtectSystem= option gained a new value "strict", which causes the
11142 whole file system tree with the exception of /dev, /proc, and /sys,
11143 to be remounted read-only for a service.
11145 * The new ProtectKernelTunables= option can be used to disable
11146 modification of configuration files in /sys and /proc by a service.
11147 Various directories and files are remounted read-only, so access is
11148 restricted even if the file permissions would allow it.
11150 * The new ProtectControlGroups= option can be used to disable write
11151 access by a service to /sys/fs/cgroup.
11153 * Various systemd services have been hardened with
11154 ProtectKernelTunables=yes, ProtectControlGroups=yes,
11155 RestrictAddressFamilies=.
11157 * Support for dynamically creating users for the lifetime of a service
11158 has been added. If DynamicUser=yes is specified, user and group IDs
11159 will be allocated from the range 61184…65519 for the lifetime of the
11160 service. They can be resolved using the new nss-systemd.so NSS
11161 module. The module must be enabled in /etc/nsswitch.conf. Services
11162 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
11163 any resources allocated by the service will be cleaned up when the
11164 service exits. They also have ProtectHome=read-only and
11165 ProtectSystem=strict enabled, so they are not able to make any
11166 permanent modifications to the system.
11168 * The nss-systemd module also always resolves root and nobody, making
11169 it possible to have no /etc/passwd or /etc/group files in minimal
11170 container or chroot environments.
11172 * Services may be started with their own user namespace using the new
11173 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
11174 under which the service is running are mapped. All other users are
11177 * Support for the cgroup namespace has been added to systemd-nspawn. If
11178 supported by kernel, the container system started by systemd-nspawn
11179 will have its own view of the cgroup hierarchy. This new behaviour
11180 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
11182 * The new MemorySwapMax= option can be used to limit the maximum swap
11183 usage under the unified cgroup hierarchy.
11185 * Support for the CPU controller in the unified cgroup hierarchy has
11186 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
11187 options. This controller requires out-of-tree patches for the kernel
11188 and the support is provisional.
11190 * Mount and automount units may now be created transiently
11191 (i.e. dynamically at runtime via the bus API, instead of requiring
11192 unit files in the file system).
11194 * systemd-mount is a new tool which may mount file systems – much like
11195 mount(8), optionally pulling in additional dependencies through
11196 transient .mount and .automount units. For example, this tool
11197 automatically runs fsck on a backing block device before mounting,
11198 and allows the automount logic to be used dynamically from the
11199 command line for establishing mount points. This tool is particularly
11200 useful when dealing with removable media, as it will ensure fsck is
11201 run – if necessary – before the first access and that the file system
11202 is quickly unmounted after each access by utilizing the automount
11203 logic. This maximizes the chance that the file system on the
11204 removable media stays in a clean state, and if it isn't in a clean
11205 state is fixed automatically.
11207 * LazyUnmount=yes option for mount units has been added to expose the
11208 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
11211 * /efi will be used as the mount point of the EFI boot partition, if
11212 the directory is present, and the mount point was not configured
11213 through other means (e.g. fstab). If /efi directory does not exist,
11214 /boot will be used as before. This makes it easier to automatically
11215 mount the EFI partition on systems where /boot is used for something
11218 * When operating on GPT disk images for containers, systemd-nspawn will
11219 now mount the ESP to /boot or /efi according to the same rules as PID
11220 1 running on a host. This allows tools like "bootctl" to operate
11221 correctly within such containers, in order to make container images
11222 bootable on physical systems.
11224 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
11226 * Two new user session targets have been added to support running
11227 graphical sessions under the systemd --user instance:
11228 graphical-session.target and graphical-session-pre.target. See
11229 systemd.special(7) for a description of how those targets should be
11232 * The vconsole initialization code has been significantly reworked to
11233 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
11234 support unicode keymaps. Font and keymap configuration will now be
11235 copied to all allocated virtual consoles.
11237 * FreeBSD's bhyve virtualization is now detected.
11239 * Information recorded in the journal for core dumps now includes the
11240 contents of /proc/mountinfo and the command line of the process at
11241 the top of the process hierarchy (which is usually the init process
11244 * systemd-journal-gatewayd learned the --directory= option to serve
11245 files from the specified location.
11247 * journalctl --root=… can be used to peruse the journal in the
11248 /var/log/ directories inside of a container tree. This is similar to
11249 the existing --machine= option, but does not require the container to
11252 * The hardware database has been extended to support
11253 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
11256 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
11257 specify the click rate for mice which include a horizontal wheel with
11258 a click rate that is different than the one for the vertical wheel.
11260 * systemd-run gained a new --wait option that makes service execution
11261 synchronous. (Specifically, the command will not return until the
11262 specified service binary exited.)
11264 * systemctl gained a new --wait option that causes the start command to
11265 wait until the units being started have terminated again.
11267 * A new journal output mode "short-full" has been added which displays
11268 timestamps with abbreviated English day names and adds a timezone
11269 suffix. Those timestamps include more information than the default
11270 "short" output mode, and can be passed directly to journalctl's
11271 --since= and --until= options.
11273 * /etc/resolv.conf will be bind-mounted into containers started by
11274 systemd-nspawn, if possible, so any changes to resolv.conf contents
11275 are automatically propagated to the container.
11277 * The number of instances for socket-activated services originating
11278 from a single IP address can be limited with
11279 MaxConnectionsPerSource=, extending the existing setting of
11282 * systemd-networkd gained support for vcan ("Virtual CAN") interface
11285 * .netdev and .network configuration can now be extended through
11288 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
11289 Segmentation Offload, Generic Receive Offload, Large Receive Offload
11290 can be enabled and disabled using the new UDPSegmentationOffload=,
11291 TCPSegmentationOffload=, GenericSegmentationOffload=,
11292 GenericReceiveOffload=, LargeReceiveOffload= options in the
11293 [Link] section of .link files.
11295 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
11296 Port VLAN ID can be configured for bridge devices using the new STP=,
11297 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
11298 section of .netdev files.
11300 * The route table to which routes received over DHCP or RA should be
11301 added can be configured with the new RouteTable= option in the [DHCP]
11302 and [IPv6AcceptRA] sections of .network files.
11304 * The Address Resolution Protocol can be disabled on links managed by
11305 systemd-networkd using the ARP=no setting in the [Link] section of
11308 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
11309 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
11310 encode information about the result and exit codes of the current
11311 service runtime cycle.
11313 * systemd-sysctl will now configure kernel parameters in the order
11314 they occur in the configuration files. This matches what sysctl
11315 has been traditionally doing.
11317 * kernel-install "plugins" that are executed to perform various
11318 tasks after a new kernel is added and before an old one is removed
11319 can now return a special value to terminate the procedure and
11320 prevent any later plugins from running.
11322 * Journald's SplitMode=login setting has been deprecated. It has been
11323 removed from documentation, and its use is discouraged. In a future
11324 release it will be completely removed, and made equivalent to current
11325 default of SplitMode=uid.
11327 * Storage=both option setting in /etc/systemd/coredump.conf has been
11328 removed. With fast LZ4 compression storing the core dump twice is not
11331 * The --share-system systemd-nspawn option has been replaced with an
11332 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
11333 this functionality is discouraged. In addition the variables
11334 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
11335 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
11336 individual namespaces.
11338 * "machinectl list" now shows the IP address of running containers in
11339 the output, as well as OS release information.
11341 * "loginctl list" now shows the TTY of each session in the output.
11343 * sd-bus gained new API calls sd_bus_track_set_recursive(),
11344 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
11345 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
11346 tracking objects in a "recursive" mode, where a single client can be
11347 counted multiple times, if it takes multiple references.
11349 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
11350 sd_bus_get_exit_on_disconnect(). They may be used to make a
11351 process using sd-bus automatically exit if the bus connection is
11354 * Bus clients of the service manager may now "pin" loaded units into
11355 memory, by taking an explicit reference on them. This is useful to
11356 ensure the client can retrieve runtime data about the service even
11357 after the service completed execution. Taking such a reference is
11358 available only for privileged clients and should be helpful to watch
11359 running services in a race-free manner, and in particular collect
11360 information about exit statuses and results.
11362 * The nss-resolve module has been changed to strictly return UNAVAIL
11363 when communication via D-Bus with resolved failed, and NOTFOUND when
11364 a lookup completed but was negative. This means it is now possible to
11365 neatly configure fallbacks using nsswitch.conf result checking
11366 expressions. Taking benefit of this, the new recommended
11367 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
11369 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
11371 * A new setting CtrlAltDelBurstAction= has been added to
11372 /etc/systemd/system.conf which may be used to configure the precise
11373 behaviour if the user on the console presses Ctrl-Alt-Del more often
11374 than 7 times in 2s. Previously this would unconditionally result in
11375 an expedited, immediate reboot. With this new setting the precise
11376 operation may be configured in more detail, and also turned off
11379 * In .netdev files two new settings RemoteChecksumTx= and
11380 RemoteChecksumRx= are now understood that permit configuring the
11381 remote checksumming logic for VXLAN networks.
11383 * The service manager learnt a new "invocation ID" concept for invoked
11384 services. Each runtime cycle of a service will get a new invocation
11385 ID (a 128-bit random UUID) assigned that identifies the current
11386 run of the service uniquely and globally. A new invocation ID
11387 is generated each time a service starts up. The journal will store
11388 the invocation ID of a service along with any logged messages, thus
11389 making the invocation ID useful for matching the online runtime of a
11390 service with the offline log data it generated in a safe way without
11391 relying on synchronized timestamps. In many ways this new service
11392 invocation ID concept is similar to the kernel's boot ID concept that
11393 uniquely and globally identifies the runtime of each boot. The
11394 invocation ID of a service is passed to the service itself via an
11395 environment variable ($INVOCATION_ID). A new bus call
11396 GetUnitByInvocationID() has been added that is similar to GetUnit()
11397 but instead of retrieving the bus path for a unit by its name
11398 retrieves it by its invocation ID. The returned path is valid only as
11399 long as the passed invocation ID is current.
11401 * systemd-resolved gained a new "DNSStubListener" setting in
11402 resolved.conf. It either takes a boolean value or the special values
11403 "udp" and "tcp", and configures whether to enable the stub DNS
11404 listener on 127.0.0.53:53.
11406 * IP addresses configured via networkd may now carry additional
11407 configuration settings supported by the kernel. New options include:
11408 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
11409 PrefixRoute=, AutoJoin=.
11411 * The PAM configuration fragment file for "user@.service" shipped with
11412 systemd (i.e. the --user instance of systemd) has been stripped to
11413 the minimum necessary to make the system boot. Previously, it
11414 contained Fedora-specific stanzas that did not apply to other
11415 distributions. It is expected that downstream distributions add
11416 additional configuration lines, matching their needs to this file,
11417 using it only as rough template of what systemd itself needs. Note
11418 that this reduced fragment does not even include an invocation of
11419 pam_limits which most distributions probably want to add, even though
11420 systemd itself does not need it. (There's also the new build time
11421 option --with-pamconfdir=no to disable installation of the PAM
11422 fragment entirely.)
11424 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
11425 capability is now also dropped from its set (in addition to
11426 CAP_SYS_MKNOD as before).
11428 * In service unit files it is now possible to connect a specific named
11429 file descriptor with stdin/stdout/stdout of an executed service. The
11430 name may be specified in matching .socket units using the
11431 FileDescriptorName= setting.
11433 * A number of journal settings may now be configured on the kernel
11434 command line. Specifically, the following options are now understood:
11435 systemd.journald.max_level_console=,
11436 systemd.journald.max_level_store=,
11437 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
11438 systemd.journald.max_level_wall=.
11440 * "systemctl is-enabled --full" will now show by which symlinks a unit
11441 file is enabled in the unit dependency tree.
11443 * Support for VeraCrypt encrypted partitions has been added to the
11444 "cryptsetup" logic and /etc/crypttab.
11446 * systemd-detect-virt gained support for a new --private-users switch
11447 that checks whether the invoking processes are running inside a user
11448 namespace. Similar, a new special value "private-users" for the
11449 existing ConditionVirtualization= setting has been added, permitting
11450 skipping of specific units in user namespace environments.
11452 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
11453 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
11454 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
11455 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
11456 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
11457 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
11458 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
11459 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
11460 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
11461 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
11462 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
11463 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
11464 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
11465 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
11466 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
11467 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
11468 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
11469 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
11470 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
11471 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
11472 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
11473 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
11474 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
11475 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
11476 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
11477 Jędrzejewski-Szmek, Zeal Jagannatha
11479 — Santa Fe, 2016-11-03
11483 * In service units the various ExecXYZ= settings have been extended
11484 with an additional special character as first argument of the
11485 assigned value: if the character '+' is used the specified command
11486 line it will be run with full privileges, regardless of User=,
11487 Group=, CapabilityBoundingSet= and similar options. The effect is
11488 similar to the existing PermissionsStartOnly= option, but allows
11489 configuration of this concept for each executed command line
11492 * Services may now alter the service watchdog timeout at runtime by
11493 sending a WATCHDOG_USEC= message via sd_notify().
11495 * MemoryLimit= and related unit settings now optionally take percentage
11496 specifications. The percentage is taken relative to the amount of
11497 physical memory in the system (or in case of containers, the assigned
11498 amount of memory). This allows scaling service resources neatly with
11499 the amount of RAM available on the system. Similarly, systemd-logind's
11500 RuntimeDirectorySize= option now also optionally takes percentage
11503 * In similar fashion TasksMax= takes percentage values now, too. The
11504 value is taken relative to the configured maximum number of processes
11505 on the system. The per-service task maximum has been changed to 15%
11506 using this functionality. (Effectively this is an increase of 512 →
11507 4915 for service units, given the kernel's default pid_max setting.)
11509 * Calendar time specifications in .timer units now understand a ".."
11510 syntax for time ranges. Example: "4..7:10" may now be used for
11511 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
11514 * The InaccessableDirectories=, ReadOnlyDirectories= and
11515 ReadWriteDirectories= unit file settings have been renamed to
11516 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
11517 applied to all kinds of file nodes, and not just directories, with
11518 the exception of symlinks. Specifically these settings may now be
11519 used on block and character device nodes, UNIX sockets and FIFOS as
11520 well as regular files. The old names of these settings remain
11521 available for compatibility.
11523 * systemd will now log about all service processes it kills forcibly
11524 (using SIGKILL) because they remained after the clean shutdown phase
11525 of the service completed. This should help identifying services that
11526 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
11527 systemd-logind's configuration a similar log message is generated for
11528 processes killed at the end of each session due to this setting.
11530 * systemd will now set the $JOURNAL_STREAM environment variable for all
11531 services whose stdout/stderr are connected to the Journal (which
11532 effectively means by default: all services). The variable contains
11533 the device and inode number of the file descriptor used for
11534 stdout/stderr. This may be used by invoked programs to detect whether
11535 their stdout/stderr is connected to the Journal, in which case they
11536 can switch over to direct Journal communication, thus being able to
11537 pass extended, structured metadata along with their log messages. As
11538 one example, this is now used by glib's logging primitives.
11540 * When using systemd's default tmp.mount unit for /tmp, the mount point
11541 will now be established with the "nosuid" and "nodev" options. This
11542 avoids privilege escalation attacks that put traps and exploits into
11543 /tmp. However, this might cause problems if you e.g. put container
11544 images or overlays into /tmp; if you need this, override tmp.mount's
11545 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
11548 * systemd now supports the "memory" cgroup controller also on
11551 * The systemd-cgtop tool now optionally takes a control group path as
11552 command line argument. If specified, the control group list shown is
11553 limited to subgroups of that group.
11555 * The SystemCallFilter= unit file setting gained support for
11556 pre-defined, named system call filter sets. For example
11557 SystemCallFilter=@clock is now an effective way to make all clock
11558 changing-related system calls unavailable to a service. A number of
11559 similar pre-defined groups are defined. Writing system call filters
11560 for system services is simplified substantially with this new
11561 concept. Accordingly, all of systemd's own, long-running services now
11562 enable system call filtering based on this, by default.
11564 * A new service setting MemoryDenyWriteExecute= has been added, taking
11565 a boolean value. If turned on, a service may no longer create memory
11566 mappings that are writable and executable at the same time. This
11567 enhances security for services where this is enabled as it becomes
11568 harder to dynamically write and then execute memory in exploited
11569 service processes. This option has been enabled for all of systemd's
11570 own long-running services.
11572 * A new RestrictRealtime= service setting has been added, taking a
11573 boolean argument. If set the service's processes may no longer
11574 acquire realtime scheduling. This improves security as realtime
11575 scheduling may otherwise be used to easily freeze the system.
11577 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
11578 value. This may be used for requesting that the system manager inside
11579 of the container reports start-up completion to nspawn which then
11580 propagates this notification further to the service manager
11581 supervising nspawn itself. A related option NotifyReady= in .nspawn
11582 files has been added too. This functionality allows ordering of the
11583 start-up of multiple containers using the usual systemd ordering
11586 * machinectl gained a new command "stop" that is an alias for
11589 * systemd-resolved gained support for contacting DNS servers on
11590 link-local IPv6 addresses.
11592 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
11593 its caches. A method call for requesting the same operation has been
11594 added to the bus API too, and is made available via "systemd-resolve
11597 * systemd-resolve gained a new --status switch. If passed a brief
11598 summary of the used DNS configuration with per-interface information
11601 * resolved.conf gained a new Cache= boolean option, defaulting to
11602 on. If turned off local DNS caching is disabled. This comes with a
11603 performance penalty in particular when DNSSEC is enabled. Note that
11604 resolved disables its internal caching implicitly anyway, when the
11605 configured DNS server is on a host-local IP address such as ::1 or
11606 127.0.0.1, thus automatically avoiding double local caching.
11608 * systemd-resolved now listens on the local IP address 127.0.0.53:53
11609 for DNS requests. This improves compatibility with local programs
11610 that do not use the libc NSS or systemd-resolved's bus APIs for name
11611 resolution. This minimal DNS service is only available to local
11612 programs and does not implement the full DNS protocol, but enough to
11613 cover local DNS clients. A new, static resolv.conf file, listing just
11614 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
11615 now recommended to make /etc/resolv.conf a symlink to this file in
11616 order to route all DNS lookups to systemd-resolved, regardless if
11617 done via NSS, the bus API or raw DNS packets. Note that this local
11618 DNS service is not as fully featured as the libc NSS or
11619 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
11620 used to deliver link-local address information (as this implies
11621 sending a local interface index along), LLMNR/mDNS support via this
11622 interface is severely restricted. It is thus strongly recommended for
11623 all applications to use the libc NSS API or native systemd-resolved
11626 * systemd-networkd's bridge support learned a new setting
11627 VLANFiltering= for controlling VLAN filtering. Moreover a new section
11628 in .network files has been added for configuring VLAN bridging in
11629 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
11631 * systemd-networkd's IPv6 Router Advertisement code now makes use of
11632 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
11633 now be acquired without relying on DHCPv6. Two new options
11634 UseDomains= and UseDNS= have been added to configure this behaviour.
11636 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
11637 renamed IPv6AcceptRA=, without altering its behaviour. The old
11638 setting name remains available for compatibility reasons.
11640 * The systemd-networkd VTI/VTI6 tunneling support gained new options
11641 Key=, InputKey= and OutputKey=.
11643 * systemd-networkd gained support for VRF ("Virtual Routing Function")
11644 interface configuration.
11646 * "systemctl edit" may now be used to create new unit files by
11647 specifying the --force switch.
11649 * sd-event gained a new function sd_event_get_iteration() for
11650 requesting the current iteration counter of the event loop. It starts
11651 at zero and is increased by one with each event loop iteration.
11653 * A new rpm macro %systemd_ordering is provided by the macros.systemd
11654 file. It can be used in lieu of %systemd_requires in packages which
11655 don't use any systemd functionality and are intended to be installed
11656 in minimal containers without systemd present. This macro provides
11657 ordering dependencies to ensure that if the package is installed in
11658 the same rpm transaction as systemd, systemd will be installed before
11659 the scriptlets for the package are executed, allowing unit presets
11662 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
11663 been added to simplify packaging of generators.
11665 * The os-release file gained VERSION_CODENAME field for the
11666 distribution nickname (e.g. VERSION_CODENAME=woody).
11668 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
11669 can be set to disable parsing of metadata and the creation
11670 of persistent symlinks for that device.
11672 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
11673 to make them available to logged-in users has been reverted.
11675 * Much of the common code of the various systemd components is now
11676 built into an internal shared library libsystemd-shared-231.so
11677 (incorporating the systemd version number in the name, to be updated
11678 with future releases) that the components link to. This should
11679 decrease systemd footprint both in memory during runtime and on
11680 disk. Note that the shared library is not for public use, and is
11681 neither API nor ABI stable, but is likely to change with every new
11682 released update. Packagers need to make sure that binaries
11683 linking to libsystemd-shared.so are updated in step with the
11686 * Configuration for "mkosi" is now part of the systemd
11687 repository. mkosi is a tool to easily build legacy-free OS images,
11688 and is available on github: https://github.com/systemd/mkosi. If
11689 "mkosi" is invoked in the build tree a new raw OS image is generated
11690 incorporating the systemd sources currently being worked on and a
11691 clean, fresh distribution installation. The generated OS image may be
11692 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
11693 UEFI PC. This functionality is particularly useful to easily test
11694 local changes made to systemd in a pristine, defined environment. See
11695 doc/HACKING for details.
11697 * configure learned the --with-support-url= option to specify the
11698 distribution's bugtracker.
11700 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
11701 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
11702 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
11703 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
11704 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
11705 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
11706 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
11707 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
11708 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
11709 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
11710 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
11711 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
11712 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
11713 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
11714 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
11715 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
11716 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
11717 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
11718 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
11720 — Berlin, 2016-07-25
11724 * DNSSEC is now turned on by default in systemd-resolved (in
11725 "allow-downgrade" mode), but may be turned off during compile time by
11726 passing "--with-default-dnssec=no" to "configure" (and of course,
11727 during runtime with DNSSEC= in resolved.conf). We recommend
11728 downstreams to leave this on at least during development cycles and
11729 report any issues with the DNSSEC logic upstream. We are very
11730 interested in collecting feedback about the DNSSEC validator and its
11731 limitations in the wild. Note however, that DNSSEC support is
11732 probably nothing downstreams should turn on in stable distros just
11733 yet, as it might create incompatibilities with a few DNS servers and
11734 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
11735 automatically whenever we detect such incompatible setups, but there
11736 might be systems we do not cover yet. Hence: please help us testing
11737 the DNSSEC code, leave this on where you can, report back, but then
11738 again don't consider turning this on in your stable, LTS or
11739 production release just yet. (Note that you have to enable
11740 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
11741 and its DNSSEC mode for hostname resolution from local
11744 * systemd-resolve conveniently resolves DANE records with the --tlsa
11745 option and OPENPGPKEY records with the --openpgp option. It also
11746 supports dumping raw DNS record data via the new --raw= switch.
11748 * systemd-logind will now by default terminate user processes that are
11749 part of the user session scope unit (session-XX.scope) when the user
11750 logs out. This behavior is controlled by the KillUserProcesses=
11751 setting in logind.conf, and the previous default of "no" is now
11752 changed to "yes". This means that user sessions will be properly
11753 cleaned up after, but additional steps are necessary to allow
11754 intentionally long-running processes to survive logout.
11756 While the user is logged in at least once, user@.service is running,
11757 and any service that should survive the end of any individual login
11758 session can be started at a user service or scope using systemd-run.
11759 systemd-run(1) man page has been extended with an example which shows
11760 how to run screen in a scope unit underneath user@.service. The same
11761 command works for tmux.
11763 After the user logs out of all sessions, user@.service will be
11764 terminated too, by default, unless the user has "lingering" enabled.
11765 To effectively allow users to run long-term tasks even if they are
11766 logged out, lingering must be enabled for them. See loginctl(1) for
11767 details. The default polkit policy was modified to allow users to
11768 set lingering for themselves without authentication.
11770 Previous defaults can be restored at compile time by the
11771 --without-kill-user-processes option to "configure".
11773 * systemd-logind gained new configuration settings SessionsMax= and
11774 InhibitorsMax=, both with a default of 8192. It will not register new
11775 user sessions or inhibitors above this limit.
11777 * systemd-logind will now reload configuration on SIGHUP.
11779 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
11780 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
11781 enable. Also, support for the "io" cgroup controller in the unified
11782 hierarchy has been added, so that the "memory", "pids" and "io" are
11783 now the controllers that are supported on the unified hierarchy.
11785 WARNING: it is not possible to use previous systemd versions with
11786 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
11787 is necessary to also update systemd in the initramfs if using the
11788 unified hierarchy. An updated SELinux policy is also required.
11790 * LLDP support has been extended, and both passive (receive-only) and
11791 active (sender) modes are supported. Passive mode ("routers-only") is
11792 enabled by default in systemd-networkd. Active LLDP mode is enabled
11793 by default for containers on the internal network. The "networkctl
11794 lldp" command may be used to list information gathered. "networkctl
11795 status" will also show basic LLDP information on connected peers now.
11797 * The IAID and DUID unique identifier sent in DHCP requests may now be
11798 configured for the system and each .network file managed by
11799 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
11801 * systemd-networkd gained support for configuring proxy ARP support for
11802 each interface, via the ProxyArp= setting in .network files. It also
11803 gained support for configuring the multicast querier feature of
11804 bridge devices, via the new MulticastQuerier= setting in .netdev
11805 files. Similarly, snooping on the IGMP traffic can be controlled
11806 via the new setting MulticastSnooping=.
11808 A new setting PreferredLifetime= has been added for addresses
11809 configured in .network file to configure the lifetime intended for an
11812 The systemd-networkd DHCP server gained the option EmitRouter=, which
11813 defaults to yes, to configure whether the DHCP Option 3 (Router)
11816 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
11817 systemd-socket-activate and installed into /usr/bin. It is now fully
11820 * systemd-journald now uses separate threads to flush changes to disk
11821 when closing journal files, thus reducing impact of slow disk I/O on
11822 logging performance.
11824 * The sd-journal API gained two new calls
11825 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
11826 can be used to open journal files using file descriptors instead of
11827 file or directory paths. sd_journal_open_container() has been
11828 deprecated, sd_journal_open_directory_fd() should be used instead
11829 with the flag SD_JOURNAL_OS_ROOT.
11831 * journalctl learned a new output mode "-o short-unix" that outputs log
11832 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
11833 UTC). It also gained support for a new --no-hostname setting to
11834 suppress the hostname column in the family of "short" output modes.
11836 * systemd-ask-password now optionally skips printing of the password to
11837 stdout with --no-output which can be useful in scripts.
11839 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
11840 (devices tagged with ID_MAKER_TOOL) are now tagged with
11841 "uaccess" and are available to logged in users.
11843 * The DeviceAllow= unit setting now supports specifiers (with "%").
11845 * "systemctl show" gained a new --value switch, which allows print a
11846 only the contents of a specific unit property, without also printing
11847 the property's name. Similar support was added to "show*" verbs
11848 of loginctl and machinectl that output "key=value" lists.
11850 * A new unit type "generated" was added for files dynamically generated
11851 by generator tools. Similarly, a new unit type "transient" is used
11852 for unit files created using the runtime API. "systemctl enable" will
11853 refuse to operate on such files.
11855 * A new command "systemctl revert" has been added that may be used to
11856 revert to the vendor version of a unit file, in case local changes
11857 have been made by adding drop-ins or overriding the unit file.
11859 * "machinectl clean" gained a new verb to automatically remove all or
11860 just hidden container images.
11862 * systemd-tmpfiles gained support for a new line type "e" for emptying
11863 directories, if they exist, without creating them if they don't.
11865 * systemd-nspawn gained support for automatically patching the UID/GIDs
11866 of the owners and the ACLs of all files and directories in a
11867 container tree to match the UID/GID user namespacing range selected
11868 for the container invocation. This mode is enabled via the new
11869 --private-users-chown switch. It also gained support for
11870 automatically choosing a free, previously unused UID/GID range when
11871 starting a container, via the new --private-users=pick setting (which
11872 implies --private-users-chown). Together, these options for the first
11873 time make user namespacing for nspawn containers fully automatic and
11874 thus deployable. The systemd-nspawn@.service template unit file has
11875 been changed to use this functionality by default.
11877 * systemd-nspawn gained a new --network-zone= switch, that allows
11878 creating ad-hoc virtual Ethernet links between multiple containers,
11879 that only exist as long as at least one container referencing them is
11880 running. This allows easy connecting of multiple containers with a
11881 common link that implements an Ethernet broadcast domain. Each of
11882 these network "zones" may be named relatively freely by the user, and
11883 may be referenced by any number of containers, but each container may
11884 only reference one of these "zones". On the lower level, this is
11885 implemented by an automatically managed bridge network interface for
11886 each zone, that is created when the first container referencing its
11887 zone is created and removed when the last one referencing its zone
11890 * The default start timeout may now be configured on the kernel command
11891 line via systemd.default_timeout_start_sec=. It was already
11892 configurable via the DefaultTimeoutStartSec= option in
11893 /etc/systemd/system.conf.
11895 * Socket units gained a new TriggerLimitIntervalSec= and
11896 TriggerLimitBurst= setting to configure a limit on the activation
11897 rate of the socket unit.
11899 * The LimitNICE= setting now optionally takes normal UNIX nice values
11900 in addition to the raw integer limit value. If the specified
11901 parameter is prefixed with "+" or "-" and is in the range -20…19 the
11902 value is understood as UNIX nice value. If not prefixed like this it
11903 is understood as raw RLIMIT_NICE limit.
11905 * Note that the effect of the PrivateDevices= unit file setting changed
11906 slightly with this release: the per-device /dev file system will be
11907 mounted read-only from this version on, and will have "noexec"
11908 set. This (minor) change of behavior might cause some (exceptional)
11909 legacy software to break, when PrivateDevices=yes is set for its
11910 service. Please leave PrivateDevices= off if you run into problems
11913 * systemd-bootchart has been split out to a separate repository:
11914 https://github.com/systemd/systemd-bootchart
11916 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
11917 merged into the kernel in its current form.
11919 * The compatibility libraries libsystemd-daemon.so,
11920 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
11921 which have been deprecated since systemd-209 have been removed along
11922 with the corresponding pkg-config files. All symbols provided by
11923 those libraries are provided by libsystemd.so.
11925 * The Capabilities= unit file setting has been removed (it is ignored
11926 for backwards compatibility). AmbientCapabilities= and
11927 CapabilityBoundingSet= should be used instead.
11929 * A new special target has been added, initrd-root-device.target,
11930 which creates a synchronization point for dependencies of the root
11931 device in early userspace. Initramfs builders must ensure that this
11932 target is now included in early userspace.
11934 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
11935 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
11936 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
11937 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
11938 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
11939 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
11940 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
11941 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
11942 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
11943 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
11944 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
11945 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
11946 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
11947 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
11948 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
11949 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
11950 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
11951 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
11952 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
11953 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
11954 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
11955 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
11956 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
11957 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
11958 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
11961 — Fairfax, 2016-05-21
11965 * The systemd-resolved DNS resolver service has gained a substantial
11966 set of new features, most prominently it may now act as a DNSSEC
11967 validating stub resolver. DNSSEC mode is currently turned off by
11968 default, but is expected to be turned on by default in one of the
11969 next releases. For now, we invite everybody to test the DNSSEC logic
11970 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
11971 service also gained a full set of D-Bus interfaces, including calls
11972 to configure DNS and DNSSEC settings per link (for use by external
11973 network management software). systemd-resolved and systemd-networkd
11974 now distinguish between "search" and "routing" domains. The former
11975 are used to qualify single-label names, the latter are used purely
11976 for routing lookups within certain domains to specific links.
11977 resolved now also synthesizes RRs for all entries from /etc/hosts.
11979 * The systemd-resolve tool (which is a client utility for
11980 systemd-resolved) has been improved considerably and is now fully
11981 supported and documented. Hence it has moved from /usr/lib/systemd to
11984 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
11987 * The coredump collection logic has been reworked: when a coredump is
11988 collected it is now written to disk, compressed and processed
11989 (including stacktrace extraction) from a new instantiated service
11990 systemd-coredump@.service, instead of directly from the
11991 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
11992 processing large coredumps can take up a substantial amount of
11993 resources and time, and this previously happened entirely outside of
11994 systemd's service supervision. With the new logic the core_pattern
11995 hook only does minimal metadata collection before passing off control
11996 to the new instantiated service, which is configured with a time
11997 limit, a nice level and other settings to minimize negative impact on
11998 the rest of the system. Also note that the new logic will honour the
11999 RLIMIT_CORE setting of the crashed process, which now allows users
12000 and processes to turn off coredumping for their processes by setting
12003 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
12004 and all forked processes by default. Previously, PID 1 would leave
12005 the setting at "0" for all processes, as set by the kernel. Note that
12006 the resource limit traditionally has no effect on the generated
12007 coredumps on the system if the /proc/sys/kernel/core_pattern hook
12008 logic is used. Since the limit is now honoured (see above) its
12009 default has been changed so that the coredumping logic is enabled by
12010 default for all processes, while allowing specific opt-out.
12012 * When the stacktrace is extracted from processes of system users, this
12013 is now done as "systemd-coredump" user, in order to sandbox this
12014 potentially security sensitive parsing operation. (Note that when
12015 processing coredumps of normal users this is done under the user ID
12016 of process that crashed, as before.) Packagers should take notice
12017 that it is now necessary to create the "systemd-coredump" system user
12018 and group at package installation time.
12020 * The systemd-activate socket activation testing tool gained support
12021 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
12022 and --seqpacket switches. It also has been extended to support both
12023 new-style and inetd-style file descriptor passing. Use the new
12024 --inetd switch to request inetd-style file descriptor passing.
12026 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
12027 variable, which takes a boolean value. If set to false, ANSI color
12028 output is disabled in the tools even when run on a terminal that
12031 * The VXLAN support in networkd now supports two new settings
12032 DestinationPort= and PortRange=.
12034 * A new systemd.machine_id= kernel command line switch has been added,
12035 that may be used to set the machine ID in /etc/machine-id if it is
12036 not initialized yet. This command line option has no effect if the
12037 file is already initialized.
12039 * systemd-nspawn gained a new --as-pid2 switch that invokes any
12040 specified command line as PID 2 rather than PID 1 in the
12041 container. In this mode PID 1 is a minimal stub init process that
12042 implements the special POSIX and Linux semantics of PID 1 regarding
12043 signal and child process management. Note that this stub init process
12044 is implemented in nspawn itself and requires no support from the
12045 container image. This new logic is useful to support running
12046 arbitrary commands in the container, as normal processes are
12047 generally not prepared to run as PID 1.
12049 * systemd-nspawn gained a new --chdir= switch for setting the current
12050 working directory for the process started in the container.
12052 * "journalctl /dev/sda" will now output all kernel log messages for
12053 specified device from the current boot, in addition to all devices
12054 that are parents of it. This should make log output about devices
12055 pretty useful, as long as kernel drivers attach enough metadata to
12056 the log messages. (The usual SATA drivers do.)
12058 * The sd-journal API gained two new calls
12059 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
12060 that report whether log data from /run or /var has been found.
12062 * journalctl gained a new switch "--fields" that prints all journal
12063 record field names currently in use in the journal. This is backed
12064 by two new sd-journal API calls sd_journal_enumerate_fields() and
12065 sd_journal_restart_fields().
12067 * Most configurable timeouts in systemd now expect an argument of
12068 "infinity" to turn them off, instead of "0" as before. The semantics
12069 from now on is that a timeout of "0" means "now", and "infinity"
12070 means "never". To maintain backwards compatibility, "0" continues to
12071 turn off previously existing timeout settings.
12073 * "systemctl reload-or-try-restart" has been renamed to "systemctl
12074 try-reload-or-restart" to clarify what it actually does: the "try"
12075 logic applies to both reloading and restarting, not just restarting.
12076 The old name continues to be accepted for compatibility.
12078 * On boot-up, when PID 1 detects that the system clock is behind the
12079 release date of the systemd version in use, the clock is now set
12080 to the latter. Previously, this was already done in timesyncd, in order
12081 to avoid running with clocks set to the various clock epochs such as
12082 1902, 1938 or 1970. With this change the logic is now done in PID 1
12083 in addition to timesyncd during early boot-up, so that it is enforced
12084 before the first process is spawned by systemd. Note that the logic
12085 in timesyncd remains, as it is more comprehensive and ensures
12086 clock monotonicity by maintaining a persistent timestamp file in
12087 /var. Since /var is generally not available in earliest boot or the
12088 initrd, this part of the logic remains in timesyncd, and is not done
12091 * Support for tweaking details in net_cls.class_id through the
12092 NetClass= configuration directive has been removed, as the kernel
12093 people have decided to deprecate that controller in cgroup v2.
12094 Userspace tools such as nftables are moving over to setting rules
12095 that are specific to the full cgroup path of a task, which obsoletes
12096 these controllers anyway. The NetClass= directive is kept around for
12097 legacy compatibility reasons. For a more in-depth description of the
12098 kernel change, please refer to the respective upstream commit:
12100 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
12102 * A new service setting RuntimeMaxSec= has been added that may be used
12103 to specify a maximum runtime for a service. If the timeout is hit, the
12104 service is terminated and put into a failure state.
12106 * A new service setting AmbientCapabilities= has been added. It allows
12107 configuration of additional Linux process capabilities that are
12108 passed to the activated processes. This is only available on very
12111 * The process resource limit settings in service units may now be used
12112 to configure hard and soft limits individually.
12114 * The various libsystemd APIs such as sd-bus or sd-event now publicly
12115 expose support for gcc's __attribute__((cleanup())) C extension.
12116 Specifically, for many object destructor functions alternative
12117 versions have been added that have names suffixed with "p" and take a
12118 pointer to a pointer to the object to destroy, instead of just a
12119 pointer to the object itself. This is useful because these destructor
12120 functions may be used directly as parameters to the cleanup
12121 construct. Internally, systemd has been a heavy user of this GCC
12122 extension for a long time, and with this change similar support is
12123 now available to consumers of the library outside of systemd. Note
12124 that by using this extension in your sources compatibility with old
12125 and strictly ANSI compatible C compilers is lost. However, all gcc or
12126 LLVM versions of recent years support this extension.
12128 * Timer units gained support for a new setting RandomizedDelaySec= that
12129 allows configuring some additional randomized delay to the configured
12130 time. This is useful to spread out timer events to avoid load peaks in
12131 clusters or larger setups.
12133 * Calendar time specifications now support sub-second accuracy.
12135 * Socket units now support listening on SCTP and UDP-lite protocol
12138 * The sd-event API now comes with a full set of man pages.
12140 * Older versions of systemd contained experimental support for
12141 compressing journal files and coredumps with the LZ4 compressor that
12142 was not compatible with the lz4 binary (due to API limitations of the
12143 lz4 library). This support has been removed; only support for files
12144 compatible with the lz4 binary remains. This LZ4 logic is now
12145 officially supported and no longer considered experimental.
12147 * The dkr image import logic has been removed again from importd. dkr's
12148 micro-services focus doesn't fit into the machine image focus of
12149 importd, and quickly got out of date with the upstream dkr API.
12151 * Creation of the /run/lock/lockdev/ directory was dropped from
12152 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
12153 been available for many years. If you still need this, you need to
12154 create your own tmpfiles.d config file with:
12156 d /run/lock/lockdev 0775 root lock -
12158 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
12159 and RebootArgument= have been moved from the [Service] section of
12160 unit files to [Unit], and they are now supported on all unit types,
12161 not just service units. Of course, systemd will continue to
12162 understand these settings also at the old location, in order to
12163 maintain compatibility.
12165 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
12166 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
12167 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
12168 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
12169 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
12170 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
12171 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
12172 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
12173 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
12174 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
12175 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
12176 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
12177 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
12178 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
12179 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
12180 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
12181 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
12182 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
12183 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
12185 — Berlin, 2016-02-11
12189 * A number of properties previously only settable in unit
12190 files are now also available as properties to set when
12191 creating transient units programmatically via the bus, as it
12192 is exposed with systemd-run's --property=
12193 setting. Specifically, these are: SyslogIdentifier=,
12194 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
12195 EnvironmentFile=, ReadWriteDirectories=,
12196 ReadOnlyDirectories=, InaccessibleDirectories=,
12197 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
12199 * When creating transient services via the bus API it is now
12200 possible to pass in a set of file descriptors to use as
12201 STDIN/STDOUT/STDERR for the invoked process.
12203 * Slice units may now be created transiently via the bus APIs,
12204 similar to the way service and scope units may already be
12205 created transiently.
12207 * Wherever systemd expects a calendar timestamp specification
12208 (like in journalctl's --since= and --until= switches) UTC
12209 timestamps are now supported. Timestamps suffixed with "UTC"
12210 are now considered to be in Universal Time Coordinated
12211 instead of the local timezone. Also, timestamps may now
12212 optionally be specified with sub-second accuracy. Both of
12213 these additions also apply to recurring calendar event
12214 specification, such as OnCalendar= in timer units.
12216 * journalctl gained a new "--sync" switch that asks the
12217 journal daemon to write all so far unwritten log messages to
12218 disk and sync the files, before returning.
12220 * systemd-tmpfiles learned two new line types "q" and "Q" that
12221 operate like "v", but also set up a basic btrfs quota
12222 hierarchy when used on a btrfs file system with quota
12225 * tmpfiles' "v", "q" and "Q" will now create a plain directory
12226 instead of a subvolume (even on a btrfs file system) if the
12227 root directory is a plain directory, and not a
12228 subvolume. This should simplify things with certain chroot()
12229 environments which are not aware of the concept of btrfs
12232 * systemd-detect-virt gained a new --chroot switch to detect
12233 whether execution takes place in a chroot() environment.
12235 * CPUAffinity= now takes CPU index ranges in addition to
12236 individual indexes.
12238 * The various memory-related resource limit settings (such as
12239 LimitAS=) now understand the usual K, M, G, … suffixes to
12240 the base of 1024 (IEC). Similar, the time-related resource
12241 limit settings understand the usual min, h, day, … suffixes
12244 * There's a new system.conf setting DefaultTasksMax= to
12245 control the default TasksMax= setting for services and
12246 scopes running on the system. (TasksMax= is the primary
12247 setting that exposes the "pids" cgroup controller on systemd
12248 and was introduced in the previous systemd release.) The
12249 setting now defaults to 512, which means services that are
12250 not explicitly configured otherwise will only be able to
12251 create 512 processes or threads at maximum, from this
12252 version on. Note that this means that thread- or
12253 process-heavy services might need to be reconfigured to set
12254 TasksMax= to a higher value. It is sufficient to set
12255 TasksMax= in these specific unit files to a higher value, or
12256 even "infinity". Similar, there's now a logind.conf setting
12257 UserTasksMax= that defaults to 4096 and limits the total
12258 number of processes or tasks each user may own
12259 concurrently. nspawn containers also have the TasksMax=
12260 value set by default now, to 8192. Note that all of this
12261 only has an effect if the "pids" cgroup controller is
12262 enabled in the kernel. The general benefit of these changes
12263 should be a more robust and safer system, that provides a
12264 certain amount of per-service fork() bomb protection.
12266 * systemd-nspawn gained the new --network-veth-extra= switch
12267 to define additional and arbitrarily-named virtual Ethernet
12268 links between the host and the container.
12270 * A new service execution setting PassEnvironment= has been
12271 added that allows importing select environment variables
12272 from PID1's environment block into the environment block of
12275 * Timer units gained support for a new RemainAfterElapse=
12276 setting which takes a boolean argument. It defaults to on,
12277 exposing behaviour unchanged to previous releases. If set to
12278 off, timer units are unloaded after they elapsed if they
12279 cannot elapse again. This is particularly useful for
12280 transient timer units, which shall not stay around longer
12281 than until they first elapse.
12283 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
12284 default now (the kernel default is 16). This is beneficial
12285 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
12286 allows substantially larger numbers of queued
12287 datagrams. This should increase the capability of systemd to
12288 parallelize boot-up, as logging and sd_notify() are unlikely
12289 to stall execution anymore. If you need to change the value
12290 from the new defaults, use the usual sysctl.d/ snippets.
12292 * The compression framing format used by the journal or
12293 coredump processing has changed to be in line with what the
12294 official LZ4 tools generate. LZ4 compression support in
12295 systemd was considered unsupported previously, as the format
12296 was not compatible with the normal tools. With this release
12297 this has changed now, and it is hence safe for downstream
12298 distributions to turn it on. While not compressing as well
12299 as the XZ, LZ4 is substantially faster, which makes
12300 it a good default choice for the compression logic in the
12301 journal and in coredump handling.
12303 * Any reference to /etc/mtab has been dropped from
12304 systemd. The file has been obsolete since a while, but
12305 systemd refused to work on systems where it was incorrectly
12306 set up (it should be a symlink or non-existent). Please make
12307 sure to update to util-linux 2.27.1 or newer in conjunction
12308 with this systemd release, which also drops any reference to
12309 /etc/mtab. If you maintain a distribution make sure that no
12310 software you package still references it, as this is a
12311 likely source of bugs. There's also a glibc bug pending,
12312 asking for removal of any reference to this obsolete file:
12314 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
12316 Note that only util-linux versions built with
12317 --enable-libmount-force-mountinfo are supported.
12319 * Support for the ".snapshot" unit type has been removed. This
12320 feature turned out to be little useful and little used, and
12321 has now been removed from the core and from systemctl.
12323 * The dependency types RequiresOverridable= and
12324 RequisiteOverridable= have been removed from systemd. They
12325 have been used only very sparingly to our knowledge and
12326 other options that provide a similar effect (such as
12327 systemctl --mode=ignore-dependencies) are much more useful
12328 and commonly used. Moreover, they were only half-way
12329 implemented as the option to control behaviour regarding
12330 these dependencies was never added to systemctl. By removing
12331 these dependency types the execution engine becomes a bit
12332 simpler. Unit files that use these dependencies should be
12333 changed to use the non-Overridable dependency types
12334 instead. In fact, when parsing unit files with these
12335 options, that's what systemd will automatically convert them
12336 too, but it will also warn, asking users to fix the unit
12337 files accordingly. Removal of these dependency types should
12338 only affect a negligible number of unit files in the wild.
12340 * Behaviour of networkd's IPForward= option changed
12341 (again). It will no longer maintain a per-interface setting,
12342 but propagate one way from interfaces where this is enabled
12343 to the global kernel setting. The global setting will be
12344 enabled when requested by a network that is set up, but
12345 never be disabled again. This change was made to make sure
12346 IPv4 and IPv6 behaviour regarding packet forwarding is
12347 similar (as the Linux IPv6 stack does not support
12348 per-interface control of this setting) and to minimize
12351 * In unit files the behaviour of %u, %U, %h, %s has
12352 changed. These specifiers will now unconditionally resolve
12353 to the various user database fields of the user that the
12354 systemd instance is running as, instead of the user
12355 configured in the specific unit via User=. Note that this
12356 effectively doesn't change much, as resolving of these
12357 specifiers was already turned off in the --system instance
12358 of systemd, as we cannot do NSS lookups from PID 1. In the
12359 --user instance of systemd these specifiers where correctly
12360 resolved, but hardly made any sense, since the user instance
12361 lacks privileges to do user switches anyway, and User= is
12362 hence useless. Moreover, even in the --user instance of
12363 systemd behaviour was awkward as it would only take settings
12364 from User= assignment placed before the specifier into
12365 account. In order to unify and simplify the logic around
12366 this the specifiers will now always resolve to the
12367 credentials of the user invoking the manager (which in case
12368 of PID 1 is the root user).
12370 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
12371 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
12372 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
12373 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
12374 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
12375 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
12376 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
12377 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
12378 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
12379 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
12380 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
12381 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
12382 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
12383 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
12386 — Berlin, 2015-11-18
12390 * systemd now depends on util-linux v2.27. More specifically,
12391 the newly added mount monitor feature in libmount now
12392 replaces systemd's former own implementation.
12394 * libmount mandates /etc/mtab not to be regular file, and
12395 systemd now enforces this condition at early boot.
12396 /etc/mtab has been deprecated and warned about for a very
12397 long time, so systems running systemd should already have
12398 stopped having this file around as anything else than a
12399 symlink to /proc/self/mounts.
12401 * Support for the "pids" cgroup controller has been added. It
12402 allows accounting the number of tasks in a cgroup and
12403 enforcing limits on it. This adds two new setting
12404 TasksAccounting= and TasksMax= to each unit, as well as a
12405 global option DefaultTasksAccounting=.
12407 * Support for the "net_cls" cgroup controller has been added.
12408 It allows assigning a net class ID to each task in the
12409 cgroup, which can then be used in firewall rules and traffic
12410 shaping configurations. Note that the kernel netfilter net
12411 class code does not currently work reliably for ingress
12412 packets on unestablished sockets.
12414 This adds a new config directive called NetClass= to CGroup
12415 enabled units. Allowed values are positive numbers for fixed
12416 assignments and "auto" for picking a free value
12419 * 'systemctl is-system-running' now returns 'offline' if the
12420 system is not booted with systemd. This command can now be
12421 used as a substitute for 'systemd-notify --booted'.
12423 * Watchdog timeouts have been increased to 3 minutes for all
12424 in-tree service files. Apparently, disk IO issues are more
12425 frequent than we hoped, and user reported >1 minute waiting
12428 * 'machine-id-commit' functionality has been merged into
12429 'machine-id-setup --commit'. The separate binary has been
12432 * The WorkingDirectory= directive in unit files may now be set
12433 to the special value '~'. In this case, the working
12434 directory is set to the home directory of the user
12435 configured in User=.
12437 * "machinectl shell" will now open the shell in the home
12438 directory of the selected user by default.
12440 * The CrashChVT= configuration file setting is renamed to
12441 CrashChangeVT=, following our usual logic of not
12442 abbreviating unnecessarily. The old directive is still
12443 supported for compat reasons. Also, this directive now takes
12444 an integer value between 1 and 63, or a boolean value. The
12445 formerly supported '-1' value for disabling stays around for
12448 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
12449 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
12450 RootDirectory= properties can now be set for transient
12453 * The systemd-analyze tool gained a new "set-log-target" verb
12454 to change the logging target the system manager logs to
12455 dynamically during runtime. This is similar to how
12456 "systemd-analyze set-log-level" already changes the log
12459 * In nspawn /sys is now mounted as tmpfs, with only a selected
12460 set of subdirectories mounted in from the real sysfs. This
12461 enhances security slightly, and is useful for ensuring user
12462 namespaces work correctly.
12464 * Support for USB FunctionFS activation has been added. This
12465 allows implementation of USB gadget services that are
12466 activated as soon as they are requested, so that they don't
12467 have to run continuously, similar to classic socket
12470 * The "systemctl exit" command now optionally takes an
12471 additional parameter that sets the exit code to return from
12472 the systemd manager when exiting. This is only relevant when
12473 running the systemd user instance, or when running the
12474 system instance in a container.
12476 * sd-bus gained the new API calls sd_bus_path_encode_many()
12477 and sd_bus_path_decode_many() that allow easy encoding and
12478 decoding of multiple identifier strings inside a D-Bus
12479 object path. Another new call sd_bus_default_flush_close()
12480 has been added to flush and close per-thread default
12483 * systemd-cgtop gained support for a -M/--machine= switch to
12484 show the control groups within a certain container only.
12486 * "systemctl kill" gained support for an optional --fail
12487 switch. If specified the requested operation will fail of no
12488 processes have been killed, because the unit had no
12489 processes attached, or similar.
12491 * A new systemd.crash_reboot=1 kernel command line option has
12492 been added that triggers a reboot after crashing. This can
12493 also be set through CrashReboot= in systemd.conf.
12495 * The RuntimeDirectory= setting now understands unit
12496 specifiers like %i or %f.
12498 * A new (still internal) library API sd-ipv4acd has been added,
12499 that implements address conflict detection for IPv4. It's
12500 based on code from sd-ipv4ll, and will be useful for
12501 detecting DHCP address conflicts.
12503 * File descriptors passed during socket activation may now be
12504 named. A new API sd_listen_fds_with_names() is added to
12505 access the names. The default names may be overridden,
12506 either in the .socket file using the FileDescriptorName=
12507 parameter, or by passing FDNAME= when storing the file
12508 descriptors using sd_notify().
12510 * systemd-networkd gained support for:
12512 - Setting the IPv6 Router Advertisement settings via
12513 IPv6AcceptRouterAdvertisements= in .network files.
12515 - Configuring the HelloTimeSec=, MaxAgeSec= and
12516 ForwardDelaySec= bridge parameters in .netdev files.
12518 - Configuring PreferredSource= for static routes in
12521 * The "ask-password" framework used to query for LUKS harddisk
12522 passwords or SSL passwords during boot gained support for
12523 caching passwords in the kernel keyring, if it is
12524 available. This makes sure that the user only has to type in
12525 a passphrase once if there are multiple objects to unlock
12526 with the same one. Previously, such password caching was
12527 available only when Plymouth was used; this moves the
12528 caching logic into the systemd codebase itself. The
12529 "systemd-ask-password" utility gained a new --keyname=
12530 switch to control which kernel keyring key to use for
12531 caching a password in. This functionality is also useful for
12532 enabling display managers such as gdm to automatically
12533 unlock the user's GNOME keyring if its passphrase, the
12534 user's password and the harddisk password are the same, if
12535 gdm-autologin is used.
12537 * When downloading tar or raw images using "machinectl
12538 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
12539 file is now also downloaded, if it is available and stored
12540 next to the image file.
12542 * Units of type ".socket" gained a new boolean setting
12543 Writable= which is only useful in conjunction with
12544 ListenSpecial=. If true, enables opening the specified
12545 special file in O_RDWR mode rather than O_RDONLY mode.
12547 * systemd-rfkill has been reworked to become a singleton
12548 service that is activated through /dev/rfkill on each rfkill
12549 state change and saves the settings to disk. This way,
12550 systemd-rfkill is now compatible with devices that exist
12551 only intermittendly, and even restores state if the previous
12552 system shutdown was abrupt rather than clean.
12554 * The journal daemon gained support for vacuuming old journal
12555 files controlled by the number of files that shall remain,
12556 in addition to the already existing control by size and by
12557 date. This is useful as journal interleaving performance
12558 degrades with too many separate journal files, and allows
12559 putting an effective limit on them. The new setting defaults
12560 to 100, but this may be changed by setting SystemMaxFiles=
12561 and RuntimeMaxFiles= in journald.conf. Also, the
12562 "journalctl" tool gained the new --vacuum-files= switch to
12563 manually vacuum journal files to leave only the specified
12564 number of files in place.
12566 * udev will now create /dev/disk/by-path links for ATA devices
12567 on kernels where that is supported.
12569 * Galician, Serbian, Turkish and Korean translations were added.
12571 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
12572 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
12573 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
12574 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
12575 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
12576 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
12577 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
12578 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
12579 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
12580 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
12581 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
12582 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
12583 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
12584 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
12585 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
12586 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
12587 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
12588 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
12590 — Berlin, 2015-10-07
12594 * The DHCP implementation of systemd-networkd gained a set of
12597 - The DHCP server now supports emitting DNS and NTP
12598 information. It may be enabled and configured via
12599 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
12600 and NTP information is enabled, but no servers are
12601 configured, the corresponding uplink information (if there
12602 is any) is propagated.
12604 - Server and client now support transmission and reception
12605 of timezone information. It can be configured via the
12606 newly introduced network options UseTimezone=,
12607 EmitTimezone=, and Timezone=. Transmission of timezone
12608 information is enabled between host and containers by
12609 default now: the container will change its local timezone
12610 to what the host has set.
12612 - Lease timeouts can now be configured via
12613 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
12615 - The DHCP server improved on the stability of
12616 leases. Clients are more likely to get the same lease
12617 information back, even if the server loses state.
12619 - The DHCP server supports two new configuration options to
12620 control the lease address pool metrics, PoolOffset= and
12623 * The encapsulation limit of tunnels in systemd-networkd may
12624 now be configured via 'EncapsulationLimit='. It allows
12625 modifying the maximum additional levels of encapsulation
12626 that are permitted to be prepended to a packet.
12628 * systemd now supports the concept of user buses replacing
12629 session buses, if used with dbus-1.10 (and enabled via dbus
12630 --enable-user-session). It previously only supported this on
12631 kdbus-enabled systems, and this release expands this to
12632 'dbus-daemon' systems.
12634 * systemd-networkd now supports predictable interface names
12635 for virtio devices.
12637 * systemd now optionally supports the new Linux kernel
12638 "unified" control group hierarchy. If enabled via the kernel
12639 command-line option 'systemd.unified_cgroup_hierarchy=1',
12640 systemd will try to mount the unified cgroup hierarchy
12641 directly on /sys/fs/cgroup. If not enabled, or not
12642 available, systemd will fall back to the legacy cgroup
12643 hierarchy setup, as before. Host system and containers can
12644 mix and match legacy and unified hierarchies as they
12645 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
12646 environment variable to individually select the hierarchy to
12647 use for executed containers. By default, nspawn will use the
12648 unified hierarchy for the containers if the host uses the
12649 unified hierarchy, and the legacy hierarchy otherwise.
12650 Please note that at this point the unified hierarchy is an
12651 experimental kernel feature and is likely to change in one
12652 of the next kernel releases. Therefore, it should not be
12653 enabled by default in downstream distributions yet. The
12654 minimum required kernel version for the unified hierarchy to
12655 work is 4.2. Note that when the unified hierarchy is used
12656 for the first time delegated access to controllers is
12657 safe. Because of this systemd-nspawn containers will get
12658 access to controllers now, as will systemd user
12659 sessions. This means containers and user sessions may now
12660 manage their own resources, partitioning up what the system
12663 * A new special scope unit "init.scope" has been introduced
12664 that encapsulates PID 1 of the system. It may be used to
12665 determine resource usage and enforce resource limits on PID
12666 1 itself. PID 1 hence moved out of the root of the control
12669 * The cgtop tool gained support for filtering out kernel
12670 threads when counting tasks in a control group. Also, the
12671 count of processes is now recursively summed up by
12672 default. Two options -k and --recursive= have been added to
12673 revert to old behaviour. The tool has also been updated to
12674 work correctly in containers now.
12676 * systemd-nspawn's --bind= and --bind-ro= options have been
12677 extended to allow creation of non-recursive bind mounts.
12679 * libsystemd gained two new calls sd_pid_get_cgroup() and
12680 sd_peer_get_cgroup() which return the control group path of
12681 a process or peer of a connected AF_UNIX socket. This
12682 function call is particularly useful when implementing
12683 delegated subtrees support in the control group hierarchy.
12685 * The "sd-event" event loop API of libsystemd now supports
12686 correct dequeuing of real-time signals, without losing
12689 * When systemd requests a polkit decision when managing units it
12690 will now add additional fields to the request, including unit
12691 name and desired operation. This enables more powerful polkit
12692 policies, that make decisions depending on these parameters.
12694 * nspawn learnt support for .nspawn settings files, that may
12695 accompany the image files or directories of containers, and
12696 may contain additional settings for the container. This is
12697 an alternative to configuring container parameters via the
12698 nspawn command line.
12700 Contributions from: Cristian Rodríguez, Daniel Mack, David
12701 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
12702 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
12703 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
12704 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
12705 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
12706 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
12707 Andersen, Tom Gundersen, Torstein Husebø
12709 — Berlin, 2015-09-08
12713 * machinectl gained a new verb 'shell' which opens a fresh
12714 shell on the target container or the host. It is similar to
12715 the existing 'login' command of machinectl, but spawns the
12716 shell directly without prompting for username or
12717 password. The pseudo machine '.host' now refers to the local
12718 host and is used by default. Hence, 'machinectl shell' can
12719 be used as replacement for 'su -' which spawns a session as
12720 a fresh systemd unit in a way that is fully isolated from
12721 the originating session.
12723 * systemd-networkd learned to cope with private-zone DHCP
12724 options and allows other programs to query the values.
12726 * SELinux access control when enabling/disabling units is no
12727 longer enforced with this release. The previous implementation
12728 was incorrect, and a new corrected implementation is not yet
12729 available. As unit file operations are still protected via
12730 polkit and D-Bus policy this is not a security problem. Yet,
12731 distributions which care about optimal SELinux support should
12732 probably not stabilize on this release.
12734 * sd-bus gained support for matches of type "arg0has=", that
12735 test for membership of strings in string arrays sent in bus
12738 * systemd-resolved now dumps the contents of its DNS and LLMNR
12739 caches to the logs on reception of the SIGUSR1 signal. This
12740 is useful to debug DNS behaviour.
12742 * The coredumpctl tool gained a new --directory= option to
12743 operate on journal files in a specific directory.
12745 * "systemctl reboot" and related commands gained a new
12746 "--message=" option which may be used to set a free-text
12747 wall message when shutting down or rebooting the
12748 system. This message is also logged, which is useful for
12749 figuring out the reason for a reboot or shutdown a
12752 * The "systemd-resolve-host" tool's -i switch now takes
12753 network interface numbers as alternative to interface names.
12755 * A new unit file setting for services has been introduced:
12756 UtmpMode= allows configuration of how precisely systemd
12757 handles utmp and wtmp entries for the service if this is
12758 enabled. This allows writing services that appear similar to
12759 user sessions in the output of the "w", "who", "last" and
12762 * systemd-resolved will now locally synthesize DNS resource
12763 records for the "localhost" and "gateway" domains as well as
12764 the local hostname. This should ensure that clients querying
12765 RRs via resolved will get similar results as those going via
12766 NSS, if nss-myhostname is enabled.
12768 Contributions from: Alastair Hughes, Alex Crawford, Daniel
12769 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
12770 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
12771 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
12772 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
12773 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
12774 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
12775 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
12776 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
12777 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
12778 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
12779 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
12781 — Berlin, 2015-08-27
12785 * The systemd-efi-boot-generator functionality was merged into
12786 systemd-gpt-auto-generator.
12788 * systemd-networkd now supports Group Policy for vxlan
12789 devices. It can be enabled via the new boolean configuration
12790 option called 'GroupPolicyExtension='.
12792 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
12793 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
12794 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
12796 — Berlin, 2015-07-31
12800 * The python-systemd code has been removed from the systemd repository.
12801 A new repository has been created which accommodates the code from
12802 now on, and we kindly ask distributions to create a separate package
12803 for this: https://github.com/systemd/python-systemd
12805 * The systemd daemon will now reload its main configuration
12806 (/etc/systemd/system.conf) on daemon-reload.
12808 * sd-dhcp now exposes vendor specific extensions via
12809 sd_dhcp_lease_get_vendor_specific().
12811 * systemd-networkd gained a number of new configuration options.
12813 - A new boolean configuration option for TAP devices called
12814 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
12815 device, thus allowing to send and receive GSO packets.
12817 - A new tunnel configuration option called 'CopyDSCP='.
12818 If enabled, the DSCP field of ip6 tunnels is copied into the
12819 decapsulated packet.
12821 - A set of boolean bridge configuration options were added.
12822 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
12823 and 'UnicastFlood=' are now parsed by networkd and applied to the
12824 respective bridge link device via the respective IFLA_BRPORT_*
12827 - A new string configuration option to override the hostname sent
12828 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
12829 is true, networkd will use the configured hostname instead of the
12830 system hostname when sending DHCP requests.
12832 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
12833 networkd will configure the IPv6 flow-label of the tunnel device
12834 according to RFC2460.
12836 - The 'macvtap' virtual network devices are now supported, similar to
12837 the already supported 'macvlan' devices.
12839 * systemd-resolved now implements RFC5452 to improve resilience against
12840 cache poisoning. Additionally, source port randomization is enabled
12841 by default to further protect against DNS spoofing attacks.
12843 * nss-mymachines now supports translating UIDs and GIDs of running
12844 containers with user-namespaces enabled. If a container 'foo'
12845 translates a host uid 'UID' to the container uid 'TUID', then
12846 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
12847 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
12848 mapped as 'vg-foo-TGID'.
12850 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
12851 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
12852 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
12853 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
12854 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
12855 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
12856 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
12857 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
12858 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
12859 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
12861 — Berlin, 2015-07-29
12865 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
12866 There are no known issues with current sysfs, and udev does not need
12867 or should be used to work around such bugs.
12869 * udev does no longer enable USB HID power management. Several reports
12870 indicate, that some devices cannot handle that setting.
12872 * The udev accelerometer helper was removed. The functionality
12873 is now fully included in iio-sensor-proxy. But this means,
12874 older iio-sensor-proxy versions will no longer provide
12875 accelerometer/orientation data with this systemd version.
12876 Please upgrade iio-sensor-proxy to version 1.0.
12878 * networkd gained a new configuration option IPv6PrivacyExtensions=
12879 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
12880 for Stateless Address") on selected networks.
12882 * For the sake of fewer build-time dependencies and less code in the
12883 main repository, the python bindings are about to be removed in the
12884 next release. A new repository has been created which accommodates
12885 the code from now on, and we kindly ask distributions to create a
12886 separate package for this. The removal will take place in v223.
12888 https://github.com/systemd/python-systemd
12890 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
12891 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
12892 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
12893 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
12894 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
12895 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
12896 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
12897 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
12898 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12899 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
12901 — Berlin, 2015-07-07
12905 * The sd-bus.h and sd-event.h APIs have now been declared
12906 stable and have been added to the official interface of
12907 libsystemd.so. sd-bus implements an alternative D-Bus client
12908 library, that is relatively easy to use, very efficient and
12909 supports both classic D-Bus as well as kdbus as transport
12910 backend. sd-event is a generic event loop abstraction that
12911 is built around Linux epoll, but adds features such as event
12912 prioritization or efficient timer handling. Both APIs are good
12913 choices for C programs looking for a bus and/or event loop
12914 implementation that is minimal and does not have to be
12915 portable to other kernels.
12917 * kdbus support is no longer compile-time optional. It is now
12918 always built-in. However, it can still be disabled at
12919 runtime using the kdbus=0 kernel command line setting, and
12920 that setting may be changed to default to off, by specifying
12921 --disable-kdbus at build-time. Note though that the kernel
12922 command line setting has no effect if the kdbus.ko kernel
12923 module is not installed, in which case kdbus is (obviously)
12924 also disabled. We encourage all downstream distributions to
12925 begin testing kdbus by adding it to the kernel images in the
12926 development distributions, and leaving kdbus support in
12929 * The minimal required util-linux version has been bumped to
12932 * Support for chkconfig (--enable-chkconfig) was removed in
12933 favor of calling an abstraction tool
12934 /lib/systemd/systemd-sysv-install. This needs to be
12935 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
12936 in README for details.
12938 * If there's a systemd unit and a SysV init script for the
12939 same service name, and the user executes "systemctl enable"
12940 for it (or a related call), then this will now enable both
12941 (or execute the related operation on both), not just the
12944 * The libudev API documentation has been converted from gtkdoc
12947 * gudev has been removed from the systemd tree, it is now an
12950 * The systemd-cgtop tool learnt a new --raw switch to generate
12951 "raw" (machine parsable) output.
12953 * networkd's IPForwarding= .network file setting learnt the
12954 new setting "kernel", which ensures that networkd does not
12955 change the IP forwarding sysctl from the default kernel
12958 * The systemd-logind bus API now exposes a new boolean
12959 property "Docked" that reports whether logind considers the
12960 system "docked", i.e. connected to a docking station or not.
12962 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
12963 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
12964 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
12965 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
12966 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
12967 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
12968 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
12969 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
12970 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
12971 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
12972 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
12973 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
12974 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
12975 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
12976 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
12977 Fink, Zbigniew Jędrzejewski-Szmek
12979 — Berlin, 2015-06-19
12983 * The gudev library has been extracted into a separate repository
12984 available at: https://git.gnome.org/browse/libgudev/
12985 It is now managed as part of the Gnome project. Distributions
12986 are recommended to pass --disable-gudev to systemd and use
12987 gudev from the Gnome project instead. gudev is still included
12988 in systemd, for now. It will be removed soon, though. Please
12989 also see the announcement-thread on systemd-devel:
12990 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
12992 * systemd now exposes a CPUUsageNSec= property for each
12993 service unit on the bus, that contains the overall consumed
12994 CPU time of a service (the sum of what each process of the
12995 service consumed). This value is only available if
12996 CPUAccounting= is turned on for a service, and is then shown
12997 in the "systemctl status" output.
12999 * Support for configuring alternative mappings of the old SysV
13000 runlevels to systemd targets has been removed. They are now
13001 hardcoded in a way that runlevels 2, 3, 4 all map to
13002 multi-user.target and 5 to graphical.target (which
13003 previously was already the default behaviour).
13005 * The auto-mounter logic gained support for mount point
13006 expiry, using a new TimeoutIdleSec= setting in .automount
13007 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
13009 * The EFI System Partition (ESP) as mounted to /boot by
13010 systemd-efi-boot-generator will now be unmounted
13011 automatically after 2 minutes of not being used. This should
13012 minimize the risk of ESP corruptions.
13014 * New /etc/fstab options x-systemd.requires= and
13015 x-systemd.requires-mounts-for= are now supported to express
13016 additional dependencies for mounts. This is useful for
13017 journaling file systems that support external journal
13018 devices or overlay file systems that require underlying file
13019 systems to be mounted.
13021 * systemd does not support direct live-upgrades (via systemctl
13022 daemon-reexec) from versions older than v44 anymore. As no
13023 distribution we are aware of shipped such old versions in a
13024 stable release this should not be problematic.
13026 * When systemd forks off a new per-connection service instance
13027 it will now set the $REMOTE_ADDR environment variable to the
13028 remote IP address, and $REMOTE_PORT environment variable to
13029 the remote IP port. This behaviour is similar to the
13030 corresponding environment variables defined by CGI.
13032 * systemd-networkd gained support for uplink failure
13033 detection. The BindCarrier= option allows binding interface
13034 configuration dynamically to the link sense of other
13035 interfaces. This is useful to achieve behaviour like in
13038 * systemd-networkd gained support for configuring the DHCP
13039 client identifier to use when requesting leases.
13041 * systemd-networkd now has a per-network UseNTP= option to
13042 configure whether NTP server information acquired via DHCP
13043 is passed on to services like systemd-timesyncd.
13045 * systemd-networkd gained support for vti6 tunnels.
13047 * Note that systemd-networkd manages the sysctl variable
13048 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
13049 it is configured for since v219. The variable controls IP
13050 forwarding, and is a per-interface alternative to the global
13051 /proc/sys/net/ipv[46]/ip_forward. This setting is
13052 configurable in the IPForward= option, which defaults to
13053 "no". This means if networkd is used for an interface it is
13054 no longer sufficient to set the global sysctl option to turn
13055 on IP forwarding! Instead, the .network file option
13056 IPForward= needs to be turned on! Note that the
13057 implementation of this behaviour was broken in v219 and has
13058 been fixed in v220.
13060 * Many bonding and vxlan options are now configurable in
13063 * systemd-nspawn gained a new --property= setting to set unit
13064 properties for the container scope. This is useful for
13065 setting resource parameters (e.g. "CPUShares=500") on
13066 containers started from the command line.
13068 * systemd-nspawn gained a new --private-users= switch to make
13069 use of user namespacing available on recent Linux kernels.
13071 * systemd-nspawn may now be called as part of a shell pipeline
13072 in which case the pipes used for stdin and stdout are passed
13073 directly to the process invoked in the container, without
13074 indirection via a pseudo tty.
13076 * systemd-nspawn gained a new switch to control the UNIX
13077 signal to use when killing the init process of the container
13078 when shutting down.
13080 * systemd-nspawn gained a new --overlay= switch for mounting
13081 overlay file systems into the container using the new kernel
13084 * When a container image is imported via systemd-importd and
13085 the host file system is not btrfs, a loopback block device
13086 file is created in /var/lib/machines.raw with a btrfs file
13087 system inside. It is then mounted to /var/lib/machines to
13088 enable btrfs features for container management. The loopback
13089 file and btrfs file system is grown as needed when container
13090 images are imported via systemd-importd.
13092 * systemd-machined/systemd-importd gained support for btrfs
13093 quota, to enforce container disk space limits on disk. This
13094 is exposed in "machinectl set-limit".
13096 * systemd-importd now can import containers from local .tar,
13097 .raw and .qcow2 images, and export them to .tar and .raw. It
13098 can also import dkr v2 images now from the network (on top
13101 * systemd-importd gained support for verifying downloaded
13102 images with gpg2 (previously only gpg1 was supported).
13104 * systemd-machined, systemd-logind, systemd: most bus calls are
13105 now accessible to unprivileged processes via polkit. Also,
13106 systemd-logind will now allow users to kill their own sessions
13107 without further privileges or authorization.
13109 * systemd-shutdownd has been removed. This service was
13110 previously responsible for implementing scheduled shutdowns
13111 as exposed in /usr/bin/shutdown's time parameter. This
13112 functionality has now been moved into systemd-logind and is
13113 accessible via a bus interface.
13115 * "systemctl reboot" gained a new switch --firmware-setup that
13116 can be used to reboot into the EFI firmware setup, if that
13117 is available. systemd-logind now exposes an API on the bus
13118 to trigger such reboots, in case graphical desktop UIs want
13119 to cover this functionality.
13121 * "systemctl enable", "systemctl disable" and "systemctl mask"
13122 now support a new "--now" switch. If specified the units
13123 that are enabled will also be started, and the ones
13124 disabled/masked also stopped.
13126 * The Gummiboot EFI boot loader tool has been merged into
13127 systemd, and renamed to "systemd-boot". The bootctl tool has been
13128 updated to support systemd-boot.
13130 * An EFI kernel stub has been added that may be used to create
13131 kernel EFI binaries that contain not only the actual kernel,
13132 but also an initrd, boot splash, command line and OS release
13133 information. This combined binary can then be signed as a
13134 single image, so that the firmware can verify it all in one
13135 step. systemd-boot has special support for EFI binaries created
13136 like this and can extract OS release information from them
13137 and show them in the boot menu. This functionality is useful
13138 to implement cryptographically verified boot schemes.
13140 * Optional support has been added to systemd-fsck to pass
13141 fsck's progress report to an AF_UNIX socket in the file
13144 * udev will no longer create device symlinks for all block devices by
13145 default. A deny list for excluding special block devices from this
13146 logic has been turned into an allow list that requires picking block
13147 devices explicitly that require device symlinks.
13149 * A new (currently still internal) API sd-device.h has been
13150 added to libsystemd. This modernized API is supposed to
13151 replace libudev eventually. In fact, already much of libudev
13152 is now just a wrapper around sd-device.h.
13154 * A new hwdb database for storing metadata about pointing
13155 stick devices has been added.
13157 * systemd-tmpfiles gained support for setting file attributes
13158 similar to the "chattr" tool with new 'h' and 'H' lines.
13160 * systemd-journald will no longer unconditionally set the
13161 btrfs NOCOW flag on new journal files. This is instead done
13162 with tmpfiles snippet using the new 'h' line type. This
13163 allows easy disabling of this logic, by masking the
13164 journal-nocow.conf tmpfiles file.
13166 * systemd-journald will now translate audit message types to
13167 human readable identifiers when writing them to the
13168 journal. This should improve readability of audit messages.
13170 * The LUKS logic gained support for the offset= and skip=
13171 options in /etc/crypttab, as previously implemented by
13174 * /usr/lib/os-release gained a new optional field VARIANT= for
13175 distributions that support multiple variants (such as a
13176 desktop edition, a server edition, …)
13178 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
13179 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
13180 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
13181 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
13182 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
13183 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
13184 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
13185 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
13186 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
13187 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
13188 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
13189 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
13190 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
13191 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
13192 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
13193 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
13194 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
13195 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
13196 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
13197 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
13198 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
13199 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
13200 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
13201 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
13202 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
13203 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
13204 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
13206 — Berlin, 2015-05-22
13210 * Introduce a new API "sd-hwdb.h" for querying the hardware
13211 metadata database. With this minimal interface one can query
13212 and enumerate the udev hwdb, decoupled from the old libudev
13213 library. libudev's interface for this is now only a wrapper
13214 around sd-hwdb. A new tool systemd-hwdb has been added to
13215 interface with and update the database.
13217 * When any of systemd's tools copies files (for example due to
13218 tmpfiles' C lines) a btrfs reflink will attempted first,
13219 before bytewise copying is done.
13221 * systemd-nspawn gained a new --ephemeral switch. When
13222 specified a btrfs snapshot is taken of the container's root
13223 directory, and immediately removed when the container
13224 terminates again. Thus, a container can be started whose
13225 changes never alter the container's root directory, and are
13226 lost on container termination. This switch can also be used
13227 for starting a container off the root file system of the
13228 host without affecting the host OS. This switch is only
13229 available on btrfs file systems.
13231 * systemd-nspawn gained a new --template= switch. It takes the
13232 path to a container tree to use as template for the tree
13233 specified via --directory=, should that directory be
13234 missing. This allows instantiating containers dynamically,
13235 on first run. This switch is only available on btrfs file
13238 * When a .mount unit refers to a mount point on which multiple
13239 mounts are stacked, and the .mount unit is stopped all of
13240 the stacked mount points will now be unmounted until no
13241 mount point remains.
13243 * systemd now has an explicit notion of supported and
13244 unsupported unit types. Jobs enqueued for unsupported unit
13245 types will now fail with an "unsupported" error code. More
13246 specifically .swap, .automount and .device units are not
13247 supported in containers, .busname units are not supported on
13248 non-kdbus systems. .swap and .automount are also not
13249 supported if their respective kernel compile time options
13252 * machinectl gained support for two new "copy-from" and
13253 "copy-to" commands for copying files from a running
13254 container to the host or vice versa.
13256 * machinectl gained support for a new "bind" command to bind
13257 mount host directories into local containers. This is
13258 currently only supported for nspawn containers.
13260 * networkd gained support for configuring bridge forwarding
13261 database entries (fdb) from .network files.
13263 * A new tiny daemon "systemd-importd" has been added that can
13264 download container images in tar, raw, qcow2 or dkr formats,
13265 and make them available locally in /var/lib/machines, so
13266 that they can run as nspawn containers. The daemon can GPG
13267 verify the downloads (not supported for dkr, since it has no
13268 provisions for verifying downloads). It will transparently
13269 decompress bz2, xz, gzip compressed downloads if necessary,
13270 and restore sparse files on disk. The daemon uses privilege
13271 separation to ensure the actual download logic runs with
13272 fewer privileges than the daemon itself. machinectl has
13273 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
13274 make the functionality of importd available to the
13275 user. With this in place the Fedora and Ubuntu "Cloud"
13276 images can be downloaded and booted as containers unmodified
13277 (the Fedora images lack the appropriate GPG signature files
13278 currently, so they cannot be verified, but this will change
13279 soon, hopefully). Note that downloading images is currently
13280 only fully supported on btrfs.
13282 * machinectl is now able to list container images found in
13283 /var/lib/machines, along with some metadata about sizes of
13284 disk and similar. If the directory is located on btrfs and
13285 quota is enabled, this includes quota display. A new command
13286 "image-status" has been added that shows additional
13287 information about images.
13289 * machinectl is now able to clone container images
13290 efficiently, if the underlying file system (btrfs) supports
13291 it, with the new "machinectl clone" command. It also
13292 gained commands for renaming and removing images, as well as
13293 marking them read-only or read-write (supported also on
13294 legacy file systems).
13296 * networkd gained support for collecting LLDP network
13297 announcements, from hardware that supports this. This is
13298 shown in networkctl output.
13300 * systemd-run gained support for a new -t (--pty) switch for
13301 invoking a binary on a pty whose input and output is
13302 connected to the invoking terminal. This allows executing
13303 processes as system services while interactively
13304 communicating with them via the terminal. Most interestingly
13305 this is supported across container boundaries. Invoking
13306 "systemd-run -t /bin/bash" is an alternative to running a
13307 full login session, the difference being that the former
13308 will not register a session, nor go through the PAM session
13311 * tmpfiles gained support for a new "v" line type for creating
13312 btrfs subvolumes. If the underlying file system is a legacy
13313 file system, this automatically degrades to creating a
13314 normal directory. Among others /var/lib/machines is now
13315 created like this at boot, should it be missing.
13317 * The directory /var/lib/containers/ has been deprecated and
13318 been replaced by /var/lib/machines. The term "machines" has
13319 been used in the systemd context as generic term for both
13320 VMs and containers, and hence appears more appropriate for
13321 this, as the directory can also contain raw images bootable
13324 * systemd-nspawn when invoked with -M but without --directory=
13325 or --image= is now capable of searching for the container
13326 root directory, subvolume or disk image automatically, in
13327 /var/lib/machines. systemd-nspawn@.service has been updated
13328 to make use of this, thus allowing it to be used for raw
13331 * A new machines.target unit has been introduced that is
13332 supposed to group all containers/VMs invoked as services on
13333 the system. systemd-nspawn@.service has been updated to
13334 integrate with that.
13336 * machinectl gained a new "start" command, for invoking a
13337 container as a service. "machinectl start foo" is mostly
13338 equivalent to "systemctl start systemd-nspawn@foo.service",
13339 but handles escaping in a nicer way.
13341 * systemd-nspawn will now mount most of the cgroupfs tree
13342 read-only into each container, with the exception of the
13343 container's own subtree in the name=systemd hierarchy.
13345 * journald now sets the special FS_NOCOW file flag for its
13346 journal files. This should improve performance on btrfs, by
13347 avoiding heavy fragmentation when journald's write-pattern
13348 is used on COW file systems. It degrades btrfs' data
13349 integrity guarantees for the files to the same levels as for
13350 ext3/ext4 however. This should be OK though as journald does
13351 its own data integrity checks and all its objects are
13352 checksummed on disk. Also, journald should handle btrfs disk
13353 full events a lot more gracefully now, by processing SIGBUS
13354 errors, and not relying on fallocate() anymore.
13356 * When journald detects that journal files it is writing to
13357 have been deleted it will immediately start new journal
13360 * systemd now provides a way to store file descriptors
13361 per-service in PID 1. This is useful for daemons to ensure
13362 that fds they require are not lost during a daemon
13363 restart. The fds are passed to the daemon on the next
13364 invocation in the same way socket activation fds are
13365 passed. This is now used by journald to ensure that the
13366 various sockets connected to all the system's stdout/stderr
13367 are not lost when journald is restarted. File descriptors
13368 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
13369 an extension to sd_notify(). Note that a limit is enforced
13370 on the number of fds a service can store in PID 1, and it
13371 defaults to 0, so that no fds may be stored, unless this is
13372 explicitly turned on.
13374 * The default TERM variable to use for units connected to a
13375 terminal, when no other value is explicitly is set is now
13376 vt220 rather than vt102. This should be fairly safe still,
13377 but allows PgUp/PgDn work.
13379 * The /etc/crypttab option header= as known from Debian is now
13382 * "loginctl user-status" and "loginctl session-status" will
13383 now show the last 10 lines of log messages of the
13384 user/session following the status output. Similar,
13385 "machinectl status" will show the last 10 log lines
13386 associated with a virtual machine or container
13387 service. (Note that this is usually not the log messages
13388 done in the VM/container itself, but simply what the
13389 container manager logs. For nspawn this includes all console
13392 * "loginctl session-status" without further argument will now
13393 show the status of the session of the caller. Similar,
13394 "lock-session", "unlock-session", "activate",
13395 "enable-linger", "disable-linger" may now be called without
13396 session/user parameter in which case they apply to the
13397 caller's session/user.
13399 * An X11 session scriptlet is now shipped that uploads
13400 $DISPLAY and $XAUTHORITY into the environment of the systemd
13401 --user daemon if a session begins. This should improve
13402 compatibility with X11 enabled applications run as systemd
13405 * Generators are now subject to masking via /etc and /run, the
13406 same way as unit files.
13408 * networkd .network files gained support for configuring
13409 per-link IPv4/IPv6 packet forwarding as well as IPv4
13410 masquerading. This is by default turned on for veth links to
13411 containers, as registered by systemd-nspawn. This means that
13412 nspawn containers run with --network-veth will now get
13413 automatic routed access to the host's networks without any
13414 further configuration or setup, as long as networkd runs on
13417 * systemd-nspawn gained the --port= (-p) switch to expose TCP
13418 or UDP posts of a container on the host. With this in place
13419 it is possible to run containers with private veth links
13420 (--network-veth), and have their functionality exposed on
13421 the host as if their services were running directly on the
13424 * systemd-nspawn's --network-veth switch now gained a short
13425 version "-n", since with the changes above it is now truly
13426 useful out-of-the-box. The systemd-nspawn@.service has been
13427 updated to make use of it too by default.
13429 * systemd-nspawn will now maintain a per-image R/W lock, to
13430 ensure that the same image is not started more than once
13431 writable. (It's OK to run an image multiple times
13432 simultaneously in read-only mode.)
13434 * systemd-nspawn's --image= option is now capable of
13435 dissecting and booting MBR and GPT disk images that contain
13436 only a single active Linux partition. Previously it
13437 supported only GPT disk images with proper GPT type
13438 IDs. This allows running cloud images from major
13439 distributions directly with systemd-nspawn, without
13442 * In addition to collecting mouse dpi data in the udev
13443 hardware database, there's now support for collecting angle
13444 information for mouse scroll wheels. The database is
13445 supposed to guarantee similar scrolling behavior on mice
13446 that it knows about. There's also support for collecting
13447 information about Touchpad types.
13449 * udev's input_id built-in will now also collect touch screen
13450 dimension data and attach it to probed devices.
13452 * /etc/os-release gained support for a Distribution Privacy
13455 * networkd gained support for creating "ipvlan", "gretap",
13456 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
13458 * systemd-tmpfiles gained support for "a" lines for setting
13461 * systemd-nspawn will now mount /tmp in the container to
13462 tmpfs, automatically.
13464 * systemd now exposes the memory.usage_in_bytes cgroup
13465 attribute and shows it for each service in the "systemctl
13466 status" output, if available.
13468 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
13469 immediate reboot is triggered. This useful if shutdown is
13470 hung and is unable to complete, to expedite the
13471 operation. Note that this kind of reboot will still unmount
13472 all file systems, and hence should not result in fsck being
13473 run on next reboot.
13475 * A .device unit for an optical block device will now be
13476 considered active only when a medium is in the drive. Also,
13477 mount units are now bound to their backing devices thus
13478 triggering automatic unmounting when devices become
13479 unavailable. With this in place systemd will now
13480 automatically unmount left-over mounts when a CD-ROM is
13481 ejected or a USB stick is yanked from the system.
13483 * networkd-wait-online now has support for waiting for
13484 specific interfaces only (with globbing), and for giving up
13485 after a configurable timeout.
13487 * networkd now exits when idle. It will be automatically
13488 restarted as soon as interfaces show up, are removed or
13489 change state. networkd will stay around as long as there is
13490 at least one DHCP state machine or similar around, that keep
13493 * networkd may now configure IPv6 link-local addressing in
13494 addition to IPv4 link-local addressing.
13496 * The IPv6 "token" for use in SLAAC may now be configured for
13497 each .network interface in networkd.
13499 * Routes configured with networkd may now be assigned a scope
13502 * networkd's [Match] sections now support globbing and lists
13503 of multiple space-separated matches per item.
13505 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
13506 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
13507 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
13508 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
13509 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
13510 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
13511 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
13512 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
13513 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
13514 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
13515 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
13516 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
13517 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
13518 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
13519 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
13520 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
13521 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
13522 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
13523 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
13524 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
13525 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
13526 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
13527 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
13528 Hoffmann, Zbigniew Jędrzejewski-Szmek
13530 — Berlin, 2015-02-16
13534 * When querying unit file enablement status (for example via
13535 "systemctl is-enabled"), a new state "indirect" is now known
13536 which indicates that a unit might not be enabled itself, but
13537 another unit listed in its Also= setting might be.
13539 * Similar to the various existing ConditionXYZ= settings for
13540 units, there are now matching AssertXYZ= settings. While
13541 failing conditions cause a unit to be skipped, but its job
13542 to succeed, failing assertions declared like this will cause
13543 a unit start operation and its job to fail.
13545 * hostnamed now knows a new chassis type "embedded".
13547 * systemctl gained a new "edit" command. When used on a unit
13548 file, this allows extending unit files with .d/ drop-in
13549 configuration snippets or editing the full file (after
13550 copying it from /usr/lib to /etc). This will invoke the
13551 user's editor (as configured with $EDITOR), and reload the
13552 modified configuration after editing.
13554 * "systemctl status" now shows the suggested enablement state
13555 for a unit, as declared in the (usually vendor-supplied)
13556 system preset files.
13558 * nss-myhostname will now resolve the single-label hostname
13559 "gateway" to the locally configured default IP routing
13560 gateways, ordered by their metrics. This assigns a stable
13561 name to the used gateways, regardless which ones are
13562 currently configured. Note that the name will only be
13563 resolved after all other name sources (if nss-myhostname is
13564 configured properly) and should hence not negatively impact
13565 systems that use the single-label hostname "gateway" in
13568 * systemd-inhibit now allows filtering by mode when listing
13571 * Scope and service units gained a new "Delegate" boolean
13572 property, which, when set, allows processes running inside the
13573 unit to further partition resources. This is primarily
13574 useful for systemd user instances as well as container
13577 * journald will now pick up audit messages directly from
13578 the kernel, and log them like any other log message. The
13579 audit fields are split up and fully indexed. This means that
13580 journalctl in many ways is now a (nicer!) alternative to
13581 ausearch, the traditional audit client. Note that this
13582 implements only a minimal audit client. If you want the
13583 special audit modes like reboot-on-log-overflow, please use
13584 the traditional auditd instead, which can be used in
13585 parallel to journald.
13587 * The ConditionSecurity= unit file option now understands the
13588 special string "audit" to check whether auditing is
13591 * journalctl gained two new commands --vacuum-size= and
13592 --vacuum-time= to delete old journal files until the
13593 remaining ones take up no more than the specified size on disk,
13594 or are not older than the specified time.
13596 * A new, native PPPoE library has been added to sd-network,
13597 systemd's library of light-weight networking protocols. This
13598 library will be used in a future version of networkd to
13599 enable PPPoE communication without an external pppd daemon.
13601 * The busctl tool now understands a new "capture" verb that
13602 works similar to "monitor", but writes a packet capture
13603 trace to STDOUT that can be redirected to a file which is
13604 compatible with libcap's capture file format. This can then
13605 be loaded in Wireshark and similar tools to inspect bus
13608 * The busctl tool now understands a new "tree" verb that shows
13609 the object trees of a specific service on the bus, or of all
13612 * The busctl tool now understands a new "introspect" verb that
13613 shows all interfaces and members of objects on the bus,
13614 including their signature and values. This is particularly
13615 useful to get more information about bus objects shown by
13616 the new "busctl tree" command.
13618 * The busctl tool now understands new verbs "call",
13619 "set-property" and "get-property" for invoking bus method
13620 calls, setting and getting bus object properties in a
13623 * busctl gained a new --augment-creds= argument that controls
13624 whether the tool shall augment credential information it
13625 gets from the bus with data from /proc, in a possibly
13628 * nspawn's --link-journal= switch gained two new values
13629 "try-guest" and "try-host" that work like "guest" and
13630 "host", but do not fail if the host has no persistent
13631 journaling enabled. -j is now equivalent to
13632 --link-journal=try-guest.
13634 * macvlan network devices created by nspawn will now have
13635 stable MAC addresses.
13637 * A new SmackProcessLabel= unit setting has been added, which
13638 controls the SMACK security label processes forked off by
13639 the respective unit shall use.
13641 * If compiled with --enable-xkbcommon, systemd-localed will
13642 verify x11 keymap settings by compiling the given keymap. It
13643 will spew out warnings if the compilation fails. This
13644 requires libxkbcommon to be installed.
13646 * When a coredump is collected, a larger number of metadata
13647 fields is now collected and included in the journal records
13648 created for it. More specifically, control group membership,
13649 environment variables, memory maps, working directory,
13650 chroot directory, /proc/$PID/status, and a list of open file
13651 descriptors is now stored in the log entry.
13653 * The udev hwdb now contains DPI information for mice. For
13656 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
13658 * All systemd programs that read standalone configuration
13659 files in /etc now also support a corresponding series of
13660 .conf.d configuration directories in /etc/, /run/,
13661 /usr/local/lib/, /usr/lib/, and (if configured with
13662 --enable-split-usr) /lib/. In particular, the following
13663 configuration files now have corresponding configuration
13664 directories: system.conf user.conf, logind.conf,
13665 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
13666 resolved.conf, timesyncd.conf, journal-remote.conf, and
13667 journal-upload.conf. Note that distributions should use the
13668 configuration directories in /usr/lib/; the directories in
13669 /etc/ are reserved for the system administrator.
13671 * systemd-rfkill will no longer take the rfkill device name
13672 into account when storing rfkill state on disk, as the name
13673 might be dynamically assigned and not stable. Instead, the
13674 ID_PATH udev variable combined with the rfkill type (wlan,
13675 bluetooth, …) is used.
13677 * A new service systemd-machine-id-commit.service has been
13678 added. When used on systems where /etc is read-only during
13679 boot, and /etc/machine-id is not initialized (but an empty
13680 file), this service will copy the temporary machine ID
13681 created as replacement into /etc after the system is fully
13682 booted up. This is useful for systems that are freshly
13683 installed with a non-initialized machine ID, but should get
13684 a fixed machine ID for subsequent boots.
13686 * networkd's .netdev files now provide a large set of
13687 configuration parameters for VXLAN devices. Similarly, the
13688 bridge port cost parameter is now configurable in .network
13689 files. There's also new support for configuring IP source
13690 routing. networkd .link files gained support for a new
13691 OriginalName= match that is useful to match against the
13692 original interface name the kernel assigned. .network files
13693 may include MTU= and MACAddress= fields for altering the MTU
13694 and MAC address while being connected to a specific network
13697 * The LUKS logic gained supported for configuring
13698 UUID-specific key files. There's also new support for naming
13699 LUKS device from the kernel command line, using the new
13700 luks.name= argument.
13702 * Timer units may now be transiently created via the bus API
13703 (this was previously already available for scope and service
13704 units). In addition it is now possible to create multiple
13705 transient units at the same time with a single bus call. The
13706 "systemd-run" tool has been updated to make use of this for
13707 running commands on a specified time, in at(1)-style.
13709 * tmpfiles gained support for "t" lines, for assigning
13710 extended attributes to files. Among other uses this may be
13711 used to assign SMACK labels to files.
13713 Contributions from: Alin Rauta, Alison Chaiken, Andrej
13714 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
13715 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
13716 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
13717 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
13718 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
13719 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
13720 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
13721 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
13722 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
13723 Michael Biebl, Michael Chapman, Michael Marineau, Michal
13724 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
13725 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
13726 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
13727 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
13728 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
13729 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
13730 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
13732 — Berlin, 2014-12-10
13736 * journalctl gained the new options -t/--identifier= to match
13737 on the syslog identifier (aka "tag"), as well as --utc to
13738 show log timestamps in the UTC timezone. journalctl now also
13739 accepts -n/--lines=all to disable line capping in a pager.
13741 * journalctl gained a new switch, --flush, that synchronously
13742 flushes logs from /run/log/journal to /var/log/journal if
13743 persistent storage is enabled. systemd-journal-flush.service
13744 now waits until the operation is complete.
13746 * Services can notify the manager before they start a reload
13747 (by sending RELOADING=1) or shutdown (by sending
13748 STOPPING=1). This allows the manager to track and show the
13749 internal state of daemons and closes a race condition when
13750 the process is still running but has closed its D-Bus
13753 * Services with Type=oneshot do not have to have any ExecStart
13756 * User units are now loaded also from
13757 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
13758 /run/systemd/user directory that was already previously
13759 supported, but is under the control of the user.
13761 * Job timeouts (i.e. timeouts on the time a job that is
13762 queued stays in the run queue) can now optionally result in
13763 immediate reboot or power-off actions (JobTimeoutAction= and
13764 JobTimeoutRebootArgument=). This is useful on ".target"
13765 units, to limit the maximum time a target remains
13766 undispatched in the run queue, and to trigger an emergency
13767 operation in such a case. This is now used by default to
13768 turn off the system if boot-up (as defined by everything in
13769 basic.target) hangs and does not complete for at least
13770 15min. Also, if power-off or reboot hang for at least 30min
13771 an immediate power-off/reboot operation is triggered. This
13772 functionality is particularly useful to increase reliability
13773 on embedded devices, but also on laptops which might
13774 accidentally get powered on when carried in a backpack and
13775 whose boot stays stuck in a hard disk encryption passphrase
13778 * systemd-logind can be configured to also handle lid switch
13779 events even when the machine is docked or multiple displays
13780 are attached (HandleLidSwitchDocked= option).
13782 * A helper binary and a service have been added which can be
13783 used to resume from hibernation in the initramfs. A
13784 generator will parse the resume= option on the kernel
13785 command line to trigger resume.
13787 * A user console daemon systemd-consoled has been
13788 added. Currently, it is a preview, and will so far open a
13789 single terminal on each session of the user marked as
13790 Desktop=systemd-console.
13792 * Route metrics can be specified for DHCP routes added by
13795 * The SELinux context of socket-activated services can be set
13796 from the information provided by the networking stack
13797 (SELinuxContextFromNet= option).
13799 * Userspace firmware loading support has been removed and
13800 the minimum supported kernel version is thus bumped to 3.7.
13802 * Timeout for udev workers has been increased from 1 to 3
13803 minutes, but a warning will be printed after 1 minute to
13804 help diagnose kernel modules that take a long time to load.
13806 * Udev rules can now remove tags on devices with TAG-="foobar".
13808 * systemd's readahead implementation has been removed. In many
13809 circumstances it didn't give expected benefits even for
13810 rotational disk drives and was becoming less relevant in the
13811 age of SSDs. As none of the developers has been using
13812 rotating media anymore, and nobody stepped up to actively
13813 maintain this component of systemd it has now been removed.
13815 * Swap units can use Options= to specify discard options.
13816 Discard options specified for swaps in /etc/fstab are now
13819 * Docker containers are now detected as a separate type of
13822 * The Password Agent protocol gained support for queries where
13823 the user input is shown, useful e.g. for user names.
13824 systemd-ask-password gained a new --echo option to turn that
13827 * The default sysctl.d/ snippets will now set:
13829 net.core.default_qdisc = fq_codel
13831 This selects Fair Queuing Controlled Delay as the default
13832 queuing discipline for network interfaces. fq_codel helps
13833 fight the network bufferbloat problem. It is believed to be
13834 a good default with no tuning required for most workloads.
13835 Downstream distributions may override this choice. On 10Gbit
13836 servers that do not do forwarding, "fq" may perform better.
13837 Systems without a good clocksource should use "pfifo_fast".
13839 * If kdbus is enabled during build a new option BusPolicy= is
13840 available for service units, that allows locking all service
13841 processes into a stricter bus policy, in order to limit
13842 access to various bus services, or even hide most of them
13843 from the service's view entirely.
13845 * networkctl will now show the .network and .link file
13846 networkd has applied to a specific interface.
13848 * sd-login gained a new API call sd_session_get_desktop() to
13849 query which desktop environment has been selected for a
13852 * UNIX utmp support is now compile-time optional to support
13853 legacy-free systems.
13855 * systemctl gained two new commands "add-wants" and
13856 "add-requires" for pulling in units from specific targets
13859 * If the word "rescue" is specified on the kernel command line
13860 the system will now boot into rescue mode (aka
13861 rescue.target), which was previously available only by
13862 specifying "1" or "systemd.unit=rescue.target" on the kernel
13863 command line. This new kernel command line option nicely
13864 mirrors the already existing "emergency" kernel command line
13867 * New kernel command line options mount.usr=, mount.usrflags=,
13868 mount.usrfstype= have been added that match root=, rootflags=,
13869 rootfstype= but allow mounting a specific file system to
13872 * The $NOTIFY_SOCKET is now also passed to control processes of
13873 services, not only the main process.
13875 * This version reenables support for fsck's -l switch. This
13876 means at least version v2.25 of util-linux is required for
13877 operation, otherwise dead-locks on device nodes may
13878 occur. Again: you need to update util-linux to at least
13879 v2.25 when updating systemd to v217.
13881 * The "multi-seat-x" tool has been removed from systemd, as
13882 its functionality has been integrated into X servers 1.16,
13883 and the tool is hence redundant. It is recommended to update
13884 display managers invoking this tool to simply invoke X
13885 directly from now on, again.
13887 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
13888 message flag has been added for all of systemd's polkit
13889 authenticated method calls has been added. In particular this
13890 now allows optional interactive authorization via polkit for
13891 many of PID1's privileged operations such as unit file
13892 enabling and disabling.
13894 * "udevadm hwdb --update" learnt a new switch "--usr" for
13895 placing the rebuilt hardware database in /usr instead of
13896 /etc. When used only hardware database entries stored in
13897 /usr will be used, and any user database entries in /etc are
13898 ignored. This functionality is useful for vendors to ship a
13899 pre-built database on systems where local configuration is
13900 unnecessary or unlikely.
13902 * Calendar time specifications in .timer units now also
13903 understand the strings "semi-annually", "quarterly" and
13904 "minutely" as shortcuts (in addition to the preexisting
13905 "annually", "hourly", …).
13907 * systemd-tmpfiles will now correctly create files in /dev
13908 at boot which are marked for creation only at boot. It is
13909 recommended to always create static device nodes with 'c!'
13910 and 'b!', so that they are created only at boot and not
13911 overwritten at runtime.
13913 * When the watchdog logic is used for a service (WatchdogSec=)
13914 and the watchdog timeout is hit the service will now be
13915 terminated with SIGABRT (instead of just SIGTERM), in order
13916 to make sure a proper coredump and backtrace is
13917 generated. This ensures that hanging services will result in
13918 similar coredump/backtrace behaviour as services that hit a
13919 segmentation fault.
13921 Contributions from: Andreas Henriksson, Andrei Borzenkov,
13922 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
13923 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
13924 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
13925 Herrmann, David Sommerseth, David Strauss, Emil Renner
13926 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
13927 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
13928 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
13929 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
13930 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
13931 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
13932 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
13933 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
13934 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
13935 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
13936 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
13937 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
13938 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
13939 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
13940 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
13941 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
13944 — Berlin, 2014-10-28
13948 * timedated no longer reads NTP implementation unit names from
13949 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
13950 implementations should add a
13952 Conflicts=systemd-timesyncd.service
13954 to their unit files to take over and replace systemd's NTP
13955 default functionality.
13957 * systemd-sysusers gained a new line type "r" for configuring
13958 which UID/GID ranges to allocate system users/groups
13959 from. Lines of type "u" may now add an additional column
13960 that specifies the home directory for the system user to be
13961 created. Also, systemd-sysusers may now optionally read user
13962 information from STDIN instead of a file. This is useful for
13963 invoking it from RPM preinst scriptlets that need to create
13964 users before the first RPM file is installed since these
13965 files might need to be owned by them. A new
13966 %sysusers_create_inline RPM macro has been introduced to do
13967 just that. systemd-sysusers now updates the shadow files as
13968 well as the user/group databases, which should enhance
13969 compatibility with certain tools like grpck.
13971 * A number of bus APIs of PID 1 now optionally consult polkit to
13972 permit access for otherwise unprivileged clients under certain
13973 conditions. Note that this currently doesn't support
13974 interactive authentication yet, but this is expected to be
13975 added eventually, too.
13977 * /etc/machine-info now has new fields for configuring the
13978 deployment environment of the machine, as well as the
13979 location of the machine. hostnamectl has been updated with
13980 new command to update these fields.
13982 * systemd-timesyncd has been updated to automatically acquire
13983 NTP server information from systemd-networkd, which might
13984 have been discovered via DHCP.
13986 * systemd-resolved now includes a caching DNS stub resolver
13987 and a complete LLMNR name resolution implementation. A new
13988 NSS module "nss-resolve" has been added which can be used
13989 instead of glibc's own "nss-dns" to resolve hostnames via
13990 systemd-resolved. Hostnames, addresses and arbitrary RRs may
13991 be resolved via systemd-resolved D-Bus APIs. In contrast to
13992 the glibc internal resolver systemd-resolved is aware of
13993 multi-homed system, and keeps DNS server and caches separate
13994 and per-interface. Queries are sent simultaneously on all
13995 interfaces that have DNS servers configured, in order to
13996 properly handle VPNs and local LANs which might resolve
13997 separate sets of domain names. systemd-resolved may acquire
13998 DNS server information from systemd-networkd automatically,
13999 which in turn might have discovered them via DHCP. A tool
14000 "systemd-resolve-host" has been added that may be used to
14001 query the DNS logic in resolved. systemd-resolved implements
14002 IDNA and automatically uses IDNA or UTF-8 encoding depending
14003 on whether classic DNS or LLMNR is used as transport. In the
14004 next releases we intend to add a DNSSEC and mDNS/DNS-SD
14005 implementation to systemd-resolved.
14007 * A new NSS module nss-mymachines has been added, that
14008 automatically resolves the names of all local registered
14009 containers to their respective IP addresses.
14011 * A new client tool "networkctl" for systemd-networkd has been
14012 added. It currently is entirely passive and will query
14013 networking configuration from udev, rtnetlink and networkd,
14014 and present it to the user in a very friendly
14015 way. Eventually, we hope to extend it to become a full
14016 control utility for networkd.
14018 * .socket units gained a new DeferAcceptSec= setting that
14019 controls the kernels' TCP_DEFER_ACCEPT sockopt for
14020 TCP. Similarly, support for controlling TCP keep-alive
14021 settings has been added (KeepAliveTimeSec=,
14022 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
14023 turning off Nagle's algorithm on TCP has been added
14026 * logind learned a new session type "web", for use in projects
14027 like Cockpit which register web clients as PAM sessions.
14029 * timer units with at least one OnCalendar= setting will now
14030 be started only after time-sync.target has been
14031 reached. This way they will not elapse before the system
14032 clock has been corrected by a local NTP client or
14033 similar. This is particular useful on RTC-less embedded
14034 machines, that come up with an invalid system clock.
14036 * systemd-nspawn's --network-veth= switch should now result in
14037 stable MAC addresses for both the outer and the inner side
14040 * systemd-nspawn gained a new --volatile= switch for running
14041 container instances with /etc or /var unpopulated.
14043 * The kdbus client code has been updated to use the new Linux
14044 3.17 memfd subsystem instead of the old kdbus-specific one.
14046 * systemd-networkd's DHCP client and server now support
14047 FORCERENEW. There are also new configuration options to
14048 configure the vendor client identifier and broadcast mode
14051 * systemd will no longer inform the kernel about the current
14052 timezone, as this is necessarily incorrect and racy as the
14053 kernel has no understanding of DST and similar
14054 concepts. This hence means FAT timestamps will be always
14055 considered UTC, similar to what Android is already
14056 doing. Also, when the RTC is configured to the local time
14057 (rather than UTC) systemd will never synchronize back to it,
14058 as this might confuse Windows at a later boot.
14060 * systemd-analyze gained a new command "verify" for offline
14061 validation of unit files.
14063 * systemd-networkd gained support for a couple of additional
14064 settings for bonding networking setups. Also, the metric for
14065 statically configured routes may now be configured. For
14066 network interfaces where this is appropriate the peer IP
14067 address may now be configured.
14069 * systemd-networkd's DHCP client will no longer request
14070 broadcasting by default, as this tripped up some networks.
14071 For hardware where broadcast is required the feature should
14072 be switched back on using RequestBroadcast=yes.
14074 * systemd-networkd will now set up IPv4LL addresses (when
14075 enabled) even if DHCP is configured successfully.
14077 * udev will now default to respect network device names given
14078 by the kernel when the kernel indicates that these are
14079 predictable. This behavior can be tweaked by changing
14080 NamePolicy= in the relevant .link file.
14082 * A new library systemd-terminal has been added that
14083 implements full TTY stream parsing and rendering. This
14084 library is supposed to be used later on for implementing a
14085 full userspace VT subsystem, replacing the current kernel
14088 * A new tool systemd-journal-upload has been added to push
14089 journal data to a remote system running
14090 systemd-journal-remote.
14092 * journald will no longer forward all local data to another
14093 running syslog daemon. This change has been made because
14094 rsyslog (which appears to be the most commonly used syslog
14095 implementation these days) no longer makes use of this, and
14096 instead pulls the data out of the journal on its own. Since
14097 forwarding the messages to a non-existent syslog server is
14098 more expensive than we assumed we have now turned this
14099 off. If you run a syslog server that is not a recent rsyslog
14100 version, you have to turn this option on again
14101 (ForwardToSyslog= in journald.conf).
14103 * journald now optionally supports the LZ4 compressor for
14104 larger journal fields. This compressor should perform much
14105 better than XZ which was the previous default.
14107 * machinectl now shows the IP addresses of local containers,
14108 if it knows them, plus the interface name of the container.
14110 * A new tool "systemd-escape" has been added that makes it
14111 easy to escape strings to build unit names and similar.
14113 * sd_notify() messages may now include a new ERRNO= field
14114 which is parsed and collected by systemd and shown among the
14115 "systemctl status" output for a service.
14117 * A new component "systemd-firstboot" has been added that
14118 queries the most basic systemd information (timezone,
14119 hostname, root password) interactively on first
14120 boot. Alternatively it may also be used to provision these
14121 things offline on OS images installed into directories.
14123 * The default sysctl.d/ snippets will now set
14125 net.ipv4.conf.default.promote_secondaries=1
14127 This has the benefit of no flushing secondary IP addresses
14128 when primary addresses are removed.
14130 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
14131 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
14132 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
14133 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
14134 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
14135 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
14136 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
14137 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
14138 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
14139 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
14140 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
14141 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
14142 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
14143 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
14144 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
14146 — Berlin, 2014-08-19
14150 * A new tool systemd-sysusers has been added. This tool
14151 creates system users and groups in /etc/passwd and
14152 /etc/group, based on static declarative system user/group
14153 definitions in /usr/lib/sysusers.d/. This is useful to
14154 enable factory resets and volatile systems that boot up with
14155 an empty /etc directory, and thus need system users and
14156 groups created during early boot. systemd now also ships
14157 with two default sysusers.d/ files for the most basic
14158 users and groups systemd and the core operating system
14161 * A new tmpfiles snippet has been added that rebuilds the
14162 essential files in /etc on boot, should they be missing.
14164 * A directive for ensuring automatic clean-up of
14165 /var/cache/man/ has been removed from the default
14166 configuration. This line should now be shipped by the man
14167 implementation. The necessary change has been made to the
14168 man-db implementation. Note that you need to update your man
14169 implementation to one that ships this line, otherwise no
14170 automatic clean-up of /var/cache/man will take place.
14172 * A new condition ConditionNeedsUpdate= has been added that
14173 may conditionalize services to only run when /etc or /var
14174 are "older" than the vendor operating system resources in
14175 /usr. This is useful for reconstructing or updating /etc
14176 after an offline update of /usr or a factory reset, on the
14177 next reboot. Services that want to run once after such an
14178 update or reset should use this condition and order
14179 themselves before the new systemd-update-done.service, which
14180 will mark the two directories as fully updated. A number of
14181 service files have been added making use of this, to rebuild
14182 the udev hardware database, the journald message catalog and
14183 dynamic loader cache (ldconfig). The systemd-sysusers tool
14184 described above also makes use of this now. With this in
14185 place it is now possible to start up a minimal operating
14186 system with /etc empty cleanly. For more information on the
14187 concepts involved see this recent blog story:
14189 https://0pointer.de/blog/projects/stateless.html
14191 * A new system group "input" has been introduced, and all
14192 input device nodes get this group assigned. This is useful
14193 for system-level software to get access to input devices. It
14194 complements what is already done for "audio" and "video".
14196 * systemd-networkd learnt minimal DHCPv4 server support in
14197 addition to the existing DHCPv4 client support. It also
14198 learnt DHCPv6 client and IPv6 Router Solicitation client
14199 support. The DHCPv4 client gained support for static routes
14200 passed in from the server. Note that the [DHCPv4] section
14201 known in older systemd-networkd versions has been renamed to
14202 [DHCP] and is now also used by the DHCPv6 client. Existing
14203 .network files using settings of this section should be
14204 updated, though compatibility is maintained. Optionally, the
14205 client hostname may now be sent to the DHCP server.
14207 * networkd gained support for vxlan virtual networks as well
14208 as tun/tap and dummy devices.
14210 * networkd gained support for automatic allocation of address
14211 ranges for interfaces from a system-wide pool of
14212 addresses. This is useful for dynamically managing a large
14213 number of interfaces with a single network configuration
14214 file. In particular this is useful to easily assign
14215 appropriate IP addresses to the veth links of a large number
14216 of nspawn instances.
14218 * RPM macros for processing sysusers, sysctl and binfmt
14219 drop-in snippets at package installation time have been
14222 * The /etc/os-release file should now be placed in
14223 /usr/lib/os-release. The old location is automatically
14224 created as symlink. /usr/lib is the more appropriate
14225 location of this file, since it shall actually describe the
14226 vendor operating system shipped in /usr, and not the
14227 configuration stored in /etc.
14229 * .mount units gained a new boolean SloppyOptions= setting
14230 that maps to mount(8)'s -s option which enables permissive
14231 parsing of unknown mount options.
14233 * tmpfiles learnt a new "L+" directive which creates a symlink
14234 but (unlike "L") deletes a pre-existing file first, should
14235 it already exist and not already be the correct
14236 symlink. Similarly, "b+", "c+" and "p+" directives have been
14237 added as well, which create block and character devices, as
14238 well as fifos in the filesystem, possibly removing any
14239 pre-existing files of different types.
14241 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
14242 'argument' field (which so far specified the source to
14243 symlink/copy the files from) is now optional. If omitted the
14244 same file os copied from /usr/share/factory/ suffixed by the
14245 full destination path. This is useful for populating /etc
14246 with essential files, by copying them from vendor defaults
14247 shipped in /usr/share/factory/etc.
14249 * A new command "systemctl preset-all" has been added that
14250 applies the service preset settings to all installed unit
14251 files. A new switch --preset-mode= has been added that
14252 controls whether only enable or only disable operations
14255 * A new command "systemctl is-system-running" has been added
14256 that allows checking the overall state of the system, for
14257 example whether it is fully up and running.
14259 * When the system boots up with an empty /etc, the equivalent
14260 to "systemctl preset-all" is executed during early boot, to
14261 make sure all default services are enabled after a factory
14264 * systemd now contains a minimal preset file that enables the
14265 most basic services systemd ships by default.
14267 * Unit files' [Install] section gained a new DefaultInstance=
14268 field for defining the default instance to create if a
14269 template unit is enabled with no instance specified.
14271 * A new passive target cryptsetup-pre.target has been added
14272 that may be used by services that need to make they run and
14273 finish before the first LUKS cryptographic device is set up.
14275 * The /dev/loop-control and /dev/btrfs-control device nodes
14276 are now owned by the "disk" group by default, opening up
14277 access to this group.
14279 * systemd-coredump will now automatically generate a
14280 stack trace of all core dumps taking place on the system,
14281 based on elfutils' libdw library. This stack trace is logged
14284 * systemd-coredump may now optionally store coredumps directly
14285 on disk (in /var/lib/systemd/coredump, possibly compressed),
14286 instead of storing them unconditionally in the journal. This
14287 mode is the new default. A new configuration file
14288 /etc/systemd/coredump.conf has been added to configure this
14289 and other parameters of systemd-coredump.
14291 * coredumpctl gained a new "info" verb to show details about a
14292 specific coredump. A new switch "-1" has also been added
14293 that makes sure to only show information about the most
14294 recent entry instead of all entries. Also, as the tool is
14295 generally useful now the "systemd-" prefix of the binary
14296 name has been removed. Distributions that want to maintain
14297 compatibility with the old name should add a symlink from
14298 the old name to the new name.
14300 * journald's SplitMode= now defaults to "uid". This makes sure
14301 that unprivileged users can access their own coredumps with
14302 coredumpctl without restrictions.
14304 * New kernel command line options "systemd.wants=" (for
14305 pulling an additional unit during boot), "systemd.mask="
14306 (for masking a specific unit for the boot), and
14307 "systemd.debug-shell" (for enabling the debug shell on tty9)
14308 have been added. This is implemented in the new generator
14309 "systemd-debug-generator".
14311 * systemd-nspawn will now by default filter a couple of
14312 syscalls for containers, among them those required for
14313 kernel module loading, direct x86 IO port access, swap
14314 management, and kexec. Most importantly though
14315 open_by_handle_at() is now prohibited for containers,
14316 closing a hole similar to a recently discussed vulnerability
14317 in docker regarding access to files on file hierarchies the
14318 container should normally not have access to. Note that, for
14319 nspawn, we generally make no security claims anyway (and
14320 this is explicitly documented in the man page), so this is
14321 just a fix for one of the most obvious problems.
14323 * A new man page file-hierarchy(7) has been added that
14324 contains a minimized, modernized version of the file system
14325 layout systemd expects, similar in style to the FHS
14326 specification or hier(5). A new tool systemd-path(1) has
14327 been added to query many of these paths for the local
14330 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
14331 longer done. Since the directory now has a per-user size
14332 limit, and is cleaned on logout this appears unnecessary,
14333 in particular since this now brings the lifecycle of this
14334 directory closer in line with how IPC objects are handled.
14336 * systemd.pc now exports a number of additional directories,
14337 including $libdir (which is useful to identify the library
14338 path for the primary architecture of the system), and a
14339 couple of drop-in directories.
14341 * udev's predictable network interface names now use the dev_port
14342 sysfs attribute, introduced in linux 3.15 instead of dev_id to
14343 distinguish between ports of the same PCI function. dev_id should
14344 only be used for ports using the same HW address, hence the need
14347 * machined has been updated to export the OS version of a
14348 container (read from /etc/os-release and
14349 /usr/lib/os-release) on the bus. This is now shown in
14350 "machinectl status" for a machine.
14352 * A new service setting RestartForceExitStatus= has been
14353 added. If configured to a set of exit signals or process
14354 return values, the service will be restarted when the main
14355 daemon process exits with any of them, regardless of the
14358 * systemctl's -H switch for connecting to remote systemd
14359 machines has been extended so that it may be used to
14360 directly connect to a specific container on the
14361 host. "systemctl -H root@foobar:waldi" will now connect as
14362 user "root" to host "foobar", and then proceed directly to
14363 the container named "waldi". Note that currently you have to
14364 authenticate as user "root" for this to work, as entering
14365 containers is a privileged operation.
14367 Contributions from: Andreas Henriksson, Benjamin Steinwender,
14368 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
14369 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
14370 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
14371 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
14372 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
14373 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
14374 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
14375 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
14376 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
14377 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
14378 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
14380 — Berlin, 2014-07-03
14384 * As an experimental feature, udev now tries to lock the
14385 disk device node (flock(LOCK_SH|LOCK_NB)) while it
14386 executes events for the disk or any of its partitions.
14387 Applications like partitioning programs can lock the
14388 disk device node (flock(LOCK_EX)) and claim temporary
14389 device ownership that way; udev will entirely skip all event
14390 handling for this disk and its partitions. If the disk
14391 was opened for writing, the close will trigger a partition
14392 table rescan in udev's "watch" facility, and if needed
14393 synthesize "change" events for the disk and all its partitions.
14394 This is now unconditionally enabled, and if it turns out to
14395 cause major problems, we might turn it on only for specific
14396 devices, or might need to disable it entirely. Device Mapper
14397 devices are excluded from this logic.
14399 * We temporarily dropped the "-l" switch for fsck invocations,
14400 since they collide with the flock() logic above. util-linux
14401 upstream has been changed already to avoid this conflict,
14402 and we will re-add "-l" as soon as util-linux with this
14403 change has been released.
14405 * The dependency on libattr has been removed. Since a long
14406 time, the extended attribute calls have moved to glibc, and
14407 libattr is thus unnecessary.
14409 * Virtualization detection works without privileges now. This
14410 means the systemd-detect-virt binary no longer requires
14411 CAP_SYS_PTRACE file capabilities, and our daemons can run
14412 with fewer privileges.
14414 * systemd-networkd now runs under its own "systemd-network"
14415 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
14416 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
14417 loses the ability to write to files owned by root this way.
14419 * Similarly, systemd-resolved now runs under its own
14420 "systemd-resolve" user with no capabilities remaining.
14422 * Similarly, systemd-bus-proxyd now runs under its own
14423 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
14425 * systemd-networkd gained support for setting up "veth"
14426 virtual Ethernet devices for container connectivity, as well
14427 as GRE and VTI tunnels.
14429 * systemd-networkd will no longer automatically attempt to
14430 manually load kernel modules necessary for certain tunnel
14431 transports. Instead, it is assumed the kernel loads them
14432 automatically when required. This only works correctly on
14433 very new kernels. On older kernels, please consider adding
14434 the kernel modules to /etc/modules-load.d/ as a work-around.
14436 * The resolv.conf file systemd-resolved generates has been
14437 moved to /run/systemd/resolve/. If you have a symlink from
14438 /etc/resolv.conf, it might be necessary to correct it.
14440 * Two new service settings, ProtectHome= and ProtectSystem=,
14441 have been added. When enabled, they will make the user data
14442 (such as /home) inaccessible or read-only and the system
14443 (such as /usr) read-only, for specific services. This allows
14444 very light-weight per-service sandboxing to avoid
14445 modifications of user data or system files from
14446 services. These two new switches have been enabled for all
14447 of systemd's long-running services, where appropriate.
14449 * Socket units gained new SocketUser= and SocketGroup=
14450 settings to set the owner user and group of AF_UNIX sockets
14451 and FIFOs in the file system.
14453 * Socket units gained a new RemoveOnStop= setting. If enabled,
14454 all FIFOS and sockets in the file system will be removed
14455 when the specific socket unit is stopped.
14457 * Socket units gained a new Symlinks= setting. It takes a list
14458 of symlinks to create to file system sockets or FIFOs
14459 created by the specific Unix sockets. This is useful to
14460 manage symlinks to socket nodes with the same lifecycle as
14463 * The /dev/log socket and /dev/initctl FIFO have been moved to
14464 /run, and have been replaced by symlinks. This allows
14465 connecting to these facilities even if PrivateDevices=yes is
14466 used for a service (which makes /dev/log itself unavailable,
14467 but /run is left). This also has the benefit of ensuring
14468 that /dev only contains device nodes, directories and
14469 symlinks, and nothing else.
14471 * sd-daemon gained two new calls sd_pid_notify() and
14472 sd_pid_notifyf(). They are similar to sd_notify() and
14473 sd_notifyf(), but allow overriding of the source PID of
14474 notification messages if permissions permit this. This is
14475 useful to send notify messages on behalf of a different
14476 process (for example, the parent process). The
14477 systemd-notify tool has been updated to make use of this
14478 when sending messages (so that notification messages now
14479 originate from the shell script invoking systemd-notify and
14480 not the systemd-notify process itself. This should minimize
14481 a race where systemd fails to associate notification
14482 messages to services when the originating process already
14485 * A new "on-abnormal" setting for Restart= has been added. If
14486 set, it will result in automatic restarts on all "abnormal"
14487 reasons for a process to exit, which includes unclean
14488 signals, core dumps, timeouts and watchdog timeouts, but
14489 does not include clean and unclean exit codes or clean
14490 signals. Restart=on-abnormal is an alternative for
14491 Restart=on-failure for services that shall be able to
14492 terminate and avoid restarts on certain errors, by
14493 indicating so with an unclean exit code. Restart=on-failure
14494 or Restart=on-abnormal is now the recommended setting for
14495 all long-running services.
14497 * If the InaccessibleDirectories= service setting points to a
14498 mount point (or if there are any submounts contained within
14499 it), it is now attempted to completely unmount it, to make
14500 the file systems truly unavailable for the respective
14503 * The ReadOnlyDirectories= service setting and
14504 systemd-nspawn's --read-only parameter are now recursively
14505 applied to all submounts, too.
14507 * Mount units may now be created transiently via the bus APIs.
14509 * The support for SysV and LSB init scripts has been removed
14510 from the systemd daemon itself. Instead, it is now
14511 implemented as a generator that creates native systemd units
14512 from these scripts when needed. This enables us to remove a
14513 substantial amount of legacy code from PID 1, following the
14514 fact that many distributions only ship a very small number
14515 of LSB/SysV init scripts nowadays.
14517 * Privileged Xen (dom0) domains are not considered
14518 virtualization anymore by the virtualization detection
14519 logic. After all, they generally have unrestricted access to
14520 the hardware and usually are used to manage the unprivileged
14523 * systemd-tmpfiles gained a new "C" line type, for copying
14524 files or entire directories.
14526 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
14527 lines. So far, they have been non-globbing versions of the
14528 latter, and have thus been redundant. In future, it is
14529 recommended to only use "z". "m" has hence been removed
14530 from the documentation, even though it stays supported.
14532 * A tmpfiles snippet to recreate the most basic structure in
14533 /var has been added. This is enough to create the /var/run →
14534 /run symlink and create a couple of structural
14535 directories. This allows systems to boot up with an empty or
14536 volatile /var. Of course, while with this change, the core OS
14537 now is capable with dealing with a volatile /var, not all
14538 user services are ready for it. However, we hope that sooner
14539 or later, many service daemons will be changed upstream so
14540 that they are able to automatically create their necessary
14541 directories in /var at boot, should they be missing. This is
14542 the first step to allow state-less systems that only require
14543 the vendor image for /usr to boot.
14545 * systemd-nspawn has gained a new --tmpfs= switch to mount an
14546 empty tmpfs instance to a specific directory. This is
14547 particularly useful for making use of the automatic
14548 reconstruction of /var (see above), by passing --tmpfs=/var.
14550 * Access modes specified in tmpfiles snippets may now be
14551 prefixed with "~", which indicates that they shall be masked
14552 by whether the existing file or directory is currently
14553 writable, readable or executable at all. Also, if specified,
14554 the sgid/suid/sticky bits will be masked for all
14557 * A new passive target unit "network-pre.target" has been
14558 added which is useful for services that shall run before any
14559 network is configured, for example firewall scripts.
14561 * The "floppy" group that previously owned the /dev/fd*
14562 devices is no longer used. The "disk" group is now used
14563 instead. Distributions should probably deprecate usage of
14566 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
14567 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
14568 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
14569 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
14570 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
14571 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
14574 — Berlin, 2014-06-11
14578 * A new "systemd-timesyncd" daemon has been added for
14579 synchronizing the system clock across the network. It
14580 implements an SNTP client. In contrast to NTP
14581 implementations such as chrony or the NTP reference server,
14582 this only implements a client side, and does not bother with
14583 the full NTP complexity, focusing only on querying time from
14584 one remote server and synchronizing the local clock to
14585 it. Unless you intend to serve NTP to networked clients or
14586 want to connect to local hardware clocks, this simple NTP
14587 client should be more than appropriate for most
14588 installations. The daemon runs with minimal privileges, and
14589 has been hooked up with networkd to only operate when
14590 network connectivity is available. The daemon saves the
14591 current clock to disk every time a new NTP sync has been
14592 acquired, and uses this to possibly correct the system clock
14593 early at bootup, in order to accommodate for systems that
14594 lack an RTC such as the Raspberry Pi and embedded devices,
14595 and to make sure that time monotonically progresses on these
14596 systems, even if it is not always correct. To make use of
14597 this daemon, a new system user and group "systemd-timesync"
14598 needs to be created on installation of systemd.
14600 * The queue "seqnum" interface of libudev has been disabled, as
14601 it was generally incompatible with device namespacing as
14602 sequence numbers of devices go "missing" if the devices are
14603 part of a different namespace.
14605 * "systemctl list-timers" and "systemctl list-sockets" gained
14606 a --recursive switch for showing units of these types also
14607 for all local containers, similar in style to the already
14608 supported --recursive switch for "systemctl list-units".
14610 * A new RebootArgument= setting has been added for service
14611 units, which may be used to specify a kernel reboot argument
14612 to use when triggering reboots with StartLimitAction=.
14614 * A new FailureAction= setting has been added for service
14615 units which may be used to specify an operation to trigger
14616 when a service fails. This works similarly to
14617 StartLimitAction=, but unlike it, controls what is done
14618 immediately rather than only after several attempts to
14619 restart the service in question.
14621 * hostnamed got updated to also expose the kernel name,
14622 release, and version on the bus. This is useful for
14623 executing commands like hostnamectl with the -H switch.
14624 systemd-analyze makes use of this to properly display
14625 details when running non-locally.
14627 * The bootchart tool can now show cgroup information in the
14628 graphs it generates.
14630 * The CFS CPU quota cgroup attribute is now exposed for
14631 services. The new CPUQuota= switch has been added for this
14632 which takes a percentage value. Setting this will have the
14633 result that a service may never get more CPU time than the
14634 specified percentage, even if the machine is otherwise idle.
14636 * systemd-networkd learned IPIP and SIT tunnel support.
14638 * LSB init scripts exposing a dependency on $network will now
14639 get a dependency on network-online.target rather than simply
14640 network.target. This should bring LSB handling closer to
14641 what it was on SysV systems.
14643 * A new fsck.repair= kernel option has been added to control
14644 how fsck shall deal with unclean file systems at boot.
14646 * The (.ini) configuration file parser will now silently ignore
14647 sections whose names begin with "X-". This may be used to maintain
14648 application-specific extension sections in unit files.
14650 * machined gained a new API to query the IP addresses of
14651 registered containers. "machinectl status" has been updated
14652 to show these addresses in its output.
14654 * A new call sd_uid_get_display() has been added to the
14655 sd-login APIs for querying the "primary" session of a
14656 user. The "primary" session of the user is elected from the
14657 user's sessions and generally a graphical session is
14658 preferred over a text one.
14660 * A minimal systemd-resolved daemon has been added. It
14661 currently simply acts as a companion to systemd-networkd and
14662 manages resolv.conf based on per-interface DNS
14663 configuration, possibly supplied via DHCP. In the long run
14664 we hope to extend this into a local DNSSEC enabled DNS and
14667 * The systemd-networkd-wait-online tool is now enabled by
14668 default. It will delay network-online.target until a network
14669 connection has been configured. The tool primarily integrates
14670 with networkd, but will also make a best effort to make sense
14671 of network configuration performed in some other way.
14673 * Two new service options StartupCPUShares= and
14674 StartupBlockIOWeight= have been added that work similarly to
14675 CPUShares= and BlockIOWeight= however only apply during
14676 system startup. This is useful to prioritize certain services
14677 differently during bootup than during normal runtime.
14679 * hostnamed has been changed to prefer the statically
14680 configured hostname in /etc/hostname (unless set to
14681 'localhost' or empty) over any dynamic one supplied by
14682 dhcp. With this change, the rules for picking the hostname
14683 match more closely the rules of other configuration settings
14684 where the local administrator's configuration in /etc always
14685 overrides any other settings.
14687 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
14688 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
14689 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
14690 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
14691 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
14692 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
14693 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
14694 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
14695 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
14696 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
14697 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
14698 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
14699 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
14700 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
14701 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
14702 Lindskog, WaLyong Cho, Will Woods, Zbigniew
14705 — Beijing, 2014-05-28
14709 * When restoring the screen brightness at boot, stay away from
14710 the darkest setting or from the lowest 5% of the available
14711 range, depending on which is the larger value of both. This
14712 should effectively protect the user from rebooting into a
14713 black screen, should the brightness have been set to minimum
14716 * sd-login gained a new sd_machine_get_class() call to
14717 determine the class ("vm" or "container") of a machine
14718 registered with machined.
14720 * sd-login gained new calls
14721 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
14722 to query the identity of the peer of a local AF_UNIX
14723 connection. They operate similarly to their sd_pid_get_xyz()
14726 * PID 1 will now maintain a system-wide system state engine
14727 with the states "starting", "running", "degraded",
14728 "maintenance", "stopping". These states are bound to system
14729 startup, normal runtime, runtime with at least one failed
14730 service, rescue/emergency mode and system shutdown. This
14731 state is shown in the "systemctl status" output when no unit
14732 name is passed. It is useful to determine system state, in
14733 particularly when doing so for many systems or containers at
14736 * A new command "list-machines" has been added to "systemctl"
14737 that lists all local OS containers and shows their system
14738 state (see above), if systemd runs inside of them.
14740 * systemctl gained a new "-r" switch to recursively enumerate
14741 units on all local containers, when used with the
14742 "list-unit" command (which is the default one that is
14743 executed when no parameters are specified).
14745 * The GPT automatic partition discovery logic will now honour
14746 two GPT partition flags: one may be set on a partition to
14747 cause it to be mounted read-only, and the other may be set
14748 on a partition to ignore it during automatic discovery.
14750 * Two new GPT type UUIDs have been added for automatic root
14751 partition discovery, for 32-bit and 64-bit ARM. This is not
14752 particularly useful for discovering the root directory on
14753 these architectures during bare-metal boots (since UEFI is
14754 not common there), but still very useful to allow booting of
14755 ARM disk images in nspawn with the -i option.
14757 * MAC addresses of interfaces created with nspawn's
14758 --network-interface= switch will now be generated from the
14759 machine name, and thus be stable between multiple invocations
14762 * logind will now automatically remove all IPC objects owned
14763 by a user if she or he fully logs out. This makes sure that
14764 users who are logged out cannot continue to consume IPC
14765 resources. This covers SysV memory, semaphores and message
14766 queues as well as POSIX shared memory and message
14767 queues. Traditionally, SysV and POSIX IPC had no lifecycle
14768 limits. With this functionality, that is corrected. This may
14769 be turned off by using the RemoveIPC= switch of logind.conf.
14771 * The systemd-machine-id-setup and tmpfiles tools gained a
14772 --root= switch to operate on a specific root directory,
14775 * journald can now forward logged messages to the TTYs of all
14776 logged in users ("wall"). This is the default for all
14777 emergency messages now.
14779 * A new tool systemd-journal-remote has been added to stream
14780 journal log messages across the network.
14782 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
14783 controller trees are mounted into it. Note that the
14784 directories mounted beneath it are not read-only. This is a
14785 security measure and is particularly useful because glibc
14786 actually includes a search logic to pick any tmpfs it can
14787 find to implement shm_open() if /dev/shm is not available
14788 (which it might very well be in namespaced setups).
14790 * machinectl gained a new "poweroff" command to cleanly power
14791 down a local OS container.
14793 * The PrivateDevices= unit file setting will now also drop the
14794 CAP_MKNOD capability from the capability bound set, and
14795 imply DevicePolicy=closed.
14797 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
14798 comprehensively on all long-running systemd services where
14799 this is appropriate.
14801 * systemd-udevd will now run in a disassociated mount
14802 namespace. To mount directories from udev rules, make sure to
14803 pull in mount units via SYSTEMD_WANTS properties.
14805 * The kdbus support gained support for uploading policy into
14806 the kernel. sd-bus gained support for creating "monitoring"
14807 connections that can eavesdrop into all bus communication
14808 for debugging purposes.
14810 * Timestamps may now be specified in seconds since the UNIX
14811 epoch Jan 1st, 1970 by specifying "@" followed by the value
14814 * Native tcpwrap support in systemd has been removed. tcpwrap
14815 is old code, not really maintained anymore and has serious
14816 shortcomings, and better options such as firewalls
14817 exist. For setups that require tcpwrap usage, please
14818 consider invoking your socket-activated service via tcpd,
14819 like on traditional inetd.
14821 * A new system.conf configuration option
14822 DefaultTimerAccuracySec= has been added that controls the
14823 default AccuracySec= setting of .timer units.
14825 * Timer units gained a new WakeSystem= switch. If enabled,
14826 timers configured this way will cause the system to resume
14827 from system suspend (if the system supports that, which most
14830 * Timer units gained a new Persistent= switch. If enabled,
14831 timers configured this way will save to disk when they have
14832 been last triggered. This information is then used on next
14833 reboot to possible execute overdue timer events, that
14834 could not take place because the system was powered off.
14835 This enables simple anacron-like behaviour for timer units.
14837 * systemctl's "list-timers" will now also list the time a
14838 timer unit was last triggered in addition to the next time
14839 it will be triggered.
14841 * systemd-networkd will now assign predictable IPv4LL
14842 addresses to its local interfaces.
14844 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
14845 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
14846 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
14847 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
14848 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
14849 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
14850 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
14851 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
14854 — Berlin, 2014-03-25
14858 * A new unit file setting RestrictAddressFamilies= has been
14859 added to restrict which socket address families unit
14860 processes gain access to. This takes address family names
14861 like "AF_INET" or "AF_UNIX", and is useful to minimize the
14862 attack surface of services via exotic protocol stacks. This
14863 is built on seccomp system call filters.
14865 * Two new unit file settings RuntimeDirectory= and
14866 RuntimeDirectoryMode= have been added that may be used to
14867 manage a per-daemon runtime directories below /run. This is
14868 an alternative for setting up directory permissions with
14869 tmpfiles snippets, and has the advantage that the runtime
14870 directory's lifetime is bound to the daemon runtime and that
14871 the daemon starts up with an empty directory each time. This
14872 is particularly useful when writing services that drop
14873 privileges using the User= or Group= setting.
14875 * The DeviceAllow= unit setting now supports globbing for
14876 matching against device group names.
14878 * The systemd configuration file system.conf gained new
14879 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
14880 DefaultMemoryAccounting= to globally turn on/off accounting
14881 for specific resources (cgroups) for all units. These
14882 settings may still be overridden individually in each unit
14885 * systemd-gpt-auto-generator is now able to discover /srv and
14886 root partitions in addition to /home and swap partitions. It
14887 also supports LUKS-encrypted partitions now. With this in
14888 place, automatic discovery of partitions to mount following
14889 the Discoverable Partitions Specification
14890 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
14891 is now a lot more complete. This allows booting without
14892 /etc/fstab and without root= on the kernel command line on
14893 systems prepared appropriately.
14895 * systemd-nspawn gained a new --image= switch which allows
14896 booting up disk images and Linux installations on any block
14897 device that follow the Discoverable Partitions Specification
14898 (see above). This means that installations made with
14899 appropriately updated installers may now be started and
14900 deployed using container managers, completely
14901 unmodified. (We hope that libvirt-lxc will add support for
14902 this feature soon, too.)
14904 * systemd-nspawn gained a new --network-macvlan= setting to
14905 set up a private macvlan interface for the
14906 container. Similarly, systemd-networkd gained a new
14907 Kind=macvlan setting in .netdev files.
14909 * systemd-networkd now supports configuring local addresses
14912 * A new tool systemd-network-wait-online has been added to
14913 synchronously wait for network connectivity using
14916 * The sd-bus.h bus API gained a new sd_bus_track object for
14917 tracking the lifecycle of bus peers. Note that sd-bus.h is
14918 still not a public API though (unless you specify
14919 --enable-kdbus on the configure command line, which however
14920 voids your warranty and you get no API stability guarantee).
14922 * The $XDG_RUNTIME_DIR runtime directories for each user are
14923 now individual tmpfs instances, which has the benefit of
14924 introducing separate pools for each user, with individual
14925 size limits, and thus making sure that unprivileged clients
14926 can no longer negatively impact the system or other users by
14927 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
14928 RuntimeDirectorySize= has been introduced that allows
14929 controlling the default size limit for all users. It
14930 defaults to 10% of the available physical memory. This is no
14931 replacement for quotas on tmpfs though (which the kernel
14932 still does not support), as /dev/shm and /tmp are still
14933 shared resources used by both the system and unprivileged
14936 * logind will now automatically turn off automatic suspending
14937 on laptop lid close when more than one display is
14938 connected. This was previously expected to be implemented
14939 individually in desktop environments (such as GNOME),
14940 however has been added to logind now, in order to fix a
14941 boot-time race where a desktop environment might not have
14942 been started yet and thus not been able to take an inhibitor
14943 lock at the time where logind already suspends the system
14944 due to a closed lid.
14946 * logind will now wait at least 30s after each system
14947 suspend/resume cycle, and 3min after system boot before
14948 suspending the system due to a closed laptop lid. This
14949 should give USB docking stations and similar enough time to
14950 be probed and configured after system resume and boot in
14951 order to then act as suspend blocker.
14953 * systemd-run gained a new --property= setting which allows
14954 initialization of resource control properties (and others)
14955 for the created scope or service unit. Example: "systemd-run
14956 --property=BlockIOWeight=10 updatedb" may be used to run
14957 updatedb at a low block IO scheduling weight.
14959 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
14960 now also work in --scope mode.
14962 * When systemd is compiled with kdbus support, basic support
14963 for enforced policies is now in place. (Note that enabling
14964 kdbus still voids your warranty and no API compatibility
14965 promises are made.)
14967 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
14968 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
14969 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
14970 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
14971 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
14972 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
14973 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
14974 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
14975 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
14976 Zbigniew Jędrzejewski-Szmek
14978 — Berlin, 2014-03-12
14982 * systemd will now relabel /dev after loading the SMACK policy
14983 according to SMACK rules.
14985 * A new unit file option AppArmorProfile= has been added to
14986 set the AppArmor profile for the processes of a unit.
14988 * A new condition check ConditionArchitecture= has been added
14989 to conditionalize units based on the system architecture, as
14990 reported by uname()'s "machine" field.
14992 * systemd-networkd now supports matching on the system
14993 virtualization, architecture, kernel command line, hostname
14996 * logind is now a lot more aggressive when suspending the
14997 machine due to a closed laptop lid. Instead of acting only
14998 on the lid close action, it will continuously watch the lid
14999 status and act on it. This is useful for laptops where the
15000 power button is on the outside of the chassis so that it can
15001 be reached without opening the lid (such as the Lenovo
15002 Yoga). On those machines, logind will now immediately
15003 re-suspend the machine if the power button has been
15004 accidentally pressed while the laptop was suspended and in a
15005 backpack or similar.
15007 * logind will now watch SW_DOCK switches and inhibit reaction
15008 to the lid switch if it is pressed. This means that logind
15009 will not suspend the machine anymore if the lid is closed
15010 and the system is docked, if the laptop supports SW_DOCK
15011 notifications via the input layer. Note that ACPI docking
15012 stations do not generate this currently. Also note that this
15013 logic is usually not fully sufficient and Desktop
15014 Environments should take a lid switch inhibitor lock when an
15015 external display is connected, as systemd will not watch
15018 * nspawn will now make use of the devices cgroup controller by
15019 default, and only permit creation of and access to the usual
15020 API device nodes like /dev/null or /dev/random, as well as
15021 access to (but not creation of) the pty devices.
15023 * We will now ship a default .network file for
15024 systemd-networkd that automatically configures DHCP for
15025 network interfaces created by nspawn's --network-veth or
15026 --network-bridge= switches.
15028 * systemd will now understand the usual M, K, G, T suffixes
15029 according to SI conventions (i.e. to the base 1000) when
15030 referring to throughput and hardware metrics. It will stay
15031 with IEC conventions (i.e. to the base 1024) for software
15032 metrics, according to what is customary according to
15033 Wikipedia. We explicitly document which base applies for
15034 each configuration option.
15036 * The DeviceAllow= setting in unit files now supports a syntax to
15037 allow-list an entire group of devices node majors at once, based on
15038 the /proc/devices listing. For example, with the string "char-pts",
15039 it is now possible to allow-list all current and future pseudo-TTYs
15042 * sd-event learned a new "post" event source. Event sources of
15043 this type are triggered by the dispatching of any event
15044 source of a type that is not "post". This is useful for
15045 implementing clean-up and check event sources that are
15046 triggered by other work being done in the program.
15048 * systemd-networkd is no longer statically enabled, but uses
15049 the usual [Install] sections so that it can be
15050 enabled/disabled using systemctl. It still is enabled by
15053 * When creating a veth interface pair with systemd-nspawn, the
15054 host side will now be prefixed with "vb-" if
15055 --network-bridge= is used, and with "ve-" if --network-veth
15056 is used. This way, it is easy to distinguish these cases on
15057 the host, for example to apply different configuration to
15058 them with systemd-networkd.
15060 * The compatibility libraries for libsystemd-journal.so,
15061 libsystem-id128.so, libsystemd-login.so and
15062 libsystemd-daemon.so do not make use of IFUNC
15063 anymore. Instead, we now build libsystemd.so multiple times
15064 under these alternative names. This means that the footprint
15065 is drastically increased, but given that these are
15066 transitional compatibility libraries, this should not matter
15067 much. This change has been made necessary to support the ARM
15068 platform for these compatibility libraries, as the ARM
15069 toolchain is not really at the same level as the toolchain
15070 for other architectures like x86 and does not support
15071 IFUNC. Please make sure to use --enable-compat-libs only
15072 during a transitional period!
15074 * The .include syntax has been deprecated and is not documented
15075 anymore. Drop-in files in .d directories should be used instead.
15077 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
15078 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
15079 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
15080 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
15081 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
15082 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
15083 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
15084 Zbigniew Jędrzejewski-Szmek
15086 — Berlin, 2014-02-24
15090 * A new component "systemd-networkd" has been added that can
15091 be used to configure local network interfaces statically or
15092 via DHCP. It is capable of bringing up bridges, VLANs, and
15093 bonding. Currently, no hook-ups for interactive network
15094 configuration are provided. Use this for your initrd,
15095 container, embedded, or server setup if you need a simple,
15096 yet powerful, network configuration solution. This
15097 configuration subsystem is quite nifty, as it allows wildcard
15098 hotplug matching in interfaces. For example, with a single
15099 configuration snippet, you can configure that all Ethernet
15100 interfaces showing up are automatically added to a bridge,
15101 or similar. It supports link-sensing and more.
15103 * A new tool "systemd-socket-proxyd" has been added which can
15104 act as a bidirectional proxy for TCP sockets. This is
15105 useful for adding socket activation support to services that
15106 do not actually support socket activation, including virtual
15107 machines and the like.
15109 * Add a new tool to save/restore rfkill state on
15112 * Save/restore state of keyboard backlights in addition to
15113 display backlights on shutdown/boot.
15115 * udev learned a new SECLABEL{} construct to label device
15116 nodes with a specific security label when they appear. For
15117 now, only SECLABEL{selinux} is supported, but the syntax is
15118 prepared for additional security frameworks.
15120 * udev gained a new scheme to configure link-level attributes
15121 from files in /etc/systemd/network/*.link. These files can
15122 match against MAC address, device path, driver name and type,
15123 and will apply attributes like the naming policy, link speed,
15124 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
15125 address assignment policy (randomized, …).
15127 * The configuration of network interface naming rules for
15128 "permanent interface names" has changed: a new NamePolicy=
15129 setting in the [Link] section of .link files determines the
15130 priority of possible naming schemes (onboard, slot, MAC,
15131 path). The default value of this setting is determined by
15132 /usr/lib/net/links/99-default.link. Old
15133 80-net-name-slot.rules udev configuration file has been
15134 removed, so local configuration overriding this file should
15135 be adapted to override 99-default.link instead.
15137 * When the User= switch is used in a unit file, also
15138 initialize $SHELL= based on the user database entry.
15140 * systemd no longer depends on libdbus. All communication is
15141 now done with sd-bus, systemd's low-level bus library
15144 * kdbus support has been added to PID 1 itself. When kdbus is
15145 enabled, this causes PID 1 to set up the system bus and
15146 enable support for a new ".busname" unit type that
15147 encapsulates bus name activation on kdbus. It works a little
15148 bit like ".socket" units, except for bus names. A new
15149 generator has been added that converts classic dbus1 service
15150 activation files automatically into native systemd .busname
15151 and .service units.
15153 * sd-bus: add a light-weight vtable implementation that allows
15154 defining objects on the bus with a simple static const
15155 vtable array of its methods, signals and properties.
15157 * systemd will not generate or install static dbus
15158 introspection data anymore to /usr/share/dbus-1/interfaces,
15159 as the precise format of these files is unclear, and
15160 nothing makes use of it.
15162 * A proxy daemon is now provided to proxy clients connecting
15163 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
15164 compatibility with classic D-Bus.
15166 * A bus driver implementation has been added that supports the
15167 classic D-Bus bus driver calls on kdbus, also for
15168 compatibility purposes.
15170 * A new API "sd-event.h" has been added that implements a
15171 minimal event loop API built around epoll. It provides a
15172 couple of features that direct epoll usage is lacking:
15173 prioritization of events, scales to large numbers of timer
15174 events, per-event timer slack (accuracy), system-wide
15175 coalescing of timer events, exit handlers, watchdog
15176 supervision support using systemd's sd_notify() API, child
15179 * A new API "sd-rntl.h" has been added that provides an API
15180 around the route netlink interface of the kernel, similar in
15181 style to "sd-bus.h".
15183 * A new API "sd-dhcp-client.h" has been added that provides a
15184 small DHCPv4 client-side implementation. This is used by
15185 "systemd-networkd".
15187 * There is a new kernel command line option
15188 "systemd.restore_state=0|1". When set to "0", none of the
15189 systemd tools will restore saved runtime state to hardware
15190 devices. More specifically, the rfkill and backlight states
15193 * The FsckPassNo= compatibility option in mount/service units
15194 has been removed. The fstab generator will now add the
15195 necessary dependencies automatically, and does not require
15196 PID1's support for that anymore.
15198 * journalctl gained a new switch, --list-boots, that lists
15199 recent boots with their times and boot IDs.
15201 * The various tools like systemctl, loginctl, timedatectl,
15202 busctl, systemd-run, … have gained a new switch "-M" to
15203 connect to a specific, local OS container (as direct
15204 connection, without requiring SSH). This works on any
15205 container that is registered with machined, such as those
15206 created by libvirt-lxc or nspawn.
15208 * systemd-run and systemd-analyze also gained support for "-H"
15209 to connect to remote hosts via SSH. This is particularly
15210 useful for systemd-run because it enables queuing of jobs
15211 onto remote systems.
15213 * machinectl gained a new command "login" to open a getty
15214 login in any local container. This works with any container
15215 that is registered with machined (such as those created by
15216 libvirt-lxc or nspawn), and which runs systemd inside.
15218 * machinectl gained a new "reboot" command that may be used to
15219 trigger a reboot on a specific container that is registered
15220 with machined. This works on any container that runs an init
15221 system of some kind.
15223 * systemctl gained a new "list-timers" command to print a nice
15224 listing of installed timer units with the times they elapse
15227 * Alternative reboot() parameters may now be specified on the
15228 "systemctl reboot" command line and are passed to the
15229 reboot() system call.
15231 * systemctl gained a new --job-mode= switch to configure the
15232 mode to queue a job with. This is a more generic version of
15233 --fail, --irreversible, and --ignore-dependencies, which are
15234 still available but not advertised anymore.
15236 * /etc/systemd/system.conf gained new settings to configure
15237 various default timeouts of units, as well as the default
15238 start limit interval and burst. These may still be overridden
15241 * PID1 will now export on the bus profile data of the security
15242 policy upload process (such as the SELinux policy upload to
15245 * journald: when forwarding logs to the console, include
15246 timestamps (following the setting in
15247 /sys/module/printk/parameters/time).
15249 * OnCalendar= in timer units now understands the special
15250 strings "yearly" and "annually". (Both are equivalent)
15252 * The accuracy of timer units is now configurable with the new
15253 AccuracySec= setting. It defaults to 1min.
15255 * A new dependency type JoinsNamespaceOf= has been added that
15256 allows running two services within the same /tmp and network
15257 namespace, if PrivateNetwork= or PrivateTmp= are used.
15259 * A new command "cat" has been added to systemctl. It outputs
15260 the original unit file of a unit, and concatenates the
15261 contents of additional "drop-in" unit file snippets, so that
15262 the full configuration is shown.
15264 * systemctl now supports globbing on the various "list-xyz"
15265 commands, like "list-units" or "list-sockets", as well as on
15266 those commands which take multiple unit names.
15268 * journalctl's --unit= switch gained support for globbing.
15270 * All systemd daemons now make use of the watchdog logic so
15271 that systemd automatically notices when they hang.
15273 * If the $container_ttys environment variable is set,
15274 getty-generator will automatically spawn a getty for each
15275 listed tty. This is useful for container managers to request
15276 login gettys to be spawned on as many ttys as needed.
15278 * %h, %s, %U specifier support is not available anymore when
15279 used in unit files for PID 1. This is because NSS calls are
15280 not safe from PID 1. They stay available for --user
15281 instances of systemd, and as special case for the root user.
15283 * loginctl gained a new "--no-legend" switch to turn off output
15284 of the legend text.
15286 * The "sd-login.h" API gained three new calls:
15287 sd_session_is_remote(), sd_session_get_remote_user(),
15288 sd_session_get_remote_host() to query information about
15291 * The udev hardware database now also carries vendor/product
15292 information of SDIO devices.
15294 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
15295 determine whether watchdog notifications are requested by
15296 the system manager.
15298 * Socket-activated per-connection services now include a
15299 short description of the connection parameters in the
15302 * tmpfiles gained a new "--boot" option. When this is not used,
15303 only lines where the command character is not suffixed with
15304 "!" are executed. When this option is specified, those
15305 options are executed too. This partitions tmpfiles
15306 directives into those that can be safely executed at any
15307 time, and those which should be run only at boot (for
15308 example, a line that creates /run/nologin).
15310 * A new API "sd-resolve.h" has been added which provides a simple
15311 asynchronous wrapper around glibc NSS hostname resolution
15312 calls, such as getaddrinfo(). In contrast to glibc's
15313 getaddrinfo_a(), it does not use signals. In contrast to most
15314 other asynchronous name resolution libraries, this one does
15315 not reimplement DNS, but reuses NSS, so that alternate
15316 hostname resolution systems continue to work, such as mDNS,
15317 LDAP, etc. This API is based on libasyncns, but it has been
15318 cleaned up for inclusion in systemd.
15320 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
15321 "sd-daemon.h" are no longer found in individual libraries
15322 libsystemd-journal.so, libsystemd-login.so,
15323 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
15324 merged them into a single library, libsystemd.so, which
15325 provides all symbols. The reason for this is cyclic
15326 dependencies, as these libraries tend to use each other's
15327 symbols. So far, we have managed to workaround that by linking
15328 a copy of a good part of our code into each of these
15329 libraries again and again, which, however, makes certain
15330 things hard to do, like sharing static variables. Also, it
15331 substantially increases footprint. With this change, there
15332 is only one library for the basic APIs systemd
15333 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
15334 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
15335 library as well, however are subject to the --enable-kdbus
15336 switch (see below). Note that "sd-dhcp-client.h" is not part
15337 of this library (this is because it only consumes, never
15338 provides, services of/to other APIs). To make the transition
15339 easy from the separate libraries to the unified one, we
15340 provide the --enable-compat-libs compile-time switch which
15341 will generate stub libraries that are compatible with the
15342 old ones but redirect all calls to the new one.
15344 * All of the kdbus logic and the new APIs "sd-bus.h",
15345 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
15346 and "sd-utf8.h" are compile-time optional via the
15347 "--enable-kdbus" switch, and they are not compiled in by
15348 default. To make use of kdbus, you have to explicitly enable
15349 the switch. Note however, that neither the kernel nor the
15350 userspace API for all of this is considered stable yet. We
15351 want to maintain the freedom to still change the APIs for
15352 now. By specifying this build-time switch, you acknowledge
15353 that you are aware of the instability of the current
15356 * Also, note that while kdbus is pretty much complete,
15357 it lacks one thing: proper policy support. This means you
15358 can build a fully working system with all features; however,
15359 it will be highly insecure. Policy support will be added in
15360 one of the next releases, at the same time that we will
15361 declare the APIs stable.
15363 * When the kernel command line argument "kdbus" is specified,
15364 systemd will automatically load the kdbus.ko kernel module. At
15365 this stage of development, it is only useful for testing kdbus
15366 and should not be used in production. Note: if "--enable-kdbus"
15367 is specified, and the kdbus.ko kernel module is available, and
15368 "kdbus" is added to the kernel command line, the entire system
15369 runs with kdbus instead of dbus-daemon, with the above mentioned
15370 problem of missing the system policy enforcement. Also a future
15371 version of kdbus.ko or a newer systemd will not be compatible with
15372 each other, and will unlikely be able to boot the machine if only
15373 one of them is updated.
15375 * systemctl gained a new "import-environment" command which
15376 uploads the caller's environment (or parts thereof) into the
15377 service manager so that it is inherited by services started
15378 by the manager. This is useful to upload variables like
15379 $DISPLAY into the user service manager.
15381 * A new PrivateDevices= switch has been added to service units
15382 which allows running a service with a namespaced /dev
15383 directory that does not contain any device nodes for
15384 physical devices. More specifically, it only includes devices
15385 such as /dev/null, /dev/urandom, and /dev/zero which are API
15388 * logind has been extended to support behaviour like VT
15389 switching on seats that do not support a VT. This makes
15390 multi-session available on seats that are not the first seat
15391 (seat0), and on systems where kernel support for VTs has
15392 been disabled at compile-time.
15394 * If a process holds a delay lock for system sleep or shutdown
15395 and fails to release it in time, we will now log its
15396 identity. This makes it easier to identify processes that
15397 cause slow suspends or power-offs.
15399 * When parsing /etc/crypttab, support for a new key-slot=
15400 option as supported by Debian is added. It allows indicating
15401 which LUKS slot to use on disk, speeding up key loading.
15403 * The sd_journal_sendv() API call has been checked and
15404 officially declared to be async-signal-safe so that it may
15405 be invoked from signal handlers for logging purposes.
15407 * Boot-time status output is now enabled automatically after a
15408 short timeout if boot does not progress, in order to give
15409 the user an indication what she or he is waiting for.
15411 * The boot-time output has been improved to show how much time
15412 remains until jobs expire.
15414 * The KillMode= switch in service units gained a new possible
15415 value "mixed". If set, and the unit is shut down, then the
15416 initial SIGTERM signal is sent only to the main daemon
15417 process, while the following SIGKILL signal is sent to
15418 all remaining processes of the service.
15420 * When a scope unit is registered, a new property "Controller"
15421 may be set. If set to a valid bus name, systemd will send a
15422 RequestStop() signal to this name when it would like to shut
15423 down the scope. This may be used to hook manager logic into
15424 the shutdown logic of scope units. Also, scope units may now
15425 be put in a special "abandoned" state, in which case the
15426 manager process which created them takes no further
15427 responsibilities for it.
15429 * When reading unit files, systemd will now verify
15430 the access mode of these files, and warn about certain
15431 suspicious combinations. This has been added to make it
15432 easier to track down packaging bugs where unit files are
15433 marked executable or world-writable.
15435 * systemd-nspawn gained a new "--setenv=" switch to set
15436 container-wide environment variables. The similar option in
15437 systemd-activate was renamed from "--environment=" to
15438 "--setenv=" for consistency.
15440 * systemd-nspawn has been updated to create a new kdbus domain
15441 for each container that is invoked, thus allowing each
15442 container to have its own set of system and user buses,
15443 independent of the host.
15445 * systemd-nspawn gained a new --drop-capability= switch to run
15446 the container with less capabilities than the default. Both
15447 --drop-capability= and --capability= now take the special
15448 string "all" for dropping or keeping all capabilities.
15450 * systemd-nspawn gained new switches for executing containers
15451 with specific SELinux labels set.
15453 * systemd-nspawn gained a new --quiet switch to not generate
15454 any additional output but the container's own console
15457 * systemd-nspawn gained a new --share-system switch to run a
15458 container without PID namespacing enabled.
15460 * systemd-nspawn gained a new --register= switch to control
15461 whether the container is registered with systemd-machined or
15462 not. This is useful for containers that do not run full
15463 OS images, but only specific apps.
15465 * systemd-nspawn gained a new --keep-unit which may be used
15466 when invoked as the only program from a service unit, and
15467 results in registration of the unit service itself in
15468 systemd-machined, instead of a newly opened scope unit.
15470 * systemd-nspawn gained a new --network-interface= switch for
15471 moving arbitrary interfaces to the container. The new
15472 --network-veth switch creates a virtual Ethernet connection
15473 between host and container. The new --network-bridge=
15474 switch then allows assigning the host side of this virtual
15475 Ethernet connection to a bridge device.
15477 * systemd-nspawn gained a new --personality= switch for
15478 setting the kernel personality for the container. This is
15479 useful when running a 32-bit container on a 64-bit host. A
15480 similar option Personality= is now also available for service
15483 * logind will now also track a "Desktop" identifier for each
15484 session which encodes the desktop environment of it. This is
15485 useful for desktop environments that want to identify
15486 multiple running sessions of itself easily.
15488 * A new SELinuxContext= setting for service units has been
15489 added that allows setting a specific SELinux execution
15490 context for a service.
15492 * Most systemd client tools will now honour $SYSTEMD_LESS for
15493 settings of the "less" pager. By default, these tools will
15494 override $LESS to allow certain operations to work, such as
15495 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
15496 influence this logic.
15498 * systemd's "seccomp" hook-up has been changed to make use of
15499 the libseccomp library instead of using its own
15500 implementation. This has benefits for portability among
15503 * For usage together with SystemCallFilter=, a new
15504 SystemCallErrorNumber= setting has been introduced that
15505 allows configuration of a system error number to be returned
15506 on filtered system calls, instead of immediately killing the
15507 process. Also, SystemCallArchitectures= has been added to
15508 limit access to system calls of a particular architecture
15509 (in order to turn off support for unused secondary
15510 architectures). There is also a global
15511 SystemCallArchitectures= setting in system.conf now to turn
15512 off support for non-native system calls system-wide.
15514 * systemd requires a kernel with a working name_to_handle_at(),
15515 please see the kernel config requirements in the README file.
15517 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
15518 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
15519 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
15520 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
15521 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
15522 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
15523 Elia Pinto, Florian Weimer, George McCollister, Goffredo
15524 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
15525 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
15526 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
15527 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
15528 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
15529 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
15530 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
15531 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
15532 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
15533 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
15534 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
15535 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
15536 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
15537 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
15538 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
15539 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
15540 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
15542 — Berlin, 2014-02-20
15546 * logind has gained support for facilitating privileged input
15547 and drm device access for unprivileged clients. This work is
15548 useful to allow Wayland display servers (and similar
15549 programs, such as kmscon) to run under the user's ID and
15550 access input and drm devices which are normally
15551 protected. When this is used (and the kernel is new enough)
15552 logind will "mute" IO on the file descriptors passed to
15553 Wayland as long as it is in the background and "unmute" it
15554 if it returns into the foreground. This allows secure
15555 session switching without allowing background sessions to
15556 eavesdrop on input and display data. This also introduces
15557 session switching support if VT support is turned off in the
15558 kernel, and on seats that are not seat0.
15560 * A new kernel command line option luks.options= is understood
15561 now which allows specifying LUKS options for usage for LUKS
15562 encrypted partitions specified with luks.uuid=.
15564 * tmpfiles.d(5) snippets may now use specifier expansion in
15565 path names. More specifically %m, %b, %H, %v, are now
15566 replaced by the local machine id, boot id, hostname, and
15567 kernel version number.
15569 * A new tmpfiles.d(5) command "m" has been introduced which
15570 may be used to change the owner/group/access mode of a file
15571 or directory if it exists, but do nothing if it does not.
15573 * This release removes high-level support for the
15574 MemorySoftLimit= cgroup setting. The underlying kernel
15575 cgroup attribute memory.soft_limit= is currently badly
15576 designed and likely to be removed from the kernel API in its
15577 current form, hence we should not expose it for now.
15579 * The memory.use_hierarchy cgroup attribute is now enabled for
15580 all cgroups systemd creates in the memory cgroup
15581 hierarchy. This option is likely to be come the built-in
15582 default in the kernel anyway, and the non-hierarchical mode
15583 never made much sense in the intrinsically hierarchical
15586 * A new field _SYSTEMD_SLICE= is logged along with all journal
15587 messages containing the slice a message was generated
15588 from. This is useful to allow easy per-customer filtering of
15589 logs among other things.
15591 * systemd-journald will no longer adjust the group of journal
15592 files it creates to the "systemd-journal" group. Instead we
15593 rely on the journal directory to be owned by the
15594 "systemd-journal" group, and its setgid bit set, so that the
15595 kernel file system layer will automatically enforce that
15596 journal files inherit this group assignment. The reason for
15597 this change is that we cannot allow NSS look-ups from
15598 journald which would be necessary to resolve
15599 "systemd-journal" to a numeric GID, because this might
15600 create deadlocks if NSS involves synchronous queries to
15601 other daemons (such as nscd, or sssd) which in turn are
15602 logging clients of journald and might block on it, which
15603 would then dead lock. A tmpfiles.d(5) snippet included in
15604 systemd will make sure the setgid bit and group are
15605 properly set on the journal directory if it exists on every
15606 boot. However, we recommend adjusting it manually after
15607 upgrades too (or from RPM scriptlets), so that the change is
15608 not delayed until next reboot.
15610 * Backlight and random seed files in /var/lib/ have moved into
15611 the /var/lib/systemd/ directory, in order to centralize all
15612 systemd generated files in one directory.
15614 * Boot time performance measurements (as displayed by
15615 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
15616 performance information if that's available to determine how
15617 much time BIOS and boot loader initialization required. With
15618 a sufficiently new BIOS you hence no longer need to boot
15619 with Gummiboot to get access to such information.
15621 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
15622 Cristian Rodríguez, Dave Reisner, David Herrmann, David
15623 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
15624 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
15625 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
15626 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
15627 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15629 — Berlin, 2013-10-02
15633 * The Restart= option for services now understands a new
15634 on-watchdog setting, which will restart the service
15635 automatically if the service stops sending out watchdog keep
15636 alive messages (as configured with WatchdogSec=).
15638 * The getty generator (which is responsible for bringing up a
15639 getty on configured serial consoles) will no longer only
15640 start a getty on the primary kernel console but on all
15641 others, too. This makes the order in which console= is
15642 specified on the kernel command line less important.
15644 * libsystemd-logind gained a new sd_session_get_vt() call to
15645 retrieve the VT number of a session.
15647 * If the option "tries=0" is set for an entry of /etc/crypttab
15648 its passphrase is queried indefinitely instead of any
15649 maximum number of tries.
15651 * If a service with a configure PID file terminates its PID
15652 file will now be removed automatically if it still exists
15653 afterwards. This should put an end to stale PID files.
15655 * systemd-run will now also take relative binary path names
15656 for execution and no longer insists on absolute paths.
15658 * InaccessibleDirectories= and ReadOnlyDirectories= now take
15659 paths that are optionally prefixed with "-" to indicate that
15660 it should not be considered a failure if they do not exist.
15662 * journalctl -o (and similar commands) now understands a new
15663 output mode "short-precise", it is similar to "short" but
15664 shows timestamps with usec accuracy.
15666 * The option "discard" (as known from Debian) is now
15667 synonymous to "allow-discards" in /etc/crypttab. In fact,
15668 "discard" is preferred now (since it is easier to remember
15671 * Some licensing clean-ups were made, so that more code is now
15672 LGPL-2.1 licensed than before.
15674 * A minimal tool to save/restore the display backlight
15675 brightness across reboots has been added. It will store the
15676 backlight setting as late as possible at shutdown, and
15677 restore it as early as possible during reboot.
15679 * A logic to automatically discover and enable home and swap
15680 partitions on GPT disks has been added. With this in place
15681 /etc/fstab becomes optional for many setups as systemd can
15682 discover certain partitions located on the root disk
15683 automatically. Home partitions are recognized under their
15684 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
15685 partitions are recognized under their GPT type ID
15686 0657fd6da4ab43c484e50933c84b4f4f.
15688 * systemd will no longer pass any environment from the kernel
15689 or initrd to system services. If you want to set an
15690 environment for all services, do so via the kernel command
15691 line systemd.setenv= assignment.
15693 * The systemd-sysctl tool no longer natively reads the file
15694 /etc/sysctl.conf. If desired, the file should be symlinked
15695 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
15696 legacy support by a symlink rather than built-in code, it
15697 also makes the otherwise hidden order of application of the
15698 different files visible. (Note that this partly reverts to a
15699 pre-198 application order of sysctl knobs!)
15701 * The "systemctl set-log-level" and "systemctl dump" commands
15702 have been moved to systemd-analyze.
15704 * systemd-run learned the new --remain-after-exit switch,
15705 which causes the scope unit not to be cleaned up
15706 automatically after the process terminated.
15708 * tmpfiles learned a new --exclude-prefix= switch to exclude
15709 certain paths from operation.
15711 * journald will now automatically flush all messages to disk
15712 as soon as a message at the log level CRIT, ALERT or EMERG
15715 Contributions from: Andrew Cook, Brandon Philips, Christian
15716 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
15717 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
15718 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
15719 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
15720 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
15721 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
15722 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
15723 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
15724 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
15725 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
15726 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
15727 William Giokas, Zbigniew Jędrzejewski-Szmek
15729 — Berlin, 2013-09-13
15733 * The documentation has been updated to cover the various new
15734 concepts introduced with 205.
15736 * Unit files now understand the new %v specifier which
15737 resolves to the kernel version string as returned by "uname
15740 * systemctl now supports filtering the unit list output by
15741 load state, active state and sub state, using the new
15742 --state= parameter.
15744 * "systemctl status" will now show the results of the
15745 condition checks (like ConditionPathExists= and similar) of
15746 the last start attempts of the unit. They are also logged to
15749 * "journalctl -b" may now be used to look for boot output of a
15750 specific boot. Try "journalctl -b -1" for the previous boot,
15751 but the syntax is substantially more powerful.
15753 * "journalctl --show-cursor" has been added which prints the
15754 cursor string the last shown log line. This may then be used
15755 with the new "journalctl --after-cursor=" switch to continue
15756 browsing logs from that point on.
15758 * "journalctl --force" may now be used to force regeneration
15761 * Creation of "dead" device nodes has been moved from udev
15762 into kmod and tmpfiles. Previously, udev would read the kmod
15763 databases to pre-generate dead device nodes based on meta
15764 information contained in kernel modules, so that these would
15765 be auto-loaded on access rather then at boot. As this
15766 does not really have much to do with the exposing actual
15767 kernel devices to userspace this has always been slightly
15768 alien in the udev codebase. Following the new scheme kmod
15769 will now generate a runtime snippet for tmpfiles from the
15770 module meta information and it now is tmpfiles' job to the
15771 create the nodes. This also allows overriding access and
15772 other parameters for the nodes using the usual tmpfiles
15773 facilities. As side effect this allows us to remove the
15774 CAP_SYS_MKNOD capability bit from udevd entirely.
15776 * logind's device ACLs may now be applied to these "dead"
15777 devices nodes too, thus finally allowing managed access to
15778 devices such as /dev/snd/sequencer without loading the
15779 backing module right-away.
15781 * A new RPM macro has been added that may be used to apply
15782 tmpfiles configuration during package installation.
15784 * systemd-detect-virt and ConditionVirtualization= now can
15785 detect User-Mode-Linux machines (UML).
15787 * journald will now implicitly log the effective capabilities
15788 set of processes in the message metadata.
15790 * systemd-cryptsetup has gained support for TrueCrypt volumes.
15792 * The initrd interface has been simplified (more specifically,
15793 support for passing performance data via environment
15794 variables and fsck results via files in /run has been
15795 removed). These features were non-essential, and are
15796 nowadays available in a much nicer way by having systemd in
15797 the initrd serialize its state and have the hosts systemd
15798 deserialize it again.
15800 * The udev "keymap" data files and tools to apply keyboard
15801 specific mappings of scan to key codes, and force-release
15802 scan code lists have been entirely replaced by a udev
15803 "keyboard" builtin and a hwdb data file.
15805 * systemd will now honour the kernel's "quiet" command line
15806 argument also during late shutdown, resulting in a
15807 completely silent shutdown when used.
15809 * There's now an option to control the SO_REUSEPORT socket
15810 option in .socket units.
15812 * Instance units will now automatically get a per-template
15813 subslice of system.slice unless something else is explicitly
15814 configured. For example, instances of sshd@.service will now
15815 implicitly be placed in system-sshd.slice rather than
15816 system.slice as before.
15818 * Test coverage support may now be enabled at build time.
15820 Contributions from: Dave Reisner, Frederic Crozat, Harald
15821 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
15822 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
15823 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
15824 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
15825 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
15826 Giokas, Zbigniew Jędrzejewski-Szmek
15828 — Berlin, 2013-07-23
15832 * Two new unit types have been introduced:
15834 Scope units are very similar to service units, however, are
15835 created out of pre-existing processes — instead of PID 1
15836 forking off the processes. By using scope units it is
15837 possible for system services and applications to group their
15838 own child processes (worker processes) in a powerful way
15839 which then maybe used to organize them, or kill them
15840 together, or apply resource limits on them.
15842 Slice units may be used to partition system resources in an
15843 hierarchical fashion and then assign other units to them. By
15844 default there are now three slices: system.slice (for all
15845 system services), user.slice (for all user sessions),
15846 machine.slice (for VMs and containers).
15848 Slices and scopes have been introduced primarily in
15849 context of the work to move cgroup handling to a
15850 single-writer scheme, where only PID 1
15851 creates/removes/manages cgroups.
15853 * There's a new concept of "transient" units. In contrast to
15854 normal units these units are created via an API at runtime,
15855 not from configuration from disk. More specifically this
15856 means it is now possible to run arbitrary programs as
15857 independent services, with all execution parameters passed
15858 in via bus APIs rather than read from disk. Transient units
15859 make systemd substantially more dynamic then it ever was,
15860 and useful as a general batch manager.
15862 * logind has been updated to make use of scope and slice units
15863 for managing user sessions. As a user logs in he will get
15864 his own private slice unit, to which all sessions are added
15865 as scope units. We also added support for automatically
15866 adding an instance of user@.service for the user into the
15867 slice. Effectively logind will no longer create cgroup
15868 hierarchies on its own now, it will defer entirely to PID 1
15869 for this by means of scope, service and slice units. Since
15870 user sessions this way become entities managed by PID 1
15871 the output of "systemctl" is now a lot more comprehensive.
15873 * A new mini-daemon "systemd-machined" has been added which
15874 may be used by virtualization managers to register local
15875 VMs/containers. nspawn has been updated accordingly, and
15876 libvirt will be updated shortly. machined will collect a bit
15877 of meta information about the VMs/containers, and assign
15878 them their own scope unit (see above). The collected
15879 meta-data is then made available via the "machinectl" tool,
15880 and exposed in "ps" and similar tools. machined/machinectl
15881 is compile-time optional.
15883 * As discussed earlier, the low-level cgroup configuration
15884 options ControlGroup=, ControlGroupModify=,
15885 ControlGroupPersistent=, ControlGroupAttribute= have been
15886 removed. Please use high-level attribute settings instead as
15887 well as slice units.
15889 * A new bus call SetUnitProperties() has been added to alter
15890 various runtime parameters of a unit. This is primarily
15891 useful to alter cgroup parameters dynamically in a nice way,
15892 but will be extended later on to make more properties
15893 modifiable at runtime. systemctl gained a new set-properties
15894 command that wraps this call.
15896 * A new tool "systemd-run" has been added which can be used to
15897 run arbitrary command lines as transient services or scopes,
15898 while configuring a number of settings via the command
15899 line. This tool is currently very basic, however already
15900 very useful. We plan to extend this tool to even allow
15901 queuing of execution jobs with time triggers from the
15902 command line, similar in fashion to "at".
15904 * nspawn will now inform the user explicitly that kernels with
15905 audit enabled break containers, and suggest the user to turn
15908 * Support for detecting the IMA and AppArmor security
15909 frameworks with ConditionSecurity= has been added.
15911 * journalctl gained a new "-k" switch for showing only kernel
15912 messages, mimicking dmesg output; in addition to "--user"
15913 and "--system" switches for showing only user's own logs
15916 * systemd-delta can now show information about drop-in
15917 snippets extending unit files.
15919 * libsystemd-bus has been substantially updated but is still
15920 not available as public API.
15922 * systemd will now look for the "debug" argument on the kernel
15923 command line and enable debug logging, similar to what
15924 "systemd.log_level=debug" already did before.
15926 * "systemctl set-default", "systemctl get-default" has been
15927 added to configure the default.target symlink, which
15928 controls what to boot into by default.
15930 * "systemctl set-log-level" has been added as a convenient
15931 way to raise and lower systemd logging threshold.
15933 * "systemd-analyze plot" will now show the time the various
15934 generators needed for execution, as well as information
15935 about the unit file loading.
15937 * libsystemd-journal gained a new sd_journal_open_files() call
15938 for opening specific journal files. journactl also gained a
15939 new switch to expose this new functionality. Previously we
15940 only supported opening all files from a directory, or all
15941 files from the system, as opening individual files only is
15942 racy due to journal file rotation.
15944 * systemd gained the new DefaultEnvironment= setting in
15945 /etc/systemd/system.conf to set environment variables for
15948 * If a privileged process logs a journal message with the
15949 OBJECT_PID= field set, then journald will automatically
15950 augment this with additional OBJECT_UID=, OBJECT_GID=,
15951 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
15952 system services want to log events about specific client
15953 processes. journactl/systemctl has been updated to make use
15954 of this information if all log messages regarding a specific
15957 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
15958 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
15959 Reisner, David Coppa, David King, David Strauss, Eelco
15960 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
15961 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
15962 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
15963 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
15964 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
15965 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
15966 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
15967 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
15968 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
15969 Łukasz Stelmach, 장동준
15973 * The Python bindings gained some minimal support for the APIs
15974 exposed by libsystemd-logind.
15976 * ConditionSecurity= gained support for detecting SMACK. Since
15977 this condition already supports SELinux and AppArmor we only
15978 miss IMA for this. Patches welcome!
15980 Contributions from: Karol Lewandowski, Lennart Poettering,
15981 Zbigniew Jędrzejewski-Szmek
15985 * systemd-nspawn will now create /etc/resolv.conf if
15986 necessary, before bind-mounting the host's file onto it.
15988 * systemd-nspawn will now store meta information about a
15989 container on the container's cgroup as extended attribute
15990 fields, including the root directory.
15992 * The cgroup hierarchy has been reworked in many ways. All
15993 objects any of the components systemd creates in the cgroup
15994 tree are now suffixed. More specifically, user sessions are
15995 now placed in cgroups suffixed with ".session", users in
15996 cgroups suffixed with ".user", and nspawn containers in
15997 cgroups suffixed with ".nspawn". Furthermore, all cgroup
15998 names are now escaped in a simple scheme to avoid collision
15999 of userspace object names with kernel filenames. This work
16000 is preparation for making these objects relocatable in the
16001 cgroup tree, in order to allow easy resource partitioning of
16002 these objects without causing naming conflicts.
16004 * systemctl list-dependencies gained the new switches
16005 --plain, --reverse, --after and --before.
16007 * systemd-inhibit now shows the process name of processes that
16008 have taken an inhibitor lock.
16010 * nss-myhostname will now also resolve "localhost"
16011 implicitly. This makes /etc/hosts an optional file and
16012 nicely handles that on IPv6 ::1 maps to both "localhost" and
16013 the local hostname.
16015 * libsystemd-logind.so gained a new call
16016 sd_get_machine_names() to enumerate running containers and
16017 VMs (currently only supported by very new libvirt and
16018 nspawn). sd_login_monitor can now be used to watch
16019 VMs/containers coming and going.
16021 * .include is not allowed recursively anymore, and only in
16022 unit files. Usually it is better to use drop-in snippets in
16023 .d/*.conf anyway, as introduced with systemd 198.
16025 * systemd-analyze gained a new "critical-chain" command that
16026 determines the slowest chain of units run during system
16027 boot-up. It is very useful for tracking down where
16028 optimizing boot time is the most beneficial.
16030 * systemd will no longer allow manipulating service paths in
16031 the name=systemd:/system cgroup tree using ControlGroup= in
16032 units. (But is still fine with it in all other dirs.)
16034 * There's a new systemd-nspawn@.service service file that may
16035 be used to easily run nspawn containers as system
16036 services. With the container's root directory in
16037 /var/lib/container/foobar it is now sufficient to run
16038 "systemctl start systemd-nspawn@foobar.service" to boot it.
16040 * systemd-cgls gained a new parameter "--machine" to list only
16041 the processes within a certain container.
16043 * ConditionSecurity= now can check for "apparmor". We still
16044 are lacking checks for SMACK and IMA for this condition
16045 check though. Patches welcome!
16047 * A new configuration file /etc/systemd/sleep.conf has been
16048 added that may be used to configure which kernel operation
16049 systemd is supposed to execute when "suspend", "hibernate"
16050 or "hybrid-sleep" is requested. This makes the new kernel
16051 "freeze" state accessible to the user.
16053 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
16054 the passed argument if applicable.
16056 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
16057 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
16058 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
16059 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
16060 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
16061 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
16062 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
16067 * The output of 'systemctl list-jobs' got some polishing. The
16068 '--type=' argument may now be passed more than once. A new
16069 command 'systemctl list-sockets' has been added which shows
16070 a list of kernel sockets systemd is listening on with the
16071 socket units they belong to, plus the units these socket
16074 * The experimental libsystemd-bus library got substantial
16075 updates to work in conjunction with the (also experimental)
16076 kdbus kernel project. It works well enough to exchange
16077 messages with some sophistication. Note that kdbus is not
16078 ready yet, and the library is mostly an elaborate test case
16079 for now, and not installable.
16081 * systemd gained a new unit 'systemd-static-nodes.service'
16082 that generates static device nodes earlier during boot, and
16083 can run in conjunction with udev.
16085 * libsystemd-login gained a new call sd_pid_get_user_unit()
16086 to retrieve the user systemd unit a process is running
16087 in. This is useful for systems where systemd is used as
16090 * systemd-nspawn now places all containers in the new /machine
16091 top-level cgroup directory in the name=systemd
16092 hierarchy. libvirt will soon do the same, so that we get a
16093 uniform separation of /system, /user and /machine for system
16094 services, user processes and containers/virtual
16095 machines. This new cgroup hierarchy is also useful to stick
16096 stable names to specific container instances, which can be
16097 recognized later this way (this name may be controlled
16098 via systemd-nspawn's new -M switch). libsystemd-login also
16099 gained a new call sd_pid_get_machine_name() to retrieve the
16100 name of the container/VM a specific process belongs to.
16102 * bootchart can now store its data in the journal.
16104 * libsystemd-journal gained a new call
16105 sd_journal_add_conjunction() for AND expressions to the
16106 matching logic. This can be used to express more complex
16107 logical expressions.
16109 * journactl can now take multiple --unit= and --user-unit=
16112 * The cryptsetup logic now understands the "luks.key=" kernel
16113 command line switch for specifying a file to read the
16114 decryption key from. Also, if a configured key file is not
16115 found the tool will now automatically fall back to prompting
16118 * Python systemd.journal module was updated to wrap recently
16119 added functions from libsystemd-journal. The interface was
16120 changed to bring the low level interface in s.j._Reader
16121 closer to the C API, and the high level interface in
16122 s.j.Reader was updated to wrap and convert all data about
16125 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
16126 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
16127 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
16128 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
16129 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
16130 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
16134 * journalctl --update-catalog now understands a new --root=
16135 option to operate on catalogs found in a different root
16138 * During shutdown after systemd has terminated all running
16139 services a final killing loop kills all remaining left-over
16140 processes. We will now print the name of these processes
16141 when we send SIGKILL to them, since this usually indicates a
16144 * If /etc/crypttab refers to password files stored on
16145 configured mount points automatic dependencies will now be
16146 generated to ensure the specific mount is established first
16147 before the key file is attempted to be read.
16149 * 'systemctl status' will now show information about the
16150 network sockets a socket unit is listening on.
16152 * 'systemctl status' will also shown information about any
16153 drop-in configuration file for units. (Drop-In configuration
16154 files in this context are files such as
16155 /etc/systemd/system/foobar.service.d/*.conf)
16157 * systemd-cgtop now optionally shows summed up CPU times of
16158 cgroups. Press '%' while running cgtop to switch between
16159 percentage and absolute mode. This is useful to determine
16160 which cgroups use up the most CPU time over the entire
16161 runtime of the system. systemd-cgtop has also been updated
16162 to be 'pipeable' for processing with further shell tools.
16164 * 'hostnamectl set-hostname' will now allow setting of FQDN
16167 * The formatting and parsing of time span values has been
16168 changed. The parser now understands fractional expressions
16169 such as "5.5h". The formatter will now output fractional
16170 expressions for all time spans under 1min, i.e. "5.123456s"
16171 rather than "5s 123ms 456us". For time spans under 1s
16172 millisecond values are shown, for those under 1ms
16173 microsecond values are shown. This should greatly improve
16174 all time-related output of systemd.
16176 * libsystemd-login and libsystemd-journal gained new
16177 functions for querying the poll() events mask and poll()
16178 timeout value for integration into arbitrary event
16181 * localectl gained the ability to list available X11 keymaps
16182 (models, layouts, variants, options).
16184 * 'systemd-analyze dot' gained the ability to filter for
16185 specific units via shell-style globs, to create smaller,
16186 more useful graphs. I.e. it is now possible to create simple
16187 graphs of all the dependencies between only target units, or
16188 of all units that Avahi has dependencies with.
16190 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
16191 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
16192 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
16193 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
16194 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
16195 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
16196 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
16200 * The boot-time readahead implementation for rotating media
16201 will now read the read-ahead data in multiple passes which
16202 consist of all read requests made in equidistant time
16203 intervals. This means instead of strictly reading read-ahead
16204 data in its physical order on disk we now try to find a
16205 middle ground between physical and access time order.
16207 * /etc/os-release files gained a new BUILD_ID= field for usage
16208 on operating systems that provide continuous builds of OS
16211 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
16212 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
16213 William Douglas, Zbigniew Jędrzejewski-Szmek
16217 * systemd-python gained an API exposing libsystemd-daemon.
16219 * The SMACK setup logic gained support for uploading CIPSO
16222 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
16223 ReadOnlyDirectories= and InaccessibleDirectories= has
16224 changed. The private /tmp and /var/tmp directories are now
16225 shared by all processes of a service (which means
16226 ExecStartPre= may now leave data in /tmp that ExecStart= of
16227 the same service can still access). When a service is
16228 stopped its temporary directories are immediately deleted
16229 (normal clean-up with tmpfiles is still done in addition to
16232 * By default, systemd will now set a couple of sysctl
16233 variables in the kernel: the safe sysrq options are turned
16234 on, IP route verification is turned on, and source routing
16235 disabled. The recently added hardlink and softlink
16236 protection of the kernel is turned on. These settings should
16237 be reasonably safe, and good defaults for all new systems.
16239 * The predictable network naming logic may now be turned off
16240 with a new kernel command line switch: net.ifnames=0.
16242 * A new libsystemd-bus module has been added that implements a
16243 pretty complete D-Bus client library. For details see:
16245 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
16247 * journald will now explicitly flush the journal files to disk
16248 at the latest 5min after each write. The file will then also
16249 be marked offline until the next write. This should increase
16250 reliability in case of a crash. The synchronization delay
16251 can be configured via SyncIntervalSec= in journald.conf.
16253 * There's a new remote-fs-setup.target unit that can be used
16254 to pull in specific services when at least one remote file
16255 system is to be mounted.
16257 * There are new targets timers.target and paths.target as
16258 canonical targets to pull user timer and path units in
16259 from. This complements sockets.target with a similar
16260 purpose for socket units.
16262 * libudev gained a new call udev_device_set_attribute_value()
16263 to set sysfs attributes of a device.
16265 * The udev daemon now sets the default number of worker
16266 processes executed in parallel based on the number of available
16267 CPUs instead of the amount of available RAM. This is supposed
16268 to provide a more reliable default and limit a too aggressive
16269 parallelism for setups with 1000s of devices connected.
16271 Contributions from: Auke Kok, Colin Walters, Cristian
16272 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
16273 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
16274 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
16275 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
16276 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
16277 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
16278 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
16279 Zbigniew Jędrzejewski-Szmek
16283 * Configuration of unit files may now be extended via drop-in
16284 files without having to edit/override the unit files
16285 themselves. More specifically, if the administrator wants to
16286 change one value for a service file foobar.service he can
16287 now do so by dropping in a configuration snippet into
16288 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
16289 will load all these snippets and apply them on top of the
16290 main unit configuration file, possibly extending or
16291 overriding its settings. Using these drop-in snippets is
16292 generally nicer than the two earlier options for changing
16293 unit files locally: copying the files from
16294 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
16295 them there; or creating a new file in /etc/systemd/system/
16296 that incorporates the original one via ".include". Drop-in
16297 snippets into these .d/ directories can be placed in any
16298 directory systemd looks for units in, and the usual
16299 overriding semantics between /usr/lib, /etc and /run apply
16302 * Most unit file settings which take lists of items can now be
16303 reset by assigning the empty string to them. For example,
16304 normally, settings such as Environment=FOO=BAR append a new
16305 environment variable assignment to the environment block,
16306 each time they are used. By assigning Environment= the empty
16307 string the environment block can be reset to empty. This is
16308 particularly useful with the .d/*.conf drop-in snippets
16309 mentioned above, since this adds the ability to reset list
16310 settings from vendor unit files via these drop-ins.
16312 * systemctl gained a new "list-dependencies" command for
16313 listing the dependencies of a unit recursively.
16315 * Inhibitors are now honored and listed by "systemctl
16316 suspend", "systemctl poweroff" (and similar) too, not only
16317 GNOME. These commands will also list active sessions by
16320 * Resource limits (as exposed by the various control group
16321 controllers) can now be controlled dynamically at runtime
16322 for all units. More specifically, you can now use a command
16323 like "systemctl set-cgroup-attr foobar.service cpu.shares
16324 2000" to alter the CPU shares a specific service gets. These
16325 settings are stored persistently on disk, and thus allow the
16326 administrator to easily adjust the resource usage of
16327 services with a few simple commands. This dynamic resource
16328 management logic is also available to other programs via the
16329 bus. Almost any kernel cgroup attribute and controller is
16332 * systemd-vconsole-setup will now copy all font settings to
16333 all allocated VTs, where it previously applied them only to
16336 * libsystemd-login gained the new sd_session_get_tty() API
16339 * This release drops support for a few legacy or
16340 distribution-specific LSB facility names when parsing init
16341 scripts: $x-display-manager, $mail-transfer-agent,
16342 $mail-transport-agent, $mail-transfer-agent, $smtp,
16343 $null. Also, the mail-transfer-agent.target unit backing
16344 this has been removed. Distributions which want to retain
16345 compatibility with this should carry the burden for
16346 supporting this themselves and patch support for these back
16347 in, if they really need to. Also, the facilities $syslog and
16348 $local_fs are now ignored, since systemd does not support
16349 early-boot LSB init scripts anymore, and these facilities
16350 are implied anyway for normal services. syslog.target has
16353 * There are new bus calls on PID1's Manager object for
16354 cancelling jobs, and removing snapshot units. Previously,
16355 both calls were only available on the Job and Snapshot
16356 objects themselves.
16358 * systemd-journal-gatewayd gained SSL support.
16360 * The various "environment" files, such as /etc/locale.conf
16361 now support continuation lines with a backslash ("\") as
16362 last character in the line, similarly in style (but different)
16363 to how this is supported in shells.
16365 * For normal user processes the _SYSTEMD_USER_UNIT= field is
16366 now implicitly appended to every log entry logged. systemctl
16367 has been updated to filter by this field when operating on a
16368 user systemd instance.
16370 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
16371 CAP_AUDIT_CONTROL capabilities to the capabilities set for
16372 the container. This makes it easier to boot unmodified
16373 Fedora systems in a container, which however still requires
16374 audit=0 to be passed on the kernel command line. Auditing in
16375 kernel and userspace is unfortunately still too broken in
16376 context of containers, hence we recommend compiling it out
16377 of the kernel or using audit=0. Hopefully this will be fixed
16378 one day for good in the kernel.
16380 * nspawn gained the new --bind= and --bind-ro= parameters to
16381 bind mount specific directories from the host into the
16384 * nspawn will now mount its own devpts file system instance
16385 into the container, in order not to leak pty devices from
16386 the host into the container.
16388 * systemd will now read the firmware boot time performance
16389 information from the EFI variables, if the used boot loader
16390 supports this, and takes it into account for boot performance
16391 analysis via "systemd-analyze". This is currently supported
16392 only in conjunction with Gummiboot, but could be supported
16393 by other boot loaders too. For details see:
16395 https://systemd.io/BOOT_LOADER_INTERFACE
16397 * A new generator has been added that automatically mounts the
16398 EFI System Partition (ESP) to /boot, if that directory
16399 exists, is empty, and no other file system has been
16400 configured to be mounted there.
16402 * logind will now send out PrepareForSleep(false) out
16403 unconditionally, after coming back from suspend. This may be
16404 used by applications as asynchronous notification for
16405 system resume events.
16407 * "systemctl unlock-sessions" has been added, that allows
16408 unlocking the screens of all user sessions at once, similar
16409 to how "systemctl lock-sessions" already locked all users
16410 sessions. This is backed by a new D-Bus call UnlockSessions().
16412 * "loginctl seat-status" will now show the master device of a
16413 seat. (i.e. the device of a seat that needs to be around for
16414 the seat to be considered available, usually the graphics
16417 * tmpfiles gained a new "X" line type, that allows
16418 configuration of files and directories (with wildcards) that
16419 shall be excluded from automatic cleanup ("aging").
16421 * udev default rules set the device node permissions now only
16422 at "add" events, and do not change them any longer with a
16423 later "change" event.
16425 * The log messages for lid events and power/sleep keypresses
16426 now carry a message ID.
16428 * We now have a substantially larger unit test suite, but this
16429 continues to be work in progress.
16431 * udevadm hwdb gained a new --root= parameter to change the
16432 root directory to operate relative to.
16434 * logind will now issue a background sync() request to the kernel
16435 early at shutdown, so that dirty buffers are flushed to disk early
16436 instead of at the last moment, in order to optimize shutdown
16439 * A new bootctl tool has been added that is an interface for
16440 certain boot loader operations. This is currently a preview
16441 and is likely to be extended into a small mechanism daemon
16442 like timedated, localed, hostnamed, and can be used by
16443 graphical UIs to enumerate available boot options, and
16444 request boot into firmware operations.
16446 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
16447 the rest of the package. It also has been updated to work
16448 correctly in initrds.
16450 * polkit previously has been runtime optional, and is now also
16451 compile time optional via a configure switch.
16453 * systemd-analyze has been reimplemented in C. Also "systemctl
16454 dot" has moved into systemd-analyze.
16456 * "systemctl status" with no further parameters will now print
16457 the status of all active or failed units.
16459 * Operations such as "systemctl start" can now be executed
16460 with a new mode "--irreversible" which may be used to queue
16461 operations that cannot accidentally be reversed by a later
16462 job queuing. This is by default used to make shutdown
16463 requests more robust.
16465 * The Python API of systemd now gained a new module for
16466 reading journal files.
16468 * A new tool kernel-install has been added that can install
16469 kernel images according to the Boot Loader Specification:
16471 https://systemd.io/BOOT_LOADER_SPECIFICATION
16473 * Boot time console output has been improved to provide
16474 animated boot time output for hanging jobs.
16476 * A new tool systemd-activate has been added which can be used
16477 to test socket activation with, directly from the command
16478 line. This should make it much easier to test and debug
16479 socket activation in daemons.
16481 * journalctl gained a new "--reverse" (or -r) option to show
16482 journal output in reverse order (i.e. newest line first).
16484 * journalctl gained a new "--pager-end" (or -e) option to jump
16485 to immediately jump to the end of the journal in the
16486 pager. This is only supported in conjunction with "less".
16488 * journalctl gained a new "--user-unit=" option, that works
16489 similarly to "--unit=" but filters for user units rather than
16492 * A number of unit files to ease adoption of systemd in
16493 initrds has been added. This moves some minimal logic from
16494 the various initrd implementations into systemd proper.
16496 * The journal files are now owned by a new group
16497 "systemd-journal", which exists specifically to allow access
16498 to the journal, and nothing else. Previously, we used the
16499 "adm" group for that, which however possibly covers more
16500 than just journal/log file access. This new group is now
16501 already used by systemd-journal-gatewayd to ensure this
16502 daemon gets access to the journal files and as little else
16503 as possible. Note that "make install" will also set FS ACLs
16504 up for /var/log/journal to give "adm" and "wheel" read
16505 access to it, in addition to "systemd-journal" which owns
16506 the journal files. We recommend that packaging scripts also
16507 add read access to "adm" + "wheel" to /var/log/journal, and
16508 all existing/future journal files. To normal users and
16509 administrators little changes, however packagers need to
16510 ensure to create the "systemd-journal" system group at
16511 package installation time.
16513 * The systemd-journal-gatewayd now runs as unprivileged user
16514 systemd-journal-gateway:systemd-journal-gateway. Packaging
16515 scripts need to create these system user/group at
16518 * timedated now exposes a new boolean property CanNTP that
16519 indicates whether a local NTP service is available or not.
16521 * systemd-detect-virt will now also detect xen PVs
16523 * The pstore file system is now mounted by default, if it is
16526 * In addition to the SELinux and IMA policies we will now also
16527 load SMACK policies at early boot.
16529 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
16530 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
16531 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
16532 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
16533 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
16534 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
16535 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
16536 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
16537 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
16538 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
16539 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
16540 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
16541 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
16542 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
16546 * Timer units now support calendar time events in addition to
16547 monotonic time events. That means you can now trigger a unit
16548 based on a calendar time specification such as "Thu,Fri
16549 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
16550 or fifth day of any month of the year 2013, given that it is
16551 a Thursday or a Friday. This brings timer event support
16552 considerably closer to cron's capabilities. For details on
16553 the supported calendar time specification language see
16556 * udev now supports a number of different naming policies for
16557 network interfaces for predictable names, and a combination
16558 of these policies is now the default. Please see this wiki
16559 document for details:
16561 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
16563 * Auke Kok's bootchart implementation has been added to the
16564 systemd tree. It is an optional component that can graph the
16565 boot in quite some detail. It is one of the best bootchart
16566 implementations around and minimal in its code and
16569 * nss-myhostname has been integrated into the systemd source
16570 tree. nss-myhostname guarantees that the local hostname
16571 always stays resolvable via NSS. It has been a weak
16572 requirement of systemd-hostnamed since a long time, and
16573 since its code is actually trivial we decided to just
16574 include it in systemd's source tree. It can be turned off
16575 with a configure switch.
16577 * The read-ahead logic is now capable of properly detecting
16578 whether a btrfs file system is on SSD or rotating media, in
16579 order to optimize the read-ahead scheme. Previously, it was
16580 only capable of detecting this on traditional file systems
16583 * In udev, additional device properties are now read from the
16584 IAB in addition to the OUI database. Also, Bluetooth company
16585 identities are attached to the devices as well.
16587 * In service files %U may be used as specifier that is
16588 replaced by the configured user name of the service.
16590 * nspawn may now be invoked without a controlling TTY. This
16591 makes it suitable for invocation as its own service. This
16592 may be used to set up a simple containerized server system
16593 using only core OS tools.
16595 * systemd and nspawn can now accept socket file descriptors
16596 when they are started for socket activation. This enables
16597 implementation of socket activated nspawn
16598 containers. i.e. think about autospawning an entire OS image
16599 when the first SSH or HTTP connection is received. We expect
16600 that similar functionality will also be added to libvirt-lxc
16603 * journalctl will now suppress ANSI color codes when
16604 presenting log data.
16606 * systemctl will no longer show control group information for
16607 a unit if the control group is empty anyway.
16609 * logind can now automatically suspend/hibernate/shutdown the
16612 * /etc/machine-info and hostnamed now also expose the chassis
16613 type of the system. This can be used to determine whether
16614 the local system is a laptop, desktop, handset or
16615 tablet. This information may either be configured by the
16616 user/vendor or is automatically determined from ACPI and DMI
16617 information if possible.
16619 * A number of polkit actions are now bound together with "imply"
16620 rules. This should simplify creating UIs because many actions
16621 will now authenticate similar ones as well.
16623 * Unit files learnt a new condition ConditionACPower= which
16624 may be used to conditionalize a unit depending on whether an
16625 AC power source is connected or not, of whether the system
16626 is running on battery power.
16628 * systemctl gained a new "is-failed" verb that may be used in
16629 shell scripts and suchlike to check whether a specific unit
16630 is in the "failed" state.
16632 * The EnvironmentFile= setting in unit files now supports file
16633 globbing, and can hence be used to easily read a number of
16634 environment files at once.
16636 * systemd will no longer detect and recognize specific
16637 distributions. All distribution-specific #ifdeffery has been
16638 removed, systemd is now fully generic and
16639 distribution-agnostic. Effectively, not too much is lost as
16640 a lot of the code is still accessible via explicit configure
16641 switches. However, support for some distribution specific
16642 legacy configuration file formats has been dropped. We
16643 recommend distributions to simply adopt the configuration
16644 files everybody else uses now and convert the old
16645 configuration from packaging scripts. Most distributions
16646 already did that. If that's not possible or desirable,
16647 distributions are welcome to forward port the specific
16648 pieces of code locally from the git history.
16650 * When logging a message about a unit systemd will now always
16651 log the unit name in the message meta data.
16653 * localectl will now also discover system locale data that is
16654 not stored in locale archives, but directly unpacked.
16656 * logind will no longer unconditionally use framebuffer
16657 devices as seat masters, i.e. as devices that are required
16658 to be existing before a seat is considered preset. Instead,
16659 it will now look for all devices that are tagged as
16660 "seat-master" in udev. By default, framebuffer devices will
16661 be marked as such, but depending on local systems, other
16662 devices might be marked as well. This may be used to
16663 integrate graphics cards using closed source drivers (such
16664 as NVidia ones) more nicely into logind. Note however, that
16665 we recommend using the open source NVidia drivers instead,
16666 and no udev rules for the closed-source drivers will be
16667 shipped from us upstream.
16669 Contributions from: Adam Williamson, Alessandro Crismani, Auke
16670 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
16671 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
16672 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
16673 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
16674 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
16675 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
16676 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
16677 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
16678 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
16679 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
16680 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
16685 * udev gained support for loading additional device properties
16686 from an indexed database that is keyed by vendor/product IDs
16687 and similar device identifiers. For the beginning this
16688 "hwdb" is populated with data from the well-known PCI and
16689 USB database, but also includes PNP, ACPI and OID data. In
16690 the longer run this indexed database shall grow into
16691 becoming the one central database for non-essential
16692 userspace device metadata. Previously, data from the PCI/USB
16693 database was only attached to select devices, since the
16694 lookup was a relatively expensive operation due to O(n) time
16695 complexity (with n being the number of entries in the
16696 database). Since this is now O(1), we decided to add in this
16697 data for all devices where this is available, by
16698 default. Note that the indexed database needs to be rebuilt
16699 when new data files are installed. To achieve this you need
16700 to update your packaging scripts to invoke "udevadm hwdb
16701 --update" after installation of hwdb data files. For
16702 RPM-based distributions we introduced the new
16703 %udev_hwdb_update macro for this purpose.
16705 * The Journal gained support for the "Message Catalog", an
16706 indexed database to link up additional information with
16707 journal entries. For further details please check:
16709 https://www.freedesktop.org/wiki/Software/systemd/catalog
16711 The indexed message catalog database also needs to be
16712 rebuilt after installation of message catalog files. Use
16713 "journalctl --update-catalog" for this. For RPM-based
16714 distributions we introduced the %journal_catalog_update
16715 macro for this purpose.
16717 * The Python Journal bindings gained support for the standard
16718 Python logging framework.
16720 * The Journal API gained new functions for checking whether
16721 the underlying file system of a journal file is capable of
16722 properly reporting file change notifications, or whether
16723 applications that want to reflect journal changes "live"
16724 need to recheck journal files continuously in appropriate
16727 * It is now possible to set the "age" field for tmpfiles
16728 entries to 0, indicating that files matching this entry
16729 shall always be removed when the directories are cleaned up.
16731 * coredumpctl gained a new "gdb" verb which invokes gdb
16732 right-away on the selected coredump.
16734 * There's now support for "hybrid sleep" on kernels that
16735 support this, in addition to "suspend" and "hibernate". Use
16736 "systemctl hybrid-sleep" to make use of this.
16738 * logind's HandleSuspendKey= setting (and related settings)
16739 now gained support for a new "lock" setting to simply
16740 request the screen lock on all local sessions, instead of
16741 actually executing a suspend or hibernation.
16743 * systemd will now mount the EFI variables file system by
16746 * Socket units now gained support for configuration of the
16747 SMACK security label.
16749 * timedatectl will now output the time of the last and next
16750 daylight saving change.
16752 * We dropped support for various legacy and distro-specific
16753 concepts, such as insserv, early-boot SysV services
16754 (i.e. those for non-standard runlevels such as 'b' or 'S')
16755 or ArchLinux /etc/rc.conf support. We recommend the
16756 distributions who still need support this to either continue
16757 to maintain the necessary patches downstream, or find a
16758 different solution. (Talk to us if you have questions!)
16760 * Various systemd components will now bypass polkit checks for
16761 root and otherwise handle properly if polkit is not found to
16762 be around. This should fix most issues for polkit-less
16763 systems. Quite frankly this should have been this way since
16764 day one. It is absolutely our intention to make systemd work
16765 fine on polkit-less systems, and we consider it a bug if
16766 something does not work as it should if polkit is not around.
16768 * For embedded systems it is now possible to build udev and
16769 systemd without blkid and/or kmod support.
16771 * "systemctl switch-root" is now capable of switching root
16772 more than once. I.e. in addition to transitions from the
16773 initrd to the host OS it is now possible to transition to
16774 further OS images from the host. This is useful to implement
16775 offline updating tools.
16777 * Various other additions have been made to the RPM macros
16778 shipped with systemd. Use %udev_rules_update() after
16779 installing new udev rules files. %_udevhwdbdir,
16780 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
16781 %_sysctldir are now available which resolve to the right
16782 directories for packages to place various data files in.
16784 * journalctl gained the new --full switch (in addition to
16785 --all, to disable ellipsation for long messages.
16787 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
16788 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
16789 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
16790 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
16791 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
16792 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
16793 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
16794 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
16795 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
16799 * journalctl gained new --since= and --until= switches to
16800 filter by time. It also now supports nice filtering for
16801 units via --unit=/-u.
16803 * Type=oneshot services may use ExecReload= and do the
16806 * The journal daemon now supports time-based rotation and
16807 vacuuming, in addition to the usual disk-space based
16810 * The journal will now index the available field values for
16811 each field name. This enables clients to show pretty drop
16812 downs of available match values when filtering. The bash
16813 completion of journalctl has been updated
16814 accordingly. journalctl gained a new switch -F to list all
16815 values a certain field takes in the journal database.
16817 * More service events are now written as structured messages
16818 to the journal, and made recognizable via message IDs.
16820 * The timedated, localed and hostnamed mini-services which
16821 previously only provided support for changing time, locale
16822 and hostname settings from graphical DEs such as GNOME now
16823 also have a minimal (but very useful) text-based client
16824 utility each. This is probably the nicest way to changing
16825 these settings from the command line now, especially since
16826 it lists available options and is fully integrated with bash
16829 * There's now a new tool "systemd-coredumpctl" to list and
16830 extract coredumps from the journal.
16832 * We now install a README each in /var/log/ and
16833 /etc/rc.d/init.d explaining where the system logs and init
16834 scripts went. This hopefully should help folks who go to
16835 that dirs and look into the otherwise now empty void and
16836 scratch their heads.
16838 * When user-services are invoked (by systemd --user) the
16839 $MANAGERPID env var is set to the PID of systemd.
16841 * SIGRTMIN+24 when sent to a --user instance will now result
16842 in immediate termination of systemd.
16844 * gatewayd received numerous feature additions such as a
16845 "follow" mode, for live syncing and filtering.
16847 * browse.html now allows filtering and showing detailed
16848 information on specific entries. Keyboard navigation and
16849 mouse screen support has been added.
16851 * gatewayd/journalctl now supports HTML5/JSON
16852 Server-Sent-Events as output.
16854 * The SysV init script compatibility logic will now
16855 heuristically determine whether a script supports the
16856 "reload" verb, and only then make this available as
16857 "systemctl reload".
16859 * "systemctl status --follow" has been removed, use "journalctl
16862 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
16863 have been removed since they are hardly useful to be
16866 * And I'd like to take the opportunity to specifically mention
16867 Zbigniew for his great contributions. Zbigniew, you rock!
16869 Contributions from: Andrew Eikum, Christian Hesse, Colin
16870 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
16871 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
16872 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
16873 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
16874 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
16875 Jędrzejewski-Szmek, Сковорода Никита Андреевич
16879 * If /etc/vconsole.conf is non-existent or empty we will no
16880 longer load any console font or key map at boot by
16881 default. Instead the kernel defaults will be left
16882 intact. This is definitely the right thing to do, as no
16883 configuration should mean no configuration, and hard-coding
16884 font names that are different on all archs is probably a bad
16885 idea. Also, the kernel default key map and font should be
16886 good enough for most cases anyway, and mostly identical to
16887 the userspace fonts/key maps we previously overloaded them
16888 with. If distributions want to continue to default to a
16889 non-kernel font or key map they should ship a default
16890 /etc/vconsole.conf with the appropriate contents.
16892 Contributions from: Colin Walters, Daniel J Walsh, Dave
16893 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
16894 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
16898 * journalctl gained a new --cursor= switch to show entries
16899 starting from the specified location in the journal.
16901 * We now enforce a size limit on journal entry fields exported
16902 with "-o json" in journalctl. Fields larger than 4K will be
16903 assigned null. This can be turned off with --all.
16905 * An (optional) journal gateway daemon is now available as
16906 "systemd-journal-gatewayd.service". This service provides
16907 access to the journal via HTTP and JSON. This functionality
16908 will be used to implement live log synchronization in both
16909 pull and push modes, but has various other users too, such
16910 as easy log access for debugging of embedded devices. Right
16911 now it is already useful to retrieve the journal via HTTP:
16913 # systemctl start systemd-journal-gatewayd.service
16914 # wget http://localhost:19531/entries
16916 This will download the journal contents in a
16917 /var/log/messages compatible format. The same as JSON:
16919 # curl -H"Accept: application/json" http://localhost:19531/entries
16921 This service is also accessible via a web browser where a
16922 single static HTML5 app is served that uses the JSON logic
16923 to enable the user to do some basic browsing of the
16924 journal. This will be extended later on. Here's an example
16925 screenshot of this app in its current state:
16927 https://0pointer.de/public/journal-gatewayd
16929 Contributions from: Kay Sievers, Lennart Poettering, Robert
16930 Milasan, Tom Gundersen
16934 * The bash completion logic is now available for journalctl
16937 * We do not mount the "cpuset" controller anymore together with
16938 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
16939 started if no parameters are assigned to it. "cpuset" hence
16940 broke code that assumed it could create "cpu" groups and
16943 * journalctl -f will now subscribe to terminal size changes,
16944 and line break accordingly.
16946 Contributions from: Dave Reisner, Kay Sievers, Lennart
16947 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
16951 * nspawn will now create a symlink /etc/localtime in the
16952 container environment, copying the host's timezone
16953 setting. Previously this has been done via a bind mount, but
16954 since symlinks cannot be bind mounted this has now been
16955 changed to create/update the appropriate symlink.
16957 * journalctl -n's line number argument is now optional, and
16958 will default to 10 if omitted.
16960 * journald will now log the maximum size the journal files may
16961 take up on disk. This is particularly useful if the default
16962 built-in logic of determining this parameter from the file
16963 system size is used. Use "systemctl status
16964 systemd-journald.service" to see this information.
16966 * The multi-seat X wrapper tool has been stripped down. As X
16967 is now capable of enumerating graphics devices via udev in a
16968 seat-aware way the wrapper is not strictly necessary
16969 anymore. A stripped down temporary stop-gap is still shipped
16970 until the upstream display managers have been updated to
16971 fully support the new X logic. Expect this wrapper to be
16972 removed entirely in one of the next releases.
16974 * HandleSleepKey= in logind.conf has been split up into
16975 HandleSuspendKey= and HandleHibernateKey=. The old setting
16976 is not available anymore. X11 and the kernel are
16977 distinguishing between these keys and we should too. This
16978 also means the inhibition lock for these keys has been split
16981 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
16982 Poettering, Lukas Nykryn, Václav Pavlín
16986 * Whenever a unit changes state we will now log this to the
16987 journal and show along the unit's own log output in
16988 "systemctl status".
16990 * ConditionPathIsMountPoint= can now properly detect bind
16991 mount points too. (Previously, a bind mount of one file
16992 system to another place in the same file system could not be
16993 detected as mount, since they shared struct stat's st_dev
16996 * We will now mount the cgroup controllers cpu, cpuacct,
16997 cpuset and the controllers net_cls, net_prio together by
17000 * nspawn containers will now have a virtualized boot
17001 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
17002 over with a randomized ID at container initialization). This
17003 has the effect of making "journalctl -b" do the right thing
17006 * The JSON output journal serialization has been updated not
17007 to generate "endless" list objects anymore, but rather one
17008 JSON object per line. This is more in line how most JSON
17009 parsers expect JSON objects. The new output mode
17010 "json-pretty" has been added to provide similar output, but
17011 neatly aligned for readability by humans.
17013 * We dropped all explicit sync() invocations in the shutdown
17014 code. The kernel does this implicitly anyway in the kernel
17015 reboot() syscall. halt(8)'s -n option is now a compatibility
17018 * We now support virtualized reboot() in containers, as
17019 supported by newer kernels. We will fall back to exit() if
17020 CAP_SYS_REBOOT is not available to the container. Also,
17021 nspawn makes use of this now and will actually reboot the
17022 container if the containerized OS asks for that.
17024 * journalctl will only show local log output by default
17025 now. Use --merge (-m) to show remote log output, too.
17027 * libsystemd-journal gained the new sd_journal_get_usage()
17028 call to determine the current disk usage of all journal
17029 files. This is exposed in the new "journalctl --disk-usage"
17032 * journald gained a new configuration setting SplitMode= in
17033 journald.conf which may be used to control how user journals
17034 are split off. See journald.conf(5) for details.
17036 * A new condition type ConditionFileNotEmpty= has been added.
17038 * tmpfiles' "w" lines now support file globbing, to write
17039 multiple files at once.
17041 * We added Python bindings for the journal submission
17042 APIs. More Python APIs for a number of selected APIs will
17043 likely follow. Note that we intend to add native bindings
17044 only for the Python language, as we consider it common
17045 enough to deserve bindings shipped within systemd. There are
17046 various projects outside of systemd that provide bindings
17047 for languages such as PHP or Lua.
17049 * Many conditions will now resolve specifiers such as %i. In
17050 addition, PathChanged= and related directives of .path units
17051 now support specifiers as well.
17053 * There's now a new RPM macro definition for the system preset
17056 * journald will now warn if it ca not forward a message to the
17057 syslog daemon because its socket is full.
17059 * timedated will no longer write or process /etc/timezone,
17060 except on Debian. As we do not support late mounted /usr
17061 anymore /etc/localtime always being a symlink is now safe,
17062 and hence the information in /etc/timezone is not necessary
17065 * logind will now always reserve one VT for a text getty (VT6
17066 by default). Previously if more than 6 X sessions where
17067 started they took up all the VTs with auto-spawned gettys,
17068 so that no text gettys were available anymore.
17070 * udev will now automatically inform the btrfs kernel logic
17071 about btrfs RAID components showing up. This should make
17072 simple hotplug based btrfs RAID assembly work.
17074 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
17075 (but not for its children which will stay at the kernel
17076 default). This should allow setups with a lot more listening
17079 * systemd will now always pass the configured timezone to the
17080 kernel at boot. timedated will do the same when the timezone
17083 * logind's inhibition logic has been updated. By default,
17084 logind will now handle the lid switch, the power and sleep
17085 keys all the time, even in graphical sessions. If DEs want
17086 to handle these events on their own they should take the new
17087 handle-power-key, handle-sleep-key and handle-lid-switch
17088 inhibitors during their runtime. A simple way to achieve
17089 that is to invoke the DE wrapped in an invocation of:
17091 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
17093 * Access to unit operations is now checked via SELinux taking
17094 the unit file label and client process label into account.
17096 * systemd will now notify the administrator in the journal
17097 when he over-mounts a non-empty directory.
17099 * There are new specifiers that are resolved in unit files,
17100 for the hostname (%H), the machine ID (%m) and the boot ID
17103 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
17104 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
17105 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
17106 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
17107 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
17108 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
17109 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
17113 * Support for reading structured kernel messages from
17114 /dev/kmsg has now been added and is enabled by default.
17116 * Support for reading kernel messages from /proc/kmsg has now
17117 been removed. If you want kernel messages in the journal
17118 make sure to run a recent kernel (>= 3.5) that supports
17119 reading structured messages from /dev/kmsg (see
17120 above). /proc/kmsg is now exclusive property of classic
17121 syslog daemons again.
17123 * The libudev API gained the new
17124 udev_device_new_from_device_id() call.
17126 * The logic for file system namespace (ReadOnlyDirectory=,
17127 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
17128 require pivot_root() anymore. This means fewer temporary
17129 directories are created below /tmp for this feature.
17131 * nspawn containers will now see and receive all submounts
17132 made on the host OS below the root file system of the
17135 * Forward Secure Sealing is now supported for Journal files,
17136 which provide cryptographical sealing of journal files so
17137 that attackers cannot alter log history anymore without this
17138 being detectable. Lennart will soon post a blog story about
17139 this explaining it in more detail.
17141 * There are two new service settings RestartPreventExitStatus=
17142 and SuccessExitStatus= which allow configuration of exit
17143 status (exit code or signal) which will be excepted from the
17144 restart logic, resp. consider successful.
17146 * journalctl gained the new --verify switch that can be used
17147 to check the integrity of the structure of journal files and
17148 (if Forward Secure Sealing is enabled) the contents of
17151 * nspawn containers will now be run with /dev/stdin, /dev/fd/
17152 and similar symlinks pre-created. This makes running shells
17153 as container init process a lot more fun.
17155 * The fstab support can now handle PARTUUID= and PARTLABEL=
17158 * A new ConditionHost= condition has been added to match
17159 against the hostname (with globs) and machine ID. This is
17160 useful for clusters where a single OS image is used to
17161 provision a large number of hosts which shall run slightly
17162 different sets of services.
17164 * Services which hit the restart limit will now be placed in a
17167 Contributions from: Bertram Poettering, Dave Reisner, Huang
17168 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
17169 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
17173 * When running in --user mode systemd will now become a
17174 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
17175 tree a lot more organized.
17177 * A new PartOf= unit dependency type has been introduced that
17178 may be used to group services in a natural way.
17180 * "systemctl enable" may now be used to enable instances of
17183 * journalctl now prints error log levels in red, and
17184 warning/notice log levels in bright white. It also supports
17185 filtering by log level now.
17187 * cgtop gained a new -n switch (similar to top), to configure
17188 the maximum number of iterations to run for. It also gained
17189 -b, to run in batch mode (accepting no input).
17191 * The suffix ".service" may now be omitted on most systemctl
17192 command lines involving service unit names.
17194 * There's a new bus call in logind to lock all sessions, as
17195 well as a loginctl verb for it "lock-sessions".
17197 * libsystemd-logind.so gained a new call sd_journal_perror()
17198 that works similar to libc perror() but logs to the journal
17199 and encodes structured information about the error number.
17201 * /etc/crypttab entries now understand the new keyfile-size=
17204 * shutdown(8) now can send a (configurable) wall message when
17205 a shutdown is cancelled.
17207 * The mount propagation mode for the root file system will now
17208 default to "shared", which is useful to make containers work
17209 nicely out-of-the-box so that they receive new mounts from
17210 the host. This can be undone locally by running "mount
17211 --make-rprivate /" if needed.
17213 * The prefdm.service file has been removed. Distributions
17214 should maintain this unit downstream if they intend to keep
17215 it around. However, we recommend writing normal unit files
17216 for display managers instead.
17218 * Since systemd is a crucial part of the OS we will now
17219 default to a number of compiler switches that improve
17220 security (hardening) such as read-only relocations, stack
17221 protection, and suchlike.
17223 * The TimeoutSec= setting for services is now split into
17224 TimeoutStartSec= and TimeoutStopSec= to allow configuration
17225 of individual time outs for the start and the stop phase of
17228 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
17229 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
17230 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
17231 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
17232 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
17233 Gundersen, Zbigniew Jędrzejewski-Szmek
17237 * The journal and id128 C APIs are now fully documented as man
17240 * Extra safety checks have been added when transitioning from
17241 the initial RAM disk to the main system to avoid accidental
17244 * /etc/crypttab entries now understand the new keyfile-offset=
17247 * systemctl -t can now be used to filter by unit load state.
17249 * The journal C API gained the new sd_journal_wait() call to
17250 make writing synchronous journal clients easier.
17252 * journalctl gained the new -D switch to show journals from a
17253 specific directory.
17255 * journalctl now displays a special marker between log
17256 messages of two different boots.
17258 * The journal is now explicitly flushed to /var via a service
17259 systemd-journal-flush.service, rather than implicitly simply
17260 by seeing /var/log/journal to be writable.
17262 * journalctl (and the journal C APIs) can now match for much
17263 more complex expressions, with alternatives and
17266 * When transitioning from the initial RAM disk to the main
17267 system we will now kill all processes in a killing spree to
17268 ensure no processes stay around by accident.
17270 * Three new specifiers may be used in unit files: %u, %h, %s
17271 resolve to the user name, user home directory resp. user
17272 shell. This is useful for running systemd user instances.
17274 * We now automatically rotate journal files if their data
17275 object hash table gets a fill level > 75%. We also size the
17276 hash table based on the configured maximum file size. This
17277 together should lower hash collisions drastically and thus
17278 speed things up a bit.
17280 * journalctl gained the new "--header" switch to introspect
17281 header data of journal files.
17283 * A new setting SystemCallFilters= has been added to services which may
17284 be used to apply deny lists or allow lists to system calls. This is
17285 based on SECCOMP Mode 2 of Linux 3.5.
17287 * nspawn gained a new --link-journal= switch (and quicker: -j)
17288 to link the container journal with the host. This makes it
17289 very easy to centralize log viewing on the host for all
17290 guests while still keeping the journal files separated.
17292 * Many bugfixes and optimizations
17294 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
17295 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
17296 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
17301 * Several tools now understand kernel command line arguments,
17302 which are only read when run in an initial RAM disk. They
17303 usually follow closely their normal counterparts, but are
17306 * There's a new tool to analyze the readahead files that are
17307 automatically generated at boot. Use:
17309 /usr/lib/systemd/systemd-readahead analyze /.readahead
17311 * We now provide an early debug shell on tty9 if this enabled. Use:
17313 systemctl enable debug-shell.service
17315 * All plymouth related units have been moved into the Plymouth
17316 package. Please make sure to upgrade your Plymouth version
17319 * systemd-tmpfiles now supports getting passed the basename of
17320 a configuration file only, in which case it will look for it
17321 in all appropriate directories automatically.
17323 * udevadm info now takes a /dev or /sys path as argument, and
17324 does the right thing. Example:
17326 udevadm info /dev/sda
17327 udevadm info /sys/class/block/sda
17329 * systemctl now prints a warning if a unit is stopped but a
17330 unit that might trigger it continues to run. Example: a
17331 service is stopped but the socket that activates it is left
17334 * "systemctl status" will now mention if the log output was
17335 shortened due to rotation since a service has been started.
17337 * The journal API now exposes functions to determine the
17338 "cutoff" times due to rotation.
17340 * journald now understands SIGUSR1 and SIGUSR2 for triggering
17341 immediately flushing of runtime logs to /var if possible,
17342 resp. for triggering immediate rotation of the journal
17345 * It is now considered an error if a service is attempted to
17346 be stopped that is not loaded.
17348 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
17350 * systemd-analyze now supports Python 3
17352 * tmpfiles now supports cleaning up directories via aging
17353 where the first level dirs are always kept around but
17354 directories beneath it automatically aged. This is enabled
17355 by prefixing the age field with '~'.
17357 * Seat objects now expose CanGraphical, CanTTY properties
17358 which is required to deal with very fast bootups where the
17359 display manager might be running before the graphics drivers
17360 completed initialization.
17362 * Seat objects now expose a State property.
17364 * We now include RPM macros for service enabling/disabling
17365 based on the preset logic. We recommend RPM based
17366 distributions to make use of these macros if possible. This
17367 makes it simpler to reuse RPM spec files across
17370 * We now make sure that the collected systemd unit name is
17371 always valid when services log to the journal via
17374 * There's a new man page kernel-command-line(7) detailing all
17375 command line options we understand.
17377 * The fstab generator may now be disabled at boot by passing
17378 fstab=0 on the kernel command line.
17380 * A new kernel command line option modules-load= is now understood
17381 to load a specific kernel module statically, early at boot.
17383 * Unit names specified on the systemctl command line are now
17384 automatically escaped as needed. Also, if file system or
17385 device paths are specified they are automatically turned
17386 into the appropriate mount or device unit names. Example:
17388 systemctl status /home
17389 systemctl status /dev/sda
17391 * The SysVConsole= configuration option has been removed from
17392 system.conf parsing.
17394 * The SysV search path is no longer exported on the D-Bus
17397 * The Names= option has been removed from unit file parsing.
17399 * There's a new man page bootup(7) detailing the boot process.
17401 * Every unit and every generator we ship with systemd now
17402 comes with full documentation. The self-explanatory boot is
17405 * A couple of services gained "systemd-" prefixes in their
17406 name if they wrap systemd code, rather than only external
17407 code. Among them fsck@.service which is now
17408 systemd-fsck@.service.
17410 * The HaveWatchdog property has been removed from the D-Bus
17413 * systemd.confirm_spawn= on the kernel command line should now
17416 * There's a new man page crypttab(5) which details all options
17417 we actually understand.
17419 * systemd-nspawn gained a new --capability= switch to pass
17420 additional capabilities to the container.
17422 * timedated will now read known NTP implementation unit names
17423 from /usr/lib/systemd/ntp-units.d/*.list,
17424 systemd-timedated-ntp.target has been removed.
17426 * journalctl gained a new switch "-b" that lists log data of
17427 the current boot only.
17429 * The notify socket is in the abstract namespace again, in
17430 order to support daemons which chroot() at start-up.
17432 * There is a new Storage= configuration option for journald
17433 which allows configuration of where log data should go. This
17434 also provides a way to disable journal logging entirely, so
17435 that data collected is only forwarded to the console, the
17436 kernel log buffer or another syslog implementation.
17438 * Many bugfixes and optimizations
17440 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
17441 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
17442 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
17443 Shawn Landden, Tom Gundersen
17447 * "systemctl help <unit>" now shows the man page if one is
17450 * Several new man pages have been added.
17452 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
17453 MaxLevelConsole= can now be specified in
17454 journald.conf. These options allow reducing the amount of
17455 data stored on disk or forwarded by the log level.
17457 * TimerSlackNSec= can now be specified in system.conf for
17458 PID1. This allows system-wide power savings.
17460 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
17461 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
17466 * logind is now capable of (optionally) handling power and
17467 sleep keys as well as the lid switch.
17469 * journalctl now understands the syntax "journalctl
17470 /usr/bin/avahi-daemon" to get all log output of a specific
17473 * CapabilityBoundingSet= in system.conf now also influences
17474 the capability bound set of usermode helpers of the kernel.
17476 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
17477 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
17478 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
17479 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
17483 * Note that we skipped 139 releases here in order to set the
17484 new version to something that is greater than both udev's
17485 and systemd's most recent version number.
17487 * udev: all udev sources are merged into the systemd source tree now.
17488 All future udev development will happen in the systemd tree. It
17489 is still fully supported to use the udev daemon and tools without
17490 systemd running, like in initramfs or other init systems. Building
17491 udev though, will require the *build* of the systemd tree, but
17492 udev can be properly *run* without systemd.
17494 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
17495 should be used to create dead device nodes as workarounds for broken
17498 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
17499 no longer supported. udev_monitor_new_from_netlink() needs to be
17500 used to subscribe to events.
17502 * udev: when udevd is started by systemd, processes which are left
17503 behind by forking them off of udev rules, are unconditionally cleaned
17504 up and killed now after the event handling has finished. Services or
17505 daemons must be started as systemd services. Services can be
17506 pulled-in by udev to get started, but they can no longer be directly
17507 forked by udev rules.
17509 * udev: the daemon binary is called systemd-udevd now and installed
17510 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
17511 to adapt to that, create symlink, or rename the binary after building
17514 * libudev no longer provides these symbols:
17515 udev_monitor_from_socket()
17516 udev_queue_get_failed_list_entry()
17517 udev_get_{dev,sys,run}_path()
17518 The versions number was bumped and symbol versioning introduced.
17520 * systemd-loginctl and systemd-journalctl have been renamed
17521 to loginctl and journalctl to match systemctl.
17523 * The config files: /etc/systemd/systemd-logind.conf and
17524 /etc/systemd/systemd-journald.conf have been renamed to
17525 logind.conf and journald.conf. Package updates should rename
17526 the files to the new names on upgrade.
17528 * For almost all files the license is now LGPL2.1+, changed
17529 from the previous GPL2.0+. Exceptions are some minor stuff
17530 of udev (which will be changed to LGPL2.1 eventually, too),
17531 and the MIT licensed sd-daemon.[ch] library that is suitable
17532 to be used as drop-in files.
17534 * systemd and logind now handle system sleep states, in
17535 particular suspending and hibernating.
17537 * logind now implements a sleep/shutdown/idle inhibiting logic
17538 suitable for a variety of uses. Soonishly Lennart will blog
17539 about this in more detail.
17541 * var-run.mount and var-lock.mount are no longer provided
17542 (which previously bind mounted these directories to their new
17543 places). Distributions which have not converted these
17544 directories to symlinks should consider stealing these files
17545 from git history and add them downstream.
17547 * We introduced the Documentation= field for units and added
17548 this to all our shipped units. This is useful to make it
17549 easier to explore the boot and the purpose of the various
17552 * All smaller setup units (such as
17553 systemd-vconsole-setup.service) now detect properly if they
17554 are run in a container and are skipped when
17555 appropriate. This guarantees an entirely noise-free boot in
17556 Linux container environments such as systemd-nspawn.
17558 * A framework for implementing offline system updates is now
17559 integrated, for details see:
17560 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
17562 * A new service type Type=idle is available now which helps us
17563 avoiding ugly interleaving of getty output and boot status
17566 * There's now a system-wide CapabilityBoundingSet= option to
17567 globally reduce the set of capabilities for the
17568 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
17569 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
17570 even CAP_NET_ADMIN system-wide for secure systems.
17572 * There are now system-wide DefaultLimitXXX= options to
17573 globally change the defaults of the various resource limits
17574 for all units started by PID 1.
17576 * Harald Hoyer's systemd test suite has been integrated into
17577 systemd which allows easy testing of systemd builds in qemu
17578 and nspawn. (This is really awesome! Ask us for details!)
17580 * The fstab parser is now implemented as generator, not inside
17583 * systemctl will now warn you if .mount units generated from
17584 /etc/fstab are out of date due to changes in fstab that
17585 have not been read by systemd yet.
17587 * systemd is now suitable for usage in initrds. Dracut has
17588 already been updated to make use of this. With this in place
17589 initrds get a slight bit faster but primarily are much
17590 easier to introspect and debug since "systemctl status" in
17591 the host system can be used to introspect initrd services,
17592 and the journal from the initrd is kept around too.
17594 * systemd-delta has been added, a tool to explore differences
17595 between user/admin configuration and vendor defaults.
17597 * PrivateTmp= now affects both /tmp and /var/tmp.
17599 * Boot time status messages are now much prettier and feature
17600 proper english language. Booting up systemd has never been
17603 * Read-ahead pack files now include the inode number of all
17604 files to pre-cache. When the inode changes the pre-caching
17605 is not attempted. This should be nicer to deal with updated
17606 packages which might result in changes of read-ahead
17609 * We now temporaritly lower the kernel's read_ahead_kb variable
17610 when collecting read-ahead data to ensure the kernel's
17611 built-in read-ahead does not add noise to our measurements
17612 of necessary blocks to pre-cache.
17614 * There's now RequiresMountsFor= to add automatic dependencies
17615 for all mounts necessary for a specific file system path.
17617 * MountAuto= and SwapAuto= have been removed from
17618 system.conf. Mounting file systems at boot has to take place
17621 * nspawn now learned a new switch --uuid= to set the machine
17622 ID on the command line.
17624 * nspawn now learned the -b switch to automatically search
17625 for an init system.
17627 * vt102 is now the default TERM for serial TTYs, upgraded from
17630 * systemd-logind now works on VT-less systems.
17632 * The build tree has been reorganized. The individual
17633 components now have directories of their own.
17635 * A new condition type ConditionPathIsReadWrite= is now available.
17637 * nspawn learned the new -C switch to create cgroups for the
17638 container in other hierarchies.
17640 * We now have support for hardware watchdogs, configurable in
17643 * The scheduled shutdown logic now has a public API.
17645 * We now mount /tmp as tmpfs by default, but this can be
17646 masked and /etc/fstab can override it.
17648 * Since udisks does not make use of /media anymore we are not
17649 mounting a tmpfs on it anymore.
17651 * journalctl gained a new --local switch to only interleave
17652 locally generated journal files.
17654 * We can now load the IMA policy at boot automatically.
17656 * The GTK tools have been split off into a systemd-ui.
17658 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
17659 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
17660 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
17661 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
17662 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
17663 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
17664 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
17665 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
17666 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
17671 * This is mostly a bugfix release
17673 * Support optional initialization of the machine ID from the
17674 KVM or container configured UUID.
17676 * Support immediate reboots with "systemctl reboot -ff"
17678 * Show /etc/os-release data in systemd-analyze output
17680 * Many bugfixes for the journal, including endianness fixes and
17681 ensuring that disk space enforcement works
17683 * sd-login.h is C++ compatible again
17685 * Extend the /etc/os-release format on request of the Debian
17688 * We now refuse non-UTF8 strings used in various configuration
17689 and unit files. This is done to ensure we do not pass invalid
17690 data over D-Bus or expose it elsewhere.
17692 * Register Mimo USB Screens as suitable for automatic seat
17695 * Read SELinux client context from journal clients in a race
17698 * Reorder configuration file lookup order. /etc now always
17699 overrides /run in order to allow the administrator to always
17700 and unconditionally override vendor-supplied or
17701 automatically generated data.
17703 * The various user visible bits of the journal now have man
17704 pages. We still lack man pages for the journal API calls
17707 * We now ship all man pages in HTML format again in the
17710 Contributions from: Dave Reisner, Dirk Eibach, Frederic
17711 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
17712 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
17717 * This is mostly a bugfix release
17719 * systems lacking /etc/os-release are no longer supported.
17721 * Various functionality updates to libsystemd-login.so
17723 * Track class of PAM logins to distinguish greeters from
17724 normal user logins.
17726 Contributions from: Kay Sievers, Lennart Poettering, Michael
17731 * This is an important bugfix release for v41.
17733 * Building man pages is now optional which should be useful
17734 for those building systemd from git but unwilling to install
17737 * Watchdog support for supervising services is now usable. In
17738 a future release support for hardware watchdogs
17739 (i.e. /dev/watchdog) will be added building on this.
17741 * Service start rate limiting is now configurable and can be
17742 turned off per service. When a start rate limit is hit a
17743 reboot can automatically be triggered.
17745 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
17747 Contributions from: Benjamin Franzke, Bill Nottingham,
17748 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
17749 Schmidt, Michał Górny, Piotr Drąg
17753 * The systemd binary is installed /usr/lib/systemd/systemd now;
17754 An existing /sbin/init symlink needs to be adapted with the
17757 * The code that loads kernel modules has been ported to invoke
17758 libkmod directly, instead of modprobe. This means we do not
17759 support systems with module-init-tools anymore.
17761 * Watchdog support is now already useful, but still not
17764 * A new kernel command line option systemd.setenv= is
17765 understood to set system wide environment variables
17766 dynamically at boot.
17768 * We now limit the set of capabilities of systemd-journald.
17770 * We now set SIGPIPE to ignore by default, since it only is
17771 useful in shell pipelines, and has little use in general
17772 code. This can be disabled with IgnoreSIPIPE=no in unit
17775 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
17776 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
17781 * This is mostly a bugfix release
17783 * We now expose the reason why a service failed in the
17784 "Result" D-Bus property.
17786 * Rudimentary service watchdog support (will be completed over
17787 the next few releases.)
17789 * When systemd forks off in order execute some service we will
17790 now immediately changes its argv[0] to reflect which process
17791 it will execute. This is useful to minimize the time window
17792 with a generic argv[0], which makes bootcharts more useful
17794 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
17795 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
17796 Mike Kazantsev, Ray Strode
17800 * This is mostly a test release, but incorporates many
17803 * New systemd-cgtop tool to show control groups by their
17806 * Linking against libacl for ACLs is optional again. If
17807 disabled, support tracking device access for active logins
17808 goes becomes unavailable, and so does access to the user
17809 journals by the respective users.
17811 * If a group "adm" exists, journal files are automatically
17812 owned by them, thus allow members of this group full access
17813 to the system journal as well as all user journals.
17815 * The journal now stores the SELinux context of the logging
17816 client for all entries.
17818 * Add C++ inclusion guards to all public headers
17820 * New output mode "cat" in the journal to print only text
17821 messages, without any meta data like date or time.
17823 * Include tiny X server wrapper as a temporary stop-gap to
17824 teach XOrg udev display enumeration. This is used by display
17825 managers such as gdm, and will go away as soon as XOrg
17826 learned native udev hotplugging for display devices.
17828 * Add new systemd-cat tool for executing arbitrary programs
17829 with STDERR/STDOUT connected to the journal. Can also act as
17830 BSD logger replacement, and does so by default.
17832 * Optionally store all locally generated coredumps in the
17833 journal along with meta data.
17835 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
17836 writing short strings to files (for usage for /sys), and for
17837 creating symlinks, character and block device nodes.
17839 * New unit file option ControlGroupPersistent= to make cgroups
17840 persistent, following the mechanisms outlined in
17841 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
17843 * Support multiple local RTCs in a sane way
17845 * No longer monopolize IO when replaying readahead data on
17846 rotating disks, since we might starve non-file-system IO to
17847 death, since fanotify() will not see accesses done by blkid,
17850 * Do not show kernel threads in systemd-cgls anymore, unless
17851 requested with new -k switch.
17853 Contributions from: Dan Horák, Kay Sievers, Lennart
17854 Poettering, Michal Schmidt
17858 * This is mostly a test release, but incorporates many
17861 * The git repository moved to:
17862 git://anongit.freedesktop.org/systemd/systemd
17863 ssh://git.freedesktop.org/git/systemd/systemd
17865 * First release with the journal
17866 https://0pointer.de/blog/projects/the-journal.html
17868 * The journal replaces both systemd-kmsg-syslogd and
17869 systemd-stdout-bridge.
17871 * New sd_pid_get_unit() API call in libsystemd-logind
17873 * Many systemadm clean-ups
17875 * Introduce remote-fs-pre.target which is ordered before all
17876 remote mounts and may be used to start services before all
17879 * Added Mageia support
17881 * Add bash completion for systemd-loginctl
17883 * Actively monitor PID file creation for daemons which exit in
17884 the parent process before having finished writing the PID
17885 file in the daemon process. Daemons which do this need to be
17886 fixed (i.e. PID file creation must have finished before the
17887 parent exits), but we now react a bit more gracefully to them.
17889 * Add colourful boot output, mimicking the well-known output
17890 of existing distributions.
17892 * New option PassCredentials= for socket units, for
17893 compatibility with a recent kernel ABI breakage.
17895 * /etc/rc.local is now hooked in via a generator binary, and
17896 thus will no longer act as synchronization point during
17899 * systemctl list-unit-files now supports --root=.
17901 * systemd-tmpfiles now understands two new commands: z, Z for
17902 relabelling files according to the SELinux database. This is
17903 useful to apply SELinux labels to specific files in /sys,
17904 among other things.
17906 * Output of SysV services is now forwarded to both the console
17907 and the journal by default, not only just the console.
17909 * New man pages for all APIs from libsystemd-login.
17911 * The build tree got reorganized and the build system is a
17912 lot more modular allowing embedded setups to specifically
17913 select the components of systemd they are interested in.
17915 * Support for Linux systems lacking the kernel VT subsystem is
17918 * configure's --with-rootdir= got renamed to
17919 --with-rootprefix= to follow the naming used by udev and
17922 * Unless specified otherwise we will now install to /usr instead
17923 of /usr/local by default.
17925 * Processes with '@' in argv[0][0] are now excluded from the
17926 final shut-down killing spree, following the logic explained
17928 https://systemd.io/ROOT_STORAGE_DAEMONS/
17930 * All processes remaining in a service cgroup when we enter
17931 the START or START_PRE states are now killed with
17932 SIGKILL. That means it is no longer possible to spawn
17933 background processes from ExecStart= lines (which was never
17934 supported anyway, and bad style).
17936 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
17937 reloading of units together.
17939 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
17940 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
17941 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
17942 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
17943 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek