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 * Support for System V service scripts is deprecated and will be
19 removed in a future release. Please make sure to update your software
20 *now* to include a native systemd unit file instead of a legacy
21 System V script to retain compatibility with future systemd releases.
23 * Support for the SystemdOptions EFI variable is deprecated.
24 'bootctl systemd-efi-options' will emit a warning when used. It seems
25 that this feature is little-used and it is better to use alternative
26 approaches like credentials and confexts. The plan is to drop support
27 altogether at a later point, but this might be revisited based on
30 * systemd-run's switch --expand-environment= which currently is disabled
31 by default when combined with --scope, will be changed in a future
32 release to be enabled by default.
34 * Previously, systemd-networkd did not explicitly remove any bridge
35 VLAN IDs assigned on bridge master and ports. Since version 256, if a
36 .network file for an interface has at least one valid setting in the
37 [BridgeVLAN] section, then all assigned VLAN IDs on the interface
38 that are not configured in the .network file are removed.
40 * systemd-gpt-auto-generator will stop generating units for ESP or
41 XBOOTLDR partitions if it finds mount entries for or below the /boot/
42 or /efi/ hierarchies in /etc/fstab. This is to prevent the generator
43 from interfering with systems where the ESP is explicitly configured
44 to be mounted at some path, for example /boot/efi/ (this type of
45 setup is obsolete, but still commonly found).
47 * The behavior of systemd-sleep and systemd-homed has been updated to
48 freeze user sessions when entering the various sleep modes or when
49 locking a homed-managed home area. This is known to cause issues with
50 the proprietary NVIDIA drivers. Packagers of the NVIDIA proprietary
51 drivers may want to add drop-in configuration files that set
52 SYSTEMD_SLEEP_FREEZE_USER_SESSION=false for systemd-suspend.service
53 and related services, and SYSTEMD_HOME_LOCK_FREEZE_SESSION=false for
54 systemd-homed.service.
56 * systemd-tmpfiles and systemd-sysusers, when given a relative
57 configuration file path (with at least one directory separator '/'),
58 will open the file directly, instead of searching for the given
59 partial path in the standard locations. The old mode wasn't useful
60 because tmpfiles.d/ and sysusers.d/ configuration has a flat
61 structure with no subdirectories under the standard locations and
62 this change makes it easier to work with local files with those
65 * systemd-tmpfiles now properly applies nested configuration to 'R' and
66 'D' stanzas. For example, with the combination of 'R /foo' and 'x
67 /foo/bar', /foo/bar will now be excluded from removal.
69 * systemd.crash_reboot and related settings are deprecated in favor of
70 systemd.crash_action=.
72 General Changes and New Features:
74 * Various programs will now attempt to load the main configuration file
75 from locations below /usr/lib/, /usr/local/lib/, and /run/, not just
76 below /etc/. For example, systemd-logind will look for
77 /etc/systemd/logind.conf, /run/systemd/logind.conf,
78 /usr/local/lib/systemd/logind.conf, and /usr/lib/systemd/logind.conf,
79 and use the first file that is found. This means that the search
80 logic for the main config file and for drop-ins is now the same.
82 Similarly, kernel-install will look for the config files in
83 /usr/lib/kernel/ and the other search locations, and now also
86 systemd-udevd now supports drop-ins for udev.conf.
88 * A new 'systemd-vpick' binary has been added. It implements the new
89 vpick protocol, where a "*.v/" directory may contain multiple files
90 which have versions (following the UAPI version format specification)
91 embedded in the file name. The files are ordered by version and
92 the newest one is selected.
94 systemd-nspawn --image=/--directory=, systemd-dissect,
95 systemd-portabled, and the RootDirectory=, RootImage=,
96 ExtensionImages=, and ExtensionDirectories= settings for units now
97 support the vpick protocol and allow the latest version to be
98 selected automatically if a "*.v/" directory is specified as the
101 * Encrypted service credentials can now be made accessible to
102 unprivileged users. systemd-creds gained new options --user/--uid=
103 for encrypting/decrypting a credential for a specific user.
105 * New command-line tool 'importctl' to download, import, and export
106 disk images via systemd-importd is added with the following verbs:
107 pull-tar, pull-raw, import-tar, import-raw, import-fs, export-tar,
108 export-raw, list-transfers, and cancel-transfer. This functionality
109 was previously available in "machinectl", where it was used
110 exclusively for machine images. The new "importctl" generalizes this
111 for sysext, confext, and portable service images.
113 * The systemd sources may now be compiled cleanly with all OpenSSL 3.0
114 deprecations removed, including the OpenSSL engine logic turned off.
118 * New system manager setting ProtectSystem= has been added. It is
119 analogous to the unit setting, but applies to the whole system. It is
120 enabled by default in the initrd.
122 Note that this means that code executed in the initrd cannot naively
123 expect to be able to write to /usr/ during boot. This affects
124 dracut <= 101, which wrote "hooks" to /lib/dracut/hooks/. See
125 https://github.com/dracut-ng/dracut-ng/commit/a45048b80c27ee5a45a380.
127 * New unit setting WantsMountsFor= has been added. It is analogous to
128 RequiresMountsFor=, but creates a Wants= dependency instead of
129 Requires=. This new logic is now used in various places where mounts
130 were added as dependencies for other settings (WorkingDirectory=-…,
131 PrivateTmp=yes, cryptsetup lines with 'nofail').
133 * New unit setting MemoryZSwapWriteback= can be used to control the new
134 memory.zswap.writeback cgroup knob added in kernel 6.8.
136 * The manager gained a org.freedesktop.systemd1.StartAuxiliaryScope()
137 D-Bus method to devolve some processes from a service into a new
138 scope. This new scope will remain running, even when the original
139 service unit is restarted or stopped. This allows a service unit to
140 split out some worker processes which need to continue running.
141 Control group properties of the new scope are copied from the
142 originating unit, so various limits are retained.
144 * Units now expose properties EffectiveMemoryMax=,
145 EffectiveMemoryHigh=, and EffectiveTasksMax=, which report the
146 most stringent limit systemd is aware of for the given unit.
148 * A new unit file specifier %D expands to $XDG_DATA_HOME (for user
149 services) or /usr/share/ (for system services).
151 * AllowedCPUs= now supports specifier expansion.
153 * What= setting in .mount and .swap units now accepts fstab-style
154 identifiers, for example UUID=… or LABEL=….
156 * RestrictNetworkInterfaces= now supports alternative network interface
159 * PAMName= now implies SetLoginEnvironment=yes.
161 * systemd.firstboot=no can be used on the kernel command-line to
162 disable interactive queries, but allow other first boot configuration
163 to happen based on credentials.
165 * The system's hostname can be configured via the systemd.hostname
168 * The systemd binary will no longer chainload sysvinit's "telinit"
169 binary when called under the init/telinit name on a system that isn't
170 booted with systemd. This previously has been supported to make sure
171 a distribution that has both init systems installed can reasonably
172 switch from one to the other via a simple reboot. Distributions
173 apparently have lost interest in this, and the functionality has not
174 been supported on the primary distribution this was still intended
175 for a long time, and hence has been removed now.
177 * A new concept called "capsules" has been introduced. "Capsules" wrap
178 additional per-user service managers, whose users are transient and
179 are only defined as long as the service manager is running. (This is
180 implemented via DynamicUser=1), allowing a user manager to be used to
181 manager a group of processes without needing to create an actual user
182 account. These service managers run with home directories of
183 /var/lib/capsules/<capsule-name> and can contain regular services and
184 other units. A capsule is started via a simple "systemctl start
185 capsule@<name>.service". See the capsule@.service(5) man page for
188 Various systemd tools (including, and most importantly, systemctl and
189 systemd-run) have been updated to interact with capsules via the new
190 "--capsule="/"-C" switch.
192 * .socket units gained a new setting PassFileDescriptorsToExec=, taking
193 a boolean value. If set to true the file descriptors the socket unit
194 encapsulates are passed to the ExecStartPost=, ExecStopPre=,
195 ExecStopPost= using the usual $LISTEN_FDS interface. This may be used
196 for doing additional initializations on the sockets once they are
197 allocated. (For example, to install an additional eBPF program on
200 * The .socket setting MaxConnectionsPerSource= (which so far put a
201 limit on concurrent connections per IP in Accept=yes socket units),
202 now also has an effect on AF_UNIX sockets: it will put a limit on the
203 number of simultaneous connections from the same source UID (as
204 determined via SO_PEERCRED). This is useful for implementing IPC
205 services in a simple Accept=yes mode.
207 * The service manager will now maintain a counter of soft reboot cycles
208 the system went through. It may be queried via the D-Bus APIs.
210 * systemd's execution logic now supports the new pidfd_spawn() API
211 introduced by glibc 2.39, which allows us to invoke a subprocess in a
212 target cgroup and get a pidfd back in a single operation.
214 * systemd/PID 1 will now send an additional sd_notify() message to its
215 supervising VMM or container manager reporting the selected hostname
216 ("X_SYSTEMD_HOSTNAME=") and machine ID ("X_SYSTEMD_MACHINE_ID=") at
217 boot. Moreover, the service manager will send additional sd_notify()
218 messages ("X_SYSTEMD_UNIT_ACTIVE=") whenever a target unit is
219 reached. This can be used by VMMs/container managers to schedule
220 access to the system precisely. For example, the moment a system
221 reports "ssh-access.target" being reached a VMM/container manager
222 knows it can now connect to the system via SSH. Finally, a new
223 sd_notify() message ("X_SYSTEMD_SIGNALS_LEVEL=2") is sent the moment
224 PID 1 has successfully completed installation of its various UNIX
225 process signal handlers (i.e. the moment where SIGRTMIN+4 sent to
226 PID 1 will start to have the effect of shutting down the system
227 cleanly). X_SYSTEMD_SHUTDOWN= is sent shortly before the system shuts
228 down, and carries a string identifying the type of shutdown,
229 i.e. "poweroff", "halt", "reboot". X_SYSTEMD_REBOOT_PARAMETER= is
230 sent at the same time and carries the string passed to "systemctl
231 --reboot-argument=" if there was one.
233 * New D-Bus properties ExecMainHandoffTimestamp and
234 ExecMainHandoffTimestampMonotonic are now published by services
235 units. This timestamp is taken as the very last operation before
236 handing off control to invoked binaries. This information is
237 available for other unit types that fork off processes (i.e. mount,
238 swap, socket units), but currently only via "systemd-analyze dump".
240 * An additional timestamp is now taken by the service manager when a
241 system shutdown operation is initiated. It can be queried via D-Bus
242 during the shutdown phase. It's passed to the following service
243 manager invocation on soft reboots, which will then use it to log the
244 overall "grey-out" time of the soft reboot operation, i.e. the time
245 when the shutdown began until the system is fully up again.
247 * "systemctl status" will now display the invocation ID in its usual
248 output, i.e. the 128bit ID uniquely assigned to the current runtime
249 cycle of the unit. The ID has been supported for a long time, but is
250 now more prominently displayed, as it is a very useful handle to a
251 specific invocation of a service.
253 * systemd now generates a new "taint" string "unmerged-bin" for systems
254 that have /usr/bin/ and /usr/sbin/ separate. It's generally
255 recommended to make the latter a symlink to the former these days.
257 * A new systemd.crash_action= kernel command line option has been added
258 that configures what to do after the system manager (PID 1) crashes.
259 This can also be configured through CrashAction= in systemd.conf.
261 * "systemctl kill" now supports --wait which will make the command wait
262 until the signalled services terminate.
266 * systemd-journald can now forward journal entries to a socket
267 (AF_INET, AF_INET6, AF_UNIX, or AF_VSOCK). The socket can be
268 specified in journald.conf via a new option ForwardToSocket= or via
269 the 'journald.forward_to_socket' credential. Log records are sent in
270 the Journal Export Format. A related setting MaxLevelSocket= has been
271 added to control the maximum log levels for the messages sent to this
274 * systemd-journald now also reads the journal.storage credential when
275 determining where to store journal files.
277 * systemd-vmspawn gained a new --forward-journal= option to forward the
278 virtual machine's journal entries to the host. This is done over a
279 AF_VSOCK socket, i.e. it does not require networking in the guest.
281 * journalctl gained option '-i' as a shortcut for --file=.
283 * journalctl gained a new -T/--exclude-identifier= option to filter
284 out certain syslog identifiers.
286 * journalctl gained a new --list-namespaces option.
288 * systemd-journal-remote now also accepts AF_VSOCK and AF_UNIX sockets
289 (so it can be used to receive entries forwarded by systemd-journald).
291 * systemd-journal-gatewayd allows restricting the time range of
292 retrieved entries with a new "realtime=[<since>]:[<until>]" URL
295 * systemd-cat gained a new option --namespace= to specify the target
296 journal namespace to which the output shall be connected.
298 * systemd-bsod gained a new option --tty= to specify the output TTY
302 * /dev/ now contains symlinks that combine by-path and by-{label,uuid}
305 /dev/disk/by-path/<path>/by-<label|uuid|…>/<label|uuid|…>
307 This allows distinguishing partitions with identical contents on
308 multiple storage devices. This is useful, for example, when copying
309 raw disk contents between devices.
311 * systemd-udevd now creates persistent /dev/media/by-path/ symlinks for
312 media controllers. For example, the uvcvideo driver may create
313 /dev/media0 which will be linked as
314 /dev/media/by-path/pci-0000:04:00.3-usb-0:1:1.0-media-controller.
316 * A new unit systemd-udev-load-credentials.service has been added
317 to pick up udev.conf drop-ins and udev rules from credentials.
319 * An allowlist/denylist may be specified to filter which sysfs
320 attributes are used when crafting network interface names. Those
321 lists are stored as hwdb entries
322 ID_NET_NAME_ALLOW_<sysfsattr>=0|1
324 ID_NET_NAME_ALLOW=0|1.
326 The goal is to avoid unexpected changes to interface names when the
327 kernel is updated and new sysfs attributes become visible.
329 * A new unit tpm2.target has been added to provide a synchronization
330 point for units which expect the TPM hardware to be available. A new
331 generator "systemd-tpm2-generator" has been added that will insert
332 this target whenever it detects that the firmware has initialized a
333 TPM, but Linux hasn't loaded a driver for it yet.
335 * systemd-backlight now properly supports numbered devices which the
336 kernel creates to avoid collisions in the leds subsystem.
338 * systemd-hwdb update operation can be disabled with a new environment
339 variable SYSTEMD_HWDB_UPDATE_BYPASS=1.
343 * systemd-hostnamed now exposes the machine ID and boot ID via
344 D-Bus. It also exposes the hosts AF_VSOCK CID, if available.
346 * systemd-hostnamed now provides a basic Varlink interface.
348 * systemd-hostnamed exports the full data in os-release(5) and
349 machine-info(5) via D-Bus and Varlink.
351 * hostnamectl now shows the system's product UUID and hardware serial
356 * systemd-networkd now provides a basic Varlink interface.
358 * systemd-networkd's ARP proxy support gained a new option to configure
359 a private VLAN variant of the proxy ARP supported by the kernel under
360 the name IPv4ProxyARPPrivateVLAN=.
362 * systemd-networkd now exports the NamespaceId and NamespaceNSID
363 properties via D-Bus and Varlink. (which expose the inode and NSID of
364 the network namespace the networkd instance manages)
366 * systemd-networkd now supports IPv6RetransmissionTimeSec= and
367 UseRetransmissionTime= settings in .network files to configure
368 retransmission time for IPv6 neighbor solicitation messages.
370 * networkctl gained new verbs 'mask' and 'unmask' for masking networkd
371 configuration files such as .network files.
373 * 'networkctl edit --runtime' allows editing volatile configuration
374 under /run/systemd/network/.
376 * The implementation behind TTLPropagate= network setting has been
377 removed and the setting is now ignored.
379 * systemd-network-generator will now pick up .netdev/.link/.network/
380 networkd.conf configuration from system credentials.
382 * systemd-networkd will now pick up wireguard secrets from
385 * systemd-networkd's Varlink API now supports enumerating LLDP peers.
387 * .link files now support new Property=, ImportProperty=,
388 UnsetProperty= fields for setting udev properties on a link.
390 * The various .link files that systemd ships for interfaces that are
391 supposed to be managed by systemd-networkd only now carry a
392 ID_NET_MANAGED_BY=io.systemd.Network udev property ensuring that
393 other network management solutions honouring this udev property do
394 not come into conflict with networkd, trying to manage these
397 * .link files now support a new ReceivePacketSteeringCPUMask= setting
398 for configuring which CPUs to steer incoming packets to.
400 * The [Network] section in .network files gained a new setting
401 UseDomains=, which is a single generic knob for controlling the
402 settings of the same name in the [DHCPv4], [DHCPv6] and
405 * The 99-default.link file we ship by default (that defines the policy
406 for all network devices to which no other .link file applies) now
407 lists "mac" among AlternativeNamesPolicy=. This means that network
408 interfaces will now by default gain an additional MAC-address based
409 alternative device name. (i.e. enx…)
413 * systemd-nspawn now provides a /run/systemd/nspawn/unix-export/
414 directory where the container payload can expose AF_UNIX sockets to
415 allow them to be accessed from outside.
417 * systemd-nspawn will tint the terminal background for containers in a
418 blueish color. This can be controller with the new --background=
419 switch or the new $SYSTEMD_TINT_BACKGROUND environment variable.
421 * systemd-nspawn gained support for the 'owneridmap' option for --bind=
422 mounts to map the target directory owner from inside the container to
423 the owner of the directory bound from the host filesystem.
425 * systemd-nspawn now supports moving Wi-Fi network devices into a
426 container, just like other network interfaces.
430 * systemd-resolved now reads RFC 8914 EDE error codes provided by
431 upstream DNS services.
433 * systemd-resolved and resolvectl now support RFC 9460 SVCB and HTTPS
434 records, as well as RFC 2915 NAPTR records.
436 * resolvectl gained a new option --relax-single-label= to allow
437 querying single-label hostnames via unicast DNS on a per-query basis.
439 * systemd-resolved's Varlink IPC interface now supports resolving
440 DNS-SD services as well as an API for resolving raw DNS RRs.
442 * systemd-resolved's .dnssd DNS_SD service description files now
443 support DNS-SD "subtypes" via the new SubType= setting.
445 * systemd-resolved's configuration may now be reloaded without
446 restarting the service. (i.e. "systemctl reload systemd-resolved" is
451 * An sshd config drop-in to allow ssh keys acquired via userdbctl (for
452 example expose by homed accounts) to be used for authorization of
453 incoming SSH connections.
455 * A small new unit generator "systemd-ssh-generator" has been added. It
456 checks if the sshd binary is installed. If so, it binds it via
457 per-connection socket activation to various sockets depending on the
460 • If the system is run in a VM providing AF_VSOCK support, it
461 automatically binds sshd to AF_VSOCK port 22.
463 • If the system is invoked as a full-OS container and the container
464 manager pre-mounts a directory /run/host/unix-export/, it will
465 bind sshd to an AF_UNIX socket /run/host/unix-export/ssh. The
466 idea is the container manager bind mounts the directory to an
467 appropriate place on the host as well, so that the AF_UNIX socket
468 may be used to easily connect from the host to the container.
470 • sshd is also bound to an AF_UNIX socket
471 /run/ssh-unix-local/socket, which may be to use ssh/sftp in a
472 "sudo"-like fashion to access resources of other local users.
474 • Via the kernel command line option "systemd.ssh_listen=" and the
475 system credential "ssh.listen" sshd may be bound to additional,
476 explicitly configured options, including AF_INET/AF_INET6 ports.
478 In particular the first two mechanisms should make dealing with local
479 VMs and full OS containers a lot easier, as SSH connections will
480 *just* *work* from the host – even if no networking is available
483 systemd-ssh-generator optionally generates a per-connection
484 socket activation service file wrapping sshd. This is only done if
485 the distribution does not provide one on its own under the name
486 "sshd@.service". The generated unit only works correctly if the SSH
487 privilege separation ("privsep") directory exists. Unfortunately
488 distributions vary wildly where they place this directory. An
489 incomprehensive list:
491 • /usr/share/empty.sshd/ (new fedora)
494 • /run/sshd/ (debian/ubuntu?)
496 If the SSH privsep directory is placed below /var/ or /run/ care
497 needs to be taken that the directory is created automatically at boot
498 if needed, since these directories possibly or always come up
499 empty. This can be done via a tmpfiles.d/ drop-in. You may use the
500 "sshdprivsepdir" meson option provided by systemd to configure the
501 directory, in case you want systemd to create the directory as needed
502 automatically, if your distribution does not cover this natively.
504 Recommendations to distributions, in order to make things just work:
506 • Please provide a per-connection SSH service file under the name
509 • Please move the SSH privsep dir into /usr/ (so that it is truly
510 immutable on image-based operating systems, is strictly under
511 package manager control, and never requires recreation if the
512 system boots up with an empty /run/ or /var/).
514 • As an extension of this: please consider following Fedora's lead
515 here, and use /usr/share/empty.sshd/ to minimize needless
516 differences between distributions.
518 • If your distribution insists on placing the directory in /var/ or
519 /run/ then please at least provide a tmpfiles.d/ drop-in to
520 recreate it automatically at boot, so that the sshd binary just
521 works, regardless in which context it is called.
523 * A small tool "systemd-ssh-proxy" has been added, which is supposed to
524 act as counterpart to "systemd-ssh-generator". It's a small plug-in
525 for the SSH client (via ProxyCommand/ProxyUseFdpass) to allow it to
526 connect to AF_VSOCK or AF_UNIX sockets. Example: "ssh vsock/4711"
527 connects to a local VM with cid 4711, or "ssh
528 unix/run/ssh-unix-local/socket" to connect to the local host via the
529 AF_UNIX socket /run/ssh-unix-local/socket.
531 systemd-boot and systemd-stub and Related Tools:
533 * TPM 1.2 PCR measurement support has been removed from systemd-stub.
534 TPM 1.2 is obsolete and – due to the (by today's standards) weak
535 cryptographic algorithms it only supports – does not actually provide
536 the security benefits it's supposed to provide. Given that the rest
537 of systemd's codebase never supported TPM 1.2, the support has now
538 been removed from systemd-stub as well.
540 * systemd-stub will now measure its payload via the new EFI
541 Confidential Computing APIs (CC), in addition to the pre-existing
544 * confexts are loaded by systemd-stub from the ESP as well.
546 * kernel-install gained support for --root= for the 'list' verb.
548 * bootctl now provides a basic Varlink interface and can be run as a
549 daemon via a template unit.
551 * systemd-measure gained new options --certificate=, --private-key=,
552 and --private-key-source= to allow using OpenSSL's "engines" or
553 "providers" as the signing mechanism to use when creating signed
554 TPM2 PCR measurement values.
556 * ukify gained support for signing of PCR signatures via OpenSSL's
557 engines and providers.
559 * ukify now supports zboot kernels.
561 * systemd-boot now supports passing additional kernel command line
562 switches to invoked kernels via an SMBIOS Type #11 string
563 "io.systemd.boot.kernel-cmdline-extra". This is similar to the
564 pre-existing support for this in systemd-stub, but also applies to
565 Type #1 Boot Loader Specification Entries.
567 * systemd-boot's automatic SecureBoot enrollment support gained support
568 for enrolling "dbx" too (Previously, only db/KEK/PK enrollment was
569 supported). It also now supports UEFI "Custom" and "Audit" modes.
571 * The pcrlock policy is saved in an unencrypted credential file
572 "pcrlock.<entry-token>.cred" under XBOOTLDR/ESP in the
573 /loader/credentials/ directory. It will be picked up at boot by
574 systemd-stub and passed to the initrd, where it can be used to unlock
575 the root file system.
577 * systemd-pcrlock gained an --entry-token= option to configure the
580 * systemd-pcrlock now provides a basic Varlink interface and can be run
581 as a daemon via a template unit.
583 * systemd-pcrlock's TPM nvindex access policy has been modified, this
584 means that previous pcrlock policies stored in nvindexes are
585 invalidated. They must be removed (systemd-pcrlock remove-policy) and
586 recreated (systemd-pcrlock make-policy). For the time being
587 systemd-pcrlock remains an experimental feature, but it is expected
588 to become stable in the next release, i.e. v257.
590 * systemd-pcrlock's --recovery-pin= switch now takes three values:
591 "hide", "show", "query". If "show" is selected the automatically
592 generated recovery PIN is shown to the user. If "query" is selected
593 then the PIN is queried from the user.
595 * sd-stub gained support for the new ".ucode" PE section in UKIs, that
596 may contain CPU microcode data. When control is handed over to the
597 Linux kernel this data is prepended to the set of initrds passed.
601 * systemd-run is now a multi-call binary. When invoked as 'run0', it
602 provides as interface similar to 'sudo', with all arguments starting
603 at the first non-option parameter being treated the command to invoke
604 as root. Unlike 'sudo' and similar tools, it does not make use of
605 setuid binaries or other privilege escalation methods, but instead
606 runs the specified command as a transient unit, which is started by
607 the system service manager, so privileges are dropped, rather than
608 gained, thus implementing a much more robust and safe security
609 model. As usual, authorization is managed via Polkit.
611 * systemd-run/run0 will now tint the terminal background on supported
612 terminals: in a reddish tone when invoking a root service, in a
613 yellowish tone otherwise. This may be controlled and turned off via
614 the new --background= switch or the new $SYSTEMD_TINT_BACKGROUND
615 environment variable.
617 * systemd-run gained a new option '--ignore-failure' to suppress
622 * 'systemctl edit --stdin' allows creation of unit files and drop-ins
623 with contents supplied via standard input. This is useful when creating
624 configuration programmatically; the tool takes care of figuring out
625 the file name, creating any directories, and reloading the manager
628 * 'systemctl disable --now' and 'systemctl mask --now' now work
629 correctly with template units.
631 * 'systemd-analyze architectures' lists known CPU architectures.
633 * 'systemd-analyze --json=…' is supported for 'architectures',
634 'capability', 'exit-status'.
636 * 'systemd-tmpfiles --purge' will purge (remove) all files and
637 directories created via tmpfiles.d configuration.
639 * systemd-id128 gained new options --no-pager, --no-legend, and
642 * hostnamectl gained '-j' as shortcut for '--json=pretty' or
645 * loginctl now supports -j/--json=.
647 * resolvectl now supports -j/--json= for --type=.
649 * systemd-tmpfiles gained a new option --dry-run to print what would be
650 done without actually taking action.
652 * varlinkctl gained a new --collect switch to collect all responses of
653 a method call that supports multiple replies and turns it into a
656 * systemd-dissect gained a new --make-archive option to generate an
657 archive file (tar.gz and similar) from a disk image.
661 * systemd-vmspawn gained a new --firmware= option to configure or list
662 firmware definitions for Qemu, a new --tpm= option to enable or
663 disable the use of a software TPM, a new --linux= option to specify a
664 kernel binary for direct kernel boot, a new --initrd= option to
665 specify an initrd for direct kernel boot, a new -D/--directory option
666 to use a plain directory as the root file system, a new
667 --private-users option similar to the one in systemd-nspawn, new
668 options --bind= and --bind-ro= to bind part of the host's file system
669 hierarchy into the guest, a new --extra-drive= option to attach
670 additional storage, and -n/--network-tap/--network-user-mode to
671 configure networking.
673 * A new systemd-vmspawn@.service can be used to launch systemd-vmspawn
676 * systemd-vmspawn gained the new --console= and --background= switches
677 that control how to interact with the VM. As before, by default an
678 interactive terminal interface is provided, but now with a background
679 tinted with a greenish hue.
681 * systemd-vmspawn can now register its VMs with systemd-machined,
682 controlled via the --register= switch.
684 * machinectl's start command (and related) can now invoke images either
685 as containers via `systemd-nspawn` (switch is --runner=nspawn, the
686 default) or as VMs via `systemd-vmspawn` (switch is --runner=vmspawn,
689 * systemd-vmspawn now supports two switches --pass-ssh-key= and
690 --ssh-key-type= to optionally set up transient SSH keys to pass to the
691 invoked VMs in order to be able to SSH into them once booted.
693 * systemd-vmspawn will now enable various "HyperV enlightenments" and
694 the "VM Generation ID" on the VMs.
696 * A new environment variable $SYSTEMD_VMSPAWN_QEMU_EXTRA may carry
697 additional qemu command line options to pass to qemu.
699 * systemd-machined gained a new GetMachineSSHInfo() D-Bus method that is
700 used by systemd-vmspawn to fetch the information needed to ssh into the
703 * systemd-machined gained a new Varlink interface that is used by
704 systemd-vmspawn to register machines with additional information and
709 * systemd-repart gained new options --generate-fstab= and
710 --generate-crypttab= to write out fstab and crypttab files matching the
711 generated partitions.
713 * systemd-repart gained a new option --private-key-source= to allow
714 using OpenSSL's "engines" or "providers" as the signing mechanism to
715 use when creating verity signature partitions.
717 * systemd-repart gained a new DefaultSubvolume= setting in repart.d/
718 drop-ins that allow configuring the default btrfs subvolume for newly
719 formatted btrfs file systems.
723 * libsystemd gained new call sd_bus_creds_new_from_pidfd() to get a
724 credentials object for a pidfd and sd_bus_creds_get_pidfd_dup() to
725 retrieve the pidfd from a credentials object.
727 * sd-bus' credentials logic will now also acquire peer's UNIX group
728 lists and peer's pidfd if supported and requested.
730 * RPM macro %_kernel_install_dir has been added with the path
731 to the directory for kernel-install plugins.
733 * The liblz4, libzstd, liblzma, libkmod, libgcrypt dependencies have
734 been changed from regular shared library dependencies into dlopen()
737 Note that this means that those libraries might not be automatically
738 pulled in when ELF dependencies are resolved. In particular lack of
739 libkmod might cause problems with boot. This affects dracut <= 101,
740 see https://github.com/dracut-ng/dracut-ng/commit/04b362d713235459cf.
742 * systemd ELF binaries that use libraries via dlopen() are now built with
743 a new ELF header note section, following a new specification defined at
744 docs/ELF_DLOPEN_METADATA.md, that provides information about which
745 sonames are loaded and used if found at runtime. This allows tools and
746 packagers to programmatically discover the list of optional
747 dependencies used by all systemd ELF binaries. A parser with packaging
748 integration tools is available at
749 https://github.com/systemd/package-notes
751 * The sd-journal API gained a new call
752 sd_journal_stream_fd_with_namespace() which is just like
753 sd_journal_stream_fd() but creates a log stream targeted at a
754 specific log namespace.
756 * The sd-id128 API gained a new API call
757 sd_id128_get_invocation_app_specific() for acquiring an app-specific
758 ID that is derived from the service invocation ID.
760 * The sd-event API gained a new API call
761 sd_event_source_get_inotify_path() that returns the file system path
762 an inotify event source was created for.
764 systemd-cryptsetup/systemd-cryptenroll:
766 * The device node argument to systemd-cryptenroll is now optional. If
767 omitted it will be derived automatically from the backing block
768 device of /var/ (which quite likely is the same as the root file
769 system, hence effectively means if you don't specify things otherwise
770 the tool will now default to enrolling a key into the root file
771 system's LUKS device).
773 * systemd-cryptenroll can now enroll directly with a PKCS11 public key
774 (instead of a certificate).
776 * systemd-cryptsetup/systemd-cryptenroll now may lock a disk against a
777 PKCS#11 provided EC key (before it only supported RSA).
779 * systemd-cryptsetup gained support for crypttab option
780 link-volume-key= to link the volume key into the kernel keyring when
781 the volume is opened.
783 * systemd-cryptenroll will no longer enable Dictionary Attack
784 Protection (i.e. turn on NO_DA) for TPM enrollments that do not
785 involve a PIN. DA should not be necessary in that case (since key
786 entropy is high enough to make this unnecessary), but risks
787 accidental lock-out in case of unexpected PCR changes.
789 * systemd-cryptenroll now supports enrolling a new slot while unlocking
790 the old slot via TPM2 (previously unlocking only worked via password
795 * The remaining documentation that was on
796 https://freedesktop.org/wiki/Software/systemd/ has been moved to
799 * A new text describing the VM integration interfaces of systemd has
802 https://systemd.io/VM_INTERFACE
804 * The sd_notify() man page has gained examples with C and Python code
805 that shows how to implement the interface in those languages without
806 involving libsystemd.
808 systemd-homed, systemd-logind, systemd-userdbd:
810 * systemd-homed now supports unlocking of home directories when logging
811 in via SSH. Previously home directories needed to be unlocked before
812 an SSH login is attempted.
814 * JSON User Records have been extended with a separate public storage
815 area called "User Record Blob Directories". This is intended to store
816 the user's background image, avatar picture, and other similar items
817 which are too large to fit into the User Record itself.
819 systemd-homed, userdbctl, and homectl gained support for blob
820 directories. homectl gained --avatar= and --login-background= to
821 control two specific items of the blob directories.
823 * A new "additionalLanguages" field has been added to JSON user records
824 (as supported by systemd-homed and systemd-userdbd), which is closely
825 related to the pre-existing "preferredLanguage", and allows
826 specifying multiple additional languages for the user account. It is
827 used to initialize the $LANGUAGES environment variable when used.
829 * A new pair of "preferredSessionType" and "preferredSessionLauncher"
830 fields have been added to JSON user records, that may be used to
831 control which kind of desktop session to preferable activate on
834 * homectl gained a new verb 'firstboot', and a new
835 systemd-homed-firstboot.service unit uses this verb to create users
836 in a first boot environment, either from system credentials or by
837 querying interactively.
839 * systemd-logind now supports a new "background-light" session class
840 which does not pull in the user@.service unit. This is intended in
841 particular for lighter weight per-user cron jobs which do require any
842 per-user service manager to be around.
844 * The per-user service manager will now be tracked as a distinct "manager"
845 session type among logind sessions of each user.
847 * homectl now supports an --offline mode, by which certain account
848 properties can be changed without unlocking the home directory.
850 * systemd-logind gained a new
851 org.freedesktop.login1.Manager.ListSessionsEx() method that provides
852 additional metadata compared to ListSessions(). loginctl makes use of
853 this to list additional fields in list-sessions.
855 * systemd-logind gained a new org.freedesktop.login1.Manager.Sleep()
856 method that automatically redirects to SuspendThenHibernate(),
857 Suspend(), HybridSleep(), or Hibernate(), depending on what is
858 supported and configured, a new configuration setting SleepOperation=,
859 and an accompanying helper method
860 org.freedesktop.login1.Manager.CanSleep() and property
861 org.freedesktop.login1.Manager.SleepOperation.
863 'systemctl sleep' calls the new method to automatically put the
864 machine to sleep in the most appropriate way.
866 Credential Management:
868 * systemd-creds now provides a Varlink IPC API for encrypting and
869 decrypting credentials.
871 * systemd-creds' "tpm2-absent" key selection has been renamed to
872 "null", since that's what it actually does: "encrypt" and "sign"
873 with a fixed null key. --with-key=null should only be used in very
874 specific cases, as it provides zero integrity or confidentiality
875 protections. (i.e. it's only safe to use as fallback in environments
876 lacking both a TPM and access to the root fs to use the host
877 encryption key, or when integrity is provided some other way.)
879 * systemd-creds gained a new switch --allow-null. If specified, the
880 "decrypt" verb will decode encrypted credentials that use the "null"
881 key (by default this is refused, since using the "null" key defeats
882 the authenticated encryption normally done).
886 * The sleep.conf configuration file gained a new MemorySleepMode=
887 setting for configuring the sleep mode in more detail.
889 * A tiny new service systemd-hibernate-clear.service has been added
890 which clears hibernation information from the HibernateLocation EFI
891 variable, in case the resume device is gone. Normally, this variable
892 is supposed to be cleaned up by the code that initiates the resume
893 from hibernation image. But when the device is missing and that code
894 doesn't run, this service will now do the necessary work, ensuring
895 that no outdated hibernation image information remains on subsequent
898 Unprivileged User Namespaces & Mounts:
900 * A small new service systemd-nsresourced.service has been added. It
901 provides a Varlink IPC API that assigns a free, transiently allocated
902 64K UID/GID range to an uninitialized user namespace a client
903 provides. It may be used to implement unprivileged container managers
904 and other programs that need dynamic user ID ranges. It also provides
905 interfaces to then delegate mount file descriptors, control groups
906 and network interfaces to user namespaces set up this way.
908 * A small new service systemd-mountfsd.service has been added. It
909 provides a Varlink IPC API for mounting DDI images, and returning a set
910 of mount file descriptors for it. If a user namespace fd is provided
911 as input, then the mounts are registered with the user namespace. To
912 ensure trust in the image it must provide Verity information (or
913 alternatively interactive polkit authentication is required).
915 * The systemd-dissect tool now can access DDIs fully unprivileged by
916 using systemd-nsresourced/systemd-mountfsd.
918 * If the service manager runs unprivileged (i.e. systemd --user) it now
919 supports RootImage= for accessing DDI images, also implemented via
920 the systemd-nsresourced/systemd-mountfsd.
922 * systemd-nspawn may now operate without privileges, if a suitable DDI
923 is provided via --image=, again implemented via
924 systemd-nsresourced/systemd-mountfsd.
928 * timedatectl and machinectl gained option '-P', an alias for
929 '--value --property=…'.
931 * Various tools that pretty-print config files will now highlight
932 configuration directives.
934 * varlinkctl gained support for the "ssh:" transport. This requires
935 OpenSSH 9.4 or newer.
937 * systemd-sysext gained support for enabling system extensions in
938 mutable fashion, where a writeable upperdir is stored under
939 /var/lib/extensions.mutable/, and a new --mutable= option to
940 configure this behaviour. An "ephemeral" mode is not also supported
941 where the mutable layer is configured to be a tmpfs that is
942 automatically released when the system extensions are reattached.
944 * Coredumps are now retained for two weeks by default (instead of three
947 * portablectl --copy= parameter gained a new 'mixed' argument, that will
948 result in resources owned by the OS (e.g.: portable profiles) to be linked
949 but resources owned by the portable image (e.g.: the unit files and the
950 images themselves) to be copied.
952 * systemd will now register MIME types for various of its file types
953 (e.g. journal files, DDIs, encrypted credentials …) via the XDG
954 shared-mime-info infrastructure. (Files of these types will thus be
955 recognized as their own thing in desktop file managers such as GNOME
958 * systemd-dissect will now show the detected sector size of a given DDI
959 in its default output.
961 * systemd-portabled now generates recognizable structured log messages
962 whenever a portable service is attached or detached.
964 * Verity signature checking in userspace (i.e. checking against
965 /etc/verity.d/ keys) when activating DDIs can now be turned on/off
966 via a kernel command line option systemd.allow_userspace_verity= and
967 an environment variable SYSTEMD_ALLOW_USERSPACE_VERITY=.
969 * ext4/xfs file system quota handling has been reworked, so that
970 quotacheck and quotaon are now invoked as per-file-system templated
971 services (as opposed to single system-wide singletons), similar in
972 style to the fsck, growfs, pcrfs logic. This means file systems with
973 quota enabled can now be reasonably enabled at runtime of the system,
976 * "systemd-analyze dot" will now also show BindsTo= dependencies.
978 * systemd-debug-generator gained the ability add in arbitrary units
979 based on them being passed in via system credentials.
981 * A new kernel command-line option systemd.default_debug_tty= can be
982 used to specify the TTY for the debug shell, independently of
983 enabling or disabling it.
985 * portablectl gained a new --clean switch that clears a portable
986 service's data (cache, logs, state, runtime, fdstore) when detaching
989 Contributions from: A S Alam, AKHIL KUMAR,
990 Abraham Samuel Adekunle, Adrian Vovk, Adrian Wannenmacher,
991 Alan Liang, Alberto Planas, Alexander Zavyalov, Anders Jonsson,
992 Andika Triwidada, Andres Beltran, Andrew Sayers,
993 Antonio Alvarez Feijoo, Arthur Zamarin, Artur Pak, AtariDreams,
994 Benjamin Franzke, Bernhard M. Wiedemann, Black-Hole1, Bryan Jacobs,
995 Burak Gerz, Carlos Garnacho, Chandra Pratap, Chris Simons,
996 Christian Wesselhoeft, Clayton Craft, Colin Geniet, Colin Walters,
997 Costa Tsaousis, Cristian Rodríguez, Daan De Meyer,
998 Damien Challet, Dan Streetman, David Tardon, David Venhoek,
999 Diego Viola, Dionna Amalie Glaze, Dmitry Konishchev,
1000 Edson Juliano Drosdeck, Eisuke Kawashima, Eli Schwartz,
1001 Emanuele Giuseppe Esposito, Eric Daigle, Evgeny Vereshchagin,
1002 Felix Riemann, Fernando Fernandez Mancera, Florian Schmaus,
1003 Franck Bui, Frantisek Sumsal, Friedrich Altheide,
1004 Gabríel Arthúr Pétursson, Gaël Donval, Georges Basile Stavracas Neto,
1005 Gerd Hoffmann, GNOME Foundation, Guido Leenders,
1006 Guilhem Lettron, Göran Uddeborg, Hans de Goede, Harald Brinkmann,
1007 Heinrich Schuchardt, Henry Li, Holger Assmann, Ivan Kruglov,
1008 Ivan Shapovalov, Jakub Sitnicki, James Muir, Jan Engelhardt,
1009 Jan Macku, Jeff King, JmbFountain, Joakim Nohlgård,
1010 Jonathan Conder, Julius Alexandre, Jörg Behrmann, Keian, Kirk,
1011 Kristian Klausen, Krzesimir Nowak, Lars Ellenberg,
1012 Lennart Poettering, Luca Boccassi, Ludwig Nussel, Lukáš Nykrýn,
1013 Luna Jernberg, Luxiter, Maanya Goenka, Mariano Giménez,
1014 Markus Merklinger, Martin Ivicic, Martin Srebotnjak,
1015 Martin Trigaux, Martin Wilck, Matt Layher, Matt Muggeridge,
1016 Matteo Croce, Matthias Lisin, Max Gautier, Max Staudt, MaxHearnden,
1017 Michael Biebl, Michal Koutný, Michal Sekletár, Mike Gilbert,
1018 Mike Yuan, Mikko Ylinen, MkfsSion, MrSmör, Nandakumar Raghavan,
1019 Nick Cao, Nick Rosbrook, Norbert Lange, Ole Peder Brandtzæg,
1020 Ondrej Kozina, Oğuz Ersen, Pablo Méndez Hernández,
1021 Pierre GRASSER, Piotr Drąg, QuonXF, Rafaël Kooi, Raito Bezarius,
1022 Rasmus Villemoes, Reid Wahl, Renjaya Raga Zenta, Richard Maw,
1023 Roland Hieber, Ronan Pigott, Rose, Ross Burton, Sam Leonard,
1024 Samuel BF, Sarvajith Adyanthaya, Sergei Zhmylev, Sergey A, Shulhan,
1025 SidhuRupinder, Simon Fowler, Sludge, Stuart Hayhurst, Susant Sahani,
1026 Takashi Sakamoto, Temuri Doghonadze, Thilo Fromm, Thomas Blume,
1027 TobiPeterG, Tobias Fleig, Tomáš Pecka, Topi Miettinen,
1028 Tycho Andersen, Unique-Usman, Usman Akinyemi, Vasiliy Kovalev,
1029 Vasiliy Stelmachenok, Vishal Chillara Srinivas, Vitaly Kuznetsov,
1030 Vito Caputo, Vladimir Stoiakin, Werner Sembach, Will Springer,
1031 Winterhuman, Xiaotian Wu, Yu Watanabe, Yuri Chornoivan,
1032 Zbigniew Jędrzejewski-Szmek, Zmyeir, aslepykh, chenjiayi,
1033 cpackham-atlnz, cunshunxia, djantti, hfavisado, hulkoba, ksaleem,
1034 medusalix, mille-feuille, mkubiak, mooo, msizanoen, networkException,
1035 nl6720, r-vdp, runiq, sam-leonard-ct, samuelvw01, sharad3001, sushmbha,
1036 wangyuhang, zzywysm, İ. Ensar Gülşen, Łukasz Stelmach,
1037 Štěpán Němec, 我超厉害, 김인수
1039 — Edinburgh, 2024-04-25
1043 Announcements of Future Feature Removals and Incompatible Changes:
1045 * Support for split-usr (/usr/ mounted separately during late boot,
1046 instead of being mounted by the initrd before switching to the rootfs)
1047 and unmerged-usr (parallel directories /bin/ and /usr/bin/, /lib/ and
1048 /usr/lib/, …) has been removed. For more details, see:
1049 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1051 * We intend to remove cgroup v1 support from a systemd release after
1052 the end of 2023. If you run services that make explicit use of
1053 cgroup v1 features (i.e. the "legacy hierarchy" with separate
1054 hierarchies for each controller), please implement compatibility with
1055 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
1056 Most of Linux userspace has been ported over already.
1058 * Support for System V service scripts is now deprecated and will be
1059 removed in a future release. Please make sure to update your software
1060 *now* to include a native systemd unit file instead of a legacy
1061 System V script to retain compatibility with future systemd releases.
1063 * Support for the SystemdOptions EFI variable is deprecated.
1064 'bootctl systemd-efi-options' will emit a warning when used. It seems
1065 that this feature is little-used and it is better to use alternative
1066 approaches like credentials and confexts. The plan is to drop support
1067 altogether at a later point, but this might be revisited based on
1070 * systemd-run's switch --expand-environment= which currently is disabled
1071 by default when combined with --scope, will be changed in a future
1072 release to be enabled by default.
1074 * "systemctl switch-root" is now restricted to initrd transitions only.
1076 Transitions between real systems should be done with
1077 "systemctl soft-reboot" instead.
1079 * The "ip=off" and "ip=none" kernel command line options interpreted by
1080 systemd-network-generator will now result in IPv6RA + link-local
1081 addressing being disabled, too. Previously DHCP was turned off, but
1082 IPv6RA and IPv6 link-local addressing was left enabled.
1084 * The NAMING_BRIDGE_MULTIFUNCTION_SLOT naming scheme has been deprecated
1085 and is now disabled.
1087 * SuspendMode=, HibernateState= and HybridSleepState= in the [Sleep]
1088 section of systemd-sleep.conf are now deprecated and have no effect.
1089 They did not (and could not) take any value other than the respective
1090 default. HybridSleepMode= is also deprecated, and will now always use
1091 the 'suspend' disk mode.
1095 * The way services are spawned has been overhauled. Previously, a
1096 process was forked that shared all of the manager's memory (via
1097 copy-on-write) while doing all the required setup (e.g.: mount
1098 namespaces, CGroup configuration, etc.) before exec'ing the target
1099 executable. This was problematic for various reasons: several glibc
1100 APIs were called that are not supposed to be used after a fork but
1101 before an exec, copy-on-write meant that if either process (the
1102 manager or the child) touched a memory page a copy was triggered, and
1103 also the memory footprint of the child process was that of the
1104 manager, but with the memory limits of the service. From this version
1105 onward, the new process is spawned using CLONE_VM and CLONE_VFORK
1106 semantics via posix_spawn(3), and it immediately execs a new internal
1107 binary, systemd-executor, that receives the configuration to apply
1108 via memfd, and sets up the process before exec'ing the target
1109 executable. The systemd-executor binary is pinned by file descriptor
1110 by each manager instance (system and users), and the reference is
1111 updated on daemon-reexec - it is thus important to reexec all running
1112 manager instances when the systemd-executor and/or libsystemd*
1113 libraries are updated on the filesystem.
1115 * Most of the internal process tracking is being changed to use PIDFDs
1116 instead of PIDs when the kernel supports it, to improve robustness
1119 * A new option SurviveFinalKillSignal= can be used to configure the
1120 unit to be skipped in the final SIGTERM/SIGKILL spree on shutdown.
1121 This is part of the required configuration to let a unit's processes
1122 survive a soft-reboot operation.
1124 * System extension images (sysext) can now set
1125 EXTENSION_RELOAD_MANAGER=1 in their extension-release files to
1126 automatically reload the service manager (PID 1) when
1127 merging/refreshing/unmerging on boot. Generally, while this can be
1128 used to ship services in system extension images it's recommended to
1129 do that via portable services instead.
1131 * The ExtensionImages= and ExtensionDirectories= options now support
1132 confexts images/directories.
1134 * A new option NFTSet= provides a method for integrating dynamic cgroup
1135 IDs into firewall rules with NFT sets. The benefit of using this
1136 setting is to be able to use control group as a selector in firewall
1137 rules easily and this in turn allows more fine grained filtering.
1138 Also, NFT rules for cgroup matching use numeric cgroup IDs, which
1139 change every time a service is restarted, making them hard to use in
1140 systemd environment.
1142 * A new option CoredumpReceive= can be set for service and scope units,
1143 together with Delegate=yes, to make systemd-coredump on the host
1144 forward core files from processes crashing inside the delegated
1145 CGroup subtree to systemd-coredump running in the container. This new
1146 option is by default used by systemd-nspawn containers that use the
1149 * A new ConditionSecurity=measured-uki option is now available, to ensure
1150 a unit can only run when the system has been booted from a measured UKI.
1152 * MemoryAvailable= now considers physical memory if there are no CGroup
1153 memory limits set anywhere in the tree.
1155 * The $USER environment variable is now always set for services, while
1156 previously it was only set if User= was specified. A new option
1157 SetLoginEnvironment= is now supported to determine whether to also set
1158 $HOME, $LOGNAME, and $SHELL.
1160 * Socket units now support a new pair of
1161 PollLimitBurst=/PollLimitInterval= options to configure a limit on
1162 how often polling events on the file descriptors backing this unit
1163 will be considered within a time window.
1165 * Scope units can now be created using PIDFDs instead of PIDs to select
1166 the processes they should include.
1168 * Sending SIGRTMIN+18 with 0x500 as sigqueue() value will now cause the
1169 manager to dump the list of currently pending jobs.
1171 * If the kernel supports MOVE_MOUNT_BENEATH, the systemctl and
1172 machinectl bind and mount-image verbs will now cause the new mount to
1173 replace the old mount (if any), instead of overmounting it.
1175 * Units now have MemoryPeak, MemorySwapPeak, MemorySwapCurrent and
1176 MemoryZSwapCurrent properties, which respectively contain the values
1177 of the cgroup v2's memory.peak, memory.swap.peak, memory.swap.current
1178 and memory.zswap.current properties. This information is also shown in
1179 "systemctl status" output, if available.
1181 TPM2 Support + Disk Encryption & Authentication:
1183 * systemd-cryptenroll now allows specifying a PCR bank and explicit hash
1184 value in the --tpm2-pcrs= option.
1186 * systemd-cryptenroll now allows specifying a TPM2 key handle (nv
1187 index) to be used instead of the default SRK via the new
1188 --tpm2-seal-key-handle= option.
1190 * systemd-cryptenroll now allows TPM2 enrollment using only a TPM2
1191 public key (in TPM2B_PUBLIC format) – without access to the TPM2
1192 device itself – which enables offline sealing of LUKS images for a
1193 specific TPM2 chip, as long as the SRK public key is known. Pass the
1194 public to the tool via the new --tpm2-device-key= switch.
1196 * systemd-cryptsetup is now installed in /usr/bin/ and is no longer an
1197 internal-only executable.
1199 * The TPM2 Storage Root Key will now be set up, if not already present,
1200 by a new systemd-tpm2-setup.service early boot service. The SRK will
1201 be stored in PEM format and TPM2_PUBLIC format (the latter is useful
1202 for systemd-cryptenroll --tpm2-device-key=, as mentioned above) for
1203 easier access. A new "srk" verb has been added to systemd-analyze to
1204 allow extracting it on demand if it is already set up.
1206 * The internal systemd-pcrphase executable has been renamed to
1209 * The systemd-pcrextend tool gained a new --pcr= switch to override
1210 which PCR to measure into.
1212 * systemd-pcrextend now exposes a Varlink interface at
1213 io.systemd.PCRExtend that can be used to do measurements and event
1216 * TPM measurements are now also written to an event log at
1217 /run/log/systemd/tpm2-measure.log, using a derivative of the TCG
1218 Canonical Event Log format. Previously we'd only log them to the
1219 journal, where they however were subject to rotation and similar.
1221 * A new component "systemd-pcrlock" has been added that allows managing
1222 local TPM2 PCR policies for PCRs 0-7 and similar, which are hard to
1223 predict by the OS vendor because of the inherently local nature of
1224 what measurements they contain, such as firmware versions of the
1225 system and extension cards and suchlike. pcrlock can predict PCR
1226 measurements ahead of time based on various inputs, such as the local
1227 TPM2 event log, GPT partition tables, PE binaries, UKI kernels, and
1228 various other things. It can then pre-calculate a TPM2 policy from
1229 this, which it stores in an TPM2 NV index. TPM2 objects (such as disk
1230 encryption keys) can be locked against this NV index, so that they
1231 are locked against a specific combination of system firmware and
1232 state. Alternatives for each component are supported to allowlist
1233 multiple kernel versions or boot loader version simultaneously
1234 without losing access to the disk encryption keys. The tool can also
1235 be used to analyze and validate the local TPM2 event log.
1236 systemd-cryptsetup, systemd-cryptenroll, systemd-repart have all been
1237 updated to support such policies. There's currently no support for
1238 locking the system's root disk against a pcrlock policy, this will be
1239 added soon. Moreover, it is currently not possible to combine a
1240 pcrlock policy with a signed PCR policy. This component is
1241 experimental and its public interface is subject to change.
1243 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
1245 * bootctl will now show whether the system was booted from a UKI in its
1248 * systemd-boot and systemd-stub now use different project keys in their
1249 respective SBAT sections, so that they can be revoked individually if
1252 * systemd-boot will no longer load unverified Devicetree blobs when UEFI
1253 SecureBoot is enabled. For more details see:
1254 https://github.com/systemd/systemd/security/advisories/GHSA-6m6p-rjcq-334c
1256 * systemd-boot gained new hotkeys to reboot and power off the system
1257 from the boot menu ("B" and "O"). If the "auto-poweroff" and
1258 "auto-reboot" options in loader.conf are set these entries are also
1259 shown as menu items (which is useful on devices lacking a regular
1262 * systemd-boot gained a new configuration value "menu-disabled" for the
1263 set-timeout option, to allow completely disabling the boot menu,
1264 including the hotkey.
1266 * systemd-boot will now measure the content of loader.conf in TPM2
1269 * systemd-stub will now concatenate the content of all kernel
1270 command-line addons before measuring them in TPM2 PCR 12, in a single
1271 measurement, instead of measuring them individually.
1273 * systemd-stub will now measure and load Devicetree Blob addons, which
1274 are searched and loaded following the same model as the existing
1275 kernel command-line addons.
1277 * systemd-stub will now ignore unauthenticated kernel command line options
1278 passed from systemd-boot when running inside Confidential VMs with UEFI
1281 * systemd-stub will now load a Devicetree blob even if the firmware did
1282 not load any beforehand (e.g.: for ACPI systems).
1284 * ukify is no longer considered experimental, and now ships in /usr/bin/.
1286 * ukify gained a new verb inspect to describe the sections of a UKI and
1287 print the contents of the well-known sections.
1289 * ukify gained a new verb genkey to generate a set of key pairs for
1290 signing UKIs and their PCR data.
1292 * The 90-loaderentry kernel-install hook now supports installing device
1295 * kernel-install now supports the --json=, --root=, --image=, and
1296 --image-policy= options for the inspect verb.
1298 * kernel-install now supports new list and add-all verbs. The former
1299 lists all installed kernel images (if those are available in
1300 /usr/lib/modules/). The latter will install all the kernels it can
1305 * A new option --copy-from= has been added that synthesizes partition
1306 definitions from the given image, which are then applied by the
1307 systemd-repart algorithm.
1309 * A new option --copy-source= has been added, which can be used to specify
1310 a directory to which CopyFiles= is considered relative to.
1312 * New --make-ddi=confext, --make-ddi=sysext, and --make-ddi=portable
1313 options have been added to make it easier to generate these types of
1314 DDIs, without having to provide repart.d definitions for them.
1316 * The dm-verity salt and UUID will now be derived from the specified
1319 * New VerityDataBlockSizeBytes= and VerityHashBlockSizeBytes= can now be
1320 configured in repart.d/ configuration files.
1322 * A new Subvolumes= setting is now supported in repart.d/ configuration
1323 files, to indicate which directories in the target partition should be
1326 * A new --tpm2-device-key= option can be used to lock a disk against a
1327 specific TPM2 public key. This matches the same switch the
1328 systemd-cryptenroll tool now supports (see above).
1332 * The journalctl --lines= parameter now accepts +N to show the oldest N
1333 entries instead of the newest.
1335 * journald now ensures that sealing happens once per epoch, and sets a
1336 new compatibility flag to distinguish old journal files that were
1337 created before this change, for backward compatibility.
1341 * udev will now create symlinks to loopback block devices in the
1342 /dev/disk/by-loop-ref/ directory that are based on the .lo_file_name
1343 string field selected during allocation. The systemd-dissect tool and
1344 the util-linux losetup command now supports a complementing new switch
1345 --loop-ref= for selecting the string. This means a loopback block
1346 device may now be allocated under a caller-chosen reference and can
1347 subsequently be referenced without first having to look up the block
1348 device name the caller ended up with.
1350 * udev also creates symlinks to loopback block devices in the
1351 /dev/disk/by-loop-inode/ directory based on the .st_dev/st_ino fields
1352 of the inode attached to the loopback block device. This means that
1353 attaching a file to a loopback device will implicitly make a handle
1354 available to be found via that file's inode information.
1356 * udevadm info gained support for JSON output via a new --json= flag, and
1357 for filtering output using the same mechanism that udevadm trigger
1360 * The predictable network interface naming logic is extended to include
1361 the SR-IOV-R "representor" information in network interface names.
1362 This feature was intended for v254, but even though the code was
1363 merged, the part that actually enabled the feature was forgotten.
1364 It is now enabled by default and is part of the new "v255" naming
1367 * A new hwdb/rules file has been added that sets the
1368 ID_NET_AUTO_LINK_LOCAL_ONLY=1 udev property on all network interfaces
1369 that should usually only be configured with link-local addressing
1370 (IPv4LL + IPv6LL), i.e. for PC-to-PC cables ("laplink") or
1371 Thunderbolt networking. systemd-networkd and NetworkManager (soon)
1372 will make use of this information to apply an appropriate network
1373 configuration by default.
1375 * The ID_NET_DRIVER property on network interfaces is now set
1376 relatively early in the udev rule set so that other rules may rely on
1377 its use. This is implemented in a new "net-driver" udev built-in.
1381 * The "duid-only" option for DHCPv4 client's ClientIdentifier= setting
1382 is now dropped, as it never worked, hence it should not be used by
1385 * The 'prefixstable' ipv6 address generation mode now considers the SSID
1386 when generating stable addresses, so that a different stable address
1387 is used when roaming between wireless networks. If you already use
1388 'prefixstable' addresses with wireless networks, the stable address
1389 will be changed by the update.
1391 * The DHCPv4 client gained a RapidCommit option, true by default, which
1392 enables RFC4039 Rapid Commit behavior to obtain a lease in a
1393 simplified 2-message exchange instead of the typical 4-message
1394 exchange, if also supported by the DHCP server.
1396 * The DHCPv4 client gained new InitialCongestionWindow= and
1397 InitialAdvertisedReceiveWindow= options for route configurations.
1399 * The DHCPv4 client gained a new RequestAddress= option that allows
1400 to send a preferred IP address in the initial DHCPDISCOVER message.
1402 * The DHCPv4 server and client gained support for IPv6-only mode
1405 * The SendHostname= and Hostname= options are now available for the
1406 DHCPv6 client, independently of the DHCPv4= option, so that these
1407 configuration values can be set independently for each client.
1409 * The DHCPv4 and DHCPv6 client state can now be queried via D-Bus,
1410 including lease information.
1412 * The DHCPv6 client can now be configured to use a custom DUID type.
1414 * .network files gained a new IPv4ReversePathFilter= setting in the
1415 [Network] section, to control sysctl's rp_filter setting.
1417 * .network files gaiend a new HopLimit= setting in the [Route] section,
1418 to configure a per-route hop limit.
1420 * .network files gained a new TCPRetransmissionTimeoutSec= setting in
1421 the [Route] section, to configure a per-route TCP retransmission
1424 * A new directive NFTSet= provides a method for integrating network
1425 configuration into firewall rules with NFT sets. The benefit of using
1426 this setting is that static network configuration or dynamically
1427 obtained network addresses can be used in firewall rules with the
1428 indirection of NFT set types.
1430 * The [IPv6AcceptRA] section supports the following new options:
1431 UsePREF64=, UseHopLimit=, UseICMP6RateLimit=, and NFTSet=.
1433 * The [IPv6SendRA] section supports the following new options:
1434 RetransmitSec=, HopLimit=, HomeAgent=, HomeAgentLifetimeSec=, and
1435 HomeAgentPreference=.
1437 * A new [IPv6PREF64Prefix] set of options, containing Prefix= and
1438 LifetimeSec=, has been introduced to append pref64 options in router
1439 advertisements (RFC8781).
1441 * The network generator now configures the interfaces with only
1442 link-local addressing if "ip=link-local" is specified on the kernel
1445 * The prefix of the configuration files generated by the network
1446 generator from the kernel command line is now prefixed with '70-',
1447 to make them have higher precedence over the default configuration
1450 * Added a new -Ddefault-network=BOOL meson option, that causes more
1451 .network files to be installed as enabled by default. These configuration
1452 files will which match generic setups, e.g. 89-ethernet.network matches
1453 all Ethernet interfaces and enables both DHCPv4 and DHCPv6 clients.
1455 * If a ID_NET_MANAGED_BY= udev property is set on a network device and
1456 it is any other string than "io.systemd.Network" then networkd will
1457 not manage this device. This may be used to allow multiple network
1458 management services to run in parallel and assign ownership of
1459 specific devices explicitly. NetworkManager will soon implement a
1464 * systemctl is-failed now checks the system state if no unit is
1467 * systemctl will now automatically soft-reboot if a new root file system
1468 is found under /run/nextroot/ when a reboot operation is invoked.
1472 * Wall messages now work even when utmp support is disabled, using
1473 systemd-logind to query the necessary information.
1475 * systemd-logind now sends a new PrepareForShutdownWithMetadata D-Bus
1476 signal before shutdown/reboot/soft-reboot that includes additional
1477 information compared to the PrepareForShutdown signal. Currently the
1478 additional information is the type of operation that is about to be
1481 Hibernation & Suspend:
1483 * The kernel and OS versions will no longer be checked on resume from
1486 * Hibernation into swap files backed by btrfs are now
1487 supported. (Previously this was supported only for other file
1492 * A new systemd-vmspawn tool has been added, that aims to provide for VMs
1493 the same interfaces and functionality that systemd-nspawn provides for
1494 containers. For now it supports QEMU as a backend, and exposes some of
1495 its options to the user. This component is experimental and its public
1496 interface is subject to change.
1498 * "systemd-analyze plot" has gained tooltips on each unit name with
1499 related-unit information in its svg output, such as Before=,
1500 Requires=, and similar properties.
1502 * A new varlinkctl tool has been added to allow interfacing with
1503 Varlink services, and introspection has been added to all such
1504 services. This component is experimental and its public interface is
1507 * systemd-sysext and systemd-confext now expose a Varlink service
1508 at io.systemd.sysext.
1510 * portable services now accept confexts as extensions.
1512 * systemd-sysupdate now accepts directories in the MatchPattern= option.
1514 * systemd-run will now output the invocation ID of the launched
1515 transient unit and its peak memory usage.
1517 * systemd-analyze, systemd-tmpfiles, systemd-sysusers, systemd-sysctl,
1518 and systemd-binfmt gained a new --tldr option that can be used instead
1519 of --cat-config to suppress uninteresting configuration lines, such as
1520 comments and whitespace.
1522 * resolvectl gained a new "show-server-state" command that shows
1523 current statistics of the resolver. This is backed by a new
1524 DumpStatistics() Varlink method provided by systemd-resolved.
1526 * systemd-timesyncd will now emit a D-Bus signal when the LinkNTPServers
1529 * vconsole now supports KEYMAP=@kernel for preserving the kernel keymap
1532 * seccomp now supports the LoongArch64 architecture.
1534 * seccomp may now be enabled for services running as a non-root User=
1535 without NoNewPrivileges=yes.
1537 * systemd-id128 now supports a new -P option to show only values. The
1538 combination of -P and --app options is also supported.
1540 * A new pam_systemd_loadkey.so PAM module is now available, which will
1541 automatically fetch the passphrase used by cryptsetup to unlock the
1542 root file system and set it as the PAM authtok. This enables, among
1543 other things, configuring auto-unlock of the GNOME Keyring / KDE
1544 Wallet when autologin is configured.
1546 * Many meson options now use the 'feature' type, which means they
1547 take enabled/disabled/auto as values.
1549 * A new meson option -Dconfigfiledir= can be used to change where
1550 configuration files with default values are installed to.
1552 * Options and verbs in man pages are now tagged with the version they
1553 were first introduced in.
1555 * A new component "systemd-storagetm" has been added, which exposes all
1556 local block devices as NVMe-TCP devices, fully automatically. It's
1557 hooked into a new target unit storage-target-mode.target that is
1558 suppsoed to be booted into via
1559 rd.systemd.unit=storage-target-mode.target on the kernel command
1560 line. This is intended to be used for installers and debugging to
1561 quickly get access to the local disk. It's inspired by MacOS "target
1562 disk mode". This component is experimental and its public interface is
1565 * A new component "systemd-bsod" has been added, which can show logged
1566 error messages full screen, if they have a log level of LOG_EMERG log
1567 level. This component is experimental and its public interface is
1570 * The systemd-dissect tool's --with command will now set the
1571 $SYSTEMD_DISSECT_DEVICE environment variable to the block device it
1572 operates on for the invoked process.
1574 * The systemd-mount tool gained a new --tmpfs switch for mounting a new
1575 'tmpfs' instance. This is useful since it does so via .mount units
1576 and thus can be executed remotely or in containers.
1578 * The various tools in systemd that take "verbs" (such as systemctl,
1579 loginctl, machinectl, …) now will suggest a close verb name in case
1580 the user specified an unrecognized one.
1582 * libsystemd now exports a new function sd_id128_get_app_specific()
1583 that generates "app-specific" 128bit IDs from any ID. It's similar to
1584 sd_id128_get_machine_app_specific() and
1585 sd_id128_get_boot_app_specific() but takes the ID to base calculation
1586 on as input. This new functionality is also exposed in the
1587 "systemd-id128" tool where you can now combine --app= with `show`.
1589 * All tools that parse timestamps now can also parse RFC3339 style
1590 timestamps that include the "T" and Z" characters.
1592 * New documentation has been added:
1594 https://systemd.io/FILE_DESCRIPTOR_STORE
1595 https://systemd.io/TPM2_PCR_MEASUREMENTS
1596 https://systemd.io/MOUNT_REQUIREMENTS
1598 * The codebase now recognizes the suffix .confext.raw and .sysext.raw
1599 as alternative to the .raw suffix generally accepted for DDIs. It is
1600 recommended to name configuration extensions and system extensions
1601 with such suffixes, to indicate their purpose in the name.
1603 * The sd-device API gained a new function
1604 sd_device_enumerator_add_match_property_required() which allows
1605 configuring matches on properties that are strictly required. This is
1606 different from the existing sd_device_enumerator_add_match_property()
1607 matches of which one needs to apply.
1609 * The MAC address the veth side of an nspawn container shall get
1610 assigned may now be controlled via the $SYSTEMD_NSPAWN_NETWORK_MAC
1611 environment variable.
1613 * The libiptc dependency is now implemented via dlopen(), so that tools
1614 such as networkd and nspawn no longer have a hard dependency on the
1615 shared library when compiled with support for libiptc.
1617 * New rpm macros have been added: %systemd_user_daemon_reexec does
1618 daemon-reexec for all user managers, and %systemd_postun_with_reload
1619 and %systemd_user_postun_with_reload do a reload for system and user
1622 * coredumpctl now propagates SIGTERM to the debugger process.
1624 Contributions from: 김인수, Abderrahim Kitouni, Adam Goldman,
1625 Adam Williamson, Alexandre Peixoto Ferreira, Alex Hudspith,
1626 Alvin Alvarado, André Paiusco, Antonio Alvarez Feijoo,
1627 Anton Lundin, Arian van Putten, Arseny Maslennikov, Arthur Shau,
1628 Balázs Úr, beh_10257, Benjamin Peterson, Bertrand Jacquin,
1629 Brian Norris, Charles Lee, Cheng-Chia Tseng, Chris Patterson,
1630 Christian Hergert, Christian Hesse, Christian Kirbach,
1631 Clayton Craft, commondservice, cunshunxia, Curtis Klein, cvlc12,
1632 Daan De Meyer, Daniele Medri, Daniel P. Berrangé, Daniel Rusek,
1633 Daniel Thompson, Dan Nicholson, Dan Streetman, David Rheinsberg,
1634 David Santamaría Rogado, David Tardon, dependabot[bot],
1635 Diego Viola, Dmitry V. Levin, Emanuele Giuseppe Esposito,
1636 Emil Renner Berthing, Emil Velikov, Etienne Dechamps, Fabian Vogt,
1637 felixdoerre, Felix Dörre, Florian Schmaus, Franck Bui,
1638 Frantisek Sumsal, G2-Games, Gioele Barabucci, Hugo Carvalho,
1639 huyubiao, Iago López Galeiras, IllusionMan1212, Jade Lovelace,
1640 janana, Jan Janssen, Jan Kuparinen, Jan Macku, Jeremy Fleischman,
1641 Jin Liu, jjimbo137, Joerg Behrmann, Johannes Segitz, Jordan Rome,
1642 Jordan Williams, Julien Malka, Juno Computers, Khem Raj, khm,
1643 Kingbom Dou, Kiran Vemula, Krzesimir Nowak, Laszlo Gombos,
1644 Lennart Poettering, linuxlion, Luca Boccassi, Lucas Adriano Salles,
1645 Lukas, Lukáš Nykrýn, Maanya Goenka, Maarten, Malte Poll,
1646 Marc Pervaz Boocha, Martin Beneš, Martin Joerg, Martin Wilck,
1647 Mathieu Tortuyaux, Matthias Schiffer, Maxim Mikityanskiy,
1648 Max Kellermann, Michael A Cassaniti, Michael Biebl, Michael Kuhn,
1649 Michael Vasseur, Michal Koutný, Michal Sekletár, Mike Yuan,
1650 Milton D. Miller II, mordner, msizanoen, NAHO, Nandakumar Raghavan,
1651 Neil Wilson, Nick Rosbrook, Nils K, NRK, Oğuz Ersen,
1652 Omojola Joshua, onenowy, Paul Meyer, Paymon MARANDI, pelaufer,
1653 Peter Hutterer, PhylLu, Pierre GRASSER, Piotr Drąg, Priit Laes,
1654 Rahil Bhimjiani, Raito Bezarius, Raul Cheleguini, Reto Schneider,
1655 Richard Maw, Robby Red, RoepLuke, Roland Hieber, Roland Singer,
1656 Ronan Pigott, Sam James, Sam Leonard, Sergey A, Susant Sahani,
1657 Sven Joachim, Tad Fisher, Takashi Sakamoto, Thorsten Kukuk, Tj,
1658 Tomasz Świątek, Topi Miettinen, Valentin David,
1659 Valentin Lefebvre, Victor Westerhuis, Vincent Haupert,
1660 Vishal Chillara Srinivas, Vito Caputo, Warren, Weblate,
1661 Xiaotian Wu, xinpeng wang, Yaron Shahrabani, Yo-Jung Lin,
1662 Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zeroskyx,
1663 Дамјан Георгиевски, наб
1665 — Edinburgh, 2023-12-06
1669 Announcements of Future Feature Removals and Incompatible Changes:
1671 * The next release (v255) will remove support for split-usr (/usr/
1672 mounted separately during late boot, instead of being mounted by the
1673 initrd before switching to the rootfs) and unmerged-usr (parallel
1674 directories /bin/ and /usr/bin/, /lib/ and /usr/lib/, …). For more
1676 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
1678 * We intend to remove cgroup v1 support from a systemd release after
1679 the end of 2023. If you run services that make explicit use of
1680 cgroup v1 features (i.e. the "legacy hierarchy" with separate
1681 hierarchies for each controller), please implement compatibility with
1682 cgroup v2 (i.e. the "unified hierarchy") sooner rather than later.
1683 Most of Linux userspace has been ported over already.
1685 * Support for System V service scripts is now deprecated and will be
1686 removed in a future release. Please make sure to update your software
1687 *now* to include a native systemd unit file instead of a legacy
1688 System V script to retain compatibility with future systemd releases.
1690 * Support for the SystemdOptions EFI variable is deprecated.
1691 'bootctl systemd-efi-options' will emit a warning when used. It seems
1692 that this feature is little-used and it is better to use alternative
1693 approaches like credentials and confexts. The plan is to drop support
1694 altogether at a later point, but this might be revisited based on
1697 * EnvironmentFile= now treats the line following a comment line
1698 trailing with escape as a non comment line. For details, see:
1699 https://github.com/systemd/systemd/issues/27975
1701 * PrivateNetwork=yes and NetworkNamespacePath= now imply
1702 PrivateMounts=yes unless PrivateMounts=no is explicitly specified.
1704 * Behaviour of sandboxing options for the per-user service manager
1705 units has changed. They now imply PrivateUsers=yes, which means user
1706 namespaces will be implicitly enabled when a sandboxing option is
1707 enabled in a user unit. Enabling user namespaces has the drawback
1708 that system users will no longer be visible (and processes/files will
1709 appear as owned by 'nobody') in the user unit.
1711 By definition a sandboxed user unit should run with reduced
1712 privileges, so impact should be small. This will remove a great
1713 source of confusion that has been reported by users over the years,
1714 due to how these options require an extra setting to be manually
1715 enabled when used in the per-user service manager, which is not
1716 needed in the system service manager. For more details, see:
1717 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
1719 * systemd-run's switch --expand-environment= which currently is disabled
1720 by default when combined with --scope, will be changed in a future
1721 release to be enabled by default.
1723 Security Relevant Changes:
1725 * pam_systemd will now by default pass the CAP_WAKE_ALARM ambient
1726 process capability to invoked session processes of regular users on
1727 local seats (as well as to systemd --user), unless configured
1728 otherwise via data from JSON user records, or via the PAM module's
1729 parameter list. This is useful in order allow desktop tools such as
1730 GNOME's Alarm Clock application to set a timer for
1731 CLOCK_REALTIME_ALARM that wakes up the system when it elapses. A
1732 per-user service unit file may thus use AmbientCapability= to pass
1733 the capability to invoked processes. Note that this capability is
1734 relatively narrow in focus (in particular compared to other process
1735 capabilities such as CAP_SYS_ADMIN) and we already — by default —
1736 permit more impactful operations such as system suspend to local
1741 * Memory limits that apply while the unit is activating are now
1742 supported. Previously IO and CPU settings were already supported via
1743 StartupCPUWeight= and similar. The same logic has been added for the
1744 various manager and unit memory settings (DefaultStartupMemoryLow=,
1745 StartupMemoryLow=, StartupMemoryHigh=, StartupMemoryMax=,
1746 StartupMemorySwapMax=, StartupMemoryZSwapMax=).
1748 * The service manager gained support for enqueuing POSIX signals to
1749 services that carry an additional integer value, exposing the
1750 sigqueue() system call. This is accessible via new D-Bus calls
1751 org.freedesktop.systemd1.Manager.QueueSignalUnit() and
1752 org.freedesktop.systemd1.Unit.QueueSignal(), as well as in systemctl
1753 via the new --kill-value= option.
1755 * systemctl gained a new "list-paths" verb, which shows all currently
1756 active .path units, similarly to how "systemctl list-timers" shows
1757 active timers, and "systemctl list-sockets" shows active sockets.
1759 * systemctl gained a new --when= switch which is honoured by the various
1760 forms of shutdown (i.e. reboot, kexec, poweroff, halt) and allows
1761 scheduling these operations by time, similar in fashion to how this
1762 has been supported by SysV shutdown.
1764 * If MemoryDenyWriteExecute= is enabled for a service and the kernel
1765 supports the new PR_SET_MDWE prctl() call, it is used instead of the
1766 seccomp()-based system call filter to achieve the same effect.
1768 * A new set of kernel command line options is now understood:
1769 systemd.tty.term.<name>=, systemd.tty.rows.<name>=,
1770 systemd.tty.columns.<name>= allow configuring the TTY type and
1771 dimensions for the tty specified via <name>. When systemd invokes a
1772 service on a tty (via TTYName=) it will look for these and configure
1773 the TTY accordingly. This is particularly useful in VM environments
1774 to propagate host terminal settings into the appropriate TTYs of the
1777 * A new RootEphemeral= setting is now understood in service units. It
1778 takes a boolean argument. If enabled for services that use RootImage=
1779 or RootDirectory= an ephemeral copy of the disk image or directory
1780 tree is made when the service is started. It is removed automatically
1781 when the service is stopped. That ephemeral copy is made using
1782 btrfs/xfs reflinks or btrfs snapshots, if available.
1784 * The service activation logic gained new settings RestartSteps= and
1785 RestartMaxDelaySec= which allow exponentially-growing restart
1786 intervals for Restart=.
1788 * The service activation logic gained a new setting RestartMode= which
1789 can be set to 'direct' to skip the inactive/failed states when
1790 restarting, so that dependent units are not notified until the service
1791 converges to a final (successful or failed) state. For example, this
1792 means that OnSuccess=/OnFailure= units will not be triggered until the
1793 service state has converged.
1795 * PID 1 will now automatically load the virtio_console kernel module
1796 during early initialization if running in a suitable VM. This is done
1797 so that early-boot logging can be written to the console if available.
1799 * Similarly, virtio-vsock support is loaded early in suitable VM
1800 environments. PID 1 will send sd_notify() notifications via AF_VSOCK
1801 to the VMM if configured, thus loading this early is beneficial.
1803 * A new verb "fdstore" has been added to systemd-analyze to show the
1804 current contents of the file descriptor store of a unit. This is
1805 backed by a new D-Bus call DumpUnitFileDescriptorStore() provided by
1806 the service manager.
1808 * The service manager will now set a new $FDSTORE environment variable
1809 when invoking processes for services that have the file descriptor
1812 * A new service option FileDescriptorStorePreserve= has been added that
1813 allows tuning the lifecycle of the per-service file descriptor store.
1814 If set to "yes", the entries in the fd store are retained even after
1815 the service has been fully stopped.
1817 * The "systemctl clean" command may now be used to clear the fdstore of
1820 * Unit *.preset files gained a new directive "ignore", in addition to
1821 the existing "enable" and "disable". As the name suggests, matching
1822 units are left unchanged, i.e. neither enabled nor disabled.
1824 * Service units gained a new setting DelegateSubgroup=. It takes the
1825 name of a sub-cgroup to place any processes the service manager forks
1826 off in. Previously, the service manager would place all service
1827 processes directly in the top-level cgroup it created for the
1828 service. This usually meant that main process in a service with
1829 delegation enabled would first have to create a subgroup and move
1830 itself down into it, in order to not conflict with the "no processes
1831 in inner cgroups" rule of cgroup v2. With this option, this step is
1832 now handled by PID 1.
1834 * The service manager will now look for .upholds/ directories,
1835 similarly to the existing support for .wants/ and .requires/
1836 directories. Symlinks in this directory result in Upholds=
1839 The [Install] section of unit files gained support for a new
1840 UpheldBy= directive to generate .upholds/ symlinks automatically when
1843 * The service manager now supports a new kernel command line option
1844 systemd.default_device_timeout_sec=, which may be used to override
1845 the default timeout for .device units.
1847 * A new "soft-reboot" mechanism has been added to the service manager.
1848 A "soft reboot" is similar to a regular reboot, except that it
1849 affects userspace only: the service manager shuts down any running
1850 services and other units, then optionally switches into a new root
1851 file system (mounted to /run/nextroot/), and then passes control to a
1852 systemd instance in the new file system which then starts the system
1853 up again. The kernel is not rebooted and neither is the hardware,
1854 firmware or boot loader. This provides a fast, lightweight mechanism
1855 to quickly reset or update userspace, without the latency that a full
1856 system reset involves. Moreover, open file descriptors may be passed
1857 across the soft reboot into the new system where they will be passed
1858 back to the originating services. This allows pinning resources
1859 across the reboot, thus minimizing grey-out time further. This new
1860 reboot mechanism is accessible via the new "systemctl soft-reboot"
1863 * Services using RootDirectory= or RootImage= will now have read-only
1864 access to a copy of the host's os-release file under
1865 /run/host/os-release, which will be kept up-to-date on 'soft-reboot'.
1866 This was already the case for Portable Services, and the feature has
1867 now been extended to all services that do not run off the host's
1870 * A new service setting MemoryKSM= has been added to enable kernel
1871 same-page merging individually for services.
1873 * A new service setting ImportCredentials= has been added that augments
1874 LoadCredential= and LoadCredentialEncrypted= and searches for
1875 credentials to import from the system, and supports globbing.
1877 * A new job mode "restart-dependencies" has been added to the service
1878 manager (exposed via systemctl --job-mode=). It is only valid when
1879 used with "start" jobs, and has the effect that the "start" job will
1880 be propagated as "restart" jobs to currently running units that have
1881 a BindsTo= or Requires= dependency on the started unit.
1883 * A new verb "whoami" has been added to "systemctl" which determines as
1884 part of which unit the command is being invoked. It writes the unit
1885 name to standard output. If one or more PIDs are specified reports
1886 the unit names the processes referenced by the PIDs belong to.
1888 * The system and service credential logic has been improved: there's
1889 now a clearly defined place where system provisioning tools running
1890 in the initrd can place credentials that will be imported into the
1891 system's set of credentials during the initrd → host transition: the
1892 /run/credentials/@initrd/ directory. Once the credentials placed
1893 there are imported into the system credential set they are deleted
1894 from this directory, and the directory itself is deleted afterwards
1897 * A new kernel command line option systemd.set_credential_binary= has
1898 been added, that is similar to the pre-existing
1899 systemd.set_credential= but accepts arbitrary binary credential data,
1900 encoded in Base64. Note that the kernel command line is not a
1901 recommend way to transfer credentials into a system, since it is
1902 world-readable from userspace.
1904 * The default machine ID to use may now be configured via the
1905 system.machine_id system credential. It will only be used if no
1906 machine ID was set yet on the host.
1908 * On Linux kernel 6.4 and newer system and service credentials will now
1909 be placed in a tmpfs instance that has the "noswap" mount option
1910 set. Previously, a "ramfs" instance was used. By switching to tmpfs
1911 ACL support and overall size limits can now be enforced, without
1912 compromising on security, as the memory is never paged out either
1915 * The service manager now can detect when it is running in a
1916 'Confidential Virtual Machine', and a corresponding 'cvm' value is now
1917 accepted by ConditionSecurity= for units that want to conditionalize
1918 themselves on this. systemd-detect-virt gained new 'cvm' and
1919 '--list-cvm' switches to respectively perform the detection or list
1920 all known flavours of confidential VM, depending on the vendor. The
1921 manager will publish a 'ConfidentialVirtualization' D-Bus property,
1922 and will also set a SYSTEMD_CONFIDENTIAL_VIRTUALIZATION= environment
1923 variable for unit generators. Finally, udev rules can match on a new
1924 'cvm' key that will be set when in a confidential VM.
1925 Additionally, when running in a 'Confidential Virtual Machine', SMBIOS
1926 strings and QEMU's fw_cfg protocol will not be used to import
1927 credentials and kernel command line parameters by the system manager,
1928 systemd-boot and systemd-stub, because the hypervisor is considered
1929 untrusted in this particular setting.
1933 * The sd-journal API gained a new call sd_journal_get_seqnum() to
1934 retrieve the current log record's sequence number and sequence number
1935 ID, which allows applications to order records the same way as
1936 journal does internally. The sequence number is now also exported in
1937 the JSON and "export" output of the journal.
1939 * journalctl gained a new switch --truncate-newline. If specified
1940 multi-line log records will be truncated at the first newline,
1941 i.e. only the first line of each log message will be shown.
1943 * systemd-journal-upload gained support for --namespace=, similar to
1944 the switch of the same name of journalctl.
1948 * systemd-repart's drop-in files gained a new ExcludeFiles= option which
1949 may be used to exclude certain files from the effect of CopyFiles=.
1951 * systemd-repart's Verity support now implements the Minimize= setting
1952 to minimize the size of the resulting partition.
1954 * systemd-repart gained a new --offline= switch, which may be used to
1955 control whether images shall be built "online" or "offline",
1956 i.e. whether to make use of kernel facilities such as loopback block
1957 devices and device mapper or not.
1959 * If systemd-repart is told to populate a newly created ESP or XBOOTLDR
1960 partition with some files, it will now default to VFAT rather than
1963 * systemd-repart gained a new --architecture= switch. If specified, the
1964 per-architecture GPT partition types (i.e. the root and /usr/
1965 partitions) configured in the partition drop-in files are
1966 automatically adjusted to match the specified CPU architecture, in
1967 order to simplify cross-architecture DDI building.
1969 * systemd-repart will now default to a minimum size of 300MB for XFS
1970 filesystems if no size parameter is specified. This matches what the
1971 XFS tools (xfsprogs) can support.
1973 systemd-boot, systemd-stub, ukify, bootctl, kernel-install:
1975 * gnu-efi is no longer required to build systemd-boot and systemd-stub.
1976 Instead, pyelftools is now needed, and it will be used to perform the
1977 ELF -> PE relocations at build time.
1979 * bootctl gained a new switch --print-root-device/-R that prints the
1980 block device the root file system is backed by. If specified twice,
1981 it returns the whole disk block device (as opposed to partition block
1982 device) the root file system is on. It's useful for invocations such
1983 as "cfdisk $(bootctl -RR)" to quickly show the partition table of the
1986 * systemd-stub will now look for the SMBIOS Type 1 field
1987 "io.systemd.stub.kernel-cmdline-extra" and append its value to the
1988 kernel command line it invokes. This is useful for VMMs such as qemu
1989 to pass additional kernel command lines into the system even when
1990 booting via full UEFI. The contents of the field are measured into
1993 * The KERNEL_INSTALL_LAYOUT= setting for kernel-install gained a new
1994 value "auto". With this value, a kernel will be automatically
1995 analyzed, and if it qualifies as UKI, it will be installed as if the
1996 setting was to set to "uki", otherwise as "bls".
1998 * systemd-stub can now optionally load UEFI PE "add-on" images that may
1999 contain additional kernel command line information. These "add-ons"
2000 superficially look like a regular UEFI executable, and are expected
2001 to be signed via SecureBoot/shim. However, they do not actually
2002 contain code, but instead a subset of the PE sections that UKIs
2003 support. They are supposed to provide a way to extend UKIs with
2004 additional resources in a secure and authenticated way. Currently,
2005 only the .cmdline PE section may be used in add-ons, in which case
2006 any specified string is appended to the command line embedded into
2007 the UKI itself. A new 'addon<EFI-ARCH>.efi.stub' is now provided that
2008 can be used to trivially create addons, via 'ukify' or 'objcopy'. In
2009 the future we expect other sections to be made extensible like this as
2012 * ukify has been updated to allow building these UEFI PE "add-on"
2013 images, using the new 'addon<EFI-ARCH>.efi.stub'.
2015 * ukify now accepts SBAT information to place in the .sbat PE section
2016 of UKIs and addons. If a UKI is built the SBAT information from the
2017 inner kernel is merged with any SBAT information associated with
2018 systemd-stub and the SBAT data specified on the ukify command line.
2020 * The kernel-install script has been rewritten in C, and reuses much of
2021 the infrastructure of existing tools such as bootctl. It also gained
2022 --esp-path= and --boot-path= options to override the path to the ESP,
2023 and the $BOOT partition. Options --make-entry-directory= and
2024 --entry-token= have been added as well, similar to bootctl's options
2027 * A new kernel-install plugin 60-ukify has been added which will
2028 combine kernel/initrd locally into a UKI and optionally sign them
2029 with a local key. This may be used to switch to UKI mode even on
2030 systems where a local kernel or initrd is used. (Typically UKIs are
2031 built and signed by the vendor.)
2033 * The ukify tool now supports "pesign" in addition to the pre-existing
2034 "sbsign" for signing UKIs.
2036 * systemd-measure and systemd-stub now look for the .uname PE section
2037 that should contain the kernel's "uname -r" string.
2039 * systemd-measure and ukify now calculate expected PCR hashes for a UKI
2040 "offline", i.e. without access to a TPM (physical or
2043 Memory Pressure & Control:
2045 * The sd-event API gained new calls sd_event_add_memory_pressure(),
2046 sd_event_source_set_memory_pressure_type(),
2047 sd_event_source_set_memory_pressure_period() to create and configure
2048 an event source that is called whenever the OS signals memory
2049 pressure. Another call sd_event_trim_memory() is provided that
2050 compacts the process' memory use by releasing allocated but unused
2051 malloc() memory back to the kernel. Services can also provide their
2052 own custom callback to do memory trimming. This should improve system
2053 behaviour under memory pressure, as on Linux traditionally provided
2054 no mechanism to return process memory back to the kernel if the
2055 kernel was under memory pressure. This makes use of the kernel's PSI
2056 interface. Most long-running services in systemd have been hooked up
2057 with this, and in particular systems with low memory should benefit
2060 * Service units gained new settings MemoryPressureWatch= and
2061 MemoryPressureThresholdSec= to configure the PSI memory pressure
2062 logic individually. If these options are used, the
2063 $MEMORY_PRESSURE_WATCH and $MEMORY_PRESSURE_WRITE environment
2064 variables will be set for the invoked processes to inform them about
2065 the requested memory pressure behaviour. (This is used by the
2066 aforementioned sd-events API additions, if set.)
2068 * systemd-analyze gained a new "malloc" verb that shows the output
2069 generated by glibc's malloc_info() on services that support it. Right
2070 now, only the service manager has been updated accordingly. This
2071 call requires privileges.
2073 User & Session Management:
2075 * The sd-login API gained a new call sd_session_get_username() to
2076 return the user name of the owner of a login session. It also gained
2077 a new call sd_session_get_start_time() to retrieve the time the login
2078 session started. A new call sd_session_get_leader() has been added to
2079 return the PID of the "leader" process of a session. A new call
2080 sd_uid_get_login_time() returns the time since the specified user has
2081 most recently been continuously logged in with at least one session.
2083 * JSON user records gained a new set of fields capabilityAmbientSet and
2084 capabilityBoundingSet which contain a list of POSIX capabilities to
2085 set for the logged in users in the ambient and bounding sets,
2086 respectively. homectl gained the ability to configure these two sets
2087 for users via --capability-bounding-set=/--capability-ambient-set=.
2089 * pam_systemd learnt two new module options
2090 default-capability-bounding-set= and default-capability-ambient-set=,
2091 which configure the default bounding sets for users as they are
2092 logging in, if the JSON user record doesn't specify this explicitly
2093 (see above). The built-in default for the ambient set now contains
2094 the CAP_WAKE_ALARM, thus allowing regular users who may log in
2095 locally to resume from a system suspend via a timer.
2097 * The Session D-Bus objects systemd-logind gained a new SetTTY() method
2098 call to update the TTY of a session after it has been allocated. This
2099 is useful for SSH sessions which are typically allocated first, and
2100 for which a TTY is added later.
2102 * The sd-login API gained a new call sd_pid_notifyf_with_fds() which
2103 combines the various other sd_pid_notify() flavours into one: takes a
2104 format string, an overriding PID, and a set of file descriptors to
2105 send. It also gained a new call sd_pid_notify_barrier() call which is
2106 equivalent to sd_notify_barrier() but allows the originating PID to
2109 * "loginctl list-users" and "loginctl list-sessions" will now show the
2110 state of each logged in user/session in their tabular output. It will
2111 also show the current idle state of sessions.
2115 * systemd-dissect will now show the intended CPU architecture of an
2118 * systemd-dissect will now install itself as mount helper for the "ddi"
2119 pseudo-file system type. This means you may now mount DDIs directly
2120 via /bin/mount or /etc/fstab, making full use of embedded Verity
2121 information and all other DDI features.
2123 Example: mount -t ddi myimage.raw /some/where
2125 * The systemd-dissect tool gained the new switches --attach/--detach to
2126 attach/detach a DDI to a loopback block device without mounting it.
2127 It will automatically derive the right sector size from the image
2128 and set up Verity and similar, but not mount the file systems in it.
2130 * When systemd-gpt-auto-generator or the DDI mounting logic mount an
2131 ESP or XBOOTLDR partition the MS_NOSYMFOLLOW mount option is now
2132 implied. Given that these file systems are typically untrusted, this
2133 should make mounting them automatically have less of a security
2136 * All tools that parse DDIs (such as systemd-nspawn, systemd-dissect,
2137 systemd-tmpfiles, …) now understand a new switch --image-policy= which
2138 takes a string encoding image dissection policy. With this mechanism
2139 automatic discovery and use of specific partition types and the
2140 cryptographic requirements on the partitions (Verity, LUKS, …) can be
2141 restricted, permitting better control of the exposed attack surfaces
2142 when mounting disk images. systemd-gpt-auto-generator will honour such
2143 an image policy too, configurable via the systemd.image_policy= kernel
2144 command line option. Unit files gained the RootImagePolicy=,
2145 MountImagePolicy= and ExtensionImagePolicy= to configure the same for
2146 disk images a service runs off.
2148 * systemd-analyze gained a new verb "image-policy" to validate and
2149 parse image policy strings.
2151 * systemd-dissect gained support for a new --validate switch to
2152 superficially validate DDI structure, and check whether a specific
2153 image policy allows the DDI.
2155 * systemd-dissect gained support for a new --mtree-hash switch to
2156 optionally disable calculating mtree hashes, which can be slow on
2159 * systemd-dissect --copy-to, --copy-from, --list and --mtree switches
2160 are now able to operate on directories too, other than images.
2164 * networkd's GENEVE support as gained a new .network option
2165 InheritInnerProtocol=.
2167 * The [Tunnel] section in .netdev files has gained a new setting
2168 IgnoreDontFragment for controlling the IPv4 "DF" flag of datagrams.
2170 * A new global IPv6PrivacyExtensions= setting has been added that
2171 selects the default value of the per-network setting of the same
2174 * The predictable network interface naming logic was extended to
2175 include SR-IOV-R "representor" information in network interface
2176 names. Unfortunately, this feature was not enabled by default and can
2177 only be enabled at compilation time by setting
2178 -Ddefault-net-naming-scheme=v254.
2180 * The DHCPv4 + DHCPv6 + IPv6 RA logic in networkd gained support for
2181 the RFC8910 captive portal option.
2185 * udevadm gained the new "verify" verb for validating udev rules files
2188 * udev gained a new tool "iocost" that can be used to configure QoS IO
2189 cost data based on hwdb information onto suitable block devices. Also
2190 see https://github.com/iocost-benchmark/iocost-benchmarks.
2192 TPM2 Support + Disk Encryption & Authentication:
2194 * systemd-cryptenroll/systemd-cryptsetup will now install a TPM2 SRK
2195 ("Storage Root Key") as first step in the TPM2, and then use that
2196 for binding FDE to, if TPM2 support is used. This matches
2197 recommendations of TCG (see
2198 https://trustedcomputinggroup.org/wp-content/uploads/TCG-TPM-v2.0-Provisioning-Guidance-Published-v1r1.pdf)
2200 * systemd-cryptenroll and other tools that take TPM2 PCR parameters now
2201 understand textual identifiers for these PCRs.
2203 * systemd-veritysetup + /etc/veritytab gained support for a series of
2204 new options: hash-offset=, superblock=, format=, data-block-size=,
2205 hash-block-size=, data-blocks=, salt=, uuid=, hash=, fec-device=,
2206 fec-offset=, fec-roots= to configure various aspects of a Verity
2209 * systemd-cryptsetup + /etc/crypttab gained support for a new
2210 veracrypt-pim= option for setting the Personal Iteration Multiplier
2211 of veracrypt volumes.
2213 * systemd-integritysetup + /etc/integritytab gained support for a new
2214 mode= setting for controlling the dm-integrity mode (journal, bitmap,
2215 direct) for the volume.
2217 * systemd-analyze gained a new verb "pcrs" that shows the known TPM PCR
2218 registers, their symbolic names and current values.
2222 * The ACL support in tmpfiles.d/ has been updated: if an uppercase "X"
2223 access right is specified this is equivalent to "x" but only if the
2224 inode in question already has the executable bit set for at least
2225 some user/group. Otherwise the "x" bit will be turned off.
2227 * tmpfiles.d/'s C line type now understands a new modifier "+": a line
2228 with C+ will result in a "merge" copy, i.e. all files of the source
2229 tree are copied into the target tree, even if that tree already
2230 exists, resulting in a combined tree of files already present in the
2231 target tree and those copied in.
2233 * systemd-tmpfiles gained a new --graceful switch. If specified lines
2234 with unknown users/groups will silently be skipped.
2238 * systemd-notify gained two new options --fd= and --fdname= for sending
2239 arbitrary file descriptors to the service manager (while specifying an
2240 explicit name for it).
2242 * systemd-notify gained a new --exec switch, which makes it execute the
2243 specified command line after sending the requested messages. This is
2244 useful for sending out READY=1 first, and then continuing invocation
2245 without changing process ID, so that the tool can be nicely used
2246 within an ExecStart= line of a unit file that uses Type=notify.
2248 sd-event + sd-bus APIs:
2250 * The sd-event API gained a new call sd_event_source_leave_ratelimit()
2251 which may be used to explicitly end a rate-limit state an event
2252 source might be in, resetting all rate limiting counters.
2254 * When the sd-bus library is used to make connections to AF_UNIX D-Bus
2255 sockets, it will now encode the "description" set via
2256 sd_bus_set_description() into the source socket address. It will also
2257 look for this information when accepting a connection. This is useful
2258 to track individual D-Bus connections on a D-Bus broker for debug
2263 * systemd-resolved gained a new resolved.conf setting
2264 StateRetentionSec= which may be used to retain cached DNS records
2265 even after their nominal TTL, and use them in case upstream DNS
2266 servers cannot be reached. This can be used to make name resolution
2267 more resilient in case of network problems.
2269 * resolvectl gained a new verb "show-cache" to show the current cache
2270 contents of systemd-resolved. This verb communicates with the
2271 systemd-resolved daemon and requires privileges.
2275 * Meson >= 0.60.0 is now required to build systemd.
2277 * The default keymap to apply may now be chosen at build-time via the
2278 new -Ddefault-keymap= meson option.
2280 * Most of systemd's long-running services now have a generic handler of
2281 the SIGRTMIN+18 signal handler which executes various operations
2282 depending on the sigqueue() parameter sent along. For example, values
2283 0x100…0x107 allow changing the maximum log level of such
2284 services. 0x200…0x203 allow changing the log target of such
2285 services. 0x300 make the services trim their memory similarly to the
2286 automatic PSI-triggered action, see above. 0x301 make the services
2287 output their malloc_info() data to the logs.
2289 * machinectl gained new "edit" and "cat" verbs for editing .nspawn
2290 files, inspired by systemctl's verbs of the same name which edit unit
2291 files. Similarly, networkctl gained the same verbs for editing
2292 .network, .netdev, .link files.
2294 * A new syscall filter group "@sandbox" has been added that contains
2295 syscalls for sandboxing system calls such as those for seccomp and
2298 * New documentation has been added:
2300 https://systemd.io/COREDUMP
2301 https://systemd.io/MEMORY_PRESSURE
2304 * systemd-firstboot gained a new --reset option. If specified, the
2305 settings in /etc/ it knows how to initialize are reset.
2307 * systemd-sysext is now a multi-call binary and is also installed under
2308 the systemd-confext alias name (via a symlink). When invoked that way
2309 it will operate on /etc/ instead of /usr/ + /opt/. It thus becomes a
2310 powerful, atomic, secure configuration management of sorts, that
2311 locally can merge configuration from multiple confext configuration
2312 images into a single immutable tree.
2314 * The --network-macvlan=, --network-ipvlan=, --network-interface=
2315 switches of systemd-nspawn may now optionally take the intended
2316 network interface inside the container.
2318 * All our programs will now send an sd_notify() message with their exit
2319 status in the EXIT_STATUS= field when exiting, using the usual
2320 protocol, including PID 1. This is useful for VMMs and container
2321 managers to collect an exit status from a system as it shuts down, as
2322 set via "systemctl exit …". This is particularly useful in test cases
2323 and similar, as invocations via a VM can now nicely propagate an exit
2324 status to the host, similar to local processes.
2326 * systemd-run gained a new switch --expand-environment=no to disable
2327 server-side environment variable expansion in specified command
2328 lines. Expansion defaults to enabled for all execution types except
2329 --scope, where it defaults to off (and prints a warning) for backward
2330 compatibility reasons. --scope will be flipped to enabled by default
2331 too in a future release. If you are using --scope and passing a '$'
2332 character in the payload you should start explicitly using
2333 --expand-environment=yes/no according to the use case.
2335 * The systemd-system-update-generator has been updated to also look for
2336 the special flag file /etc/system-update in addition to the existing
2337 support for /system-update to decide whether to enter system update
2340 * The /dev/hugepages/ file system is now mounted with nosuid + nodev
2341 mount options by default.
2343 * systemd-fstab-generator now understands two new kernel command line
2344 options systemd.mount-extra= and systemd.swap-extra=, which configure
2345 additional mounts or swaps in a format similar to /etc/fstab. 'fsck'
2346 will be ran on these block devices, like it already happens for
2347 'root='. It also now supports the new fstab.extra and
2348 fstab.extra.initrd credentials that may contain additional /etc/fstab
2349 lines to apply at boot.
2351 * systemd-getty-generator now understands two new credentials
2352 getty.ttys.container and getty.ttys.serial. These credentials may
2353 contain a list of TTY devices – one per line – to instantiate
2354 container-getty@.service and serial-getty@.service on.
2356 * The getty/serial-getty/container-getty units now import the 'agetty.*'
2357 and 'login.*' credentials, which are consumed by the 'login' and
2358 'agetty' programs starting from util-linux v2.40.
2360 * systemd-sysupdate's sysupdate.d/ drop-ins gained a new setting
2361 PathRelativeTo=, which can be set to "esp", "xbootldr", "boot", in
2362 which case the Path= setting is taken relative to the ESP or XBOOTLDR
2363 partitions, rather than the system's root directory /. The relevant
2364 directories are automatically discovered.
2366 * The systemd-ac-power tool gained a new switch --low, which reports
2367 whether the battery charge is considered "low", similar to how the
2368 s2h suspend logic checks this state to decide whether to enter system
2369 suspend or hibernation.
2371 * The /etc/os-release file can now have two new optional fields
2372 VENDOR_NAME= and VENDOR_URL= to carry information about the vendor of
2375 * When the system hibernates, information about the device and offset
2376 used is now written to a non-volatile EFI variable. On next boot the
2377 system will attempt to resume from the location indicated in this EFI
2378 variable. This should make hibernation a lot more robust, while
2379 requiring no manual configuration of the resume location.
2381 * The $XDG_STATE_HOME environment variable (added in more recent
2382 versions of the XDG basedir specification) is now honoured to
2383 implement the StateDirectory= setting in user services.
2385 * A new component "systemd-battery-check" has been added. It may run
2386 during early boot (usually in the initrd), and checks the battery
2387 charge level of the system. In case the charge level is very low the
2388 user is notified (graphically via Plymouth – if available – as well
2389 as in text form on the console), and the system is turned off after a
2390 10s delay. The feature can be disabled by passing
2391 systemd.battery_check=0 through the kernel command line.
2393 * The 'passwdqc' library is now supported as an alternative to the
2394 'pwquality' library and can be selected at build time.
2396 Contributions from: 김인수, 07416, Addison Snelling, Adrian Vovk,
2397 Aidan Dang, Alexander Krabler, Alfred Klomp, Anatoli Babenia,
2398 Andrei Stepanov, Andrew Baxter, Antonio Alvarez Feijoo,
2399 Arian van Putten, Arthur Shau, A S Alam,
2400 Asier Sarasua Garmendia, Balló György, Bastien Nocera,
2401 Benjamin Herrenschmidt, Benjamin Raison, Bill Peterson,
2402 Brad Fitzpatrick, Brett Holman, bri, Chen Qi, Chitoku,
2403 Christian Hesse, Christoph Anton Mitterer, Christopher Gurnee,
2404 Colin Walters, Cornelius Hoffmann, Cristian Rodríguez, cunshunxia,
2405 cvlc12, Cyril Roelandt, Daan De Meyer, Daniele Medri,
2406 Daniel P. Berrangé, Daniel Rusek, Dan Streetman, David Edmundson,
2407 David Schroeder, David Tardon, dependabot[bot],
2408 Dimitri John Ledkov, Dmitrii Fomchenkov, Dmitry V. Levin, dmkUK,
2409 Dominique Martinet, don bright, drosdeck, Edson Juliano Drosdeck,
2410 Egor Ignatov, EinBaum, Emanuele Giuseppe Esposito, Eric Curtin,
2411 Erik Sjölund, Evgeny Vereshchagin, Florian Klink, Franck Bui,
2412 François Rigault, Fran Diéguez, Franklin Yu, Frantisek Sumsal,
2413 Fuminobu TAKEYAMA, Gaël PORTAY, Gerd Hoffmann, Gertalitec,
2414 Gibeom Gwon, Gustavo Noronha Silva, Hannu Lounento,
2415 Hans de Goede, Haochen Tong, HATAYAMA Daisuke, Henrik Holst,
2416 Hoe Hao Cheng, Igor Tsiglyar, Ivan Vecera, James Hilliard,
2417 Jan Engelhardt, Jan Janssen, Jan Luebbe, Jan Macku, Janne Sirén,
2418 jcg, Jeidnx, Joan Bruguera, Joerg Behrmann, jonathanmetzman,
2419 Jordan Rome, Josef Miegl, Joshua Goins, Joyce, Joyce Brum,
2420 Juno Computers, Kai Lueke, Kevin P. Fleming, Kiran Vemula, Klaus,
2421 Klaus Zipfel, Lawrence Thorpe, Lennart Poettering, licunlong,
2422 Lily Foster, Luca Boccassi, Ludwig Nussel, Luna Jernberg,
2423 maanyagoenka, Maanya Goenka, Maksim Kliazovich, Malte Poll,
2424 Marko Korhonen, Masatake YAMATO, Mateusz Poliwczak, Matt Johnston,
2425 Miao Wang, Micah Abbott, Michael A Cassaniti, Michal Koutný,
2426 Michal Sekletár, Mike Yuan, mooo, Morten Linderud, msizanoen,
2427 Nick Rosbrook, nikstur, Olivier Gayot, Omojola Joshua,
2428 Paolo Velati, Paul Barker, Pavel Borecki, Petr Menšík,
2429 Philipp Kern, Philip Withnall, Piotr Drąg, Quintin Hill,
2430 Rene Hollander, Richard Phibel, Robert Meijers, Robert Scheck,
2431 Roger Gammans, Romain Geissler, Ronan Pigott, Russell Harmon,
2432 saikat0511, Samanta Navarro, Sam James, Sam Morris,
2433 Simon Braunschmidt, Sjoerd Simons, Sorah Fukumori,
2434 Stanislaw Gruszka, Stefan Roesch, Steven Luo, Steve Ramage,
2435 Susant Sahani, taniishkaaa, Tanishka, Temuri Doghonadze,
2436 Thierry Martin, Thomas Blume, Thomas Genty, Thomas Weißschuh,
2437 Thorsten Kukuk, Times-Z, Tobias Powalowski, tofylion,
2438 Topi Miettinen, Uwe Kleine-König, Velislav Ivanov,
2439 Vitaly Kuznetsov, Vít Zikmund, Weblate, Will Fancher,
2440 William Roberts, Winterhuman, Wolfgang Müller, Xeonacid,
2441 Xiaotian Wu, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Yuxiang Zhu,
2442 Zbigniew Jędrzejewski-Szmek, zhmylove, ZjYwMj,
2443 Дамјан Георгиевски, наб
2445 — Edinburgh, 2023-07-28
2449 Announcements of Future Feature Removals and Incompatible Changes:
2451 * We intend to remove cgroup v1 support from systemd release after the
2452 end of 2023. If you run services that make explicit use of cgroup v1
2453 features (i.e. the "legacy hierarchy" with separate hierarchies for
2454 each controller), please implement compatibility with cgroup v2 (i.e.
2455 the "unified hierarchy") sooner rather than later. Most of Linux
2456 userspace has been ported over already.
2458 * We intend to remove support for split-usr (/usr mounted separately
2459 during boot) and unmerged-usr (parallel directories /bin and
2460 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
2461 half of 2023, in the first release that falls into that time window.
2462 For more details, see:
2463 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
2465 * We intend to change behaviour w.r.t. units of the per-user service
2466 manager and sandboxing options, so that they work without having to
2467 manually enable PrivateUsers= as well, which is not required for
2468 system units. To make this work, we will implicitly enable user
2469 namespaces (PrivateUsers=yes) when a sandboxing option is enabled in a
2470 user unit. The drawback is that system users will no longer be visible
2471 (and appear as 'nobody') to the user unit when a sandboxing option is
2472 enabled. By definition a sandboxed user unit should run with reduced
2473 privileges, so impact should be small. This will remove a great source
2474 of confusion that has been reported by users over the years, due to
2475 how these options require an extra setting to be manually enabled when
2476 used in the per-user service manager, as opposed as to the system
2477 service manager. We plan to enable this change in the next release
2478 later this year. For more details, see:
2479 https://lists.freedesktop.org/archives/systemd-devel/2022-December/048682.html
2481 Deprecations and incompatible changes:
2483 * systemctl will now warn when invoked without /proc/ mounted
2484 (e.g. when invoked after chroot() into an directory tree without the
2485 API mount points like /proc/ being set up.) Operation in such an
2486 environment is not fully supported.
2488 * The return value of 'systemctl is-active|is-enabled|is-failed' for
2489 unknown units is changed: previously 1 or 3 were returned, but now 4
2490 (EXIT_PROGRAM_OR_SERVICES_STATUS_UNKNOWN) is used as documented.
2492 * 'udevadm hwdb' subcommand is deprecated and will emit a warning.
2493 systemd-hwdb (added in 2014) should be used instead.
2495 * 'bootctl --json' now outputs a single JSON array, instead of a stream
2496 of newline-separated JSON objects.
2498 * Udev rules in 60-evdev.rules have been changed to load hwdb
2499 properties for all modalias patterns. Previously only the first
2500 matching pattern was used. This could change what properties are
2501 assigned if the user has more and less specific patterns that could
2502 match the same device, but it is expected that the change will have
2503 no effect for most users.
2505 * systemd-networkd-wait-online exits successfully when all interfaces
2506 are ready or unmanaged. Previously, if neither '--any' nor
2507 '--interface=' options were used, at least one interface had to be in
2508 configured state. This change allows the case where systemd-networkd
2509 is enabled, but no interfaces are configured, to be handled
2510 gracefully. It may occur in particular when a different network
2511 manager is also enabled and used.
2513 * Some compatibility helpers were dropped: EmergencyAction= in the user
2514 manager, as well as measuring kernel command line into PCR 8 in
2515 systemd-stub, along with the -Defi-tpm-pcr-compat compile-time
2518 * The '-Dupdate-helper-user-timeout=' build-time option has been
2519 renamed to '-Dupdate-helper-user-timeout-sec=', and now takes an
2520 integer as parameter instead of a string.
2522 * The DDI image dissection logic (which backs RootImage= in service
2523 unit files, the --image= switch in various tools such as
2524 systemd-nspawn, as well as systemd-dissect) will now only mount file
2525 systems of types btrfs, ext4, xfs, erofs, squashfs, vfat. This list
2526 can be overridden via the $SYSTEMD_DISSECT_FILE_SYSTEMS environment
2527 variable. These file systems are fairly well supported and maintained
2528 in current kernels, while others are usually more niche, exotic or
2529 legacy and thus typically do not receive the same level of security
2532 * The default per-link multicast DNS mode is changed to "yes"
2533 (that was previously "no"). As the default global multicast DNS mode
2534 has been "yes" (but can be changed by the build option), now the
2535 multicast DNS is enabled on all links by default. You can disable the
2536 multicast DNS on all links by setting MulticastDNS= in resolved.conf,
2537 or on an interface by calling "resolvectl mdns INTERFACE no".
2541 * A tool 'ukify' tool to build, measure, and sign Unified Kernel Images
2542 (UKIs) has been added. This replaces functionality provided by
2543 'dracut --uefi' and extends it with automatic calculation of PE file
2544 offsets, insertion of signed PCR policies generated by
2545 systemd-measure, support for initrd concatenation, signing of the
2546 embedded Linux image and the combined image with sbsign, and
2547 heuristics to autodetect the kernel uname and verify the splash
2550 Changes in systemd and units:
2552 * A new service type Type=notify-reload is defined. When such a unit is
2553 reloaded a UNIX process signal (typically SIGHUP) is sent to the main
2554 service process. The manager will then wait until it receives a
2555 "RELOADING=1" followed by a "READY=1" notification from the unit as
2556 response (via sd_notify()). Otherwise, this type is the same as
2557 Type=notify. A new setting ReloadSignal= may be used to change the
2558 signal to send from the default of SIGHUP.
2560 user@.service, systemd-networkd.service, systemd-udevd.service, and
2561 systemd-logind have been updated to this type.
2563 * Initrd environments which are not on a pure memory file system (e.g.
2564 overlayfs combination as opposed to tmpfs) are now supported. With
2565 this change, during the initrd → host transition ("switch root")
2566 systemd will erase all files of the initrd only when the initrd is
2567 backed by a memory file system such as tmpfs.
2569 * New per-unit MemoryZSwapMax= option has been added to configure
2570 memory.zswap.max cgroup properties (the maximum amount of zswap
2573 * A new LogFilterPatterns= option has been added for units. It may be
2574 used to specify accept/deny regular expressions for log messages
2575 generated by the unit, that shall be enforced by systemd-journald.
2576 Rejected messages are neither stored in the journal nor forwarded.
2577 This option may be used to suppress noisy or uninteresting messages
2580 * The manager has a new
2581 org.freedesktop.systemd1.Manager.GetUnitByPIDFD() D-Bus method to
2582 query process ownership via a PIDFD, which is more resilient against
2583 PID recycling issues.
2585 * Scope units now support OOMPolicy=. Login session scopes default to
2586 OOMPolicy=continue, allowing login scopes to survive the OOM killer
2587 terminating some processes in the scope.
2589 * systemd-fstab-generator now supports x-systemd.makefs option for
2590 /sysroot/ (in the initrd).
2592 * The maximum rate at which daemon reloads are executed can now be
2593 limited with the new ReloadLimitIntervalSec=/ReloadLimitBurst=
2594 options. (Or the equivalent on the kernel command line:
2595 systemd.reload_limit_interval_sec=/systemd.reload_limit_burst=). In
2596 addition, systemd now logs the originating unit and PID when a reload
2597 request is received over D-Bus.
2599 * When enabling a swap device systemd will now reinitialize the device
2600 when the page size of the swap space does not match the page size of
2601 the running kernel. Note that this requires the 'swapon' utility to
2602 provide the '--fixpgsz' option, as implemented by util-linux, and it
2603 is not supported by busybox at the time of writing.
2605 * systemd now executes generator programs in a mount namespace
2606 "sandbox" with most of the file system read-only and write access
2607 restricted to the output directories, and with a temporary /tmp/
2608 mount provided. This provides a safeguard against programming errors
2609 in the generators, but also fixes here-docs in shells, which
2610 previously didn't work in early boot when /tmp/ wasn't available
2611 yet. (This feature has no security implications, because the code is
2612 still privileged and can trivially exit the sandbox.)
2614 * The system manager will now parse a new "vmm.notify_socket"
2615 system credential, which may be supplied to a VM via SMBIOS. If
2616 found, the manager will send a "READY=1" notification on the
2617 specified socket after boot is complete. This allows readiness
2618 notification to be sent from a VM guest to the VM host over a VSOCK
2621 * The sample PAM configuration file for systemd-user@.service now
2622 includes a call to pam_namespace. This puts children of user@.service
2623 in the expected namespace. (Many distributions replace their file
2624 with something custom, so this change has limited effect.)
2626 * A new environment variable $SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST
2627 can be used to override the mount units burst late limit for
2628 parsing '/proc/self/mountinfo', which was introduced in v249.
2631 * Drop-ins for init.scope changing control group resource limits are
2632 now applied, while they were previously ignored.
2634 * New build-time configuration options '-Ddefault-timeout-sec=' and
2635 '-Ddefault-user-timeout-sec=' have been added, to let distributions
2636 choose the default timeout for starting/stopping/aborting system and
2637 user units respectively.
2639 * Service units gained a new setting OpenFile= which may be used to
2640 open arbitrary files in the file system (or connect to arbitrary
2641 AF_UNIX sockets in the file system), and pass the open file
2642 descriptor to the invoked process via the usual file descriptor
2643 passing protocol. This is useful to give unprivileged services access
2644 to select files which have restrictive access modes that would
2645 normally not allow this. It's also useful in case RootDirectory= or
2646 RootImage= is used to allow access to files from the host environment
2647 (which is after all not visible from the service if these two options
2652 * The new net naming scheme "v253" has been introduced. In the new
2653 scheme, ID_NET_NAME_PATH is also set for USB devices not connected via
2654 a PCI bus. This extends the coverage of predictable interface names
2655 in some embedded systems.
2657 The "amba" bus path is now included in ID_NET_NAME_PATH, resulting in
2658 a more informative path on some embedded systems.
2660 * Partition block devices will now also get symlinks in
2661 /dev/disk/by-diskseq/<seq>-part<n>, which may be used to reference
2662 block device nodes via the kernel's "diskseq" value. Previously those
2663 symlinks were only created for the main block device.
2665 * A new operator '-=' is supported for SYMLINK variables. This allows
2666 symlinks to be unconfigured even if an earlier rule added them.
2668 * 'udevadm --trigger --settle' now also works for network devices
2669 that are being renamed.
2671 Changes in sd-boot, bootctl, and the Boot Loader Specification:
2673 * systemd-boot now passes its random seed directly to the kernel's RNG
2674 via the LINUX_EFI_RANDOM_SEED_TABLE_GUID configuration table, which
2675 means the RNG gets seeded very early in boot before userspace has
2678 * systemd-boot will pass a disk-backed random seed – even when secure
2679 boot is enabled – if it can additionally get a random seed from EFI
2680 itself (via EFI's RNG protocol), or a prior seed in
2681 LINUX_EFI_RANDOM_SEED_TABLE_GUID from a preceding bootloader.
2683 * systemd-boot-system-token.service was renamed to
2684 systemd-boot-random-seed.service and extended to always save a random
2685 seed to ESP on every boot when a compatible boot loader is used. This
2686 allows a refreshed random seed to be used in the boot loader.
2688 * systemd-boot handles various seed inputs using a domain- and
2689 field-separated hashing scheme.
2691 * systemd-boot's 'random-seed-mode' option has been removed. A system
2692 token is now always required to be present for random seeds to be
2695 * systemd-boot now supports being loaded from other locations than the
2696 ESP, for example for direct kernel boot under QEMU or when embedded
2699 * systemd-boot now parses SMBIOS information to detect
2700 virtualization. This information is used to skip some warnings which
2701 are not useful in a VM and to conditionalize other aspects of
2704 * systemd-boot now supports a new 'if-safe' mode that will perform UEFI
2705 Secure Boot automated certificate enrollment from the ESP only if it
2706 is considered 'safe' to do so. At the moment 'safe' means running in
2709 * systemd-stub now processes random seeds in the same way as
2710 systemd-boot already does, in case a unified kernel image is being
2711 used from a different bootloader than systemd-boot, or without any
2714 * bootctl will now generate a system token on all EFI systems, even
2715 virtualized ones, and is activated in the case that the system token
2716 is missing from either sd-boot and sd-stub booted systems.
2718 * bootctl now implements two new verbs: 'kernel-identify' prints the
2719 type of a kernel image file, and 'kernel-inspect' provides
2720 information about the embedded command line and kernel version of
2723 * bootctl now honours $KERNEL_INSTALL_CONF_ROOT with the same meaning
2724 as for kernel-install.
2726 * The JSON output of "bootctl list" will now contain two more fields:
2727 isDefault and isSelected are boolean fields set to true on the
2728 default and currently booted boot menu entries.
2730 * bootctl gained a new verb "unlink" for removing a boot loader entry
2731 type #1 file from disk in a safe and robust way.
2733 * bootctl also gained a new verb "cleanup" that automatically removes
2734 all files from the ESP's and XBOOTLDR's "entry-token" directory, that
2735 is not referenced anymore by any installed Type #1 boot loader
2736 specification entry. This is particularly useful in environments where
2737 a large number of entries reference the same or partly the same
2738 resources (for example, for snapshot-based setups).
2740 Changes in kernel-install:
2742 * A new "installation layout" can be configured as layout=uki. With
2743 this setting, a Boot Loader Specification Type#1 entry will not be
2744 created. Instead, a new kernel-install plugin 90-uki-copy.install
2745 will copy any .efi files from the staging area into the boot
2746 partition. A plugin to generate the UKI .efi file must be provided
2749 Changes in systemctl:
2751 * 'systemctl reboot' has dropped support for accepting a positional
2752 argument as the argument to the reboot(2) syscall. Please use the
2753 --reboot-argument= option instead.
2755 * 'systemctl disable' will now warn when called on units without
2756 install information. A new --no-warn option has been added that
2757 silences this warning.
2759 * New option '--drop-in=' can be used to tell 'systemctl edit' the name
2760 of the drop-in to edit. (Previously, 'override.conf' was always
2763 * 'systemctl list-dependencies' now respects --type= and --state=.
2765 * 'systemctl kexec' now supports XEN VMM environments.
2767 * 'systemctl edit' will now tell the invoked editor to jump into the
2768 first line with actual unit file data, skipping over synthesized
2771 Changes in systemd-networkd and related tools:
2773 * The [DHCPv4] section in .network file gained new SocketPriority=
2774 setting that assigns the Linux socket priority used by the DHCPv4 raw
2775 socket. This may be used in conjunction with the
2776 EgressQOSMaps=setting in [VLAN] section of .netdev file to send the
2777 desired ethernet 802.1Q frame priority for DHCPv4 initial
2778 packets. This cannot be achieved with netfilter mangle tables because
2779 of the raw socket bypass.
2781 * The [DHCPv4] and [IPv6AcceptRA] sections in .network file gained a
2782 new QuickAck= boolean setting that enables the TCP quick ACK mode for
2783 the routes configured by the acquired DHCPv4 lease or received router
2784 advertisements (RAs).
2786 * The RouteMetric= option (for DHCPv4, DHCPv6, and IPv6 advertised
2787 routes) now accepts three values, for high, medium, and low preference
2788 of the router (which can be set with the RouterPreference=) setting.
2790 * systemd-networkd-wait-online now supports matching via alternative
2793 * The [DHCPv6] section in .network file gained new SendRelease=
2794 setting which enables the DHCPv6 client to send release when
2795 it stops. This is the analog of the [DHCPv4] SendRelease= setting.
2796 It is enabled by default.
2798 * If the Address= setting in [Network] or [Address] sections in .network
2799 specified without its prefix length, then now systemd-networkd assumes
2800 /32 for IPv4 or /128 for IPv6 addresses.
2802 * networkctl shows network and link file dropins in status output.
2804 Changes in systemd-dissect:
2806 * systemd-dissect gained a new option --list, to print the paths of
2807 all files and directories in a DDI.
2809 * systemd-dissect gained a new option --mtree, to generate a file
2810 manifest compatible with BSD mtree(5) of a DDI
2812 * systemd-dissect gained a new option --with, to execute a command with
2813 the specified DDI temporarily mounted and used as working
2814 directory. This is for example useful to convert a DDI to "tar"
2815 simply by running it within a "systemd-dissect --with" invocation.
2817 * systemd-dissect gained a new option --discover, to search for
2818 Discoverable Disk Images (DDIs) in well-known directories of the
2819 system. This will list machine, portable service and system extension
2822 * systemd-dissect now understands 2nd stage initrd images stored as a
2823 Discoverable Disk Image (DDI).
2825 * systemd-dissect will now display the main UUID of GPT DDIs (i.e. the
2826 disk UUID stored in the GPT header) among the other data it can show.
2828 * systemd-dissect gained a new --in-memory switch to operate on an
2829 in-memory copy of the specified DDI file. This is useful to access a
2830 DDI with write access without persisting any changes. It's also
2831 useful for accessing a DDI without keeping the originating file
2834 * The DDI dissection logic will now automatically detect the intended
2835 sector size of disk images stored in files, based on the GPT
2836 partition table arrangement. Loopback block devices for such DDIs
2837 will then be configured automatically for the right sector size. This
2838 is useful to make dealing with modern 4K sector size DDIs fully
2839 automatic. The systemd-dissect tool will now show the detected sector
2840 size among the other DDI information in its output.
2842 Changes in systemd-repart:
2844 * systemd-repart gained new options --include-partitions= and
2845 --exclude-partitions= to filter operation on partitions by type UUID.
2846 This allows systemd-repart to be used to build images in which the
2847 type of one partition is set based on the contents of another
2848 partition (for example when the boot partition shall include a verity
2849 hash of the root partition).
2851 * systemd-repart also gained a --defer-partitions= option that is
2852 similar to --exclude-partitions=, but the size of the partition is
2853 still taken into account when sizing partitions, but without
2856 * systemd-repart gained a new --sector-size= option to specify what
2857 sector size should be used when an image is created.
2859 * systemd-repart now supports generating erofs file systems via
2860 CopyFiles= (a read-only file system similar to squashfs).
2862 * The Minimize= option was extended to accept "best" (which means the
2863 most minimal image possible, but may require multiple attempts) and
2864 "guess" (which means a reasonably small image).
2866 * The systemd-growfs binary now comes with a regular unit file template
2867 systemd-growfs@.service which can be instantiated directly for any
2868 desired file system. (Previously, the unit was generated dynamically
2869 by various generators, but no regular unit file template was
2872 Changes in journal tools:
2874 * Various systemd tools will append extra fields to log messages when
2875 in debug mode, or when SYSTEMD_ENABLE_LOG_CONTEXT=1 is set. Currently
2876 this includes information about D-Bus messages when sd-bus is used,
2877 e.g. DBUS_SENDER=, DBUS_DESTINATION=, and DBUS_PATH=, and information
2878 about devices when sd-device is used, e.g. DEVNAME= and DRIVER=.
2879 Details of what is logged and when are subject to change.
2881 * The systemd-journald-audit.socket can now be disabled via the usual
2882 "systemctl disable" mechanism to stop collection of audit
2883 messages. Please note that it is not enabled statically anymore and
2884 must be handled by the preset/enablement logic in package
2885 installation scripts.
2887 * New options MaxUse=, KeepFree=, MaxFileSize=, and MaxFiles= can
2888 be used to curtail disk use by systemd-journal-remote. This is
2889 similar to the options supported by systemd-journald.
2891 Changes in systemd-cryptenroll, systemd-cryptsetup, and related
2894 * When enrolling new keys systemd-cryptenroll now supports unlocking
2895 via FIDO2 tokens (option --unlock-fido2-device=). Previously, a
2896 password was strictly required to be specified.
2898 * systemd-cryptsetup now supports pre-flight requests for FIDO2 tokens
2899 (except for tokens with user verification, UV) to identify tokens
2900 before authentication. Multiple FIDO2 tokens can now be enrolled at
2901 the same time, and systemd-cryptsetup will automatically select one
2902 that corresponds to one of the available LUKS key slots.
2904 * systemd-cryptsetup now supports new options tpm2-measure-bank= and
2905 tpm2-measure-pcr= in crypttab(5). These allow specifying the TPM2 PCR
2906 bank and number into which the volume key should be measured. This is
2907 automatically enabled for the encrypted root volume discovered and
2908 activated by systemd-gpt-auto-generator.
2910 * systemd-gpt-auto-generator mounts the ESP and XBOOTLDR partitions with
2911 "noexec,nosuid,nodev".
2913 * systemd-gpt-auto-generator will now honour the rootfstype= and
2914 rootflags= kernel command line switches for root file systems it
2915 discovers, to match behaviour in case an explicit root fs is
2916 specified via root=.
2918 * systemd-pcrphase gained new options --machine-id and --file-system=
2919 to measure the machine-id and mount point information into PCR 15.
2920 New service unit files systemd-pcrmachine.service and
2921 systemd-pcrfs@.service have been added that invoke the tool with
2922 these switches during early boot.
2924 * systemd-pcrphase gained a --graceful switch will make it exit cleanly
2925 with a success exit code even if no TPM device is detected.
2927 * systemd-cryptenroll now stores the user-supplied PIN with a salt,
2928 making it harder to brute-force.
2930 Changes in other tools:
2932 * systemd-homed gained support for luksPbkdfForceIterations (the
2933 intended number of iterations for the PBKDF operation on LUKS).
2935 * Environment variables $SYSTEMD_HOME_MKFS_OPTIONS_BTRFS,
2936 $SYSTEMD_HOME_MKFS_OPTIONS_EXT4, and $SYSTEMD_HOME_MKFS_OPTIONS_XFS
2937 may now be used to specify additional arguments for mkfs when
2938 systemd-homed formats a file system.
2940 * systemd-hostnamed now exports the contents of
2941 /sys/class/dmi/id/bios_vendor and /sys/class/dmi/id/bios_date via two
2942 new D-Bus properties: FirmwareVendor and FirmwareDate. This allows
2943 unprivileged code to access those values.
2945 systemd-hostnamed also exports the SUPPORT_END= field from
2946 os-release(5) as OperatingSystemSupportEnd. hostnamectl make uses of
2947 this to show the status of the installed system.
2949 * systemd-measure gained an --append= option to sign multiple phase
2950 paths with different signing keys. This allows secrets to be
2951 accessible only in certain parts of the boot sequence. Note that
2952 'ukify' provides similar functionality in a more accessible form.
2954 * systemd-timesyncd will now write a structured log message with
2955 MESSAGE_ID set to SD_MESSAGE_TIME_BUMP when it bumps the clock based
2956 on a on-disk timestamp, similarly to what it did when reaching
2957 synchronization via NTP.
2959 * systemd-timesyncd will now update the on-disk timestamp file on each
2960 boot at least once, making it more likely that the system time
2961 increases in subsequent boots.
2963 * systemd-vconsole-setup gained support for system/service credentials:
2964 vconsole.keymap/vconsole.keymap_toggle and
2965 vconsole.font/vconsole.font_map/vconsole.font_unimap are analogous
2966 the similarly-named options in vconsole.conf.
2968 * systemd-localed will now save the XKB keyboard configuration to
2969 /etc/vconsole.conf, and also read it from there with a higher
2970 preference than the /etc/X11/xorg.conf.d/00-keyboard.conf config
2971 file. Previously, this information was stored in the former file in
2972 converted form, and only in latter file in the original form. Tools
2973 which want to access keyboard configuration can now do so from a
2976 * systemd-resolved gained support for configuring the nameservers and
2977 search domains via kernel command line (nameserver=, domain=) and
2978 credentials (network.dns, network.search_domains).
2980 * systemd-resolved will now synthesize host names for the DNS stub
2981 addresses it supports. Specifically when "_localdnsstub" is resolved,
2982 127.0.0.53 is returned, and if "_localdnsproxy" is resolved
2983 127.0.0.54 is returned.
2985 * systemd-notify will now send a "RELOADING=1" notification when called
2986 with --reloading, and "STOPPING=1" when called with --stopping. This
2987 can be used to implement notifications from units where it's easier
2988 to call a program than to use the sd-daemon library.
2990 * systemd-analyze's 'plot' command can now output its information in
2991 JSON, controlled via the --json= switch. Also, new --table, and
2992 --no-legend options have been added.
2994 * 'machinectl enable' will now automatically enable machines.target
2995 unit in addition to adding the machine unit to the target.
2997 Similarly, 'machinectl start|stop' gained a --now option to enable or
2998 disable the machine unit when starting or stopping it.
3000 * systemd-sysusers will now create /etc/ if it is missing.
3002 * systemd-sleep 'HibernateDelaySec=' setting is changed back to
3003 pre-v252's behaviour, and a new 'SuspendEstimationSec=' setting is
3004 added to provide the new initial value for the new automated battery
3005 estimation functionality. If 'HibernateDelaySec=' is set to any value,
3006 the automated estimate (and thus the automated hibernation on low
3007 battery to avoid data loss) functionality will be disabled.
3009 * Default tmpfiles.d/ configuration will now automatically create
3010 credentials storage directory '/etc/credstore/' with the appropriate,
3011 secure permissions. If '/run/credstore/' exists, its permissions will
3012 be fixed too in case they are not correct.
3014 Changes in libsystemd and shared code:
3016 * sd-bus gained new convenience functions sd_bus_emit_signal_to(),
3017 sd_bus_emit_signal_tov(), and sd_bus_message_new_signal_to().
3019 * sd-id128 functions now return -EUCLEAN (instead of -EIO) when the
3020 128-bit ID in files such as /etc/machine-id has an invalid
3021 format. They also accept NULL as output parameter in more places,
3022 which is useful when the caller only wants to validate the inputs and
3023 does not need the output value.
3025 * sd-login gained new functions sd_pidfd_get_session(),
3026 sd_pidfd_get_owner_uid(), sd_pidfd_get_unit(),
3027 sd_pidfd_get_user_unit(), sd_pidfd_get_slice(),
3028 sd_pidfd_get_user_slice(), sd_pidfd_get_machine_name(), and
3029 sd_pidfd_get_cgroup(), that are analogous to sd_pid_get_*(),
3030 but accept a PIDFD instead of a PID.
3032 * sd-path (and systemd-path) now export four new paths:
3033 SD_PATH_SYSTEMD_SYSTEM_ENVIRONMENT_GENERATOR,
3034 SD_PATH_SYSTEMD_USER_ENVIRONMENT_GENERATOR,
3035 SD_PATH_SYSTEMD_SEARCH_SYSTEM_ENVIRONMENT_GENERATOR, and
3036 SD_PATH_SYSTEMD_SEARCH_USER_ENVIRONMENT_GENERATOR,
3038 * sd_notify() now supports AF_VSOCK as transport for notification
3039 messages (in addition to the existing AF_UNIX support). This is
3040 enabled if $NOTIFY_SOCKET is set in a "vsock:CID:port" format.
3042 * Detection of chroot() environments now works if /proc/ is not
3043 mounted. This affects systemd-detect-virt --chroot, but also means
3044 that systemd tools will silently skip various operations in such an
3047 * "Lockheed Martin Hardened Security for Intel Processors" (HS SRE)
3048 virtualization is now detected.
3050 Changes in the build system:
3052 * Standalone variants of systemd-repart and systemd-shutdown may now be
3053 built (if -Dstandalone=true).
3055 * systemd-ac-power has been moved from /usr/lib/ to /usr/bin/, to, for
3056 example, allow scripts to conditionalize execution on AC power
3059 * The libp11kit library is now loaded through dlopen(3).
3061 Changes in the documentation:
3063 * Specifications that are not closely tied to systemd have moved to
3064 https://uapi-group.org/specifications/: the Boot Loader Specification
3065 and the Discoverable Partitions Specification.
3067 Contributions from: 김인수, 13r0ck, Aidan Dang, Alberto Planas,
3068 Alvin Šipraga, Andika Triwidada, AndyChi, angus-p, Anita Zhang,
3069 Antonio Alvarez Feijoo, Arsen Arsenović, asavah, Benjamin Fogle,
3070 Benjamin Tissoires, berenddeschouwer, BerndAdameit,
3071 Bernd Steinhauser, blutch112, cake03, Callum Farmer, Carlo Teubner,
3072 Charles Hardin, chris, Christian Brauner, Christian Göttsche,
3073 Cristian Rodríguez, Daan De Meyer, Dan Streetman, DaPigGuy,
3074 Darrell Kavanagh, David Tardon, dependabot[bot], Dirk Su,
3075 Dmitry V. Levin, drosdeck, Edson Juliano Drosdeck, edupont,
3076 Eric DeVolder, Erik Moqvist, Evgeny Vereshchagin, Fabian Gurtner,
3077 Felix Riemann, Franck Bui, Frantisek Sumsal, Geert Lorang,
3078 Gerd Hoffmann, Gio, Hannoskaj, Hans de Goede, Hugo Carvalho,
3079 igo95862, Ilya Leoshkevich, Ivan Shapovalov, Jacek Migacz,
3080 Jade Lovelace, Jan Engelhardt, Jan Janssen, Jan Macku, January,
3081 Jason A. Donenfeld, jcg, Jean-Tiare Le Bigot, Jelle van der Waa,
3082 Jeremy Linton, Jian Zhang, Jiayi Chen, Jia Zhang, Joerg Behrmann,
3083 Jörg Thalheim, Joshua Goins, joshuazivkovic, Joshua Zivkovic,
3084 Kai-Chuan Hsieh, Khem Raj, Koba Ko, Lennart Poettering, lichao,
3085 Li kunyu, Luca Boccassi, Luca BRUNO, Ludwig Nussel,
3086 Łukasz Stelmach, Lycowolf, marcel151, Marcus Schäfer, Marek Vasut,
3087 Mark Laws, Michael Biebl, Michał Kotyla, Michal Koutný,
3088 Michal Sekletár, Mike Gilbert, Mike Yuan, MkfsSion, ml,
3089 msizanoen1, mvzlb, MVZ Ludwigsburg, Neil Moore, Nick Rosbrook,
3090 noodlejetski, Pasha Vorobyev, Peter Cai, p-fpv, Phaedrus Leeds,
3091 Philipp Jungkamp, Quentin Deslandes, Raul Tambre, Ray Strode,
3092 reuben olinsky, Richard E. van der Luit, Richard Phibel,
3093 Ricky Tigg, Robin Humble, rogg, Rudi Heitbaum, Sam James,
3094 Samuel Cabrero, Samuel Thibault, Siddhesh Poyarekar, Simon Brand,
3095 Space Meyer, Spindle Security, Steve Ramage, Takashi Sakamoto,
3096 Thomas Haller, Tonći Galić, Topi Miettinen, Torsten Hilbrich,
3097 Tuetuopay, uerdogan, Ulrich Ölmann, Valentin David,
3098 Vitaly Kuznetsov, Vito Caputo, Waltibaba, Will Fancher,
3099 William Roberts, wouter bolsterlee, Youfu Zhang, Yu Watanabe,
3100 Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски,
3103 — Warsaw, 2023-02-15
3107 Announcements of Future Feature Removals:
3109 * We intend to remove cgroup v1 support from systemd release after the
3110 end of 2023. If you run services that make explicit use of cgroup v1
3111 features (i.e. the "legacy hierarchy" with separate hierarchies for
3112 each controller), please implement compatibility with cgroup v2 (i.e.
3113 the "unified hierarchy") sooner rather than later. Most of Linux
3114 userspace has been ported over already.
3116 * We intend to remove support for split-usr (/usr mounted separately
3117 during boot) and unmerged-usr (parallel directories /bin and
3118 /usr/bin, /lib and /usr/lib, etc). This will happen in the second
3119 half of 2023, in the first release that falls into that time window.
3120 For more details, see:
3121 https://lists.freedesktop.org/archives/systemd-devel/2022-September/048352.html
3123 Compatibility Breaks:
3125 * ConditionKernelVersion= checks that use the '=' or '!=' operators
3126 will now do simple string comparisons (instead of version comparisons
3127 à la stverscmp()). Version comparisons are still done for the
3128 ordering operators '<', '>', '<=', '>='. Moreover, if no operator is
3129 specified, a shell-style glob match is now done. This creates a minor
3130 incompatibility compared to older systemd versions when the '*', '?',
3131 '[', ']' characters are used, as these will now match as shell globs
3132 instead of literally. Given that kernel version strings typically do
3133 not include these characters we expect little breakage through this
3136 * The service manager will now read the SELinux label used for SELinux
3137 access checks from the unit file at the time it loads the file.
3138 Previously, the label would be read at the moment of the access
3139 check, which was problematic since at that time the unit file might
3140 already have been updated or removed.
3144 * systemd-measure is a new tool for calculating and signing expected
3145 TPM2 PCR values for a given unified kernel image (UKI) booted via
3146 sd-stub. The public key used for the signature and the signed
3147 expected PCR information can be embedded inside the UKI. This
3148 information can be extracted from the UKI by external tools and code
3149 in the image itself and is made available to userspace in the booted
3152 systemd-cryptsetup, systemd-cryptenroll, and systemd-creds have been
3153 updated to make use of this information if available in the booted
3154 kernel: when locking an encrypted volume/credential to the TPM
3155 systemd-cryptenroll/systemd-creds will use the public key to bind the
3156 volume/credential to any kernel that carries PCR information signed
3157 by the same key pair. When unlocking such volumes/credentials
3158 systemd-cryptsetup/systemd-creds will use the signature embedded in
3159 the booted UKI to gain access.
3161 Binding TPM-based disk encryption to public keys/signatures of PCR
3162 values — instead of literal PCR values — addresses the inherent
3163 "brittleness" of traditional PCR-bound TPM disk encryption schemes:
3164 disks remain accessible even if the UKI is updated, without any TPM
3165 specific preparation during the OS update — as long as each UKI
3166 carries the necessary PCR signature information.
3168 Net effect: if you boot a properly prepared kernel, TPM-bound disk
3169 encryption now defaults to be locked to kernels which carry PCR
3170 signatures from the same key pair. Example: if a hypothetical distro
3171 FooOS prepares its UKIs like this, TPM-based disk encryption is now –
3172 by default – bound to only FooOS kernels, and encrypted volumes bound
3173 to the TPM cannot be unlocked on kernels from other sources. (But do
3174 note this behaviour requires preparation/enabling in the UKI, and of
3175 course users can always enroll non-TPM ways to unlock the volume.)
3177 * systemd-pcrphase is a new tool that is invoked at six places during
3178 system runtime, and measures additional words into TPM2 PCR 11, to
3179 mark milestones of the boot process. This allows binding access to
3180 specific TPM2-encrypted secrets to specific phases of the boot
3181 process. (Example: LUKS2 disk encryption key only accessible in the
3182 initrd, but not later.)
3184 Changes in systemd itself, i.e. the manager and units
3186 * The cpu controller is delegated to user manager units by default, and
3187 CPUWeight= settings are applied to the top-level user slice units
3188 (app.slice, background.slice, session.slice). This provides a degree
3189 of resource isolation between different user services competing for
3192 * Systemd can optionally do a full preset in the "first boot" condition
3193 (instead of just enable-only). This behaviour is controlled by the
3194 compile-time option -Dfirst-boot-full-preset. Right now it defaults
3195 to 'false', but the plan is to switch it to 'true' for the subsequent
3198 * Drop-ins are now allowed for transient units too.
3200 * Systemd will set the taint flag 'support-ended' if it detects that
3201 the OS image is past its end-of-support date. This date is declared
3202 in a new /etc/os-release field SUPPORT_END= described below.
3204 * Two new settings ConditionCredential= and AssertCredential= can be
3205 used to skip or fail units if a certain system credential is not
3208 * ConditionMemory= accepts size suffixes (K, M, G, T, …).
3210 * DefaultSmackProcessLabel= can be used in system.conf and user.conf to
3211 specify the SMACK security label to use when not specified in a unit
3214 * DefaultDeviceTimeoutSec= can be used in system.conf and user.conf to
3215 specify the default timeout when waiting for device units to
3218 * C.UTF-8 is used as the default locale if nothing else has been
3221 * [Condition|Assert]Firmware= have been extended to support certain
3222 SMBIOS fields. For example
3224 ConditionFirmware=smbios-field(board_name = "Custom Board")
3226 conditionalizes the unit to run only when
3227 /sys/class/dmi/id/board_name contains "Custom Board" (without the
3230 * ConditionFirstBoot= now correctly evaluates as true only during the
3231 boot phase of the first boot. A unit executed later, after booting
3232 has completed, will no longer evaluate this condition as true.
3234 * Socket units will now create sockets in the SELinuxContext= of the
3235 associated service unit, if any.
3237 * Boot phase transitions (start initrd → exit initrd → boot complete →
3238 shutdown) will be measured into TPM2 PCR 11, so that secrets can be
3239 bound to a specific runtime phase. E.g.: a LUKS encryption key can be
3240 unsealed only in the initrd.
3242 * Service credentials (i.e. SetCredential=/LoadCredential=/…) will now
3243 also be provided to ExecStartPre= processes.
3245 * Various units are now correctly ordered against
3246 initrd-switch-root.target where previously a conflict without
3247 ordering was configured. A stop job for those units would be queued,
3248 but without the ordering it could be executed only after
3249 initrd-switch-root.service, leading to units not being restarted in
3250 the host system as expected.
3252 * In order to fully support the IPMI watchdog driver, which has not yet
3253 been ported to the new common watchdog device interface,
3254 /dev/watchdog0 will be tried first and systemd will silently fallback
3255 to /dev/watchdog if it is not found.
3257 * New watchdog-related D-Bus properties are now published by systemd:
3258 WatchdogDevice, WatchdogLastPingTimestamp,
3259 WatchdogLastPingTimestampMonotonic.
3261 * At shutdown, API virtual files systems (proc, sys, etc.) will be
3264 * At shutdown, systemd will now log about processes blocking unmounting
3267 * A new meson build option 'clock-valid-range-usec-max' was added to
3268 allow disabling system time correction if RTC returns a timestamp far
3271 * Propagated restart jobs will no longer be discarded while a unit is
3274 * PID 1 will now import system credentials from SMBIOS Type 11 fields
3275 ("OEM vendor strings"), in addition to qemu_fwcfg. This provides a
3276 simple, fast and generic path for supplying credentials to a VM,
3277 without involving external tools such as cloud-init/ignition.
3279 * The CPUWeight= setting of unit files now accepts a new special value
3280 "idle", which configures "idle" level scheduling for the unit.
3282 * Service processes that are activated due to a .timer or .path unit
3283 triggering will now receive information about this via environment
3284 variables. Note that this is information is lossy, as activation
3285 might be coalesced and only one of the activating triggers will be
3286 reported. This is hence more suited for debugging or tracing rather
3287 than for behaviour decisions.
3289 * The riscv_flush_icache(2) system call has been added to the list of
3290 system calls allowed by default when SystemCallFilter= is used.
3292 * The selinux context derived from the target executable, instead of
3293 'init_t' used for the manager itself, is now used when creating
3294 listening sockets for units that specify SELinuxContextFromNet=yes.
3296 Changes in sd-boot, bootctl, and the Boot Loader Specification:
3298 * The Boot Loader Specification has been cleaned up and clarified.
3299 Various corner cases in version string comparisons have been fixed
3300 (e.g. comparisons for empty strings). Boot counting is now part of
3301 the main specification.
3303 * New PCRs measurements are performed during boot: PCR 11 for the
3304 kernel+initrd combo, PCR 13 for any sysext images. If a measurement
3305 took place this is now reported to userspace via the new
3306 StubPcrKernelImage and StubPcrInitRDSysExts EFI variables.
3308 * As before, systemd-stub will measure kernel parameters and system
3309 credentials into PCR 12. It will now report this fact via the
3310 StubPcrKernelParameters EFI variable to userspace.
3312 * The UEFI monotonic boot counter is now included in the updated random
3313 seed file maintained by sd-boot, providing some additional entropy.
3315 * sd-stub will use LoadImage/StartImage to execute the kernel, instead
3316 of arranging the image manually and jumping to the kernel entry
3317 point. sd-stub also installs a temporary UEFI SecurityOverride to
3318 allow the (unsigned) nested image to be booted. This is safe because
3319 the outer (signed) stub+kernel binary must have been verified before
3320 the stub was executed.
3322 * Booting in EFI mixed mode (a 64-bit kernel over 32-bit UEFI firmware)
3323 is now supported by sd-boot.
3325 * bootctl gained a bunch of new options: --all-architectures to install
3326 binaries for all supported EFI architectures, --root= and --image=
3327 options to operate on a directory or disk image, and
3328 --install-source= to specify the source for binaries to install,
3329 --efi-boot-option-description= to control the name of the boot entry.
3331 * The sd-boot stub exports a StubFeatures flag, which is used by
3332 bootctl to show features supported by the stub that was used to boot.
3334 * The PE section offsets that are used by tools that assemble unified
3335 kernel images have historically been hard-coded. This may lead to
3336 overlapping PE sections which may break on boot. The UKI will now try
3337 to detect and warn about this.
3339 Any tools that assemble UKIs must update to calculate these offsets
3340 dynamically. Future sd-stub versions may use offsets that will not
3341 work with the currently used set of hard-coded offsets!
3343 * sd-stub now accepts (and passes to the initrd and then to the full
3344 OS) new PE sections '.pcrsig' and '.pcrkey' that can be used to embed
3345 signatures of expected PCR values, to allow sealing secrets via the
3346 TPM2 against pre-calculated PCR measurements.
3348 Changes in the hardware database:
3350 * 'systemd-hwdb query' now supports the --root= option.
3352 Changes in systemctl:
3354 * systemctl now supports --state= and --type= options for the 'show'
3357 * systemctl gained a new verb 'list-automounts' to list automount
3360 * systemctl gained support for a new --image= switch to be able to
3361 operate on the specified disk image (similar to the existing --root=
3362 which operates relative to some directory).
3364 Changes in systemd-networkd:
3366 * networkd can set Linux NetLabel labels for integration with the
3367 network control in security modules via a new NetLabel= option.
3369 * The RapidCommit= is (re-)introduced to enable faster configuration
3370 via DHCPv6 (RFC 3315).
3372 * networkd gained a new option TCPCongestionControlAlgorithm= that
3373 allows setting a per-route TCP algorithm.
3375 * networkd gained a new option KeepFileDescriptor= to allow keeping a
3376 reference (file descriptor) open on TUN/TAP interfaces, which is
3377 useful to avoid link flaps while the underlying service providing the
3378 interface is being serviced.
3380 * RouteTable= now also accepts route table names.
3382 Changes in systemd-nspawn:
3384 * The --bind= and --overlay= options now support relative paths.
3386 * The --bind= option now supports a 'rootidmap' value, which will
3387 use id-mapped mounts to map the root user inside the container to the
3388 owner of the mounted directory on the host.
3390 Changes in systemd-resolved:
3392 * systemd-resolved now persists DNSOverTLS in its state file too. This
3393 fixes a problem when used in combination with NetworkManager, which
3394 sends the setting only once, causing it to be lost if resolved was
3395 restarted at any point.
3397 * systemd-resolved now exposes a Varlink socket at
3398 /run/systemd/resolve/io.systemd.Resolve.Monitor, accessible only for
3399 root. Processed DNS requests in a JSON format will be published to
3400 any clients connected to this socket.
3402 resolvectl gained a 'monitor' verb to make use of this.
3404 * systemd-resolved now treats unsupported DNSSEC algorithms as INSECURE
3405 instead of returning SERVFAIL, as per RFC:
3406 https://datatracker.ietf.org/doc/html/rfc6840#section-5.2
3408 * OpenSSL is the default crypto backend for systemd-resolved. (gnutls
3409 is still supported.)
3411 Changes in libsystemd and other libraries:
3413 * libsystemd now exports sd_bus_error_setfv() (a convenience function
3414 for setting bus errors), sd_id128_string_equal (a convenience
3415 function for 128-bit ID string comparisons), and
3416 sd_bus_message_read_strv_extend() (a function to incrementally read
3419 * libsystemd now exports sd_device_get_child_first()/_next() as a
3420 high-level interface for enumerating child devices. It also supports
3421 sd_device_new_child() for opening a child device given a device
3424 * libsystemd now exports sd_device_monitor_set()/get_description()
3425 which allow setting a custom description that will be used in log
3426 messages by sd_device_monitor*.
3428 * Private shared libraries (libsystemd-shared-nnn.so,
3429 libsystemd-core-nnn.so) are now installed into arch-specific
3430 directories to allow multi-arch installs.
3432 * A new sd-gpt.h header is now published, listing GUIDs from the
3433 Discoverable Partitions specification. For more details see:
3434 https://systemd.io/DISCOVERABLE_PARTITIONS/
3436 * A new function sd_hwdb_new_from_path() has been added to open a hwdb
3437 database given an explicit path to the file.
3439 * The signal number argument to sd_event_add_signal() now can now be
3440 ORed with the SD_EVENT_SIGNAL_PROCMASK flag, causing sigprocmask() to
3441 be automatically invoked to block the specified signal. This is
3442 useful to simplify invocations as the caller doesn't have to do this
3445 * A new convenience call sd_event_set_signal_exit() has been added to
3446 sd-event to set up signal handling so that the event loop
3447 automatically terminates cleanly on SIGTERM/SIGINT.
3449 Changes in other components:
3451 * systemd-sysusers, systemd-tmpfiles, and systemd-sysctl configuration
3452 can now be provided via the credential mechanism.
3454 * systemd-analyze gained a new verb 'compare-versions' that implements
3455 comparisons for versions strings (similarly to 'rpmdev-vercmp' and
3456 'dpkg --compare-versions').
3458 * 'systemd-analyze dump' is extended to accept glob patterns for unit
3459 names to limit the output to matching units.
3461 * tmpfiles.d/ lines can read file contents to write from a credential.
3462 The new modifier char '^' is used to specify that the argument is a
3463 credential name. This mechanism is used to automatically populate
3464 /etc/motd, /etc/issue, and /etc/hosts from credentials.
3466 * tmpfiles.d/ may now be configured to avoid changing uid/gid/mode of
3467 an inode if the specification is prefixed with ':' and the inode
3470 * Default tmpfiles.d/ configuration now carries a line to automatically
3471 use an 'ssh.authorized_keys.root' credential if provided to set up
3472 the SSH authorized_keys file for the root user.
3474 * systemd-tmpfiles will now gracefully handle absent source of "C" copy
3477 * tmpfiles.d/ F/w lines now optionally permit encoding of the payload
3478 in base64. This is useful to write arbitrary binary data into files.
3480 * The pkgconfig and rpm macros files now export the directory for user
3481 units as 'user_tmpfiles_dir' and '%_user_tmpfilesdir'.
3483 * Detection of Apple Virtualization and detection of Parallels and
3484 KubeVirt virtualization on non-x86 archs have been added.
3486 * os-release gained a new field SUPPORT_END=YYYY-MM-DD to inform the
3487 user when their system will become unsupported.
3489 * When performing suspend-then-hibernate, the system will estimate the
3490 discharge rate and use that to set the delay until hibernation and
3491 hibernate immediately instead of suspending when running from a
3492 battery and the capacity is below 5%.
3494 * systemd-sysctl gained a --strict option to fail when a sysctl
3495 setting is unknown to the kernel.
3497 * machinectl supports --force for the 'copy-to' and 'copy-from'
3500 * coredumpctl gained the --root and --image options to look for journal
3501 files under the specified root directory, image, or block device.
3503 * 'journalctl -o' and similar commands now implement a new output mode
3504 "short-delta". It is similar to "short-monotonic", but also shows the
3505 time delta between subsequent messages.
3507 * journalctl now respects the --quiet flag when verifying consistency
3510 * Journal log messages gained a new implicit field _RUNTIME_SCOPE= that
3511 will indicate whether a message was logged in the 'initrd' phase or
3512 in the 'system' phase of the boot process.
3514 * Journal files gained a new compatibility flag
3515 'HEADER_INCOMPATIBLE_COMPACT'. Files with this flag implement changes
3516 to the storage format that allow reducing size on disk. As with other
3517 compatibility flags, older journalctl versions will not be able to
3518 read journal files using this new format. The environment variable
3519 'SYSTEMD_JOURNAL_COMPACT=0' can be passed to systemd-journald to
3520 disable this functionality. It is enabled by default.
3522 * systemd-run's --working-directory= switch now works when used in
3523 combination with --scope.
3525 * portablectl gained a --force flag to skip certain sanity checks. This
3526 is implemented using new flags accepted by systemd-portabled for the
3527 *WithExtensions() D-Bus methods: SD_SYSTEMD_PORTABLE_FORCE_ATTACH
3528 flag now means that the attach/detach checks whether the units are
3529 already present and running will be skipped. Similarly,
3530 SD_SYSTEMD_PORTABLE_FORCE_SYSEXT flag means that the check whether
3531 image name matches the name declared inside of the image will be
3532 skipped. Callers must be sure to do those checks themselves if
3535 * systemd-portabled will now use the original filename to check
3536 extension-release.NAME for correctness, in case it is passed a
3539 * systemd-portabled now uses PrivateTmp=yes in the 'trusted' profile
3542 * sysext's extension-release files now support '_any' as a special
3543 value for the ID= field, to allow distribution-independent extensions
3544 (e.g.: fully statically compiled binaries, scripts). It also gained
3545 support for a new ARCHITECTURE= field that may be used to explicitly
3546 restrict an image to hosts of a specific architecture.
3548 * systemd-repart now supports creating squashfs partitions. This
3549 requires mksquashfs from squashfs-tools.
3551 * systemd-repart gained a --split flag to also generate split
3552 artifacts, i.e. a separate file for each partition. This is useful in
3553 conjunction with systemd-sysupdate or other tools, or to generate
3554 split dm-verity artifacts.
3556 * systemd-repart is now able to generate dm-verity partitions, including
3559 * systemd-repart can now set a partition UUID to zero, allowing it to
3560 be filled in later, such as when using verity partitions.
3562 * systemd-repart now supports drop-ins for its configuration files.
3564 * Package metadata logged by systemd-coredump in the system journal is
3567 * xdg-autostart-service now expands 'tilde' characters in Exec lines.
3569 * systemd-oomd now automatically links against libatomic, if available.
3571 * systemd-oomd now sends out a 'Killed' D-Bus signal when a cgroup is
3574 * scope units now also provide oom-kill status.
3576 * systemd-pstore will now try to load only the efi_pstore kernel module
3577 before running, ensuring that pstore can be used.
3579 * systemd-logind gained a new StopIdleSessionSec= option to stop an idle
3580 session after a preconfigure timeout.
3582 * systemd-homed will now wait up to 30 seconds for workers to terminate,
3583 rather than indefinitely.
3585 * homectl gained a new '--luks-sector-size=' flag that allows users to
3586 select the preferred LUKS sector size. Must be a power of 2 between 512
3587 and 4096. systemd-userdbd records gained a corresponding field.
3589 * systemd-sysusers will now respect the 'SOURCE_DATE_EPOCH' environment
3590 variable when generating the 'sp_lstchg' field, to ensure an image
3591 build can be reproducible.
3593 * 'udevadm wait' will now listen to kernel uevents too when called with
3596 * When naming network devices udev will now consult the Devicetree
3597 "alias" fields for the device.
3599 * systemd-udev will now create infiniband/by-path and
3600 infiniband/by-ibdev links for Infiniband verbs devices.
3602 * systemd-udev-trigger.service will now also prioritize input devices.
3604 * ConditionACPower= and systemd-ac-power will now assume the system is
3605 running on AC power if no battery can be found.
3607 * All features and tools using the TPM2 will now communicate with it
3608 using a bind key. Beforehand, the tpm2 support used encrypted sessions
3609 by creating a primary key that was used to encrypt traffic. This
3610 creates a problem as the key created for encrypting the traffic could
3611 be faked by an active interposer on the bus. In cases when a pin is
3612 used, a bind key will be used. The pin is used as the auth value for
3613 the seal key, aka the disk encryption key, and that auth value will be
3614 used in the session establishment. An attacker would need the pin
3615 value to create the secure session and thus an active interposer
3616 without the pin cannot interpose on TPM2 traffic.
3618 * systemd-growfs no longer requires udev to run.
3620 * systemd-backlight now will better support systems with multiple
3623 * systemd-cryptsetup's keyfile-timeout= option now also works when a
3624 device is used as a keyfile.
3626 * systemd-cryptenroll gained a new --unlock-key-file= option to get the
3627 unlocking key from a key file (instead of prompting the user). Note
3628 that this is the key for unlocking the volume in order to be able to
3629 enroll a new key, but it is not the key that is enrolled.
3631 * systemd-dissect gained a new --umount switch that will safely and
3632 synchronously unmount all partitions of an image previously mounted
3633 with 'systemd-dissect --mount'.
3635 * When using gcrypt, all systemd tools and services will now configure
3636 it to prefer the OS random number generator if present.
3638 * All example code shipped with documentation has been relicensed from CC0
3641 * Unit tests will no longer fail when running on a system without
3644 Experimental features:
3646 * BPF programs can now be compiled with bpf-gcc (requires libbpf >= 1.0
3647 and bpftool >= 7.0).
3649 * sd-boot can automatically enroll SecureBoot keys from files found on
3650 the ESP. This enrollment can be either automatic ('force' mode) or
3651 controlled by the user ('manual' mode). It is sufficient to place the
3652 SecureBoot keys in the right place in the ESP and they will be picked
3653 up by sd-boot and shown in the boot menu.
3655 * The mkosi config in systemd gained support for automatically
3656 compiling a kernel with the configuration appropriate for testing
3657 systemd. This may be useful when developing or testing systemd in
3658 tandem with the kernel.
3660 Contributions from: 김인수, Adam Williamson, adrian5, Aidan Dang,
3661 Akihiko Odaki, Alban Bedel, Albert Mikaelyan, Aleksey Vasenev,
3662 Alexander Graf, Alexander Shopov, Alexander Wilson,
3663 Alper Nebi Yasak, anarcat, Anders Jonsson, Andre Kalb,
3664 Andrew Stone, Andrey Albershteyn, Anita Zhang, Ansgar Burchardt,
3665 Antonio Alvarez Feijoo, Arnaud Ferraris, Aryan singh, asavah,
3666 Avamander, Avram Lubkin, Balázs Meskó, Bastien Nocera,
3667 Benjamin Franzke, BerndAdameit, bin456789, Celeste Liu,
3668 Chih-Hsuan Yen, Christian Brauner, Christian Göttsche,
3669 Christian Hesse, Clyde Byrd III, codefiles, Colin Walters,
3670 Cristian Rodríguez, Daan De Meyer, Daniel Braunwarth,
3671 Daniel Rusek, Dan Streetman, Darsey Litzenberger, David Edmundson,
3672 David Jaša, David Rheinsberg, David Seifert, David Tardon,
3673 dependabot[bot], Devendra Tewari, Dominique Martinet, drosdeck,
3674 Edson Juliano Drosdeck, Eduard Tolosa, eggfly, Einsler Lee,
3675 Elias Probst, Eli Schwartz, Evgeny Vereshchagin, exploide, Fei Li,
3676 Foster Snowhill, Franck Bui, Frank Dana, Frantisek Sumsal,
3677 Gerd Hoffmann, Gio, Goffredo Baroncelli, gtwang01,
3678 Guillaume W. Bres, H A, Hans de Goede, Heinrich Schuchardt,
3679 Hugo Carvalho, i-do-cpp, igo95862, j00512545, Jacek Migacz,
3680 Jade Bilkey, James Hilliard, Jan B, Janis Goldschmidt,
3681 Jan Janssen, Jan Kuparinen, Jan Luebbe, Jan Macku,
3682 Jason A. Donenfeld, Javkhlanbayar Khongorzul, Jeremy Soller,
3683 JeroenHD, jiangchuangang, João Loureiro,
3684 Joaquín Ignacio Aramendía, Jochen Sprickerhof,
3685 Johannes Schauer Marin Rodrigues, Jonas Kümmerlin,
3686 Jonas Witschel, Jonathan Kang, Jonathan Lebon, Joost Heitbrink,
3687 Jörg Thalheim, josh-gordon-fb, Joyce, Kai Lueke, lastkrick,
3688 Lennart Poettering, Leon M. George, licunlong, Li kunyu,
3689 LockBlock-dev, Loïc Collignon, Lubomir Rintel, Luca Boccassi,
3690 Luca BRUNO, Ludwig Nussel, Łukasz Stelmach, Maccraft123,
3691 Marc Kleine-Budde, Marius Vollmer, Martin Wilck, matoro,
3692 Matthias Lisin, Max Gautier, Maxim Mikityanskiy, Michael Biebl,
3693 Michal Koutný, Michal Sekletár, Michal Stanke, Mike Gilbert,
3694 Mitchell Freiderich, msizanoen1, Nick Rosbrook, nl6720, Oğuz Ersen,
3695 Oleg Solovyov, Olga Smirnova, Pablo Ceballos, Pavel Zhukov,
3696 Phaedrus Leeds, Philipp Gortan, Piotr Drąg, Pyfisch,
3697 Quentin Deslandes, Rahil Bhimjiani, Rene Hollander, Richard Huang,
3698 Richard Phibel, Rudi Heitbaum, Sam James, Sarah Brofeldt,
3699 Sean Anderson, Sebastian Scheibner, Shreenidhi Shedi,
3700 Sonali Srivastava, Steve Ramage, Suraj Krishnan, Swapnil Devesh,
3701 Takashi Sakamoto, Ted X. Toth, Temuri Doghonadze, Thomas Blume,
3702 Thomas Haller, Thomas Hebb, Tomáš Hnyk, Tomasz Paweł Gajc,
3703 Topi Miettinen, Ulrich Ölmann, undef, Uriel Corfa,
3704 Victor Westerhuis, Vincent Dagonneau, Vishal Chillara Srinivas,
3705 Vito Caputo, Weblate, Wenchao Hao, William Roberts, williamsumendap,
3706 wineway, xiaoyang, Yuri Chornoivan, Yu Watanabe,
3707 Zbigniew Jędrzejewski-Szmek, Zhaofeng Li, наб
3709 – The Great Beyond, 2022-10-31 👻
3713 Backwards-incompatible changes:
3715 * The minimum kernel version required has been bumped from 3.13 to 4.15,
3716 and CLOCK_BOOTTIME is now assumed to always exist.
3718 * C11 with GNU extensions (aka "gnu11") is now used to build our
3719 components. Public API headers are still restricted to ISO C89.
3721 * In v250, a systemd-networkd feature that automatically configures
3722 routes to addresses specified in AllowedIPs= was added and enabled by
3723 default. However, this causes network connectivity issues in many
3724 existing setups. Hence, it has been disabled by default since
3725 systemd-stable 250.3. The feature can still be used by explicitly
3726 configuring RouteTable= setting in .netdev files.
3728 * Jobs started via StartUnitWithFlags() will no longer return 'skipped'
3729 when a Condition*= check does not succeed, restoring the JobRemoved
3730 signal to the behaviour it had before v250.
3732 * The org.freedesktop.portable1 methods GetMetadataWithExtensions() and
3733 GetImageMetadataWithExtensions() have been fixed to provide an extra
3734 return parameter, containing the actual extension release metadata.
3735 The current implementation was judged to be broken and unusable, and
3736 thus the usual procedure of adding a new set of methods was skipped,
3737 and backward compatibility broken instead on the assumption that
3738 nobody can be affected given the current state of this interface.
3740 * All kernels supported by systemd mix bytes returned by RDRAND (or
3741 similar) into the entropy pool at early boot. This means that on
3742 those systems, even if /dev/urandom is not yet initialized, it still
3743 returns bytes that are of at least RDRAND quality. For that reason,
3744 we no longer have reason to invoke RDRAND from systemd itself, which
3745 has historically been a source of bugs. Furthermore, kernels ≥5.6
3746 provide the getrandom(GRND_INSECURE) interface for returning random
3747 bytes before the entropy pool is initialized without warning into
3748 kmsg, which is what we attempt to use if available. systemd's direct
3749 usage of RDRAND has been removed. x86 systems ≥Broadwell that are
3750 running an older kernel may experience kmsg warnings that were not
3751 seen with 250. For newer kernels, non-x86 systems, or older x86
3752 systems, there should be no visible changes.
3754 * sd-boot will now measure the kernel command line into TPM PCR 12
3755 rather than PCR 8. This improves usefulness of the measurements on
3756 systems where sd-boot is chainloaded from Grub. Grub measures all
3757 commands its executes into PCR 8, which makes it very hard to use
3758 reasonably, hence separate ourselves from that and use PCR 12
3759 instead, which is what certain Ubuntu editions already do. To retain
3760 compatibility with systems running older systemd systems a new meson
3761 option 'efi-tpm-pcr-compat' has been added (which defaults to false).
3762 If enabled, the measurement is done twice: into the new-style PCR 12
3763 *and* the old-style PCR 8. It's strongly advised to migrate all users
3764 to PCR 12 for this purpose in the long run, as we intend to remove
3765 this compatibility feature in two years' time.
3767 * busctl capture now writes output in the newer pcapng format instead
3770 * A udev rule that imported hwdb matches for USB devices with lowercase
3771 hexadecimal vendor/product ID digits was added in systemd 250. This
3772 has been reverted, since uppercase hexadecimal digits are supposed to
3773 be used, and we already had a rule with the appropriate match.
3775 Users might need to adjust their local hwdb entries.
3777 * arch_prctl(2) has been moved to the @default set in the syscall filters
3778 (as exposed via the SystemCallFilter= setting in service unit files).
3779 It is apparently used by the linker now.
3781 * The tmpfiles entries that create the /run/systemd/netif directory and
3782 its subdirectories were moved from tmpfiles.d/systemd.conf to
3783 tmpfiles.d/systemd-network.conf.
3785 Users might need to adjust their files that override tmpfiles.d/systemd.conf
3786 to account for this change.
3788 * The requirement for Portable Services images to contain a well-formed
3789 os-release file (i.e.: contain at least an ID field) is now enforced.
3790 This applies to base images and extensions, and also to systemd-sysext.
3792 Changes in the Boot Loader Specification, kernel-install and sd-boot:
3794 * kernel-install's and bootctl's Boot Loader Specification Type #1
3795 entry generation logic has been reworked. The user may now pick
3796 explicitly by which "token" string to name the installation's boot
3797 entries, via the new /etc/kernel/entry-token file or the new
3798 --entry-token= switch to bootctl. By default — as before — the
3799 entries are named after the local machine ID. However, in "golden
3800 image" environments, where the machine ID shall be initialized on
3801 first boot (as opposed to at installation time before first boot) the
3802 machine ID will not be available at build time. In this case the
3803 --entry-token= switch to bootctl (or the /etc/kernel/entry-token
3804 file) may be used to override the "token" for the entries, for
3805 example the IMAGE_ID= or ID= fields from /etc/os-release. This will
3806 make the OS images independent of any machine ID, and ensure that the
3807 images will not carry any identifiable information before first boot,
3808 but on the other hand means that multiple parallel installations of
3809 the very same image on the same disk cannot be supported.
3811 Summary: if you are building golden images that shall acquire
3812 identity information exclusively on first boot, make sure to both
3813 remove /etc/machine-id *and* to write /etc/kernel/entry-token to the
3814 value of the IMAGE_ID= or ID= field of /etc/os-release or another
3815 suitable identifier before deploying the image.
3817 * The Boot Loader Specification has been extended with
3818 /loader/entries.srel file located in the EFI System Partition (ESP)
3819 that disambiguates the format of the entries in the /loader/entries/
3820 directory (in order to discern them from incompatible uses of this
3821 directory by other projects). For entries that follow the
3822 Specification, the string "type1" is stored in this file.
3824 bootctl will now write this file automatically when installing the
3825 systemd-boot boot loader.
3827 * kernel-install supports a new initrd_generator= setting in
3828 /etc/kernel/install.conf, that is exported as
3829 $KERNEL_INSTALL_INITRD_GENERATOR to kernel-install plugins. This
3830 allows choosing different initrd generators.
3832 * kernel-install will now create a "staging area" (an initially-empty
3833 directory to gather files for a Boot Loader Specification Type #1
3834 entry). The path to this directory is exported as
3835 $KERNEL_INSTALL_STAGING_AREA to kernel-install plugins, which should
3836 drop files there instead of writing them directly to the final
3837 location. kernel-install will move them when all files have been
3838 prepared successfully.
3840 * New option sort-key= has been added to the Boot Loader Specification
3841 to override the sorting order of the entries in the boot menu. It is
3842 read by sd-boot and bootctl, and will be written by kernel-install,
3843 with the default value of IMAGE_ID= or ID= fields from
3844 os-release. Together, this means that on multiboot installations,
3845 entries should be grouped and sorted in a predictable way.
3847 * The sort order of boot entries has been updated: entries which have
3848 the new field sort-key= are sorted by it first, and all entries
3849 without it are ordered later. After that, entries are sorted by
3850 version so that newest entries are towards the beginning of the list.
3852 * The kernel-install tool gained a new 'inspect' verb which shows the
3853 paths and other settings used.
3855 * sd-boot can now optionally beep when the menu is shown and menu
3856 entries are selected, which can be useful on machines without a
3857 working display. (Controllable via a loader.conf setting.)
3859 * The --make-machine-id-directory= switch to bootctl has been replaced
3860 by --make-entry-directory=, given that the entry directory is not
3861 necessarily named after the machine ID, but after some other suitable
3862 ID as selected via --entry-token= described above. The old name of
3863 the option is still understood to maximize compatibility.
3865 * 'bootctl list' gained support for a new --json= switch to output boot
3866 menu entries in JSON format.
3868 * 'bootctl is-installed' now supports the --graceful, and various verbs
3869 omit output with the new option --quiet.
3871 Changes in systemd-homed:
3873 * Starting with v250 systemd-homed uses UID/GID mapping on the mounts
3874 of activated home directories it manages (if the kernel and selected
3875 file systems support it). So far it mapped three UID ranges: the
3876 range from 0…60000, the user's own UID, and the range 60514…65534,
3877 leaving everything else unmapped (in other words, the 16-bit UID range
3878 is mapped almost fully, with the exception of the UID subrange used
3879 for systemd-homed users, with one exception: the user's own UID).
3880 Unmapped UIDs may not be used for file ownership in the home
3881 directory — any chown() attempts with them will fail. With this
3882 release a fourth range is added to these mappings:
3883 524288…1879048191. This range is the UID range intended for container
3886 https://systemd.io/UIDS-GIDS
3888 This range may be used for container managers that place container OS
3889 trees in the home directory (which is a questionable approach, for
3890 quota, permission, SUID handling and network file system
3891 compatibility reasons, but nonetheless apparently commonplace). Note
3892 that this mapping is mapped 1:1 in a pass-through fashion, i.e. the
3893 UID assignments from the range are not managed or mapped by
3894 `systemd-homed`, and must be managed with other mechanisms, in the
3895 context of the local system.
3897 Typically, a better approach to user namespacing in relevant
3898 container managers would be to leave container OS trees on disk at
3899 UID offset 0, but then map them to a dynamically allocated runtime
3900 UID range via another UID mount map at container invocation
3901 time. That way user namespace UID ranges become strictly a runtime
3902 concept, and do not leak into persistent file systems, persistent
3903 user databases or persistent configuration, thus greatly simplifying
3904 handling, and improving compatibility with home directories intended
3905 to be portable like the ones managed by systemd-homed.
3907 Changes in shared libraries:
3909 * A new libsystemd-core-<version>.so private shared library is
3910 installed under /usr/lib/systemd/system, mirroring the existing
3911 libsystemd-shared-<version>.so library. This allows the total
3912 installation size to be reduced by binary code reuse.
3914 * The <version> tag used in the name of libsystemd-shared.so and
3915 libsystemd-core.so can be configured via the meson option
3916 'shared-lib-tag'. Distributions may build subsequent versions of the
3917 systemd package with unique tags (e.g. the full package version),
3918 thus allowing multiple installations of those shared libraries to be
3919 available at the same time. This is intended to fix an issue where
3920 programs that link to those libraries would fail to execute because
3921 they were installed earlier or later than the appropriate version of
3924 * The sd-id128 API gained a new call sd_id128_to_uuid_string() that is
3925 similar to sd_id128_to_string() but formats the ID in RFC 4122 UUID
3926 format instead of as a simple series of hex characters.
3928 * The sd-device API gained two new calls sd_device_new_from_devname()
3929 and sd_device_new_from_path() which permit allocating an sd_device
3930 object from a device node name or file system path.
3932 * sd-device also gained a new call sd_device_open() which will open the
3933 device node associated with a device for which an sd_device object
3934 has been allocated. The call is supposed to address races around
3935 device nodes being removed/recycled due to hotplug events, or media
3936 change events: the call checks internally whether the major/minor of
3937 the device node and the "diskseq" (in case of block devices) match
3938 with the metadata loaded in the sd_device object, thus ensuring that
3939 the device once opened really matches the provided sd_device object.
3941 Changes in PID1, systemctl, and systemd-oomd:
3943 * A new set of service monitor environment variables will be passed to
3944 OnFailure=/OnSuccess= handlers, but only if exactly one unit lists the
3945 handler unit as OnFailure=/OnSuccess=. The variables are:
3946 $MONITOR_SERVICE_RESULT, $MONITOR_EXIT_CODE, $MONITOR_EXIT_STATUS,
3947 $MONITOR_INVOCATION_ID and $MONITOR_UNIT. For cases when a single
3948 handler needs to watch multiple units, use a templated handler.
3950 * A new ExtensionDirectories= setting in service unit files allows
3951 system extensions to be loaded from a directory. (It is similar to
3952 ExtensionImages=, but takes paths to directories, instead of
3955 'portablectl attach --extension=' now also accepts directory paths.
3957 * The user.delegate and user.invocation_id extended attributes on
3958 cgroups are used in addition to trusted.delegate and
3959 trusted.invocation_id. The latter pair requires privileges to set,
3960 but the former doesn't and can be also set by the unprivileged user
3963 (Only supported on kernels ≥5.6.)
3965 * Units that were killed by systemd-oomd will now have a service result
3966 of 'oom-kill'. The number of times a service was killed is tallied
3967 in the 'user.oomd_ooms' extended attribute.
3969 The OOMPolicy= unit file setting is now also honoured by
3972 * In unit files the new %y/%Y specifiers can be used to refer to
3973 normalized unit file path, which is particularly useful for symlinked
3976 The new %q specifier resolves to the pretty hostname
3977 (i.e. PRETTY_HOSTNAME= from /etc/machine-info).
3979 The new %d specifier resolves to the credentials directory of a
3980 service (same as $CREDENTIALS_DIRECTORY).
3982 * The RootDirectory=, MountAPIVFS=, ExtensionDirectories=,
3983 *Capabilities*=, ProtectHome=, *Directory=, TemporaryFileSystem=,
3984 PrivateTmp=, PrivateDevices=, PrivateNetwork=, NetworkNamespacePath=,
3985 PrivateIPC=, IPCNamespacePath=, PrivateUsers=, ProtectClock=,
3986 ProtectKernelTunables=, ProtectKernelModules=, ProtectKernelLogs=,
3987 MountFlags= service settings now also work in unprivileged user
3988 services, i.e. those run by the user's --user service manager, as long
3989 as user namespaces are enabled on the system.
3991 * Services with Restart=always and a failing ExecCondition= will no
3992 longer be restarted, to bring ExecCondition= behaviour in line with
3993 Condition*= settings.
3995 * LoadCredential= now accepts a directory as the argument; all files
3996 from the directory will be loaded as credentials.
3998 * A new D-Bus property ControlGroupId is now exposed on service units,
3999 that encapsulates the service's numeric cgroup ID that newer kernels
4000 assign to each cgroup.
4002 * PID 1 gained support for configuring the "pre-timeout" of watchdog
4003 devices and the associated governor, via the new
4004 RuntimeWatchdogPreSec= and RuntimeWatchdogPreGovernor= configuration
4005 options in /etc/systemd/system.conf.
4007 * systemctl's --timestamp= option gained a new choice "unix", to show
4008 timestamp as unix times, i.e. seconds since 1970, Jan 1st.
4010 * A new "taint" flag named "old-kernel" is introduced which is set when
4011 the kernel systemd runs on is older then the current baseline version
4012 (see above). The flag is shown in "systemctl status" output.
4014 * Two additional taint flags "short-uid-range" and "short-gid-range"
4015 have been added as well, which are set when systemd notices it is run
4016 within a userns namespace that does not define the full 0…65535 UID
4019 * A new "unmerged-usr" taint flag has been added that is set whenever
4020 running on systems where /bin/ + /sbin/ are *not* symlinks to their
4021 counterparts in /usr/, i.e. on systems where the /usr/-merge has not
4024 * Generators invoked by PID 1 will now have a couple of useful
4025 environment variables set describing the execution context a
4026 bit. $SYSTEMD_SCOPE encodes whether the generator is called from the
4027 system service manager, or from the per-user service
4028 manager. $SYSTEMD_IN_INITRD encodes whether the generator is invoked
4029 in initrd context or on the host. $SYSTEMD_FIRST_BOOT encodes whether
4030 systemd considers the current boot to be a "first"
4031 boot. $SYSTEMD_VIRTUALIZATION encode whether virtualization is
4032 detected and which type of hypervisor/container
4033 manager. $SYSTEMD_ARCHITECTURE indicates which architecture the
4034 kernel is built for.
4036 * PID 1 will now automatically pick up system credentials from qemu's
4037 fw_cfg interface, thus allowing passing arbitrary data into VM
4038 systems similar to how this is already supported for passing them
4039 into `systemd-nspawn` containers. Credentials may now also be passed
4040 in via the new kernel command line option `systemd.set_credential=`
4041 (note that kernel command line options are world-readable during
4042 runtime, and only useful for credentials that require no
4043 confidentiality). The credentials that can be passed to unified
4044 kernels that use the `systemd-stub` UEFI stub are now similarly
4045 picked up automatically. Automatic importing of system credentials
4046 this way can be turned off via the new
4047 `systemd.import_credentials=no` kernel command line option.
4049 * LoadCredential= will now automatically look for credentials in the
4050 /etc/credstore/, /run/credstore/, /usr/lib/credstore/ directories if
4051 the argument is not an absolute path. Similarly,
4052 LoadCredentialEncrypted= will check the same directories plus
4053 /etc/credstore.encrypted/, /run/credstore.encrypted/ and
4054 /usr/lib/credstore.encrypted/. The idea is to use those directories
4055 as the system-wide location for credentials that services should pick
4058 * System and service credentials are described in great detail in a new
4061 https://systemd.io/CREDENTIALS
4063 Changes in systemd-journald:
4065 * The journal JSON export format has been added to listed of stable
4066 interfaces (https://systemd.io/PORTABILITY_AND_STABILITY/).
4068 * journalctl --list-boots now supports JSON output and the --reverse option.
4070 * Under docs/: JOURNAL_EXPORT_FORMATS was imported from the wiki and
4071 updated, BUILDING_IMAGES is new:
4073 https://systemd.io/JOURNAL_EXPORT_FORMATS
4074 https://systemd.io/BUILDING_IMAGES
4078 * Two new hwdb files have been added. One lists "handhelds" (PDAs,
4079 calculators, etc.), the other AV production devices (DJ tables,
4080 keypads, etc.) that should accessible to the seat owner user by
4083 * udevadm trigger gained a new --prioritized-subsystem= option to
4084 process certain subsystems (and all their parent devices) earlier.
4086 systemd-udev-trigger.service now uses this new option to trigger
4087 block and TPM devices first, hopefully making the boot a bit faster.
4089 * udevadm trigger now implements --type=all, --initialized-match,
4090 --initialized-nomatch to trigger both subsystems and devices, only
4091 already-initialized devices, and only devices which haven't been
4092 initialized yet, respectively.
4094 * udevadm gained a new "wait" command for safely waiting for a specific
4095 device to show up in the udev device database. This is useful in
4096 scripts that asynchronously allocate a block device (e.g. through
4097 repartitioning, or allocating a loopback device or similar) and need
4098 to synchronize on the creation to complete.
4100 * udevadm gained a new "lock" command for locking one or more block
4101 devices while formatting it or writing a partition table to it. It is
4102 an implementation of https://systemd.io/BLOCK_DEVICE_LOCKING and
4103 usable in scripts dealing with block devices.
4105 * udevadm info will show a couple of additional device fields in its
4106 output, and will not apply a limited set of coloring to line types.
4108 * udevadm info --tree will now show a tree of objects (i.e. devices and
4109 suchlike) in the /sys/ hierarchy.
4111 * Block devices will now get a new set of device symlinks in
4112 /dev/disk/by-diskseq/<nr>, which may be used to reference block
4113 device nodes via the kernel's "diskseq" value. Note that this does
4114 not guarantee that opening a device by a symlink like this will
4115 guarantee that the opened device actually matches the specified
4116 diskseq value. To be safe against races, the actual diskseq value of
4117 the opened device (BLKGETDISKSEQ ioctl()) must still be compred with
4118 the one in the symlink path.
4120 * .link files gained support for setting MDI/MID-X on a link.
4122 * .link files gained support for [Match] Firmware= setting to match on
4123 the device firmware description string. By mistake, it was previously
4124 only supported in .network files.
4126 * .link files gained support for [Link] SR-IOVVirtualFunctions= setting
4127 and [SR-IOV] section to configure SR-IOV virtual functions.
4129 Changes in systemd-networkd:
4131 * The default scope for unicast routes configured through [Route]
4132 section is changed to "link", to make the behavior consistent with
4133 "ip route" command. The manual configuration of [Route] Scope= is
4136 * A new unit systemd-networkd-wait-online@<interface>.service has been
4137 added that can be used to wait for a specific network interface to be
4140 * systemd-networkd gained a new [Bridge] Isolated=true|false setting
4141 that configures the eponymous kernel attribute on the bridge.
4143 * .netdev files now can be used to create virtual WLAN devices, and
4144 configure various settings on them, via the [WLAN] section.
4146 * .link/.network files gained support for [Match] Kind= setting to match
4147 on device kind ("bond", "bridge", "gre", "tun", "veth", etc.)
4149 This value is also shown by 'networkctl status'.
4151 * The Local= setting in .netdev files for various virtual network
4152 devices gained support for specifying, in addition to the network
4153 address, the name of a local interface which must have the specified
4156 * systemd-networkd gained a new [Tunnel] External= setting in .netdev
4157 files, to configure tunnels in external mode (a.k.a. collect metadata
4160 * [Network] L2TP= setting was removed. Please use interface specifier in
4161 Local= setting in .netdev files of corresponding L2TP interface.
4163 * New [DHCPServer] BootServerName=, BootServerAddress=, and
4164 BootFilename= settings can be used to configure the server address,
4165 server name, and file name sent in the DHCP packet (e.g. to configure
4168 Changes in systemd-resolved:
4170 * systemd-resolved is started earlier (in sysinit.target), so it
4171 available earlier and will also be started in the initrd if installed
4174 Changes in disk encryption:
4176 * systemd-cryptenroll can now control whether to require the user to
4177 enter a PIN when using TPM-based unlocking of a volume via the new
4178 --tpm2-with-pin= option.
4180 Option tpm2-pin= can be used in /etc/crypttab.
4182 * When unlocking devices via TPM, TPM2 parameter encryption is now
4183 used, to ensure that communication between CPU and discrete TPM chips
4184 cannot be eavesdropped to acquire disk encryption keys.
4186 * A new switch --fido2-credential-algorithm= has been added to
4187 systemd-cryptenroll allowing selection of the credential algorithm to
4188 use when binding encryption to FIDO2 tokens.
4190 Changes in systemd-hostnamed:
4192 * HARDWARE_VENDOR= and HARDWARE_MODEL= can be set in /etc/machine-info
4193 to override the values gleaned from the hwdb.
4195 * A ID_CHASSIS property can be set in the hwdb (for the DMI device
4196 /sys/class/dmi/id) to override the chassis that is reported by
4199 * hostnamed's D-Bus interface gained a new method GetHardwareSerial()
4200 for reading the hardware serial number, as reportd by DMI. It also
4201 exposes a new method D-Bus property FirmwareVersion that encode the
4202 firmware version of the system.
4204 Changes in other components:
4206 * /etc/locale.conf is now populated through tmpfiles.d factory /etc/
4207 handling with the values that were configured during systemd build
4208 (if /etc/locale.conf has not been created through some other
4209 mechanism). This means that /etc/locale.conf should always have
4210 reasonable contents and we avoid a potential mismatch in defaults.
4212 * The userdbctl tool will now show UID range information as part of the
4213 list of known users.
4215 * A new build-time configuration setting default-user-shell= can be
4216 used to set the default shell for user records and nspawn shell
4217 invocations (instead of the default /bin/bash).
4219 * systemd-timesyncd now provides a D-Bus API for receiving NTP server
4220 information dynamically at runtime via IPC.
4222 * The systemd-creds tool gained a new "has-tpm2" verb, which reports
4223 whether a functioning TPM2 infrastructure is available, i.e. if
4224 firmware, kernel driver and systemd all have TPM2 support enabled and
4227 * The systemd-creds tool gained support for generating encrypted
4228 credentials that are using an empty encryption key. While this
4229 provides no integrity nor confidentiality it's useful to implement
4230 codeflows that work the same on TPM-ful and TPM2-less systems. The
4231 service manager will only accept credentials "encrypted" that way if
4232 a TPM2 device cannot be detected, to ensure that credentials
4233 "encrypted" like that cannot be used to trick TPM2 systems.
4235 * When deciding whether to colorize output, all systemd programs now
4236 also check $COLORTERM (in addition to $NO_COLOR, $SYSTEMD_COLORS, and
4239 * Meson's new install_tag feature is now in use for several components,
4240 allowing to build and install select binaries only: pam, nss, devel
4241 (pkg-config files), systemd-boot, libsystemd, libudev. Example:
4242 $ meson build systemd-boot
4243 $ meson install --tags systemd-boot --no-rebuild
4244 https://mesonbuild.com/Installing.html#installation-tags
4246 * A new build configuration option has been added, to allow selecting the
4247 default compression algorithm used by systemd-journald and systemd-coredump.
4248 This allows to build-in support for decompressing all supported formats,
4249 but choose a specific one for compression. E.g.:
4250 $ meson -Ddefault-compression=xz
4252 Experimental features:
4254 * sd-boot gained a new *experimental* setting "reboot-for-bitlocker" in
4255 loader.conf that implements booting Microsoft Windows from the
4256 sd-boot in a way that first reboots the system, to reset the TPM
4257 PCRs. This improves compatibility with BitLocker's TPM use, as the
4258 PCRs will only record the Windows boot process, and not sd-boot
4259 itself, thus retaining the PCR measurements not involving sd-boot.
4260 Note that this feature is experimental for now, and is likely going
4261 to be generalized and renamed in a future release, without retaining
4262 compatibility with the current implementation.
4264 * A new systemd-sysupdate component has been added that automatically
4265 discovers, downloads, and installs A/B-style updates for the host
4266 installation itself, or container images, portable service images,
4267 and other assets. See the new systemd-sysupdate man page for updates.
4269 Contributions from: 4piu, Adam Williamson, adrian5, Albert Brox,
4270 AlexCatze, Alex Henrie, Alfonso Sánchez-Beato, Alice S,
4271 Alvin Šipraga, amarjargal, Amarjargal, Andrea Pappacoda,
4272 Andreas Rammhold, Andy Chi, Anita Zhang, Antonio Alvarez Feijoo,
4273 Arfrever Frehtes Taifersar Arahesis, ash, Bastien Nocera, Be,
4274 bearhoney, Ben Efros, Benjamin Berg, Benjamin Franzke,
4275 Brett Holman, Christian Brauner, Clyde Byrd III, Curtis Klein,
4276 Daan De Meyer, Daniele Medri, Daniel Mack, Danilo Krummrich,
4277 David, David Bond, Davide Cavalca, David Tardon, davijosw,
4278 dependabot[bot], Donald Chan, Dorian Clay, Eduard Tolosa,
4279 Elias Probst, Eli Schwartz, Erik Sjölund, Evgeny Vereshchagin,
4280 Federico Ceratto, Franck Bui, Frantisek Sumsal, Gaël PORTAY,
4281 Georges Basile Stavracas Neto, Gibeom Gwon, Goffredo Baroncelli,
4282 Grigori Goronzy, Hans de Goede, Heiko Becker, Hugo Carvalho,
4283 Jakob Lell, James Hilliard, Jan Janssen, Jason A. Donenfeld,
4284 Joan Bruguera, Joerie de Gram, Josh Triplett, Julia Kartseva,
4285 Kazuo Moriwaka, Khem Raj, ksa678491784, Lance, Lan Tian,
4286 Laura Barcziova, Lennart Poettering, Leviticoh, licunlong,
4287 Lidong Zhong, lincoln auster, Lubomir Rintel, Luca Boccassi,
4288 Luca BRUNO, lucagoc, Ludwig Nussel, Marcel Hellwig, march1993,
4289 Marco Scardovi, Mario Limonciello, Mariusz Tkaczyk,
4290 Markus Weippert, Martin, Martin Liska, Martin Wilck, Matija Skala,
4291 Matthew Blythe, Matthias Lisin, Matthijs van Duin, Matt Walton,
4292 Max Gautier, Michael Biebl, Michael Olbrich, Michal Koutný,
4293 Michal Sekletár, Mike Gilbert, MkfsSion, Morten Linderud,
4294 Nick Rosbrook, Nikolai Grigoriev, Nikolai Kostrigin,
4295 Nishal Kulkarni, Noel Kuntze, Pablo Ceballos, Peter Hutterer,
4296 Peter Morrow, Pigmy-penguin, Piotr Drąg, prumian, Richard Neill,
4297 Rike-Benjamin Schuppner, rodin-ia, Romain Naour, Ruben Kerkhof,
4298 Ryan Hendrickson, Santa Wiryaman, Sebastian Pucilowski, Seth Falco,
4299 Simon Ellmann, Sonali Srivastava, Stefan Seering,
4300 Stephen Hemminger, tawefogo, techtino, Temuri Doghonadze,
4301 Thomas Batten, Thomas Haller, Thomas Weißschuh, Tobias Stoeckmann,
4302 Tomasz Pala, Tyson Whitehead, Vishal Chillara Srinivas,
4303 Vivien Didelot, w30023233, wangyuhang, Weblate, Xiaotian Wu,
4304 yangmingtai, YmrDtnJu, Yonathan Randolph, Yutsuten, Yu Watanabe,
4305 Zbigniew Jędrzejewski-Szmek, наб
4307 — Edinburgh, 2022-05-21
4311 * Support for encrypted and authenticated credentials has been added.
4312 This extends the credential logic introduced with v247 to support
4313 non-interactive symmetric encryption and authentication, based on a
4314 key that is stored on the /var/ file system or in the TPM2 chip (if
4315 available), or the combination of both (by default if a TPM2 chip
4316 exists the combination is used, otherwise the /var/ key only). The
4317 credentials are automatically decrypted at the moment a service is
4318 started, and are made accessible to the service itself in unencrypted
4319 form. A new tool 'systemd-creds' encrypts credentials for this
4320 purpose, and two new service file settings LoadCredentialEncrypted=
4321 and SetCredentialEncrypted= configure such credentials.
4323 This feature is useful to store sensitive material such as SSL
4324 certificates, passwords and similar securely at rest and only decrypt
4325 them when needed, and in a way that is tied to the local OS
4326 installation or hardware.
4328 * systemd-gpt-auto-generator can now automatically set up discoverable
4329 LUKS2 encrypted swap partitions.
4331 * The GPT Discoverable Partitions Specification has been substantially
4332 extended with support for root and /usr/ partitions for the majority
4333 of architectures systemd supports. This includes platforms that do
4334 not natively support UEFI, because even though GPT is specified under
4335 UEFI umbrella, it is useful on other systems too. Specifically,
4336 systemd-nspawn, systemd-sysext, systemd-gpt-auto-generator and
4337 Portable Services use the concept without requiring UEFI.
4339 * The GPT Discoverable Partitions Specifications has been extended with
4340 a new set of partitions that may carry PKCS#7 signatures for Verity
4341 partitions, encoded in a simple JSON format. This implements a simple
4342 mechanism for building disk images that are fully authenticated and
4343 can be tested against a set of cryptographic certificates. This is
4344 now implemented for the various systemd tools that can operate with
4345 disk images, such as systemd-nspawn, systemd-sysext, systemd-dissect,
4346 Portable services/RootImage=, systemd-tmpfiles, and systemd-sysusers.
4347 The PKCS#7 signatures are passed to the kernel (where they are
4348 checked against certificates from the kernel keyring), or can be
4349 verified against certificates provided in userspace (via a simple
4350 drop-in file mechanism).
4352 * systemd-dissect's inspection logic will now report for which uses a
4353 disk image is intended. Specifically, it will display whether an
4354 image is suitable for booting on UEFI or in a container (using
4355 systemd-nspawn's --image= switch), whether it can be used as portable
4356 service, or attached as system extension.
4358 * The system-extension.d/ drop-in files now support a new field
4359 SYSEXT_SCOPE= that may encode which purpose a system extension image
4360 is for: one of "initrd", "system" or "portable". This is useful to
4361 make images more self-descriptive, and to ensure system extensions
4362 cannot be attached in the wrong contexts.
4364 * The os-release file learnt a new PORTABLE_PREFIXES= field which may
4365 be used in portable service images to indicate which unit prefixes
4368 * The GPT image dissection logic in systemd-nspawn/systemd-dissect/…
4369 now is able to decode images for non-native architectures as well.
4370 This allows systemd-nspawn to boot images of non-native architectures
4371 if the corresponding user mode emulator is installed and
4372 systemd-binfmtd is running.
4374 * systemd-logind gained new settings HandlePowerKeyLongPress=,
4375 HandleRebootKeyLongPress=, HandleSuspendKeyLongPress= and
4376 HandleHibernateKeyLongPress= which may be used to configure actions
4377 when the relevant keys are pressed for more than 5s. This is useful
4378 on devices that only have hardware for a subset of these keys. By
4379 default, if the reboot key is pressed long the poweroff operation is
4380 now triggered, and when the suspend key is pressed long the hibernate
4381 operation is triggered. Long pressing the other two keys currently
4382 does not trigger any operation by default.
4384 * When showing unit status updates on the console during boot and
4385 shutdown, and a service is slow to start so that the cylon animation
4386 is shown, the most recent sd_notify() STATUS= text is now shown as
4387 well. Services may use this to make the boot/shutdown output easier
4388 to understand, and to indicate what precisely a service that is slow
4389 to start or stop is waiting for. In particular, the per-user service
4390 manager instance now reports what it is doing and which service it is
4391 waiting for this way to the system service manager.
4393 * The service manager will now re-execute on reception of the
4394 SIGRTMIN+25 signal. It previously already did that on SIGTERM — but
4395 only when running as PID 1. There was no signal to request this when
4396 running as per-user service manager, i.e. as any other PID than 1.
4397 SIGRTMIN+25 works for both system and user managers.
4399 * The hardware watchdog logic in PID 1 gained support for operating
4400 with the default timeout configured in the hardware, instead of
4401 insisting on re-configuring it. Set RuntimeWatchdogSec=default to
4402 request this behavior.
4404 * A new kernel command line option systemd.watchdog_sec= is now
4405 understood which may be used to override the hardware watchdog
4406 time-out for the boot.
4408 * A new setting DefaultOOMScoreAdjust= is now supported in
4409 /etc/systemd/system.conf and /etc/systemd/user.conf. It may be used
4410 to set the default process OOM score adjustment value for processes
4411 started by the service manager. For per-user service managers this
4412 now defaults to 100, but for per-system service managers is left as
4413 is. This means that by default now services forked off the user
4414 service manager are more likely to be killed by the OOM killer than
4415 system services or the managers themselves.
4417 * A new per-service setting RestrictFileSystems= as been added that
4418 restricts the file systems a service has access to by their type.
4419 This is based on the new BPF LSM of the Linux kernel. It provides an
4420 effective way to make certain API file systems unavailable to
4421 services (and thus minimizing attack surface). A new command
4422 "systemd-analyze filesystems" has been added that lists all known
4423 file system types (and how they are grouped together under useful
4426 * Services now support a new setting RestrictNetworkInterfaces= for
4427 restricting access to specific network interfaces.
4429 * Service unit files gained new settings StartupAllowedCPUs= and
4430 StartupAllowedMemoryNodes=. These are similar to their counterparts
4431 without the "Startup" prefix and apply during the boot process
4432 only. This is useful to improve boot-time behavior of the system and
4433 assign resources differently during boot than during regular
4434 runtime. This is similar to the preexisting StartupCPUWeight=
4437 * Related to this: the various StartupXYZ= settings
4438 (i.e. StartupCPUWeight=, StartupAllowedCPUs=, …) are now also applied
4439 during shutdown. The settings not prefixed with "Startup" hence apply
4440 during regular runtime, and those that are prefixed like that apply
4441 during boot and shutdown.
4443 * A new per-unit set of conditions/asserts
4444 [Condition|Assert][Memory|CPU|IO]Pressure= have been added to make a
4445 unit skip/fail activation if the system's (or a slice's) memory/cpu/io
4446 pressure is above the configured threshold, using the kernel PSI
4447 feature. For more details see systemd.unit(5) and
4448 https://docs.kernel.org/accounting/psi.html
4450 * The combination of ProcSubset=pid and ProtectKernelTunables=yes and/or
4451 ProtectKernelLogs=yes can now be used.
4453 * The default maximum numbers of inodes have been raised from 64k to 1M
4454 for /dev/, and from 400k to 1M for /tmp/.
4456 * The per-user service manager learnt support for communicating with
4457 systemd-oomd to acquire OOM kill information.
4459 * A new service setting ExecSearchPath= has been added that allows
4460 changing the search path for executables for services. It affects
4461 where we look for the binaries specified in ExecStart= and similar,
4462 and the specified directories are also added the $PATH environment
4463 variable passed to invoked processes.
4465 * A new setting RuntimeRandomizedExtraSec= has been added for service
4466 and scope units that allows extending the runtime time-out as
4467 configured by RuntimeMaxSec= with a randomized amount.
4469 * The syntax of the service unit settings RuntimeDirectory=,
4470 StateDirectory=, CacheDirectory=, LogsDirectory= has been extended:
4471 if the specified value is now suffixed with a colon, followed by
4472 another filename, the latter will be created as symbolic link to the
4473 specified directory. This allows creating these service directories
4474 together with alias symlinks to make them available under multiple
4477 * Service unit files gained two new settings TTYRows=/TTYColumns= for
4478 configuring rows/columns of the TTY device passed to
4479 stdin/stdout/stderr of the service. This is useful to propagate TTY
4480 dimensions to a virtual machine.
4482 * A new service unit file setting ExitType= has been added that
4483 specifies when to assume a service has exited. By default systemd
4484 only watches the main process of a service. By setting
4485 ExitType=cgroup it can be told to wait for the last process in a
4488 * Automount unit files gained a new setting ExtraOptions= that can be
4489 used to configure additional mount options to pass to the kernel when
4490 mounting the autofs instance.
4492 * "Urlification" (generation of ESC sequences that generate clickable
4493 hyperlinks in modern terminals) may now be turned off altogether
4496 * Path units gained new TriggerLimitBurst= and TriggerLimitIntervalSec=
4497 settings that default to 200 and 2 s respectively. The ratelimit
4498 ensures that a path unit cannot cause PID1 to busy-loop when it is
4499 trying to trigger a service that is skipped because of a Condition*=
4500 not being satisfied. This matches the configuration and behaviour of
4503 * The TPM2/FIDO2/PKCS11 support in systemd-cryptsetup is now also built
4504 as a plug-in for cryptsetup. This means the plain cryptsetup command
4505 may now be used to unlock volumes set up this way.
4507 * The TPM2 logic in cryptsetup will now automatically detect systems
4508 where the TPM2 chip advertises SHA256 PCR banks but the firmware only
4509 updates the SHA1 banks. In such a case PCR policies will be
4510 automatically bound to the latter, not the former. This makes the PCR
4511 policies reliable, but of course do not provide the same level of
4512 trust as SHA256 banks.
4514 * The TPM2 logic in systemd-cryptsetup/systemd-cryptsetup now supports
4515 RSA primary keys in addition to ECC, improving compatibility with
4516 TPM2 chips that do not support ECC. RSA keys are much slower to use
4517 than ECC, and hence are only used if ECC is not available.
4519 * /etc/crypttab gained support for a new token-timeout= setting for
4520 encrypted volumes that allows configuration of the maximum time to
4521 wait for PKCS#11/FIDO2 tokens to be plugged in. If the time elapses
4522 the logic will query the user for a regular passphrase/recovery key
4525 * Support for activating dm-integrity volumes at boot via a new file
4526 /etc/integritytab and the tool systemd-integritysetup have been
4527 added. This is similar to /etc/crypttab and /etc/veritytab, but deals
4528 with dm-integrity instead of dm-crypt/dm-verity.
4530 * The systemd-veritysetup-generator now understands a new usrhash=
4531 kernel command line option for specifying the Verity root hash for
4532 the partition backing the /usr/ file system. A matching set of
4533 systemd.verity_usr_* kernel command line options has been added as
4534 well. These all work similar to the corresponding options for the
4537 * The sd-device API gained a new API call sd_device_get_diskseq() to
4538 return the DISKSEQ property of a device structure. The "disk
4539 sequence" concept is a new feature recently introduced to the Linux
4540 kernel that allows detecting reuse cycles of block devices, i.e. can
4541 be used to recognize when loopback block devices are reused for a
4542 different purpose or CD-ROM drives get their media changed.
4544 * A new unit systemd-boot-update.service has been added. If enabled
4545 (the default) and the sd-boot loader is detected to be installed, it
4546 is automatically updated to the newest version when out of date. This
4547 is useful to ensure the boot loader remains up-to-date, and updates
4548 automatically propagate from the OS tree in /usr/.
4550 * sd-boot will now build with SBAT by default in order to facilitate
4551 working with recent versions of Shim that require it to be present.
4553 * sd-boot can now parse Microsoft Windows' Boot Configuration Data.
4554 This is used to robustly generate boot entry titles for Windows.
4556 * A new generic target unit factory-reset.target has been added. It is
4557 hooked into systemd-logind similar in fashion to
4558 reboot/poweroff/suspend/hibernate, and is supposed to be used to
4559 initiate a factory reset operation. What precisely this operation
4560 entails is up for the implementer to decide, the primary goal of the
4561 new unit is provide a framework where to plug in the implementation
4562 and how to trigger it.
4564 * A new meson build-time option 'clock-valid-range-usec-max' has been
4565 added which takes a time in µs and defaults to 15 years. If the RTC
4566 time is noticed to be more than the specified time ahead of the
4567 built-in epoch of systemd (which by default is the release timestamp
4568 of systemd) it is assumed that the RTC is not working correctly, and
4569 the RTC is reset to the epoch. (It already is reset to the epoch when
4570 noticed to be before it.) This should increase the chance that time
4571 doesn't accidentally jump too far ahead due to faulty hardware or
4574 * A new setting SaveIntervalSec= has been added to systemd-timesyncd,
4575 which may be used to automatically save the current system time to
4576 disk in regular intervals. This is useful to maintain a roughly
4577 monotonic clock even without RTC hardware and with some robustness
4578 against abnormal system shutdown.
4580 * systemd-analyze verify gained support for a pair of new --image= +
4581 --root= switches for verifying units below a specific root
4582 directory/image instead of on the host.
4584 * systemd-analyze verify gained support for verifying unit files under
4585 an explicitly specified unit name, independently of what the filename
4588 * systemd-analyze verify gained a new switch --recursive-errors= which
4589 controls whether to only fail on errors found in the specified units
4590 or recursively any dependent units.
4592 * systemd-analyze security now supports a new --offline mode for
4593 analyzing unit files stored on disk instead of loaded units. It may
4594 be combined with --root=/--image to analyze unit files under a root
4595 directory or disk image. It also learnt a new --threshold= parameter
4596 for specifying an exposure level threshold: if the exposure level
4597 exceeds the specified value the call will fail. It also gained a new
4598 --security-policy= switch for configuring security policies to
4599 enforce on the units. A policy is a JSON file that lists which tests
4600 shall be weighted how much to determine the overall exposure
4601 level. Altogether these new features are useful for fully automatic
4602 analysis and enforcement of security policies on unit files.
4604 * systemd-analyze security gain a new --json= switch for JSON output.
4606 * systemd-analyze learnt a new --quiet switch for reducing
4607 non-essential output. It's honored by the "dot", "syscall-filter",
4608 "filesystems" commands.
4610 * systemd-analyze security gained a --profile= option that can be used
4611 to take into account a portable profile when analyzing portable
4612 services, since a lot of the security-related settings are enabled
4615 * systemd-analyze learnt a new inspect-elf verb that parses ELF core
4616 files, binaries and executables and prints metadata information,
4617 including the build-id and other info described on:
4618 https://systemd.io/COREDUMP_PACKAGE_METADATA/
4620 * .network files gained a new UplinkInterface= in the [IPv6SendRA]
4621 section, for automatically propagating DNS settings from other
4624 * The static lease DHCP server logic in systemd-networkd may now serve
4625 IP addresses outside of the configured IP pool range for the server.
4627 * CAN support in systemd-networkd gained four new settings Loopback=,
4628 OneShot=, PresumeAck=, ClassicDataLengthCode= for tweaking CAN
4629 control modes. It gained a number of further settings for tweaking
4632 * The [CAN] section in .network file gained new TimeQuantaNSec=,
4633 PropagationSegment=, PhaseBufferSegment1=, PhaseBufferSegment2=,
4634 SyncJumpWidth=, DataTimeQuantaNSec=, DataPropagationSegment=,
4635 DataPhaseBufferSegment1=, DataPhaseBufferSegment2=, and
4636 DataSyncJumpWidth= settings to control bit-timing processed by the
4639 * DHCPv4 client support in systemd-networkd learnt a new Label= option
4640 for configuring the address label to apply to configure IPv4
4643 * The [IPv6AcceptRA] section of .network files gained support for a new
4644 UseMTU= setting that may be used to control whether to apply the
4645 announced MTU settings to the local interface.
4647 * The [DHCPv4] section in .network file gained a new Use6RD= boolean
4648 setting to control whether the DHCPv4 client request and process the
4651 * The [DHCPv6PrefixDelegation] section in .network file is renamed to
4652 [DHCPPrefixDelegation], as now the prefix delegation is also supported
4653 with DHCPv4 protocol by enabling the Use6RD= setting.
4655 * The [DHCPPrefixDelegation] section in .network file gained a new
4656 setting UplinkInterface= to specify the upstream interface.
4658 * The [DHCPv6] section in .network file gained a new setting
4659 UseDelegatedPrefix= to control whether the delegated prefixes will be
4660 propagated to the downstream interfaces.
4662 * The [IPv6AcceptRA] section of .network files now understands two new
4663 settings UseGateway=/UseRoutePrefix= for explicitly configuring
4664 whether to use the relevant fields from the IPv6 Router Advertisement
4667 * The ForceDHCPv6PDOtherInformation= setting in the [DHCPv6] section
4668 has been removed. Please use the WithoutRA= and UseDelegatedPrefix=
4669 settings in the [DHCPv6] section and the DHCPv6Client= setting in the
4670 [IPv6AcceptRA] section to control when the DHCPv6 client is started
4671 and how the delegated prefixes are handled by the DHCPv6 client.
4673 * The IPv6Token= section in the [Network] section is deprecated, and
4674 the [IPv6AcceptRA] section gained the Token= setting for its
4675 replacement. The [IPv6Prefix] section also gained the Token= setting.
4676 The Token= setting gained 'eui64' mode to explicitly configure an
4677 address with the EUI64 algorithm based on the interface MAC address.
4678 The 'prefixstable' mode can now optionally take a secret key. The
4679 Token= setting in the [DHCPPrefixDelegation] section now supports all
4680 algorithms supported by the same settings in the other sections.
4682 * The [RoutingPolicyRule] section of .network file gained a new
4683 SuppressInterfaceGroup= setting.
4685 * The IgnoreCarrierLoss= setting in the [Network] section of .network
4686 files now allows a duration to be specified, controlling how long to
4687 wait before reacting to carrier loss.
4689 * The [DHCPServer] section of .network file gained a new Router=
4690 setting to specify the router address.
4692 * The [CAKE] section of .network files gained various new settings
4693 AutoRateIngress=, CompensationMode=, FlowIsolationMode=, NAT=,
4694 MPUBytes=, PriorityQueueingPreset=, FirewallMark=, Wash=, SplitGSO=,
4695 and UseRawPacketSize= for configuring CAKE.
4697 * systemd-networkd now ships with new default .network files:
4698 80-container-vb.network which matches host-side network bridge device
4699 created by systemd-nspawn's --network-bridge or --network-zone
4700 switch, and 80-6rd-tunnel.network which matches automatically created
4701 sit tunnel with 6rd prefix when the DHCP 6RD option is received.
4703 * systemd-networkd's handling of Endpoint= resolution for WireGuard
4704 interfaces has been improved.
4706 * systemd-networkd will now automatically configure routes to addresses
4707 specified in AllowedIPs=. This feature can be controlled via
4708 RouteTable= and RouteMetric= settings in [WireGuard] or
4709 [WireGuardPeer] sections.
4711 * systemd-networkd will now once again automatically generate persistent
4712 MAC addresses for batadv and bridge interfaces. Users can disable this
4713 by using MACAddress=none in .netdev files.
4715 * systemd-networkd and systemd-udevd now support IP over InfiniBand
4716 interfaces. The Kind= setting in .netdev file accepts "ipoib". And
4717 systemd.netdev files gained the [IPoIB] section.
4719 * systemd-networkd and systemd-udevd now support net.ifname_policy=
4720 option on the kernel command-line. This is implemented through the
4721 systemd-network-generator service that automatically generates
4722 appropriate .link, .network, and .netdev files.
4724 * The various systemd-udevd "ethtool" buffer settings now understand
4725 the special value "max" to configure the buffers to the maximum the
4728 * systemd-udevd's .link files may now configure a large variety of
4729 NIC coalescing settings, plus more hardware offload settings.
4731 * .link files gained a new WakeOnLanPassword= setting in the [Link]
4732 section that allows to specify a WoL "SecureOn" password on hardware
4735 * systemd-nspawn's --setenv= switch now supports an additional syntax:
4736 if only a variable name is specified (i.e. without being suffixed by
4737 a '=' character and a value) the current value of the environment
4738 variable is propagated to the container. e.g. --setenv=FOO will
4739 lookup the current value of $FOO in the environment, and pass it down
4740 to the container. Similar behavior has been added to homectl's,
4741 machinectl's and systemd-run's --setenv= switch.
4743 * systemd-nspawn gained a new switch --suppress-sync= which may be used
4744 to optionally suppress the effect of the sync()/fsync()/fdatasync()
4745 system calls for the container payload. This is useful for build
4746 system environments where safety against abnormal system shutdown is
4747 not essential as all build artifacts can be regenerated any time, but
4748 the performance win is beneficial.
4750 * systemd-nspawn will now raise the RLIMIT_NOFILE hard limit to the
4751 same value that PID 1 uses for most forked off processes.
4753 * systemd-nspawn's --bind=/--bind-ro= switches now optionally take
4754 uidmap/nouidmap options as last parameter. If "uidmap" is used the
4755 bind mounts are created with UID mapping taking place that ensures
4756 the host's file ownerships are mapped 1:1 to container file
4757 ownerships, even if user namespacing is used. This way
4758 files/directories bound into containers will no longer show up as
4759 owned by the nobody user as they typically did if no special care was
4760 taken to shift them manually.
4762 * When discovering Windows installations sd-boot will now attempt to
4763 show the Windows version.
4765 * The color scheme to use in sd-boot may now be configured at
4768 * sd-boot gained the ability to change screen resolution during
4769 boot-time, by hitting the "r" key. This will cycle through available
4770 resolutions and save the last selection.
4772 * sd-boot learnt a new hotkey "f". When pressed the system will enter
4773 firmware setup. This is useful in environments where it is difficult
4774 to hit the right keys early enough to enter the firmware, and works
4775 on any firmware regardless which key it natively uses.
4777 * sd-boot gained support for automatically booting into the menu item
4778 selected on the last boot (using the "@saved" identifier for menu
4781 * sd-boot gained support for automatically loading all EFI drivers
4782 placed in the /EFI/systemd/drivers/ subdirectory of the EFI System
4783 Partition (ESP). These drivers are loaded before the menu entries are
4784 loaded. This is useful e.g. to load additional file system drivers
4785 for the XBOOTLDR partition.
4787 * systemd-boot will now paint the input cursor on its own instead of
4788 relying on the firmware to do so, increasing compatibility with broken
4789 firmware that doesn't make the cursor reasonably visible.
4791 * sd-boot now embeds a .osrel PE section like we expect from Boot
4792 Loader Specification Type #2 Unified Kernels. This means sd-boot
4793 itself may be used in place of a Type #2 Unified Kernel. This is
4794 useful for debugging purposes as it allows chain-loading one a
4795 (development) sd-boot instance from another.
4797 * sd-boot now supports a new "devicetree" field in Boot Loader
4798 Specification Type #1 entries: if configured the specified device
4799 tree file is installed before the kernel is invoked. This is useful
4800 for installing/applying new devicetree files without updating the
4803 * Similarly, sd-stub now can read devicetree data from a PE section
4804 ".dtb" and apply it before invoking the kernel.
4806 * sd-stub (the EFI stub that can be glued in front of a Linux kernel)
4807 gained the ability to pick up credentials and sysext files, wrap them
4808 in a cpio archive, and pass as an additional initrd to the invoked
4809 Linux kernel, in effect placing those files in the /.extra/ directory
4810 of the initrd environment. This is useful to implement trusted initrd
4811 environments which are fully authenticated but still can be extended
4812 (via sysexts) and parameterized (via encrypted/authenticated
4813 credentials, see above).
4815 Credentials can be located next to the kernel image file (credentials
4816 specific to a single boot entry), or in one of the shared directories
4817 (credentials applicable to multiple boot entries).
4819 * sd-stub now comes with a full man page, that explains its feature set
4820 and how to combine a kernel image, an initrd and the stub to build a
4821 complete EFI unified kernel image, implementing Boot Loader
4822 Specification Type #2.
4824 * sd-stub may now provide the initrd to the executed kernel via the
4825 LINUX_EFI_INITRD_MEDIA_GUID EFI protocol, adding compatibility for
4826 non-x86 architectures.
4828 * bootctl learnt new set-timeout and set-timeout-oneshot commands that
4829 may be used to set the boot menu time-out of the boot loader (for all
4830 or just the subsequent boot).
4832 * bootctl and kernel-install will now read variables
4833 KERNEL_INSTALL_LAYOUT= from /etc/machine-info and layout= from
4834 /etc/kernel/install.conf. When set, it specifies the layout to use
4835 for installation directories on the boot partition, so that tools
4836 don't need to guess it based on the already-existing directories. The
4837 only value that is defined natively is "bls", corresponding to the
4839 https://systemd.io/BOOT_LOADER_SPECIFICATION/. Plugins for
4840 kernel-install that implement a different layout can declare other
4841 values for this variable.
4843 'bootctl install' will now write KERNEL_INSTALL_LAYOUT=bls, on the
4844 assumption that if the user installed sd-boot to the ESP, they intend
4845 to use the entry layout understood by sd-boot. It'll also write
4846 KERNEL_INSTALL_MACHINE_ID= if it creates any directories using the ID
4847 (and it wasn't specified in the config file yet). Similarly,
4848 kernel-install will now write KERNEL_INSTALL_MACHINE_ID= (if it
4849 wasn't specified in the config file yet). Effectively, those changes
4850 mean that the machine-id used for boot loader entry installation is
4851 "frozen" upon first use and becomes independent of the actual
4854 Configuring KERNEL_INSTALL_MACHINE_ID fixes the following problem:
4855 images created for distribution ("golden images") are built with no
4856 machine-id, so that a unique machine-id can be created on the first
4857 boot. But those images may contain boot loader entries with the
4858 machine-id used during build included in paths. Using a "frozen"
4859 value allows unambiguously identifying entries that match the
4860 specific installation, while still permitting parallel installations
4863 Configuring KERNEL_INSTALL_LAYOUT obviates the need for
4864 kernel-install to guess the installation layout. This fixes the
4865 problem where a (possibly empty) directory in the boot partition is
4866 created from a different layout causing kernel-install plugins to
4867 assume the wrong layout. A particular example of how this may happen
4868 is the grub2 package in Fedora which includes directories under /boot
4869 directly in its file list. Various other packages pull in grub2 as a
4870 dependency, so it may be installed even if unused, breaking
4871 installations that use the bls layout.
4873 * bootctl and systemd-bless-boot can now be linked statically.
4875 * systemd-sysext now optionally doesn't insist on extension-release.d/
4876 files being placed in the image under the image's file name. If the
4877 file system xattr user.extension-release.strict is set on the
4878 extension release file, it is accepted regardless of its name. This
4879 relaxes security restrictions a bit, as system extension may be
4880 attached under a wrong name this way.
4882 * udevadm's test-builtin command learnt a new --action= switch for
4883 testing the built-in with the specified action (in place of the
4886 * udevadm info gained new switches --property=/--value for showing only
4887 specific udev properties/values instead of all.
4889 * A new hwdb database has been added that contains matches for various
4890 types of signal analyzers (protocol analyzers, logic analyzers,
4891 oscilloscopes, multimeters, bench power supplies, etc.) that should
4892 be accessible to regular users.
4894 * A new hwdb database entry has been added that carries information
4895 about types of cameras (regular or infrared), and in which direction
4896 they point (front or back).
4898 * A new rule to allow console users access to rfkill by default has been
4901 * Device nodes for the Software Guard eXtension enclaves (sgx_vepc) are
4902 now also owned by the system group "sgx".
4904 * A new build-time meson option "extra-net-naming-schemes=" has been
4905 added to define additional naming schemes for udev's network
4906 interface naming logic. This is useful for enterprise distributions
4907 and similar which want to pin the schemes of certain distribution
4908 releases under a specific name and previously had to patch the
4909 sources to introduce new named schemes.
4911 * The predictable naming logic for network interfaces has been extended
4912 to generate stable names from Xen netfront device information.
4914 * hostnamed's chassis property can now be sourced from chassis-type
4915 field encoded in devicetree (in addition to the existing DMI
4918 * systemd-cgls now optionally displays cgroup IDs and extended
4919 attributes for each cgroup. (Controllable via the new --xattr= +
4920 --cgroup-id= switches.)
4922 * coredumpctl gained a new --all switch for operating on all
4923 Journal files instead of just the local ones.
4925 * systemd-coredump will now use libdw/libelf via dlopen() rather than
4926 directly linking, allowing users to easily opt-out of backtrace/metadata
4927 analysis of core files, and reduce image sizes when this is not needed.
4929 * systemd-coredump will now analyze core files with libdw/libelf in a
4930 forked, sandboxed process.
4932 * systemd-homed will now try to unmount an activate home area in
4933 regular intervals once the user logged out fully. Previously this was
4934 attempted exactly once but if the home directory was busy for some
4935 reason it was not tried again.
4937 * systemd-homed's LUKS2 home area backend will now create a BSD file
4938 system lock on the image file while the home area is active
4939 (i.e. mounted). If a home area is found to be locked, logins are
4940 politely refused. This should improve behavior when using home areas
4941 images that are accessible via the network from multiple clients, and
4942 reduce the chance of accidental file system corruption in that case.
4944 * Optionally, systemd-homed will now drop the kernel buffer cache once
4945 a user has fully logged out, configurable via the new --drop-caches=
4948 * systemd-homed now makes use of UID mapped mounts for the home areas.
4949 If the kernel and used file system support it, files are now
4950 internally owned by the "nobody" user (i.e. the user typically used
4951 for indicating "this ownership is not mapped"), and dynamically
4952 mapped to the UID used locally on the system via the UID mapping
4953 mount logic of recent kernels. This makes migrating home areas
4954 between different systems cheaper because recursively chown()ing file
4955 system trees is no longer necessary.
4957 * systemd-homed's CIFS backend now optionally supports CIFS service
4958 names with a directory suffix, in order to place home directories in
4959 a subdirectory of a CIFS share, instead of the top-level directory.
4961 * systemd-homed's CIFS backend gained support for specifying additional
4962 mount options in the JSON user record (cifsExtraMountOptions field,
4963 and --cifs-extra-mount-options= homectl switch). This is for example
4964 useful for configuring mount options such as "noserverino" that some
4965 SMB3 services require (use that to run a homed home directory from a
4966 FritzBox SMB3 share this way).
4968 * systemd-homed will now default to btrfs' zstd compression for home
4969 areas. This is inspired by Fedora's recent decision to switch to zstd
4972 * Additional mount options to use when mounting the file system of
4973 LUKS2 volumes in systemd-homed has been added. Via the
4974 $SYSTEMD_HOME_MOUNT_OPTIONS_BTRFS, $SYSTEMD_HOME_MOUNT_OPTIONS_EXT4,
4975 $SYSTEMD_HOME_MOUNT_OPTIONS_XFS environment variables to
4976 systemd-homed or via the luksExtraMountOptions user record JSON
4977 property. (Exposed via homectl --luks-extra-mount-options)
4979 * homectl's resize command now takes the special size specifications
4980 "min" and "max" to shrink/grow the home area to the minimum/maximum
4981 size possible, taking disk usage/space constraints and file system
4982 limitations into account. Resizing is now generally graceful: the
4983 logic will try to get as close to the specified size as possible, but
4984 not consider it a failure if the request couldn't be fulfilled
4987 * systemd-homed gained the ability to automatically shrink home areas
4988 on logout to their minimal size and grow them again on next
4989 login. This ensures that while inactive, a home area only takes up
4990 the minimal space necessary, but once activated, it provides
4991 sufficient space for the user's needs. This behavior is only
4992 supported if btrfs is used as file system inside the home area
4993 (because only for btrfs online growing/shrinking is implemented in
4994 the kernel). This behavior is now enabled by default, but may be
4995 controlled via the new --auto-resize-mode= setting of homectl.
4997 * systemd-homed gained support for automatically re-balancing free disk
4998 space among active home areas, in case the LUKS2 backends are used,
4999 and no explicit disk size was requested. This way disk space is
5000 automatically managed and home areas resized in regular intervals and
5001 manual resizing when disk space becomes scarce should not be
5002 necessary anymore. This behavior is only supported if btrfs is used
5003 within the home areas (as only then online shrinking and growing is
5004 supported), and may be configured via the new rebalanceWeight JSON
5005 user record field (as exposed via the new --rebalance-weight= homectl
5006 setting). Re-balancing is mostly automatic, but can also be requested
5007 explicitly via "homectl rebalance", which is synchronous, and thus
5008 may be used to wait until the rebalance run is complete.
5010 * userdbctl gained a --json= switch for configured the JSON formatting
5011 to use when outputting user or group records.
5013 * userdbctl gained a new --multiplexer= switch for explicitly
5014 configuring whether to use the systemd-userdbd server side user
5015 record resolution logic.
5017 * userdbctl's ssh-authorized-keys command learnt a new --chain switch,
5018 for chaining up another command to execute after completing the
5019 look-up. Since the OpenSSH's AuthorizedKeysCommand only allows
5020 configuration of a single command to invoke, this maybe used to
5021 invoke multiple: first userdbctl's own implementation, and then any
5022 other also configured in the command line.
5024 * The sd-event API gained a new function sd_event_add_inotify_fd() that
5025 is similar to sd_event_add_inotify() but accepts a file descriptor
5026 instead of a path in the file system for referencing the inode to
5029 * The sd-event API gained a new function
5030 sd_event_source_set_ratelimit_expire_callback() that may be used to
5031 define a callback function that is called whenever an event source
5032 leaves the rate limiting phase.
5034 * New documentation has been added explaining which steps are necessary
5035 to port systemd to a new architecture:
5037 https://systemd.io/PORTING_TO_NEW_ARCHITECTURES
5039 * The x-systemd.makefs option in /etc/fstab now explicitly supports
5040 ext2, ext3, and f2fs file systems.
5042 * Mount units and units generated from /etc/fstab entries with 'noauto'
5043 are now ordered the same as other units. Effectively, they will be
5044 started earlier (if something actually pulled them in) and stopped
5045 later, similarly to normal mount units that are part of
5046 fs-local.target. This change should be invisible to users, but
5047 should prevent those units from being stopped too early during
5050 * The systemd-getty-generator now honors a new kernel command line
5051 argument systemd.getty_auto= and a new environment variable
5052 $SYSTEMD_GETTY_AUTO that allows turning it off at boot. This is for
5053 example useful to turn off gettys inside of containers or similar
5056 * systemd-resolved now listens on a second DNS stub address: 127.0.0.54
5057 (in addition to 127.0.0.53, as before). If DNS requests are sent to
5058 this address they are propagated in "bypass" mode only, i.e. are
5059 almost not processed locally, but mostly forwarded as-is to the
5060 current upstream DNS servers. This provides a stable DNS server
5061 address that proxies all requests dynamically to the right upstream
5062 DNS servers even if these dynamically change. This stub does not do
5063 mDNS/LLMNR resolution. However, it will translate look-ups to
5064 DNS-over-TLS if necessary. This new stub is particularly useful in
5065 container/VM environments, or for tethering setups: use DNAT to
5066 redirect traffic to any IP address to this stub.
5068 * systemd-importd now honors new environment variables
5069 $SYSTEMD_IMPORT_BTRFS_SUBVOL, $SYSTEMD_IMPORT_BTRFS_QUOTA,
5070 $SYSTEMD_IMPORT_SYNC, which may be used disable btrfs subvolume
5071 generation, btrfs quota setup and disk synchronization.
5073 * systemd-importd and systemd-resolved can now be optionally built with
5074 OpenSSL instead of libgcrypt.
5076 * systemd-repart no longer requires OpenSSL.
5078 * systemd-sysusers will no longer create the redundant 'nobody' group
5079 by default, as the 'nobody' user is already created with an
5080 appropriate primary group.
5082 * If a unit uses RuntimeMaxSec, systemctl show will now display it.
5084 * systemctl show-environment gained support for --output=json.
5086 * pam_systemd will now first try to use the X11 abstract socket, and
5087 fallback to the socket file in /tmp/.X11-unix/ only if that does not
5090 * systemd-journald will no longer go back to volatile storage
5091 regardless of configuration when its unit is restarted.
5093 * Initial support for the LoongArch architecture has been added (system
5094 call lists, GPT partition table UUIDs, etc).
5096 * systemd-journald's own logging messages are now also logged to the
5097 journal itself when systemd-journald logs to /dev/kmsg.
5099 * systemd-journald now re-enables COW for archived journal files on
5100 filesystems that support COW. One benefit of this change is that
5101 archived journal files will now get compressed on btrfs filesystems
5102 that have compression enabled.
5104 * systemd-journald now deduplicates fields in a single log message
5105 before adding it to the journal. In archived journal files, it will
5106 also punch holes for unused parts and truncate the file as
5107 appropriate, leading to reductions in disk usage.
5109 * journalctl --verify was extended with more informative error
5112 * More of sd-journal's functions are now resistant against journal file
5115 * The shutdown command learnt a new option --show, to display the
5118 * A LICENSES/ directory is now included in the git tree. It contains a
5119 README.md file that explains the licenses used by source files in
5120 this repository. It also contains the text of all applicable
5121 licenses as they appear on spdx.org.
5123 Contributions from: Aakash Singh, acsfer, Adolfo Jayme Barrientos,
5124 Adrian Vovk, Albert Brox, Alberto Mardegan, Alexander Kanavin,
5125 alexlzhu, Alfonso Sánchez-Beato, Alvin Šipraga, Alyssa Ross,
5126 Amir Omidi, Anatol Pomozov, Andika Triwidada, Andreas Rammhold,
5127 Andreas Valder, Andrej Lajovic, Andrew Soutar, Andrew Stone, Andy Chi,
5128 Anita Zhang, Anssi Hannula, Antonio Alvarez Feijoo,
5129 Antony Deepak Thomas, Arnaud Ferraris, Arvid E. Picciani,
5130 Bastien Nocera, Benjamin Berg, Benjamin Herrenschmidt, Ben Stockett,
5131 Bogdan Seniuc, Boqun Feng, Carl Lei, chlorophyll-zz, Chris Packham,
5132 Christian Brauner, Christian Göttsche, Christian Wehrli,
5133 Christoph Anton Mitterer, Cristian Rodríguez, Daan De Meyer,
5134 Daniel Maixner, Dann Frazier, Dan Streetman, Davide Cavalca,
5135 David Seifert, David Tardon, dependabot[bot], Dimitri John Ledkov,
5136 Dimitri Papadopoulos, Dimitry Ishenko, Dmitry Khlebnikov,
5137 Dominique Martinet, duament, Egor, Egor Ignatov, Emil Renner Berthing,
5138 Emily Gonyer, Ettore Atalan, Evgeny Vereshchagin, Florian Klink,
5139 Franck Bui, Frantisek Sumsal, Geass-LL, Gibeom Gwon, GnunuX,
5140 Gogo Gogsi, gregzuro, Greg Zuro, Gustavo Costa, Hans de Goede,
5141 Hela Basa, Henri Chain, hikigaya58, Hugo Carvalho,
5142 Hugo Osvaldo Barrera, Iago Lopez Galeiras, Iago López Galeiras,
5143 I-dont-need-name, igo95862, Jack Dähn, James Hilliard, Jan Janssen,
5144 Jan Kuparinen, Jan Macku, Jan Palus, Jarkko Sakkinen, Jayce Fayne,
5145 jiangchuangang, jlempen, John Lindgren, Jonas Dreßler, Jonas Jelten,
5146 Jonas Witschel, Joris Hartog, José Expósito, Julia Kartseva,
5147 Kai-Heng Feng, Kai Wohlfahrt, Kay Siver Bø, KennthStailey,
5148 Kevin Kuehler, Kevin Orr, Khem Raj, Kristian Klausen, Kyle Laker,
5149 lainahai, LaserEyess, Lennart Poettering, Lia Lenckowski, longpanda,
5150 Luca Boccassi, Luca BRUNO, Ludwig Nussel, Lukas Senionis,
5151 Maanya Goenka, Maciek Borzecki, Marcel Menzel, Marco Scardovi,
5152 Marcus Harrison, Mark Boudreau, Matthijs van Duin, Mauricio Vásquez,
5153 Maxime de Roucy, Max Resch, MertsA, Michael Biebl, Michael Catanzaro,
5154 Michal Koutný, Michal Sekletár, Miika Karanki, Mike Gilbert,
5155 Milo Turner, ml, monosans, Nacho Barrientos, nassir90, Nishal Kulkarni,
5156 nl6720, Ondrej Kozina, Paulo Neves, Pavel Březina, pedro martelletto,
5157 Peter Hutterer, Peter Morrow, Piotr Drąg, Rasmus Villemoes, ratijas,
5158 Raul Tambre, rene, Riccardo Schirone, Robert-L-Turner, Robert Scheck,
5159 Ross Jennings, saikat0511, Scott Lamb, Scott Worley,
5160 Sergei Trofimovich, Sho Iizuka, Slava Bacherikov, Slimane Selyan Amiri,
5161 StefanBruens, Steven Siloti, svonohr, Taiki Sugawara, Takashi Sakamoto,
5162 Takuro Onoue, Thomas Blume, Thomas Haller, Thomas Mühlbacher,
5163 Tianlu Shao, Toke Høiland-Jørgensen, Tom Yan, Tony Asleson,
5164 Topi Miettinen, Ulrich Ölmann, Urs Ritzmann, Vincent Bernat,
5165 Vito Caputo, Vladimir Panteleev, WANG Xuerui, Wind/owZ, Wu Xiaotian,
5166 xdavidwu, Xiaotian Wu, xujing, yangmingtai, Yao Wei, Yao Wei (魏銘廷),
5167 Yegor Alexeyev, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
5168 Дамјан Георгиевски, наб
5170 — Warsaw, 2021-12-23
5174 * When operating on disk images via the --image= switch of various
5175 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
5176 no 'root=' parameter on the kernel command line, and multiple
5177 suitable root or /usr/ partitions exist in the image, then a simple
5178 comparison inspired by strverscmp() is done on the GPT partition
5179 label, and the newest partition is picked. This permits a simple and
5180 generic whole-file-system A/B update logic where new operating system
5181 versions are dropped into partitions whose label is then updated with
5182 a matching version identifier.
5184 * systemd-sysusers now supports querying the passwords to set for the
5185 users it creates via the "credentials" logic introduced in v247: the
5186 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
5187 credentials are consulted for the password to use (either in UNIX
5188 hashed form, or literally). By default these credentials are inherited
5189 down from PID1 (which in turn imports it from a container manager if
5190 there is one). This permits easy configuration of user passwords
5191 during first boot. Example:
5193 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
5195 Note that systemd-sysusers operates in purely additive mode: it
5196 executes no operation if the declared users already exist, and hence
5197 doesn't set any passwords as effect of the command line above if the
5198 specified root user exists already in the image. (Note that
5199 --volatile=yes ensures it doesn't, though.)
5201 * systemd-firstboot now also supports querying various system
5202 parameters via the credential subsystems. Thus, as above this may be
5203 used to initialize important system parameters on first boot of
5204 previously unprovisioned images (i.e. images with a mostly empty
5207 * PID 1 may now show both the unit name and the unit description
5208 strings in its status output during boot. This may be configured with
5209 StatusUnitFormat=combined in system.conf or
5210 systemd.status-unit-format=combined on the kernel command line.
5212 * The systemd-machine-id-setup tool now supports a --image= switch for
5213 provisioning a machine ID file into an OS disk image, similar to how
5214 --root= operates on an OS file tree. This matches the existing switch
5215 of the same name for systemd-tmpfiles, systemd-firstboot, and
5216 systemd-sysusers tools.
5218 * Similarly, systemd-repart gained support for the --image= switch too.
5219 In combination with the existing --size= option, this makes the tool
5220 particularly useful for easily growing disk images in a single
5221 invocation, following the declarative rules included in the image
5224 * systemd-repart's partition configuration files gained support for a
5225 new switch MakeDirectories= which may be used to create arbitrary
5226 directories inside file systems that are created, before registering
5227 them in the partition table. This is useful in particular for root
5228 partitions to create mount point directories for other partitions
5229 included in the image. For example, a disk image that contains a
5230 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
5231 create /home/ and /var/ as empty directories in the root file system
5232 on its creation, so that the resulting image can be mounted
5233 immediately, even in read-only mode.
5235 * systemd-repart's CopyBlocks= setting gained support for the special
5236 value "auto". If used, a suitable matching partition on the booted OS
5237 is found as source to copy blocks from. This is useful when
5238 implementing replicating installers, that are booted from one medium
5239 and then stream their own root partition onto the target medium.
5241 * systemd-repart's partition configuration files gained support for a
5242 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
5243 GPT partition flags for the created partitions: this is useful for
5244 marking newly created partitions as read-only, or as not being
5245 subject for automatic mounting from creation on.
5247 * The /etc/os-release file has been extended with two new (optional)
5248 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
5249 information for OS images that are updated comprehensively and
5250 atomically as one image. Two new specifiers %M, %A now resolve to
5251 these two fields in the various configuration options that resolve
5254 * portablectl gained a new switch --extension= for enabling portable
5255 service images with extensions that follow the extension image
5256 concept introduced with v248, and thus allows layering multiple
5257 images when setting up the root filesystem of the service.
5259 * systemd-coredump will now extract ELF build-id information from
5260 processes dumping core and include it in the coredump report.
5261 Moreover, it will look for ELF .note.package sections with
5262 distribution packaging meta-information about the crashing process.
5263 This is useful to directly embed the rpm or deb (or any other)
5264 package name and version in ELF files, making it easy to match
5265 coredump reports with the specific package for which the software was
5266 compiled. This is particularly useful on environments with ELF files
5267 from multiple vendors, different distributions and versions, as is
5268 common today in our containerized and sand-boxed world. For further
5271 https://systemd.io/COREDUMP_PACKAGE_METADATA
5273 * A new udev hardware database has been added for FireWire devices
5276 * The "net_id" built-in of udev has been updated with three
5277 backwards-incompatible changes:
5279 - PCI hotplug slot names on s390 systems are now parsed as
5280 hexadecimal numbers. They were incorrectly parsed as decimal
5281 previously, or ignored if the name was not a valid decimal
5284 - PCI onboard indices up to 65535 are allowed. Previously, numbers
5285 above 16383 were rejected. This primarily impacts s390 systems,
5286 where values up to 65535 are used.
5288 - Invalid characters in interface names are replaced with "_".
5290 The new version of the net naming scheme is "v249". The previous
5291 scheme can be selected via the "net.naming_scheme=v247" kernel
5292 command line parameter.
5294 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
5295 NULL bus object, for which they will return false. Or in other words,
5296 an unallocated bus connection is neither ready nor open.
5298 * The sd-device API acquired a new API function
5299 sd_device_get_usec_initialized() that returns the monotonic time when
5300 the udev device first appeared in the database.
5302 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
5303 sd_device_get_trigger_uuid(). The former is similar to
5304 sd_device_trigger() but returns a randomly generated UUID that is
5305 associated with the synthetic uevent generated by the call. This UUID
5306 may be read from the sd_device object a monitor eventually receives,
5307 via the sd_device_get_trigger_uuid(). This interface requires kernel
5308 4.13 or above to work, and allows tracking a synthetic uevent through
5309 the entire device management stack. The "udevadm trigger --settle"
5310 logic has been updated to make use of this concept if available to
5311 wait precisely for the uevents it generates. "udevadm trigger" also
5312 gained a new parameter --uuid that prints the UUID for each generated
5315 * sd-device also gained new APIs sd_device_new_from_ifname() and
5316 sd_device_new_from_ifindex() for allocating an sd-device object for
5317 the specified network interface. The former accepts an interface name
5318 (either a primary or an alternative name), the latter an interface
5321 * The native Journal protocol has been documented. Clients may talk
5322 this as alternative to the classic BSD syslog protocol for locally
5323 delivering log records to the Journal. The protocol has been stable
5324 for a long time and in fact been implemented already in a variety
5325 of alternative client libraries. This documentation makes the support
5328 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
5330 * A new BPFProgram= setting has been added to service files. It may be
5331 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
5332 file, or a bind mount or symlink to one. This may be used to upload
5333 and manage BPF programs externally and then hook arbitrary systemd
5336 * The "home.arpa" domain that has been officially declared as the
5337 choice for domain for local home networks per RFC 8375 has been added
5338 to the default NTA list of resolved, since DNSSEC is generally not
5339 available on private domains.
5341 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
5343 * A new ManageForeignRoutingPolicyRules= setting has been added to
5344 .network files which may be used to exclude foreign-created routing
5345 policy rules from systemd-networkd management.
5347 * systemd-network-wait-online gained two new switches -4 and -6 that
5348 may be used to tweak whether to wait for only IPv4 or only IPv6
5351 * .network files gained a new RequiredFamilyForOnline= setting to
5352 fine-tune whether to require an IPv4 or IPv6 address in order to
5353 consider an interface "online".
5355 * networkctl will now show an over-all "online" state in the per-link
5358 * In .network files a new OutgoingInterface= setting has been added to
5359 specify the output interface in bridge FDB setups.
5361 * In .network files the Multipath group ID may now be configured for
5362 [NextHop] entries, via the new Group= setting.
5364 * The DHCP server logic configured in .network files gained a new
5365 setting RelayTarget= that turns the server into a DHCP server relay.
5366 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
5367 to further tweak the DHCP relay behaviour.
5369 * The DHCP server logic also gained a new ServerAddress= setting in
5370 .network files that explicitly specifies the server IP address to
5371 use. If not specified, the address is determined automatically, as
5374 * The DHCP server logic in systemd-networkd gained support for static
5375 DHCP leases, configurable via the [DHCPServerStaticLease]
5376 section. This allows explicitly mapping specific MAC addresses to
5377 fixed IP addresses and vice versa.
5379 * The RestrictAddressFamilies= setting in service files now supports a
5380 new special value "none". If specified sockets of all address
5381 families will be made unavailable to services configured that way.
5383 * systemd-fstab-generator and systemd-repart have been updated to
5384 support booting from disks that carry only a /usr/ partition but no
5385 root partition yet, and where systemd-repart can add it in on the
5386 first boot. This is useful for implementing systems that ship with a
5387 single /usr/ file system, and whose root file system shall be set up
5388 and formatted on a LUKS-encrypted volume whose key is generated
5389 locally (and possibly enrolled in the TPM) during the first boot.
5391 * The [Address] section of .network files now accepts a new
5392 RouteMetric= setting that configures the routing metric to use for
5393 the prefix route created as effect of the address configuration.
5394 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
5395 gained matching settings for their prefix routes. (The option of the
5396 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
5397 it conceptually belongs there; the old option is still understood for
5400 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
5403 * A new udev property ID_NET_DHCP_BROADCAST on network interface
5404 devices is now honoured by systemd-networkd, controlling whether to
5405 issue DHCP offers via broadcasting. This is used to ensure that s390
5406 layer 3 network interfaces work out-of-the-box with systemd-networkd.
5408 * nss-myhostname and systemd-resolved will now synthesize address
5409 records for a new special hostname "_outbound". The name will always
5410 resolve to the local IP addresses most likely used for outbound
5411 connections towards the default routes. On multi-homed hosts this is
5412 useful to have a stable handle referring to "the" local IP address
5413 that matters most, to the point where this is defined.
5415 * The Discoverable Partition Specification has been updated with a new
5416 GPT partition flag "grow-file-system" defined for its partition
5417 types. Whenever partitions with this flag set are automatically
5418 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
5419 of systemd-nspawn or other tools; and as opposed to explicit mounting
5420 via /etc/fstab), the file system within the partition is
5421 automatically grown to the full size of the partition. If the file
5422 system size already matches the partition size this flag has no
5423 effect. Previously, this functionality has been available via the
5424 explicit x-systemd.growfs mount option, and this new flag extends
5425 this to automatically discovered mounts. A new GrowFileSystem=
5426 setting has been added to systemd-repart drop-in files that allows
5427 configuring this partition flag. This new flag defaults to on for
5428 partitions automatically created by systemd-repart, except if they
5429 are marked read-only. See the specification for further details:
5431 https://systemd.io/DISCOVERABLE_PARTITIONS
5433 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
5434 section. If enabled (which is the default), and an NTP server address
5435 is acquired through a DHCP lease on this interface an explicit route
5436 to this address is created on this interface to ensure that NTP
5437 traffic to the NTP server acquired on an interface is also routed
5438 through that interface. The pre-existing RoutesToDNS= setting that
5439 implements the same for DNS servers is now enabled by default.
5441 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
5442 been added that may be used to restrict the network interfaces
5443 sockets created by the service may be bound to. This is implemented
5446 * A new ConditionFirmware= setting has been added to unit files to
5447 conditionalize on certain firmware features. At the moment it may
5448 check whether running on a UEFI system, a device.tree system, or if
5449 the system is compatible with some specified device-tree feature.
5451 * A new ConditionOSRelease= setting has been added to unit files to
5452 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
5453 operators may be used to check if some field has some specific value
5454 or do an alphanumerical comparison. Equality comparisons are useful
5455 for fields like ID, but relative comparisons for fields like
5456 VERSION_ID or IMAGE_VERSION.
5458 * hostnamed gained a new Describe() D-Bus method that returns a JSON
5459 serialization of the host data it exposes. This is exposed via
5460 "hostnamectl --json=" to acquire a host identity description in JSON.
5461 It's our intention to add a similar features to most services and
5462 objects systemd manages, in order to simplify integration with
5463 program code that can consume JSON.
5465 * Similarly, networkd gained a Describe() method on its Manager and
5466 Link bus objects. This is exposed via "networkctl --json=".
5468 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
5469 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
5470 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
5471 that is used both to get the value (when no argument is given), and
5472 to set the value (when an argument is specified). The old names
5473 continue to be supported for compatibility.
5475 * systemd-detect-virt and ConditionVirtualization= are now able to
5476 correctly identify Amazon EC2 environments.
5478 * The LogLevelMax= setting of unit files now applies not only to log
5479 messages generated *by* the service, but also to log messages
5480 generated *about* the service by PID 1. To suppress logs concerning a
5481 specific service comprehensively, set this option to a high log
5484 * bootctl gained support for a new --make-machine-id-directory= switch
5485 that allows precise control on whether to create the top-level
5486 per-machine directory in the boot partition that typically contains
5487 Type 1 boot loader entries.
5489 * During build SBAT data to include in the systemd-boot EFI PE binaries
5490 may be specified now.
5492 * /etc/crypttab learnt a new option "headless". If specified any
5493 requests to query the user interactively for passwords or PINs will
5494 be skipped. This is useful on systems that are headless, i.e. where
5495 an interactive user is generally not present.
5497 * /etc/crypttab also learnt a new option "password-echo=" that allows
5498 configuring whether the encryption password prompt shall echo the
5499 typed password and if so, do so literally or via asterisks. (The
5500 default is the same behaviour as before: provide echo feedback via
5503 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
5504 systemd-homed has been updated to allow explicit configuration of the
5505 "user presence" and "user verification" checks, as well as whether a
5506 PIN is required for authentication, via the new switches
5507 --fido2-with-user-presence=, --fido2-with-user-verification=,
5508 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
5509 features are available, and may be enabled or disabled depends on the
5512 * systemd-nspawn's --private-user= switch now accepts the special value
5513 "identity" which configures a user namespacing environment with an
5514 identity mapping of 65535 UIDs. This means the container UID 0 is
5515 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
5516 this doesn't appear to be useful, however it does reduce the attack
5517 surface a bit, since the resulting container will possess process
5518 capabilities only within its namespace and not on the host.
5520 * systemd-nspawn's --private-user-chown switch has been replaced by a
5521 more generic --private-user-ownership= switch that accepts one of
5522 three values: "chown" is equivalent to the old --private-user-chown,
5523 and "off" is equivalent to the absence of the old switch. The value
5524 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
5525 of files and directories of the underlying image to the chosen UID
5526 range for the container. "auto" is equivalent to "map" if UID mapping
5527 mount are supported, otherwise it is equivalent to "chown". The short
5528 -U switch systemd-nspawn now implies --private-user-ownership=auto
5529 instead of the old --private-user-chown. Effectively this means: if
5530 the backing file system supports UID mapping mounts the feature is
5531 now used by default if -U is used. Generally, it's a good idea to use
5532 UID mapping mounts instead of recursive chown()ing, since it allows
5533 running containers off immutable images (since no modifications of
5534 the images need to take place), and share images between multiple
5535 instances. Moreover, the recursive chown()ing operation is slow and
5536 can be avoided. Conceptually it's also a good thing if transient UID
5537 range uses do not leak into persistent file ownership anymore. TLDR:
5538 finally, the last major drawback of user namespacing has been
5539 removed, and -U should always be used (unless you use btrfs, where
5540 UID mapped mounts do not exist; or your container actually needs
5541 privileges on the host).
5543 * nss-systemd now synthesizes user and group shadow records in addition
5544 to the main user and group records. Thus, hashed passwords managed by
5545 systemd-homed are now accessible via the shadow database.
5547 * The userdb logic (and thus nss-systemd, and so on) now read
5548 additional user/group definitions in JSON format from the drop-in
5549 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
5550 /usr/lib/userdb/. This is a simple and powerful mechanism for making
5551 additional users available to the system, with full integration into
5552 NSS including the shadow databases. Since the full JSON user/group
5553 record format is supported this may also be used to define users with
5554 resource management settings and other runtime settings that
5555 pam_systemd and systemd-logind enforce at login.
5557 * The userdbctl tool gained two new switches --with-dropin= and
5558 --with-varlink= which can be used to fine-tune the sources used for
5559 user database lookups.
5561 * systemd-nspawn gained a new switch --bind-user= for binding a host
5562 user account into the container. This does three things: the user's
5563 home directory is bind mounted from the host into the container,
5564 below the /run/userdb/home/ hierarchy. A free UID is picked in the
5565 container, and a user namespacing UID mapping to the host user's UID
5566 installed. And finally, a minimal JSON user and group record (along
5567 with its hashed password) is dropped into /run/host/userdb/. These
5568 records are picked up automatically by the userdb drop-in logic
5569 describe above, and allow the user to login with the same password as
5570 on the host. Effectively this means: if host and container run new
5571 enough systemd versions making a host user available to the container
5572 is trivially simple.
5574 * systemd-journal-gatewayd now supports the switches --user, --system,
5575 --merge, --file= that are equivalent to the same switches of
5576 journalctl, and permit exposing only the specified subset of the
5579 * The OnFailure= dependency between units is now augmented with a
5580 implicit reverse dependency OnFailureOf= (this new dependency cannot
5581 be configured directly it's only created as effect of an OnFailure=
5582 dependency in the reverse order — it's visible in "systemctl show"
5583 however). Similar, Slice= now has an reverse dependency SliceOf=,
5584 that is also not configurable directly, but useful to determine all
5585 units that are members of a slice.
5587 * A pair of new dependency types between units PropagatesStopTo= +
5588 StopPropagatedFrom= has been added, that allows propagation of unit
5589 stop events between two units. It operates similar to the existing
5590 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
5592 * A new dependency type OnSuccess= has been added (plus the reverse
5593 dependency OnSuccessOf=, which cannot be configured directly, but
5594 exists only as effect of the reverse OnSuccess=). It is similar to
5595 OnFailure=, but triggers in the opposite case: when a service exits
5596 cleanly. This allows "chaining up" of services where one or more
5597 services are started once another service has successfully completed.
5599 * A new dependency type Upholds= has been added (plus the reverse
5600 dependency UpheldBy=, which cannot be configured directly, but exists
5601 only as effect of Upholds=). This dependency type is a stronger form
5602 of Wants=: if a unit has an UpHolds= dependency on some other unit
5603 and the former is active then the latter is started whenever it is
5604 found inactive (and no job is queued for it). This is an alternative
5605 to Restart= inside service units, but less configurable, and the
5606 request to uphold a unit is not encoded in the unit itself but in
5607 another unit that intends to uphold it.
5609 * The systemd-ask-password tool now also supports reading passwords
5610 from the credentials subsystem, via the new --credential= switch.
5612 * The systemd-ask-password tool learnt a new switch --emoji= which may
5613 be used to explicit control whether the lock and key emoji (🔐) is
5614 shown in the password prompt on suitable TTYs.
5616 * The --echo switch of systemd-ask-password now optionally takes a
5617 parameter that controls character echo. It may either show asterisks
5618 (default, as before), turn echo off entirely, or echo the typed
5619 characters literally.
5621 * The systemd-ask-password tool also gained a new -n switch for
5622 suppressing output of a trailing newline character when writing the
5623 acquired password to standard output, similar to /bin/echo's -n
5626 * New documentation has been added that describes the organization of
5627 the systemd source code tree:
5629 https://systemd.io/ARCHITECTURE
5631 * Units using ConditionNeedsUpdate= will no longer be activated in
5634 * It is now possible to list a template unit in the WantedBy= or
5635 RequiredBy= settings of the [Install] section of another template
5636 unit, which will be instantiated using the same instance name.
5638 * A new MemoryAvailable property is available for units. If the unit,
5639 or the slices it is part of, have a memory limit set via MemoryMax=/
5640 MemoryHigh=, MemoryAvailable will indicate how much more memory the
5641 unit can claim before hitting the limits.
5643 * systemd-coredump will now try to stay below the cgroup memory limit
5644 placed on itself or one of the slices it runs under, if the storage
5645 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
5646 since files written on such filesystems count toward the cgroup memory
5647 limit. If there is not enough available memory in such cases to store
5648 the core file uncompressed, systemd-coredump will skip to compressed
5649 storage directly (if enabled) and it will avoid analyzing the core file
5650 to print backtrace and metadata in the journal.
5652 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
5653 of a path matches the configured expectations, and remove it if not.
5655 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
5656 specify which of the several available filesystem timestamps (access
5657 time, birth time, change time, modification time) to look at when
5658 deciding whether a path has aged enough to be cleaned.
5660 * A new IPv6StableSecretAddress= setting has been added to .network
5661 files, which takes an IPv6 address to use as secret for IPv6 address
5664 * The [DHCPServer] logic in .network files gained support for a new
5665 UplinkInterface= setting that permits configuration of the uplink
5666 interface name to propagate DHCP lease information from.
5668 * The WakeOnLan= setting in .link files now accepts a list of flags
5669 instead of a single one, to configure multiple wake-on-LAN policies.
5671 * User-space defined tracepoints (USDT) have been added to udev at
5672 strategic locations. This is useful for tracing udev behaviour and
5673 performance with bpftrace and similar tools.
5675 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
5676 setting a network timeout time.
5678 * If a system service is running in a new mount namespace (RootDirectory=
5679 and friends), all file systems will be mounted with MS_NOSUID by
5680 default, unless the system is running with SELinux enabled.
5682 * When enumerating time zones the timedatectl tool will now consult the
5683 'tzdata.zi' file shipped by the IANA time zone database package, in
5684 addition to 'zone1970.tab', as before. This makes sure time zone
5685 aliases are now correctly supported. Some distributions so far did
5686 not install this additional file, most do however. If you
5687 distribution does not install it yet, it might make sense to change
5690 * Intel HID rfkill event is no longer masked, since it's the only
5691 source of rfkill event on newer HP laptops. To have both backward and
5692 forward compatibility, userspace daemon needs to debounce duplicated
5693 events in a short time window.
5695 Contributions from: Aakash Singh, adrian5, Albert Brox,
5696 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
5697 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
5698 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
5699 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
5700 Christian Hesse, Daniel Schaefer, Dan Streetman,
5701 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
5702 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
5703 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
5704 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
5705 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
5706 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
5707 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
5708 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
5709 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
5710 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
5711 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
5712 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
5713 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
5714 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
5715 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
5716 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
5717 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
5718 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
5719 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
5720 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
5721 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
5722 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
5723 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
5724 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
5725 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
5727 — Edinburgh, 2021-07-07
5731 * A concept of system extension images is introduced. Such images may
5732 be used to extend the /usr/ and /opt/ directory hierarchies at
5733 runtime with additional files (even if the file system is read-only).
5734 When a system extension image is activated, its /usr/ and /opt/
5735 hierarchies and os-release information are combined via overlayfs
5736 with the file system hierarchy of the host OS.
5738 A new systemd-sysext tool can be used to merge, unmerge, list, and
5739 refresh system extension hierarchies. See
5740 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
5742 The systemd-sysext.service automatically merges installed system
5743 extensions during boot (before basic.target, but not in very early
5744 boot, since various file systems have to be mounted first).
5746 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
5747 supported system extension level.
5749 * A new ExtensionImages= unit setting can be used to apply the same
5750 system extension image concept from systemd-sysext to the namespaced
5751 file hierarchy of specific services, following the same rules and
5754 * Support for a new special "root=tmpfs" kernel command-line option has
5755 been added. When specified, a tmpfs is mounted on /, and mount.usr=
5756 should be used to point to the operating system implementation.
5758 * A new configuration file /etc/veritytab may be used to configure
5759 dm-verity integrity protection for block devices. Each line is in the
5760 format "volume-name data-device hash-device roothash options",
5761 similar to /etc/crypttab.
5763 * A new kernel command-line option systemd.verity.root_options= may be
5764 used to configure dm-verity behaviour for the root device.
5766 * The key file specified in /etc/crypttab (the third field) may now
5767 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
5768 acquired by connecting to that socket and reading from it. This
5769 allows the implementation of a service to provide key information
5770 dynamically, at the moment when it is needed.
5772 * When the hostname is set explicitly to "localhost", systemd-hostnamed
5773 will respect this. Previously such a setting would be mostly silently
5774 ignored. The goal is to honour configuration as specified by the
5777 * The fallback hostname that will be used by the system manager and
5778 systemd-hostnamed can now be configured in two new ways: by setting
5779 DEFAULT_HOSTNAME= in os-release(5), or by setting
5780 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
5781 also be configured during compilation. The environment variable is
5782 intended for testing and local overrides, the os-release(5) field is
5783 intended to allow customization by different variants of a
5784 distribution that share the same compiled packages.
5786 * The environment block of the manager itself may be configured through
5787 a new ManagerEnvironment= setting in system.conf or user.conf. This
5788 complements existing ways to set the environment block (the kernel
5789 command line for the system manager, the inherited environment and
5790 user@.service unit file settings for the user manager).
5792 * systemd-hostnamed now exports the default hostname and the source of
5793 the configured hostname ("static", "transient", or "default") as
5796 * systemd-hostnamed now exports the "HardwareVendor" and
5797 "HardwareModel" D-Bus properties, which are supposed to contain a
5798 pair of cleaned up, human readable strings describing the system's
5799 vendor and model. It's typically sourced from the firmware's DMI
5800 tables, but may be augmented from a new hwdb database. hostnamectl
5801 shows this in the status output.
5803 * Support has been added to systemd-cryptsetup for extracting the
5804 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
5805 metadata header. This allows the information how to open the
5806 encrypted device to be embedded directly in the device and obviates
5807 the need for configuration in an external file.
5809 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
5810 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
5811 pre-existing support for PKCS#11 security tokens).
5813 * systemd-repart may enroll encrypted partitions using TPM2
5814 hardware. This may be useful for example to create an encrypted /var
5815 partition bound to the machine on first boot.
5817 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
5818 and PKCS#11 security tokens to LUKS volumes, list and destroy
5821 https://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
5823 It also supports enrolling "recovery keys" and regular passphrases.
5825 * The libfido2 dependency is now based on dlopen(), so that the library
5826 is used at runtime when installed, but is not a hard runtime
5829 * systemd-cryptsetup gained support for two new options in
5830 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
5831 request synchronous processing of encryption/decryption IO.
5833 * The manager may be configured at compile time to use the fexecve()
5834 instead of the execve() system call when spawning processes. Using
5835 fexecve() closes a window between checking the security context of an
5836 executable and spawning it, but unfortunately the kernel displays
5837 stale information in the process' "comm" field, which impacts ps
5840 * The configuration option -Dcompat-gateway-hostname has been dropped.
5841 "_gateway" is now the only supported name.
5843 * The ConditionSecurity=tpm2 unit file setting may be used to check if
5844 the system has at least one TPM2 (tpmrm class) device.
5846 * A new ConditionCPUFeature= has been added that may be used to
5847 conditionalize units based on CPU features. For example,
5848 ConditionCPUFeature=rdrand will condition a unit so that it is only
5849 run when the system CPU supports the RDRAND opcode.
5851 * The existing ConditionControlGroupController= setting has been
5852 extended with two new values "v1" and "v2". "v2" means that the
5853 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
5854 hierarchy or the hybrid hierarchy are used.
5856 * A new PrivateIPC= setting on a unit file allows executed processes to
5857 be moved into a private IPC namespace, with separate System V IPC
5858 identifiers and POSIX message queues.
5860 A new IPCNamespacePath= allows the unit to be joined to an existing
5863 * The tables of system calls in seccomp filters are now automatically
5864 generated from kernel lists exported on
5865 https://fedora.juszkiewicz.com.pl/syscalls.html.
5867 The following architectures should now have complete lists:
5868 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
5869 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
5871 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
5872 on /run/ if it is not already a mount point. A writable /run/ has
5873 always been a requirement for a functioning system, but this was not
5874 guaranteed when using a read-only image.
5876 Users can always specify BindPaths= or InaccessiblePaths= as
5877 overrides, and they will take precedence. If the host's root mount
5878 point is used, there is no change in behaviour.
5880 * New bind mounts and file system image mounts may be injected into the
5881 mount namespace of a service (without restarting it). This is exposed
5882 respectively as 'systemctl bind <unit> <path>…' and
5883 'systemctl mount-image <unit> <image>…'.
5885 * The StandardOutput= and StandardError= settings can now specify files
5886 to be truncated for output (as "truncate:<path>").
5888 * The ExecPaths= and NoExecPaths= settings may be used to specify
5889 noexec for parts of the file system.
5891 * sd-bus has a new function sd_bus_open_user_machine() to open a
5892 connection to the session bus of a specific user in a local container
5893 or on the local host. This is exposed in the existing -M switch to
5894 systemctl and similar tools:
5896 systemctl --user -M lennart@foobar start foo
5898 This will connect to the user bus of a user "lennart" in container
5899 "foobar". If no container name is specified, the specified user on
5900 the host itself is connected to
5902 systemctl --user -M lennart@ start quux
5904 * sd-bus also gained a convenience function sd_bus_message_send() to
5905 simplify invocations of sd_bus_send(), taking only a single
5906 parameter: the message to send.
5908 * sd-event allows rate limits to be set on event sources, for dealing
5909 with high-priority event sources that might starve out others. See
5910 the new man page sd_event_source_set_ratelimit(3) for details.
5912 * systemd.link files gained a [Link] Promiscuous= switch, which allows
5913 the device to be raised in promiscuous mode.
5915 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
5916 number of TX and RX queues to be configured.
5918 New [Link] TransmitQueueLength= setting allows the size of the TX
5919 queue to be configured.
5921 New [Link] GenericSegmentOffloadMaxBytes= and
5922 GenericSegmentOffloadMaxSegments= allow capping the packet size and
5923 the number of segments accepted in Generic Segment Offload.
5925 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
5926 wireless routing protocol that operates on ISO/OSI Layer 2 only and
5927 uses ethernet frames to route/bridge packets. This encompasses a new
5928 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
5929 new settings in .netdev files, and a new BatmanAdvanced= setting in
5932 * systemd.network files gained a [Network] RouteTable= configuration
5933 switch to select the routing policy table.
5935 systemd.network files gained a [RoutingPolicyRule] Type=
5936 configuration switch (one of "blackhole, "unreachable", "prohibit").
5938 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
5939 RouteAllowList= settings to ignore/accept route advertisements from
5940 routers matching specified prefixes. The DenyList= setting has been
5941 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
5944 systemd.network files gained a [DHCPv6] UseAddress= setting to
5945 optionally ignore the address provided in the lease.
5947 systemd.network files gained a [DHCPv6PrefixDelegation]
5948 ManageTemporaryAddress= switch.
5950 systemd.network files gained a new ActivationPolicy= setting which
5951 allows configuring how the UP state of an interface shall be managed,
5952 i.e. whether the interface is always upped, always downed, or may be
5953 upped/downed by the user using "ip link set dev".
5955 * The default for the Broadcast= setting in .network files has slightly
5956 changed: the broadcast address will not be configured for wireguard
5959 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
5960 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
5961 configuration options for VLAN packet handling.
5963 * udev rules may now set log_level= option. This allows debug logs to
5964 be enabled for select events, e.g. just for a specific subsystem or
5965 even a single device.
5967 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
5968 DATA_PREPARED_ID properties for block devices with ISO9660 file
5971 * udev now exports decoded DMI information about installed memory slots
5972 as device properties under the /sys/class/dmi/id/ pseudo device.
5974 * /dev/ is not mounted noexec anymore. This didn't provide any
5975 significant security benefits and would conflict with the executable
5976 mappings used with /dev/sgx device nodes. The previous behaviour can
5977 be restored for individual services with NoExecPaths=/dev (or by allow-
5978 listing and excluding /dev from ExecPaths=).
5980 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
5981 and /dev/vhost-net are owned by the kvm group.
5983 * The hardware database has been extended with a list of fingerprint
5984 readers that correctly support USB auto-suspend using data from
5987 * systemd-resolved can now answer DNSSEC questions through the stub
5988 resolver interface in a way that allows local clients to do DNSSEC
5989 validation themselves. For a question with DO+CD set, it'll proxy the
5990 DNS query and respond with a mostly unmodified packet received from
5991 the upstream server.
5993 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
5994 resolved.conf. If true the service will provide caching even for DNS
5995 lookups made to an upstream DNS server on the 127.0.0.1/::1
5996 addresses. By default (and when the option is false) systemd-resolved
5997 will not cache such lookups, in order to avoid duplicate local
5998 caching, under the assumption the local upstream server caches
6001 * systemd-resolved now implements RFC5001 NSID in its local DNS
6002 stub. This may be used by local clients to determine whether they are
6003 talking to the DNS resolver stub or a different DNS server.
6005 * When resolving host names and other records resolvectl will now
6006 report where the data was acquired from (i.e. the local cache, the
6007 network, locally synthesized, …) and whether the network traffic it
6008 effected was encrypted or not. Moreover the tool acquired a number of
6009 new options --cache=, --synthesize=, --network=, --zone=,
6010 --trust-anchor=, --validate= that take booleans and may be used to
6011 tweak a lookup, i.e. whether it may be answered from cached
6012 information, locally synthesized information, information acquired
6013 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
6014 anchor, and whether DNSSEC validation shall be executed for the
6017 * systemd-nspawn gained a new --ambient-capability= setting
6018 (AmbientCapability= in .nspawn files) to configure ambient
6019 capabilities passed to the container payload.
6021 * systemd-nspawn gained the ability to configure the firewall using the
6022 nftables subsystem (in addition to the existing iptables
6023 support). Similarly, systemd-networkd's IPMasquerade= option now
6024 supports nftables as back-end, too. In both cases NAT on IPv6 is now
6025 supported too, in addition to IPv4 (the iptables back-end still is
6028 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
6029 retains its meaning, but has been deprecated. Please switch to either
6030 "ivp4" or "both" (if covering IPv6 is desired).
6032 * systemd-importd will now download .verity and .roothash.p7s files
6033 along with the machine image (as exposed via machinectl pull-raw).
6035 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
6036 setting to configure the time a unit's cgroup needs to exceed memory
6037 pressure limits before action will be taken, and a new
6038 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
6041 systemd-oomd is now considered fully supported (the usual
6042 backwards-compatibility promises apply). Swap is not required for
6043 operation, but it is still recommended.
6045 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
6046 configures the retry delay when trying to contact servers.
6048 * systemd-stdio-bridge gained --system/--user options to connect to the
6049 system bus (previous default) or the user session bus.
6051 * systemd-localed may now call locale-gen to generate missing locales
6052 on-demand (UTF-8-only). This improves integration with Debian-based
6053 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
6055 * systemctl --check-inhibitors=true may now be used to obey inhibitors
6056 even when invoked non-interactively. The old --ignore-inhibitors
6057 switch is now deprecated and replaced by --check-inhibitors=false.
6059 * systemctl import-environment will now emit a warning when called
6060 without any arguments (i.e. to import the full environment block of
6061 the called program). This command will usually be invoked from a
6062 shell, which means that it'll inherit a bunch of variables which are
6063 specific to that shell, and usually to the TTY the shell is connected
6064 to, and don't have any meaning in the global context of the system or
6065 user service manager. Instead, only specific variables should be
6066 imported into the manager environment block.
6068 Similarly, programs which update the manager environment block by
6069 directly calling the D-Bus API of the manager, should also push
6070 specific variables, and not the full inherited environment.
6072 * systemctl's status output now shows unit state with a more careful
6073 choice of Unicode characters: units in maintenance show a "○" symbol
6074 instead of the usual "●", failed units show "×", and services being
6077 * coredumpctl gained a --debugger-arguments= switch to pass arguments
6078 to the debugger. It also gained support for showing coredump info in
6079 a simple JSON format.
6081 * systemctl/loginctl/machinectl's --signal= option now accept a special
6082 value "list", which may be used to show a brief table with known
6083 process signals and their numbers.
6085 * networkctl now shows the link activation policy in status.
6087 * Various tools gained --pager/--no-pager/--json= switches to
6088 enable/disable the pager and provide JSON output.
6090 * Various tools now accept two new values for the SYSTEMD_COLORS
6091 environment variable: "16" and "256", to configure how many terminal
6092 colors are used in output.
6094 * less 568 or newer is now required for the auto-paging logic of the
6095 various tools. Hyperlink ANSI sequences in terminal output are now
6096 used even if a pager is used, and older versions of less are not able
6097 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
6098 disable this output again.
6100 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
6101 builds, non-merged-usr builds) are now officially deprecated. A
6102 warning is emitted during build. Support is slated to be removed in
6103 about a year (when the Debian Bookworm release development starts).
6105 * Systems with the legacy cgroup v1 hierarchy are now marked as
6106 "tainted", to make it clearer that using the legacy hierarchy is not
6109 * systemd-localed will now refuse to configure a keymap which is not
6110 installed in the file system. This is intended as a bug fix, but
6111 could break cases where systemd-localed was used to configure the
6112 keymap in advanced of it being installed. It is necessary to install
6113 the keymap file first.
6115 * The main git development branch has been renamed to 'main'.
6117 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
6118 for partitions, as in the vast majority of cases they contain none
6119 and are used internally by the bootloader (eg: uboot).
6121 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
6122 spawned processes to the PID of the process itself. This may be used
6123 by programs for detecting whether they were forked off by the service
6124 manager itself or are a process forked off further down the tree.
6126 * The sd-device API gained four new calls: sd_device_get_action() to
6127 determine the uevent add/remove/change/… action the device object has
6128 been seen for, sd_device_get_seqno() to determine the uevent sequence
6129 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
6130 object from stat(2) data of a device node, and sd_device_trigger() to
6131 write to the 'uevent' attribute of a device.
6133 * For most tools the --no-legend= switch has been replaced by
6134 --legend=no and --legend=yes, to force whether tables are shown with
6137 * Units acquired a new property "Markers" that takes a list of zero,
6138 one or two of the following strings: "needs-reload" and
6139 "needs-restart". These markers may be set via "systemctl
6140 set-property". Once a marker is set, "systemctl reload-or-restart
6141 --marked" may be invoked to execute the operation the units are
6142 marked for. This is useful for package managers that want to mark
6143 units for restart/reload while updating, but effect the actual
6144 operations at a later step at once.
6146 * The sd_bus_message_read_strv() API call of sd-bus may now also be
6147 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
6150 * bootctl will now report whether the UEFI firmware used a TPM2 device
6151 and measured the boot process into it.
6153 * systemd-tmpfiles learnt support for a new environment variable
6154 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
6155 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
6156 even if the root fs of the system is not itself a btrfs volume.
6158 * systemd-detect-virt/ConditionVirtualization= will now explicitly
6159 detect Docker/Podman environments where possible. Moreover, they
6160 should be able to generically detect any container manager as long as
6161 it assigns the container a cgroup.
6163 * portablectl gained a new "reattach" verb for detaching/reattaching a
6164 portable service image, useful for updating images on-the-fly.
6166 * Intel SGX enclave device nodes (which expose a security feature of
6167 newer Intel CPUs) will now be owned by a new system group "sgx".
6169 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
6170 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
6171 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
6172 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
6173 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
6174 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
6175 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
6176 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
6177 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
6178 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
6179 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
6180 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
6181 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
6182 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
6183 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
6184 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
6185 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
6186 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
6187 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
6188 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
6189 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
6190 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
6191 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
6192 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
6193 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
6194 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
6195 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
6196 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
6197 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
6198 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
6199 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
6200 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
6201 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
6202 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
6203 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
6204 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
6205 Zmicer Turok, Дамјан Георгиевски
6207 — Berlin, 2021-03-30
6211 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
6212 "bind" and "unbind" to the Linux device model. When this kernel
6213 change was made, systemd-udevd was only minimally updated to handle
6214 and propagate these new event types. The introduction of these new
6215 uevents (which are typically generated for USB devices and devices
6216 needing a firmware upload before being functional) resulted in a
6217 number of issues which we so far didn't address. We hoped the kernel
6218 maintainers would themselves address these issues in some form, but
6219 that did not happen. To handle them properly, many (if not most) udev
6220 rules files shipped in various packages need updating, and so do many
6221 programs that monitor or enumerate devices with libudev or sd-device,
6222 or otherwise process uevents. Please note that this incompatibility
6223 is not fault of systemd or udev, but caused by an incompatible kernel
6224 change that happened back in Linux 4.14, but is becoming more and
6225 more visible as the new uevents are generated by more kernel drivers.
6227 To minimize issues resulting from this kernel change (but not avoid
6228 them entirely) starting with systemd-udevd 247 the udev "tags"
6229 concept (which is a concept for marking and filtering devices during
6230 enumeration and monitoring) has been reworked: udev tags are now
6231 "sticky", meaning that once a tag is assigned to a device it will not
6232 be removed from the device again until the device itself is removed
6233 (i.e. unplugged). This makes sure that any application monitoring
6234 devices that match a specific tag is guaranteed to both see uevents
6235 where the device starts being relevant, and those where it stops
6236 being relevant (the latter now regularly happening due to the new
6237 "unbind" uevent type). The udev tags concept is hence now a concept
6238 tied to a *device* instead of a device *event* — unlike for example
6239 udev properties whose lifecycle (as before) is generally tied to a
6240 device event, meaning that the previously determined properties are
6241 forgotten whenever a new uevent is processed.
6243 With the newly redefined udev tags concept, sometimes it's necessary
6244 to determine which tags are the ones applied by the most recent
6245 uevent/database update, in order to discern them from those
6246 originating from earlier uevents/database updates of the same
6247 device. To accommodate for this a new automatic property CURRENT_TAGS
6248 has been added that works similar to the existing TAGS property but
6249 only lists tags set by the most recent uevent/database
6250 update. Similarly, the libudev/sd-device API has been updated with
6251 new functions to enumerate these 'current' tags, in addition to the
6252 existing APIs that now enumerate the 'sticky' ones.
6254 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
6255 essential that all udev rules files and applications are updated to
6256 handle the new events. Specifically:
6258 • All rule files that currently use a header guard similar to
6259 ACTION!="add|change",GOTO="xyz_end" should be updated to use
6260 ACTION=="remove",GOTO="xyz_end" instead, so that the
6261 properties/tags they add are also applied whenever "bind" (or
6262 "unbind") is seen. (This is most important for all physical device
6263 types — those for which "bind" and "unbind" are currently
6264 generated, for all other device types this change is still
6265 recommended but not as important — but certainly prepares for
6266 future kernel uevent type additions).
6268 • Similarly, all code monitoring devices that contains an 'if' branch
6269 discerning the "add" + "change" uevent actions from all other
6270 uevents actions (i.e. considering devices only relevant after "add"
6271 or "change", and irrelevant on all other events) should be reworked
6272 to instead negatively check for "remove" only (i.e. considering
6273 devices relevant after all event types, except for "remove", which
6274 invalidates the device). Note that this also means that devices
6275 should be considered relevant on "unbind", even though conceptually
6276 this — in some form — invalidates the device. Since the precise
6277 effect of "unbind" is not generically defined, devices should be
6278 considered relevant even after "unbind", however I/O errors
6279 accessing the device should then be handled gracefully.
6281 • Any code that uses device tags for deciding whether a device is
6282 relevant or not most likely needs to be updated to use the new
6283 udev_device_has_current_tag() API (or sd_device_has_current_tag()
6284 in case sd-device is used), to check whether the tag is set at the
6285 moment an uevent is seen (as opposed to the existing
6286 udev_device_has_tag() API which checks if the tag ever existed on
6287 the device, following the API concept redefinition explained
6290 We are very sorry for this breakage and the requirement to update
6291 packages using these interfaces. We'd again like to underline that
6292 this is not caused by systemd/udev changes, but result of a kernel
6295 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
6296 packages have not retriggered devices once the udev package (or any
6297 auxiliary package installing additional udev rules) is updated. We
6298 intend to work with major distributions to change this, so that
6299 "udevadm trigger -c change" is issued on such upgrades, ensuring that
6300 the updated ruleset is applied to the devices already discovered, so
6301 that (asynchronously) after the upgrade completed the udev database
6302 is consistent with the updated rule set. This means udev rules must
6303 be ready to be retriggered with a "change" action any time, and
6304 result in correct and complete udev database entries. While the
6305 majority of udev rule files known to us currently get this right,
6306 some don't. Specifically, there are udev rules files included in
6307 various packages that only set udev properties on the "add" action,
6308 but do not handle the "change" action. If a device matching those
6309 rules is retriggered with the "change" action (as is intended here)
6310 it would suddenly lose the relevant properties. This always has been
6311 problematic, but as soon as all udev devices are triggered on relevant
6312 package upgrades this will become particularly so. It is strongly
6313 recommended to fix offending rules so that they can handle a "change"
6314 action at any time, and acquire all necessary udev properties even
6315 then. Or in other words: the header guard mentioned above
6316 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
6317 this, as it makes sure rules are rerun on "change" correctly, and
6318 accumulate the correct and complete set of udev properties. udev rule
6319 definitions that cannot handle "change" events being triggered at
6320 arbitrary times should be considered buggy.
6322 * The MountAPIVFS= service file setting now defaults to on if
6323 RootImage= and RootDirectory= are used, which means that with those
6324 two settings /proc/, /sys/ and /dev/ are automatically properly set
6325 up for services. Previous behaviour may be restored by explicitly
6326 setting MountAPIVFS=off.
6328 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
6329 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
6330 latter it takes precedence over the former, similar to how most of
6331 systemd's own configuration is handled. Given that PAM stack
6332 definitions are primarily put together by OS vendors/distributions
6333 (though possibly overridden by users), this systemd release moves its
6334 own PAM stack configuration for the "systemd-user" PAM service (i.e.
6335 for the PAM session invoked by the per-user user@.service instance)
6336 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
6337 packages' vendor versions of their PAM stack definitions from
6338 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
6339 desired the location to which systemd installs its PAM stack
6340 configuration may be changed via the -Dpamconfdir Meson option.
6342 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
6343 libpwquality and libcryptsetup have been changed to be based on
6344 dlopen(): instead of regular dynamic library dependencies declared in
6345 the binary ELF headers, these libraries are now loaded on demand
6346 only, if they are available. If the libraries cannot be found the
6347 relevant operations will fail gracefully, or a suitable fallback
6348 logic is chosen. This is supposed to be useful for general purpose
6349 distributions, as it allows minimizing the list of dependencies the
6350 systemd packages pull in, permitting building of more minimal OS
6351 images, while still making use of these "weak" dependencies should
6352 they be installed. Since many package managers automatically
6353 synthesize package dependencies from ELF shared library dependencies,
6354 some additional manual packaging work has to be done now to replace
6355 those (slightly downgraded from "required" to "recommended" or
6356 whatever is conceptually suitable for the package manager). Note that
6357 this change does not alter build-time behaviour: as before the
6358 build-time dependencies have to be installed during build, even if
6359 they now are optional during runtime.
6361 * sd-event.h gained a new call sd_event_add_time_relative() for
6362 installing timers relative to the current time. This is mostly a
6363 convenience wrapper around the pre-existing sd_event_add_time() call
6364 which installs absolute timers.
6366 * sd-event event sources may now be placed in a new "exit-on-failure"
6367 mode, which may be controlled via the new
6368 sd_event_source_get_exit_on_failure() and
6369 sd_event_source_set_exit_on_failure() functions. If enabled, any
6370 failure returned by the event source handler functions will result in
6371 exiting the event loop (unlike the default behaviour of just
6372 disabling the event source but continuing with the event loop). This
6373 feature is useful to set for all event sources that define "primary"
6374 program behaviour (where failure should be fatal) in contrast to
6375 "auxiliary" behaviour (where failure should remain local).
6377 * Most event source types sd-event supports now accept a NULL handler
6378 function, in which case the event loop is exited once the event
6379 source is to be dispatched, using the userdata pointer — converted to
6380 a signed integer — as exit code of the event loop. Previously this
6381 was supported for IO and signal event sources already. Exit event
6382 sources still do not support this (simply because it makes little
6383 sense there, as the event loop is already exiting when they are
6386 * A new per-unit setting RootImageOptions= has been added which allows
6387 tweaking the mount options for any file system mounted as effect of
6388 the RootImage= setting.
6390 * Another new per-unit setting MountImages= has been added, that allows
6391 mounting additional disk images into the file system tree accessible
6394 * Timer units gained a new FixedRandomDelay= boolean setting. If
6395 enabled, the random delay configured with RandomizedDelaySec= is
6396 selected in a way that is stable on a given system (though still
6397 different for different units).
6399 * Socket units gained a new setting Timestamping= that takes "us", "ns"
6400 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
6403 * systemd-repart now generates JSON output when requested with the new
6406 * systemd-machined's OpenMachineShell() bus call will now pass
6407 additional policy metadata data fields to the PolicyKit
6408 authentication request.
6410 * systemd-tmpfiles gained a new -E switch, which is equivalent to
6411 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
6412 --exclude=/sys. It's particularly useful in combination with --root=,
6413 when operating on OS trees that do not have any of these four runtime
6414 directories mounted, as this means no files below these subtrees are
6415 created or modified, since those mount points should probably remain
6418 * systemd-tmpfiles gained a new --image= switch which is like --root=,
6419 but takes a disk image instead of a directory as argument. The
6420 specified disk image is mounted inside a temporary mount namespace
6421 and the tmpfiles.d/ drop-ins stored in the image are executed and
6422 applied to the image. systemd-sysusers similarly gained a new
6423 --image= switch, that allows the sysusers.d/ drop-ins stored in the
6424 image to be applied onto the image.
6426 * Similarly, the journalctl command also gained an --image= switch,
6427 which is a quick one-step solution to look at the log data included
6430 * journalctl's --output=cat option (which outputs the log content
6431 without any metadata, just the pure text messages) will now make use
6432 of terminal colors when run on a suitable terminal, similarly to the
6435 * JSON group records now support a "description" string that may be
6436 used to add a human-readable textual description to such groups. This
6437 is supposed to match the user's GECOS field which traditionally
6438 didn't have a counterpart for group records.
6440 * The "systemd-dissect" tool that may be used to inspect OS disk images
6441 and that was previously installed to /usr/lib/systemd/ has now been
6442 moved to /usr/bin/, reflecting its updated status of an officially
6443 supported tool with a stable interface. It gained support for a new
6444 --mkdir switch which when combined with --mount has the effect of
6445 creating the directory to mount the image to if it is missing
6446 first. It also gained two new commands --copy-from and --copy-to for
6447 copying files and directories in and out of an OS image without the
6448 need to manually mount it. It also acquired support for a new option
6449 --json= to generate JSON output when inspecting an OS image.
6451 * The cgroup2 file system is now mounted with the
6452 "memory_recursiveprot" mount option, supported since kernel 5.7. This
6453 means that the MemoryLow= and MemoryMin= unit file settings now apply
6454 recursively to whole subtrees.
6456 * systemd-homed now defaults to using the btrfs file system — if
6457 available — when creating home directories in LUKS volumes. This may
6458 be changed with the DefaultFileSystemType= setting in homed.conf.
6459 It's now the default file system in various major distributions and
6460 has the major benefit for homed that it can be grown and shrunk while
6461 mounted, unlike the other contenders ext4 and xfs, which can both be
6462 grown online, but not shrunk (in fact xfs is the technically most
6463 limited option here, as it cannot be shrunk at all).
6465 * JSON user records managed by systemd-homed gained support for
6466 "recovery keys". These are basically secondary passphrases that can
6467 unlock user accounts/home directories. They are computer-generated
6468 rather than user-chosen, and typically have greater entropy.
6469 homectl's --recovery-key= option may be used to add a recovery key to
6470 a user account. The generated recovery key is displayed as a QR code,
6471 so that it can be scanned to be kept in a safe place. This feature is
6472 particularly useful in combination with systemd-homed's support for
6473 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
6474 security tokens are lost. Recovery keys may be entered wherever the
6475 system asks for a password.
6477 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
6478 home directory which indicates that a home directory has not been
6479 deactivated cleanly when offline. This flag is useful to identify
6480 home directories for which the offline discard logic did not run when
6481 offlining, and where it would be a good idea to log in again to catch
6484 * systemctl gained a new parameter --timestamp= which may be used to
6485 change the style in which timestamps are output, i.e. whether to show
6486 them in local timezone or UTC, or whether to show µs granularity.
6488 * Alibaba's "pouch" container manager is now detected by
6489 systemd-detect-virt, ConditionVirtualization= and similar
6490 constructs. Similar, they now also recognize IBM PowerVM machine
6493 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
6494 place to use for propagating external mounts into the
6495 container. Similarly /run/host/notify is now used as the socket path
6496 for container payloads to communicate with the container manager
6497 using sd_notify(). The container manager now uses the
6498 /run/host/inaccessible/ directory to place "inaccessible" file nodes
6499 of all relevant types which may be used by the container payload as
6500 bind mount source to over-mount inodes to make them inaccessible.
6501 /run/host/container-manager will now be initialized with the same
6502 string as the $container environment variable passed to the
6503 container's PID 1. /run/host/container-uuid will be initialized with
6504 the same string as $container_uuid. This means the /run/host/
6505 hierarchy is now the primary way to make host resources available to
6506 the container. The Container Interface documents these new files and
6509 https://systemd.io/CONTAINER_INTERFACE
6511 * Support for the "ConditionNull=" unit file condition has been
6512 deprecated and undocumented for 6 years. systemd started to warn
6513 about its use 1.5 years ago. It has now been removed entirely.
6515 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
6516 a sd_bus_error struct and a list of error names, and checks if the
6517 error matches one of these names. It's a convenience wrapper that is
6518 useful in cases where multiple errors shall be handled the same way.
6520 * A new system call filter list "@known" has been added, that contains
6521 all system calls known at the time systemd was built.
6523 * Behaviour of system call filter allow lists has changed slightly:
6524 system calls that are contained in @known will result in EPERM by
6525 default, while those not contained in it result in ENOSYS. This
6526 should improve compatibility because known system calls will thus be
6527 communicated as prohibited, while unknown (and thus newer ones) will
6528 be communicated as not implemented, which hopefully has the greatest
6529 chance of triggering the right fallback code paths in client
6532 * "systemd-analyze syscall-filter" will now show two separate sections
6533 at the bottom of the output: system calls known during systemd build
6534 time but not included in any of the filter groups shown above, and
6535 system calls defined on the local kernel but known during systemd
6538 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
6539 systemd-nspawn all system call filter violations will be logged by
6540 the kernel (audit). This is useful for tracking down system calls
6541 invoked by container payloads that are prohibited by the container's
6542 system call filter policy.
6544 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
6545 systemd-nspawn (and other programs that use seccomp) all seccomp
6546 filtering is turned off.
6548 * Two new unit file settings ProtectProc= and ProcSubset= have been
6549 added that expose the hidepid= and subset= mount options of procfs.
6550 All processes of the unit will only see processes in /proc that are
6551 are owned by the unit's user. This is an important new sandboxing
6552 option that is recommended to be set on all system services. All
6553 long-running system services that are included in systemd itself set
6554 this option now. This option is only supported on kernel 5.8 and
6555 above, since the hidepid= option supported on older kernels was not a
6556 per-mount option but actually applied to the whole PID namespace.
6558 * Socket units gained a new boolean setting FlushPending=. If enabled
6559 all pending socket data/connections are flushed whenever the socket
6560 unit enters the "listening" state, i.e. after the associated service
6563 * The unit file setting NUMAMask= gained a new "all" value: when used,
6564 all existing NUMA nodes are added to the NUMA mask.
6566 * A new "credentials" logic has been added to system services. This is
6567 a simple mechanism to pass privileged data to services in a safe and
6568 secure way. It's supposed to be used to pass per-service secret data
6569 such as passwords or cryptographic keys but also associated less
6570 private information such as user names, certificates, and similar to
6571 system services. Each credential is identified by a short user-chosen
6572 name and may contain arbitrary binary data. Two new unit file
6573 settings have been added: SetCredential= and LoadCredential=. The
6574 former allows setting a credential to a literal string, the latter
6575 sets a credential to the contents of a file (or data read from a
6576 user-chosen AF_UNIX stream socket). Credentials are passed to the
6577 service via a special credentials directory, one file for each
6578 credential. The path to the credentials directory is passed in a new
6579 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
6580 are passed in the file system they may be easily referenced in
6581 ExecStart= command lines too, thus no explicit support for the
6582 credentials logic in daemons is required (though ideally daemons
6583 would look for the bits they need in $CREDENTIALS_DIRECTORY
6584 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
6585 backed by unswappable memory if privileges allow it, immutable if
6586 privileges allow it, is accessible only to the service's UID, and is
6587 automatically destroyed when the service stops.
6589 * systemd-nspawn supports the same credentials logic. It can both
6590 consume credentials passed to it via the aforementioned
6591 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
6592 to its payload. The service manager/PID 1 has been updated to match
6593 this: it can also accept credentials from the container manager that
6594 invokes it (in fact: any process that invokes it), and passes them on
6595 to its services. Thus, credentials can be propagated recursively down
6596 the tree: from a system's service manager to a systemd-nspawn
6597 service, to the service manager that runs as container payload and to
6598 the service it runs below. Credentials may also be added on the
6599 systemd-nspawn command line, using new --set-credential= and
6600 --load-credential= command line switches that match the
6601 aforementioned service settings.
6603 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
6604 the partition drop-ins which may be used to format/LUKS
6605 encrypt/populate any created partitions. The partitions are
6606 encrypted/formatted/populated before they are registered in the
6607 partition table, so that they appear atomically: either the
6608 partitions do not exist yet or they exist fully encrypted, formatted,
6609 and populated — there is no time window where they are
6610 "half-initialized". Thus the system is robust to abrupt shutdown: if
6611 the tool is terminated half-way during its operations on next boot it
6612 will start from the beginning.
6614 * systemd-repart's --size= operation gained a new "auto" value. If
6615 specified, and operating on a loopback file it is automatically sized
6616 to the minimal size the size constraints permit. This is useful to
6617 use "systemd-repart" as an image builder for minimally sized images.
6619 * systemd-resolved now gained a third IPC interface for requesting name
6620 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
6621 interface is now supported. The nss-resolve NSS module has been
6622 modified to use this new interface instead of D-Bus. Using Varlink
6623 has a major benefit over D-Bus: it works without a broker service,
6624 and thus already during earliest boot, before the dbus daemon has
6625 been started. This means name resolution via systemd-resolved now
6626 works at the same time systemd-networkd operates: from earliest boot
6627 on, including in the initrd.
6629 * systemd-resolved gained support for a new DNSStubListenerExtra=
6630 configuration file setting which may be used to specify additional IP
6631 addresses the built-in DNS stub shall listen on, in addition to the
6632 main one on 127.0.0.53:53.
6634 * Name lookups issued via systemd-resolved's D-Bus and Varlink
6635 interfaces (and thus also via glibc NSS if nss-resolve is used) will
6636 now honour a trailing dot in the hostname: if specified the search
6637 path logic is turned off. Thus "resolvectl query foo." is now
6638 equivalent to "resolvectl query --search=off foo.".
6640 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
6641 exposes how /etc/resolv.conf is currently managed: by resolved (and
6642 in which mode if so) or another subsystem. "resolvctl" will display
6643 this property in its status output.
6645 * The resolv.conf snippets systemd-resolved provides will now set "."
6646 as the search domain if no other search domain is known. This turns
6647 off the derivation of an implicit search domain by nss-dns for the
6648 hostname, when the hostname is set to an FQDN. This change is done to
6649 make nss-dns using resolv.conf provided by systemd-resolved behave
6650 more similarly to nss-resolve.
6652 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
6653 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
6654 "birth" time (btime) of a file in addition to the atime, mtime, and
6657 * systemd-analyze gained a new verb "capability" that lists all known
6658 capabilities by the systemd build and by the kernel.
6660 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
6661 advance the system clock to it at boot if it is noticed to be before
6662 that time. Previously, PID 1 would only advance the time to an epoch
6663 time that is set during build-time. With this new file OS builders
6664 can change this epoch timestamp on individual OS images without
6665 having to rebuild systemd.
6667 * systemd-logind will now listen to the KEY_RESTART key from the Linux
6668 input layer and reboot the system if it is pressed, similarly to how
6669 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
6670 was originally defined in the Multimedia context (to restart playback
6671 of a song or film), but is now primarily used in various embedded
6672 devices for "Reboot" buttons. Accordingly, systemd-logind will now
6673 honour it as such. This may configured in more detail via the new
6674 HandleRebootKey= and RebootKeyIgnoreInhibited=.
6676 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
6677 copying OS trees, for example in "systemd-nspawn --ephemeral",
6678 "systemd-nspawn --template=", "machinectl clone" and similar. This is
6679 useful when operating with OSTree images, which use hardlinks heavily
6680 throughout, and where such copies previously resulting in "exploding"
6683 * systemd-nspawn's --console= setting gained support for a new
6684 "autopipe" value, which is identical to "interactive" when invoked on
6685 a TTY, and "pipe" otherwise.
6687 * systemd-networkd's .network files gained support for explicitly
6688 configuring the multicast membership entries of bridge devices in the
6689 [BridgeMDB] section. It also gained support for the PIE queuing
6690 discipline in the [FlowQueuePIE] sections.
6692 * systemd-networkd's .netdev files may now be used to create "BareUDP"
6693 tunnels, configured in the new [BareUDP] setting.
6695 * systemd-networkd's Gateway= setting in .network files now accepts the
6696 special values "_dhcp4" and "_ipv6ra" to configure additional,
6697 locally defined, explicit routes to the gateway acquired via DHCP or
6698 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
6699 but still accepted for backwards compatibility.
6701 * systemd-networkd's [IPv6PrefixDelegation] section and
6702 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
6703 IPv6SendRA= (the old names are still accepted for backwards
6706 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
6707 boolean setting in [Network] section. If enabled, the delegated prefix
6708 gained by another link will be configured, and an address within the
6709 prefix will be assigned.
6711 * systemd-networkd's .network files gained the Announce= boolean setting
6712 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
6713 prefix will be announced through IPv6 router advertisement (IPv6 RA).
6714 The setting is enabled by default.
6716 * VXLAN tunnels may now be marked as independent of any underlying
6717 network interface via the new Independent= boolean setting.
6719 * systemctl gained support for two new verbs: "service-log-level" and
6720 "service-log-target" may be used on services that implement the
6721 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
6722 adjust the log level and target. All of systemd's long-running
6723 services support this now, but ideally all system services would
6724 implement this interface to make the system more uniformly
6727 * The SystemCallErrorNumber= unit file setting now accepts the new
6728 "kill" and "log" actions, in addition to arbitrary error number
6729 specifications as before. If "kill" the processes are killed on the
6730 event, if "log" the offending system call is audit logged.
6732 * A new SystemCallLog= unit file setting has been added that accepts a
6733 list of system calls that shall be logged about (audit).
6735 * The OS image dissection logic (as used by RootImage= in unit files or
6736 systemd-nspawn's --image= switch) has gained support for identifying
6737 and mounting explicit /usr/ partitions, which are now defined in the
6738 discoverable partition specification. This should be useful for
6739 environments where the root file system is
6740 generated/formatted/populated dynamically on first boot and combined
6741 with an immutable /usr/ tree that is supplied by the vendor.
6743 * In the final phase of shutdown, within the systemd-shutdown binary
6744 we'll now try to detach MD devices (i.e software RAID) in addition to
6745 loopback block devices and DM devices as before. This is supposed to
6746 be a safety net only, in order to increase robustness if things go
6747 wrong. Storage subsystems are expected to properly detach their
6748 storage volumes during regular shutdown already (or in case of
6749 storage backing the root file system: in the initrd hook we return to
6752 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
6753 will now log the thread ID in their log output. This is useful when
6754 working with heavily threaded programs.
6756 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
6757 not use the RDRAND CPU instruction. This is useful in environments
6758 such as replay debuggers where non-deterministic behaviour is not
6761 * The autopaging logic in systemd's various tools (such as systemctl)
6762 has been updated to turn on "secure" mode in "less"
6763 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
6764 detected. This disables invoking external programs from the pager,
6765 via the pipe logic. This behaviour may be overridden via the new
6766 $SYSTEMD_PAGERSECURE environment variable.
6768 * Units which have resource limits (.service, .mount, .swap, .slice,
6769 .socket, and .slice) gained new configuration settings
6770 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
6771 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
6772 limits and optional action taken by systemd-oomd.
6774 * A new service systemd-oomd has been added. It monitors resource
6775 contention for selected parts of the unit hierarchy using the PSI
6776 information reported by the kernel, and kills processes when memory
6777 or swap pressure is above configured limits. This service is only
6778 enabled by default in developer mode (see below) and should be
6779 considered a preview in this release. Behaviour details and option
6780 names are subject to change without the usual backwards-compatibility
6783 * A new helper oomctl has been added to introspect systemd-oomd state.
6784 It is only enabled by default in developer mode and should be
6785 considered a preview without the usual backwards-compatibility
6788 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
6789 enabled, systemd reads the system UID boundaries from /etc/login.defs
6790 at runtime, instead of using the built-in values selected during
6791 build. This is an option to improve compatibility for upgrades from
6792 old systems. It's strongly recommended not to make use of this
6793 functionality on new systems (or even enable it during build), as it
6794 makes something runtime-configurable that is mostly an implementation
6795 detail of the OS, and permits avoidable differences in deployments
6796 that create all kinds of problems in the long run.
6798 * New meson option '-Dmode=developer|release' has been added. When
6799 'developer', additional checks and features are enabled that are
6800 relevant during upstream development, e.g. verification that
6801 semi-automatically-generated documentation has been properly updated
6802 following API changes. Those checks are considered hints for
6803 developers and are not actionable in downstream builds. In addition,
6804 extra features that are not ready for general consumption may be
6805 enabled in developer mode. It is thus recommended to set
6806 '-Dmode=release' in end-user and distro builds.
6808 * systemd-cryptsetup gained support for processing detached LUKS
6809 headers specified on the kernel command line via the header=
6810 parameter of the luks.options= kernel command line option. The same
6811 device/path syntax as for key files is supported for header files
6814 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
6815 slot index data for devices that are connected through a PCI bridge
6816 where the _SUN index is associated with the bridge instead of the
6817 network device itself. Previously this would create ambiguous device
6818 naming if multiple network interfaces were connected to the same PCI
6819 bridge. Since this is a naming scheme incompatibility on systems that
6820 possess hardware like this it has been introduced as new naming
6821 scheme "v247". The previous scheme can be selected via the
6822 "net.naming_scheme=v245" kernel command line parameter.
6824 * ConditionFirstBoot= semantics have been modified to be safe towards
6825 abnormal system power-off during first boot. Specifically, the
6826 "systemd-machine-id-commit.service" service now acts as boot
6827 milestone indicating when the first boot process is sufficiently
6828 complete in order to not consider the next following boot also a
6829 first boot. If the system is reset before this unit is reached the
6830 first time, the next boot will still be considered a first boot; once
6831 it has been reached, no further boots will be considered a first
6832 boot. The "first-boot-complete.target" unit now acts as official hook
6833 point to order against this. If a service shall be run on every boot
6834 until the first boot fully succeeds it may thus be ordered before
6835 this target unit (and pull it in) and carry ConditionFirstBoot=
6838 * bootctl's set-default and set-oneshot commands now accept the three
6839 special strings "@default", "@oneshot", "@current" in place of a boot
6840 entry id. These strings are resolved to the current default and
6841 oneshot boot loader entry, as well as the currently booted one. Thus
6842 a command "bootctl set-default @current" may be used to make the
6843 currently boot menu item the new default for all subsequent boots.
6845 * "systemctl edit" has been updated to show the original effective unit
6846 contents in commented form in the text editor.
6848 * Units in user mode are now segregated into three new slices:
6849 session.slice (units that form the core of graphical session),
6850 app.slice ("normal" user applications), and background.slice
6851 (low-priority tasks). Unless otherwise configured, user units are
6852 placed in app.slice. The plan is to add resource limits and
6853 protections for the different slices in the future.
6855 * New GPT partition types for RISCV32/64 for the root and /usr
6856 partitions, and their associated Verity partitions have been defined,
6857 and are now understood by systemd-gpt-auto-generator, and the OS
6858 image dissection logic.
6860 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
6861 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
6862 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
6863 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
6864 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
6865 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
6866 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
6867 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
6868 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
6869 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
6870 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
6871 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
6872 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
6873 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
6874 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
6875 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
6876 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
6877 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
6878 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
6879 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
6880 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
6881 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
6882 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
6883 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
6884 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
6885 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
6886 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
6887 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
6888 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
6889 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
6890 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
6891 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
6892 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
6894 – Warsaw, 2020-11-26
6898 * The service manager gained basic support for cgroup v2 freezer. Units
6899 can now be suspended or resumed either using new systemctl verbs,
6900 freeze and thaw respectively, or via D-Bus.
6902 * PID 1 may now automatically load pre-compiled AppArmor policies from
6903 /etc/apparmor/earlypolicy during early boot.
6905 * The CPUAffinity= setting in service unit files now supports a new
6906 special value "numa" that causes the CPU affinity masked to be set
6907 based on the NUMA mask.
6909 * systemd will now log about all left-over processes remaining in a
6910 unit when the unit is stopped. It will now warn about services using
6911 KillMode=none, as this is generally an unsafe thing to make use of.
6913 * Two new unit file settings
6914 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
6915 added. They may be used to check whether a specific file system path
6916 resides on a block device that is encrypted on the block level
6917 (i.e. using dm-crypt/LUKS).
6919 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
6920 has been added that may be used for simple environment checks. This
6921 is particularly useful when passing in environment variables from a
6922 container manager (or from PAM in case of the systemd --user
6925 * .service unit files now accept a new setting CoredumpFilter= which
6926 allows configuration of the memory sections coredumps of the
6927 service's processes shall include.
6929 * .mount units gained a new ReadWriteOnly= boolean option. If set
6930 it will not be attempted to mount a file system read-only if mounting
6931 in read-write mode doesn't succeed. An option x-systemd.rw-only is
6932 available in /etc/fstab to control the same.
6934 * .socket units gained a new boolean setting PassPacketInfo=. If
6935 enabled, the kernel will attach additional per-packet metadata to all
6936 packets read from the socket, as an ancillary message. This controls
6937 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
6938 depending on socket type.
6940 * .service units gained a new setting RootHash= which may be used to
6941 specify the root hash for verity enabled disk images which are
6942 specified in RootImage=. RootVerity= may be used to specify a path to
6943 the Verity data matching a RootImage= file system. (The latter is
6944 only useful for images that do not contain the Verity data embedded
6945 into the same image that carries a GPT partition table following the
6946 Discoverable Partition Specification). Similarly, systemd-nspawn
6947 gained a new switch --verity-data= that takes a path to a file with
6948 the verity data of the disk image supplied in --image=, if the image
6949 doesn't contain the verity data itself.
6951 * .service units gained a new setting RootHashSignature= which takes
6952 either a base64 encoded PKCS#7 signature of the root hash specified
6953 with RootHash=, or a path to a file to read the signature from. This
6954 allows validation of the root hash against public keys available in
6955 the kernel keyring, and is only supported on recent kernels
6956 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
6957 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
6958 this mechanism has also been added to systemd-veritysetup.
6960 * .service unit files gained two new options
6961 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
6962 tune behaviour if a start or stop timeout is hit, i.e. whether to
6963 terminate the service with SIGTERM, SIGABRT or SIGKILL.
6965 * Most options in systemd that accept hexadecimal values prefixed with
6966 0x in additional to the usual decimal notation now also support octal
6967 notation when the 0o prefix is used and binary notation if the 0b
6970 * Various command line parameters and configuration file settings that
6971 configure key or certificate files now optionally take paths to
6972 AF_UNIX sockets in the file system. If configured that way a stream
6973 connection is made to the socket and the required data read from
6974 it. This is a simple and natural extension to the existing regular
6975 file logic, and permits other software to provide keys or
6976 certificates via simple IPC services, for example when unencrypted
6977 storage on disk is not desired. Specifically, systemd-networkd's
6978 Wireguard and MACSEC key file settings as well as
6979 systemd-journal-gatewayd's and systemd-journal-remote's PEM
6980 key/certificate parameters support this now.
6982 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
6983 configuration files that support specifier expansion learnt six new
6984 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
6985 resolve to the various ID fields from /etc/os-release, %l resolves to
6986 the "short" hostname of the system, i.e. the hostname configured in
6987 the kernel truncated at the first dot.
6989 * Support for the .include syntax in unit files has been removed. The
6990 concept has been obsolete for 6 years and we started warning about
6991 its pending removal 2 years ago (also see NEWS file below). It's
6994 * StandardError= and StandardOutput= in unit files no longer support
6995 the "syslog" and "syslog-console" switches. They were long removed
6996 from the documentation, but will now result in warnings when used,
6997 and be converted to "journal" and "journal+console" automatically.
6999 * If the service setting User= is set to the "nobody" user, a warning
7000 message is now written to the logs (but the value is nonetheless
7001 accepted). Setting User=nobody is unsafe, since the primary purpose
7002 of the "nobody" user is to own all files whose owner cannot be mapped
7003 locally. It's in particular used by the NFS subsystem and in user
7004 namespacing. By running a service under this user's UID it might get
7005 read and even write access to all these otherwise unmappable files,
7006 which is quite likely a major security problem.
7008 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
7009 and others) now have a size and inode limits applied (50% of RAM for
7010 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
7011 that the implicit kernel default is 50% too, so there is no change
7012 in the size limit for /tmp and /dev/shm.
7014 * nss-mymachines lost support for resolution of users and groups, and
7015 now only does resolution of hostnames. This functionality is now
7016 provided by nss-systemd. Thus, the 'mymachines' entry should be
7017 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
7018 (and 'systemd' added if it is not already there).
7020 * A new kernel command line option systemd.hostname= has been added
7021 that allows controlling the hostname that is initialized early during
7024 * A kernel command line option "udev.blockdev_read_only" has been
7025 added. If specified all hardware block devices that show up are
7026 immediately marked as read-only by udev. This option is useful for
7027 making sure that a specific boot under no circumstances modifies data
7028 on disk. Use "blockdev --setrw" to undo the effect of this, per
7031 * A new boolean kernel command line option systemd.swap= has been
7032 added, which may be used to turn off automatic activation of swap
7033 devices listed in /etc/fstab.
7035 * New kernel command line options systemd.condition_needs_update= and
7036 systemd.condition_first_boot= have been added, which override the
7037 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
7040 * A new kernel command line option systemd.clock_usec= has been added
7041 that allows setting the system clock to the specified time in µs
7042 since Jan 1st, 1970 early during boot. This is in particular useful
7043 in order to make test cases more reliable.
7045 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
7046 systemd-coredump to save core files for suid processes. When saving
7047 the core file, systemd-coredump will use the effective uid and gid of
7048 the process that faulted.
7050 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
7051 now automatically set to "Y" at boot, in order to enable pstore
7052 generation for collection with systemd-pstore.
7054 * We provide a set of udev rules to enable auto-suspend on PCI and USB
7055 devices that were tested to correctly support it. Previously, this
7056 was distributed as a set of udev rules, but has now been replaced by
7057 by a set of hwdb entries (and a much shorter udev rule to take action
7058 if the device modalias matches one of the new hwdb entries).
7060 As before, entries are periodically imported from the database
7061 maintained by the ChromiumOS project. If you have a device that
7062 supports auto-suspend correctly and where it should be enabled by
7063 default, please submit a patch that adds it to the database (see
7064 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
7066 * systemd-udevd gained the new configuration option timeout_signal= as well
7067 as a corresponding kernel command line option udev.timeout_signal=.
7068 The option can be used to configure the UNIX signal that the main
7069 daemon sends to the worker processes on timeout. Setting the signal
7070 to SIGABRT is useful for debugging.
7072 * .link files managed by systemd-udevd gained options RxFlowControl=,
7073 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
7074 order to configure various flow control parameters. They also gained
7075 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
7076 frame ring buffer sizes.
7078 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
7079 enabled systemd-networkd manages all routes configured by other tools.
7081 * .network files managed by systemd-networkd gained a new section
7082 [SR-IOV], in order to configure SR-IOV capable network devices.
7084 * systemd-networkd's [IPv6Prefix] section in .network files gained a
7085 new boolean setting Assign=. If enabled an address from the prefix is
7086 automatically assigned to the interface.
7088 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
7089 controls delegated prefixes assigned by DHCPv6 client. The section
7090 has three settings: SubnetID=, Assign=, and Token=. The setting
7091 SubnetID= allows explicit configuration of the preferred subnet that
7092 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
7093 Assign= is enabled (which is the default) an address from any acquired
7094 delegated prefix is automatically chosen and assigned to the
7095 interface. The setting Token= specifies an optional address generation
7098 * systemd-networkd's [Network] section gained a new setting
7099 IPv4AcceptLocal=. If enabled the interface accepts packets with local
7102 * systemd-networkd gained support for configuring the HTB queuing
7103 discipline in the [HierarchyTokenBucket] and
7104 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
7105 be configured in the [PFIFO] section, "GRED" in
7106 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
7107 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
7108 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
7109 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
7110 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
7111 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
7113 * systemd-networkd gained support for a new Termination= setting in the
7114 [CAN] section for configuring the termination resistor. It also
7115 gained a new ListenOnly= setting for controlling whether to only
7116 listen on CAN interfaces, without interfering with traffic otherwise
7117 (which is useful for debugging/monitoring CAN network
7118 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
7119 been added to configure various CAN-FD aspects.
7121 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
7122 When enabled, DHCPv6 will be attempted right-away without requiring an
7123 Router Advertisement packet suggesting it first (i.e. without the 'M'
7124 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
7125 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
7126 the RA packets suggest it.
7128 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
7129 which may be used to turn off use of the gateway information provided
7130 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
7131 used to configure how to process leases that lack a lifetime option.
7133 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
7134 setting SendVendorOption= allowing configuration of additional vendor
7135 options to send in the DHCP requests/responses. The [DHCPv6] section
7136 gained a new SendOption= setting for sending arbitrary DHCP
7137 options. RequestOptions= has been added to request arbitrary options
7138 from the server. UserClass= has been added to set the DHCP user class
7141 * systemd-networkd's [DHCPServer] section gained a new set of options
7142 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
7143 information about these three protocols in the DHCP lease. It also
7144 gained support for including "MUD" URLs ("Manufacturer Usage
7145 Description"). Support for "MUD" URLs was also added to the LLDP
7146 stack, configurable in the [LLDP] section in .network files.
7148 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
7149 mode. Also, the sections now support a new setting SourceMACAddress=.
7151 * systemd-networkd's .netdev files now support a new setting
7152 VLANProtocol= in the [Bridge] section that allows configuration of
7153 the VLAN protocol to use.
7155 * systemd-networkd supports a new Group= setting in the [Link] section
7156 of the .network files, to control the link group.
7158 * systemd-networkd's [Network] section gained a new
7159 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
7160 link local address is generated.
7162 * A new default .network file is now shipped that matches TUN/TAP
7163 devices that begin with "vt-" in their name. Such interfaces will
7164 have IP routing onto the host links set up automatically. This is
7165 supposed to be used by VM managers to trivially acquire a network
7166 interface which is fully set up for host communication, simply by
7167 carefully picking an interface name to use.
7169 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
7170 which sets the route priority for routes specified by the DHCP server.
7172 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
7173 which configures the vendor class information sent to DHCP server.
7175 * The BlackList= settings in .network files' [DHCPv4] and
7176 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
7177 are still understood to provide compatibility.
7179 * networkctl gained the new "forcerenew" command for forcing all DHCP
7180 server clients to renew their lease. The interface "status" output
7181 will now show numerous additional fields of information about an
7182 interface. There are new "up" and "down" commands to bring specific
7183 interfaces up or down.
7185 * systemd-resolved's DNS= configuration option now optionally accepts a
7186 port number (after ":") and a host name (after "#"). When the host
7187 name is specified, the DNS-over-TLS certificate is validated to match
7188 the specified hostname. Additionally, in case of IPv6 addresses, an
7189 interface may be specified (after "%").
7191 * systemd-resolved may be configured to forward single-label DNS names.
7192 This is not standard-conformant, but may make sense in setups where
7193 public DNS servers are not used.
7195 * systemd-resolved's DNS-over-TLS support gained SNI validation.
7197 * systemd-nspawn's --resolv-conf= switch gained a number of new
7198 supported values. Specifically, options starting with "replace-" are
7199 like those prefixed "copy-" but replace any existing resolv.conf
7200 file. And options ending in "-uplink" and "-stub" can now be used to
7201 propagate other flavours of resolv.conf into the container (as
7202 defined by systemd-resolved).
7204 * The various programs included in systemd can now optionally output
7205 their log messages on stderr prefixed with a timestamp, controlled by
7206 the $SYSTEMD_LOG_TIME environment variable.
7208 * systemctl gained a new "-P" switch that is a shortcut for "--value
7211 * "systemctl list-units" and "systemctl list-machines" no longer hide
7212 their first output column with --no-legend. To hide the first column,
7215 * "systemctl reboot" takes the option "--reboot-argument=".
7216 The optional positional argument to "systemctl reboot" is now
7217 being deprecated in favor of this option.
7219 * systemd-run gained a new switch --slice-inherit. If specified the
7220 unit it generates is placed in the same slice as the systemd-run
7223 * systemd-journald gained support for zstd compression of large fields
7224 in journal files. The hash tables in journal files have been hardened
7225 against hash collisions. This is an incompatible change and means
7226 that journal files created with new systemd versions are not readable
7227 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
7228 environment variable for systemd-journald.service is set to 0 this
7229 new hardening functionality may be turned off, so that generated
7230 journal files remain compatible with older journalctl
7233 * journalctl will now include a clickable link in the default output for
7234 each log message for which a URL with further documentation is
7235 known. This is only supported on terminal emulators that support
7236 clickable hyperlinks, and is turned off if a pager is used (since
7237 "less" still doesn't support hyperlinks,
7238 unfortunately). Documentation URLs may be included in log messages
7239 either by including a DOCUMENTATION= journal field in it, or by
7240 associating a journal message catalog entry with the log message's
7241 MESSAGE_ID, which then carries a "Documentation:" tag.
7243 * journald.conf gained a new boolean setting Audit= that may be used to
7244 control whether systemd-journald will enable audit during
7247 * when systemd-journald's log stream is broken up into multiple lines
7248 because the PID of the sender changed this is indicated in the
7249 generated log records via the _LINE_BREAK=pid-change field.
7251 * journalctl's "-o cat" output mode will now show one or more journal
7252 fields specified with --output-fields= instead of unconditionally
7253 MESSAGE=. This is useful to retrieve a very specific set of fields
7254 without any decoration.
7256 * The sd-journal.h API gained two new functions:
7257 sd_journal_enumerate_available_unique() and
7258 sd_journal_enumerate_available_data() that operate like their
7259 counterparts that lack the _available_ in the name, but skip items
7260 that cannot be read and processed by the local implementation
7261 (i.e. are compressed in an unsupported format or such),
7263 * coredumpctl gained a new --file= switch, matching the same one in
7264 journalctl: a specific journal file may be specified to read the
7267 * coredumps collected by systemd-coredump may now be compressed using
7270 * systemd-binfmt gained a new switch --unregister for unregistering all
7271 registered entries at once. This is now invoked automatically at
7272 shutdown, so that binary formats registered with the "F" flag will
7273 not block clean file system unmounting.
7275 * systemd-notify's --pid= switch gained new values: "parent", "self",
7276 "auto" for controlling which PID to send to the service manager: the
7277 systemd-notify process' PID, or the one of the process invoking it.
7279 * systemd-logind's Session bus object learnt a new method call
7280 SetType() for temporarily updating the session type of an already
7281 allocated session. This is useful for upgrading tty sessions to
7282 graphical ones once a compositor is invoked.
7284 * systemd-socket-proxy gained a new switch --exit-idle-time= for
7285 configuring an exit-on-idle time.
7287 * systemd-repart's --empty= setting gained a new value "create". If
7288 specified a new empty regular disk image file is created under the
7289 specified name. Its size may be specified with the new --size=
7290 option. The latter is also supported without the "create" mode, in
7291 order to grow existing disk image files to the specified size. These
7292 two new options are useful when creating or manipulating disk images
7293 instead of operating on actual block devices.
7295 * systemd-repart drop-ins now support a new UUID= setting to control
7296 the UUID to assign to a newly created partition.
7298 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
7301 * systemd-repart's Label= setting now support the usual, simple
7302 specifier expansion.
7304 * systemd-homed's LUKS backend gained the ability to discard empty file
7305 system blocks automatically when the user logs out. This is enabled
7306 by default to ensure that home directories take minimal space when
7307 logged out but get full size guarantees when logged in. This may be
7308 controlled with the new --luks-offline-discard= switch to homectl.
7310 * If systemd-homed detects that /home/ is encrypted as a whole it will
7311 now default to the directory or subvolume backends instead of the
7312 LUKS backend, in order to avoid double encryption. The default
7313 storage and file system may now be configured explicitly, too, via
7314 the new /etc/systemd/homed.conf configuration file.
7316 * systemd-homed now supports unlocking home directories with FIDO2
7317 security tokens that support the 'hmac-secret' extension, in addition
7318 to the existing support for PKCS#11 security token unlocking
7319 support. Note that many recent hardware security tokens support both
7320 interfaces. The FIDO2 support is accessible via homectl's
7321 --fido2-device= option.
7323 * homectl's --pkcs11-uri= setting now accepts two special parameters:
7324 if "auto" is specified and only one suitable PKCS#11 security token
7325 is plugged in, its URL is automatically determined and enrolled for
7326 unlocking the home directory. If "list" is specified a brief table of
7327 suitable PKCS#11 security tokens is shown. Similar, the new
7328 --fido2-device= option also supports these two special values, for
7329 automatically selecting and listing suitable FIDO2 devices.
7331 * The /etc/crypttab tmp option now optionally takes an argument
7332 selecting the file system to use. Moreover, the default is now
7333 changed from ext2 to ext4.
7335 * There's a new /etc/crypttab option "keyfile-erase". If specified the
7336 key file listed in the same line is removed after use, regardless if
7337 volume activation was successful or not. This is useful if the key
7338 file is only acquired transiently at runtime and shall be erased
7339 before the system continues to boot.
7341 * There's also a new /etc/crypttab option "try-empty-password". If
7342 specified, before asking the user for a password it is attempted to
7343 unlock the volume with an empty password. This is useful for
7344 installing encrypted images whose password shall be set on first boot
7345 instead of at installation time.
7347 * systemd-cryptsetup will now attempt to load the keys to unlock
7348 volumes with automatically from files in
7349 /etc/cryptsetup-keys.d/<volume>.key and
7350 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
7352 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
7353 /etc/crypttab, during boot.
7355 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
7356 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
7359 * A new generator systemd-xdg-autostart-generator has been added. It
7360 generates systemd unit files from XDG autostart .desktop files, and
7361 may be used to let the systemd user instance manage services that are
7362 started automatically as part of the desktop session.
7364 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
7365 to query and change the firmware's 'Reboot Into Firmware Interface'
7368 * systemd-firstboot gained a new switch --kernel-command-line= that may
7369 be used to initialize the /etc/kernel/cmdline file of the image. It
7370 also gained a new switch --root-password-hashed= which is like
7371 --root-password= but accepts a pre-hashed UNIX password as
7372 argument. The new option --delete-root-password may be used to unset
7373 any password for the root user (dangerous!). The --root-shell= switch
7374 may be used to control the shell to use for the root account. A new
7375 --force option may be used to override any already set settings with
7376 the parameters specified on the command line (by default, the tool
7377 will not override what has already been set before, i.e. is purely
7380 * systemd-firstboot gained support for a new --image= switch, which is
7381 similar to --root= but accepts the path to a disk image file, on
7382 which it then operates.
7384 * A new sd-path.h API has been added to libsystemd. It provides a
7385 simple API for retrieving various search paths and primary
7386 directories for various resources.
7388 * A new call sd_notify_barrier() has been added to the sd-daemon.h
7389 API. The call will block until all previously sent sd_notify()
7390 messages have been processed by the service manager. This is useful
7391 to remove races caused by a process already having disappeared at the
7392 time a notification message is processed by the service manager,
7393 making correct attribution impossible. The systemd-notify tool will
7394 now make use of this call implicitly, but this can be turned off again
7395 via the new --no-block switch.
7397 * When sending a file descriptor (fd) to the service manager to keep
7398 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
7399 may be specified. If passed the service manager will refrain from
7400 poll()ing on the file descriptor. Traditionally (and when the
7401 parameter is not specified), the service manager will poll it for
7402 POLLHUP or POLLERR events, and immediately close the fds in that
7405 * The service manager (PID1) gained a new D-Bus method call
7406 SetShowStatus() which may be used to control whether it shall show
7407 boot-time status output on the console. This method has a similar
7408 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
7410 * The sd-bus API gained a number of convenience functions that take
7411 va_list arguments rather than "...". For example, there's now
7412 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
7413 it easier to build wrappers that accept variadic arguments and want
7414 to pass a ready va_list structure to sd-bus.
7416 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
7417 flag which alters how the userdata pointer to pass to the callbacks
7418 is determined. When the flag is set, the offset field is converted
7419 as-is into a pointer, without adding it to the object pointer the
7420 vtable is associated with.
7422 * sd-bus now exposes four new functions:
7423 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
7424 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
7425 validate strings to check if they qualify as various D-Bus concepts.
7427 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
7428 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
7429 that simplify adding argument names to D-Bus methods and signals.
7431 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
7433 * Various D-Bus APIs of systemd daemons now have man pages that
7434 document the methods, signals and properties.
7436 * The expectations on user/group name syntax are now documented in
7437 detail; documentation on how classic home directories may be
7438 converted into home directories managed by homed has been added;
7439 documentation regarding integration of homed/userdb functionality in
7440 desktops has been added:
7442 https://systemd.io/USER_NAMES
7443 https://systemd.io/CONVERTING_TO_HOMED
7444 https://systemd.io/USERDB_AND_DESKTOPS
7446 * Documentation for the on-disk Journal file format has been updated
7447 and has now moved to:
7449 https://systemd.io/JOURNAL_FILE_FORMAT
7451 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
7452 has been extended by a set of environment variables that expose
7453 select fields from the host's os-release file to the container
7454 payload. Similarly, host's os-release files can be mounted into the
7455 container underneath /run/host. Together, those mechanisms provide a
7456 standardized way to expose information about the host to the
7457 container payload. Both interfaces are implemented in systemd-nspawn.
7459 * All D-Bus services shipped in systemd now implement the generic
7460 LogControl1 D-Bus API which allows clients to change log level +
7461 target of the service during runtime.
7463 * Only relevant for developers: the mkosi.default symlink has been
7464 dropped from version control. Please create a symlink to one of the
7465 distribution-specific defaults in .mkosi/ based on your preference.
7467 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
7468 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
7469 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
7470 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
7471 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
7472 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
7473 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
7474 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
7475 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
7476 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
7477 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
7478 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
7479 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
7480 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
7481 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
7482 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
7483 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
7484 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
7485 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
7486 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
7487 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
7488 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
7489 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
7490 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
7491 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
7492 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
7493 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
7494 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
7495 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
7496 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
7497 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
7498 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
7499 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
7500 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
7501 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
7502 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
7503 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
7504 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
7506 – Warsaw, 2020-07-30
7510 * A new tool "systemd-repart" has been added, that operates as an
7511 idempotent declarative repartitioner for GPT partition tables.
7512 Specifically, a set of partitions that must or may exist can be
7513 configured via drop-in files, and during every boot the partition
7514 table on disk is compared with these files, creating missing
7515 partitions or growing existing ones based on configurable relative
7516 and absolute size constraints. The tool is strictly incremental,
7517 i.e. does not delete, shrink or move partitions, but only adds and
7518 grows them. The primary use-case is OS images that ship in minimized
7519 form, that on first boot are grown to the size of the underlying
7520 block device or augmented with additional partitions. For example,
7521 the root partition could be extended to cover the whole disk, or a
7522 swap or /home partitions could be added on first boot. It can also be
7523 used for systems that use an A/B update scheme but ship images with
7524 just the A partition, with B added on first boot. The tool is
7525 primarily intended to be run in the initrd, shortly before
7526 transitioning into the host OS, but can also be run after the
7527 transition took place. It automatically discovers the disk backing
7528 the root file system, and should hence not require any additional
7529 configuration besides the partition definition drop-ins. If no
7530 configuration drop-ins are present, no action is taken.
7532 * A new component "userdb" has been added, along with a small daemon
7533 "systemd-userdbd.service" and a client tool "userdbctl". The framework
7534 allows defining rich user and group records in a JSON format,
7535 extending on the classic "struct passwd" and "struct group"
7536 structures. Various components in systemd have been updated to
7537 process records in this format, including systemd-logind and
7538 pam-systemd. The user records are intended to be extensible, and
7539 allow setting various resource management, security and runtime
7540 parameters that shall be applied to processes and sessions of the
7541 user as they log in. This facility is intended to allow associating
7542 such metadata directly with user/group records so that they can be
7543 produced, extended and consumed in unified form. We hope that
7544 eventually frameworks such as sssd will generate records this way, so
7545 that for the first time resource management and various other
7546 per-user settings can be configured in LDAP directories and then
7547 provided to systemd (specifically to systemd-logind and pam-system)
7548 to apply on login. For further details see:
7550 https://systemd.io/USER_RECORD
7551 https://systemd.io/GROUP_RECORD
7552 https://systemd.io/USER_GROUP_API
7554 * A small new service systemd-homed.service has been added, that may be
7555 used to securely manage home directories with built-in encryption.
7556 The complete user record data is unified with the home directory,
7557 thus making home directories naturally migratable. Its primary
7558 back-end is based on LUKS volumes, but fscrypt, plain directories,
7559 and other storage schemes are also supported. This solves a couple of
7560 problems we saw with traditional ways to manage home directories, in
7561 particular when it comes to encryption. For further discussion of
7562 this, see the video of Lennart's talk at AllSystemsGo! 2019:
7564 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
7566 For further details about the format and expectations on home
7567 directories this new daemon makes, see:
7569 https://systemd.io/HOME_DIRECTORY
7571 * systemd-journald is now multi-instantiable. In addition to the main
7572 instance systemd-journald.service there's now a template unit
7573 systemd-journald@.service, with each instance defining a new named
7574 log 'namespace' (whose name is specified via the instance part of the
7575 unit name). A new unit file setting LogNamespace= has been added,
7576 taking such a namespace name, that assigns services to the specified
7577 log namespaces. As each log namespace is serviced by its own
7578 independent journal daemon, this functionality may be used to improve
7579 performance and increase isolation of applications, at the price of
7580 losing global message ordering. Each instance of journald has a
7581 separate set of configuration files, with possibly different disk
7582 usage limitations and other settings.
7584 journalctl now takes a new option --namespace= to show logs from a
7585 specific log namespace. The sd-journal.h API gained
7586 sd_journal_open_namespace() for opening the log stream of a specific
7587 log namespace. systemd-journald also gained the ability to exit on
7588 idle, which is useful in the context of log namespaces, as this means
7589 log daemons for log namespaces can be activated automatically on
7590 demand and will stop automatically when no longer used, minimizing
7593 * When systemd-tmpfiles copies a file tree using the 'C' line type it
7594 will now label every copied file according to the SELinux database.
7596 * When systemd/PID 1 detects it is used in the initrd it will now boot
7597 into initrd.target rather than default.target by default. This should
7598 make it simpler to build initrds with systemd as for many cases the
7599 only difference between a host OS image and an initrd image now is
7600 the presence of the /etc/initrd-release file.
7602 * A new kernel command line option systemd.cpu_affinity= is now
7603 understood. It's equivalent to the CPUAffinity= option in
7604 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
7605 itself and the default for all other processes.
7607 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
7608 equivalent), the SELinux database is now reloaded, ensuring that
7609 sockets and other file system objects are generated taking the new
7610 database into account.
7612 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
7613 "quiet" has been changed to imply that instead of
7614 "systemd.show-status=auto". In this mode, only messages about errors
7615 and significant delays in boot are shown on the console.
7617 * The sd-event.h API gained native support for the new Linux "pidfd"
7618 concept. This permits watching processes using file descriptors
7619 instead of PID numbers, which fixes a number of races and makes
7620 process supervision more robust and efficient. All of systemd's
7621 components will now use pidfds if the kernel supports it for process
7622 watching, with the exception of PID 1 itself, unfortunately. We hope
7623 to move PID 1 to exclusively using pidfds too eventually, but this
7624 requires some more kernel work first. (Background: PID 1 watches
7625 processes using waitid() with the P_ALL flag, and that does not play
7626 together nicely with pidfds yet.)
7628 * Closely related to this, the sd-event.h API gained two new calls
7629 sd_event_source_send_child_signal() (for sending a signal to a
7630 watched process) and sd_event_source_get_child_process_own() (for
7631 marking a process so that it is killed automatically whenever the
7632 event source watching it is freed).
7634 * systemd-networkd gained support for configuring Token Bucket Filter
7635 (TBF) parameters in its qdisc configuration support. Similarly,
7636 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
7637 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
7639 * systemd-networkd gained support for Intermediate Functional Block
7640 (IFB) network devices.
7642 * systemd-networkd gained support for configuring multi-path IP routes,
7643 using the new MultiPathRoute= setting in the [Route] section.
7645 * systemd-networkd's DHCPv4 client has been updated to support a new
7646 SendDecline= option. If enabled, duplicate address detection is done
7647 after a DHCP offer is received from the server. If a conflict is
7648 detected, the address is declined. The DHCPv4 client also gained
7649 support for a new RouteMTUBytes= setting that allows to configure the
7650 MTU size to be used for routes generated from DHCPv4 leases.
7652 * The PrefixRoute= setting in systemd-networkd's [Address] section of
7653 .network files has been deprecated, and replaced by AddPrefixRoute=,
7654 with its sense inverted.
7656 * The Gateway= setting of [Route] sections of .network files gained
7657 support for a special new value "_dhcp". If set, the configured
7658 static route uses the gateway host configured via DHCP.
7660 * New User= and SuppressPrefixLength= settings have been implemented
7661 for the [RoutingPolicyRule] section of .network files to configure
7662 source routing based on UID ranges and prefix length, respectively.
7664 * The Type= match property of .link files has been generalized to
7665 always match the device type shown by 'networkctl status', even for
7666 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
7669 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
7670 D-Bus message object as "sensitive". Those objects are erased from
7671 memory when they are freed. This concept is intended to be used for
7672 messages that contain security sensitive data. A new flag
7673 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
7674 in sd-bus vtables, causing any incoming and outgoing messages of
7675 those methods to be implicitly marked as "sensitive".
7677 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
7678 contents of a message (or parts thereof) to standard output for
7681 * systemd-sysusers gained support for creating users with the primary
7682 group named differently than the user.
7684 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
7685 gained support for growing XFS partitions. Previously it supported
7686 only ext4 and btrfs partitions.
7688 * The support for /etc/crypttab gained a new x-initrd.attach option. If
7689 set, the specified encrypted volume is unlocked already in the
7690 initrd. This concept corresponds to the x-initrd.mount option in
7693 * systemd-cryptsetup gained native support for unlocking encrypted
7694 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
7695 encryption of volumes to YubiKeys. This is exposed in the new
7696 pkcs11-uri= option in /etc/crypttab.
7698 * The /etc/fstab support in systemd now supports two new mount options
7699 x-systemd.{required,wanted}-by=, for explicitly configuring the units
7700 that the specified mount shall be pulled in by, in place of
7701 the usual local-fs.target/remote-fs.target.
7703 * The https://systemd.io/ web site has been relaunched, directly
7704 populated with most of the documentation included in the systemd
7705 repository. systemd also acquired a new logo, thanks to Tobias
7708 * systemd-udevd gained support for managing "alternative" network
7709 interface names, as supported by new Linux kernels. For the first
7710 time this permits assigning multiple (and longer!) names to a network
7711 interface. systemd-udevd will now by default assign the names
7712 generated via all supported naming schemes to each interface. This
7713 may be further tweaked with .link files and the AlternativeName= and
7714 AlternativeNamesPolicy= settings. Other components of systemd have
7715 been updated to support the new alternative names wherever
7716 appropriate. For example, systemd-nspawn will now generate
7717 alternative interface names for the host-facing side of container
7718 veth links based on the full container name without truncation.
7720 * systemd-nspawn interface naming logic has been updated in another way
7721 too: if the main interface name (i.e. as opposed to new-style
7722 "alternative" names) based on the container name is truncated, a
7723 simple hashing scheme is used to give different interface names to
7724 multiple containers whose names all begin with the same prefix. Since
7725 this changes the primary interface names pointing to containers if
7726 truncation happens, the old scheme may still be requested by
7727 selecting an older naming scheme, via the net.naming_scheme= kernel
7728 command line option.
7730 * PrivateUsers= in service files now works in services run by the
7731 systemd --user per-user instance of the service manager.
7733 * A new per-service sandboxing option ProtectClock= has been added that
7734 locks down write access to the system clock. It takes away device
7735 node access to /dev/rtc as well as the system calls that set the
7736 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
7737 Note that this option does not affect access to auxiliary services
7738 that allow changing the clock, for example access to
7741 * The systemd-id128 tool gained a new "show" verb for listing or
7742 resolving a number of well-known UUIDs/128-bit IDs, currently mostly
7743 GPT partition table types.
7745 * The Discoverable Partitions Specification has been updated to support
7746 /var and /var/tmp partition discovery. Support for this has been
7747 added to systemd-gpt-auto-generator. For details see:
7749 https://systemd.io/DISCOVERABLE_PARTITIONS
7751 * "systemctl list-unit-files" has been updated to show a new column
7752 with the suggested enablement state based on the vendor preset files
7753 for the respective units.
7755 * "systemctl" gained a new option "--with-dependencies". If specified
7756 commands such as "systemctl status" or "systemctl cat" will now show
7757 all specified units along with all units they depend on.
7759 * networkctl gained support for showing per-interface logs in its
7762 * systemd-networkd-wait-online gained support for specifying the maximum
7763 operational state to wait for, and to wait for interfaces to
7766 * The [Match] section of .link and .network files now supports a new
7767 option PermanentMACAddress= which may be used to check against the
7768 permanent MAC address of a network device even if a randomized MAC
7771 * The [TrafficControlQueueingDiscipline] section in .network files has
7772 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
7773 dropped from the individual setting names.
7775 * Any .link and .network files that have an empty [Match] section (this
7776 also includes empty and commented-out files) will now be
7777 rejected. systemd-udev and systemd-networkd started warning about
7778 such files in version 243.
7780 * systemd-logind will now validate access to the operation of changing
7781 the virtual terminal via a polkit action. By default, only users
7782 with at least one session on a local VT are granted permission.
7784 * When systemd sets up PAM sessions that invoked service processes
7785 shall run in, the pam_setcred() API is now invoked, thus permitting
7786 PAM modules to set additional credentials for the processes.
7788 * portablectl attach/detach verbs now accept --now and --enable options
7789 to combine attachment with enablement and invocation, or detachment
7790 with stopping and disablement.
7792 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
7793 fixed, which in turn exposed bugs in unit configuration of services
7794 which have Type=oneshot and should only run once, but do not have
7795 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
7796 service may be started again after exiting successfully, for example
7797 as a dependency in another transaction. Affected services included
7798 some internal systemd services (most notably
7799 systemd-vconsole-setup.service, which was updated to have
7800 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
7801 plymouth has been suitably updated or patched before upgrading to
7802 this systemd release. See
7803 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
7804 additional discussion.
7806 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
7807 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
7808 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
7809 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
7810 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
7811 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
7812 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
7813 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
7814 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
7815 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
7816 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
7817 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
7818 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
7819 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
7820 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
7821 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
7822 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
7823 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
7824 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
7825 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
7826 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
7827 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
7828 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
7829 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
7832 – Warsaw, 2020-03-06
7836 * Support for the cpuset cgroups v2 controller has been added.
7837 Processes may be restricted to specific CPUs using the new
7838 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
7839 AllowedMemoryNodes= setting.
7841 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
7842 now be configured using a new RestartKillSignal= setting. This
7843 allows units which signals to request termination to implement
7844 different behaviour when stopping in preparation for a restart.
7846 * "systemctl clean" may now be used also for socket, mount, and swap
7849 * systemd will also read configuration options from the EFI variable
7850 SystemdOptions. This may be used to configure systemd behaviour when
7851 modifying the kernel command line is inconvenient, but configuration
7852 on disk is read too late, for example for the options related to
7853 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
7854 set the EFI variable.
7856 * systemd will now disable printk ratelimits in early boot. This should
7857 allow us to capture more logs from the early boot phase where normal
7858 storage is not available and the kernel ring buffer is used for
7859 logging. Configuration on the kernel command line has higher priority
7860 and overrides the systemd setting.
7862 systemd programs which log to /dev/kmsg directly use internal
7863 ratelimits to prevent runaway logging. (Normally this is only used
7864 during early boot, so in practice this change has very little
7867 * Unit files now support top level dropin directories of the form
7868 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
7869 that affects all corresponding unit files.
7871 * systemctl gained support for 'stop --job-mode=triggering' which will
7872 stop the specified unit and any units which could trigger it.
7874 * Unit status display now includes units triggering and triggered by
7875 the unit being shown.
7877 * The RuntimeMaxSec= setting is now supported by scopes, not just
7878 .service units. This is particularly useful for PAM sessions which
7879 create a scope unit for the user login. systemd.runtime_max_sec=
7880 setting may used with the pam_systemd module to limit the duration
7881 of the PAM session, for example for time-limited logins.
7883 * A new @pkey system call group is now defined to make it easier to
7884 allow-list memory protection syscalls for containers and services
7885 which need to use them.
7887 * systemd-udevd: removed the 30s timeout for killing stale workers on
7888 exit. systemd-udevd now waits for workers to finish. The hard-coded
7889 exit timeout of 30s was too short for some large installations, where
7890 driver initialization could be prematurely interrupted during initrd
7891 processing if the root file system had been mounted and init was
7892 preparing to switch root. If udevd is run without systemd and workers
7893 are hanging while udevd receives an exit signal, udevd will now exit
7894 when udev.event_timeout is reached for the last hanging worker. With
7895 systemd, the exit timeout can additionally be configured using
7896 TimeoutStopSec= in systemd-udevd.service.
7898 * udev now provides a program (fido_id) that identifies FIDO CTAP1
7899 ("U2F")/CTAP2 security tokens based on the usage declared in their
7900 report and descriptor and outputs suitable environment variables.
7901 This replaces the externally maintained allow lists of all known
7902 security tokens that were used previously.
7904 * Automatically generated autosuspend udev rules for allow-listed
7905 devices have been imported from the Chromium OS project. This should
7906 improve power saving with many more devices.
7908 * udev gained a new "CONST{key}=value" setting that allows matching
7909 against system-wide constants without forking a helper binary.
7910 Currently "arch" and "virt" keys are supported.
7912 * udev now opens CDROMs in non-exclusive mode when querying their
7913 capabilities. This should fix issues where other programs trying to
7914 use the CDROM cannot gain access to it, but carries a risk of
7915 interfering with programs writing to the disk, if they did not open
7916 the device in exclusive mode as they should.
7918 * systemd-networkd does not create a default route for IPv4 link local
7919 addressing anymore. The creation of the route was unexpected and was
7920 breaking routing in various cases, but people who rely on it being
7921 created implicitly will need to adjust. Such a route may be requested
7922 with DefaultRouteOnDevice=yes.
7924 Similarly, systemd-networkd will not assign a link-local IPv6 address
7925 when IPv6 link-local routing is not enabled.
7927 * Receive and transmit buffers may now be configured on links with
7928 the new RxBufferSize= and TxBufferSize= settings.
7930 * systemd-networkd may now advertise additional IPv6 routes. A new
7931 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
7934 * systemd-networkd may now configure "next hop" routes using the
7935 [NextHop] section and Gateway= and Id= settings.
7937 * systemd-networkd will now retain DHCP config on restarts by default
7938 (but this may be overridden using the KeepConfiguration= setting).
7939 The default for SendRelease= has been changed to true.
7941 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
7942 received from the server.
7944 The client will use the received SIP server list if UseSIP=yes is
7947 The client may be configured to request specific options from the
7948 server using a new RequestOptions= setting.
7950 The client may be configured to send arbitrary options to the server
7951 using a new SendOption= setting.
7953 A new IPServiceType= setting has been added to configure the "IP
7954 service type" value used by the client.
7956 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
7957 request prefix hints in the DHCPv6 solicitation.
7959 * The DHCPv4 server may be configured to send arbitrary options using
7960 a new SendOption= setting.
7962 * The DHCPv4 server may now be configured to emit SIP server list using
7963 the new EmitSIP= and SIP= settings.
7965 * systemd-networkd and networkctl may now renew DHCP leases on demand.
7966 networkctl has a new 'networkctl renew' verb.
7968 * systemd-networkd may now reconfigure links on demand. networkctl
7969 gained two new verbs: "reload" will reload the configuration, and
7970 "reconfigure DEVICE…" will reconfigure one or more devices.
7972 * .network files may now match on SSID and BSSID of a wireless network,
7973 i.e. the access point name and hardware address using the new SSID=
7974 and BSSID= options. networkctl will display the current SSID and
7975 BSSID for wireless links.
7977 .network files may also match on the wireless network type using the
7978 new WLANInterfaceType= option.
7980 * systemd-networkd now includes default configuration that enables
7981 link-local addressing when connected to an ad-hoc wireless network.
7983 * systemd-networkd may configure the Traffic Control queueing
7984 disciplines in the kernel using the new
7985 [TrafficControlQueueingDiscipline] section and Parent=,
7986 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
7987 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
7988 NetworkEmulatorDuplicateRate= settings.
7990 * systemd-tmpfiles gained a new w+ setting to append to files.
7992 * systemd-analyze dump will now report when the memory configuration in
7993 the kernel does not match what systemd has configured (usually,
7994 because some external program has modified the kernel configuration
7997 * systemd-analyze gained a new --base-time= switch instructs the
7998 'calendar' verb to resolve times relative to that timestamp instead
7999 of the present time.
8001 * journalctl --update-catalog now produces deterministic output (making
8002 reproducible image builds easier).
8004 * A new devicetree-overlay setting is now documented in the Boot Loader
8007 * The default value of the WatchdogSec= setting used in systemd
8008 services (the ones bundled with the project itself) may be set at
8009 configuration time using the -Dservice-watchdog= setting. If set to
8010 empty, the watchdogs will be disabled.
8012 * systemd-resolved validates IP addresses in certificates now when GnuTLS
8015 * libcryptsetup >= 2.0.1 is now required.
8017 * A configuration option -Duser-path= may be used to override the $PATH
8018 used by the user service manager. The default is again to use the same
8019 path as the system manager.
8021 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
8022 outputting the 128-bit IDs in UUID format (i.e. in the "canonical
8025 * Service units gained a new sandboxing option ProtectKernelLogs= which
8026 makes sure the program cannot get direct access to the kernel log
8027 buffer anymore, i.e. the syslog() system call (not to be confused
8028 with the API of the same name in libc, which is not affected), the
8029 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
8030 inaccessible to the service. It's recommended to enable this setting
8031 for all services that should not be able to read from or write to the
8032 kernel log buffer, which are probably almost all.
8034 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
8035 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
8036 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
8037 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
8038 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
8039 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
8040 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
8041 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
8042 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
8043 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
8044 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
8045 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
8046 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
8047 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
8048 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
8049 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
8050 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
8051 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
8052 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
8053 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
8054 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
8055 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
8056 Zach Smith, Zbigniew Jędrzejewski-Szmek
8058 – Warsaw, 2019-11-29
8062 * This release enables unprivileged programs (i.e. requiring neither
8063 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
8064 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
8065 kernel for the whole UNIX group range, i.e. all processes. This
8066 change should be reasonably safe, as the kernel support for it was
8067 specifically implemented to allow safe access to ICMP Echo for
8068 processes lacking any privileges. If this is not desirable, it can be
8069 disabled again by setting the parameter to "1 0".
8071 * Previously, filters defined with SystemCallFilter= would have the
8072 effect that any calling of an offending system call would terminate
8073 the calling thread. This behaviour never made much sense, since
8074 killing individual threads of unsuspecting processes is likely to
8075 create more problems than it solves. With this release the default
8076 action changed from killing the thread to killing the whole
8077 process. For this to work correctly both a kernel version (>= 4.14)
8078 and a libseccomp version (>= 2.4.0) supporting this new seccomp
8079 action is required. If an older kernel or libseccomp is used the old
8080 behaviour continues to be used. This change does not affect any
8081 services that have no system call filters defined, or that use
8082 SystemCallErrorNumber= (and thus see EPERM or another error instead
8083 of being killed when calling an offending system call). Note that
8084 systemd documentation always claimed that the whole process is
8085 killed. With this change behaviour is thus adjusted to match the
8088 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
8089 4194304 by default, i.e. the full 22bit range the kernel allows, up
8090 from the old 16-bit range. This should improve security and
8091 robustness, as PID collisions are made less likely (though certainly
8092 still possible). There are rumours this might create compatibility
8093 problems, though at this moment no practical ones are known to
8094 us. Downstream distributions are hence advised to undo this change in
8095 their builds if they are concerned about maximum compatibility, but
8096 for everybody else we recommend leaving the value bumped. Besides
8097 improving security and robustness this should also simplify things as
8098 the maximum number of allowed concurrent tasks was previously bounded
8099 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
8100 only a single knob is left ("kernel.threads-max"). There have been
8101 concerns that usability is affected by this change because larger PID
8102 numbers are harder to type, but we believe the change from 5 digits
8103 to 7 digits doesn't hamper usability.
8105 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
8106 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
8107 hierarchically set default memory protection values for a particular
8108 subtree of the unit hierarchy.
8110 * Memory protection directives can now take a value of zero, allowing
8111 explicit opting out of a default value propagated by an ancestor.
8113 * systemd now defaults to the "unified" cgroup hierarchy setup during
8114 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
8115 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
8116 change reflects the fact that cgroupsv2 support has matured
8117 substantially in both systemd and in the kernel, and is clearly the
8118 way forward. Downstream production distributions might want to
8119 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
8120 their builds as unfortunately the popular container managers have not
8121 caught up with the kernel API changes.
8123 * Man pages are not built by default anymore (html pages were already
8124 disabled by default), to make development builds quicker. When
8125 building systemd for a full installation with documentation, meson
8126 should be called with -Dman=true and/or -Dhtml=true as appropriate.
8127 The default was changed based on the assumption that quick one-off or
8128 repeated development builds are much more common than full optimized
8129 builds for installation, and people need to pass various other
8130 options to when doing "proper" builds anyway, so the gain from making
8131 development builds quicker is bigger than the one time disruption for
8134 Two scripts are created in the *build* directory to generate and
8135 preview man and html pages on demand, e.g.:
8137 build/man/man systemctl
8138 build/man/html systemd.index
8140 * libidn2 is used by default if both libidn2 and libidn are installed.
8141 Please use -Dlibidn=true if libidn is preferred.
8143 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
8144 big-endian machines. Before, bytes were written and read in native
8145 machine order as exposed by the native libc __cpu_mask interface.
8146 Now, little-endian order is always used (CPUs 0–7 are described by
8147 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
8148 This change fixes D-Bus calls that cross endianness boundary.
8150 The presentation format used for CPUAffinity= by "systemctl show" and
8151 "systemd-analyze dump" is changed to present CPU indices instead of
8152 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
8153 shown as CPUAffinity=03000000000000000000000000000… (on
8154 little-endian) or CPUAffinity=00000000000000300000000000000… (on
8155 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
8156 input format. The maximum integer that will be printed in the new
8157 format is 8191 (four digits), while the old format always used a very
8158 long number (with the length varying by architecture), so they can be
8159 unambiguously distinguished.
8161 * /usr/sbin/halt.local is no longer supported. Implementation in
8162 distributions was inconsistent and it seems this functionality was
8165 To replace this functionality, users should:
8166 - either define a new unit and make it a dependency of final.target
8167 (systemctl add-wants final.target my-halt-local.service)
8168 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
8169 and ensure that it accepts "halt", "poweroff", "reboot", and
8170 "kexec" as an argument, see the description in systemd-shutdown(8).
8172 * When a [Match] section in .link or .network file is empty (contains
8173 no match patterns), a warning will be emitted. Please add any "match
8174 all" pattern instead, e.g. OriginalName=* or Name=* in case all
8175 interfaces should really be matched.
8177 * A new setting NUMAPolicy= may be used to set process memory
8178 allocation policy. This setting can be specified in
8179 /etc/systemd/system.conf and hence will set the default policy for
8180 PID1. The default policy can be overridden on a per-service
8181 basis. The related setting NUMAMask= is used to specify NUMA node
8182 mask that should be associated with the selected policy.
8184 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
8185 generates when processes it manages are reaching their memory limits,
8186 and will place their units in a special state, and optionally kill or
8187 stop the whole unit.
8189 * The service manager will now expose bus properties for the IO
8190 resources used by units. This information is also shown in "systemctl
8191 status" now (for services that have IOAccounting=yes set). Moreover,
8192 the IO accounting data is included in the resource log message
8193 generated whenever a unit stops.
8195 * Units may now configure an explicit timeout to wait for when killed
8196 with SIGABRT, for example when a service watchdog is hit. Previously,
8197 the regular TimeoutStopSec= timeout was applied in this case too —
8198 now a separate timeout may be set using TimeoutAbortSec=.
8200 * Services may now send a special WATCHDOG=trigger message with
8201 sd_notify() to trigger an immediate "watchdog missed" event, and thus
8202 trigger service termination. This is useful both for testing watchdog
8203 handling, but also for defining error paths in services, that shall
8204 be handled the same way as watchdog events.
8206 * There are two new per-unit settings IPIngressFilterPath= and
8207 IPEgressFilterPath= which allow configuration of a BPF program
8208 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
8209 to apply to the IP packet ingress/egress path of all processes of a
8210 unit. This is useful to allow running systemd services with BPF
8211 programs set up externally.
8213 * systemctl gained a new "clean" verb for removing the state, cache,
8214 runtime or logs directories of a service while it is terminated. The
8215 new verb may also be used to remove the state maintained on disk for
8216 timer units that have Persistent= configured.
8218 * During the last phase of shutdown systemd will now automatically
8219 increase the log level configured in the "kernel.printk" sysctl so
8220 that any relevant loggable events happening during late shutdown are
8221 made visible. Previously, loggable events happening so late during
8222 shutdown were generally lost if the "kernel.printk" sysctl was set to
8223 high thresholds, as regular logging daemons are terminated at that
8224 time and thus nothing is written to disk.
8226 * If processes terminated during the last phase of shutdown do not exit
8227 quickly systemd will now show their names after a short time, to make
8228 debugging easier. After a longer timeout they are forcibly killed,
8231 * journalctl (and the other tools that display logs) will now highlight
8232 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
8233 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
8234 are now shown in blue color, to separate them visually from regular
8235 logs. References to configuration files are now turned into clickable
8236 links on terminals that support that.
8238 * systemd-journald will now stop logging to /var/log/journal during
8239 shutdown when /var/ is on a separate mount, so that it can be
8240 unmounted safely during shutdown.
8242 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
8244 * systemd-resolved "Cache=" configuration option in resolved.conf has
8245 been extended to also accept the 'no-negative' value. Previously,
8246 only a boolean option was allowed (yes/no), having yes as the
8247 default. If this option is set to 'no-negative', negative answers are
8248 not cached while the old cache heuristics are used positive answers.
8249 The default remains unchanged.
8251 * The predictable naming scheme for network devices now supports
8252 generating predictable names for "netdevsim" devices.
8254 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
8257 Those two changes form a new net.naming_scheme= entry. Distributions
8258 which want to preserve naming stability may want to set the
8259 -Ddefault-net-naming-scheme= configuration option.
8261 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
8262 interfaces natively.
8264 * systemd-networkd's bridge FDB support now allows configuration of a
8265 destination address for each entry (Destination=), as well as the
8266 VXLAN VNI (VNI=), as well as an option to declare what an entry is
8267 associated with (AssociatedWith=).
8269 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
8270 option for configuring the maximum number of DHCP lease requests. It
8271 also learnt a new BlackList= option for deny-listing DHCP servers (a
8272 similar setting has also been added to the IPv6 RA client), as well
8273 as a SendRelease= option for configuring whether to send a DHCP
8274 RELEASE message when terminating.
8276 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
8277 separately in the [DHCPv4] and [DHCPv6] sections.
8279 * systemd-networkd's DHCP support will now optionally create an
8280 implicit host route to the DNS server specified in the DHCP lease, in
8281 addition to the routes listed explicitly in the lease. This should
8282 ensure that in multi-homed systems DNS traffic leaves the systems on
8283 the interface that acquired the DNS server information even if other
8284 routes such as default routes exist. This behaviour may be turned on
8285 with the new RoutesToDNS= option.
8287 * systemd-networkd's VXLAN support gained a new option
8288 GenericProtocolExtension= for enabling VXLAN Generic Protocol
8289 Extension support, as well as IPDoNotFragment= for setting the IP
8290 "Don't fragment" bit on outgoing packets. A similar option has been
8291 added to the GENEVE support.
8293 * In systemd-networkd's [Route] section you may now configure
8294 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
8295 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
8296 propagation. The Type= setting now supports local, broadcast,
8297 anycast, multicast, any, xresolve routes, too.
8299 * systemd-networkd's [Network] section learnt a new option
8300 DefaultRouteOnDevice= for automatically configuring a default route
8301 onto the network device.
8303 * systemd-networkd's bridging support gained two new options ProxyARP=
8304 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
8305 MulticastRouter= for configuring multicast routing behaviour. A new
8306 option MulticastIGMPVersion= may be used to change bridge's multicast
8307 Internet Group Management Protocol (IGMP) version.
8309 * systemd-networkd's FooOverUDP support gained the ability to configure
8310 local and peer IP addresses via Local= and Peer=. A new option
8311 PeerPort= may be used to configure the peer's IP port.
8313 * systemd-networkd's TUN support gained a new setting VnetHeader= for
8314 tweaking Generic Segment Offload support.
8316 * The address family for policy rules may be specified using the new
8317 Family= option in the [RoutingPolicyRule] section.
8319 * networkctl gained a new "delete" command for removing virtual network
8320 devices, as well as a new "--stats" switch for showing device
8323 * networkd.conf gained a new setting SpeedMeter= and
8324 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
8325 measured speed may be shown by 'networkctl status'.
8327 * "networkctl status" now displays MTU and queue lengths, and more
8328 detailed information about VXLAN and bridge devices.
8330 * systemd-networkd's .network and .link files gained a new Property=
8331 setting in the [Match] section, to match against devices with
8332 specific udev properties.
8334 * systemd-networkd's tunnel support gained a new option
8335 AssignToLoopback= for selecting whether to use the loopback device
8336 "lo" as underlying device.
8338 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
8339 been renamed to LinkLayerAddress=, and it now allows configuration of
8342 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
8343 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
8344 IPv6 configuration (static or DHCPv6) was found for a given
8345 interface. It will not touch the sysctl otherwise.
8347 * The order of entries is $PATH used by the user manager instance was
8348 changed to put bin/ entries before the corresponding sbin/ entries.
8349 It is recommended to not rely on this order, and only ever have one
8350 binary with a given name in the system paths under /usr.
8352 * A new tool systemd-network-generator has been added that may generate
8353 .network, .netdev and .link files from IP configuration specified on
8354 the kernel command line in the format used by Dracut.
8356 * The CriticalConnection= setting in .network files is now deprecated,
8357 and replaced by a new KeepConfiguration= setting which allows more
8358 detailed configuration of the IP configuration to keep in place.
8360 * systemd-analyze gained a few new verbs:
8362 - "systemd-analyze timestamp" parses and converts timestamps. This is
8363 similar to the existing "systemd-analyze calendar" command which
8364 does the same for recurring calendar events.
8366 - "systemd-analyze timespan" parses and converts timespans (i.e.
8367 durations as opposed to points in time).
8369 - "systemd-analyze condition" will parse and test ConditionXYZ=
8372 - "systemd-analyze exit-status" will parse and convert exit status
8373 codes to their names and back.
8375 - "systemd-analyze unit-files" will print a list of all unit
8376 file paths and unit aliases.
8378 * SuccessExitStatus=, RestartPreventExitStatus=, and
8379 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
8380 is equivalent to "65"). Those exit status name mappings may be
8381 displayed with the systemd-analyze exit-status verb describe above.
8383 * systemd-logind now exposes a per-session SetBrightness() bus call,
8384 which may be used to securely change the brightness of a kernel
8385 brightness device, if it belongs to the session's seat. By using this
8386 call unprivileged clients can make changes to "backlight" and "leds"
8387 devices securely with strict requirements on session membership.
8388 Desktop environments may use this to generically make brightness
8389 changes to such devices without shipping private SUID binaries or
8390 udev rules for that purpose.
8392 * "udevadm info" gained a --wait-for-initialization switch to wait for
8393 a device to be initialized.
8395 * systemd-hibernate-resume-generator will now look for resumeflags= on
8396 the kernel command line, which is similar to rootflags= and may be
8397 used to configure device timeout for the hibernation device.
8399 * sd-event learnt a new API call sd_event_source_disable_unref() for
8400 disabling and unref'ing an event source in a single function. A
8401 related call sd_event_source_disable_unrefp() has been added for use
8402 with gcc's cleanup extension.
8404 * The sd-id128.h public API gained a new definition
8405 SD_ID128_UUID_FORMAT_STR for formatting a 128-bit ID in UUID format
8408 * "busctl introspect" gained a new switch --xml-interface for dumping
8409 XML introspection data unmodified.
8411 * PID 1 may now show the unit name instead of the unit description
8412 string in its status output during boot. This may be configured in
8413 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
8414 kernel command line option systemd.status_unit_format=.
8416 * PID 1 now understands a new option KExecWatchdogSec= in
8417 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
8418 Previously watchdog functionality was only available for regular
8419 reboots. The new setting defaults to off, because we don't know in
8420 the general case if the watchdog will be reset after kexec (some
8421 drivers do reset it, but not all), and the new userspace might not be
8422 configured to handle the watchdog.
8424 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
8425 RebootWatchdogSec= to more clearly communicate what it is about. The
8426 old name is still accepted for compatibility.
8428 * The systemd.debug_shell kernel command line option now optionally
8429 takes a tty name to spawn the debug shell on, which allows a
8430 different tty to be selected than the built-in default.
8432 * Service units gained a new ExecCondition= setting which will run
8433 before ExecStartPre= and either continue execution of the unit (for
8434 clean exit codes), stop execution without marking the unit failed
8435 (for exit codes 1 through 254), or stop execution and fail the unit
8436 (for exit code 255 or abnormal termination).
8438 * A new service systemd-pstore.service has been added that pulls data
8439 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
8442 * timedatectl gained new verbs for configuring per-interface NTP
8443 service configuration for systemd-timesyncd.
8445 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
8446 2019. (You can set non-UTF-8 locales though, if you know their name.)
8448 * If variable assignments in sysctl.d/ files are prefixed with "-" any
8449 failures to apply them are now ignored.
8451 * systemd-random-seed.service now optionally credits entropy when
8452 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
8453 true for the service to enable this behaviour, but please consult the
8454 documentation first, since this comes with a couple of caveats.
8456 * systemd-random-seed.service is now a synchronization point for full
8457 initialization of the kernel's entropy pool. Services that require
8458 /dev/urandom to be correctly initialized should be ordered after this
8461 * The systemd-boot boot loader has been updated to optionally maintain
8462 a random seed file in the EFI System Partition (ESP). During the boot
8463 phase, this random seed is read and updated with a new seed
8464 cryptographically derived from it. Another derived seed is passed to
8465 the OS. The latter seed is then credited to the kernel's entropy pool
8466 very early during userspace initialization (from PID 1). This allows
8467 systems to boot up with a fully initialized kernel entropy pool from
8468 earliest boot on, and thus entirely removes all entropy pool
8469 initialization delays from systems using systemd-boot. Special care
8470 is taken to ensure different seeds are derived on system images
8471 replicated to multiple systems. "bootctl status" will show whether
8472 a seed was received from the boot loader.
8474 * bootctl gained two new verbs:
8476 - "bootctl random-seed" will generate the file in ESP and an EFI
8477 variable to allow a random seed to be passed to the OS as described
8480 - "bootctl is-installed" checks whether systemd-boot is currently
8483 * bootctl will warn if it detects that boot entries are misconfigured
8484 (for example if the kernel image was removed without purging the
8487 * A new document has been added describing systemd's use and support
8488 for the kernel's entropy pool subsystem:
8490 https://systemd.io/RANDOM_SEEDS
8492 * When the system is hibernated the swap device to write the
8493 hibernation image to is now automatically picked from all available
8494 swap devices, preferring the swap device with the highest configured
8495 priority over all others, and picking the device with the most free
8496 space if there are multiple devices with the highest priority.
8498 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
8499 option that permits selecting the timeout how long to wait for a
8500 device with an encryption key before asking for the password.
8502 * IOWeight= has learnt to properly set the IO weight when using the
8503 BFQ scheduler officially found in kernels 5.0+.
8505 * A new mailing list has been created for reporting of security issues:
8506 systemd-security@redhat.com. For mode details, see
8507 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
8509 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
8510 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
8511 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
8512 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
8513 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
8514 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
8515 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
8516 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
8517 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
8518 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
8519 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
8520 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
8521 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
8522 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
8523 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
8524 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
8525 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
8526 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
8527 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
8528 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
8529 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
8530 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
8531 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
8532 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
8533 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
8534 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
8535 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
8536 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
8537 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
8538 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
8540 – Camerino, 2019-09-03
8544 * In .link files, MACAddressPolicy=persistent (the default) is changed
8545 to cover more devices. For devices like bridges, tun, tap, bond, and
8546 similar interfaces that do not have other identifying information,
8547 the interface name is used as the basis for persistent seed for MAC
8548 and IPv4LL addresses. The way that devices that were handled
8549 previously is not changed, and this change is about covering more
8550 devices then previously by the "persistent" policy.
8552 MACAddressPolicy=random may be used to force randomized MACs and
8553 IPv4LL addresses for a device if desired.
8555 Hint: the log output from udev (at debug level) was enhanced to
8556 clarify what policy is followed and which attributes are used.
8557 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
8558 may be used to view this.
8560 Hint: if a bridge interface is created without any slaves, and gains
8561 a slave later, then now the bridge does not inherit slave's MAC.
8562 To inherit slave's MAC, for example, create the following file:
8564 # /etc/systemd/network/98-bridge-inherit-mac.link
8569 MACAddressPolicy=none
8572 * The .device units generated by systemd-fstab-generator and other
8573 generators do not automatically pull in the corresponding .mount unit
8574 as a Wants= dependency. This means that simply plugging in the device
8575 will not cause the mount unit to be started automatically. But please
8576 note that the mount unit may be started for other reasons, in
8577 particular if it is part of local-fs.target, and any unit which
8578 (transitively) depends on local-fs.target is started.
8580 * networkctl list/status/lldp now accept globbing wildcards for network
8581 interface names to match against all existing interfaces.
8583 * The $PIDFILE environment variable is set to point the absolute path
8584 configured with PIDFile= for processes of that service.
8586 * The fallback DNS server list was augmented with Cloudflare public DNS
8587 servers. Use `-Ddns-servers=` to set a different fallback.
8589 * A new special target usb-gadget.target will be started automatically
8590 when a USB Device Controller is detected (which means that the system
8591 is a USB peripheral).
8593 * A new unit setting CPUQuotaPeriodSec= assigns the time period
8594 relatively to which the CPU time quota specified by CPUQuota= is
8597 * A new unit setting ProtectHostname= may be used to prevent services
8598 from modifying hostname information (even if they otherwise would
8599 have privileges to do so).
8601 * A new unit setting NetworkNamespacePath= may be used to specify a
8602 namespace for service or socket units through a path referring to a
8603 Linux network namespace pseudo-file.
8605 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
8606 have an effect on .socket units: when used the listening socket is
8607 created within the configured network namespace instead of the host
8610 * ExecStart= command lines in unit files may now be prefixed with ':'
8611 in which case environment variable substitution is
8612 disabled. (Supported for the other ExecXYZ= settings, too.)
8614 * .timer units gained two new boolean settings OnClockChange= and
8615 OnTimezoneChange= which may be used to also trigger a unit when the
8616 system clock is changed or the local timezone is
8617 modified. systemd-run has been updated to make these options easily
8618 accessible from the command line for transient timers.
8620 * Two new conditions for units have been added: ConditionMemory= may be
8621 used to conditionalize a unit based on installed system
8622 RAM. ConditionCPUs= may be used to conditionalize a unit based on
8623 installed CPU cores.
8625 * The @default system call filter group understood by SystemCallFilter=
8626 has been updated to include the new rseq() system call introduced in
8629 * A new time-set.target has been added that indicates that the system
8630 time has been set from a local source (possibly imprecise). The
8631 existing time-sync.target is stronger and indicates that the time has
8632 been synchronized with a precise external source. Services where
8633 approximate time is sufficient should use the new target.
8635 * "systemctl start" (and related commands) learnt a new
8636 --show-transaction option. If specified brief information about all
8637 jobs queued because of the requested operation is shown.
8639 * systemd-networkd recognizes a new operation state 'enslaved', used
8640 (instead of 'degraded' or 'carrier') for interfaces which form a
8641 bridge, bond, or similar, and an new 'degraded-carrier' operational
8642 state used for the bond or bridge master interface when one of the
8643 enslaved devices is not operational.
8645 * .network files learnt the new IgnoreCarrierLoss= option for leaving
8646 networks configured even if the carrier is lost.
8648 * The RequiredForOnline= setting in .network files may now specify a
8649 minimum operational state required for the interface to be considered
8650 "online" by systemd-networkd-wait-online. Related to this
8651 systemd-networkd-wait-online gained a new option --operational-state=
8652 to configure the same, and its --interface= option was updated to
8653 optionally also take an operational state specific for an interface.
8655 * systemd-networkd-wait-online gained a new setting --any for waiting
8656 for only one of the requested interfaces instead of all of them.
8658 * systemd-networkd now implements L2TP tunnels.
8660 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
8661 may be used to cause autonomous and onlink prefixes received in IPv6
8662 Router Advertisements to be ignored.
8664 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
8665 file settings may be used to tweak bridge behaviour.
8667 * The new TripleSampling= option in .network files may be used to
8668 configure CAN triple sampling.
8670 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
8671 used to point to private or preshared key for a WireGuard interface.
8673 * /etc/crypttab now supports the same-cpu-crypt and
8674 submit-from-crypt-cpus options to tweak encryption work scheduling
8677 * systemd-tmpfiles will now take a BSD file lock before operating on a
8678 contents of directory. This may be used to temporarily exclude
8679 directories from aging by taking the same lock (useful for example
8680 when extracting a tarball into /tmp or /var/tmp as a privileged user,
8681 which might create files with really old timestamps, which
8682 nevertheless should not be deleted). For further details, see:
8684 https://systemd.io/TEMPORARY_DIRECTORIES
8686 * systemd-tmpfiles' h line type gained support for the
8687 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
8688 controlling project quota inheritance.
8690 * sd-boot and bootctl now implement support for an Extended Boot Loader
8691 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
8692 addition to the ESP partition mounted to /efi or /boot/efi.
8693 Configuration file fragments, kernels, initrds and other EFI images
8694 to boot will be loaded from both the ESP and XBOOTLDR partitions.
8695 The XBOOTLDR partition was previously described by the Boot Loader
8696 Specification, but implementation was missing in sd-boot. Support for
8697 this concept allows using the sd-boot boot loader in more
8698 conservative scenarios where the boot loader itself is placed in the
8699 ESP but the kernels to boot (and their metadata) in a separate
8702 * A system may now be booted with systemd.volatile=overlay on the
8703 kernel command line, which causes the root file system to be set up
8704 an overlayfs mount combining the root-only root directory with a
8705 writable tmpfs. In this setup, the underlying root device is not
8706 modified, and any changes are lost at reboot.
8708 * Similar, systemd-nspawn can now boot containers with a volatile
8709 overlayfs root with the new --volatile=overlay switch.
8711 * systemd-nspawn can now consume OCI runtime bundles using a new
8712 --oci-bundle= option. This implementation is fully usable, with most
8713 features in the specification implemented, but since this a lot of
8714 new code and functionality, this feature should most likely not
8715 be used in production yet.
8717 * systemd-nspawn now supports various options described by the OCI
8718 runtime specification on the command-line and in .nspawn files:
8719 --inaccessible=/Inaccessible= may be used to mask parts of the file
8720 system tree, --console=/--pipe may be used to configure how standard
8721 input, output, and error are set up.
8723 * busctl learned the `emit` verb to generate D-Bus signals.
8725 * systemd-analyze cat-config may be used to gather and display
8726 configuration spread over multiple files, for example system and user
8727 presets, tmpfiles.d, sysusers.d, udev rules, etc.
8729 * systemd-analyze calendar now takes an optional new parameter
8730 --iterations= which may be used to show a maximum number of iterations
8731 the specified expression will elapse next.
8733 * The sd-bus C API gained support for naming method parameters in the
8736 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
8737 the reboot() system call expects.
8739 * journalctl learnt a new --cursor-file= option that points to a file
8740 from which a cursor should be loaded in the beginning and to which
8741 the updated cursor should be stored at the end.
8743 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
8744 detected by systemd-detect-virt (and may also be used in
8745 ConditionVirtualization=).
8747 * The behaviour of systemd-logind may now be modified with environment
8748 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
8749 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
8750 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
8751 skip the relevant operation completely (when set to false), or to
8752 create a flag file in /run/systemd (when set to true), instead of
8753 actually commencing the real operation when requested. The presence
8754 of /run/systemd/reboot-to-firmware-setup,
8755 /run/systemd/reboot-to-boot-loader-menu, and
8756 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
8757 boot loader implementations to replace some steps logind performs
8758 during reboot with their own operations.
8760 * systemctl can be used to request a reboot into the boot loader menu
8761 or a specific boot loader entry with the new --boot-load-menu= and
8762 --boot-loader-entry= options to a reboot command. (This requires a
8763 boot loader that supports this, for example sd-boot.)
8765 * kernel-install will no longer unconditionally create the output
8766 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
8767 snippets, but will do only if the machine-specific parent directory
8768 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
8769 to create this parent directory during sd-boot installation.
8771 This makes it easier to use kernel-install with plugins which support
8772 a different layout of the bootloader partitions (for example grub2).
8774 * During package installation (with `ninja install`), we would create
8775 symlinks for getty@tty1.service, systemd-networkd.service,
8776 systemd-networkd.socket, systemd-resolved.service,
8777 remote-cryptsetup.target, remote-fs.target,
8778 systemd-networkd-wait-online.service, and systemd-timesyncd.service
8779 in /etc, as if `systemctl enable` was called for those units, to make
8780 the system usable immediately after installation. Now this is not
8781 done anymore, and instead calling `systemctl preset-all` is
8782 recommended after the first installation of systemd.
8784 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
8785 is built on seccomp. When turned on creation of SUID/SGID files is
8788 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
8789 implied if DynamicUser= is turned on for a service. This hardens
8790 these services, so that they neither can benefit from nor create
8791 SUID/SGID executables. This is a minor compatibility breakage, given
8792 that when DynamicUser= was first introduced SUID/SGID behaviour was
8793 unaffected. However, the security benefit of these two options is
8794 substantial, and the setting is still relatively new, hence we opted
8795 to make it mandatory for services with dynamic users.
8797 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
8798 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
8799 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
8800 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
8801 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
8802 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
8803 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
8804 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
8805 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
8806 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
8807 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
8808 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
8809 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
8810 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
8811 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
8812 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
8813 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
8814 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
8816 — Warsaw, 2019-04-11
8820 * The default locale can now be configured at compile time. Otherwise,
8821 a suitable default will be selected automatically (one of C.UTF-8,
8822 en_US.UTF-8, and C).
8824 * The version string shown by systemd and other tools now includes the
8825 git commit hash when built from git. An override may be specified
8826 during compilation, which is intended to be used by distributions to
8827 include the package release information.
8829 * systemd-cat can now filter standard input and standard error streams
8830 for different syslog priorities using the new --stderr-priority=
8833 * systemd-journald and systemd-journal-remote reject entries which
8834 contain too many fields (CVE-2018-16865) and set limits on the
8835 process' command line length (CVE-2018-16864).
8837 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
8840 * A new network device NamePolicy "keep" is implemented for link files,
8841 and used by default in 99-default.link (the fallback configuration
8842 provided by systemd). With this policy, if the network device name
8843 was already set by userspace, the device will not be renamed again.
8844 This matches the naming scheme that was implemented before
8845 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
8846 is also enabled by default, even if not specified. Effectively, this
8847 means that if naming-scheme >= 240 is specified, network devices will
8848 be renamed according to the configuration, even if they have been
8849 renamed already, if "keep" is not specified as the naming policy in
8850 the .link file. The 99-default.link file provided by systemd includes
8851 "keep" for backwards compatibility, but it is recommended for user
8852 installed .link files to *not* include it.
8854 The "kernel" policy, which keeps kernel names declared to be
8855 "persistent", now works again as documented.
8857 * kernel-install script now optionally takes the paths to one or more
8858 initrd files, and passes them to all plugins.
8860 * The mincore() system call has been dropped from the @system-service
8861 system call filter group, as it is pretty exotic and may potentially
8862 used for side-channel attacks.
8864 * -fPIE is dropped from compiler and linker options. Please specify
8865 -Db_pie=true option to meson to build position-independent
8866 executables. Note that the meson option is supported since meson-0.49.
8868 * The fs.protected_regular and fs.protected_fifos sysctls, which were
8869 added in Linux 4.19 to make some data spoofing attacks harder, are
8870 now enabled by default. While this will hopefully improve the
8871 security of most installations, it is technically a backwards
8872 incompatible change; to disable these sysctls again, place the
8873 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
8875 fs.protected_regular = 0
8876 fs.protected_fifos = 0
8878 Note that the similar hardlink and symlink protection has been
8879 enabled since v199, and may be disabled likewise.
8881 * The files read from the EnvironmentFile= setting in unit files now
8882 parse backslashes inside quotes literally, matching the behaviour of
8885 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
8886 now automatically become NOPs when run in a chroot() environment.
8888 * The tmpfiles.d/ "C" line type will now copy directory trees not only
8889 when the destination is so far missing, but also if it already exists
8890 as a directory and is empty. This is useful to cater for systems
8891 where directory trees are put together from multiple separate mount
8892 points but otherwise empty.
8894 * A new function sd_bus_close_unref() (and the associated
8895 sd_bus_close_unrefp()) has been added to libsystemd, that combines
8896 sd_bus_close() and sd_bus_unref() in one.
8898 * udevadm control learnt a new option for --ping for testing whether a
8899 systemd-udevd instance is running and reacting.
8901 * udevadm trigger learnt a new option for --wait-daemon for waiting
8902 systemd-udevd daemon to be initialized.
8904 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
8905 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
8906 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
8907 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
8908 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
8909 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
8910 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
8911 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
8912 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
8913 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
8914 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
8915 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
8916 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
8917 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
8918 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
8919 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
8920 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
8922 — Berlin, 2019-02-14
8926 * NoNewPrivileges=yes has been set for all long-running services
8927 implemented by systemd. Previously, this was problematic due to
8928 SELinux (as this would also prohibit the transition from PID1's label
8929 to the service's label). This restriction has since been lifted, but
8930 an SELinux policy update is required.
8931 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
8933 * DynamicUser=yes is dropped from systemd-networkd.service,
8934 systemd-resolved.service and systemd-timesyncd.service, which was
8935 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
8936 and since v236 for systemd-timesyncd.service. The users and groups
8937 systemd-network, systemd-resolve and systemd-timesync are created
8938 by systemd-sysusers again. Distributors or system administrators
8939 may need to create these users and groups if they not exist (or need
8940 to re-enable DynamicUser= for those units) while upgrading systemd.
8941 Also, the clock file for systemd-timesyncd may need to move from
8942 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
8944 * When unit files are loaded from disk, previously systemd would
8945 sometimes (depending on the unit loading order) load units from the
8946 target path of symlinks in .wants/ or .requires/ directories of other
8947 units. This meant that unit could be loaded from different paths
8948 depending on whether the unit was requested explicitly or as a
8949 dependency of another unit, not honouring the priority of directories
8950 in search path. It also meant that it was possible to successfully
8951 load and start units which are not found in the unit search path, as
8952 long as they were requested as a dependency and linked to from
8953 .wants/ or .requires/. The target paths of those symlinks are not
8954 used for loading units anymore and the unit file must be found in
8957 * A new service type has been added: Type=exec. It's very similar to
8958 Type=simple but ensures the service manager will wait for both fork()
8959 and execve() of the main service binary to complete before proceeding
8960 with follow-up units. This is primarily useful so that the manager
8961 propagates any errors in the preparation phase of service execution
8962 back to the job that requested the unit to be started. For example,
8963 consider a service that has ExecStart= set to a file system binary
8964 that doesn't exist. With Type=simple starting the unit would be
8965 considered instantly successful, as only fork() has to complete
8966 successfully and the manager does not wait for execve(), and hence
8967 its failure is seen "too late". With the new Type=exec service type
8968 starting the unit will fail, as the manager will wait for the
8969 execve() and notice its failure, which is then propagated back to the
8972 NOTE: with the next release 241 of systemd we intend to change the
8973 systemd-run tool to default to Type=exec for transient services
8974 started by it. This should be mostly safe, but in specific corner
8975 cases might result in problems, as the systemd-run tool will then
8976 block on NSS calls (such as user name look-ups due to User=) done
8977 between the fork() and execve(), which under specific circumstances
8978 might cause problems. It is recommended to specify "-p Type=simple"
8979 explicitly in the few cases where this applies. For regular,
8980 non-transient services (i.e. those defined with unit files on disk)
8981 we will continue to default to Type=simple.
8983 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
8984 userspace processes is set to 1024 (soft) and 4096
8985 (hard). Previously, systemd passed this on unmodified to all
8986 processes it forked off. With this systemd release the hard limit
8987 systemd passes on is increased to 512K, overriding the kernel's
8988 defaults and substantially increasing the number of simultaneous file
8989 descriptors unprivileged userspace processes can allocate. Note that
8990 the soft limit remains at 1024 for compatibility reasons: the
8991 traditional UNIX select() call cannot deal with file descriptors >=
8992 1024 and increasing the soft limit globally might thus result in
8993 programs unexpectedly allocating a high file descriptor and thus
8994 failing abnormally when attempting to use it with select() (of
8995 course, programs shouldn't use select() anymore, and prefer
8996 poll()/epoll, but the call unfortunately remains undeservedly popular
8997 at this time). This change reflects the fact that file descriptor
8998 handling in the Linux kernel has been optimized in more recent
8999 kernels and allocating large numbers of them should be much cheaper
9000 both in memory and in performance than it used to be. Programs that
9001 want to take benefit of the increased limit have to "opt-in" into
9002 high file descriptors explicitly by raising their soft limit. Of
9003 course, when they do that they must acknowledge that they cannot use
9004 select() anymore (and neither can any shared library they use — or
9005 any shared library used by any shared library they use and so on).
9006 Which default hard limit is most appropriate is of course hard to
9007 decide. However, given reports that ~300K file descriptors are used
9008 in real-life applications we believe 512K is sufficiently high as new
9009 default for now. Note that there are also reports that using very
9010 high hard limits (e.g. 1G) is problematic: some software allocates
9011 large arrays with one element for each potential file descriptor
9012 (Java, …) — a high hard limit thus triggers excessively large memory
9013 allocations in these applications. Hopefully, the new default of 512K
9014 is a good middle ground: higher than what real-life applications
9015 currently need, and low enough for avoid triggering excessively large
9016 allocations in problematic software. (And yes, somebody should fix
9019 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
9020 to the highest possible values, as separate accounting of file
9021 descriptors is no longer necessary, as memcg tracks them correctly as
9022 part of the memory accounting anyway. Thus, from the four limits on
9023 file descriptors currently enforced (fs.file-max, fs.nr_open,
9024 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
9025 and keep only the latter two. A set of build-time options
9026 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
9027 has been added to revert this change in behaviour, which might be
9028 an option for systems that turn off memcg in the kernel.
9030 * When no /etc/locale.conf file exists (and hence no locale settings
9031 are in place), systemd will now use the "C.UTF-8" locale by default,
9032 and set LANG= to it. This locale is supported by various
9033 distributions including Fedora, with clear indications that upstream
9034 glibc is going to make it available too. This locale enables UTF-8
9035 mode by default, which appears appropriate for 2018.
9037 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
9038 default. This effectively switches the RFC3704 Reverse Path filtering
9039 from Strict mode to Loose mode. This is more appropriate for hosts
9040 that have multiple links with routes to the same networks (e.g.
9041 a client with a Wi-Fi and Ethernet both connected to the internet).
9043 Consult the kernel documentation for details on this sysctl:
9044 https://docs.kernel.org/networking/ip-sysctl.html
9046 * The v239 change to turn on "net.ipv4.tcp_ecn" by default has been
9049 * CPUAccounting=yes no longer enables the CPU controller when using
9050 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
9051 statistics are now provided independently from the CPU controller.
9053 * Support for disabling a particular cgroup controller within a sub-tree
9054 has been added through the DisableControllers= directive.
9056 * cgroup_no_v1=all on the kernel command line now also implies
9057 using the unified cgroup hierarchy, unless one explicitly passes
9058 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
9060 * The new "MemoryMin=" unit file property may now be used to set the
9061 memory usage protection limit of processes invoked by the unit. This
9062 controls the cgroup v2 memory.min attribute. Similarly, the new
9063 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
9064 cgroup v2 io.latency cgroup property for configuring per-service I/O
9067 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
9068 to the cgroup v1 "devices" cgroup controller.
9070 * systemd-escape now is able to combine --unescape with --template. It
9071 also learnt a new option --instance for extracting and unescaping the
9072 instance part of a unit name.
9074 * sd-bus now provides the sd_bus_message_readv() which is similar to
9075 sd_bus_message_read() but takes a va_list object. The pair
9076 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
9077 has been added for configuring the default method call timeout to
9078 use. sd_bus_error_move() may be used to efficiently move the contents
9079 from one sd_bus_error structure to another, invalidating the
9080 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
9081 be used to control whether a bus connection object is automatically
9082 flushed when an sd-event loop is exited.
9084 * When processing classic BSD syslog log messages, journald will now
9085 save the original time-stamp string supplied in the new
9086 SYSLOG_TIMESTAMP= journal field. This permits consumers to
9087 reconstruct the original BSD syslog message more correctly.
9089 * StandardOutput=/StandardError= in service files gained support for
9090 new "append:…" parameters, for connecting STDOUT/STDERR of a service
9091 to a file, and appending to it.
9093 * The signal to use as last step of killing of unit processes is now
9094 configurable. Previously it was hard-coded to SIGKILL, which may now
9095 be overridden with the new KillSignal= setting. Note that this is the
9096 signal used when regular termination (i.e. SIGTERM) does not suffice.
9097 Similarly, the signal used when aborting a program in case of a
9098 watchdog timeout may now be configured too (WatchdogSignal=).
9100 * The XDG_SESSION_DESKTOP environment variable may now be configured in
9101 the pam_systemd argument line, using the new desktop= switch. This is
9102 useful to initialize it properly from a display manager without
9103 having to touch C code.
9105 * Most configuration options that previously accepted percentage values
9106 now also accept permille values with the '‰' suffix (instead of '%').
9108 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
9111 * systemd-resolved's configuration file resolved.conf gained a new
9112 option ReadEtcHosts= which may be used to turn off processing and
9113 honoring /etc/hosts entries.
9115 * The "--wait" switch may now be passed to "systemctl
9116 is-system-running", in which case the tool will synchronously wait
9117 until the system finished start-up.
9119 * hostnamed gained a new bus call to determine the DMI product UUID.
9121 * On x86-64 systemd will now prefer using the RDRAND processor
9122 instruction over /dev/urandom whenever it requires randomness that
9123 neither has to be crypto-grade nor should be reproducible. This
9124 should substantially reduce the amount of entropy systemd requests
9125 from the kernel during initialization on such systems, though not
9126 reduce it to zero. (Why not zero? systemd still needs to allocate
9127 UUIDs and such uniquely, which require high-quality randomness.)
9129 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
9130 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
9131 for forcing the "Other Information" bit in IPv6 RA messages. The
9132 bonding logic gained four new options AdActorSystemPriority=,
9133 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
9134 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
9135 shuffling of flows. The tunnel logic gained a new
9136 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
9137 Deployment. The policy rule logic gained four new options IPProtocol=,
9138 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
9139 support for the MulticastToUnicast= option. networkd also gained
9140 support for configuring static IPv4 ARP or IPv6 neighbor entries.
9142 * .preset files (as read by 'systemctl preset') may now be used to
9143 instantiate services.
9145 * /etc/crypttab now understands the sector-size= option to configure
9146 the sector size for an encrypted partition.
9148 * Key material for encrypted disks may now be placed on a formatted
9149 medium, and referenced from /etc/crypttab by the UUID of the file
9150 system, followed by "=" suffixed by the path to the key file.
9152 * The "collect" udev component has been removed without replacement, as
9153 it is neither used nor maintained.
9155 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
9156 LogsDirectory=, ConfigurationDirectory= settings are used in a
9157 service the executed processes will now receive a set of environment
9158 variables containing the full paths of these directories.
9159 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
9160 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
9161 are used. Note that these options may be used multiple times per
9162 service in which case the resulting paths will be concatenated and
9163 separated by colons.
9165 * Predictable interface naming has been extended to cover InfiniBand
9166 NICs. They will be exposed with an "ib" prefix.
9168 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
9169 which case the respective line failing is ignored.
9171 * .link files may now be used to configure the equivalent to the
9172 "ethtool advertise" commands.
9174 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
9175 alternative to libudev.h. Previously, the latter was just an internal
9176 wrapper around the former, but now these two APIs are exposed
9179 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
9180 which calculates an app-specific boot ID similar to how
9181 sd_id128_get_machine_app_specific() generates an app-specific machine
9184 * A new tool systemd-id128 has been added that can be used to determine
9185 and generate various 128-bit IDs.
9187 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
9190 * systemd-hibernate-resume-generator will now honor the "noresume"
9191 kernel command line option, in which case it will bypass resuming
9192 from any hibernated image.
9194 * The systemd-sleep.conf configuration file gained new options
9195 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
9196 AllowHybridSleep= for prohibiting specific sleep modes even if the
9197 kernel exports them.
9199 * portablectl is now officially supported and has thus moved to
9202 * bootctl learnt the two new commands "set-default" and "set-oneshot"
9203 for setting the default boot loader item to boot to (either
9204 persistently or only for the next boot). This is currently only
9205 compatible with sd-boot, but may be implemented on other boot loaders
9206 too, that follow the boot loader interface. The updated interface is
9207 now documented here:
9209 https://systemd.io/BOOT_LOADER_INTERFACE
9211 * A new kernel command line option systemd.early_core_pattern= is now
9212 understood which may be used to influence the core_pattern PID 1
9213 installs during early boot.
9215 * busctl learnt two new options -j and --json= for outputting method
9216 call replies, properties and monitoring output in JSON.
9218 * journalctl's JSON output now supports simple ANSI coloring as well as
9219 a new "json-seq" mode for generating RFC7464 output.
9221 * Unit files now support the %g/%G specifiers that resolve to the UNIX
9222 group/GID of the service manager runs as, similar to the existing
9223 %u/%U specifiers that resolve to the UNIX user/UID.
9225 * systemd-logind learnt a new global configuration option
9226 UserStopDelaySec= that may be set in logind.conf. It specifies how
9227 long the systemd --user instance shall remain started after a user
9228 logs out. This is useful to speed up repetitive re-connections of the
9229 same user, as it means the user's service manager doesn't have to be
9230 stopped/restarted on each iteration, but can be reused between
9231 subsequent options. This setting defaults to 10s. systemd-logind also
9232 exports two new properties on its Manager D-Bus objects indicating
9233 whether the system's lid is currently closed, and whether the system
9236 * systemd gained support for a generic boot counting logic, which
9237 generically permits automatic reverting to older boot loader entries
9238 if newer updated ones don't work. The boot loader side is implemented
9239 in sd-boot, but is kept open for other boot loaders too. For details
9242 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
9244 * The SuccessAction=/FailureAction= unit file settings now learnt two
9245 new parameters: "exit" and "exit-force", which result in immediate
9246 exiting of the service manager, and are only useful in systemd --user
9247 and container environments.
9249 * Unit files gained support for a pair of options
9250 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
9251 exit status to use as service manager exit status when
9252 SuccessAction=/FailureAction= is set to exit or exit-force.
9254 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
9255 options may now be used to configure the log rate limiting applied by
9256 journald per-service.
9258 * systemd-analyze gained a new verb "timespan" for parsing and
9259 normalizing time span values (i.e. strings like "5min 7s 8us").
9261 * systemd-analyze also gained a new verb "security" for analyzing the
9262 security and sand-boxing settings of services in order to determine an
9263 "exposure level" for them, indicating whether a service would benefit
9264 from more sand-boxing options turned on for them.
9266 * "systemd-analyze syscall-filter" will now also show system calls
9267 supported by the local kernel but not included in any of the defined
9270 * .nspawn files now understand the Ephemeral= setting, matching the
9271 --ephemeral command line switch.
9273 * sd-event gained the new APIs sd_event_source_get_floating() and
9274 sd_event_source_set_floating() for controlling whether a specific
9275 event source is "floating", i.e. destroyed along with the even loop
9278 * Unit objects on D-Bus gained a new "Refs" property that lists all
9279 clients that currently have a reference on the unit (to ensure it is
9282 * The JoinControllers= option in system.conf is no longer supported, as
9283 it didn't work correctly, is hard to support properly, is legacy (as
9284 the concept only exists on cgroup v1) and apparently wasn't used.
9286 * Journal messages that are generated whenever a unit enters the failed
9287 state are now tagged with a unique MESSAGE_ID. Similarly, messages
9288 generated whenever a service process exits are now made recognizable,
9289 too. A tagged message is also emitted whenever a unit enters the
9290 "dead" state on success.
9292 * systemd-run gained a new switch --working-directory= for configuring
9293 the working directory of the service to start. A shortcut -d is
9294 equivalent, setting the working directory of the service to the
9295 current working directory of the invoking program. The new --shell
9296 (or just -S) option has been added for invoking the $SHELL of the
9297 caller as a service, and implies --pty --same-dir --wait --collect
9298 --service-type=exec. Or in other words, "systemd-run -S" is now the
9299 quickest way to quickly get an interactive in a fully clean and
9300 well-defined system service context.
9302 * machinectl gained a new verb "import-fs" for importing an OS tree
9303 from a directory. Moreover, when a directory or tarball is imported
9304 and single top-level directory found with the OS itself below the OS
9305 tree is automatically mangled and moved one level up.
9307 * systemd-importd will no longer set up an implicit btrfs loop-back
9308 file system on /var/lib/machines. If one is already set up, it will
9309 continue to be used.
9311 * A new generator "systemd-run-generator" has been added. It will
9312 synthesize a unit from one or more program command lines included in
9313 the kernel command line. This is very useful in container managers
9316 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
9318 This will run "systemd-nspawn" on an image, invoke the specified
9319 command line and immediately shut down the container again, returning
9320 the command line's exit code.
9322 * The block device locking logic is now documented:
9324 https://systemd.io/BLOCK_DEVICE_LOCKING
9326 * loginctl and machinectl now optionally output the various tables in
9327 JSON using the --output= switch. It is our intention to add similar
9328 support to systemctl and all other commands.
9330 * udevadm's query and trigger verb now optionally take a .device unit
9333 * systemd-udevd's network naming logic now understands a new
9334 net.naming_scheme= kernel command line switch, which may be used to
9335 pick a specific version of the naming scheme. This helps stabilizing
9336 interface names even as systemd/udev are updated and the naming logic
9339 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
9340 SD_ID128_ALLF to test if a 128-bit ID is set to all 0xFF bytes, and to
9341 initialize one to all 0xFF.
9343 * After loading the SELinux policy systemd will now recursively relabel
9344 all files and directories listed in
9345 /run/systemd/relabel-extra.d/*.relabel (which should be simple
9346 newline separated lists of paths) in addition to the ones it already
9347 implicitly relabels in /run, /dev and /sys. After the relabelling is
9348 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
9349 removed. This is useful to permit initrds (i.e. code running before
9350 the SELinux policy is in effect) to generate files in the host
9351 filesystem safely and ensure that the correct label is applied during
9352 the transition to the host OS.
9354 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
9355 mknod() handling in user namespaces. Previously mknod() would always
9356 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
9357 but device nodes generated that way cannot be opened, and attempts to
9358 open them result in EPERM. This breaks the "graceful fallback" logic
9359 in systemd's PrivateDevices= sand-boxing option. This option is
9360 implemented defensively, so that when systemd detects it runs in a
9361 restricted environment (such as a user namespace, or an environment
9362 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
9363 where device nodes cannot be created the effect of PrivateDevices= is
9364 bypassed (following the logic that 2nd-level sand-boxing is not
9365 essential if the system systemd runs in is itself already sand-boxed
9366 as a whole). This logic breaks with 4.18 in container managers where
9367 user namespacing is used: suddenly PrivateDevices= succeeds setting
9368 up a private /dev/ file system containing devices nodes — but when
9369 these are opened they don't work.
9371 At this point it is recommended that container managers utilizing
9372 user namespaces that intend to run systemd in the payload explicitly
9373 block mknod() with seccomp or similar, so that the graceful fallback
9376 We are very sorry for the breakage and the requirement to change
9377 container configurations for newer kernels. It's purely caused by an
9378 incompatible kernel change. The relevant kernel developers have been
9379 notified about this userspace breakage quickly, but they chose to
9382 * PermissionsStartOnly= setting is deprecated (but is still supported
9383 for backwards compatibility). The same functionality is provided by
9384 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
9387 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
9388 pam_systemd anymore.
9390 * The naming scheme for network devices was changed to always rename
9391 devices, even if they were already renamed by userspace. The "kernel"
9392 policy was changed to only apply as a fallback, if no other naming
9395 * The requirements to build systemd is bumped to meson-0.46 and
9398 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
9399 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
9400 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
9401 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
9402 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
9403 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
9404 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
9405 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
9406 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
9407 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
9408 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
9409 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
9410 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
9411 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
9412 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
9413 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
9414 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
9415 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
9416 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
9417 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
9418 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
9419 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
9420 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
9421 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
9422 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
9423 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
9424 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
9425 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
9426 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
9427 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
9428 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
9429 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
9430 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
9431 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
9432 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
9433 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
9434 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
9435 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
9436 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
9437 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
9438 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
9439 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
9440 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
9441 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
9442 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
9444 — Warsaw, 2018-12-21
9448 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
9449 builtin will name network interfaces differently than in previous
9450 versions for virtual network interfaces created with SR-IOV and NPAR
9451 and for devices where the PCI network controller device does not have
9452 a slot number associated.
9454 SR-IOV virtual devices are now named based on the name of the parent
9455 interface, with a suffix of "v<N>", where <N> is the virtual device
9456 number. Previously those virtual devices were named as if completely
9459 The ninth and later NPAR virtual devices will be named following the
9460 scheme used for the first eight NPAR partitions. Previously those
9461 devices were not renamed and the kernel default (eth<n>) was used.
9463 "net_id" will also generate names for PCI devices where the PCI
9464 network controller device does not have an associated slot number
9465 itself, but one of its parents does. Previously those devices were
9466 not renamed and the kernel default (eth<n>) was used.
9468 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
9469 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
9470 the unit. So, it is expected that the default behavior of
9471 systemd-logind is not changed. However, if distribution packagers or
9472 administrators disabled or modified IPAddressDeny= setting by a
9473 drop-in config file, then it may be necessary to update the file to
9474 re-enable AF_INET and AF_INET6 to support network user name services,
9477 * When the RestrictNamespaces= unit property is specified multiple
9478 times, then the specified types are merged now. Previously, only the
9479 last assignment was used. So, if distribution packagers or
9480 administrators modified the setting by a drop-in config file, then it
9481 may be necessary to update the file.
9483 * When OnFailure= is used in combination with Restart= on a service
9484 unit, then the specified units will no longer be triggered on
9485 failures that result in restarting. Previously, the specified units
9486 would be activated each time the unit failed, even when the unit was
9487 going to be restarted automatically. This behaviour contradicted the
9488 documentation. With this release the code is adjusted to match the
9491 * systemd-tmpfiles will now print a notice whenever it encounters
9492 tmpfiles.d/ lines referencing the /var/run/ directory. It will
9493 recommend reworking them to use the /run/ directory instead (for
9494 which /var/run/ is simply a symlinked compatibility alias). This way
9495 systemd-tmpfiles can properly detect line conflicts and merge lines
9496 referencing the same file by two paths, without having to access
9499 * systemctl disable/unmask/preset/preset-all cannot be used with
9500 --runtime. Previously this was allowed, but resulted in unintuitive
9501 behaviour that wasn't useful. systemctl disable/unmask will now undo
9502 both runtime and persistent enablement/masking, i.e. it will remove
9503 any relevant symlinks both in /run and /etc.
9505 * Note that all long-running system services shipped with systemd will
9506 now default to a system call allow list (rather than a deny list, as
9507 before). In particular, systemd-udevd will now enforce one too. For
9508 most cases this should be safe, however downstream distributions
9509 which disabled sandboxing of systemd-udevd (specifically the
9510 MountFlags= setting), might want to disable this security feature
9511 too, as the default allow-listing will prohibit all mount, swap,
9512 reboot and clock changing operations from udev rules.
9514 * sd-boot acquired new loader configuration settings to optionally turn
9515 off Windows and MacOS boot partition discovery as well as
9516 reboot-into-firmware menu items. It is also able to pick a better
9517 screen resolution for HiDPI systems, and now provides loader
9518 configuration settings to change the resolution explicitly.
9520 * systemd-resolved now supports DNS-over-TLS. It's still
9521 turned off by default, use DNSOverTLS=opportunistic to turn it on in
9522 resolved.conf. We intend to make this the default as soon as couple
9523 of additional techniques for optimizing the initial latency caused by
9524 establishing a TLS/TCP connection are implemented.
9526 * systemd-resolved.service and systemd-networkd.service now set
9527 DynamicUser=yes. The users systemd-resolve and systemd-network are
9528 not created by systemd-sysusers anymore.
9530 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
9531 that embed a network facing module into any process using getpwuid()
9532 or related call: the dynamic allocation of the user ID for
9533 systemd-resolved.service means the service manager has to check NSS
9534 if the user name is already taken when forking off the service. Since
9535 the user in the common case won't be defined in /etc/passwd the
9536 lookup is likely to trigger nss-ldap which in turn might use NSS to
9537 ask systemd-resolved for hostname lookups. This will hence result in
9538 a deadlock: a user name lookup in order to start
9539 systemd-resolved.service will result in a hostname lookup for which
9540 systemd-resolved.service needs to be started already. There are
9541 multiple ways to work around this problem: pre-allocate the
9542 "systemd-resolve" user on such systems, so that nss-ldap won't be
9543 triggered; or use a different NSS package that doesn't do networking
9544 in-process but provides a local asynchronous name cache; or configure
9545 the NSS package to avoid lookups for UIDs in the range `pkg-config
9546 systemd --variable=dynamicuidmin` … `pkg-config systemd
9547 --variable=dynamicuidmax`, so that it does not consider itself
9548 authoritative for the same UID range systemd allocates dynamic users
9551 * The systemd-resolve tool has been renamed to resolvectl (it also
9552 remains available under the old name, for compatibility), and its
9553 interface is now verb-based, similar in style to the other <xyz>ctl
9554 tools, such as systemctl or loginctl.
9556 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
9557 compatibility. It may be symlinked under the 'resolvconf' name, in
9558 which case it will take arguments and input compatible with the
9559 Debian and FreeBSD resolvconf tool.
9561 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
9562 where the system initially suspends, and after a timeout resumes and
9565 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
9566 set the client will only send a DUID as client identifier. (EDIT: the
9567 option was broken, and was dropped in v255.)
9569 * The nss-systemd glibc NSS module will now enumerate dynamic users and
9570 groups in effect. Previously, it could resolve UIDs/GIDs to user
9571 names/groups and vice versa, but did not support enumeration.
9573 * journald's Compress= configuration setting now optionally accepts a
9574 byte threshold value. All journal objects larger than this threshold
9575 will be compressed, smaller ones will not. Previously this threshold
9576 was not configurable and set to 512.
9578 * A new system.conf setting NoNewPrivileges= is now available which may
9579 be used to turn off acquisition of new privileges system-wide
9580 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
9581 for all its children). Note that turning this option on means setuid
9582 binaries and file system capabilities lose their special powers.
9583 While turning on this option is a big step towards a more secure
9584 system, doing so is likely to break numerous pre-existing UNIX tools,
9585 in particular su and sudo.
9587 * A new service systemd-time-sync-wait.service has been added. If
9588 enabled it will delay the time-sync.target unit at boot until time
9589 synchronization has been received from the network. This
9590 functionality is useful on systems lacking a local RTC or where it is
9591 acceptable that the boot process shall be delayed by external network
9594 * When hibernating, systemd will now inform the kernel of the image
9595 write offset, on kernels new enough to support this. This means swap
9596 files should work for hibernation now.
9598 * When loading unit files, systemd will now look for drop-in unit files
9599 extensions in additional places. Previously, for a unit file name
9600 "foo-bar-baz.service" it would look for dropin files in
9601 "foo-bar-baz.service.d/*.conf". Now, it will also look in
9602 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
9603 service name truncated after all inner dashes. This scheme allows
9604 writing drop-ins easily that apply to a whole set of unit files at
9605 once. It's particularly useful for mount and slice units (as their
9606 naming is prefix based), but is also useful for service and other
9607 units, for packages that install multiple unit files at once,
9608 following a strict naming regime of beginning the unit file name with
9609 the package's name. Two new specifiers are now supported in unit
9610 files to match this: %j and %J are replaced by the part of the unit
9611 name following the last dash.
9613 * Unit files and other configuration files that support specifier
9614 expansion now understand another three new specifiers: %T and %V will
9615 resolve to /tmp and /var/tmp respectively, or whatever temporary
9616 directory has been set for the calling user. %E will expand to either
9617 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
9619 * The ExecStart= lines of unit files are no longer required to
9620 reference absolute paths. If non-absolute paths are specified the
9621 specified binary name is searched within the service manager's
9622 built-in $PATH, which may be queried with 'systemd-path
9623 search-binaries-default'. It's generally recommended to continue to
9624 use absolute paths for all binaries specified in unit files.
9626 * Units gained a new load state "bad-setting", which is used when a
9627 unit file was loaded, but contained fatal errors which prevent it
9628 from being started (for example, a service unit has been defined
9629 lacking both ExecStart= and ExecStop= lines).
9631 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
9632 support alternative debuggers, for example lldb. The old name
9633 continues to be available however, for compatibility reasons. Use the
9634 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
9635 to pick an alternative debugger instead of the default gdb.
9637 * systemctl and the other tools will now output escape sequences that
9638 generate proper clickable hyperlinks in various terminal emulators
9639 where useful (for example, in the "systemctl status" output you can
9640 now click on the unit file name to quickly open it in the
9641 editor/viewer of your choice). Note that not all terminal emulators
9642 support this functionality yet, but many do. Unfortunately, the
9643 "less" pager doesn't support this yet, hence this functionality is
9644 currently automatically turned off when a pager is started (which
9645 happens quite often due to auto-paging). We hope to remove this
9646 limitation as soon as "less" learns these escape sequences. This new
9647 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
9648 environment variable. For details on these escape sequences see:
9649 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
9651 * networkd's .network files now support a new IPv6MTUBytes= option for
9652 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
9653 option in the [Route] section to configure the MTU to use for
9654 specific routes. It also gained support for configuration of the DHCP
9655 "UserClass" option through the new UserClass= setting. It gained
9656 three new options in the new [CAN] section for configuring CAN
9657 networks. The MULTICAST and ALLMULTI interface flags may now be
9658 controlled explicitly with the new Multicast= and AllMulticast=
9661 * networkd will now automatically make use of the kernel's route
9662 expiration feature, if it is available.
9664 * udevd's .link files now support setting the number of receive and
9665 transmit channels, using the RxChannels=, TxChannels=,
9666 OtherChannels=, CombinedChannels= settings.
9668 * Support for UDPSegmentationOffload= has been removed, given its
9669 limited support in hardware, and waning software support.
9671 * networkd's .netdev files now support creating "netdevsim" interfaces.
9673 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
9674 to query the unit belonging to a specific kernel control group.
9676 * systemd-analyze gained a new verb "cat-config", which may be used to
9677 dump the contents of any configuration file, with all its matching
9678 drop-in files added in, and honouring the usual search and masking
9679 logic applied to systemd configuration files. For example use
9680 "systemd-analyze cat-config systemd/system.conf" to get the complete
9681 system configuration file of systemd how it would be loaded by PID 1
9682 itself. Similar to this, various tools such as systemd-tmpfiles or
9683 systemd-sysusers, gained a new option "--cat-config", which does the
9684 corresponding operation for their own configuration settings. For
9685 example, "systemd-tmpfiles --cat-config" will now output the full
9686 list of tmpfiles.d/ lines in place.
9688 * timedatectl gained three new verbs: "show" shows bus properties of
9689 systemd-timedated, "timesync-status" shows the current NTP
9690 synchronization state of systemd-timesyncd, and "show-timesync"
9691 shows bus properties of systemd-timesyncd.
9693 * systemd-timesyncd gained a bus interface on which it exposes details
9696 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
9697 understood by systemd-timedated. It takes a colon-separated list of
9698 unit names of NTP client services. The list is used by
9699 "timedatectl set-ntp".
9701 * systemd-nspawn gained a new --rlimit= switch for setting initial
9702 resource limits for the container payload. There's a new switch
9703 --hostname= to explicitly override the container's hostname. A new
9704 --no-new-privileges= switch may be used to control the
9705 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
9706 --oom-score-adjust= switch controls the OOM scoring adjustment value
9707 for the payload. The new --cpu-affinity= switch controls the CPU
9708 affinity of the container payload. The new --resolv-conf= switch
9709 allows more detailed control of /etc/resolv.conf handling of the
9710 container. Similarly, the new --timezone= switch allows more detailed
9711 control of /etc/localtime handling of the container.
9713 * systemd-detect-virt gained a new --list switch, which will print a
9714 list of all currently known VM and container environments.
9716 * Support for "Portable Services" has been added, see
9717 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
9718 experimental, but this is expected to change soon. Reflecting this
9719 experimental state, the "portablectl" binary is not installed into
9720 /usr/bin yet. The binary has to be called with the full path
9721 /usr/lib/systemd/portablectl instead.
9723 * journalctl's and systemctl's -o switch now knows a new log output
9724 mode "with-unit". The output it generates is very similar to the
9725 regular "short" mode, but displays the unit name instead of the
9726 syslog tag for each log line. Also, the date is shown with timezone
9727 information. This mode is probably more useful than the classic
9728 "short" output mode for most purposes, except where pixel-perfect
9729 compatibility with classic /var/log/messages formatting is required.
9731 * A new --dump-bus-properties switch has been added to the systemd
9732 binary, which may be used to dump all supported D-Bus properties.
9733 (Options which are still supported, but are deprecated, are *not*
9736 * sd-bus gained a set of new calls:
9737 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
9738 enable/disable the "floating" state of a bus slot object,
9739 i.e. whether the slot object pins the bus it is allocated for into
9740 memory or if the bus slot object gets disconnected when the bus goes
9741 away. sd_bus_open_with_description(),
9742 sd_bus_open_user_with_description(),
9743 sd_bus_open_system_with_description() may be used to allocate bus
9744 objects and set their description string already during allocation.
9746 * sd-event gained support for watching inotify events from the event
9747 loop, in an efficient way, sharing inotify handles between multiple
9748 users. For this a new function sd_event_add_inotify() has been added.
9750 * sd-event and sd-bus gained support for calling special user-supplied
9751 destructor functions for userdata pointers associated with
9752 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
9753 functions sd_bus_slot_set_destroy_callback,
9754 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
9755 sd_bus_track_get_destroy_callback,
9756 sd_event_source_set_destroy_callback,
9757 sd_event_source_get_destroy_callback have been added.
9759 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
9761 * PID 1 will now automatically reschedule .timer units whenever the
9762 local timezone changes. (They previously got rescheduled
9763 automatically when the system clock changed.)
9765 * New documentation has been added to document cgroups delegation,
9766 portable services and the various code quality tools we have set up:
9768 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
9769 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
9770 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
9772 * The Boot Loader Specification has been added to the source tree.
9774 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
9776 While moving it into our source tree we have updated it and further
9777 changes are now accepted through the usual github PR workflow.
9779 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
9780 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
9781 earlier PAM modules. The data in these fields is used to initialize
9782 the session scope's resource properties. Thus external PAM modules
9783 may now configure per-session limits, for example sourced from
9784 external user databases.
9786 * socket units with Accept=yes will now maintain a "refused" counter in
9787 addition to the existing "accepted" counter, counting connections
9788 refused due to the enforced limits.
9790 * The "systemd-path search-binaries-default" command may now be use to
9791 query the default, built-in $PATH PID 1 will pass to the services it
9794 * A new unit file setting PrivateMounts= has been added. It's a boolean
9795 option. If enabled the unit's processes are invoked in their own file
9796 system namespace. Note that this behaviour is also implied if any
9797 other file system namespacing options (such as PrivateTmp=,
9798 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
9799 primarily useful for services that do not use any of the other file
9800 system namespacing options. One such service is systemd-udevd.service
9801 where this is now used by default.
9803 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
9804 when the system is booted in UEFI "secure mode".
9806 * A new unit "system-update-pre.target" is added, which defines an
9807 optional synchronization point for offline system updates, as
9808 implemented by the pre-existing "system-update.target" unit. It
9809 allows ordering services before the service that executes the actual
9810 update process in a generic way.
9812 * Systemd now emits warnings whenever .include syntax is used.
9814 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
9815 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
9816 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
9817 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
9818 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
9819 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
9820 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
9821 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
9822 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
9823 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
9824 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
9825 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
9826 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
9827 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
9828 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
9829 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
9830 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
9831 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
9832 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
9833 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
9834 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
9835 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
9836 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
9837 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
9838 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
9839 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
9840 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
9841 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
9842 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
9844 — Berlin, 2018-06-22
9848 * The MemoryAccounting= unit property now defaults to on. After
9849 discussions with the upstream control group maintainers we learnt
9850 that the negative impact of cgroup memory accounting on current
9851 kernels is finally relatively minimal, so that it should be safe to
9852 enable this by default without affecting system performance. Besides
9853 memory accounting only task accounting is turned on by default, all
9854 other forms of resource accounting (CPU, IO, IP) remain off for now,
9855 because it's not clear yet that their impact is small enough to move
9856 from opt-in to opt-out. We recommend downstreams to leave memory
9857 accounting on by default if kernel 4.14 or higher is primarily
9858 used. On very resource constrained systems or when support for old
9859 kernels is a necessity, -Dmemory-accounting-default=false can be used
9860 to revert this change.
9862 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
9863 %udev_rules_update) and the journal catalog (%journal_catalog_update)
9864 from the upgrade scriptlets of individual packages now do nothing.
9865 Transfiletriggers have been added which will perform those updates
9866 once at the end of the transaction.
9868 Similar transfiletriggers have been added to execute any sysctl.d
9869 and binfmt.d rules. Thus, it should be unnecessary to provide any
9870 scriptlets to execute this configuration from package installation
9873 * systemd-sysusers gained a mode where the configuration to execute is
9874 specified on the command line, but this configuration is not executed
9875 directly, but instead it is merged with the configuration on disk,
9876 and the result is executed. This is useful for package installation
9877 scripts which want to create the user before installing any files on
9878 disk (in case some of those files are owned by that user), while
9879 still allowing local admin overrides.
9881 This functionality is exposed to rpm scriptlets through a new
9882 %sysusers_create_package macro. Old %sysusers_create and
9883 %sysusers_create_inline macros are deprecated.
9885 A transfiletrigger for sysusers.d configuration is now installed,
9886 which means that it should be unnecessary to call systemd-sysusers from
9887 package installation scripts, unless the package installs any files
9888 owned by those newly-created users, in which case
9889 %sysusers_create_package should be used.
9891 * Analogous change has been done for systemd-tmpfiles: it gained a mode
9892 where the command-line configuration is merged with the configuration
9893 on disk. This is exposed as the new %tmpfiles_create_package macro,
9894 and %tmpfiles_create is deprecated. A transfiletrigger is installed
9895 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
9896 from package installation scripts.
9898 * sysusers.d configuration for a user may now also specify the group
9899 number, in addition to the user number ("u username 123:456"), or
9900 without the user number ("u username -:456").
9902 * Configution items for systemd-sysusers can now be specified as
9903 positional arguments when the new --inline switch is used.
9905 * The login shell of users created through sysusers.d may now be
9906 specified (previously, it was always /bin/sh for root and
9907 /sbin/nologin for other users).
9909 * systemd-analyze gained a new --global switch to look at global user
9910 configuration. It also gained a unit-paths verb to list the unit load
9911 paths that are compiled into systemd (which can be used with
9912 --systemd, --user, or --global).
9914 * udevadm trigger gained a new --settle/-w option to wait for any
9915 triggered events to finish (but just those, and not any other events
9916 which are triggered meanwhile).
9918 * The action that systemd-logind takes when the lid is closed and the
9919 machine is connected to external power can now be configured using
9920 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
9921 was determined by HandleLidSwitch=, and, for backwards compatibility,
9922 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
9924 * journalctl will periodically call sd_journal_process() to make it
9925 resilient against inotify queue overruns when journal files are
9926 rotated very quickly.
9928 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
9929 sd_bus_get_n_queued_write — may be used to check the number of
9930 pending bus messages.
9932 * systemd gained a new
9933 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
9934 which can be used to migrate foreign processes to scope and service
9935 units. The primary user for this new API is systemd itself: the
9936 systemd --user instance uses this call of the systemd --system
9937 instance to migrate processes if it itself gets the request to
9938 migrate processes and the kernel refuses this due to access
9939 restrictions. Thanks to this "systemd-run --scope --user …" works
9940 again in pure cgroup v2 environments when invoked from the user
9943 * A new TemporaryFileSystem= setting can be used to mask out part of
9944 the real file system tree with tmpfs mounts. This may be combined
9945 with BindPaths= and BindReadOnlyPaths= to hide files or directories
9946 not relevant to the unit, while still allowing some paths lower in
9947 the tree to be accessed.
9949 ProtectHome=tmpfs may now be used to hide user home and runtime
9950 directories from units, in a way that is mostly equivalent to
9951 "TemporaryFileSystem=/home /run/user /root".
9953 * Non-service units are now started with KeyringMode=shared by default.
9954 This means that mount and swapon and other mount tools have access
9955 to keys in the main keyring.
9957 * /sys/fs/bpf is now mounted automatically.
9959 * QNX virtualization is now detected by systemd-detect-virt and may
9960 be used in ConditionVirtualization=.
9962 * IPAccounting= may now be enabled also for slice units.
9964 * A new -Dsplit-bin= build configuration switch may be used to specify
9965 whether bin and sbin directories are merged, or if they should be
9966 included separately in $PATH and various listings of executable
9967 directories. The build configuration scripts will try to autodetect
9968 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
9969 system, but distributions are encouraged to configure this
9972 * A new -Dok-color= build configuration switch may be used to change
9973 the colour of "OK" status messages.
9975 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
9976 PrivateNetwork=yes was buggy in previous versions of systemd. This
9977 means that after the upgrade and daemon-reexec, any such units must
9980 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
9981 will not exclude read-only files owned by root from cleanup.
9983 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
9984 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
9985 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
9986 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
9987 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
9988 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
9989 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
9990 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
9991 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
9992 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
9993 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
9994 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
9995 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
9996 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
9997 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
9998 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
10000 — Warsaw, 2018-03-05
10004 * Some keyboards come with a zoom see-saw or rocker which until now got
10005 mapped to the Linux "zoomin/out" keys in hwdb. However, these
10006 keycodes are not recognized by any major desktop. They now produce
10007 Up/Down key events so that they can be used for scrolling.
10009 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
10010 slightly: previously, if an argument was specified for lines of this
10011 type (i.e. the right-most column was set) this string was appended to
10012 existing files each time systemd-tmpfiles was run. This behaviour was
10013 different from what the documentation said, and not particularly
10014 useful, as repeated systemd-tmpfiles invocations would not be
10015 idempotent and grow such files without bounds. With this release
10016 behaviour has been altered to match what the documentation says:
10017 lines of this type only have an effect if the indicated files don't
10018 exist yet, and only then the argument string is written to the file.
10020 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
10021 systemd-tmpfiles behaviour: previously, read-only files owned by root
10022 were always excluded from the file "aging" algorithm (i.e. the
10023 automatic clean-up of directories like /tmp based on
10024 atime/mtime/ctime). We intend to drop this restriction, and age files
10025 by default even when owned by root and read-only. This behaviour was
10026 inherited from older tools, but there have been requests to remove
10027 it, and it's not obvious why this restriction was made in the first
10028 place. Please speak up now, if you are aware of software that requires
10029 this behaviour, otherwise we'll remove the restriction in v238.
10031 * A new environment variable $SYSTEMD_OFFLINE is now understood by
10032 systemctl. It takes a boolean argument. If on, systemctl assumes it
10033 operates on an "offline" OS tree, and will not attempt to talk to the
10034 service manager. Previously, this mode was implicitly enabled if a
10035 chroot() environment was detected, and this new environment variable
10036 now provides explicit control.
10038 * .path and .socket units may now be created transiently, too.
10039 Previously only service, mount, automount and timer units were
10040 supported as transient units. The systemd-run tool has been updated
10041 to expose this new functionality, you may hence use it now to bind
10042 arbitrary commands to path or socket activation on-the-fly from the
10043 command line. Moreover, almost all properties are now exposed for the
10044 unit types that already supported transient operation.
10046 * The systemd-mount command gained support for a new --owner= parameter
10047 which takes a user name, which is then resolved and included in uid=
10048 and gid= mount options string of the file system to mount.
10050 * A new unit condition ConditionControlGroupController= has been added
10051 that checks whether a specific cgroup controller is available.
10053 * Unit files, udev's .link files, and systemd-networkd's .netdev and
10054 .network files all gained support for a new condition
10055 ConditionKernelVersion= for checking against specific kernel
10058 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
10059 support for configuring device flags in the Flags= setting. In the
10060 same files, the [Tunnel] section gained support for configuring
10061 AllowLocalRemote=. The [Route] section in .network files gained
10062 support for configuring InitialCongestionWindow=,
10063 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
10064 understands RapidCommit=.
10066 * systemd-networkd's DHCPv6 support gained support for Prefix
10069 * sd-bus gained support for a new "watch-bind" feature. When this
10070 feature is enabled, an sd_bus connection may be set up to connect to
10071 an AF_UNIX socket in the file system as soon as it is created. This
10072 functionality is useful for writing early-boot services that
10073 automatically connect to the system bus as soon as it is started,
10074 without ugly time-based polling. systemd-networkd and
10075 systemd-resolved have been updated to make use of this
10076 functionality. busctl exposes this functionality in a new
10077 --watch-bind= command line switch.
10079 * sd-bus will now optionally synthesize a local "Connected" signal as
10080 soon as a D-Bus connection is set up fully. This message mirrors the
10081 already existing "Disconnected" signal which is synthesized when the
10082 connection is terminated. This signal is generally useful but
10083 particularly handy in combination with the "watch-bind" feature
10084 described above. Synthesizing of this message has to be requested
10085 explicitly through the new API call sd_bus_set_connected_signal(). In
10086 addition a new call sd_bus_is_ready() has been added that checks
10087 whether a connection is fully set up (i.e. between the "Connected" and
10088 "Disconnected" signals).
10090 * sd-bus gained two new calls sd_bus_request_name_async() and
10091 sd_bus_release_name_async() for asynchronously registering bus
10092 names. Similar, there is now sd_bus_add_match_async() for installing
10093 a signal match asynchronously. All of systemd's own services have
10094 been updated to make use of these calls. Doing these operations
10095 asynchronously has two benefits: it reduces the risk of deadlocks in
10096 case of cyclic dependencies between bus services, and it speeds up
10097 service initialization since synchronization points for bus
10098 round-trips are removed.
10100 * sd-bus gained two new calls sd_bus_match_signal() and
10101 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
10102 and sd_bus_add_match_async() but instead of taking a D-Bus match
10103 string take match fields as normal function parameters.
10105 * sd-bus gained two new calls sd_bus_set_sender() and
10106 sd_bus_message_set_sender() for setting the sender name of outgoing
10107 messages (either for all outgoing messages or for just one specific
10108 one). These calls are only useful in direct connections as on
10109 brokered connections the broker fills in the sender anyway,
10110 overwriting whatever the client filled in.
10112 * sd-event gained a new pseudo-handle that may be specified on all API
10113 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
10114 used this refers to the default event loop object of the calling
10115 thread. Note however that this does not implicitly allocate one —
10116 which has to be done prior by using sd_event_default(). Similarly
10117 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
10118 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
10119 to the default bus of the specified type of the calling thread. Here
10120 too this does not implicitly allocate bus connection objects, this
10121 has to be done prior with sd_bus_default() and friends.
10123 * sd-event gained a new call pair
10124 sd_event_source_{get|set}_io_fd_own(). This may be used to request
10125 automatic closure of the file descriptor an IO event source watches
10126 when the event source is destroyed.
10128 * systemd-networkd gained support for natively configuring WireGuard
10131 * In previous versions systemd synthesized user records both for the
10132 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
10133 internally. In order to simplify distribution-wide renames of the
10134 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
10135 new transitional flag file has been added: if
10136 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
10137 user and group record within the systemd codebase is disabled.
10139 * systemd-notify gained a new --uid= option for selecting the source
10140 user/UID to use for notification messages sent to the service
10143 * journalctl gained a new --grep= option to list only entries in which
10144 the message matches a certain pattern. By default matching is case
10145 insensitive if the pattern is lowercase, and case sensitive
10146 otherwise. Option --case-sensitive=yes|no can be used to override
10147 this an specify case sensitivity or case insensitivity.
10149 * There's now a "systemd-analyze service-watchdogs" command for printing
10150 the current state of the service runtime watchdog, and optionally
10151 enabling or disabling the per-service watchdogs system-wide if given a
10152 boolean argument (i.e. the concept you configure in WatchdogSec=), for
10153 debugging purposes. There's also a kernel command line option
10154 systemd.service_watchdogs= for controlling the same.
10156 * Two new "log-level" and "log-target" options for systemd-analyze were
10157 added that merge the now deprecated get-log-level, set-log-level and
10158 get-log-target, set-log-target pairs. The deprecated options are still
10159 understood for backwards compatibility. The two new options print the
10160 current value when no arguments are given, and set them when a
10161 level/target is given as an argument.
10163 * sysusers.d's "u" lines now optionally accept both a UID and a GID
10164 specification, separated by a ":" character, in order to create users
10165 where UID and GID do not match.
10167 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
10168 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
10169 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
10170 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
10171 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
10172 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
10173 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
10174 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
10175 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
10176 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
10177 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
10178 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
10179 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
10180 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
10181 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
10182 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
10183 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
10184 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
10185 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
10186 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
10193 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
10194 in v235 has been extended to also set the dummy.ko module option
10195 numdummies=0, preventing the kernel from automatically creating
10196 dummy0. All dummy interfaces must now be explicitly created.
10198 * Unknown '%' specifiers in configuration files are now rejected. This
10199 applies to units and tmpfiles.d configuration. Any percent characters
10200 that are followed by a letter or digit that are not supposed to be
10201 interpreted as the beginning of a specifier should be escaped by
10202 doubling ("%%"). (So "size=5%" is still accepted, as well as
10203 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
10204 valid specifiers today.)
10206 * systemd-resolved now maintains a new dynamic
10207 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
10208 recommended to make /etc/resolv.conf a symlink to it. This file
10209 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
10210 includes dynamically acquired search domains, achieving more correct
10211 DNS resolution by software that bypasses local DNS APIs such as NSS.
10213 * The "uaccess" udev tag has been dropped from /dev/kvm and
10214 /dev/dri/renderD*. These devices now have the 0666 permissions by
10215 default (but this may be changed at build-time). /dev/dri/renderD*
10216 will now be owned by the "render" group along with /dev/kfd.
10218 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
10219 systemd-journal-gatewayd.service and
10220 systemd-journal-upload.service. This means "nss-systemd" must be
10221 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
10222 services are resolved properly.
10224 * In /etc/fstab two new mount options are now understood:
10225 x-systemd.makefs and x-systemd.growfs. The former has the effect that
10226 the configured file system is formatted before it is mounted, the
10227 latter that the file system is resized to the full block device size
10228 after it is mounted (i.e. if the file system is smaller than the
10229 partition it resides on, it's grown). This is similar to the fsck
10230 logic in /etc/fstab, and pulls in systemd-makefs@.service and
10231 systemd-growfs@.service as necessary, similar to
10232 systemd-fsck@.service. Resizing is currently only supported on ext4
10235 * In systemd-networkd, the IPv6 RA logic now optionally may announce
10236 DNS server and domain information.
10238 * Support for the LUKS2 on-disk format for encrypted partitions has
10239 been added. This requires libcryptsetup2 during compilation and
10242 * The systemd --user instance will now signal "readiness" when its
10243 basic.target unit has been reached, instead of when the run queue ran
10244 empty for the first time.
10246 * Tmpfiles.d with user configuration are now also supported.
10247 systemd-tmpfiles gained a new --user switch, and snippets placed in
10248 ~/.config/user-tmpfiles.d/ and corresponding directories will be
10249 executed by systemd-tmpfiles --user running in the new
10250 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
10251 running in the user session.
10253 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
10254 %S resolves to the top-level state directory (/var/lib for the system
10255 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
10256 top-level cache directory (/var/cache for the system instance,
10257 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
10258 logs directory (/var/log for the system instance,
10259 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
10260 existing %t specifier, that resolves to the top-level runtime
10261 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
10264 * journalctl learnt a new parameter --output-fields= for limiting the
10265 set of journal fields to output in verbose and JSON output modes.
10267 * systemd-timesyncd's configuration file gained a new option
10268 RootDistanceMaxSec= for setting the maximum root distance of servers
10269 it'll use, as well as the new options PollIntervalMinSec= and
10270 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
10272 * bootctl gained a new command "list" for listing all available boot
10273 menu items on systems that follow the boot loader specification.
10275 * systemctl gained a new --dry-run switch that shows what would be done
10276 instead of doing it, and is currently supported by the shutdown and
10279 * ConditionSecurity= can now detect the TOMOYO security module.
10281 * Unit file [Install] sections are now also respected in unit drop-in
10282 files. This is intended to be used by drop-ins under /usr/lib/.
10284 * systemd-firstboot may now also set the initial keyboard mapping.
10286 * Udev "changed" events for devices which are exposed as systemd
10287 .device units are now propagated to units specified in
10288 ReloadPropagatedFrom= as reload requests.
10290 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
10291 unit template name (i.e. a name in the form of 'foobar@.service',
10292 without the instance component between the '@' and - the '.'), then
10293 the escaped sysfs path of the device is automatically used as the
10296 * SystemCallFilter= in unit files has been extended so that an "errno"
10297 can be specified individually for each system call. Example:
10298 SystemCallFilter=~uname:EILSEQ.
10300 * The cgroup delegation logic has been substantially updated. Delegate=
10301 now optionally takes a list of controllers (instead of a boolean, as
10302 before), which lists the controllers to delegate at least.
10304 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
10306 * A new LogLevelMax= setting configures the maximum log level any
10307 process of the service may log at (i.e. anything with a lesser
10308 priority than what is specified is automatically dropped). A new
10309 LogExtraFields= setting allows configuration of additional journal
10310 fields to attach to all log records generated by any of the unit's
10313 * New StandardInputData= and StandardInputText= settings along with the
10314 new option StandardInput=data may be used to configure textual or
10315 binary data that shall be passed to the executed service process via
10316 standard input, encoded in-line in the unit file.
10318 * StandardInput=, StandardOutput= and StandardError= may now be used to
10319 connect stdin/stdout/stderr of executed processes directly with a
10320 file or AF_UNIX socket in the file system, using the new "file:" option.
10322 * A new unit file option CollectMode= has been added, that allows
10323 tweaking the garbage collection logic for units. It may be used to
10324 tell systemd to garbage collect units that have failed automatically
10325 (normally it only GCs units that exited successfully). systemd-run
10326 and systemd-mount expose this new functionality with a new -G option.
10328 * "machinectl bind" may now be used to bind mount non-directories
10329 (i.e. regularfiles, devices, fifos, sockets).
10331 * systemd-analyze gained a new verb "calendar" for validating and
10332 testing calendar time specifications to use for OnCalendar= in timer
10333 units. Besides validating the expression it will calculate the next
10334 time the specified expression would elapse.
10336 * In addition to the pre-existing FailureAction= unit file setting
10337 there's now SuccessAction=, for configuring a shutdown action to
10338 execute when a unit completes successfully. This is useful in
10339 particular inside containers that shall terminate after some workload
10340 has been completed. Also, both options are now supported for all unit
10341 types, not just services.
10343 * networkds's IP rule support gained two new options
10344 IncomingInterface= and OutgoingInterface= for configuring the incoming
10345 and outgoing interfaces of configured rules. systemd-networkd also
10346 gained support for "vxcan" network devices.
10348 * networkd gained a new setting RequiredForOnline=, taking a
10349 boolean. If set, systemd-wait-online will take it into consideration
10350 when determining that the system is up, otherwise it will ignore the
10351 interface for this purpose.
10353 * The sd_notify() protocol gained support for a new operation: with
10354 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
10355 store again, ahead of POLLHUP or POLLERR when they are removed
10358 * A new document doc/UIDS-GIDS.md has been added to the source tree,
10359 that documents the UID/GID range and assignment assumptions and
10360 requirements of systemd.
10362 * The watchdog device PID 1 will ping may now be configured through the
10363 WatchdogDevice= configuration file setting, or by setting the
10364 systemd.watchdog_service= kernel command line option.
10366 * systemd-resolved's gained support for registering DNS-SD services on
10367 the local network using MulticastDNS. Services may either be
10368 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
10369 the same dir below /run, /usr/lib), or through its D-Bus API.
10371 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
10372 extend the effective start, runtime, and stop time. The service must
10373 continue to send EXTEND_TIMEOUT_USEC within the period specified to
10374 prevent the service manager from making the service as timedout.
10376 * systemd-resolved's DNSSEC support gained support for RFC 8080
10377 (Ed25519 keys and signatures).
10379 * The systemd-resolve command line tool gained a new set of options
10380 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
10381 --set-nta= and --revert to configure per-interface DNS configuration
10382 dynamically during runtime. It's useful for pushing DNS information
10383 into systemd-resolved from DNS hook scripts that various interface
10384 managing software supports (such as pppd).
10386 * systemd-nspawn gained a new --network-namespace-path= command line
10387 option, which may be used to make a container join an existing
10388 network namespace, by specifying a path to a "netns" file.
10390 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
10391 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
10392 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
10393 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
10394 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
10395 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
10396 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
10397 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
10398 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
10399 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
10400 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
10401 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
10402 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
10403 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
10404 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
10405 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
10406 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
10407 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
10408 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
10409 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
10410 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
10411 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
10412 Jędrzejewski-Szmek, Zeal Jagannatha
10414 — Berlin, 2017-12-14
10418 * INCOMPATIBILITY: systemd-logind.service and other long-running
10419 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
10420 communication with the outside. This generally improves security of
10421 the system, and is in almost all cases a safe and good choice, as
10422 these services do not and should not provide any network-facing
10423 functionality. However, systemd-logind uses the glibc NSS API to
10424 query the user database. This creates problems on systems where NSS
10425 is set up to directly consult network services for user database
10426 lookups. In particular, this creates incompatibilities with the
10427 "nss-nis" module, which attempts to directly contact the NIS/YP
10428 network servers it is configured for, and will now consistently
10429 fail. In such cases, it is possible to turn off IP sandboxing for
10430 systemd-logind.service (set IPAddressDeny= in its [Service] section
10431 to the empty string, via a .d/ unit file drop-in). Downstream
10432 distributions might want to update their nss-nis packaging to include
10433 such a drop-in snippet, accordingly, to hide this incompatibility
10434 from the user. Another option is to make use of glibc's nscd service
10435 to proxy such network requests through a privilege-separated, minimal
10436 local caching daemon, or to switch to more modern technologies such
10437 sssd, whose NSS hook-ups generally do not involve direct network
10438 access. In general, we think it's definitely time to question the
10439 implementation choices of nss-nis, i.e. whether it's a good idea
10440 today to embed a network-facing loadable module into all local
10441 processes that need to query the user database, including the most
10442 trivial and benign ones, such as "ls". For more details about
10443 IPAddressDeny= see below.
10445 * A new modprobe.d drop-in is now shipped by default that sets the
10446 bonding module option max_bonds=0. This overrides the kernel default,
10447 to avoid conflicts and ambiguity as to whether or not bond0 should be
10448 managed by systemd-networkd or not. This resolves multiple issues
10449 with bond0 properties not being applied, when bond0 is configured
10450 with systemd-networkd. Distributors may choose to not package this,
10451 however in that case users will be prevented from correctly managing
10452 bond0 interface using systemd-networkd.
10454 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
10455 which print the logging level and target of the system manager. They
10456 complement the existing "set-log-level" and "set-log-target" verbs
10457 used to change those values.
10459 * journald.conf gained a new boolean setting ReadKMsg= which defaults
10460 to on. If turned off kernel log messages will not be read by
10461 systemd-journald or included in the logs. It also gained a new
10462 setting LineMax= for configuring the maximum line length in
10463 STDOUT/STDERR log streams. The new default for this value is 48K, up
10464 from the previous hardcoded 2048.
10466 * A new unit setting RuntimeDirectoryPreserve= has been added, which
10467 allows more detailed control of what to do with a runtime directory
10468 configured with RuntimeDirectory= (i.e. a directory below /run or
10469 $XDG_RUNTIME_DIR) after a unit is stopped.
10471 * The RuntimeDirectory= setting for units gained support for creating
10472 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
10473 one top-level directory.
10475 * Units gained new options StateDirectory=, CacheDirectory=,
10476 LogsDirectory= and ConfigurationDirectory= which are closely related
10477 to RuntimeDirectory= but manage per-service directories below
10478 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
10479 possible to write unit files which when activated automatically gain
10480 properly owned service specific directories in these locations, thus
10481 making unit files self-contained and increasing compatibility with
10482 stateless systems and factory reset where /etc or /var are
10483 unpopulated at boot. Matching these new settings there's also
10484 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
10485 ConfigurationDirectoryMode= for configuring the access mode of these
10486 directories. These settings are particularly useful in combination
10487 with DynamicUser=yes as they provide secure, properly-owned,
10488 writable, and stateful locations for storage, excluded from the
10489 sandbox that such services live in otherwise.
10491 * Automake support has been removed from this release. systemd is now
10494 * systemd-journald will now aggressively cache client metadata during
10495 runtime, speeding up log write performance under pressure. This comes
10496 at a small price though: as much of the metadata is read
10497 asynchronously from /proc/ (and isn't implicitly attached to log
10498 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
10499 metadata stored alongside a log entry might be slightly
10500 out-of-date. Previously it could only be slightly newer than the log
10501 message. The time window is small however, and given that the kernel
10502 is unlikely to be improved anytime soon in this regard, this appears
10505 * nss-myhostname/systemd-resolved will now by default synthesize an
10506 A/AAAA resource record for the "_gateway" hostname, pointing to the
10507 current default IP gateway. Previously it did that for the "gateway"
10508 name, hampering adoption, as some distributions wanted to leave that
10509 hostname open for local use. The old behaviour may still be
10510 requested at build time.
10512 * systemd-networkd's [Address] section in .network files gained a new
10513 Scope= setting for configuring the IP address scope. The [Network]
10514 section gained a new boolean setting ConfigureWithoutCarrier= that
10515 tells systemd-networkd to ignore link sensing when configuring the
10516 device. The [DHCP] section gained a new Anonymize= boolean option for
10517 turning on a number of options suggested in RFC 7844. A new
10518 [RoutingPolicyRule] section has been added for configuring the IP
10519 routing policy. The [Route] section has gained support for a new
10520 Type= setting which permits configuring
10521 blackhole/unreachable/prohibit routes.
10523 * The [VRF] section in .netdev files gained a new Table= setting for
10524 configuring the routing table to use. The [Tunnel] section gained a
10525 new Independent= boolean field for configuring tunnels independent of
10526 an underlying network interface. The [Bridge] section gained a new
10527 GroupForwardMask= option for configuration of propagation of link
10528 local frames between bridge ports.
10530 * The WakeOnLan= setting in .link files gained support for a number of
10531 new modes. A new TCP6SegmentationOffload= setting has been added for
10532 configuring TCP/IPv6 hardware segmentation offload.
10534 * The IPv6 RA sender implementation may now optionally send out RDNSS
10535 and RDNSSL records to supply DNS configuration to peers.
10537 * systemd-nspawn gained support for a new --system-call-filter= command
10538 line option for adding and removing entries in the default system
10539 call filter it applies. Moreover systemd-nspawn has been changed to
10540 implement a system call allow list instead of a deny list.
10542 * systemd-run gained support for a new --pipe command line option. If
10543 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
10544 are directly passed on to the activated transient service
10545 executable. This allows invoking arbitrary processes as systemd
10546 services (for example to take benefit of dependency management,
10547 accounting management, resource management or log management that is
10548 done automatically for services) — while still allowing them to be
10549 integrated in a classic UNIX shell pipeline.
10551 * When a service sends RELOAD=1 via sd_notify() and reload propagation
10552 using ReloadPropagationTo= is configured, a reload is now propagated
10553 to configured units. (Previously this was only done on explicitly
10554 requested reloads, using "systemctl reload" or an equivalent
10557 * For each service unit a restart counter is now kept: it is increased
10558 each time the service is restarted due to Restart=, and may be
10559 queried using "systemctl show -p NRestarts …".
10561 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
10562 @signal and @timer have been added, for usage with SystemCallFilter=
10563 in unit files and the new --system-call-filter= command line option
10564 of systemd-nspawn (see above).
10566 * ExecStart= lines in unit files gained two new modifiers: when a
10567 command line is prefixed with "!" the command will be executed as
10568 configured, except for the credentials applied by
10569 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
10570 "+", but does still apply namespacing options unlike "+". There's
10571 also "!!" now, which is mostly identical, but becomes a NOP on
10572 systems that support ambient capabilities. This is useful to write
10573 unit files that work with ambient capabilities where possible but
10574 automatically fall back to traditional privilege dropping mechanisms
10575 on systems where this is not supported.
10577 * ListenNetlink= settings in socket units now support RDMA netlink
10580 * A new unit file setting LockPersonality= has been added which permits
10581 locking down the chosen execution domain ("personality") of a service
10584 * A new special target "getty-pre.target" has been added, which is
10585 ordered before all text logins, and may be used to order services
10586 before textual logins acquire access to the console.
10588 * systemd will now attempt to load the virtio-rng.ko kernel module very
10589 early on if a VM environment supporting this is detected. This should
10590 improve entropy during early boot in virtualized environments.
10592 * A _netdev option is now supported in /etc/crypttab that operates in a
10593 similar way as the same option in /etc/fstab: it permits configuring
10594 encrypted devices that need to be ordered after the network is up.
10595 Following this logic, two new special targets
10596 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
10597 added that are to cryptsetup.target what remote-fs.target and
10598 remote-fs-pre.target are to local-fs.target.
10600 * Service units gained a new UnsetEnvironment= setting which permits
10601 unsetting specific environment variables for services that are
10602 normally passed to it (for example in order to mask out locale
10603 settings for specific services that can't deal with it).
10605 * Units acquired a new boolean option IPAccounting=. When turned on, IP
10606 traffic accounting (packet count as well as byte count) is done for
10607 the service, and shown as part of "systemctl status" or "systemd-run
10610 * Service units acquired two new options IPAddressAllow= and
10611 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
10612 for configuring a simple IP access control list for all sockets of
10613 the unit. These options are available also on .slice and .socket
10614 units, permitting flexible access list configuration for individual
10615 services as well as groups of services (as defined by a slice unit),
10616 including system-wide. Note that IP ACLs configured this way are
10617 enforced on every single IPv4 and IPv6 socket created by any process
10618 of the service unit, and apply to ingress as well as egress traffic.
10620 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
10621 structured log message is generated each time the unit is stopped,
10622 containing information about the consumed resources of this
10625 * A new setting KeyringMode= has been added to unit files, which may be
10626 used to control how the kernel keyring is set up for executed
10629 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
10630 "systemctl kexec" and "systemctl exit" are now always asynchronous in
10631 behaviour (that is: these commands return immediately after the
10632 operation was enqueued instead of waiting for the operation to
10633 complete). Previously, "systemctl poweroff" and "systemctl reboot"
10634 were asynchronous on systems using systemd-logind (i.e. almost
10635 always, and like they were on sysvinit), and the other three commands
10636 were unconditionally synchronous. With this release this is cleaned
10637 up, and callers will see the same asynchronous behaviour on all
10638 systems for all five operations.
10640 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
10643 * .timer units now accept calendar specifications in other timezones
10644 than UTC or the local timezone.
10646 * The tmpfiles snippet var.conf has been changed to create
10647 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
10648 the "utmp" group already, and it appears to be generally understood
10649 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
10650 databases. Previously this was implemented correctly for all these
10651 databases excepts btmp, which has been opened up like this now
10652 too. Note that while the other databases are world-readable
10653 (i.e. 0644), btmp is not and remains more restrictive.
10655 * The systemd-resolve tool gained a new --reset-server-features
10656 switch. When invoked like this systemd-resolved will forget
10657 everything it learnt about the features supported by the configured
10658 upstream DNS servers, and restarts the feature probing logic on the
10659 next resolver look-up for them at the highest feature level
10662 * The status dump systemd-resolved sends to the logs upon receiving
10663 SIGUSR1 now also includes information about all DNS servers it is
10664 configured to use, and the features levels it probed for them.
10666 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
10667 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
10668 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
10669 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
10670 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
10671 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
10672 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
10673 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
10674 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
10675 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
10676 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
10677 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
10678 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
10679 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
10680 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
10681 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
10682 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
10683 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
10684 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
10685 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
10687 — Berlin, 2017-10-06
10691 * Meson is now supported as build system in addition to Automake. It is
10692 our plan to remove Automake in one of our next releases, so that
10693 Meson becomes our exclusive build system. Hence, please start using
10694 the Meson build system in your downstream packaging. There's plenty
10695 of documentation around how to use Meson, the extremely brief
10698 ./autogen.sh && ./configure && make && sudo make install
10702 meson build && ninja -C build && sudo ninja -C build install
10704 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
10705 which permits configuring a timeout on the time a job is
10706 running. This is particularly useful for setting timeouts on jobs for
10709 * Unit files gained two new options ConditionUser= and ConditionGroup=
10710 for conditionalizing units based on the identity of the user/group
10711 running a systemd user instance.
10713 * systemd-networkd now understands a new FlowLabel= setting in the
10714 [VXLAN] section of .network files, as well as a Priority= in
10715 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
10716 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
10717 gained support for configuration of GENEVE links, and IPv6 address
10718 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
10720 * .link files now understand a new Port= setting.
10722 * systemd-networkd's DHCP support gained support for DHCP option 119
10723 (domain search list).
10725 * systemd-networkd gained support for serving IPv6 address ranges using
10726 the Router Advertisement protocol. The new .network configuration
10727 section [IPv6Prefix] may be used to configure the ranges to
10728 serve. This is implemented based on a new, minimal, native server
10729 implementation of RA.
10731 * journalctl's --output= switch gained support for a new parameter
10732 "short-iso-precise" for a mode where timestamps are shown as precise
10735 * systemd-udevd's "net_id" builtin may now generate stable network
10736 interface names from IBM PowerVM VIO devices as well as ACPI platform
10739 * MulticastDNS support in systemd-resolved may now be explicitly
10740 enabled/disabled using the new MulticastDNS= configuration file
10743 * systemd-resolved may now optionally use libidn2 instead of the libidn
10744 for processing internationalized domain names. Support for libidn2
10745 should be considered experimental and should not be enabled by
10748 * "machinectl pull-tar" and related call may now do verification of
10749 downloaded images using SUSE-style .sha256 checksum files in addition
10750 to the already existing support for validating using Ubuntu-style
10753 * sd-bus gained support for a new sd_bus_message_appendv() call which
10754 is va_list equivalent of sd_bus_message_append().
10756 * sd-boot gained support for validating images using SHIM/MOK.
10758 * The SMACK code learnt support for "onlycap".
10760 * systemd-mount --umount is now much smarter in figuring out how to
10761 properly unmount a device given its mount or device path.
10763 * The code to call libnss_dns as a fallback from libnss_resolve when
10764 the communication with systemd-resolved fails was removed. This
10765 fallback was redundant and interfered with the [!UNAVAIL=return]
10766 suffix. See nss-resolve(8) for the recommended configuration.
10768 * systemd-logind may now be restarted without losing state. It stores
10769 the file descriptors for devices it manages in the system manager
10770 using the FDSTORE= mechanism. Please note that further changes in
10771 other components may be required to make use of this (for example
10772 Xorg has code to listen for stops of systemd-logind and terminate
10773 itself when logind is stopped or restarted, in order to avoid using
10774 stale file descriptors for graphical devices, which is now
10775 counterproductive and must be reverted in order for restarts of
10776 systemd-logind to be safe. See
10777 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
10779 * All kernel-install plugins are called with the environment variable
10780 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
10781 /etc/machine-id. If the machine ID could not be determined,
10782 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
10783 anything in the entry directory (passed as the second argument) if
10784 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
10785 temporary directory is passed as the entry directory and removed
10786 after all the plugins exit.
10788 * If KERNEL_INSTALL_MACHINE_ID is set in /etc/machine-info, kernel-install
10789 will now use its value as the machine ID instead of the machine ID
10790 from /etc/machine-id. If KERNEL_INSTALL_MACHINE_ID isn't set in
10791 /etc/machine-info and no machine ID is set in /etc/machine-id,
10792 kernel-install will try to store the current machine ID there as
10793 KERNEL_INSTALL_MACHINE_ID. If there is no machine ID, kernel-install
10794 will generate a new UUID, store it in /etc/machine-info as
10795 KERNEL_INSTALL_MACHINE_ID and use it as the machine ID.
10797 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
10798 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
10799 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
10800 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
10801 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
10802 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
10803 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
10804 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
10805 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
10806 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
10807 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
10808 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
10809 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
10810 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
10811 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
10812 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
10813 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
10814 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
10815 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
10816 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
10817 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
10818 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
10819 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
10820 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
10821 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
10822 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
10823 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
10824 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
10825 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
10828 — Berlin, 2017-07-12
10832 * The "hybrid" control group mode has been modified to improve
10833 compatibility with "legacy" cgroups-v1 setups. Specifically, the
10834 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
10835 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
10836 cgroups-v1 hierarchy), the only externally visible change being that
10837 the cgroups-v2 hierarchy is also mounted, to
10838 /sys/fs/cgroup/unified. This should provide a large degree of
10839 compatibility with "legacy" cgroups-v1, while taking benefit of the
10840 better management capabilities of cgroups-v2.
10842 * The default control group setup mode may be selected both a boot-time
10843 via a set of kernel command line parameters (specifically:
10844 systemd.unified_cgroup_hierarchy= and
10845 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
10846 default selected on the configure command line
10847 (--with-default-hierarchy=). The upstream default is "hybrid"
10848 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
10849 this will change in a future systemd version to be "unified" (pure
10850 cgroups-v2 mode). The third option for the compile time option is
10851 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
10852 distributions to default to "hybrid" mode for release distributions,
10853 starting with v233. We recommend "unified" for development
10854 distributions (specifically: distributions such as Fedora's rawhide)
10855 as that's where things are headed in the long run. Use "legacy" for
10856 greatest stability and compatibility only.
10858 * Note one current limitation of "unified" and "hybrid" control group
10859 setup modes: the kernel currently does not permit the systemd --user
10860 instance (i.e. unprivileged code) to migrate processes between two
10861 disconnected cgroup subtrees, even if both are managed and owned by
10862 the user. This effectively means "systemd-run --user --scope" doesn't
10863 work when invoked from outside of any "systemd --user" service or
10864 scope. Specifically, it is not supported from session scopes. We are
10865 working on fixing this in a future systemd version. (See #3388 for
10866 further details about this.)
10868 * DBus policy files are now installed into /usr rather than /etc. Make
10869 sure your system has dbus >= 1.9.18 running before upgrading to this
10870 version, or override the install path with --with-dbuspolicydir= .
10872 * All python scripts shipped with systemd (specifically: the various
10873 tests written in Python) now require Python 3.
10875 * systemd unit tests can now run standalone (without the source or
10876 build directories), and can be installed into /usr/lib/systemd/tests/
10877 with 'make install-tests'.
10879 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
10880 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
10883 * Support for the %c, %r, %R specifiers in unit files has been
10884 removed. Specifiers are not supposed to be dependent on configuration
10885 in the unit file itself (so that they resolve the same regardless
10886 where used in the unit files), but these specifiers were influenced
10887 by the Slice= option.
10889 * The shell invoked by debug-shell.service now defaults to /bin/sh in
10890 all cases. If distributions want to use a different shell for this
10891 purpose (for example Fedora's /sbin/sushell) they need to specify
10892 this explicitly at configure time using --with-debug-shell=.
10894 * The confirmation spawn prompt has been reworked to offer the
10897 (c)ontinue, proceed without asking anymore
10898 (D)ump, show the state of the unit
10899 (f)ail, don't execute the command and pretend it failed
10901 (i)nfo, show a short summary of the unit
10902 (j)obs, show jobs that are in progress
10903 (s)kip, don't execute the command and pretend it succeeded
10904 (y)es, execute the command
10906 The 'n' choice for the confirmation spawn prompt has been removed,
10907 because its meaning was confusing.
10909 The prompt may now also be redirected to an alternative console by
10910 specifying the console as parameter to systemd.confirm_spawn=.
10912 * Services of Type=notify require a READY=1 notification to be sent
10913 during startup. If no such message is sent, the service now fails,
10914 even if the main process exited with a successful exit code.
10916 * Services that fail to start up correctly now always have their
10917 ExecStopPost= commands executed. Previously, they'd enter "failed"
10918 state directly, without executing these commands.
10920 * The option MulticastDNS= of network configuration files has acquired
10921 an actual implementation. With MulticastDNS=yes a host can resolve
10922 names of remote hosts and reply to mDNS A and AAAA requests.
10924 * When units are about to be started an additional check is now done to
10925 ensure that all dependencies of type BindsTo= (when used in
10926 combination with After=) have been started.
10928 * systemd-analyze gained a new verb "syscall-filter" which shows which
10929 system call groups are defined for the SystemCallFilter= unit file
10930 setting, and which system calls they contain.
10932 * A new system call filter group "@filesystem" has been added,
10933 consisting of various file system related system calls. Group
10934 "@reboot" has been added, covering reboot, kexec and shutdown related
10935 calls. Finally, group "@swap" has been added covering swap
10936 configuration related calls.
10938 * A new unit file option RestrictNamespaces= has been added that may be
10939 used to restrict access to the various process namespace types the
10940 Linux kernel provides. Specifically, it may be used to take away the
10941 right for a service unit to create additional file system, network,
10942 user, and other namespaces. This sandboxing option is particularly
10943 relevant due to the high amount of recently discovered namespacing
10944 related vulnerabilities in the kernel.
10946 * systemd-udev's .link files gained support for a new AutoNegotiation=
10947 setting for configuring Ethernet auto-negotiation.
10949 * systemd-networkd's .network files gained support for a new
10950 ListenPort= setting in the [DHCP] section to explicitly configure the
10951 UDP client port the DHCP client shall listen on.
10953 * .network files gained a new Unmanaged= boolean setting for explicitly
10954 excluding one or more interfaces from management by systemd-networkd.
10956 * The systemd-networkd ProxyARP= option has been renamed to
10957 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
10958 renamed to ReduceARPProxy=. The old names continue to be available
10961 * systemd-networkd gained support for configuring IPv6 Proxy NDP
10962 addresses via the new IPv6ProxyNDPAddress= .network file setting.
10964 * systemd-networkd's bonding device support gained support for two new
10965 configuration options ActiveSlave= and PrimarySlave=.
10967 * The various options in the [Match] section of .network files gained
10968 support for negative matching.
10970 * New systemd-specific mount options are now understood in /etc/fstab:
10972 x-systemd.mount-timeout= may be used to configure the maximum
10973 permitted runtime of the mount command.
10975 x-systemd.device-bound may be set to bind a mount point to its
10976 backing device unit, in order to automatically remove a mount point
10977 if its backing device is unplugged. This option may also be
10978 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
10979 on the block device, which is now automatically set for all CDROM
10980 drives, so that mounted CDs are automatically unmounted when they are
10981 removed from the drive.
10983 x-systemd.after= and x-systemd.before= may be used to explicitly
10984 order a mount after or before another unit or mount point.
10986 * Enqueued start jobs for device units are now automatically garbage
10987 collected if there are no jobs waiting for them anymore.
10989 * systemctl list-jobs gained two new switches: with --after, for every
10990 queued job the jobs it's waiting for are shown; with --before the
10991 jobs which it's blocking are shown.
10993 * systemd-nspawn gained support for ephemeral boots from disk images
10994 (or in other words: --ephemeral and --image= may now be
10995 combined). Moreover, ephemeral boots are now supported for normal
10996 directories, even if the backing file system is not btrfs. Of course,
10997 if the file system does not support file system snapshots or
10998 reflinks, the initial copy operation will be relatively expensive, but
10999 this should still be suitable for many use cases.
11001 * Calendar time specifications in .timer units now support
11002 specifications relative to the end of a month by using "~" instead of
11003 "-" as separator between month and day. For example, "*-02~03" means
11004 "the third last day in February". In addition a new syntax for
11005 repeated events has been added using the "/" character. For example,
11006 "9..17/2:00" means "every two hours from 9am to 5pm".
11008 * systemd-socket-proxyd gained a new parameter --connections-max= for
11009 configuring the maximum number of concurrent connections.
11011 * sd-id128 gained a new API for generating unique IDs for the host in a
11012 way that does not leak the machine ID. Specifically,
11013 sd_id128_get_machine_app_specific() derives an ID based on the
11014 machine ID in a well-defined, non-reversible, stable way. This is
11015 useful whenever an identifier for the host is needed but where the
11016 identifier shall not be useful to identify the system beyond the
11017 scope of the application itself. (Internally this uses HMAC-SHA256 as
11018 keyed hash function using the machine ID as input.)
11020 * NotifyAccess= gained a new supported value "exec". When set
11021 notifications are accepted from all processes systemd itself invoked,
11022 including all control processes.
11024 * .nspawn files gained support for defining overlay mounts using the
11025 Overlay= and OverlayReadOnly= options. Previously this functionality
11026 was only available on the systemd-nspawn command line.
11028 * systemd-nspawn's --bind= and --overlay= options gained support for
11029 bind/overlay mounts whose source lies within the container tree by
11030 prefixing the source path with "+".
11032 * systemd-nspawn's --bind= and --overlay= options gained support for
11033 automatically allocating a temporary source directory in /var/tmp
11034 that is removed when the container dies. Specifically, if the source
11035 directory is specified as empty string this mechanism is selected. An
11036 example usage is --overlay=+/var::/var, which creates an overlay
11037 mount based on the original /var contained in the image, overlaid
11038 with a temporary directory in the host's /var/tmp. This way changes
11039 to /var are automatically flushed when the container shuts down.
11041 * systemd-nspawn --image= option does now permit raw file system block
11042 devices (in addition to images containing partition tables, as
11045 * The disk image dissection logic in systemd-nspawn gained support for
11046 automatically setting up LUKS encrypted as well as Verity protected
11047 partitions. When a container is booted from an encrypted image the
11048 passphrase is queried at start-up time. When a container with Verity
11049 data is started, the root hash is search in a ".roothash" file
11050 accompanying the disk image (alternatively, pass the root hash via
11051 the new --root-hash= command line option).
11053 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
11054 be used to dissect disk images the same way as systemd-nspawn does
11055 it, following the Bootable Partition Specification. It may even be
11056 used to mount disk images with complex partition setups (including
11057 LUKS and Verity partitions) to a local host directory, in order to
11058 inspect them. This tool is not considered public API (yet), and is
11059 thus not installed into /usr/bin. Please do not rely on its
11060 existence, since it might go away or be changed in later systemd
11063 * A new generator "systemd-verity-generator" has been added, similar in
11064 style to "systemd-cryptsetup-generator", permitting automatic setup of
11065 Verity root partitions when systemd boots up. In order to make use of
11066 this your partition setup should follow the Discoverable Partitions
11067 Specification, and the GPT partition ID of the root file system
11068 partition should be identical to the upper 128-bit of the Verity root
11069 hash. The GPT partition ID of the Verity partition protecting it
11070 should be the lower 128-bit of the Verity root hash. If the partition
11071 image follows this model it is sufficient to specify a single
11072 "roothash=" kernel command line argument to both configure which root
11073 image and verity partition to use as well as the root hash for
11074 it. Note that systemd-nspawn's Verity support follows the same
11075 semantics, meaning that disk images with proper Verity data in place
11076 may be booted in containers with systemd-nspawn as well as on
11077 physical systems via the verity generator. Also note that the "mkosi"
11078 tool available at https://github.com/systemd/mkosi has been updated
11079 to generate Verity protected disk images following this scheme. In
11080 fact, it has been updated to generate disk images that optionally
11081 implement a complete UEFI SecureBoot trust chain, involving a signed
11082 kernel and initrd image that incorporates such a root hash as well as
11083 a Verity-enabled root partition.
11085 * The hardware database (hwdb) udev supports has been updated to carry
11086 accelerometer quirks.
11088 * All system services are now run with a fresh kernel keyring set up
11089 for them. The invocation ID is stored by default in it, thus
11090 providing a safe, non-overridable way to determine the invocation
11091 ID of each service.
11093 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
11094 options for bind mounting arbitrary paths in a service-specific
11095 way. When these options are used, arbitrary host or service files and
11096 directories may be mounted to arbitrary locations in the service's
11099 * Documentation has been added that lists all of systemd's low-level
11100 environment variables:
11102 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
11104 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
11105 whether a specific socket file descriptor matches a specified socket
11108 * systemd-firstboot has been updated to check for the
11109 systemd.firstboot= kernel command line option. It accepts a boolean
11110 and when set to false the first boot questions are skipped.
11112 * systemd-fstab-generator has been updated to check for the
11113 systemd.volatile= kernel command line option, which either takes an
11114 optional boolean parameter or the special value "state". If used the
11115 system may be booted in a "volatile" boot mode. Specifically,
11116 "systemd.volatile" is used, the root directory will be mounted as
11117 tmpfs, and only /usr is mounted from the actual root file system. If
11118 "systemd.volatile=state" is used, the root directory will be mounted
11119 as usual, but /var is mounted as tmpfs. This concept provides similar
11120 functionality as systemd-nspawn's --volatile= option, but provides it
11121 on physical boots. Use this option for implementing stateless
11122 systems, or testing systems with all state and/or configuration reset
11123 to the defaults. (Note though that many distributions are not
11124 prepared to boot up without a populated /etc or /var, though.)
11126 * systemd-gpt-auto-generator gained support for LUKS encrypted root
11127 partitions. Previously it only supported LUKS encrypted partitions
11128 for all other uses, except for the root partition itself.
11130 * Socket units gained support for listening on AF_VSOCK sockets for
11131 communication in virtualized QEMU environments.
11133 * The "configure" script gained a new option --with-fallback-hostname=
11134 for specifying the fallback hostname to use if none is configured in
11135 /etc/hostname. For example, by specifying
11136 --with-fallback-hostname=fedora it is possible to default to a
11137 hostname of "fedora" on pristine installations.
11139 * systemd-cgls gained support for a new --unit= switch for listing only
11140 the control groups of a specific unit. Similar --user-unit= has been
11141 added for listing only the control groups of a specific user unit.
11143 * systemd-mount gained a new --umount switch for unmounting a mount or
11144 automount point (and all mount/automount points below it).
11146 * systemd will now refuse full configuration reloads (via systemctl
11147 daemon-reload and related calls) unless at least 16MiB of free space
11148 are available in /run. This is a safety precaution in order to ensure
11149 that generators can safely operate after the reload completed.
11151 * A new unit file option RootImage= has been added, which has a similar
11152 effect as RootDirectory= but mounts the service's root directory from
11153 a disk image instead of plain directory. This logic reuses the same
11154 image dissection and mount logic that systemd-nspawn already uses,
11155 and hence supports any disk images systemd-nspawn supports, including
11156 those following the Discoverable Partition Specification, as well as
11157 Verity enabled images. This option enables systemd to run system
11158 services directly off disk images acting as resource bundles,
11159 possibly even including full integrity data.
11161 * A new MountAPIVFS= unit file option has been added, taking a boolean
11162 argument. If enabled /proc, /sys and /dev (collectively called the
11163 "API VFS") will be mounted for the service. This is only relevant if
11164 RootDirectory= or RootImage= is used for the service, as these mounts
11165 are of course in place in the host mount namespace anyway.
11167 * systemd-nspawn gained support for a new --pivot-root= switch. If
11168 specified the root directory within the container image is pivoted to
11169 the specified mount point, while the original root disk is moved to a
11170 different place. This option enables booting of ostree images
11171 directly with systemd-nspawn.
11173 * The systemd build scripts will no longer complain if the NTP server
11174 addresses are not changed from the defaults. Google now supports
11175 these NTP servers officially. We still recommend downstreams to
11176 properly register an NTP pool with the NTP pool project though.
11178 * coredumpctl gained a new "--reverse" option for printing the list
11179 of coredumps in reverse order.
11181 * coredumpctl will now show additional information about truncated and
11182 inaccessible coredumps, as well as coredumps that are still being
11183 processed. It also gained a new --quiet switch for suppressing
11184 additional informational message in its output.
11186 * coredumpctl gained support for only showing coredumps newer and/or
11187 older than specific timestamps, using the new --since= and --until=
11188 options, reminiscent of journalctl's options by the same name.
11190 * The systemd-coredump logic has been improved so that it may be reused
11191 to collect backtraces in non-compiled languages, for example in
11192 scripting languages such as Python.
11194 * machinectl will now show the UID shift of local containers, if user
11195 namespacing is enabled for them.
11197 * systemd will now optionally run "environment generator" binaries at
11198 configuration load time. They may be used to add environment
11199 variables to the environment block passed to services invoked. One
11200 user environment generator is shipped by default that sets up
11201 environment variables based on files dropped into /etc/environment.d
11202 and ~/.config/environment.d/.
11204 * systemd-resolved now includes the new, recently published 2017 DNSSEC
11207 * hostnamed has been updated to report a new chassis type of
11208 "convertible" to cover "foldable" laptops that can both act as a
11209 tablet and as a laptop, such as various Lenovo Yoga devices.
11211 Contributions from: Adrián López, Alexander Galanin, Alexander
11212 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
11213 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
11214 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
11215 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
11216 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
11217 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
11218 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
11219 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
11220 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
11221 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
11222 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
11223 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
11224 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
11225 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
11226 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
11227 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
11228 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
11229 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
11230 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
11231 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
11232 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
11233 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
11234 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
11235 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
11236 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
11237 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
11238 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
11241 — Berlin, 2017-03-01
11245 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
11246 RestrictAddressFamilies= enabled. These sandboxing options should
11247 generally be compatible with the various external udev call-out
11248 binaries we are aware of, however there may be exceptions, in
11249 particular when exotic languages for these call-outs are used. In
11250 this case, consider turning off these settings locally.
11252 * The new RemoveIPC= option can be used to remove IPC objects owned by
11253 the user or group of a service when that service exits.
11255 * The new ProtectKernelModules= option can be used to disable explicit
11256 load and unload operations of kernel modules by a service. In
11257 addition access to /usr/lib/modules is removed if this option is set.
11259 * ProtectSystem= option gained a new value "strict", which causes the
11260 whole file system tree with the exception of /dev, /proc, and /sys,
11261 to be remounted read-only for a service.
11263 * The new ProtectKernelTunables= option can be used to disable
11264 modification of configuration files in /sys and /proc by a service.
11265 Various directories and files are remounted read-only, so access is
11266 restricted even if the file permissions would allow it.
11268 * The new ProtectControlGroups= option can be used to disable write
11269 access by a service to /sys/fs/cgroup.
11271 * Various systemd services have been hardened with
11272 ProtectKernelTunables=yes, ProtectControlGroups=yes,
11273 RestrictAddressFamilies=.
11275 * Support for dynamically creating users for the lifetime of a service
11276 has been added. If DynamicUser=yes is specified, user and group IDs
11277 will be allocated from the range 61184…65519 for the lifetime of the
11278 service. They can be resolved using the new nss-systemd.so NSS
11279 module. The module must be enabled in /etc/nsswitch.conf. Services
11280 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
11281 any resources allocated by the service will be cleaned up when the
11282 service exits. They also have ProtectHome=read-only and
11283 ProtectSystem=strict enabled, so they are not able to make any
11284 permanent modifications to the system.
11286 * The nss-systemd module also always resolves root and nobody, making
11287 it possible to have no /etc/passwd or /etc/group files in minimal
11288 container or chroot environments.
11290 * Services may be started with their own user namespace using the new
11291 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
11292 under which the service is running are mapped. All other users are
11295 * Support for the cgroup namespace has been added to systemd-nspawn. If
11296 supported by kernel, the container system started by systemd-nspawn
11297 will have its own view of the cgroup hierarchy. This new behaviour
11298 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
11300 * The new MemorySwapMax= option can be used to limit the maximum swap
11301 usage under the unified cgroup hierarchy.
11303 * Support for the CPU controller in the unified cgroup hierarchy has
11304 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
11305 options. This controller requires out-of-tree patches for the kernel
11306 and the support is provisional.
11308 * Mount and automount units may now be created transiently
11309 (i.e. dynamically at runtime via the bus API, instead of requiring
11310 unit files in the file system).
11312 * systemd-mount is a new tool which may mount file systems – much like
11313 mount(8), optionally pulling in additional dependencies through
11314 transient .mount and .automount units. For example, this tool
11315 automatically runs fsck on a backing block device before mounting,
11316 and allows the automount logic to be used dynamically from the
11317 command line for establishing mount points. This tool is particularly
11318 useful when dealing with removable media, as it will ensure fsck is
11319 run – if necessary – before the first access and that the file system
11320 is quickly unmounted after each access by utilizing the automount
11321 logic. This maximizes the chance that the file system on the
11322 removable media stays in a clean state, and if it isn't in a clean
11323 state is fixed automatically.
11325 * LazyUnmount=yes option for mount units has been added to expose the
11326 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
11329 * /efi will be used as the mount point of the EFI boot partition, if
11330 the directory is present, and the mount point was not configured
11331 through other means (e.g. fstab). If /efi directory does not exist,
11332 /boot will be used as before. This makes it easier to automatically
11333 mount the EFI partition on systems where /boot is used for something
11336 * When operating on GPT disk images for containers, systemd-nspawn will
11337 now mount the ESP to /boot or /efi according to the same rules as PID
11338 1 running on a host. This allows tools like "bootctl" to operate
11339 correctly within such containers, in order to make container images
11340 bootable on physical systems.
11342 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
11344 * Two new user session targets have been added to support running
11345 graphical sessions under the systemd --user instance:
11346 graphical-session.target and graphical-session-pre.target. See
11347 systemd.special(7) for a description of how those targets should be
11350 * The vconsole initialization code has been significantly reworked to
11351 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
11352 support unicode keymaps. Font and keymap configuration will now be
11353 copied to all allocated virtual consoles.
11355 * FreeBSD's bhyve virtualization is now detected.
11357 * Information recorded in the journal for core dumps now includes the
11358 contents of /proc/mountinfo and the command line of the process at
11359 the top of the process hierarchy (which is usually the init process
11362 * systemd-journal-gatewayd learned the --directory= option to serve
11363 files from the specified location.
11365 * journalctl --root=… can be used to peruse the journal in the
11366 /var/log/ directories inside of a container tree. This is similar to
11367 the existing --machine= option, but does not require the container to
11370 * The hardware database has been extended to support
11371 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
11374 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
11375 specify the click rate for mice which include a horizontal wheel with
11376 a click rate that is different than the one for the vertical wheel.
11378 * systemd-run gained a new --wait option that makes service execution
11379 synchronous. (Specifically, the command will not return until the
11380 specified service binary exited.)
11382 * systemctl gained a new --wait option that causes the start command to
11383 wait until the units being started have terminated again.
11385 * A new journal output mode "short-full" has been added which displays
11386 timestamps with abbreviated English day names and adds a timezone
11387 suffix. Those timestamps include more information than the default
11388 "short" output mode, and can be passed directly to journalctl's
11389 --since= and --until= options.
11391 * /etc/resolv.conf will be bind-mounted into containers started by
11392 systemd-nspawn, if possible, so any changes to resolv.conf contents
11393 are automatically propagated to the container.
11395 * The number of instances for socket-activated services originating
11396 from a single IP address can be limited with
11397 MaxConnectionsPerSource=, extending the existing setting of
11400 * systemd-networkd gained support for vcan ("Virtual CAN") interface
11403 * .netdev and .network configuration can now be extended through
11406 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
11407 Segmentation Offload, Generic Receive Offload, Large Receive Offload
11408 can be enabled and disabled using the new UDPSegmentationOffload=,
11409 TCPSegmentationOffload=, GenericSegmentationOffload=,
11410 GenericReceiveOffload=, LargeReceiveOffload= options in the
11411 [Link] section of .link files.
11413 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
11414 Port VLAN ID can be configured for bridge devices using the new STP=,
11415 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
11416 section of .netdev files.
11418 * The route table to which routes received over DHCP or RA should be
11419 added can be configured with the new RouteTable= option in the [DHCP]
11420 and [IPv6AcceptRA] sections of .network files.
11422 * The Address Resolution Protocol can be disabled on links managed by
11423 systemd-networkd using the ARP=no setting in the [Link] section of
11426 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
11427 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
11428 encode information about the result and exit codes of the current
11429 service runtime cycle.
11431 * systemd-sysctl will now configure kernel parameters in the order
11432 they occur in the configuration files. This matches what sysctl
11433 has been traditionally doing.
11435 * kernel-install "plugins" that are executed to perform various
11436 tasks after a new kernel is added and before an old one is removed
11437 can now return a special value to terminate the procedure and
11438 prevent any later plugins from running.
11440 * Journald's SplitMode=login setting has been deprecated. It has been
11441 removed from documentation, and its use is discouraged. In a future
11442 release it will be completely removed, and made equivalent to current
11443 default of SplitMode=uid.
11445 * Storage=both option setting in /etc/systemd/coredump.conf has been
11446 removed. With fast LZ4 compression storing the core dump twice is not
11449 * The --share-system systemd-nspawn option has been replaced with an
11450 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
11451 this functionality is discouraged. In addition the variables
11452 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
11453 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
11454 individual namespaces.
11456 * "machinectl list" now shows the IP address of running containers in
11457 the output, as well as OS release information.
11459 * "loginctl list" now shows the TTY of each session in the output.
11461 * sd-bus gained new API calls sd_bus_track_set_recursive(),
11462 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
11463 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
11464 tracking objects in a "recursive" mode, where a single client can be
11465 counted multiple times, if it takes multiple references.
11467 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
11468 sd_bus_get_exit_on_disconnect(). They may be used to make a
11469 process using sd-bus automatically exit if the bus connection is
11472 * Bus clients of the service manager may now "pin" loaded units into
11473 memory, by taking an explicit reference on them. This is useful to
11474 ensure the client can retrieve runtime data about the service even
11475 after the service completed execution. Taking such a reference is
11476 available only for privileged clients and should be helpful to watch
11477 running services in a race-free manner, and in particular collect
11478 information about exit statuses and results.
11480 * The nss-resolve module has been changed to strictly return UNAVAIL
11481 when communication via D-Bus with resolved failed, and NOTFOUND when
11482 a lookup completed but was negative. This means it is now possible to
11483 neatly configure fallbacks using nsswitch.conf result checking
11484 expressions. Taking benefit of this, the new recommended
11485 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
11487 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
11489 * A new setting CtrlAltDelBurstAction= has been added to
11490 /etc/systemd/system.conf which may be used to configure the precise
11491 behaviour if the user on the console presses Ctrl-Alt-Del more often
11492 than 7 times in 2s. Previously this would unconditionally result in
11493 an expedited, immediate reboot. With this new setting the precise
11494 operation may be configured in more detail, and also turned off
11497 * In .netdev files two new settings RemoteChecksumTx= and
11498 RemoteChecksumRx= are now understood that permit configuring the
11499 remote checksumming logic for VXLAN networks.
11501 * The service manager learnt a new "invocation ID" concept for invoked
11502 services. Each runtime cycle of a service will get a new invocation
11503 ID (a 128-bit random UUID) assigned that identifies the current
11504 run of the service uniquely and globally. A new invocation ID
11505 is generated each time a service starts up. The journal will store
11506 the invocation ID of a service along with any logged messages, thus
11507 making the invocation ID useful for matching the online runtime of a
11508 service with the offline log data it generated in a safe way without
11509 relying on synchronized timestamps. In many ways this new service
11510 invocation ID concept is similar to the kernel's boot ID concept that
11511 uniquely and globally identifies the runtime of each boot. The
11512 invocation ID of a service is passed to the service itself via an
11513 environment variable ($INVOCATION_ID). A new bus call
11514 GetUnitByInvocationID() has been added that is similar to GetUnit()
11515 but instead of retrieving the bus path for a unit by its name
11516 retrieves it by its invocation ID. The returned path is valid only as
11517 long as the passed invocation ID is current.
11519 * systemd-resolved gained a new "DNSStubListener" setting in
11520 resolved.conf. It either takes a boolean value or the special values
11521 "udp" and "tcp", and configures whether to enable the stub DNS
11522 listener on 127.0.0.53:53.
11524 * IP addresses configured via networkd may now carry additional
11525 configuration settings supported by the kernel. New options include:
11526 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
11527 PrefixRoute=, AutoJoin=.
11529 * The PAM configuration fragment file for "user@.service" shipped with
11530 systemd (i.e. the --user instance of systemd) has been stripped to
11531 the minimum necessary to make the system boot. Previously, it
11532 contained Fedora-specific stanzas that did not apply to other
11533 distributions. It is expected that downstream distributions add
11534 additional configuration lines, matching their needs to this file,
11535 using it only as rough template of what systemd itself needs. Note
11536 that this reduced fragment does not even include an invocation of
11537 pam_limits which most distributions probably want to add, even though
11538 systemd itself does not need it. (There's also the new build time
11539 option --with-pamconfdir=no to disable installation of the PAM
11540 fragment entirely.)
11542 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
11543 capability is now also dropped from its set (in addition to
11544 CAP_SYS_MKNOD as before).
11546 * In service unit files it is now possible to connect a specific named
11547 file descriptor with stdin/stdout/stdout of an executed service. The
11548 name may be specified in matching .socket units using the
11549 FileDescriptorName= setting.
11551 * A number of journal settings may now be configured on the kernel
11552 command line. Specifically, the following options are now understood:
11553 systemd.journald.max_level_console=,
11554 systemd.journald.max_level_store=,
11555 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
11556 systemd.journald.max_level_wall=.
11558 * "systemctl is-enabled --full" will now show by which symlinks a unit
11559 file is enabled in the unit dependency tree.
11561 * Support for VeraCrypt encrypted partitions has been added to the
11562 "cryptsetup" logic and /etc/crypttab.
11564 * systemd-detect-virt gained support for a new --private-users switch
11565 that checks whether the invoking processes are running inside a user
11566 namespace. Similar, a new special value "private-users" for the
11567 existing ConditionVirtualization= setting has been added, permitting
11568 skipping of specific units in user namespace environments.
11570 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
11571 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
11572 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
11573 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
11574 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
11575 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
11576 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
11577 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
11578 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
11579 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
11580 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
11581 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
11582 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
11583 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
11584 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
11585 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
11586 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
11587 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
11588 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
11589 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
11590 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
11591 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
11592 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
11593 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
11594 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
11595 Jędrzejewski-Szmek, Zeal Jagannatha
11597 — Santa Fe, 2016-11-03
11601 * In service units the various ExecXYZ= settings have been extended
11602 with an additional special character as first argument of the
11603 assigned value: if the character '+' is used the specified command
11604 line it will be run with full privileges, regardless of User=,
11605 Group=, CapabilityBoundingSet= and similar options. The effect is
11606 similar to the existing PermissionsStartOnly= option, but allows
11607 configuration of this concept for each executed command line
11610 * Services may now alter the service watchdog timeout at runtime by
11611 sending a WATCHDOG_USEC= message via sd_notify().
11613 * MemoryLimit= and related unit settings now optionally take percentage
11614 specifications. The percentage is taken relative to the amount of
11615 physical memory in the system (or in case of containers, the assigned
11616 amount of memory). This allows scaling service resources neatly with
11617 the amount of RAM available on the system. Similarly, systemd-logind's
11618 RuntimeDirectorySize= option now also optionally takes percentage
11621 * In similar fashion TasksMax= takes percentage values now, too. The
11622 value is taken relative to the configured maximum number of processes
11623 on the system. The per-service task maximum has been changed to 15%
11624 using this functionality. (Effectively this is an increase of 512 →
11625 4915 for service units, given the kernel's default pid_max setting.)
11627 * Calendar time specifications in .timer units now understand a ".."
11628 syntax for time ranges. Example: "4..7:10" may now be used for
11629 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
11632 * The InaccessableDirectories=, ReadOnlyDirectories= and
11633 ReadWriteDirectories= unit file settings have been renamed to
11634 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
11635 applied to all kinds of file nodes, and not just directories, with
11636 the exception of symlinks. Specifically these settings may now be
11637 used on block and character device nodes, UNIX sockets and FIFOS as
11638 well as regular files. The old names of these settings remain
11639 available for compatibility.
11641 * systemd will now log about all service processes it kills forcibly
11642 (using SIGKILL) because they remained after the clean shutdown phase
11643 of the service completed. This should help identifying services that
11644 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
11645 systemd-logind's configuration a similar log message is generated for
11646 processes killed at the end of each session due to this setting.
11648 * systemd will now set the $JOURNAL_STREAM environment variable for all
11649 services whose stdout/stderr are connected to the Journal (which
11650 effectively means by default: all services). The variable contains
11651 the device and inode number of the file descriptor used for
11652 stdout/stderr. This may be used by invoked programs to detect whether
11653 their stdout/stderr is connected to the Journal, in which case they
11654 can switch over to direct Journal communication, thus being able to
11655 pass extended, structured metadata along with their log messages. As
11656 one example, this is now used by glib's logging primitives.
11658 * When using systemd's default tmp.mount unit for /tmp, the mount point
11659 will now be established with the "nosuid" and "nodev" options. This
11660 avoids privilege escalation attacks that put traps and exploits into
11661 /tmp. However, this might cause problems if you e.g. put container
11662 images or overlays into /tmp; if you need this, override tmp.mount's
11663 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
11666 * systemd now supports the "memory" cgroup controller also on
11669 * The systemd-cgtop tool now optionally takes a control group path as
11670 command line argument. If specified, the control group list shown is
11671 limited to subgroups of that group.
11673 * The SystemCallFilter= unit file setting gained support for
11674 pre-defined, named system call filter sets. For example
11675 SystemCallFilter=@clock is now an effective way to make all clock
11676 changing-related system calls unavailable to a service. A number of
11677 similar pre-defined groups are defined. Writing system call filters
11678 for system services is simplified substantially with this new
11679 concept. Accordingly, all of systemd's own, long-running services now
11680 enable system call filtering based on this, by default.
11682 * A new service setting MemoryDenyWriteExecute= has been added, taking
11683 a boolean value. If turned on, a service may no longer create memory
11684 mappings that are writable and executable at the same time. This
11685 enhances security for services where this is enabled as it becomes
11686 harder to dynamically write and then execute memory in exploited
11687 service processes. This option has been enabled for all of systemd's
11688 own long-running services.
11690 * A new RestrictRealtime= service setting has been added, taking a
11691 boolean argument. If set the service's processes may no longer
11692 acquire realtime scheduling. This improves security as realtime
11693 scheduling may otherwise be used to easily freeze the system.
11695 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
11696 value. This may be used for requesting that the system manager inside
11697 of the container reports start-up completion to nspawn which then
11698 propagates this notification further to the service manager
11699 supervising nspawn itself. A related option NotifyReady= in .nspawn
11700 files has been added too. This functionality allows ordering of the
11701 start-up of multiple containers using the usual systemd ordering
11704 * machinectl gained a new command "stop" that is an alias for
11707 * systemd-resolved gained support for contacting DNS servers on
11708 link-local IPv6 addresses.
11710 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
11711 its caches. A method call for requesting the same operation has been
11712 added to the bus API too, and is made available via "systemd-resolve
11715 * systemd-resolve gained a new --status switch. If passed a brief
11716 summary of the used DNS configuration with per-interface information
11719 * resolved.conf gained a new Cache= boolean option, defaulting to
11720 on. If turned off local DNS caching is disabled. This comes with a
11721 performance penalty in particular when DNSSEC is enabled. Note that
11722 resolved disables its internal caching implicitly anyway, when the
11723 configured DNS server is on a host-local IP address such as ::1 or
11724 127.0.0.1, thus automatically avoiding double local caching.
11726 * systemd-resolved now listens on the local IP address 127.0.0.53:53
11727 for DNS requests. This improves compatibility with local programs
11728 that do not use the libc NSS or systemd-resolved's bus APIs for name
11729 resolution. This minimal DNS service is only available to local
11730 programs and does not implement the full DNS protocol, but enough to
11731 cover local DNS clients. A new, static resolv.conf file, listing just
11732 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
11733 now recommended to make /etc/resolv.conf a symlink to this file in
11734 order to route all DNS lookups to systemd-resolved, regardless if
11735 done via NSS, the bus API or raw DNS packets. Note that this local
11736 DNS service is not as fully featured as the libc NSS or
11737 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
11738 used to deliver link-local address information (as this implies
11739 sending a local interface index along), LLMNR/mDNS support via this
11740 interface is severely restricted. It is thus strongly recommended for
11741 all applications to use the libc NSS API or native systemd-resolved
11744 * systemd-networkd's bridge support learned a new setting
11745 VLANFiltering= for controlling VLAN filtering. Moreover a new section
11746 in .network files has been added for configuring VLAN bridging in
11747 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
11749 * systemd-networkd's IPv6 Router Advertisement code now makes use of
11750 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
11751 now be acquired without relying on DHCPv6. Two new options
11752 UseDomains= and UseDNS= have been added to configure this behaviour.
11754 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
11755 renamed IPv6AcceptRA=, without altering its behaviour. The old
11756 setting name remains available for compatibility reasons.
11758 * The systemd-networkd VTI/VTI6 tunneling support gained new options
11759 Key=, InputKey= and OutputKey=.
11761 * systemd-networkd gained support for VRF ("Virtual Routing Function")
11762 interface configuration.
11764 * "systemctl edit" may now be used to create new unit files by
11765 specifying the --force switch.
11767 * sd-event gained a new function sd_event_get_iteration() for
11768 requesting the current iteration counter of the event loop. It starts
11769 at zero and is increased by one with each event loop iteration.
11771 * A new rpm macro %systemd_ordering is provided by the macros.systemd
11772 file. It can be used in lieu of %systemd_requires in packages which
11773 don't use any systemd functionality and are intended to be installed
11774 in minimal containers without systemd present. This macro provides
11775 ordering dependencies to ensure that if the package is installed in
11776 the same rpm transaction as systemd, systemd will be installed before
11777 the scriptlets for the package are executed, allowing unit presets
11780 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
11781 been added to simplify packaging of generators.
11783 * The os-release file gained VERSION_CODENAME field for the
11784 distribution nickname (e.g. VERSION_CODENAME=woody).
11786 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
11787 can be set to disable parsing of metadata and the creation
11788 of persistent symlinks for that device.
11790 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
11791 to make them available to logged-in users has been reverted.
11793 * Much of the common code of the various systemd components is now
11794 built into an internal shared library libsystemd-shared-231.so
11795 (incorporating the systemd version number in the name, to be updated
11796 with future releases) that the components link to. This should
11797 decrease systemd footprint both in memory during runtime and on
11798 disk. Note that the shared library is not for public use, and is
11799 neither API nor ABI stable, but is likely to change with every new
11800 released update. Packagers need to make sure that binaries
11801 linking to libsystemd-shared.so are updated in step with the
11804 * Configuration for "mkosi" is now part of the systemd
11805 repository. mkosi is a tool to easily build legacy-free OS images,
11806 and is available on github: https://github.com/systemd/mkosi. If
11807 "mkosi" is invoked in the build tree a new raw OS image is generated
11808 incorporating the systemd sources currently being worked on and a
11809 clean, fresh distribution installation. The generated OS image may be
11810 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
11811 UEFI PC. This functionality is particularly useful to easily test
11812 local changes made to systemd in a pristine, defined environment. See
11813 doc/HACKING for details.
11815 * configure learned the --with-support-url= option to specify the
11816 distribution's bugtracker.
11818 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
11819 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
11820 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
11821 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
11822 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
11823 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
11824 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
11825 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
11826 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
11827 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
11828 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
11829 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
11830 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
11831 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
11832 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
11833 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
11834 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
11835 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
11836 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
11838 — Berlin, 2016-07-25
11842 * DNSSEC is now turned on by default in systemd-resolved (in
11843 "allow-downgrade" mode), but may be turned off during compile time by
11844 passing "--with-default-dnssec=no" to "configure" (and of course,
11845 during runtime with DNSSEC= in resolved.conf). We recommend
11846 downstreams to leave this on at least during development cycles and
11847 report any issues with the DNSSEC logic upstream. We are very
11848 interested in collecting feedback about the DNSSEC validator and its
11849 limitations in the wild. Note however, that DNSSEC support is
11850 probably nothing downstreams should turn on in stable distros just
11851 yet, as it might create incompatibilities with a few DNS servers and
11852 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
11853 automatically whenever we detect such incompatible setups, but there
11854 might be systems we do not cover yet. Hence: please help us testing
11855 the DNSSEC code, leave this on where you can, report back, but then
11856 again don't consider turning this on in your stable, LTS or
11857 production release just yet. (Note that you have to enable
11858 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
11859 and its DNSSEC mode for hostname resolution from local
11862 * systemd-resolve conveniently resolves DANE records with the --tlsa
11863 option and OPENPGPKEY records with the --openpgp option. It also
11864 supports dumping raw DNS record data via the new --raw= switch.
11866 * systemd-logind will now by default terminate user processes that are
11867 part of the user session scope unit (session-XX.scope) when the user
11868 logs out. This behavior is controlled by the KillUserProcesses=
11869 setting in logind.conf, and the previous default of "no" is now
11870 changed to "yes". This means that user sessions will be properly
11871 cleaned up after, but additional steps are necessary to allow
11872 intentionally long-running processes to survive logout.
11874 While the user is logged in at least once, user@.service is running,
11875 and any service that should survive the end of any individual login
11876 session can be started at a user service or scope using systemd-run.
11877 systemd-run(1) man page has been extended with an example which shows
11878 how to run screen in a scope unit underneath user@.service. The same
11879 command works for tmux.
11881 After the user logs out of all sessions, user@.service will be
11882 terminated too, by default, unless the user has "lingering" enabled.
11883 To effectively allow users to run long-term tasks even if they are
11884 logged out, lingering must be enabled for them. See loginctl(1) for
11885 details. The default polkit policy was modified to allow users to
11886 set lingering for themselves without authentication.
11888 Previous defaults can be restored at compile time by the
11889 --without-kill-user-processes option to "configure".
11891 * systemd-logind gained new configuration settings SessionsMax= and
11892 InhibitorsMax=, both with a default of 8192. It will not register new
11893 user sessions or inhibitors above this limit.
11895 * systemd-logind will now reload configuration on SIGHUP.
11897 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
11898 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
11899 enable. Also, support for the "io" cgroup controller in the unified
11900 hierarchy has been added, so that the "memory", "pids" and "io" are
11901 now the controllers that are supported on the unified hierarchy.
11903 WARNING: it is not possible to use previous systemd versions with
11904 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
11905 is necessary to also update systemd in the initramfs if using the
11906 unified hierarchy. An updated SELinux policy is also required.
11908 * LLDP support has been extended, and both passive (receive-only) and
11909 active (sender) modes are supported. Passive mode ("routers-only") is
11910 enabled by default in systemd-networkd. Active LLDP mode is enabled
11911 by default for containers on the internal network. The "networkctl
11912 lldp" command may be used to list information gathered. "networkctl
11913 status" will also show basic LLDP information on connected peers now.
11915 * The IAID and DUID unique identifier sent in DHCP requests may now be
11916 configured for the system and each .network file managed by
11917 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
11919 * systemd-networkd gained support for configuring proxy ARP support for
11920 each interface, via the ProxyArp= setting in .network files. It also
11921 gained support for configuring the multicast querier feature of
11922 bridge devices, via the new MulticastQuerier= setting in .netdev
11923 files. Similarly, snooping on the IGMP traffic can be controlled
11924 via the new setting MulticastSnooping=.
11926 A new setting PreferredLifetime= has been added for addresses
11927 configured in .network file to configure the lifetime intended for an
11930 The systemd-networkd DHCP server gained the option EmitRouter=, which
11931 defaults to yes, to configure whether the DHCP Option 3 (Router)
11934 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
11935 systemd-socket-activate and installed into /usr/bin. It is now fully
11938 * systemd-journald now uses separate threads to flush changes to disk
11939 when closing journal files, thus reducing impact of slow disk I/O on
11940 logging performance.
11942 * The sd-journal API gained two new calls
11943 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
11944 can be used to open journal files using file descriptors instead of
11945 file or directory paths. sd_journal_open_container() has been
11946 deprecated, sd_journal_open_directory_fd() should be used instead
11947 with the flag SD_JOURNAL_OS_ROOT.
11949 * journalctl learned a new output mode "-o short-unix" that outputs log
11950 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
11951 UTC). It also gained support for a new --no-hostname setting to
11952 suppress the hostname column in the family of "short" output modes.
11954 * systemd-ask-password now optionally skips printing of the password to
11955 stdout with --no-output which can be useful in scripts.
11957 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
11958 (devices tagged with ID_MAKER_TOOL) are now tagged with
11959 "uaccess" and are available to logged in users.
11961 * The DeviceAllow= unit setting now supports specifiers (with "%").
11963 * "systemctl show" gained a new --value switch, which allows print a
11964 only the contents of a specific unit property, without also printing
11965 the property's name. Similar support was added to "show*" verbs
11966 of loginctl and machinectl that output "key=value" lists.
11968 * A new unit type "generated" was added for files dynamically generated
11969 by generator tools. Similarly, a new unit type "transient" is used
11970 for unit files created using the runtime API. "systemctl enable" will
11971 refuse to operate on such files.
11973 * A new command "systemctl revert" has been added that may be used to
11974 revert to the vendor version of a unit file, in case local changes
11975 have been made by adding drop-ins or overriding the unit file.
11977 * "machinectl clean" gained a new verb to automatically remove all or
11978 just hidden container images.
11980 * systemd-tmpfiles gained support for a new line type "e" for emptying
11981 directories, if they exist, without creating them if they don't.
11983 * systemd-nspawn gained support for automatically patching the UID/GIDs
11984 of the owners and the ACLs of all files and directories in a
11985 container tree to match the UID/GID user namespacing range selected
11986 for the container invocation. This mode is enabled via the new
11987 --private-users-chown switch. It also gained support for
11988 automatically choosing a free, previously unused UID/GID range when
11989 starting a container, via the new --private-users=pick setting (which
11990 implies --private-users-chown). Together, these options for the first
11991 time make user namespacing for nspawn containers fully automatic and
11992 thus deployable. The systemd-nspawn@.service template unit file has
11993 been changed to use this functionality by default.
11995 * systemd-nspawn gained a new --network-zone= switch, that allows
11996 creating ad-hoc virtual Ethernet links between multiple containers,
11997 that only exist as long as at least one container referencing them is
11998 running. This allows easy connecting of multiple containers with a
11999 common link that implements an Ethernet broadcast domain. Each of
12000 these network "zones" may be named relatively freely by the user, and
12001 may be referenced by any number of containers, but each container may
12002 only reference one of these "zones". On the lower level, this is
12003 implemented by an automatically managed bridge network interface for
12004 each zone, that is created when the first container referencing its
12005 zone is created and removed when the last one referencing its zone
12008 * The default start timeout may now be configured on the kernel command
12009 line via systemd.default_timeout_start_sec=. It was already
12010 configurable via the DefaultTimeoutStartSec= option in
12011 /etc/systemd/system.conf.
12013 * Socket units gained a new TriggerLimitIntervalSec= and
12014 TriggerLimitBurst= setting to configure a limit on the activation
12015 rate of the socket unit.
12017 * The LimitNICE= setting now optionally takes normal UNIX nice values
12018 in addition to the raw integer limit value. If the specified
12019 parameter is prefixed with "+" or "-" and is in the range -20…19 the
12020 value is understood as UNIX nice value. If not prefixed like this it
12021 is understood as raw RLIMIT_NICE limit.
12023 * Note that the effect of the PrivateDevices= unit file setting changed
12024 slightly with this release: the per-device /dev file system will be
12025 mounted read-only from this version on, and will have "noexec"
12026 set. This (minor) change of behavior might cause some (exceptional)
12027 legacy software to break, when PrivateDevices=yes is set for its
12028 service. Please leave PrivateDevices= off if you run into problems
12031 * systemd-bootchart has been split out to a separate repository:
12032 https://github.com/systemd/systemd-bootchart
12034 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
12035 merged into the kernel in its current form.
12037 * The compatibility libraries libsystemd-daemon.so,
12038 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
12039 which have been deprecated since systemd-209 have been removed along
12040 with the corresponding pkg-config files. All symbols provided by
12041 those libraries are provided by libsystemd.so.
12043 * The Capabilities= unit file setting has been removed (it is ignored
12044 for backwards compatibility). AmbientCapabilities= and
12045 CapabilityBoundingSet= should be used instead.
12047 * A new special target has been added, initrd-root-device.target,
12048 which creates a synchronization point for dependencies of the root
12049 device in early userspace. Initramfs builders must ensure that this
12050 target is now included in early userspace.
12052 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
12053 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
12054 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
12055 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
12056 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
12057 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
12058 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
12059 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
12060 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
12061 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
12062 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
12063 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
12064 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
12065 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
12066 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
12067 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
12068 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
12069 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
12070 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
12071 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
12072 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
12073 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
12074 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
12075 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
12076 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
12079 — Fairfax, 2016-05-21
12083 * The systemd-resolved DNS resolver service has gained a substantial
12084 set of new features, most prominently it may now act as a DNSSEC
12085 validating stub resolver. DNSSEC mode is currently turned off by
12086 default, but is expected to be turned on by default in one of the
12087 next releases. For now, we invite everybody to test the DNSSEC logic
12088 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
12089 service also gained a full set of D-Bus interfaces, including calls
12090 to configure DNS and DNSSEC settings per link (for use by external
12091 network management software). systemd-resolved and systemd-networkd
12092 now distinguish between "search" and "routing" domains. The former
12093 are used to qualify single-label names, the latter are used purely
12094 for routing lookups within certain domains to specific links.
12095 resolved now also synthesizes RRs for all entries from /etc/hosts.
12097 * The systemd-resolve tool (which is a client utility for
12098 systemd-resolved) has been improved considerably and is now fully
12099 supported and documented. Hence it has moved from /usr/lib/systemd to
12102 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
12105 * The coredump collection logic has been reworked: when a coredump is
12106 collected it is now written to disk, compressed and processed
12107 (including stacktrace extraction) from a new instantiated service
12108 systemd-coredump@.service, instead of directly from the
12109 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
12110 processing large coredumps can take up a substantial amount of
12111 resources and time, and this previously happened entirely outside of
12112 systemd's service supervision. With the new logic the core_pattern
12113 hook only does minimal metadata collection before passing off control
12114 to the new instantiated service, which is configured with a time
12115 limit, a nice level and other settings to minimize negative impact on
12116 the rest of the system. Also note that the new logic will honour the
12117 RLIMIT_CORE setting of the crashed process, which now allows users
12118 and processes to turn off coredumping for their processes by setting
12121 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
12122 and all forked processes by default. Previously, PID 1 would leave
12123 the setting at "0" for all processes, as set by the kernel. Note that
12124 the resource limit traditionally has no effect on the generated
12125 coredumps on the system if the /proc/sys/kernel/core_pattern hook
12126 logic is used. Since the limit is now honoured (see above) its
12127 default has been changed so that the coredumping logic is enabled by
12128 default for all processes, while allowing specific opt-out.
12130 * When the stacktrace is extracted from processes of system users, this
12131 is now done as "systemd-coredump" user, in order to sandbox this
12132 potentially security sensitive parsing operation. (Note that when
12133 processing coredumps of normal users this is done under the user ID
12134 of process that crashed, as before.) Packagers should take notice
12135 that it is now necessary to create the "systemd-coredump" system user
12136 and group at package installation time.
12138 * The systemd-activate socket activation testing tool gained support
12139 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
12140 and --seqpacket switches. It also has been extended to support both
12141 new-style and inetd-style file descriptor passing. Use the new
12142 --inetd switch to request inetd-style file descriptor passing.
12144 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
12145 variable, which takes a boolean value. If set to false, ANSI color
12146 output is disabled in the tools even when run on a terminal that
12149 * The VXLAN support in networkd now supports two new settings
12150 DestinationPort= and PortRange=.
12152 * A new systemd.machine_id= kernel command line switch has been added,
12153 that may be used to set the machine ID in /etc/machine-id if it is
12154 not initialized yet. This command line option has no effect if the
12155 file is already initialized.
12157 * systemd-nspawn gained a new --as-pid2 switch that invokes any
12158 specified command line as PID 2 rather than PID 1 in the
12159 container. In this mode PID 1 is a minimal stub init process that
12160 implements the special POSIX and Linux semantics of PID 1 regarding
12161 signal and child process management. Note that this stub init process
12162 is implemented in nspawn itself and requires no support from the
12163 container image. This new logic is useful to support running
12164 arbitrary commands in the container, as normal processes are
12165 generally not prepared to run as PID 1.
12167 * systemd-nspawn gained a new --chdir= switch for setting the current
12168 working directory for the process started in the container.
12170 * "journalctl /dev/sda" will now output all kernel log messages for
12171 specified device from the current boot, in addition to all devices
12172 that are parents of it. This should make log output about devices
12173 pretty useful, as long as kernel drivers attach enough metadata to
12174 the log messages. (The usual SATA drivers do.)
12176 * The sd-journal API gained two new calls
12177 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
12178 that report whether log data from /run or /var has been found.
12180 * journalctl gained a new switch "--fields" that prints all journal
12181 record field names currently in use in the journal. This is backed
12182 by two new sd-journal API calls sd_journal_enumerate_fields() and
12183 sd_journal_restart_fields().
12185 * Most configurable timeouts in systemd now expect an argument of
12186 "infinity" to turn them off, instead of "0" as before. The semantics
12187 from now on is that a timeout of "0" means "now", and "infinity"
12188 means "never". To maintain backwards compatibility, "0" continues to
12189 turn off previously existing timeout settings.
12191 * "systemctl reload-or-try-restart" has been renamed to "systemctl
12192 try-reload-or-restart" to clarify what it actually does: the "try"
12193 logic applies to both reloading and restarting, not just restarting.
12194 The old name continues to be accepted for compatibility.
12196 * On boot-up, when PID 1 detects that the system clock is behind the
12197 release date of the systemd version in use, the clock is now set
12198 to the latter. Previously, this was already done in timesyncd, in order
12199 to avoid running with clocks set to the various clock epochs such as
12200 1902, 1938 or 1970. With this change the logic is now done in PID 1
12201 in addition to timesyncd during early boot-up, so that it is enforced
12202 before the first process is spawned by systemd. Note that the logic
12203 in timesyncd remains, as it is more comprehensive and ensures
12204 clock monotonicity by maintaining a persistent timestamp file in
12205 /var. Since /var is generally not available in earliest boot or the
12206 initrd, this part of the logic remains in timesyncd, and is not done
12209 * Support for tweaking details in net_cls.class_id through the
12210 NetClass= configuration directive has been removed, as the kernel
12211 people have decided to deprecate that controller in cgroup v2.
12212 Userspace tools such as nftables are moving over to setting rules
12213 that are specific to the full cgroup path of a task, which obsoletes
12214 these controllers anyway. The NetClass= directive is kept around for
12215 legacy compatibility reasons. For a more in-depth description of the
12216 kernel change, please refer to the respective upstream commit:
12218 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
12220 * A new service setting RuntimeMaxSec= has been added that may be used
12221 to specify a maximum runtime for a service. If the timeout is hit, the
12222 service is terminated and put into a failure state.
12224 * A new service setting AmbientCapabilities= has been added. It allows
12225 configuration of additional Linux process capabilities that are
12226 passed to the activated processes. This is only available on very
12229 * The process resource limit settings in service units may now be used
12230 to configure hard and soft limits individually.
12232 * The various libsystemd APIs such as sd-bus or sd-event now publicly
12233 expose support for gcc's __attribute__((cleanup())) C extension.
12234 Specifically, for many object destructor functions alternative
12235 versions have been added that have names suffixed with "p" and take a
12236 pointer to a pointer to the object to destroy, instead of just a
12237 pointer to the object itself. This is useful because these destructor
12238 functions may be used directly as parameters to the cleanup
12239 construct. Internally, systemd has been a heavy user of this GCC
12240 extension for a long time, and with this change similar support is
12241 now available to consumers of the library outside of systemd. Note
12242 that by using this extension in your sources compatibility with old
12243 and strictly ANSI compatible C compilers is lost. However, all gcc or
12244 LLVM versions of recent years support this extension.
12246 * Timer units gained support for a new setting RandomizedDelaySec= that
12247 allows configuring some additional randomized delay to the configured
12248 time. This is useful to spread out timer events to avoid load peaks in
12249 clusters or larger setups.
12251 * Calendar time specifications now support sub-second accuracy.
12253 * Socket units now support listening on SCTP and UDP-lite protocol
12256 * The sd-event API now comes with a full set of man pages.
12258 * Older versions of systemd contained experimental support for
12259 compressing journal files and coredumps with the LZ4 compressor that
12260 was not compatible with the lz4 binary (due to API limitations of the
12261 lz4 library). This support has been removed; only support for files
12262 compatible with the lz4 binary remains. This LZ4 logic is now
12263 officially supported and no longer considered experimental.
12265 * The dkr image import logic has been removed again from importd. dkr's
12266 micro-services focus doesn't fit into the machine image focus of
12267 importd, and quickly got out of date with the upstream dkr API.
12269 * Creation of the /run/lock/lockdev/ directory was dropped from
12270 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
12271 been available for many years. If you still need this, you need to
12272 create your own tmpfiles.d config file with:
12274 d /run/lock/lockdev 0775 root lock -
12276 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
12277 and RebootArgument= have been moved from the [Service] section of
12278 unit files to [Unit], and they are now supported on all unit types,
12279 not just service units. Of course, systemd will continue to
12280 understand these settings also at the old location, in order to
12281 maintain compatibility.
12283 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
12284 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
12285 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
12286 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
12287 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
12288 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
12289 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
12290 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
12291 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
12292 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
12293 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
12294 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
12295 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
12296 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
12297 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
12298 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
12299 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
12300 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
12301 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
12303 — Berlin, 2016-02-11
12307 * A number of properties previously only settable in unit
12308 files are now also available as properties to set when
12309 creating transient units programmatically via the bus, as it
12310 is exposed with systemd-run's --property=
12311 setting. Specifically, these are: SyslogIdentifier=,
12312 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
12313 EnvironmentFile=, ReadWriteDirectories=,
12314 ReadOnlyDirectories=, InaccessibleDirectories=,
12315 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
12317 * When creating transient services via the bus API it is now
12318 possible to pass in a set of file descriptors to use as
12319 STDIN/STDOUT/STDERR for the invoked process.
12321 * Slice units may now be created transiently via the bus APIs,
12322 similar to the way service and scope units may already be
12323 created transiently.
12325 * Wherever systemd expects a calendar timestamp specification
12326 (like in journalctl's --since= and --until= switches) UTC
12327 timestamps are now supported. Timestamps suffixed with "UTC"
12328 are now considered to be in Universal Time Coordinated
12329 instead of the local timezone. Also, timestamps may now
12330 optionally be specified with sub-second accuracy. Both of
12331 these additions also apply to recurring calendar event
12332 specification, such as OnCalendar= in timer units.
12334 * journalctl gained a new "--sync" switch that asks the
12335 journal daemon to write all so far unwritten log messages to
12336 disk and sync the files, before returning.
12338 * systemd-tmpfiles learned two new line types "q" and "Q" that
12339 operate like "v", but also set up a basic btrfs quota
12340 hierarchy when used on a btrfs file system with quota
12343 * tmpfiles' "v", "q" and "Q" will now create a plain directory
12344 instead of a subvolume (even on a btrfs file system) if the
12345 root directory is a plain directory, and not a
12346 subvolume. This should simplify things with certain chroot()
12347 environments which are not aware of the concept of btrfs
12350 * systemd-detect-virt gained a new --chroot switch to detect
12351 whether execution takes place in a chroot() environment.
12353 * CPUAffinity= now takes CPU index ranges in addition to
12354 individual indexes.
12356 * The various memory-related resource limit settings (such as
12357 LimitAS=) now understand the usual K, M, G, … suffixes to
12358 the base of 1024 (IEC). Similar, the time-related resource
12359 limit settings understand the usual min, h, day, … suffixes
12362 * There's a new system.conf setting DefaultTasksMax= to
12363 control the default TasksMax= setting for services and
12364 scopes running on the system. (TasksMax= is the primary
12365 setting that exposes the "pids" cgroup controller on systemd
12366 and was introduced in the previous systemd release.) The
12367 setting now defaults to 512, which means services that are
12368 not explicitly configured otherwise will only be able to
12369 create 512 processes or threads at maximum, from this
12370 version on. Note that this means that thread- or
12371 process-heavy services might need to be reconfigured to set
12372 TasksMax= to a higher value. It is sufficient to set
12373 TasksMax= in these specific unit files to a higher value, or
12374 even "infinity". Similar, there's now a logind.conf setting
12375 UserTasksMax= that defaults to 4096 and limits the total
12376 number of processes or tasks each user may own
12377 concurrently. nspawn containers also have the TasksMax=
12378 value set by default now, to 8192. Note that all of this
12379 only has an effect if the "pids" cgroup controller is
12380 enabled in the kernel. The general benefit of these changes
12381 should be a more robust and safer system, that provides a
12382 certain amount of per-service fork() bomb protection.
12384 * systemd-nspawn gained the new --network-veth-extra= switch
12385 to define additional and arbitrarily-named virtual Ethernet
12386 links between the host and the container.
12388 * A new service execution setting PassEnvironment= has been
12389 added that allows importing select environment variables
12390 from PID1's environment block into the environment block of
12393 * Timer units gained support for a new RemainAfterElapse=
12394 setting which takes a boolean argument. It defaults to on,
12395 exposing behaviour unchanged to previous releases. If set to
12396 off, timer units are unloaded after they elapsed if they
12397 cannot elapse again. This is particularly useful for
12398 transient timer units, which shall not stay around longer
12399 than until they first elapse.
12401 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
12402 default now (the kernel default is 16). This is beneficial
12403 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
12404 allows substantially larger numbers of queued
12405 datagrams. This should increase the capability of systemd to
12406 parallelize boot-up, as logging and sd_notify() are unlikely
12407 to stall execution anymore. If you need to change the value
12408 from the new defaults, use the usual sysctl.d/ snippets.
12410 * The compression framing format used by the journal or
12411 coredump processing has changed to be in line with what the
12412 official LZ4 tools generate. LZ4 compression support in
12413 systemd was considered unsupported previously, as the format
12414 was not compatible with the normal tools. With this release
12415 this has changed now, and it is hence safe for downstream
12416 distributions to turn it on. While not compressing as well
12417 as the XZ, LZ4 is substantially faster, which makes
12418 it a good default choice for the compression logic in the
12419 journal and in coredump handling.
12421 * Any reference to /etc/mtab has been dropped from
12422 systemd. The file has been obsolete since a while, but
12423 systemd refused to work on systems where it was incorrectly
12424 set up (it should be a symlink or non-existent). Please make
12425 sure to update to util-linux 2.27.1 or newer in conjunction
12426 with this systemd release, which also drops any reference to
12427 /etc/mtab. If you maintain a distribution make sure that no
12428 software you package still references it, as this is a
12429 likely source of bugs. There's also a glibc bug pending,
12430 asking for removal of any reference to this obsolete file:
12432 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
12434 Note that only util-linux versions built with
12435 --enable-libmount-force-mountinfo are supported.
12437 * Support for the ".snapshot" unit type has been removed. This
12438 feature turned out to be little useful and little used, and
12439 has now been removed from the core and from systemctl.
12441 * The dependency types RequiresOverridable= and
12442 RequisiteOverridable= have been removed from systemd. They
12443 have been used only very sparingly to our knowledge and
12444 other options that provide a similar effect (such as
12445 systemctl --mode=ignore-dependencies) are much more useful
12446 and commonly used. Moreover, they were only half-way
12447 implemented as the option to control behaviour regarding
12448 these dependencies was never added to systemctl. By removing
12449 these dependency types the execution engine becomes a bit
12450 simpler. Unit files that use these dependencies should be
12451 changed to use the non-Overridable dependency types
12452 instead. In fact, when parsing unit files with these
12453 options, that's what systemd will automatically convert them
12454 too, but it will also warn, asking users to fix the unit
12455 files accordingly. Removal of these dependency types should
12456 only affect a negligible number of unit files in the wild.
12458 * Behaviour of networkd's IPForward= option changed
12459 (again). It will no longer maintain a per-interface setting,
12460 but propagate one way from interfaces where this is enabled
12461 to the global kernel setting. The global setting will be
12462 enabled when requested by a network that is set up, but
12463 never be disabled again. This change was made to make sure
12464 IPv4 and IPv6 behaviour regarding packet forwarding is
12465 similar (as the Linux IPv6 stack does not support
12466 per-interface control of this setting) and to minimize
12469 * In unit files the behaviour of %u, %U, %h, %s has
12470 changed. These specifiers will now unconditionally resolve
12471 to the various user database fields of the user that the
12472 systemd instance is running as, instead of the user
12473 configured in the specific unit via User=. Note that this
12474 effectively doesn't change much, as resolving of these
12475 specifiers was already turned off in the --system instance
12476 of systemd, as we cannot do NSS lookups from PID 1. In the
12477 --user instance of systemd these specifiers where correctly
12478 resolved, but hardly made any sense, since the user instance
12479 lacks privileges to do user switches anyway, and User= is
12480 hence useless. Moreover, even in the --user instance of
12481 systemd behaviour was awkward as it would only take settings
12482 from User= assignment placed before the specifier into
12483 account. In order to unify and simplify the logic around
12484 this the specifiers will now always resolve to the
12485 credentials of the user invoking the manager (which in case
12486 of PID 1 is the root user).
12488 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
12489 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
12490 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
12491 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
12492 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
12493 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
12494 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
12495 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
12496 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
12497 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
12498 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
12499 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
12500 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
12501 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
12504 — Berlin, 2015-11-18
12508 * systemd now depends on util-linux v2.27. More specifically,
12509 the newly added mount monitor feature in libmount now
12510 replaces systemd's former own implementation.
12512 * libmount mandates /etc/mtab not to be regular file, and
12513 systemd now enforces this condition at early boot.
12514 /etc/mtab has been deprecated and warned about for a very
12515 long time, so systems running systemd should already have
12516 stopped having this file around as anything else than a
12517 symlink to /proc/self/mounts.
12519 * Support for the "pids" cgroup controller has been added. It
12520 allows accounting the number of tasks in a cgroup and
12521 enforcing limits on it. This adds two new setting
12522 TasksAccounting= and TasksMax= to each unit, as well as a
12523 global option DefaultTasksAccounting=.
12525 * Support for the "net_cls" cgroup controller has been added.
12526 It allows assigning a net class ID to each task in the
12527 cgroup, which can then be used in firewall rules and traffic
12528 shaping configurations. Note that the kernel netfilter net
12529 class code does not currently work reliably for ingress
12530 packets on unestablished sockets.
12532 This adds a new config directive called NetClass= to CGroup
12533 enabled units. Allowed values are positive numbers for fixed
12534 assignments and "auto" for picking a free value
12537 * 'systemctl is-system-running' now returns 'offline' if the
12538 system is not booted with systemd. This command can now be
12539 used as a substitute for 'systemd-notify --booted'.
12541 * Watchdog timeouts have been increased to 3 minutes for all
12542 in-tree service files. Apparently, disk IO issues are more
12543 frequent than we hoped, and user reported >1 minute waiting
12546 * 'machine-id-commit' functionality has been merged into
12547 'machine-id-setup --commit'. The separate binary has been
12550 * The WorkingDirectory= directive in unit files may now be set
12551 to the special value '~'. In this case, the working
12552 directory is set to the home directory of the user
12553 configured in User=.
12555 * "machinectl shell" will now open the shell in the home
12556 directory of the selected user by default.
12558 * The CrashChVT= configuration file setting is renamed to
12559 CrashChangeVT=, following our usual logic of not
12560 abbreviating unnecessarily. The old directive is still
12561 supported for compat reasons. Also, this directive now takes
12562 an integer value between 1 and 63, or a boolean value. The
12563 formerly supported '-1' value for disabling stays around for
12566 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
12567 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
12568 RootDirectory= properties can now be set for transient
12571 * The systemd-analyze tool gained a new "set-log-target" verb
12572 to change the logging target the system manager logs to
12573 dynamically during runtime. This is similar to how
12574 "systemd-analyze set-log-level" already changes the log
12577 * In nspawn /sys is now mounted as tmpfs, with only a selected
12578 set of subdirectories mounted in from the real sysfs. This
12579 enhances security slightly, and is useful for ensuring user
12580 namespaces work correctly.
12582 * Support for USB FunctionFS activation has been added. This
12583 allows implementation of USB gadget services that are
12584 activated as soon as they are requested, so that they don't
12585 have to run continuously, similar to classic socket
12588 * The "systemctl exit" command now optionally takes an
12589 additional parameter that sets the exit code to return from
12590 the systemd manager when exiting. This is only relevant when
12591 running the systemd user instance, or when running the
12592 system instance in a container.
12594 * sd-bus gained the new API calls sd_bus_path_encode_many()
12595 and sd_bus_path_decode_many() that allow easy encoding and
12596 decoding of multiple identifier strings inside a D-Bus
12597 object path. Another new call sd_bus_default_flush_close()
12598 has been added to flush and close per-thread default
12601 * systemd-cgtop gained support for a -M/--machine= switch to
12602 show the control groups within a certain container only.
12604 * "systemctl kill" gained support for an optional --fail
12605 switch. If specified the requested operation will fail of no
12606 processes have been killed, because the unit had no
12607 processes attached, or similar.
12609 * A new systemd.crash_reboot=1 kernel command line option has
12610 been added that triggers a reboot after crashing. This can
12611 also be set through CrashReboot= in systemd.conf.
12613 * The RuntimeDirectory= setting now understands unit
12614 specifiers like %i or %f.
12616 * A new (still internal) library API sd-ipv4acd has been added,
12617 that implements address conflict detection for IPv4. It's
12618 based on code from sd-ipv4ll, and will be useful for
12619 detecting DHCP address conflicts.
12621 * File descriptors passed during socket activation may now be
12622 named. A new API sd_listen_fds_with_names() is added to
12623 access the names. The default names may be overridden,
12624 either in the .socket file using the FileDescriptorName=
12625 parameter, or by passing FDNAME= when storing the file
12626 descriptors using sd_notify().
12628 * systemd-networkd gained support for:
12630 - Setting the IPv6 Router Advertisement settings via
12631 IPv6AcceptRouterAdvertisements= in .network files.
12633 - Configuring the HelloTimeSec=, MaxAgeSec= and
12634 ForwardDelaySec= bridge parameters in .netdev files.
12636 - Configuring PreferredSource= for static routes in
12639 * The "ask-password" framework used to query for LUKS harddisk
12640 passwords or SSL passwords during boot gained support for
12641 caching passwords in the kernel keyring, if it is
12642 available. This makes sure that the user only has to type in
12643 a passphrase once if there are multiple objects to unlock
12644 with the same one. Previously, such password caching was
12645 available only when Plymouth was used; this moves the
12646 caching logic into the systemd codebase itself. The
12647 "systemd-ask-password" utility gained a new --keyname=
12648 switch to control which kernel keyring key to use for
12649 caching a password in. This functionality is also useful for
12650 enabling display managers such as gdm to automatically
12651 unlock the user's GNOME keyring if its passphrase, the
12652 user's password and the harddisk password are the same, if
12653 gdm-autologin is used.
12655 * When downloading tar or raw images using "machinectl
12656 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
12657 file is now also downloaded, if it is available and stored
12658 next to the image file.
12660 * Units of type ".socket" gained a new boolean setting
12661 Writable= which is only useful in conjunction with
12662 ListenSpecial=. If true, enables opening the specified
12663 special file in O_RDWR mode rather than O_RDONLY mode.
12665 * systemd-rfkill has been reworked to become a singleton
12666 service that is activated through /dev/rfkill on each rfkill
12667 state change and saves the settings to disk. This way,
12668 systemd-rfkill is now compatible with devices that exist
12669 only intermittendly, and even restores state if the previous
12670 system shutdown was abrupt rather than clean.
12672 * The journal daemon gained support for vacuuming old journal
12673 files controlled by the number of files that shall remain,
12674 in addition to the already existing control by size and by
12675 date. This is useful as journal interleaving performance
12676 degrades with too many separate journal files, and allows
12677 putting an effective limit on them. The new setting defaults
12678 to 100, but this may be changed by setting SystemMaxFiles=
12679 and RuntimeMaxFiles= in journald.conf. Also, the
12680 "journalctl" tool gained the new --vacuum-files= switch to
12681 manually vacuum journal files to leave only the specified
12682 number of files in place.
12684 * udev will now create /dev/disk/by-path links for ATA devices
12685 on kernels where that is supported.
12687 * Galician, Serbian, Turkish and Korean translations were added.
12689 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
12690 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
12691 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
12692 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
12693 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
12694 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
12695 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
12696 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
12697 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
12698 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
12699 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
12700 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
12701 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
12702 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
12703 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
12704 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
12705 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
12706 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
12708 — Berlin, 2015-10-07
12712 * The DHCP implementation of systemd-networkd gained a set of
12715 - The DHCP server now supports emitting DNS and NTP
12716 information. It may be enabled and configured via
12717 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
12718 and NTP information is enabled, but no servers are
12719 configured, the corresponding uplink information (if there
12720 is any) is propagated.
12722 - Server and client now support transmission and reception
12723 of timezone information. It can be configured via the
12724 newly introduced network options UseTimezone=,
12725 EmitTimezone=, and Timezone=. Transmission of timezone
12726 information is enabled between host and containers by
12727 default now: the container will change its local timezone
12728 to what the host has set.
12730 - Lease timeouts can now be configured via
12731 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
12733 - The DHCP server improved on the stability of
12734 leases. Clients are more likely to get the same lease
12735 information back, even if the server loses state.
12737 - The DHCP server supports two new configuration options to
12738 control the lease address pool metrics, PoolOffset= and
12741 * The encapsulation limit of tunnels in systemd-networkd may
12742 now be configured via 'EncapsulationLimit='. It allows
12743 modifying the maximum additional levels of encapsulation
12744 that are permitted to be prepended to a packet.
12746 * systemd now supports the concept of user buses replacing
12747 session buses, if used with dbus-1.10 (and enabled via dbus
12748 --enable-user-session). It previously only supported this on
12749 kdbus-enabled systems, and this release expands this to
12750 'dbus-daemon' systems.
12752 * systemd-networkd now supports predictable interface names
12753 for virtio devices.
12755 * systemd now optionally supports the new Linux kernel
12756 "unified" control group hierarchy. If enabled via the kernel
12757 command-line option 'systemd.unified_cgroup_hierarchy=1',
12758 systemd will try to mount the unified cgroup hierarchy
12759 directly on /sys/fs/cgroup. If not enabled, or not
12760 available, systemd will fall back to the legacy cgroup
12761 hierarchy setup, as before. Host system and containers can
12762 mix and match legacy and unified hierarchies as they
12763 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
12764 environment variable to individually select the hierarchy to
12765 use for executed containers. By default, nspawn will use the
12766 unified hierarchy for the containers if the host uses the
12767 unified hierarchy, and the legacy hierarchy otherwise.
12768 Please note that at this point the unified hierarchy is an
12769 experimental kernel feature and is likely to change in one
12770 of the next kernel releases. Therefore, it should not be
12771 enabled by default in downstream distributions yet. The
12772 minimum required kernel version for the unified hierarchy to
12773 work is 4.2. Note that when the unified hierarchy is used
12774 for the first time delegated access to controllers is
12775 safe. Because of this systemd-nspawn containers will get
12776 access to controllers now, as will systemd user
12777 sessions. This means containers and user sessions may now
12778 manage their own resources, partitioning up what the system
12781 * A new special scope unit "init.scope" has been introduced
12782 that encapsulates PID 1 of the system. It may be used to
12783 determine resource usage and enforce resource limits on PID
12784 1 itself. PID 1 hence moved out of the root of the control
12787 * The cgtop tool gained support for filtering out kernel
12788 threads when counting tasks in a control group. Also, the
12789 count of processes is now recursively summed up by
12790 default. Two options -k and --recursive= have been added to
12791 revert to old behaviour. The tool has also been updated to
12792 work correctly in containers now.
12794 * systemd-nspawn's --bind= and --bind-ro= options have been
12795 extended to allow creation of non-recursive bind mounts.
12797 * libsystemd gained two new calls sd_pid_get_cgroup() and
12798 sd_peer_get_cgroup() which return the control group path of
12799 a process or peer of a connected AF_UNIX socket. This
12800 function call is particularly useful when implementing
12801 delegated subtrees support in the control group hierarchy.
12803 * The "sd-event" event loop API of libsystemd now supports
12804 correct dequeuing of real-time signals, without losing
12807 * When systemd requests a polkit decision when managing units it
12808 will now add additional fields to the request, including unit
12809 name and desired operation. This enables more powerful polkit
12810 policies, that make decisions depending on these parameters.
12812 * nspawn learnt support for .nspawn settings files, that may
12813 accompany the image files or directories of containers, and
12814 may contain additional settings for the container. This is
12815 an alternative to configuring container parameters via the
12816 nspawn command line.
12818 Contributions from: Cristian Rodríguez, Daniel Mack, David
12819 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
12820 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
12821 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
12822 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
12823 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
12824 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
12825 Andersen, Tom Gundersen, Torstein Husebø
12827 — Berlin, 2015-09-08
12831 * machinectl gained a new verb 'shell' which opens a fresh
12832 shell on the target container or the host. It is similar to
12833 the existing 'login' command of machinectl, but spawns the
12834 shell directly without prompting for username or
12835 password. The pseudo machine '.host' now refers to the local
12836 host and is used by default. Hence, 'machinectl shell' can
12837 be used as replacement for 'su -' which spawns a session as
12838 a fresh systemd unit in a way that is fully isolated from
12839 the originating session.
12841 * systemd-networkd learned to cope with private-zone DHCP
12842 options and allows other programs to query the values.
12844 * SELinux access control when enabling/disabling units is no
12845 longer enforced with this release. The previous implementation
12846 was incorrect, and a new corrected implementation is not yet
12847 available. As unit file operations are still protected via
12848 polkit and D-Bus policy this is not a security problem. Yet,
12849 distributions which care about optimal SELinux support should
12850 probably not stabilize on this release.
12852 * sd-bus gained support for matches of type "arg0has=", that
12853 test for membership of strings in string arrays sent in bus
12856 * systemd-resolved now dumps the contents of its DNS and LLMNR
12857 caches to the logs on reception of the SIGUSR1 signal. This
12858 is useful to debug DNS behaviour.
12860 * The coredumpctl tool gained a new --directory= option to
12861 operate on journal files in a specific directory.
12863 * "systemctl reboot" and related commands gained a new
12864 "--message=" option which may be used to set a free-text
12865 wall message when shutting down or rebooting the
12866 system. This message is also logged, which is useful for
12867 figuring out the reason for a reboot or shutdown a
12870 * The "systemd-resolve-host" tool's -i switch now takes
12871 network interface numbers as alternative to interface names.
12873 * A new unit file setting for services has been introduced:
12874 UtmpMode= allows configuration of how precisely systemd
12875 handles utmp and wtmp entries for the service if this is
12876 enabled. This allows writing services that appear similar to
12877 user sessions in the output of the "w", "who", "last" and
12880 * systemd-resolved will now locally synthesize DNS resource
12881 records for the "localhost" and "gateway" domains as well as
12882 the local hostname. This should ensure that clients querying
12883 RRs via resolved will get similar results as those going via
12884 NSS, if nss-myhostname is enabled.
12886 Contributions from: Alastair Hughes, Alex Crawford, Daniel
12887 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
12888 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
12889 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
12890 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
12891 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
12892 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
12893 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
12894 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
12895 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
12896 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
12897 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
12899 — Berlin, 2015-08-27
12903 * The systemd-efi-boot-generator functionality was merged into
12904 systemd-gpt-auto-generator.
12906 * systemd-networkd now supports Group Policy for vxlan
12907 devices. It can be enabled via the new boolean configuration
12908 option called 'GroupPolicyExtension='.
12910 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
12911 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
12912 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
12914 — Berlin, 2015-07-31
12918 * The python-systemd code has been removed from the systemd repository.
12919 A new repository has been created which accommodates the code from
12920 now on, and we kindly ask distributions to create a separate package
12921 for this: https://github.com/systemd/python-systemd
12923 * The systemd daemon will now reload its main configuration
12924 (/etc/systemd/system.conf) on daemon-reload.
12926 * sd-dhcp now exposes vendor specific extensions via
12927 sd_dhcp_lease_get_vendor_specific().
12929 * systemd-networkd gained a number of new configuration options.
12931 - A new boolean configuration option for TAP devices called
12932 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
12933 device, thus allowing to send and receive GSO packets.
12935 - A new tunnel configuration option called 'CopyDSCP='.
12936 If enabled, the DSCP field of ip6 tunnels is copied into the
12937 decapsulated packet.
12939 - A set of boolean bridge configuration options were added.
12940 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
12941 and 'UnicastFlood=' are now parsed by networkd and applied to the
12942 respective bridge link device via the respective IFLA_BRPORT_*
12945 - A new string configuration option to override the hostname sent
12946 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
12947 is true, networkd will use the configured hostname instead of the
12948 system hostname when sending DHCP requests.
12950 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
12951 networkd will configure the IPv6 flow-label of the tunnel device
12952 according to RFC2460.
12954 - The 'macvtap' virtual network devices are now supported, similar to
12955 the already supported 'macvlan' devices.
12957 * systemd-resolved now implements RFC5452 to improve resilience against
12958 cache poisoning. Additionally, source port randomization is enabled
12959 by default to further protect against DNS spoofing attacks.
12961 * nss-mymachines now supports translating UIDs and GIDs of running
12962 containers with user-namespaces enabled. If a container 'foo'
12963 translates a host uid 'UID' to the container uid 'TUID', then
12964 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
12965 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
12966 mapped as 'vg-foo-TGID'.
12968 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
12969 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
12970 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
12971 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
12972 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
12973 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
12974 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
12975 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
12976 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
12977 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
12979 — Berlin, 2015-07-29
12983 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
12984 There are no known issues with current sysfs, and udev does not need
12985 or should be used to work around such bugs.
12987 * udev does no longer enable USB HID power management. Several reports
12988 indicate, that some devices cannot handle that setting.
12990 * The udev accelerometer helper was removed. The functionality
12991 is now fully included in iio-sensor-proxy. But this means,
12992 older iio-sensor-proxy versions will no longer provide
12993 accelerometer/orientation data with this systemd version.
12994 Please upgrade iio-sensor-proxy to version 1.0.
12996 * networkd gained a new configuration option IPv6PrivacyExtensions=
12997 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
12998 for Stateless Address") on selected networks.
13000 * For the sake of fewer build-time dependencies and less code in the
13001 main repository, the python bindings are about to be removed in the
13002 next release. A new repository has been created which accommodates
13003 the code from now on, and we kindly ask distributions to create a
13004 separate package for this. The removal will take place in v223.
13006 https://github.com/systemd/python-systemd
13008 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
13009 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
13010 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
13011 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
13012 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
13013 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
13014 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
13015 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
13016 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
13017 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
13019 — Berlin, 2015-07-07
13023 * The sd-bus.h and sd-event.h APIs have now been declared
13024 stable and have been added to the official interface of
13025 libsystemd.so. sd-bus implements an alternative D-Bus client
13026 library, that is relatively easy to use, very efficient and
13027 supports both classic D-Bus as well as kdbus as transport
13028 backend. sd-event is a generic event loop abstraction that
13029 is built around Linux epoll, but adds features such as event
13030 prioritization or efficient timer handling. Both APIs are good
13031 choices for C programs looking for a bus and/or event loop
13032 implementation that is minimal and does not have to be
13033 portable to other kernels.
13035 * kdbus support is no longer compile-time optional. It is now
13036 always built-in. However, it can still be disabled at
13037 runtime using the kdbus=0 kernel command line setting, and
13038 that setting may be changed to default to off, by specifying
13039 --disable-kdbus at build-time. Note though that the kernel
13040 command line setting has no effect if the kdbus.ko kernel
13041 module is not installed, in which case kdbus is (obviously)
13042 also disabled. We encourage all downstream distributions to
13043 begin testing kdbus by adding it to the kernel images in the
13044 development distributions, and leaving kdbus support in
13047 * The minimal required util-linux version has been bumped to
13050 * Support for chkconfig (--enable-chkconfig) was removed in
13051 favor of calling an abstraction tool
13052 /lib/systemd/systemd-sysv-install. This needs to be
13053 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
13054 in README for details.
13056 * If there's a systemd unit and a SysV init script for the
13057 same service name, and the user executes "systemctl enable"
13058 for it (or a related call), then this will now enable both
13059 (or execute the related operation on both), not just the
13062 * The libudev API documentation has been converted from gtkdoc
13065 * gudev has been removed from the systemd tree, it is now an
13068 * The systemd-cgtop tool learnt a new --raw switch to generate
13069 "raw" (machine parsable) output.
13071 * networkd's IPForwarding= .network file setting learnt the
13072 new setting "kernel", which ensures that networkd does not
13073 change the IP forwarding sysctl from the default kernel
13076 * The systemd-logind bus API now exposes a new boolean
13077 property "Docked" that reports whether logind considers the
13078 system "docked", i.e. connected to a docking station or not.
13080 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
13081 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
13082 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
13083 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
13084 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
13085 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
13086 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
13087 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
13088 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
13089 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
13090 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
13091 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
13092 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
13093 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
13094 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
13095 Fink, Zbigniew Jędrzejewski-Szmek
13097 — Berlin, 2015-06-19
13101 * The gudev library has been extracted into a separate repository
13102 available at: https://git.gnome.org/browse/libgudev/
13103 It is now managed as part of the Gnome project. Distributions
13104 are recommended to pass --disable-gudev to systemd and use
13105 gudev from the Gnome project instead. gudev is still included
13106 in systemd, for now. It will be removed soon, though. Please
13107 also see the announcement-thread on systemd-devel:
13108 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
13110 * systemd now exposes a CPUUsageNSec= property for each
13111 service unit on the bus, that contains the overall consumed
13112 CPU time of a service (the sum of what each process of the
13113 service consumed). This value is only available if
13114 CPUAccounting= is turned on for a service, and is then shown
13115 in the "systemctl status" output.
13117 * Support for configuring alternative mappings of the old SysV
13118 runlevels to systemd targets has been removed. They are now
13119 hardcoded in a way that runlevels 2, 3, 4 all map to
13120 multi-user.target and 5 to graphical.target (which
13121 previously was already the default behaviour).
13123 * The auto-mounter logic gained support for mount point
13124 expiry, using a new TimeoutIdleSec= setting in .automount
13125 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
13127 * The EFI System Partition (ESP) as mounted to /boot by
13128 systemd-efi-boot-generator will now be unmounted
13129 automatically after 2 minutes of not being used. This should
13130 minimize the risk of ESP corruptions.
13132 * New /etc/fstab options x-systemd.requires= and
13133 x-systemd.requires-mounts-for= are now supported to express
13134 additional dependencies for mounts. This is useful for
13135 journaling file systems that support external journal
13136 devices or overlay file systems that require underlying file
13137 systems to be mounted.
13139 * systemd does not support direct live-upgrades (via systemctl
13140 daemon-reexec) from versions older than v44 anymore. As no
13141 distribution we are aware of shipped such old versions in a
13142 stable release this should not be problematic.
13144 * When systemd forks off a new per-connection service instance
13145 it will now set the $REMOTE_ADDR environment variable to the
13146 remote IP address, and $REMOTE_PORT environment variable to
13147 the remote IP port. This behaviour is similar to the
13148 corresponding environment variables defined by CGI.
13150 * systemd-networkd gained support for uplink failure
13151 detection. The BindCarrier= option allows binding interface
13152 configuration dynamically to the link sense of other
13153 interfaces. This is useful to achieve behaviour like in
13156 * systemd-networkd gained support for configuring the DHCP
13157 client identifier to use when requesting leases.
13159 * systemd-networkd now has a per-network UseNTP= option to
13160 configure whether NTP server information acquired via DHCP
13161 is passed on to services like systemd-timesyncd.
13163 * systemd-networkd gained support for vti6 tunnels.
13165 * Note that systemd-networkd manages the sysctl variable
13166 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
13167 it is configured for since v219. The variable controls IP
13168 forwarding, and is a per-interface alternative to the global
13169 /proc/sys/net/ipv[46]/ip_forward. This setting is
13170 configurable in the IPForward= option, which defaults to
13171 "no". This means if networkd is used for an interface it is
13172 no longer sufficient to set the global sysctl option to turn
13173 on IP forwarding! Instead, the .network file option
13174 IPForward= needs to be turned on! Note that the
13175 implementation of this behaviour was broken in v219 and has
13176 been fixed in v220.
13178 * Many bonding and vxlan options are now configurable in
13181 * systemd-nspawn gained a new --property= setting to set unit
13182 properties for the container scope. This is useful for
13183 setting resource parameters (e.g. "CPUShares=500") on
13184 containers started from the command line.
13186 * systemd-nspawn gained a new --private-users= switch to make
13187 use of user namespacing available on recent Linux kernels.
13189 * systemd-nspawn may now be called as part of a shell pipeline
13190 in which case the pipes used for stdin and stdout are passed
13191 directly to the process invoked in the container, without
13192 indirection via a pseudo tty.
13194 * systemd-nspawn gained a new switch to control the UNIX
13195 signal to use when killing the init process of the container
13196 when shutting down.
13198 * systemd-nspawn gained a new --overlay= switch for mounting
13199 overlay file systems into the container using the new kernel
13202 * When a container image is imported via systemd-importd and
13203 the host file system is not btrfs, a loopback block device
13204 file is created in /var/lib/machines.raw with a btrfs file
13205 system inside. It is then mounted to /var/lib/machines to
13206 enable btrfs features for container management. The loopback
13207 file and btrfs file system is grown as needed when container
13208 images are imported via systemd-importd.
13210 * systemd-machined/systemd-importd gained support for btrfs
13211 quota, to enforce container disk space limits on disk. This
13212 is exposed in "machinectl set-limit".
13214 * systemd-importd now can import containers from local .tar,
13215 .raw and .qcow2 images, and export them to .tar and .raw. It
13216 can also import dkr v2 images now from the network (on top
13219 * systemd-importd gained support for verifying downloaded
13220 images with gpg2 (previously only gpg1 was supported).
13222 * systemd-machined, systemd-logind, systemd: most bus calls are
13223 now accessible to unprivileged processes via polkit. Also,
13224 systemd-logind will now allow users to kill their own sessions
13225 without further privileges or authorization.
13227 * systemd-shutdownd has been removed. This service was
13228 previously responsible for implementing scheduled shutdowns
13229 as exposed in /usr/bin/shutdown's time parameter. This
13230 functionality has now been moved into systemd-logind and is
13231 accessible via a bus interface.
13233 * "systemctl reboot" gained a new switch --firmware-setup that
13234 can be used to reboot into the EFI firmware setup, if that
13235 is available. systemd-logind now exposes an API on the bus
13236 to trigger such reboots, in case graphical desktop UIs want
13237 to cover this functionality.
13239 * "systemctl enable", "systemctl disable" and "systemctl mask"
13240 now support a new "--now" switch. If specified the units
13241 that are enabled will also be started, and the ones
13242 disabled/masked also stopped.
13244 * The Gummiboot EFI boot loader tool has been merged into
13245 systemd, and renamed to "systemd-boot". The bootctl tool has been
13246 updated to support systemd-boot.
13248 * An EFI kernel stub has been added that may be used to create
13249 kernel EFI binaries that contain not only the actual kernel,
13250 but also an initrd, boot splash, command line and OS release
13251 information. This combined binary can then be signed as a
13252 single image, so that the firmware can verify it all in one
13253 step. systemd-boot has special support for EFI binaries created
13254 like this and can extract OS release information from them
13255 and show them in the boot menu. This functionality is useful
13256 to implement cryptographically verified boot schemes.
13258 * Optional support has been added to systemd-fsck to pass
13259 fsck's progress report to an AF_UNIX socket in the file
13262 * udev will no longer create device symlinks for all block devices by
13263 default. A deny list for excluding special block devices from this
13264 logic has been turned into an allow list that requires picking block
13265 devices explicitly that require device symlinks.
13267 * A new (currently still internal) API sd-device.h has been
13268 added to libsystemd. This modernized API is supposed to
13269 replace libudev eventually. In fact, already much of libudev
13270 is now just a wrapper around sd-device.h.
13272 * A new hwdb database for storing metadata about pointing
13273 stick devices has been added.
13275 * systemd-tmpfiles gained support for setting file attributes
13276 similar to the "chattr" tool with new 'h' and 'H' lines.
13278 * systemd-journald will no longer unconditionally set the
13279 btrfs NOCOW flag on new journal files. This is instead done
13280 with tmpfiles snippet using the new 'h' line type. This
13281 allows easy disabling of this logic, by masking the
13282 journal-nocow.conf tmpfiles file.
13284 * systemd-journald will now translate audit message types to
13285 human readable identifiers when writing them to the
13286 journal. This should improve readability of audit messages.
13288 * The LUKS logic gained support for the offset= and skip=
13289 options in /etc/crypttab, as previously implemented by
13292 * /usr/lib/os-release gained a new optional field VARIANT= for
13293 distributions that support multiple variants (such as a
13294 desktop edition, a server edition, …)
13296 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
13297 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
13298 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
13299 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
13300 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
13301 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
13302 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
13303 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
13304 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
13305 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
13306 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
13307 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
13308 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
13309 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
13310 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
13311 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
13312 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
13313 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
13314 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
13315 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
13316 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
13317 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
13318 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
13319 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
13320 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
13321 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
13322 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
13324 — Berlin, 2015-05-22
13328 * Introduce a new API "sd-hwdb.h" for querying the hardware
13329 metadata database. With this minimal interface one can query
13330 and enumerate the udev hwdb, decoupled from the old libudev
13331 library. libudev's interface for this is now only a wrapper
13332 around sd-hwdb. A new tool systemd-hwdb has been added to
13333 interface with and update the database.
13335 * When any of systemd's tools copies files (for example due to
13336 tmpfiles' C lines) a btrfs reflink will attempted first,
13337 before bytewise copying is done.
13339 * systemd-nspawn gained a new --ephemeral switch. When
13340 specified a btrfs snapshot is taken of the container's root
13341 directory, and immediately removed when the container
13342 terminates again. Thus, a container can be started whose
13343 changes never alter the container's root directory, and are
13344 lost on container termination. This switch can also be used
13345 for starting a container off the root file system of the
13346 host without affecting the host OS. This switch is only
13347 available on btrfs file systems.
13349 * systemd-nspawn gained a new --template= switch. It takes the
13350 path to a container tree to use as template for the tree
13351 specified via --directory=, should that directory be
13352 missing. This allows instantiating containers dynamically,
13353 on first run. This switch is only available on btrfs file
13356 * When a .mount unit refers to a mount point on which multiple
13357 mounts are stacked, and the .mount unit is stopped all of
13358 the stacked mount points will now be unmounted until no
13359 mount point remains.
13361 * systemd now has an explicit notion of supported and
13362 unsupported unit types. Jobs enqueued for unsupported unit
13363 types will now fail with an "unsupported" error code. More
13364 specifically .swap, .automount and .device units are not
13365 supported in containers, .busname units are not supported on
13366 non-kdbus systems. .swap and .automount are also not
13367 supported if their respective kernel compile time options
13370 * machinectl gained support for two new "copy-from" and
13371 "copy-to" commands for copying files from a running
13372 container to the host or vice versa.
13374 * machinectl gained support for a new "bind" command to bind
13375 mount host directories into local containers. This is
13376 currently only supported for nspawn containers.
13378 * networkd gained support for configuring bridge forwarding
13379 database entries (fdb) from .network files.
13381 * A new tiny daemon "systemd-importd" has been added that can
13382 download container images in tar, raw, qcow2 or dkr formats,
13383 and make them available locally in /var/lib/machines, so
13384 that they can run as nspawn containers. The daemon can GPG
13385 verify the downloads (not supported for dkr, since it has no
13386 provisions for verifying downloads). It will transparently
13387 decompress bz2, xz, gzip compressed downloads if necessary,
13388 and restore sparse files on disk. The daemon uses privilege
13389 separation to ensure the actual download logic runs with
13390 fewer privileges than the daemon itself. machinectl has
13391 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
13392 make the functionality of importd available to the
13393 user. With this in place the Fedora and Ubuntu "Cloud"
13394 images can be downloaded and booted as containers unmodified
13395 (the Fedora images lack the appropriate GPG signature files
13396 currently, so they cannot be verified, but this will change
13397 soon, hopefully). Note that downloading images is currently
13398 only fully supported on btrfs.
13400 * machinectl is now able to list container images found in
13401 /var/lib/machines, along with some metadata about sizes of
13402 disk and similar. If the directory is located on btrfs and
13403 quota is enabled, this includes quota display. A new command
13404 "image-status" has been added that shows additional
13405 information about images.
13407 * machinectl is now able to clone container images
13408 efficiently, if the underlying file system (btrfs) supports
13409 it, with the new "machinectl clone" command. It also
13410 gained commands for renaming and removing images, as well as
13411 marking them read-only or read-write (supported also on
13412 legacy file systems).
13414 * networkd gained support for collecting LLDP network
13415 announcements, from hardware that supports this. This is
13416 shown in networkctl output.
13418 * systemd-run gained support for a new -t (--pty) switch for
13419 invoking a binary on a pty whose input and output is
13420 connected to the invoking terminal. This allows executing
13421 processes as system services while interactively
13422 communicating with them via the terminal. Most interestingly
13423 this is supported across container boundaries. Invoking
13424 "systemd-run -t /bin/bash" is an alternative to running a
13425 full login session, the difference being that the former
13426 will not register a session, nor go through the PAM session
13429 * tmpfiles gained support for a new "v" line type for creating
13430 btrfs subvolumes. If the underlying file system is a legacy
13431 file system, this automatically degrades to creating a
13432 normal directory. Among others /var/lib/machines is now
13433 created like this at boot, should it be missing.
13435 * The directory /var/lib/containers/ has been deprecated and
13436 been replaced by /var/lib/machines. The term "machines" has
13437 been used in the systemd context as generic term for both
13438 VMs and containers, and hence appears more appropriate for
13439 this, as the directory can also contain raw images bootable
13442 * systemd-nspawn when invoked with -M but without --directory=
13443 or --image= is now capable of searching for the container
13444 root directory, subvolume or disk image automatically, in
13445 /var/lib/machines. systemd-nspawn@.service has been updated
13446 to make use of this, thus allowing it to be used for raw
13449 * A new machines.target unit has been introduced that is
13450 supposed to group all containers/VMs invoked as services on
13451 the system. systemd-nspawn@.service has been updated to
13452 integrate with that.
13454 * machinectl gained a new "start" command, for invoking a
13455 container as a service. "machinectl start foo" is mostly
13456 equivalent to "systemctl start systemd-nspawn@foo.service",
13457 but handles escaping in a nicer way.
13459 * systemd-nspawn will now mount most of the cgroupfs tree
13460 read-only into each container, with the exception of the
13461 container's own subtree in the name=systemd hierarchy.
13463 * journald now sets the special FS_NOCOW file flag for its
13464 journal files. This should improve performance on btrfs, by
13465 avoiding heavy fragmentation when journald's write-pattern
13466 is used on COW file systems. It degrades btrfs' data
13467 integrity guarantees for the files to the same levels as for
13468 ext3/ext4 however. This should be OK though as journald does
13469 its own data integrity checks and all its objects are
13470 checksummed on disk. Also, journald should handle btrfs disk
13471 full events a lot more gracefully now, by processing SIGBUS
13472 errors, and not relying on fallocate() anymore.
13474 * When journald detects that journal files it is writing to
13475 have been deleted it will immediately start new journal
13478 * systemd now provides a way to store file descriptors
13479 per-service in PID 1. This is useful for daemons to ensure
13480 that fds they require are not lost during a daemon
13481 restart. The fds are passed to the daemon on the next
13482 invocation in the same way socket activation fds are
13483 passed. This is now used by journald to ensure that the
13484 various sockets connected to all the system's stdout/stderr
13485 are not lost when journald is restarted. File descriptors
13486 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
13487 an extension to sd_notify(). Note that a limit is enforced
13488 on the number of fds a service can store in PID 1, and it
13489 defaults to 0, so that no fds may be stored, unless this is
13490 explicitly turned on.
13492 * The default TERM variable to use for units connected to a
13493 terminal, when no other value is explicitly is set is now
13494 vt220 rather than vt102. This should be fairly safe still,
13495 but allows PgUp/PgDn work.
13497 * The /etc/crypttab option header= as known from Debian is now
13500 * "loginctl user-status" and "loginctl session-status" will
13501 now show the last 10 lines of log messages of the
13502 user/session following the status output. Similar,
13503 "machinectl status" will show the last 10 log lines
13504 associated with a virtual machine or container
13505 service. (Note that this is usually not the log messages
13506 done in the VM/container itself, but simply what the
13507 container manager logs. For nspawn this includes all console
13510 * "loginctl session-status" without further argument will now
13511 show the status of the session of the caller. Similar,
13512 "lock-session", "unlock-session", "activate",
13513 "enable-linger", "disable-linger" may now be called without
13514 session/user parameter in which case they apply to the
13515 caller's session/user.
13517 * An X11 session scriptlet is now shipped that uploads
13518 $DISPLAY and $XAUTHORITY into the environment of the systemd
13519 --user daemon if a session begins. This should improve
13520 compatibility with X11 enabled applications run as systemd
13523 * Generators are now subject to masking via /etc and /run, the
13524 same way as unit files.
13526 * networkd .network files gained support for configuring
13527 per-link IPv4/IPv6 packet forwarding as well as IPv4
13528 masquerading. This is by default turned on for veth links to
13529 containers, as registered by systemd-nspawn. This means that
13530 nspawn containers run with --network-veth will now get
13531 automatic routed access to the host's networks without any
13532 further configuration or setup, as long as networkd runs on
13535 * systemd-nspawn gained the --port= (-p) switch to expose TCP
13536 or UDP posts of a container on the host. With this in place
13537 it is possible to run containers with private veth links
13538 (--network-veth), and have their functionality exposed on
13539 the host as if their services were running directly on the
13542 * systemd-nspawn's --network-veth switch now gained a short
13543 version "-n", since with the changes above it is now truly
13544 useful out-of-the-box. The systemd-nspawn@.service has been
13545 updated to make use of it too by default.
13547 * systemd-nspawn will now maintain a per-image R/W lock, to
13548 ensure that the same image is not started more than once
13549 writable. (It's OK to run an image multiple times
13550 simultaneously in read-only mode.)
13552 * systemd-nspawn's --image= option is now capable of
13553 dissecting and booting MBR and GPT disk images that contain
13554 only a single active Linux partition. Previously it
13555 supported only GPT disk images with proper GPT type
13556 IDs. This allows running cloud images from major
13557 distributions directly with systemd-nspawn, without
13560 * In addition to collecting mouse dpi data in the udev
13561 hardware database, there's now support for collecting angle
13562 information for mouse scroll wheels. The database is
13563 supposed to guarantee similar scrolling behavior on mice
13564 that it knows about. There's also support for collecting
13565 information about Touchpad types.
13567 * udev's input_id built-in will now also collect touch screen
13568 dimension data and attach it to probed devices.
13570 * /etc/os-release gained support for a Distribution Privacy
13573 * networkd gained support for creating "ipvlan", "gretap",
13574 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
13576 * systemd-tmpfiles gained support for "a" lines for setting
13579 * systemd-nspawn will now mount /tmp in the container to
13580 tmpfs, automatically.
13582 * systemd now exposes the memory.usage_in_bytes cgroup
13583 attribute and shows it for each service in the "systemctl
13584 status" output, if available.
13586 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
13587 immediate reboot is triggered. This useful if shutdown is
13588 hung and is unable to complete, to expedite the
13589 operation. Note that this kind of reboot will still unmount
13590 all file systems, and hence should not result in fsck being
13591 run on next reboot.
13593 * A .device unit for an optical block device will now be
13594 considered active only when a medium is in the drive. Also,
13595 mount units are now bound to their backing devices thus
13596 triggering automatic unmounting when devices become
13597 unavailable. With this in place systemd will now
13598 automatically unmount left-over mounts when a CD-ROM is
13599 ejected or a USB stick is yanked from the system.
13601 * networkd-wait-online now has support for waiting for
13602 specific interfaces only (with globbing), and for giving up
13603 after a configurable timeout.
13605 * networkd now exits when idle. It will be automatically
13606 restarted as soon as interfaces show up, are removed or
13607 change state. networkd will stay around as long as there is
13608 at least one DHCP state machine or similar around, that keep
13611 * networkd may now configure IPv6 link-local addressing in
13612 addition to IPv4 link-local addressing.
13614 * The IPv6 "token" for use in SLAAC may now be configured for
13615 each .network interface in networkd.
13617 * Routes configured with networkd may now be assigned a scope
13620 * networkd's [Match] sections now support globbing and lists
13621 of multiple space-separated matches per item.
13623 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
13624 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
13625 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
13626 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
13627 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
13628 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
13629 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
13630 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
13631 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
13632 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
13633 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
13634 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
13635 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
13636 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
13637 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
13638 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
13639 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
13640 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
13641 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
13642 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
13643 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
13644 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
13645 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
13646 Hoffmann, Zbigniew Jędrzejewski-Szmek
13648 — Berlin, 2015-02-16
13652 * When querying unit file enablement status (for example via
13653 "systemctl is-enabled"), a new state "indirect" is now known
13654 which indicates that a unit might not be enabled itself, but
13655 another unit listed in its Also= setting might be.
13657 * Similar to the various existing ConditionXYZ= settings for
13658 units, there are now matching AssertXYZ= settings. While
13659 failing conditions cause a unit to be skipped, but its job
13660 to succeed, failing assertions declared like this will cause
13661 a unit start operation and its job to fail.
13663 * hostnamed now knows a new chassis type "embedded".
13665 * systemctl gained a new "edit" command. When used on a unit
13666 file, this allows extending unit files with .d/ drop-in
13667 configuration snippets or editing the full file (after
13668 copying it from /usr/lib to /etc). This will invoke the
13669 user's editor (as configured with $EDITOR), and reload the
13670 modified configuration after editing.
13672 * "systemctl status" now shows the suggested enablement state
13673 for a unit, as declared in the (usually vendor-supplied)
13674 system preset files.
13676 * nss-myhostname will now resolve the single-label hostname
13677 "gateway" to the locally configured default IP routing
13678 gateways, ordered by their metrics. This assigns a stable
13679 name to the used gateways, regardless which ones are
13680 currently configured. Note that the name will only be
13681 resolved after all other name sources (if nss-myhostname is
13682 configured properly) and should hence not negatively impact
13683 systems that use the single-label hostname "gateway" in
13686 * systemd-inhibit now allows filtering by mode when listing
13689 * Scope and service units gained a new "Delegate" boolean
13690 property, which, when set, allows processes running inside the
13691 unit to further partition resources. This is primarily
13692 useful for systemd user instances as well as container
13695 * journald will now pick up audit messages directly from
13696 the kernel, and log them like any other log message. The
13697 audit fields are split up and fully indexed. This means that
13698 journalctl in many ways is now a (nicer!) alternative to
13699 ausearch, the traditional audit client. Note that this
13700 implements only a minimal audit client. If you want the
13701 special audit modes like reboot-on-log-overflow, please use
13702 the traditional auditd instead, which can be used in
13703 parallel to journald.
13705 * The ConditionSecurity= unit file option now understands the
13706 special string "audit" to check whether auditing is
13709 * journalctl gained two new commands --vacuum-size= and
13710 --vacuum-time= to delete old journal files until the
13711 remaining ones take up no more than the specified size on disk,
13712 or are not older than the specified time.
13714 * A new, native PPPoE library has been added to sd-network,
13715 systemd's library of light-weight networking protocols. This
13716 library will be used in a future version of networkd to
13717 enable PPPoE communication without an external pppd daemon.
13719 * The busctl tool now understands a new "capture" verb that
13720 works similar to "monitor", but writes a packet capture
13721 trace to STDOUT that can be redirected to a file which is
13722 compatible with libcap's capture file format. This can then
13723 be loaded in Wireshark and similar tools to inspect bus
13726 * The busctl tool now understands a new "tree" verb that shows
13727 the object trees of a specific service on the bus, or of all
13730 * The busctl tool now understands a new "introspect" verb that
13731 shows all interfaces and members of objects on the bus,
13732 including their signature and values. This is particularly
13733 useful to get more information about bus objects shown by
13734 the new "busctl tree" command.
13736 * The busctl tool now understands new verbs "call",
13737 "set-property" and "get-property" for invoking bus method
13738 calls, setting and getting bus object properties in a
13741 * busctl gained a new --augment-creds= argument that controls
13742 whether the tool shall augment credential information it
13743 gets from the bus with data from /proc, in a possibly
13746 * nspawn's --link-journal= switch gained two new values
13747 "try-guest" and "try-host" that work like "guest" and
13748 "host", but do not fail if the host has no persistent
13749 journaling enabled. -j is now equivalent to
13750 --link-journal=try-guest.
13752 * macvlan network devices created by nspawn will now have
13753 stable MAC addresses.
13755 * A new SmackProcessLabel= unit setting has been added, which
13756 controls the SMACK security label processes forked off by
13757 the respective unit shall use.
13759 * If compiled with --enable-xkbcommon, systemd-localed will
13760 verify x11 keymap settings by compiling the given keymap. It
13761 will spew out warnings if the compilation fails. This
13762 requires libxkbcommon to be installed.
13764 * When a coredump is collected, a larger number of metadata
13765 fields is now collected and included in the journal records
13766 created for it. More specifically, control group membership,
13767 environment variables, memory maps, working directory,
13768 chroot directory, /proc/$PID/status, and a list of open file
13769 descriptors is now stored in the log entry.
13771 * The udev hwdb now contains DPI information for mice. For
13774 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
13776 * All systemd programs that read standalone configuration
13777 files in /etc now also support a corresponding series of
13778 .conf.d configuration directories in /etc/, /run/,
13779 /usr/local/lib/, /usr/lib/, and (if configured with
13780 --enable-split-usr) /lib/. In particular, the following
13781 configuration files now have corresponding configuration
13782 directories: system.conf user.conf, logind.conf,
13783 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
13784 resolved.conf, timesyncd.conf, journal-remote.conf, and
13785 journal-upload.conf. Note that distributions should use the
13786 configuration directories in /usr/lib/; the directories in
13787 /etc/ are reserved for the system administrator.
13789 * systemd-rfkill will no longer take the rfkill device name
13790 into account when storing rfkill state on disk, as the name
13791 might be dynamically assigned and not stable. Instead, the
13792 ID_PATH udev variable combined with the rfkill type (wlan,
13793 bluetooth, …) is used.
13795 * A new service systemd-machine-id-commit.service has been
13796 added. When used on systems where /etc is read-only during
13797 boot, and /etc/machine-id is not initialized (but an empty
13798 file), this service will copy the temporary machine ID
13799 created as replacement into /etc after the system is fully
13800 booted up. This is useful for systems that are freshly
13801 installed with a non-initialized machine ID, but should get
13802 a fixed machine ID for subsequent boots.
13804 * networkd's .netdev files now provide a large set of
13805 configuration parameters for VXLAN devices. Similarly, the
13806 bridge port cost parameter is now configurable in .network
13807 files. There's also new support for configuring IP source
13808 routing. networkd .link files gained support for a new
13809 OriginalName= match that is useful to match against the
13810 original interface name the kernel assigned. .network files
13811 may include MTU= and MACAddress= fields for altering the MTU
13812 and MAC address while being connected to a specific network
13815 * The LUKS logic gained supported for configuring
13816 UUID-specific key files. There's also new support for naming
13817 LUKS device from the kernel command line, using the new
13818 luks.name= argument.
13820 * Timer units may now be transiently created via the bus API
13821 (this was previously already available for scope and service
13822 units). In addition it is now possible to create multiple
13823 transient units at the same time with a single bus call. The
13824 "systemd-run" tool has been updated to make use of this for
13825 running commands on a specified time, in at(1)-style.
13827 * tmpfiles gained support for "t" lines, for assigning
13828 extended attributes to files. Among other uses this may be
13829 used to assign SMACK labels to files.
13831 Contributions from: Alin Rauta, Alison Chaiken, Andrej
13832 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
13833 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
13834 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
13835 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
13836 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
13837 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
13838 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
13839 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
13840 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
13841 Michael Biebl, Michael Chapman, Michael Marineau, Michal
13842 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
13843 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
13844 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
13845 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
13846 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
13847 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
13848 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
13850 — Berlin, 2014-12-10
13854 * journalctl gained the new options -t/--identifier= to match
13855 on the syslog identifier (aka "tag"), as well as --utc to
13856 show log timestamps in the UTC timezone. journalctl now also
13857 accepts -n/--lines=all to disable line capping in a pager.
13859 * journalctl gained a new switch, --flush, that synchronously
13860 flushes logs from /run/log/journal to /var/log/journal if
13861 persistent storage is enabled. systemd-journal-flush.service
13862 now waits until the operation is complete.
13864 * Services can notify the manager before they start a reload
13865 (by sending RELOADING=1) or shutdown (by sending
13866 STOPPING=1). This allows the manager to track and show the
13867 internal state of daemons and closes a race condition when
13868 the process is still running but has closed its D-Bus
13871 * Services with Type=oneshot do not have to have any ExecStart
13874 * User units are now loaded also from
13875 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
13876 /run/systemd/user directory that was already previously
13877 supported, but is under the control of the user.
13879 * Job timeouts (i.e. timeouts on the time a job that is
13880 queued stays in the run queue) can now optionally result in
13881 immediate reboot or power-off actions (JobTimeoutAction= and
13882 JobTimeoutRebootArgument=). This is useful on ".target"
13883 units, to limit the maximum time a target remains
13884 undispatched in the run queue, and to trigger an emergency
13885 operation in such a case. This is now used by default to
13886 turn off the system if boot-up (as defined by everything in
13887 basic.target) hangs and does not complete for at least
13888 15min. Also, if power-off or reboot hang for at least 30min
13889 an immediate power-off/reboot operation is triggered. This
13890 functionality is particularly useful to increase reliability
13891 on embedded devices, but also on laptops which might
13892 accidentally get powered on when carried in a backpack and
13893 whose boot stays stuck in a hard disk encryption passphrase
13896 * systemd-logind can be configured to also handle lid switch
13897 events even when the machine is docked or multiple displays
13898 are attached (HandleLidSwitchDocked= option).
13900 * A helper binary and a service have been added which can be
13901 used to resume from hibernation in the initramfs. A
13902 generator will parse the resume= option on the kernel
13903 command line to trigger resume.
13905 * A user console daemon systemd-consoled has been
13906 added. Currently, it is a preview, and will so far open a
13907 single terminal on each session of the user marked as
13908 Desktop=systemd-console.
13910 * Route metrics can be specified for DHCP routes added by
13913 * The SELinux context of socket-activated services can be set
13914 from the information provided by the networking stack
13915 (SELinuxContextFromNet= option).
13917 * Userspace firmware loading support has been removed and
13918 the minimum supported kernel version is thus bumped to 3.7.
13920 * Timeout for udev workers has been increased from 1 to 3
13921 minutes, but a warning will be printed after 1 minute to
13922 help diagnose kernel modules that take a long time to load.
13924 * Udev rules can now remove tags on devices with TAG-="foobar".
13926 * systemd's readahead implementation has been removed. In many
13927 circumstances it didn't give expected benefits even for
13928 rotational disk drives and was becoming less relevant in the
13929 age of SSDs. As none of the developers has been using
13930 rotating media anymore, and nobody stepped up to actively
13931 maintain this component of systemd it has now been removed.
13933 * Swap units can use Options= to specify discard options.
13934 Discard options specified for swaps in /etc/fstab are now
13937 * Docker containers are now detected as a separate type of
13940 * The Password Agent protocol gained support for queries where
13941 the user input is shown, useful e.g. for user names.
13942 systemd-ask-password gained a new --echo option to turn that
13945 * The default sysctl.d/ snippets will now set:
13947 net.core.default_qdisc = fq_codel
13949 This selects Fair Queuing Controlled Delay as the default
13950 queuing discipline for network interfaces. fq_codel helps
13951 fight the network bufferbloat problem. It is believed to be
13952 a good default with no tuning required for most workloads.
13953 Downstream distributions may override this choice. On 10Gbit
13954 servers that do not do forwarding, "fq" may perform better.
13955 Systems without a good clocksource should use "pfifo_fast".
13957 * If kdbus is enabled during build a new option BusPolicy= is
13958 available for service units, that allows locking all service
13959 processes into a stricter bus policy, in order to limit
13960 access to various bus services, or even hide most of them
13961 from the service's view entirely.
13963 * networkctl will now show the .network and .link file
13964 networkd has applied to a specific interface.
13966 * sd-login gained a new API call sd_session_get_desktop() to
13967 query which desktop environment has been selected for a
13970 * UNIX utmp support is now compile-time optional to support
13971 legacy-free systems.
13973 * systemctl gained two new commands "add-wants" and
13974 "add-requires" for pulling in units from specific targets
13977 * If the word "rescue" is specified on the kernel command line
13978 the system will now boot into rescue mode (aka
13979 rescue.target), which was previously available only by
13980 specifying "1" or "systemd.unit=rescue.target" on the kernel
13981 command line. This new kernel command line option nicely
13982 mirrors the already existing "emergency" kernel command line
13985 * New kernel command line options mount.usr=, mount.usrflags=,
13986 mount.usrfstype= have been added that match root=, rootflags=,
13987 rootfstype= but allow mounting a specific file system to
13990 * The $NOTIFY_SOCKET is now also passed to control processes of
13991 services, not only the main process.
13993 * This version reenables support for fsck's -l switch. This
13994 means at least version v2.25 of util-linux is required for
13995 operation, otherwise dead-locks on device nodes may
13996 occur. Again: you need to update util-linux to at least
13997 v2.25 when updating systemd to v217.
13999 * The "multi-seat-x" tool has been removed from systemd, as
14000 its functionality has been integrated into X servers 1.16,
14001 and the tool is hence redundant. It is recommended to update
14002 display managers invoking this tool to simply invoke X
14003 directly from now on, again.
14005 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
14006 message flag has been added for all of systemd's polkit
14007 authenticated method calls has been added. In particular this
14008 now allows optional interactive authorization via polkit for
14009 many of PID1's privileged operations such as unit file
14010 enabling and disabling.
14012 * "udevadm hwdb --update" learnt a new switch "--usr" for
14013 placing the rebuilt hardware database in /usr instead of
14014 /etc. When used only hardware database entries stored in
14015 /usr will be used, and any user database entries in /etc are
14016 ignored. This functionality is useful for vendors to ship a
14017 pre-built database on systems where local configuration is
14018 unnecessary or unlikely.
14020 * Calendar time specifications in .timer units now also
14021 understand the strings "semi-annually", "quarterly" and
14022 "minutely" as shortcuts (in addition to the preexisting
14023 "annually", "hourly", …).
14025 * systemd-tmpfiles will now correctly create files in /dev
14026 at boot which are marked for creation only at boot. It is
14027 recommended to always create static device nodes with 'c!'
14028 and 'b!', so that they are created only at boot and not
14029 overwritten at runtime.
14031 * When the watchdog logic is used for a service (WatchdogSec=)
14032 and the watchdog timeout is hit the service will now be
14033 terminated with SIGABRT (instead of just SIGTERM), in order
14034 to make sure a proper coredump and backtrace is
14035 generated. This ensures that hanging services will result in
14036 similar coredump/backtrace behaviour as services that hit a
14037 segmentation fault.
14039 Contributions from: Andreas Henriksson, Andrei Borzenkov,
14040 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
14041 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
14042 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
14043 Herrmann, David Sommerseth, David Strauss, Emil Renner
14044 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
14045 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
14046 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
14047 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
14048 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
14049 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
14050 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
14051 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
14052 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
14053 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
14054 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
14055 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
14056 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
14057 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
14058 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
14059 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
14062 — Berlin, 2014-10-28
14066 * timedated no longer reads NTP implementation unit names from
14067 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
14068 implementations should add a
14070 Conflicts=systemd-timesyncd.service
14072 to their unit files to take over and replace systemd's NTP
14073 default functionality.
14075 * systemd-sysusers gained a new line type "r" for configuring
14076 which UID/GID ranges to allocate system users/groups
14077 from. Lines of type "u" may now add an additional column
14078 that specifies the home directory for the system user to be
14079 created. Also, systemd-sysusers may now optionally read user
14080 information from STDIN instead of a file. This is useful for
14081 invoking it from RPM preinst scriptlets that need to create
14082 users before the first RPM file is installed since these
14083 files might need to be owned by them. A new
14084 %sysusers_create_inline RPM macro has been introduced to do
14085 just that. systemd-sysusers now updates the shadow files as
14086 well as the user/group databases, which should enhance
14087 compatibility with certain tools like grpck.
14089 * A number of bus APIs of PID 1 now optionally consult polkit to
14090 permit access for otherwise unprivileged clients under certain
14091 conditions. Note that this currently doesn't support
14092 interactive authentication yet, but this is expected to be
14093 added eventually, too.
14095 * /etc/machine-info now has new fields for configuring the
14096 deployment environment of the machine, as well as the
14097 location of the machine. hostnamectl has been updated with
14098 new command to update these fields.
14100 * systemd-timesyncd has been updated to automatically acquire
14101 NTP server information from systemd-networkd, which might
14102 have been discovered via DHCP.
14104 * systemd-resolved now includes a caching DNS stub resolver
14105 and a complete LLMNR name resolution implementation. A new
14106 NSS module "nss-resolve" has been added which can be used
14107 instead of glibc's own "nss-dns" to resolve hostnames via
14108 systemd-resolved. Hostnames, addresses and arbitrary RRs may
14109 be resolved via systemd-resolved D-Bus APIs. In contrast to
14110 the glibc internal resolver systemd-resolved is aware of
14111 multi-homed system, and keeps DNS server and caches separate
14112 and per-interface. Queries are sent simultaneously on all
14113 interfaces that have DNS servers configured, in order to
14114 properly handle VPNs and local LANs which might resolve
14115 separate sets of domain names. systemd-resolved may acquire
14116 DNS server information from systemd-networkd automatically,
14117 which in turn might have discovered them via DHCP. A tool
14118 "systemd-resolve-host" has been added that may be used to
14119 query the DNS logic in resolved. systemd-resolved implements
14120 IDNA and automatically uses IDNA or UTF-8 encoding depending
14121 on whether classic DNS or LLMNR is used as transport. In the
14122 next releases we intend to add a DNSSEC and mDNS/DNS-SD
14123 implementation to systemd-resolved.
14125 * A new NSS module nss-mymachines has been added, that
14126 automatically resolves the names of all local registered
14127 containers to their respective IP addresses.
14129 * A new client tool "networkctl" for systemd-networkd has been
14130 added. It currently is entirely passive and will query
14131 networking configuration from udev, rtnetlink and networkd,
14132 and present it to the user in a very friendly
14133 way. Eventually, we hope to extend it to become a full
14134 control utility for networkd.
14136 * .socket units gained a new DeferAcceptSec= setting that
14137 controls the kernels' TCP_DEFER_ACCEPT sockopt for
14138 TCP. Similarly, support for controlling TCP keep-alive
14139 settings has been added (KeepAliveTimeSec=,
14140 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
14141 turning off Nagle's algorithm on TCP has been added
14144 * logind learned a new session type "web", for use in projects
14145 like Cockpit which register web clients as PAM sessions.
14147 * timer units with at least one OnCalendar= setting will now
14148 be started only after time-sync.target has been
14149 reached. This way they will not elapse before the system
14150 clock has been corrected by a local NTP client or
14151 similar. This is particular useful on RTC-less embedded
14152 machines, that come up with an invalid system clock.
14154 * systemd-nspawn's --network-veth= switch should now result in
14155 stable MAC addresses for both the outer and the inner side
14158 * systemd-nspawn gained a new --volatile= switch for running
14159 container instances with /etc or /var unpopulated.
14161 * The kdbus client code has been updated to use the new Linux
14162 3.17 memfd subsystem instead of the old kdbus-specific one.
14164 * systemd-networkd's DHCP client and server now support
14165 FORCERENEW. There are also new configuration options to
14166 configure the vendor client identifier and broadcast mode
14169 * systemd will no longer inform the kernel about the current
14170 timezone, as this is necessarily incorrect and racy as the
14171 kernel has no understanding of DST and similar
14172 concepts. This hence means FAT timestamps will be always
14173 considered UTC, similar to what Android is already
14174 doing. Also, when the RTC is configured to the local time
14175 (rather than UTC) systemd will never synchronize back to it,
14176 as this might confuse Windows at a later boot.
14178 * systemd-analyze gained a new command "verify" for offline
14179 validation of unit files.
14181 * systemd-networkd gained support for a couple of additional
14182 settings for bonding networking setups. Also, the metric for
14183 statically configured routes may now be configured. For
14184 network interfaces where this is appropriate the peer IP
14185 address may now be configured.
14187 * systemd-networkd's DHCP client will no longer request
14188 broadcasting by default, as this tripped up some networks.
14189 For hardware where broadcast is required the feature should
14190 be switched back on using RequestBroadcast=yes.
14192 * systemd-networkd will now set up IPv4LL addresses (when
14193 enabled) even if DHCP is configured successfully.
14195 * udev will now default to respect network device names given
14196 by the kernel when the kernel indicates that these are
14197 predictable. This behavior can be tweaked by changing
14198 NamePolicy= in the relevant .link file.
14200 * A new library systemd-terminal has been added that
14201 implements full TTY stream parsing and rendering. This
14202 library is supposed to be used later on for implementing a
14203 full userspace VT subsystem, replacing the current kernel
14206 * A new tool systemd-journal-upload has been added to push
14207 journal data to a remote system running
14208 systemd-journal-remote.
14210 * journald will no longer forward all local data to another
14211 running syslog daemon. This change has been made because
14212 rsyslog (which appears to be the most commonly used syslog
14213 implementation these days) no longer makes use of this, and
14214 instead pulls the data out of the journal on its own. Since
14215 forwarding the messages to a non-existent syslog server is
14216 more expensive than we assumed we have now turned this
14217 off. If you run a syslog server that is not a recent rsyslog
14218 version, you have to turn this option on again
14219 (ForwardToSyslog= in journald.conf).
14221 * journald now optionally supports the LZ4 compressor for
14222 larger journal fields. This compressor should perform much
14223 better than XZ which was the previous default.
14225 * machinectl now shows the IP addresses of local containers,
14226 if it knows them, plus the interface name of the container.
14228 * A new tool "systemd-escape" has been added that makes it
14229 easy to escape strings to build unit names and similar.
14231 * sd_notify() messages may now include a new ERRNO= field
14232 which is parsed and collected by systemd and shown among the
14233 "systemctl status" output for a service.
14235 * A new component "systemd-firstboot" has been added that
14236 queries the most basic systemd information (timezone,
14237 hostname, root password) interactively on first
14238 boot. Alternatively it may also be used to provision these
14239 things offline on OS images installed into directories.
14241 * The default sysctl.d/ snippets will now set
14243 net.ipv4.conf.default.promote_secondaries=1
14245 This has the benefit of no flushing secondary IP addresses
14246 when primary addresses are removed.
14248 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
14249 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
14250 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
14251 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
14252 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
14253 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
14254 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
14255 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
14256 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
14257 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
14258 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
14259 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
14260 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
14261 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
14262 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
14264 — Berlin, 2014-08-19
14268 * A new tool systemd-sysusers has been added. This tool
14269 creates system users and groups in /etc/passwd and
14270 /etc/group, based on static declarative system user/group
14271 definitions in /usr/lib/sysusers.d/. This is useful to
14272 enable factory resets and volatile systems that boot up with
14273 an empty /etc directory, and thus need system users and
14274 groups created during early boot. systemd now also ships
14275 with two default sysusers.d/ files for the most basic
14276 users and groups systemd and the core operating system
14279 * A new tmpfiles snippet has been added that rebuilds the
14280 essential files in /etc on boot, should they be missing.
14282 * A directive for ensuring automatic clean-up of
14283 /var/cache/man/ has been removed from the default
14284 configuration. This line should now be shipped by the man
14285 implementation. The necessary change has been made to the
14286 man-db implementation. Note that you need to update your man
14287 implementation to one that ships this line, otherwise no
14288 automatic clean-up of /var/cache/man will take place.
14290 * A new condition ConditionNeedsUpdate= has been added that
14291 may conditionalize services to only run when /etc or /var
14292 are "older" than the vendor operating system resources in
14293 /usr. This is useful for reconstructing or updating /etc
14294 after an offline update of /usr or a factory reset, on the
14295 next reboot. Services that want to run once after such an
14296 update or reset should use this condition and order
14297 themselves before the new systemd-update-done.service, which
14298 will mark the two directories as fully updated. A number of
14299 service files have been added making use of this, to rebuild
14300 the udev hardware database, the journald message catalog and
14301 dynamic loader cache (ldconfig). The systemd-sysusers tool
14302 described above also makes use of this now. With this in
14303 place it is now possible to start up a minimal operating
14304 system with /etc empty cleanly. For more information on the
14305 concepts involved see this recent blog story:
14307 https://0pointer.de/blog/projects/stateless.html
14309 * A new system group "input" has been introduced, and all
14310 input device nodes get this group assigned. This is useful
14311 for system-level software to get access to input devices. It
14312 complements what is already done for "audio" and "video".
14314 * systemd-networkd learnt minimal DHCPv4 server support in
14315 addition to the existing DHCPv4 client support. It also
14316 learnt DHCPv6 client and IPv6 Router Solicitation client
14317 support. The DHCPv4 client gained support for static routes
14318 passed in from the server. Note that the [DHCPv4] section
14319 known in older systemd-networkd versions has been renamed to
14320 [DHCP] and is now also used by the DHCPv6 client. Existing
14321 .network files using settings of this section should be
14322 updated, though compatibility is maintained. Optionally, the
14323 client hostname may now be sent to the DHCP server.
14325 * networkd gained support for vxlan virtual networks as well
14326 as tun/tap and dummy devices.
14328 * networkd gained support for automatic allocation of address
14329 ranges for interfaces from a system-wide pool of
14330 addresses. This is useful for dynamically managing a large
14331 number of interfaces with a single network configuration
14332 file. In particular this is useful to easily assign
14333 appropriate IP addresses to the veth links of a large number
14334 of nspawn instances.
14336 * RPM macros for processing sysusers, sysctl and binfmt
14337 drop-in snippets at package installation time have been
14340 * The /etc/os-release file should now be placed in
14341 /usr/lib/os-release. The old location is automatically
14342 created as symlink. /usr/lib is the more appropriate
14343 location of this file, since it shall actually describe the
14344 vendor operating system shipped in /usr, and not the
14345 configuration stored in /etc.
14347 * .mount units gained a new boolean SloppyOptions= setting
14348 that maps to mount(8)'s -s option which enables permissive
14349 parsing of unknown mount options.
14351 * tmpfiles learnt a new "L+" directive which creates a symlink
14352 but (unlike "L") deletes a pre-existing file first, should
14353 it already exist and not already be the correct
14354 symlink. Similarly, "b+", "c+" and "p+" directives have been
14355 added as well, which create block and character devices, as
14356 well as fifos in the filesystem, possibly removing any
14357 pre-existing files of different types.
14359 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
14360 'argument' field (which so far specified the source to
14361 symlink/copy the files from) is now optional. If omitted the
14362 same file os copied from /usr/share/factory/ suffixed by the
14363 full destination path. This is useful for populating /etc
14364 with essential files, by copying them from vendor defaults
14365 shipped in /usr/share/factory/etc.
14367 * A new command "systemctl preset-all" has been added that
14368 applies the service preset settings to all installed unit
14369 files. A new switch --preset-mode= has been added that
14370 controls whether only enable or only disable operations
14373 * A new command "systemctl is-system-running" has been added
14374 that allows checking the overall state of the system, for
14375 example whether it is fully up and running.
14377 * When the system boots up with an empty /etc, the equivalent
14378 to "systemctl preset-all" is executed during early boot, to
14379 make sure all default services are enabled after a factory
14382 * systemd now contains a minimal preset file that enables the
14383 most basic services systemd ships by default.
14385 * Unit files' [Install] section gained a new DefaultInstance=
14386 field for defining the default instance to create if a
14387 template unit is enabled with no instance specified.
14389 * A new passive target cryptsetup-pre.target has been added
14390 that may be used by services that need to make they run and
14391 finish before the first LUKS cryptographic device is set up.
14393 * The /dev/loop-control and /dev/btrfs-control device nodes
14394 are now owned by the "disk" group by default, opening up
14395 access to this group.
14397 * systemd-coredump will now automatically generate a
14398 stack trace of all core dumps taking place on the system,
14399 based on elfutils' libdw library. This stack trace is logged
14402 * systemd-coredump may now optionally store coredumps directly
14403 on disk (in /var/lib/systemd/coredump, possibly compressed),
14404 instead of storing them unconditionally in the journal. This
14405 mode is the new default. A new configuration file
14406 /etc/systemd/coredump.conf has been added to configure this
14407 and other parameters of systemd-coredump.
14409 * coredumpctl gained a new "info" verb to show details about a
14410 specific coredump. A new switch "-1" has also been added
14411 that makes sure to only show information about the most
14412 recent entry instead of all entries. Also, as the tool is
14413 generally useful now the "systemd-" prefix of the binary
14414 name has been removed. Distributions that want to maintain
14415 compatibility with the old name should add a symlink from
14416 the old name to the new name.
14418 * journald's SplitMode= now defaults to "uid". This makes sure
14419 that unprivileged users can access their own coredumps with
14420 coredumpctl without restrictions.
14422 * New kernel command line options "systemd.wants=" (for
14423 pulling an additional unit during boot), "systemd.mask="
14424 (for masking a specific unit for the boot), and
14425 "systemd.debug-shell" (for enabling the debug shell on tty9)
14426 have been added. This is implemented in the new generator
14427 "systemd-debug-generator".
14429 * systemd-nspawn will now by default filter a couple of
14430 syscalls for containers, among them those required for
14431 kernel module loading, direct x86 IO port access, swap
14432 management, and kexec. Most importantly though
14433 open_by_handle_at() is now prohibited for containers,
14434 closing a hole similar to a recently discussed vulnerability
14435 in docker regarding access to files on file hierarchies the
14436 container should normally not have access to. Note that, for
14437 nspawn, we generally make no security claims anyway (and
14438 this is explicitly documented in the man page), so this is
14439 just a fix for one of the most obvious problems.
14441 * A new man page file-hierarchy(7) has been added that
14442 contains a minimized, modernized version of the file system
14443 layout systemd expects, similar in style to the FHS
14444 specification or hier(5). A new tool systemd-path(1) has
14445 been added to query many of these paths for the local
14448 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
14449 longer done. Since the directory now has a per-user size
14450 limit, and is cleaned on logout this appears unnecessary,
14451 in particular since this now brings the lifecycle of this
14452 directory closer in line with how IPC objects are handled.
14454 * systemd.pc now exports a number of additional directories,
14455 including $libdir (which is useful to identify the library
14456 path for the primary architecture of the system), and a
14457 couple of drop-in directories.
14459 * udev's predictable network interface names now use the dev_port
14460 sysfs attribute, introduced in linux 3.15 instead of dev_id to
14461 distinguish between ports of the same PCI function. dev_id should
14462 only be used for ports using the same HW address, hence the need
14465 * machined has been updated to export the OS version of a
14466 container (read from /etc/os-release and
14467 /usr/lib/os-release) on the bus. This is now shown in
14468 "machinectl status" for a machine.
14470 * A new service setting RestartForceExitStatus= has been
14471 added. If configured to a set of exit signals or process
14472 return values, the service will be restarted when the main
14473 daemon process exits with any of them, regardless of the
14476 * systemctl's -H switch for connecting to remote systemd
14477 machines has been extended so that it may be used to
14478 directly connect to a specific container on the
14479 host. "systemctl -H root@foobar:waldi" will now connect as
14480 user "root" to host "foobar", and then proceed directly to
14481 the container named "waldi". Note that currently you have to
14482 authenticate as user "root" for this to work, as entering
14483 containers is a privileged operation.
14485 Contributions from: Andreas Henriksson, Benjamin Steinwender,
14486 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
14487 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
14488 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
14489 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
14490 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
14491 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
14492 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
14493 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
14494 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
14495 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
14496 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
14498 — Berlin, 2014-07-03
14502 * As an experimental feature, udev now tries to lock the
14503 disk device node (flock(LOCK_SH|LOCK_NB)) while it
14504 executes events for the disk or any of its partitions.
14505 Applications like partitioning programs can lock the
14506 disk device node (flock(LOCK_EX)) and claim temporary
14507 device ownership that way; udev will entirely skip all event
14508 handling for this disk and its partitions. If the disk
14509 was opened for writing, the close will trigger a partition
14510 table rescan in udev's "watch" facility, and if needed
14511 synthesize "change" events for the disk and all its partitions.
14512 This is now unconditionally enabled, and if it turns out to
14513 cause major problems, we might turn it on only for specific
14514 devices, or might need to disable it entirely. Device Mapper
14515 devices are excluded from this logic.
14517 * We temporarily dropped the "-l" switch for fsck invocations,
14518 since they collide with the flock() logic above. util-linux
14519 upstream has been changed already to avoid this conflict,
14520 and we will re-add "-l" as soon as util-linux with this
14521 change has been released.
14523 * The dependency on libattr has been removed. Since a long
14524 time, the extended attribute calls have moved to glibc, and
14525 libattr is thus unnecessary.
14527 * Virtualization detection works without privileges now. This
14528 means the systemd-detect-virt binary no longer requires
14529 CAP_SYS_PTRACE file capabilities, and our daemons can run
14530 with fewer privileges.
14532 * systemd-networkd now runs under its own "systemd-network"
14533 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
14534 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
14535 loses the ability to write to files owned by root this way.
14537 * Similarly, systemd-resolved now runs under its own
14538 "systemd-resolve" user with no capabilities remaining.
14540 * Similarly, systemd-bus-proxyd now runs under its own
14541 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
14543 * systemd-networkd gained support for setting up "veth"
14544 virtual Ethernet devices for container connectivity, as well
14545 as GRE and VTI tunnels.
14547 * systemd-networkd will no longer automatically attempt to
14548 manually load kernel modules necessary for certain tunnel
14549 transports. Instead, it is assumed the kernel loads them
14550 automatically when required. This only works correctly on
14551 very new kernels. On older kernels, please consider adding
14552 the kernel modules to /etc/modules-load.d/ as a work-around.
14554 * The resolv.conf file systemd-resolved generates has been
14555 moved to /run/systemd/resolve/. If you have a symlink from
14556 /etc/resolv.conf, it might be necessary to correct it.
14558 * Two new service settings, ProtectHome= and ProtectSystem=,
14559 have been added. When enabled, they will make the user data
14560 (such as /home) inaccessible or read-only and the system
14561 (such as /usr) read-only, for specific services. This allows
14562 very light-weight per-service sandboxing to avoid
14563 modifications of user data or system files from
14564 services. These two new switches have been enabled for all
14565 of systemd's long-running services, where appropriate.
14567 * Socket units gained new SocketUser= and SocketGroup=
14568 settings to set the owner user and group of AF_UNIX sockets
14569 and FIFOs in the file system.
14571 * Socket units gained a new RemoveOnStop= setting. If enabled,
14572 all FIFOS and sockets in the file system will be removed
14573 when the specific socket unit is stopped.
14575 * Socket units gained a new Symlinks= setting. It takes a list
14576 of symlinks to create to file system sockets or FIFOs
14577 created by the specific Unix sockets. This is useful to
14578 manage symlinks to socket nodes with the same lifecycle as
14581 * The /dev/log socket and /dev/initctl FIFO have been moved to
14582 /run, and have been replaced by symlinks. This allows
14583 connecting to these facilities even if PrivateDevices=yes is
14584 used for a service (which makes /dev/log itself unavailable,
14585 but /run is left). This also has the benefit of ensuring
14586 that /dev only contains device nodes, directories and
14587 symlinks, and nothing else.
14589 * sd-daemon gained two new calls sd_pid_notify() and
14590 sd_pid_notifyf(). They are similar to sd_notify() and
14591 sd_notifyf(), but allow overriding of the source PID of
14592 notification messages if permissions permit this. This is
14593 useful to send notify messages on behalf of a different
14594 process (for example, the parent process). The
14595 systemd-notify tool has been updated to make use of this
14596 when sending messages (so that notification messages now
14597 originate from the shell script invoking systemd-notify and
14598 not the systemd-notify process itself. This should minimize
14599 a race where systemd fails to associate notification
14600 messages to services when the originating process already
14603 * A new "on-abnormal" setting for Restart= has been added. If
14604 set, it will result in automatic restarts on all "abnormal"
14605 reasons for a process to exit, which includes unclean
14606 signals, core dumps, timeouts and watchdog timeouts, but
14607 does not include clean and unclean exit codes or clean
14608 signals. Restart=on-abnormal is an alternative for
14609 Restart=on-failure for services that shall be able to
14610 terminate and avoid restarts on certain errors, by
14611 indicating so with an unclean exit code. Restart=on-failure
14612 or Restart=on-abnormal is now the recommended setting for
14613 all long-running services.
14615 * If the InaccessibleDirectories= service setting points to a
14616 mount point (or if there are any submounts contained within
14617 it), it is now attempted to completely unmount it, to make
14618 the file systems truly unavailable for the respective
14621 * The ReadOnlyDirectories= service setting and
14622 systemd-nspawn's --read-only parameter are now recursively
14623 applied to all submounts, too.
14625 * Mount units may now be created transiently via the bus APIs.
14627 * The support for SysV and LSB init scripts has been removed
14628 from the systemd daemon itself. Instead, it is now
14629 implemented as a generator that creates native systemd units
14630 from these scripts when needed. This enables us to remove a
14631 substantial amount of legacy code from PID 1, following the
14632 fact that many distributions only ship a very small number
14633 of LSB/SysV init scripts nowadays.
14635 * Privileged Xen (dom0) domains are not considered
14636 virtualization anymore by the virtualization detection
14637 logic. After all, they generally have unrestricted access to
14638 the hardware and usually are used to manage the unprivileged
14641 * systemd-tmpfiles gained a new "C" line type, for copying
14642 files or entire directories.
14644 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
14645 lines. So far, they have been non-globbing versions of the
14646 latter, and have thus been redundant. In future, it is
14647 recommended to only use "z". "m" has hence been removed
14648 from the documentation, even though it stays supported.
14650 * A tmpfiles snippet to recreate the most basic structure in
14651 /var has been added. This is enough to create the /var/run →
14652 /run symlink and create a couple of structural
14653 directories. This allows systems to boot up with an empty or
14654 volatile /var. Of course, while with this change, the core OS
14655 now is capable with dealing with a volatile /var, not all
14656 user services are ready for it. However, we hope that sooner
14657 or later, many service daemons will be changed upstream so
14658 that they are able to automatically create their necessary
14659 directories in /var at boot, should they be missing. This is
14660 the first step to allow state-less systems that only require
14661 the vendor image for /usr to boot.
14663 * systemd-nspawn has gained a new --tmpfs= switch to mount an
14664 empty tmpfs instance to a specific directory. This is
14665 particularly useful for making use of the automatic
14666 reconstruction of /var (see above), by passing --tmpfs=/var.
14668 * Access modes specified in tmpfiles snippets may now be
14669 prefixed with "~", which indicates that they shall be masked
14670 by whether the existing file or directory is currently
14671 writable, readable or executable at all. Also, if specified,
14672 the sgid/suid/sticky bits will be masked for all
14675 * A new passive target unit "network-pre.target" has been
14676 added which is useful for services that shall run before any
14677 network is configured, for example firewall scripts.
14679 * The "floppy" group that previously owned the /dev/fd*
14680 devices is no longer used. The "disk" group is now used
14681 instead. Distributions should probably deprecate usage of
14684 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
14685 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
14686 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
14687 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
14688 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
14689 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
14692 — Berlin, 2014-06-11
14696 * A new "systemd-timesyncd" daemon has been added for
14697 synchronizing the system clock across the network. It
14698 implements an SNTP client. In contrast to NTP
14699 implementations such as chrony or the NTP reference server,
14700 this only implements a client side, and does not bother with
14701 the full NTP complexity, focusing only on querying time from
14702 one remote server and synchronizing the local clock to
14703 it. Unless you intend to serve NTP to networked clients or
14704 want to connect to local hardware clocks, this simple NTP
14705 client should be more than appropriate for most
14706 installations. The daemon runs with minimal privileges, and
14707 has been hooked up with networkd to only operate when
14708 network connectivity is available. The daemon saves the
14709 current clock to disk every time a new NTP sync has been
14710 acquired, and uses this to possibly correct the system clock
14711 early at bootup, in order to accommodate for systems that
14712 lack an RTC such as the Raspberry Pi and embedded devices,
14713 and to make sure that time monotonically progresses on these
14714 systems, even if it is not always correct. To make use of
14715 this daemon, a new system user and group "systemd-timesync"
14716 needs to be created on installation of systemd.
14718 * The queue "seqnum" interface of libudev has been disabled, as
14719 it was generally incompatible with device namespacing as
14720 sequence numbers of devices go "missing" if the devices are
14721 part of a different namespace.
14723 * "systemctl list-timers" and "systemctl list-sockets" gained
14724 a --recursive switch for showing units of these types also
14725 for all local containers, similar in style to the already
14726 supported --recursive switch for "systemctl list-units".
14728 * A new RebootArgument= setting has been added for service
14729 units, which may be used to specify a kernel reboot argument
14730 to use when triggering reboots with StartLimitAction=.
14732 * A new FailureAction= setting has been added for service
14733 units which may be used to specify an operation to trigger
14734 when a service fails. This works similarly to
14735 StartLimitAction=, but unlike it, controls what is done
14736 immediately rather than only after several attempts to
14737 restart the service in question.
14739 * hostnamed got updated to also expose the kernel name,
14740 release, and version on the bus. This is useful for
14741 executing commands like hostnamectl with the -H switch.
14742 systemd-analyze makes use of this to properly display
14743 details when running non-locally.
14745 * The bootchart tool can now show cgroup information in the
14746 graphs it generates.
14748 * The CFS CPU quota cgroup attribute is now exposed for
14749 services. The new CPUQuota= switch has been added for this
14750 which takes a percentage value. Setting this will have the
14751 result that a service may never get more CPU time than the
14752 specified percentage, even if the machine is otherwise idle.
14754 * systemd-networkd learned IPIP and SIT tunnel support.
14756 * LSB init scripts exposing a dependency on $network will now
14757 get a dependency on network-online.target rather than simply
14758 network.target. This should bring LSB handling closer to
14759 what it was on SysV systems.
14761 * A new fsck.repair= kernel option has been added to control
14762 how fsck shall deal with unclean file systems at boot.
14764 * The (.ini) configuration file parser will now silently ignore
14765 sections whose names begin with "X-". This may be used to maintain
14766 application-specific extension sections in unit files.
14768 * machined gained a new API to query the IP addresses of
14769 registered containers. "machinectl status" has been updated
14770 to show these addresses in its output.
14772 * A new call sd_uid_get_display() has been added to the
14773 sd-login APIs for querying the "primary" session of a
14774 user. The "primary" session of the user is elected from the
14775 user's sessions and generally a graphical session is
14776 preferred over a text one.
14778 * A minimal systemd-resolved daemon has been added. It
14779 currently simply acts as a companion to systemd-networkd and
14780 manages resolv.conf based on per-interface DNS
14781 configuration, possibly supplied via DHCP. In the long run
14782 we hope to extend this into a local DNSSEC enabled DNS and
14785 * The systemd-networkd-wait-online tool is now enabled by
14786 default. It will delay network-online.target until a network
14787 connection has been configured. The tool primarily integrates
14788 with networkd, but will also make a best effort to make sense
14789 of network configuration performed in some other way.
14791 * Two new service options StartupCPUShares= and
14792 StartupBlockIOWeight= have been added that work similarly to
14793 CPUShares= and BlockIOWeight= however only apply during
14794 system startup. This is useful to prioritize certain services
14795 differently during bootup than during normal runtime.
14797 * hostnamed has been changed to prefer the statically
14798 configured hostname in /etc/hostname (unless set to
14799 'localhost' or empty) over any dynamic one supplied by
14800 dhcp. With this change, the rules for picking the hostname
14801 match more closely the rules of other configuration settings
14802 where the local administrator's configuration in /etc always
14803 overrides any other settings.
14805 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
14806 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
14807 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
14808 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
14809 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
14810 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
14811 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
14812 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
14813 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
14814 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
14815 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
14816 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
14817 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
14818 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
14819 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
14820 Lindskog, WaLyong Cho, Will Woods, Zbigniew
14823 — Beijing, 2014-05-28
14827 * When restoring the screen brightness at boot, stay away from
14828 the darkest setting or from the lowest 5% of the available
14829 range, depending on which is the larger value of both. This
14830 should effectively protect the user from rebooting into a
14831 black screen, should the brightness have been set to minimum
14834 * sd-login gained a new sd_machine_get_class() call to
14835 determine the class ("vm" or "container") of a machine
14836 registered with machined.
14838 * sd-login gained new calls
14839 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
14840 to query the identity of the peer of a local AF_UNIX
14841 connection. They operate similarly to their sd_pid_get_xyz()
14844 * PID 1 will now maintain a system-wide system state engine
14845 with the states "starting", "running", "degraded",
14846 "maintenance", "stopping". These states are bound to system
14847 startup, normal runtime, runtime with at least one failed
14848 service, rescue/emergency mode and system shutdown. This
14849 state is shown in the "systemctl status" output when no unit
14850 name is passed. It is useful to determine system state, in
14851 particularly when doing so for many systems or containers at
14854 * A new command "list-machines" has been added to "systemctl"
14855 that lists all local OS containers and shows their system
14856 state (see above), if systemd runs inside of them.
14858 * systemctl gained a new "-r" switch to recursively enumerate
14859 units on all local containers, when used with the
14860 "list-unit" command (which is the default one that is
14861 executed when no parameters are specified).
14863 * The GPT automatic partition discovery logic will now honour
14864 two GPT partition flags: one may be set on a partition to
14865 cause it to be mounted read-only, and the other may be set
14866 on a partition to ignore it during automatic discovery.
14868 * Two new GPT type UUIDs have been added for automatic root
14869 partition discovery, for 32-bit and 64-bit ARM. This is not
14870 particularly useful for discovering the root directory on
14871 these architectures during bare-metal boots (since UEFI is
14872 not common there), but still very useful to allow booting of
14873 ARM disk images in nspawn with the -i option.
14875 * MAC addresses of interfaces created with nspawn's
14876 --network-interface= switch will now be generated from the
14877 machine name, and thus be stable between multiple invocations
14880 * logind will now automatically remove all IPC objects owned
14881 by a user if she or he fully logs out. This makes sure that
14882 users who are logged out cannot continue to consume IPC
14883 resources. This covers SysV memory, semaphores and message
14884 queues as well as POSIX shared memory and message
14885 queues. Traditionally, SysV and POSIX IPC had no lifecycle
14886 limits. With this functionality, that is corrected. This may
14887 be turned off by using the RemoveIPC= switch of logind.conf.
14889 * The systemd-machine-id-setup and tmpfiles tools gained a
14890 --root= switch to operate on a specific root directory,
14893 * journald can now forward logged messages to the TTYs of all
14894 logged in users ("wall"). This is the default for all
14895 emergency messages now.
14897 * A new tool systemd-journal-remote has been added to stream
14898 journal log messages across the network.
14900 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
14901 controller trees are mounted into it. Note that the
14902 directories mounted beneath it are not read-only. This is a
14903 security measure and is particularly useful because glibc
14904 actually includes a search logic to pick any tmpfs it can
14905 find to implement shm_open() if /dev/shm is not available
14906 (which it might very well be in namespaced setups).
14908 * machinectl gained a new "poweroff" command to cleanly power
14909 down a local OS container.
14911 * The PrivateDevices= unit file setting will now also drop the
14912 CAP_MKNOD capability from the capability bound set, and
14913 imply DevicePolicy=closed.
14915 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
14916 comprehensively on all long-running systemd services where
14917 this is appropriate.
14919 * systemd-udevd will now run in a disassociated mount
14920 namespace. To mount directories from udev rules, make sure to
14921 pull in mount units via SYSTEMD_WANTS properties.
14923 * The kdbus support gained support for uploading policy into
14924 the kernel. sd-bus gained support for creating "monitoring"
14925 connections that can eavesdrop into all bus communication
14926 for debugging purposes.
14928 * Timestamps may now be specified in seconds since the UNIX
14929 epoch Jan 1st, 1970 by specifying "@" followed by the value
14932 * Native tcpwrap support in systemd has been removed. tcpwrap
14933 is old code, not really maintained anymore and has serious
14934 shortcomings, and better options such as firewalls
14935 exist. For setups that require tcpwrap usage, please
14936 consider invoking your socket-activated service via tcpd,
14937 like on traditional inetd.
14939 * A new system.conf configuration option
14940 DefaultTimerAccuracySec= has been added that controls the
14941 default AccuracySec= setting of .timer units.
14943 * Timer units gained a new WakeSystem= switch. If enabled,
14944 timers configured this way will cause the system to resume
14945 from system suspend (if the system supports that, which most
14948 * Timer units gained a new Persistent= switch. If enabled,
14949 timers configured this way will save to disk when they have
14950 been last triggered. This information is then used on next
14951 reboot to possible execute overdue timer events, that
14952 could not take place because the system was powered off.
14953 This enables simple anacron-like behaviour for timer units.
14955 * systemctl's "list-timers" will now also list the time a
14956 timer unit was last triggered in addition to the next time
14957 it will be triggered.
14959 * systemd-networkd will now assign predictable IPv4LL
14960 addresses to its local interfaces.
14962 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
14963 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
14964 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
14965 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
14966 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
14967 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
14968 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
14969 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
14972 — Berlin, 2014-03-25
14976 * A new unit file setting RestrictAddressFamilies= has been
14977 added to restrict which socket address families unit
14978 processes gain access to. This takes address family names
14979 like "AF_INET" or "AF_UNIX", and is useful to minimize the
14980 attack surface of services via exotic protocol stacks. This
14981 is built on seccomp system call filters.
14983 * Two new unit file settings RuntimeDirectory= and
14984 RuntimeDirectoryMode= have been added that may be used to
14985 manage a per-daemon runtime directories below /run. This is
14986 an alternative for setting up directory permissions with
14987 tmpfiles snippets, and has the advantage that the runtime
14988 directory's lifetime is bound to the daemon runtime and that
14989 the daemon starts up with an empty directory each time. This
14990 is particularly useful when writing services that drop
14991 privileges using the User= or Group= setting.
14993 * The DeviceAllow= unit setting now supports globbing for
14994 matching against device group names.
14996 * The systemd configuration file system.conf gained new
14997 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
14998 DefaultMemoryAccounting= to globally turn on/off accounting
14999 for specific resources (cgroups) for all units. These
15000 settings may still be overridden individually in each unit
15003 * systemd-gpt-auto-generator is now able to discover /srv and
15004 root partitions in addition to /home and swap partitions. It
15005 also supports LUKS-encrypted partitions now. With this in
15006 place, automatic discovery of partitions to mount following
15007 the Discoverable Partitions Specification
15008 (https://systemd.io/DISCOVERABLE_PARTITIONS/)
15009 is now a lot more complete. This allows booting without
15010 /etc/fstab and without root= on the kernel command line on
15011 systems prepared appropriately.
15013 * systemd-nspawn gained a new --image= switch which allows
15014 booting up disk images and Linux installations on any block
15015 device that follow the Discoverable Partitions Specification
15016 (see above). This means that installations made with
15017 appropriately updated installers may now be started and
15018 deployed using container managers, completely
15019 unmodified. (We hope that libvirt-lxc will add support for
15020 this feature soon, too.)
15022 * systemd-nspawn gained a new --network-macvlan= setting to
15023 set up a private macvlan interface for the
15024 container. Similarly, systemd-networkd gained a new
15025 Kind=macvlan setting in .netdev files.
15027 * systemd-networkd now supports configuring local addresses
15030 * A new tool systemd-network-wait-online has been added to
15031 synchronously wait for network connectivity using
15034 * The sd-bus.h bus API gained a new sd_bus_track object for
15035 tracking the lifecycle of bus peers. Note that sd-bus.h is
15036 still not a public API though (unless you specify
15037 --enable-kdbus on the configure command line, which however
15038 voids your warranty and you get no API stability guarantee).
15040 * The $XDG_RUNTIME_DIR runtime directories for each user are
15041 now individual tmpfs instances, which has the benefit of
15042 introducing separate pools for each user, with individual
15043 size limits, and thus making sure that unprivileged clients
15044 can no longer negatively impact the system or other users by
15045 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
15046 RuntimeDirectorySize= has been introduced that allows
15047 controlling the default size limit for all users. It
15048 defaults to 10% of the available physical memory. This is no
15049 replacement for quotas on tmpfs though (which the kernel
15050 still does not support), as /dev/shm and /tmp are still
15051 shared resources used by both the system and unprivileged
15054 * logind will now automatically turn off automatic suspending
15055 on laptop lid close when more than one display is
15056 connected. This was previously expected to be implemented
15057 individually in desktop environments (such as GNOME),
15058 however has been added to logind now, in order to fix a
15059 boot-time race where a desktop environment might not have
15060 been started yet and thus not been able to take an inhibitor
15061 lock at the time where logind already suspends the system
15062 due to a closed lid.
15064 * logind will now wait at least 30s after each system
15065 suspend/resume cycle, and 3min after system boot before
15066 suspending the system due to a closed laptop lid. This
15067 should give USB docking stations and similar enough time to
15068 be probed and configured after system resume and boot in
15069 order to then act as suspend blocker.
15071 * systemd-run gained a new --property= setting which allows
15072 initialization of resource control properties (and others)
15073 for the created scope or service unit. Example: "systemd-run
15074 --property=BlockIOWeight=10 updatedb" may be used to run
15075 updatedb at a low block IO scheduling weight.
15077 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
15078 now also work in --scope mode.
15080 * When systemd is compiled with kdbus support, basic support
15081 for enforced policies is now in place. (Note that enabling
15082 kdbus still voids your warranty and no API compatibility
15083 promises are made.)
15085 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
15086 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
15087 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
15088 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
15089 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
15090 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
15091 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
15092 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
15093 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
15094 Zbigniew Jędrzejewski-Szmek
15096 — Berlin, 2014-03-12
15100 * systemd will now relabel /dev after loading the SMACK policy
15101 according to SMACK rules.
15103 * A new unit file option AppArmorProfile= has been added to
15104 set the AppArmor profile for the processes of a unit.
15106 * A new condition check ConditionArchitecture= has been added
15107 to conditionalize units based on the system architecture, as
15108 reported by uname()'s "machine" field.
15110 * systemd-networkd now supports matching on the system
15111 virtualization, architecture, kernel command line, hostname
15114 * logind is now a lot more aggressive when suspending the
15115 machine due to a closed laptop lid. Instead of acting only
15116 on the lid close action, it will continuously watch the lid
15117 status and act on it. This is useful for laptops where the
15118 power button is on the outside of the chassis so that it can
15119 be reached without opening the lid (such as the Lenovo
15120 Yoga). On those machines, logind will now immediately
15121 re-suspend the machine if the power button has been
15122 accidentally pressed while the laptop was suspended and in a
15123 backpack or similar.
15125 * logind will now watch SW_DOCK switches and inhibit reaction
15126 to the lid switch if it is pressed. This means that logind
15127 will not suspend the machine anymore if the lid is closed
15128 and the system is docked, if the laptop supports SW_DOCK
15129 notifications via the input layer. Note that ACPI docking
15130 stations do not generate this currently. Also note that this
15131 logic is usually not fully sufficient and Desktop
15132 Environments should take a lid switch inhibitor lock when an
15133 external display is connected, as systemd will not watch
15136 * nspawn will now make use of the devices cgroup controller by
15137 default, and only permit creation of and access to the usual
15138 API device nodes like /dev/null or /dev/random, as well as
15139 access to (but not creation of) the pty devices.
15141 * We will now ship a default .network file for
15142 systemd-networkd that automatically configures DHCP for
15143 network interfaces created by nspawn's --network-veth or
15144 --network-bridge= switches.
15146 * systemd will now understand the usual M, K, G, T suffixes
15147 according to SI conventions (i.e. to the base 1000) when
15148 referring to throughput and hardware metrics. It will stay
15149 with IEC conventions (i.e. to the base 1024) for software
15150 metrics, according to what is customary according to
15151 Wikipedia. We explicitly document which base applies for
15152 each configuration option.
15154 * The DeviceAllow= setting in unit files now supports a syntax to
15155 allow-list an entire group of devices node majors at once, based on
15156 the /proc/devices listing. For example, with the string "char-pts",
15157 it is now possible to allow-list all current and future pseudo-TTYs
15160 * sd-event learned a new "post" event source. Event sources of
15161 this type are triggered by the dispatching of any event
15162 source of a type that is not "post". This is useful for
15163 implementing clean-up and check event sources that are
15164 triggered by other work being done in the program.
15166 * systemd-networkd is no longer statically enabled, but uses
15167 the usual [Install] sections so that it can be
15168 enabled/disabled using systemctl. It still is enabled by
15171 * When creating a veth interface pair with systemd-nspawn, the
15172 host side will now be prefixed with "vb-" if
15173 --network-bridge= is used, and with "ve-" if --network-veth
15174 is used. This way, it is easy to distinguish these cases on
15175 the host, for example to apply different configuration to
15176 them with systemd-networkd.
15178 * The compatibility libraries for libsystemd-journal.so,
15179 libsystem-id128.so, libsystemd-login.so and
15180 libsystemd-daemon.so do not make use of IFUNC
15181 anymore. Instead, we now build libsystemd.so multiple times
15182 under these alternative names. This means that the footprint
15183 is drastically increased, but given that these are
15184 transitional compatibility libraries, this should not matter
15185 much. This change has been made necessary to support the ARM
15186 platform for these compatibility libraries, as the ARM
15187 toolchain is not really at the same level as the toolchain
15188 for other architectures like x86 and does not support
15189 IFUNC. Please make sure to use --enable-compat-libs only
15190 during a transitional period!
15192 * The .include syntax has been deprecated and is not documented
15193 anymore. Drop-in files in .d directories should be used instead.
15195 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
15196 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
15197 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
15198 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
15199 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
15200 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
15201 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
15202 Zbigniew Jędrzejewski-Szmek
15204 — Berlin, 2014-02-24
15208 * A new component "systemd-networkd" has been added that can
15209 be used to configure local network interfaces statically or
15210 via DHCP. It is capable of bringing up bridges, VLANs, and
15211 bonding. Currently, no hook-ups for interactive network
15212 configuration are provided. Use this for your initrd,
15213 container, embedded, or server setup if you need a simple,
15214 yet powerful, network configuration solution. This
15215 configuration subsystem is quite nifty, as it allows wildcard
15216 hotplug matching in interfaces. For example, with a single
15217 configuration snippet, you can configure that all Ethernet
15218 interfaces showing up are automatically added to a bridge,
15219 or similar. It supports link-sensing and more.
15221 * A new tool "systemd-socket-proxyd" has been added which can
15222 act as a bidirectional proxy for TCP sockets. This is
15223 useful for adding socket activation support to services that
15224 do not actually support socket activation, including virtual
15225 machines and the like.
15227 * Add a new tool to save/restore rfkill state on
15230 * Save/restore state of keyboard backlights in addition to
15231 display backlights on shutdown/boot.
15233 * udev learned a new SECLABEL{} construct to label device
15234 nodes with a specific security label when they appear. For
15235 now, only SECLABEL{selinux} is supported, but the syntax is
15236 prepared for additional security frameworks.
15238 * udev gained a new scheme to configure link-level attributes
15239 from files in /etc/systemd/network/*.link. These files can
15240 match against MAC address, device path, driver name and type,
15241 and will apply attributes like the naming policy, link speed,
15242 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
15243 address assignment policy (randomized, …).
15245 * The configuration of network interface naming rules for
15246 "permanent interface names" has changed: a new NamePolicy=
15247 setting in the [Link] section of .link files determines the
15248 priority of possible naming schemes (onboard, slot, MAC,
15249 path). The default value of this setting is determined by
15250 /usr/lib/net/links/99-default.link. Old
15251 80-net-name-slot.rules udev configuration file has been
15252 removed, so local configuration overriding this file should
15253 be adapted to override 99-default.link instead.
15255 * When the User= switch is used in a unit file, also
15256 initialize $SHELL= based on the user database entry.
15258 * systemd no longer depends on libdbus. All communication is
15259 now done with sd-bus, systemd's low-level bus library
15262 * kdbus support has been added to PID 1 itself. When kdbus is
15263 enabled, this causes PID 1 to set up the system bus and
15264 enable support for a new ".busname" unit type that
15265 encapsulates bus name activation on kdbus. It works a little
15266 bit like ".socket" units, except for bus names. A new
15267 generator has been added that converts classic dbus1 service
15268 activation files automatically into native systemd .busname
15269 and .service units.
15271 * sd-bus: add a light-weight vtable implementation that allows
15272 defining objects on the bus with a simple static const
15273 vtable array of its methods, signals and properties.
15275 * systemd will not generate or install static dbus
15276 introspection data anymore to /usr/share/dbus-1/interfaces,
15277 as the precise format of these files is unclear, and
15278 nothing makes use of it.
15280 * A proxy daemon is now provided to proxy clients connecting
15281 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
15282 compatibility with classic D-Bus.
15284 * A bus driver implementation has been added that supports the
15285 classic D-Bus bus driver calls on kdbus, also for
15286 compatibility purposes.
15288 * A new API "sd-event.h" has been added that implements a
15289 minimal event loop API built around epoll. It provides a
15290 couple of features that direct epoll usage is lacking:
15291 prioritization of events, scales to large numbers of timer
15292 events, per-event timer slack (accuracy), system-wide
15293 coalescing of timer events, exit handlers, watchdog
15294 supervision support using systemd's sd_notify() API, child
15297 * A new API "sd-rntl.h" has been added that provides an API
15298 around the route netlink interface of the kernel, similar in
15299 style to "sd-bus.h".
15301 * A new API "sd-dhcp-client.h" has been added that provides a
15302 small DHCPv4 client-side implementation. This is used by
15303 "systemd-networkd".
15305 * There is a new kernel command line option
15306 "systemd.restore_state=0|1". When set to "0", none of the
15307 systemd tools will restore saved runtime state to hardware
15308 devices. More specifically, the rfkill and backlight states
15311 * The FsckPassNo= compatibility option in mount/service units
15312 has been removed. The fstab generator will now add the
15313 necessary dependencies automatically, and does not require
15314 PID1's support for that anymore.
15316 * journalctl gained a new switch, --list-boots, that lists
15317 recent boots with their times and boot IDs.
15319 * The various tools like systemctl, loginctl, timedatectl,
15320 busctl, systemd-run, … have gained a new switch "-M" to
15321 connect to a specific, local OS container (as direct
15322 connection, without requiring SSH). This works on any
15323 container that is registered with machined, such as those
15324 created by libvirt-lxc or nspawn.
15326 * systemd-run and systemd-analyze also gained support for "-H"
15327 to connect to remote hosts via SSH. This is particularly
15328 useful for systemd-run because it enables queuing of jobs
15329 onto remote systems.
15331 * machinectl gained a new command "login" to open a getty
15332 login in any local container. This works with any container
15333 that is registered with machined (such as those created by
15334 libvirt-lxc or nspawn), and which runs systemd inside.
15336 * machinectl gained a new "reboot" command that may be used to
15337 trigger a reboot on a specific container that is registered
15338 with machined. This works on any container that runs an init
15339 system of some kind.
15341 * systemctl gained a new "list-timers" command to print a nice
15342 listing of installed timer units with the times they elapse
15345 * Alternative reboot() parameters may now be specified on the
15346 "systemctl reboot" command line and are passed to the
15347 reboot() system call.
15349 * systemctl gained a new --job-mode= switch to configure the
15350 mode to queue a job with. This is a more generic version of
15351 --fail, --irreversible, and --ignore-dependencies, which are
15352 still available but not advertised anymore.
15354 * /etc/systemd/system.conf gained new settings to configure
15355 various default timeouts of units, as well as the default
15356 start limit interval and burst. These may still be overridden
15359 * PID1 will now export on the bus profile data of the security
15360 policy upload process (such as the SELinux policy upload to
15363 * journald: when forwarding logs to the console, include
15364 timestamps (following the setting in
15365 /sys/module/printk/parameters/time).
15367 * OnCalendar= in timer units now understands the special
15368 strings "yearly" and "annually". (Both are equivalent)
15370 * The accuracy of timer units is now configurable with the new
15371 AccuracySec= setting. It defaults to 1min.
15373 * A new dependency type JoinsNamespaceOf= has been added that
15374 allows running two services within the same /tmp and network
15375 namespace, if PrivateNetwork= or PrivateTmp= are used.
15377 * A new command "cat" has been added to systemctl. It outputs
15378 the original unit file of a unit, and concatenates the
15379 contents of additional "drop-in" unit file snippets, so that
15380 the full configuration is shown.
15382 * systemctl now supports globbing on the various "list-xyz"
15383 commands, like "list-units" or "list-sockets", as well as on
15384 those commands which take multiple unit names.
15386 * journalctl's --unit= switch gained support for globbing.
15388 * All systemd daemons now make use of the watchdog logic so
15389 that systemd automatically notices when they hang.
15391 * If the $container_ttys environment variable is set,
15392 getty-generator will automatically spawn a getty for each
15393 listed tty. This is useful for container managers to request
15394 login gettys to be spawned on as many ttys as needed.
15396 * %h, %s, %U specifier support is not available anymore when
15397 used in unit files for PID 1. This is because NSS calls are
15398 not safe from PID 1. They stay available for --user
15399 instances of systemd, and as special case for the root user.
15401 * loginctl gained a new "--no-legend" switch to turn off output
15402 of the legend text.
15404 * The "sd-login.h" API gained three new calls:
15405 sd_session_is_remote(), sd_session_get_remote_user(),
15406 sd_session_get_remote_host() to query information about
15409 * The udev hardware database now also carries vendor/product
15410 information of SDIO devices.
15412 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
15413 determine whether watchdog notifications are requested by
15414 the system manager.
15416 * Socket-activated per-connection services now include a
15417 short description of the connection parameters in the
15420 * tmpfiles gained a new "--boot" option. When this is not used,
15421 only lines where the command character is not suffixed with
15422 "!" are executed. When this option is specified, those
15423 options are executed too. This partitions tmpfiles
15424 directives into those that can be safely executed at any
15425 time, and those which should be run only at boot (for
15426 example, a line that creates /run/nologin).
15428 * A new API "sd-resolve.h" has been added which provides a simple
15429 asynchronous wrapper around glibc NSS hostname resolution
15430 calls, such as getaddrinfo(). In contrast to glibc's
15431 getaddrinfo_a(), it does not use signals. In contrast to most
15432 other asynchronous name resolution libraries, this one does
15433 not reimplement DNS, but reuses NSS, so that alternate
15434 hostname resolution systems continue to work, such as mDNS,
15435 LDAP, etc. This API is based on libasyncns, but it has been
15436 cleaned up for inclusion in systemd.
15438 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
15439 "sd-daemon.h" are no longer found in individual libraries
15440 libsystemd-journal.so, libsystemd-login.so,
15441 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
15442 merged them into a single library, libsystemd.so, which
15443 provides all symbols. The reason for this is cyclic
15444 dependencies, as these libraries tend to use each other's
15445 symbols. So far, we have managed to workaround that by linking
15446 a copy of a good part of our code into each of these
15447 libraries again and again, which, however, makes certain
15448 things hard to do, like sharing static variables. Also, it
15449 substantially increases footprint. With this change, there
15450 is only one library for the basic APIs systemd
15451 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
15452 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
15453 library as well, however are subject to the --enable-kdbus
15454 switch (see below). Note that "sd-dhcp-client.h" is not part
15455 of this library (this is because it only consumes, never
15456 provides, services of/to other APIs). To make the transition
15457 easy from the separate libraries to the unified one, we
15458 provide the --enable-compat-libs compile-time switch which
15459 will generate stub libraries that are compatible with the
15460 old ones but redirect all calls to the new one.
15462 * All of the kdbus logic and the new APIs "sd-bus.h",
15463 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
15464 and "sd-utf8.h" are compile-time optional via the
15465 "--enable-kdbus" switch, and they are not compiled in by
15466 default. To make use of kdbus, you have to explicitly enable
15467 the switch. Note however, that neither the kernel nor the
15468 userspace API for all of this is considered stable yet. We
15469 want to maintain the freedom to still change the APIs for
15470 now. By specifying this build-time switch, you acknowledge
15471 that you are aware of the instability of the current
15474 * Also, note that while kdbus is pretty much complete,
15475 it lacks one thing: proper policy support. This means you
15476 can build a fully working system with all features; however,
15477 it will be highly insecure. Policy support will be added in
15478 one of the next releases, at the same time that we will
15479 declare the APIs stable.
15481 * When the kernel command line argument "kdbus" is specified,
15482 systemd will automatically load the kdbus.ko kernel module. At
15483 this stage of development, it is only useful for testing kdbus
15484 and should not be used in production. Note: if "--enable-kdbus"
15485 is specified, and the kdbus.ko kernel module is available, and
15486 "kdbus" is added to the kernel command line, the entire system
15487 runs with kdbus instead of dbus-daemon, with the above mentioned
15488 problem of missing the system policy enforcement. Also a future
15489 version of kdbus.ko or a newer systemd will not be compatible with
15490 each other, and will unlikely be able to boot the machine if only
15491 one of them is updated.
15493 * systemctl gained a new "import-environment" command which
15494 uploads the caller's environment (or parts thereof) into the
15495 service manager so that it is inherited by services started
15496 by the manager. This is useful to upload variables like
15497 $DISPLAY into the user service manager.
15499 * A new PrivateDevices= switch has been added to service units
15500 which allows running a service with a namespaced /dev
15501 directory that does not contain any device nodes for
15502 physical devices. More specifically, it only includes devices
15503 such as /dev/null, /dev/urandom, and /dev/zero which are API
15506 * logind has been extended to support behaviour like VT
15507 switching on seats that do not support a VT. This makes
15508 multi-session available on seats that are not the first seat
15509 (seat0), and on systems where kernel support for VTs has
15510 been disabled at compile-time.
15512 * If a process holds a delay lock for system sleep or shutdown
15513 and fails to release it in time, we will now log its
15514 identity. This makes it easier to identify processes that
15515 cause slow suspends or power-offs.
15517 * When parsing /etc/crypttab, support for a new key-slot=
15518 option as supported by Debian is added. It allows indicating
15519 which LUKS slot to use on disk, speeding up key loading.
15521 * The sd_journal_sendv() API call has been checked and
15522 officially declared to be async-signal-safe so that it may
15523 be invoked from signal handlers for logging purposes.
15525 * Boot-time status output is now enabled automatically after a
15526 short timeout if boot does not progress, in order to give
15527 the user an indication what she or he is waiting for.
15529 * The boot-time output has been improved to show how much time
15530 remains until jobs expire.
15532 * The KillMode= switch in service units gained a new possible
15533 value "mixed". If set, and the unit is shut down, then the
15534 initial SIGTERM signal is sent only to the main daemon
15535 process, while the following SIGKILL signal is sent to
15536 all remaining processes of the service.
15538 * When a scope unit is registered, a new property "Controller"
15539 may be set. If set to a valid bus name, systemd will send a
15540 RequestStop() signal to this name when it would like to shut
15541 down the scope. This may be used to hook manager logic into
15542 the shutdown logic of scope units. Also, scope units may now
15543 be put in a special "abandoned" state, in which case the
15544 manager process which created them takes no further
15545 responsibilities for it.
15547 * When reading unit files, systemd will now verify
15548 the access mode of these files, and warn about certain
15549 suspicious combinations. This has been added to make it
15550 easier to track down packaging bugs where unit files are
15551 marked executable or world-writable.
15553 * systemd-nspawn gained a new "--setenv=" switch to set
15554 container-wide environment variables. The similar option in
15555 systemd-activate was renamed from "--environment=" to
15556 "--setenv=" for consistency.
15558 * systemd-nspawn has been updated to create a new kdbus domain
15559 for each container that is invoked, thus allowing each
15560 container to have its own set of system and user buses,
15561 independent of the host.
15563 * systemd-nspawn gained a new --drop-capability= switch to run
15564 the container with less capabilities than the default. Both
15565 --drop-capability= and --capability= now take the special
15566 string "all" for dropping or keeping all capabilities.
15568 * systemd-nspawn gained new switches for executing containers
15569 with specific SELinux labels set.
15571 * systemd-nspawn gained a new --quiet switch to not generate
15572 any additional output but the container's own console
15575 * systemd-nspawn gained a new --share-system switch to run a
15576 container without PID namespacing enabled.
15578 * systemd-nspawn gained a new --register= switch to control
15579 whether the container is registered with systemd-machined or
15580 not. This is useful for containers that do not run full
15581 OS images, but only specific apps.
15583 * systemd-nspawn gained a new --keep-unit which may be used
15584 when invoked as the only program from a service unit, and
15585 results in registration of the unit service itself in
15586 systemd-machined, instead of a newly opened scope unit.
15588 * systemd-nspawn gained a new --network-interface= switch for
15589 moving arbitrary interfaces to the container. The new
15590 --network-veth switch creates a virtual Ethernet connection
15591 between host and container. The new --network-bridge=
15592 switch then allows assigning the host side of this virtual
15593 Ethernet connection to a bridge device.
15595 * systemd-nspawn gained a new --personality= switch for
15596 setting the kernel personality for the container. This is
15597 useful when running a 32-bit container on a 64-bit host. A
15598 similar option Personality= is now also available for service
15601 * logind will now also track a "Desktop" identifier for each
15602 session which encodes the desktop environment of it. This is
15603 useful for desktop environments that want to identify
15604 multiple running sessions of itself easily.
15606 * A new SELinuxContext= setting for service units has been
15607 added that allows setting a specific SELinux execution
15608 context for a service.
15610 * Most systemd client tools will now honour $SYSTEMD_LESS for
15611 settings of the "less" pager. By default, these tools will
15612 override $LESS to allow certain operations to work, such as
15613 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
15614 influence this logic.
15616 * systemd's "seccomp" hook-up has been changed to make use of
15617 the libseccomp library instead of using its own
15618 implementation. This has benefits for portability among
15621 * For usage together with SystemCallFilter=, a new
15622 SystemCallErrorNumber= setting has been introduced that
15623 allows configuration of a system error number to be returned
15624 on filtered system calls, instead of immediately killing the
15625 process. Also, SystemCallArchitectures= has been added to
15626 limit access to system calls of a particular architecture
15627 (in order to turn off support for unused secondary
15628 architectures). There is also a global
15629 SystemCallArchitectures= setting in system.conf now to turn
15630 off support for non-native system calls system-wide.
15632 * systemd requires a kernel with a working name_to_handle_at(),
15633 please see the kernel config requirements in the README file.
15635 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
15636 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
15637 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
15638 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
15639 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
15640 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
15641 Elia Pinto, Florian Weimer, George McCollister, Goffredo
15642 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
15643 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
15644 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
15645 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
15646 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
15647 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
15648 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
15649 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
15650 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
15651 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
15652 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
15653 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
15654 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
15655 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
15656 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
15657 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
15658 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
15660 — Berlin, 2014-02-20
15664 * logind has gained support for facilitating privileged input
15665 and drm device access for unprivileged clients. This work is
15666 useful to allow Wayland display servers (and similar
15667 programs, such as kmscon) to run under the user's ID and
15668 access input and drm devices which are normally
15669 protected. When this is used (and the kernel is new enough)
15670 logind will "mute" IO on the file descriptors passed to
15671 Wayland as long as it is in the background and "unmute" it
15672 if it returns into the foreground. This allows secure
15673 session switching without allowing background sessions to
15674 eavesdrop on input and display data. This also introduces
15675 session switching support if VT support is turned off in the
15676 kernel, and on seats that are not seat0.
15678 * A new kernel command line option luks.options= is understood
15679 now which allows specifying LUKS options for usage for LUKS
15680 encrypted partitions specified with luks.uuid=.
15682 * tmpfiles.d(5) snippets may now use specifier expansion in
15683 path names. More specifically %m, %b, %H, %v, are now
15684 replaced by the local machine id, boot id, hostname, and
15685 kernel version number.
15687 * A new tmpfiles.d(5) command "m" has been introduced which
15688 may be used to change the owner/group/access mode of a file
15689 or directory if it exists, but do nothing if it does not.
15691 * This release removes high-level support for the
15692 MemorySoftLimit= cgroup setting. The underlying kernel
15693 cgroup attribute memory.soft_limit= is currently badly
15694 designed and likely to be removed from the kernel API in its
15695 current form, hence we should not expose it for now.
15697 * The memory.use_hierarchy cgroup attribute is now enabled for
15698 all cgroups systemd creates in the memory cgroup
15699 hierarchy. This option is likely to be come the built-in
15700 default in the kernel anyway, and the non-hierarchical mode
15701 never made much sense in the intrinsically hierarchical
15704 * A new field _SYSTEMD_SLICE= is logged along with all journal
15705 messages containing the slice a message was generated
15706 from. This is useful to allow easy per-customer filtering of
15707 logs among other things.
15709 * systemd-journald will no longer adjust the group of journal
15710 files it creates to the "systemd-journal" group. Instead we
15711 rely on the journal directory to be owned by the
15712 "systemd-journal" group, and its setgid bit set, so that the
15713 kernel file system layer will automatically enforce that
15714 journal files inherit this group assignment. The reason for
15715 this change is that we cannot allow NSS look-ups from
15716 journald which would be necessary to resolve
15717 "systemd-journal" to a numeric GID, because this might
15718 create deadlocks if NSS involves synchronous queries to
15719 other daemons (such as nscd, or sssd) which in turn are
15720 logging clients of journald and might block on it, which
15721 would then dead lock. A tmpfiles.d(5) snippet included in
15722 systemd will make sure the setgid bit and group are
15723 properly set on the journal directory if it exists on every
15724 boot. However, we recommend adjusting it manually after
15725 upgrades too (or from RPM scriptlets), so that the change is
15726 not delayed until next reboot.
15728 * Backlight and random seed files in /var/lib/ have moved into
15729 the /var/lib/systemd/ directory, in order to centralize all
15730 systemd generated files in one directory.
15732 * Boot time performance measurements (as displayed by
15733 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
15734 performance information if that's available to determine how
15735 much time BIOS and boot loader initialization required. With
15736 a sufficiently new BIOS you hence no longer need to boot
15737 with Gummiboot to get access to such information.
15739 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
15740 Cristian Rodríguez, Dave Reisner, David Herrmann, David
15741 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
15742 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
15743 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
15744 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
15745 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
15747 — Berlin, 2013-10-02
15751 * The Restart= option for services now understands a new
15752 on-watchdog setting, which will restart the service
15753 automatically if the service stops sending out watchdog keep
15754 alive messages (as configured with WatchdogSec=).
15756 * The getty generator (which is responsible for bringing up a
15757 getty on configured serial consoles) will no longer only
15758 start a getty on the primary kernel console but on all
15759 others, too. This makes the order in which console= is
15760 specified on the kernel command line less important.
15762 * libsystemd-logind gained a new sd_session_get_vt() call to
15763 retrieve the VT number of a session.
15765 * If the option "tries=0" is set for an entry of /etc/crypttab
15766 its passphrase is queried indefinitely instead of any
15767 maximum number of tries.
15769 * If a service with a configure PID file terminates its PID
15770 file will now be removed automatically if it still exists
15771 afterwards. This should put an end to stale PID files.
15773 * systemd-run will now also take relative binary path names
15774 for execution and no longer insists on absolute paths.
15776 * InaccessibleDirectories= and ReadOnlyDirectories= now take
15777 paths that are optionally prefixed with "-" to indicate that
15778 it should not be considered a failure if they do not exist.
15780 * journalctl -o (and similar commands) now understands a new
15781 output mode "short-precise", it is similar to "short" but
15782 shows timestamps with usec accuracy.
15784 * The option "discard" (as known from Debian) is now
15785 synonymous to "allow-discards" in /etc/crypttab. In fact,
15786 "discard" is preferred now (since it is easier to remember
15789 * Some licensing clean-ups were made, so that more code is now
15790 LGPL-2.1 licensed than before.
15792 * A minimal tool to save/restore the display backlight
15793 brightness across reboots has been added. It will store the
15794 backlight setting as late as possible at shutdown, and
15795 restore it as early as possible during reboot.
15797 * A logic to automatically discover and enable home and swap
15798 partitions on GPT disks has been added. With this in place
15799 /etc/fstab becomes optional for many setups as systemd can
15800 discover certain partitions located on the root disk
15801 automatically. Home partitions are recognized under their
15802 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
15803 partitions are recognized under their GPT type ID
15804 0657fd6da4ab43c484e50933c84b4f4f.
15806 * systemd will no longer pass any environment from the kernel
15807 or initrd to system services. If you want to set an
15808 environment for all services, do so via the kernel command
15809 line systemd.setenv= assignment.
15811 * The systemd-sysctl tool no longer natively reads the file
15812 /etc/sysctl.conf. If desired, the file should be symlinked
15813 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
15814 legacy support by a symlink rather than built-in code, it
15815 also makes the otherwise hidden order of application of the
15816 different files visible. (Note that this partly reverts to a
15817 pre-198 application order of sysctl knobs!)
15819 * The "systemctl set-log-level" and "systemctl dump" commands
15820 have been moved to systemd-analyze.
15822 * systemd-run learned the new --remain-after-exit switch,
15823 which causes the scope unit not to be cleaned up
15824 automatically after the process terminated.
15826 * tmpfiles learned a new --exclude-prefix= switch to exclude
15827 certain paths from operation.
15829 * journald will now automatically flush all messages to disk
15830 as soon as a message at the log level CRIT, ALERT or EMERG
15833 Contributions from: Andrew Cook, Brandon Philips, Christian
15834 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
15835 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
15836 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
15837 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
15838 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
15839 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
15840 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
15841 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
15842 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
15843 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
15844 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
15845 William Giokas, Zbigniew Jędrzejewski-Szmek
15847 — Berlin, 2013-09-13
15851 * The documentation has been updated to cover the various new
15852 concepts introduced with 205.
15854 * Unit files now understand the new %v specifier which
15855 resolves to the kernel version string as returned by "uname
15858 * systemctl now supports filtering the unit list output by
15859 load state, active state and sub state, using the new
15860 --state= parameter.
15862 * "systemctl status" will now show the results of the
15863 condition checks (like ConditionPathExists= and similar) of
15864 the last start attempts of the unit. They are also logged to
15867 * "journalctl -b" may now be used to look for boot output of a
15868 specific boot. Try "journalctl -b -1" for the previous boot,
15869 but the syntax is substantially more powerful.
15871 * "journalctl --show-cursor" has been added which prints the
15872 cursor string the last shown log line. This may then be used
15873 with the new "journalctl --after-cursor=" switch to continue
15874 browsing logs from that point on.
15876 * "journalctl --force" may now be used to force regeneration
15879 * Creation of "dead" device nodes has been moved from udev
15880 into kmod and tmpfiles. Previously, udev would read the kmod
15881 databases to pre-generate dead device nodes based on meta
15882 information contained in kernel modules, so that these would
15883 be auto-loaded on access rather then at boot. As this
15884 does not really have much to do with the exposing actual
15885 kernel devices to userspace this has always been slightly
15886 alien in the udev codebase. Following the new scheme kmod
15887 will now generate a runtime snippet for tmpfiles from the
15888 module meta information and it now is tmpfiles' job to the
15889 create the nodes. This also allows overriding access and
15890 other parameters for the nodes using the usual tmpfiles
15891 facilities. As side effect this allows us to remove the
15892 CAP_SYS_MKNOD capability bit from udevd entirely.
15894 * logind's device ACLs may now be applied to these "dead"
15895 devices nodes too, thus finally allowing managed access to
15896 devices such as /dev/snd/sequencer without loading the
15897 backing module right-away.
15899 * A new RPM macro has been added that may be used to apply
15900 tmpfiles configuration during package installation.
15902 * systemd-detect-virt and ConditionVirtualization= now can
15903 detect User-Mode-Linux machines (UML).
15905 * journald will now implicitly log the effective capabilities
15906 set of processes in the message metadata.
15908 * systemd-cryptsetup has gained support for TrueCrypt volumes.
15910 * The initrd interface has been simplified (more specifically,
15911 support for passing performance data via environment
15912 variables and fsck results via files in /run has been
15913 removed). These features were non-essential, and are
15914 nowadays available in a much nicer way by having systemd in
15915 the initrd serialize its state and have the hosts systemd
15916 deserialize it again.
15918 * The udev "keymap" data files and tools to apply keyboard
15919 specific mappings of scan to key codes, and force-release
15920 scan code lists have been entirely replaced by a udev
15921 "keyboard" builtin and a hwdb data file.
15923 * systemd will now honour the kernel's "quiet" command line
15924 argument also during late shutdown, resulting in a
15925 completely silent shutdown when used.
15927 * There's now an option to control the SO_REUSEPORT socket
15928 option in .socket units.
15930 * Instance units will now automatically get a per-template
15931 subslice of system.slice unless something else is explicitly
15932 configured. For example, instances of sshd@.service will now
15933 implicitly be placed in system-sshd.slice rather than
15934 system.slice as before.
15936 * Test coverage support may now be enabled at build time.
15938 Contributions from: Dave Reisner, Frederic Crozat, Harald
15939 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
15940 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
15941 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
15942 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
15943 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
15944 Giokas, Zbigniew Jędrzejewski-Szmek
15946 — Berlin, 2013-07-23
15950 * Two new unit types have been introduced:
15952 Scope units are very similar to service units, however, are
15953 created out of pre-existing processes — instead of PID 1
15954 forking off the processes. By using scope units it is
15955 possible for system services and applications to group their
15956 own child processes (worker processes) in a powerful way
15957 which then maybe used to organize them, or kill them
15958 together, or apply resource limits on them.
15960 Slice units may be used to partition system resources in an
15961 hierarchical fashion and then assign other units to them. By
15962 default there are now three slices: system.slice (for all
15963 system services), user.slice (for all user sessions),
15964 machine.slice (for VMs and containers).
15966 Slices and scopes have been introduced primarily in
15967 context of the work to move cgroup handling to a
15968 single-writer scheme, where only PID 1
15969 creates/removes/manages cgroups.
15971 * There's a new concept of "transient" units. In contrast to
15972 normal units these units are created via an API at runtime,
15973 not from configuration from disk. More specifically this
15974 means it is now possible to run arbitrary programs as
15975 independent services, with all execution parameters passed
15976 in via bus APIs rather than read from disk. Transient units
15977 make systemd substantially more dynamic then it ever was,
15978 and useful as a general batch manager.
15980 * logind has been updated to make use of scope and slice units
15981 for managing user sessions. As a user logs in he will get
15982 his own private slice unit, to which all sessions are added
15983 as scope units. We also added support for automatically
15984 adding an instance of user@.service for the user into the
15985 slice. Effectively logind will no longer create cgroup
15986 hierarchies on its own now, it will defer entirely to PID 1
15987 for this by means of scope, service and slice units. Since
15988 user sessions this way become entities managed by PID 1
15989 the output of "systemctl" is now a lot more comprehensive.
15991 * A new mini-daemon "systemd-machined" has been added which
15992 may be used by virtualization managers to register local
15993 VMs/containers. nspawn has been updated accordingly, and
15994 libvirt will be updated shortly. machined will collect a bit
15995 of meta information about the VMs/containers, and assign
15996 them their own scope unit (see above). The collected
15997 meta-data is then made available via the "machinectl" tool,
15998 and exposed in "ps" and similar tools. machined/machinectl
15999 is compile-time optional.
16001 * As discussed earlier, the low-level cgroup configuration
16002 options ControlGroup=, ControlGroupModify=,
16003 ControlGroupPersistent=, ControlGroupAttribute= have been
16004 removed. Please use high-level attribute settings instead as
16005 well as slice units.
16007 * A new bus call SetUnitProperties() has been added to alter
16008 various runtime parameters of a unit. This is primarily
16009 useful to alter cgroup parameters dynamically in a nice way,
16010 but will be extended later on to make more properties
16011 modifiable at runtime. systemctl gained a new set-properties
16012 command that wraps this call.
16014 * A new tool "systemd-run" has been added which can be used to
16015 run arbitrary command lines as transient services or scopes,
16016 while configuring a number of settings via the command
16017 line. This tool is currently very basic, however already
16018 very useful. We plan to extend this tool to even allow
16019 queuing of execution jobs with time triggers from the
16020 command line, similar in fashion to "at".
16022 * nspawn will now inform the user explicitly that kernels with
16023 audit enabled break containers, and suggest the user to turn
16026 * Support for detecting the IMA and AppArmor security
16027 frameworks with ConditionSecurity= has been added.
16029 * journalctl gained a new "-k" switch for showing only kernel
16030 messages, mimicking dmesg output; in addition to "--user"
16031 and "--system" switches for showing only user's own logs
16034 * systemd-delta can now show information about drop-in
16035 snippets extending unit files.
16037 * libsystemd-bus has been substantially updated but is still
16038 not available as public API.
16040 * systemd will now look for the "debug" argument on the kernel
16041 command line and enable debug logging, similar to what
16042 "systemd.log_level=debug" already did before.
16044 * "systemctl set-default", "systemctl get-default" has been
16045 added to configure the default.target symlink, which
16046 controls what to boot into by default.
16048 * "systemctl set-log-level" has been added as a convenient
16049 way to raise and lower systemd logging threshold.
16051 * "systemd-analyze plot" will now show the time the various
16052 generators needed for execution, as well as information
16053 about the unit file loading.
16055 * libsystemd-journal gained a new sd_journal_open_files() call
16056 for opening specific journal files. journactl also gained a
16057 new switch to expose this new functionality. Previously we
16058 only supported opening all files from a directory, or all
16059 files from the system, as opening individual files only is
16060 racy due to journal file rotation.
16062 * systemd gained the new DefaultEnvironment= setting in
16063 /etc/systemd/system.conf to set environment variables for
16066 * If a privileged process logs a journal message with the
16067 OBJECT_PID= field set, then journald will automatically
16068 augment this with additional OBJECT_UID=, OBJECT_GID=,
16069 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
16070 system services want to log events about specific client
16071 processes. journactl/systemctl has been updated to make use
16072 of this information if all log messages regarding a specific
16075 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
16076 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
16077 Reisner, David Coppa, David King, David Strauss, Eelco
16078 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
16079 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
16080 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
16081 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
16082 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
16083 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
16084 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
16085 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
16086 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
16087 Łukasz Stelmach, 장동준
16091 * The Python bindings gained some minimal support for the APIs
16092 exposed by libsystemd-logind.
16094 * ConditionSecurity= gained support for detecting SMACK. Since
16095 this condition already supports SELinux and AppArmor we only
16096 miss IMA for this. Patches welcome!
16098 Contributions from: Karol Lewandowski, Lennart Poettering,
16099 Zbigniew Jędrzejewski-Szmek
16103 * systemd-nspawn will now create /etc/resolv.conf if
16104 necessary, before bind-mounting the host's file onto it.
16106 * systemd-nspawn will now store meta information about a
16107 container on the container's cgroup as extended attribute
16108 fields, including the root directory.
16110 * The cgroup hierarchy has been reworked in many ways. All
16111 objects any of the components systemd creates in the cgroup
16112 tree are now suffixed. More specifically, user sessions are
16113 now placed in cgroups suffixed with ".session", users in
16114 cgroups suffixed with ".user", and nspawn containers in
16115 cgroups suffixed with ".nspawn". Furthermore, all cgroup
16116 names are now escaped in a simple scheme to avoid collision
16117 of userspace object names with kernel filenames. This work
16118 is preparation for making these objects relocatable in the
16119 cgroup tree, in order to allow easy resource partitioning of
16120 these objects without causing naming conflicts.
16122 * systemctl list-dependencies gained the new switches
16123 --plain, --reverse, --after and --before.
16125 * systemd-inhibit now shows the process name of processes that
16126 have taken an inhibitor lock.
16128 * nss-myhostname will now also resolve "localhost"
16129 implicitly. This makes /etc/hosts an optional file and
16130 nicely handles that on IPv6 ::1 maps to both "localhost" and
16131 the local hostname.
16133 * libsystemd-logind.so gained a new call
16134 sd_get_machine_names() to enumerate running containers and
16135 VMs (currently only supported by very new libvirt and
16136 nspawn). sd_login_monitor can now be used to watch
16137 VMs/containers coming and going.
16139 * .include is not allowed recursively anymore, and only in
16140 unit files. Usually it is better to use drop-in snippets in
16141 .d/*.conf anyway, as introduced with systemd 198.
16143 * systemd-analyze gained a new "critical-chain" command that
16144 determines the slowest chain of units run during system
16145 boot-up. It is very useful for tracking down where
16146 optimizing boot time is the most beneficial.
16148 * systemd will no longer allow manipulating service paths in
16149 the name=systemd:/system cgroup tree using ControlGroup= in
16150 units. (But is still fine with it in all other dirs.)
16152 * There's a new systemd-nspawn@.service service file that may
16153 be used to easily run nspawn containers as system
16154 services. With the container's root directory in
16155 /var/lib/container/foobar it is now sufficient to run
16156 "systemctl start systemd-nspawn@foobar.service" to boot it.
16158 * systemd-cgls gained a new parameter "--machine" to list only
16159 the processes within a certain container.
16161 * ConditionSecurity= now can check for "apparmor". We still
16162 are lacking checks for SMACK and IMA for this condition
16163 check though. Patches welcome!
16165 * A new configuration file /etc/systemd/sleep.conf has been
16166 added that may be used to configure which kernel operation
16167 systemd is supposed to execute when "suspend", "hibernate"
16168 or "hybrid-sleep" is requested. This makes the new kernel
16169 "freeze" state accessible to the user.
16171 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
16172 the passed argument if applicable.
16174 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
16175 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
16176 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
16177 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
16178 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
16179 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
16180 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
16185 * The output of 'systemctl list-jobs' got some polishing. The
16186 '--type=' argument may now be passed more than once. A new
16187 command 'systemctl list-sockets' has been added which shows
16188 a list of kernel sockets systemd is listening on with the
16189 socket units they belong to, plus the units these socket
16192 * The experimental libsystemd-bus library got substantial
16193 updates to work in conjunction with the (also experimental)
16194 kdbus kernel project. It works well enough to exchange
16195 messages with some sophistication. Note that kdbus is not
16196 ready yet, and the library is mostly an elaborate test case
16197 for now, and not installable.
16199 * systemd gained a new unit 'systemd-static-nodes.service'
16200 that generates static device nodes earlier during boot, and
16201 can run in conjunction with udev.
16203 * libsystemd-login gained a new call sd_pid_get_user_unit()
16204 to retrieve the user systemd unit a process is running
16205 in. This is useful for systems where systemd is used as
16208 * systemd-nspawn now places all containers in the new /machine
16209 top-level cgroup directory in the name=systemd
16210 hierarchy. libvirt will soon do the same, so that we get a
16211 uniform separation of /system, /user and /machine for system
16212 services, user processes and containers/virtual
16213 machines. This new cgroup hierarchy is also useful to stick
16214 stable names to specific container instances, which can be
16215 recognized later this way (this name may be controlled
16216 via systemd-nspawn's new -M switch). libsystemd-login also
16217 gained a new call sd_pid_get_machine_name() to retrieve the
16218 name of the container/VM a specific process belongs to.
16220 * bootchart can now store its data in the journal.
16222 * libsystemd-journal gained a new call
16223 sd_journal_add_conjunction() for AND expressions to the
16224 matching logic. This can be used to express more complex
16225 logical expressions.
16227 * journactl can now take multiple --unit= and --user-unit=
16230 * The cryptsetup logic now understands the "luks.key=" kernel
16231 command line switch for specifying a file to read the
16232 decryption key from. Also, if a configured key file is not
16233 found the tool will now automatically fall back to prompting
16236 * Python systemd.journal module was updated to wrap recently
16237 added functions from libsystemd-journal. The interface was
16238 changed to bring the low level interface in s.j._Reader
16239 closer to the C API, and the high level interface in
16240 s.j.Reader was updated to wrap and convert all data about
16243 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
16244 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
16245 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
16246 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
16247 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
16248 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
16252 * journalctl --update-catalog now understands a new --root=
16253 option to operate on catalogs found in a different root
16256 * During shutdown after systemd has terminated all running
16257 services a final killing loop kills all remaining left-over
16258 processes. We will now print the name of these processes
16259 when we send SIGKILL to them, since this usually indicates a
16262 * If /etc/crypttab refers to password files stored on
16263 configured mount points automatic dependencies will now be
16264 generated to ensure the specific mount is established first
16265 before the key file is attempted to be read.
16267 * 'systemctl status' will now show information about the
16268 network sockets a socket unit is listening on.
16270 * 'systemctl status' will also shown information about any
16271 drop-in configuration file for units. (Drop-In configuration
16272 files in this context are files such as
16273 /etc/systemd/system/foobar.service.d/*.conf)
16275 * systemd-cgtop now optionally shows summed up CPU times of
16276 cgroups. Press '%' while running cgtop to switch between
16277 percentage and absolute mode. This is useful to determine
16278 which cgroups use up the most CPU time over the entire
16279 runtime of the system. systemd-cgtop has also been updated
16280 to be 'pipeable' for processing with further shell tools.
16282 * 'hostnamectl set-hostname' will now allow setting of FQDN
16285 * The formatting and parsing of time span values has been
16286 changed. The parser now understands fractional expressions
16287 such as "5.5h". The formatter will now output fractional
16288 expressions for all time spans under 1min, i.e. "5.123456s"
16289 rather than "5s 123ms 456us". For time spans under 1s
16290 millisecond values are shown, for those under 1ms
16291 microsecond values are shown. This should greatly improve
16292 all time-related output of systemd.
16294 * libsystemd-login and libsystemd-journal gained new
16295 functions for querying the poll() events mask and poll()
16296 timeout value for integration into arbitrary event
16299 * localectl gained the ability to list available X11 keymaps
16300 (models, layouts, variants, options).
16302 * 'systemd-analyze dot' gained the ability to filter for
16303 specific units via shell-style globs, to create smaller,
16304 more useful graphs. I.e. it is now possible to create simple
16305 graphs of all the dependencies between only target units, or
16306 of all units that Avahi has dependencies with.
16308 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
16309 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
16310 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
16311 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
16312 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
16313 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
16314 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
16318 * The boot-time readahead implementation for rotating media
16319 will now read the read-ahead data in multiple passes which
16320 consist of all read requests made in equidistant time
16321 intervals. This means instead of strictly reading read-ahead
16322 data in its physical order on disk we now try to find a
16323 middle ground between physical and access time order.
16325 * /etc/os-release files gained a new BUILD_ID= field for usage
16326 on operating systems that provide continuous builds of OS
16329 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
16330 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
16331 William Douglas, Zbigniew Jędrzejewski-Szmek
16335 * systemd-python gained an API exposing libsystemd-daemon.
16337 * The SMACK setup logic gained support for uploading CIPSO
16340 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
16341 ReadOnlyDirectories= and InaccessibleDirectories= has
16342 changed. The private /tmp and /var/tmp directories are now
16343 shared by all processes of a service (which means
16344 ExecStartPre= may now leave data in /tmp that ExecStart= of
16345 the same service can still access). When a service is
16346 stopped its temporary directories are immediately deleted
16347 (normal clean-up with tmpfiles is still done in addition to
16350 * By default, systemd will now set a couple of sysctl
16351 variables in the kernel: the safe sysrq options are turned
16352 on, IP route verification is turned on, and source routing
16353 disabled. The recently added hardlink and softlink
16354 protection of the kernel is turned on. These settings should
16355 be reasonably safe, and good defaults for all new systems.
16357 * The predictable network naming logic may now be turned off
16358 with a new kernel command line switch: net.ifnames=0.
16360 * A new libsystemd-bus module has been added that implements a
16361 pretty complete D-Bus client library. For details see:
16363 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
16365 * journald will now explicitly flush the journal files to disk
16366 at the latest 5min after each write. The file will then also
16367 be marked offline until the next write. This should increase
16368 reliability in case of a crash. The synchronization delay
16369 can be configured via SyncIntervalSec= in journald.conf.
16371 * There's a new remote-fs-setup.target unit that can be used
16372 to pull in specific services when at least one remote file
16373 system is to be mounted.
16375 * There are new targets timers.target and paths.target as
16376 canonical targets to pull user timer and path units in
16377 from. This complements sockets.target with a similar
16378 purpose for socket units.
16380 * libudev gained a new call udev_device_set_attribute_value()
16381 to set sysfs attributes of a device.
16383 * The udev daemon now sets the default number of worker
16384 processes executed in parallel based on the number of available
16385 CPUs instead of the amount of available RAM. This is supposed
16386 to provide a more reliable default and limit a too aggressive
16387 parallelism for setups with 1000s of devices connected.
16389 Contributions from: Auke Kok, Colin Walters, Cristian
16390 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
16391 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
16392 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
16393 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
16394 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
16395 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
16396 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
16397 Zbigniew Jędrzejewski-Szmek
16401 * Configuration of unit files may now be extended via drop-in
16402 files without having to edit/override the unit files
16403 themselves. More specifically, if the administrator wants to
16404 change one value for a service file foobar.service he can
16405 now do so by dropping in a configuration snippet into
16406 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
16407 will load all these snippets and apply them on top of the
16408 main unit configuration file, possibly extending or
16409 overriding its settings. Using these drop-in snippets is
16410 generally nicer than the two earlier options for changing
16411 unit files locally: copying the files from
16412 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
16413 them there; or creating a new file in /etc/systemd/system/
16414 that incorporates the original one via ".include". Drop-in
16415 snippets into these .d/ directories can be placed in any
16416 directory systemd looks for units in, and the usual
16417 overriding semantics between /usr/lib, /etc and /run apply
16420 * Most unit file settings which take lists of items can now be
16421 reset by assigning the empty string to them. For example,
16422 normally, settings such as Environment=FOO=BAR append a new
16423 environment variable assignment to the environment block,
16424 each time they are used. By assigning Environment= the empty
16425 string the environment block can be reset to empty. This is
16426 particularly useful with the .d/*.conf drop-in snippets
16427 mentioned above, since this adds the ability to reset list
16428 settings from vendor unit files via these drop-ins.
16430 * systemctl gained a new "list-dependencies" command for
16431 listing the dependencies of a unit recursively.
16433 * Inhibitors are now honored and listed by "systemctl
16434 suspend", "systemctl poweroff" (and similar) too, not only
16435 GNOME. These commands will also list active sessions by
16438 * Resource limits (as exposed by the various control group
16439 controllers) can now be controlled dynamically at runtime
16440 for all units. More specifically, you can now use a command
16441 like "systemctl set-cgroup-attr foobar.service cpu.shares
16442 2000" to alter the CPU shares a specific service gets. These
16443 settings are stored persistently on disk, and thus allow the
16444 administrator to easily adjust the resource usage of
16445 services with a few simple commands. This dynamic resource
16446 management logic is also available to other programs via the
16447 bus. Almost any kernel cgroup attribute and controller is
16450 * systemd-vconsole-setup will now copy all font settings to
16451 all allocated VTs, where it previously applied them only to
16454 * libsystemd-login gained the new sd_session_get_tty() API
16457 * This release drops support for a few legacy or
16458 distribution-specific LSB facility names when parsing init
16459 scripts: $x-display-manager, $mail-transfer-agent,
16460 $mail-transport-agent, $mail-transfer-agent, $smtp,
16461 $null. Also, the mail-transfer-agent.target unit backing
16462 this has been removed. Distributions which want to retain
16463 compatibility with this should carry the burden for
16464 supporting this themselves and patch support for these back
16465 in, if they really need to. Also, the facilities $syslog and
16466 $local_fs are now ignored, since systemd does not support
16467 early-boot LSB init scripts anymore, and these facilities
16468 are implied anyway for normal services. syslog.target has
16471 * There are new bus calls on PID1's Manager object for
16472 cancelling jobs, and removing snapshot units. Previously,
16473 both calls were only available on the Job and Snapshot
16474 objects themselves.
16476 * systemd-journal-gatewayd gained SSL support.
16478 * The various "environment" files, such as /etc/locale.conf
16479 now support continuation lines with a backslash ("\") as
16480 last character in the line, similarly in style (but different)
16481 to how this is supported in shells.
16483 * For normal user processes the _SYSTEMD_USER_UNIT= field is
16484 now implicitly appended to every log entry logged. systemctl
16485 has been updated to filter by this field when operating on a
16486 user systemd instance.
16488 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
16489 CAP_AUDIT_CONTROL capabilities to the capabilities set for
16490 the container. This makes it easier to boot unmodified
16491 Fedora systems in a container, which however still requires
16492 audit=0 to be passed on the kernel command line. Auditing in
16493 kernel and userspace is unfortunately still too broken in
16494 context of containers, hence we recommend compiling it out
16495 of the kernel or using audit=0. Hopefully this will be fixed
16496 one day for good in the kernel.
16498 * nspawn gained the new --bind= and --bind-ro= parameters to
16499 bind mount specific directories from the host into the
16502 * nspawn will now mount its own devpts file system instance
16503 into the container, in order not to leak pty devices from
16504 the host into the container.
16506 * systemd will now read the firmware boot time performance
16507 information from the EFI variables, if the used boot loader
16508 supports this, and takes it into account for boot performance
16509 analysis via "systemd-analyze". This is currently supported
16510 only in conjunction with Gummiboot, but could be supported
16511 by other boot loaders too. For details see:
16513 https://systemd.io/BOOT_LOADER_INTERFACE
16515 * A new generator has been added that automatically mounts the
16516 EFI System Partition (ESP) to /boot, if that directory
16517 exists, is empty, and no other file system has been
16518 configured to be mounted there.
16520 * logind will now send out PrepareForSleep(false) out
16521 unconditionally, after coming back from suspend. This may be
16522 used by applications as asynchronous notification for
16523 system resume events.
16525 * "systemctl unlock-sessions" has been added, that allows
16526 unlocking the screens of all user sessions at once, similar
16527 to how "systemctl lock-sessions" already locked all users
16528 sessions. This is backed by a new D-Bus call UnlockSessions().
16530 * "loginctl seat-status" will now show the master device of a
16531 seat. (i.e. the device of a seat that needs to be around for
16532 the seat to be considered available, usually the graphics
16535 * tmpfiles gained a new "X" line type, that allows
16536 configuration of files and directories (with wildcards) that
16537 shall be excluded from automatic cleanup ("aging").
16539 * udev default rules set the device node permissions now only
16540 at "add" events, and do not change them any longer with a
16541 later "change" event.
16543 * The log messages for lid events and power/sleep keypresses
16544 now carry a message ID.
16546 * We now have a substantially larger unit test suite, but this
16547 continues to be work in progress.
16549 * udevadm hwdb gained a new --root= parameter to change the
16550 root directory to operate relative to.
16552 * logind will now issue a background sync() request to the kernel
16553 early at shutdown, so that dirty buffers are flushed to disk early
16554 instead of at the last moment, in order to optimize shutdown
16557 * A new bootctl tool has been added that is an interface for
16558 certain boot loader operations. This is currently a preview
16559 and is likely to be extended into a small mechanism daemon
16560 like timedated, localed, hostnamed, and can be used by
16561 graphical UIs to enumerate available boot options, and
16562 request boot into firmware operations.
16564 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
16565 the rest of the package. It also has been updated to work
16566 correctly in initrds.
16568 * polkit previously has been runtime optional, and is now also
16569 compile time optional via a configure switch.
16571 * systemd-analyze has been reimplemented in C. Also "systemctl
16572 dot" has moved into systemd-analyze.
16574 * "systemctl status" with no further parameters will now print
16575 the status of all active or failed units.
16577 * Operations such as "systemctl start" can now be executed
16578 with a new mode "--irreversible" which may be used to queue
16579 operations that cannot accidentally be reversed by a later
16580 job queuing. This is by default used to make shutdown
16581 requests more robust.
16583 * The Python API of systemd now gained a new module for
16584 reading journal files.
16586 * A new tool kernel-install has been added that can install
16587 kernel images according to the Boot Loader Specification:
16589 https://systemd.io/BOOT_LOADER_SPECIFICATION
16591 * Boot time console output has been improved to provide
16592 animated boot time output for hanging jobs.
16594 * A new tool systemd-activate has been added which can be used
16595 to test socket activation with, directly from the command
16596 line. This should make it much easier to test and debug
16597 socket activation in daemons.
16599 * journalctl gained a new "--reverse" (or -r) option to show
16600 journal output in reverse order (i.e. newest line first).
16602 * journalctl gained a new "--pager-end" (or -e) option to jump
16603 to immediately jump to the end of the journal in the
16604 pager. This is only supported in conjunction with "less".
16606 * journalctl gained a new "--user-unit=" option, that works
16607 similarly to "--unit=" but filters for user units rather than
16610 * A number of unit files to ease adoption of systemd in
16611 initrds has been added. This moves some minimal logic from
16612 the various initrd implementations into systemd proper.
16614 * The journal files are now owned by a new group
16615 "systemd-journal", which exists specifically to allow access
16616 to the journal, and nothing else. Previously, we used the
16617 "adm" group for that, which however possibly covers more
16618 than just journal/log file access. This new group is now
16619 already used by systemd-journal-gatewayd to ensure this
16620 daemon gets access to the journal files and as little else
16621 as possible. Note that "make install" will also set FS ACLs
16622 up for /var/log/journal to give "adm" and "wheel" read
16623 access to it, in addition to "systemd-journal" which owns
16624 the journal files. We recommend that packaging scripts also
16625 add read access to "adm" + "wheel" to /var/log/journal, and
16626 all existing/future journal files. To normal users and
16627 administrators little changes, however packagers need to
16628 ensure to create the "systemd-journal" system group at
16629 package installation time.
16631 * The systemd-journal-gatewayd now runs as unprivileged user
16632 systemd-journal-gateway:systemd-journal-gateway. Packaging
16633 scripts need to create these system user/group at
16636 * timedated now exposes a new boolean property CanNTP that
16637 indicates whether a local NTP service is available or not.
16639 * systemd-detect-virt will now also detect xen PVs
16641 * The pstore file system is now mounted by default, if it is
16644 * In addition to the SELinux and IMA policies we will now also
16645 load SMACK policies at early boot.
16647 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
16648 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
16649 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
16650 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
16651 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
16652 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
16653 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
16654 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
16655 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
16656 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
16657 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
16658 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
16659 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
16660 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
16664 * Timer units now support calendar time events in addition to
16665 monotonic time events. That means you can now trigger a unit
16666 based on a calendar time specification such as "Thu,Fri
16667 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
16668 or fifth day of any month of the year 2013, given that it is
16669 a Thursday or a Friday. This brings timer event support
16670 considerably closer to cron's capabilities. For details on
16671 the supported calendar time specification language see
16674 * udev now supports a number of different naming policies for
16675 network interfaces for predictable names, and a combination
16676 of these policies is now the default. Please see this wiki
16677 document for details:
16679 https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html
16681 * Auke Kok's bootchart implementation has been added to the
16682 systemd tree. It is an optional component that can graph the
16683 boot in quite some detail. It is one of the best bootchart
16684 implementations around and minimal in its code and
16687 * nss-myhostname has been integrated into the systemd source
16688 tree. nss-myhostname guarantees that the local hostname
16689 always stays resolvable via NSS. It has been a weak
16690 requirement of systemd-hostnamed since a long time, and
16691 since its code is actually trivial we decided to just
16692 include it in systemd's source tree. It can be turned off
16693 with a configure switch.
16695 * The read-ahead logic is now capable of properly detecting
16696 whether a btrfs file system is on SSD or rotating media, in
16697 order to optimize the read-ahead scheme. Previously, it was
16698 only capable of detecting this on traditional file systems
16701 * In udev, additional device properties are now read from the
16702 IAB in addition to the OUI database. Also, Bluetooth company
16703 identities are attached to the devices as well.
16705 * In service files %U may be used as specifier that is
16706 replaced by the configured user name of the service.
16708 * nspawn may now be invoked without a controlling TTY. This
16709 makes it suitable for invocation as its own service. This
16710 may be used to set up a simple containerized server system
16711 using only core OS tools.
16713 * systemd and nspawn can now accept socket file descriptors
16714 when they are started for socket activation. This enables
16715 implementation of socket activated nspawn
16716 containers. i.e. think about autospawning an entire OS image
16717 when the first SSH or HTTP connection is received. We expect
16718 that similar functionality will also be added to libvirt-lxc
16721 * journalctl will now suppress ANSI color codes when
16722 presenting log data.
16724 * systemctl will no longer show control group information for
16725 a unit if the control group is empty anyway.
16727 * logind can now automatically suspend/hibernate/shutdown the
16730 * /etc/machine-info and hostnamed now also expose the chassis
16731 type of the system. This can be used to determine whether
16732 the local system is a laptop, desktop, handset or
16733 tablet. This information may either be configured by the
16734 user/vendor or is automatically determined from ACPI and DMI
16735 information if possible.
16737 * A number of polkit actions are now bound together with "imply"
16738 rules. This should simplify creating UIs because many actions
16739 will now authenticate similar ones as well.
16741 * Unit files learnt a new condition ConditionACPower= which
16742 may be used to conditionalize a unit depending on whether an
16743 AC power source is connected or not, of whether the system
16744 is running on battery power.
16746 * systemctl gained a new "is-failed" verb that may be used in
16747 shell scripts and suchlike to check whether a specific unit
16748 is in the "failed" state.
16750 * The EnvironmentFile= setting in unit files now supports file
16751 globbing, and can hence be used to easily read a number of
16752 environment files at once.
16754 * systemd will no longer detect and recognize specific
16755 distributions. All distribution-specific #ifdeffery has been
16756 removed, systemd is now fully generic and
16757 distribution-agnostic. Effectively, not too much is lost as
16758 a lot of the code is still accessible via explicit configure
16759 switches. However, support for some distribution specific
16760 legacy configuration file formats has been dropped. We
16761 recommend distributions to simply adopt the configuration
16762 files everybody else uses now and convert the old
16763 configuration from packaging scripts. Most distributions
16764 already did that. If that's not possible or desirable,
16765 distributions are welcome to forward port the specific
16766 pieces of code locally from the git history.
16768 * When logging a message about a unit systemd will now always
16769 log the unit name in the message meta data.
16771 * localectl will now also discover system locale data that is
16772 not stored in locale archives, but directly unpacked.
16774 * logind will no longer unconditionally use framebuffer
16775 devices as seat masters, i.e. as devices that are required
16776 to be existing before a seat is considered preset. Instead,
16777 it will now look for all devices that are tagged as
16778 "seat-master" in udev. By default, framebuffer devices will
16779 be marked as such, but depending on local systems, other
16780 devices might be marked as well. This may be used to
16781 integrate graphics cards using closed source drivers (such
16782 as NVidia ones) more nicely into logind. Note however, that
16783 we recommend using the open source NVidia drivers instead,
16784 and no udev rules for the closed-source drivers will be
16785 shipped from us upstream.
16787 Contributions from: Adam Williamson, Alessandro Crismani, Auke
16788 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
16789 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
16790 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
16791 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
16792 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
16793 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
16794 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
16795 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
16796 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
16797 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
16798 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
16803 * udev gained support for loading additional device properties
16804 from an indexed database that is keyed by vendor/product IDs
16805 and similar device identifiers. For the beginning this
16806 "hwdb" is populated with data from the well-known PCI and
16807 USB database, but also includes PNP, ACPI and OID data. In
16808 the longer run this indexed database shall grow into
16809 becoming the one central database for non-essential
16810 userspace device metadata. Previously, data from the PCI/USB
16811 database was only attached to select devices, since the
16812 lookup was a relatively expensive operation due to O(n) time
16813 complexity (with n being the number of entries in the
16814 database). Since this is now O(1), we decided to add in this
16815 data for all devices where this is available, by
16816 default. Note that the indexed database needs to be rebuilt
16817 when new data files are installed. To achieve this you need
16818 to update your packaging scripts to invoke "udevadm hwdb
16819 --update" after installation of hwdb data files. For
16820 RPM-based distributions we introduced the new
16821 %udev_hwdb_update macro for this purpose.
16823 * The Journal gained support for the "Message Catalog", an
16824 indexed database to link up additional information with
16825 journal entries. For further details please check:
16827 https://www.freedesktop.org/wiki/Software/systemd/catalog
16829 The indexed message catalog database also needs to be
16830 rebuilt after installation of message catalog files. Use
16831 "journalctl --update-catalog" for this. For RPM-based
16832 distributions we introduced the %journal_catalog_update
16833 macro for this purpose.
16835 * The Python Journal bindings gained support for the standard
16836 Python logging framework.
16838 * The Journal API gained new functions for checking whether
16839 the underlying file system of a journal file is capable of
16840 properly reporting file change notifications, or whether
16841 applications that want to reflect journal changes "live"
16842 need to recheck journal files continuously in appropriate
16845 * It is now possible to set the "age" field for tmpfiles
16846 entries to 0, indicating that files matching this entry
16847 shall always be removed when the directories are cleaned up.
16849 * coredumpctl gained a new "gdb" verb which invokes gdb
16850 right-away on the selected coredump.
16852 * There's now support for "hybrid sleep" on kernels that
16853 support this, in addition to "suspend" and "hibernate". Use
16854 "systemctl hybrid-sleep" to make use of this.
16856 * logind's HandleSuspendKey= setting (and related settings)
16857 now gained support for a new "lock" setting to simply
16858 request the screen lock on all local sessions, instead of
16859 actually executing a suspend or hibernation.
16861 * systemd will now mount the EFI variables file system by
16864 * Socket units now gained support for configuration of the
16865 SMACK security label.
16867 * timedatectl will now output the time of the last and next
16868 daylight saving change.
16870 * We dropped support for various legacy and distro-specific
16871 concepts, such as insserv, early-boot SysV services
16872 (i.e. those for non-standard runlevels such as 'b' or 'S')
16873 or ArchLinux /etc/rc.conf support. We recommend the
16874 distributions who still need support this to either continue
16875 to maintain the necessary patches downstream, or find a
16876 different solution. (Talk to us if you have questions!)
16878 * Various systemd components will now bypass polkit checks for
16879 root and otherwise handle properly if polkit is not found to
16880 be around. This should fix most issues for polkit-less
16881 systems. Quite frankly this should have been this way since
16882 day one. It is absolutely our intention to make systemd work
16883 fine on polkit-less systems, and we consider it a bug if
16884 something does not work as it should if polkit is not around.
16886 * For embedded systems it is now possible to build udev and
16887 systemd without blkid and/or kmod support.
16889 * "systemctl switch-root" is now capable of switching root
16890 more than once. I.e. in addition to transitions from the
16891 initrd to the host OS it is now possible to transition to
16892 further OS images from the host. This is useful to implement
16893 offline updating tools.
16895 * Various other additions have been made to the RPM macros
16896 shipped with systemd. Use %udev_rules_update() after
16897 installing new udev rules files. %_udevhwdbdir,
16898 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
16899 %_sysctldir are now available which resolve to the right
16900 directories for packages to place various data files in.
16902 * journalctl gained the new --full switch (in addition to
16903 --all, to disable ellipsation for long messages.
16905 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
16906 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
16907 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
16908 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
16909 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
16910 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
16911 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
16912 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
16913 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
16917 * journalctl gained new --since= and --until= switches to
16918 filter by time. It also now supports nice filtering for
16919 units via --unit=/-u.
16921 * Type=oneshot services may use ExecReload= and do the
16924 * The journal daemon now supports time-based rotation and
16925 vacuuming, in addition to the usual disk-space based
16928 * The journal will now index the available field values for
16929 each field name. This enables clients to show pretty drop
16930 downs of available match values when filtering. The bash
16931 completion of journalctl has been updated
16932 accordingly. journalctl gained a new switch -F to list all
16933 values a certain field takes in the journal database.
16935 * More service events are now written as structured messages
16936 to the journal, and made recognizable via message IDs.
16938 * The timedated, localed and hostnamed mini-services which
16939 previously only provided support for changing time, locale
16940 and hostname settings from graphical DEs such as GNOME now
16941 also have a minimal (but very useful) text-based client
16942 utility each. This is probably the nicest way to changing
16943 these settings from the command line now, especially since
16944 it lists available options and is fully integrated with bash
16947 * There's now a new tool "systemd-coredumpctl" to list and
16948 extract coredumps from the journal.
16950 * We now install a README each in /var/log/ and
16951 /etc/rc.d/init.d explaining where the system logs and init
16952 scripts went. This hopefully should help folks who go to
16953 that dirs and look into the otherwise now empty void and
16954 scratch their heads.
16956 * When user-services are invoked (by systemd --user) the
16957 $MANAGERPID env var is set to the PID of systemd.
16959 * SIGRTMIN+24 when sent to a --user instance will now result
16960 in immediate termination of systemd.
16962 * gatewayd received numerous feature additions such as a
16963 "follow" mode, for live syncing and filtering.
16965 * browse.html now allows filtering and showing detailed
16966 information on specific entries. Keyboard navigation and
16967 mouse screen support has been added.
16969 * gatewayd/journalctl now supports HTML5/JSON
16970 Server-Sent-Events as output.
16972 * The SysV init script compatibility logic will now
16973 heuristically determine whether a script supports the
16974 "reload" verb, and only then make this available as
16975 "systemctl reload".
16977 * "systemctl status --follow" has been removed, use "journalctl
16980 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
16981 have been removed since they are hardly useful to be
16984 * And I'd like to take the opportunity to specifically mention
16985 Zbigniew for his great contributions. Zbigniew, you rock!
16987 Contributions from: Andrew Eikum, Christian Hesse, Colin
16988 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
16989 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
16990 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
16991 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
16992 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
16993 Jędrzejewski-Szmek, Сковорода Никита Андреевич
16997 * If /etc/vconsole.conf is non-existent or empty we will no
16998 longer load any console font or key map at boot by
16999 default. Instead the kernel defaults will be left
17000 intact. This is definitely the right thing to do, as no
17001 configuration should mean no configuration, and hard-coding
17002 font names that are different on all archs is probably a bad
17003 idea. Also, the kernel default key map and font should be
17004 good enough for most cases anyway, and mostly identical to
17005 the userspace fonts/key maps we previously overloaded them
17006 with. If distributions want to continue to default to a
17007 non-kernel font or key map they should ship a default
17008 /etc/vconsole.conf with the appropriate contents.
17010 Contributions from: Colin Walters, Daniel J Walsh, Dave
17011 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
17012 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
17016 * journalctl gained a new --cursor= switch to show entries
17017 starting from the specified location in the journal.
17019 * We now enforce a size limit on journal entry fields exported
17020 with "-o json" in journalctl. Fields larger than 4K will be
17021 assigned null. This can be turned off with --all.
17023 * An (optional) journal gateway daemon is now available as
17024 "systemd-journal-gatewayd.service". This service provides
17025 access to the journal via HTTP and JSON. This functionality
17026 will be used to implement live log synchronization in both
17027 pull and push modes, but has various other users too, such
17028 as easy log access for debugging of embedded devices. Right
17029 now it is already useful to retrieve the journal via HTTP:
17031 # systemctl start systemd-journal-gatewayd.service
17032 # wget http://localhost:19531/entries
17034 This will download the journal contents in a
17035 /var/log/messages compatible format. The same as JSON:
17037 # curl -H"Accept: application/json" http://localhost:19531/entries
17039 This service is also accessible via a web browser where a
17040 single static HTML5 app is served that uses the JSON logic
17041 to enable the user to do some basic browsing of the
17042 journal. This will be extended later on. Here's an example
17043 screenshot of this app in its current state:
17045 https://0pointer.de/public/journal-gatewayd
17047 Contributions from: Kay Sievers, Lennart Poettering, Robert
17048 Milasan, Tom Gundersen
17052 * The bash completion logic is now available for journalctl
17055 * We do not mount the "cpuset" controller anymore together with
17056 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
17057 started if no parameters are assigned to it. "cpuset" hence
17058 broke code that assumed it could create "cpu" groups and
17061 * journalctl -f will now subscribe to terminal size changes,
17062 and line break accordingly.
17064 Contributions from: Dave Reisner, Kay Sievers, Lennart
17065 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
17069 * nspawn will now create a symlink /etc/localtime in the
17070 container environment, copying the host's timezone
17071 setting. Previously this has been done via a bind mount, but
17072 since symlinks cannot be bind mounted this has now been
17073 changed to create/update the appropriate symlink.
17075 * journalctl -n's line number argument is now optional, and
17076 will default to 10 if omitted.
17078 * journald will now log the maximum size the journal files may
17079 take up on disk. This is particularly useful if the default
17080 built-in logic of determining this parameter from the file
17081 system size is used. Use "systemctl status
17082 systemd-journald.service" to see this information.
17084 * The multi-seat X wrapper tool has been stripped down. As X
17085 is now capable of enumerating graphics devices via udev in a
17086 seat-aware way the wrapper is not strictly necessary
17087 anymore. A stripped down temporary stop-gap is still shipped
17088 until the upstream display managers have been updated to
17089 fully support the new X logic. Expect this wrapper to be
17090 removed entirely in one of the next releases.
17092 * HandleSleepKey= in logind.conf has been split up into
17093 HandleSuspendKey= and HandleHibernateKey=. The old setting
17094 is not available anymore. X11 and the kernel are
17095 distinguishing between these keys and we should too. This
17096 also means the inhibition lock for these keys has been split
17099 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
17100 Poettering, Lukas Nykryn, Václav Pavlín
17104 * Whenever a unit changes state we will now log this to the
17105 journal and show along the unit's own log output in
17106 "systemctl status".
17108 * ConditionPathIsMountPoint= can now properly detect bind
17109 mount points too. (Previously, a bind mount of one file
17110 system to another place in the same file system could not be
17111 detected as mount, since they shared struct stat's st_dev
17114 * We will now mount the cgroup controllers cpu, cpuacct,
17115 cpuset and the controllers net_cls, net_prio together by
17118 * nspawn containers will now have a virtualized boot
17119 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
17120 over with a randomized ID at container initialization). This
17121 has the effect of making "journalctl -b" do the right thing
17124 * The JSON output journal serialization has been updated not
17125 to generate "endless" list objects anymore, but rather one
17126 JSON object per line. This is more in line how most JSON
17127 parsers expect JSON objects. The new output mode
17128 "json-pretty" has been added to provide similar output, but
17129 neatly aligned for readability by humans.
17131 * We dropped all explicit sync() invocations in the shutdown
17132 code. The kernel does this implicitly anyway in the kernel
17133 reboot() syscall. halt(8)'s -n option is now a compatibility
17136 * We now support virtualized reboot() in containers, as
17137 supported by newer kernels. We will fall back to exit() if
17138 CAP_SYS_REBOOT is not available to the container. Also,
17139 nspawn makes use of this now and will actually reboot the
17140 container if the containerized OS asks for that.
17142 * journalctl will only show local log output by default
17143 now. Use --merge (-m) to show remote log output, too.
17145 * libsystemd-journal gained the new sd_journal_get_usage()
17146 call to determine the current disk usage of all journal
17147 files. This is exposed in the new "journalctl --disk-usage"
17150 * journald gained a new configuration setting SplitMode= in
17151 journald.conf which may be used to control how user journals
17152 are split off. See journald.conf(5) for details.
17154 * A new condition type ConditionFileNotEmpty= has been added.
17156 * tmpfiles' "w" lines now support file globbing, to write
17157 multiple files at once.
17159 * We added Python bindings for the journal submission
17160 APIs. More Python APIs for a number of selected APIs will
17161 likely follow. Note that we intend to add native bindings
17162 only for the Python language, as we consider it common
17163 enough to deserve bindings shipped within systemd. There are
17164 various projects outside of systemd that provide bindings
17165 for languages such as PHP or Lua.
17167 * Many conditions will now resolve specifiers such as %i. In
17168 addition, PathChanged= and related directives of .path units
17169 now support specifiers as well.
17171 * There's now a new RPM macro definition for the system preset
17174 * journald will now warn if it ca not forward a message to the
17175 syslog daemon because its socket is full.
17177 * timedated will no longer write or process /etc/timezone,
17178 except on Debian. As we do not support late mounted /usr
17179 anymore /etc/localtime always being a symlink is now safe,
17180 and hence the information in /etc/timezone is not necessary
17183 * logind will now always reserve one VT for a text getty (VT6
17184 by default). Previously if more than 6 X sessions where
17185 started they took up all the VTs with auto-spawned gettys,
17186 so that no text gettys were available anymore.
17188 * udev will now automatically inform the btrfs kernel logic
17189 about btrfs RAID components showing up. This should make
17190 simple hotplug based btrfs RAID assembly work.
17192 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
17193 (but not for its children which will stay at the kernel
17194 default). This should allow setups with a lot more listening
17197 * systemd will now always pass the configured timezone to the
17198 kernel at boot. timedated will do the same when the timezone
17201 * logind's inhibition logic has been updated. By default,
17202 logind will now handle the lid switch, the power and sleep
17203 keys all the time, even in graphical sessions. If DEs want
17204 to handle these events on their own they should take the new
17205 handle-power-key, handle-sleep-key and handle-lid-switch
17206 inhibitors during their runtime. A simple way to achieve
17207 that is to invoke the DE wrapped in an invocation of:
17209 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
17211 * Access to unit operations is now checked via SELinux taking
17212 the unit file label and client process label into account.
17214 * systemd will now notify the administrator in the journal
17215 when he over-mounts a non-empty directory.
17217 * There are new specifiers that are resolved in unit files,
17218 for the hostname (%H), the machine ID (%m) and the boot ID
17221 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
17222 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
17223 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
17224 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
17225 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
17226 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
17227 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
17231 * Support for reading structured kernel messages from
17232 /dev/kmsg has now been added and is enabled by default.
17234 * Support for reading kernel messages from /proc/kmsg has now
17235 been removed. If you want kernel messages in the journal
17236 make sure to run a recent kernel (>= 3.5) that supports
17237 reading structured messages from /dev/kmsg (see
17238 above). /proc/kmsg is now exclusive property of classic
17239 syslog daemons again.
17241 * The libudev API gained the new
17242 udev_device_new_from_device_id() call.
17244 * The logic for file system namespace (ReadOnlyDirectory=,
17245 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
17246 require pivot_root() anymore. This means fewer temporary
17247 directories are created below /tmp for this feature.
17249 * nspawn containers will now see and receive all submounts
17250 made on the host OS below the root file system of the
17253 * Forward Secure Sealing is now supported for Journal files,
17254 which provide cryptographical sealing of journal files so
17255 that attackers cannot alter log history anymore without this
17256 being detectable. Lennart will soon post a blog story about
17257 this explaining it in more detail.
17259 * There are two new service settings RestartPreventExitStatus=
17260 and SuccessExitStatus= which allow configuration of exit
17261 status (exit code or signal) which will be excepted from the
17262 restart logic, resp. consider successful.
17264 * journalctl gained the new --verify switch that can be used
17265 to check the integrity of the structure of journal files and
17266 (if Forward Secure Sealing is enabled) the contents of
17269 * nspawn containers will now be run with /dev/stdin, /dev/fd/
17270 and similar symlinks pre-created. This makes running shells
17271 as container init process a lot more fun.
17273 * The fstab support can now handle PARTUUID= and PARTLABEL=
17276 * A new ConditionHost= condition has been added to match
17277 against the hostname (with globs) and machine ID. This is
17278 useful for clusters where a single OS image is used to
17279 provision a large number of hosts which shall run slightly
17280 different sets of services.
17282 * Services which hit the restart limit will now be placed in a
17285 Contributions from: Bertram Poettering, Dave Reisner, Huang
17286 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
17287 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
17291 * When running in --user mode systemd will now become a
17292 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
17293 tree a lot more organized.
17295 * A new PartOf= unit dependency type has been introduced that
17296 may be used to group services in a natural way.
17298 * "systemctl enable" may now be used to enable instances of
17301 * journalctl now prints error log levels in red, and
17302 warning/notice log levels in bright white. It also supports
17303 filtering by log level now.
17305 * cgtop gained a new -n switch (similar to top), to configure
17306 the maximum number of iterations to run for. It also gained
17307 -b, to run in batch mode (accepting no input).
17309 * The suffix ".service" may now be omitted on most systemctl
17310 command lines involving service unit names.
17312 * There's a new bus call in logind to lock all sessions, as
17313 well as a loginctl verb for it "lock-sessions".
17315 * libsystemd-logind.so gained a new call sd_journal_perror()
17316 that works similar to libc perror() but logs to the journal
17317 and encodes structured information about the error number.
17319 * /etc/crypttab entries now understand the new keyfile-size=
17322 * shutdown(8) now can send a (configurable) wall message when
17323 a shutdown is cancelled.
17325 * The mount propagation mode for the root file system will now
17326 default to "shared", which is useful to make containers work
17327 nicely out-of-the-box so that they receive new mounts from
17328 the host. This can be undone locally by running "mount
17329 --make-rprivate /" if needed.
17331 * The prefdm.service file has been removed. Distributions
17332 should maintain this unit downstream if they intend to keep
17333 it around. However, we recommend writing normal unit files
17334 for display managers instead.
17336 * Since systemd is a crucial part of the OS we will now
17337 default to a number of compiler switches that improve
17338 security (hardening) such as read-only relocations, stack
17339 protection, and suchlike.
17341 * The TimeoutSec= setting for services is now split into
17342 TimeoutStartSec= and TimeoutStopSec= to allow configuration
17343 of individual time outs for the start and the stop phase of
17346 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
17347 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
17348 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
17349 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
17350 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
17351 Gundersen, Zbigniew Jędrzejewski-Szmek
17355 * The journal and id128 C APIs are now fully documented as man
17358 * Extra safety checks have been added when transitioning from
17359 the initial RAM disk to the main system to avoid accidental
17362 * /etc/crypttab entries now understand the new keyfile-offset=
17365 * systemctl -t can now be used to filter by unit load state.
17367 * The journal C API gained the new sd_journal_wait() call to
17368 make writing synchronous journal clients easier.
17370 * journalctl gained the new -D switch to show journals from a
17371 specific directory.
17373 * journalctl now displays a special marker between log
17374 messages of two different boots.
17376 * The journal is now explicitly flushed to /var via a service
17377 systemd-journal-flush.service, rather than implicitly simply
17378 by seeing /var/log/journal to be writable.
17380 * journalctl (and the journal C APIs) can now match for much
17381 more complex expressions, with alternatives and
17384 * When transitioning from the initial RAM disk to the main
17385 system we will now kill all processes in a killing spree to
17386 ensure no processes stay around by accident.
17388 * Three new specifiers may be used in unit files: %u, %h, %s
17389 resolve to the user name, user home directory resp. user
17390 shell. This is useful for running systemd user instances.
17392 * We now automatically rotate journal files if their data
17393 object hash table gets a fill level > 75%. We also size the
17394 hash table based on the configured maximum file size. This
17395 together should lower hash collisions drastically and thus
17396 speed things up a bit.
17398 * journalctl gained the new "--header" switch to introspect
17399 header data of journal files.
17401 * A new setting SystemCallFilters= has been added to services which may
17402 be used to apply deny lists or allow lists to system calls. This is
17403 based on SECCOMP Mode 2 of Linux 3.5.
17405 * nspawn gained a new --link-journal= switch (and quicker: -j)
17406 to link the container journal with the host. This makes it
17407 very easy to centralize log viewing on the host for all
17408 guests while still keeping the journal files separated.
17410 * Many bugfixes and optimizations
17412 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
17413 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
17414 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
17419 * Several tools now understand kernel command line arguments,
17420 which are only read when run in an initial RAM disk. They
17421 usually follow closely their normal counterparts, but are
17424 * There's a new tool to analyze the readahead files that are
17425 automatically generated at boot. Use:
17427 /usr/lib/systemd/systemd-readahead analyze /.readahead
17429 * We now provide an early debug shell on tty9 if this enabled. Use:
17431 systemctl enable debug-shell.service
17433 * All plymouth related units have been moved into the Plymouth
17434 package. Please make sure to upgrade your Plymouth version
17437 * systemd-tmpfiles now supports getting passed the basename of
17438 a configuration file only, in which case it will look for it
17439 in all appropriate directories automatically.
17441 * udevadm info now takes a /dev or /sys path as argument, and
17442 does the right thing. Example:
17444 udevadm info /dev/sda
17445 udevadm info /sys/class/block/sda
17447 * systemctl now prints a warning if a unit is stopped but a
17448 unit that might trigger it continues to run. Example: a
17449 service is stopped but the socket that activates it is left
17452 * "systemctl status" will now mention if the log output was
17453 shortened due to rotation since a service has been started.
17455 * The journal API now exposes functions to determine the
17456 "cutoff" times due to rotation.
17458 * journald now understands SIGUSR1 and SIGUSR2 for triggering
17459 immediately flushing of runtime logs to /var if possible,
17460 resp. for triggering immediate rotation of the journal
17463 * It is now considered an error if a service is attempted to
17464 be stopped that is not loaded.
17466 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
17468 * systemd-analyze now supports Python 3
17470 * tmpfiles now supports cleaning up directories via aging
17471 where the first level dirs are always kept around but
17472 directories beneath it automatically aged. This is enabled
17473 by prefixing the age field with '~'.
17475 * Seat objects now expose CanGraphical, CanTTY properties
17476 which is required to deal with very fast bootups where the
17477 display manager might be running before the graphics drivers
17478 completed initialization.
17480 * Seat objects now expose a State property.
17482 * We now include RPM macros for service enabling/disabling
17483 based on the preset logic. We recommend RPM based
17484 distributions to make use of these macros if possible. This
17485 makes it simpler to reuse RPM spec files across
17488 * We now make sure that the collected systemd unit name is
17489 always valid when services log to the journal via
17492 * There's a new man page kernel-command-line(7) detailing all
17493 command line options we understand.
17495 * The fstab generator may now be disabled at boot by passing
17496 fstab=0 on the kernel command line.
17498 * A new kernel command line option modules-load= is now understood
17499 to load a specific kernel module statically, early at boot.
17501 * Unit names specified on the systemctl command line are now
17502 automatically escaped as needed. Also, if file system or
17503 device paths are specified they are automatically turned
17504 into the appropriate mount or device unit names. Example:
17506 systemctl status /home
17507 systemctl status /dev/sda
17509 * The SysVConsole= configuration option has been removed from
17510 system.conf parsing.
17512 * The SysV search path is no longer exported on the D-Bus
17515 * The Names= option has been removed from unit file parsing.
17517 * There's a new man page bootup(7) detailing the boot process.
17519 * Every unit and every generator we ship with systemd now
17520 comes with full documentation. The self-explanatory boot is
17523 * A couple of services gained "systemd-" prefixes in their
17524 name if they wrap systemd code, rather than only external
17525 code. Among them fsck@.service which is now
17526 systemd-fsck@.service.
17528 * The HaveWatchdog property has been removed from the D-Bus
17531 * systemd.confirm_spawn= on the kernel command line should now
17534 * There's a new man page crypttab(5) which details all options
17535 we actually understand.
17537 * systemd-nspawn gained a new --capability= switch to pass
17538 additional capabilities to the container.
17540 * timedated will now read known NTP implementation unit names
17541 from /usr/lib/systemd/ntp-units.d/*.list,
17542 systemd-timedated-ntp.target has been removed.
17544 * journalctl gained a new switch "-b" that lists log data of
17545 the current boot only.
17547 * The notify socket is in the abstract namespace again, in
17548 order to support daemons which chroot() at start-up.
17550 * There is a new Storage= configuration option for journald
17551 which allows configuration of where log data should go. This
17552 also provides a way to disable journal logging entirely, so
17553 that data collected is only forwarded to the console, the
17554 kernel log buffer or another syslog implementation.
17556 * Many bugfixes and optimizations
17558 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
17559 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
17560 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
17561 Shawn Landden, Tom Gundersen
17565 * "systemctl help <unit>" now shows the man page if one is
17568 * Several new man pages have been added.
17570 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
17571 MaxLevelConsole= can now be specified in
17572 journald.conf. These options allow reducing the amount of
17573 data stored on disk or forwarded by the log level.
17575 * TimerSlackNSec= can now be specified in system.conf for
17576 PID1. This allows system-wide power savings.
17578 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
17579 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
17584 * logind is now capable of (optionally) handling power and
17585 sleep keys as well as the lid switch.
17587 * journalctl now understands the syntax "journalctl
17588 /usr/bin/avahi-daemon" to get all log output of a specific
17591 * CapabilityBoundingSet= in system.conf now also influences
17592 the capability bound set of usermode helpers of the kernel.
17594 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
17595 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
17596 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
17597 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
17601 * Note that we skipped 139 releases here in order to set the
17602 new version to something that is greater than both udev's
17603 and systemd's most recent version number.
17605 * udev: all udev sources are merged into the systemd source tree now.
17606 All future udev development will happen in the systemd tree. It
17607 is still fully supported to use the udev daemon and tools without
17608 systemd running, like in initramfs or other init systems. Building
17609 udev though, will require the *build* of the systemd tree, but
17610 udev can be properly *run* without systemd.
17612 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
17613 should be used to create dead device nodes as workarounds for broken
17616 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
17617 no longer supported. udev_monitor_new_from_netlink() needs to be
17618 used to subscribe to events.
17620 * udev: when udevd is started by systemd, processes which are left
17621 behind by forking them off of udev rules, are unconditionally cleaned
17622 up and killed now after the event handling has finished. Services or
17623 daemons must be started as systemd services. Services can be
17624 pulled-in by udev to get started, but they can no longer be directly
17625 forked by udev rules.
17627 * udev: the daemon binary is called systemd-udevd now and installed
17628 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
17629 to adapt to that, create symlink, or rename the binary after building
17632 * libudev no longer provides these symbols:
17633 udev_monitor_from_socket()
17634 udev_queue_get_failed_list_entry()
17635 udev_get_{dev,sys,run}_path()
17636 The versions number was bumped and symbol versioning introduced.
17638 * systemd-loginctl and systemd-journalctl have been renamed
17639 to loginctl and journalctl to match systemctl.
17641 * The config files: /etc/systemd/systemd-logind.conf and
17642 /etc/systemd/systemd-journald.conf have been renamed to
17643 logind.conf and journald.conf. Package updates should rename
17644 the files to the new names on upgrade.
17646 * For almost all files the license is now LGPL2.1+, changed
17647 from the previous GPL2.0+. Exceptions are some minor stuff
17648 of udev (which will be changed to LGPL2.1 eventually, too),
17649 and the MIT licensed sd-daemon.[ch] library that is suitable
17650 to be used as drop-in files.
17652 * systemd and logind now handle system sleep states, in
17653 particular suspending and hibernating.
17655 * logind now implements a sleep/shutdown/idle inhibiting logic
17656 suitable for a variety of uses. Soonishly Lennart will blog
17657 about this in more detail.
17659 * var-run.mount and var-lock.mount are no longer provided
17660 (which previously bind mounted these directories to their new
17661 places). Distributions which have not converted these
17662 directories to symlinks should consider stealing these files
17663 from git history and add them downstream.
17665 * We introduced the Documentation= field for units and added
17666 this to all our shipped units. This is useful to make it
17667 easier to explore the boot and the purpose of the various
17670 * All smaller setup units (such as
17671 systemd-vconsole-setup.service) now detect properly if they
17672 are run in a container and are skipped when
17673 appropriate. This guarantees an entirely noise-free boot in
17674 Linux container environments such as systemd-nspawn.
17676 * A framework for implementing offline system updates is now
17677 integrated, for details see:
17678 https://www.freedesktop.org/software/systemd/man/systemd.offline-updates.html
17680 * A new service type Type=idle is available now which helps us
17681 avoiding ugly interleaving of getty output and boot status
17684 * There's now a system-wide CapabilityBoundingSet= option to
17685 globally reduce the set of capabilities for the
17686 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
17687 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
17688 even CAP_NET_ADMIN system-wide for secure systems.
17690 * There are now system-wide DefaultLimitXXX= options to
17691 globally change the defaults of the various resource limits
17692 for all units started by PID 1.
17694 * Harald Hoyer's systemd test suite has been integrated into
17695 systemd which allows easy testing of systemd builds in qemu
17696 and nspawn. (This is really awesome! Ask us for details!)
17698 * The fstab parser is now implemented as generator, not inside
17701 * systemctl will now warn you if .mount units generated from
17702 /etc/fstab are out of date due to changes in fstab that
17703 have not been read by systemd yet.
17705 * systemd is now suitable for usage in initrds. Dracut has
17706 already been updated to make use of this. With this in place
17707 initrds get a slight bit faster but primarily are much
17708 easier to introspect and debug since "systemctl status" in
17709 the host system can be used to introspect initrd services,
17710 and the journal from the initrd is kept around too.
17712 * systemd-delta has been added, a tool to explore differences
17713 between user/admin configuration and vendor defaults.
17715 * PrivateTmp= now affects both /tmp and /var/tmp.
17717 * Boot time status messages are now much prettier and feature
17718 proper english language. Booting up systemd has never been
17721 * Read-ahead pack files now include the inode number of all
17722 files to pre-cache. When the inode changes the pre-caching
17723 is not attempted. This should be nicer to deal with updated
17724 packages which might result in changes of read-ahead
17727 * We now temporaritly lower the kernel's read_ahead_kb variable
17728 when collecting read-ahead data to ensure the kernel's
17729 built-in read-ahead does not add noise to our measurements
17730 of necessary blocks to pre-cache.
17732 * There's now RequiresMountsFor= to add automatic dependencies
17733 for all mounts necessary for a specific file system path.
17735 * MountAuto= and SwapAuto= have been removed from
17736 system.conf. Mounting file systems at boot has to take place
17739 * nspawn now learned a new switch --uuid= to set the machine
17740 ID on the command line.
17742 * nspawn now learned the -b switch to automatically search
17743 for an init system.
17745 * vt102 is now the default TERM for serial TTYs, upgraded from
17748 * systemd-logind now works on VT-less systems.
17750 * The build tree has been reorganized. The individual
17751 components now have directories of their own.
17753 * A new condition type ConditionPathIsReadWrite= is now available.
17755 * nspawn learned the new -C switch to create cgroups for the
17756 container in other hierarchies.
17758 * We now have support for hardware watchdogs, configurable in
17761 * The scheduled shutdown logic now has a public API.
17763 * We now mount /tmp as tmpfs by default, but this can be
17764 masked and /etc/fstab can override it.
17766 * Since udisks does not make use of /media anymore we are not
17767 mounting a tmpfs on it anymore.
17769 * journalctl gained a new --local switch to only interleave
17770 locally generated journal files.
17772 * We can now load the IMA policy at boot automatically.
17774 * The GTK tools have been split off into a systemd-ui.
17776 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
17777 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
17778 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
17779 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
17780 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
17781 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
17782 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
17783 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
17784 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
17789 * This is mostly a bugfix release
17791 * Support optional initialization of the machine ID from the
17792 KVM or container configured UUID.
17794 * Support immediate reboots with "systemctl reboot -ff"
17796 * Show /etc/os-release data in systemd-analyze output
17798 * Many bugfixes for the journal, including endianness fixes and
17799 ensuring that disk space enforcement works
17801 * sd-login.h is C++ compatible again
17803 * Extend the /etc/os-release format on request of the Debian
17806 * We now refuse non-UTF8 strings used in various configuration
17807 and unit files. This is done to ensure we do not pass invalid
17808 data over D-Bus or expose it elsewhere.
17810 * Register Mimo USB Screens as suitable for automatic seat
17813 * Read SELinux client context from journal clients in a race
17816 * Reorder configuration file lookup order. /etc now always
17817 overrides /run in order to allow the administrator to always
17818 and unconditionally override vendor-supplied or
17819 automatically generated data.
17821 * The various user visible bits of the journal now have man
17822 pages. We still lack man pages for the journal API calls
17825 * We now ship all man pages in HTML format again in the
17828 Contributions from: Dave Reisner, Dirk Eibach, Frederic
17829 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
17830 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
17835 * This is mostly a bugfix release
17837 * systems lacking /etc/os-release are no longer supported.
17839 * Various functionality updates to libsystemd-login.so
17841 * Track class of PAM logins to distinguish greeters from
17842 normal user logins.
17844 Contributions from: Kay Sievers, Lennart Poettering, Michael
17849 * This is an important bugfix release for v41.
17851 * Building man pages is now optional which should be useful
17852 for those building systemd from git but unwilling to install
17855 * Watchdog support for supervising services is now usable. In
17856 a future release support for hardware watchdogs
17857 (i.e. /dev/watchdog) will be added building on this.
17859 * Service start rate limiting is now configurable and can be
17860 turned off per service. When a start rate limit is hit a
17861 reboot can automatically be triggered.
17863 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
17865 Contributions from: Benjamin Franzke, Bill Nottingham,
17866 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
17867 Schmidt, Michał Górny, Piotr Drąg
17871 * The systemd binary is installed /usr/lib/systemd/systemd now;
17872 An existing /sbin/init symlink needs to be adapted with the
17875 * The code that loads kernel modules has been ported to invoke
17876 libkmod directly, instead of modprobe. This means we do not
17877 support systems with module-init-tools anymore.
17879 * Watchdog support is now already useful, but still not
17882 * A new kernel command line option systemd.setenv= is
17883 understood to set system wide environment variables
17884 dynamically at boot.
17886 * We now limit the set of capabilities of systemd-journald.
17888 * We now set SIGPIPE to ignore by default, since it only is
17889 useful in shell pipelines, and has little use in general
17890 code. This can be disabled with IgnoreSIPIPE=no in unit
17893 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
17894 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
17899 * This is mostly a bugfix release
17901 * We now expose the reason why a service failed in the
17902 "Result" D-Bus property.
17904 * Rudimentary service watchdog support (will be completed over
17905 the next few releases.)
17907 * When systemd forks off in order execute some service we will
17908 now immediately changes its argv[0] to reflect which process
17909 it will execute. This is useful to minimize the time window
17910 with a generic argv[0], which makes bootcharts more useful
17912 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
17913 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
17914 Mike Kazantsev, Ray Strode
17918 * This is mostly a test release, but incorporates many
17921 * New systemd-cgtop tool to show control groups by their
17924 * Linking against libacl for ACLs is optional again. If
17925 disabled, support tracking device access for active logins
17926 goes becomes unavailable, and so does access to the user
17927 journals by the respective users.
17929 * If a group "adm" exists, journal files are automatically
17930 owned by them, thus allow members of this group full access
17931 to the system journal as well as all user journals.
17933 * The journal now stores the SELinux context of the logging
17934 client for all entries.
17936 * Add C++ inclusion guards to all public headers
17938 * New output mode "cat" in the journal to print only text
17939 messages, without any meta data like date or time.
17941 * Include tiny X server wrapper as a temporary stop-gap to
17942 teach XOrg udev display enumeration. This is used by display
17943 managers such as gdm, and will go away as soon as XOrg
17944 learned native udev hotplugging for display devices.
17946 * Add new systemd-cat tool for executing arbitrary programs
17947 with STDERR/STDOUT connected to the journal. Can also act as
17948 BSD logger replacement, and does so by default.
17950 * Optionally store all locally generated coredumps in the
17951 journal along with meta data.
17953 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
17954 writing short strings to files (for usage for /sys), and for
17955 creating symlinks, character and block device nodes.
17957 * New unit file option ControlGroupPersistent= to make cgroups
17958 persistent, following the mechanisms outlined in
17959 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
17961 * Support multiple local RTCs in a sane way
17963 * No longer monopolize IO when replaying readahead data on
17964 rotating disks, since we might starve non-file-system IO to
17965 death, since fanotify() will not see accesses done by blkid,
17968 * Do not show kernel threads in systemd-cgls anymore, unless
17969 requested with new -k switch.
17971 Contributions from: Dan Horák, Kay Sievers, Lennart
17972 Poettering, Michal Schmidt
17976 * This is mostly a test release, but incorporates many
17979 * The git repository moved to:
17980 git://anongit.freedesktop.org/systemd/systemd
17981 ssh://git.freedesktop.org/git/systemd/systemd
17983 * First release with the journal
17984 https://0pointer.de/blog/projects/the-journal.html
17986 * The journal replaces both systemd-kmsg-syslogd and
17987 systemd-stdout-bridge.
17989 * New sd_pid_get_unit() API call in libsystemd-logind
17991 * Many systemadm clean-ups
17993 * Introduce remote-fs-pre.target which is ordered before all
17994 remote mounts and may be used to start services before all
17997 * Added Mageia support
17999 * Add bash completion for systemd-loginctl
18001 * Actively monitor PID file creation for daemons which exit in
18002 the parent process before having finished writing the PID
18003 file in the daemon process. Daemons which do this need to be
18004 fixed (i.e. PID file creation must have finished before the
18005 parent exits), but we now react a bit more gracefully to them.
18007 * Add colourful boot output, mimicking the well-known output
18008 of existing distributions.
18010 * New option PassCredentials= for socket units, for
18011 compatibility with a recent kernel ABI breakage.
18013 * /etc/rc.local is now hooked in via a generator binary, and
18014 thus will no longer act as synchronization point during
18017 * systemctl list-unit-files now supports --root=.
18019 * systemd-tmpfiles now understands two new commands: z, Z for
18020 relabelling files according to the SELinux database. This is
18021 useful to apply SELinux labels to specific files in /sys,
18022 among other things.
18024 * Output of SysV services is now forwarded to both the console
18025 and the journal by default, not only just the console.
18027 * New man pages for all APIs from libsystemd-login.
18029 * The build tree got reorganized and the build system is a
18030 lot more modular allowing embedded setups to specifically
18031 select the components of systemd they are interested in.
18033 * Support for Linux systems lacking the kernel VT subsystem is
18036 * configure's --with-rootdir= got renamed to
18037 --with-rootprefix= to follow the naming used by udev and
18040 * Unless specified otherwise we will now install to /usr instead
18041 of /usr/local by default.
18043 * Processes with '@' in argv[0][0] are now excluded from the
18044 final shut-down killing spree, following the logic explained
18046 https://systemd.io/ROOT_STORAGE_DAEMONS/
18048 * All processes remaining in a service cgroup when we enter
18049 the START or START_PRE states are now killed with
18050 SIGKILL. That means it is no longer possible to spawn
18051 background processes from ExecStart= lines (which was never
18052 supported anyway, and bad style).
18054 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
18055 reloading of units together.
18057 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
18058 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
18059 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
18060 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
18061 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek